The animal care and use procedures were approved by the Animal Care Committee of Osaka Medical Center for Cancer and Cardiovascular Diseases. The generation of KO mice has been described previously[26]. We mated wild-type (WT) littermate mice produced by backcrossing to the C57BL/6J strain for five generations and used their offspring as WT controls in this study. Mice were maintained on a 12-h light/dark cycle with free access to drinking water and a standard diet. We injected 10-wk-old male mice with AOM (Sigma Chemical Co., St. Louis, MO) at a dose of 7.5 mg/kg body weight intraperitoneally once a week for 8 wk, and control mice received equal volume of saline injection (WT + saline, n = 9; WT + AOM, n = 23; KO + saline, n = 13; and KO + AOM, n = 24). The mice were sacrificed 53 wk after the first AOM injection, and the colons and small intestines were removed immediately. The harvested specimens were opened longitudinally, and the number and size of tumors were recorded. Using calipers, we measured the length (L), width (W), and depth (D) of each intestinal tumor and calculated the tumor volume using the formula V = L×W×D×π/6, as described previously[27]. We also noted the development of liver tumors.

Tumors were fixed in 10% buffered formalin, embedded in paraffin blocks, and sectioned at 3.0-μm thickness. Some sections of the colon tumors were subjected to hematoxylin and eosin (HE) staining for histopathology, and others were used for immunohistochemistry. proliferating cell nuclear antigen (PCNA) was visualized by staining with rat anti-mouse PCNA monoclonal antibody (Dakocytomation, Glostrup, Denmark). To determine the PCNA labeling index, we selected five representative PCNA-positive fields in each section, counted more than 200 tumor cells in each field, and then calculated the percentage of PCNA-positive cells. cyclooxygenase-2 (COX-2) was visualized by staining with rabbit anti-mouse COX-2 polyclonal antibody (Cayman, Ann Arbor, MI). We observed immunoreactive COX-2 expression in the periluminal stromal cells and epithelium of the colorectal tumors. Therefore, we evaluated COX-2 expression; both the intensity of immunoreactivity and the percentages of positively-stained areas in relation to the circumference of the tumor. We graded COX-2 expression of each immunostained section on a 0 to 4+ scale; no immunoreactivity (0), weak immunoreactivity and 1% to 25% positive regions (1+), mild immunoreactivity and 26% to 50% positive regions (2+), moderate immunoreactivity and 51% to 75% positive regions (3+), strong immunoreactivity and 76% to 100% positive regions (4+). We regarded a case displaying very weak immunoreactivity, or less than 1% positive regions, as negative. Sections of the liver tumors were subjected to HE staining for histopathology.

Results are expressed as mean ± SE. Statistical analyses of data were performed using the Student’s t-test, the chi-square test, Fisher’s exact probability test, Wilcoxon rank sum test, or the Spearman’s rank correlation. Statistical significance was defined as P < 0.05.

We treated WT and KO mice with AOM at a dose of 7.5 mg/kg or with saline vehicle once a week for 8 wk. At 53 wk after the first AOM treatment, the mice were sacrificed to evaluate the development of colorectal tumors. Body weight changes did not differ between WT and KO mice after AOM treatment, although AOM treatment resulted in a significant reduction in body weight gain of both WT and KO mice (16.4% vs 14.4% reduction at sacrifice, Figure 1A). We observed no colorectal tumors in either WT or KO mice treated with only saline (data not shown). Figure 1B shows representative pictures of the colorectal tumors arising in WT and KO mice treated with AOM. There was no difference in the incidence of colorectal tumor formation between WT and KO mice (39.1% vs 37.5%, Figure 1C). However, the number of colorectal tumors per mouse was significantly greater in KO mice (2.33 ± 0.47, range: 1-4) than in WT mice (1.11 ± 0.11, 1-2) (P < 0.05; Figure 1D), and the average volume of colorectal tumors was markedly larger in KO mice (79.2 ± 29.2 mm³) than in WT mice (9.02 ± 1.46 mm³) (P < 0.05; Figure 1E). Histological analysis revealed that KO mice developed more progressive tumors in the colon than WT mice (Figure 2). We observed one adenoma (1 of 9, 11%), three carcinomas in situ, as assessed by the findings of high-grade dysplasia (3 of 9, 33%) and five adenocarcinomas (5 of 9, 56%) of the colorectal tumors arising in WT mice (Figure 2A). In contrast, of the colorectal tumors in KO mice examined, all tumors (14 of 14) were classified as adenocarcinomas (2 well- and 12 moderately-differentiated adenocarcinomas) (Figure 2B). These findings suggest that adiponectin deficiency enhances AOM-induced colorectal carcinogenesis. One KO mouse developed a small intestinal tumor, classified as a moderately-differentiated adenocarcinoma, in the duodenum (data not shown), whereas none of WT mice developed small intestinal tumors.

To characterize the influence of adiponectin deficiency on colorectal tumor growth, we evaluated cell proliferation of colorectal tumor cells using the PCNA labeling index assessed by immunohistochemistry. Figure 3A shows representative sections of PCNA-labeled nuclei in colorectal tumors from WT and KO mice. Expression of PCNA was identified by cell nuclei that stained brown to PCNA. For quantitative analysis, we determined the PCNA labeling index by calculating the percentage of PCNA-positive cells. The PCNA-labeling index of KO mice (68.5% ± 2.3%) was greater than that of WT mice (44.8% ± 6.3%) (P < 0.01; Figure 3B), suggesting that adiponectin deficiency increased cell proliferation of colorectal tumor cells. To investigate the mechanisms by which adiponectin deficiency increased cell proliferation, we first examined the expression of cyclin-dependent kinase inhibitors (CDKIs), p21^CIP and p27^KIP, in colorectal tumors by immunohistochemistry. However, WT and KO mice showed absence or only focal positivity of both proteins in the colorectal tumors, and there was no significant difference (data not shown).

Next, we examined the expression of COX-2, a well-established pathogenic factor in colorectal carcinogenesis[28,29], by immunohistochemistry. Figure 4 depicts representative sections stained for COX-2. We found no COX-2 positive staining in normal colonic tissues of WT and KO mice. In the colorectal tumor tissues, we observed immunoreactive COX-2 expression in the epithelium and periluminal stromal cells, predominantly in the myofibroblasts. We detected very weak COX-2 staining in 22% (2 of 9) and only local COX-2 staining in 56% (5 of 9) of the colorectal tumors in WT mice (Figure 4A), whereas colorectal tumors in KO mice showed higher levels of COX-2 staining (Figure 4B). In 36% (4 of 11) of the colorectal tumors in KO mice, we detected marked COX-2 staining almost all over the surface of the tumor. Statistical analysis revealed that the expression of COX-2 in colorectal tumors of KO mice was higher than in WT mice (Figure 4C, P < 0.05; Wilcoxon rank sum test). Moreover, we found a significant correlation (r = 0.89, P < 0.001; Spearman’s rank correlation) between the PCNA-labeling index and COX-2 expression level (Figure 4D). These results suggest that adiponectin deficiency may promote proliferation of colorectal tumor cells, which may be associated with overexpression of COX-2 in the periluminal stromal cells.

AOM is primarily metabolized by the liver and induces DNA damage in both the colon and the liver[30]. While AOM can induce tumor formation in the liver[31,32], tumors are reported to be formed almost exclusively in the colon[30]. Consistent with the previous observation[32], the incidence of liver tumor formation after AOM treatment was only 13% (3 of 23) in WT mice (Figure 5A), and all three WT mice had only a single tumor (Figure 5B). In contrast, the incidence of liver tumor formation was 50% (12 of 24) in KO mice (Figure 5A); 5 mice with a single tumor (42%), 2 mice with two tumors (16%), and 5 mice with more than three tumors (42%) (Figure 5B). Figure 5C and D show representative HE-stained sections of the liver tumors in KO mice. The majority of the liver tumors were identified as hepatocellular neoplastic nodules (Figure 5C), and one was a hepatocellular carcinoma (Figure 5D). We also observed that WT mice developed 2 hepatocellular neoplastic nodules and one hepatocellular carcinoma (data not shown). These results suggest that adiponectin deficiency may promote AOM-induced liver tumor formation.

Adiponectin has attracted much attention because of its potential role in the development and progression of obesity-related malignancies. It has been reported that hypoadiponectinemia is an independent risk factor for colorectal cancer and its precursory adenoma.

The anti-proliferative effect of adiponectin on several types of cancer cells has been well documented in cultured cells and/or xenograft models. The molecular pathways regulated by AMP-activated protein kinase (AMPK) have been reported to be the potential mechanisms through which adiponectin regulate carcinogenesis.

This study represented the causal relationship between hypoadiponectinemia and colorectal carcinogenesis in an in vivo model. Adiponectin deficient (KO) mice developed more advanced colorectal tumors, and showed enhanced liver tumor formation after azoxymethane (AOM) treatment.

This study demonstrated overexpression of cyclooxygenase-2 (COX-2) in the stromal myofiroblasts of the colorectal tumors in knockout (KO) mice. COX-2 overexpression might be a potential mechanism through which adiponectin deficiency was involved in the colorectal carcinogenesis.

Adiponectin is one of the major adipocytokines, and its plasma levels are reduced in obesity, type-2 diabetes, and coronary artery disease.

The present paper shows highest growth and progression of AOM-induced colon cancer in KO mice, with respect to wild type (WT) control mice. The results are of certain interest. Nevertheless, some points need to be further clarified and considered in depth to increase the interest of the paper.