Poly(ADP-ribosy)ation of nuclear proteins. Enzymatic elongation of chemically synthesized ADP-ribose-histone adducts.

ADP-ribose, an NAD metabolite, binds nonenzymatically to histones by forming a Schiff base at the terminal ribose (Kun, E., Chang, A. C. Y., Sharma, M. L., Ferro, A. M., and Nitecki, D. (1976) Proc. N&Z. Acud. Sci. U. S. A. 73,3131-3135). The ADP-ribose*histone Hl adduct thus formed activates purified rat liver poly(ADP-ribose) synthetase more effectively than unmodified Hl in the presence of DNA. Analyses of the reaction products using the protease treatment and electrophoreses suggest that the synthetase transfers the ADP-ribose moiety of NAD to the adduct, but little, if any, to unmodified Hl. The transferred ADP-ribose forms a linear polymer, poly(ADP-ribose), having the average chain length of 2 to 11 ADP-ribosyl units under various conditions. The fact that the chemically bound ADP-ribose is the site of poly(ADP-ribosyl)ation, that is, the ADP-ribose chain is elongated, is shown by the production of labeled isoADP-ribose upon digestion with snake venom phosphodiesterase of an adeninelabeled ADP-ribose*Hl adduct preincubated with the synthetase and unlabeled NAD. Comparative studies using variously labeled adducts and NADs reveal that the elongation proceeds exclusively by a terminal addition mechanism, i.e. ADP-ribose being transferred to the adenosine terminus. In the presence of ADP-riboseahistone HI adduct, a majority (50 to 90%) of new chains originates from the preattached ADP-ribose and the remainder from similar structures on an endogenous acceptor contained in the synthetase preparation. These results taken together suggest that poly(ADPribose) synthetase is primarily engaged in chain elongation, but not in direct ADP-ribosylation of histones (chain initiation); the latter reaction is probably catalyzed by an as yet unidentified enzyme or requires other factor(s) or conditions.

The ADP-ribose*histone Hl adduct thus formed activates purified rat liver poly(ADP-ribose) synthetase more effectively than unmodified Hl in the presence of DNA. Analyses of the reaction products using the protease treatment and electrophoreses suggest that the synthetase transfers the ADP-ribose moiety of NAD to the adduct, but little, if any, to unmodified Hl. The transferred ADP-ribose forms a linear polymer, poly(ADP-ribose), having the average chain length of 2 to 11 ADP-ribosyl units under various conditions. The fact that the chemically bound ADP-ribose is the site of poly(ADP-ribosyl)ation, that is, the ADP-ribose chain is elongated, is shown by the production of labeled isoADP-ribose upon digestion with snake venom phosphodiesterase of an adeninelabeled ADP-ribose*Hl adduct preincubated with the synthetase and unlabeled NAD. Comparative studies using variously labeled adducts and NADs reveal that the elongation proceeds exclusively by a terminal addition mechanism, i.e. ADP-ribose being transferred to the adenosine terminus.
In the presence of ADP-riboseahistone HI adduct, a majority (50 to 90%) of new chains originates from the preattached ADP-ribose and the remainder from similar structures on an endogenous acceptor contained in the synthetase preparation. These results taken together suggest that poly(ADPribose) synthetase is primarily engaged in chain elongation, but not in direct ADP-ribosylation of histones (chain initiation); the latter reaction is probably catalyzed by an as yet unidentified enzyme or requires other factor(s) or conditions.
Poly(ADP-ribosyl)ation' is a unique post-translational modification of nuclear proteins (l-3). The modification reaction is supposed to proceed by an initial attachment of an ADP-ribose monomer, followed by successive elongation using NAD as the donor of ADP-ribose.
Both the modifying enzyme (poly(ADP-ribose) synthetase) and modified proteins have been demonstrated in various eukaryotic organisms and tissues in vitro (l-7) as well as in viuo (8)(9)(10)(11)(12). Although there * This work was supported in part by grants-in-aid for scientific and cancer research from the Ministry of Education, Science and Culture, Japan, and by a research grant from the Princess Takamatsu Cancer Research Fund. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' The term "poly(ADP-ribosyl)ation" refers to all modifications by (ADP-ribose),, chains (n 2 2).
are some indications suggesting a relationship between this modification and the regulation of gene expression (13)(14)(15)), its precise biological function has not yet been elucidated (16,17). Among many proteins reported to be modified, histone was the first example (l-3), and the subgroups Hl and H2B have been documented to be the major acceptors (4-11, 18). Poly(ADP-ribose) synthetase has been partially purified from rat liver (19)(20)(21)(22), calf thymus (23, 24), pig thymus (25) or ascites tumor cells (26). Common unique features of these enzyme preparations are the absolute dependence of the activity on DNA and further stimulation by histones added along with DNA (16)(17)(18)(19). The latter effect of histones is not due to a supply of acceptors since our recent analysis on an extensively purified rat liver synthetase suggested that histones did not serve as ADP-ribose acceptors but acted as allosteric activators and that poly(ADP-ribosyl)ation occurred exclusively on a non-histone-type acceptor which co-purified with the enzyme (22, 27). Recently, Yoshihara et al. reported with a calf thymus enzyme that the enzyme itself may accept ADP-ribose (24). The apparent discrepancy concerning histone utilization between the in uiuo (or crude nuclei) and in vitro (purified) systems suggests various possibilities, such as the existence of two kinds of enzyme, one for initiation and the other for subsequent elongation, or a certain factor(s) or structure necessary for initiation but lost during the purification of the synthetase. In order to examine these possibilities and to obtain a closer insight into the mechanism of poly(ADP-ribosyl)ation, the present study was carried out. Using a nonenzymatic ADP-ribose. histone adduct described by Kun  Schiff base formation, the treatment with NaBH, resulted in stabilization of the linkage at alkaline pH values, and the treatment with snake venom phosphodiesterase released AMP from the bound ADP-ribose but no ribose 5-phosphate. Furthermore, when the ADP-ribose.Hl adduct reduced by NaBH, was digested with proteinase K (1:20 (w/w) to protein; 37"C, 4 h) and analyzed by paper electrophoresis, no free ADP-ribose was recovered, further supporting the view that the linkage between ADP-ribose and Hl was of a covalent nature. The ADP-ribose adducts used in the following studies were all treated with NaBH4 and extensively dialyzed in order to remove unreacted ADP-ribose.
Activation of Poly(ADP-ribose) Synthesis by ADPribose. Histone Adduct-As previously described (21,22), poly(ADP-ribose) synthetase purified from rat liver nuclei absolutely required DNA and, in addition, histone for full activity. The extent of activation by histone varied with the preparation of enzyme, the kind of histone as well as DNA, the ratio of histone to DNA, and other conditions, such as salt concentrations and temperature. Under all conditions tested, an ADP-riboseeH1 adduct was more stimulatory than un- modified Hl. The time course (Fig. 3) showed that the adduct was approximately 4 times as potent as unmodified Hl at any time points up to 3 h. Furthermore, as shown in Fig. 4, an ADP-ribose.
Hl adduct was more stimulatory than unmodified Hl at any given concentrations as well as at respective optima, in both lo-and 60-min incubations, indicating that the difference was not due to a shift of concentration optimum of histone. Free ADP-ribose showed no stimulatory effect on poly(ADP-ribose) synthetase in the absence and presence of Hl.
The activation of poly(ADP-ribose) synthetase was relatively specific for histones even in the ADP-ribose adduct form; an ADP-ribose adduct of albumin gave little, if any, effect on the synthetase (Table I). A preliminary survey has shown that the capacity to bind to DNA is essential for a of ADP-ribose.  protein to stimulate the DNA-activated synthetase." The stimulation by ADP-ribose. histone adducts varied with the content of ADP-ribose.

Histone Adducts
In the case of Hl adducts, the adduct having about 2 ADP-ribose residues/m01 of protein was most stimulatory. These observations, together with our previous finding that histones did not accept ADP-ribose in the purified system (22, 27), suggest that an ADP-ribose.
histone adduct may play two separate roles in poly(ADP-ribosyl)ation, one as a general activator as does unmodified histone and the other as a specific activator depending on its ADP-ribose residue. ADP-ribose. Histone Adduct as ADP-ribose Acceptor-Evidence for the latter activator function of ADP-rihose. histone adducts ascribable to accepting ADP-ribose was obtained by analyses of the reaction products with the protease treatment and polyacrylamide gel electrophoresis. As shown in Fig. 5 (A and B), pronase, a broad spectrum protease, solubilized essentially no poly(ADP-ribosyl)ated material synthesized in the absence of exogenous protein or in the presence of unmodified histone Hl, the result confirming our previous view that purified poly(ADP-ribose) synthetase did not ADPribosylate any of histones but modified an endogenous acceptor (22). The latter acceptor is known to be very resistant to proteases (27). In contrast, the same protease solubilized a majority of the ADP-ribose incorporated in the presence of ADP-ribose.Hl adduct (Fig. 5C). Although there was a difference in the average chain length among these products (about 12, 10, and 4 ADP-ribosyl units for the products of A, B, and C, respectively), the difference in protease susceptibility did not appear due to the difference in polymer size, since a polymer having about 10 ADP-ribosyl units is only partly acid-insoluble." It seemed, therefore, that the different susceptibilities reflected the different chemical nature of acceptor molecules and that the acceptor in the reaction with an ADPribose. histone adduct would be the adduct itself. remaining at the top in Fig. 6C was indicative of poly(ADPribosyl)ation occurring also on an endogenous acceptor as in two other cases. This mobility of the endogenous acceptor (RF = 0) was different from the value obtained earlier (RF = 0.58) (22). The difference appeared to be due to the different solubility of the poly(ADP-ribosyl)ated endogenous acceptor in these two experiments since the average chain length of the products differed markedly (>lO (see below) uersus ~3.7 ADP-ribosyl units (22)), and the acceptor material with longer polymers has been shown to be less soluble." This notion was substantiated by a combination of electrophoretic and proteolytic analyses. While the products synthesized with no added protein or with unmodified Hl remained at the top of the gel apart from histone bands (Fig. 6, A and  B), a majority of ADP-ribose incorporated in the presence of ADP-ribose.Hl adduct migrated close to the protein bands (Fig. 6C). Our ADP-ribose.
histone Hl adduct preparation contained an Hl dimer (as judged by its molecular weight in the SDS gel); this dimer also accompanied a small amount of incorporated ADP-ribose. A minor portion of ADP-ribose Treatment of all these products with proteinase K, a nonspecific protease (37), further disclosed the difference; the treatment converted to smaller products little, if any, of the material synthesized with no addition (Fig. 6D) or with unmodified Hl (Fig. 6E), but totally converted the material which had been synthesized with ADP-ribose. Hl adduct and migrated close to histone bands (Fig. 6F). Since the average chain lengths of the products of A, B, and C, determined in a parallel experiment, were 12.7, 10.5, and 3.9, respectively, and free oligomers and polymers (up to about 15 ADP-ribosyl units) migrated faster than the marker dye under the present conditionq7 it did not appear plausible that small peptides accompanying poly(ADP-ribose) remained at the top, but rather that the products in the former cases (A/D and B/E) were resistant, whereas the product in the latter (C/E;3 was sensitive to the protease. This conclusion was compatible with  the view that the synthetase-associated acceptor and ADPribose.Hl adduct were, respectively, the main acceptors in the absence and presence of the adduct. The ADP-ribose remaining at the top in Fig. 6F was indicative of partial poly(ADP-ribosyl)ation of an endogenous acceptor, as supposed above.
Enzymatic poly(ADP-ribosyl)ation of ADP-ribose. histone adducts was also implied by a change in the electrophoretic mobility after the synthetase reaction (Fig. 7). The adduct preparation used in this experiment gave one major and several minor bands, all accompanied by ADP-ribose. When this preparation was incubated with purified poly(ADP-ribose) synthetase and ["HINAD followed by extraction with 0.25 N HCl, a slight shift of the 14C peak to a slowly migrating range or a shoulder of 14C coincident with the peak of "H emerged, suggesting that a part (approximately 20%) of ADPribose.Hl adducts was poly(ADP-ribosyhated and changed the mobility.
In order to determine the size of polymers synthesized on the adducts, the reaction products were extracted with 0.25 N HCl and purified by electrophoresis.
Two different conditions of poly(ADP-ribosyl)ation, 0°C for 30 min (Fig. 8B) or 37°C for 30 min (Fig. 8C), were examined. The products synthesized with unmodified Hl were similarly treated for reference (Fig. 8A). Recovery of acid-insoluble ADP-ribose into the HCl extract was quantitative for both products with ADP-ribose.
Hl adducts, but only one-third for the products made with unmodified Hl. The latter products migrated at a distance from the main band of histone (Fig. 8A)  Elongation of ADP-ribose -Histone Adducts synthesized with the adducts at 0°C migrated close to the histone bands (Fig. 8B), whereas those of the 37°C incubation retarded and were distributed in several peaks (Fig. 8C). Analyses of average chain lengths of the peak fractions (indicated by arrows) revealed that the products synthesized at 37°C was 10.7 ADP-ribosyl units long and that of 0°C incubation was 2.5 units long. These results indicated that not only oligomers but also polymers were synthesized on ADPribose. histone adducts and that the larger the polymer was, the more slowly the modified adduct moved.
All above results led to the conclusion that an ADP-rihose. histone adduct, but not unmodified histone, served as an acceptor for enzymatic poly(ADP-ribosyl)ation.
This conclusion, however, did not necessarily mean that poly(ADPribosyl)ation took place at preattached ADP-ribose. Elongation from ADP-ribose Attached to Histone by Terminal Addition Mechanism-In order to investigate whether the ADP-ribose residue bound to histone is the very site of enzymatic modification (that is, elongation), we examined the fate of radioactivity of [Ade-'4C]ADP-ribose. histone adduct in the poly(ADP-ribose) synthetase reaction. The principle of the experiment is illustrated in Fig. 9. If the ADP-ribose transfer from NAD by poly(ADP-ribose) synthetase does not occur to chemically bound ['4C]ADP-ribose ("not elongated"), the radioactivity is expected to be recovered in AMP upon digestion of the reaction products with snake venom phosphodiesterase. On the other hand, if the enzymatic ADP-ribosylation with unlabeled NAD takes place on the prebound ["'CIADP-ribose ("elongated"), radioactive isoADP-ribose will be produced by the phosphodiesterase digestion; hence the formation of ['*C]isoADP-ribose provides an estimate of elongation from prebound ADP-ribose.
The experimental results (Fig. 10) indicated the latter idea to be the case. Among the phosphodiesterase digests of ['4C]ADP-ribose.
Hl adduct incubated with poly(ADP-ribose) synthetase and NAD, 6.1% of the radioactivity eluting from a Dowex 1 column was recovered in the fractions of isoADP-ribose preceding the marker ADP-ribose. From the adduct not incubated, no radioactivity was recovered in the corresponding fractions. Other labeled products were AMP and adenosine; the latter was probably produced from AMP by phosphatases contaminating the phosphodiesterase preparation used (38). Radioactive isoADP-ribose was identified by paper chromatography in two solvent systems (Fig. 11, A and C) and further by conversion to ribosyladenosine by treatment with alkaline phosphatase (Fig. 11, B and D). These results provided unequivocal evidence for the elongation of the ADP-ribose residue preattached to histone. Above analyses also indicated that chain elongation proceeded by a terminal addition mechanism, i.e. the ADP-ribose units being transferred to free, adenosine termini. Whether    . 11. Identification of isoADP-ribose by paper chromatography. The putative isoADP-ribose material isolated as in Fig. 10 was chromatographed on a filter paper directly (A and C) or after phosphatase digestion ("Materials and Methods") (B and D). Solvent System 1 was used in A and B and System 2, in C and D.  this was the only mechanism for chain elongation or other mechanisms, such as basal addition (an insertion of ADPribose between histone and bound ADP-ribose) or a transfer of prebound ADP-ribose to other sites, might also work, was examined by similar analyses using the ADP-ribose.
Hl adduct labeled in the opposite position, ribose (Fig. 12). Among the phosphodiesterase digests of incubated as well as not incubated ADP-["Hlribose. Hl adduct, neither radioactive isoADP-ribose nor ribose 5-phosphate (which should elute out at about 1.5 N HCOOH) was recovered. These results excluded the possibility of alternative mechanisms mentioned above and strongly supported the terminal addition mecha- nism.
Quantitative estimation of chain elongation from adducts was performed by incubating [a-""PIADP-ribose.Hl adduct with [Ade-"H]NAD (Table II). Elongation and phosphodiesterase digestion yielded double-labeled isoADP-ribose (Figs. 13 and 14). Approximately 7.4% of original "2P was recovered in isoADP-ribose, and the average chain length of "H-labeled portion was calculated to be 4.0. From these values, about one-half of new ADP-ribose chains was estimated to have origin on the adducts.
Per cent elongation from adducts, estimated by the produc-tion of isoADP-ribose from preattached ADP-ribose, varied markedly depending on the reaction conditions such as the concentrations of adducts and NAD and the preparation of enzyme; under various conditions, we observed 6 to 30% elongation. These values corresponded to 50 to 90% of newly synthesized chains. In all cases preliminarily examined, the number of elongated chains was in good accordance with the number of chains starting from the adducts (as examined by electrophoresis and chain length analysis), indicating that poly(ADP-ribose) on adducts originates exclusively from preattached ADP-ribose and not from any other sites.
identity of degradation products (Figs. 13 and 14), the first and subsequent stages of chain elongation appear to proceed by the same mechanism. This mechanism, terminal addition, is similar to primer-oriented DNA synthesis but dissimilar to tRNA-mediated polypeptide synthesis or coenzyme A-linked fatty acid synthesis. The same mechanism has also been proposed by Yoshihara et al. on the basis of preliminary evidence with an enzyme-bound early product (24). DISCUSSION Taking advantage of the specificity of isoADP-ribose production from polymerized ADP-ribose, the present study showed that poly(ADP-ribose) synthetase purified from rat liver transfers ADP-ribose from NAD to the ADP-ribose residue of a chemically prepared ADP-ribose.
histone adduct and elongates it into a polymer. The purified enzyme is not able to utilize histone as an acceptor but modifies exclusively an endogenous acceptor (22, 27). Our analysis on the latter acceptor using poly(ADP-ribose) glycohydrolase and snake venom phosphodiesterase has revealed that it has ADP-ribose or alike structure at the ADP-ribosylation site (39). Recently, Yoshihara et al. reported that poly(ADP-ribose) synthetase purified from calf thymus was also incapable of modifying histones (24). In this case, however, ADP-ribose appeared to be transferred on to the enzyme itself. Whether this means only a very tight association of an endogenous acceptor similar to ours with the enzyme protein or represents a reaction intermediate of poly(ADP-ribose) synthetase is not clear at present.
These observations appear to indicate that partially degraded poly(ADP-ribose) may be re-elongated in uivo and, most probably, such an event occurs very frequently (46). For a polymer to be re-elongated it will be necessary that the polymer remains bound to an acceptor at one end and preserves the adenosine terminus at the other. Poly(ADP-ribose) glycohydrolase, the enzyme responsible for in viuo degradation of the polymer (47,48), meets these requirements.
A preliminary experiment in our laboratory has suggested that mono-as well as oligo(ADP-ribosyl) histones (Hl and H2B) that are synthesized in rat liver nuclei are elongated by poly(ADP-ribose) synthetase as are the nonenzymatic adducts described in this paper (45).