Introduction

Extracellular vesicles derived from cells of every type, including tumour cells, can be released in the circulation and convey functional information to distant sites. On the basis of their origin, size, physical properties, and functions, different types of extracellular vesicles have been described, among which exosomes and microvesicles (MVs).1

Exosomes (30–150 nm) form within the intraluminal vesicles generated in the multivesicular bodies as part of late endosomes. Multivesicular bodies can fuse with lysosomes for degradation or with the plasma membrane, releasing exosomes. Sorting of cargo during the internal budding of the membrane that leads to intraluminal vesicle formation is an essential step in exosome biogenesis.1

Microvesicle (0.1–1 μm) generation starts from the formation of outward buds in specific sites of the plasma membrane followed by fission and subsequent vesicle release into the extracellular space. In addition to rearrangements in the plasma membrane composition, proteins responsible for cell shape maintenance may be involved in MV biogenesis.1

Extracellular vesicle content consists, in variable proportions, of proteins, DNA, messenger RNAs (mRNAs), microRNAs (miRs), lipids, and surface molecules. In particular, miRs enclosed into extracellular vesicles are released into the bloodstream or other body fluids to transfer specific paracrine or endocrine messages thus altering gene expression of distant recipient cells.

Few observations suggest that extracellular vesicles are involved in the pathogenesis of cancer-induced muscle wasting.2 Indeed, MVs derived from lung or pancreatic cancer cells induce myoblast apoptosis. These MVs contain miR-21 that, activating Toll-like receptor 7 on murine myoblasts, promotes apoptosis.3 Exosomes produced by osteosarcoma cells have been shown to contain factors able to activate the notch-dependent signalling in muscle progenitors, resulting in impaired myogenesis.4 More recently, muscle wasting in tumour-bearing mice has been shown to result from exosomal release of heat shock proteins,5 while extracellular vesicles derived from Lewis lung carcinoma (LLC) cells have been reported to induce protein degradation in C2C12 myotubes and lipolysis in 3T3-L1 adipocytes.6

Muscle wasting, typically occurring in cancer patients, is one of the main features characterizing cancer cachexia, a complex syndrome associated with reduced survival, poor quality of life, and reduced tolerance to anticancer treatments.

Cancer-induced muscle protein depletion is mainly associated with hyperactivation of proteasome and lysosome-dependent degradation, although reduced protein synthesis occurs as well. Besides enhanced protein turnover, muscle wasting in cancer hosts also reflects an energy crisis mainly resulting from impaired oxidative capacity due to mitochondrial dysfunction.7 Indeed, a proteomic profiling performed on the muscle of mice bearing the C26 colon carcinoma has shown an altered expression of proteins involved in energy homeostasis and mitochondrial function.8 Consistently, mitochondria in the muscle of tumour-bearing animals display ultrastructural and functional changes,8, 9 ultimately resulting in decreased mitochondrial number and impaired oxidative capacity, as shown by reduced succinate dehydrogenase (SDH) activity and proportion of oxidative fibres in the muscle of tumour bearers.9 Altered mitochondrial fusion, fission, and biogenesis, frequently associated with a pro-inflammatory environment, have also been reported.10 Consistently with the observations above, markers of mitophagy and mitochondrial dynamics are altered in cachectic cancer patients, a picture that is further complicated by anticancer treatments.11

The mechanisms underlying muscle energy crisis in cancer cachexia are still unclear. The present study is aimed to investigate the occurrence of a causal connection among TMVs, muscle wasting, and derangements of energy metabolism in the skeletal muscle. The study demonstrates that the exposure to TMVs markedly affects myotube oxidative metabolism and that TMVs infused into healthy mice partially recapitulate some of the features of cancer cachexia.

Methods Cell cultures

C2C12 and LLC cell lines were purchased from American Type Culture Collection, and C26 cells were obtained from Prof M. P. Colombo (National Cancer Institute, Milano, Italy). Yoshida AH-130 (AH-130) ascites hepatoma cells were a gift from Prof U. Del Monte (University of Milano, Italy). NIH-3T3 fibroblasts were kindly provided by Prof P. Porporato (University of Torino, Italy).

Murine C2C12, C26, LLC, and NIH-3T3 cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) (Sigma-Aldrich) supplemented with 10% foetal bovine serum (FBS), 100 U/mL penicillin, 100 μg/mL streptomycin, 100 μg/mL sodium pyruvate, and 2 mmol/L l-glutamine and maintained at 37°C in a humidified atmosphere of 5% CO2 in air. C2C12 myoblast differentiation was induced by shifting subconfluent cultures to DMEM supplemented with 2% horse serum (differentiation medium). For MV isolation, cells were maintained in either serum-free medium (C26, LLC) or 5% MVs-free FBS (NIH-3T3) for 48 h.

Plasmid cloning and cell transfection

Plasmids overexpressing miR-21-5p, miR-148a-3p, and miR-181-5p were designed and cloned using BLOCK-iT™ Pol II miR RNAi Expression Vector Kit (K4937, Invitrogen) following manufacturer instructions. Plasmids were then transfected to either NIH-3T3 fibroblast or C2C12 myoblasts using Attractene Transfection Reagent (301005, Qiagen). Cells transfected with a plasmid coding for a SCR were used as control. The day after cell transfection, C2C12 myoblasts were shifted to differentiation medium and myogenic gene expression analysed after 2 and 4 days. As for NIH-3T3 fibroblasts, they were shifted to DMEM containing 5% MVs-free FBS. After 48 h, the conditioned medium was processed for MV isolation (3T3-MVs) as described below.

Animals

Experimental animals were cared for in compliance with the Italian Ministry of Health Guidelines (n°63/2014-B; 20/02/2014) and the Policy on Humane Care and Use of Laboratory Animals (NRC 2011). The experimental protocol was approved by the Bioethical Committee of the University of Torino. Male Wistar rats weighing approximately 150 g and Balb-c mice weighing approximately 20 g (Charles River Wilmington, MA) were maintained on a 12:12 h dark–light cycle with controlled temperature (18–23°C) and free access to food and water during the whole experimental period. Tumour-bearing mice (n = 8) were subcutaneously inoculated on the back with 5 × 105 C26 cells whereas tumour-bearing rats (n = 8) received an intraperitoneal inoculum of 108 AH-130 cells.

Tumour-bearing mice and rats were sacrificed under isoflurane anaesthesia 14 and 7 days after tumour transplantation, respectively. Total blood was obtained by cardiac puncture and processed to separate the plasma, subsequently stored at −80°C.

Microvesicle isolation

Microvesicles were isolated from (i) conditioned medium obtained from 250 × 106 C26 or LLC cells cultured for 48 h, (ii) plasma derived from healthy mice and rats, and (iii) plasma derived from C26 host or rats bearing the AH-130 hepatoma. Conditioned medium or plasma was centrifuged (300g, 10 min) to eliminate cell debris. The supernatant was recovered, and successive centrifugations at increasing speed were performed: one at 2000g for 20 min to eliminate dead cells, then one at 10 000g for 30 min to remove cell debris, and finally one at 100 000g for 70 min to pellet MVs. The resulting pellet was washed in phosphate-buffered saline (PBS) and ultracentrifuged again at the same speed (Supporting Information, Figure S1A). The pellet was resuspended in 1 mL PBS. Nanosight characterization of isolated particles revealed a population ranging from 90 to 600 nm, with a peak of abundancy at 200 nm (Figure S1B), thus confirming the enrichment in MVs.12 MV internalization into C2C12 myotubes was assessed by fluorescence microscopy using the PKH26 fluorescent dye (Sigma-Aldrich).

Treatment with microvesicles C2C12 myoblasts

Tumour-derived microvesicles were added to the culture medium of C2C12 myoblasts during 6 days from the beginning of differentiation. Within this period, cultures were analysed at 2, 4, and 6 days, roughly corresponding to differentiating myoblasts, early differentiated, and fully differentiated myotubes, respectively. Comparable experiments were performed exposing C2C12 myoblasts to 3T3-MV for 2 and 4 days after shifting to differentiation medium.

C2C12 myotubes

Fully differentiated C2C12 myotubes were treated with C26-derived or LLC-derived MVs for 24, 48, and 72 h.

In vivo studies

Healthy Balb-c mice were infused (i.v.) daily with a suspension of MVs derived from healthy mice or rats (used as controls) or from C26 tumour-bearing mice or AH-130 bearing rats. As for dosage, all the MVs obtained from 1.5 mL of blood (for both mice and rats) were infused each day in each animal.

Extracellular flux analysis

Mitochondrial oxygen consumption rate (OCR) was assessed by using the Mito-Stress Test Kit in a Seahorse XF Extracellular Flux Analyser (Agilent Technologies, Santa Clara, CA, USA) according to the manufacturer instructions. C2C12 myotubes were treated or not with LLC-MVs or C26-MVs for 24 h. One hour prior to assay, cultures were incubated in Seahorse XF Base Medium (Agilent Technologies) containing 1 mM Na-pyruvate, 2 mM glutamine, and 25 mM glucose in the absence of CO2 supplementation for 1 h at 37°C. Basal respiration was monitored prior to addition of 2 μM oligomycin. The OCR values were obtained at baseline and after the addition of 2 μM oligomycin, 1 μM carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone), and 0.5 μM rotenone. OCR was automatically calculated and recorded by the Seahorse XF software (Agilent Technologies), and data were expressed as picomoles per minute.

Lactate assay

Lactate was measured in the cultured media of C2C12 myotubes treated or untreated with TMVs using the Lactate Colorimetric/Fluorimetric Assay kit (BioVision) according to the manufacturer's instructions.

Primary cultures

Balb-c mice were sacrificed under anaesthesia, and extensor digitorum longus and soleus muscles were rapidly excised and digested with type I collagenase (2 mg/mL, Sigma C0130) for 1 h at 37°C. Digested muscle were plated on matrigel-coated dishes (Sigma E1270) in DMEM supplemented with 20% FBS, 10% horse serum, 0.5% chick embryo extract, and 1% penicillin–streptomycin, allowing satellite cells release and attachment to the matrigel. Three days later, the medium was replaced with proliferation medium (DMEM 20% FBS, 10% horse serum, and 1% chick embryo extract). After 5 days, the medium was replaced with differentiation medium (2% horse serum and 0.5% chick embryo extract in DMEM). TMVs were added to primary cultures 48 h after the beginning of differentiation.

Protein synthesis

To assess the relative rate of protein synthesis, SUnSET methodology was used.13 Briefly, 1 μM puromycin was added to the C2C12 culture medium 30 min before trypsinization. The amount of puromycin incorporated into nascent peptides was detected by western blotting, using a specific anti-puromycin antibody.

Western blotting

Total protein from C2C12 monolayers or from muscle tissues were extracted in RIPA buffer (50 mM Tris-HCl at pH 7.4, 150 mM NaCl, 1% NP40, 0.25% Na-deoxycholate, and 1 mM PMSF) with freshly added protease and phosphatase inhibitor cocktails, sonicated, and centrifuged at 3000g for 5 min at 4°C, and then the supernatant was collected. Protein concentration was assayed using BSA as working standard. Equal amounts of protein were heat denaturated in sample-loading buffer (50 mM Tris-HCl, pH 6.8, 100 mM DTT, 2% SDS, 0.1% bromophenol blue, and 10% glycerol), resolved by SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad). The filters were blocked with Tris-buffered saline containing 0.05% Tween and 5% non-fat dry milk and then incubated overnight with antibodies directed against: PGC-1 alpha (rabbit polyclonal antibody, Millipore), Cytochrome C (mouse antibody, BD Pharmingen), SDH-A (mouse monoclonal antibody Santa Cruz, CA), Caspase3 (rabbit polyclonal antibody, Sigma-Aldrich), myosin heavy chain (MyHC; mouse monoclonal antibody, clone MY32, Sigma-Aldrich), myogenin (mouse monoclonal antibody, clone F5D, Santa Cruz, CA), Pax7 (Developmental Studies Hybridoma Bank, University of Iowa), MyoD (rabbit polyclonal antibody, clone M318, Santa Cruz Biotechnology, CA), p-ERK (Thr202/Tyr204, rabbit monoclonal antibody, clone 20G111, Cell Signaling), ERK (goat polyclonal antibody, clone sc-94, Santa Cruz, CA), puromycin (mouse monoclonal antibody, clone 12D10, Millipore), and tubulin (mouse monoclonal antibody, clone T5168, Sigma-Aldrich). Peroxidase conjugated IgG (Bio-Rad) was used as secondary antibodies. Quantification of the bands was performed by densitometric analysis using a specific software (TotalLab, NonLinear Dynamics, Newcastle upon Tyne, UK).

Immunofluorescence

C2C12 monolayers were washed with PBS and fixed in 3% paraformaldehyde. Samples were then probed with the anti-MyHC antibody MF-20 (Developmental Studies Hybridoma Bank, Iowa City, IA, USA). Detection was performed using a fluorescein isothiocyanate-conjugated mouse or rabbit IgG secondary antibody. Nuclei were stained with Hoechst 33342 fluorochrome, and the images were captured using an epi-illuminated fluorescence microscope (Axiovert 35; Carl Zeiss MicroImaging GmbH, Jena, Germany).

MitoTracker Red CMXRos

Mitochondria polarization was assessed using the MitoTracker Red CMXRos selective probe (Molecular Probes, Barcelona, Spain). C2C12 cells were incubated in a medium without FBS and containing 100 nM MitoTracker for 30 min at 37 °C. The medium was then replaced with a complete medium without MitoTracker. Images were captured using an epi-illuminated fluorescence microscope (Axiovert 35; Carl Zeiss MicroImaging GmbH, Jena, Germany). Red CMXRos quantification was made by red fluorescence quantification. Myotubes were lysed on RIPA buffer (50 mmol/L Tris-HCl, pH 7.4, 150 mmol/L NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, and 1 mmol/L phenyl-methylsulfonyl fluoride). After sonication, the lysate were centrifuged at 1500 g for 5 min. Supernatant fluorescence for Red CMXRos (∼579 nm excitation and ∼599 nm emission) and DAPI (∼358 nm excitation and ∼461 nm emission) was evaluated in a spectrofluorometer (Perkin-Elmer, Italy).

Real-time PCR Total RNA was obtained using TriPure isolation reagent (Roche Applied Science, Indianapolis, IN) following the manufacturer's instructions. RNA concentration was determined fluorometrically using RiboGreen reagent (Invitrogen, Carlsbad, CA). RNA integrity was checked by electrophoresis on 1.2% agarose gel containing 0.02 mol/L morpholinopropanesulfonic acid and 18% formaldehyde. Total mRNA was retrotranscribed using an iScript cDNA synthesis kit (Bio-Rad Laboratories). Transcript levels were determined by real-time PCR using the SsoFast EvaGreen supermix and the MiniOpticon thermal cycler (Bio-Rad Laboratories). Primer sequences (forward and reverse) were as follows: Bnip35'-CAGAGCGGGGAGGAGAAC-3' and 5'-GAGGCTGGAACGCTGCTC-3'; Sqtm15'-CCCAGTGTCTTGGCATTCTT-3' and 5'-AGGGAAAGCAGAGGAAGCTC-3'; Becn15'-TGAAATCAATGCTGCCTGGG-3' and 5'-CCAGAACAGTATAACGGCAACTCC-3'; Cycs5'-GGAGGCAAGCATAAGACTGG-3' and 5'-TCCATCAGGGTATCCTCTCC-3'; Fbxo325'-TCACAGCTCACATCCCTGAG-3' and 5'-AGACTTGCCGACTCTTTGGA-3'; Trim635'-TGTCTGGAGGTCGTTTCCG-3' and 5'-ATGCCGGTCCATGATCACTT-3'; Ppargc1a5'-GCCAAAGCTGAAGCCCTCTTG-3' and 5'-TGCAAGGAGAGACCTGCTTGCT-3' Myf55'-TGAAGGATGGACATGACGGACG-3' and 5'-TTGTGTGCTCCGAAGGCTGCTA-3'; Mef2c5'-AGATCTGACATCCGGTGC-3' and 5'-TCTTGTTCAGGTTACCAGCT-3′; Myog5'-GCAAATGATCTCCTGGGTTG-3' and 5'-ATCCAGTACATTGAGCGCCT-3'; Myh45'-CACCTGGACGATGCTCTCAGA-3' and 5'-GCTCTTGCTCGGCCACTCT-3'; Myh75'-CTCAAGCTGCTCAGCAATCTATTT-3' and 5'-GGAGCGCAAGTTTGTCATAAGT-3'

For mature miR analysis, all miRs were retrotranscribed using the miScript II RT Kit (Qiagen, Hilden, Germany). Transcript levels were determined by quantitative real-time PCR using the miScript SYBR Green PCR Kit (Qiagen, Hilden, Germany) and the CFX Connect Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). Fifteen seconds of denaturation at 95°C were followed by 30 s of annealing at 55°C and extension at 70°C for 30 s. These steps were repeated for 40 cycles. A universal primer was used together with a miR specific primer in each reaction (miScript Primer Assay, Qiagen, Table 3). For samples with low amount of total RNA, a step of pre-amplification was added before real-time quantification by using the miScript PreAMP PCR Kit (Qiagen, Hilden, Germany) in combination to a selection of miR specific primers. Every quantitative real-time PCR was validated by analysing the respective melting curve. Gene expression were normalized to either SNORD72 expression or the mean of a group of selected housekeeping miRs (let-7i-5p, miR-27b-3p, and miR-152-3p), and results are expressed as –ΔCt.

Small RNA-sequencing

Total RNA concentration was quantified by Qubit® 2.0 Fluorometer with Qubit® microRNA Assay Kit (Invitrogen, Carlsbad, CA, USA). Small RNA transcripts were converted into barcoded cDNA libraries. Library setting was performed using NEBNext Multiplex Small RNA Library Prep Set for Illumina (New England BioLabs Inc., Ipswich, MA, USA). Each library was prepared with a unique indexed primer, according to previous studies.14 The obtained sequence libraries were subjected to the Illumina sequencing pipeline, passing through clonal cluster generation on a single-read flow cell (Illumina Inc., San Diego, CA, USA) by bridge amplification and were sequenced at ultrahigh throughput on NextSeq500 (Illumina Inc., San Diego, CA, USA). One flowcell in 24-plex could produce about six million single reads per sample, 75 base pairs long. Sequencing results were validated using digital PCR. Housekeeping genes were chosen following a modified version of the methodology described by Eisenberg and Levanon.15

Bulk small RNA-sequencing (sRNA-Seq) on MVs isolated from C26 cell cultures and from the plasma of C26-bearing mice was performed. The expression of small non-coding (snc)RNAs in the skeletal muscle was analysed using two library sets originating from two different animal experiments, referred to as ML-1 and ML-2. As for plasma MVs, library was prepared on material contained in MVs isolated from the plasma of ML-2 animals. In this particular case, only n = 4 controls reached the minimal RNA concentration needed and were loaded for sequencing, while all n = 6 C26 samples were sequenced. Additionally, we included the sequencing of three samples containing MVs isolated from the culture medium of C26 cells in step-down condition. All libraries were run at the same sequencing facility, and bioinformatic analysis was performed by the same operator, following the pipeline represented in Figure S1C. Raw sequencing data were deposited on Gene Expression Omnibus with the identifier GSE165531 (token for Reviewer access: slobyoauntyrtkx).

Digital PCR

The expression of miRs in plasma-derived MVs was quantified by means of the QX200 TM Droplet Digital TM PCR System (ddPCR, Bio-Rad) using EvaGreen Digital PCR Supermix (Bio-Rad) with Droplet Generation Oil for EvaGreen (Bio-Rad) following manufacturer's instruction. Briefly, 20 μL of EvaGreen Supermix, 2 μM primers, and pre-amplified cDNA were transferred in a droplet generation cartridge and mixed with 70 μL of droplet generation oil to obtain an emulsion. Droplets (40 μL) were transferred to a new ddPCR 96-well plate. Amplification of specific miRs was obtained with the following protocol: 95°C 5 min 1 cycle enzyme activation, 95°C 30 s 40 cycles denaturation, 54°C 1 min 40 cycles annealing/extension, 4°C 5 min 1 cycle signal stabilization, 90°C 5 min 1 cycle signal stabilization, 4°C 40 min 1 cycle, 4°C. Fluorescence of single droplets was read using QX200 Droplet Reader, and data were analysed using the QuantaSoft Version 1.7.4 Software (Bio-Rad) and expressed in copy number per microlitre.

Data analysis and presentation

Data are expressed as means ± SD, with the exception of gene expression (means ± SEM). The significance of the differences was evaluated by Students t-test or ANOVA followed by uncorrected Fisher's least significant difference. sRNA-Seq pipeline analyses were performed using a methodology described in Sabo et al.14

Results Tumour-derived microvesicles affect C2C12 myoblast differentiation

C2C12 myoblast were exposed to TMVs derived from C26-MVs or LLC-MVs conditioned medium starting from the beginning of differentiation (Day 0) and analysed after 2, 4, and 6 days, roughly corresponding, respectively, to differentiating myoblasts, early, and fully differentiated myotubes. In comparison with untreated cultures, the morphological appearance of C2C12 cells treated for 2 and 4 days with MVs suggested a delayed differentiation no longer appreciable after 6 days (Figure 1A). A comparable pattern was observed in primary cultures of muscle progenitors isolated from both the soleus and the extensor digitorum longus muscles of healthy mice exposed to C26-MVs for 24 and 48 h (Figure S2A and S2B).

Effect of tumour-derived microvesicles on C2C12 myoblast differentiation. (A) Phase-contrast microscopy of C2C12 myoblast cultures during differentiation in the presence or in the absence of C26-MV or LLC-MV (scale bar: 100 μm); (B) heat map representing changes in protein expression. Data are expressed as arbitrary units (AI); n = 3 for each experimental condition, statistical significance is set at P < 0.05; (C) representative western blotting pattern of protein expression reported in panel (B); (D) western blotting showing caspase-3 expression; (E) representative blots and quantification of Sunset analysis for the assessment of protein synthesis 4 and 24 h after exposure to C26-MV or LLC-MV. Data are expressed as % of controls, n = 3. Significance of the differences: *P < 0.05, **P < 0.01 vs. controls. LLC, Lewis lung carcinoma; MV, microvesicles.

The delayed differentiation was partially consistent with the expression of molecules driving myogenesis. Indeed, myogenin, a marker of late differentiation, is down-regulated with respect to control values at Days 4 and 6 (Figure 1B and 1C) while unchanged Pax7 and MyoD levels were detected throughout the experiment (Figure 1B and 1C). Differentiating C2C12 myoblast did not die by apoptosis, as demonstrated by both the morphological appearance (Figure 1A) and the lack of activated caspase-3 after exposure to C26-MVs/LLC-MVs at any time point (Figure 1D). Previous observations showed that muscle wasting in C26-bearing mice was associated with impaired myogenesis and could be corrected by MEK inhibitors.16, 17 Consistently, the results reported in Figure 1B and 1C show a trend to increase of the p-ERK/ERK ratio in C2C12 myoblasts exposed to both C26-MVs and LLC-MVs with respect to control cultures.

The expression of MyHC paralleled the morphological data, being lower in TMV-treated cultures than in control cells at 2 and 4 days after differentiation, still showing a non-significant trend to reduction at Day 6 (Figure 1B and 1C). Reduced MyHC levels were associated with decreased de novo protein synthesis (Figure 1E), C26-MVs appearing more effective than LLC-MVs (Figure 1E).

Remodelling of the mitochondrial network is an important step during myogenic differentiation.18 Along this line, the impact of TMVs on mitochondria during C2C12 differentiation was assessed. The amount of polarized mitochondria was reduced in differentiating C2C12 cultures exposed to TMVs (Figure S3A). The expression of PGC-1α, the master regulator of mitochondrial biogenesis, was unchanged at all the experimental points (Figure S3C), although a not significant trend towards reduction could be envisaged. SDH-A expression was significantly reduced at Day 6 (Figure S3B), indicating a reduced mitochondrial function.

Tumour-derived microvesicles modulate energy metabolism in C2C12 myotubes

To assess if TMVs could affect mature myotubes, fully differentiated C2C12 cultures were exposed to C26-MVs or LLC-MVs for 24 h. Figure S4A demonstrates that both TMVs were able to directly interact with and enter C2C12 myotubes.

Immunostaining for MyHC showed that myotube size was not affected by C26-MVs or LLC-MVs (Figure 2A). Consistently, the mRNA levels of Fbxo32/Atrogin1 and Trim63/Murf1, muscle-specific ubiquitin ligases frequently up-regulated in myotubes exposed to catabolic stimuli, remained comparable to control values (normalized fold change 24–48 h: C = 1; C26-MV = 1.13–1.40; LLC-MV = 1.08–0.99).

Effect of tumour-derived microvesicles on C2C12 myotubes. (A) Immunofluorescence analysis of MyHC expression after 24 h of tumour-derived microvesicle exposure; (B) heat map (ΔΔCt; n = 3) showing the expression of mitochondria-related genes. (C–D) Mitochondrial respiration and oxygen consumption rate (OCR) in C2C12 myotubes exposed to LLC and C26-MVs for 24 h (n = 3). ATP, adenosine triphosphate; BR, basal respiration; MR, maximal respiration; PL, proton leak; SRC, spare respiratory capacity. Data expressed as picomoles per minute. (E) Lactate release from C2C12 myotubes exposed to LLC-MVs or C26-MVs. Data are expressed as means ± SD, n = 3. Significance of the difference: *P < 0.05, **P < 0.01, ***P < 0.001 vs. controls.

In order to evaluate the effects of TMVs on mitochondria homeostasis, markers of mitochondrial biogenesis and function were analysed. As shown in Figure 2B, the mRNA expression of both Ppargc1a and Cycs was reduced after 24 h exposure to C26-MVs or LLC-MVs when compared with untreated myotubes. By contrast, the expression of Bnip3, a molecule involved in mitophagy, which is the leading mechanism for the disposal of dysfunctional mitochondria, was increased. Such a pattern was associated with reduced amount of polarized mitochondria, at least in myotubes exposed to C26-MVs (Figure S4B). Consistently, a significant impairment of mitochondrial respiration occurred (Figure 2C). Both LLC-MVs and C26-MVs were able to reduce basal respiration, maximal respiration, and spare respiratory capacity (Figure 2D). Only the C26-MVs negatively modulated the ATP production that was unchanged in the presence of LLC-MVs (Figure 2D). Furthermore, TMVs stimulated lactate release into the culture medium of C2C12 myotubes (Figure 2E), suggesting a shift from oxidative to glycolytic metabolism.

Microvesicles isolated from the plasma of tumour-bearing animals partially recapitulate cachexia in healthy mice

The ability of the MVs derived from the plasma of tumour-bearing mice in reproducing cachexia in healthy animals was assessed. MVs isolated from C26 hosts and rats bearing the AH-130 hepatoma, a tumour known to rapidly induce cachexia in the host,19 were infused into healthy mice with daily i.v. injection for a maximum of 4 days. As shown in Figure 3, MV infusion resulted in slightly reduced GSN mass in comparison with controls. A different kinetic was observed among animals infused with C26-MVs (Figure 3A) or AH130-MVs (Figure 3C). As for the tibialis anterior, only AH130-MVs led to a reduction in size (Figure 3B and 3D).

Tumour-derived microvesicles infused into healthy mice partially recapitulate cachexia. Changes in (A, C) gastrocnemius and (B, D) tibialis anterior mass in healthy mice infused for 48, 72, and 96 h with C26-MV or AH130-MV. Data (mg/10 g initial body weight, means ± SD) are expressed as % of controls, n = 5. Significance of the differences: *P < 0.05, **P < 0.01 vs. controls. MV, microvesicles.

The expression of genes involved in mitochondrial homeostasis and protein degradation did not change in the GSN of mice receiving MVs from the C26 host, with the exception of increased Becn1/Beclin1 levels (Figure S5A). When mice were treated with AH130-MVs for 72 h, Ppargc1a expression did not change, whereas Cycs was reduced. Similarly, the levels of both Bnip3 and Trim63/Murf1 were reduced, while no differences could be observed as for Fbxo32/Atrogin1, Becn1/Beclin1, and p62/Sqstm1. By contrast, most of these markers were overexpressed in comparison with untreated animals at 96 h, with the exception of atrogin-1 and beclin-1 that remained unchanged (Figure S5B).

Tumour-derived microvesicles show a peculiar small non-coding RNA profile

Tumour-derived microvesicles are one of the routes by which tumour cells convey signals to both surrounding and distant target tissues. In this study, the attention was focused on sncRNAs by analysing MVs isolated from the plasma of the C26 hosts (Figure 4) and from C26 culture medium (Figure 5) by sRNA-Seq.

F miR expression profile in circulating MVs of C26-bearing mice. (A) PCA plot (PC1 vs. PC2) representing sample variability in the library; (B) raw read abundance and molecular diversity of each sncRNA biotype; (C) top 20 miRs detected in circulating MVs; (D) volcano plot reporting all detected miRNAs in the gastrocnemius, according to P value (−log10) and fold change (log2) distribution (controls: n = 4; C26: n = 5). Significance of the difference set at P < 0.05; (E) expression of selected myomiRs in the gastrocnemius of controls and C26 hosts (means ± SEM); (F) panel representing the changes in the levels of the snoRNA Gm23011 in the gastrocnemius of the C26 hosts. For all ‘box and whisker’ plots, data are represented as: line = median, whiskers = min to max (controls: n = 4; C26: n = 5; P < 0.05). miR, microRNA; MV, microvesicles.

miR expression profile in MVs isolated from the C26 CM. (A) Top 20 miRs detected in MVs isolated from the C26 CM; (B) molecular diversity of each sncRNA biotype; (C) top 5 rRNAs and (D) top 5 snoRNAs in MVs isolated from the C26 CM. For all ‘box and whisker; plots, data are represented as: line = median, whiskers = min to max (n = 3). miR, microRNA; MV, microvesicles.

Plasma MV samples showed poor group clustering in principal component analysis, indicating high sample variability (Figure 4A and 4B). Although MV content was represented mostly by miRs (86.93% of reads aligned to miRs), MV cargo was further enriched with other types of sncRNAs such as lincRNAs (2.20%), rRNAs (1.87%), snRNAs (1.73%), and snoRNAs (1.43%), along with high diversity of sequences for each biotype (Figure 4B). Up to 122 different miR sequences were detected in plasma MVs, the top 20 being reported in Figure 4C. From all the detected miRs, only miR-181a-5p, miR-375-3p, and miR-455-5p were significantly down-regulated in MVs isolated from the plasma of the C26 hosts with respect to those deriving from healthy, control mice (Figure 4D). Although the average fold changes were high for other miRs, sample variance in both control mice and C26 hosts compromised statistics assessment, as exemplified by miR-1a-3p and miR-21a-5p expression (Figure 4E). Indeed, while miR-21a-5p is not significant when correcting for multiple testing, it is nominally significant (P = 0.005) and still one of the top miRs (#7), so with a larger sample size changes would likely be significant. By contrast, miR-1a-3p is weaker and not nominally significant (although close to, P value being 0.057). Additionally, other myomiRs such as miR-133a-3p and miR-206-3p, which were described to be modulated in the blood of patients suffering from different myopathies,20 were detected in very low amounts and apparently not altered in the present experimental setting (Figure 4E). Among the three miRs found significantly modulated, validation by digital droplet PCR confirmed the down-regulation just for miR-375-3p (Figure S6).

Microvesicles isolated from the plasma of the C26-bearing mice are produced by both tumour and host cells. In terms of miR content, the differences observed between MVs obtained from tumour-bearing and control mice could reflect the contribution of the C26 cells but also host cell response to the tumour. To investigate this point, sncRNA cargo was assessed in MVs released in C26 culture medium, showing that they contained several let-7 family members as well as miR-21a-5p (Figure 5A). C26-MVs also contained large number of miR-148a-3p reads. By contrast, low read counts for miR-181a-3p (mean counts = 127) were collected, while both miR-375-5p and miR-455-5p were absent (Figure 5A). Despite these results suffer from the limitation of lacking an appropriate control, for example, normal enterocytes, they reasonably demonstrate that the down-regulation of miR-181a-5p, miR-375-3p, and miR-455-5p observed in MVs isolated from the plasma of the C26 hosts unlikely reflects the contribution of MVs released by the tumour cells. However, as an additional consideration and note of care, the possibility that the C26 cells could release MVs with different miR content in vitro and in vivo cannot be discarded.

Besides miRs, the sRNA-Seq data set used in the present study also included annotations for several other types of sncRNA, including snoRNAs, mt-tRNAs, snRNAs, and rRNAs among the other biotypes (Figure 5B). As for plasma-derived MVs, only Gm23011, a sequence predicted to be a snoRNA (Ensambl entry: ENSMUST00000104148.3), was found up-regulated in MVs isolated from C26 hosts compared with controls (Figure 4E). Figure 5C and 5D shows the top 5 rRNAs and snoRNAs, respectively, in MVs isolated from C26 CM.

MicroRNAs contained into microvesicles isolated from cultured C26 cells slightly impinge on myogenesis

C2C12 myoblasts were either transfected with plasmids overexpressing miR-21-5p, miR-148a-3p, and miR-181-5p or exposed to MVs isolated from NIH3T3 fibroblasts overexpressing the same miRs (3T3-MVs; refer to Methods section), and the effects on markers of myogenesis were evaluated. The results reported in Figure 6 show that overexpressing miR-148a-3p into C2C12 myoblasts resulted in increased mRNA levels of Myf5 and Myog at Day 2 of differentiation, although not at Day 4. By contrast, overexpressing miR-181a-5p reduced the mRNA levels of genes coding for MyHC, namely, Myh4 and Myh7. Decreased levels of Myh7 were observed also in C2C12 cells overexpressing miR-21-5p or miR-148a-3p at Day 4 of differentiation. These observations are partially consistent with the reduced MyHC protein levels observed in differentiating C2C12 cultures (refer to Figure 1B). No appreciable effect could be detected in C2C12 cultures exposed to 3T3-MVs (data not shown).

Effect of miRs contained into C26-MVs on the expression of genes involved in myogenesis. (A) Heat map (ΔΔCt; n = 3) showing the expression of genes involved in C2C12 overexpressing specific miRs at Days 2 and 4 of differentiation. Significance of the differences: *P < 0.05, **P < 0.01 vs. SCR (scramble) for each time point. SCR, scrambled sequence.

Discussion

This is the first study reporting that MVs secreted from tumour cells impair differentiation of murine myoblasts in vitro and significantly affect energy metabolism in myotube cultures. Moreover, MVs isolated from the plasma of tumour-bearing animals partially recapitulate cachexia when infused into healthy animals.

The first relevant observation reported in the present study is that cytotoxicity does not occur in TMV-exposed C2C12, in contrast with results reported by He et al.3 This latter study is performed adding TMVs to myoblasts maintained in growing medium, while the results here presented refer to cultures in differentiating medium (refer to Methods section), suggesting that growing, but not differentiating, myocytes are prone to TMV-induced cytotoxicity. Indeed, myotubes were reported to be more resistant than myoblasts to death, due to enhanced anti-apoptosis pathways driven by apoptosis repressor with caspase recruitment domain, HSP70, and HSP25.21 In addition, recent observations showed that both apoptosis induction and apoptosome assembly are required for myoblast differentiation.22 In this regard, the apoptotic stimulus provided by TMVs could likely result in cell death only when applied to proliferating myoblasts.

Tumour-derived microvesicle exposure in differentiating C2C12 cultures results in delayed, although not inhibited, shift from myoblasts to myotubes, as demonstrated by the morphologic analysis, by the steadily reduced levels of myogenin and by the transient decrease of MyHC expression. To the best of our knowledge, this is the very first evidence that TMVs can impinge on in vitro myogenesis. There are just few studies published on this topic. Extracellular vesicles isolated from C2C12 myotubes exposed to pro-oxidant agents impaired myoblast differentiation.23 Alternatively, extracellular vesicles derived from mesenchymal stem cells, endothelial, and inflammatory cells were required for an efficient muscle regeneration.24 In the present study, the delayed myogenic differentiation induced by TMVs is associated with reduced protein synthesis and with a trend to increased levels of the p-ERK/ERK ratio. This latter, in particular, suggests a consistency with previous observations showing that increased p-ERK levels in the muscle of the C26 hosts, causally contributed to wasting also by impinging on regeneration.16 In addition, increased p-ERK levels were reported in TNFα-treated myoblast cultures, characterized by delayed differentiation.16 The possibility that TMVs could delay myoblast differentiation by releasing TNFα is unlikely because C26 cells do not produce this cytokine in significant amounts.25

In addition to modulate in vitro myogenic differentiation, TMVs also impinge on energy metabolism in myotube cultures, by reducing mitochondrial respiration and increasing lactate production. Such functional alterations are paralleled by reduced expression of proteins involved in mitochondrial biogenesis/content and by increased levels of the marker of mitophagy Bnip3. These observations do not conform to the data available in the literature. Indeed, extracellular vesicles derived from mesenchymal stem cells infused into mice with kidney injury were reported to improve mitochondrial antioxidant defence and ATP production, to reduce mitochondrial fragmentation and to preserve both mitochondrial membrane potential and mitochondrial number.26 Exosomes derived from endothelial progenitors added to homotypic cells exposed to hypoxia and reoxygenation were shown to decrease mitochondrial fragmentation and to elevate mitochondrial membrane potential and ATP levels.27 Extracellular vesicles isolated from a hippocampal cell line exposed to TNFα were able to increase the mitochondrial OCR in naive cells. However, also proton leak and the production of reactive oxygen species were enhanced, suggesting that, despite the improved oxidative phosphorylation, mitochondria were not properly working.28 Finally, extracellular vesicles derived from C2C12 myotube cultures suppressed the enhanced oxygen consumption and mRNA expression of markers of mitochondrial biogenesis induced by RANKL in murine bone marrow cells and preosteoclastic Raw264.7 cells.29 Consistently with this last observation, the results reported in the present study demonstrate that TMVs exert a negative action on mitochondrial function in C2C12 myotube cultures, supporting their contribution to energy crisis in tumour-bearing mice. Consistently, exposure to TMVs partially recapitulates cachexia in healthy mice. However, depending on the host by which MVs are derived (C26 or AH-130 bearing mice), this cachexia-inducing effect differently impinges on the expression of genes involved in mitochondrial homeostasis. In this regard, one of the main limitations of this experiment is that TMVs were delivered to the recipient animals with a daily injection, in comparison with the continuous perfusion occurring in the tumour hosts, resulting in differences in the kinetics of exposure.

The present study assessed sncRNAs contained into MVs isolated from C26 cell culture medium or from the plasma of mice bearing the C26 tumour. Strikingly, only three miRs were differentially expressed in MVs isolated from the circulation of the C26 hosts compared with control mice, and only miR-375-3p was confirmed to be down-regulated. This miR behaves as a tumour suppressor targeting oncogenes and inhibiting proliferation in hepatocellular carcinomas, colorectal, gastric, and cervical cancers.30 Reduced hepatic miR-375-3p was reported to activate the TGF-β1/Smad system through JAK2/STAT3. The same signalling pathways were activated in the muscle of the C26-bearing mice,31 being potentially consistent with the reduced amount of miR-375-3p in circulating MVs.

The sncRNA profile of C26-MVs is markedly different from that assessed in MVs circulating in C26-bearing mice. This observation clearly demonstrates that tumour cells may behave quite differently in vitro and in vivo, corroborating that data deriving from cell cultures might not reflect pre-clinical results. In addition, MVs in the bloodstream derive from the tumour, but also from other cell types, including those pertaining to the inflammatory response, adipocytes, hepatocytes and myofibres themselves. In this regard, TMVs in the circulation could be diluted, resulting in concentrations of specific sncRNAs below the detection limit.

The most abundant miRs contained in MVs isolated from C26 cell cultures are miR-21a-5p, let-7i-5p, and let-7c-5p. Up-regulation of miR-21 was reported in both aged muscles and satellite cells as well as in the presence of oxidative stress or pro-inflammatory cytokines.20 Consistently with the results shown in this study, miR-21 was proposed to act as a negative regulator of in vitro myogenesis. Indeed, its inhibition by a specific anti-miR was able to abrogate the impaired myogenesis induced in