Alterations in epidermal growth factor binding to hepatic membranes following dietary exposure of rats to known hepatocarcinogens.

Administration of the chemical carcinogen 2-acetylaminofluorene (2-AAF) has previously been shown to lower hepatic epidermal growth factor (EGF) binding levels during chemically induced hepatocarcinogenesis. To further characterize the specificity of this response, EGF binding levels for liver microsomes were determined after a 3-week administration of subacute doses of 2-AAF and five other known hepatocarcinogens: 3'-methyl-4-dimethylaminoazobenzene (3'Me-DAB), 2-AAF, aflatoxin B1 (AFB1), thioacetamide (TA), ethionine, benzidine (Benz), as well as four non-hepatocarcinogens: fluorene, p-aminoazobenzene, 4-acetylaminofluorene (4-AAF), and 3-methylcholanthrene. Five of six of the hepatocarcinogens tested (3'Me-DAB, 2-AAF, TA, AFB1 and Benz) caused significant lowering of EGF binding levels, and one of the four non-hepatocarcinogens (4-AAF) caused significant lowering of EGF binding levels. Paired feeding studies indicated that the decreases in EGF binding levels were not a result of differences in net diet consumption. These findings show that decreases in EGF binding capacity are caused by a diverse group of known hepatocarcinogenic compounds at an early stage in the carcinogenesis process.

Changes in the binding capacity of hepatic membranes for epidermal growth factor during multistage hepatocarcinogenesis in rats.

To study changes in hepatic capacity for binding epidermal growth factor (EGF) during 2-acetylaminofluorene (2-AAF)-induced, multistage hepatocarcinogenesis, a 5 cycle protocol of discontinuous 2-AAF administration was used to produce hepatocarcinogenesis in rats. The hallmark of the 5 cycle protocol is that rats fed 1 to 3 cycles of 2-AAF are at low risk for cancer, while rats fed 2-AAF for 4 or 5 cycles are at high risk for cancer. EGF binding by liver membranes was found to be lowered to 20-25% of control throughout the 5 cycle regimen. When the persistence of lowered EGF binding was tested by returning rats fed 2-AAF for 1 to 3 cycles to diet without 2-AAF for 3 weeks, binding was found to recover to 80 to 90% of values for control rats. In contrast, for rats fed 2-AAF for 4 or 5 cycles, EGF binding capacity remained low, 30 to 40% of control, following placement of rats on diet without 2-AAF for 3 weeks. Immunochemical analysis indicated a close correspondence between changes in EGF receptor levels and changes in the above EGF binding levels. These studies show that during the 2-AAF protocol, the 2-AAF-mediated loss in hepatic EGF binding capacity and EGF receptor protein undergo a transition from a reversible loss to a persistent loss in binding capacity, and EGF receptor protein, as rats underwent a change from low to high risk for developing hepatocarcinomas. The persistent decrease in hepatic EGF binding level may be associated with the progression stage of hepatocarcinogenesis.

Poly(ADP-ribose) quantification at the femtomole level in mammalian cells.

ADP-ribose polymers were isolated from living mammalian cells, separated by polyacrylamide gel electrophoresis, visualized in the gel with a novel silver staining agent, and quantified by computer-aided scanning densitometry. This method detects as little as approximately 40 fmol of ADP-ribose polymers of a particular size class and reduces the gel exposure times required for conventionally radiolabeled polymers from 2 months to about 1 h. The method also detects polymers with slow turnover which may be underestimated by techniques requiring metabolic radiolabeling of poly(ADP-ribose).

Interactions of poly(ADP-ribose) with nuclear proteins.

The molecular mechanisms whereby poly(ADP-ribosyl)ation primes chromatin proteins for an active role in DNA excision repair are not understood. The prevalent view is that the covalent linkage of ADP-ribose polymers is essential for the modification of target protein function. By contrast, we have focused on the possibility that ADP-ribose polymers interact non-covalently with nuclear proteins and thereby modulate their function. The results show that ADP-ribose polymers engage in highly specific and strong non-covalent interactions with a small number of nuclear proteins, predominantly histones, and among these only with specific polypeptide domains. The binding affinities were largely determined by two factors, ie the polymer sizes and the presence of branches. This provides an explanation for the target specificity of the histone shuttle mechanism that was previously reported by our laboratory. Interestingly, the polymer molecules being most effective in protein targeting in vitro, are strictly regulated in mammalian cells during DNA repair in vivo.

Histone shuttling by poly ADP-ribosylation.

The enzymes poly(ADP-ribose)polymerase and poly(ADP-ribose) glycohydrolase may cooperate to drive a histone shuttle mechanism in chromatin. The mechanism is triggered by binding of the N-terminal zinc-finger domain of the polymerase to DNA strand breaks, which activates the catalytic activities residing in the C-terminal domain. The polymerase converts into a protein carrying multiple ADP-ribose polymers which displace histones from DNA by specifically targeting the histone tails responsible for DNA condensation. As a result, the domains surrounding DNA strand breaks become accessible to other proteins. Poly(ADP-ribose)glycohydrolase attacks ADP-ribose polymers in a specific order and thereby releases histones for reassociation with DNA. Increasing evidence from different model systems suggests that histone shuttling participates in DNA repair in vivo as a catalyst for nucleosomal unfolding.

DNA strand break-mediated partitioning of poly(ADP-ribose) polymerase function.

The nuclear enzyme poly(ADP-ribose) polymerase participates in DNA excision repair. Following binding to DNA strand breaks through its amino-terminal Zn(2+)-finger domain, the enzyme is activated to form polymerase-associated ADP-ribose polymers of various sizes. Focusing on this "automodification" reaction, we observed that optimal enzyme activity and maximal polymer formation were attained only at a strict stoichiometry of two polymerase molecules per DNA fragment. Using various linearized DNAs and nicked circular DNA, we show that this stoichiometric dependence is dictated by the number of enzyme activating sites, i.e., DNA strand breaks. Deviations from the optimal ratio inevitably resulted in decreased polymer formation, ruling out a strict automodification mechanism of poly(ADP-ribosyl)ation. Our results suggest that the mechanism of poly(ADP-ribose) formation on polymerase molecules entails DNA strand break-mediated partitioning of the polymerase into two functional populations: one bound to the DNA breaks and catalytically active, the other, catalytically inactive, functioning as polymer acceptors.

Endoglycosidic cleavage of branched polymers by poly(ADP-ribose) glycohydrolase.

Post-translational modification of nuclear proteins with poly(ADP-ribose) modules chromatin structure and may be required for DNA processing events such as replication, repair and transcription. The polymer-catabolizing enzyme, poly(ADP-ribose) glycohydrolase, is crucial for the regulation of polymer metabolism and the reversibility of the protein modification. Previous reports have shown that glycohydrolase digests poly(ADP-ribose) via an exoglycosidic mechanism progressing from the protein-distal end of the polymer. Using two independent approaches, we investigated the possibility that poly(ADP-ribose) glycohydrolase also engages in endoglycosidic cleavage of polymers. First, partial glycohydrolase digestion of protein-bound poly(ADP-ribose) led to the production of protein-free oligomers of ADP-ribose. Second, partial glycohydrolase digestion of a fixed number of protein-free poly(ADP-ribose) polymers resulted in a transient increase in the absolute number of polymers while polymer size continuously decreased. Furthermore, endoglycosidic activity produced linear polymers from branched polymers although branch points themselves were not a preferential target of cleavage. From these data, we propose a mechanism whereby poly(ADP-ribose) glycohydrolase degrades polymers in three distinct phases; (a) endoglycosidic cleavage, (b) endoglycosidic cleavage plus exoglycosidic, processive degradation, (c) exoglycosidic, distributive degradation.

DNA precursors are channelled to nuclear matrix DNA replication sites.

Studies of replicative DNA synthesis using DNA precursors have shown that the DNA that was replicated most recently is that associated with the nuclear matrix. Consequently, precursors arising via the salvage and the de novo metabolic pathways are first incorporated into a small percentage of the total nuclear DNA that is termed nuclear matrix-associated DNA. These results have been substantiated in cell culture, as well as in intact mammalian systems. Furthermore, when DNA precursors were injected intravenously into regenerating rat liver, a significant lag in the incorporation of orotic acid-derived nucleotides (de novo pathway precursors) into nuclear DNA was observed, when compared with deoxythymidine-derived nucleotides (salvage pathway precursors). This lag in incorporation kinetics was also evident at the nuclear matrix level, although, once incorporated into nuclear matrix-associated DNA, the distribution patterns of both precursors into extra-matrix nuclear DNA fractions were identical. To determine the basis for this kinetic lag, we compared the incorporation kinetics of orotic acid and of deoxythymidine into dTTP and into nuclear matrix-associated DNA, respectively. Orotic acid-derived nucleotides entered the cytosolic dTTP pool before being incorporated into nuclear matrix-associated DNA, that is, traversing the classical metabolic route of DNA precursors. Conversely, deoxythymidine-derived nucleotides by-passed the soluble dTTP cellular pool and engaged directly in DNA synthesis at the nuclear matrix. Not only is this the first evidence for nucleotide channelling in an intact mammalian system, but it also forms direct evidence that salvage pathway DNA precursors are channelled to nuclear matrix-associated sites of DNA replication.

Targeting of histone tails by poly(ADP-ribose).

After Zn2+ finger-mediated binding to a DNA break, poly(ADP-ribose) polymerase becomes automodified with long polymers of ADP-ribose. These nucleic acid-like polymers may facilitate DNA repair by noncovalently interacting with neighboring proteins. Using a novel screening technique, we have identified histones as the predominant poly(ADP-ribose)-binding species in human keratinocytes, rat hepatocytes, frog eggs, and yeast. Polymer binding is confined specifically to the histone domains responsible for DNA condensation, i.e. histone tails. Our results indicate that polymers of ADP-ribose are targeted to sites of DNA strand breaks by poly(ADP-ribose) polymerase and subsequently function to alter chromatin conformation through noncovalent interactions with histone tails.

The carboxyl-terminal domain of human poly(ADP-ribose) polymerase. Overproduction in Escherichia coli, large scale purification, and characterization.

The cDNA encoding the carboxyl-terminal 40-kDa domain of human poly(ADP-ribose) polymerase was inserted into an expression vector. The recombinant protein was overproduced in Escherichia coli, and purified to homogeneity. The 40-kDa domain had the same affinity (Km) for NAD+ as the full-length enzyme, expressed abortive NAD+ glycohydrolase activity, catalyzed the initiation, elongation, and branching of ADP-ribose polymers, but exhibited no DNA dependence. Its specific activity was approximately 500-fold lower than that of the whole enzyme activated by DNA strand breaks. Surprisingly, the carboxyl-terminal 40-kDa domain exhibited the processive mode of polymer attachment typical of full-length poly(ADP-ribose) polymerase and was able to modify histones H1 and H2B. Finally, the polymer sizes formed by the 40-kDa domain were influenced by histone H1.

Synthesis of poly(ADP-ribose)-agarose beads: an affinity resin for studying (ADP-ribose)n-protein interactions.

Polymers of ADP-ribose bind chromatosomal histones in solution and may play a role in chromatin accessibility in vivo. We have enzymatically synthesized a poly(ADP-ribose) affinity resin to further characterize binding of nuclear proteins to ADP-ribose polymers. NAD+- and (ADP-ribose)-derivatized agarose beads were recognized as polymer acceptors by the nuclear enzyme poly(ADP-ribose) polymerase. This polymerase elongated the existing ligands by successive addition of exogenously available ADP-ribose residues to form polymers covalently linked to the agarose beads. Poly(ADP-ribose) formation on the beads was dependent on incubation time and the mode of ligand attachment to the agarose. The resulting poly(ADP-ribose)-derivatized agarose beads possessed polymers which closely resembled those modifying the ADP-ribose polymerase by the automodification reaction. Fractionation of rat liver nuclear lysate over the poly(ADP-ribose) resin revealed a strong affinity of H1 for ADP-ribose polymers, thereby supporting a role for poly(ADP-ribose) in chromatin functions. Poly(ADP-ribose)-agarose beads are extremely stable and will be useful not only for affinity studies, but also for mechanistic studies involving polymer elongation and catabolism.

The alpha-glycosidic bonds of poly(ADP-ribose) are acid-labile.

The poly(ADP-ribosyl)ation system of higher eukaryotes produces multiple ADP-ribose polymers of distinct sizes which exhibit different binding affinities for histones. Although precipitation with trichloroacetic acid (TCA) is the standard procedure for isolation of poly(ADP-ribose) from biological material, we show here that poly(ADP-ribose) is not stable under acidic conditions. Storage of poly(ADP-ribose) as TCA pellets results in acid hydrolysis of polymers, the extent of which is dependent on storage time and temperature. The alpha-glycosidic, inter-residue bonds are the preferred sites of attack, thus reducing polymer sizes by integral numbers of ADP-ribose to yield artefactually more and smaller polymers than originally present. Therefore, poly(ADP-ribosyl)ation studies involving TCA precipitation, histone extraction with acids, or acidic incubations of ADP-ribose polymers must account for the impact of acids on resulting polymer populations.

Noncovalent interactions of poly(adenosine diphosphate ribose) with histones.

Covalent linkage of ADP-ribose polymers to proteins is generally considered essential for the posttranslational modification of protein function by poly(ADP-ribosyl)ation. Here we demonstrate an alternative way by which ADP-ribose polymers may modify protein function. Using a highly stringent binding assay in combination with DNA sequencing gels, we found that ADP-ribose polymers bind noncovalently to a specific group of chromatin proteins, i.e., histones H1, H2A, H2B, H3, and H4 and protamine. This binding resisted strong acids, chaotropes, detergents, and high salt concentrations but was readily reversible by DNA. When the interactions of variously sized linear and branched polymer molecules with individual histone species were tested, the hierarchies of binding were branched polymers greater than long, linear polymers greater than short, linear polymers and H1 greater than H2A greater than H2B = H3 greater than H4. For histone H1, the target of polymer binding was the carboxy-terminal domain, which is also the domain most effective in inducing higher order structure of chromatin. Thus, noncovalent interactions may be involved in the modification of histone functions in chromatin.

Evidence for nucleoside channeling in vivo: deoxythymidine incorporation into rat liver dTTP and nuclear matrix DNA.

Previous studies in prokaryotes and in eukaryotic cell lines have indicated the possible existence of more than one dTTP pool accessible to DNA synthesis. To investigate this possibility in eukaryotes in vivo, the incorporation of [3H] deoxythymidine into nuclear matrix-attached DNA and intracellular dTTP was examined in regenerating rat liver. The labeling of matrix DNA reached a maximum after a 5 min pulse and then began to rapidly decrease. Conversely, [3H] deoxythymidine incorporation into dTTP began to increase after 5 min and peaked 10 min after injection. Since the peak specific activity for [3H] deoxythymidine incorporation into matrix DNA precedes that into dTTP, there seems to be channeling of exogenous thymidine directly to sites of DNA replication, bypassing existing nucleotide pools.