Inhibition by transforming growth factor (34-43)-alpha, a TGF-alpha antagonist, of gastric carcinogenesis induced by N-methyl-N'-nitro-N-nitrosoguanidine in Wistar rats.

The effect of prolonged administration of transforming growth factor (34-43)-alpha, an antagonist of TGF-alpha, on gastric carcinogenesis induced by N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) and on the labelling and apoptotic indices and TGF-alpha immunoreactivity of gastric mucosa and gastric cancers was examined in Wistar rats. The rats received intraperitoneal injections of 10 or 20 microg kg(-1) body weight of TGF(34-43)-alpha every other day after oral treatment with MNNG for 25 weeks. Long-term administration of TGF(34-43)-alpha at both doses significantly reduced the incidence of gastric cancers at the end of the experiment in week 52. However, TGF(34-43)-alpha had no significant effect on the number, histological type or depth of involvement of gastric cancers. Administration of TGF(34-43)-alpha also significantly decreased the bromodeoxyuridine labelling index and TGF-alpha immunoreactivity, and significantly increased the apoptotic index of antral mucosa and gastric cancers. These findings indicate that TGF(34-43)-alpha inhibits gastric carcinogenesis, and that its effects are mediated through decreased cell proliferation and TGF-alpha immunoreactivity and increased apoptosis induction in the gastric cancers.

logical effects on a variety of epithelial cells (Liu et al. 1994). It is a cvtokine that increases cell proliferation and transformation of x-arious cells Ilihara et al. 1993: Liu et al. 1994: Llio et al. 1995: Ciacci et al. 1996: Gogusev et al. 1996: Taga et al. 1996. How-ever. the role that exogenous TGF-a ma) plax in cell proliferation in Xivo is poorlN understood. Hormi and Lehy 11996) prox ed for the first time the stimulatorx effect in vivo of exouenous rat TGF-a on epithelial cell proliferation in antral. duodenal and colonic mucosae.
Perez-Tomas et al 11992) examined the distribution pattern of TGF-a in experimental hepatocarcinogenesis induced by dimethx Initrosamine and found that TGF-a x-as obserxed immunohistochemicallx in hepatic tumour cells. Wang et al 1996) also found increased lexels of TGF-ca mRrNA and protein products in papillomas and in pronounced hx perplastic and dysplastic lesions in rats treated with the chemical carcinocen N-nitroso-methN lbenzvlamine. and concluded that TGF-a may play an important role in experimental oesophageal tumorigenesis in rats. Livingstone et al (1994) reported that the carcinogen ANT-methx v-Vnitro-N-nitrosoguanidine ( MNNG) caused a significant increase in the intensity of TGF-a expression in the gastric mucosa after as little as 16 wxeeks' exposure. These findings indicate that TGF-cx mav be involx-ed in gastric carcinogenesis. but there wxere no reports on the possible role of exogenous TGF-a in gastric carcinogenesis. TGF(34-43 )-c is an antagronist of TGF-a (Nestor

MATERIALS AND METHODS Animals
Sixtv 6-xweek-old male Wistar rats w-ere purchased from Japan SLC ( Shizuoka. Japan). Two rats each w ere housed under standard conditions at a room temperature maintained at 2' 1-22-C with a 12-h liahtldark cvcle.

Experimental design
The animals were given MINNG (50 ig ml-': Aldrich Chemical Co. Milwaukee. W'I. USA) in drinking wxater for 25 Nx-eeks and regular chox pellets (Nihon Nosan. Yokohama. Japan) oxer the entire study period. The MNNG x-as dissolved in deionized xwater at a concentration of 1 mg ml-' and kept in a cool (4 C). dark place. Just before use. the stock solution >-as diluted to 50 PLg ml' with tap wxater. Forty millilitres of MNNG solution (less than a single rat consumes in 48 h: this procedure did not affect normal bodx weight gain) xas rixven to each rat from bottles coxered with aluminium foil to prexvent photolx-sis of the MNNG. The bottles were refilled exenother dav. From wxeek 26. the rats had free access to ordinary tap water from an automatic wxatering system. At this point. the animals wxere dixided randomly into three groups (20 rats in each). Each group receixved i.p. injections even-other dax until the end of the experiment at week 52 as followxs: group 1. the control group. xxas gixen the xehicle. 0.9%; sodium chloride solution. only: groups 2 and 3 wxere rixven 10 and 20 lgg kgi body weight of TGF(31-43-(-C peptide purity > 99%c: Bachem Fine  (100)  0(0) aFor an explanation of treatment. see Table 1.
Chemiicals. Bubendorf. Sw itzerland) respectively. The TGF( 34-43)-a wvas dissolved in 0.9%e sodium chloride solution just before use. Injections were given at a volume of 2 ml ko-' body weight between 14.00 and 15.00 h each day i.p. to strengthen the pharmacological action of TGF(34-43i-cx. All experimental procedures were approved bx the Animal Care Committee of the Osaka Medical Centre for Cancer and Cardiovascular Diseases.

Histological observations
Animals that survived for more than 50 weeks were included in the effective numbers because the first tumour of the glandular stomach was found in a rat in group 1 that died in week 50. All survivin2 animals were killed and examined at the end of the experiment at A-eek 52. At necropsv. the stomach and other major organs wvere subjected to careful macroscopic examination. The stomach % as opened along the greater curvature. pinned to a cork mat and fixed with a buffered picric acid-formaldehyde solution.
After processing with a method routinely used for histological exarirnation. sections were stained with haematoxvlin and eosin.
Sections w-ere examined x ithout know-ledge of the group to w-hich thev belonged.
Definition and classification of gastric cancers Histolo,icallv. adenocarcinomas w-ere defined as tumours in which the neoplastic glandular tissue had entered the submucosa or deeper lavers. As in a previous study (Tatsuta et al. 1988).
adenocarcinomas A-ere subclassified into three types: very Axelldifferentiated. A ell-differentiated or poorly differentiated.
index w-as expressed as the percentage of labelled cells amonc the cells examined.

Measurement of the apoptotic index
The 3'-end labelling of apoptotic cell DNA A as performed A ith an ApopTag in situ apoptosis detection kit (Oncor. Gaithersburg. MD. USA) (Tormanen et al. 19951. Briefly . after de%vaxing and dehN-dration. sections w ere incubated xx ith 20 go ml-' proteinase K (Boehringer Mannheim. Mannheim. Germanv) at room temperature for 15 min. Endogenous peroxidase activity was quenched in 2c hydrogen peroxidase in phospate-buffered saline (pH 7.2). Terminal transferase enzyme was used to catalyse the addition of digoxigenin-labelled nucleotides to the 3'-hvdroxv ends of fragmented DNA. Antidigoxigeninm-peroxidase solution was then applied to the slides. Diaminobenzidine-hydrogen peroxide w-as used to develop the colour reaction. The specimens were lightly counterstained with haematoxvlin. The apoptotic index was determined as described above.
Immunohistochemical observation of TGF-a Immunohistochermistry was performed with the mouse monoclonal antibodv AB-2 (Oncogene Science. Cambridge. UK). which is specific for human and rat TGF-a and exhibits no crossreactivitx to epidermal growth factor (Lixingstone et al. 1994). Sections were predigested w-ith trypsin for 15 min to expose the antigenic sites before incubation with AB-2 at a dilution of 5:100 overnight at 4 C. After washina with Tris-buffered saline. rabbit anti-mouse serum (Dak-o. UK) and streptavidin-peroxidase complex (Dako) were added at dilutions of 1:333 and 1:400. respectively. for 30 min each before application of diaminobenzidine and counterstained w ith haematoxy lin. After dehydration in alcohol. the sections w-ere cleared with xvlene. and mounted in diphthalate xylene. A positixve control section was incubated in each batch to ensure consistency of staining. Two txypes of negatixe controls were used: in the first. the primary antibodx was replaced by Tris-buffered saline: in the second. specific controls were performed by preincubation of the sections with an excess of the TGF-a peptide PF 008 (Oncogene Science). The relative number of cells that were immunoreactixe for TGF-a was determined as described aboxe.
Statistical analysis Statistical analvsis w as performed w-ith the chi-squared test.
Fisher's exact probability test. or one-wav analvsis of xariance w%vith Dunnms multiple comparison (Miller. 1966). Data are presented as the means ± s.e. Differences w ith calculated P-x alues less than 0.05 w ere regarded as significant.

RESULTS
Incidence, number, histological type and depth of involvement of gastric cancers Administration of TGF( 34-43 )-a had no significant effect on the bodx weiaht of the rats in w-eek 52 (Table 1).
Macroscopically. there xxere no abdominal tissue reactions or damage as a result of direct exposure of TGF(34-43 -a in week 52. In group 1 (control). gastric cancers were found in 19 (95%e) of the 20 rats examined. The incidence of gastric cancers in groups [TGF(34-43i-a at 10 jgc ko-'] and 3 [TGF(34-43 i-a at 20 k(-'] was significantlv lower than in group 1 (Table 1). In group 1. the axerage number of gastric cancers per tumour-bearing rat xxwas 1.9 ± 0.2. Howexer. the difference in the number of gastric cancers amonr the three groups was not significant.
All tumours induced in the glandular stomach were histoloeically determined to be adenocarcinomas (Table 2). Virtuallx all of the adenocarcinomas were xen-well differentiated. The incidence of xerv x-ell-differentiated adenocarcinomas was slightly. but not significantlv. higher in group 3 i-a at 20 jg kg-'] than in group 1 (control). No poorlv differentiated cancers were found in this series. Furthermore. neither dose of TGF 34-43 -a had any effect on the depth of Mixolxrement of the castric cancers (Table 2i. All cancers were found in the antral mucosa. and no macroscopic metastases ere seen in any rat.

Labelling and apoptotic indices and TGF-a immunoreactivity
Administration of TGF(34-43 i-a at 10 jig kgr-(group 2) and 20 ig-kg-l (aroup 3) body xxeight significantl-decreased the BrdU labelling, index and TGF-a immunoreactixits-and significantly increased the apoptotic index of antral mucosa and gastric cancers. as compared with those in control group 1 (Table 3).

DISCUSSION
There are sex eral reports on transgenic mice oxverexpressing TGF-a. Takagi et al (1992) and Sharp et al (1995) established a transgenic line bearing a human TGF-a cDNA driven bx the mouse metallothionein I promoter in the inbred mouse line FVB/N. These mice British Joumal of Cancer (1998) 78(7). 857-861 0 Cancer Research Campaign 1998 develop severe cystic hyperplasia containing mucus-laden secretions in the fundic mucosa of the stomach. Foci of dysplastic cells were seen in the lesions of mice surviving until the later stages of life. However. gastric cancers have never been described until now.
The present study showed that prolonged administration of TGF(34-43-t. an antagonist of TGF-ac at both high and low doses significantly decreased the incidence of gastric cancers induced by MNNG. The exact mechanism by which TGF(34-43)-a inhibits gastric carcinogenesis is not clear. but at least two possible explanations may be considered. One involves the effect of TGF(34-43)-a on cell proliferation. TGF-a is a cytokine which increases cell proliferation of various cells (Taga et al. 1996). Bishop et al (1995) found that TGF-a antisense oligodeoxynucleotides markedly inhibited proliferation of Caco 2 cells. and reported that cholesterol-modified oligodeoxynucleotides were more effective and specific than unmodified oligodeoxynucleotides. Seki et al (1997) reported that culture of a human hepatocellular carcinoma cell line  in the presence of a neutralizing antibody to TGF-a inhibited cell proliferation. These findings indicate that inhibition of TGF-ct may inhibit cell proliferation. The results of the present work show that long-term administration of TGF(34-43)-a significantly decreases the BrdU labelling index of gastric cancers.
TGF-a stimulates cell proliferation through interaction with its receptor. the epidermal growth factor receptor. by activating its tyrosine kinase activities (Wang et al. 1996). Tyrosine kinases are important in the signal transduction of a number of growth factors. In a study of tyrosine phosphorylation in type II pneumocytes exposed to TGF-c. Chess et al (1994) found that after addition of TGF-ct phosphorylation of a tyrosine protein with a molecular mass of 170 kDa. presumed to be the epidermal growth factor receptor. peaked by 5 min and that the tyrosine kinase inhibitor genistein and tyrphostin decreased the TGF-a-induced phosphorylation of the epidermal growth factor receptor.
A second possible explanation for the inhibitory effect of TGF(34-43)-t on gastric carcinogenesis relates to apoptosis. The integrity of the gastrointestinal mucosa is guaranteed by a regulated balance of proliferation, differentiation and physiological cell death of its main constituents. Physiological cell death is known as apoptosis. In a study on the effect of epidermal growth factor and TGF-a on apoptosis of an astrocyte progenitor cell line (AP-16). Yoshida et al ( 1993) found that epidermal growth factor deprivation caused the death of AP-16 cells by apoptosis and that TGF-a prevented apoptosis occurring in the absence of epidermal growth factor. Reinartz et al (1996) reported that induction of apoptosis by tumour necrosis factor-a in the human keratinocyte cell line HaCaT was reduced by preincubation of the cells with TGF-a and that the protective effect of TGF-a was abrogated by translation inhibition. indicating that it depended on de novo protein synthesis. More recently. Seki et al (1997) also found that culture of cells in the presence of a neutralizing antibody to TGFa induced apoptosis of large numbers of cells. The findings of the present study show that prolonged administration of TGF(34-43)-a significantly increases the frequency of apoptosis induction in gastric cancers. Increased induction of apoptosis decreases the susceptibility of an individual to malignancy.
The results presented indicate that administration of TGF(34-43)-a inhibits the development of gastric cancers. and that inhibition of gastric carcinogenesis by TGF(34-43)-ct may be mediated by decreased cell proliferation and TGF-a immunoreactivity and enhanced induction of apoptosis.