Differential impact of cancer- and chemotherapy-induced cachexia: a comparative analysis in a pre-clinical model of colorectal cancer by biological sex

Abstract

Cancer cachexia is a wasting condition characterized by muscle loss and reduced quality of life in cancer patients. Although biological sex differences in the progression of cancer cachexia have been increasingly recognized, their role during chemotherapy-treated cancer cachexia remains largely unexplored. We evaluated such potential differences using male and female mice implanted subcutaneously with Colon-26 allografts. Our novel approach to disentangle the effects of chemotherapy consisted of administering two cycles of 75% of the maximum tolerated dose of 5-fluorouracil, cisplatin, or paclitaxel, in the presence and absence of cancer (n=6-11/condition). Muscles and organs were collected 25 days following tumor implant, and protein turnover markers, gene expression through RNA sequencing, and mitochondrial function were evaluated in gastrocnemius. A two-way factorial analysis was conducted to assess main effects and interactions across groups (p<0.05). We demonstrate chemotherapy exhibited preserved fat mass in males while maintained body weight in females. Chemotherapy elicited negative impacts on muscle in both sexes, even in a reduced or absent tumor burden. In males, protein synthesis was lower in the presence of cancer and chemotherapy without corresponding differences in atrogenes. Cluster analysis revealed largest differences in muscle transcriptome between cancer control and C26-paclitaxel in male mice, highlighting altered regulation of ubiquitin-mediated proteolysis. These findings point to divergent mechanisms during protein processing in endoplasmic reticulum regulation in muscle atrophy associated with cancer in the presence of chemotherapy. Our results highlight distinct mechanisms underlying cancer-cachexia alone versus the addition of chemotherapy and further indicate responses to chemotherapy differ between biological sexes.

New & Noteworthy

Our study highlights critical differences in the mechanisms mediating muscle loss between cancer and chemotherapy-treated cancer, particularly those by which proteins are degraded and processed. Our data point to biological sex differences in the responsiveness to chemotherapy agents as well as the wasting response. Major differences were observed between sexes, specifically fat mass preservation in male mice and the preservation of body weight in female mice; both despite significant reduction in muscle mass.

Graphical Abstract

Introduction

Between 15 and 37% of patients reporting unintentional body weight loss are diagnosed with cancer (1). Cancer patients with body weight loss of >5% in a 6-month window, or a loss of >2% and a body mass index (BMI) of <20 kg/m² are diagnosed with cancer cachexia (2). Cachexia is primarily defined by muscle loss and often accompanied by weakness and fatigue, irrespective of fat loss, and is present in up to 48% of colorectal cancer patients, and associated with up to 40% of cancer-related deaths (3-8). Chemotherapy is the current standard of care for colorectal cancer patients; however, it significantly decreases patient quality of life (QoL) (9). More than 80% of cancer patients undergoing chemotherapy lose their functional independence and suffer from the drug’s toxicity, which results in adverse effects such as lethargy and fatigue (10). Chemotherapy alone can induce muscle wasting, and in the presence of cancer, chemotherapy may exacerbate muscle wasting, weakness, and fatigue (11-15).

Cancer- and chemotherapy-induced cachexia compromise treatment efficacy, prognosis, and survival rates (15-21). To date, effective therapies to prevent or reverse cachexia are limited, and management typically involves supportive and palliative care, usually starting after the onset of muscle loss. The underlying mechanisms driving cachexia have not been fully characterized, complicating the design of new therapies. Mechanisms of cancer cachexia are known to be heterogeneous, depending on multiple variables such as the type of cancer and biological sex (22). Sex-based variations are evident at early stages of the disease in which females often present delayed onset of muscle wasting compared to males; however, during cancer treatments, any such differences have largely been overlooked (16-19). Thus, biological sex differences during cancer-induced and chemotherapy-amplified cachexia warrant further scientific investigation.

The basic mechanism of most chemotherapeutic drugs involves cell cycle disruption to prevent cancer cells from proliferation (23). 5-fluorouracil (5FU) is a main component of standard first-line therapy in colorectal cancer patients including the chemotherapy cocktails FOLFOX and FOLFIRI (24). 5FU is an antimetabolite disrupting thymine production, thereby altering DNA synthesis and leading to cell arrest and apoptosis (25). Other compounds frequently used include platinum-based drugs, such as oxaliplatin and cisplatin (26). Cisplatin (Cis) induces the intrinsic apoptotic pathway by activating the DNA damage response (27). Cisplatin is effective in treating colon cancer, however, patients are likely to develop resistance to this chemotherapy due to the expression of multidrug resistance transporters (MDRs), ATP7B, and CTR1, among other potential mechanisms (26, 27). Other compounds, such as paclitaxel (PTX), have controversial effects regarding their usage during colorectal cancer treatment (28, 29). PTX is a taxane that inhibits the G2/M phase of the cell cycle, promotes microtubule stability, and induces apoptosis (30). PTX is effective in lung and breast cancers, although it presents low antitumor activity in colorectal cancer due to multidrug resistance (28, 29).

Although chemotherapy helps mitigate tumor burden, around 45% of colorectal cancer patients die despite receiving treatment (31). To date, studies suggest cancer and chemotherapy activate common signaling pathways associated with muscle atrophy (32). Specifically, 5FU- and Cis-based treatments may activate the NF-κB pathway to promote MAPK-dependent muscle atrophy and mitochondrial depletion independent of cancer (33, 34). Disentangling the effects of chemotherapy on skeletal muscle from those of cancer itself presents a critical challenge in the literature. To address this, we aimed to evaluate how chemotherapy and cancer interact relative to development of cachexia across biological sex using a pre-clinical colorectal cancer model. Specifically, to accomplish this goal, our study was designed to examine the impact of: (1) effective chemotherapy facilitating tumor reduction (5FU), (2) low colorectal tumor responsive chemotherapy (Cis and PTX), and (3) chemotherapy administration in the absence of cancer. Our results help unravel the complex mechanisms of muscle atrophy associated with chemotherapy in both sexes to guide future investigations to identify potential therapeutic targets to prevent or ameliorate cachexia symptoms in cancer patients undergoing chemotherapy treatments.

Methods

Animals and cancer intervention

All animal methods were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Arkansas. BALB/c mice (78 males and 71 females) were purchased from Jackson Laboratory (Bar Harbor, ME; Stock No.: 001026). Animals were housed in a temperature-controlled environment and received ad libitum access to water and chow for the duration of the study. At day 0, 8-week-old mice were subcutaneously given 1×10⁶ Colon-26 Carcinoma cells passage 5 (C26, National Cancer Institute (35)). Injections were performed bilaterally, each consisting of 5×10⁵ cells suspended in 100 μL sterile phosphate-buffered saline (PBS) for a total of 1×10⁶ cells, control mice received an equal volume of sterile PBS as outlined in previous work (18). Euthanasia was performed 25 days after PBS or tumor injections, as this was the timepoint biological sex differences were observed in our prior work (18).

Initial sample sizes were estimated using a combination of a priori power analysis and our prior works in cancer-induced cachexia (16, 18). As several animals did not complete study endpoints assessments of phenotypic outcomes were performed to determine if sufficient statistical power had been reached to allow robust statistical conclusions consistent with principles of preclinical research. Each experimental group included 6-11 mice at the study endpoint with a total of 8 groups per sex for all combinations of tumor state and chemotherapy treatment: PBS-Saline (n=10 males, n=10 females), PBS-5FU (n=11 males, n=10 females), PBS-PTX (n=9 males, n=10 females), PBS-Cis (n=11 males, n=10 females), C26-Saline (n=9 males, n=8 females), C26-5FU (n=11 males, n=8 females), C26-PTX (n=6 males, n=8 females), C26-Cis (n=10 males, n=7 females) (further explained below and Figure 1). The exact sample size used for each downstream analysis has now been provided in Supplemental File 1 and is also indicated in the corresponding figure legends. Mice were housed in sex-segregated cages, with four animals per cage. Each cage contained mice of the same cancer status (either C26 or PBS). Within each cage, animals were randomly assigned to one of the treatment conditions: saline, 5FU, PTX, or Cis as to represent all chemotherapy conditions within a cage and reduce potential for cage effects. Animals that did not develop detectable tumors or failed to reach the expected tumor burden range were excluded from downstream analyses and were not included in the reported sample sizes. Specifically, across females, two C26-PBS, two C26-5FU, and three C26-Cis. In males, three C26-PTX, one C26-Cis, and one C26-saline animals. In addition, one PBS-5FU animal was excluded due to an initial bigger body weight outlier. Animals were humanely euthanized, and tissues and organs were collected and weighed at endpoint. Heparinized blood was collected by cardiac puncture, and plasma was separated by high-speed centrifugation. Tissues and plasma were snap-frozen in liquid nitrogen and stored at −80°C until further analysis. The gastrocnemius muscle was utilized for all subsequent muscle analyses. The gastrocnemius was selected due to its mixed fiber type, which provides a better representation of the full musculature and for its susceptibility to muscle loss during cancer.

Figure 1. Experimental design.

Balb/c female and male mice of 8-weeks-old received either PBS or C26 allograft at day 0. They received 2 cycles of 75% of the maximum tolerated dose (MTD) of 5-fluorouracil (5FU), paclitaxel (PTX), or cisplatin (Cis) or saline as vehicle control. Each cycle consisted in 2 injections of 37.5% MTD each. First cycle took place on days 12 and 15 and second cycle on days 19 and 22.

Chemotherapy treatment

Twelve days after PBS or C26 implantation, when tumors were visual and palpable (~ pea size), mice received intraperitoneal (IP) injections of two cycles of 75% of the maximum tolerated dose (MTD) per kg/week of 5FU, PTX, Cis, or sterile saline 0.9% (vehicle control). MTD 75% was chosen to maintain the therapeutic dose between chemotherapeutics and avoid the risk of toxicity (36). To accomplish this, each cycle consisted of two injections per week of 37.5% of the MTD per dose. IP injections took place during the light phase on days 12, 15, 19, and 22 after tumor implantation, regardless of tumor size (Figure 1). Stock solutions of 5FU (MilliporeSigma, Burlington, MA, USA; Cat. No. F6627-1G, CAS No. 200-085-6) and Cis (cis-Diammineplatinum (II) dichloride, MilliporeSigma, Burlington, MA, USA; Cat. No. 232120-50MG, CAS No. 15663-27-1) were prepared at concentrations of 37.5 μg/μL and 2.25 μg/μL, respectively. Based on the compound’s solubility, stock solution of 5FU was dissolved in dimethyl sulfoxide (DMSO) and stock solution of Cis in ultrapure water. Both stock solutions were aliquoted and stored at −20°C until use. On the day of injection, aliquots were thawed and diluted in 0.9% sterile saline (Alpha Teknova, Inc., Hollister, CA, USA; Cat No. S5825) to achieve final concentrations of 3.75 μg/μL for 5FU and 0.225 μg/μL for Cis, corresponding to 37.5% of their respective MTD in mice (5FU MTD = 100mg/kg (37); cisplatin MTD = 6 mg/kg (38)). Each drug was administered at 10 μL per gram of body weight, resulting in final dosages of 37.5 mg/kg for 5FU and 2.25 mg/kg for Cis per injection. Stock concentration of PTX (MilliporeSigma, Burlington, MA, USA; Cat. No. T7191-5MG, CAS No. 33069-62-4) was prepared at 2.5 μg/μL of DMSO. Aliquots were stored at −20°C until use. On the day of injection, 3 μL per gram of body weight was administered without further dilution due to the lack of PTX solubility in water. This dose corresponds to 37.5% of the MTD in mice (MTD = 20 mg/kg (39)) and yields a final dosage of 7.5 mg/kg per injection. For healthy and cancer control groups, mice received 10 μL/g of body weight of 0.9% saline via IP injection on the same day as chemotherapy treatments to approximate volume in chemotherapy-treated mice as vehicle control. Body weight was monitored daily and animals experiencing greater than 10% weight loss in relation to measurements from the previous day were provided a hydration and electrolyte supplement DietGel^® Recovery gel (ClearH2O, Portland, ME, USA; Cat. No. 7206-5022). Humane euthanasia was performed if the body weight loss was > 20%, however, nearly all animals reached the experimental endpoint.

Protein fractional synthesis rate (FSR)

Protein FSR was measured in the gastrocnemius muscle using in vivo labeling with 99.9% deuterium oxide (D₂O) (MilliporeSigma, Burlington, MA, USA; Cat. No. 151882-1L). To attain 2% of D₂O body enrichment, mice were given an IP injection of 20 μL/g of body weight 24 hours before tissue collection and provided 4% D₂O drinking water to maintain plasma enrichment for the 24 hour period, as reported previously (40). Protein-bound alanine enrichment was used as an indicator of alanine incorporation during new protein synthesis and measured using gas chromatography-mass spectrometry (GC-MS; 7890A and 5977A, Agilent) and analyzed as we reported (16, 18, 41). Briefly, to precipitate proteins and remove soluble cytosolic components, ~15 mg of muscle tissue was dissolved in 10% cold trichloroacetic acid (TCA) and homogenized using the Bead Mill 24 homogenizer (Fisherbrand, Thermo Fisher Scientific, Waltham, MA, USA) with the homogenization settings of two cycles of 15 seconds each, displacement D=0.10, and speed setting S=6. The protein pellet was isolated by centrifugation and washed repeatedly with cold TCA to ensure removal of free amino acids. Proteins were then hydrolyzed with 6N HCl to liberate individual amino acids. After hydrolysis, a 3:2:1 solution (methyl-8, methanol, and acetonitrile) was added to determine alanine enrichment (²H labeling of alanine) using the GC-MS. The ratio 100/99 of peak abundances was used to calculate percent enrichment using a regression formula generated by [²H] alanine standards. Deuterium incorporation into protein synthesis is relative to deuterated plasma enrichment. Deuterium from plasma was transferred to acetone using 10 N NaOH and 5% acetone in acetonitrile. Acetone was measured using the GC-MS. FSR of mixed proteins per hour was calculated as %/hour using the equation EA x [EBW x 3.7 x t(h)] – 1 x 100, where EA is the amount of protein-bound [2H] alanine (mol% excess), EBW is the 2H2O in the body water (mol% excess), 3.7 the exchange of 2H between body water and alanine, and t (h) is the time in hours (16, 18, 41).

RNA extraction and real-time PCR (qPCR)

qPCR was used a priori to assess RNA content of specific targets previously validated to be impacted during cancer-induced cachexia (16, 18, 42). For RNA isolation, ~15-20 mg of gastrocnemius muscle was homogenized with TRIzol^™ Reagent (Thermo Fisher Scientific, Waltham, MA, USA; Cat. No. 15596018) and Bead Mill 24 homogenizer (Fisherbrand, Thermo Fisher Scientific, Waltham, MA, USA) with the settings of two cycles of 15 seconds each, displacement D=0.10, and speed setting S=6. Chloroform (100%) was used to accomplish phase separation and RNA precipitation from the aqueous phase was achieved by the addition of 70% diethyl pyrocarbonate (DEPC) ethanol. mRNA was isolated using the RNeasy columns of the PureLink RNA Mini Kit (Thermo Fisher Scientific, Waltham, MA, USA; Cat. No. 12183025) as previously described (18). RNA samples were eluted in RNase-free water. The RNA concentration and purity (260/280 nm ratio > 1.8) were measured using the Take3 micro-volume plate reader and Gen5 software (BioTek Instruments, VT). Samples were stored at −80°C until further use. cDNA was attained by reverse transcription using cDNA SuperScript VILO Master Mix (Thermo Fisher Scientific; Cat No. 11755500) and 1 μg/μL of RNA sample as described (18). Dilutions of 1:100 of cDNA with sterile H2O were made and stored at −20°C until use for qPCR. cDNA quantification through cycle threshold (Ct) was measured using the QuantStudio 3 Real-Time PCR system (Applied Biosystems, MA). PCR reaction was prepared with TaqMan Gene Expression Assay (FAM) (Applied Biosystems, 4331182) using TaqMan probes and TaqMan Fast Advanced Master Mix (Applied Biosystems, 4444558). Amplification was conducted as described (18). As values of 18S (Applied Biosystems, Rn18s ID Mm03928990_g1) were not different between groups, it was used as a housekeeping gene. ΔCt values for each sample were calculated as the difference between the targeted gene Ct value and the 18S RNA Ct value (target– 18S). Final quantification of gene expression was estimated using the ΔΔCt method. ΔΔCt values were calculated by the difference between the sample ΔCt and the average of the group’s ΔCt (ΔCt– average of ΔCt). Relative quantification was calculated as 2^ΔΔCt. The fluoresce-labeled TaqMan probes used in this study were the following: Fbxo32 (Atrogin1; Mm00499523_m1), Trim63 (MuRF1; Mm01185221_m1), and Bnip3 (Mm01275600_g1). All targets were normalized to the 18S Ct value. Corrected final fold change for each group was reported as the relationship between each group’s fold change and the average fold change from the PBS-control group.

RNA sequencing

RNA sequencing of cancer-chemotherapy groups and PBS-Saline controls was performed to query the potential mechanisms of cachexia when cancer is treated by chemotherapy. Given the large number of experimental groups, RNA sequencing was prioritized for groups exhibiting the strongest phenotypic responses; therefore, PBS-chemotherapy groups were not included in this analysis. RNA library preparation and sequencing from gastrocnemius muscle were performed by the Clinical Genomics Center from the Oklahoma Medical Research Foundation (OMRF, USA) on an Illumina platform (NovaSeq X Plus) using 150 bp paired-end reads to a depth of ~20 million reads per sample. Sequence data was provided as FASTQ files. Quality control was performed using FastQC and MultiQC as previously done (43). Raw reads were preprocessed using fastp to remove low-quality bases (Phred score < 25) and automatically detecting and removing adapter sequences from paired-end reads. Reads shorter than 20 bp after trimming were discarded. Transcript quantification and differential gene expression analysis were performed as previously described (43). Briefly, transcript abundances were quantified using Salmon mapping-based mode and imported into R with the tximport package (v 1.36.1) and using Mus musculus mm39 (GRCm39) transcriptome. Gene annotation was obtained from Ensembl 113 (EnsDb.Mmusculus.v113, AnnotationHub ID: AH119358). Transcripts per million (TPM) were generated and raw counts were imported using edgeR. Counts per million (CPM) were calculated and genes with low expression were excluded based on the CPM threshold as previously reported (43). Data were normalized using trimmed mean of M-values (TMM) and the design matrix ~0 + group was utilized to consider cancer and chemotherapy as variables. Each sex was compared to their sex-matched PBS-saline controls. Gene expression differences were calculated using the limma package and voom transformation (43). Significantly differentially expressed genes (DEGs) were identified using an adjusted p-value (false discovery rate, FDR) threshold of 0.05 and without a fold change cutoff for up- and down-regulated genes. Pathway enrichment analysis was performed using the kegga function from the limma package to identify enriched KEGG pathways relative to the background gene set (alpha set at FDR<0.05). Pathview visualizations were generated using the pathview R Package.

Mitochondrial respiration

Mitochondrial oxygen consumption was measured in the white head of the gastrocnemius (white gastrocnemius) with fiber bundles using an Oxygraph+ (Hansatech Instruments, UK) with a method adapted from Min et al., (44) and described by us (18). The white head was specifically selected for this analysis to help ensure a more consistent consideration for the metabolic phenotype of the muscle across all samples. Muscles were placed in ice-cold buffer X (60 mM K-MES, 35 mM KCl, 7.23 mM K2EGTA, 2.77 mM CaK2EGTA, 20 mM imidazole, 0.5 mM DTT, 20 mM taurine, 5.7 mM ATP, 15 mM phosphocreatine, and 6.56 mM MgCl2 at pH 7.1). Fiber bundles were permeabilized in cold saponin (50 μg/mL of buffer X) followed by washes with cold buffer Z (110 mM K-MES, 35 mM KCl, 1 mM EGTA, 5 mM K2HPO4, 3 mM MgCl2, 0.005 mM glutamate, 0.02 mM malate, and 0.5 mg/mL BSA at pH 7.1) (18). Mitochondrial respiration states 2 and 3 (maximal respiration rate) were induced in the permeabilized bundles by adding 5 mM of malate and pyruvate and 0.25 mM of ADP, respectively. State 4 (non-ADP-stimulated, proton leak respiration) was induced by adding oligomycin (10 μg/mL). State 3 was inferred by a 30-second interval of the steepest slope of the oxygen consumption curve after ADP addition, whereas state 4 was assessed during a one-minute interval of the shallowest slope of the curve after oligomycin incorporation. The ratio of states 3 and 4 was calculated to determine mitochondrial respiratory control ratio (RCR) as an index measure of coupled mitochondrial respiration (18).

Statistics

A linear mixed-effects model (LMM) was used to evaluate longitudinal changes over time in daily body weight measurements. Body weight measurements included the baseline (Day 0) value obtained prior to treatment initiation and were not different between groups within the same sex. The function lmer from the lme4 package was used in R (v. 4.5.1), with Day, Group, and their interaction (Day × Group) as fixed effects. The interaction term was included to estimate whether the slope of weight change over time differed across treatments. Mouse ID was included as a random intercept to account for repeated measures within animals. Significance of fixed effects was evaluated using Satterthwaite’s degrees of freedom as implemented in lmerTest, yielding adjusted t statistics and p-values. Estimated marginal means (EMMs) and pairwise contrasts were computed using emmeans R package when needed for post hoc interpretation. All statistical tests were two-sided, and significance was set at p<0.05.

To assess the effects of cancer and chemotherapy treatments on remaining dependent variables, unless otherwise specified, a two-way factorial analysis was performed using a mixed-effects model. Statistical outliers for each variable were identified using the outlier identification function in GraphPad Prism (v.10.6.1) and excluded from the analysis. The dataset was stratified by sex with separate analyses conducted for males and females. The factors included were Cancer (2 levels: PBS or C26) and Chemotherapy (4 levels: Saline, 5FU, PTX, or Cis). A mixed-effects model was employed to evaluate the main effects of cancer and chemotherapy, as well as their interaction (Cancer × Chemo) on the outcome variables. Post-hoc comparisons were conducted using Tukey’s adjustment to control for multiple comparisons between the treatment groups and adjusted p-values (adj.p) are reported throughout. For this statistical approach, significance was denoted at alpha < 0.05. Descriptive statistics, including mean, standard deviation (SD), variance, and standard error (SE), were calculated for each sex and treatment group using the PROC MEANS procedure. The analyses were performed using SAS software (SAS Institute Inc., Cary, NC). Differences between sexes were assessed visually and by comparing their separate outcomes. All data analysis was performed analyzing single measurements of biological replicates. The number of samples used for each assay is described in the Supplemental File 1. Figures were compiled using GraphPad Prism 10.6.1 (GraphPad Software, La Jolla, CA), R (v. 4.5.1), or Biorender Murach, K. (2026). Data are expressed as least square means ± SE.

Differential expression for the RNA sequencing data using limma-voom was detailed in the RNA sequencing section.

Results

Female mice preserve tumor-free body mass but exhibit greater muscle vulnerability to chemotherapy

To characterize the effects of cancer and chemotherapy on the cachexia phenotype, body weights were monitored daily through the duration of the study. In male mice, the LMM showed a day × group interaction (p<0.05) in the slope of weight change over time across treatments in all groups compared to PBS-saline control (Figure 2A and Supplemental File 2A). This interaction suggests that cancer-chemotherapy treated males lost weight (between 0.06 to 0.1 g/day, p<0.001) faster than the PBS-chemotherapy groups (between 0.04 to 0.06 g/day, p<0.0262) and similarly to the cancer control group (0.1g/day, p <0.001) (Figure 2A and Supplemental File 2B). In female mice, the body weight of the cancer control group (C26-saline) was similar to the PBS-control group. Body weight in female mice declined following both the first and second cycles of chemotherapy, with the most pronounced effects observed in the Cis- and PTX-treated groups (Figure 2A, Supplemental File 2C, and Supplemental File 2D). The LMM showed a day × group interaction in the slope of weight change over time across PTX and Cis treatments regardless of cancer status (p<0.01). Female C26-PTX and C26-Cis groups presented body weight losses of 0.7 and 0.9 g/day (p<0.001), respectively; with corresponding losses between 0.4 and 0.5 g/day (p<0.0089) for PBS-PTX and PBS-Cis, respectively (Supplemental File 2). The slope of weight change in the cancer control group (C26-saline) in females was not statistically different from PBS-saline control. To define total body weight changes despite the tumor mass, we examined the tumor-free body weight (TF-BW) as a percentage relative to PBS-Saline at the experimental endpoint (25 days following tumor implantation). In male mice, the TF-BW showed a cancer × chemotherapy interaction (p=0.009). The TF-BW was between 2.2 to 4.5% lower in the PBS-chemotherapy treated groups; however, was not statistically different from PBS-control. In the cancer control group, the TF-BW was 16% (adj.p<0.0001) lower than the PBS-control group, PBS-chemotherapy treated groups (adj.p<0.0017), and C26-Cis (adj.p=0.0192). C26-5FU and C26-PTX were also lower than PBS-control with TF-BW ~10.4% (adj.p<0.0183) lower than the PBS-controls (Figure 2B). Despite the cancer × chemotherapy interaction observed, the C26-Cis group was not different from PBS-control, which is likely tied to the large factorial statistical design within each sex (2X4 ANOVA). In female mice, even though C26-saline, C26-PTX, and C26-Cis had mean TF-BW ~4 to 8% lower than PBS control, they were not statistically different from any of the other groups (Figure 2B). We next examined whether these alterations were accompanied by differences in skeletal muscle mass. Gastrocnemius muscle presented a cancer main effect (p<0.0001) in which the C26 groups had 8.3 to 15.8% lower mass than PBS-control (Figure 2C). Similarly, female mice presented a cancer main effect (p=0.0021) in which cancer groups had 7.4% to 13.1% lower gastrocnemius mass than PBS-control. Female mice also exhibited a chemotherapy main effect (p=0.0256) whereby a difference between 5FU- and PTX-treated groups approached significance (adj.p=0.0662), wherein PTX-treated groups were ~6.5% smaller than the 5FU-treated groups (Figure 2C).

Figure 2. Phenotypic characterization of cachexia.

A) Daily raw body weights of male and female mice. Arrow represents the day of injections. B) Percentage loss of tumor-free body weight (TF-BW) relative to sex-matched PBS controls. C) Percentage difference of gastrocnemius mass loss against the average of PBS control. D) Tumor mass of C26 mice. Groups are marked by different letters the groups are statistically different from one another. Data are expressed as least square means ± SE. Cancer*chemo represents a statistical interaction, Cancer ME represents a main effect of cancer, and Chemo ME represents a main effect of chemotherapy with the detailed differences below. Sample sizes are provided in Supplemental File 1.

Tumor masses were assessed as a marker of the efficacy of chemotherapy in reducing tumor burden. In males, all three chemotherapeutic regimens exhibited lower tumor burden. The tumor masses were 1.4 g for the C26-saline group and significantly smaller (~0.7 g, adj.p<0.005) in all C26-chemotherapy-treated groups (Figure 2D). In females, poor responsiveness to chemotherapy closely reflected our expectations, particularly PTX. Tumors in females were 0.9 g in the C26-saline group, 0.4 g in C26-5FU, 1.0 g in C26-PTX, and 0.5 g in C26-Cis. Tumors in C26-PTX were significantly larger than the C26-5FU tumors (p=0.0287) (Figure 2D). Overall, these findings suggest a sex-dependent divergence in susceptibility to chemotherapy-induced cachexia and anti-cancer efficacy, whereby cancerchemotherapy-treated males exhibit skeletal muscle loss comparable to only tumor-bearing animals despite having tumors of approximately half the size, whereas females display a heightened chemosensitivity characterized by muscle loss irrespective of tumor presence.

Chemotherapy leads to multi-organ alterations in female mice with preserved fat mass in male mice

Given that cachexia is a systemic syndrome with multi-organ involvement, we next assessed the phenotypic changes in non-skeletal muscle tissues. Fat mass presented a cancer × chemotherapy interaction (p=0.002) in male mice, in which the C26-saline group was 69.6% (adj.p<0.0001) lower than the PBS-control group and statistically different from all other groups, except C26-PTX (Figure 3A). In female mice, a chemotherapy main effect (p=0.0464) was observed in fat mass; however, no significant differences between groups were noted, although the mean fat mass of the 5FU groups was slightly greater than the PTX groups (adj.p=0.0638). Fat mass in female mice also showed a near-significant cancer main effect for lower fat mass in cancer groups (p=0.0505) (Figure 3A). Spleen mass, a surrogate marker for systemic inflammation, exhibited a cancer × chemotherapy interaction in both male and female mice (p<0.0001). In males, C26-Saline was 114.6%, C26-5FU 52.2%, and C26-PTX 80.2% bigger than PBS-control (adj.p<0.0001); the statistical significance of the difference between C26-Cis and PBS-control was marginally achieved (−32.7%; adj.p=0.0559). Moreover, the spleen of the C26-Cis was 46.2% (adj.p=0.0023) greater than the PBS-Cis and smaller than the C26-Saline (~38.2%, adj.p<0.0001) and C26-PTX (~26.4%, adj.p=0.0037) (Figure 3B). In female mice, the spleen mass of the C26-saline group was 126.7% larger than the PBS-control and statistically different than all other PBS groups and the C26-5FU (adj.p≤0.0001). In addition, the spleen mass from the C26-5FU group was 64.0% (adj.p=0.0102) bigger than the PBS-Cis group and the C26-PTX group was ~92% greater than all the PBS groups (adj.p<0.0001) (Figure 3B). The gonad mass (testicles) in male mice was not statistically affected by cancer or chemotherapy, while in female mice there was a cancer main effect (p=0.0001) in which the gonads (ovaries) of the C26 groups were ~26% smaller than the PBS groups. A significant main effect of chemotherapy was observed in ovarian mass (p=0.0362). No significant pairwise differences were detected among the chemotherapeutic treatments. The cisplatin-treated groups exhibited a trend toward lower ovarian mass compared with both the saline-treated (p=0.0771) and 5FU–treated groups (p=0.0813) (Figure 3C). The heart mass of male mice had a cancer × chemotherapy interaction (p=0.0012) in which the heart mass was smaller in PBS-PTX (11.3%, adj.p=0.0455), C26-Saline (~14.4%, adj.p=0.0053), and C26-5FU (−16.5%, adj.p=0.0002) compared to PBS-Control (Supplemental File 3). C26-5FU heart mass was 11.8% smaller than PBS-5FU (adj.p=0.021). No statistical differences were observed in the heart of female mice (Supplemental File 2). Liver mass of female mice presented a cancer main effect (p=0.0181) with masses ~8.7% bigger in the C26 groups than the PBS groups, while no statistical differences were observed in male mice (Supplemental File 3). Collectively, chemotherapy was associated with preservation of fat mass in males and treatment-specific reductions in splenomegaly in cancer-bearing mice of both sexes, while exerting measurable effects on the ovaries but not on the testicles.

Figure 3. Systemic characterization of cachexia.

Raw A) Fat, B) Spleen, and C) Gonads mass (testicles for male or ovaries for female mice). Data are expressed as least square means ± SE. Groups are marked by different letters the groups are statistically different from one another. Cancer*chemo represents a statistical interaction, Cancer ME represents a main effect of cancer, and Chemo ME represents a main effect of chemotherapy with the detailed differences below. Sample sizes are provided in Supplemental File 1.

Chemotherapy-induced cachexia alters protein synthesis but not atrogenes expression

Efficient protein turnover is essential for maintaining muscle plasticity. We measured protein FSR and mRNA content of commonly measured markers of muscle atrophy (atrogenes) in gastrocnemius muscle. FSR in male mice presented a cancer × chemotherapy interaction (p<0.0001) in which FSR was 37.5% (adj.p<0.0001) lower in PBS-Cis, 33.2% (adj.p<0.0001) in C26-saline, 22.0% (adj.p=0.0123) in C26-5FU, 64.0% (adj.p<0.0001) in C26-PTX, and 59.0% (adj.p<0.0001) in C26-Cis compared to PBS-control. In addition, C26-PTX and C26-Cis FSR was ~35-40 % less compared to PBS-Cis (adj.p<0.0161) (Figure 4A). Female mice had a cancer × chemotherapy interaction (p=0.0007) in which PBS-5FU was ~ 98-151% higher than all other groups, with no additional statistical differences (Figure 4A). Atrogin1 (Fbxo32) in male mice exhibited a cancer × chemotherapy interaction (p=0.0023) with the mRNA content in C26-saline being 6.2-fold higher (adj.p<0.0001) than PBS control and different than all other groups, except for C26-PTX. C26-PTX was 5.9-fold greater than PBS-5FU (adj.p=0.0280) (Figure 4B). In females, Atrogin1 mRNA content presented a cancer × chemotherapy interaction (p=0.0235) with ~2.6-fold greater content in the C26-saline group than all other groups (adj.p<0.0181) (Figure 4B). MuRF1 (Trim63) mRNA content displayed a similar trend to Atrogin1, with a cancer × chemotherapy interaction in male (adj.p=0.0236) and female mice (adj.p=0.0155) (Figure 4C). mRNA content of MuRF1 in C26-saline male mice was ~6.4-fold higher than PBS groups (adj.p<0.0005) and ~2.8-fold greater than C26-5FU and C26-Cis (adj.p<0.0138). C26-PTX was ~5.4-fold greater than PBS-saline, PBS-5FU, and PBS-Cis (adj.p<0.0343). In female mice, C26-saline MuRF1 values were ~2.2-fold higher than all other groups (adj.p<0.0208) without any other statistical significances (Figure 4C).

Figure 4. Protein turnover.

A) fractional synthesis rate (FSR, %/h) after 2% of body enrichment of deuterium oxide (D2O). Data are presented as percentage difference relative to PBS control group and expressed as least square means ± SE. B) mRNA content of atrogin1 (Fbxo32) and C) MuRF1 (Trim63) for gastrocnemius muscle. Groups are marked by different letters the groups are statistically different from one another. Cancer×chemo represents a statistical interaction. Sample sizes are provided in Supplemental File 1.

C26 and PTX Distinctly Modulate Protein Processing Pathways in Male Mice

We next sought to identify the potential mechanisms mediating cancer-induced cachexia in the presence of chemotherapy utilizing RNA sequencing. Due to the lack of a strong cachectic phenotype in PBS-chemotherapy groups, sequencing was limited to cancerchemotherapy treated groups and compared to PBS-saline controls. To account for the 11% of variance attributable to sex observed in the PCA plot, differential expression analysis was performed using within sex-matched comparisons (Supplemental File 4). The summary of the analysis of gene expression differences is shown in Figure 5. The total number of differentially expressed genes (DEGs) was markedly higher in males than in females (Figure 5A). Despite the cancer main effect observed on the percent change in gastrocnemius mass, in both sexes, the greatest number of DEGs were observed in the C26-saline and C26-PTX groups. These groups also exhibited the highest proportion of unique DEGs (Figure 5B). Despite that we used 75% of the MTD, we cannot rule out that the higher amount of DEGs for C26-PTX is due to systemic toxicity, as this compound is rarely used in clinical practice and demonstrates limited clinical translatability for colorectal cancer treatment. Comprehensive DEG lists, including group-specific and shared gene sets are represented in the Venn diagrams, and included in the Supplemental File 5.

Figure 5. Differential expression analysis.

A) Total number of up- and down-regulated differentially expressed genes (DEGs, FDR < 0.05) compared to their sex-matched PBS-saline control. B) Venn diagram representation of unique and common shared DEGs. RNA-sequencing analyses were limited to PBS-saline and C26 tumor-bearing groups. Sample sizes are provided in Supplemental File 1.

Our cluster analysis revealed C26-saline and C26-PTX groups had the strongest variations, particularly in males (Figure 6A). The functional annotation of the clusters showed differences in Ubiquitin mediated proteolysis, Proteasome, and Lysosome pathways, among others (Figure 6A, Supplemental File 6). To gain further insight into the differences between these two groups, we performed KEGG pathway enrichment analysis. In C26-saline males, Metabolic pathways, Thermogenesis, and Chemical carcinogenesis – reactive oxygen species pathways were downregulated, while Ubiquitin mediated proteolysis was the only downregulated pathway in C26-PTX males (Figure 6B). On the other hand, all the upregulated pathways annotated in male C26-PTX were also annotated in male C26-saline, including Lysosome and Coronavirus disease – COVID 19 (Figure 6B, Supplemental File 6). The cancer control group exhibited upregulation in additional pathways, such as Autophagy – animal, Proteasome, and NF-kappa B signaling pathway, among others. Given that the lysosome represents the terminal degradative step of autophagy and considering the concurrent downregulation of protein synthesis observed with chemotherapy in males, we decided to explore the proteostasis regulation toward lysosome-dependent degradation pathways in C26-saline and C26-PTX (Figure 7). Our Pathview analysis of the protein processing in the endoplasmic reticulum (ER) pathway in C26-PTX revealed a downregulation in markers of the protein folding quality control system, such as calnexin (CNX), UDP-glucose:glycoprotein glucosyltransferase (UGGT), and molecular chaperones such as Hsp40. Downregulation in the ER-Associated Degradation (ERAD) pathway and structural components of the Skp1–Cullin–F-box (SCF) E3 Ubiquitin ligase complex, such as RBX1, Cul1, and Skp1 were also depicted. These alterations could be associated with a suppressed proteasome-mediated protein degradation and a shift toward alternative degradation pathways. Interestingly, all of these processes are upregulated in C26-saline Pathview, along with others such as the Unfolded protein response (UPR) (Supplemental File 7). Overall, our findings suggest a compensatory response to ER stress and an attempt to eliminate aberrant proteins during cancer, while chemotherapy (PTX) presents a global repression of ER protein processing.

Figure 6. Hierarchical organization and functional analysis.

A) Hierarchical cluster analysis. B) Downregulated and C) upregulated KEGG pathways in C26-saline and C26-PTX of differentially expressed genes compared to PBS-saline male mice. RNA-sequencing analyses were limited to PBS-saline and C26 tumor-bearing groups. Sample sizes are provided in Supplemental File 1.

Figure 7. Overview of protein processing in endoplasmic reticulum.

Schematic adaptation from the Pathview-rendered KEGG pathway shown in Supplementary File 7. The general mechanisms of protein folding, quality control, and degradation is shown. Detailed pathway topology and gene-level changes are provided in the Pathview render. Directionality of differential expression is summarized in the heatmap where purple represents upregulation, green downregulation, and unchanged direction marked by an x.

Sex-dependent effects of cancer and chemotherapy on mitochondrial function

To determine whether the observed molecular alterations translated into impaired cellular energy metabolism and mitochondrial quality control, mitochondrial respiration (RCR) and Bnip3 expression were assessed (Figure 8). The RCR in male mice showed a cancer × chemotherapy interaction (p=0.0034), in which PBS-PTX was ~55-133% higher (adj.p<0.0214) than the other groups. In female mice, the ratio analysis yielded nonsignificant results. In addition, mRNA content of Bnip3 in male mice exhibited a cancer × chemotherapy interaction (p=0.0088) in which C26-saline was ~2.4-fold (adj.p<0.0001) higher than PBS groups and ~1.8-fold larger than the C26-5FU and C26-Cis groups (adj.p<0.0003). In female mice, Bnip3 mRNA content presented a cancer main effect (p=0.0006) with the C26 groups showing more than ~1.31-fold change than the PBS groups. Collectively, these results suggest cancer as the predominant driver of alterations in mitochondrial function, whereas chemotherapy exerts a more limited effect, in a sex-dependent manner.

Figure 8. Mitochondrial Function.

A) Respiratory control ratio (RCR) of gastrocnemius muscle. Data are expressed as least square means ± SE. B) mRNA content of Bnip3 for gastrocnemius muscle. Groups are marked by different letters the groups are statistically different from one another. Cancer×chemo represents a statistical interaction and ME a main effect of cancer. Sample sizes are provided in Supplemental File 1.

Discussion

Despite the known potential for cancer and chemotherapy to independently induce cachexia, the combined effects remain poorly understood. Understanding chemotherapy-induced cachexia in pre-clinical models remains particularly challenging, as the anti-neoplastic effects of these compounds may differ in the absence of cancer or necessitate high chemotherapeutic doses capable of inducing cytotoxicity, potentially limiting clinical translatability. In studies of chemotherapy alone, body weight loss and cachexia have typically been observed at high doses of chemotherapy (34, 45-49). On the other hand, cancer-induced cachexia in preclinical models commonly relies on tumor burden. Therefore, studying cachexia resulting from chemotherapy treatment during the presence of cancer can result in confounded outcomes, as the treatment may reduce tumor burden, leading to a mild tumor model and consequently attenuating cachexia-related alterations as a result of study conditions (49). As such, we aimed to utilize an assortment of chemotherapies including standard of care components (5FU) and agents reported to be less effective for colorectal tumors (PTX and Cis). Our data highlight critical differences in the likely mechanisms mediating muscle loss between cancer and chemotherapeutics, particularly those by which proteins are degraded and processed. The current study demonstrated biological sex differences in the wasting response when cancer is considered in the context of different chemotherapeutic agents, as well as the responsiveness to chemotherapy agents.

While our initial hypothesis on tumor response to chemotherapeutics was validated in female mice, it was not supported in males. In male mice, chemotherapy attenuated tumor burden regardless of the agent used, whereas females did not display an anti-cancer effect of PTX, indicating a potential sex-specific variation in therapeutic efficacy within this model. Despite notable differences in tumor size, male mice receiving anti-cancer treatment exhibited a similar degree of muscle mass loss to cancer alone while almost completely preserving fat mass, suggesting chemotherapy may protect adipose loss at the cost of a higher induction of muscle loss relative to tumor burden. In female mice, the cancer-chemotherapy cachectic phenotype was consistent with findings observed in cancer alone (in the absence of chemotherapy) (18), in which total body weight was maintained during chemotherapy despite significant reductions in muscle, adipose, and gonad masses. However, at this time we lack sufficient data to speculate on the potential mechanisms herein regarding the selective loss of adipose in females. While the present study was conducted in a pre-clinical colorectal cancer model, clinical evidence from PDAC patients indicates that survival among those with fat wasting is comparable to patients with combined muscle and fat loss, supporting a central role for adipose tissue in cachexia-related mortality that warrants further investigation (50). In the present study, female mice presented a progressive decline in daily body weight in groups treated with Cis and PTX regardless of cancer status. Our findings indicate female mice may be more vulnerable to chemotherapy-induced cachexia, in agreement with previous findings which demonstrate lower tolerability to chemotherapy with more adverse drug reactions in female mice and cancer patients (51-54). Data from our laboratory and others point to a sex-specific cachectic phenotype especially during early stages (16, 18, 19), and cessation of estrous cycling in female mice has been associated to severe cachexia (55). Despite greater recent attention to biological sex differences during cancer cachexia the exact mechanisms underlying those differences remain largely unknown. Survival is also affected and variations between sexes have been reported during chronomodulated therapies due to potential differences in the molecular circadian clocks (56). Our outcomes of exacerbated impacts of chemotherapy in female mice appear to be translationally relevant and may provide a model to explore chemotherapy toxicities therein. Current dosing of anti-cancer treatments is not sex-specific and is usually administered based on body weight or body surface area (51). Future studies may aim to explore alternative dosing indicators, such as lean mass, to enhance treatment tolerability in the female population (51, 57).

Our results regarding skeletal muscle protein anabolism point to sex-specific differences in mechanisms of muscle loss during the chemotherapeutic regimen. While FSR was low in males, with apparent exacerbation by PTX and Cis, females did not show significant effects. Although no significant, there was a trend toward reduced FSR in females across most chemotherapy-treated groups. The exception was the PBS-5FU group, which exhibited unexpectedly higher FSR values. This finding should be interpreted with caution, as the present study does not provide a mechanistic explanation for this observation. This sex-specific difference in protein synthesis recapitulates our previous findings in cancer alone (18). In both sexes, as has been consistently reported, mRNA expression of Atrogin1 and MuRF1 was induced in tumor-bearing animals (16, 18). By contrast, when chemotherapy was concurrently administered, this induction of common atrogenes was not observed (except for C26-PTX male mice), suggesting the mechanism of muscle wasting under chemotherapy-treated cancer differs from that induced by cancer alone (34). These findings are consistent with previous publications where common atrogene expression in chemotherapy-treated mice was not different from controls, highlighting that any induction of catabolism by chemotherapy is not reliant upon these classic muscle E3 ligases but rather upon alternate mechanisms (33, 58). The higher expression of atrogenes in C26-PTX males supports the notion that the mechanisms of cachexia during chemotherapy-treated cancer are compound specific.

As evidenced previously, cancer triggered ubiquitin-mediated proteolysis, UPR, and lysosome and proteasome activity in the muscle. Here, we interpret this response as a compensatory ER stress mechanism intended to remove aberrant proteins, which fails to function effectively during antineoplastic therapy. In conjunction with the a priori assessment of atrogenes, our RNA sequencing data provide evidence of a suppressed proteasome-mediated protein degradation when treated with chemotherapy with a concurrent global repression of ER protein processing altering protein folding quality control systems. Even though our RNA sequencing was limited to chemotherapy+cancer-induced cachexia, this is consistent with the downregulation or the lack of change reported in the ubiquitin-proteasome-dependent in other models of chemotherapy-related cachexia (33, 59). Using proteomic analyses, Barreto et al. (32) showed that during chemotherapy treatment the mechanisms activating proteasome-mediated proteolysis differ by approximately 40% from the classical pathways involved in muscle protein degradation. Additionally, cancer cachexia–associated muscle atrophy has been linked to activation of ER stress and the UPR system (60). Tumor-derived signals can activate ER stress–mediated pathways contributing to muscle atrophy and apoptosis (60). This meaningful distinction between muscle wasting mechanisms during the presence or absence of chemotherapy points to cellular protein-processing regulatory processes that at this moment we cannot be certain as to how they may differ in more mature animals, opening new avenues for future research. Notably, as these were young animals we cannot be certain as to whether these effects reflect a true muscle atrophy compared to impairments in normal growth.

Although only subtle alterations in mitochondrial function were observed in our current study, evidence from prior studies with extended chemotherapy schedules in the absence of cancer suggests mitochondrial activity may become compromised over time (33). Chemotherapy significantly upregulates mitophagy regulator Bnip3 expression during high chemotherapy doses (61, 62). Here, Bnip3 had a statistical main effect of cancer in female mice, and in males it was elevated in C26-saline and C26-PTX. Bnip3, implicated in autophagic pathways, reinforces the relevance of the autophagy-related markers identified in our study as a potential mechanism of catabolism during cachexia in chemotherapy-treated cancer. This observation suggests potential relevance of Bnip3 targeting as we recently assessed in cancer alone (63). During cancer alone, females tend to better preserve mitochondrial quality and function at early disease stages and depending on cancer type (16, 18). Given these prior observations, longer experimental time courses and disease severity may be required to better resolve these sex-specific mitochondrial adaptations over disease progression and chemotherapy.

Conclusion and limitations

Our findings suggest divergent mechanisms of cachexia especially during the protein processing in endoplasmic reticulum in muscle atrophy associated with cancer in the presence of chemotherapy compared to cancer alone. However, functional validation, temporal progression, and causality of these molecular alterations require further exploration. Our results further underscore the need for sex-specific dosing strategies and improved indicators to optimize chemotherapy efficacy while minimizing adverse effects. Current tolerability thresholds are mainly based on studies focused on the male population (51), emphasizing the need of more trials investigating chemotherapy dosing and cachexia effects in female populations (64). Future studies should validate these findings in older mice, as the observed atrophy may partly reflect impaired developmental growth. Although necessary for experimental control, the use of a constant chemotherapy dose in our study does not fully capture the clinical reality, where treatment regimens are regularly adjusted to meet patient needs. Chemotherapy was administered during the light phase (inactive period), and prior work indicates that chronomodulated treatment can influence therapeutic response in a time-of-day–dependent manner, with potential sex-specific effects driven by circadian clocks (56). In addition, incorporating tumor burden–matched controls will be important to account for the contribution of tumor size to atrophy. Despite these limitations, our findings provide a framework for future studies aimed at identifying therapeutic targets to improve treatment outcomes.

Supplementary Material

Supplemental Files available at (https://figshare.com/s/a411f7a12f47091c12c2) DOI 10.6084/m9.figshare.31931304

Data Availability Statement

Raw sequencing data are available in Gene Expression Omnibus: GSE325415. Other data will be made available upon reasonable request.