There is a fivefold greater incidence of breast cancers and a threefold greater incidence of colon cancers in Western populations compared with Asian populations in which soybean foods comprise ~10% of the total per capita protein consumption in the form of fermented and nonfermented products (Hawrylewicz et al. 1995). Numerous investigators have examined the effectiveness of soy in the diet in animal models of cancer (Barnes 1995, Hawrylewicz et al. 1995, Kennedy 1995) to test the hypothesis that specific components of soy contribute to this epidemiologic effect. The soy products or derivatives in these studies vary widely in their preparation, and the chemical composition of the dietary additives has not been reliably reported.

Components of soybeans shown to have an inhibitory effect on cancer cells include trypsin inhibitors, chief of which is the Bowman Birk inhibitor (BBI,⁴ 6% of the total protein in soy), saponins, phytoestrogens or the isoflavones (genistein, another major component of soybean and widely studied in its purified form), lignans and inositol hexaphosphates (Kennedy 1995). Soy isolates or hydrolysates are prepared by water or acid washing, or by ethanol extraction of soy flour. Acid-washed hydrolysates contain the same amount of isoflavone as soy flour, but ethanol extraction removes a great deal of the isoflavone from soy flour. The anticancer effect of soy diets is destroyed with the removal of isoflavones (Barnes 1995). Variations in experimental results of animal studies may be due to differences in isoflavone content of the experimental diets used. For this reason, most investigators have switched to use of purified genistein for studies of anti-cancer effects of soy products.

Soy protein isolates (SP) added to the diets of rats and mice reduce the incidence and number of breast, skin and colon tumors (Barnes 1995, Barnes et al. 1995). Soybean isolates have been implicated in the suppression of liver and colon carcinomas in rats and mice (Hawrylewicz et al. 1995), the biotherapy of B-cell leukemias in mice (Uckun et al. 1995), and the prevention of breast and prostate cancer in humans (Barnes et al. 1995). There have also been reports of the inhibition of proliferation of human hepatoma, neuroblastoma, leukemia, breast and prostate tumor cell lines by genistein (Peterson 1995).

Because isolated components of soy, such as genistein, have been shown to have an inhibitory effect on cultured cells, it was of great interest to test partially hydrolyzed soy protein isolates on cultured intestinal adenocarcinoma cells to determine the existence of any inhibitory effect in vitro. Two hydrolysates were chosen for this purpose. The starting material, macronutrient distribution and hydrolysis conditions of the two soy protein hydrolysates (SP) used were similar (Table 1). However, one of the hydrolysates (SP-A) was more highly processed in that it was exposed to posthydrolysis fractionation, removing high and very low molecular weight species. As a result, the largest and smallest peptides, as well as the soy-derived phenolics, were greatly reduced in SP-A relative to the less processed hydrolysate, SP-B. SP-A has been shown in preliminary in vivo studies to result in inhibition of enterocyte apoptosis (Cope et al. 1992 and 1996). SP-A and SP-B were selected to see if we could correlate the in vivo observation of inhibition of apoptosis with an in vitro assay and to determine if apoptosis would be enhanced or diminished in the less processed hydrolysate.

Table 1. Characteristics of soy protein hydrolysates (SP) added to C2BBe cell culture media [View Table]

The cells selected for these studies are C2BBe cells, a subclone of the human colonic adenocarcinoma Caco-2 cell line. This subclone is distinct from the parent Caco-2 cell in its expression of enterocytic markers such as brush border myosin I, characteristic of late differentiation (Peterson and Mooseker 1992). Because of their brush border development, these cells more closely resemble enterocytes than colonocytes. They can be grown in confluent monolayers on solid or microporous matrices within 7-10 d, which facilitates the rapid testing of multiple samples.

In addition to the potential of soy hydrolysate for cancer prevention and therapy, these supplements may be useful in postsurgical therapy by enhancing the biological integrity of the intestinal epithelium after surgical trauma, thereby reducing the risk of postsurgical infections (Fine et al. 1959, Fink 1969). Gut-derived systemic sepsis is a major clinical complication of major surgery, shock, pharmacologic immunosuppression, and also occurs after severe trauma or burn injury (Lane 1912, Lillehei 1957). These events may lead to disruption of the mucosal barrier due to hypovolemia, ischemia and cytokine response and permit translocation of luminal bacteria into the bowel wall. To this end, partially purified soy hydrolysates were added to the culture media of C2BBe monolayers. These supplemented cultured cells were tested for their ability to resist bacterial translocation in the hopes that this model would also serve to assist in selection of hydrolysates for clinical trials on the effect of soy supplements on postsurgical infections. For this purpose, we employed the following two enteric pathogens that have been thoroughly studied in in vitro systems: Salmonella typhi Ty2, which is one of the most enteroinvasive pathogens, and S. typhimurium, which is also highly invasive but appears to traverse the mucosal barrier by a different mechanism (Finlay and Falkow 1990, Kops et al. 1996).

Soy protein hydrolysates may be economical therapeutic agents compared with purified materials such as genistein (Barnes et al 1995). The investigations presented here were done to explore whether soy hydrolysates could be efficiently and effectively studied in preclinical tests in which cultured C2BBe intestinal cells were exposed to partially purified hydrolysates in vitro.

MATERIALS AND METHODS

Flasks were treated with trypsin-EDTA (Sigma) to split the cells. After two washes with complete medium, the concentration of the cells was adjusted by counting in a hemocytometer, and 2 × 10⁵ cells were applied to collagen-coated 3-µm pore, 6.5-mm diameter Transwell filter units (Costar, Cambridge, MA). Collagen coating of the filters was accomplished by incubation in 100 µL of Type I collagen solution (Sigma) (0.1 g/L collagen in 0.1 mol/L acetic acid, stored over chloroform to maintain sterility) for 3 h at 37°C, and irradiation with UV light overnight. Medium was added to both lower and upper chambers of the Transwell filter units, with the volumes kept constant at 1.0 mL and 200 µL, respectively, so that fluid levels in both chambers were equal.

Cell cultures were maintained at 37°C in an atmosphere of 5% CO₂ . Cells were fed with fresh medium every other day for 3-4 wk. Confluent monolayers were allowed to form, followed by full differentiation of the apical brush border and intercellular tight junctions over a 28 d period (Peterson and Mooseker 1992).

Electrical resistance across the monolayers was measured with an ohmvoltmeter (World Precision Instruments, New Haven CT) by inserting electrodes into the medium in the upper and lower chambers of the Transwell filter units (Madara and Dharmsathaphorn 1985). Base-line measurements were made on monolayer-free filters. The readings were corrected for the surface area of the monolayer and the data expressed as m·cm². Background readings of the monolayer-free control filters were subtracted from the corrected readings to determine TER. Great care was taken to avoid injuring the monolayers with the probe during measurement of TER, and they were subsequently examined with an inverted microscope to check for possible mechanical disruptions from manipulations during measurements. Filters with mechanical injury were discarded.

All monolayer-containing filters in each experiment contained C2BBe cells of the same passage number and stage of maturation. Four to twelve hours before the start, the medium in both lower and upper chambers of the filter units was replaced with antibiotic-free complete cell culture medium. Salmonellae, grown in overnight cultures, were diluted either in antibiotic-free media or in fresh LB broth. Logarithmic growth phase was determined spectrophotometrically at 595 nm. When cultures reached log phase and an optical density equivalent to between 10¹⁰ and 10¹¹ organisms/L, the medium in the upper chamber of the Transwell unit was withdrawn; then 10 µL of the bacterial suspension was applied directly to the apical surface of the C2BBe cells and 190 µL fresh antibiotic-free medium was replaced over the infected cells. Equivalent numbers of the normally nonenteroinvasive E. coli DH5 and enteroinvasive Salmonella were used to inoculate the monolayers. E. coli were used to confirm integrity of the monolayers because this strain is noninvasive and will not pass through confluent monolayers unless they are damaged, and to monitor the development of leakiness resulting from exposure to Salmonella. In these experiments, six filters were assigned to each treatment group.

Immunofluorescent localization and confocal microscopy. C2BBe monolayers were grown for 3 wk on glass cover slips in control media and media with SP-A or SP-B soy protein hydrolysates. At 21 d, coverslips were fixed in paraformaldehyde (40 g/L) for 20 min and permeabilized with Triton in PBS (5g/L) containing 1 g/L CaCl₂ . After two washes with PBS/CaCl₂ the monolayers were incubated in propidium iodide (1 mg/L) for localization of karyorrhexic (segmented) nuclei. Monolayers were then washed three times and mounted on glass slides with DABCO mounting medium (Sigma). Slides were stored in UV-protectant containers under refrigeration and examined with a Bio-Rad (North Yorkshire, UK) dual channel laser scanning confocal microscope within 24 h.

RESULTS

Determination of cell number by MTT assay at 5 d showed that, although monolayer formation was inhibited by the addition of SP-A, the number of cells per well in SP-A-treated cultures did not differ from that in control media (Fig. 3). Cells treated with SP-B proliferated significantly more than those in control media (P  0.05), and this level of cell expansion did not differ from that seen in cultures treated with EGF.

Transepithelial electrical resistance, an index of monolayer integrity, determined in cells cultured on microporous membranes in Transwell units, was significantly reduced in monolayers treated with media to which SP-A had been added (Fig. 4). C2BBe cells grown on microporous filters were judged to be confluent at 16 d by visual examination with phase-contrast microscopy and shortening of the time between necessary feedings of the monolayers on filter units, compared with 21 d needed by cells grown in flasks. C2BBe 16-d cultures grown on microporous filters were equivalent to 21-d cultures grown in flasks for confluence and brush border enzyme assays. There was no appreciable enhancement of TER in cell cultures after 16 d (21, 28 and 35 d), and 16-d cultures were judged to be fully mature (data not shown). No inhibition or enhancement of monolayer integrity (TER) was seen by the addition of SP-B (Fig. 4).

The addition of soy protein hydrolysates SP-A and SP-B to the culture media of C2BBe cells enhanced the development of brush border enzymes, as assayed by levels of AP (Fig. 5A) and SI (Fig. 5B). Detectable levels of both of these enzymes appeared at 10 d and reached maximum levels at 28 d following seeding of the flasks. Thereafter, enzyme levels declined by 35 d as cultures aged (data not shown). Therefore, 21-28 d was designated as the optimal time of monolayer differentiation and maturation for monolayers grown in flasks. Although levels of AP in SP-supplemented cultures were significantly enhanced by 10 d, on through 28 d, SI levels were not significantly enhanced until 28 d (Fig. 5B).

Figure 6A, B reports the results of inoculation of C2BBe monolayers grown on microporous filters with two species of Salmonella at 21 d of maturation. Both S. typhimurium and S. typhi crossed the C2BBe monolayers in substantial numbers over a 2-h period after inoculation. Nonenteroinvasive E. coli DH5 penetrated the C2BBe monolayers in similar numbers when applied simultaneously with Salmonella, but E.coli alone did not penetrate monolayers.

Addition of soy hydrolysates SP-A and SP-B to the culture media did not affect the number of S. typhimurium passing across the monolayers. S. typhi passage was significantly less in C2BBe monolayers cultured in either of the soy hydrolysates SP-A and SP-B.

There was a greater number of karyorrhexic nuclei in cells cultured in media for 21 d with the addition of soy hydrolysates SP-A and SP-B, compared with cells cultured in control media (Fig. 7). SP-A cultures contained ~3 abnormal cell nuclei per field compared with <1 in control cultures. SP-B cultured cells showed a greater increase (8/field) over levels seen in control cultures.

DISCUSSION

Although achievement of monolayer confluence was delayed at 10 d in cultures exposed to SP-A, eventually confluence was achieved by 21 d. The total number of viable cells in the culture was not affected by the presence of SP-A. EGF or SP-B enhanced the number of cells in culture as assessed by the MTT assay. It appears that cell division was not inhibited in media enriched with SP-A, only the ability of the cells to form confluent monolayers and develop tight junctional complexes, as evaluated by TER measurements.

TER, which is an indicator of the development and integrity of intercellular tight junctions, was inhibited at 16 d by the presence of SP-A but was unaffected by the presence of SP-B. It was expected that TER would be decreased in SP-A cultures because the electrical resistance develops after confluence is achieved and the cell junctional complexes mature. However, even after confluence was achieved at 16 d in monolayers cultured on microporous filters in the presence of SP-A, TER levels did not increase thereafter at 21 and 28 d. Similarly, TER measurements for control and SP-B cultures did not show changes after 16 d. Maximum maturation potential was therefore achieved by 16 d in the monolayer-bearing filters.

It is interesting that the functional development of cellular brush borders, marked by the enzymes alkaline phosphatase and isomaltase, was enhanced by the addition of the soy protein hydrolysates. The later development and enhancement of sucrase/isomaltase in the presence of SP (at 28 d) compared with alkaline phosphatase levels was not unexpected because this enzyme activity develops much later in normal cultures (Jumarie and Malo 1991). The formation of mature enzymes at a much later stage in cultures enriched with SP-A is consistent with the delay in the formation of TER.

SP-A has been investigated in preliminary in vivo studies in animal models (Cope et al. 1992, Funk-Ashuleta et al. 1995a and 1995b, Hennebold et al. 1995) and clinical trials in patient populations (Cope et al. 1996). SP-A has been found to provide complete protection against radiation toxicity in mice (Cope et al. 1992, Hennebold et al. 1995) and against methotrexate toxicity in mice (Funk-Ashuleta et al. 1995a) and rats (Funk-Ashuleta et al. 1995b), assessed by decreased weight loss and absence of diarrhea. Histology of the ileum of mice fed soy protein after irradiation showed less crypt necrosis. In canine and murine models, SP-A was shown to suppress the Alzheimer gene in intestinal crypt cells, produce a reduction in mitotic index and eliminate apoptotic cell death in apical intestinal cells (Cope et al. 1992). Preliminary clinical tests of this peptide in bone marrow transplant patients suggested that SP-A inhibits mucositis or apoptotic cell death in the esophagus and gastrointestinal tract (Cope et al. 1996). These findings suggest that SP-A is antiapoptotic.

The present in vitro studies showed a mild increase in degenerate nuclei, suggestive of late-stage apoptotis, in monolayers cultured in SP-A (3/field) and a somewhat larger increase in apoptotic nuclei in SP-B cultures (8/field). The difference in the degree of apoptosis found between in vitro and in vivo studies of SP-A may be due to the interaction of hydrolyzed soy protein components with gastrointestinal enzymes and intrinsic regulators of cell longevity in the whole-animal model that are not present in single cell culture. Another explanation of the differences between animal models and this cell model might be the adenocarcinoma origin of the C2BBe cells. An increase in apoptosis is characteristic of regressing tumor cells (Symonds et al. 1994) and C2BBe cells are tumor cells. A possible increase in programmed cell death within the culture cell population may contribute to the retarded development of mature monolayers and points to an inhibitory effect of SP-B on cancer cells. The results of animal studies may reflect the dynamics of apoptosis or apoptotic cells within a normal cell population or reflect differing levels of apoptosis at variable sites in the gastrointestinal tract.

There are many studies on the intracellular effects of purified soy-derived compounds, chiefly BBI, a 8000-kDa protein, and the isoflavinoid, genistein. BBI has been shown to suppress cancers experimentally induced in animals by X-irradiation and chemical carcinogens (Billings et al. 1991, Kennedy and Little 1978, St. Clair and St. Clair 1991, St. Clair et al. 1990, Weed et al. 1995). Parallel in vitro studies using the intestinal epithelial cell line IEC-17 have shown that BBI can suppress malignant transformation by inhibiting intracellular proteases as BBI is internalized by the colonic cells (Richards et al. 1989). These studies taken together demonstrate that BBI and other soybean-derived protease inhibitors have intracellular actions that function in the cell growth cycle and may have an intervening function in carcinogenesis.

Proteases may play a role in carcinogenesis by their effects on collagen, thereby promoting altered interactions with the cell matrix and invasiveness of tumor cells. Protease inhibitors capable of inhibiting collagenase or blocking the protease cascade that leads to its activation have been shown to block metastatic progress (Roughley et al. 1978). It is possible that the hydrolysate SP-A contains components that act on cell adhesion molecules and influence cell-matrix interactions, which would affect confluence in the cultures evaluated in these studies.

The isoflavone, genistein, is known to inhibit tyrosine protein kinases, which are key elements in the carcinogenic process. In addition, genistein inhibits several other enzymes, specifically cAMP-dependent protein kinase, phosphorylase kinase and Ca²⁺/phospholipid-dependent enzyme protein kinase involved in signal transduction events that regulate cell growth (Aikyama et al. 1987). Furthermore, genistein down-regulated the phosphorylation of the EGF receptors in cultured A431 cells derived from a human epidermal carcinoma (Aikyama et al. 1987). The phytoestrogens also influence sex hormones, and in particular, they may have an antiestrogenic effect. They also inhibit intracellular enzymes, protein synthesis, growth factor action, malignant cell proliferation, differentiation and angiogenesis (Herman et al. 1995). They have an effect on a wide variety of tumor cell lines including human breast and prostate, leukemia, neuroblastomas and melanomas. In spite of all of the information accumulated on the activity of genistein, the mechanism of its action remains unknown (Barnes et al. 1990).

SP-A and SP-B have not been analyzed for BBI protein, but these hydrolysates, especially SP-A, are highly processed, and this component is likely to be completely extracted. The genistein content of both hydrolysates was high, and this may contribute to the activities observed in the C2BBe culture. It is possible that there are other factors in the soy hydrolysates that contributed to their effects on proliferation and differentiation described here. Unlike its companion hydrolysate SP-A, SP-B has not been extensively studied in animal models and clinical trials. This work is the first test of SP-B in a biological system. Further studies to isolate the active factors in SP-A and SP-B and to compare in vitro with in vivo effects have to be undertaken.

In addition to the antiproliferative effect of SP-A and SP-B, bacterial transmigration studies using the nontyphoid Salmonella, S. typhimurium, and the typhoid strain, S. typhi, demonstrated that soy supplements in culture media impart a resistance to certain enteropathogenic bacteria by gastrointestinal cells. This laboratory has demonstrated that S. typhimurium and S. typhi have different mechanisms of invasion of cultured gastrointestinal cell monolayers (Kops et al. 1996). Different invasion mechanisms might explain the species-specific effect of soy hydrolysate supplementation on bacterial passage through C2BBe monolayers. The presence of higher levels of brush border enzymes in cells cultured in SP-A and SP-B compared with control cultures in these and other investigations (Jumarie and Malo 1991) may indicate a difference in the cell interface where interaction with enteroinvasive bacteria occurs. Brush border enzyme differentiation might be an important factor in inoculation with S. typhi and may not be as important in exposure to S. typhimurium.

These studies show the feasibility of testing semipurified soy-based derivatives in a cultured cell system at minimal cost. They demonstrate the capability of soy hydrolysates to have an inhibitory effect on cancer cells and a potential to enhance the resistance of the mucosal barrier of the intestinal tract to invasion by pathogens.

ACKNOWLEDGMENTS