Obesity presents a significant public health concern given its association with increased cancer incidence, unfavorable prognosis, and metastasis. However, there is very little literature on the effects of weight loss, following obesity, on risk for colon cancer or liver cancer. Therefore, we sought to study whether intentional weight loss through diet manipulation was capable of mitigating colon and liver cancer in mice. We fed mice with a high-fat diet (HFD) comprised of 47% carbohydrates, 40% fat, and 13% protein for 20 wk to mimic human obesity. Subsequently, azoxymethane (AOM) was used to promote colon and liver carcinogenesis. A subset of obese mice was then switched to a low-fat diet (LFD) containing 67.5% carbohydrate, 12.2% fat, and 20% protein to promote intentional weight loss. Body weight loss and excess fat reduction did not protect mice from colon cancer progression and liver dysplastic lesion in the AOM-chemical-cancer model even though these mice had improved blood glucose and leptin levels. Intentional weight loss in AOM-treated mice actually produced histological changes that resemble dysplastic alterations in the liver and presented a higher percentage of F4/80+CD206+ macrophages and activated T cells (CD4+CD69+) in the spleen and lymph nodes, respectively. In addition, the liver of AOM-treated mice exposed to a HFD during the entire period of the experiment exhibited a marked increase in proliferation and pNF-κB activation. Altogether, these data suggest that intentional weight loss following chemical-induced carcinogenesis does not affect colon tumorigenesis but may in fact negatively impact liver repair mechanisms.
- colon carcinogenesis
- weight loss
- liver pathology
- azoxymethane-induced carcinogenesis
epidemiological studies have associated obesity with increased cancer incidence (4, 36). This presents a significant public health concern given that the prevalence of obesity, defined by a body mass index (BMI) >30 kg/m2, has now reached 35% in the U.S. with no indication of a decline (14, 26). Colon and liver cancer are among the cancers in which augmented adiposity is believed to promote cancer progression and metastasis (36). For instance, a large cohort study in the U.S. reported that the relative risk of dying from colon cancer was 1.82 times higher and from liver cancer was 4.52 times higher in men with a BMI of ≥35 mg/m2 than those with normal body weight (5). As such, the majority of colon cancer and liver cancer cases are likely preventable through lifestyle changes (15).
Consistent with the epidemiological literature, a number of controlled experimental studies in rodents have supported the link between diet-induced obesity and cancer risk. We have reported that consumption of a high-fat diet was associated with an increase in the number of large polyps in the Apcmin/+ mouse model of intestinal tumorigenesis (8). Using a chemically induced model of colon cancer, Park et. al. (30) found that diet (45% kcal from fat)-induced obesity in male A/J mice facilitated tumor development. Similarly, Yoshimoto et al. (41) reported that diet-induced obesity promotes liver cancer development in mice after exposure to a chemical carcinogen; mice fed a high-fat diet for 30 wk had a 100% incidence of liver tumors whereas none of the normal diet fed mice exhibited tumorigenesis. An earlier study indicated that diet-induced obesity promotes liver tumorigenesis as an approximate doubling in tumor number and size was observed in both male and female mice following 34 wk on a 60% high-fat diet after exposure to a carcinogen (29).
Although the link between obesity and cancer has not been questioned, the exact mechanisms that are driving this relationship have not yet been fully elucidated. Given the plethora of biological processes that become dysfunctional during the obese state, it is likely that this relationship is complex and multifaceted. Recent advances have proposed a number of plausible mechanisms including, inflammatory processes, metabolic factors, adipokines, hormones, immune function, and gut microbiota; all have been reported to affect the tumorigenic response and are known to be altered in obesity. For example, we have reported that high-fat diet increased the expression of inflammatory mediators in the adipose tissue [F4/80, CD11c, Toll-like receptor-4 (TLR-4), and monocyte chemotactic protein-1 (MCP-1)] and tumor microenvironment (IL-12, MCP-1, IL-6, and TNF-α) in the Apcmin/+ mouse model of intestinal tumorigenesis, which was associated with an increase in the number of large polyps (8).
To date, the link between obesity and cancer is well established and significant advances have been made at determining plausible biological mechanisms. However, there is very little literature on the effects of weight loss, following obesity, on risk for colon cancer or liver cancer. To our knowledge, evidence is limited to prospective and observational studies (22, 25) with no reports in animals. Therefore, we sought to investigate whether a reduction in adipose tissue in obese mice was capable of protecting against carcinogenesis. Obesity manipulations were performed using dietary approaches; a high-fat diet was administered to mice to induce obesity and a low-fat diet was administered to implement weight loss. We used a chemical model, azoxymethane, to promote colon carcinogenesis. Given that azoxymethane has also been reported to induce liver cancer, we extended our findings to examine any effects that weight loss may have on liver tumorigenesis. Inflammation, metabolism, adipokines, and immune function have all been implicated in the link between obesity and cancer; therefore, we examined changes in these parameters in association with the tumorigenic outcomes.
MATERIALS AND METHODS
Animals and high-fat diet.
C57BL/6 male mice were purchased from Jackson Laboratories and housed in ventilated cages in accordance with established institutional guidance and approved protocols from the Institutional Animal Care and Use Committee of the University of South Carolina. At 10 wk of age all mice were fed a high-fat diet (HFD) comprised of 47% carbohydrates, 40% fat, and 13% protein (BioServ, Frenchtown, NJ) as previously described (12). Once mice achieved an obese phenotype (after 20 wk of HFD feeding), they were divided into three groups following stratified random sampling for body weight, fat mass, and lean mass: 1) Control-HFD group (n = 6), mice were given 6 weekly consecutive intraperitoneal injections of saline and were maintained on a HFD for 30 additional wk; 2) H-AOM-LFD group (n = 13), mice were injected intraperitoneally for 6 consecutive wk with AOM to induce colon cancer (12 mg/kg, week 1; 10 mg/kg, week 2; and 7.5 mg/kg, weeks 3–6) and were subsequently switched to a purified low-fat diet (LFD) for 30 wk (LFD contained 67.5% carbohydrate, 12.2% fat, and 20% protein; Ain76A from BioServ); and 3) H-AOM-HFD group (n = 11), mice followed the same AOM treatment as mentioned above and continued with the same HFD for an additional 30 wk. As we sought to intentionally alter calorie and macronutrient content to achieve obesity or weight loss, we did not utilize isocaloric diets and neither did we intentionally use calorie restriction in H-AOM-LFD mice at any stage of the study. The AOM treatment was selected based off of previous AOM doses used in rodent studies (7, 20, 27, 35) and all AOM-treated mice received a similar amount of the carcinogen, as there were no differences in body weights among the treatment groups at the time of administration. Mice were euthanized with isoflurane at 60 wk of age, colons were dissected, polyps were counted using a stereoscope, and a portion of the colon was fixed in 4% paraformaldehyde, paraffin embedded and sectioned for hematoxylin and eosin (H&E) staining. Fat pads and liver were extracted, weighed, and stored in fixative or frozen at −80°C for further analysis. Spleen and lymph nodes were dissected for flow cytometry analysis.
Body composition and food intake.
Body weight and food consumption were monitored weekly over the course of the study. Body composition was assessed twice (before AOM injection and before death) via dual-energy X-ray absorptiometry (DEXA; Lunar PIXImus) as previously described (12).
A week before death, blood samples were collected from the tip of the tail after a 5-h fast. Blood glucose concentrations were determined in whole blood using a glucometer (Bayer Contour). Collected blood was centrifuged at 2,000 g for 10 min at 4°C. Plasma insulin concentrations were determined by a commercially available ELISA kit (Mercodia). Insulin resistance was estimated using the following equation homeostasis model of assessment (HOMA) index = fasting glucose (μU/ml) × fasting insulin (mmol/l)/22.5. At death (nonfasted) blood was collected from the inferior vena cava and leptin concentrations were assessed using an ELISA kit (R&D Systems).
Two hours before death, mice received an intraperitoneal injection of 5-bromo-2-deoxyuridine (BrdU; 40 mg/kg body wt; Sigma) to later assess BrdU incorporation in the liver.
Paraffin block sections of the colon, epididymal fat, and liver were stained with H&E for histological examination. Liver specimens, were also stained for picrosirius red for further analysis. Histological sections from the colonic polyps (adenomas) were evaluated for the level of dysplasia and were graded as low and high grade. Histological examination of the liver specimens was also performed for the presence of nonalcoholic fatty liver disease (NAFLD) according to the scoring system designed by the Pathology Committee of the NASH Clinical Research Network, which addresses the full spectrum of lesions of NAFLD (21). The scoring system is comprised of four histological features that are evaluated semiquantitatively [steatosis (0–3), inflammation (0–2), ballooning (0–2), and fibrosis (0–4)] and another nine that are recorded as either present or absent (21).
BrdU and terminal deoxynucleotidyl transferase dUTP-mediated nick-end labeling (TUNEL) staining were performed in liver specimens to assess liver proliferation and apoptosis, respectively. Liver slides were deparaffined and stained using a BrdU In-Situ Detection Kit from BD Biosciences (cat. no. 550803) and TUNEL In-Situ Apoptosis Detection Kit from Millipore (cat. no. S7100).
Western blot analysis.
The frozen colon (tumor free whole colon), liver (left lateral lobe), and epididymal fat were homogenized on the same day using identical mueller buffer (6) and total protein was determined via the Bradford method. All samples were run in triplicate on the same plate with a coefficient of variation of <3%. Colon, liver, and epididymal fat homogenates (10–20 μg) were fractioned on 10–12% SDS polyacrylamide gels. Gels were transferred to PVDF membrane at 18 V for 1 h (Genie electrophoretic transfer; Idea Scientific). Before proceeding with Western blot analysis, all membranes were stained with ponceau solution to assure equal loading and quality of the transfer. Subsequently, membranes were blocked for 1 h in 5% milk in Tris-buffered saline, 0.1% Tween 20 (TBST). Primary antibodies were diluted 1:1,000 in a 5% milk-TBST or 5% BSA-TBST solution according to antibody specifications for either a 1 h incubation at room temperature [Cell Signaling: NF-κB (4764), Stat3 (4904), and Jnk (9258); Abcam: PCNA (2426)] or overnight incubation at 4°C [Cell Signaling: phospho-NF-kB (3033), phospho-Stat3 (9145), phospho-Jnk (4671), and cleaved-caspase 3 (9664)]. An anti-rabbit IgG horseradish peroxidase conjugated secondary antibody (Cell Signaling 7047) was diluted 1:2,000 in 5% milk and incubated for 1 h at room temperature. An enhanced chemiluminescent substrate for detection of horseradish peroxidase (Thermo Scientific) was used to visualize the antibody-antigen interaction. All samples were run on the same gel. Autoradiography films were scanned and blots were quantified using scientific imaging software (ImageJ). After completion of the Western blot, all membranes were stained with Amido black for 5 min and then destained for 30 min in 25% isopropanol and 10% acetic acid solution to allow for total protein normalization for each well using Quantity One Software (Bio-Rad). We analyzed the entire lane of each sample (low and high molecular weight proteins) by densitometry and used this value to standardize its corresponding target antibody. This method of normalization has been shown to be more accurate than typically used loading controls (2). All Western blot data are presented as 1) total protein, 2) phosphorylated protein, and 3) phosphorylated protein:total protein.
Flow cytometry analysis.
Spleen and lymph node tissues from pooled mice (n = 3/group) were dissociated (gentleMACS dissociator), filtrated through a 70-μm cell strainer (Fisher Scientific), washed with 10% FBS in RPMI, and resuspended in FACS buffer. Cells were incubated with the following antibodies from Biolegend: APC anti-CD4 (GK1.5), FITC anti-CD8 (LY-2 53-6.7), PE anti-CD69 (H1.2F3), FITC anti-CD11b (M1/70), PE anti-CD11c (HL3), FITC anti-F4/80 (BM8), and PE anti-CD206 for 45 min in a shaker at room temperature. Lymphocytes were then washed in FACS buffer and fixed in 2% paraformaldehyde and analyzed by flow cytometry (DB FACS Aria II). Given that samples were pooled for these analyses, groups were not statistically compared.
Data were analyzed using commercially available statistical software: Prism 6 (GraphPad Software). A one-way ANOVA, following Tukey post hoc analyses were used to determine differences among Control-HFD, H-AOM-LFD, and H-AOM-HFD mice. The nonparametric Kruskal-Wallis test was used to compare level of dysplasia in the colon and NAFLD in the liver. Any statistical test that did not pass the equal-variance test (Bartlett's test for equal variances) was log transformed and then reanalyzed. Data are presented as the mean ± SE and the level of significance was set at P < 0.05.
Morphometric and metabolic characterization of obese AOM-treated mice.
To confirm that HFD feeding increased adiposity we measured body weights throughout the study and baseline and terminal body fat percentage via DEXA. As expected, 20 wk of HFD consumption resulted in an obese phenotype; mice weighed, on average, 47 g with a body fat percentage of 45%.
We next corroborated that switching the diet of obese mice from high-fat to low-fat content would result in a lean phenotype. According to previous data from our laboratory, a “lean mouse” (mouse in a low-fat diet for 20 wk) has ∼2 g of visceral fat, blood leptin levels around 25 ng/ml, and an average fasting metabolic panel of 110 mg/dl of glucose, 1.15 μg/l of insulin, and HOMA score of 12 (12). In this study, obese mice switched to a LFD showed a dramatic drop in body weight, which may be explained, in part, by the initial reduction in food intake due to an adjustment to a novel diet (Fig. 1, A and B). Following the first week of diet manipulation however, H-AOM-LFD mice consumed a relatively similar quantity of LFD comparable to other studies performed in our laboratory (data not shown); the small apparent difference between the groups is most likely due to common spillage of the HFD, given its poor consistency. It is important to note that during AOM injections (week 1–6) none of the mice lost body weight; thus mice received a similar amount of the carcinogen (Fig. 1A). At baseline all groups had similar fat and lean mass (Fig. 1C). Regarding body composition and metabolic measurements, at death H-AOM-LFD mice displayed a lean phenotype, with significantly less fat mass (P < 0.0001), fasting glucose (P = 0.0076), and leptin (P < 0.001) and increased lean mass (P < 0.00001) compared with Control-HFD and H-AOM-HFD mice (Fig. 1, D, E, and H). However, no differences were seen in insulin (P = 0.2963) and HOMA index (P = 0.1442) between groups (Fig. F and G). In general, Control-HFD and H-AOM-HFD mice presented similar morphological (fat mass, percent body fat, and lean mass) and metabolic characteristics.
Additionally, we examined the weight of each individual fat pad, liver, and spleen tissues given their relevance to obesity-induced cancer. In general, all visceral fat pads (P < 0.0001) and liver mass were reduced in H-AOM-LFD mice compared with Control-HFD and H-AOM-HFD mice (Fig. 2, A–E). H-AOM-LFD mice showed a significant decrease in epididymal (P < 0.0001) and kidney fat (P < 0.0001) compared with Control-HFD and H-AOM-HFD. Interestingly, mesentery fat pad (P < 0.0001) and liver mass (P < 0.0001) were significantly different between all groups (Fig. 2, D and E) with Control-HFD mice having the highest and H-AOM-LFD mice with the lowest amount of mesentery fat and liver mass. Spleen weight, a marker of systemic inflammation, was found to be increased in the AOM-treated mice significantly more than Control mice (P = 0.0028; Fig. 2F). Even more, H-AOM-HFD mice showed a higher increase in spleen mass than H-AOM-LFD mice (P = 0.0488).
To determine changes in adipocyte morphology and degree of inflammation in the adipose tissue, we performed H&E staining and Western blot for Jnk in the epididymal fat pad. As seen in the representative images in Fig. 2G, Control-HFD and H-AOM-HFD mice exhibit larger adipocytes than H-AOM-LFD mice. We next assessed protein expression of Jnk, to corroborate a less proinflammatory state in the epididymal fat pad of H-AOM-LFD mice (Fig. 1, H–J). A significant reduction in total Jnk (P = 0.0196) was observed in H-AOM-LFD and H-AOM-HFD compared with Control-LFD mice (Fig. 2H). As expected, H-AOM-LFD mice showed a reduction in Jnk activation (Fig. 2I) compared with H-AOM-HFD mice (P = 0.0004) but not Control-HFD (P = 0.1351). When we calculate the ratio between phosphorylated Jnk and total Jnk, we observed a significant increase in H-AOM-HFD mice compared with Control-HFD (P = 0.0343) and H-AOM-LFD (P = 0.0004), but no difference was observed between Control-HFD and H-AOM-LFD mice (P = 0.3223; Fig. 2H).
Effects of obese-to-lean phenotype on histopathology and cancer-related signaling pathways in colonic tissue of AOM-treated mice.
To evaluate whether body weight loss in obese mice protects against colon cancer, we next examined mucosal histopathology and protein expression of several proteins and transcription factors involved in cell proliferation, apoptosis, and inflammation. There were no significant differences in the polyp (adenoma) number (P = 0.7813), polyp size (P = 0.7478, 0.7301, 0.6050, 0.2678, 0.0947, and 0.4608, respectively), and grade of dysplasia across (P = 0.0014, 0.8667, and 0.4526, respectively) the AOM-treated groups (Fig. 3, A–D). The Control-HFD group was not compared with the AOM-treated groups; this group is depicted in Fig. 3B simply to document that these mice did not develop polyps. The polyp results were consistent with the protein expression of cleaved caspase-3, as there was no significant difference in its content between H-AOM-LFD and H-AOM-HFD groups nor in AOM-treated groups vs. control (P = 0.0875; Fig. 3F). However, a significant difference was observed in cell proliferation (PCNA expression) between H-AOM-LFD and H-AOM-HFD groups (P = 0.0122) but not between H-AOM-LFD and Control-HFD (P = 0.2575) nor between H-AOM-HFD and Control-HFD (P = 0.4427; Fig. 3E).
We next measured the protein expression of key transcription factors involved in inflammatory signaling pathways. No significant differences were observed in total, activated and the ratio of phosphorylated and total Stat3 (P = 0.0908, 0.9481, and 0.4999, respectively), and NF-κB (P = 0.8020, 0.2909, and 0.0976, respectively) between groups in colonic tissue (Fig. 4, A and B). However, a significant difference was observed between the ratio of phosphorylated and total Jnk between H-AOM-LFD and H-AOM-HFD mice (P = 0.0072), without significant differences in total or phosphorylated protein between groups (P = 0.2059 and 0.0529, respectively; Fig. 4C).
Effects of obese-to-lean phenotype on the histopathology and cancer-related signaling pathways in liver tissue of AOM-treated mice.
Since AOM is metabolized in the liver (34) and previous studies in rodents have reported that a single injection of AOM is capable of inducing liver injury and hepatocellular carcinoma in rodents (31, 37), we next investigated whether the liver of AOM-treated mice was affected by the multiple AOM injections. We performed H&E (Fig. 5A) and picrosirius red (Fig. 5B) staining on the liver tissue of Control-HFD, H-AOM-LFD, and H-AOM-HFD mice. Subsequently, liver specimens were evaluated and scored for pathologic changes of nonalcoholic fatty liver disease (NAFLD), as well as the presence of dysplastic changes of the liver (Fig. 5, C and D). In general, NAFLD scoring showed similar pathological manifestations between Control-HFD and H-AOM-HFD mice (Fig. 5C). H-AOM-LFD mice presented a significant decrease in steatosis (P = 0.0235 and 0.0045) and ballooning degeneration of hepatocytes (P = 0.0080), without visible changes in inflammation (P = 0.2680) and fibrosis of the parenchyma (P = 0.1164) compared with Control-HFD mice. With respect to other types of histological changes, a higher percentage of H-AOM-LFD mice displayed numerous aggregations of Kupffer cells (73%; P = 0.0017) and atypic enlarged hepatocytes with prominent nuclear pleomorphism (54%; P = 0.0280) compared with Control-HFD and H-AOM-HFD mice (Fig. 5D).
Additionally, we measured the protein expression of markers for cell proliferation, cellular apoptosis, and inflammation in the liver tissue. We observed that cell proliferation (BrdU staining and PCNA Western blot, Fig. 6, A and C) was significantly elevated in the liver of H-AOM-HFD mice compared with Control-HFD and H-AOM-LFD mice (P = 0.0308 and 0.0083, respectively). However, no significant differences in cellular apoptosis (TUNEL staining and cleaved caspase-3 Western blot, Fig. 6, B–D) were observed between any of the groups (P = 0.2400).
In Fig. 7A, we show a significant increase in total Stat3 in H-AOM-LFD mice when compared with Control-HFD mice (P = 0.0273) but not H-AOM-HFD mice (P = 0.9095). No significant difference was observed between Control-HFD and H-AOM-HFD mice in total Stat3 (P = 0.1200). On the other hand, Stat3 activation was significantly increased in H-AOM-LFD and H-AOM-HFD mice compared with Control-HFD mice (P = 0.0262 and 0.0344, respectively). However, no changes were observed in the calculated ratio of phosphorylated and total Stat3 between groups (P = 0.2203). In the NF-κB blots (Fig. 7B) we observed similar expression of total, phosphorylated, and calculated ratio of phosphorylated and total NF-κB protein between Control-HFD and H-AOM-LFD groups (P = 0.4226, 0.9953, and 0.5099, respectively). A significant increase was observed in total and phosphorylated NF-κB in H-AOM-HFD mice (P = 0.0118 and 0.0095, respectively) compared with Control-HFD mice. Calculations of the ratio of phosphorylated and total NF-κB showed a significant difference between H-AOM-LFD and H-AOM-HFD mice (P = 0.0061). However, no differences were observed in NF-κB ratio between Control-HFD mice and H-AOM-HFD groups (P = 0.1504). Figure 7C shows a significant increase of total Jnk in H-AOM-HFD mice compared with Control-HFD and H-AOM-LFD mice (P < 0.0001). No difference in total Jnk was observed between Control-HFD and H-AOM-LFD mice (P = 0.5618). In contrast, a decrease in Jnk activation and ratio was seen in AOM-treated groups compared with Control-HFD mice (P = 0.0272 and 0.0052, respectively).
Effects of obese-to-lean phenotype on spleen and lymph node macrophages and T cells of AOM-treated mice.
To further examine whether the immune cells in AOM-treated mice were compromised, we next analyzed splenic and mesenteric lymph node macrophages, dendritic cells and T cells by flow cytometry. As shown in Fig. 8A, Control-HFD and H-AOM-HFD mice exhibited a similar percentage of macrophages in the spleen and mesenteric lymph nodes. However, a dramatic increase of F4/80+CD206+ macrophages was seen in the spleen of H-AOM-LFD mice (Fig. 8A). We then analyzed the presence of dendritic cells in the splenic and mesenteric lymph node of AOM-treated mice and found minor changes in the percentage of CD11b+CD11c+ dendritic cells between all groups (Fig. 8B).
We next looked for T cells in splenic and mesenteric lymph nodes (Fig. 9, A and B). Interestingly, a drastic change was observed in T-cell percentage of H-AOM-LFD mice. In general, we observed an increase in the percentage of CD8+ T cells, CD4+ T cells, and CD4+CD69+ activated T cells in the spleen and lymph nodes of H-AOM-LFD mice compared with Control-HFD and H-AOM-HFD mice.
The World Cancer Research Fund and the American Institute for Cancer Research has linked excess adiposity with at least 15% of colon and liver cancer incidence. These epidemiological data, in part, imply that a commensurable portion of colon and liver cancers may be preventable. However, it is known that intentional body weight loss is difficult to achieve in an obese population, due to imprinted sedentary lifestyles and poor eating habits. Therefore, it is expected that obesity-related cancer will continue to rise. Arguably, the most convincing evidence for the role of obesity in cancer comes from a meta-analysis report of four prospective cohorts showing that bariatric surgery decreases the risk for cancer among obese individuals (1). Despite this however, data on the effects of weight loss on risk for gastrointestinal or liver cancers are limited and to our knowledge there are no reported studies using controlled experimental mouse models. Thus, using azoxymethane, a chemical model of colon carcinogenesis, we sought to determine whether intentional weight loss, through diet manipulation, was capable of mitigating colon cancer in mice. Given that azoxymethane has also been reported to induce liver cancer, we extended our findings to examine any effects that weight loss may have on liver tumorigenesis. As inflammation, metabolism, adipokines, and immune function have all been implicated in the link between obesity and cancer, we examined changes in these parameters in association with the tumorigenic outcomes.
In general, we found that body weight and excess fat reduction did not protect mice from colon cancer progression and liver dysplastic lesion in the AOM chemical cancer model even though these mice had improved blood glucose and leptin. Interestingly, we found that intentional weight loss in AOM-treated mice actually incited histological changes that resemble dysplastic alterations in the liver and presented a higher percentage of F4/80+CD206+ macrophages and activated T cells (CD4+CD69+) in the spleen and mesenteric lymph nodes, respectively. In addition, the liver of AOM-treated mice exposed to a HFD during the entire period of the experiment exhibited a marked increase in proliferation and pNF-κB activation. One possible explanation for these results could be that the liver of H-AOM-HFD mice may be undergoing liver regeneration; both proliferation and pNF-κB activation are commonly exhibited in liver regenerative processes. Altogether, these data suggest that intentional weight loss after carcinogen exposure may negatively impact liver pathology in mice without affecting colon tumorigenesis.
Generally, animal studies in obesity-induced colon cancer have been performed on mice that are obese before carcinogen (AOM) administration (7, 11). This type of cancer induction, in particular, is not a suitable comparison to study the difference between obese and lean phenotypes in cancer, as the dose of AOM is administered based on body weight. Thus, an obese mouse will receive a higher amount of the carcinogen than their lean counterpart. A dose response study utilizing a single injection of AOM in rats found that a higher dose of AOM translates to more tumorigenesis in the gastrointestinal tract and liver (37). To avoid giving a higher dose of AOM to obese mice, Tuominen et al. (35) selected to administer the same amount of AOM to lean and obese mice. This experimental design revealed no significant differences in tumor number, tumor diameter, nor abnormal crypt foci formation between obese and lean mice (35). Consistent with this finding there were no changes in cleaved caspase-3 across any of the groups. On the other hand, a recent finding that showed a decrease in apoptosis in genetically damaged colon epithelial cells following HFD feedings (33). Similarly, HFD feedings have been reported to decrease apoptosis in the AOM model (9). Interestingly, we revealed a significant increase in PCNA in H-AOM-LFD mice compared with H-AOM-HFD. This finding is in contrast to a report by DeClercq et al. (9) that reported an increase in proliferation in colonic stem cells following HFD feedings in the AOM model. However, the differences in the administration of the HFD feedings in relation to carcinogen exposure along with the fact that we did not specifically examine cell types that may be undergoing proliferation or apoptosis make it difficult to compare our results with previously reported studies.
Questions still remain on the appropriate AOM dosage regime for obesity-related studies in rodents. In the current investigation, we administered AOM to diet-induced obese mice per body weight and subsequently sought to determine whether implementation of weight loss was capable of protecting against colon tumorigenesis. As there were no differences in the AOM dose administered between groups, we can rule out any influence of dose of chemical administration on our findings. Our data indicate that the switch from a HFD to a LFD produced a leaner phenotype. Mice in the H-AOM-LFD group showed a 39% reduction in body weight and 49% drop in fat mass. In addition, H-AOM-LFD mice also reduced blood glucose and leptin concentrations. Despite this however, we conclude that weight loss following exposure to a carcinogen does not protect against colon tumorigenesis as no differences in pathology or molecular markers of cancer progression were found. It is important to note that there were no significant differences observed in fasting insulin nor in the HOMA index between groups. These results could be due to a permanent impartment of insulin signaling or incomplete recovery of insulin metabolism in H-AOM-LFD mice, which could also be contributing to the unfavorable liver pathology. It is of interest that H-AOM-HFD mice exhibited decreased tissue weight compared with Control-HFD mice. While we did not further explore this finding, it is possible that cachexia may be contributing to this effect.
Inflammation plays an important role in the progression of both obesity and cancer, and it is believed to be one of the reasons for the increased incidence of obesity-associated colon cancer (38). In an attempt to link high-fat diet consumption with tumor formation other scientists and our group have reported an increase in gene expression of inflammatory cytokines, such as, IL-6, IL-12, and TNF-α, in the colon of chemical and genetic models of colon cancer (8, 28). Although we did not see differences in colon tumorigenesis across groups, we proceeded with analysis of inflammatory mediators in the colon tissue in an attempt to understand any role that weight loss may have on colon inflammation. In general, we observed no differences in the activation of common transcription factors involved in inflammation. These findings coincide with data from Bugni's group, where they observed no differences in colonic expression of TNF-α, IL-6, IFNγ, and IL-1β genes following dietary switches in the AOM model (35). In summary, our findings indicate that switching from an obese to a lean phenotype does not protect against inflammation in the colon, at least in this model. It is possible that the initial exposure to a HFD may lead to a permanent impairment in inflammatory processes, or alternatively, more time on a LFD may be necessary to reverse the inflammation.
Consistent with the current findings, other scientists have revealed that AOM administration can result in liver injury, tumor formation, and fulminant hepatic failure, which is dependent on the dose of AOM administered (24, 37). Histological and biochemical reports have found microvesicular steatosis and lobular necrosis associated with mitochondria injury and increased circulating alanine aminotransferase between 2 and 20 h after administration of a lethal dose of AOM (100 mg/kg) (24). Lower doses of AOM (15, 20, and 50 mg/kg) were not able to produce acute hepatotoxicity in a short period of time (24, 42). However, the long-term effects of low doses of AOM have been shown to produce steatosis, preneoplastic, and neoplastic hepatic lesions (3). In our study, we found that low doses of AOM to obese mice result in liver injury, and intentional weight loss did not prevent hepatic lesions, instead it seems to exacerbate them.
Scientists have clearly established the remarkable ability of the liver to respond to an injury through a regenerative response. For example, human and rodent hepatocytes exposed to excessive lipid accumulation, as seen in nonalcoholic steatohepatitis, have exhibited a regenerative capacity (10, 18). Furthermore, recent studies have shown activation of STAT3, NF-κB, and Jnk signaling transduction pathways to be increased in liver regenerative processes after partial hepatectomy (13, 19, 40). In the present study, we found a marked increase in BrdU incorporation, cell proliferation, and NF-κB phosphorylation in the liver of H-AOM-HFD mice compared with H-AOM-LFD or Control-HFD groups. Others have reported that NF-κB activation promotes survival of hepatic stem cell differentiation, which facilitates repair-promoting effects (32). Thus the combined increase in BrdU incorporation, PCNA expression, and activation of NF-κB may suggest that H-AOM-HFD mice are capable of inducing liver regeneration, which might be explained by exposure of this group to multiple insults (i.e., HFD and AOM). It's unlikely that STAT3 and Jnk are playing a significant role in liver inflammation in this AOM model, at least at this time point, as there was no difference in hepatitis between the H-AOM-HFD and H-AOM-LFD mice as determined by histopathological analysis. As fibrosis has been seen in patients with hepatocellular carcinoma (23), we also examined fibrosis. In our study, we observed a similar fibrotic score between all groups; thus we do not believe that the difference seen in liver proliferation in the H-AOM-HFD mice was due to liver fibrosis. Further experiments need to be performed using additional mouse models of liver tumorigenesis and liver regeneration to corroborate these findings.
Immune evasion is an emerging hallmark of cancer. A competent immune system is necessary not only for prevention of cancer development but also to slow its progression. Thus we next determined the effects of weight loss on select immune cells in AOM-treated mice. Examination of immune cells in the spleen and lymph node of Control-HFD, H-AOM-LFD, and H-AOM-HFD mice revealed modulation of macrophages and T cells in H-AOM-LFD mice. Specifically, we observed a 33% increase in splenic M2 macrophages (CD206+F4/80+) in H-AOM-LFD mice compared with the other two groups. An interesting study by Yeung et al. (39) observed an elevated number of M2 macrophages (CD163+) in peritumoral tissue of patients with hepatocellular carcinoma (HCC) (39). Additionally, they found that injection of M2 macrophages in an orthotopic tumor was capable of increasing tumor mass and rate of metastasis in MHCC97L tumor-bearing mice. We also observed an increase in lymph node CD4+CD69+ T cells, which also coincided with data shown in another orthotopic model of liver cancer (16, 17). In this model, Han et al. (16, 17) uncovered CD4+CD69+ T cells as a new subset of T cells involved in tumor-induced immunosuppression. Furthermore, patients with HCC have shown a positive correlation between CD4+CD69+ T cells and HCC progression (43). Thus, the increase in spleen CD206+F4/80+ M2 macrophages and mesenteric lymph node CD4+CD69+-activated T cells may be the cause of prominent nuclear atypia in the liver of H-AOM-LFD mice, possibly by promoting liver immunosuppression. Conversely, liver nuclear atypia may be modulating macrophages and T cells in this model. It would have been of further interest to examine immune cell populations in the colon. However, given the limited available tissue, we were not able to run FACS on intestinal tissue. Thus our findings on immune processes in this model are limited.
In summary, this is the first study to examine the effects of intentional body weight loss after chemical exposure to a carcinogen. Our results suggest that intentional body weight loss by diet manipulation does not provide any beneficial effects on colon tumorigenesis and it may in fact aggravate liver capacity of repair. However, further studies are necessary to corroborate the effects of weight loss in the liver and using additional models of liver tumorigenesis to fully understand the mechanisms involved.
This work was supported by National Cancer Institute Grants R21-CA-167058, R21-CA-175636, and R21CA191966 and National Center for Complementary and Alternative Medicine Grant K01-AT-007824 (all to E. Angela Murphy) and R01-CA-121249 from the National Cancer Institute to James A. Carson.
No conflicts of interest, financial or otherwise, are declared by the author(s).
K.T.V., R.T.E., T.L.C., J.B., I.C., U.S., P.S.N., M.N., J.M.D., J.A.C., and E.A.M. conception and design of research; K.T.V., R.T.E., M.S.C., T.L.C., J.B., and U.S. performed experiments; K.T.V., M.S.C., and I.C. analyzed data; K.T.V., R.T.E., M.S.C., T.L.C., J.B., I.C., U.S., P.S.N., M.N., J.M.D., J.A.C., and E.A.M. interpreted results of experiments; K.T.V. prepared figures; K.T.V. and E.A.M. drafted manuscript; K.T.V., R.T.E., M.S.C., T.L.C., J.B., I.C., U.S., P.S.N., M.N., J.M.D., J.A.C., and E.A.M. edited and revised manuscript; K.T.V., R.T.E., M.S.C., T.L.C., J.B., I.C., U.S., P.S.N., M.N., J.M.D., J.A.C., and E.A.M. approved final version of manuscript.
- Copyright © 2016 the American Physiological Society