Introduction

Colorectal cancer, the third most common cancer in the world, is linked to several genetic and environmental factors1,2. Epidemiological studies have shown that the age-standardized incidence rate of colorectal cancer is higher in men than in women3. In addition, the correlation between colorectal cancer and obesity is more pronounced in men than in women4,5,6. Nevertheless, colorectal cancer is the second most common cancer in women aged > 65 years3 and older women have a relatively lower overall survival compared to age-matched men7, possibly owing to the loss of estrogen protection.5

Recently, the US National Institutes of Health has mandated the inclusion of sex as a biological variable while conducting biomedical research8. Biological sex plays an important role in the detection, prevention, and treatment of many diseases, including colorectal cancer9. In addition, fat distribution differs with sex, with women having less visceral and more subcutaneous adipose tissue than men10. As the accumulation of visceral adipose tissue increases the risk of colorectal cancer11,12, differences in incidence of colorectal cancer between men and women may be partly explained by differences in fat distribution. However, studies that reveal sex differences in the mechanism of colorectal cancer, particularly in relation to obesity, are lacking.

Sex differences in colorectal cancer have been mostly explained by differences in concentration of sex hormones. The Women Health Initiative study has revealed that hormone replacement therapy reduces the risk of colorectal cancer by 30% in postmenopausal women13,14. Estrogen treatment in mice with colitis-associated colorectal cancer has proven effective in reducing both the number and size of colon tumors15. The efficacy suggests that estrogen depletion plays a pivotal role in colorectal cancer development, particularly in postmenopausal women. Given that estrogen has the capacity to suppress excess fat accumulation, a factor positively correlated with the risk of colorectal cancer16,17, the rapid increase in colorectal cancer incidence in postmenopausal women may be attributed to estrogen depletion-induced fat accumulation. Recent insights propose that sex differences in cancer are genome-wide, involving genes located on both sex chromosome and autosomes18. Nevertheless, a comprehensive understanding of sex differences in the mechanism of obesity-associated colorectal cancer, particularly regarding chromosomal variances and estrogen-depletion-induced obesity, remains limited.

Chronic inflammation induced by obesity, alterations in adipokine concentration and gut dysbiosis have been suggested as causative factors in the occurrence of body fat-associated colorectal cancer19. The gut microbiota may influence the pathogenesis of obesity and related metabolic disorders, including colorectal cancer20,21,22. The gut microbiota can serve as a biomarker for early detection of colorectal cancer23. The proportion of Hafnia alvei belonging to the Proteobacteria phylum and that of Akkermansia muciniphila belonging to the Verrucomicrobia phylum increase in obesity and CRC patients24. In a recent study, AOM-treated APCMin/+ mice transplanted with feces from obese individuals exhibited increased colon tumor formation25. This observation implies that alterations in the gut microbial composition induced by obesity may impact the susceptibility to colorectal cancer. Sex differences exist in gut microbial composition26, and therefore, responses to high-fat diet (HFD) may vary by biological sex27. For example, the proportions of Lactobacillus, Alistipes, Lachnospiraceae, and Clostridium have been found to be increased by feeding HFD in the male group only28. In a separate investigation employing the AOM-DSS model for colorectal tumorigenesis, a reduction in microbial diversity and compositional changes were noted in female mouse compared to their male counterparts29.

This study aims to offer mechanistic insights into observed differences in the incidence of colorectal cancer among men, premenopausal women and postmenopausal women. We examined sex- and hormone-dependent variations by employing male (M), female (F) and ovariectomized (OVX)-F AOM-DSS colon tumorigenesis murine fed either LFD or HFD. Genes and gut bacteria composition associated with sex, diet and estrogen availability were identified. We utilized a high-fat diet-induced obesity model, widely acknowledged for its ability to closely emulate human obesity.

Results

Body and organ weights and number of tumors in experimental animals

The final body weight (g) was significantly higher (p < 0.05) in the male M + HFD group (36.42 ± 4.29) than in the M + LFD group (30.64 ± 0.81) and all F groups (Fig. 1A). The final body weight (g) for the female groups F + LFD, F + HFD, F(OVX) + LFD, and F(OVX) + HFD was as follows: 22.32 ± 1.52, 25.82 ± 2.61, 23.10 ± 1.45, and 29.98 ± 5.02, respectively. Differences between LFD and HFD groups within female groups were statistically significant (p < 0.05). The average daily food intake for the groups M + LFD, M + HFD, F + LFD, F + HFD, F(OVX) + LFD and F(OVX) + HFD was as follows: 7.66 ± 0.26, 15.01 ± 3.63, 6.17 ± 0.20, 6.14 ± 0.17, 5.80 ± 0.07 and 6.73 ± 0.07, respectively. The experimental diet, which was freely accessible, contained a high saturated fat content, enhancing the diet’s palatability30. Consequently, we observed an increased average food intake in the M + HFD group (p < 0.05) compared to other groups. However, no statistically significant differences were observed within the female groups. These results may be partially explained by the observed hyperphagic response in male mice when fed a high-fat diet, which was not observed in male mice fed a low-fat diet or female mice fed either low-fat or high fat diet31. Consequently, body weight gain in the M + HFD mice can be attributed to higher energy intake. However, an increased body weight gain in females fed a HFD cannot be solely explained by hyperphagia. The weights of total white adipose tissue (g) for the groups M + LFD, M + HFD, F + LFD, F + HFD, F(OVX) + LFD and F(OVX) + HFD was as follows: 4.28 ± 1.79, 10.18 ± 4.22, 3.22 ± 1.52, 7.42 ± 3.47, 5.29 ± 1.21, and 11.77 ± 5.02, respectively. HFD-fed groups had significantly higher amount of white adipose tissue that their counterparts fed LFD (p < 0.05). (Fig. 1B–D). Estrogen deprivation through ovariectomy significantly reduced the uterine weights of OVX female mice. The liver weight was significantly higher in the M + HFD group than those in the other groups. The number of colon tumors in the M + HFD group was significantly higher than that in all female groups, and the F + HFD group had a significantly lower number of tumors than the M + LFD and M + HFD groups (Table 1). Two-way ANOVA showed a significant interaction between sex and diet (p < 0.05), indicating that HFD may be a risk factor in the male group only (Fig. 2). However, the effect of estrogen depletion on the number of colon tumors was not significant.

Figure 1
figure 1

Final body weight and percentage of fat mass of experimental animals. C57BL/6J mice were fed LFD or HFD for 14 weeks. (A) Final body weight and the weights of adipose tissues including (B) visceral white adipose tissue (WAT), (C) subcutaneous WAT, and (D) whole WAT were measured at the end of the experiment. Results were analyzed using a one-way ANOVA and were expressed as mean ± standard deviation (SD). Bars labeled with different letters are significantly different based on Duncan's multiple range tests (p < 0.05).

Full size image
Table 1 Number of tumors in all experimental groups.
Full size table
Figure 2
figure 2

Effects of sex and diet and their interactions on the number of colon tumors. The number of colon tumors was measured at 75 days after AOM injection. Results were analyzed using two-way ANOVA and were expressed as mean ± standard deviation (SD). The p-values for main effects and their interactions are reported in box. Bars labeled with different letters are significantly different based on Duncan's multiple range tests (p < 0.05).

Full size image

Colonic MUC2 expression

Colonic MUC2 expression as a protective factor against epithelial damage was measured as an indicator of the integrity of mucosal barrier. Immunohistochemical staining indicated that MUC2 expression (%) for the groups M + LFD, M + HFD, F + LFD, F + HFD, F(OVX) + LFD and F(OVX) + HFD was as follows: 6.12 ± 1.25, 4.17 ± 3.03, 7.98 ± 1.30, 7.91 ± 0.98, 6.96 ± 1.47 and 7.09 ± 1.28. (Fig. 3A,B). Results indicated that the percentage of MUC2 positive area is significantly lower in the M + HFD group than in the female groups (p < 0.05). No significant difference was observed between the M + LFD and M + HFD groups.

Figure 3
figure 3

Immunohistochemical analysis of MUC2 expression. (A) Representative images of MUC2 expression in the colon (20 × magnification) and (B) percentage of MUC2 staining area. MUC2 staining is visualized by a brown color. Results were analyzed using one-way ANOVA and are expressed as mean ± standard deviation (SD). Bars labeled with different letters are significantly different based on Duncan's multiple range tests (p < 0.05).

Full size image

Sex differences in gene expression of HFD-modulated colon tissues

Microarray analysis of colon tissues was performed to determine differentially expressed genes (DEGs) in HFD-fed vs. LFD-fed groups in males, females or OVX females. The DEGs obtained from three pairwise comparisons were plotted in a Venn diagram (Fig. 4A,B) revealing that the number of HFD-modulated DEGs was 290 for males, 64 for females, and 137 for OVX females. Only one up-regulated gene (Gfra3) overlapped between females and OVX females, whereas 2 down-regulated genes (Thrsp and Gbp11) overlapped between males and OVX females. Therefore, HFD affected more genes in male mice than in female mice, which might be associated with tumor development.

Figure 4
figure 4

Venn diagram comparing (A) upregulated DEGs and (B) downregulated DEGs in colon between LFD- and HFD-fed groups in male, female, and OVX mice. Fold change values > 1.5 were considered upregulated whereas those < -1.5 were downregulated along with p-value threshold of 0.05 for statistically significant differences.

Full size image

Kyoto encyclopedia of Genes and Genome (KEGG) enrichment analysis of male-specific upregulated or downregulated DEGs

A total of 288 male-specific DEGs, including 183 upregulated and 105 downregulated genes, were subjected to KEGG analysis to identify the major pathways these genes are involved in. We aimed to identify genes and their functions that may accelerate tumor formation in the M + HFD group. The top 15 significant KEGG pathways related to male-specific upregulated DEGs included several cancer-related pathways, such as interaction between cytokines and cytokine receptors, hypoxia-inducible factor (HIF)-1 signaling pathway, and central carbon metabolism in cancer (Fig. 5A). Male-specific downregulated DEGs were mainly involved in viral infection-related signaling pathways such as infection by Epstein-Barr virus, herpes simplex virus 1, human T-cell leukemia virus 1, and human immunodeficiency virus 1 (Fig. 5B).

Figure 5
figure 5

(A) Top 15 enriched KEGG pathways of male-specific upregulated DEGs. (B) Top 10 enriched KEGG pathways of male-specific downregulated DEGs.

Full size image

Hub gene identification in male-specific DEGs using protein–protein interaction (PPI) network construction

PPI networks of the male-specific upregulated or downregulated DEGs were constructed based on the information obtained from the STRING database. The PPI network of upregulated DEGs included 139 nodes and 97 edges. Among them, 10 genes (Rhob, Hif1a, Met, Pkm, Eno1, Mapk6, Serpine1, Arhgap35, Junb, and Exoc5) with the highest degree scores were identified as male-specific upregulated hub genes by applying the cytoHubba plugin (Fig. 6A). The PPI network of downregulated DEGs included 86 nodes and 87 edges. The top 10 male-specific downregulated hub genes with higher degrees were filtered, including Irf7, B2m, H2-K1, Helz2, Ddx58, Ifi47, Cd3d, H2-Q7, Cd3g, and Lgals3bp (Fig. 6B).

Figure 6
figure 6

Top 10 hub genes among the (A) upregulated and (B) downregulated DEGs in male mice. The top 10 hub genes were screened by cytoHubba according to degree. Red color represents the highest degree, and yellow color represents the lowest degree.

Full size image

Sex-specific changes in the composition of gut microbiota in response to HFD

Effects of sex and diet, and their interactions on the composition of gut microbiota were determined by two-way ANOVA at each level of microbial taxonomy (Fig. 7). At the class level, diet had a significant effect on the Deltaproteobacteria [F(1,12) = 8.97, p = 0.011]. At the order level, diet had a significant effect on the relative abundance of Desulfovibrionales [F(1,12) = 8.97, p = 0.011]. At the family level, diet had a significant effect on the relative abundance of Desulfovibrionaceae [F(1,12) = 8.97, p = 0.011] and Odoribacteraceae [F(1,12) = 10.20, p = 0.007]. There were significant effects of both sexes [F(1,12) = 7.07, p = 0.021] and diet [F(1,12) = 10.61, p = 0.007] on the relative abundance of the family Prevotellaceae. We also observed significant effects of both sex [F(1,12) = 5.04, p = 0.044] and diet [F(1,12) = 8.52, p = 0.013] on the relative abundance of the family Bifidobacteriaceae. The proportion of Deltaproteobacteria (class level), Desulfovibrionales (order level), and Desulfovibrionaceae (family level), corresponding to sulfate-reducing bacteria, significantly increased in the M + HFD group compared to that in the M + LFD group. At the genus level, taxa with significant interaction effects between sex and diet included Ruminococcaceae_uc [F(1,12) = 7.20, p = 0.012] and Turicibacter [F(1,12) = 16.34, p = 0.002]. Diet had a significant effect on the relative abundance of bacterial genera, including Odoribacter [F(1,12) = 10.20, p = 0.008] and Parvibacter [F(1,12) = 6.02, p = 0.030]. In the male group, the ratio of Odoribacter was significantly increased by HFD, and the ratio of Parvibacter was significantly decreased. At the species level, a significant effect of interaction of sex and diet was noticed on the relative abundance of Bifidobacterium_uc [F(1,12) = 18.65, p = 0.001].

Figure 7
figure 7

Effects of sex and diet and their interactions on gut microbial composition at (A) class, (B) order, (C–F) family, (G–J) genus, and (K) species levels. Results were analyzed using two-way ANOVA and are expressed as mean ± standard deviation (SD). The p-values for main effects owing to sex and diet and their interaction are reported in box. Bars labeled with different letters are significantly different based on Duncan's multiple range tests (p < 0.05).

Full size image

Correlation of bacteria at family, genus, species levels with the number of tumors and weight of adipose tissue

Ruminococcaceae_uc (genus level), the abundance of which was significantly increased by HFD in the male group, showed a positive correlation with the number of colon tumors (R = 0.55) (Table 2). The abundance of Turicibacter (genus level) was significantly higher in the F + LFD group than in all other groups, and relative abundance of Turicibacter showed a negative correlation with the weight of whole white adipose tissue (R = − 0.48). Odoribacteraceae (family level) and Odoribacter (genus level), which were significantly increased by HFD in the male group, showed a positive correlation with the weight of whole white adipose tissue (R = 0.42).

Table 2 Correlation of bacteria with the number of tumors and weight of adipose tissue.
Full size table

Functional differences in the gut microbiota between male and female mice

The potential functions of gut microbiota that showed compositional differences were determined using the PICRUSt and MinPath algorithms and are presented in Fig. 8A. Microbial gene functions, lipopolysaccharide biosynthesis, and phosphoinositide-3-kinase–Akt signaling pathway were significantly higher in the M + LFD group than in the F + HFD group (Fig. 8B). Microbial gene functions, AMP-activated protein kinase signaling pathway, and retinoic acid-inducible gene I-like receptor signaling pathway were significantly lower in the M + LFD group than in the F + HFD group (Fig. 8B).

Figure 8
figure 8

(A) Heatmap of metabolic pathways of all experimental groups obtained from PICRUSt analysis of 16S rRNA gene sequence data. Pathways are colored based on their abundance relative to that of all measured genes. (B) Boxplots showing the significantly different KEGG pathways between male and female mice.

Full size image

Discussion

In this study, the number of colon tumors in the M + HFD group was significantly higher than that in all female groups. The effect of sex on the number of colon tumors and the effect of interaction between sex and diet were significant. This may support the findings of previous epidemiological studies showing that the correlation between obesity and colorectal cancer was stronger in men than in women32,33. We also found that the expression of MUC2 that forms the protective mucus layer in the large intestine was significantly higher in all female groups than in the M + HFD group. MUC2 knockout mice have been previously found to be more susceptible to colon tumorigenesis, and MUC2 deficiency resulted in increased cell proliferation, reduced apoptosis, and elevated c-Myc levels in tumors34. The downregulation of MUC2 increases the levels of Ki67 and matrix metalloprotease-9, suggesting an association between MUC2 and cellular proliferation and invasion35. Induction of murine colorectal cancer through oncogenic K-Ras expression results in a significant reduction of MUC2 expression in HFD group compared to that in the normal diet-fed group36. In Winnie mice, MUC2 mutations cause endoplasmic reticulum stress in intestinal goblet cells, resulting in depletion of the mucus barrier and increased bacterial translocation leading to spontaneous colitis37. Taken together, this study reveals that diet- and sex-specific changes in MUC2 expression may be associated with an increased risk of colorectal cancer.

In this study, male-specific upregulated DEGs were mainly involved in cancer-related pathways, such as cytokine–cytokine receptor interaction, HIF-1 signaling, and central carbon metabolism. Indeed, obesity-induced chronic inflammation is a major link between obesity and colorectal cancer19. Inflammatory cytokines such as interleukin (IL)-6, tumor necrosis factor alpha, and CC motif chemokine ligand 2 produced in adipose tissue are major factors that induce adiposity-related chronic inflammation38,39. Specifically, IL-6 is overexpressed in colorectal cancer tissues, and an increase in serum IL-6 concentration has been found to correlate with an increase in tumor size and reduced survival rate40. HIF-1 is a key factor that regulates cellular adaptation to hypoxia, and elevated HIF-1 levels are strongly correlated with tumor metastasis, angiogenesis, and poor patient prognosis41. To determine the major genes that play a role in HFD-linked colorectal cancer development, male-specific upregulated hub genes were selected using the degree algorithm of the cytoHubba plug-in. The top 10 hub genes were Rhob, Hif1a, Met, Pkm, Eno1, Mapk6, Serpine1, Arhgap35, Junb, and Exoc5. The expression of RhoB is increased by stress-induced signaling events, such as lipopolysaccharides, inflammatory cytokines, and growth factors42. An increased RhoB expression contributes to the development of ulcerative colitis by regulating cell signals and inducing changes in intestinal microbial composition and metabolites43. Met expression is significantly increased in the early stages of colorectal cancer44, and MAPK6 promotes cancer development by activating Akt in an mTORC2-independent manner45. Pyruvate kinase isoenzyme type M2 (PKM2) encoded by PKM increases anabolic synthesis of macromolecules through the pentose phosphate pathway and consequently promotes proliferation and growth of cancer cells46,47. Taken together, RhoB, HIF1A, Met, MAPK6, and PKM may play key roles in promoting colorectal tumor formation in males.

In this study, we found that changes in the composition of gut microbes caused by HFD differed according to sex. In the male group, the proportions of Odoribacter (genus level) and Ruminococcaceae_uc (genus level) were significantly increased, and the abundance of Parvibacter (genus level) was significantly decreased by HFD. In the female group, however, the proportions of Bifidobacteriaceae (family level) and Bifidobacterium_uc (species level) were significantly increased, and the abundance of Turicibacter (genus level) was significantly decreased by HFD. The proportion of Odoribacter has been found to be higher in AOM/DSS-treated mouse model of cancer than in control mice48, and patients with colorectal cancer show the same pattern49. We observed a positive correlation between Odoribacter and the weight of whole white adipose tissue, suggesting that Odoribacter may play a role in tumor formation in males. Ruminococcaceae is increased in patients with colorectal cancer compared to that in healthy people50. Moreover, the proportion of Ruminococcaceae is increased in HFD-fed mice compared to that in mice fed a chow diet51. Since we observed a positive correlation between Ruminococcaceae_uc and the number of colon tumors, Ruminococcaceae_uc seems to be closely related to tumor formation in males. Parvibacter improves the function of gut barrier, increases the expression of ZO-1 corresponding to the tight junction protein, and reduces the entry of LPS into the body52. The proportion of Parvibacter is negatively correlated with body weight and weight of white adipose tissue53. In our study, the abundance of Parvibacter decreased significantly in the M + HFD group compared with that in the M + LFD group. This indicates that a decrease in Parvibacter ratio may also have contributed to tumor development in males. In addition, we found that the abundance of sulfate-reducing bacteria, including Deltaproteobacteria (class level), Desulfovibrionales (order level), and Desulfovibrionaceae (family level), was significantly increased by HFD in the male group. Hydrogen sulfide produced by sulfate-reducing bacteria can promote cancer development by inducing oxidation and DNA damage54. In mice with aberrant colonic crypt induced by AOM, significant increases in Deltaproteobacteria and Desulfovibrionales have been observed in the HFD group compared to that in the control diet group55. Taken together, sulfate-reducing bacteria may also play a role in tumor development.

Bifidobacterium_uc that belongs to the family Bifidobacteriaceae was significantly more abundant in the F + HFD group than in all other groups, and Turicibacter was significantly more abundant in the F + LFD group than in all other groups. In mice, Bifidobacterium and Turicibacter have been reported to be more abundant in females than in males48. Bifidobacteriaceae increases the production of short-chain fatty acids and strengthens the function of gut barrier56. Turicibacter has been reported to have a positive correlation with the butyric acid production57. Moreover, the abundance of Turicibacter has been found to be decreased in animal models of inflammatory bowel disease58,59 and in HFD-induced obese mice60. In our study, Turicibacter was negatively correlated with the weight of whole white adipose tissue. Therefore, Turicibacter may have a diet-dependent effect on sex differences in colorectal cancer. In addition, our results showed the F + HFD group possessed less microbiota involved in tumor-promoting pathways and more microbiota involved in tumor-suppressing pathways than did the M + LFD group, suggesting the possible involvement of gut microbiota in colon carcinogenesis.

The strength of this study lies in its capacity to provide indispensable data essential for tailored colorectal cancer prevention and treatment, extending beyond the consideration of estrogen levels and incorporating a genomic perspective. It aims to elucidate the mechanisms contributing to the higher incidence of colorectal cancer in men and the possible increase in postmenopausal women compared to their premenopausal counterparts. A limitation, however, is that the selection of genes and bacteria is based on group-to-group comparisons. To identify actionable targets, it would be necessary to delve further into functional studies to establish causality.

In summary, high-fat diet accelerates colorectal tumor formation in male mice not in both ovary intact or ovariectomized female mice suggesting estrogen depletion might not be solely responsible for increased tumor formation. The number of HFD-modulated group-specific DEGs was 290 for males, 64 for females, and 137 for OVX females. Genes up-regulated HFD males include pathways related to cytokine-cytokine interaction, HIF-1 signaling pathway, central carbon metabolism in cancer, all of which are positively associated with cancer development. Sex-specific alterations in gut microbial composition in response to HFD in males involve elevated proportions of Odoribacter and Ruminococcaceae_uc, accompanied by a reduction in Parvibacter.

Methods

Animals and experimental design

In this study, we employed a high-fat diet-induced obesity model to explore the influence of sex hormones and genes as factors contributing to sex-specific differences in obesity-related colorectal cancer. We also investigated their associations with the composition of gut bacteria. This choice of model was made in preference to commonly used genetically modified obesity models, such leptin-deleted ob/ob, leptin receptor-deleted db/db or adiponectin-deleted model. The decision was based on previous reports suggesting that leptin and its receptor, as well as adiponectin, possess signaling capabilities that promote the development of colorectal tumors61,62,63. Simultaneously, we considered the composition of gut microbiota, which can be significantly altered by dietary factors. This parameter served as another indicator of the mechanisms at play, further justifying the used of the diet-induced obesity model as a more human-relevant approach. Seven-week-old male and female C57BL/6J mice purchased from Charles River Laboratories (Yokohama, Japan) were housed in plastic cages (2 ~ 3 mice/cage) under controlled temperature (21 ± 2 °C), humidity (50 ± 10%) and light (12 h light/dark cycle) conditions. After one-week acclimatization, sham surgery was performed on male mice (n = 14), and ovariectomy (n = 17) or sham surgery (n = 19) was performed on female mice. After 14 days of recovery period, male, female or ovariectomized female animals were allocated into two dietary groups: a low-fat diet (LFD; 10 kcal% fat) group or a HFD (45 kcal% fat) diet group (Fig. 9). The experimental groups included M + LFD (n = 6), M + HFD (n = 8), F + LFD (n = 10), F + HFD (n = 9), OVX + LFD (n = 9), and OVX + HFD (n = 8). After sorting by body weight, mice were alternately allocated into two groups in descending order to ensure similar baseline average body weight in both LFD group and HFD group. Colitis-associated colorectal cancer was induced by a single intraperitoneal injection of 10 mg/kg azoxymethane (AOM) at day 15, followed by 1.5% (w/v) DSS in drinking water starting one week after AOM injection. Mice received three cycles of DSS treatment, each consisting of 5 consecutive DSS supply at 10-day intervals. A healthy control group was not included in this study, as the primary objective was to investigate hormonal and genetic determinants in colitis-associated CRC. The experimental diet was supplied for 14 weeks and body weight and food intake were measured three times a week. At 75 days after AOM injection, mice were sacrificed under anesthesia using a 2:1 mixture of Zoletil (Virbac, Magny-en-Vexin, France) and Rompun (Bayer, Seoul, Republic of Korea). We used the “weight loss exceeding 20% of the initial weight” as the humane end point. Colon tissues from the anus to below the cecum were excised using scissors and opened longitudinally along the main axis64.Feces were washed with phosphate-buffered saline, and the colons were laid flat on a glass plate. The total number of tumors (nodular, polypoid, or caterpillar-like) in the entire colon was counted65. After macroscopic evaluation of tumors, approximately 1 cm of the distal colon (1.5 cm proximal to the anus) was fixed with formalin at room temperature for histological analysis (n = 4). We selected the four animals with the closest average body weight for each group. Rest of the distal colon was immersed in RNA later (Thermo Fisher, Waltham, MA, USA) solution for microarray analysis (n = 5) and stored at -80 °C. Fecal samples were collected from all the animals in each group three days before sacrifice and stored at − 80 °C until further analysis. The sample size was determined through power analysis based on the number of colon tumor, which served as a primary outcome. This study was approved by the Institutional Animal Care and Use Committee of the Sookmyung Women’s University, Seoul, Republic of Korea (SMWU-IACUC-2012–015-02) and followed the Korean Institutional Animal Care and Use Committee (IACUC) guidelines (Korea Food & Drug Administration, Chungju, Republic of Korea). This study is reported in accordance with Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

Figure 9
figure 9

Experimental design of the study.

Full size image

Immunohistochemical analysis

Immunohistochemical staining was performed on colon tissue samples (n = 4 per group). We selected four animals with the closest average body weight for each group. Colon tissues were fixed in 10% buffered formalin and embedded in paraffin. Paraffin-embedded sections (5 µm) were analyzed using rabbit anti-MUC2 antibody (Abcam, Cambridge, MA, USA). Slides were examined using an Olympus BX51 microscope (Olympus, Tokyo, Japan) at 20–400 × magnification. Quantitative analyses were performed using the Image-Pro 10 program, and the ratio of positively stained area to the total area was calculated for each sample. Three random scans per section were analyzed and averaged.

Microarray analysis and identification of DEGs

Colon tissue microarray analysis was conducted using five animals from each group, selected based on the closest average body weight. Total RNA was extracted from colon tissues using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA, USA). RNA purity was assessed from the ratio of absorbances at 260 and 280 nm using a ND-2000 NanoDrop (Thermo Fisher Scientific, USA), with values between 1.7 and 2 considered acceptable. Total RNA integrity was checked using an Agilent Technologies 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), with an RNA Integrity Number value ≥ 7. Raw data were extracted using the Agilent Feature Extraction Software (v11.0.1.1) and were summarized for the same gene automatically using the Agilent feature extraction protocol to generate a raw data text file, providing expression data for each gene probed on the array. Array probes with Flag A in samples were filtered out. The selected gProcessedSignal value was logarithmically transformed and normalized using the quantile method. Statistical significance of the expression data was determined using fold change and independent t-test, in which the null hypothesis was that no difference existed among groups. Fold change ≥ 1.5 and p-value < 0.05 were regarded as the cut-off criteria to determine DEGs. Three comparison groups of DEGs were generated: M + LFD vs. M + HFD, F + LFD vs. F + HFD, and F(OVX) + LFD vs. F(OVX) + HFD. We generated a Venn diagram that showed the relationships among DEGs in these three pairwise comparison groups.

Functional enrichment analysis of DEGs

The KEGG database contains network-related information on genes and molecules66. In our study, KEGG enrichment analysis was performed using the g:Profiler tool (https://biit.cs.ut.ee/gprofiler/) to map DEGs to specific biochemical pathways in the KEGG database67. The statistical significance of KEGG pathways was determined using genes with false discovery rate (FDR) < 0.05, and the enrichment score [− log10 (FDR)] was calculated. The top ten KEGG pathways were sorted in descending order based on the enrichment score. When multiple pathways had the same enrichment score, more than 10 pathways were selected.

Construction of PPI network and hub gene identification

To confirm the interactions among DEGs, a PPI network was constructed using the STRING database (https://string-db.org). The criterion for PPI network included a confidence score ≥ 0.4. In a PPI network, each gene is represented as a node, and the interactions between genes are represented by edges68. Degree represents the number of edges connected to a given node, and nodes with a high degree are commonly referred to as hub genes69. In our study, hub genes were selected in the order of high-degree scores using the degree algorithm of cytoHubba that identifies hub objects and subnetworks in a complex interactome70.

Analysis of the gut microbiota

Fecal bacterial DNA was extracted using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany). Amplification by polymerase chain reaction (PCR) was performed using primers targeting the V3–V4 region of the 16S rRNA gene. Thermal cycling consisted of an initial denaturation at 95 °C for 3 min, followed by 25 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, and extension at 72 °C for 30 s, and a final elongation at 72 °C for 5 min. Non-target products were removed using CleanPCR (CleanNA), and the quality and size of PCR products were confirmed using a Bioanalyzer 2100 (Agilent, Palo Alto, CA, USA). Mixed amplicons were pooled and sequenced using an Illumina MiSeq Sequencing system (Illumina, San Diego, CA, USA). Data analysis was performed using the EzBioCloud database (ChunLab Inc., Seoul, Republic of Korea). Based on 16 s rRNA gene sequences, differences in the composition of gut microbiota between experimental groups were verified, and the potential function of gut microbiota was predicted using the PICRUSt and MinPath algorithms. The basic functional profiles were estimated using PICRUSt, and MinPath was used to predict the metabolic pathways. MinPath is used to minimize the number of pathways required to map all functions by handling the problem of pathway redundancy. KEGG pathways that were differentially expressed in all experimental groups were identified and visualized as a heatmap.

Statistical analysis

Statistical analysis was performed using SAS 9.4 program (SAS Institute Inc., Cary, NC, USA), and all values are presented as mean ± standard deviation (SD). Data were analyzed using one-way analysis of variance (ANOVA) or two-way ANOVA followed by Duncan's multiple range test. p < 0.05 was considered statistically significant. Two-way ANOVA was performed on the four groups (M + LFD, M + HFD, F + LFD, and F + HFD) to determine the main effects of sex and diet separately and their interactions. The Mann–Whitney U-test was used to compare differences between two groups without normal distributions, and the Kruskal–Wallis H test (non-parametric ANOVA) was used to compare differences between more than two groups. Correlations between the relative abundance of gut bacteria and the number of tumors or weight of adipose tissue were evaluated using the Pearson’s correlation test.

Ethics approval

The protocols involving mice were approved by the Institutional Animal Care and Use Committee of Sookmyung Women`s University, Seoul, Republic of Korea (SMWU-IACUC-2012-015-02).