Promotion of intestinal tumor formation by inulin is associated with an accumulation of cytosolic β-catenin in Min mice

Authors

  • Anne-Maria Pajari,

    1. Department of Applied Chemistry and Microbiology, Division of Nutrition, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Johanna Rajakangas,

    1. Department of Applied Chemistry and Microbiology, Division of Nutrition, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Essi Päivärinta,

    1. Department of Applied Chemistry and Microbiology, Division of Nutrition, University of Helsinki, Helsinki, Finland
    Search for more papers by this author
  • Veli-Matti Kosma,

    1. Department of Pathology and Forensic Medicine, University of Kuopio and Kuopio University Hospital, Kuopio, Finland
    2. Department of Pathology, Tampere University Hospital and University of Tampere, Tampere, Finland
    Search for more papers by this author
  • Joseph Rafter,

    1. Department of Medical Nutrition, Karolinska Institutet, Huddinge, Sweden
    Search for more papers by this author
  • Marja Mutanen

    Corresponding author
    1. Department of Applied Chemistry and Microbiology, Division of Nutrition, University of Helsinki, Helsinki, Finland
    • Department of Applied Chemistry and Microbiology, Division of Nutrition, P.O. Box 27, FIN-00014 University of Helsinki, Finland
    Search for more papers by this author
    • Fax: +358-9-19158269


Abstract

Inulin, polydisperse β (2-1) fructan, has been suggested to protect against colon carcinogenesis and is currently used in a number of food applications. However, the data regarding the role of inulin in intestinal carcinogenesis remains controversial since the results of our previous study suggested that inulin promotes intestinal tumor formation in Min mice, an animal model for intestinal cancer with a mutation in the Apc tumor suppressor gene (Carcinogenesis 2000;21:1167–73). In our present study, we further examined the effects of inulin on intestinal tumor formation in Min mice by carefully analyzing β-catenin expression and cellular localization at 3 different time points during the tumorigenic process. Min mice were fed a high-fat inulin-enriched (10% w/w) diet or the high-fat diet without any added fiber from the age of 6 weeks to the ages of 9, 12 or 15 weeks. The results showed that inulin significantly increased the number (by 20%) and especially the size (by 44%) of adenomas in the small intestine. At week 15, the promotion of tumor development was accompanied by an accumulation of cytosolic β-catenin in the adenoma tissue. In the normal appearing mucosa, levels of membrane β-catenin and PCNA were reduced in the inulin-fed mice, possibly indicating impaired enterocyte migration. These data do not support the earlier suggestions on the cancer preventive effects of inulin and emphasize the need for further research and evaluation where health claims for inulin are concerned. © 2003 Wiley-Liss, Inc.

Colon cancer is known to progress from normal tissue to adenoma and carcinoma through accumulation of genetic alterations.1 Mutations in the adenomatous polyposis coli (APC) tumor suppressor gene occur very early and are found in a majority of sporadic colorectal tumors2 as well as in familial adenomatous polyposis (FAP), a hereditary form of colon cancer. The main tumor suppressing activity of APC protein is its ability to downregulate cellular β-catenin.

β-catenin is a multifunctional protein involved in cell-cell adhesion and Wnt signaling.3, 4 In unstimulated cells, β-catenin is complexed with E-cadherin at the cell junctions, whereas the levels of β-catenin in the cytoplasm or nucleus are kept down by targeting excess β-catenin to APC-facilitated and proteasome-mediated degradation. Activation of the Wnt pathway or mutations in the APC or β-catenin result in stabilization of cytoplasmic β-catenin. The stabilized β-catenin then translocates into the nucleus where it interacts with the transcriptional factor Tcf/Lef to stimulate transcription of genes involved in cell growth, such as c-myc and cyclin D1.5, 6 Overexpression of β-catenin is oncogenic in the intestinal tract, resulting in, for example, dysplasia7 and formation of numerous adenomatous polyps in the small intestine and microadenomas in the colon.8 In a subset of colon cancer cases, a mutation in the β-catenin gene is the initiating event leading to tumor formation.9

β-catenin has been shown to participate in many key processes maintaining normal cell function and microarchitecture of the epithelia in the intestine. In a mouse model of colonic hyperplasia, increased cytosolic and nuclear expression of β-catenin was associated with hyperproliferation as well as with increased steady-state levels of c-myc and cyclin D1.10 In chimeric mice, overexpression of β-catenin stimulated cell division of undifferentiated cells and slowed cellular migration along the crypt-villus unit in the small intestine.11 In cell line studies, downregulation of β-catenin by antisense treatment led to inhibition of cell proliferation, anchorage-independent growth and cellular invasiveness.12 Furthermore, suppression of transcriptional activation of β-catenin promoted differentiation13 and restored cell polarity in colon cancer cell lines.14 These findings indicate that changes in β-catenin signaling may be important early events in colon carcinogenesis. Indeed, Yamada et al.15, 16 found β-catenin-accumulated crypts in the colon of carcinogen-treated rats, which they stated to be truly premalignant lesions for colon cancer. Similarly, Paulsen et al.17 identified lesions with accumulated β-catenin in the colon of Min mice.

The oncogenic nature of β-catenin demonstrates that prevention of its cellular accumulation could be a promising target for strategies in colon cancer prevention. Apart from the mutations in APC or β-catenin, cellular β-catenin pools also may be disturbed by epigenetic changes in either the Wnt or the integrin-linked kinase (ILK) signaling pathways.18, 19 These pathways are in connection with components of the extracellular matrix and thus could be regulated by environmental factors such as diet. Furthermore, β-catenin levels can be regulated by other major cellular pathways involved in carcinogenesis. Recent publications indicate an important role for p53 protein in β-catenin degradation.20 In cell lines, overexpression of β-catenin results in accumulation of p53,21 and Min mice deficient in p53 develop significantly more intestinal adenomas than Min mice with wild-type p53 gene.22 These observations indicate close interaction of these 2 pathways.

Because diet is known to modulate the risk for colon cancer, efforts have been made to identify dietary compounds with cancer preventive effects. Inulin, a chicory fructan, has been suggested to promote human health by favorably affecting intestinal bacteria23 and thereby also to prevent colon carcinogenesis.24, 25, 26 However, the results of our previous study suggested that inulin promotes intestinal tumor formation in Apc-mutated Min mice, a murine model of human FAP, which was accompanied by an increased cytosolic β-catenin in the intestine.27 Our present study was designed to further examine the tissue specificity (adenoma vs. normal appearing mucosa) and the timing of the changes in β-catenin expression in the small intestine of Min mice by immunoblotting in an effort to link these molecular changes with adenoma formation and responsiveness to dietary treatment. In addition to β-catenin, protein levels of p53 and proliferating cell nuclear antigen (PCNA) were analyzed in the mucosa and adenoma tissue. Our study strengthens our earlier findings that a diet enriched with inulin enhances tumor formation and growth in the intestine of Min mice and that this enhancement is accompanied by changes in β-catenin levels and cellular localization during the course of tumorigenesis.

MATERIAL AND METHODS

Animals and diets

The study protocol involving animals was approved by the Laboratory Animal Ethics Committee of the University of Helsinki. Male and female C57BL/6J Min mice were bred at the Laboratory Animal Centre of the University of Helsinki, Finland, from inbred mice originally obtained from the Jackson Laboratory (Bar Harbor, ME). Mice were genotyped by PCR assay on DNA isolated from tail to determine wild type or heterozygote for the Apc allele.28 The mice were given a standard rodent laboratory chow (Altromin, Ringsted, Denmark) until the age of 6 weeks when they were stratified by body mass and sex and assigned randomly to the experimental diets with 9–11 mice per group. Animals were housed in plastic cages in a temperature- and humidity-controlled animal facility, with 12 hr light/dark cycle. They had free access to the semisynthetic diets and tap water for the feeding periods of 3, 6 and 9 weeks. The body weights of the animals were recorded weekly.

A diet enriched with inulin at 10% (w/w) (polydisperse β (2-1) fructan, RaftilineHP®, Orafti, Tienen, Belgium) was used as an experimental diet. It was a modified AIN93-G diet29 containing 40-energy % of fat, which was a mixture of butter, rapeseed oil and sunflower seed oil providing the intake of saturated, monounsaturated and polyunsaturated fatty acid in a ratio close to 3:2:1. It corresponded to the intake of these fatty acids in an average Western-type diet. The control diet was a similar high-fat diet without any added fiber, providing similar amounts of protein, fat, vitamin and minerals on an energy basis (Table I). The diets were prepared weekly and kept at 4°C.

Table I. Composition of the Experimental Diets (G/KG)
IngredientNonfiberInulin
Casein236.2212.6
Dextrose479.0431.1
Butter148.9134.0
Sunflower-seed oil13.312.0
Rapeseed oil62.256.0
Mineral mix (AIN93G)41.637.4
Vitamin mix (AIN93GM)11.810.6
L-cystine3.63.2
Choline chloride3.63.2
Tertiary butylhydroxyquinone0.0140.014
Inulin100

Intestinal adenoma scoring and tissue samples

At the end of the feeding periods, the mice were killed by CO2 asphyxiation. The small intestine, cecum and colon were removed, opened along the longitudinal axis and rinsed with ice-cold saline. The small intestine was divided into 5 sections of equal length. The cecum and colon were kept together. The small intestine and colon and cecum were then spread flat on a microscope slide and the number, diameter and location of adenomas were determined with an inverse light microscope with a magnification of 3× by 2 observers blind to the dietary treatment. All the adenomas from each section of the intestine were excised, and adenomas from the 2 most distal parts of the small intestine were pooled to give a representative sample per mouse for immunoblot analyses. The corresponding normal appearing mucosa was then gently scraped off with a microscope slide. All the tissue samples were snap frozen in liquid nitrogen and stored at −80°C until analyzed.

SDS-PAGE and immunoblot analysis

Because subcellular localization of β-catenin is considered to be an important determinant of its function, β-catenin was analyzed separately from cytosolic, nuclear and membranous fractions of adenoma and normal appearing mucosa. PCNA and p53 were analyzed from the nuclear fraction. Tissue samples were homogenized on ice, and the cytosol and particulate fractions extracted as previously described27 with the addition of the extra centrifugation (1,500g for 10 min at 4°C) to obtain the nuclear fraction. Five milliliters of the crude tissue extracts were concentrated to 1/50 volume with Millipore Ultrafree®-4 tubes (Millipore, Bedford, MA). After protein concentration measurement (Bradford; Bio-Rad protein assay reagent, Hercules, CA), the homogenate was mixed with an equal volume of SDS sample buffer, boiled for 3 min and stored at −80°C until use. The purity of cellular fractions was controlled by determining the nuclear lamin B levels in the cellular fractions. Both the cytosol and the membrane were free of lamin B. Similarly, cyclooxygenase-2 was present in the membrane (and nuclear) fraction of the adenoma but was absent in the cytosol.

Five micrograms protein of adenoma samples and 20 μg of mucosa samples were used for β-catenin analysis, and 20 μg of both adenoma and mucosa for PCNA and p53 analyses. A constant amount of rat-brain or RAW cell homogenate was used to control inter-assay variation in immunoblotting analysis for β-catenin, PCNA and p53. Intestinal samples and rat brain homogenate (10 μg) or RAW cell homogenate (6 μg for PCNA and 13 μg for p53) were subjected to 10% SDS-PAGE and then transferred to nitrocellulose membranes (Hybond ECL membrane, Amersham Pharmacia Biotech, UK) at 100 V for 1 hr. For β-catenin and p53 analyses, the membranes were blocked with 3.5% nonfat soy flour in Tris-buffered saline containing 0.1% Tween (TBS-Tween) at 4°C overnight. Membranes were then incubated with rabbit polyclonal antibody for β-catenin or goat polyclonal for p53 (Santa Cruz Biotechnology, Santa Cruz, CA) for 2 hr and washed 3 times with TBS-Tween, after which membranes were incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit (β-catenin) or anti-goat secondary antibody (p53) (Santa Cruz Biotechnology). The bands were visualized by using the ECL detection system (Amersham Corp., Arlington Heights, IL) according to the manufacturer's instructions. For PCNA analysis, nitrocellulose membranes were blocked with 5% nonfat dry milk in 0.1% TBS-Tween for 1 hr at room temperature, incubated with HRP-conjugated mouse monoclonal antibody for PCNA for 2 hr, washed 3 times with TBS-Tween and bands were visualized in the same way as for β-catenin and p53. Blots were scanned and analyzed using a GS-710 Calibrated Imaging Densitometer and the Quantity One programme (Bio-Rad). Each sample was run twice, the duplicates being loaded on a different gel. The intensities of sample bands on different gels were controlled by loading on each gel a constant amount of rat brain homogenate or RAW cell homogenate in duplicates, which was used to normalize the sample intensities between the different gels. If the difference in results between different gels was more than 2-fold, the sample was analyzed once more. Results in duplicates are expressed as sample band intensity (optical density of the β-catenin/p53/PCNA band multiplied by band area) divided by rat brain or RAW band intensity.

Statistical analyses

The differences among the time points were analyzed by nonparametric Kruskal-Wallis and the differences between the groups by Mann-Whitney U-test (StatView, version 5.0.1, SAS Institute Inc., Cary, NC).

RESULTS

General observations

The mice grew well and there were no differences in final body weights between the dietary groups at any of the time points. Although the mice were checked daily, one mouse in the inulin group of 15 weeks was found dead 2 days before the scheduled end point. The autopsy revealed a very large tumor in the distal colon. Because of difficulties in counting adenomas reliably, this mouse was excluded from the data analyses. The male and female Min mice did not differ significantly in the number or diameter of adenomas and therefore the data from male and female mice are pooled in the results section.

Adenoma number

The majority of adenomas emerged between the weeks 6 and 9 and they were mostly located in the distal part of the small intestine (Table II), which here refers to the last 2 parts of the 5 described in Material and Methods. The mice fed the inulin diet had nearly reached the final number of adenomas already at week 9, whereas the adenoma number in the mice fed the nonfiber control diet continued to increase between the weeks 12 and 15 (p = 0.046). The inulin-fed mice had significantly (p < 0.05) more adenomas in the distal small intestine at the time points of 12 and 15 weeks as well as in the entire small intestine at week 15 than their counterparts fed the nonfiber diet (Table II). Colonic adenoma formation followed a similar pattern with respect to time except that the number of adenomas was only 1–2% of that in the small intestine (Table II). The dietary treatment had no significant effect on colonic adenoma formation, although adenoma incidence tended to be higher in the inulin mice than in the nonfiber control mice when all time points were included (p = 0.07 by Pearson χ2 test).

Table II. Adenoma Number in the Intestine of Min Mice Fed either a Nonfiber Control Diet or a Diet Enriched with Inulin (10% W/W) from the Age of 6 Weeks until the Ages of 9, 12 or 15 Weeks
Age of mice (weeks)DietNo. of miceDistal small intestine1Total small intestine1Colon1Adenoma incidence in the colon2Adenoma size in the colon (mm)
  • 1

    Adenoma number per mouse expressed as mean (SD).

  • 2

    Number of mice with colonic adenomas. P = 0.07 for the difference in incidence (12/28 vs. 18/27) between nonfiber and inulin-fed groups when data from all time points were included by Pearson χ2 test.

  • *

    P = 0.007

  • **

    P = 0.022

  • ***

    P = 0.043, in comparison to mice fed the nonfiber control diet at the same time point by Mann-Whitney test.

6Pellet92.44 (2.56)5.44 (4.88)0.000/6
9Nonfiber1127.4 (20.6)40.6 (29.8)0.73 (1.01)5/111.32 (0.38)
 Inulin941.7 (26.8)60.7 (38.1)0.56 (0.53)5/91.96 (0.73)
12Nonfiber919.9 (10.5)36.4 (18.6)0.78 (0.97)5/92.22 (0.78)
 Inulin938.9 (14.4)*61.0 (21.3)**1.33 (1.32)7/92.62 (1.29)
15Nonfiber830.6 (10.0)54.3 (18.0)0.63 (1.19)2/83.72 (0.54)
 Inulin943.8 (12.1)***66.8 (16.0)1.33 (1.66)6/92.54 (1.09)

Adenoma size

The mice fed the inulin diet had a considerable enhancement of adenoma growth in the small intestine between the weeks 12 and 15 when compared to the control mice (p = 0.004 at week 15). The difference in size was even more pronounced in the distal small intestine, where it could be seen as early as week 9 (p = 0.094; Fig. 1) and became significant at week 12 on (p = 0.004 at week 12 and p = 0.0005 at week 15). In the colon, adenoma size did not differ between the dietary groups (Table II).

Figure 1.

Adenoma size (mm), expressed as box plots, in the distal small intestine of Min mice fed either a nonfiber control diet or a diet enriched with inulin (10% w/w) from the age of 6 weeks until the ages of 9, 12 or 15 weeks. Until the age of 6 weeks, the mice were fed a standard rodent laboratory chow (Pellet). **p < 0.01, ***p < 0.001 in comparison to control mice by Mann-Whitney test.

β-Catenin

The adenoma and normal appearing mucosa samples of Min mice contained several β-catenin bands, whereas the rat brain homogenate that was used to control inter-assay variation had only one band (Fig. 2). The β-catenin band in the rat brain homogenate had a molecular weight somewhat higher, about 98 kDa, than that of 92 kDa reported for the full-length β-catenin in cell line studies.30 The main β-catenin band in the Min mice samples had a molecular weight of 92 (91–93) kDa and could be seen in almost all the adenoma samples and in a majority of normal appearing mucosa samples. In addition to this full-length band, there were 2 bands, sometimes seen as duplets, with molecular weights from 72–86 kDa. Most samples contained still another group of bands with molecular weights of 62–64 kDa or lower, which presumably represent truncated forms of β-catenin.30 The specificity of the β-catenin bands was ensured by using 2 other commercially available antibodies (Fig. 3).

Figure 2.

Representative immunoblots of β-catenin protein in intestinal tissue samples of Min mice. Lane 1 contains molecular weight marker and lanes 2 and 3 contain rat brain homogenate (10 μg), which was used to control inter-assay variation. Each of the following lanes contains cytosolic sample of adenoma tissue (5 μg protein) from one mouse at week 15: control mice in lanes 4–8 and inulin mice in lanes 9–14. Corresponding actin levels are shown below the β-catenin bands.

Figure 3.

An immunoprecipitation experiment to show the specificity of β-catenin bands. (a) Rat brain homogenate (lane 1) and an intestinal tissue sample of Min mouse (50 μg protein, lane 2) was immunoprecipitated with either mouse monoclonal β-catenin antibody from Santa Crutz (Sc7963) or rabbit polyclonal β-catenin antibody from Santa Cruz (Sc7199, lanes 3 and 4). The immunoprecipitates were then loaded on SDS-PAGE gel and immunoblotted with the mouse monoclonal β-catenin antibody from Transduction Laboratories (C19220). (b) Samples were immunoprecipitated with either mouse monoclonal β-catenin antibody from Santa Cruz (Sc7963) or monoclonal β-catenin antibody from Transduction Laboratories (C19220), and detected by rabbit polyclonal β-catenin antibody from Santa Cruz (Sc7199). (c) Samples were immunoprecipitated with either rabbit polyclonal β-catenin antibody from Santa Cruz (Sc7199) or mouse monoclonal β-catenin antibody from Transduction Laboratories (C19220), and detected by mouse monoclonal β-catenin antibody from Santa Cruz (Sc7963).

For the results, we analyzed separately the intensity for the full-length band, for all the bands and for the bands lower than 64 kDa. The β-catenin results expressed either as the intensity of the full-length band or as a sum of all bands together appeared to be similar. We present β-catenin results as the sum of all the bands since the bands proved to be specific for β-catenin. In addition, the results for the low weight forms (less than 64 kDa) are presented separately.

β-Catenin in the adenoma tissue of the distal small intestine increased in all cellular fractions from week 9 to week 12 (p < 0.01; Fig. 4), and the cytosolic β-catenin continued to increase from week 12 to 15 (p = 0.007). The mice fed the inulin diet had a significantly (p = 0.021) higher β-catenin level in the cytosolic fraction at week 15 than the mice fed the nonfiber diet. Adenoma number in the distal small intestine was correlated with nuclear β-catenin at week 9 (r = 0.46, p < 0.05), whereas adenoma size was correlated with cytosolic β-catenin at week 15 (r = 0.51, p < 0.05). Inulin feeding also resulted in an increase in β-catenin bands lower than 64 kDa in the cytosolic fraction at week 15 (p = 0.002; Fig. 5), and these bands were significantly correlated with adenoma size in the distal small intestine at week 15 (r = 0.71, p = 0.005). The low molecular weight bands were increased in the membrane and nuclear fractions between the weeks 9 and 15 (p < 0.01; Fig. 5) but did not differ between the diets.

Figure 4.

β-catenin levels (mean ± SD, n = 8–11 per group) by immunoblotting in the adenoma tissue in the distal small intestine of Min mice fed either a nonfiber control diet or a diet enriched with inulin (10% w/w) from the age of 6 weeks until the ages of 9, 12 or 15 weeks. *p < 0.05 in comparison to control mice by Mann-Whitney test.

Figure 5.

Levels of low molecular weight β-catenin (mean ± SD, n = 8–11 per group) by immunoblotting in the adenoma tissue in the distal small intestine of Min mice fed either a nonfiber control diet or a diet enriched with inulin (10% w/w) from the age of 6 weeks until the ages of 9, 12 or 15 weeks. **p < 0.01 in comparison to control mice by Mann-Whitney test.

The mice fed the inulin diet had a decrease in the membrane β-catenin in the normal appearing mucosa from week 12 on (p = 0.021; Fig. 6) and a decrease in the cytosol at week 15 (p = 0.037) compared to the mice fed the nonfiber diet. Age of Min mice had no significant effect on total β-catenin in the mucosa.

Figure 6.

β-catenin (mean ± SD, n = 8–11 per group) in the normal appearing mucosa in the distal small intestine of Min mice fed either a nonfiber control diet or a diet enriched with inulin (10% w/w) from the age of 6 weeks until the ages of 9, 12 or 15 weeks. *p < 0.05 in comparison to control mice by Mann-Whitney test.

PCNA

A representative immunoblot of PCNA is shown in Figure 7. No effect of diet on PCNA could be seen in the adenoma tissue (Table III). In the normal appearing mucosa, the mice fed inulin had significantly lower PCNA levels at week 9 (p = 0.026) as well as at week 15 (p = 0.006) than mice fed the nonfiber diet. There was no effect of time on mucosal PCNA.

Figure 7.

Representative immunoblots on p53 and PCNA in adenoma samples of inulin-fed mice at week 15.

Table III. Expression of PCNA and p53 by Immunoblotting in the Normal Appearing Mucosa and Adenoma Tissue in the Distal Small Intestine of Min Mice Fed a Nonfiber Control Diet or a Diet Enriched with Inulin (10% W/W) from the Age of 6 Weeks until the Ages of 9 or 15 Weeks
 Normal appearing mucosaAdenoma
9 weeks15 weeks9 weeks15 weeks
  • 1

    Intensity values [mean (SD), n = 8–11 per dietary group] are normalized to a constant amount of RAW cell homogenate.

  • *

    P = 0.026 and

  • **

    P = 0.006, in comparison to mice fed the nonfiber control diet at the same time point by Mann-Whitney test.

PCNA1    
 Nonfiber diet0.44 (0.36)0.37 (0.15)0.47 (0.29)2.55 (2.26)
 Inulin diet0.17 (0.13)*0.11 (0.11)**0.27 (0.18)2.04 (1.26)
p531    
 Nonfiber diet1.24 (0.89)2.26 (0.89)0.68 (0.48)2.07 (0.67)
 Inulin diet1.09 (1.22)1.76 (1.45)0.82 (0.45)1.35 (0.88)

p53

Since recent publications indicate an important role for p53 protein in regulation of β-catenin degradation,20 we were interested to see whether p53 could explain some of the changes of β-catenin in the intestine of Min mice. p53 was analyzed from the adenoma tissue at week 9 and 15 (a representative immunoblot on p53 is shown in Fig. 7), and from the normal appearing mucosa at weeks 6, 9 and 15. In the adenoma tissue, p53 was increased from week 9 to week 15 in both diet groups (p = 0.001; Table III). The mice fed inulin had a trend for lower levels of p53 at week 15 compared to the mice fed the nonfiber diet (p = 0.068). p53 was significantly correlated with the nuclear β-catenin in the adenoma tissue at week 15 (r = 0.72, p = 0.004).

DISCUSSION

In our present study, we have demonstrated that Min mice fed the diet enriched with inulin [β (2-1) fructan, 10% w/w] developed significantly more and larger tumors in the distal small intestine than their counterparts fed the nonfiber control diet. The promoting effect of inulin on tumor formation and growth was consistent over 3 feeding periods of varying length (until the ages of 9, 12 and 15 weeks) and was accompanied by changes in β-catenin levels and cellular localization during the course of tumorigenesis. These results are in accordance and extend the data of our previous study in which feeding inulin at 2.5% (w/w) for a period of 6 to 8 weeks promoted tumor formation and increased cytosolic β-catenin in the intestine of Min mice.27

In recent years, inulin has been suggested to have positive effects on human health, including a possible reduction in risk of colon cancer.23 To support this view, there are some preliminary studies in which inulin was shown to inhibit aberrant crypt foci (ACF) formation in the colon of carcinogen-treated rats.24, 25, 26 Although widely used, total number of ACF does not necessarily predict the actual tumor outcome in the colon as has been shown, for example, with certain fibers31 and fish oil.32 Furthermore, only β-catenin accumulated crypts, whether independent of or a subgroup of traditional ACF, are suggested to be truly premalignant lesions for colon cancer.15, 16, 33 In line with this suggestion, a recent study shows that more than 50% of dysplastic human ACF have increased expression of β-catenin.34 Apart from our studies, the effect of inulin as such has not been studied in Min mice previously. In a study by Pierre et al.35 short-chain fructo-oligosaccharides had no effect on tumor formation in the small intestine but reduced tumor incidence in the colon of Min mice. Our study does not confirm the colonic effects either and in fact demonstrates an opposite trend in colonic tumor formation. Pierre et al.35 suggested that their results were due to the bifidogenic effect of fructo-oligosaccharides on intestinal bacteria. The difference in comparison to our results may be explained by the different chain lengths of the fructo-oligosaccharides used as well as dietary context, i.e., fat content and composition as well as the presence of cellulose in the diet, which are likely to affect the fermentation in the intestine. It is noteworthy that in our study the mice fed inulin had a stimulation of bifidobacteria in the total bacterial community in the cecum but the major changes were within previously unknown bacterial taxa.36 Not only do our data not support the colon cancer preventive effects of inulin, they clearly show that this compound has a promoting effect on tumor formation and growth in the small intestine of Min mice. These observations raise concerns that must be addressed before any health claims for inulin are authorized.

In our previous study, we found that the Min mice fed inulin had an increased level of β-catenin in the cytosolic fraction of samples containing both normal appearing mucosa and adenoma tissue by immunoblotting.27 In our present study, we further examined the tissue specificity (adenoma vs. mucosa) and the timing of the changes in β-catenin expression in the small intestine of Min mice in an effort to link these molecular changes with adenoma formation and responsiveness to dietary treatment. The results show that the increase in cytoplasmic β-catenin by inulin feeding is specific for the adenoma tissue. Although the promotive effect of inulin on tumor growth was obvious at week 12, as assessed by adenoma size, the difference in cytosolic β-catenin in the adenoma tissue between the dietary groups was not significant until week 15. Thus, it seems that accumulation of cytosolic β-catenin is a consequence of promotion and might not have much predictive value in regard to dietary effects on adenoma formation. However, cytosolic but not nuclear or membranous β-catenin was correlated with adenoma size at week 15, suggesting that cytosolic β-catenin could be related to adenoma growth in Min mice. In this regard, it is interesting that cytoplasmic expression of β-catenin was the most frequent abnormality in human ACF, adenomas, and carcinomas.34

It is not clear what is the exact role of cytosolic and nuclear β-catenin accumulation in intestinal tumor development. As assessed by immunohistochemical studies, both have been reported to occur more or less simultaneously in tumor tissue and in much the same way in human colon cancer patients (sporadic and inherited) as well as experimental animals.37, 38, 39, 40, 41 Herter et al.42 suggested, based on their analysis of β-catenin distribution in 60 sporadic colorectal tumors at different stages, that nuclear accumulation of β-catenin is the very first event of tumor development, followed by a substantial increase of β-catenin in the entire cytosol. Our data support this view since the nuclear β-catenin level at week 9 was significantly correlated with adenoma number, supporting the oncogenic nature of nuclear accumulation of β-catenin.43 Interestingly, the 2 mice in the inulin group having more than 100 adenomas in the intestine also had high nuclear β-catenin levels. Because not only the absolute level but also phosphorylation status of β-catenin may be an important contributor to β-catenin's ability to mediate Wnt signals,44 we also analysed nuclear β-catenin in the adenoma tissue with an antibody specific for the dephosphorylated β-catenin (kindly provided by Dr. Clevers, University of Utrecht, the Netherlands). However, no differences in dephosphorylated β-catenin between the dietary groups at any time points could be seen (data not shown). In general, the levels of dephosphorylated β-catenin paralleled those analyzed with the conventional β-catenin antibody.

There also was an enhanced accumulation of truncated forms of cytosolic β-catenin over time in the adenoma tissue of mice fed the inulin diet. These truncated forms showed a strong correlation with adenoma size at week 15 (r = 0.71, p = 0.005). The physiologic function of these β-catenin forms is not yet clear. Hugh et al.40 reported that a subgroup of metastatic carcinomas from colon cancer patients contained a truncated form of β-catenin that binds to E-cadherin only weakly. However, the approximate molecular weight of these truncated forms (80 kDa) was higher than in our study (<70 kDa). A number of cell line studies have associated the β-catenin bands of molecular weight about 65–75 kDa with apoptosis-induced and caspase-dependent cleavage of β-catenin.30, 45, 46 The cleavage products were enriched in the cytoplasm, had decreased transcription activation potential and possibly a reduced ability to bind to E-cadherin and β-catenin. Kim et al.47 showed that overexpression of β-catenin caused apoptosis in normal fibroblasts and tumor cells, and they suggested that apoptosis caused by β-catenin overexpression might be a physiologic mechanism to eliminate β-catenin overexpressing cells from the population. The apoptotic effect of β-catenin was not dependent on its nuclear localization and/or interaction with Lef.47 Thus, the increase of truncated as well as the total β-catenin in the cytosolic fraction of adenoma tissue might indicate more deregulated β-catenin processing in the inulin-fed mice. Whether this is related to tumor progression can only be speculated on.

Careful evaluation of normal appearing mucosa tissue was included to understand possible early changes in cellular signaling related to adenoma formation. Inulin feeding resulted in a reduction of membranous β-catenin in the mucosa of Min mice at week 12. Reduced levels of membranous β-catenin have previously been observed in intestinal tumor tissue where it has indicated tumor progression.40, 48 However, the recent study by Hao et al.34 reported a reduced membrane expression of β-catenin in human ACF, suggesting that it could be among the earliest alterations in intestinal tumorigenesis. The membranous β-catenin is thought to indicate the β-catenin pool associated with E-cadherin in the adherent junctions, which are responsible for cell-cell adhesion and required to maintain the architecture of the epithelia.49 Interaction of β-catenin with E-cadherin appears to be necessary for cell-cell adhesion50 and therefore reduction in membrane β-catenin may lead to impaired adhesion in the inulin-fed mice. It is well known that enterocyte crypt-villus migration is decreased in the intestinal epithelium of Min mice, which is a result of impaired cell proliferation and apoptosis.51, 52 PCNA protein levels in the normal appearing mucosa were significantly lower in Min mice fed the inulin diet at weeks 9 and 15, indicating that inulin further decreased the enterocyte migration rate. The PCNA result was confirmed by immunohistochemistry with 5 mice of each group (data not shown). Though PCNA levels were greatly elevated in the adenoma tissue of Min mice from week 9 to week 15, there was no effect of the dietary treatment. The results indicate that measurement of PCNA in adenoma tissue is far beyond the sensitive point to detect differences with dietary treatments.

The recent works of 2 research groups53, 54 demonstrated that p53 may have an important role in triggering β-catenin degradation as a part of a previously unknown pathway. This pathway involves Siah-1 and is independent of GSK3β and TrCP though it requires intact APC protein. It is well known that overexpression of oncogenes, such as β-catenin, results in accumulation of p53 in cell lines.21 Therefore a p53-dependent pathway for β-catenin degradation might also be important in situations in vivo. Indeed, Min mice homozygous for a null allele were shown to develop significantly more intestinal adenomas than those homozygous for the wild-type allele for p53.22 In our study, p53 levels in the adenoma tissue were increased over time and were significantly correlated with the nuclear β-catenin at week 15. These data are in line with the cell line studies that have demonstrated an autoregulatory loop in which excess β-catenin induces p53 activation, leading to downregulation of β-catenin levels and activity.55 It is interesting that in our present study the mice fed inulin had a trend for lower (p = 0.068) p53 levels in the adenoma tissue at week 15 than their counterparts fed the nonfiber diet.

In summary, our study demonstrates that inulin promotes intestinal tumor development in Min mice, which is accompanied by an accumulation of cytosolic β-catenin and a trend for a reduction of p53 in the adenoma tissue as well as reductions in membrane β-catenin and PCNA in the normal appearing mucosa. These data do not support the suggested cancer-preventive effects of inulin and demonstrate the importance of a thorough understanding of the underlying mechanisms of action for putative chemopreventive agents.

Acknowledgements

We thank Ms. S. Oikarinen for genotyping the mice and Ms. I. Harlin for breeding and taking care of the mice.

Ancillary