Quantitative Studies of Inhibitors of ADP-ribosylation in Vitro and in Vivo*

The ADP-ribosyl moiety of NAD+ is consumed in reactions catalyzed by three classes of enzymes: poly(ADP-ribose) polymerase, protein mono(ADP-ri- bosyl)transferases, and NAD+ glycohydrolases. In this study, we have evaluated the selectivity of compounds originally identified as inhibitors of poly(ADP-ribose) polymerase on members of the three classes of en- zymes. The 50% inhibitory concentration (ICao) of more than 20 compounds was determined in vitro for both poly(ADP-ribose) polymerase and mono(ADP-ri-bosy1)transferase A in an assay containing 300 NM NAD+. Of the compounds tested, benzamide was the most potent inhibitor of poly(ADP-ribose) polymerase with an ICs0 of 3.3 p ~ . The ICs0 for benzamide for mono(ADP-ribosy1)transferaseA was 4.1 mM, and sim- ilar values were observed for four additional cellular mono(ADP-ribosy1)transferases. The ICao for NAD+ glycohydrolase for benzamide was approximately 40 For seven of the best inhibitors, inhibition of poly(ADP-ribose) polymerase in intact C3H10T1/2 cells was studied as a function of the inhibitor concentration adjusted to a final of (w/v) trichloroacetic To determine NAD', trichloroacetic acid supernatant was ml ammonium adjusted pH 8.6 with concentrated ammonium sample was applied to a DHB-Sepharose column which had been prewashed with 10 ml of 250 mM ammonium formate, The was washed with ml of mM ammonium formate, pH ml NAD' was 4 ml 250 mM ammonium formate, pH 4.5, and quantified To determine ADP-ribose the acid-insoluble

The ADP-ribosyl moiety of NAD+ is consumed in reactions catalyzed by three classes of enzymes: poly(ADP-ribose) polymerase, protein mono(ADP-ribosyl)transferases, and NAD+ glycohydrolases. In this study, we have evaluated the selectivity of compounds originally identified as inhibitors of poly(ADP-ribose) polymerase on members of the three classes of enzymes. The 50% inhibitory concentration (ICao) of more than 20 compounds was determined in vitro for both poly(ADP-ribose) polymerase and mono(ADP-ri-bosy1)transferase A in an assay containing 300 NM NAD+. Of the compounds tested, benzamide was the most potent inhibitor of poly(ADP-ribose) polymerase with an ICs0 of 3.3 p~. The ICs0 for benzamide for mono(ADP-ribosy1)transferaseA was 4.1 mM, and similar values were observed for four additional cellular mono(ADP-ribosy1)transferases. The ICao for NAD+ glycohydrolase for benzamide was approximately 40 For seven of the best inhibitors, inhibition of poly(ADP-ribose) polymerase in intact C3H10T1/2 cells was studied as a function of the inhibitor concentration of the culture medium, and the concentration for 50% inhibition (culture medium ICso) was determined. Culture medium ICao values for benzamide and its derivatives were very similar to in vitro ICso values. For other inhibitors, such as nicotinamide, 5-methylnicotinamide, and 5-bromodeoxyuridine, culture medium values were 3-5-fold higher than in vitro ICs0 values. These results suggest that micromolar levels of the benzamides in the culture medium should allow selective inhibition of poly(ADP-ribose) metabolism in intact cells. Furthermore, comparative quantitative inhibition studies should prove useful for assigning the biological effects of these inhibitors as an effect on either poly(ADP-ribose) or mono(ADP-ribose) metabolism. mM.
NAD+ is a cosubstrate in numerous hydride transfer reactions central to intermediary metabolism. It is also a substrate for a second distinct class of enzymes that catalyze the cleavage of the glycosylic linkage between nicotinamide and ribose and transfer of the ADP-ribosyl moiety to an acceptor. En-* This work was supported by Grant CA43894 from the National Institutes of Health and by the Texas Veterans of Foreign Wars Ladies Auxiliary 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 "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. zymes that use proteins as acceptors include the nuclear enzyme poly(ADP-ribose) polymerase, which also catalyzes synthesis of polymers of ADP-ribose (l), and a second group of mono(ADP-ribosy1)transferases that transfer only a single ADP-ribose residue t o acceptors (2). Although both poly(ADP-ribose) polymerase and mono(ADP-ribo-sy1)transferases can catalyze the hydrolysis of NAD' to ADPribose and nicotniamide, they preferentially transfer ADPribose to proteins. A third class of enzymes, termed NAD' glycohydrolases, preferentially catalyze the hydrolysis of NAD+ (3).
The biological functions of ADP-ribosylation reactions are poorly understood. Many studies concerned with elucidating their biological functions have used compounds originally identified as inhibitors of poly(ADP-ribose) polymerase (4, 5). Studies using these inhibitors have shown that they alter many processes including cellular recovery from DNA damage (6, 7), sister chromatid exchange (8), malignant transformation (g-ll), DNA replication (12), and cellular differentiation (13,14). Whereas the effects of these inhibitors have been generally attributed to inhibition of poly(ADP-ribose) polymerase, the lack of absolute selectivity with regard to ADPribose-transferring enzymes or other enzymes makes the link to poly(ADP-ribose) polymerase a tenuous one. In this study, we have examined compounds originally identified as inhibitors of poly(ADP-ribose) polymerase for their relative inhibition of several classes of ADP-ribose-transferring enzymes and for their efficacy of inhibition of poly(ADP-ribose) polymerase in vivo.

Methods
Determination of PolyfADP-ribose) Polymerase Activity in Vitro-Poly(ADP-ribose) polymerase was partially purified from calf thymus as described previously (17). Different preparations had specific ac-tivities of 200-500 units mg" of protein. One unit is defined as the amount of enzyme that converts 1 nmol of NAD' to poly(ADPribose)/min under the assay conditions described here. The enzyme preparation was dependent upon exogenous DNA for activity, and pBR322 DNA digested with HaeIII endonuclease (17) was used as a defined source of activating DNA. Assays of 200 pl contained 50 mM Tris-C1, pH 7.2, 1.0 mM EDTA, 300 ,UM [32P]NAD' (1 X IO6 cpm), 0.2 pg of DNA, and 1-2 units of enzyme. After incubation for 2 min at 37 "C, 3.0 ml of 20% trichloroacetic acid and 100 pg of bovine serum albumin were added, and samples were mixed and placed on ice. Acid-insoluble product was collected onto Whatman GF/A glass fiber filters and washed three times with 10 ml of 5% trichloroacetic acid and twice with 10 ml of 95% ethanol. Filters were air-dried, and radiolabel was determined by liquid scintillation counting.
Determination of Mono(ADP-ribosyl)transferaseActivity in Vitro-Mono(ADP-ribosy1)transferase A was purified to apparent homogeneity as described previously (18). Other mono(ADP-ribo-sy1)transferases were purified as described (19).' Transferase A had a specific activity of 156 units mg". Assays of 200 p l contained 100 mM Hepes? pH 7.5, 300 mM NaC1, 1.8 mg of histone, 300 p M [''PI-NAD' (1 X 106 cpm), and approximately 1 milliunit of enzyme. After incubation at 30 "C for 30 min, reactions were terminated with 3 ml of 20% trichloroacetic acid and placed on ice. Samples were filtered, and radiolabel was determined as described above.
Determination of NAD+ Glycohydrolase Activity in Vitro-Partially purified porcine brain NAD' glycohydrolase was obtained from Sigma. For assay, 5 mg (0.035 unit) of acetone powder was added to 1.5 ml of Hz0 and suspended by two 5-s bursts of sonification in an ice bath. Insoluble material was removed by centrifugation for 10 min at 2000 X g. The 200-pl assay contained 67 mM potassium phosphate buffer, pH 7.3,300 pM [3H]nicotinamide-labeled NAD+ (3 X lo5 cpm), and 20 pl of enzyme preparation. After incubation for 10 min at 37 "C, the reaction mixture was diluted to 5.0 ml with 1.0 M ammonium formate buffer, pH 9.0, containing 10 mM EDTA and immediately applied to a 0.5-ml column of dihydroxy-Bio-Rex that had been prewashed with 10 ml of the same buffer. The column flow-through containing the released nicotinamide was quantified by liquid scintillation counting. The release of nicotinamide was linear with time and did not exceed 10% of the total substrate. This enzyme preparation was also examined for poly(ADP-ribose) polymerase or protein mono(ADP-ribosy1)transferase activity by utilizing NAD+ radiolabeled in the adenine ring (21) and examining for the formation of an acid-insoluble product. No acid-insoluble product was detected.
Determination of Poly(ADP-ribose) Polymerase Activity in Vivo-C3HlOT1/2 cells were maintained as described previously (20). Confluent cultures in 35-mm dishes were radiolabeled for 16 h with 1 ml of medium containing 20 pCi of [3H]adenine and NAD+, and ADPribose polymer levels were determined using methodology whose validation is described elsewhere (21). Medium was replaced, and cells were incubated for 30 min in the presence of inhibitors and then treated with 100 p M N-methyl-N-nitroso-N'-nitroguanidine (MNNG). MNNG was prepared as a 20 mg ml" stock solution in acetone. Acetone was present in the medium at a final level of 0.5% following addition of MNNG. Incubation was terminated by removal of medium, washing with phosphate-buffered saline, and addition of 1 ml of ice-cold 20% trichloroacetic acid. Cellular residue was removed from the dishes by scraping with a rubber policeman, and the acidinsoluble materia1 was collected by centrifugation. The supernatant was removed and saved for NAD' determination. The pellet was dissolved in 0.2 ml of ice-cold 98% formic acid and diluted by the addition of 10 volumes of ice-cold deionized H20. Bovine serum albumin (1 mg) was added to facilitate reprecipitation, samples were adjusted to a final concentration of 20% (w/v) trichloroacetic acid; and the acid-insoluble fraction was collected by centrifugation. To determine NAD', the trichloroacetic acid supernatant was diluted to 10 ml with 250 mM ammonium formate, pH 8.6, and adjusted to pH 8.6 f. 0.2 with concentrated ammonium hydroxide. The sample was applied to a 0.5-ml DHB-Sepharose column which had been prewashed with 10 ml of 250 mM ammonium formate, pH 8.6. The column was washed with 10 ml of 250 mM ammonium formate, pH 8.6, followed by 2 ml of H20. NAD' was eluted with 4 ml of 250 mM ammonium formate, pH 4. pellet was dissolved in 1 ml of 6 M guanidinium chloride, 250 mM ammonium acetate, 10 mM EDTA, pH 6.0. One ml of 1 M KOH, 100 mM EDTA was added, and the samples were incubated at 37 "C for 2 h. Samples were then diluted to 10 ml with 1 M quanidinium chloride, 250 mM ammonium acetate, 10 mM EDTA, pH 9.0 (Buffer A), adjusted to pH 9.0 ? 0.2 using HCI, and applied to a 0.5-ml column of DHB-Bio-Rex that had been prewashed with 5 ml of Hz0 and 10 ml of Buffer A. Following application, the column was washed with 25 ml of Buffer A, followed by 10 ml of 1 M ammonium bicarbonate, 1 mM EDTA, pH 9.0. Poly(ADP-ribose) was eluted with 5 ml of Hz0 and quantified by liquid scintillation counting.

Polymerase and Mono(ADP-ribosy1)transferase A in Vitro-
To examine inhibition patterns in vitro, a partially purified preparation of poly(ADP-ribose) polymerase from calf thymus (17) was compared with homogeneous turkey erythrocyte NAD+:arginine mono(ADP-ribosy1)transferas.e A (18). They are referred to here as polymerase and transferase, respectively. The NAD' concentration of both assays was 300 pM, which was selected because it approximates the estimated intracellular concentration of NAD' in cultured mouse cells To evaluate each compound, varying concentrations were competed against 300 M M NAD+. Fig. 2 shows representative data of polymerase activity as a function of the concentration of 3-methoxybenzamide, 3-aminobenzamide, nicotinamide, nicotinate, and benzoate. The three amides have previously been reported to inhibit the polymerase (4, 5), and dosedependent inhibition for each of these compounds was also observed under the conditions used in this study. Nicotinate and benzoate have been reported to be noninhibitory (4, 5), and they likewise did not show significant inhibition in these assays. Data from multiple experiments such as those shown in Fig. 2 were combined, and a best fit curve was generated by computer analysis (23). From this analysis, a 50% inhibitory concentration (IG0), defined as the concentration which inhibited enzyme activity by 50% under the assay conditions, was determined. Table 1 lists ICs0 values for 23 compounds examined. In addition, an ICs0 relative to that for benzamide was calculated by dividing each IC50 by the ICs0 for benzamide, the most potent inhibitor. The most effective inhibitors were the benzamides, with ICs0 values between 3 and 6 p~. The best inhibitor of the polymerase which was not a benzamide derivative was 5-bromodeoxyuridine with an ICs0 of 15 p~.
Nicotinamide, theophylline, and thymidine were also relatively effective inhibitors with ICs0 values ranging from 31 to 46 pM. Fig. 3 shows representative data for inhibition of transferase A by 5-bromodeoxyurid1ne, 3-methoxybenzamide, 3-aminobenzamide, and benzoic acid. Benzoic acid had no effect on transferase activity, whereas dose-dependent inhibition was evident for the three other compounds. The ICs0 values of 21 compounds examined with the transferase were calculated using computer analysis and are shown in Table 2 of 590 and 1900 p~, respectively, followed by the benzamides, which yielded IC50 values of between 2700 and 4100 p~. For many compounds tested, the IC50 values were greater than 30 mM, and these compounds were designated as noninhibitory.
Comparison of Tables 1 and 2 reveals that IC50 values of all compounds tested were much higher for transferase than for polymerase. This is illustrated in Fig. 4, which compares the in vitro inhibition curves for 3-aminobenzamide, nicotinamide, 5-methylnicotinamide, and 5-bromodeoxyuridine. Table 3 lists ratios of the ICso values of transferase relative to polymerase for eight selected inhibitors. The largest difference in IC50 values was seen for benzamide, in which the IC50 value for the transferase was higher by more than three orders of magnitude. Such a marked quantitative difference in sensitivity is potentially useful for discriminating between effects of inhibitors on the polymerase and transferases in intact cells, providing that similar inhibition patterns are observed in vivo.
Inhibition Patterns of Poly(ADP-ribose) Polymerase in Vivo-The term in vivo is used here to refer to C3H10T1/2 cells maintained in culture. To study inhibition patterns in vivo, the effect of inhibitors on the accumulation of ADPribose polymers was examined following treatment of C3H10T1/2 cells with the alkylating agent MNNG. Previous studies have established that this treatment results in a rapid activation of poly(ADP-ribose) polymerase, resulting in elevated levels of ADP-ribose polymers (20,23). Fig. 5 shows a time course of the NAD+ and poly(ADP-ribose) content of cells following treatment with 100 p~ MNNG. These conditions resulted in a maximal accumulation of poly(ADP-ribose) at 30 min following treatment, and this treatment period was therefore chosen to study effects of the inhibitors. However, for nicotinamide, 5-bromodeoxyuridine, and 5methylnicotinamide, the i n vivo patterns were shifted to considerably higher concentrations. Table 4 shows calculated culture medium IC50 values for the seven inhibitors and ratios of culture medium IC50 values to in vitro IC60 values. The ratio for all the benzamides tested was close to 1.0. However, the ratios for nicotinamide, 5-methylnicotinamide, 5-bromodeoxyuridine, and theophylline ranged from 3.2 to 5.0.

Inhibition of Other Protein Mono(ADP-ribosy1)transferases and NAD' Glycohydrolase in Vitro-Two results described
here suggest that the benzamides may prove very useful in discriminating between effects on polymerase and transferases in intact cells. First, there are large quantitative differences in the relative sensitivity of the polymerase as compared to transferase A in vitro (Fig. 4). Second, there is a close correspondence between in vitro inhibition of the polymerase and the inhibition of poly(ADP-ribose) accumulation in uivo (Fig. 6). However, since multiple endogenous transferases are present in cells, the utility of this approach would depend upon whether the inhibition pattern of transferase A is representative of other endogenous transferases. Additionally, it was also of interest to determine the effects of these inhibitors on a third class of NAD+-consuming enzymes, NAD' glycohydrolase. Thus, inhibition curves were generated for four additional transferases (19) and for NAD' glycohydrolase. Fig. 7 shows in vitro inhibition by benzamide of each of the five endogenous transferases and NAD+ glycohydrolase in relation to the polymerase. The data show that all of the transferases had similar inhibition patterns, whereas the NAD+ glycohydrolase was less sensitive to inhibition than the transferases. Additional experiments using 3-methoxybenzamide and 3-aminobenzamide also demonstrated equivalent responses by the five transferases (data not shown).

DISCUSSION
A number of studies have quantitatively evaluated the inhibition of poly(ADP-ribose) polymerase in vitro (4,5,(29)(30)(31), but very little quantitative information concerning the inhibition of other classes of ADP-ribosyl-transferring enzymes or on the efficacy of inhibitors in intact cells has been reported. The existence of multiple endogenous mono(ADP-ribosy1)transferases (25) makes information concerning the selectivity of these inhibitors especially important with regard to assessing the scope of mono(ADP-ribose) metabolism in total cellular ADP-ribosylation and in assigning biological effects of these inhibitors. Whereas the use of inhibitors in intact cells has many limitations, the lack of genetic approaches to ADP-ribosylation in mammalian cells makes their use important in understanding the physiological functions of ADP-ribosylation. Whereas this study has focused on the selectivity of these inhibitors between different classes of ADP-ribosyl-transferring enzymes, the possible action of the inhibitors on unrelated enzymes must also be a concern when evaluating biological effects of these compounds. For example, studies have suggested that the benzamides alter nucleotide synthesis (26,27), although the studies of Hunting et al. (28) are most consistent with effects on nucleotide labeling that are secondary to effects on folate metabolism. In general, it has been necessary to employ medium concentrations of 5 mM and higher to observe effects on nucleotide labeling. Whereas the studies described here would argue that effects on nucleotide labeling are not likely due to inhibition of poly(ADP-ribose) polymerase, our data would be consistent with these effects being related to inhibition of mono(ADP-ribosy1)transferases or of unrelated enzymes.
The objective of the in vitro analyses described here was to generate data useful for interpreting the observed effects of inhibitors on intact cells rather than to provide a detailed kinetic analysis of each compound. Nevertheless, the ICs0 The results presented here are in general agreement with previous studies (4, 5) that have shown that benzamide and its derivatives are very potent inhibitors of poly(ADP-ribose) polymerase in vitro. This study has demonstrated that relatively low medium concentrations of these compounds effectively inhibit poly(ADP-ribose) polymerase in vivo. It is difficult to determine accurately the intracellular concentration of these compounds available to poly(ADP-ribose) polymerase due to several uncertainties including partitioning of the compound between aqueous and membrane phases and distribution between different intracellular compartments. HOW-ever, the close correspondence between in vitro and in vivo inhibition curves for the polymerase is consistent with the possibility that the concentration of the benzamides in the nucleus closely approximates the extracellular concentration. However, our results are also noteworthy in that many of the inhibitors examined did not show a close correspondence between in vitro and in vivo inhibition curves. This lack of correlation could be due to several factors including membrane permeability, transport, metabolism, and intracellular compartmentalization. Nevertheless, such observations suggest that these inhibitors have a more limited utility in intact cells than the benzamides.
The results shown here have demonstrated that the benzamides are effective inhibitors of poly(ADP-ribose) polymerase in vivo at concentrations in the medium which are much lower than those generally used in studies designed to assess the effects of ADP-ribosylation inhibitors on biological responses (4-14). The large quantitative differences between the in vitro inhibition curves of poly(ADP-ribose) polymerase and the mono(ADP-ribosy1)transferases examined suggest that it may be possible to inhibit poly(ADP-ribose) polymerase with minimal effects on mono(ADP-ribosy1)transferases. However, it should be noted that the inhibition curves of endogenous mono(ADP-ribosy1)transferases shown here (Fig.  7) have all been generated with transferases specific for arginine residues. Endogenous transferases specific for cysteine (32) and modified histidine (33) residues have also been reported, and it remains to be determined whether or not these (ADP-ribosy1)transferases have similar inhibition properties.
The large quantitative differences between ICs0 values observed for poly(ADP-ribose) polymerase and mono(ADP-ri-bosy1)transferases may prove useful in assessing the mechanism of the biological effects of the benzamides. For example, benzamide and its derivatives have been shown to be relatively nontoxic to cells alone, but they enhance the cytotoxicity of DNA-alkylating agents (6, 7). Previous studies have used concentrations in the culture medium of 1 mM and higher to achieve an enhancement of cytotoxicity. In view of the results described here, we have re-examined the co-cytotoxic property of benzamide at lower concentrations and have observed maximal enhancement of the cytotoxicity of MNNG at micromolar concentration^.^ When quantitatively comparing the effects of inhibitors on biological end points such as cell survival, an exact correspondence between inhibition curves of the target enzyme and the biological effect is not necessarily expected since the process must be inhibited to the extent that it becomes rate-limiting for the end point measured. Nevertheless, the fact that benzamide was co-cytotoxic at similar concentrations to those that inhibited poly(ADPribose) synthesis in vivo, which is far below that where effects on other ADP-ribosyl-transferring enzymes are expected, argues that poly(ADP-ribose) polymerase is the likely target for the co-cytotoxic effects of benzamide. Furthermore, it suggests that even relatively low levels in tissues may be effective in enhancing the cytotoxicity of chemotherapeutic DNAdamaging agents.