The impact of diet-induced obesity on 5 fluorouracil-induced tumor and liver immune cell cytotoxicity

Abstract

Obesity increases the risk for developing several cancers, including colorectal cancer (CRC), and is associated with liver perturbations, which likely impacts treatment tolerance. 5 fluorouracil (5FU) remains a first line treatment for CRC, but efficacy is hampered by interpatient variable responsiveness and off-target toxicities. The current study examined the impact of diet-induced obesity (DIO) on 5FU cytopenia and efficacy using two established CRC models: MC38 (C57BL/6) and C26 (CD2F1). DIO increased tumor size in both MC38 and C26. DIO reduced liver dihydropyrimidine dehydrogenase (dpyd) expression, the enzyme that catalyzes 5FU’s catabolism to become inactive, in MC38 mice, but not in C26. 5FU remained efficacious against early MC38 and C26 tumor growth; however, 5FU-induced tumor and liver immune cell death was exacerbated following three cycles of 5FU with MC38. DIO caused dramatic changes to liver Kupffer cells (KCs), wherein there were increased prometastatic, immunosuppressive KCs in Obese Control and MC38. 5FU, however, depleted these KCs and increased inflammatory KCs in both Lean and Obese MC38. DIO yielded a milder obesity phenotype in CD2F1 mice, and 5FU-induced cytopenia was not different between Lean and Obese. DIO increased total liver KCs; however, C26 tumors increased liver KCs, which were normalized with 5FU treatment, irrespective of DIO. Although 5FU remained efficacious in both models of CRC and did not reduce survival, multiple cycles of 5FU monotherapy increased liver and tumor immune cell death in DIO mice. Altogether, obesity was not protective but rather exacerbated chemotherapy-induced cytotoxicity and promoted a prometastatic liver environment.

NEW & NOTEWORTHY The current study aimed to examine the impact of obesity on tumorigenesis and 5FU safety and efficacy with two established murine models of colorectal cancer. Diet-induced obesity increased tumor burden in both models, and 5FU’s antitumor efficacy remained and extended survival with both tumor models. Obese mice demonstrated increased 5FU-induced immune cell cytotoxicity following multiple cycles of 5FU with distinct changes to liver macrophages, suggesting an increased propensity for liver metastasis.

INTRODUCTION

Colorectal cancer (CRC) remains the second most common cause of cancer death in the United States (1). More than half of CRC cases and subsequent deaths are attributable to lifestyle factors—of which increased body weight is second only to cigarette smoking (2). Indeed, not only does obesity increase the risk for developing CRC but it also leads to poorer outcomes with 5-fluorouracil (5FU)-based adjuvant chemotherapy (3). Despite this, an obesity paradox has been postulated, which suggests that patients with a higher BMI have improved survival (4, 5). Clinical explanations for this have been proposed and include less aggressive tumor biology, better response to treatment, and excess energy reserve (5). However, large participant studies have pushed back against this narrative and suggest that patients with obesity have reduced survival and worsened outcomes (6, 7). Unfortunately, there is a dearth of controlled rodent studies in this arena, precluding any mechanistic support for the obesity paradox.

5FU and 5FU-based chemotherapies remain a first line treatment for CRC and other GI-related cancers. Although survivorship with CRC continues to improve, 5FU’s efficacy is hampered by its narrow therapeutic index and pervasive off-target effects. Among these side effects are cytopenia/neutropenia, leading to anemia and increased susceptibility to infection, bruising, and fatigue (8, 9). 5FU is a fluorinated pyrimidine analog that, upon activation, serves as a thymidylate synthase inhibitor that disrupts cell division (10). 5FU is primarily (>80%) catabolized in the liver by dihydropyrimidine dehydrogenase (DPYD) enzyme to dihydrofluorouracil, which is then excreted in the urine as α-fluoro-β-alanine, ammonia, and carbon dioxide (10). The importance of DPYD in 5FU therapy is driven home with the knowledge that naturally occurring germline, inactivating mutations in the DPYD gene result in unacceptable patient toxicity. Therefore, environmental/dietary factors such as obesity that might impact DYPD expression would also be predicted to impact 5FU therapeutic efficacy as well as its off-target cytotoxicity to normal tissues (11). 5FU has historically been dosed based on body surface area (BSA); however, considerable evidence is compelling a move toward dosing based on lean mass (LM) to improve safety and efficacy (12). Despite this clinical movement, there remains little known about how obesity impacts chemotherapy efficacy or toxicity. Indeed, we previously demonstrated that obese tumor-free mice have reduced liver DPYD protein and gene expression by 40% consistent with reduced survival and increased immune cytotoxicity even when dosed for LM (13), which suggests obesity in fact has an adverse interaction with chemotherapy. In the current study, we sought to dose mice based on their LM and then a reduced dose based on their relative liver DPYD expression (40% reduced dose) (13).

Diet-induced obesity (DIO) is associated with numerous metabolic abnormalities, including impaired glucose homeostasis and liver aberrations [i.e., metabolic dysfunction-associated steatohepatitis (MASH) and metabolic dysfunction-associated steatotic liver disease (MASLD)] (14). Given that 5FU is metabolized by the liver and DPYD expression is highest in hepatocytes, it is therefore reasonable to hypothesize that DIO may impact 5FU safety and efficacy (15). In addition, 5FU itself is hepatotoxic (16, 17); however, whether pre-existing liver dysfunction exacerbates 5FU-induced toxicity is currently unknown. Using two models of CRC (MC38—syngeneic to C57BL/6 mice and C26—syngeneic to CD2F1 mice), we sought to examine the impact of obesity on 5FU anticancer efficacy as well as its cytotoxic effects on the liver and immune cells. First, we examined the impact of DIO on 5FU efficacy in the MC38 model of CRC, wherein MC38 cells are implanted in C57BL/6 mice, which are highly susceptible to DIO and liver perturbations (18). Next, we examined the impact of DIO on 5FU efficacy in the C26 model of CRC, wherein C26 cells are implanted in CD2F1 mice; although reports of DIO in this model are essentially nonexistent, our laboratory has observed that CD2F1 mice are susceptible to DIO but do not demonstrate extensive metabolic dysfunction or liver perturbations. This approach allowed for interrogation of DIO effects on 5FU safety and efficacy with (C57BL/6) and without (CD2F1) extensive metabolic dysfunction. In addition, C26 tumors grow very rapidly and are lethal within 15 days (5–7 days following initial tumor palpation) and results in severe cachexia, whereas the MC38 grows slightly slower, is lethal following 25–30 days (14–20 days following initial tumor palpation), and displays mild cachexia (19, 20). We hypothesized that DIO would increase tumor size in both models and 5FU would mitigate this response. Furthermore, we hypothesized that DIO would exacerbate 5FU-induced tumor and liver immune cell cytotoxicity to a greater extent in the DIO-sensitive C57BL/6 mice-bearing MC38 tumors versus CD2F1 mice-bearing C26 tumors.

METHODS

Animals

Thirty-five male C57BL/6 mice were purchased from Jackson Laboratories, and 35 CD2F1 mice were purchased from Charles River, both of which were purchased at 8 wk of age and housed in the Department of Laboratory Animal Resources at the University of South Carolina. Male mice were used exclusively due to the biological sex differences between male and female mice regarding DIO, as male mice demonstrate more consistent and advanced metabolic perturbations (21), as well as estrogen’s established role in tumorigenesis (22). Mice were group-housed (3–5/cage) and kept on a 12:12-h cycle. Animals were placed on a purified AIN-76A diet and allowed to acclimate to the new facility for 2 wk. Mice were then given a high-fat diet (HFD; Cat#: F3282; Pro: 20.5%, Fat: 36.0%, Carb: 35.7%, 5.49 kcal/g; Bio-serv) for 12 wk (Obese) or maintained on AIN-76A diet (Lean). Animals were given food and water ad libitum throughout the duration of the study, and body weights were measured weekly. All animals were fasted 5 h before tissue collection to control for any acute feeding response. Mice were anesthetized with isoflurane, and tumors and livers were carefully excised, weighed, and either snap-frozen in liquid nitrogen or placed in the appropriate buffers for flow cytometry analysis and histology. Animals were euthanized by isoflurane overdose following tissue excision. All animal experiments were approved by the University of South Carolina’s IACUC.

Experiment 1 Design

Male C57BL/6 mice were randomized into seven groups: Lean Control, Obese Control, Lean MC38, Obese MC38, Lean MC38 + 5FU (35 mg/kgLM), Obese MC38 + 5FU (35 mg/kgLM), and Obese MC38 + 5FU (21 mg/kgLM). MC38 cells were prepared and implanted as previously described (22). Lean and Obese Controls were given a sham injection of PBS. 5FU dosing was completed like previous studies with slight modifications (23). Briefly, before treatment, mice underwent dual-energy X-ray absorptiometry (DEXA) to assess individual LM. 5FU was solubilized in warmed PBS and sterile filtered at 3.5 mg/mL. Once MC38 tumors were palpable (10 days post implantation), 5FU was administered intraperitoneally at 35 mg/kg of LM (35 mg/kgLM) or at 35 mg/kg LM*0.6 (40% reduction based on liver DPYD expression as we previously showed; 21 mg/kgLM) (13) for 5 consecutive days (first cycle). Mice were then allowed to recover for 5 days, then five daily 5FU injections (second cycle), and then recovered for five more days and then a final five daily 5FU injections (third cycle). Tumor size was measured using vernier calipers concomitant with 5FU/PBS injections. Mice were euthanized 24 h following the third cycle. Control mice were given sterile PBS injections.

Experiment 2 Design

Male CD2F1 mice were randomized into seven groups: Lean Control, Obese Control, Lean C26, Obese C26, Lean C26 + 5FU, Obese C26 + 5FU (30 mg/kgLM), and Obese C26 + 5FU (18 mg/kgLM). C26 cells were prepared and implanted, as previously described (24, 25). Lean and Obese Controls were given a sham injection of PBS. 5FU dosing was completed, as previously described (24, 25). Similar to Experiment 1, before treatment, mice underwent DEXA analysis to assess individual LM. 5FU was solubilized in warmed PBS and sterile filtered at 3.0 mg/mL. Nine days post tumor inoculation when the tumors became palpable, 5FU was administered intraperitoneally at 30 mg/kgLM for five consecutive days (1 cycle only). Obese CD2F1 mice did not have differing amounts of liver dpyd gene expression, so this group was excluded from further analysis. Mice were euthanized 24 h following the last 5FU injection. Control mice were given sterile PBS injections.

Blood Analysis

Before anesthesia, fasting blood glucose was assessed via the tail vein. While mice were under anesthesia, whole blood was then collected via the inferior vena cava and transferred to an EDTA-coated vacutainer (Catalog No.: 454428; VWR, Suwanee, GA) and placed on ice. A complete blood count was performed using the VetScan HMT (Abaxis, Union City, CA) for determination of white blood cells and subsets including lymphocytes, monocytes, and neutrophils, as well as red blood cells (RBCs), hemoglobin, hematocrit, and platelets.

Flow Cytometry

Tumor immune cells were isolated, and flow cytometry was completed as previously described (24). To isolate immune cells from the liver, the large lobe of the liver was excised, minced, and enzymatically digested in 5 mL of media containing Collagenase D (0.2 U/mL) and DNase I (100 U/mL) for 30 min at 37°C under gentle agitation. Following enzymatic digestion, suspense was passed through a 70 µm strainer, spun at 50g to pellet hepatocytes. The supernatant was then spun at 700g, and this pellet was resuspended in 40% percoll. Eighty percent percoll was then carefully and slowly added to the base of the 40% percoll suspension, and cells were separated at 2,500g w/o acceleration or brakes to remove RBCs and debris. Cells at the interface were collected and washed, incubated in Fc Block (anti-CD16/CD31, Cat# 101302; BioLegend) before fluorescent antibody staining. Cell surface markers used were as follows: excited by 488 nm laser—CD9-FITC (Cat# 124808; BioLegend); excited by 640 nm laser—CD206-APC (Cat# 141708; BioLegend), CCR2-AF700 (Cat# FAB5538N; R&D Systems), Ly6c-APC/Fire750 (Cat# 128046; BioLegend); excited by 405 laser—CD163-BV421 (Cat# 155309; BioLegend), CD11b-BV570 (Cat# 101233; BioLegend), VSIG4-BV605 (Cat# 749504; BD Biosciences), CD3-BV650 (Cat# 100229; BioLegend), CD11c-BV711 (Cat# 117349; BioLegend), Ly6g-BV785 (Cat# 127645; BioLegend); excited by 561 laser—CD45-PE (Cat# 147712; BioLegend), MHCII-PE/Dazzle594 (Cat# 107648; BioLegend), CX3CR1-PE/Fire700 (Cat# 149052; BioLegend), F4/80-PE/Cy7 (Cat# 123114; BioLegend); excited by 355 laser—NK1.1-BUV563 (Cat# 741233; BD Biosciences), B220-BUV661 (Cat# 612972; BD Biosciences), CD31/Sca1/CD326/Ter119-BUV737 (Cat# 612802, 749199, 4092216, and 741736; BD Biosciences), and TIM4-BUV805 (Cat# 748750; BD Biosciences). Cells were incubated in the aforementioned antibodies for 20 min in the dark at 4°C. Cells were then washed with PBS and incubated in ZombieAqua (Cat# 423101; BioLegend) according to the manufacturer’s instructions. Finally, cells were washed with FBS containing flow buffer and then fixed with 2% PFA and measured with a BD FACSymphony A5 and populations were analyzed with FlowJo V10.10.1.

Liver Histopathology

A separate lobe of the liver was excised and incubated in 10% neutral buffered formalin for 18 h before being paraffin embedded. Sections (5–7 μm) were cut, and slides were used from hematoxylin and eosin staining or Masson’s Trichrome staining, as previously described (13, 26). Images (×20) were taken using a Keyence BZX800 microscope. Evidence of microvesicular and macrovesicular steatosis, inflammation, and hepatocyte hypertrophy was quantified to assess metabolic dysfunction-associated steatohepatitis (MASH), as previously described (27).

Liver RT-PCR

RNA isolation, cDNA synthesis, and RT-PCR were performed, as previously described (13). RNA isolation from another separate lobe of the liver was performed using TRIzol, isopropanol, and chloroform. RNA sample quality and quantities were verified using a Nanodrop One Microvolume UV-Vis Spectrophotometer and determined to be of good quality based on A260/A280 and 260/230 values (>1.9) before cDNA synthesis using High-capacity Reverse Transcriptase kit (Cat# 4368814; Applied Biosystems). Probes for dpyd, TGFβ, MMP2, Col3a1, Col1a1 as well as housekeeping genes Hmbs, Hprt, B2M, TBP, H2afv, and 18s were purchased from Applied Biosystems. Quantitative RT-PCR analysis was carried out as per the manufacturer’s instructions using Taq-Man Gene Expression Assays on a Qiagen Rotor-Gene Q. Data were normalized to Lean Controls and compared with five reference targets (Hmbs, B2M, TBP, H2afv, and hprt), which were evaluated for expression stability using GeNorm.

Statistics

All values shown are means ± SD. Data were graphed, and statistics were run using Prism GraphPad (San Diego, CA). Two-way ANOVAs were used to determine differences between weight status (Lean vs. Obese) and cancer condition (Control, MC38/C26, MC38/C26 + 5FU). Spearman’s test was used to assess the heteroscedasticity of the data, and D’Agostino–Pearson omnibus test was used to assess normality. If either an interaction or a main effect of cancer emerged, a post hoc Tukey’s multiple comparisons test was completed to determine where the differences occurred. In a single situation, a pre-planned t test was used to determine differences in Lean Control versus Obese Control fasting blood glucose (# signifies a pre-planned t test). Throughout the flow cytometry analysis, the data were transformed to assess the difference between tumor bearing and tumor bearing with 5FU. An unpaired t test was used to determine differences between the effects on chemotherapy across lean and obese. *Significant differences across barred groups. Significance was set at P < 0.05.

RESULTS

DIO Reduced Liver DPD and Increased Tumor Weight While 5FU Induced Cytopenia and Anemia in the MC38 Model

Fig. 1A shows the experimental design wherein male C57BL/6 mice were subjected to an HFD (Obese) and injected with 2 × 10⁴ MC38 cells followed by three cycles of 5FU. As expected, DIO increased body weight before tumor inoculation (Fig. 1B). There were no differences in tumor size before the first cycle of 5FU; however, after cycle 1, and as hypothesized, Obese MC38 tumors were significantly bigger than Lean MC38 (Fig. 1C). 5FU’s efficacy was confirmed in the Obese condition as tumor size in Obese MC38 + 5FU (35 mg/kgLM) was smaller than that in Obese MC38 after cycles 1 and 2; however, Obese MC38 + 5FU (21 mg/kgLM) was not different from Obese MC38 establishing the lack of 5FU efficacy when dosing based on liver dpyd expression. The tumors were then excised (postmortem), and DIO increased tumor size regardless of treatment (Fig. 1D). Similar to our previous study, DIO resulted in reduced liver dpyd gene expression (Fig. 1E) (13). In addition, there was a nonsignificant increase in liver dpyd expression between Lean Control and Lean MC38 + 5FU (35 mg/kgLM; P = 0.1) corroborating increased 5FU acquired resistance; however, this was not evident in Obese mice. Fasting blood glucose was elevated with DIO regardless of treatment (Fig. 1F).

Figure 1.

Diet-induced obesity increases MC38 tumor size and weight, while 5FU induced cytopenia. A: experimental design. B: body weight in grams over the course of the study. Red arrow signifies when MC38 cells were implanted, and shaded regions signify 5FU cycles. C: tumor sizes measured following each 5FU cycle in mm². D: tumor weight at euthanasia in mg. E: gene expression of liver dihydropyrimidine dehydrogenase (dpyd) normalized to lean control. F: fasting blood glucose in mg/dL collected immediately before euthanasia via the tail vein. G–N: whole blood vetscan. B and C: Lean Control (closed black circles), Lean MC38 (open circles), Lean MC38 + 5FU (halved circles), Obese Control (closed black squares), Obese MC38 (closed squares), Obese MC38 + 5FU (LM) (halved squares), and Obese MC38 + 5FU (DPD) (closed red squares). D–N: Lean and Obese Control (solid black bars), Lean and Obese MC38 (open bars), Lean and Obese MC38 + 5FU (LM) (hashed bars), Obese MC38 + 5FU (DPD) (red hashed bars). *Statistically significant groups using two-way ANOVA. Lean control n = 3, Lean MC38 n = 4, Lean MC38 + 5FU (35 mg/kgLM) n = 3, Obese control n = 5, Obese MC38 n = 5, Obese MC38 + 5FU (35 mg/kgLM) n = 5, and Obese MC38 + 5FU (21 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. Significance was set at P < 0.05.

A complete blood count was conducted to examine circulatory cells. DIO increased total white blood cells, neutrophils, monocytes, and lymphocytes with MC38; however, as expected 5FU decreased each white blood cell measure, regardless of weight status (Fig. 1, G–J). In addition, both MC38 and 5FU (35 mg/kgLM) decreased RBCs, hemoglobin, and hematocrit, regardless of weight status (Fig. 1, K–M); however, DIO had a greater reduction in hemoglobin with MC38 (Fig. 1L). Interestingly, 5FU increased platelet counts only in Lean mice (Fig. 1N).

DIO Increased 5FU-Induced Immune Cytotoxicity and Altered the Immune Microenvironment of MC38 Tumors

Immune cells were isolated from the tumors of Lean and Obese mice with or without 5FU, and numerous immune cell populations were assessed as well as the relative percent of dead or dying immune cells. Using high parameter flow cytometry, we were able to gate and analyze nine immune cell subsets (Fig. 2A). As expected, 5FU (35 mg/kgLM) increased immune cell cytotoxicity as demonstrated by increased Zombie positivity in CD45⁺ cells (Fig. 2B) regardless of weight status, resulting in a reduction in the relative amount of CD45⁺ cells within the tumor (Fig. 2C). Reducing the dose of 5FU in Obese mice based on liver dpyd expression resulted in a lessening of immune cell death and loss (Fig. 2B). No differences were observed in the relative abundance of T- or B-cells (Fig. 2, D–E). Interestingly, MC38 tumors collectively had a relative high abundance of T-cells (∼30%), supporting its previous identification of being a “warm” tumor (28). Although dendritic cells were greatly reduced after 5FU treatment (Fig. 2F), the relative abundance of antitumoral natural killer cells (NK cells) was increased, regardless of weight status (Fig. 2G). However, weight status affected the magnitude of this response; NK cells were lower in Obese MC38 compared with those in Lean MC38 and Obese MC38 + 5FU (35 mg/kgLM) and Lean MC38 + 5FU (35 mg/kgLM). It is likely that this reduction in NK cells plays a role in the observed increase in tumor size with DIO. Consistent with 5FU-induced neutropenia, 5FU reduced tumoral neutrophils, regardless of weight status (Fig. 2H).

Figure 2.

DIO increased 5FU-induced immune cytotoxicity and altered the immune microenvironment of MC38 tumors. Tumors were excised at euthanasia, and cells were isolated for flow cytometry. A: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺ live immune cells from each sample were concatenated and tSNE cell grouping was completed using CD11b, F4/80, CD206, CD163, CD3, B220, MHCII, NK1.1, CD11c, and Ly6g. B: the relative amount of dead immune cells shown by Zombie⁺ as a percentage of CD45⁺ cells. C: relative amount of live immune cells as a percentage of all single isolated cells. D: the relative percent of B220⁺MHCII⁺ B cells gated from LiveCD45⁺CD11b⁻ lymphocytes. E: the relative abundance of CD3⁺ T cells gated from LiveCD45⁺CD11b⁻MHCII⁻B220⁻ lymphocytes shown as a percentage of LiveCD45⁺CD11b⁻ lymphocytes. F: the relative abundance of CD11c⁺ dendritic cells gated from LiveCD45⁺CD11b⁻MHCII⁻B220⁻CD3⁻ shown as a percentage of LiveCD45⁺CD11b⁻ lymphocytes. G: the relative abundance of NK1.1⁺ Natural killer (NK) cells from LiveCD45⁺CD11b⁻MHCII⁻B220⁻CD3⁻CD11c⁻ shown as a percent of LiveCD45⁺CD11b⁻ lymphocytes. H: the relative percent of CD11b⁺Ly6g⁺ neutrophils gated from LiveCD45⁺ immune cells. I: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ macrophages from each sample were concatenated, and tSNE cell grouping was completed using Ly6c, CX3CR1, CCR2, CD206, CD163, and MHCII. Expression heatmaps for each cell surface protein from low (blue) to high (red) expression. J: the relative abundance of CD11b⁺F4/80⁺ macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺Ly6g⁻ cells shown as a percentage of CD45⁺ immune cells. K: the relative abundance of M2-like CD206⁺MHCII⁻ tumor-associated macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ shown as a percent of CD11b⁺F4/80⁺. L: transformed immune cells data to examine the change in immune cells within lean and obese tumors in response to 5FU. A value of 1 signifies that MC38 vs. MC38 + 5FU were not different within this weight status. >1 signifies and increases in this cell type with 5FU, while <1 signifies and decreases in this cell type with 5FU. B–K: *Statistically significant groups using two-way ANOVA. Lean and Obese MC38 (open bars), Lean and Obese MC38 + 5FU (LM) (hashed bars), and Obese MC38 + 5FU (DPD) (red hashed bars). L: *Statistically significant groups using an unpaired t test. Lean MC38 + 5FU (solid blue bars), and Obese MC38 + 5FU (open blue bars). Lean control n = 3, Lean MC38 n = 4, Lean MC38 + 5FU (35 mg/kgLM) n = 3, Obese control n = 5, Obese MC38 n = 5, Obese MC38 + 5FU (35 mg/kgLM) n = 5, and Obese MC38 + 5FU (21 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. Significance was set at P < 0.05. DIO, diet-induced obesity.

CD11b⁺Ly6g⁻F480⁺ tumoral macrophages were gated and tSNE cell grouping was completed to assess tumor-associated macrophages (TAMs; Fig. 2I). Relative expression of Ly6c, CX3CR1, CCR2, CD206, CD163, and MHCII revealed that most TAMs that are recruited are immunosuppressive (Ly6c^warm, CCR2^warm, CD206^warm, and MHCII^cold; Fig. 2I). 5FU reduced tumoral macrophages, regardless of weight status (Fig. 2J); however, DIO had a reduced relative abundance of CD206⁺MHCII⁻ TAMs, which were further decreased with 5FU (Fig. 2K). Reductions in total macrophages or TAMs were not observed in the Obese MC38 + 5FU (21 mg/kgLM) group implying that dosing based on liver dpyd is not effective in reducing TAMs.

Finally, to examine if there were differences in the response to 5FU in the Lean and Obese condition, data were transformed to look at the relative change between MC38 and MC38 + 5FU (35 mg/kgLM) within each weight status (Fig. 2L). Here, we can appreciate the greater increase in CD45⁺Zombie⁺ and NK cells in the Obese MC38 + 5FU compared with Lean MC38 + 5FU (Fig. 2L). However, relative abundance of total immune cells (CD45⁺), dendritic cells (CD45⁺CD3⁻B220⁻CD11c⁺), neutrophils (CD11b⁺Ly6g⁺), macrophages (CD45⁺CD11b⁺F480⁺), and TAMS (CD45⁺CD11b⁺F480⁺CD206⁺MHCII⁻) were all reduced similarly in Lean and Obese. There were differences in the changes to B cells with 5FU; however, as previously mentioned, neither B cells nor T cells were significantly changed from respective controls (Fig. 2, D and E). Altogether, DIO increases broad 5FU-induced immune cell cytotoxicity but did not dramatically alter relative tumor immune cell abundances compared with Lean.

DIO and 5FU Exacerbated Liver Steatosis in Mice-Bearing MC38 Tumors

The liver is the primary site of 5FU metabolism and is a major organ implicated in DIO and DIO-induced metabolic dysfunction (10, 14). As expected, exposure to an HFD resulted in dramatic increases in overall liver MASH scores (Fig. 3A). Obese MC38 + 5FU (35 mg/kgLM) had increased evidence of a shift from microvesicular to macrovesicular steatosis, but reduced liver inflammation compared with Obese MC38 (Fig. 3A). Furthermore, DIO increased steatohepatitis with mild increases in fibrosis regardless of tumor presence or 5FU administration (Fig. 3B). We confirmed increased liver lipid content with DIO; however, regardless of weight status, MC38 and MC38 + 5FU (35 mg/kgLM) had reduced liver lipid content (Fig. 3C). As expected, evidence of increased steatohepatitis and fibrosis was observed in the Lean MC38 + 5FU compared with Lean Control and Lean MC38 (Fig. 3, B, D, and E); similar trends were observed with DIO but large variability precluded statistically significant findings (29). Increased transcription of collagens, col1a1 and col3a1, was observed with DIO regardless of tumor presence or 5FU administration (Fig. 3, D and E) but within the Lean condition only Lean MC38 + 5FU showed increased expression of these collagens. MC38 reduced MMP2 gene expression and increased MMP9, regardless of weight status (Fig. 3, F and G). Importantly, MMP9 is typically transcribed by leukocytes during liver pathologies (30). Last, 5FU increased Bax expression, which suggests increased liver cell apoptosis with 5FU, regardless of weight status (Fig. 3F).

Figure 3.

DIO and 5FU exacerbated liver steatosis in mice-bearing MC38 tumors. A: liver metabolic dysfunction-associated steatohepatitis scores based on the presence of microvesicular and macrovesicular steatosis, inflammation, and hepatocyte hypertrophy. B: representative images of H&E and Masson’s trichrome (MT) stains at ×20 from each group. Scale bar: 100 μm. C: liver lipid content determined using Folch’s extraction. D–H: liver gene expression of collagens, col1a1 and col3a1, matrix metalloproteinases (MMP) 2 and 9, and apoptosis marker, Bcl-1-associated X (Bax). Lean and Obese Control (solid black bars), Lean and Obese MC38 (open bars), Lean and Obese MC38 + 5FU (LM) (hashed bars), and Obese MC38 + 5FU (DPD) (red hashed bars). Lean control n = 3, Lean MC38 n = 4, Lean MC38 + 5FU (35 mg/kgLM) n = 3, Obese control n = 5, Obese MC38 n = 5, Obese MC38 + 5FU (35 mg/kgLM) n = 5, and Obese MC38 + 5FU (21 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. Significance was set at P < 0.05. DIO, diet-induced obesity; H&E, hematoxylin and eosin.

DIO Increased Liver Immune Cells and 5FU Exacerbated Liver Immune Cytotoxicity

Immune cells were isolated from the large lobe of the liver, and populations were assessed (Fig. 4, A and B). As expected, DIO increased overall CD45⁺ liver immune cells demonstrating hepatic inflammation with MASLD. 5FU decreased CD45⁺ cells, but only in the Obese mice (Fig. 4C). There were no statistical differences in the relative amount of dead (Zombie⁺) CD45⁺ cells across all groups (Fig. 4D). Interestingly, in DIO, both MC38 and MC38 + 5FU (35 mg/kgLM) had decreased CD3⁺ T-cells compared with Control, with no differences observed in Lean mice (Fig. 4E). DIO decreased liver NK1.1⁺ NK cells, CD11c⁺ dendritic cells, and CD11b⁺Ly6g⁺ neutrophils, but increased B220⁺MHCII⁺ B-cells (Fig. 4, F—I). B-cells represent a significant portion of liver lymphocytes and have recently been implicated as a key immune cell regulating DIO-induced liver dysfunction (31). Interestingly, there was an interaction for Lean MC38 + 5FU to have reduced NK cells and neutrophils compared with Lean Control and Lean MC38 (Fig. 4, F and I). CD11b⁺Ly6c⁺ infiltrating/inflammatory monocytes were increased with MC38, regardless of weight status; however, this effect was mitigated with 5FU administration (Fig. 4J).

Figure 4.

DIO increased liver immune cells and 5FU exacerbated liver immune cytotoxicity. The large lobe of the liver was excised at euthanasia, and cells were isolated for flow cytometry. A: gating strategy to examine liver immune cell subsets. B: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺ immune cells from each sample were concatenated, and tSNE cell grouping was completed using CD11b, F4/80, CD3, B220, MHCII, NK1.1, CD11c, and Ly6g. C: relative amount of CD45⁺ immune cells as a percent of ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻. D: the relative amount of dead immune cells shown by Zombie⁺ as a percentage of CD45⁺. E: the relative abundance of CD45⁺CD3⁺ T cells gated from FSC-A × SSC-A lymphocytes shown as a percentage of lymphocytes. F: the relative abundance of NK1.1⁺ Natural killer (NK) cells from LiveCD45⁺MHCII⁻B220⁻CD3⁻CD11c⁻ shown as a percent of LiveCD45⁺CD11b⁻ lymphocytes. G: the relative abundance of CD11c⁺ dendritic cells gated from LiveCD45⁺MHCII⁻B220⁻CD3⁻ shown as a percentage of lymphocytes. H: the relative percent of B220⁺MHCII⁺ B cells gated from LiveCD45⁺CD3⁻ lymphocytes shown as a percentage of lymphocytes. I: the relative percent of CD11b⁺Ly6g⁺ neutrophils gated from LiveCD45⁺ immune cells. J: the relative percent of CD11b⁺Ly6c⁺ inflammatory monocytes gated from LiveCD45⁺Ly6g⁻ immune cells. K: transformed immune cells data to examine the change in immune cells within lean and obese tumors in response to 5FU. A value of 1 signifies that MC38 vs. MC38 + 5FU were not different within this weight status. >1 signifies and increases in this cell type with 5FU, whereas <1 signifies and decreases in this cell type with 5FU. C–J: Lean and Obese Control (solid black bars), Lean and Obese MC38 (open bars), Lean and Obese MC38 + 5FU (LM) (hashed bars), and Obese MC38 + 5FU (DPD) (red hashed bars). Lean control n = 3, Lean MC38 n = 4, Lean MC38 + 5FU (35 mg/kgLM) n = 3, Obese control n = 5, Obese MC38 n = 5, Obese MC38 + 5FU (35 mg/kgLM) n = 5, and Obese MC38 + 5FU (21 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. K: *statistically significant groups using an unpaired t test. Significance was set at P < 0.05. DIO, diet-induced obesity.

To further examine if there were differences in the response to 5FU in the Lean and Obese condition, data were transformed to examine the relative change between MC38 and MC38 + 5FU (35 mg/kgLM) within each weight status (Fig. 4K). Although the relative amount of dead CD45⁺ cells were not different across groups (Fig. 4D), the impact of 5FU (35 mg/kgLM) in Obese was different from the Lean in both the change in CD45 and the change in dead CD45 cells. Furthermore, NK cells increased to a greater degree in Obese livers with 5FU compared with Lean, and there was a greater reduction in CD11b⁺Ly6c⁺ infiltrating monocytes in the Obese compared with Lean (Fig. 4K). All other cell groupings (B-cells, T-cells, cDCs, and neutrophils) were either unchanged from MC38 alone or were not different between Lean and Obese.

DIO Increased 5FU-Induced Kupffer Cell Death and DIO Alone Promotes a Prometastatic, anti-Inflammatory Phenotype

Liver resident macrophages (Fig. 5A), or Kupffer cells (KCs), have an established role in regulating liver function and inflammatory status (32). As expected, CD11b⁺F480⁺ KCs were increased with DIO; however, 5FU (35 mg/kgLM) decreased KCs compared with MC38 and tumor-free controls, regardless of weight status (Fig. 5B). This decrease was not evident with Obese MC38 + 5FU (21 mg/kgLM) given the reduced dose. Interestingly, 5FU (35 mg/kgLM) induced greater KC death in Obese livers compared with Lean (Fig. 5C). We then further interrogated KC using eight macrophage phenotype markers: Ly6c, CCR2, CD206, CD163, TIM4, VSIG4, MHCII, and CX3CR1 (Fig. 5D). We found that a majority of KC were Ly6c⁺CCR2⁺CD206⁺CD163^IntVSIG4^IntMHCII⁻CX3CR1⁺; however, seven distinct populations emerged (Fig. 5D).

Figure 5.

DIO increased 5FU-induced Kupffer cell death and DIO alone promotes a prometastatic, anti-inflammatory phenotype. A: gating strategy to examine liver macrophages. B: the relative abundance of F4/80⁺ macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻ cells shown as a percent of CD45⁺CD11b⁺Ly6g⁻ immune cells. C: the relative amount of dead immune cells shown by Zombie⁺ as a percentage of CD11b⁺F480⁺ macrophages. D: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ macrophages from each sample were concatenated, and tSNE cell grouping was completed using Ly6c, CCR2, CD206, CD163, TIM4, VSIG4, MHCII, and CX3CR1. Expression heatmaps for each cell surface protein from low (blue) to high (red) expression. E–L: the relative abundance of Ly6c⁺, CCR2⁺, CD206⁺, CD163⁺, TIM4⁺, VSIG4⁺, MHCII⁺, and CX3CR1⁺ macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ shown as a percent of CD11b⁺F4/80⁺. M: transformed immune cells data to examine the change in liver immune cells within lean and obese tumors in response to 5FU. A value of 1 signifies that MC38 vs. MC38 + 5FU were not different within this weight status. >1 signifies and increases in this cell type with 5FU, whereas <1 signifies and decreases in this cell type with 5FU. B–L: Lean and Obese Control (solid black bars), Lean and Obese MC38 (open bars), Lean and Obese MC38 + 5FU (LM) (hashed bars), and Obese MC38 + 5FU (DPD) (red hashed bars). Lean control n = 3, Lean MC38 n = 4, Lean MC38 + 5FU (35 mg/kgLM) n = 3, Obese control n = 5, Obese MC38 n = 5, Obese MC38 + 5FU (35 mg/kgLM) n = 5, and Obese MC38 + 5FU (21 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. M: Lean MC38 + 5FU (solid blue bars) and Obese MC38 + 5FU (open blue bars). *Statistically significant groups using an unpaired t test. Significance was set at P < 0.05. DIO, diet-induced obesity.

DIO increased Ly6c⁺, CCR2⁺, CX3CR1⁺, CD206⁺, and CD163⁺ macrophages with no additive effect of MC38. Contrastingly, MC38 increased Ly6c⁺, CCR2⁺, CX3CR1⁺, CD206⁺, and CD163⁺ in the Lean livers. 5FU (35 mg/kgLM) reduced Ly6c⁺, CCR2⁺, CX3CR1⁺, CD206⁺, and CD163⁺ KCs regardless of weight status. Ly6c, CCR2, and CX3CR1 are common markers for macrophage recruitment and infiltration, whereas CD206 and CD163 are traditional KC markers or common among M2-like resident anti-inflammatory profibrotic macrophages. Both TIM4 and VSIG4 are additional KC protein markers and surprisingly, DIO decreased TIM4⁺ but increased VSIG4⁺ cells (Fig. 5, I and J). Contrastingly, MC38 decreased TIM4⁺ and increased VSIG4⁺ in the Lean livers. 5FU increased TIM4⁺ and decreased VSIG4⁺ compared with MC38 regardless of weight status. MHCII, a common inflammatory, antigen-presentation KC marker, mirrored TIM4⁺ demonstrating reduced expression with DIO and MC38 and an increased with 5FU (35 mg/kgLM; Fig. 5L).

Again, to examine if there were differences in the response to 5FU in the Lean and Obese condition, data were transformed to look at the relative change between MC38 and MC38 + 5FU (35 mg/kgLM) within each weight status (Fig. 5M). Dramatic differences can be appreciated between Lean and Obese as it pertained to TIM4 and MHCII; however, the remaining cell types changed similarly between Lean and Obese in respect to 5FU (Fig. 5M). TIM4⁺ KCs are increased with chemical-induced liver injury (33). In addition, MHCII is traditionally expressed on liver capsular macrophages, which are phenotypically distinct from KCs and regulate bacterial immunity (34). Although there was no evidence of liver metastasis (Fig. 3B), the immunosuppressive effects of obesity on liver KCs suggest that these mice may be more susceptible to metastasis.

CD2F1 Given an HFD Display Mild Metabolic Dysfunction but Increased C26 Tumor Burden

Fig. 6A shows the experimental design wherein male CD2F1 mice were subjected to a HFD and given 1 × 10⁶ C26 cells followed by 1 cycle of 5FU. As expected, DIO increased body weight before tumor inoculation (Fig. 6B). Similarly, DIO increased tumor weight; however, 1 cycle of 5FU was able to stop tumor weight increases, regardless of weight status (Fig. 6C) (24). Interestingly, contrasting the C57BL/6, CD2F1 mice exposed to an HFD did not change liver dpyd gene expression (Fig. 6D). Therefore, no 5FU reduced dose group was included in subsequent experiments. There was, however, a mild increase in fasting blood glucose with Obese Control compared with Lean Control (pre-planned t test; Fig. 6E) consistent with our expectation that this model would not demonstrate extensive metabolic dysfunction. There was a main effect for C26 to reduce fasting blood glucose, which was not apparent with C26 + 5FU (30 mg/kgLM), regardless of weight status (Fig. 6E).

Figure 6.

Diet-induced obesity increases C26 tumor size and weight, whereas 5FU induced cytopenia. A: experimental design. B: body weight in grams (g) over the course of the study. Arrow signifies when MC38 cells were implanted, and shaded regions signify 5FU cycle. C: tumor weight at euthanasia in mg. D: gene expression of liver dihydropyrimidine dehydrogenase (dpyd) normalized to lean control. E: fasting blood glucose in g/dL collected immediately before euthanasia via the tail vein. F–M: whole blood vetscan. B: Lean Control (closed black circles), Lean C26 (open circles), Lean C26 + 5FU (halved circles), Obese Control (closed black squares), Obese C26 (closed squares), and Obese C26 + 5FU (LM) (halved squares). C–M: Lean and Obese Control (solid black bars), Lean and Obese C26 (open bars), and Lean and Obese C26 + 5FU (LM) (hashed bars). Lean control n = 3, Lean C26 n = 4, Lean C26 + 5FU (30 mg/kgLM) n = 3, Obese control n = 5, Obese C26 n = 5, and Obese C26 + 5FU (30 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. Significance was set at P < 0.05.

Similar to the MC38 model, C26 increased circulating white blood cells, neutrophils, and monocytes and decreased circulating lymphocytes, regardless of weight status (Fig. 6, F–I). Furthermore, C26 + 5FU (30 mg/kgLM) mice had decreased white blood cells, neutrophils, and monocytes when compared with C26, regardless of weight status. C26 + 5FU (30 mg/kgLM) also decreased lymphocytes compared with control, regardless of weight status (Fig. 6G). Again, similar to MC38, C26 reduced RBC counts and hemoglobin, regardless of weight status demonstrating anemia (Fig. 6, J and K); however, unlike the MC38 model, RBCs and hemoglobin in C26 + 5FU (30 mg/kgLM) mice were not different from control or C26 (Fig. 6, J and K). Hematocrit was decreased in both C26 and C26 + 5FU compared with control, irrespective of weight status (Fig. 6L). As we have previously seen, C26 increased platelet counts (35). Interestingly, 5FU (30 mg/kgLM) decreased platelet counts regardless of weight status; however, Obese C26 were increased compared with Lean C26 and Obese C26 + 5FU (30 mg/kgLM) were increased compared with their Lean counterpart (Fig. 6M).

5FU Increases Tumor Immune Cell Cytotoxicity That Does Not Appear to be Impacted by DIO in the C26 Model

Immune cells were isolated from the tumors of Lean and Obese mice with or without 5FU, and numerous immune cell populations were assessed as well as the relative percent of dead or dying immune cells. Using high parameter flow cytometry, we were able to gate and analyze six immune cell subsets (Fig. 7A). 5FU (30 mg/kgLM) increased the relative percent of dead (Zombie⁺) CD45⁺ immune cell death, regardless of weight status (Fig. 7B); however, the relative abundance of live CD45⁺ immune cells within the tumor was increased in both Lean and Obese C26 (Fig. 7C). There appeared to be no differences in B220⁺MHCII⁺ B-cells with 5FU; however, CD45⁺CD3⁺ T-cells were reduced by 5FU, irrespective of weight status (Fig. 7E). 5FU increased the relative abundance of tumoral CD11c⁺ dendritic cells and NK1.1⁺ NK Cells, regardless of weight status (Fig. 7, F and G). In addition, 5FU (30 mg/kgLM) reduced tumoral CD11b⁺Ly6g⁺ neutrophils, regardless of weight status (Fig. 7H). CD11b⁺Ly6g⁻F480⁺ tumoral macrophages were gated, and tSNE cell grouping was completed to assess TAMs (Fig. 7I). Relative expressions of Ly6c, CX3CR1, CCR2, CD9, CD206, CD163, CD11c, and MHCII revealed 11 distinct macrophage populations (Fig. 7I). Consistent with the MC38 model, 5FU reduced tumoral CD11b⁺F480⁺ macrophages, regardless of weight status (Fig. 7J). Interestingly, Obese C26 had increased relative abundance of CD206⁺MHCII⁺ TAMs compared with Lean C26; however, 5FU (30 mg/kgLM) was able to reduce TAMs, irrespective of weight status (Fig. 7K). Unlike the MC38 model, no differences between Lean and Obese were observed in the response to 5FU demonstrating that obesity did not impact 5FU-induced immune cell cytotoxicity or population changes in CD2F1 mice (Fig. 7L).

Figure 7.

5FU increases tumor immune cell cytotoxicity, which does not appear to be impacted by DIO in the C26 model. Tumors were excised at euthanasia, and cells were isolated for flow cytometry. A: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺ immune cells from each sample were concatenated, and tSNE cell grouping was completed using CD11b, F4/80, CD206, CD163, CD3, B220, MHCII, NK1.1, CD11c, and Ly6g. B: the relative amount of dead immune cells shown by Zombie⁺ as a percentage of CD45⁺. C: relative amount of CD45⁺ immune cells as a percentage of ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻. D: the relative percent of B220⁺MHCII⁺ B cells gated from LiveCD45⁺CD11b⁻ lymphocytes. E: the relative abundance of CD3⁺ T cells gated from LiveCD45⁺CD11b⁻MHCII⁻B220⁻ lymphocytes shown as a percentage of lymphocytes. F: the relative abundance of CD11c⁺ dendritic cells gated from LiveCD45⁺CD11b⁻MHCII⁻B220⁻CD3⁻ shown as a percentage of lymphocytes. G: the relative abundance of NK1.1⁺ Natural killer (NK) cells from LiveCD45⁺CD11b⁻MHCII⁻B220⁻CD3⁻CD11c⁻ shown as a percent of lymphocytes. H: the relative percent of CD11b⁺Ly6g⁺ neutrophils gated from LiveCD45⁺ immune cells. I: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ macrophages from each sample were concatenated, and tSNE cell grouping was completed using Ly6c, CX3CR1, CCR2, CD9, CD206, CD163, CD11c, and MHCII. Expression heatmaps for each cell surface protein from low (blue) to high (red) expression. J: the relative abundance of CD11b⁺F4/80⁺ macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺Ly6g⁻ cells shown as a percentage of CD45⁺ immune cells. K: the relative abundance of M2-like CD206⁺MHCII⁻ tumor-associated macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ shown as a percent of CD11b⁺F4/80⁺. L: transformed immune cells data to examine the change in immune cells within lean and obese tumors in response to 5FU. A value of 1 signifies that C26 vs. C26 + 5FU were not different within this weight status. >1 signifies and increases in this cell type with 5FU, whereas <1 signifies and decreases in this cell type with 5FU. B–K: Lean and Obese C26 (open bars), and Lean and Obese C26 + 5FU (LM) (hashed bars) *Statistically significant groups using two-way ANOVA. L: *Statistically significant groups using an unpaired t test. Lean C26 + 5FU (solid blue bars) and Obese C26 + 5FU (open blue bars). Lean control n = 3, Lean C26 n = 4, Lean C26 + 5FU (30 mg/kgLM) n = 3, Obese control n = 5, Obese C26 n = 5, and Obese C26 + 5FU (30 mg/kgLM) n = 5. Values are means ± SD. Significance was set at P < 0.05. DIO, diet-induced obesity.

C26 Tumors Induced Liver Steatosis That Was Exacerbated by DIO

Ad libitum access to an HFD resulted in a modest increase in liver MASH scores in male CD2F1 mice (Fig. 8A). Interestingly, Lean C26 mice had increased evidence of inflammation, whereas Lean C26 + 5FU (30 mg/kgLM) had no evidence of inflammation but increased microvesicular steatosis (Fig. 8A). Within Obese mice, control mice had evidence of macrovesicular steatosis, mild hepatocyte hypertrophy, and mild inflammation. Interestingly, all Obese C26 tumor-bearing mice regardless of 5FU had increased MASH scores compared with Obese Control (Fig. 8A). There was evidence of a shift from microvesicular to macrovesicular steatosis and increased inflammation with Obese C26 + 5FU (30 mg/kgLM) compared with Obese C26 (Fig. 8A). The increase in MASH with HFD occurred with mild steatohepatitis, but there was no visible evidence of fibrosis; however, C26 as well as C26 + 5FU (30 mg/kgLM) appeared to increase collagen formation in both Lean and Obese mice (Fig. 8B). Evidence of worsened steatohepatitis is evident with Obese C26 and Obese C26 + 5FU (30 mg/kgLM; Fig. 8B). DIO increased liver lipid content, but within Obese, liver lipids were not different with C26 or C26 + 5FU (30 mg/kgLM; Fig. 8C). Interestingly, C26 increased col1a1 and col3a1 liver gene expression, regardless of weight status (Fig. 8, D and E). Although there was no difference in col1a1 expression with Lean C26 + 5FU (30 mg/kgLM), col1a1 remained elevated in Obese C26 + 5FU (30 mg/kgLM; Fig. 8D). Both Lean and Obese C26 and C26 + 5FU (30 mg/kgLM) had elevated col3a1 gene expression compared with Control (Fig. 8E). There was a main effect of C26 and C26 + 5FU (30 mg/kgLM) to increase MMP2 and Bax gene expression compared with control, irrespective of weight status; however, C26 increased MMP9 only in Lean mice (Fig. 8, F–H).

Figure 8.

C26 tumors induced liver steatosis, which was exacerbated by DIO. A: liver metabolic dysfunction-associated steatohepatitis scores based on the presence of microvesicular and macrovesicular steatosis, inflammation, and hepatocyte hypertrophy. B: representative images of H&E and Masson’s trichrome (MT) stains at ×20 from each group. Scale bar: 100 μm. C: liver lipid content determined using Folch’s extraction. D–H: liver gene expression of collagens, col1a1 and col3a1, matrix metalloproteinases (MMP) 2 and 9, and apoptosis marker, Bcl-1-associated X (Bax). Lean and Obese Control (solid black bars), Lean and Obese C26 (open bars), Lean and Obese C26 + 5FU (LM) (hashed bars), Lean C26 + 5FU (solid blue bars), and Obese C26 + 5FU (open blue bars). Lean control n = 3, Lean C26 n = 4, Lean C26 + 5FU (30 mg/kgLM) n = 3, Obese control n = 5, Obese C26 n = 5, and Obese C26 + 5FU (30 mg/kgLM) n = 5. Values are means ± SD. *Statistically significant groups using two-way ANOVA. Significance was set at P < 0.05. DIO, diet-induced obesity; H&E, hematoxylin and eosin.

DIO Increased Liver Immune Cells but Blunted the C26-Induced Increase in Liver Neutrophils

Immune cells were isolated from the large lobe of the liver from each group, and immune cell populations were assessed (Fig. 9, A and B). DIO increased overall abundance of CD45⁺ immune cells in the liver (Fig. 9C). C26 reduced the relative abundance of dead (Zombie⁺) CD45⁺ liver immune cells, regardless of weight status; however, C26 + 5FU (30 mg/kgLM) increased the percent of dead (Zombie⁺) CD45⁺ liver immune cells compared with C26 but was not different from Control, regardless of weight status (Fig. 9D). Interestingly, Obese C26 + 5FU (30 mg/kgLM) had a lower abundance of dead (Zombie⁺) CD45⁺ liver immune cells compared with Lean C26 + 5FU (30 mg/kgLM; Fig. 9D). There was a main effect for DIO mice to have increased relative abundance of T-cells (Fig. 9E), but no differences in B220⁺MHCII⁺ B-cells were observed (Fig. 9F). CD11c⁺ dendritic cells were reduced with Lean C26 compared with Lean Control, and Obese C26 + 5FU (30 mg/kgLM) were increased compared with Obese C26, whereas Obese C26 + 5FU (30 mg/kgLM) were increased compared with Lean C26 + 5FU (30 mg/kgLM) (Fig. 9G). NK1.1⁺ NK cells were reduced with Lean C26 compared with Lean Controls and increased with Obese C26 + 5FU (30 mg/kgLM) compared with Obese C26 (Fig. 9H). Strikingly, CD11b⁺Ly6g⁺ neutrophils were greatly increased with C26, regardless of weight status, which was reduced in C26 + 5FU (30 mg/kgLM). Interestingly, the relative abundance of liver neutrophils in Obese C26 were less than that of Lean C26 (Fig. 9I). No statistical differences were observed with CD11b⁺Ly6c⁺ infiltrating/inflammatory monocytes (Fig. 9J). Similar to tumor immune cells, there were no differences between Lean and Obese in the response to 5FU, demonstrating that obesity did not impact 5FU-induced immune cell cytotoxicity or liver immune cell population changes in CD2F1 mice (Fig. 9K).

Figure 9.

DIO increased liver immune cells but blunted the C26-induced increase in liver neutrophils. The large lobe of the liver was excised at euthanasia, and cells were isolated for flow cytometry. A: gating strategy to examine liver immune cell subsets. B: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺ immune cells from each sample were concatenated, and tSNE cell grouping was completed using CD11b, F4/80, CD3, B220, MHCII, NK1.1, CD11c, and Ly6g. C: relative amount of CD45⁺ immune cells as a percent of ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻. D: the relative amount of dead immune cells shown by Zombie⁺ as a percentage of CD45⁺. E: the relative abundance of CD45⁺CD3⁺ T cells gated from FSC-A × SSC-A lymphocytes shown as a percentage of lymphocytes. F: the relative percent of B220⁺MHCII⁺ B cells gated from Live CD45⁺CD3⁻ lymphocytes shown as a percentage of lymphocytes. G: the relative abundance of CD11c⁺ dendritic cells gated from LiveCD45⁺MHCII⁻B220⁻CD3⁻ shown as a percentage of lymphocytes. H: the relative abundance of NK1.1⁺ Natural killer (NK) cells from LiveCD45⁺MHCII⁻B220⁻CD3⁻CD11c⁻ shown as a percent of LiveCD45⁺CD11b⁻ lymphocytes. I: the relative percent of CD11b⁺Ly6g⁺ neutrophils gated from LiveCD45⁺ immune cells. J: the relative percent of CD11b⁺Ly6c^Hi inflammatory monocytes gated from LiveCD45⁺Ly6g⁻ immune cells. K: transformed immune cells data to examine the change in immune cells within lean and obese tumors in response to 5FU. A value of 1 signifies that C26 vs. C26 + 5FU were not different within this weight status. >1 signifies and increases in this cell type with 5FU, whereas <1 signifies and decreases in this cell type with 5FU. C–J: Lean and Obese Control (solid black bars), Lean and Obese C26 (open bars), and Lean and Obese C26 + 5FU (LM) (hashed bars). *Statistically significant groups using two-way ANOVA. K: Lean C26 + 5FU (solid blue bars) and Obese C26 + 5FU (open blue bars). *Statistically significant groups using an unpaired t test. Lean C26 + 5FU (solid blue bars) and Obese C26 + 5FU (open blue bars). Lean control n = 3, Lean C26 n = 4, Lean C26 + 5FU (30 mg/kgLM) n = 3, Obese control n = 5, Obese C26 n = 5, and Obese C26 + 5FU (30 mg/kgLM) n = 5. Values are means ± SD. Significance was set at P < 0.05. DIO, diet-induced obesity.

Kupffer Cells Are Increased with DIO in CD2F1 Mice, but Individual Macrophage Subsets Were Impacted Similarly with C26 and 5FU

Liver macrophages (Fig. 10A) were increased with DIO in CD2F1 mice regardless of tumor presence or 5FU (Fig. 10B). Interestingly, C26 reduced the percent of dead (Zombie⁺) CD11b⁺F4/80⁺ macrophages compared with Control and C26 + 5FU (30 mg/kgLM), regardless of weight status, but there were no differences between C26 + 5FU (30 mg/kgLM) and Controls (Fig. 10C). We then further interrogated liver KCs using seven macrophage phenotype markers: TIM4, VSIG4, CD9, CCR2, CD206, CD163, and MHCII (Fig. 10D). C26 increased the abundance of TIM4⁺ KCs compared with C26 + 5FU (30 mg/kgLM) and VSIG4⁺ KCs compared with Controls, regardless of weight status (Fig. 10, E and F). C26 + 5FU (30 mg/kgLM) reduced the abundance of CD9⁺ KCs compared with C26 and Controls, irrespective of weight status (Fig. 10G). There was a main effect for C26 to increase CCR2⁺ KCs (Fig. 10H). Although there were no differences in CD206⁺ KCs (Fig. 10I), Lean C26 increased CD163⁺ KCs compared with Lean Controls, which was then reduced back to Control levels with Lean C26 + 5FU (30 mg/kgLM; Fig. 10J). Interestingly, Lean C26 had a greater abundance of CD163⁺ KCs compared with Obese C26, and Obese C26 + 5FU (30 mg/kgLM) had reduced CD163⁺ KCs compared with Obese C26 and Obese Controls (Fig. 10J). Last, Lean C26 had reduced MHCII⁺ KCs compared with Lean Control and Lean C26 + 5FU (30 mg/kgLM; Fig. 10K). There were minimal differences between Lean and Obese in the response to 5FU, where only the increase in MHCII⁺ KCs with 5FU responded differently across Lean and Obese (Fig. 10L), again, demonstrating that DIO had a minimal impact on 5FU-induced immune cell cytotoxicity or liver immune cell population changes in CD2F1 mice (Fig. 10L).

Figure 10.

Kupffer Cells are increased with DIO in CD2F1 mice, but individual macrophage subsets were impacted similarly with C26 and 5FU. A: gating strategy to examine liver macrophages. B: the relative abundance of F4/80⁺ macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻ cells shown as the number of cells collected per 100,000 live cells. C: the relative amount of dead immune cells shown by Zombie⁺ as a percentage of CD11b⁺F480⁺ macrophages. D: ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ macrophages from each sample were concatenated, and tSNE cell grouping was completed using TIM4, VSIG4, CD9, CCR2, CD206, CD163, and MHCII. Expression heatmaps for each cell surface protein from low (blue) to high (red) expression. E–K: the relative abundance of TIM4⁺, VSIG4⁺, CD9⁺, CCR2⁺, CD206⁺, CD163⁺, and MHCII⁺ macrophages gated from ZombieAqua⁻CD31⁻EpCAM⁻Ter119⁻CD45⁺CD11b⁺Ly6g⁻F4/80⁺ shown as the number of cells per 10,000 CD11b⁺F4/80⁺ macrophages. L: transformed immune cells data to examine the change in liver immune cells within lean and obese tumors in response to 5FU. A value of 1 signifies that C26 vs. C26 + 5FU were not different within this weight status. >1 signifies and increases in this cell type with 5FU, whereas <1 signifies and decreases in this cell type with 5FU. B–K: Lean and Obese Control (solid black bars), Lean and Obese C26 (open bars), and Lean and Obese C26 + 5FU (LM) (hashed bars). *Statistically significant groups using two-way ANOVA. L: Lean C26 + 5FU (solid blue bars) and Obese C26 + 5FU (open blue bars). *Statistically significant groups using an unpaired t test. Lean C26 + 5FU (solid blue bars) and Obese C26 + 5FU (open blue bars). Lean control n = 3, Lean C26 n = 4, Lean C26 + 5FU (30 mg/kgLM) n = 3, Obese control n = 5, Obese C26 n = 5, and Obese C26 + 5FU (30 mg/kgLM) n = 5. Values are means ± SD. Significance was set at P < 0.05. DIO, diet-induced obesity.

DISCUSSION

Recently, large sample size epidemiological studies and tightly controlled preclinical investigations have pushed back against the cancer specific “obesity paradox,” wherein obesity was suggested to protect against early mortality and cancer treatment complications (6, 7, 13, 36). The current study aimed to examine the impact of DIO on colon tumorigenesis and 5FU safety and efficacy with two distinct preclinical models: the MC38, syngeneic in C57BL/6 mice, which demonstrate extensive metabolic dysregulation, and the C26, syngeneic in CD2F1 mice, which present with mild metabolic dysregulation. Importantly, DIO increased tumor growth and 5FU remained efficacious in both models. First, 5FU resulted in exacerbated tumor and liver immune cell death with distinct changes to live immune cell populations with DIO concomitant with reduced liver dpyd expression in the MC38 model. Next, 5FU induced tumor and liver immune cell cytotoxicity with C26; however, DIO had no impact on liver dpyd expression, and there was no substantial evidence of exacerbated cytotoxicity in this model. CD2F1 mice demonstrated only mild MASLD in contrast to C57BL/6, which has typically been used for models of DIO. Altogether, our findings not only corroborate DIO’s impact on CRC tumorigenesis but also suggest that a spectrum of obesity-induced perturbations exists and that more severe disruptions to the liver and whole-body metabolism (i.e., MC38) result in increased immune cytotoxicity and likely impaired cancer survivorship.

Despite advancements in CRC care and treatment, 5FU, discovered in 1957 (37), remains a staple of CRC treatment regimens. While often used in combination with other anti-neoplastic drugs, namely irinotecan, leucovorin, and oxaliplatin (FOLFOX, FOLFIRI, and FOLFIRINOX), 5FU remains the backbone of these treatment regimens. Indeed, both oxaliplatin (38) and irinotecan (39) induce immunosuppression by immune cell cytopenia and bone marrow cell cycle arrest, which would only serve to enhance the current study’s findings. In addition, when paired with the anti-folate 5FU, the folate analog leucovorin has been shown to exacerbate cytopenia (40). To our knowledge, before our first study (13), a link between DIO and exacerbated chemotherapy-induced cytopenia has not previously been suggested. However, obese CRC patients have been reported to have an increased prevalence of inoperable blood loss and more wound infections (41). We have now extended our previous work again, demonstrating that DIO exacerbated immune cell death with multiple cycles of 5FU in tumor-bearing mice.

The “Obesity Paradox,” defined typically as the increased risk for developing cancer with obesity but improved survivorship following the cancer diagnosis, remains heavily debated and contentious. Despite this, the link between obesity and cancer remains consistent and there is strong evidence to support that low BMIs (i.e., underweight) are at an elevated mortality risk (42). This has largely been linked to low treatment dosages, lethal cachexia (muscle wasting), and surgical complications (43). Higher BMIs then are thought to allow for increased chemotherapy dosages, a “more weight to spare” phenomena, and better surgical outcomes (44). However, the degree of obesity and severity of metabolic dysfunction within the highest BMI groups shrouds any clear link. A recent study demonstrated that type II diabetes, rather than obesity alone, increased mortality in CRC patients associated with cachexia progression (45). Furthermore, limiting chemotherapy dosages for patients with larger body surface areas (BSAs; 2.0–2.2 m²) is a common practice arguing against that it is the increased relative dosages than improves patient outcomes, but this is also debated (44). In attempt to better personalize dosages, it has been recommended that patients should be dosed based on lean body mass rather than either BSA or body weight, but this has not been ubiquitously adopted (46). The current study aimed to shed light on this complex phenomenon through tightly controlled preclinical studies. Although survival was not different between lean and obese tumor-bearing mice, there was evidence of exacerbated cytopenia in mice with more severe DIO-induced metabolic perturbations; however, limiting the dose of 5FU was not advantageous as the treatment efficacy was diminished. Certainly, more work is needed, clinically and preclinically, to better understand and unravel the complexity of the obesity paradox within the cancer space.

The link between obesity and CRC risk and tumorigenesis is strong (41). Here, we showed that DIO increased tumor burden in two murine models of CRC. Although the exact mechanisms linking DIO and increased tumor burden remain elusive, perturbations to inflammation, IGF-1 signaling, and hormone imbalance are the most likely candidates (47). Furthermore, DIO has demonstrated to disrupt circulating, liver, adipose tissue, and tumoral immune cells, all contributing to a tumor-promoting environment (48). We previously showed that DIO mice had increased tumor size associated with increased TAMs, decreased M1-like macrophages, and decreased apoptosis without apparent changes to proliferation (36). Although DIO increased TAMs in the C26 tumors, TAMs were reduced with DIO in MC38 tumors. Whether this is due to the differences in the duration of tumor burden, or the diversity of the CRC tumor models requires further investigation. Regardless of the models, however, 5FU reduced TAMs. Although we did not see changes in T-cells using the pan T cell marker, CD3, DIO has previously been shown to impact T cell function and metabolism, rather than abundance, contributing to its pro-tumoral environment (49). Moreover, the dysfunction of these tumor infiltrating lymphocytes (TILs) with obesity has been suggested to prime the tumor for immunotherapies, resulting in increased treatment efficacy (adding to the aforementioned obesity paradox) (49, 50). A more thorough and exhaustive T cell specific panel would be needed to rule out unique phenotype changes within lymphocyte populations with the current study. Interestingly, 5FU increased the antitumoral NK cells in both models, regardless of weight status, highlighting a likely mechanism for maintained 5FU antitumoral efficacy. Altogether, 5FU’s efficacy appeared to be similar across models and weight status wherein, 5FU maintains its cytotoxicity, decreases TAMs, and increases the relative abundance of NK cells (51).

DIO is associated with numerous comorbidities, not the least of which is MASLD (formerly termed nonalcoholic fatty liver disease; NAFLD). To the best of our knowledge, no clinical evidence exists to demonstrate DIO is associated with reduced dpyd expression and activity; however, there is a preponderance of clinical evidence demonstrating that patients with dpyd gene mutations experience severe and potentially lethal 5FU side effects (52, 53). The prevalence of DPD deficiency is estimated between 2% and 13% of the population with disparities across sex and race (54). We now have shown decreased liver dpyd expression in male C57BL/6 mice given an HFD for >12 wk with two independent studies (13); however, dpyd gene expression was not different in male CD2F1 mice on an HFD. It is well established that not all mouse strains have similar susceptibility to DIO. We are the first to show CD2F1 mice are susceptible to HFD-induced body weight increases but have much less severe metabolic perturbations (MASH and FBG); however, it should be noted that across experiment, comparisons were not explicitly done. In line with our hypothesis, reducing the dosage of 5FU in MC38 mice based on liver dpyd gene expression reduced the immune cytotoxicity; however, this dose was not efficacious and did not improve survival. Future studies utilizing the CD2F1 mouse strain may prove advantageous to examining a milder obesity phenotype and MASLD, but increased body weight (i.e., BMI).

As the prevalence of obesity continues to increase, MASLD poses a considerable public health problem worldwide affecting up to 32% of the global population (55). Indeed, MASLD has also been demonstrated to impair metabolism wherein drug clearance and catabolism may be diminished (56). Furthermore, both cancer and chemotherapy, independent of weight status, may exacerbate MASLD and liver dysfunction and worsen prognosis (57). To this end, little is known about the impact of DIO, tumor presence, and 5FU administration on liver weight and liver lipid accumulation. Although collagens were increased in C57BL/6 Obese Controls, neither MC38 nor MC38 + 5FU appeared to exacerbate any outcomes related to MASLD or MASH. Rather, liver lipids were reduced with MC38 and 5FU suggesting an increase in liver metabolism and lipid usage. In contrast, DIO increased lipids in the CD2F1 mice, but increased fibrosis and collagen gene expression (Col1a1 and Col3a1) was only evident in nontreated C26 tumor-bearing mice. This was accompanied by increases in liver MMPs and increased apoptosis. Considerable work in this area has largely been focused on liver metastasis (58); however, little is known about the mechanisms driving liver fibrosis and steatosis independent of metastasis. Although the current models did not demonstrate liver metastasis, our findings suggest an increased propensity for liver metastasis.

Liver immune cells have an appreciable role in maintaining liver function and homeostasis while also protecting against or promoting liver metastasis with cancer. As previously stated, the majority of 5FU catabolism occurs in the liver (10), and we hypothesized that altered liver immune cell microenvironment with DIO would impair 5FU metabolism resulting in exacerbated liver toxicity and immune cell toxicity. There was an increase in total immune cells and B-cells with DIO in the C57BL/6 model with reduced NK cells, dendritic cells, and neutrophils. The addition of an MC38 tumor or three cycles of 5FU did not demonstrate significant changes in these immune cells, but rather monocytes were increased with MC38 and then reduced with 5FU. Wang et al. (59) recently demonstrated that HFD-induced fatty livers promote metastasis at least in part due to an increase in immunosuppressive immune microenvironment (i.e., increased liver M2 like macrophages). Although we did not observe evidence of metastasis—owing to flank tumor cell injections versus splenic—we did observe a similar increase in immunosuppressive macrophages in the liver with both the C26 and MC38 tumors with DIO. Regardless of weight status, 5FU decreased M2-like macrophages (CD206⁺, CD163⁺) but increased VSIG4⁺ and MHCII⁺ and dramatically decreased the Ly6c⁺, CCR2⁺, and Cx3CR1⁺ infiltrating macrophages in the C57BL/6. Similar, albeit milder, changes were observed with the CD2F1 model. Altogether, DIO promoted an immunosuppressive liver immune microenvironment increasing the likelihood of liver metastasis.

In the current study, we sought to test the impact of DIO on 5FU immune cell cytotoxicity in the context of liver toxicity and antitumor efficacy. We show that DIO induced MASH and MASLD concomitant with increased immune cell death. Based on our recent investigations, we hypothesized that immune cytotoxicity would be increased with DIO and we observed several instances of increased toxicity; however, these toxicities were not acutely lethal. Consistent with previous reports, DIO increased tumor burden; however, 5FU’s antitumor efficacy remained and extended survival in both Lean and Obese animals when compared with untreated MC38 and C26 mice. These data add to the current body of literature regarding obesity-associated increases in tumorigenesis and the novel study design extend this to show 5FU remains efficacious despite reductions in liver dpyd expression (MC38 model). Although there was evidence of exacerbated cytotoxicity and DIO-specific changes to liver macrophages, these effects did not appear to impact 5FU’s antitumor efficacy. These immune changes to the liver do, however, suggest an increased propensity for liver metastasis, which has been previously reported. Future studies will focus on the antitumor mechanisms with DIO as well as the metabolic manifestations of the obese cancer condition.

DATA AVAILABILITY

Data will be made available upon request.

GRANTS

This work was supported by the National Institutes of Health K99CA276891 (B.N.V.), F31CA278490 (T.D.C.), and U01CA272977 (E.A.M.). This work was also supported by 1S10OD032271 (University of South Carolina’s Instrumentation Resource Facility).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.N.V., M.S., M.D.W., D.F., and E.A.M. conceived and designed research; B.N.V., T.D.C., B.M.B., C.A.U., and J.C.F. performed experiments; B.N.V., T.D.C., B.M.B., C.A.U., J.C.F., R.T.E., and E.A.M. analyzed data; B.N.V., T.D.C., B.M.B., C.A.U., J.C.F., R.T.E., M.S., M.D.W., D.F., and E.A.M. interpreted results of experiments; B.N.V. prepared figures; B.N.V. drafted manuscript; B.N.V., T.D.C., B.M.B., C.A.U., J.C.F., R.T.E., M.S., M.D.W., D.F., and E.A.M. edited and revised manuscript; B.N.V., T.D.C., B.M.B., C.A.U., J.C.F., R.T.E., M.S., M.D.W., D.F., and E.A.M. approved final version of manuscript.