Induction of p21 mediated by reactive oxygen species formed during the metabolism of aziridinylbenzoquinones by HCT116 cells.

Aziridinylbenzoquinones are a group of antitumor agents that elicit cytotoxicity by generating either alkylating intermediates or reactive oxygen species. The mechanism of toxicity may not always, however, involve profound damage of cellular constituents, but may involve a cytostatic effect through interference with the cell cycle. In this context, we have examined the induction of the cell cycle inhibitor p21 (WAF1, CIP1, or sdi1), whose overexpression suppresses the growth of various tumor cells, in human tumor cells metabolizing 3,6-diaziridinyl-1,4-benzoquinone (DZQ) and its C2,C5-substituted derivatives: 2,5-bis-(carboethoxyamino) (AZQ) and 2,5-bis-2(-hydroxyethylamino) (BZQ). Both DZQ and AZQ were effectively activated by HCT116 human colonic carcinoma cells; the activation of the former involved largely a dicoumarol-sensitive activity, whereas that of the latter appeared to be accomplished primarily by one-electron transfer reductases. BZQ was not a substrate for the dicoumarol-sensitive enzyme in HCT116 cells. Cellular activation of the first two quinones was associated with formation of oxygen-centered radicals as detected by EPR in conjunction with the spin trap 5,5′-dimethyl-1-pyrroline-N-oxide. The redox transitions of DZQ involved hydroxyl radical formation and were strongly inhibited by catalase, whereas those of AZQ showed a strong superoxide anion component sensitive to superoxide dismutase. These signals were suppressed by N-acetylcysteine with concomitant production of a thiyl radical adduct. This suggests an effective electron transfer between the thiol and free radicals formed during the activation of these quinones. DZQ and AZQ induced significantly the expression of p21 in HCT116 cells, but a 10-fold higher concentration of AZQ was required to achieve the level of induction elicited by DZQ. BZQ had little effect on p21 expression. p21 induction at both mRNA and protein levels correlated with the inhibition of either cyclin-dependent kinase activity or cell proliferation. p21 induction elicited by the above quinones was inhibited by N-acetylcysteine, whereas the non-sulfur analog, N-acetylalanine, was without effect. Catalase and superoxide dismutase did not effect p21 induction by aziridinylbenzoquinones in HCT116 cells, thus suggesting that extracellular sources of oxygen radicals generated by plasma membrane reductases have no influence in the expression of this gene. Hydrogen peroxide, a product of quinone redox cycling, elicited an increase of p21 mRNA levels in HCT116 and K562 human chronic myelogenous leukemia cells. The latter lacks p53, one of the activators of p21 transcription, thus suggesting that p21 expression can be accomplished in a p53-independent manner in these cells. This study suggests that p21 induction is mediated by an increase in the cellular steady-state concentration of oxygen radicals and that the greater effectiveness in p21 induction by DZQ may be related to its efficient metabolism by NAD(P)H:quinone oxidoreductase activity in HCT116 cells.

Aziridinylbenzoquinones are a group of antitumor agents that elicit cytotoxicity by generating either alkylating intermediates or reactive oxygen species. The mechanism of toxicity may not always, however, involve profound damage of cellular constituents, but may involve a cytostatic effect through interference with the cell cycle. In this context, we have examined the induction of the cell cycle inhibitor p21 (WAF1, CIP1, or sdi1), whose overexpression suppresses the growth of various tumor cells, in human tumor cells metabolizing 3,6-diaziridinyl-1,4-benzoquinone (DZQ) and its C 2 ,C 5 -substituted derivatives: 2,5-bis-(carboethoxyamino) (AZQ) and 2,5-bis-2(-hydroxyethylamino) (BZQ).
Both DZQ and AZQ were effectively activated by HCT116 human colonic carcinoma cells; the activation of the former involved largely a dicoumarol-sensitive activity, whereas that of the latter appeared to be accomplished primarily by one-electron transfer reductases. BZQ was not a substrate for the dicoumarol-sensitive enzyme in HCT116 cells. Cellular activation of the first two quinones was associated with formation of oxygen-centered radicals as detected by EPR in conjunction with the spin trap 5,5-dimethyl-1-pyrroline-N-oxide. The redox transitions of DZQ involved hydroxyl radical formation and were strongly inhibited by catalase, whereas those of AZQ showed a strong superoxide anion component sensitive to superoxide dismutase. These signals were suppressed by N-acetylcysteine with concomitant production of a thiyl radical adduct. This suggests an effective electron transfer between the thiol and free radicals formed during the activation of these quinones.
DZQ and AZQ induced significantly the expression of p21 in HCT116 cells, but a 10-fold higher concentration of AZQ was required to achieve the level of induction elicited by DZQ. BZQ had little effect on p21 expression. p21 induction at both mRNA and protein levels correlated with the inhibition of either cyclin-dependent kinase activity or cell proliferation. p21 induction elicited by the above quinones was inhibited by N-acetylcysteine, whereas the non-sulfur analog, N-acetylalanine, was without effect. Catalase and superoxide dismutase did not effect p21 induction by aziridinylbenzoquinones in HCT116 cells, thus suggesting that extracellular sources of oxygen radicals generated by plasma mem-brane reductases have no influence in the expression of this gene. Hydrogen peroxide, a product of quinone redox cycling, elicited an increase of p21 mRNA levels in HCT116 and K562 human chronic myelogenous leukemia cells. The latter lacks p53, one of the activators of p21 transcription, thus suggesting that p21 expression can be accomplished in a p53-independent manner in these cells.
This study suggests that p21 induction is mediated by an increase in the cellular steady-state concentration of oxygen radicals and that the greater effectiveness in p21 induction by DZQ may be related to its efficient metabolism by NAD(P)H:quinone oxidoreductase activity in HCT116 cells.
The cell cycle inhibitor p21 (WAF1, CIP1, or sdi1) plays a critical role in the control of the cell cycle by interacting with multiple cellular targets. It has been reported that p21 inhibited cyclin-dependent kinases (1, 2), proliferating cell nuclear antigen (3), transactivator E 2 F (4), and the stress-activated protein kinases (5). Introduction of antisense p21 RNA sequences causes G 0 -arrested cells to reenter the cell cycle and to synthesize DNA (6). Furthermore, cells lacking p21 are defective in their ability to undergo G 1 arrest following DNA damage (7,8). Diverse lines of evidence suggested that p21 may play a role in cell terminal differentiation (9) and senescence (10,11). Accordingly, ectopic expression of p21 suppresses the growth of different types of tumor cells in culture or in vivo, such as human brain, lung, prostate, osteosarcoma, and colon tumor cells (12)(13)(14)(15).
The mechanism of p21 gene activation by exogenous agents remains unclear. p21 can be induced by adriamycin, etoposide, and ␥-irradiation (16,17), all of which are DNA-damaging agents as well as generators of reactive oxygen species (such as O 2 . , H 2 O 2 , and HO ⅐ ) (18 -20).
As p21 gene expression is inducible by tumor therapeutic agents and few mutations were identified among hundreds of the tested tumor DNAs (21)(22)(23)(24), up-regulating p21 expression by anti-cancer agents could be effective in arresting the cell cycle and beneficial to cancer treatment. In this context, the production of reactive oxygen species inherent in the activation of some anticancer agents may be of relevance for p21 induction and the subsequent effects on the cell cycle. Aziridinylbenzoquinones are a group of potential antitumor compounds with both quinone and aziridine moieties: it is still not clear whether they elicit their cytotoxic effects through the formation of reactive oxygen species or of strong alkylating agents. Reports on DNA strand breakage (mediated by free radicals) and/or DNA covalent binding (mediated by electrophiles) abound in the aziridinylbenzoquinone literature.
This study was aimed at gathering information on the mo-* This work was supported by NIEHS Grant 1RO1 ES05423. 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.
§ To whom correspondence should be addressed: Dept. of Molecular Pharmacology & Toxicology, School of Pharmacy, University of Southern California, Los Angeles, CA 90033. lecular mechanisms inherent in the activation of p21 during the metabolism of several aziridinylbenzoquinones by HCT116 human colon carcinoma cells. A comprehensive approach to elucidate this mechanism(s) involved (a) characterization of the cellular activation of these quinones, (b) measurement of the ensuing steady-state concentration of oxygen radicals, and (c) establishment of a relationship between the latter and expression of p21 at mRNA and protein levels. The cellular effects of these quinones are expected to be an expression of both their functional group chemistry and mode of activation. The former led us to investigate aziridinylbenzoquinones with different substitution patterns at C 2 and C 5 : 3,6-diaziridinyl-1,4-benzoquinone (DZQ), 2,5-bis-(carboethoxyamino)-3,6-diaziridinyl-1,4-benzoquinone (AZQ), 1 and 2,5-bis-(2-hydroxyethylamino)-3,6-diaziridinyl-1,4-benzoquinone (BZQ) (see . The latter is also of interest because most cancer cells and preneoplastic nodules, among them HCT116 human colon carcinoma cells, overexpress NAD(P)H:quinone oxidoreductase (DT-diaphorase), a two-electron transfer flavoprotein (25) with salient features in quinone metabolism.

MATERIALS AND METHODS
Cell Culture and Treatments-Human colonic carcinoma HCT116 cells (ATCC CCL247) and human chronic myelogenous leukemia K562 cells (lacking p53; ATCC CCL243) were grown in McCoy's 5A medium or RPMI 1640 medium (Sigma) with 10% fetal bovine serum (Biocell Laboratories Inc., Rancho Dominguez, CA). Cells were seeded 24 h before drug treatment. All quinones were dissolved in Me 2 SO. Control cultures were treated with Me 2 SO alone. In experiments with cells supplemented with N-acetylcysteine or N-acetylalanine, the pH values of these compounds were adjusted to 7.4 before the addition. Cell viability was determined by trypan blue exclusion. Cell numbers were counted in hemacytometer by light microscopy.
RNA Isolation and Northern Blot Analysis-Extraction of total RNA (20 g/lane) from HCT116 cells and Northern blot analysis were carried out by standard procedures (26). Human p21 probe was prepared by the polymerase chain reaction as described (27) using human genomic DNA from HCT116 cells. Human GAPDH cDNA was used as control probe to determine the relative amount of RNA loaded. GAPDH cDNA was prepared from the plasmid with human GAPDH cDNA insert in Escherichia coli (ATCC 57090) by using a plasmid purification kit (Qiagen Inc., Chatsworth, CA).
Immunoblot Analysis-5 ϫ 10 6 cells were lysed in radioimmune precipitation buffer as described (28) and the amount of protein was determined using the bicinchoninic acid protein assay (Pierce). 50 g of total protein was loaded on a 12% SDS-polyacrylamide gel. After electrophoresis at 120 V for 5-6 h, the protein was transferred to a nitrocellulose membrane and incubated for 3 h at room temperature with monoclonal p21 antibody or purified p34 cdc2 antibody (PharMingen, San Diego, CA). The levels of antigens were analyzed by sensitized chemiluminescence (Amersham Corp.).
Histone H1 Kinase Assay-The cell lysate was prepared as described for immunoblot analysis. 150 g of total protein was subjected to immunoprecipitation with cdc2 or cyclin A antibody (PharMingen). The immunocomplexes were bound to protein A-agarose, washed three times with radioimmune precipitation buffer and twice with kinase buffer (50 mM Tris, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol). 25 l of kinase buffer containing 4 g of histone H1, 30 M ATP, and 5 Ci of [␥-32 P]ATP (3000 Ci mmol Ϫ1 ) was added and incubated at room temperature for 15 min. The reaction products were analyzed on a 12% acrylamide gel, quantified by autoradiography (Instant Imager Electronic Autoradiography; Packard Instrument Co.), and exposed to a Kodak X-AR film.
EPR Spectroscopy-HCT116 cells in exponential growth were harvested by trypsinizing, washed twice in phosphate-buffered saline (PBS, pH 7.4), sonicated for 1 min, and then centrifuged at 10,000 ϫ g for 5 min at 4°C to obtain the homogenates. Protein was determined by the Lowry method (29). Cell homogenates (2.17 Ϯ 0.03 mg of protein/ml) were mixed with 80 mM DMPO, 2% DMSO, 1 mM NADH, and 100 M quinone in the absence or presence of 20 M dicoumarol. EPR spectra were recorded on a Bruker ECS106 spectrometer with the following settings: microwave frequency, 9.80 GHz; microwave power, 20 mW; sweep width, 100 G; receiver gain, 1 ϫ 10 6 ; modulation amplitude, 0.86 G; time constant, 1.3 s; sweep rate, 20 G/min.

Cellular Activation of Aziridinylbenzoquinones-NAD(P)H:
quinone oxidoreductase, a two-electron flavoprotein overexpressed in most cancer cell lines and preneoplastic tumors is expected to play an important role in the bioactivation of aziri-STRUCTURE 1  (25,31,32). The specific activity of this oxidoreductase in HCT116 cells is ϳ390 nmol/min/mg of protein (measured as the dicoumarol-sensitive reduction of dichlorophenolindophenol) (33).
The dicoumarol-sensitive reduction of DZQ by HCT116 cell homogenates was 374 nmol/min/mg of protein, a value that represented 72% of the overall reducing activity of HCT116 cells. That of AZQ was only 8.6 nmol/min/mg of protein, indicating that quinone reductases other than NAD(P)H:quinone oxidoreductase were largely (ϳ95%) involved in the metabolism of AZQ (Table I). These activities were also measured in the presence of superoxide dismutase in order to distinguish the fraction of cytochrome c reduced by hydroquinone, semiquinone, or O 2 . . Superoxide dismutase had little effect on cytochrome c reduction, suggesting that most of the hemoprotein was reduced primarily by hydroquinones and/or semiquinones. In summary, DZQ was reduced by a dicoumarol-sensitive activity at a rate ϳ43-fold higher than AZQ, whereas BZQ was not a substrate for NAD(P)H:quinone oxidoreductase. This is in agreement with previous observations obtained with HT29 cells (31).
Free Radical Formation during Metabolism of Aziridinylbenzoquinones by HCT116 Cells-EPR in conjunction with the spin trap DMPO revealed a strong signal when HCT116 cell homogenates were incubated with DZQ ( Fig. 1). This signal was ascribed to the DMPO-CH 3 adduct (a N ϭ 16.4 and a H ϭ 23.52 G) originating from HO ⅐ attack on the quinone solvent (Me 2 SO) and subsequent trapping of ⅐ CH 3 by DMPO (Reaction 1).  1). The contribution of the former adduct to the overall EPR signal was abolished by superoxide dismutase. The generation of HO ⅐ by AZQ was ϳ3-fold lower than that by DZQ, in agreement with the enzymic reduction of the quinones (Table I).
The differences in the EPR signals obtained during the metabolism of DZQ and AZQ by HCT116 cells are an expression of their functional group chemistry and the type of activation. AZQ is predominantly reduced by quinone reductases other than NAD(P)H:quinone oxidoreductase, whereas the latter enzyme appears largely responsible for the activation of DZQ (Table I). The occurrence of O 2 . during AZQ metabolism appears consistent with a univalent activation of the quinone. Conversely, the absence of this radical adduct during DZQ metabolism may be rationalized as a predominant role of O 2 . as a propagating species (oxidizing the hydroquinone to semiquinone). BZQ did not produce any detectable oxygen-centered radicals in the cell homogenates within the limits of our assay procedures, but a strong semiquinone signal was detected in the absence of the spin trap (Fig. 2a). This quinone itself, in PBS buffer and in the absence of cell homogenates, produced a semiquinone signal (Fig. 2b), whose intensity was enhanced upon addition of NADH (Fig. 2c). It may be surmised that BZQ is mainly reduced nonenzymically to a "stabilized" semiquinone.
Effect of N-Acetylcysteine on Free Radical Production by HCT116 Cells-N-Acetylcysteine is a powerful reductant, which has been used as an intracellular antioxidant in several experimental models (18,35). The effect of N-acetylcysteine on the EPR signal observed during the metabolism of AZQ by HCT116 cells is shown in Fig. 3: a new signal (hyperfine splitting constants: a N ϭ 15.0 and a H ϭ 16.8 G) (36) ascribed to the N-acetylcysteinyl radical was obtained. This is consistent with the one-electron oxidation of the thiol by the free radicals formed during the redox transitions of AZQ (or DZQ, not shown) and subsequent trapping of the thiyl radical by DMPO (Reaction 2). with human p21 mRNA (12). Treatment of HCT116 cells with DZQ elicited a concentration-and time-dependent increase of p21 mRNA levels (Fig. 4). p21 protein level was significantly elevated while the level of p34 cdc2 protein, one of the cyclin-dependent kinases, remained unchanged following the treatment with 0.4 or 1.0 M DZQ for 15 h (Fig. 5A). The cdc2-associated H1 kinase activity was strongly suppressed following the above treatments (Fig. 5B). The histone H1 kinase activities of the cyclin-dependent kinases associated wtih cyclin A (immunoprecipitated by cyclin A antibody) were also substantially inhibited (Fig. 5C). These results are in agreement with the role of p21 in inhibiting cyclin-dependent kinase activities. Fig. 6 shows that the increase of p21 mRNA levels elicited by the three quinones differed markedly: DZQ produced the highest response (a 6-fold increase) over a narrow concentration range, possibly determined by the toxicity of the quinone at high doses (see below). AZQ significantly elevated (ϳ3-fold) p21 mRNA levels, but higher concentrations than those used with DZQ were required. Conversely, BZQ produced no obvious dosedependent response on p21 mRNA.
p21 Induction by Reactive Oxygen Species Formed during Quinone Metabolism-The effect of N-acetylcysteine on free radical production by the above quinones suggested an effective transfer of the radical character from an oxygen-centered radical to a less reactive sulfur-center radical (Reaction 2; Fig. 3). The exposure of HCT116 cells to N-acetylcysteine caused a small increase at constitutive levels of p21 mRNA (Fig. 7, A and  B). The thiol completely inhibited the DZQ-mediated p21 induction, partially inhibited that mediated by AZQ, and had no effect on cells treated with BZQ (Fig. 7B). The effect elicited by N-acetylcysteine appears to be dependent on its free thiol group, because N-acetylalanine, a non-thiol analog, failed to elicit any significant inhibition of p21 induction (Fig. 8).
To further elucidate if oxyradicals mediate p21 induction by quinones, the HCT116 cells were treated with H 2 O 2 , a membrane-permeable product of quinone redox cycling (37): ϳ2-fold increase in the amount of p21 mRNA was observed (Fig. 9A). H 2 O 2 also led to an increase at p21 mRNA level in human chronic myelogenous leukemia K562 cells lacking p53 (one of the activators of p21 transcription) in which p21 mRNA was elevated following the treatment of DZQ (Fig. 9B).
Effect of Superoxide Dismutase and Catalase on p21 Induction-These enzymes, when present in the medium surrounding intact cells, did not have any effect on p21 induction. This suggests that extracellular sources of oxygen radicals generated by plasma membrane reductases have no influence on the expression of this gene.

FIG. 3. Effect of N-acetylcysteine on oxyradical formation during the metabolism of AZQ by HCT116 cells.
Assay conditions were the same as described in the legend of Fig. 1a (right panel). a, control; b, plus 30 mM N-acetylcysteine. c, simulated spectrum of b corresponding to the N-acetylcysteinyl radical. p21 Induction and Cell Proliferation-Cell proliferation was also inhibited by DZQ and AZQ at the same concentrations required to induce p21. Similar to what was observed with p21 induction (Fig. 6), BZQ did not elicit any obvious effect on cell proliferation (Fig. 10A). To determine whether inhibition of cell proliferation was a transient phenomenon, cells were counted at 6, 24, 48, and 72 h following their exposure to 0.4 M DZQ (Fig. 10B). Cells were still proliferating at the early stage of the exposure. After 24 h, cell proliferation was completely inhibited by this treatment, even when the DZQ in the medium was removed after 6-h treatment. This inhibition lasted at least until 72 h after the treatment. A significant percentage of dead cells were identified only in the continuous presence of DZQ for more than 48 h after the treatment. Thus, at a concentration of 0.4 M, the effect of DZQ on cell number appeared to be due to a decreased proliferation rather than outright toxicity. DISCUSSION The role of oxygen free radicals in antitumor quinone reactivity and their implications for cancer cell and systemic cytotoxicity has been randomly established (25,32). Cytotoxicity elicited by aziridinylbenzoquinones is attributed to the quinone moiety and/or aziridinyl substituents; the former participates in redox cycling processes and oxygen radical production, whereas the latter is essential for the alkylating properties of these compounds. Purified NAD(P)H:quinone oxidoreductase reduced aziridinylbenzoquinones with different efficiency, and the redox transitions of the hydroquinone product were associated with production of oxygen-centered radicals (38). Metab- olism of AZQ by this enzyme generates radical species responsible for plasmid DNA strand breakage, cytotoxicity toward Chinese hamster ovary cells (39), and potentiation of toxicity in L5178Y murine lymphoblastic cells (40), as well as the formation of HO ⅐ in the MCF-7 S9 fraction (41). Conversely, some aziridinylbenzoquinones may be activated to electrophilic species that cause DNA cross-links (42,43). The cytotoxicity elicited by BZQ is supposed to involve the latter mechanism (31,43).
In this study, we investigated the bioactivation of several aziridinylbenzoquinones and their ability to induce p21. We have shown that the metabolism of DZQ and AZQ by HCT116 cells was associated with formation of oxyradicals and an elevation of p21 mRNA levels, which correlated with inhibition of cell proliferation. BZQ, which was not a substrate for NAD-(P)H:quinone oxidoreductase and which is expected to elicit cytotoxicity by causing DNA cross-linking, had little, if any, effect on cell proliferation and only slightly induced p21 in HCT116 cells.
Understanding the biochemical pathways leading to expression of p21 during quinone metabolism by cancer cells requires careful consideration of their bioactivation and functional group chemistry, as well as variations among different cell lines.
HCT116 cells are endowed with a high activity of NAD(P)H: quinone oxidoreductase (33). This enzyme reduces DZQ and AZQ (31,38) (Table I), but not BZQ. DZQ was largely reduced by a dicoumarol-sensitive activity associated with HO ⅐ production. AZQ, on the other hand, was primarily reduced by oneelectron transfer flavins (dicoumarol-insensitive activities) leading to a high production of O 2 . (Table I and Fig. 1). . from the equilibrium. Accordingly, semiquinones of DZQ or AZQ were not observed with HCT116 cell homogenates, whereas they were evident in less complex systems involving activation of the quinone by the purified enzymes in the absence of superoxide dismutase (38). The effect elicited by N-acetylcysteine on the EPR signals ( Fig. 3) may be rationalized as the prevalence of a reductive pathway involving oxygen-centered radicals or the semiquinone. In both instances, a low steady-state level of oxyradicals would ensue. The reactivity of N-acetylcysteine toward different radicals is partly dependent on the reduction potential of the radical: the second order rate constant for the reaction of N-acetylcysteine with HO ⅐ (1.36 ϫ 10 10 M Ϫ1 s Ϫ1 ) (45) is far higher than that with O 2 . (10 3 M Ϫ1 s Ϫ1 (35)). This may bear some relevance for the p21 induction elicited by the metabolism of AZQ and DZQ: the type of oxyradicals produced may also make some difference, since N-acetylcysteine reacts more rapidly with HO ⅐ than with O 2 . .
Although N-acetylcysteine may serve as a precursor for GSH biosynthesis and/or a reductant for cellular GSSG (and less likely for mixed disulfides), these functions are unlikely to account for the effects observed on free radical production (Fig.  3) and p21 induction (Fig. 7). An increase in cellular GSH (through cysteine) will amount to a small fraction which cannot explain the inhibitory effect observed. N-Acetyl-D-cysteine, an analog of N-acetylcysteine, had no effect on cellular GSH levels, but it exerted a protective effect on murine T-cell hybridoma DO-11.10 cells (46). The differential effects of N-acetylcysteine on the p21 inductions by DZQ and AZQ suggest that multiple mechanisms of p21 induction were possibly involved. For instance, AZQ is known to cause the inactivation of glutathione reductase and lead to the failure of GSH regeneration (47). Depletion of GSH has been suggested to induce p21 (48).
p53 is a tumor suppressor, which activates p21 expression and controls a G 1 checkpoint in cell cycle. It has been shown that DNA strand breaks (caused by introducing endonucleases) may induce an elevation in p53 protein (49,50). However, p21 can also be activated through p53-independent pathways (9,27). Both DZQ and H 2 O 2 were also able to induce p21 expression in p53-null K562 cells (Fig. 9). Hence, DNA oxidative damage seems not to be a requisite condition for p21 induction in p53-independent pathways. Oxidative stress, which can be caused by free radicals, leads to the activation of signal transducer and activator of transcription (STAT) proteins (51), which have been shown to activate p21 expression in a p53independent way (52). Therefore, it is possible that oxyradicals may elicit p21 induction through STAT-dependent pathway by causing oxidative stress. However, our data with p53-null K562 cells do not provide unequivocal evidence for a p53-independent p21 induction, a process that may occur in cells expressing p53.
Taken together, the p21 gene is a potential target for anti-  p21 Induction by Reactive Oxygen Species cancer drug development. In this study we have established a correlation between an increased steady-state concentration of oxyradicals and induction of p21 in HCT116 cells. These results suggest that the greater effectiveness in p21 induction by DZQ might be related to its efficient metabolism by NAD(P)H:quinone reductase. Although H 2 O 2 , a long-lived species, appears to be important as a cellular signal leading to p21 expression, it could be hypothesized that far more reactive and short-lived species, such as HO ⅐ , may be a more immediate trigger for this response. This concept is supported by, on the one hand, the strong induction of p21 during the metabolism of DZQ, whose redox transitions produced mainly HO ⅐ and, on the other hand, the inhibitory effect of N-acetylcysteine that reacts at diffusioncontrolled rates with HO ⅐ . The occurrence of transition metals on the vicinity of molecules engaged in the regulation of the cell cycle machinery may be of utmost importance for the generation of this species in a site-specific manner.