Mechanisms of chemotherapy-induced muscle wasting in mice with cancer cachexia

Experimental animals

All experiments were approved by the Animal Ethics Committee of The University of Melbourne and conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes as stipulated by the National Health and Medical Research Council (Australia). Male CD2F1 and Balb/c mice were obtained from the Animal Resources Centre (Canning Vale, Western Australia) and housed in the Biological Research Facility at The University of Melbourne under a 12:12 h light–dark cycle. Water and standard laboratory chow were available ad libitum, and consumption was recorded daily.

Effect of 5-fluorouracil chemotherapy in cachectic tumour-bearing mice

Frozen C-26 cells were kindly donated by Prof. Martha Belury (The Ohio State University, Columbus, OH, USA). We have previously described the procedures used to thaw and count the C-26 cells and have characterized the functional and metabolic impairments in mice injected with these cells in detail elsewhere.²⁵

On the day of inoculation (Day 1), 11- to 12-week-old CD2F1 mice were allocated randomly into one of three experimental groups: a control group injected with phosphate-buffered saline (PBS) alone (n = 8); a C-26 tumour-bearing group treated with DMSO vehicle (n = 6); and a C-26 tumour-bearing group treated with 5-FU (n = 7). Mice were anaesthetized via an intraperitoneal (i.p.) injection of a mixture of ketamine (100 mg·kg⁻¹) and xylazine (10 mg·kg⁻¹; VM Supplies, Chelsea Heights, VIC, Australia), such that they were unresponsive to tactile stimuli. Mice were shaved on the dorsal side and given a subcutaneous (s.c.) injection of 5 × 10⁵ C-26 cells suspended in 100 μL of sterilized PBS or 100 μL of sterilized PBS only (control). Mice recovered from anaesthesia on a heat pad and were given an s.c. injection of atipamezole (Antisedan; 1 mg·kg⁻¹; VM Supplies) to partially reverse the effects of xylazine and promote faster recovery from sedation. On Days 15, 18, and 21, lean body mass was assessed (EchoMRI™, Houston, TX, USA) and mice received an i.p. injection of either 60% DMSO vehicle (3.3 μL·g⁻¹ body mass; Sigma-Aldrich, Castle Hill, NSW, Australia) or 5-FU diluted in 60% DMSO (100 mg·kg⁻¹ body mass, 3.3 μL·g⁻¹ body mass; Sigma-Aldrich). The 5-FU dose was based on that used previously to reduce C-26 tumour size²⁶ and predicted to be within the clinical range (50–154 mg·kg⁻¹ lean body mass).²⁷ On Day 22, mice were anaesthetized with sodium pentobarbitone (Nembutal; 60 mg·kg⁻¹; Sigma-Aldrich) via i.p. injection and various hindlimb skeletal muscles [tibialis anterior (TA), extensor digitorum longus, soleus, and plantaris] as well as the epididymal fat, spleen, heart, and tumour were carefully excised, blotted on filter paper, and weighed on an analytical balance. The TA muscles were mounted in embedding medium and frozen in thawing isopentane, and the other tissues were frozen rapidly in liquid nitrogen. All tissues were then stored at −80°C for subsequent analyses. Mice were killed as a consequence of heart excision while still anaesthetized deeply.

Generation and intramuscular injection of rAAV6:eGFP-miR-351-3p sponge in 5-FU-treated mice with cancer cachexia

Methods for the design, cloning, and in vitro validation of the AAV:eGFP-miR-351-3p sponge and scrambled controls are provided in the Supporting Information. Transfection of the AAV:eGFP-miR-351-3p sponge and AAV:eGFP-Scrambled control plasmids with the pDGM6 packaging plasmid into HEK 293 cells generated type 6 pseudotyped viral vectors that were harvested and purified as described previously.²⁸ Briefly, HEK 293 cells were plated on a 10 cm culture dish, 8–16 h prior to transfection with 10 μg of a vector genome-containing plasmid and 20 μg of the packaging plasmid pDGM6, using the calcium phosphate precipitate method to generate pseudotype 6 vectors. Seventy-two hours after transfection, the media and cells were collected and homogenized through a microfluidizer (Microfluidics) prior to 0.22 μm clarification (Millipore). Purified vector was generated from the clarified lysate by affinity chromatography over a HiTrap heparin column (Amersham) and ultracentrifuged overnight prior to resuspension in sterile Hank's balanced salt solution.

For this experiment, we used the more readily available Balb/c mouse model that has an endpoint of 18 days after the injection of C-26 cells, at which time mice exhibit a similar percentage decrease in tumour-free body mass and loss of muscle and heart mass as C-26 tumour-bearing CD2F1 mice at endpoint (21 days after C-26 injection). On the day of inoculation (Day 1), 15- to 16-week-old male Balb/c mice (n = 8) were anaesthetized with isoflurane (induction, 3–4% oxygen–isoflurane at 0.5 L·min⁻¹; maintenance, 2–3% at 0.5 L·min⁻¹), such that they were unresponsive to tactile stimuli. They were then shaved on the dorsal side and given an s.c. injection of 5 × 10⁵ C-26 cells suspended in 100 μL of sterilized PBS. Four days later (Day 5), mice were again anaesthetized with isoflurane, both hindlimbs shaved and the right TA muscle was injected with rAAV6:eGFP-miR-351-3p sponge while the left TA muscle was injected with rAAV6:eGFP-Scrambled sponge at a dose of 5 × 10⁹ vector genomes (determined by optimization pilot studies) in 40 μL of sterilized Hank's balanced salt solution. Mice were given an s.c. injection of buprenorphine (0.05 mg·kg⁻¹) for analgesia and recovered on a heat pad. To match the timing of 5-FU treatments to the CD2F1 model where injections were given every 3 days with the final injection given the day before endpoint, on Days 12, 15, and 18, mice received an i.p. injection of 5-FU diluted in 60% DMSO (100 mg·kg⁻¹ body mass). On Day 19, mice were anaesthetized with sodium pentobarbitone (Nembutal; 60 mg·kg⁻¹) via i.p. injection and the right TA and left TA muscles were carefully excised, blotted on filter paper, and weighed on an analytical balance. The TA muscles were mounted in embedding medium, frozen in thawing isopentane, and stored at −80°C for subsequent analyses. Mice were killed by cardiac excision while still anaesthetized deeply.

Culture, transfection, and treatment of C2C12 cells

Murine C2C12 myoblasts (American Type Culture Collection, Manassas, VA, USA) were seeded in 12-well plates and cultured in growth media containing Dulbecco's modified Eagle's medium (DMEM, Life Technologies, Scoresby, VIC, Australia) supplemented with 10% FBS and 1% l-glut (DMEM/10% FBS/1% l-glut) at 37°C + 5% CO₂. Upon confluency, the media were changed to DMEM containing 2% (v/v) horse serum (HS, Life Technologies) and 1% l-glut (DMEM/2% HS/1% l-glut, differentiation media) for 4 days to induce differentiation during which time the media were changed every 48 h. For experiments assessing the effect of treatment with various chemotherapeutics, and as per our standard procedures,^{29, 30} treatment was initiated on Day 4 of differentiation, at which point we observe fully differentiated myotubes. Cells were then incubated for an additional 48 h at 37°C + 5% CO₂ in fresh DMEM/2% HS/1% l-glut containing 5-FU (5 μM in 0.0013% DMSO) or vehicle (0.0013% DMSO). Pilot dose–response experiments showed that 5 μM was the lowest dose of 5-FU to induce significant and reproducible atrophy.

For miRNA mimic and inhibitor experiments, murine C2C12 myoblasts were seeded in six-well plates and cultured in growth media at 37°C + 5% CO₂. When myoblasts reached ~75% confluency, they were transfected with 25 pmol of miRNA mimic negative control [Applied Biological Materials Inc. (ABM), Richmond, BC, Canada]; mmu-miR-351-3p mimic (ABM); miRNA inhibitor negative control (ABM); or mmu-miR-351-3p inhibitor (ABM) using Lipofectamine RNAiMAX Transfection Reagent (Invitrogen, Carlsbad, CA, USA), according to manufacturer's instructions. At 24 h after transfection, growth media were changed. At 48 h after transfection, the media were changed to differentiation media for 6 days, with media refreshed every 48 h. In these experiments, myotubes were incubated at 37°C + 5% CO₂ for 48 h (Days 4–6) in fresh differentiation media with vehicle control (0.0013% DMSO) or 5-FU (5 μM); vehicle control (0.2% DMSO), oxaliplatin (40 μM, Sigma-Aldrich), or cisplatin (50 μM, Sigma-Aldrich); and leucovorin (10 μg·ml⁻¹, Sigma-Aldrich) or irinotecan (20 μg·ml⁻¹, Sigma-Aldrich).

Statistics

All values are expressed as mean ± SEM unless stated otherwise. Groups were compared using Student's t-test or a one-way or two-way ANOVA, where appropriate. Bonferonni's post hoc test was used to determine significant differences between individual groups. The level of significance was set at P < 0.05 for all comparisons except global miRNA expression profiling where cut-offs for significant changes were a fold-change >1.5 and a P value ≤0.01.

Further details about the Materials and methods section are provided in the Supporting Information.

Cachectic mice are susceptible to chemotherapy-induced muscle wasting

As depicted in Figure 1A, cachectic CD2F1 mice bearing C-26 tumours received three intraperitoneal injections of vehicle (DMSO) or 5-FU (Days 15, 18, and 21) and, 1 day after the final injection, were anaesthetized deeply, and the tumour, various skeletal muscles, and organs were carefully excised and weighed. Lean body mass was assessed prior to each treatment, and calculations revealed that mice received an average 5-FU dose of 117 ± 2 mg·kg⁻¹ lean body mass. No signs of 5-FU-related toxicity such as diarrhoea, palmar plantar erythematosus, mucositis, or ataxia were observed. Cumulative food intake was not different between vehicle-treated and 5-FU-treated mice (Figure S1A), but 5-FU increased cumulative water intake from Day 16 (P < 0.0001; Figure S1B). 5-FU had no effect on the percentage change in tumour-free body mass from the start of the experiment (Figure S1C) but reduced relative tumour mass by 42% compared with vehicle (P < 0.001, Figure 1B). Compared with vehicle, 5-FU treatment reduced mass of the soleus (by −13%, P < 0.01), extensor digitorum longus (by −14%, P < 0.05), plantaris (by −9%, P < 0.05), TA (by −11%, P < 0.05, all Figure 1C), and heart (by −8%, P < 0.03, Figure 1D) but had no effect on mass of the epididymal fat or spleen (Figure S1D).

Tibialis anterior muscle cross-sections were reacted for myosin IIa (N2.261, green) and laminin (red) to identify type IIa fibres and to visualize all fibres (Figure 1E), respectively, for assessment of fibre-type proportions and cross-sectional area. Because mouse TA muscles contain very few detectable type I fibres,³¹ those fibres not reacting to N2.261 were assumed to be type IIx/b fibres. 5-FU induced a fibre-type shift towards more type IIa fibres (+22%, P < 0.05) and fewer type IIx/b fibres (−22%, P < 0.05, Figure 1F) and reduced cross-sectional area of the type IIx/b fibres (−15%, P < 0.05) and average of all muscle fibres compared with vehicle-treated controls (−16%, P < 0.05, Figure 1G). A histogram analysis confirmed that 5-FU reduced the proportion of larger muscle fibres and increased the proportion of smaller muscle fibres in cachectic mice (P < 0.05, Figure 1H).

Chemotherapy-induced muscle wasting is associated with miR-351-3p-dependent ERK2 inhibition

Given that recent studies have demonstrated that cancer cachexia is associated with dysregulated miR-regulated networks,^{20, 32, 33} we performed global miR expression profiling analysis to identify whether miR expression alters with chemotherapy-induced wasting in cachectic mice. Of the 641 miRNAs measured, 313 (49%) were considered not expressed in mouse plantaris muscle and were therefore excluded from further analysis. Using a cut-off for significant changes of a fold-change >1.5 and P value ≤0.01 (q value <0.05), five miRs were differentially expressed (three up-regulated and two down-regulated) in plantaris muscles from 5-FU-treated mice compared with DMSO vehicle-treated mice (Figure 2A). Of the five differentially regulated miRs, only miR-351-3p was significantly increased with 5-FU when analysed by quantitative real-time PCR (qPCR, by 7.6-fold, P < 0.05, Figure 2B). Gene expression of the three up-regulated miRs was also examined in PBS mice and compared with DMSO vehicle-treated C-26 mice to differentiate between effects of cancer cells and chemotherapy. There were no differences between groups, indicating that the increase in miR-351-3p mRNA expression in 5-FU-treated cachectic mice was due to the chemotherapy and not the cancer cells (data not shown).

The gene targets of mouse miR-351-3p (mmu-miR-351-3p) were predicted on the basis of 3′UTR sequence homology, with 2370 potential targets identified, and these were submitted to KEGG analysis of signalling pathways (Table S1). Of known pathways typically associated with regulation of muscle size,³⁴ the MAPK signalling pathway was the top pathway enriched with predicted gene targets of mmu-miR-351-3p, including Mapk1 (ERK2), Mapk14 (p38), and Mapk8 (JNK1) (Figure 2C and Table S1), and so we investigated the phosphorylation status of these proteins in muscles from vehicle-treated and 5-FU-treated cachectic mice (Figure 2D). 5-FU reduced ERK1/2 phosphorylation by 37% (P < 0.05) but had no effect on phosphorylation of p38 or p54/p46 SAPK/JNK (Figure 2E).

We used the C2C12 cell culture system to examine the direct effects of miR-351-3p modulation on myotube size and/or the phosphorylation status of ERK1/2. C2C12 myoblasts transfected with a miR-351-3p mimic and differentiated into mature myotubes resulted in a modest but significant increase in miR-351-3p expression (Figure 2F) and caused a 23% reduction in myotube size (P < 0.0001, Figure 2G) and a 43% decrease in ERK1/2 phosphorylation (P < 0.05, Figure 2H and 2I).

To further investigate the involvement of miR-351-3p in 5-FU-induced muscle atrophy, C2C12 myoblasts were transfected with a miR-351-3p inhibitor or negative control and, after 4 days of differentiation into myotubes, were treated for a further 48 h with 5-FU or vehicle control. The miR-351-3p inhibitor induced a trend for reduced miR-351-3p expression (P < 0.06, Figure 3A) and completely prevented atrophy induced by 5-FU (P < 0.01, Figure 3B). Furthermore, the miR-351-3p inhibitor prevented the 5-FU-induced decrease in ERK1/2 phosphorylation in C2C12 myotubes (P < 0.05, Figure 3C and 3D). To assess whether the findings reflected effects on proliferation, raw cell counts were taken at 24, 48, and 72 h after transfection and revealed very similar growth curves for C2C12 myoblasts transfected with the miR-351-3p inhibitor and inhibitor negative control (Figure S2A and S2B) and no difference between groups for the mean doubling time (Figure S2C). These findings demonstrate that miR-351 inhibition does not affect the proliferative capacity of C2C12 myoblasts.

Given that the MAPK pathway has been associated with increased expression of the ubiquitin ligases MuRF-1 and atrogin-1, their expression was investigated in the experiments described previously. There was no difference in MuRF-1 (P < 0.44) or atrogin-1 mRNA expression (P < 0.48) in muscles from vehicle-treated and 5-FU-treated cachectic mice (Figure S2D). There was also no effect on MuRF-1 and atrogin-1 mRNA expression of the miR-351-3p mimic (Figure S2E) or of either 5-FU or the miR-351-3p inhibitor (Figure S2F).

To determine whether miR-351-3p could directly bind and regulate the Mapk1 3′UTR, HEK 293 cells were co-transfected with a reporter plasmid containing the full-length mouse Mapk1 3′UTR alongside either a miR-351-3p mimic or an irrelevant miRNA lacking the predicted binding site on the Mapk1 3′UTR (negative control). HEK 293 cells were used for this purpose due to their high transfection efficiency. At 24 h after co-transfection, there was a 33% reduction in luciferase activity in the cells containing the miR-351-3p mimic compared with the negative control (P < 0.001, Figure 3E). We then investigated the miR-351-3p putative binding site on the Mapk1 3′UTR by co-transfecting HEK 293 cells with the miR-351-3p mimic or negative control as well as a reporter plasmid with a shorter version of the Mapk1 3′UTR containing only the putative miR-351-3p binding site (WT, Figure 3F). At 24 h after co-transfection, there was a 31% reduction in luciferase activity (P < 0.0001) in cells containing the miR-351-3p mimic compared with the negative control (Figure 3F). However, this reduction was completely prevented when cells were co-transfected with a shorter Mapk1 3′UTR reporter plasmid in which the miR-351-3p binding site was mutated (Mut, P < 0.13, Figure 3F). Taken together, these data indicate that Mapk1 is a direct target of miR-351-3p.

miR-351-3p inhibition increases muscle fibre size in chemotherapy-treated cachectic mice

To test the therapeutic potential of miR-351-3p inhibition for attenuating chemotherapy-induced muscle fibre atrophy, we generated a AAV:eGFP-miR-351-3p sponge (Figure S3A) and AAV:eGFP-Scrambled (control) sponge plasmid. Function of the AAV:eGFP-miR-351-3p sponge was confirmed with transduction of C2C12 myotubes preventing the 5-FU-associated decrease in myotube size observed with the scrambled sponge (−26%, P < 0.0001; Figure S3B). These findings also confirm that effects of the miR-351-3p inhibitor were directly on myotubes and not myoblasts.

Having confirmed the function of the vector in vitro, we generated AAV6 preparations to test whether AAV6-mediated inhibition of miR-351-3p expression could increase muscle fibre size in 5-FU-treated mice with cancer cachexia. For this experiment, we used the more readily available Balb/c mouse model, which, after the injection of C-26 cells, develops tumours at a similar rate and size as the CD2F1 model (Figure S3C) and, at endpoint (Day 19), exhibits a similar percentage decrease in tumour-free body mass (Figure S3D) and loss of muscle and heart mass as C-26 tumour-bearing CD2F1 mice (Figure S3E). A schematic of the experimental timeline is shown in Figure 4A, with intramuscular injection of AAV6 on Day 5 (4 days after C-26 injection) and 5-FU treatments on Days 12, 15, and 18 to match the timing in the CD2F1 model where injections occurred every 3 days with the final injection given the day before endpoint. At 14 days after intramuscular injection into the TA muscles of cachectic C-26 tumour-bearing mice (Day 19), western blotting for GFP confirmed effective transduction of AAV6:eGFP-Scrambled sponge and AAV6:eGFP-miR-351-3p sponge vectors (Figure 4B), and injection of the AAV6:eGFP-miR-351-3p sponge increased ERK1/2 phosphorylation (+18%, P < 0.05, Figure 4B and 4C). H&E staining revealed no evidence of toxicity after intramuscular injection of AAV6:eGFP-miR-351-3p sponge or AAV6:eGFP-Scrambled control sponge (Figure 4D). Immunohistochemistry was used to differentiate between transduced (GFP positive) and non-transduced (GFP negative) fibres and revealed that the average size of fibres transduced with the AAV6:eGFP-miR-351-3p sponge was greater than those transduced with AAV6:eGFP-Scrambled sponge (+15%, P < 0.01, Figure 4E and 4F), due to a reduced proportion of smaller muscle fibres and an increased proportion of larger muscle fibres (P < 0.05, Figure 4G). qPCR was performed to examine changes in expression of the myosin heavy chain (MHC) isoforms and found an increase in MHCIIb mRNA expression (+130%, P = 0.05, Figure 4H) but no change in MHCIIa or MHCIIx mRNA expression with the AAV6:eGFP-miR-351-3p sponge.

Relevance to other chemotherapeutic drugs and human cancer

To gain a better understanding of the therapeutic potential of miR-351-3p inhibition, we examined the ability of the miR-351-3p inhibitor to attenuate muscle wasting in response to other chemotherapeutic agents. These included leucovorin and oxaliplatin, which are often used in combination with 5-FU (Folfox) as well as irinotecan, used in combination with 5-FU and leucovorin (Folfiri). Cisplatin was also examined as it is used to treat numerous human cancers including lung, ovarian, breast, bladder, head and neck, and testicular cancer. None of the chemotherapeutic agents increased endogenous miR-351-3p mRNA expression (Figure S4A and S4B). When added to C2C12 myotubes in vitro, oxaliplatin and cisplatin reduced myotube size by 23% (P < 0.0001) and 28% (P < 0.0001), respectively; effects were completely prevented by miR-351-3p inhibition (Figure S4C and S4D). Irinotecan increased ERK1/2 phosphorylation (by 1.77-fold, P < 0.01), but neither leucovorin nor irinotecan induced C2C12 myotube atrophy (Figure S4E and S4F) and so were not combined with the miR-351-3p inhibitor. Thus, miR-351-3p inhibition may have broad therapeutic potential across multiple and specific chemotherapy regimens.

miR-351-3p is not expressed in the human genome, and so we sought to identify a human miRNA targeting similar genes and pathways. Using miRWalk software and KEGG pathway analysis, we found that 16 of the 27 pathways enriched in predicted gene targets of human miR-125a-3p (hsa-miR-125a-3p) were shared with mouse miR-351-3p (mmu-miR-351-3p, Table S2). Furthermore, the MAPK signalling pathway was the top-ranked pathway enriched in genes predicted to be targets of hsa-miR-125a-3p (Figure 5A and Table S2), including MAPK1, which contains four potential miR-125a-3p binding sites (Figure 5B). To assess the relevance of our findings to humans, we tested whether hsa-miR-125a-3p inhibition could attenuate 5-FU-induced atrophy of human primary skeletal muscle cells (Figure 5C and 5D). Transfection of the hsa-miR-125a-3p inhibitor had no effect on the size of DMSO-treated myotubes (P < 0.69) but completely prevented the 5-FU-induced reduction in muscle cell size and ERK1/2 phosphorylation (P < 0.001; Figure 5C and 5D). Therefore, inhibition of hsa-miR-125a-3p, which is predicted to target similar genes and pathways as mmu-miR-351a-3p, has therapeutic potential for attenuating chemotherapy-induced muscle wasting in patients with cancer cachexia.