Lipidomic profiling of exosomes from colorectal cancer cells and patients reveals potential biomarkers

Strong evidence suggests that differences in the molecular composition of lipids in exosomes depend on the cell type and has an influence on cancer initiation and progression. Here, we analyzed by liquid chromatography–mass spectrometry (LC‐MS) the lipidomic signature of exosomes derived from the human cell lines normal colon mucosa (NCM460D), and colorectal cancer (CRC) nonmetastatic (HCT116) and metastatic (SW620), and exosomes isolated from the plasma of nonmetastatic and metastatic CRC patients and healthy donors. Analysis of this exhaustive lipid study highlighted changes in some molecular species that were found in the cell lines and confirmed in the patients. For example, exosomes from primary cancer patients and nonmetastatic cells compared with healthy donors and control cells displayed a common marked increase in phosphatidylcholine (PC) 34 : 1, phosphatidylethanolamine (PE) 36 : 2, sphingomyelin (SM) d18 : 1/16 : 0, hexosylceramide (HexCer) d18 : 1/24 : 0 and HexCer d18 : 1/24 : 1. Interestingly, these same lipids species were decreased in the metastatic cell line and patients. Further, levels of PE 34 : 2, PE 36 : 2, and phosphorylated PE p16 : 0/20 : 4 were also significantly decreased in metastatic conditions when compared to the nonmetastatic counterparts. The only molecule species found markedly increased in metastatic conditions (in both patients and cells) when compared to controls was ceramide (Cer) d18 : 1/24 : 1. These decreases in lipid species in the extracellular vesicles might reflect function‐associated changes in the metastatic cell membrane. Although these potential biomarkers need to be validated in a larger cohort, they provide new insight toward the use of clusters of lipid biomarkers rather than a single molecule for the diagnosis of different stages of CRC.

Strong evidence suggests that differences in the molecular composition of lipids in exosomes depend on the cell type and has an influence on cancer initiation and progression. Here, we analyzed by liquid chromatographymass spectrometry (LC-MS) the lipidomic signature of exosomes derived from the human cell lines normal colon mucosa (NCM460D), and colorectal cancer (CRC) nonmetastatic (HCT116) and metastatic (SW620), and exosomes isolated from the plasma of nonmetastatic and metastatic CRC patients and healthy donors. Analysis of this exhaustive lipid study highlighted changes in some molecular species that were found in the cell lines and confirmed in the patients. For example, exosomes from primary cancer patients and nonmetastatic cells compared with healthy donors and control cells displayed a common marked increase in phosphatidylcholine (PC) 34 : 1, phosphatidylethanolamine (PE) 36 : 2, sphingomyelin (SM) d18 : 1/16 : 0, hexosylceramide (HexCer) d18 : 1/24 : 0 and HexCer d18 : 1/24 : 1. Interestingly, these same lipids species were decreased in the metastatic cell line and patients. Further, levels of PE 34 : 2, PE 36 : 2, and phosphorylated PE p16 : 0/20 : 4 were also significantly decreased in metastatic conditions when compared to the nonmetastatic counterparts. The only molecule species found markedly increased in metastatic conditions (in both patients and cells) when compared to controls was ceramide (Cer) d18 : 1/24 : 1. These decreases in lipid species in the extracellular vesicles might reflect function-associated changes in the metastatic cell membrane. Although these potential biomarkers need to be validated in a larger cohort, they provide new insight toward the use of clusters of lipid biomarkers rather than a single molecule for the diagnosis of different stages of CRC.

Introduction
Exosomes, extracellular nanovesicles (50-200 nm in diameter) of endosomal origin secreted by living cells into the extracellular environment [1], harbor a bioactive cargo of proteins, nucleic acids, and lipids [2]. These molecules can be transported by exosomes to different cell targets influencing their phenotype and physiological behavior. Tumor-derived exosomes have been reported to play a major role in cancer initiation and progression for instance in colorectal cancer (CRC) [3].
Dysregulation of lipid metabolism can affect cellular homeostasis and signaling pathways, which subsequently influences the process of cell proliferation and differentiation. Such change in the dynamic structure of the plasma membrane lipid bilayer has a major contribution to the onset of various diseases including cancer [4]. The majority of exosomal lipids are mainly localized in the membrane and have been reported to play a role in the biogenesis, secretion, fusion, and uptake of exosomes [5]. Although the molecular composition of lipids in exosomes depends on the cell type, it has been found that the membrane of the exosomes compared to that of the cell from which they originate is enriched in cholesterol (C), sphingomyelin (SM), glycosphingolipids, and glycerophospholipids [6].
Exploration of exosomal lipids as noninvasive circulant cancer biomarkers has only recently started. So far, just a few studies have analyzed the lipidomic profile of exosomes derived from breast [3], ovarian [7], and prostate [1] cancer cell lines. For CRC, only the lipid composition analysis of exosome-derived from the colorectal cancer LIM1215 cell line by mass spectrometry has been reported [8]. Therefore, further lipidomic analysis in colorectal cancer cells-derived exosomes is needed to understand in-depth the role of exosomes in cancer initiation and progression and to identify specific diagnostic/prognostic lipid biomarkers for different stages of CRC. In this pilot study, we analyzed the lipidomic signature of exosomes derived from CRC cell lines and patients by LC-MS. The results revealed that exosomes from both nonmetastatic and metastatic cell lines and those from the plasma of patients displayed similar significant variations in the lipidomic signature of certain lipid molecular species, particularly in glycerophospholipids and sphingolipids compared with their corresponding controls.

Cell lines and patients
The normal colonic epithelial cell line NCM460D (RRID:CVCL_IS47) was purchased from In Cell (San Antonio, TX, USA). HCT116 (RRID:CVCL_0291) and SW620 (RRID:CVCL_0547) cell lines were purchased from American Type Culture Collection (Manassas, VA, USA). All experiments were performed with cell-free mycoplasma using a mycoplasma detection kit (MycoAlert, Lonza Pharma&Biotech, Basel, Switzerland). Cell lines were grown in Dulbecco 0 s Modified-Eagle 0 s Medium (DMEM) supplemented with 10% FBS. The patient's blood samples were obtained from the University Hospital of Dijon (France). The study was conducted in accordance with the Declaration of Helsinki with an approved written consent form for each patient (CPP ESTI: 2014/39; N°ID: 2014-A00968-39). This study was approved by the local ethics committee (IRB 00010311).

Isolation of exosomes
Cells were cultured in DMEM supplemented with 10% FBS (exosome depleted) until reached 80% confluence. Exosomes derived from this conditioned medium and from the plasma of patients were performed by differential ultracentrifugation and filtration as previously described [9]. The concentration and size distribution of exosomes were measured by Nanoparticle tracking analysis (NanoSight NS300, Malvern, UK) and stored at À80°C until use.

Lipid extraction
LC-MS/MS quality grade chemicals were from Sigma Aldrich (Saint-Quentin Fallavier, France) and solvents were purchased from Fischer Scientific (Illkirch, France). Lipids were extracted according to the method of Bligh and Dyer as previously described [10].

Targeted lipidomics
Phospholipids and ceramides were analyzed on a 1200 6460-QqQ LC-MS/MS system equipped with an Electrospray ionization (ESI) source (Agilent Technologies) as previously described [10]. Cholesterol was measured by gas chromatography-mass spectroscopy (GC-MS) using 10 or 15 µL of the Bligh and Dyer extracts obtained from plasma or cellular exosomes, respectively [11].

Statistics
Lipid species were normalized to total cholesterol and analyzed by Two-way ANOVA followed by Tukey's multiple comparisons test. Data were considered statistically significant when P values ≤ 0.05. The statistical analysis was performed using GRAPHPAD PRISM version 8.0.0 for Windows, GraphPad Software (San Diego, CA, USA).

Results and Discussion
Analysis of exosomes derived from normal colon mucosa NCM460, nonmetastatic HCT116 and metastatic SW620 CRC cells by nanosight tracking analysis (NTA) did not reveal any significant differences in their average size. However, NCM460 and HCT116 showed a higher average concentration of exosomes compared with SW620 ( Fig. S1A). Concerning NTA analysis of plasma-derived exosomes (n = 12) of nonmetastatic, metastatic CRC patients, and healthy donors, no significant differences were found among the three groups in both size and concentration (Fig. S2A). Western blot analysis showed that the isolated nanovesicles from all cells (Fig. S1B) and patients ( Fig. S2B) were positive for the exosomal marker proteins Tsg101, Alix, syntenin-1, CD9, and CD63 while negative for the endoplasmic reticulum (ER) marker calnexin.
The lipidomic profile of exosomes, analyzed by LC-MS and normalized to total cholesterol (nm) lead to the quantification of 175 lipid species in exosomes from both NCM460 and HCT116, 132 lipid species in SW620, and 178 lipid species in the three groups of plasma-derived exosomes (healthy donors, nonmetastatic, and metastatic, Table S1). The relative distribution of lipid compositions was considerably different among the exosomes. However, all exosomes were relatively abundant in sphingolipids (Figs S1C and S2C) and PC (Figs S1D and S2D), which is in agreement with the hypothesis that exosomal membranes harbor lipid raft-like domains [12] and are enriched in PC subclasses [8,13]. In metastatic patients, like metastatic SW620, exosomes possessed a smaller mole ratio of PS compared with nonmetastatic patients (Fig. S2D). Cholesterol was chosen to normalize the lipidomic, as it was an abundant lipid in all samples and no significant differences were detected among the different exosomes in the mole ratio of cholesterol (Fig. S3).
The lipidomic analysis was next extended to the individual molecular species of the identified lipid subclasses. Figures 1-3 show the subclasses for which differences were obtained when comparing controls with cancer and/or nonmetastatic from metastatic conditions (raw data obtained for all subclasses analyzed are shown in Figs S4-S8). Considering PC subclass, the molecular species, PC 30:0, 32:1, 34:2, 34:1, and 36:2 were significantly increased in HCT116 compared with control NCM460 (Fig. 1A). Interestingly, all these PC species were decreased in SW620 along with PC 32:0, 36:1, and 38:2 when compared both with the control and the nonmetastatic HCT116 (Fig. 1C,E). Only the molecular species phosphorylated PC 34:0 (pPC 34:0) was markedly increased in SW620 (Fig. 1C,E). In accord with this result, an increased level of PC molecular species 32:1 was reported in CRC tissues [14]. For CRC plasma-derived exosomes, nonmetastatic patients revealed a significant enrichment in the PC 34:1 and 36:5 molecular species compared with the healthy controls and metastatic patients ( Fig. 1B,F). Moreover, exosomes derived from cancer patients, compared with the healthy donor, showed a decrease in the level of PC 34:2 and 36:4 individual species (Fig. 1B,D). Interestingly, the significant increase in the PC molecular species 34:1 in nonmetastatic HCT116-exosomes was also observed in exosomes derived from plasma of nonmetastatic CRC patients when compared with their corresponding normal counterparts (Fig. 1A,B). It should be noted that the level of PC 34:1 was also found to be increased in the exosomes derived from NB26 and PC-3 prostate cancer cell lines [1].
For the PE subclass, the molecular species PE 32:1, 34:2, 36:2, and 36:1 were significantly decreased in SW620 compared with NCM460 and HCT116 (Fig. 1C, E). Like the nonmetastatic HCT116-exosomes, plasma exosomes derived from nonmetastatic patients revealed a significant increase in the PE individual species 36:2, compared with the control and metastatic patients (Fig. 1A,B). Similarly, exosomes derived from metastatic both SW620 cells and patients displayed a significant decrease in the level of PE 34:2 and 36:2 molecular species compared with their nonmetastatic counterparts (Fig. 1E,F). In addition, PE 38:5 and 38:4 were also found decreased in exosomes from metastatic patients (compared with healthy donors and nonmetastatic patients; Fig. 1D,F).

Conclusion
In summary, targeted lipidomic analysis can enable the description of potential diagnostic/prognostic cancer biomarkers. Some signature profiling can already be proposed. For instance, markers when comparing controls and primary cancers might be

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Error bars represent the standard error mean (AESED) values of four independent replicates (n = 4). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Fig. S3. Determination of cholesterol by gas chromatography-mass spectroscopy (GC-MS) in exosomes derived from both colorectal cancer (CRC) cell lines and patients compared with their corresponding controls (exosomes from NCM460 cells and healthy control, respectively). As depicted in the figure, no significant change in cholesterol was observed in the exosomes derived from both cell lines and patients. Data were analyzed by two-way ANOVA followed by the Tukey's multiple comparison test. Error bars represent the standard error mean (AESED) values of four independent replicates (n = 4). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Fig. S4. Phosphatidylcholine (PC) species analysis normalized to total cholesterol of exosomes derived from (A) normal colon mucosa NCM460D, nonmetastatic HCT116, and metastatic SW620 colorectal cancer (CRC) cell lines and from (B) Plasma-derived exosomes of healthy donors and CRC patients (nonmetastatic and metastatic) illustrating an overall enrichment of 34:1 PC molecular species in both nonmetastatic cells and patients compared with their corresponding controls and metastatic counterparts. Data were analyzed by two-way ANOVA followed by the Tukey's multiple comparison test. Error bars represent standard error mean values (AESEM, n = 4). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Fig. S5. Phosphatidylethanolamine (PE) and plasmalogen (pPE) molecular species analysis normalized to total cholesterol in exosomes derived from (A) normal colon mucosa NCM460D, nonmetastatic HCT116 and metastatic SW620 colorectal cancer (CRC) cell lines and from (B) plasma-derived exosomes of healthy donors and CRC patients nonmetastatic and metastatic (n = 4 for each group, pooled). Exosomes from both nonmetastatic HCT116 cells, and patients showed a significant increase of PE species 36:2 compared with their corresponding controls and metastatic counterparts. Metastatic SW620 cells and patients revealed a significant decrease in p16:0/20:4 pPE level compared with their nonmetastatic counterparts. Data were analyzed by two-way ANOVA followed by the Tukey's multiple comparison test. Error bars represent standard deviation (AESD, n = 4) values. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Fig. S6. Phosphatidylinositol (PI) and phosphatidylserine (PS) molecular species analysis of exosomes derived from (A) normal colon mucosa NCM460D, nonmetastatic HCT116, and metastatic SW620 colorectal cancer (CRC) cell lines and from (B) plasma-derived exosomes of healthy donors and CRC patients nonmetastatic and metastatic (n = 4 for each group, pooled) normalized to total cholesterol. HCT116-derived exosomes are enriched in the PI molecular species PI 34:1, 36:2, and 36:1 compared with NCM460D and SW620. No significant change in the level of PI species was detected in all exosomes derived from the plasma of healthy donors and patients. Data were analyzed by two-way ANOVA followed by the Tukey's multiple comparison test. Error bars represent standard deviation (AESD, n = 4) values. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Fig. S7. Analysis of sphingomyelin (SM) molecular species in exosomes derived from (A) normal colon mucosa NCM460D, nonmetastatic HCT116, and metastatic SW620 colorectal cancer (CRC) cell lines and from (B) plasma-derived exosomes of healthy donors and CRC patients nonmetastatic and metastatic normalized to total cholesterol (n = 4 for each group, pooled). Both nonmetastatic HCT116-and patientderived exosomes revealed a marked increase in the level of d18:1/16:0 SM molecular species compared with their corresponding controls and metastatic counterparts. Data were analyzed by two-way ANOVA followed by the Tukey's multiple comparison test. Error bars represent standard deviation (AESD, n = 4) values. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Fig. S8. Ceramide (Cer) molecular species analysis of exosomes derived from (A) normal colon mucosa NCM460D, nonmetastatic HCT116, and metastatic SW620 colorectal cancer (CRC) cell lines and from (B) plasma-derived exosomes of healthy donors and CRC patients nonmetastatic and metastatic, normalized to total cholesterol (n = 4 for each group, pooled). Nonmetastatic HCT116-and patient-derived exosomes had an increase in the level of hexosylceramide d18:1/24:1 HexCer and d18:1/24:0 HexCer molecular species compared with their controls and metastatic counterparts. Both metastatic SW620-and patient-derived exosomes displayed a significant increase in the ceramide molecular species d18:1/24:1 compared with their controls. Data were analyzed by two-way ANOVA followed by the Tukey's multiple comparison test. Error bars AESD, n = 4. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Table S1. Total lipid ions quantified by liquid chromatography-mass spectrometry (LC-MS) in exosomes derived from normal colon mucosa NCM460D, nonmetastatic HCT116, and metastatic SW620 colorectal cancer (CRC) cell lines and from plasma-derived exosomes of healthy controls (HC), and CRC patients nonmetastatic (NM) and metastatic (M).