Proteomic analysis of the small extracellular vesicles and soluble secretory proteins from cachexia inducing and non‐inducing cancer cells

Cancer cachexia is a wasting syndrome characterised by the loss of fat and/or muscle mass in advanced cancer patients. It has been well‐established that cancer cells themselves can induce cachexia via the release of several pro‐cachectic and pro‐inflammatory factors. However, it is unclear how this process is regulated and the key cachexins that are involved. In this study, we validated C26 and EL4 as cachexic and non‐cachexic cell models, respectively. Treatment of adipocytes and myotubes with C26 conditioned medium induced lipolysis and atrophy, respectively. We profiled soluble secreted proteins (secretome) as well as small extracellular vesicles (sEVs) released from cachexia‐inducing (C26) and non‐inducing (EL4) cancer cells by label‐free quantitative proteomics. A total of 1268 and 1022 proteins were identified in the secretome of C26 and EL4, respectively. Furthermore, proteomic analysis of sEVs derived from C26 and EL4 cancer cells revealed a distinct difference in the protein cargo. Functional enrichment analysis using FunRich highlighted the enrichment of proteins that are implicated in biological processes such as muscle atrophy, lipolysis, and inflammation in both the secretome and sEVs derived from C26 cancer cells. Overall, our characterisation of the proteomic profiles of the secretory factors and sEVs from cachexia‐inducing and non‐inducing cancer cells provides insights into tumour factors that promote weight loss by mediating protein and lipid loss in various organs and tissues. Further investigation of these proteins may assist in highlighting potential therapeutic targets and biomarkers of cancer cachexia.


INTRODUCTION
Cancer cachexia is a wasting syndrome observed in late-stage cancer patients and is characterised by the progressive loss of skeletal muscle mass with or without depletion of adipose tissue [1]. Atrophy of skeletal muscle and lipolysis of adipose tissue are common phenomena seen in many types of cancer and are associated with poor prognosis [2][3][4]. It is estimated that 50%-80% of late-stage cancer patients are affected by cachexia [5]. However, clinical management of cancer cachexia remains a challenge due to the unavailability of drugs that are currently approved for the treatment of cachexia [6]. Decades of research have attributed a combination of pro-cachectic and proinflammatory factors, reduced food intake as well as metabolic changes in the pathogenesis of cancer cachexia [6]. Whilst several pro-cachectic and pro-inflammatory factors have been implicated in mediating cancer cachexia in pre-clinical models, therapies targeting a number of these factors have shown low or no clinical benefit [6][7][8][9]. Similarly, nutritional interventions that are aimed at managing cancer cachexia have limited clinical benefit, although early treatment (progesterone analogue or corticosteroid) has been proposed to increase the quality of life in cancer patients [9,10]. Hence, potential nutritional and pharmacologic interventions at earlier time points could benefit from the availability of biomarkers for cancer cachexia [9].
It has been proposed that cancer cells in coordination with cells within the tumour microenvironment secrete factors as well as extracellular vesicles (EVs) that result in the induction of wasting in distant organs such as muscles and adipose tissue [7,[11][12][13]. EVs are secreted by almost all cell types and contain proteins, lipids, and nucleic acids [14][15][16]. EVs have been implicated in several biological processes that are involved in promoting cancer progression, regulating metastasis, and priming the tumour microenvironment [17][18][19][20]. In the quest to identify the mediators of cachexia, several studies have reported pro-cachectic and pro-inflammatory factors as potential therapeutic targets and biomarkers of cancer cachexia [21]. In addition, studies have also highlighted molecules and proteins that are released into systemic circulation by muscle cells during atrophy [22]. However, reliable biomarkers that are needed to improve the clinical management of cancer cachexia are not currently available [21]. Hence, analysis of secreted factors and EVs from the cachexia-inducing tumour cells may expand our knowledge of the key mediators and further aid in the identification of potential biomarkers. However, proteomic profiling of secreted factors and EVs from cachexia-inducing cells have not been carried out previously.
In this study, we first validated C26 as a cachexia-inducing and EL4 as cachexia non-inducing cell models and we describe the protein profile of both the soluble secreted factors (secretome) and small EVs (sEVs) derived from C26 and EL4 cells. To identify the tumorigenic factors that may be involved in cancer-induced wasting, label-free quantitative proteomics analysis of the secretome and sEVs of C26 and EL4 cells was performed using high-resolution mass spectrometry. Proteomic analysis demonstrated that the secretome and sEVs of C26 are abundant in proteins and cytokines that are associated with biological processes such as muscle atrophy, lipolysis, inflammation, as well

Statement of significance
Although it is established that cancer cachexia is mediated, at least in part, by cancer cell secreted factors and extracellular vesicles, the proteomic profile of these secreted factors is uncharacterised. Here, we examined the protein profile of secreted factors and extracellular vesicles from cachexiainducing and non-inducing cancer cells. The identified proteins may be utilised in discovering potential therapeutic targets and biomarkers of cancer cachexia.
as induction of cell fate determinates such as autophagy and ubiquitinproteasome pathways (UPS). In contrast, the analysis of the secretome and sEVs of EL4 cells revealed low or undetectable levels of known

Isolation of secretory factors and sEVs from conditioned medium
For the preparation of conditioned medium (CM), C26 and EL4 cells were seeded in 150 mm diameter culture dishes in a complete culture medium. At 60%-70% confluency, cells were washed twice with 1× phosphate-buffered saline (PBS; without magnesium chloride and calcium chloride). Cells were then cultured in 15 mL of serum-free medium for 24 h. CM was then collected and centrifuged at 500 × g for 10 min followed by 2000 × g for 20 min. A tablet of protease inhibitor cocktail (Roche, USA) was added to 10 mL of CM as suggested by the manufacturer. CM was then subjected to centrifugation for 30 min at 10,000 × g followed by 1 h at 100,000 × g to separate sEVs (pellet) and secretory factors (supernatant). The supernatant was concentrated from 45 mL to 1-2 mL using an Amicon Ultra-15 centrifugal filter device with a 3000 Da nominal molecular weight limit (Millipore) at 4000 × g. sEVs pellet was washed with 1× PBS and further centrifuged at 100,000 × g for 1 h. The resultant pellet was resuspended in 0.22 μm-filtered ice-cold 1× PBS. In all the centrifugation steps, the temperature was maintained at 4 • C. Secretory factors and sEVs were stored at −80 • C until further analysis.

Nanoparticle tracking analysis
Visualisation and analysis of size distribution of sEVs were carried out using NanoSight N300 (Malvern Instruments, Malvern, UK) and results obtained were analysed using NTA software 3.0 (ATA Scientific). Samples were monitored with the monochromatic laser beam at 405 nm, camera level at 12, detection threshold at 5 and syringe pump speed at 50, and temperature at 23-25 • C.

Immunofluorescence and Oil Red O staining
Cells were seeded on to UV-sterilised glass cover slips in 6-well plates.

Database searching and identification of proteins
MaxQuant software (version 2.0.1.0) [23] was used for database search against the UniProt Mus musculus reviewed database (Taxonomy ID 10090, Feb 2023). False discovery rate (FDR) was set to 1% at both peptide and protein level with match between runs enabled. Trypsin/P was selected for enzyme specificity to allow a maximum of two missed cleavages at C-termini of arginine and lysine, also when followed by proline. Cysteine carbamido-methylation was set as fixed modification.
N-termini acetylation and methionine oxidation were considered as variable modifications. Other parameters were set to software default.
Label-free quantitation was achieved using the built-in MaxLFQ algorithm. Potential contaminants and reverse hits were removed prior to identifying proteins present in all three biological replicates (i.e., LFQ intensity of more than 0). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD041744.

Bioinformatics and statistical analysis
The analysis of quantitative proteomics data was achieved using Perseus (version 2.0.7.0) [24]. Volcano plots were generated using

C26-derived conditioned medium induces lipolysis and muscle atrophy
In a physiological setting, tumours and energy reserve tissues can be distantly located and communication between them is mediated, at  Figure 1B and C). Next, we examined the effect of C26 and EL4-derived CM on myotubes to assess muscular atrophy. To do so, C2C12 myoblast were differentiated into myotubes using horse serum and treated with either C26 or EL4-derived CM for 48 h ( Figure 1D). Immunofluorescence images showed that treatment with EL4-derived CM was had no effect on myotube diameter, whilst treatment with C26- suggest that the CM derived from C26 cells may be enriched in procachectic factors that potentially induces lipolysis as well as muscle atrophy.

Separation of sEVs and soluble proteins in CM
Given that CM from C26 cells was inducing lipolysis and atrophy, in this study, we investigated the protein signature in both the soluble secretome and sEV fractions from cachexia inducing C26 and noninducing EL4 cancer cells CM. C26 and EL4 cells were grown in serum-free medium for 24 h and the resulting CM were subjected to differential-ultracentrifugation to isolate fractions containing sEVs and soluble secreted proteins (secretome) with a molecular weight above 3 kDa (Figure 2A Bip is an endoplasmic reticulum resident protein [30] which is typically not detected in sEVs. Consistent with this, Bip was present only in the whole-cell lysates (WCLs) of C26 and EL4 cells, thus validating the purity of the isolated sEVs ( Figure 2C). Furthermore, TEM analysis confirmed the presence and expected morphology of sEVs ( Figure 2D and E). Taken together, these results suggest high composition of sEV population in the isolated samples from C26 and EL4 cells.

Overview of protein content in secretome and sEVs
To gain a deeper understanding of the differences in the protein profile of the secretome and sEVs derived from cancer cachexia-inducing C26 and non-inducing EL4 cells, mass spectrometry-based label-free quantitative proteomics analysis was conducted (Tables S1 and S2) It is well-established that proteins that are secreted via the classical pathway harbour a signal peptide, a conserved N-terminal tripartite structure, which assists in protein targeting for secretion [31]. Hence, to determine the proportion of classically secreted proteins in the secretome of C26 and EL4 cells, the list of proteins obtained through mass spectrometry was compared against the signal peptidecontaining proteins that were annotated in the UniProt (Table S3). A total of 120 and 70 signal peptide-containing proteins were identified in the secretome of C26 and EL4 cells, respectively, where 60 were commonly identified ( Figure 3C). Interestingly, a large proportion of the identified proteins in the secretome of C26 and EL4 cells do not contain a known or predicted signal peptide. Although we cannot rule out the possibility of minor traces of sEVs in the C26 and EL4-derived CM in the secretome, it has been shown that proteins lacking a signal peptide can be secreted into the extracellular space via unconventional pathways, such as plasma membrane pore formation, organelle-based translocation of leaderless proteins as well as bypassing Golgi [32][33][34][35][36][37].
Furthermore, the obtained list of signal peptide-containing proteins includes proteins from various sources, such as neuron-specific secretory proteins, which may not be present in colon and lymphoblast carcinoma cells.
Next, we compared the protein cargo of sEVs derived from C26 and EL4 cells. Using the same parameter as previously mentioned, we identified a total of 737 proteins in the sEVs derived from C26 cells and 1228 proteins in that of EL4 cells, where 578 proteins were mutually present ( Figure 3D). Subsequent quantitative Venn diagram revealed 159 and 650 proteins to be exclusively identified in the sEVs derived from C26 and EL4 cells, respectively. Furthermore, among the 578 proteins that were mutually present, 84 were detected in greater abundance in C26-derived sEVs, whereas 167 were more abundant in sEVs derived from EL4 cells. Interestingly, comparison of the protein cargo depicted a noticeable difference in the total number of proteins identified in sEVs derived from C26 and EL4 cells.
Although it is unclear why EL4-derived sEVs consist of a significantly higher number of proteins, it can be speculated that this may potentially be due to cancer cell type-specific proteins that are secreted via sEVs [13].

Differential abundance of protein expression in C26 and EL4 secretome and sEVs
In order to clearly visualise the changes in the abundance of proteins highlighted that serum derived EVs of pancreatic cancer patients display higher Itgb1 levels when compared to that of healthy individuals [39]. EVs containing Itgb1 are actively captured by the adipocytes and induce phosphorylation of hormone sensitive lipase which is known to induce lipolysis and browning [39]. Taken together, it can be speculated that, Itgb1 could be utilised as potential therapeutic target for pancreatic and colon cancer induced cachexia, and as potential biomarker for early diagnosis of the disease. However, further targeted analysis is needed to validate these preliminary findings.

Proteins implicated in cancer cachexia are enriched in the secretome of C26 cells
To identify the biological processes that are enriched among proteins identified from C26 CM, proteins that were statistically significant with fold change greater than ± 2 in the secretome of C26 and EL4 cells were subjected to functional enrichment analysis using FunRich [25,26]. The results depicted that proteins implicated in biological processes such as negative regulation of adipose tissue development, positive regulation of NIK/NF-κB, TGF-β, and IL-1 mediated signalling pathways were highly enriched in the secretome of C26 cells. Additionally, biological processes implicated protein regulation such as proteolysis and ubiquitin-dependent protein catabolic processes were significantly enriched in the secretome of C26 ( Figure 4A).

C26-derived sEVs are enriched in proteins implicated in cancer cachexia
It is well-known that sEVs mediate potential crosstalk between tumours and various tissues such as muscles and adipocytes [7,12,13,[41][42][43]. Hence to determine the functional aspects of the protein cargo of sEVs derived from C26 and EL4 cells, functional enrichment analysis was conducted using proteins that are statistically significant with fold change greater than ± 2. The results show that proteins impli-

DISCUSSION
In the current study, we aimed to profile the proteins that are secreted as well as packaged in sEVs derived from cachexia-inducing C26 and cachexia non-inducing EL4 cancer cells by mass spectrometry based quantitative proteomics analysis. Although we are aware of the limitations in variability of the cell models due to different tissue and cancer type, we examined the proteomic profile of cachexia and noncachexia-inducing cell lines to identify potential biomarkers and/or therapeutic targets. Functional enrichment analysis revealed that C26 secretome and sEVs were significantly enriched in proteins implicated in the production and release of cachectic factors, induction of muscle atrophy, lipolysis as well as NF-κB and TGF-β signalling. It is well known that upregulation of NF-κB negatively regulates skeletal myogenesis and positively regulates muscle atrophy under various cachexia conditions [44][45][46][47]. Concurring with previous reports, our data supports the notion that induction of NF-κB in muscle cells could be driven by the tumour-derived secretory factors [48]. During muscle differentiation from myoblast to myofibers, upregulation of self-renewing transcription factor Pax7 in the progenitor cells prematurely stalls the myogenic differentiation by preventing fusion and restoration of muscle growth, which eventually leads to atrophy [46,49]. Similarly, activation of NF-κB in myoblast has also been shown to inhibit MyoD synthesis, one of the key factors of myogenic lineage, by suppressing MyoD mRNA at the post-transcriptional level. In healthy individuals, upon skeletal muscle damage, MyoD expression is induced from satellite cells which increases the proliferation and re-initiation of skeletal muscle differentiation, thereby repairing the damaged muscle tissue [46,50,51]. Other processes that were depleted in C26 secretome and sEVs include negative regulation of protein catabolic process and IFNγ production. It was previously demonstrated that tumour-induced IFN-γ/TNF-α activate signal transducer and activator of transcription 3 (STAT3) by triggering its phosphorylation on Y705 residue, pY705-STAT3 collaborates with the NF-kB pathway and translocate to nucleus where it induces the transcription of inducible nitric oxide (NO) synthase (iNOS). The upregulation of iNOS/NO pathway correlates with a decrease in promyogenic factors such as MyoD and contribute to the muscle wasting [52].
Among the proteins identified in our analysis, Mfge8 protein levels were significantly abundant in both the secretome and sEVs derived from C26 cells compared to EL4. Mfge8 is a secreted extracellular glycoprotein that participates in various physiological and pathological processes [53]. Recently, Mfge8 has been shown to be involved in invasion, migration and proliferation of breast [54], liver [55], ovarian, and colorectal [53] cancers. Recently, Ikemoto-Uezumi et al. highlighted the role of Mfge8 in neuromuscular junction (NMJ) denervation and muscle weakness [56]. Increased levels of Mfge8 contributed to the NMJ denervation and muscle weakness but the mechanism underlying this process is not yet fully understood [56]. Another interesting secreted protein that was significantly abundant in the secretome of C26 cells is Spp-1. Spp-1 is a secreted extracellular matrix protein [57] and shown to be elevated in various cancers such as colorectal [58], liver, pancreatic [59], and prostate [60]. Additionally, Spp-1 is shown to accelerate tumour progression [61], invasion and metastasis [62] across various cancers. Interestingly, all the above-mentioned cancers that exhibit

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [1] partner repository with the dataset identifier PXD041744.