Sequential ADP-ribosylation pattern of nucleosomal histones. ADP-ribosylation of nucleosomal histones

The pattern of nucleosomal histones poly(ADP-ribosyl)ation is changed under conditions which affect the poly(ADP-ribosyl)ation state of the enzyme. At low NAD concentrations the enzyme can poly(ADP-ribosyl)ate histones H1 and H1, H2A, A2A, and H2B. However at NAD concentrations above 10 microM the enzyme preferentially poly(ADP-ribosyl)ates histone H1 to a hyper ADP-ribosylated form. Furthermore we have observed hyper ADP-ribosylation of histone H2B at NAD concentrations of 10 microM suggesting that histone H2B can undergo the same type of ADP-ribosylation pattern as histone H1. Also at higher NAD concentrations an elongation of the polymer attached to the enzyme and other nuclear proteins takes place.     

ADP-ribosylation of diadenosine 5',5"-P1,P4-tetraphosphate by poly(ADP-ribose) polymerase in vitro

When the effect of diadenosine 5',5"'-P1,P4-tetraphosphate on a purified poly(ADP-ribose) polymerase reaction was examined, the compound strongly inhibited ADP-ribosylation reaction of histone, while the compound was much less inhibitory of the Mg2+-dependent automodification of this enzyme. In an attempt to study the mechanism of the inhibition, we analyzed the total reaction products, which were synthesized from NAD+ in the presence of diadenosine 5',5"'-P1,P4-tetraphosphate in a reaction mixture for ADP-ribosylation of histone, and found that a new, low molecular product was predominantly synthesized instead of ADP-ribosylated histone in the reaction. Approximately 90% of added NAD+ was converted into this low molecular product under an appropriate reaction condition. Further analysis revealed that the product was mono- and oligo(ADP-ribosyl)ated diadenosine nucleotide and that the bound oligo(ADP-ribose) is elongating at one end of the product during the reaction. Thus, the present study clearly demonstrated that diadenosine 5',5"'-P1,P4-tetraphosphate functions as an acceptor for ADP-ribose in a poly(ADP-ribose) polymerase reaction in vitro. The finding that histone H1 is required in the reaction mixture for the synthesis of this new product suggests that histone H1 and the diadenosine compound interact during this modification reaction.     

Hyper(ADP-ribosyl)ation of histone H1

Nucleosomal chains of various repeat unit lengths were generated by a mild micrococcal nuclease digestion of purified pancreatic nuclei. Maximal nucleosome associated poly(ADP-ribose) polymerase activity was recovered in trimeric to tetrameric chromatin fragments, after which the enzyme activity gradually decreased and stabilized towards oligomeric periodicities of 11 to 16 nucleosomes. Electrophoresis of [32P]ADP-ribosylated histones on first-dimension acid-urea or acid-urea-Triton gels and on second-dimension acid--urea--cetyltriammonium bromide gels revealed that, of all histones, only histone H1 could be significantly poly(ADP-ribosyl)ated while only minimal modification could be recovered with histone H1(0). Furthermore, the extent of ADP-ribosylation present on pancreatic histone H1 is shown to proportionally retard this protein's electrophoretic mobility in all gel systems and to consist of a distinct series of at least 12 modification intermediates which can be evidenced, in nuclei or nucleosomes, and fully recovered along with histone H1 upon its selective extraction with 5% perchloric acid. The generation of these increasingly ADP-ribosylated forms of histone H1 is also demonstrated to be time dependent and the more complex ADP-ribosylated forms of this histone are favored at high NAD+ concentrations. Moreover, the electrophoretic mobilities of all intermediates are unaffected by the presence of the nonionic detergent Triton X-100.     

Diadenosine 5', 5"'-P1,P4-tetraphosphate stimulates processing of adp-ribosylated poly(ADP-ribose) polymerase

The effect of diadenosine 5', 5"'-P1,P4-tetraphosphate (Ap4A) on the time course and acceptors of poly(ADP-ribose) synthesis was studied in undamaged and N-methyl-N'-nitro-N-nitrosoguanidine-treated human lymphocytes. Analysis of protein acceptors of poly(ADP-ribose) revealed that treatment with Ap4A stimulated ADP-ribosylation of bands at molecular weights of 96,000, 79,000, and 62,000. Pulse-chase studies showed that these bands were produced as a result of an effect of Ap4A on the processing of ADP-ribosylated proteins rather than on the synthesis of newly ADP-ribosylated proteins. By incubating permeabilized cells in the absence or presence of Ap4A and purified poly(ADP-ribose) polymerase auto-ADP-ribosylated with [32P]NAD+, we showed that the Mr = 96,000, 79,000, and 62,000 bands were derivatives of the prelabeled enzyme. Our results indicate that normal human lymphocytes process auto-ADP-ribosylated poly(ADP-ribose) polymerase to specific lower molecular weight products and that this processing is stimulated by Ap4A.     

Production of anti-(ADP-ribose) antibodies with the aid of a dinucleotide-pyrophosphatase-resistant hapten and their application for the detection of mono(ADP-ribosyl)ated polypeptides

Previous attempts to produce anti-(ADP-ribose) antibodies by immunization of rabbits with ADP-ribose conjugated to serum albumin had resulted in the production of 5'AMP-specific antibodies [Bredehorst et al. (1978) Eur. J. Biochem. 82, 105-113]. To obtain true anti-(ADP-ribose) antibodies an antigen was constructed that was resistant to enzymic degradation at the pyrophosphate group. The enzymically active beta-methylene derivative of NAD (NAD[CH2]) was synthesized from ADP containing a methylene bridge (CH2) instead of an oxygen in the diphosphate group. NAD[CH2] was converted to its N6-[(2-carboxyethyl)thiomethyl] derivative and hydrolyzed to the corresponding ADP[CH2]-ribose derivative which was then coupled to bovine serum albumin. The antibodies obtained with this antigen were specific for free or protein-bound ADP-ribose groups, except for a cross-reaction with FAD, AMP, ADP, ATP or poly(ADP-ribose) interfered with [3H]ADP-ribose tracer binding only at higher concentrations. No interference was observed with poly(A), RNA and DNA at 6000-fold excess. The antibodies were purified on a novel type of affinity matrix. This was formed from NAD and guanidinobutyrate by a cholera-toxin-catalyzed reaction and the product, ADP-ribosyl guanidinobutyrate, was bound to Affi Gel by carbodiimide-aided condensation. The purified antibodies allowed the detection of ADP-ribose conjugated to polypeptides in amounts lower than 1 pmol as demonstrated by immunoblotting of [14C]ADP-ribosylated elongation factor 2. They also could be used to observe in situ, by indirect immunofluorescence, the increased mono(ADP-ribosyl)ation of nuclear proteins in dimethyl-sulfate-treated cells, and to show that histone H2B was the principal histone acceptor of single ADP-ribose groups in alkylated 3T3 cells.     

Enzymatic synthesis of [4R-2H]NAD (P)H and [4S-2H]NAD(P)H and determination of the stereospecificity of 7 alpha- and 12 alpha hydroxysteroid dehydrogenase

The stereospecifically labeled coenzymes [4R-2H]NADH, [4R-2H]NADPH and [4S-2H]NAD(P)H were synthesized enzymatically in high yield and high isotopic purity (greater than or equal to 95%) with 2HCOO2H/formate dehydrogenase, (CH3)2C2HOH/alchol dehydrogenase from Thermoanaerobium brockii and [1-2H]glucose/glucose dehydrogenase, respectively. This set of deuterated coenzymes was used to determine the stereospecificity of the previously unstudied 7 alpha-hydroxysteroid dehydrogenase from Escherichia coli (NAD-dependent) and 12 alpha-hydroxysteroid dehydrogenase from Clostridium group P (NADP-dependent). H-NMR and EI-MS of the nicotinamide moiety after enzymatic oxidation of deuterated NAD(P)H with dehydrocholic acid as substrate showed that both dehydrogenases are B-sterospecific.     

Amino acid sequence of histone H1 at the ADP-ribose-accepting site and ADP-ribose X histone-H1 adduct as an inhibitor of cyclic-AMP-dependent phosphorylation

The ADP-ribosylation site of histone H1 from calf thymus by purified hen liver nuclear ADP-ribosyltransferase was determined and effects of the ADP-ribose X histone-H1 adduct on cAMP-dependent phosphorylation of the histone H1 were investigated. ADP-ribosylated histone H1 was prepared by incubation of histone H1, 1 mM [adenylate-32P]NAD and the purified ADP-ribosyltransferase. N-Bromosuccinimide-directed bisection of ADP-ribosylated histone H1 showed that the NH2-terminal fragment (Mr = 6000) was modified and contained serine residue 38, the site of phosphorylation by cAMP-dependent protein kinase. Digestion of the NH2-terminal fragment with cathepsin D and trypsin, and purification of this fragment, using high-performance liquid chromatography, yielded a radiolabelled single peptide corresponding to residues 29-34 of histone H1, containing the arginine residue as the ADP-ribosylation site. These results indicate that ADP-ribosylation of histone H1 occurs at the arginine residue 34, sequenced at the NH2-terminal side of the phosphate-accepting serine residue 38. Phosphorylation of histone H1 from calf thymus by cAMP-dependent protein kinase was markedly reduced when histone H1 was ADP-ribosylated. Kinetic studies of phosphorylation revealed that ADP-ribosylated histone H1 was a linear competitive inhibitor of histone H1 and a linear non-competitive inhibitor of ATP.     

ADP-Ribosylation of Histones in Nuclei Isolated from Embryos of the Sea Urchin, Hemicentrotus pulcherrimus

Two-dimensional electrophoresis (2D-PAGE) of a histone fraction isolated from nuclei of embryos of the sea urchin Hemicentrotus pulcherrimus exhibited almost all histone species at all stages examined. At the gastrula stage, a spot of H1A became evident and three spots closely associated with one another were found in place of a single spot of H2A.1. In the histone fraction isolated from [adenylate-³²P] NAD⁺-treated nuclei of all stages examined, autoradiograms of 2D-PAGE exhibited spots of mono [ADP-ribosyl] ated H1 and polymodified H2B.2, H3.1, H3.3 and H4 but did not show ADP-ribosylated H2A.1, H2A.2 or H2B.1. Poly [ADP-ribosyl] ated H3.2, found in morulae, was not detectable in blastulae and gastrulae. Treatment with dimethylsulfate, known to activate ADP-ribosylation in other cell types, induced poly [ADP-ribosyl] ation of H2A.2 and H2B.1 in embryos at all stages examined, and also polymodification of H3.2 in gastrulae. ADP-ribosylation of H1, H2B.2, H3.1 and H3.3 was hardly affected by dimethylsulfate treatment, though modification of H4 was blocked by this treatment. Probably, strong regulation of ADP-ribosyltransferase reactions causes failures of modification of H2A.2 and H2B.1 throughout early development and also of H3.2 at the gastrula stage. Regulation of histone ADP-ribosylation is thought to alter chromatin structures and the rate of gene expression, contributing to cell differentiation.     

In vitro poly(ADP-ribosyl)ated histones H1a and H1t modulate rat testis chromatin condensation differently

Rat testis H1 proteins were poly(ADP-ribosyl)ated in vitro. The modifying product, poly(ADP-ribose), was found covalently bound to each histone variant at various extents and exhibited distinct structural features (linear and short, rather than branched and long chains). Interest was focused on the somatic H1a, particularly abundant in the testis, as compared with other tissues, and the testis-specific H1t, which appears only at the pachytene spermatocyte stage of germ cell development. These H1s were modified with poly(ADP-ribose) by means of two in vitro experimental approaches. In the first system, each variant was incubated with purified rat testis poly(ADP-ribose)polymerase in the presence of [(32)P] NAD. In parallel, poly(ADP-ribosyl)ated H1s were also prepared following incubation of intact rat testis nuclei with [(32)P] NAD. In both experiments, the poly(ADP-ribosyl)ated proteins were purified from the native forms by means of phenyl boronic agarose chromatography. The results from both analyses were in agreement and showed qualitative differences with regard to the poly(ADP-ribose) covalently associated with H1a and H1t. Comparison of the bound polymers clearly indicated that the oligomers associated with H1a were within 10-12 units long, whereas longer chains (相似文献   

Studies on some specific Ap4A-degrading enzymes with the use of various methylene analogues of P1P4-bis-(5'',5''''''-adenosyl) tetraphosphate.

Six new methylenephosphonate analogues of P1P4-bis-(5',5'-adenosyl) tetraphosphate, Ap4A, having P2-P3 carbon bridges CF2, CCl2 and CH2CH2 or P1-P2 and P3-P4 carbon bridges CF2, CCl2 and CH2CH2 in the tetraphosphate chain, were examined as substrates or inhibitors for two specific Ap4A-degrading enzymes: (asymmetrical) Ap4A hydrolase (EC 3.6.1.17) from yellow-lupin seeds and (symmetrical) Ap4A hydrolase (EC 3.6.1.41) from Escherichia coli. All analogues in which the central oxygen atom was replaced by a stable carbon bridge were hydrolysed by the asymmetrical hydrolase (CF2 greater than CCl2 greater than O greater than CHBr greater than CH2 greater than CH2CH2). As expected, these analogues were not hydrolysed by the symmetrical hydrolase, which was also unable to act on analogues having P1-P2 and P3-P4 carbon bridges.     

ADP-ribosylation of pancreatic histone H1 and of other histones

Incubation of pancreatic nuclei with high NAD concentrations resulted in increased ADP-ribosylation of histone H1. Interaction of [3H]ADP-ribosylated histone H1 with chromatin was significantly different from unmodified histone H1. The presence of a protein which is eluted at a lower salt concentration and which is ADP-ribosylated was also noticed. Pancreatic histones were isolated by column chromatography and their degree of ADP-ribosylation evaluated both by gel electrophoresis and by chromatography: histone H1 was the main acceptor while the core histones H3, H2B, and H2A were lightly labelled. Histones H1 and H1(0) have a differential binding to pancreatic chromatin and histone H1(0) is not ADP-ribosylated.     

Specific inhibitors of poly(ADP-ribose) synthetase and mono(ADP-ribosyl)transferase.

Two classes of enzymes, poly(ADP-ribose) synthetase and mono(ADP-ribosyl)transferases, catalyze covalent attachment of multiple or single residues, respectively, of the ADP-ribose moiety of NAD+ to various proteins. In order to find good inhibitors of poly(ADP-ribose) synthetase free of side actions and applicable to in vivo studies, we made a large scale survey using an in vitro assay system, and found many potent inhibitors. The four strongest were 4-amino-1,8-naphthalimide, 6(5H)- and 2-nitro-6(5H)-phenanthridinones, and 1,5-dihydroxyisoquinoline. Their 50% inhibitory concentrations, 0.18-0.39 microM, were about two orders of magnitude lower than that of 3-aminobenzamide that is currently most popularly used. A common structural feature among all potent inhibitors, including 1-hydroxyisoquinoline, chlorthenoxazin, 3-hydroxybenzamide, and 4-hydroxyquinazoline, in addition to the four mentioned above, was the presence of a carbonyl group built in a polyaromatic heterocyclic skeleton or a carbamoyl group attached to an aromatic ring. Most of the inhibitors exhibited mixed-type inhibition with respect to NAD+. Comparative studies of the effects on poly(ADP-ribose) synthetase and mono(ADP-ribosyl)transferase from hen heterophils revealed high specificity of most of the potent inhibitors for poly(ADP-ribose) synthetase. On the other hand, unsaturated long-chain fatty acids inhibited both enzymes, and saturated long-chain fatty acids and vitamin K1 acted selectively on mono(ADP-ribosyl)transferase. The finding of many inhibitors of ADP-ribosyltransferases, especially poly(ADP-ribose) synthetase, supports the view that ADP-ribosylation of proteins may be regulated by a variety of metabolites or structural constituents in the cell.     

Correlation between endogenous nucleosomal hyper(ADP-ribosyl)ation of histone H1 and the induction of chromatin relaxation.

The effect of poly(ADP-ribose) synthesis on chromatin structure was investigated by velocity sedimentation and electron microscopy. We demonstrate that locally relaxed regions can be generated within polynucleosome chains by the activity of their intrinsic poly(ADP-ribose)polymerase. This relaxation phenomenon is also shown to be NAD dependent and to be correlated with the formation of hyper(ADP-ribosyl)ated forms of histone H1. Evidence is also presented which suggests that hyper(ADP-ribosyl)ated histone H1 is neither released from the relaxed chromatin, nor does it seem to participate in polynucleosomal aggregation.     

Differentiation of 3T3-L1 pre-adipocytes induced by inhibitors of poly(ADP-ribose) polymerase and by related noninhibitory acids

To analyze a possible involvement of ADP-ribosylation reactions in 3T3-L1 pre-adipocyte differentiation. ADP-ribosyltransferase activities is permeabilized cells as well as endogenous amounts of protein-bound mono- and poly(ADP-ribose) residues were determined. Also, in vivo labeling with [3H]adenosine of ADP-ribose residues linked to high-mobility-group (HMG) proteins was performed. As an additional probe, the effects of ADP-ribosylation inhibitors and non-inhibitory analogs were studied. Basal and total poly(ADP-ribose) polymerase activities markedly increased prior to the appearance of the differentiation marker glycerol-3-phosphate dehydrogenase. Despite these apparent changes in activity, however, neither protein-bound poly(ADP-ribose) residue nor mono(ADP-ribosyl) groups in histones, nor the NAD content, changed significantly under these conditions. Furthermore, although HMG protein-associated [3H]ADP-ribose was reduced in differentiating [3H]adenosine-labeled cells, the data suggest altered precursor pool labeling rather than a specific decrease in ADP-ribosylated HMG proteins. Non-participation of ADP-ribosylation reactions in 3T3-L1 differentiation is further supported by experiments with inhibitors and non-inhibitory analogs. Benzamide at 0.3-3 mM per se without effect on differentiation, was able to induce specific gene expression when combined with insulin (10(-12)-10(-7) M). Similar effects were seen with benzoate as well as with nicotinamide, 3-aminobenzamide and their corresponding acids. The data indicate that benzamide and analogs have profound effects on chromatin functions that are not mediated by ADP-ribosylation reactions.     

The binding of adenosine(5'')tetraphospho(5'')adenosine to calf thymus histones measured by non-equilibrium dialysis.

Binding of adenosine(5')tetraphospho(5')adenosine (Ap4A) to histones of calf thymus was investigated by non-equilibrium dialysis. Histone H1 interacts with the dinucleotide via two strong sites and competes with Mg2+ ions. Intrinsic dissociation constants were 1.6 +/- 0.1 microM and 11 +/- 1 microM for zero and 0.4 mm-Mg2+ concentration respectively. Binding of poly(dT) and of other nucleotides to histone H1 was measured in an [3H]Ap4A-competition assay. The tendency to form complexes among nucleotides was highest for bisnucleoside tetraphosphates and decreased in the order poly(dT) greater than or equal to Ap4A approximately Gp4G greater than Ap4 much greater than Ap3A approximately Ap5A greater than or equal to ATP, GTP and dTTP. The co-ordination complex derived from Ap4A and cis-diammine-dichloroplatinum(II) was not reactive. The other histones of calf thymus also bound Ap4A with affinities decreasing in the order H4 approximately H3 greater than H1 greater than H2b greater than H2a. Ap4A stimulated the exchange of histone H1 between nucleosomes, but this effect was referred to ionic strength. It did not bind to assembled nucleosomes. Binding of Ap4A to histone H1 was decreased by salt (NaCl). At physiological saline concentration the value of the dissociation constant is commensurable with the value of the Ap4A concentration in the nucleus and thus indicative of complex-formation in vivo.     

Nicotinamide treatment reduces the levels of histone H3K4 trimethylation in the promoter of the mper1 circadian clock gene and blocks the ability of dexamethasone to induce the acute response

Circadian rhythms, which measure time on a scale of 24h, are generated by one of the most ubiquitous endogenous mechanisms, the circadian clock. SIRT1, a class III histone deacetylase, and PARP-1, a poly(ADP-ribose) polymerase, are two NAD(+)-dependent enzymes that have been shown to be involved in the regulation of the clock. Here we present evidence that the metabolite nicotinamide, an inhibitor of SIRT1, PARP-1 and mono(ADP-ribosyl) transferases, blocks the ability of dexamethasone to induce the acute response of the circadian clock gene, mper1, while it concomitantly reduces the levels of histone H3 trimethylation of lysine 4 (H3K4me3) in the mper1 promoter. Moreover, application of alternative inhibitors of SIRT1 and ADP-ribosylation did not lead to similar results. Therefore, inhibition of these enzymes does not seem to be the mode by which NAM exerts these effects. These results suggest the presence of a novel mechanism, not previously documented, by which NAM can alter gene expression levels via changes in the histone H3K4 trimethylation state.     

Chromosomal protein poly(ADP-ribosyl)ation in pancreatic nucleosomes

When pancreatic chromatin fragments were prepared and resolved in the presence of 80 mM NaCl, endogenous poly(ADP-ribose) polymerase activity was found to be maximal in nucleosome periodicities of four to five units and did not respond to any further increases in nucleosomal architecture. Furthermore, in nucleosome complexities spanning 1 through 14 and over unit lengths, polyacrylamide gel electrophoresis on acid-urea and acid-urea-Triton gels has shown pancreatic histone H1 to be the only actively ADP-ribosylated histone species. The extent of ADP-ribosylation of histone H1 was also demonstrated to retard the protein's mobility in acid-urea, acid-urea-Triton, and lithium dodecyl sulfate polyacrylamide gels and to consist of at least 12 distinct ADP-ribosylated species extractable in all nucleosome complexities studied. Finally, extraction and subsequent electrophoresis of total chromosomal proteins in the presence of lithium dodecyl sulfate also evidenced heavy ADP-ribosylation at the level of nonhistone chromosomal proteins of the high mobility group comigrating in the core histone region, as well as in the topmost region of the gels where poly(ADP-ribose) polymerase was found to form a poly(ADP-ribosyl)ated aggregate.     

The effect of poly(ADP-ribosyl)ation on native and H1-depleted chromatin. A role of poly(ADP-ribosyl)ation on core nucleosome structure

The effect of poly(ADP-ribosyl)ation on native and H1-depleted chromatin was analyzed by gel electrophoresis, electron microscopy, and velocity sedimentation. In parallel, the interaction of automodified poly(ADP-ribose) polymerase with native and H1-depleted chromatin was analyzed. In H1-depleted chromatin histone H2B becomes the major poly(ADP-ribose) histone acceptor protein, whereas in native chromatin histone H1 was the major histone acceptor. Poly(ADP-ribosyl)ation of H1-depleted chromatin prevented the recondensation of polynucleosomes reconstituted with exogenous histone H1. This is probably due to the presence of modified poly(ADP-ribose) polymerase and hyper(ADP-ribosyl)ated histone H2B. Indeed, about 40% of the modified enzyme remained associated with H1-depleted chromatin, while less than 1% of the modified enzyme was bound to native chromatin. The influence of poly(ADP-ribosyl)ation on the chromatin conformation was also studied at the level of nucleosome in using monoclonal and polyclonal antibodies specific for individual histones and synthetic peptides of histones. In native chromatin incubated in the presence of Mg2+ there was a drop in the accessibility of histone epitopes to monoclonal and polyclonal antibodies whereas upon poly(ADP-ribosyl)ation their accessibility was found to remain even in the presence of Mg2+. In poly(ADP-ribosyl)ated H1-depleted chromatin an increased accessibility of some histone tails to antibodies was observed.     

Inhibition of ADP-ribosylation of histone by diadenosine 5', 5"' -p (1), p(4)-tetraphosphate

Diadenosine 5′, 5?-p¹, p⁴-tetraphosphate (Ap4A) strongly inhibited ADP-ribosylation reaction of histone by purified bovine thymus poly(ADP-ribose) polymerase. This compound showed a relatively weak inhibitory effect on Mg²⁺-dependent, enzyme-bound poly(ADP-ribose) synthesis. Among various adenine nucleotides tested, including several diadenosine nucleotides with varying phosphate chain length, Ap4A was the most effective inhibitor of the histone-modification reaction. Ap5A and Ap6A showed slightly lower inhibitory effect than Ap4A. Kinetic analysis of the inhibitor (Ap4A) with varying concentration of substrate (NAD⁺) revealed that this compound is a “mixed type inhibitor”, with a Ki value of 5.1 μM.     

Adenosine diphosphate ribosylation of chicken-erythrocyte histones H1, H5 and high-mobility-group proteins by purified calf-thymus poly(adenosinediphosphate-ribose) polymerase

Poly(ADP-ribosylation) of histones H1, H5 and non-histone chromosomal high-mobility-group proteins HMG 1, 2, 14 and 17 from chicken erythrocytes by purified calf thymus poly(ADP-ribose) polymerase was studied using acid/urea/Triton gel electrophoresis and autoradiography. With histone H1, besides ADP-ribosylated H1 supporting short chains of polymer, the appearance of H1 'dimer' was observed and this reaction was dependent on NAD concentration and incubation time. In addition, highly modified and/or aggregated species of histone H1 were observed. Histone H5 was slightly ADP-ribosylated at low NAD concentrations. At higher NAD concentrations or after longer incubations the formation of H5 'dimer' and of more modified forms of H5 could be observed. HMG 1 and HMG 2 were found to be ADP-ribosylated, the reaction being dependent on NAD concentration and time. Here again some discrete intermediates appeared. HMG 14 and HMG 17 were only slightly ADP-ribosylated under our experimental conditions. These results indicate that the purified DNA-independent poly(ADP-ribose) polymerase can catalyse the formation of H1 'dimer' as in nuclei and nucleosomes and that H5 and HMG proteins can also be ADP-ribosylated and produce well-defined higher complexes. These modifications of nuclear proteins may provide a means of localized conformational changes of the chromatin structure in vivo.