Cancer Letters 143 (1999) 63±69
Dietary zinc de®ciency enhances esophageal cell proliferation and N-nitrosomethylbenzylamine (NMBA)-induced esophageal tumor incidence in C57BL/6 mouse Louise Y.Y. Fong*, Peter N. Magee Department of Microbiology and Immunology, Kimmel Cancer Institute, Thomas Jefferson University, 1020 Locust Street, Philadelphia, PA 19107, USA Received 22 February 1999; received in revised form 3 May 1999; accepted 6 May 1999
Abstract The effect of zinc de®ciency on N-nitrosomethylbenzylamine (NMBA)-induced esophageal tumor formation in rats has been well documented. Our previous work showed that zinc de®ciency and its associated increased esophageal cell proliferation were of paramount importance in esophageal tumor development in the NMBA-rat model. However, there has been no report concerning zinc de®ciency and NMBA-induced esophageal tumor formation in mice. In this study, weanling C57BL/6 mice were fed ad libitum with either a zinc-suf®cient or a zinc-de®cient diet containing 3±4 ppm of zinc, and received six intragastric doses of NMBA (2 mg/kg; twice weekly for 3 weeks). The animals were sacri®ced 46 weeks later after in vivo bromodeoxyuridine (BrDU) labeling followed by immunohistochemical detection of cells in S-phase. At 46 weeks, the tumor incidences in zinc-de®cient mice were 57, 100, and 100% respectively, in the esophagus, forestomach and squamocolumnar junction with the glandular stomach (SCJ), as compared to 17, 39, and 67% in the corresponding tissue of zinc-suf®cient mice. The difference between the two dietary groups was signi®cant at P , 0:02 for the esophagus, and P , 0:001 for the forestomach and the SCJ. BrDU labeling revealed that the esophageal labeling index and the number of labeled cells were increased by zinc de®ciency. These results support a role of increased cell proliferation in esophageal carcinogenesis in the mouse. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Mouse; Zinc de®ciency; N-nitrosomethylbenzylamine; Tumorigenesis; Cell proliferation
1. Introduction The enhancing effect of dietary zinc de®ciency on the incidence of esophageal tumors induced by Nnitrosomethylbenzylamine (NMBA) has been clearly demonstrated in rats [1±9]. Zinc de®ciency was known to produce hyperplasia and hyperkeratosis in the rat esophagus [10,11]. More recently, using in * Corresponding author. Tel.: 1 1-215-503-4763; fax: 1 1-215923-7144. E-mail address:
[email protected] (L.Y.Y. Fong)
vivo 5-bromo-2 0 -deoxyuridine (BrDU, a thymidine analog) labeling, followed by visualization of cells in S-phase by immunohistochemistry, we have established a direct relationship between cell proliferation induced by zinc de®ciency and esophageal tumor incidence in rats [6]. Our results [7] further showed that sustained, increased cell proliferation can drive an otherwise non-tumorigenic dose of NMBA into a highly tumorigenic one in the rat esophagus. However, there has been no report to date concerning zinc de®ciency and NMBA-induced tumor incidence
0304-3835/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0304-383 5(99)00191-3
64
L.Y.Y. Fong, P.N. Magee / Cancer Letters 143 (1999) 63±69
in the mouse. Since the genetics of the mouse are better known than those of the rat, a zinc-de®cient mouse-esophageal cancer model could take advantage of techniques utilizing transgenic and knock-out mice to investigate the mechanistic role of cell proliferation in esophageal tumorigenesis in vivo. This study was undertaken to examine the effect of dietary zinc de®ciency on esophageal cell proliferation and tumor formation in the mouse. In order to compare the effect of zinc de®ciency on esophageal cell proliferation and NMBA-induced tumorigenesis in mice with that in rats [6], in the present study we followed the experimental protocol for zinc-de®cient rats [6]. Thus, weanling male mice were fed a de®cient diet containing 3±4 ppm zinc, and administered six intragastric doses of NMBA at 2 mg/kg, twice weekly for 3 weeks. The mode of carcinogenicity of NMBA in mice depends on the route of administration. NMBA was reported to induce esophageal tumors (forestomachs were not examined) by gavage [12], both esophageal and forestomach tumors when administered in the drinking water [13] and only forestomach tumors when given intraperitoneally [13,14]. In the rat, however, NMBA speci®cally induces esophageal tumors [15]. 2. Materials and methods 2.1. Chemicals and animal diets NMBA was purchased from Ash Stevens Inc. (Detroit, MI). Custom-formulated, egg-white based zinc-de®cient and zinc-suf®cient diets were prepared by Teklad (Madison, WI). Zinc levels in these diets were regularly monitored by atomic absorption spectroscopy in our laboratory and were 3±4, and 74±75 ppm, respectively, for zinc-de®cient and zinc-suf®cient diets [6]. 2.2. Experimental design Weanling, 21-day old, male C57BL/6 mice (9:6 ^ 1:2 g, mean ^ SD) were purchased from Taconic Laboratory (Germantown, NY). The animals were group-housed in suspended stainless steel cages in a temperature/humidity-controlled room with a 12 h light-dark cycle. The animals were randomized into two dietary groups and were fed ad libitum with either
the zinc-de®cient or the control zinc-suf®cient diet, and had free access to deionized drinking water. All mice were weighed weekly, and the amount of food consumed by the animals was estimated twice weekly. After 5 weeks on their respective diets, NMBAtreated mice were given six intragastric doses of NMBA (Ash Stevens, Detroit, MI) over the course of 3 weeks at 2 mg/kg body wt, twice weekly. Control mice were untreated. Forty-six weeks after the ®rst dose, all animals were sacri®ced for end point tumor incidence analysis following BrDU administration so that esophageal cell proliferation was measured in all of these animals. 2.3. In vivo BrDU labeling The mice were each administered three intraperitoneal doses of BrDU, 10 mg/kg body wt, at 24 h intervals, as described previously for the rat [6]. This schedule minimizes any possible effects of diurnal variation and carcinogen treatment on cell replication response. At sacri®ce, the animals were anesthetized with iso¯urane (Ohmeda Inc., NJ), blood was collected from the retro-orbital venous plexus of each animal, and serum was prepared for zinc analysis. Subsequently, the animals were sacri®ced, whole esophagi were excised, opened longitudinally, ®xed in buffered formalin and embedded in paraf®n. Cross sections of the esophagi were cut 4-mm thick. Serial sections were prepared and were either stained with hematoxylin and eosin (HE) for histopathology, or left unstained for BrDU immunohistochemistry. 2.4. BrDU immunohistochemistry BrDU-labeled esophageal sections were mounted on Superfrost Plus slides (Fisher Scienti®c, Pittsburgh, PA) to ensure adhesion during processing. Tissue sections were deparaf®nized, rehydrated in graded alcohols and treated with 2 N HCl to denature DNA. This was followed by an incubation for 20 min in 0.01% pepsin in 0.01 N HCl at 378C. Endogenous peroxidase was inhibited with 3% hydrogen peroxide, and non-speci®c binding sites were blocked with normal rabbit serum. Slides were incubated with the primary antibody, monoclonal mouse anti-BrDU (Becton Dickinson) (1:40 dilution, 1 h), followed by incubation with biotinylated, rabbit anti-mouse antibody (Dako) (1:200 dilution, 30 min). Negative
L.Y.Y. Fong, P.N. Magee / Cancer Letters 143 (1999) 63±69
controls were not incubated with the primary antibody. Slides were then incubated with strepavidinhorseradish peroxidase (Dako) (1:1000 dilution). BrDU incorporation was localized by a ®nal incubation with 3,3 0 -diaminobenzidine tetrahydrochloride (DAB) (Sigma). Slides were then counterstained with hematoxylin, dehydrated and coverslipped. 2.5. Scoring of labeled nuclei Cells that had incorporated BrDU were identi®ed as having a dark-brown to black DAB pigment over their nuclei. The basal epithelial cells of the cross section of an entire esophagus were scored by light microscopy. Labeling index (LI) was calculated by dividing the number of BrDU-labeled cells by the total number of cells, and the result was expressed as a percentage. 2.6. Zinc determination Zinc determinations were performed by atomic absorption spectroscopy using a Perkin-Elmer 703 instrument. Serum zinc content was determined by direct aspiration of 1:10 dilution of serum in deionized water (6). 2.7. Tumor analysis At autopsy, whole esophagi and stomachs were removed carefully and opened longitudinally. Other organs were not examined. The number of animals bearing tumors in the esophagus, forestomach and SCJ were scored. Esophageal tumors with diameters greater than 0.5 mm were mapped. The esophagi were then ®xed in buffered formalin, and examined histologically for the presence or absence of hyperkeratosis, parakeratosis, dysplasia, papillomas and carcinomas. 2.8. Statistical analysis Our cell proliferation data were analyzed by oneway analysis of variance (ANOVA) using the SAS statistical computer program [16,17]. Tumor incidence differences were analyzed by two-tailed Fisher's exact test [16,17].
65
3. Results 3.1. Zinc de®ciency, food intake and body weights Weanling mice fed, ad libitum, a de®cient diet containing 3±4 ppm zinc consumed about 6±9 % less food than mice fed a control diet (75 ppm zinc). Irrespective of NMBA treatment, this reduced food intake was associated with a slightly lower body weight in de®cient mice (P . 0:05) measured at the end point for tumor analysis (Table 1). Overt signs of zinc de®ciency such as retarded growth, loss of hair and foci of alopecia, well described for rats fed a diet containing 3±4 ppm of zinc [6,7,9] were not evident in the de®cient mice fed the same diet. However, serum samples from de®cient mice had signi®cantly lower levels of zinc than those from zinc-suf®cient animals (P , 0:01, Table 1). 3.2. Tumor incidence Table 1 shows zinc-de®cient mice had a 57, 100 and 100% tumor incidence in the esophagus, forestomach and SCJ, respectively, compared with 17, 39, and 67% in the corresponding tissue for zinc-suf®cient mice. In addition, zinc-de®cient mice had signi®cantly more esophageal papillomas per mouse than zinc-suf®cient mice (0:95 ^ 1:20 vs. 0.17 ^ 0.38 tumors/mouse, mean ^ SD). Since the tumor counts were not normally distributed, a non-parametric test (Wilcoxon Rank-Sum test) was performed, and the difference between the two groups was signi®cant at P , 0:02 (two-tailed t-test, with a continuity correction factor of 0.5). Also, the highest number of esophageal tumors in a de®cient mouse was ®ve as compared with one in zinc-suf®cient animals. Forestomach tumors were not counted, but they were noticeably more numerous in de®cient mice. Also, six of 18 (33%) zinc-suf®cient mice remained tumor-free at 46 weeks. Fig. 1 shows nodular tumors in the esophagus and forestomach from a moribund, de®cient mouse sacri®ced 14 weeks after NMBA treatment. Fig. 2 illustrates the cross section of an esophageal papilloma from a zinc-de®cient mouse at 46 weeks. 3.3. Esophageal cell proliferation As in zinc-suf®cient rats [6,9], histopathologically, the esophagus from a zinc-suf®cient mouse typically
66
L.Y.Y. Fong, P.N. Magee / Cancer Letters 143 (1999) 63±69
Table 1 Incidence of esophageal and forestomach tumors induced by low doses of NMBA in male C57BL mice fed zinc-de®cient and zinc-suf®cient diets Diet (group)
Total NMBA dose, (mg/kg) Body weight (g) Serum Zn (mg/100 ml) Tumor incidence (%) Esophagus
Zn de®cient/ 2 NMBA 0 Zn suf®cient/ 2 NMBA 0 Zn de®cient/ 1 NMBA 12 Zn suf®cient/ 1 NMBA 12
30.9 ^ 2.8 34.0 ^ 4.1 32.1 ^ 3.2 34.4 ^ 4.4
66 87 68 92
^ 8b ^5 ^ 7b ^8
Forestomach Junction a
0/8 (0) 0/8 (0) 0/8 (0) 0/8 (0) 0/8 (0) 0/8 (0) 12/21 (57) c 21/21 (100) d 21/21 (100) d 3/18 (17) 7/18 (39) 12/18 (67)
a
Squamocolumnar junction with the glandular stomach. De®cient group is signi®cantly different from respective zinc-suf®cient group, P , 0:01, two-tailed t-test. c Tumor incidence: de®cient group is signi®cantly different from respective zinc-suf®cient group, P , 0:02. d Tumor incidence: de®cient group is signi®cantly different from respective zinc-suf®cient group, P , 0:001, two-tailed Fisher's exact test. b
showed a single layer of basal cells, with an overlying stratum, two to three cells thick, covered by a thin keratinous layer (Fig. 3a). The de®cient-mouse esophageal epithelium was thickened and the overlying stratum showed an increase in width averaging four to six cells in thickness (Fig. 3b). However, the esophageal hyperplasia in mice induced by zinc de®ciency was less severe than that in rats fed the same zinc-de®cient diet [6]. In vivo BrDU labeling and immunostaining revealed that de®cient mouse esophagi had substantially more labeled cells in S-phase than zinc-suf®-
cient mice (Fig. 4a,b). Quantitatively, de®cient mouse esophagus showed signi®cantly increased number of labeled cells, and LI (%) as compared with those in zinc-suf®cient group (Table 2). The esophageal LI was 40 ^ 4:5 (mean ^ SD) in de®cient mice, as compared with 27 ^ 3:2 in zinc-suf®cient animals. However, there was no difference in the total number of cells between the two groups.
4. Discussion Compared with rats [6,7,9,18], dietary zinc de®ciency does not have as profound an effect in mice. For example, overt manifestation of zinc de®ciency such as extreme growth retardation, loss of hair and
Fig. 1. Esophagus and forestomach from a zinc-de®cient mouse 14 weeks after NMBA treatment showing nodular tumors.
Fig. 2. Microphotograph showing a cross section of a pedunculated esophageal squamous cell papilloma projecting into the esophageal lumen from a NMBA-treated zinc-de®cient mouse at 46 weeks. HE. Microscope setting £ 200.
L.Y.Y. Fong, P.N. Magee / Cancer Letters 143 (1999) 63±69
67
esophagus, the forestomach and the SCJ than their similarly-treated zinc-suf®cient counterparts (Table 1). In addition, in vivo BrDU labeling and immunohistochemical visualization of cells in S-phase demonstrated that zinc-de®cient mice had signi®cantly more labeled cells and a higher LI than zincsuf®cient mice (Table 2). These results are in line with our previous ®ndings, that sustained, increased esophageal cell proliferation is associated with an increased incidence of esophageal cancer in rats [6], and lend further support to the hypothesis that increased esophageal cell proliferation is linked to an increased risk of the human cancer [19]. Upon bioactivation, NMBA produces benzaldehyde and an electrophilic methylating agent [20] which methylates DNA resulting in the formation of the promutagenic adduct, O 6-methylguanine (O 6-meG) [21±23]. The enhanced carcinogenic
Fig. 3. Representative microphotographs showing cross sections of esophagi from: (a) a zinc-suf®cient mouse, (b) a zinc-de®cient mouse, sacri®ced 46 weeks after NMBA treatment. The esophageal epithelium from (a) shows a single layer of basal cells and that from (b) is markedly thickened. HE Microscope setting £ 200.
dermal lesions were not found in mice fed the same de®cient diet containing 3±4 ppm zinc. The esophageal cell proliferation induced by this diet was less severe in mice than in rats. Typically, the esophageal epithelium from the zinc-de®cient mouse is markedly less thickened, hyperplastic or dysplastic, with an overlying suprabasal layer averaging four to six cells in thickness (Fig 2b) in comparison with the six to ten cells of the zinc-de®cient rat. It is not known if zinc de®ciency and its associated esophageal cell proliferation, per se, can induce tumors in the esophagus of the mouse as it does in the rat [6,8]. In addition, NMBA-treated zinc-de®cient mice had substantially lower incidence of esophageal tumors (57%, Table 1) than similarly-treated zinc-de®cient rats (100%) [6]. Nonetheless, NMBA-treated zinc-de®cient mice had signi®cantly higher incidence of tumors in the
Fig. 4. Representative microphotographs showing typical pattern of immunostaining with BrDU monoclonal antibody to detect cells in S-phase in esophagi from (a) a zinc-suf®cient mouse, (b) a zincde®cient mouse, sacri®ced 46 weeks after NMBA treatment. Counterstaining with hematoxylin. Microscope setting £ 200.
68
L.Y.Y. Fong, P.N. Magee / Cancer Letters 143 (1999) 63±69
Table 2 Effect of zinc de®ciency on epithelial cell proliferation in mouse esophagus as determined by in vivo BrDU immunohistochemistry a Group
No. of labeled cells
Total no. of cells
LI (%)
Zn de®cient/ 1 NMBA Zn suf®cient/ 1 NMBA
383 ^ 74 b 224 ^ 33
944 ^ 102 843 ^ 101
40 ^ 4.5 c 27 ^ 3.2
a Results are mean ^ SD on a sample size of ten animals per group. LI (%) is the ratio between the number of BrDU-labeled cells in S phase and the total number of cells per cross section of an entire esophagus. b Signi®cantly different from Zn suf®cient/ 1 NMBA group, P , 0:01. c Signi®cantly different from Zn suf®cient/ 1 NMBA group, P , 0:001.
activity of NMBA observed in zinc de®ciency is probably brought about by a combination of, (a) increased esophageal microsomal metabolism of NMBA [4], (b) increased formation of the promutagenic adduct O 6-meG in the esophagus [24], (c) depressed esophageal activities of the repair enzyme for O 6-meG, namely, O 6-alkylguanine-DNA-methyltransferase [25] and (d) increased esophageal cell proliferation, as demonstrated in this study. However, our previous results [6,7,9,18] suggest that sustained, increased cell proliferation is likely to be the rate-determining factor for tumor incidence after low doses of NMBA. On the other hand, zinc de®ciency in vivo causes a number of biochemical changes, including oxidative damage [26,27] and abnormal methyl methionine metabolism [28], which have been associated with carcinogenesis brought about by other agents. All our previous work regarding the role of cell proliferation on esophageal carcinogenesis was performed in the rat [6,7,9,18]. The zinc de®ciencyNMBA model in rats has many attractions because proliferation is induced by a reduced dietary intake of an essential trace metal, and it can be reversed by its replenishment [18] and inhibited by an anti-cancer drug, a-di¯uoromethylornithine [9]. This in vivo model is also relevant to human cancer, since epidemiological studies have implicated dietary zinc de®ciency [29±31] and exposure to carcinogenic nitrosamines, NMBA in particular, in the high incidence of esophageal cancer in northern China and parts of Iran [30,32±34]. In view of the fact that the genetics of the mouse is better known than those of the rat, we envision that a zinc-de®cient mouse model could take advantage of transgenic and knock-out mouse techniques to help to elucidate the mechanisms of cell proliferation in esophageal cancer develop-
ment, and its prevention. The present results demonstrate such a mouse model is feasible. It is also possible to get a more marked effect of dietary zinc de®ciency on esophageal cell proliferation by further reducing the level of zinc in the diet to approx. 1±2 ppm, at which level, unlike rats [11], the mouse can still be maintained for a long term experimental study [35,36]. Acknowledgements We would like to thank Laisen Qian for expert technical assistance, and Mr. Karl Smalley for help with statistical analysis of data. This study was supported by Grant No. 95A70-REN from the American Institute for Cancer Research and by Grant No. CN-123 from the American Cancer Society. References [1] L.Y.Y. Fong, A. Sivak, P.M. Newberne, Zinc de®ciency and methylbenzylnitrosamine-induced esophageal cancer in rats, J. Natl. Cancer Inst. 61 (1978) 145±150. [2] S.J. van Rensburg, D.B. du Bruyn, D.J. van Schalkwyk, Promotion of methylbenzyl-nitrosamine-induced esophageal cancer in rats by subclinical zinc de®ciency, Nutr. Rep. Int. 22 (1980) 891±899. [3] G.N. Gabrial, T.F. Schrager, P.M. Newberne, Zinc de®ciency, alcohol and retinoid: association with esophageal cancer in rats, J. Natl. Cancer Inst. 68 (1982) 785±789. [4] D.H. Barch, S. Kuemmerle, P. Hollenberg, P. Iannaccone, Esophageal microsomal metabolism of N-nitrosomethylbenzylamine in the zinc-de®cient rat, Cancer Res. 44 (1984) 5629±5633. [5] L.Y.Y. Fong, J.S.K. Lee, W.C. Chan, P.M. Newberne, Zinc de®ciency and the development of esophageal and forestomach tumors in Sprague±Dawley rats fed precursors of Nnitrosobenzylmethylamine , J. Natl. Cancer Inst. 72 (1984) 419±425.
L.Y.Y. Fong, P.N. Magee / Cancer Letters 143 (1999) 63±69 [6] L.Y.Y. Fong, J.X. Li, J.L. Farber, P.N. Magee, Cell proliferation and esophageal carcinogenesis in the zinc-de®cient rat, Carcinogenesis 17 (1996) 1841±1848. [7] L.Y.Y. Fong, K.M. Lau, K. Huebner, P.N. Magee, Induction of esophageal tumors in zinc-de®cient rats by single low doses of N-nitrosomethylbenzylamine (NMBA): analysis of cell proliferation, and mutations in H-ras and p53 genesCarcinogenesis 18 (1997) 1477±1484. [8] P.M. Newberne, T.S. Schrager, S. Broitman, Esophageal carcinogenesis in the rat: zinc de®ciency and alcohol effects on tumor induction, Pathobiology 65 (1997) 39±45. [9] L.Y.Y. Fong, A.E. Pegg, P.N. Magee, a-Di¯uoromethylornithine inhibits N-nitrosomethyl-benzylamine-induced esophageal carcinogenesis in zinc-de®cient rats: effects on esophageal cell proliferation and apoptosis, Cancer Res. 58 (1998) 5380±5388. [10] R.H. Follis Jr., The pathology of zinc de®ciency, in: A.S. Prasad (Ed.), Zinc Metabolism, Charles C. Thomas, Spring®eld, IL, 1966, pp. 129±141. [11] H. Swenerton, L.S. Hurley, Severe zinc de®ciency in male and female rats, J. Nutr. 95 (1968) 8±18. [12] O.E. Oldeleye, C.D. Eskelson, S.I. Mufti, R.R. Watson, Vitamin E protection against nitrosamine-induced esophageal tumor incidence in mice immunocompromised by retroviral infection, Carcinogenesis 13 (1992) 1811±1816. [13] J. Sander, F. Schweinsberg, Tumorinduktion bei Mausen durch N-methylbenzyl-nitrosamin in niedriger Dosierung, Z. Krebsforsch. 19 (1973) 157±161. [14] P. Schneider, S. Hinrichs, R. Zulim, R. Towery, C. Morris, S.S. Mirvish, Carcinogenesis by methylbenzylnitrosamine near the squamocolumnar junction and methylamylnitrosamine metabolism in the mouse forestomach, Cancer Lett. 102 (1996) 125±131. [15] P.N. Magee, J.M. Barnes, Carcinogenic nitroso compounds, Adv. Cancer Res. 10 (1967) 163±246. [16] P. Armitage, G. Berry, Statistical Methods in Medical Research, Blackwell, Oxford, 1987. [17] The GLM procedure, SAS/STAT User's Guide, 4th edition, vol. 2, Cary, NC, SAS Institute Inc, 1989. [18] L.Y.Y. Fong, J.L. Farber, P.N. Magee, Zinc replenishment reduces esophageal cell proliferation and N-nitrosomethylbenzylamine (NMBA)-induced esophageal tumor incidence in zinc-de®cient rats, Carcinogenesis 19 (1998) 1591±1596. [19] L.D. Wang, M. Lipkin, S.L. Qui, G.R. Yang, C.S. Yang, H.L. Newmark, Labeling index and labeling distribution of cells in esophageal epithelium of individuals at increased risk for esophageal cancer in Huixian, China, Cancer Res. 50 (1990) 2651±2653. [20] G.E. Labuc, M.C. Archer, Esophageal and hepatic microsomal metabolism of N-nitroso-methylbenzylamine and N-nitrosodimethylamine in the rat, Cancer Res. 42 (1982) 3181± 3186. [21] L.Y.Y. Fong, H.J. Lin, C.H. Lee, Methylation of DNA in
[22]
[23] [24]
[25] [26] [27] [28] [29]
[30] [31] [32]
[33] [34] [35]
[36]
69
target and non-target organs of the rat with methylbenzylnitrosamine and dimethylnitrosamine, Int. J. Cancer 23 (1979) 679±682. R.M. Hodgson, M. Wiessler, P. Kleihues, Preferential methylation of target organ DNA by the esophageal carcinogen Nnitrosomethylbenzylnitrosamine, Carcinogenesis 1 (1980) 861±866. H. Autrup, G.D. Stoner, Metabolism of N-nitrosamines by cultured human and rat esophagus, Cancer Res. 42 (1982) 1307±1311. D.H. Barch, C.C. Fox, Dietary zinc de®ciency increases the methylbenzylnitrosamine-induced formation of O 6-methylguanine in the esophageal DNA of the rat, Carcinogenesis 8 (1987) 1461±1464. L.Y.Y. Fong, T. Cheung, Y.S. Ho, Effect of nutritional zincde®ciency on O 6-alkylguanine-DNA-methyltransferase activities in rat tissues, Cancer Lett. 42 (1988) 217±223. J.F. Sullivan, M.M. Jetton, H.K. Hahn, E. Burch, Enhanced lipid peroxidation in liver microsomes of zinc-de®cient rats, Am. J. Clin. Nutr. 33 (1980) 51±56. G.H. Cao, J.D. Chen, Effects of dietary zinc on free radical generation, lipid peroxidation, and superoxide dismutase in trained mice, Arch. Biochem. Biophys. 291 (1991) 147±153. J.A. Duerre, J.C. Wallwork, Methionine metabolism in isolated perfused livers from rats fed on zinc-de®cient and restricted diets, Br. J. Nutr. 56 (1986) 395±405. Joint Iran-International Agency for Research on Cancer Study Group, Esophageal cancer studies in the Caspian littoral of Iran: results of population studies ± a prodrome, J. Natl. Cancer Inst. 59 (1977) 1127±1138. C.S. Yang, Research on esophageal cancer in China, Cancer Res. 40 (1980) 2633±2644. S.J. van Rensburg, Epidemiologic and dietary evidence for a speci®c nutritional predisposition to esophageal cancer, J. Natl. Cancer Inst. 67 (1981) 243±251. S.H. Lu, R. Montesano, M.S. Zhang, L. Feng, F.J. Luo, S.X. Chui, D. Umbenhauer, R. Saffhill, M.F. Rajewsky, Relevance of N-nitrosamines to esophageal cancer in China, J. Cellular Physiol. Suppl. 4 (1986) 51±58. S.H. Lu, S.X. Chui, W.X. Yang, X.N. Hu, L.P. Guo, F.M. Li, Relevance of N-nitrosamines to esophageal cancer in China, IARC, Lyon, 1991 pp. 11±17. P.N. Magee, The experimental basis for the role of nitroso compounds in human cancer, Cancer Surveys 8 (1989) 207± 239. D.L. Donaldson, C.C. Smith, M.S. Walker, O.W. Rennert, Tissue zinc and copper levels in diabetic C57BL/KsJ (db/ db) mice fed a zinc-de®cient: lack of evidence for speci®c depletion of tissue zinc store, J. Nutr. 118 (1988) 1502±1508. H.N. Shi, M.E. Scott, M.M. Stevenson, K.G. Koski, Energy restriction and zinc de®ciency impair the functions of murine T cells and antigen-presenting cells during gastrointestinal nematode infection, J. Nutr. 128 (1997) 20±27.