Critical role of oxidative stress in -induced carcinogenesis

Hari K. Bhat*†, Gloria Calaf‡, Tom K. Hei*‡, Theresa Loya§, and Jaydutt V. Vadgama¶

*Department of Environmental Health Sciences, Mailman School of Public Health, 60 Haven Avenue-B1, Columbia University, New York, NY 10032; ‡Center for Radiological Research, Columbia University, New York, NY 10032; and Departments of §Pathology and ¶Medicine, Charles Drew University, Los Angeles, CA 90059

Communicated by Donald C. Malins, Pacific Northwest Research Institute, Seattle, WA, December 27, 2002 (received for review August 22, 2002) Mechanisms of estrogen-induced tumorigenesis in the target quinones generates oxidative stress and potentially harmful free organ are not well understood. It has been suggested that oxida- radicals that are postulated to be required for the carcinogenic tive stress resulting from metabolic activation of carcinogenic process, and analogous to the metabolic activation of hydrocar- plays a critical role in estrogen-induced carcinogenesis. bons and other estrogen carcinogens (9, 19–22). We We tested this hypothesis by using an estrogen-induced hamster have investigated the role of oxidative stress in estrogen carci- renal tumor model, a well established animal model of hormonal nogenesis by using a well established hamster renal tumor model carcinogenesis. Hamsters were implanted with 17␤- (␤E2), that shares several characteristics with human breast and uterine 17␣-estradiol (␣E2), 17␣- (␣EE), menadione, a com- cancers, pointing to a common mechanistic origin (6, 9, 23). bination of ␣E2 and ␣EE, or a combination of ␣EE and menadione Different estrogens used in the present study differ in their for 7 months. The group treated with ␤E2 developed target organ estrogenic, carcinogenic, and metabolic activation potentials specific kidney tumors. The kidneys of hamsters treated with ␣E2, (14–17). ␤E2 is a good catechol progenitor and a potent ␣EE, or menadione alone did not show any gross evidence of estrogen; its use results in 80–100% tumor incidence in the tumor. Kidneys of hamsters treated with a combination of ␣E2 and hamster kidney (6, 10, 14, 15). 17␣-estradiol (␣E2) is a nontu- ␣EE showed early signs of proliferation in the interstitial cells. morigenic, weak estrogen with a catechol-forming potential Kidneys of hamsters treated with a combination of menadione and similar to that of ␤E2 (24, 25). 17␣-Ethinylestradiol (␣EE) is a ␣EE showed foci of tumor with congested tubules and atrophic potent estrogen, but a weak catechol progenitor that is either glomeruli. ␤E2-treated tumor-bearing kidneys showed >2-fold nontumorigenic or very weakly tumorigenic in the hamster Ͼ increase in 8-iso-prostaglandin F2␣ (8-iso-PGF2␣) levels compared model with 9 months of continuous exposure required for less with untreated controls. Kidneys of hamsters treated with a than 10% tumor incidence (15, 17). Menadione (2-methyl-1,4- combination of menadione and ␣EE showed increased 8-iso-PGF2␣ napthaquinone) is used in the present study as a model com- levels compared with untreated controls, whereas no increase in pound with known oxidant stress potential to study the influence 8-iso-PGF2␣ was detected in kidneys of ␣EE-treated group. A of oxidative stress on estrogen-induced carcinogenesis (26–28). chemical known to produce oxidative stress or a potent estrogen We used a combination of an oxidant chemical menadione and with poor ability to produce oxidative stress, were nontumorigenic a noncarcinogenic estrogen ␣EE to show the induction of renal in hamsters, when given as single agents, but induced renal tumors, tumors in a rodent model of hormonal carcinogenesis. We also when given together. Thus, these data provide evidence that oxidant demonstrated increased levels of 8-iso-prostaglandin F2␣ (8-iso- ␤ stress plays a crucial role in estrogen-induced carcinogenesis. PGF2␣), a known marker of oxidant stress (29, 30), in E2- induced tumor-bearing kidneys as well as in menadione plus tumor ͉ hormonal carcinogenesis ͉ menadione ͉ prostaglandin ͉ metabolic ␣EE-treated kidneys of hamsters. Our studies suggest that activation oxidative stress plays a critical role in estrogen-induced carcinogenesis. ex hormones are implicated in the development of a variety Materials and Methods of human cancers (1–4). Estrogen administration to post- S Treatment of Animals. Male Syrian hamsters (4–6 weeks old; menopausal women is associated with an increased risk of Harlan Sprague–Dawley, Madison, WI) were housed in our endometrial and (1–4). An increasing evidence of animal facility with Purina rodent chow and water available ad elevated breast cancer risk with increases in total lifetime libitum throughout the experiment. Hamsters were implanted exposure of women to estrogens has been presented (1–3). s.c. with 25-mg pellets of ␤E2, ␣E2, ␣EE, menadione, a com- Recently, the clinical trial of estrogen plus progestin treatment bination of ␣E2 ϩ ␣EE, or a combination of ␣EE ϩ menadione. therapy was stopped because of an increased risk of breast cancer These hamsters received a second estrogen or menadione pellet (5). Knowledge of how estrogens induce proliferation and tu- 3 months after initial treatment. Before implantation of the drug morigenesis in their target organ is not well defined (6–9). The pellets, hamsters were anesthetized with a combination of ket- mechanism of tumor induction by estrogens is being investigated amine and xylazine (ketamine, 100 mg͞kg body weight, i.p., and in rodent models of hormonal carcinogenesis. The natural ͞ ␤ ␤ xylazine, 10 mg kg body weight, i.p.). Estrogen and menadione female 17 -estradiol ( E2) and the synthetic pellets were prepared by using a hand press and implanted into estrogen induce tumors in rats, mice, and the hamsters s.c. as described (10, 12). There were 10 hamsters hamsters (10–13). It must be noted that in rodent models, in each group. A control group of 10 animals was sham operated different estrogens tested have not shown similar carcinogenic and left untreated. Hamsters were killed after 7 months and potential despite their similar hormonal potencies (6, 14, 15). inspected macroscopically for tumor nodules on the surface of However, carcinogenic and noncarcinogenic estrogens differ in each kidney as reported (31). Portions of liver and kidney tissue their metabolic activation profiles (14–17). Therefore, it is were placed on dry ice and stored at Ϫ80°C for further studies. postulated that estrogen metabolism may play a key role in hormonal carcinogenesis. Estrogens can be metabolically activated into catechol estro- Abbreviations: ␤E2, 17␤-estradiol; ␣E2, 17␣-estradiol; ␣EE, 17␣-ethinylestradiol; ER, estro-

gens by cytochrome P450 enzymes (18, 19). Metabolic redox gen receptor; 8-iso-PGF2␣, 8-iso-prostaglandin F2␣. BIOCHEMISTRY cycling between catechol estrogens and their corresponding †To whom correspondence should be addressed. E-mail: [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0437929100 PNAS ͉ April 1, 2003 ͉ vol. 100 ͉ no. 7 ͉ 3913–3918 Downloaded by guest on September 26, 2021 Other portions of liver and kidney were placed in 10% buffered wt͞vol) were prepared by using a PRO 200 homogenizer with a formalin for histopathological evaluation and immunocytochem- 5mmϫ 75 mm generator (PRO Scientific, Oxford, CT) in 2-ml ical studies. Menadione and all estrogens used in the present microfuge tubes. Homogenization was carried out by moving the study were purchased from Sigma. motor speed dial of the homogenizer from 0 to 5 (0–30,000 rpm) back and forth five times with a total homogenization time of Ϸ ␮ Tissue Preparation for Histopathology and Immunocytochemistry. 5 s. 8-iso-PGF2␣ esters in 100 l of the total kidney homoge- The formalin-fixed tissue was embedded in paraffin, and sections nate were hydrolyzed by incubation with 25 ␮lof10N NaOH at of 4- to 5-␮m thickness were cut. Paraffin-embedded sections of 45°C for 2 h. The reaction mixture was cooled on ice for 5 min the kidneys and livers were stained with hematoxylin and eosin and neutralized with 25 ␮l of 12 N HCl, and centrifuged in a for histopathological evaluation. Gross examination and histo- microcentrifuge for 5 min. The clear neutralized supernatant was logical sections were interpreted by two independent patholo- transferred into a new microfuge tube, and 50 ␮l of the neu- gists in a blinded fashion, without knowledge as to how the tralized sample was used for 8-iso-PGF2␣ assay. The samples animals were stratified. Paraffin-embedded sections were also were incubated with the 8-iso-PGF2␣ antibody for 18 h at 4°Cin used for cell-specific expression of (ER)-␣ and a 96-well format. After incubation, the contents of the wells were -␤ proteins. Deparaffinized sections were incubated with the emptied and washed with wash buffer; wash buffer was removed corresponding primary antibodies: ER-␣ (Santa Cruz Biotech- from the wells, and the color was developed by incubation with nology, SC-542) and ER-␤ (Santa Cruz Biotechnology, SC-6821) 200 ␮lofp-nitrophenyl phosphate for 45 min at room temper- at dilutions suggested by the suppliers. Incubation with the ature. The reaction was stopped by the addition of 50 ␮l of stop primary antibodies was performed overnight at 4°C. Nonspecific solution, and the plate was read at 405 nm. A standard curve was sites were blocked by covering sections with solutions of 1% BSA generated by measuring the optical density of 160–100,000 ͞ (Sigma). After washing in PBS, the sections were incubated with pg ml of 8-iso-PGF2␣ standards that were processed simulta- Ј peroxidase-conjugated, affinity-purified F(ab )2 fragment of neously with unknown samples on the same plate. Protein donkey anti-rabbit IgG (Jackson ImmunoResearch) for 60 min concentrations from the neutralized homogenates were deter- at room temperature as described (12). Slides were rinsed three mined by using a Pierce protein assay kit. 8-iso-PGF2␣ and times in PBS, and staining was developed by incubating with protein were analyzed from 10 kidney homogenates from each Ј ͞ ͞ 3,3 -diaminobenzidine tetrachloride H2O2 (Sigma) for 2–5 min. group, and data were expressed as mean 8-iso-PGF2␣ pg mg 3,3Ј-Diaminobenzidine solution was prepared according to the protein Ϯ SEM. manufacturer’s recommendations. After staining, the sections were rinsed in distilled water, dehydrated in ethanol͞water baths Statistical Analysis. Statistical analysis was performed by using with decreasing water content, and finally rinsed in xylene before SPSS statistical software package (SPSS, Chicago). Unpaired t being mounted with a permanent mounting medium. test was used to assess significance between the two different treatment groups. Tumor incidence was analyzed by Fisher’s Semiquantitative Analysis of Critical Histopathological Differences. exact test. The severity of the total kidney damage was evaluated by a scoring system that gave a semiquantitative measurement of the Results damage as described (32). The four types of damage for which Tumor Incidence. On macroscopic examination, no tumor nodules semiquantitative analysis was performed were glomerular atro- were detected in untreated hamsters and in groups of hamsters phy, glomerular congestion, tubular congestion, and nodular that were treated with ␣E2, ␣EE, menadione, or a combination proliferation. One hundred nephrons (glomeruli and adjacent of ␣E2 and ␣EE for 7 months (Table 1). In contrast, hamsters tubules) were examined histologically, and each nephron was treated with either ␤E2 or a combination of ␣EE and menadione scored from 0 to 4. For glomerular atrophy, glomeruli that for 7 months developed renal tumors. The tumor incidence was contained Ͼ20 nuclei were graded as ‘‘0,’’ and those glomeruli 90% for the group treated with ␤E2 and 30% for the group which contained Ͻ10 nuclei were scored as ‘‘4.’’ For glomerular treated with a combination of ␣EE and menadione (Table 1). congestion, glomeruli that did not contain eosionophilic and The mean number of tumor nodules was higher in the group pink deposits were graded as ‘‘0,’’ and those that had Ϸ80% of treated with ␤E2 compared with the group treated with a glomeruli with such deposits were graded as ‘‘4.’’ For tubular combination of ␣EE and menadione (Table 1). congestion, convoluted tubules that did not contain eosinophilic and pink deposits were graded as ‘‘0,’’ and those convoluted Histopathological Analysis. Sections from untreated hamster kid- tubules that contained Ϸ 80% deposits were graded as ‘‘4.’’ neys demonstrated normal convoluted tubules and glomeruli Intermediate stages were graded as 1, 2, and 3. For proliferation, within the cortex (Fig. 1). The corticomedullary junction ap- sections of kidneys were graded as ‘‘4’’ if they contained tumor peared normal, as did the renal hilum, which was lined by normal nodules, and as ‘‘0’’ if they did not have any tumor nodules. The transitional urothelium. sum of individual scores of 100 counts on each kidney section was Kidneys of hamsters treated with ␤E2 for 7 months were used as an estimate of the severity of the kidney damage. A total abnormal and contained numerous tumor nodules (Fig. 2a). The score of 1,600 was given by the sum of the four descriptive nodules were composed of a combination of round hyperchro- damages described above (each kind of damage getting a max- matic cells and round to spindled, irregular, hyperchromatic cells imum score of 400). For each type of damage, six to eight (Fig. 2b). Scattered tubules within the tumor nodules were hematoxylin and eosin-stained kidney sections were examined congested (Fig. 2a). Most of the tumor nodules were in the from each treatment group and scored, and data were expressed cortical area of the kidney and, in some of the nodules, en- as a mean Ϯ SEM of the individual scores. trapped and atrophic glomeruli were seen (Fig. 2c). Sections from the kidney away from the tumor were also markedly Analysis of 8-iso-PGF2␣ Levels. Total 8-iso-PGF2␣ levels in kidney abnormal and demonstrated large dilated congested convoluted tissue of hamsters treated with different chemicals were quan- tubules, which were lined by somewhat flattened, cuboidal tified by using a direct 8-iso-PGF2␣ enzyme immunoassay kit epithelial cells (Fig. 2d). Many of the congested tubules were from Assay Designs (Ann Arbor, MI, catalog no. 900-091) filled with pink eosinophilic deposits (Fig. 2d). A significant according to the supplier’s recommendations. Kidney tissue increase in the number of congested convoluted tubules, con- (50–100 mg) was homogenized in cold PBS (pH 7.4) containing gested glomeruli, atrophic glomeruli, and nodular proliferation 0.005% butylated hydroxytoluene. Tissue homogenates (10% was detected in the ␤E2-treated tumor-bearing kidneys com-

3914 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0437929100 Bhat et al. Downloaded by guest on September 26, 2021 Table 1. Influence of different estrogens and menadione on renal carcinogenesis in male Syrian hamsters Mean no. of No. of animals with tumor nodules Treatment tumors͞no. of animals per hamster

␤E2 9͞10* 7.0 Ϯ 3.16† ␣E2 0͞10 0 ␣EE 0͞10 0 ␣E2 ϩ ␣EE 0͞10 0 Menadione 0͞10 0 Menadione ϩ ␣EE 3͞10‡ 1.6 Ϯ 2.67§ Untreated 0͞10 0

Male Syrian hamsters were treated with implants of ␤E2, ␣E2, ␣EE, mena- dione, ␣E2 ϩ ␣EE, or menadione ϩ ␣EE. Details of the experiment are provided in Materials and Methods. The tumor incidence was evaluated by macroscopic examination of the tumor nodules on both kidneys. *P Ͻ 0.05 compared with untreated controls by Fisher’s exact test. Fig. 2. Paraffin section of a tumor-bearing kidney stained with hematoxylin †P Ͻ 0.05 when compared with untreated controls by using ␹2 test. and eosin. The tumors were induced by treatment of male Syrian hamsters ‡P Ͻ 0.05 compared with ␤E2 treated group by Fisher’s exact test; P ϭ 0.10 with ␤E2 for 7 months. (a) An abnormal kidney architecture with tumor when compared with untreated controls by Fisher’s exact test. nodules (T) and congestion of scattered convoluted tubules (arrow) within the §P Ͻ 0.05 when compared with ␤E2-treated group by using unpaired t test. tumor nodules can be observed. (b and c) The tumor nodules are composed of a combination of round to spindled hyperchromatic cells (b, arrows), and in some of the tumor nodules (T), entrapped and atrophic glomeruli (G) are pared with untreated controls by using the semiquantitative present (c). Many of the congested tubules (Tb) are filled with pink eosino- estimates of the total kidney damage (Fig. 3). philic deposits (arrow in c, arrowhead in d) and are lined by somewhat Microscopic evidence of tumor nodules was not seen in any of flattened epithelial cells (d, arrow). Magnification: a ϭϫ4; b ϭϫ40; c ϭϫ10; the kidneys of hamsters treated with ␣E2 for 7 months. Kidney d ϭϫ40. sections showed normal glomeruli and convoluted tubules within the renal cortex. No abnormal proliferations were seen. The medulla and hilum were microscopically normal (data not web site). No tumor nodules were identified, and there was no shown). evidence of atrophic glomeruli (Fig. 3). Atypical congested convoluted tubules were seen in kidneys of On microscopic examination, vascular congestion of glomeruli and of convoluted tubules was observed in the kidneys of hamsters treated with ␣EE for 7 months (Fig. 3). In some areas, hamsters treated with a combination of ␣E2 plus ␣EE for 7 the convoluted tubules were lined by crowded hyperchromatic months (Figs. 3 and 4a). Kidneys contained foci where there cuboidal cells, which had decreased cytoplasm, and the kidney were atypical collections of both interstitial cells and convoluted sections also showed some congested glomeruli (see Fig. 7, which tubules (Fig. 4b). In these areas, some of the tubules contained is published as supporting information on the PNAS web site, cuboidal cells, which had scant cytoplasm and irregular, slightly www.pnas.org). No nodular proliferation was detected in the ␣ pleomorphic, hyperchromatic nuclei (Fig. 4c). Some slightly kidney sections of hamsters treated with EE for 7 months crowded hyperchromatic renal collecting tubules were observed (Fig. 3). (Fig. 4d). Kidney sections of hamsters treated with menadione for 7 Kidney sections of hamsters treated with a combination of months demonstrated renal convoluted tubules with evidence of ␣EE plus menadione demonstrated foci of tumor that contained vascular as well as glomerular congestion (Fig. 3) and the congested tubules (Fig. 5a). The tumor nodules were character- presence of an altered epithelial layer of convoluted tubules (see ized by hyperchromatic, spindled to rounded cells with marked Fig. 8, which is published as supporting information on the PNAS nuclear crowding and overlapping (Fig. 5b). This group showed significant congestion of both renal convoluted tubules and glomeruli when compared with ␣E2 and ␣EE treated groups (Fig. 3). Glomerular atrophy and nodular proliferation were detected in the kidneys of hamsters treated either with a combination of ␣EE plus menadione, or with ␤E2 for 7 months (Figs. 3 and 5b). Microscopic examination of tissue sections from livers of hamsters treated with ␤E2, ␣E2, ␣EE, menadione, ␣E2 ϩ ␣EE, or menadione ϩ ␣EE did not show any evidence of tumor or dysplasia (data not shown).

Immunocytochemical Analysis. Renal tumors induced by a carci- nogenic estrogen or by a combination of a noncarcinogenic estrogen and a chemical known to produce oxidative stress were characterized immunocytochemically. Cell-specific expression of ER-␣ and -␤ was examined in the kidneys of hamsters treated with different estrogens, either alone or in combination with menadione. ER-␤ protein expression was not detected either in Fig. 1. Paraffin section of an untreated male Syrian hamster kidney stained control or in tumor-bearing kidneys (data not shown). A very

with hematoxylin and eosin. Normal kidney architecture with normal convoluted poor immunoreactivity for ER-␣ protein was observed in the BIOCHEMISTRY tubules (CT) and glomerulus (G) is observed. Magnification ϭϫ40. tubules of untreated control kidneys (see Fig. 9, which is

Bhat et al. PNAS ͉ April 1, 2003 ͉ vol. 100 ͉ no. 7 ͉ 3915 Downloaded by guest on September 26, 2021 Fig. 3. Semiquantitative estimate of the total kidney damage in hamsters treated with 25-mg pellets of ␤E2, ␣E2, ␣EE, menadione, a combination of ␣E2 ϩ ␣EE, or a combination of ␣EE ϩ menadione for 7 months as s.c. Fig. 4. Paraffin section of a male hamster kidney stained with hematoxylin implants. The control group was sham operated and received no estrogen and eosin. Male Syrian hamsters were treated with a combination of ␣EE and implants. Hematoxylin and eosin-stained sections of paraffin-embedded ␣E2 for 7 months. (a) Vascular congestion of glomeruli (G) and convoluted tissues were analyzed microscopically for the extent of renal damage as tubules (CT) is observed. (b) Kidneys contain foci where there are atypical described in Materials and Methods. Increased congestion of the convo- collections of both interstitial cells (arrow) and CT. (c) Some of the tubules luted tubules (c-ct) and glomeruli (c-g) was observed in kidneys of hamsters contain cuboidal cells that have scant cytoplasm and irregular, slightly pleo- treated with ␤E2, menadione, a combination of ␣E2 ϩ ␣EE, or a combina- morphic hyperchromatic nuclei (arrowheads). (d) Some slightly crowded renal tion of ␣EE ϩ menadione. Glomerular atrophy (a-g) and proliferation (p) is collecting tubules (arrows) are also observed in kidneys of hamsters treated observed in kidney tissue of the groups treated with ␤E2 or a combination with a combination of ␣EE and ␣E2 for 7 months. Magnification: of ␣EE ϩ menadione for 7 months. 1, untreated control; 2, ␣E2; 3, ␣EE; 4, a–c ϭϫ40; d ϭϫ10. ␣E2 ϩ ␣EE; 5, menadione; 6, menadione ϩ ␣EE; 7, ␤E2. a, P Ͻ 0.05 for both c-g and c-ct when compared with untreated, ␣E2-treated, or ␣EE-treated hamster kidney by an unpaired t test. b and c,PϽ 0.05 for a-g and p, respectively, when compared with any other group by an unpaired t test. increase, or other assays of estrogenic activity (6, 14, 15, 24, 33, For each type of damage, six to eight hematoxylin and eosin-stained kidney 34). A number of studies of estrogen-induced carcinogenesis sections were examined from each treatment group and scored, and data indicate that tumor induction depends on estrogen metabolism, are expressed as a mean of the individual scores. The SEM for all of the four and oxidative stress as a result of the redox cycling of estrogen types of kidney damage (c-ct, c-g, a-g, and p) in all of the seven different ␤ treatment groups varied from 0% to 22% except for c-ct for the group that metabolites (6, 16, 20, 21, 35). It has been shown that E2 in was treated with ␣E2, where the % error was 64. This group showed very target organs of estrogen-induced cancer is metabolized to 2- little kidney damage. and 4-hydroxyestradiol, commonly known as catechol estrogens (16, 36, 37). Elevated estradiol-4-hydroxylase activity has also been shown in organs that are prone to estradiol-induced published as supporting information on the PNAS web site). hyperplasia or cancer in rodents, in humans, and in human breast ␣ ␤ Intense nuclear staining for ER- protein was found in E2- cancer cell lines (16, 18, 36–40). Catechol estrogens are ortho- ␣ induced renal tumors (see Fig. 9). ER- protein expression was hydroquinones that are capable of undergoing metabolic redox not detected in the glomeruli or kidney tubules of menadione- treated animals (data not shown). Moderate to strong immuno- reactivity for ER-␣ protein was observed in the renal tumor area of ␣EE plus menadione-treated hamsters and some of the tubules within the tumor mass also expressed high levels of ER-␣ protein (see Fig. 9).

8-iso-PGF2␣ Analysis. Total 8-iso-PGF2␣ levels in kidney tissue of hamsters treated with different chemicals were quantified. A Ͼ2- fold increase in 8-iso-PGF2␣ was detected in tumor-bearing kidney homogenates of hamsters treated with ␤E2 for 7 months compared with untreated controls (Fig. 6). Kidney homogenates of hamsters treated with ␣EE for 7 months did not show any increase in 8-iso-PGF2␣ levels compared with untreated controls (Fig. 6). The fold increases in 8-iso-PGF2␣ were 1.38, 1.50, 1.52, and 1.55, respectively, for kidneys of hamsters treated with ␣E2, menadione, ␣E2 ϩ ␣EE, and menadione ϩ ␣EE compared with untreated controls (Fig. 6). There were no significant differences in 8-iso- ␣ ␣ ϩ ␣ PGF2␣ levels between E2-, menadione-, E2 EE-, and men- adione ϩ ␣EE-treated groups. Fig. 5. Paraffin section of a tumor-bearing kidney stained with hematoxylin and eosin. The tumors were induced by treatment of male Syrian hamsters Discussion with a combination of ␣EE and menadione for 7 months. (a) Sections of the kidney demonstrate foci of tumor (T) with congested renal tubules (Tb and Several potent estrogens, such as ␣EE, 2-fluoroestradiol, and ␤ arrowheads). (b) The tumor is comprised of hyperchromatic cells with nuclear 11 -, have previously been identified, which are crowding (arrows). Congested renal tubules (arrowheads) and atrophied weakly or not at all carcinogenic despite their high hormonal glomeruli (G) are also seen entrapped within the tumor nodule. Magnifica- potencies as measured by receptor binding, uterine wet weight tion: a ϭϫ10; b ϭϫ40.

3916 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0437929100 Bhat et al. Downloaded by guest on September 26, 2021 and redox cycling of estrogen-quinone metabolites generates free radicals by a mechanism similar to redox cycling of mena- dione and its semiquinone (20, 21, 26, 41). Moreover, tumors induced by ␤E2, or by a combination of ␣EE plus menadione, showed similar histological and immunocytochemical character- istics. Although early signs of proliferation in the renal intersti- tium, as demonstrated by foci with atypical collection of inter- stitial cells and increased 8-iso-PGF2␣ levels, were seen in the kidneys of hamsters treated with a combination of ␣E2 and ␣EE, tumors were not clearly visible. It may take Ͼ7 months for tumors to develop with this treatment regimen. It has been shown earlier that tumor foci and early neoplastic cells arise in the renal interstitium (12, 46, 47). It is also possible that ␣EE, which is known to inhibit cytochrome P450 enzyme activity, may reduce the catechol forming potential of both ␣E2 and ␣EE (48, ␣ 49). E2 forms 8-iso-PGF2␣ at reduced levels compared with Fig. 6. 8-iso-PGF2␣ levels in kidney homogenates of hamsters treated with ␤ ␣ ␣E2, ␤E2, ␣EE, menadione, ␣E2 ϩ ␣EE, or menadione ϩ ␣EE for 7 months. E2. This observation may indicate that E2 catechols may be ␤ 8-iso-PGF2␣ was quantified by using a commercially available kit from Assay methylated faster than E2 catechols, thus making them avail- Designs as described in Materials and Methods.AϾ2-fold increase in 8-iso- able at a reduced level for catechol-quinone redox cycling. ␤ PGF2␣ is detected in E2-treated tumor-bearing kidneys of hamsters. The fold Cotreatment with menadione and ␣EE resulted in tumor devel- increases in 8-iso-PGF2␣ are 1.38, 1.50, 1.52, and 1.55, respectively, for kidneys opment in the hamster kidney. Menadione may induce cyto- ␣ ␣ ϩ ␣ ϩ ␣ of hamsters treated with E2, menadione, E2 EE, and menadione EE chrome P450 activity, or it may antagonize the inhibitory effect compared with untreated controls. 8-iso-PGF2␣ and protein were analyzed on P450 by ␣EE. The reduced ability of ␣EE to form catechol from 10 kidney homogenates from each group, and data are expressed as Ϯ Ͻ estrogens has been suggested to be responsible for the poor mean 8-iso-PGF2␣ pg/mg protein SEM. *, P 0.05 compared with untreated ␣ controls by an unpaired t test; **, P Ͻ 0.05 compared with ␣EE treated group carcinogenic potential of EE (17). Menadione may also inhibit by an unpaired t test. the catechol-O-methyl transferase-mediated conversion of 2- and 4-hydroxylated ␣EE to methoxyestrogens, thus leading to an increased state of oxidant stress as a result of metabolic redox cycling (16, 19, 20). Metabolic redox cycling between catechol cycling of estrogen catechols and quinones. No evidence of estrogens and their corresponding quinones generates oxidative hemosiderin was observed in kidney sections of hamsters treated stress and potentially harmful free radicals (21, 22). Therefore, with menadione or menadione plus ␣EE, suggesting that the redox cycling and free radical generation represent a potential histopathological effects of menadione are not through its effects mechanism, whereby relatively low concentrations of active on iron homeostasis (50). Although subchronic treatment of metabolites might generate major cell damage. Reactive oxygen hamsters with menadione or ␣E2 also increased 8-iso-PGF2␣ species and metabolic activation have been known to modulate levels compared with untreated controls, tumors were not gene expression and several studies support the role of oxidative detected in these two treatment groups. These chemicals either stress in tumor formation (21, 22, 41–45). lack or have weak estrogenic activities (24). It appears that both We hypothesized that if we use a combination of a potent estrogenic potential and oxidative stress as a result of metabolic estrogen with poor ability to form catechol estrogens such as redox cycling of estrogen metabolites are essential for estrogen- ␣EE (15, 17) and a chemical that is known to produce oxidative induced carcinogenesis, because, if estrogen is withdrawn, tu- stress such as menadione (26–28), we should be able to mimic the mors will regress (10). The hormonal effects of estrogens may effect of ␤E2, a tumorigenic potent estrogen that is also a strong promote the development of tumors. catechol progenitor (10, 12, 16, 19, 24). A similar effect may be Our results demonstrate that tumors can be induced in a observed if hamsters are treated with a combination of a rodent model of hormonal carcinogenesis by subchronic treat- noncarcinogenic estrogen of poor hormonal potency but with ment with a combination of a noncarcinogenic estrogen and a good catechol forming capability such as ␣E2 (24, 25) and a chemical known to produce oxidative stress. Our results are in potent estrogen such as ␣EE (15, 17, 24). As expected, the group agreement with the role of oxidant stress in estrogen-induced of hamsters treated with ␤E2 showed 90% tumor incidence, carcinogenesis. It has been recently shown that hydroxylated and which is in agreement with the previously reported results (10, O-methylated estrogens account for nearly 95% of the total Ͼ 12). This group also showed a 2-fold increase in 8-iso-PGF2␣ estrogen in normal human breast and in human breast tumor levels compared with untreated controls, suggesting an increased tissue, with no difference in O-methylated estrogens between the state of oxidative stress. Chronic treatment of hamsters with ␤E2 tumor and control group (51). Both 2- and 4-hydroxy-␤E2 levels for 9 days has previously been shown to result in increased lipid have been shown to be significantly increased in the human hydroperoxide levels in kidney, the target organ of estrogen- breast tumor compared with normal tissue, with 4-hydroxy-␤E2 induced carcinogenesis in hamster, but not in liver, a nontarget showing an Ϸ20-fold increase in breast tumor tissue compared organ (44). Kidneys of hamsters treated with a combination of with normal tissue (51). The elevated expression of estrogen-4- ␣E2 and ␣EE showed signs of early neoplastic changes with hydroxylase activity has been demonstrated in organs of rodents proliferation of interstitial cells. Tumors were clearly seen in that develop estrogen-induced tumors, in MCF-7 human breast kidneys of hamsters treated with a combination of ␣EE and cancer cells, in human uterine myoma, and in human breast menadione. This treatment group also showed a significant cancer, but not in livers of these species (16, 18, 36–40). As ␣ increase in 8-iso-PGF2␣ levels compared with EE-treated suggested by several investigators, 4-hydroxy-␤E2 may undergo group and compared with untreated controls. Menadione is metabolic redox cycling between its catechol and quinone me- reduced to a semiquinone radical through one electron reduction tabolites and potentially generate harmful free radicals and catalyzed by cellular reductases (26). Redox cycling of menadi- oxidative stress (19–22). Therefore, cancer may develop only in one has been shown to generate free radicals͞reactive oxygen those organs that synthesize hydroxyl metabolites of estrogen

species, and has been widely used to investigate chemical- at the target site of carcinogenesis, which may elicit biological BIOCHEMISTRY induced oxidative stress (26–28). Metabolic activation of ␤E2 activities distinct from ␤E2; most notably, an oxidant stress

Bhat et al. PNAS ͉ April 1, 2003 ͉ vol. 100 ͉ no. 7 ͉ 3917 Downloaded by guest on September 26, 2021 response induced by free radicals generated by metabolic redox This work was supported by National Institutes of Health Grants cycling reactions. In summary, our data support the concept CA66724 and P30 ES09089 (to H.K.B.), ES05785, ES11804, and P30 that oxidative stress plays a critical role in estrogen-induced ES09089 (to G.M.C. and T.K.H.); and P20 R11145 (to J.V.V.); and the carcinogenesis. Jean Sindab͞AVON Breast Cancer Foundation grant (to H.K.B.).

1. Henderson, B. E. & Feigelson, H. S. (2000) Carcinogenesis 21, 427–433. 26. Thor, H., Smith, M. T., Hartzell, P., Bellomo, G., Jewell, S. A. & Orrenius, S. 2. Ross, R. K., Paganini-Hill, A., Wan, P. C. & Pike, M. C. (2000) J. Natl. Cancer (1982) J. Biol. Chem. 257, 12419–12425. Inst. 92, 328–332. 27. Ngo, E. O., Sun, T. P., Chang, J. Y., Wang, C. C., Chi, K. H., Cheng, A. L. & 3. Key, T. J. (1999) Endocrine-Related Cancer 6, 175–180. Nutter, L. M. (1991) Biochem. Pharmacol. 42, 1961–1968. 4. Verloop, J., Rookus, M. A. & Van Leeuwen, F. E. (2000) N. Engl. J. Med. 342, 28. Chang, M., Shi, M. & Forman, J. (1992) Am. J. Physiol. 262, L637–L643. 1838–1839. 29. Morrow, J. D., Hill, K. E., Burk, R. F., Nammour, T. M., Badr, K. F. & Roberts, 5. Rossouw, J. E., Anderson, G. L., Prentice, R. L., LaCroix, A. Z., Kooperberg, L. J., II (1990) Proc. Natl. Acad. Sci. USA 87, 9383–9387. C., Stefanick, M. L., Jackson, R. D., Beresford, S. A., Howard, B. V., Johnson, 30. Pratico, D., Lawson, J. A. & FitzGerald, G. A. (1995) J. Biol. Chem. 270, K. C., Kotchen, J. M. & Ockene, J. (2002) J. Am. Med. Assoc. 288, 321–333. 9800–9805. 6. Liehr, J. G. (2000) Endocr. Rev. 21, 40–54. 31. Liehr, J. G., Sirbasku, D. A., Jurka, E., Randerath, K. & Randerath, E. (1988) 7. Li, J. J., Gonzalez, A., Banerjee, S., Banerjee, S. K. & Li, S. A. (1993) Environ. Cancer Res. 48, 779–783. Health Perspect. 101 (Suppl. 5), 259–264. 32. Iturri, G. M. & Cox, H. W. (1969) Military Med. 134, 1119–1128. 8. Russo, I. H. & Russo, J. (1998) J. Mammary Gland Biol. Neopl. 3, 49–61. 33. Liehr, J. G. (1983) Mol. Pharmacol. 23, 278–281. 9. Cavalieri, E., Frenkel, K., Liehr, J. G., Rogan, E. & Roy, D. (2000) J. Natl. 34. Liehr, J. G., Purdy, R. H., Baren, J. S., Nutting, E. F., Colton, F. & Randerath, Cancer Inst. Monogr. 27, 75–93. E. (1987) Cancer Res. 47, 2583–2588. 10. Kirkman, H. (1959) Natl. Cancer Inst. Monogr. 1, 1–57. 35. Li, S. A., Klicka, J. K. & Li, J. J. (1985) Cancer Res. 45, 181–185. 50, 11. Newbold, P. R., Bullock, B. C. & McLachlan, J. A. (1990) Cancer Res. 36. Bunyagidj, C. & McLachlan, J. A. (1988) J. Biochem. 31, 795–801. 7677–7681. 37. Bui, Q. D. & Weisz, J. (1988) Pharmacology 36, 356–364. 12. Bhat, H. K., Hacker, H. J., Bannasch, P., Thompson, E. A. & Liehr, J. G. (1993) 38. Bui, Q. D. & Weisz, J. (1989) Endocrinology 124, 1085–1087. Cancer Res. 53, 5447–5451. 39. Liehr, J. G., Ricci, M. J., Jefcoate, C. R., Hannigan, E. V., Hokanson, J. A. & 13. Shull, J. D., Spady, T. J., Snyder, M. C., Johansson, S. L. & Pennington, K. L. Zhu, B. T. (1995) Proc. Natl. Acad. Sci. USA 92, 9220–9224. (1997) Carcinogenesis 18, 1595–1601. 40. Liehr, J. G. & Ricci, M. J. (1996) Proc. Natl. Acad. Sci. USA 93, 3294–3296. 14. Liehr, J. G., Stancel, G. M., Chorich, L. P., Bousfield, G. R. & Ulubelen, A. A. 41. Jan, S. T., Devanesan, P. D., Stack, D. E., Ramanathan, R., Byun, J., Gross, (1986) Chem. Biol. Interact. 59, 173–184. M. L., Rogan, E. G. & Cavalieri, E. L. (1998) Chem. Res. Toxicol. 11, 15. Li, J. J., Li, S. A., Oberly, T. D. & Parson, J. A. (1995) Cancer Res. 55, 4347–4351. 412–419. 16. Weisz, J., Bui, Q. D., Roy, D. & Liehr, J. G. (1992) Endocrinology 131, 655–661. 42. Wang, M. Y. & Liehr, J. G. (1995) Carcinogenesis 16, 1941–1945. 17. Zhu, B. T., Roy, D. & Liehr, J. G. (1993) Endocrinology 132, 577–583. 43. Wang, M. Y. & Liehr, J. G. (1994) J. Biol. Chem. 269, 284–291. 18. Spink, D. C., Eugster, H. P., Lincoln, D. W., Schuetz, J. D., Schuetz, E. G., 44. Malins, D. C., Polissar, N. L. & Gunselman, S. J. (1996) Proc. Natl. Acad. Sci. Johnson, J. A., Kaminsky, L. S. & Gierthy, J. F. (1992) Arch. Biochem. Biophys. USA 93, 2557–2563. 293, 342–348. 45. Amstad, P. A., Krupitza, G. & Cerutti, P. A. (1992) Cancer Res. 52, 3952–3960. 19. Liehr, J. G., Ulubelen, A. A. & Strobel, H. W. (1986) J. Biol. Chem. 261, 46. Hacker, H. J., Bannasch, P. & Liehr, J. G. (1988) Cancer Res. 48, 971–976. 16865–16870. 47. Gonzales, A., Oberley, T. D. & Li, J. J. (1989) Cancer Res. 49, 1020–1028. 20. Liehr, J. G. & Roy, D. (1990) Free Radical Biol. Med. 8, 415–423. 48. White, I. N. H. & Muller-Eberhard, U. (1977) Biochem. J. 166, 57–64. 21. Cavalieri, E. L., Rogan, E. G. & Chakravarti, D. (2002) Cell. Mol. Life Sci. 59, 49. Oritz de Montellano, P. R. & Kunze, K. L. (1980) J. Biol. Chem. 255, 665–681. 5578–5585. 22. Cavalieri, E. L. & Rogan, E. G. (1992) Pharmacol. Ther. 55, 183–199. 50. Calderaro, M., Martins, E. A. & Meneghini, R. (1993) Mol. Cell. Biochem. 126, 23. Liehr, J. G. (1997) Environ. Health Perspect. 105, 565–569. 17–23. 24. Korenman, S. G. (1969) 13, 163–177. 51. Castagnetta, L. A. M., Granata, O. M., Traina, A., Ravazzolo, B., Amoroso, M., 25. Hershey, R. M., Lloyd, T., MacLusky, N. J. & Weisz, J. (1982) Endocrinology Miele, M., Bellavia, V., Agostara, B. & Carruba, G. (2002) Clin. Cancer Res. 111, 1734–1736. 8, 3146–3155.

3918 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0437929100 Bhat et al. Downloaded by guest on September 26, 2021