Poly(ADP-ribose) Synthesis in Vitro Programmed by Damaged DNA

Poly(ADP-ribose) Synthesis in Vitro Programmed by Damaged DNA A COMPARISON OF DNA MOLECULES CONTAINING DIFFERENT TYPES OF STRAND BREAKS*

(Received for publication, March 14, 1980)

Robert C. Benjamin$ and D. Michael Gillg From the $Department of Biology, Harvard University, Cambridge, Massachusetts 02138 and the §Department of Molecular Biology and Microbiology, Tufts University, Boston, Massachusetts 021 11

The ability of DNA to support poly(ADP-ribose) synthesis is completely dependent upon the number and type of strand breaks it contains and is independent of the sequence. Single-stranded DNA is ineffective. Co- valently closed circular plasmid DNA is ineffective, but when it is enzymatically digested it activates poly(ADP-ribose) polymerase in proportion to the number of strand breaks, suggesting that the polymerase recognizes DNA ends. Double-stranded restriction fragments with flush ends are approximately 3 times more effective than are fragments with unpaired nucleotides extending from the 3‘ termini and about 10 times more effective than are either fragments with unpaired nucleotides extending from the 5’ termini or plasmids with single-strand breaks. All types of restriction fragments become more effective upon removal of terminal 5”phosphate groups. This specificity profile may relate to the proposed role of poly(ADP-ribose) synthesis in the repair of DNA strand breaks, for those which are assumed to be more difficult to repair in vivo are the more effective stimulators. Poly(ADP-ribose) polymerase has no divalent cation requirement when supplied with flush-ended DNA fragments, but magnesium may enhance the effectiveness of other types of DNA by activating magnesium-dependent nucleases.

Ineffective DNAs, such as covalently closed plasmids or synthetic homopolymers that are unable to form Watson-Crick duplexes, apparently compete with effective DNA and weakly inhibit poly(ADP-ribose) synthesis.

Since the discovery of poly(ADP-ribose), it has been known that DNA is involved in the synthesis of the polymer from NAD+ (1, 2). Thus, complete digestion of the endogenous DNA in nuclei (2,3) or ghost cells (4) prevents any subsequent formation of poly(ADP-ribose). The synthetic enzyme, poly(ADP-ribose) polymerase, can be isolated from nuclei and, when free of nucleic acids, is dependent upon added DNA for activity. With such a system, it is possible to ask what sort of DNA is most effective, and there have been several attempts to compare the stimulating activities of different types of polynucleotides. Initially, the results were not particularly illuminating. For example, it was reported that the activity of rat liver poly(ADP-ribose) polymerase was stimulated more effectively by total rat liver DNA than by calf thymus DNA. DNAs from Escherichia coli and T4 were less effective than was either mammalian DNA, but +X174 DNA was said to be as effective as that derived from calf

* 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.

thymus ( 5 ) . Caplan showed, in fact, that nucleotide sequence was probably of little importance, for unique and repeated sequence DNA fragments of the same average size were equally effective. He found, rather, that the degree of fragmentation was crucial: shearing of DNA increased its effectiveness approximately in proportion to the number of breaks introduced, as if the poly(ADP-ribose) polymerase responded only to strand ends (6). We reached a similar conclusion in the previous paper (7) where, in agreement with other reports (8-12), we demonstrated that there were increased rates of poly(ADP-ribose) synthesis in permeabdized cells after the introduction of breaks into the nuclear DNA.

To examine this issue more closely, and in particular to determine the nature of the break or breaks to which poly(ADP-ribose) polymerase responds, we have prepared poly(ADP-ribose) polymerase free of DNA and programmed its activity with plasmid DNA in a covalently closed form or digested with a variety of restriction enzymes and other endonucleases that provided a range of defined fragments of known structure. We have confirmed that the DNA requirement is in fact a requirement for double-stranded DNA containing specific types of strand breaks and that the previously observed differences in effectiveness of DNA of various types were probably due to variations in the incidence of these breaks between the DNA preparations.

EXPERIMENTAL PROCEDURES

Materials-DNase I, poly(dA), poly(dT), poly(dC) .poly(dG), E. coli DNA and tRNA, calf thymus DNA, and calf thymus histones fl, Ea, ab, and f3 were obtained from the Sigma Chemical Co. Restric- tion endonucleases and polynucleotide kinase were from New England Biolabs, and the bacterial alkaline phosphatase was from Worthing- ton. Double-stranded viral RNA isolated from Penicillium chryso- genum was generously provided by Tim Hunt and Richard Jackson of the Department of Biochemistry, University of Cambridge. Neo- carzinostatin was obtained from the Sidney Farber Cancer Institute.

Isolation of Superhelical pBR322 DNA-E. coli K12 RR1 F- pro leu thi lacy stprk- mkC containing the plasmid pBR322 (13) was grown at 34°C in medium that consisted, in grams per liter, of NHICI, 2.0; Na2HP04, 6.0; NaC1, 10.0; MgS04. 7&0,0.2; casamino acids, 40.0; CaC12.2H20, 0.5; and glucose, 6.0. An overnight culture in medium containing 10 Fg/ml of tetracycline was diluted 50-fold into medium without tetracycline. When the absorbance (550 nm) reached 0.8, chloramphenicol was added to 170 p g / d and the incubation was continued for an additional 18 h.

Plasmids were isolated at 0 4 ° C by the method of Tanaka and Weisblum (14). Cells from 1 liter were harvested by centrifugation, resuspended in 10 ml of 25% sucrose, 50 mM Tris-HCI, pH 8.0, and mixed with 2 ml of 5 mg/ml lysozyme. Following incubation for 5 min at 0-4”c, 4 ml of 0.25 M Na2EDTA, pH 8.0, was added and the mixture was incubated a further 5 min. Five milliliters of 5 M NaCl and 2 ml of 10% sodium dodecyl sulfate were added in succession followed by immediate dispersion by use of a mechanical Vortex mixer. The mixture was allowed to stand on ice for 3 h and was centrifuged at 30,000 X g for 30 min. The supernatant was mixed with 0.6 volume of isopropyl alcohol and placed at -70°C for 20 min. The

10502

Poly(ADP-ribose) and Specific DNA Strand Breaks 10503

precipitate that formed was collected by centrifugation, resuspended in 7.5 ml of 1 mM EDTA, 10 mM Tris-HC1, pH 8.0, and cleared of insoluble material by centrifugation. To the supernatant were added 7.65 g of CsCl plus 0.3 ml of 10 mg/ml ethidium bromide. Aliquots were centrifuged at 90,OOO X g for 60 h and the position of the superhelical plasmid DNA in the centrifuge tubes was determined by viewing briefly under ultraviolet illumination. The plasmid supercoil bands of several tubes were removed with a hypodermic needle punched through the sides of the tubes, combined, and recentrifuged at 90,OOO x g for 60 h to remove any contaminating relaxed or linear forms. This time, fractions were collected in very dim light and those containing supercoiled forms were identified by agarose gel electrophoresis of small portions. Ethidium bromide was removed by extracting three times with 3 volumes of isoamyl alcohol saturated with cesium chloride. Cesium chloride was removed by dialysis against 1 mM EDTA, 10 mM Tris-HC1, pH 8.0. The product was stored at 0- 4'C. It contained fewer than 1% linear and relaxed forms as measured by agarose gel electrophoresis and electron microscopy. No tRNA was detectable by electrophoresis on agarose or polyacrylamide gels.

Digestion ofpBR322 with Restriction Endonucleases-Restriction endonuclease digestions of pBR322 were performed in a standard buffer (9 m MgC12, 1 mM dithiothreitol, 10 mM Tris-HC1, pH 7.5) with salt concentrations for each enzyme as follows: none, Hue 11, Hae 111, Taq I, and Hpa 11; 50 m NaC1, HincII, Alu I, Pst I, Bum HI, Hinf, and Hha I; 100 m~ NaCl, Eco RI and Aua I; and 150 mM NaC1, Sal I. Ten-microgram amounts of pBR322 in 50 pI of buffer were incubated at 37°C for 60 min with 30 units of the appropriate restriction endonuclease (1 unit being the amount of enzyme required to degrade 1.0 pg of X-DNA in 60 min at 37°C). Taq I digestion was carried out at 55°C. Digestions involving more than one endonuclease were performed sequentially in order of increasing NaCl optima, or simultaneously when the optima were identical. Reactions were terminated by the addition of 50 p1 of 15 m~ EDTA, pH 8.0. In each case, electrophoresis of the product on an 8% polyacrylamide slab gel

a t 4OC. (15) showed that the digestion was complete. The digests were stored

Alkaline Phosphatase Treatment of Restriction Fragments-Ten micrograms of Hae 111-, Hpa 11-, or Hha I-digested pBR322 were incubated in 100 pl of 1 mM EDTA, 10 mM Tris-HC1, pH 8.0, at 37°C for 3 h with 5 units of bacterial alkaline phosphatase. The reaction was terminated by extracting twice with 100 p1 of water-saturated phenol. Samples were then extracted with 100 pl of diethyl ether, ethanol-precipitated, and stored in 1 m~ EDTA, 10 mM Tris-HC1, pH 8.0, a t 4°C.

Polynucleotide Kinase Treatment of Restriction Fragments-Five micrograms of phosphatase-treated restriction digests were each incubated in 50 p1 of 10 mM MgCL, 5 mM dithiothreitol, 1 m~ spermine, 8 mM ATP, 50 mM Tris-HC1, pH 8.0, for 30 rnin at 37'C with 2 units of polynucleotide kinase. The kinase reaction was terminated by phenol extraction and the DNA was recovered as described for the phosphatase reaction.

Poly(ADP-ribose) Polymerase-A 5-g portion of minced calf thymus was homogenized in 15 ml of ice-cold 0.25 M sucrose using a Potter-Elvehjem homogenizer with a Teflon pestle. One-third volume of 2 M NaCl was added to the homogenate and the mixture was shaken vigorously four times during the next 15 min. Following centrifugation (l00,OOO X g, 60 min), 10 ml of the supernatant was fractionated on a column of DEAE-cellulose (Fig. 1). Poly(ADP- ribose) polymerase eluted among the early fractions, ahead of the major protein peak. Providing that the column was not heavily loaded, little or no DNA was present in the enzyme peak and poly(ADP-ribose) synthesis was greatly dependent upon added DNA. Fractions whose ability to synthesize the polymer depended most strongly on exogenous DNA (Fig. 1, arrows) were pooled and stored in 25% glycerol at -70°C. Such a preparation retains more than 80% of its synthetic activity for 12 weeks.

It is possible to recover a second fraction containing poly(ADP- ribose) polymerase by eluting with 0.5 M NaCl, 10 mM potassium phosphate, pH 7.2. This fraction contains endogenous DNA fragments, and polymer synthesis is only slightly dependent upon added DNA (data not shown). We suppose that in the original extract this portion of the enzyme is bound to the DNA which in turn binds to the column matrix.

Measurement of Poly(ADP-ribose) Synthesis-In our standard assay, 50-pl samples of the DE52 pool were incubated at 27°C for 3 min in a total volume of 100 pl containing 50 p~ [a-32P]NAD' (ICN; 10 to 30 Ci/mmol), 2 mM EDTA, and 10 mM Tris-HC1, pH 8.0. Unless otherwise stated, DNA was present at 1 pg/ml. Trichloroacetic acid-

21

31

I so

I \ 600

400

="-+ I so

FIG. 1. Preparation of poly(ADP-ribose) polymerase dependent on exogenous DNA. Ten milliliters of a 100,OOO X g salt extract of calf thumus was applied to a column (2 X 30 cm) of DE52 equilibrated in 10 mM potassium phosphate, pH 7.2. The column was eluted with the same buffer and fractions were assayed for poly(ADP- ribose) polymerase with (0) or without (0) 1 pg/ml of Hae 111- digested pBR322. Arrows indicate the fractions that were pooled and used for subsequent experiments. The pool was stored in 25% glycerol at -70°C. The final protein concentration was 100 pg/ml. Sodium chloride is slightly retarded by the column, and that applied with the sample elutes with the major protein peak (upperpanel). The material eluting after 110 ml is not precipitated by 10% trichloroacetic acid and is presumed to be nucleotides and other small molecules.

insoluble material was collected by filtration and i ts radioactivity was determined by liquid scintillation counting.

RESULTS

Poly(ADP-ribose) Synthesis by a Cell-free Extract of Calf Thymus-As reported earlier, a 0.5 M NaCl extract of calf thymus contains both DNA fragments and poly(ADP-ribose) polymerase. The DNA is necessary for the activity of the polymerase, and poly(ADP-ribose) synthesis may be reduced or abolished by deoxyribonuclease digestion of the extract (16). There is usually sufficient endogenous DNA to saturate the enzyme and the addition of DNA generally produces little or no increase in the rate of poly(ADP-ribose) synthesis.

However, passage of this high salt extract of calf thymus over a DEAE-cellulose column separates a fraction of the poly(ADP-ribose) polymerase that is almost completely dependent on exogeneous DNA for activity (Fig. l). The rate of polymer synthesis is increased more than 40-fold by the addition of 1 pg/ml of pBR322 digested with Hae I11 endonuclease. This DNA-stimulated activity has a K,,, for NAD' of 40 p ~ , a temperature optimum of 35"C, and a pH optimum of 8.0. Poly(ADP-ribose) synthesis by this preparation does not require any divalent cations as cofactors and is not reduced by the inclusion of either 2 m~ EGTA' or EDTA. This point is discussed in more detail below, for it contrasts with some previous observations. Synthesis is inhibited by thymidine and nicotinamide, with K, values of 25 p~ and 15 PM, respectively, and is reduced 66% by 100 m~ NaC1. Poly(ADP-ribose) continues to accumulate for more than 60 min both at 25" and 37°C and, in contrast to the ephemeral poly(ADP-ribose) synthesized in ghost cells (4, 7), the polymer synthesized by the soluble enzyme is reasonably stable, although some turn- over is evident. Following termination of synthesis by the addition of nicotinamide or of Neurospora NAD' glycohydro- lase, the previously synthesized material is degraded with a half-life of over 90 min at 27°C.

ethyl ether)N,N,N',N'-tetraacetic acid. ' The abbreviation used is: EGTA, ethylene glycol bis(p-amino-

10504 Poly(ADP-ribose) and Specific DNA Strand Breaks

)Histones I

I

A B FIG. 2. Analysis of [:“PP]NAD+-derived products by 7 to 15%

polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate. I’oly(AI)I’-ribose) was synthesized as described under “Experimental Procedures” with 1 pg/ml of Hae 111-digested pRH322. An autoradiogram of the gel is shown. The top segment represents the stacking gel (5% acrylamide) and the second segment is the portion of the 40-cm main gel that was cut off before drying and placed separately on the gel drier. Oligomers of up to about 10 to 12 repeat units migrate with the dye front. Above this, 60 bands are resolved representing chain lengths of up to about 70 repeat units. Chains longer than 70 units can be distinguished on longer gels but are unresolved here. The very large material migrating only I to 2 cm into the stacking gel and that migrating 1 to 2 cm into the main gel each generally comprised about 15 to 20% of the total acid-insoluble product. The electrophoretic pattern is unaffected by pretreatment with DNase, RNase, pronase, or trypsin, but the product is digested completely by snake venom phosphodiesterase. Some relatively in- tense bands can be seen at positions occupied by DNA restriction fragments. as indicated by the bars at left. The particular intensifi- cation pattern is a function only of the DNA present at the time the gel is run and appears not to depend upon an enzymatic activity. The location of the restriction fragments is unchanged by the apparent interaction with free poly(ADP-ribose) and the basis for this phenom- enon is a t present unclear. A, no histone; B, with 5 pg/ml of calf thymus histone. Histones H2a, H2b, H3, and H4 migrate in the region indicated and are not labeled.

Nature of the Product-Polyacrylamide gel analysis in the presence of sodium dodecyl sulfate reveals that the NAD’- derived product is principally composed of free polymers of ADP-ribose, the majority being over 50 residues in length, and does not contain significant amounts of ADP-ribosylated protein (Fig. 2). This is in contrast to the NAD’-derived material formed by HeLa cell ghosts (4, 7) which, in addition to free polymer, contains a variety of ADP-ribosylated proteins, including all five classes of histones. Apparently, the small amount of histone H1 present in the DE52 pool is not, in general, covalently modified with either single residues or polymers of ADP-ribose when incubated with NAD’. Further, additional H1 or other calf thymus histones do not enhance

the polymerase activity; in fact, they are slightly inhibitory. This suggests that free polymer can be formed without the involvement of histones and raises the possibility that the covalent complexes of histone and poly(ADP-ribose) frequently observed in nuclei and ghost cells may actually be formed subsequent to the polymer itself.

Specificity of the DNA Requirement: Single- a n d Double- stranded DNA-The low background incorporation of the calf thymus poly(ADP-ribose) polymerase preparation made it possible to study the DNA requirement in detail. DNA from almost any source is effective to some degree and, at low concentrations a t least, the rate of poly(ADP-ribose) synthesis increases linearly with DNA concentration (Fig. 3 and 4) and the net amount of polymer increases linearly with time (Fig. 5). RNA is totally unable to substitute for DNA, as neither E. coli tRNA nor double-stranded viral RNA has any effect.

The poly(ADP-ribose) polymerase exhibits a very strong preference, and possibly an absolute requirement, for double- stranded DNA. When heat-denatured and rapidly cooled calf thymus DNA was assayed immediately, it was found to be about 15% as effective as was the native DNA (Fig. 3), approximately corresponding to the amount of rapidly reannealing sequences present. Heat-denatured E. coli DNA was about 10% as effective as was the same preparation before heating. The residual effect might be due to short, rapidly reannealing, self-complementary sequences rather than to single-stranded sequences per se. For a more stringent test of the effectiveness of single-stranded DNA, therefore, we chose to use synthetic homopolymers which could form no Watson- Crick duplex structures at all. Poly(dA) and poly(dT) were individually totally ineffective. However, a mixture of the two homopolymers, which forms duplex DNA, was extremely effective in supporting poly(ADP-ribose) polymerase activity (Table I).

Supercoils and Single-strand Breaks-In the previous paper (4), we showed that the poly(ADP-ribose) polymerase activity of ghost cells was increased markedly by agents that damage DNA. We also knew that fragmentation of calf thymus DNA increased its ability to support synthesis in extracts (6). Therefore, we wished to study the effects of particular types of strand breakage by providing DNA with defined modifications. For this purpose, it is not useful to employ cellular DNA because of its complexity and the large but variable and unknown amount of strand breakage incurred during its isolation. We therefore chose to work with the E. coli plasmid pBR322. This can be readily isolated as a covalently closed, double-stranded supercoil of 4362 base pairs. The sequence, and thus the number of cutting sites of each restriction enzyme, are known (13).

A particularly significant result is that pBR322 supercoils, even a t high concentrations, are totally unable to support poly(ADP-ribose) synthesis (Fig. 4). This is the first ineffective double-stranded DNA to be found. The result suggests either that there is an obligate requirement for strand breaks or that supercoiling prevents the DNA from stimulating poly(ADP-ribose) synthesis. Upon the introduction of one or two single-strand breaks by limited DNase I digestion, the DNA becomes able to support poly(ADP-ribose) synthesis. Significantly, the effectiveness of the DNA further increases in direct proportion to the number of single-strand breaks introduced (Fig. 5). Thus, breaks seem to be the determining factor, not relaxation of supercoils.

I t is necessary to allow for the contribution from double- strand breaks since, as we show in the next section, some of them can be more than 10 times more effective than are single-strand breaks in supporting poly(ADP-ribose) synthesis. An upper limit to the number of double-strand breaks was


- 0 Calf Thymus DNA E,

c u 200-

0 a 0

c V

L

-

DNA )Ig/rnl DNA pg/ mt Minutes FIG. 3 (left) . Poly(ADP-ribose) polymerase activity as a function of thymus DNA concentration. 0, native calf thumus DNA; 0,

FIG. 4 (center). Poly(ADP-ribose) polymerase activity as a function of the concentration of restriction fragments. Uncut pBH322

FIG. 5 (right). Kinetics of poly(ADP-ribose) synthesis supported by endonuclease digests of pBR322. No added DNA (O), 1 pgl

heat-denatured calf thymus DNA.

(V) and pBR322 digested with Eco RI ( 0 , Hpa I1 (A), or Hae 111 (0).

mi of uncut pBR322 (V), or 1 pg/rnl of pBR322 digested with E m RI (O), Hpa I1 (A), or Hue I11 (0).

TABLE I Poly(ADP-ribose) polymerase activity supported by

deoxyribonucleotide homopolymers DNA added ADP-ribose

pmol 1.9 2.4 2.4 2.0 1.2 3.8

180.0 110.0

estimated from the number of linear plasmid forms present (see the legend to Fig. 6). Fewer than 1% of the cuts introduced into pBR322 by DNase I were double-strand breaks, and even if all of these had been of the most effective type, namely flush-ended, they would have accounted for less than 10% of the activity observed. We must therefore conclude that DNA with single-strand breaks is truly able to stimulate poly(ADP- ribose) polymerase.

Other agents that introduce single-strand breaks into DNA have qualitatively similar results. Treatment of the plasmid with neocarzinostatin, an endonuclease isolated from Strep- tomyces carzinostaticus (18, 19), increases its ability to support polymer synthesis. Although this increase was generally similar to that produced by DNase I treatment, gel analysis of the digest revealed a rather higher proportion of linear pBR322 molecules, namely about 5 double-strand breaks per 100 single-strand breaks introduced. Large doses of x-rays also produce many relaxed circular forms from pBR322 and increase its ability to support poly(ADP4bose) synthesk2 However, x-irradiation produces a wide variety of structures at the strand termini (20), as well as a rather large number of linear plasmid forms, thus making i t difficult to assess the importance of particular modifications.

Stimulation by Double-strand Breaks-The introduction of double-strand breaks into the plasmid DNA also markedly increases its ability to support poly(ADP-ribose) synthesis. The effect rises in proportion to the number of double-strand breaks introduced per pIasmid moIecuIe, provided we compare

A,". Storniolo and R. C. Benjamin, unpublished observations.

l I I I I I )

0 a :: 1 2 3 4 5 6

Single-strand BreakdPlarrnid 4

FIG. 6. Stimulation of poly(ADP-ribose) polymerase activity by DNA containing single-strand breaks. SingIe-strand breaks were introduced into pBR322 by partial digestion with bovine pan- creatic DNase I, which predominantly produces single-strand breaks when magnesium is provided as the only divalent cation (17). Three hundred micrograms of "H-labeled pBR322 (1600 cpmlpg) were incubated at 27°C with 10 mM MgC12, 2 mM EGTA, 50 pg/ml of bovine serum albumin, 10 mM Tris-HC1, pH 8.0, and 0.2 pg/ml of DNase I (pretreated with 2 m~ EGTA) in a total volume of 500 pl. Sixty- microliter samples were removed after 0, 1, 2, 3, 4, 6, 10, and 15 min, and were extracted once with water-saturated phenol, and twice with diethyl ether. EDTA was then added to 15 mM and samples were stored at 0 4 ° C . Ten-microliter aliquots were analyzed by agarose gel electrophoresis (2% agarose, 60 V for 16 h). The gel was stained with ethidium bromide and viewed under ultraviolet illumination. Regions containing supercoiled, relaxed, and linear plasmid forms were ex- cised, extracted overnight with a mixture of 3% Protosol in Aquasol, and counted. The number of supercoiled plasmid forms decreased exponentially with time of DNase I digestion. After 10 min, over 98% of the former supercoils were present as relaxed plasmid circles. If we assume that supercoiled and relaxed circles are equally susceptible to DNase I nicking, so that the number of cuts per plasmid molecule forms a Poisson distribution, we can calculate the average number of single-strand breaks per plasmid from the fraction of surviving supercoiled forms. The calculated values, 0.00, 0.35, 0.68, 1.18, 1.57, 2.65, 4.05, and 6.10 single-strand breaks per plasmid, represent a linear increase in the average number of breaks introduced with time of incubation. The ability of 10 pg/ml of the pBR322, which is 10 times the usual DNA concentration, to stimulate poly(ADP-ribose) synthesis also increased linearly with time of DNase I digestion, suggesting that the assumption is valid. We determined the incidence of double- strand breaks by measuring the proportion of linear forms. There were always over 100 times more single-strand breaks than double- strand breaks produced under these conditions.


DNA fragments with similar types of ends (Fig. 7). The addition of 1 p g / d of pBR322 that has been cut into 10 fragments with Hinf has an effect equal to 10 p g / d of the plasmid cut once by Eco RI. Thus, the synthetic activity is not a measure of the concentration of DNA per se, but of the concentration of DNA ends.

In addition to the number of fragments produced, the structure of the DNA ends also determines to a great extent the effectiveness of a given digest in supporting poly(ADP- ribose) synthesis. Restriction enzymes can be classified into three groups according to the structure of the DNA ends they produce. One class cuts both strands of the DNA duplex at the same position, producing DNA fragments with flush ends. The other two classes cut one strand 2 to 4 base pairs distant from the break in the complementary strand, thereby gener- ating DNA fragments with 2 to 4 unpaired nucleotides extending from either the 3’ or 5’ strands. Although DNA fragments of all types are able to support poly(ADP-ribose) synthesis to some degree, fragments with flush ends, such as those produced by Hue 111, are significantly more effective than are those with unpaired nucleotides at the termini. The

”’1 200 -

- 0

E U V c

z E 100- 0

0 0 t - u n 0 n .- L I

n. 0 a

IO 20 30 40

Flush Double-strandcuts Per Plasmid

3’ & b b c c

n n ‘9

5 ‘ . . . . . . . . e ;Ir

Restriction Enzymes Used FIG. 7. Stimulation of poly(ADP-ribose) synthesis by re-

striction endonuclease digests of E. coli plasmid pBR322 whose fragments have flush ends 0, unpaired nucleotides attached to the 3’ termini (A), or unpaired nucleotides attached to the 5’ termini (0). The restriction enzymes used, with the number of restriction sites per plasmid given in parentheses, were as follows. Fragments with flush ends: a, HincII (2); 6, Alu I (16); and c, Hue 111 (21). DNA fragments with 2 to 4 unpaired nucleotides extending from the 3’ termini: d, Pst I (1); e, Hue I1 (11); and f , Hha I(311. Fragments with unpaired nucleotides extending from the 5’ termini: g, Eco RI (1); h, Aua I (1); i, Bum HI (l);;, Sal I (1); k, Tuy I (7); 1, Hinf (10); and m, Hpa I1 (26). The combinations of restriction enzymes used to provide other numbers of cuts per plasmid are listed below the figure. Background incorporation (0). that not dependent on exogenous DNA, was 1 to 2 pmol and has not been subtracted from other points.

presence of only 2 unpaired nucleotides extending from the 5‘ terminus reduces the ability of a fragment to support poly(ADP-ribose) synthesis by almost 90% (compare the Hinf plus Taq I to A b I digests in Fig. 7). Increasing the number of unpaired bases to 3 or 4 has no additional effect. Thus, the effectiveness of Hpa 11, Hinf, and Eco RI digests (which are composed of DNA fragments with 2, 3, and 4 unpaired nucleotides, respectively) are proportional simply to the number of fragments produced.

The DNA strand to which the unpaired nucleotides are attached is important. DNA fragments with unpaired nucleotides extending from the 5’ strand are only about one-third as effective as are those with nucleotides extending from the 3‘ strand. It should be noted that a double-strand break of the least effective variety, such as that produced by Eco RI, has a slightly greater effect than does a single-strand break discussed in the previous section.

There is no apparent effect of the terminal nucleotide sequence. This can best be seen by comparing restriction digests having the same types of ends but cut at different sequences. Thus, the lower curve in Fig. 7 represents DNA fragments with 5’-nucleotide extensions terminating in CTTAA, GGC, GGGCT, CCTAG, CAGCT, AGC, and CTNA (generated by digestion with Eco RI, Hpa 11, Ava I, Bum HI, Sal I, Tag I, and Hinf, respectively), all of which seem to be about equally effective per fragment. These fragments have 3“terminal bases of G, C, C, G, G, T, and G, respectively. Likewise, the flush-ended fragments in the upper curve with 5”terminal sequences of GG, AG, and GTC (produced by Hue 111, Alu I, and HincII) all seem about equally effective.

The poly(ADP-ribose) synthesis in response to different types of DNA strand breakage, both single- and double-strand breaks, appears to be the result of a single enzymatic activity. Thus, although the stimulatory effect of the different DNA preparations is approximately additive at low rates of poly(ADP-ribose) synthesis, the addition of 10 p g / d of circular pBR322, containing an average of five single-strand breaks per plasmid, does not increase the maximum rate of polymer synthesis supported by saturating levels of Hue I11 fragments.

Removal of the Terminal 5‘-Phosphate Group-The abilities of all three types of restriction fragments to support poly(ADP-ribose) synthesis is improved by the removal of their terminal 5’-phosphate groups with bacterial alkaline phosphatase (Table 11). Hue I11 and Hha I digests of pBR322 become approximately twice as effective, while Hpa I1 fragments are improved by nearly 5-fold. Rephosphorylation of the 5’ ends, using polynucleotide kinase, restores the original, lower abilities of the fragments.

Polynucleotide Inhibitors of Poly(ADP-ribose) Polymer- use-DNAs which are unable to activate poly(ADP-ribose) polymerase are frequently capable of inhibiting polymer synthesis supported by other DNAs. Poly(ADP-ribose) synthesis supported by 1 p g / d of Hue 111-digested pBR322 is reduced

TABLE I1 The effect of the terminal 5’-phosphate group on the ability of restriction fragments ofpBR322 to support poly(ADP-ribose)

polymerase activity Terminal 5’-phosphate groups were removed by alkaline phospha-

tase treatment and were subsequently replaced by polynucleotide kinase and ATP as described under “Experimental Procedures.”

Digest Native P h ~ ~ ~ ~ ~ ~ - Kinase-treated

pmol ADP-ribose Hue I11 74.3 180.0 69.4 Hpa I1 10.0 43.9 8.6 Hha I 42.3 76.3 43.5


50% by 50 p g / d of uncut pBR322 or 25 p g / d of poly(dA). We suppose that single-stranded and covalently closed, circular, double-stranded DNAs are capable of weakly interacting with a DNA-binding site of the poly(ADP-ribose) polymerase without activating the enzyme. We have noticed a different effect with poly(dT), which inhibits the synthesis by 50% at a concentration of only 1.7 p g / d . We attribute this to the thymine residues, since many other thymine derivatives are potent inhibitors of poly(ADP-ribose) polymerase. Thy- midine itself has a K, of 6 p g / d (25 p ~ ) . In order to have an inhibitory effect, the thymine bases of the poly(dT) must not be involved in base pairing, for an excess of poly(dA) produces a duplex molecule which is a highly effective activator of poly(ADP-ribose) polymerase (Table I). RNAs, which are similarly unable to support poly(ADP-ribose) synthesis, also act as weak inhibitors.

Effect of Magnesium Ions on Poly(ADP-ribose) Synthesis-The poly(ADP-ribose) polymerase in the calf thymus preparation does not require any divalent cations for its activity. However, the addition of magnesium to the incubation mixture may increase or decrease the rate of polymer synthesis depending upon the type of DNA fragment used. These effects are probably caused by the presence of low levels of magnesium-dependent exonucleases, which intercon- vert the various types of DNA ends and, regardless of the initial composition, produce a mixture of flush and staggered ends. If the DNA fragments provided have flush ends, the type most effective in eliciting poly(ADP-ribose) polymerase activity, the inclusion of magnesium is inhibitory, presumably because many of these ends are converted to the staggered varieties. On the other hand, if the ends of the DNA fragments have unpaired bases extending from the double-stranded region, a less effective type of end, then magnesium has a slightly positive effect which increases somewhat with time of incubation (Fig. 8). In this case, the removal of a few bases from the double-stranded portion of the ends would only slightly increase the number of unpaired bases and have little effect on the ability of the fragments to elicit polymer synthesis, but the occasional production of a flush end by the removal of the unpaired bases would be stimulatory. Such production of flush ends may account, at least in part, for the stimulatory effect of magnesium that has been frequently reported for poly(ADP-ribose) synthesis primed with DNA broken by random shearing, nicking, or exonuclease action (21, 22), the ends of which are probably mostly staggered. All incubations described in this paper were done in the presence of 2 mM EDTA to eliminate the effects of such nucleases.

- I n r H~I I I+EDTA ! I

-ribose) Minutes

FIG. 8. The effect of magnesium ions on poly(ADP oolvmerase activitv. Svnthesis supported by 0.2 pg/ml of Hae III-

a 11-digested pBR322 (El, .) in the presence bf 1 m~ EDTA (0,O) or 10 mM MgCL (0, .).

DISCUSSION

We have now shown that although DNA is an absolute requirement for poly(ADP-ribose) polymerase activity, not all types of DNA are capable of supporting polymer synthesis. In order to be effective, the DNA must contain certain specific structural features. First, it must be double-stranded. The homopolymers poly(dA) and poly(dT) are totally ineffective until combined to form Watson-Crick duplex DNA. Second, the duplex DNA must have at least one strand broken. The complete inability of the E . coli plasmid pBR322 to activate poly(ADP-ribose) polymerase when native is due neither to its supercoiled state nor to the lack of a specific nucleotide sequence, but simply to the absence of strand breaks. Once cut, the DNA can effectively support polymer synthesis and its effectiveness increases in proportion to the frequency of similar types of strand breakage (Figs. 6 and 7).

Although a wide variety of DNA strand breaks elicit poly(ADP-ribose) synthesis in vitro, the relative efficiencies of the different types may vary by up to 25-fold. Single-strand breaks are less than one-tenth as effective as are double- strand breaks which produce flush-ended DNA fragments. Further, there is a strong correlation between the structure of the end of a DNA fragment and its ability to support polymer synthesis. Breaks producing DNA fragments with 2 to 4 unpaired nucleotides extending from their 3’ strands are only one-third as effective as are those with flush ends, while a double-strand break producing DNA ends with unpaired bases attached to the 5’ strands is slightly more effective than is a single-strand break. Finally, the removal of the terminal 5’- phosphate groups markedly increases the ability of all types of DNA fragments to activate poly(ADP-ribose) polymerase.

A comparison of the different types of DNA and their relative abilities to support poly(ADP-ribose) synthesis pro- vides an indication of what features of the DNA end are recognized by the polymerase. The ineffectiveness of single- stranded DNA fragments must reflect the importance of the duplex nature of the DNA end. The ability of the polymerase molecule to distinguish between the 5’-phosphorylated and 5‘- hydroxyl varieties of all three types of DNA ends tested clearly identifies the 5’ terminus as an important recognition feature. A further indication that the 5’ strand of the DNA end is more important than is the 3’ strand is suggested by the fact that DNA fragments with 2 to 4 unpaired nucleotides extending from the 3’ strands, and thus with their 5’ termini in duplex form, are about 3 times more effective in supporting poly(ADP-ribose) synthesis than are those whose 5’ strands terminate in 2 to 4 unpaired nucleotides. A possible explana- tion of the preference of the enzyme’s for flush-ended fragments is discussed below.

We will now consider the relation of these observations made in vitro to the events within cells that have suffered damage to their DNA. The absolute dependence of the calf thymus poly(ADP-ribose) polymerase activity in uitro on DNA strand breaks is in complete agreement with the observed stimulation of polymer synthesis in mammalian cell ghosts by agents which introduce breaks into the cellular DNA (4, 7,8, 23, 24). A further indication of the similarity of the two systems can be derived from a comparison of their responses to different levels of DNA strand breakage. An x- ray dose of 10 to 12 kilorads produces a half-maximal stimulation of poly(ADP-ribose) synthesis in mammalian cell ghosts (4). This dose generates about 100,000 single-strand breaks and about 4,000 double-strand breaks per cell (25) . Since single-strand breaks are only about one-twelfth as effective in supporting poly(ADP-ribose) synthesis in vitro as are flush double-strand breaks, and if we assume that x-irradiation generates double-strand breaks with a random assortment of


DNA end structures, these numbers can be combined to give an approximate value of 10,000 flush double-strand break- equivalents per cell, a value that, given the volume of a HeLa cell nucleus as 1 X liters (26), corresponds to approximately 20 X lo-’ M flush double-strand break-equivalents. This value is reasonably similar to the concentration of flush- ended DNA fragments that gives a half-maximal rate of poly(ADP-ribose) synthesis by the calf thymus preparation, namely 6 X lo-!’ M. Although the calculations are clearly very approximate, the similarity of the derived values does lead us to suppose that the response we observe in uitro with extracted enzyme and restriction fragments reflects fairly the situation in ghost cells, We seem justified in concluding that poly(ADP-ribose) synthesis is stimulated in ghost cells by the same sort of DNA modifications as are effective in uitro.

In view of our previous conclusion that poly(ADP-ribose) synthesis may be necessary for the repair of x-ray-generated strand breaks (4, 7), it is also of particular relevance to note that the types of DNA strand breaks that are most effective in eliciting poly(ADP-ribose) synthesis in uitro are those potentially most dangerous to a living cell. Double-strand breaks are, in general, more effective than are single-strand breaks. The most effective double-strand breaks have flush DNA ends which, lacking any complementary single-stranded tails, would be most likely to separate and must be among the most urgent DNA lesions to repair. In addition, the removal of the 5’-terminal phosphate groups from the ends of DNA fragments, which prevents conventional strand ligation, further increases their effectiveness in supporting poly(ADP- ribose) synthesis.

It is not clear precisely how poly(ADP-ribose) synthesis assists in the repair of DNA strand breaks. However, it is possible to speculate about possible roles within the frame- work of the following observations. Several thousand NAD’ molecules are converted to poly(ADP-ribose) at the site of each DNA strand break. There is presently no evidence indicating a covalent attachment of the polymer to the DNA. Although proteins can be covalently modified by poly(ADP- ribose), this may not be a necessary prerequisite of polymer synthesis. Finally, because flush double-strand breaks are most effective in eliciting poly(ADP-ribose) synthesis, at this time we will restrict our models to possible roles in the repair of double-strand breaks, which occurs in mammalian cells (25, 27-31) by a mechanism which is at present poorly understood.

The localized synthesis of poly(ADP-ribose) at a DNA strand break in uiuo could increase the accessibility of the site to repair enzymes and thus accelerate the repair process. The attachment of negatively charged ADP-ribose residues to chromosomal proteins could weaken their binding to the DNA and facilitate their displacement. The nucleosomal structure of chromatin dictates that histones be by random chance the most frequent class of proteins present at the point of strand breakage and, indeed, the ADP-ribosylation of histones is a well established occurrence (21, 22). A variety of other proteins could also be located at the break site but, since their incidence of ADP-ribosylation would he roughly in proportion to the percentage of the total chromosomal protein they comprise, their identification could be anticipated to be much more difficult. Still, ADP-ribosylation of nonhistone chromosomal proteins is known (21, 22).

Alternatively, the polymer could act to stabilize the two opposing DNA ends, keeping them close to one another until ligation can be effected. This might be accomplished by form- ing links, covalent or electrostatic, between proteins associated with the two pieces of DNA or by interacting directly with the DNA itself. We have presented preliminary evidence (Fig. 2) that a noncovalent interaction between the DNA double helix and free poly(ADP-ribose), predicted by the latter model, might indeed occur, at least in uitro.

1.

2.

3.

4.

5.

6.

7. 8.

9. 10. 11.

12.

13.

14. 15.

16. 17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27. 28.

29.

30.

31.

REFERENCES Chambon, P., Weill, J. D., and Mandel, P. (1963) Biochem.

Chambon, P., Weill, J . D., Doly, J., Strosser, M. T., and Mandel,

Nishizuka, Y., Ueda, K., Nakazawa, K., and Hayaishi, 0. (1967)

Benjamin, R. C., and Gill, D. M. (1980) J. Biol. Chem. 255, 10493-

Yoshihara, K. (1972) Biochem. Biophys. Res. Commun. 47, 119-

Gill, D. M., Caplan, R. A,, Bredlau, C. E., and Meren, R. (1973)

Benjamin, R. C., and Gill, D. M. (1980) Deu. Cell Biol. 6,357-366 Berger, N. A,, Weber, G., and Kaichi, A. S. (1978) Biochim.

Miller, E. G. (1975) Biochim. Biophys. Acta 395, 191-200 Janakidevi, K., and Koh, C. (1974) Biochemistry 13, 1327-1330 Juarez-Salinas, H., Sims, J. L., and Jacobson, M. K. (1979) Nature

282, 740-741 Skidmore, C. J., Davies, M. I., Goodwin, P. M., Halldorsson, H.,

Lewis, P. J., Shall, S., and Zia’ee, A. (1979) Eur. J. Biochem. 101, 135-142

Sutcliffe, J. G. (1979) Cold Spring Harbor Symp. Quant. Biol. 43, 77-90

Tanaka, T., and Weisblum, B. (1975) J. Bacteriol. 121, 354-362 Sutcliffe, J. G., and Church, G. M. (1978) Nucleic Acids Res. 5,

Gill, D. M. (1972) J. Biol. Chem. 247, 5964-5971 Melgar, E., and Goldthwait, D. A. (1968) J. Biol. Chem. 243,

Beerman, T . A., and Goldberg, I. H. (1974) Biochem. Biophys. Res. Commun. 59, 1254-1261

Ishida, R., and Takahashi, T. (1976) Biochem. Biophys. Res. Commun. 68,256-261

Lennartz, M., Coquerelle, T., Bopp, A,, and Hagen, U. (1975) Int. J. Radiat. Biol. Relat. Stud. Phys. Chem. Med. 27, 577-587

Hayaishi, O., and Ueda, K. (1977) Annu. Reu. Biochem. 46, 95- 116

Hilz, H., and Stone, P. (1976) Reu. Physiol. Biochem. Pharmacol.

Benjamin, R. C., and Gill, D. M. (1978) ZCN-UCLA Symp. Mol.

Halldorsson, H., Gray, D. A., and Shall, S. (1978) FEBS Lett. 85,

Elkind, M. M. (1978) ZCN-UCLA Syntp. Mol. Cell. Biol. 9, 477-

Lampert, F., Hofmann, D., and Sandritter, W. (1966) Histochemie

Corry, P. M., and Cole, A. (1973) Nut. New Biol. 245, 100-101 Hartwig, M., and Handschack, W. (1975) Stud. Biophys. 50,203-

Lehmann, A. R., and Stevens, S. (1977) Biochim. Biophys. Acta

Ahnstron, T., and Edvardsson, K. (1978) ZCN- UCLA Symp. Mol.

Lett, d. T., Keng, P. C., and Sun, C. (1978) ICN-UCLA Symp.

Biophys. Res. Commun. 11,39-43

P. (1966) Biochem. Biophys. Res. Commun. 25,638-643

J. Biol. Chem. 242, 3164-3171

10501

125

Fogarty Int. Center Proc. 26,85-99

Biophys. Acta 519,87-104

2313-2319

4409-4416

76, 1-58

Cell. Biol. 9, 337-340

349-352

480

6, 370-381

209

4 7 4 , 4 9 4 0

Cell. Biol. 9, 469-472

Mol. Cell. Biol. 9, 481-484

Documents

Poly(ADP-ribose) Synthesis in Vitro Programmed by Damaged DNA