key: cord-0692920-pizqox7r
authors: Han, Zhenzhen; Peng, Cheng; Yi, Jia; Wang, Yiwen; Liu, Qi; Yang, Yi; Long, Shuping; Qiao, Liang; Shen, Yuhui
title: Matrix-assisted laser desorption ionization mass spectrometry profiling of plasma exosomes evaluates osteosarcoma metastasis
date: 2021-07-26
journal: iScience
DOI: 10.1016/j.isci.2021.102906
sha: 66c2f7ff41e1af0154983d41de1af733bc4de744
doc_id: 692920
cord_uid: pizqox7r

Osteosarcoma is the most common primary sarcoma of bone among adolescents, often characterized by early lung metastasis resulting in high mortality. Recently, exosomes have been used in liquid biopsy to monitor tumors. Herein, we used matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) to profile human plasma exosomes for the evaluation of osteosarcoma lung metastasis. Forty patients with osteosarcoma with (n = 20) or without (n = 20) lung metastasis as well as 12 heathy controls were recruited. Exosomes were isolated from human plasma for MALDI-TOF MS analysis. Multivariate statistical analyses were performed based on the MALDI-TOF mass spectra. The strategy can efficiently differentiate osteosarcomas from healthy controls and further discriminate osteosarcoma lung metastasis from non-lung metastasis. We identified seven exosomal proteins as potential biomarkers of osteosarcoma lung metastasis. The proposed method holds great promise to clinically diagnose osteosarcoma and monitor osteosarcoma lung metastasis.

Osteosarcoma is the most common primary sarcoma of bone, particularly among young children and teenagers (Behjati et al., 2017) . Although the survival rate of patients with osteosarcoma has been improved to $70% by surgery combined with adjuvant chemotherapy (Kansara et al., 2014) , still about 20% of patients with osteosarcoma suffer from metastases (Meyers et al., 2011) , among which more than 85% metastatic lesion is detected in the lungs (Bielack et al., 2014) . The early diagnosis of osteosarcoma lung metastasis is crucial for the treatment and prognosis of osteosarcoma. Computed tomography (CT) scanning combining histopathologic biopsy is usually used to diagnose osteosarcoma lung metastasis, but is less efficient in the detection of small lesions (Luetke et al., 2014) . Moreover, the traditional biopsy methods require complex operations, which are highly invasive and have the risk of arousing tumor spread (Brock et al., 2015) .

As an alternative diagnosis method, liquid biopsy is emerging in recent years. The liquid biopsy involving exosomes has recently been reported to monitor tumor progression (Gerlinger et al., 2012) and long-term treatment response . Exosomes exhibit the size of 30 nm-150 nm, containing nucleic acids, lipids, proteins, etc. They are shed by almost all types of cells and found in various body fluids, such as plasma, urine, breast milk, saliva, etc. (Chen et al., 2018; Costa-Silva et al., 2015; Di et al., 2019; Tieu et al., 2020; Yang et al., 2020) . Exosomes play an important role in cellular communication, signal transduction and immune response and serve as potential biomarkers for various diseases (Jansen et al., 2009; Saenz-Pipaon et al., 2020) . The implications of exosomes in osteosarcoma have also been studied. It has been reported that osteosarcoma-derived exosomes can work as transfer cargo and are relevant with tumor progression and metastasis (Chicon-Bosch and Tirado, 2020; Galardi et al., 2019) . To date, the most widely used platforms (Witwer et al., 2013) to characterize exosomes (see Table S1 ) include 1) dynamic light scattering (DLS) (Palmieri et al., 2014) , nanoparticle tracking analysis (NTA) (Gardiner et al., 2013) and tunable resistive pulse sensing (TRPS) (Vogel et al., 2016) for size characterization; 2) Western blot (Rat et al., 2008) , enzyme-linked immunosorbent assay (ELISA) (Hoshino et al., 2020) and nano-flow cytometry (Welsh et al., 2020) for subgroup characterization; 3) mass spectrometry for proteomic characterization (Andaluz Hoshino et al., 2020) ; and 4) sequencing for genomic and transcriptomic Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) has attracted increasing interests in clinical diagnosis with its high throughput, low sample consumption, label-free and direct analysis ability, easy operation, and low cost per sample characterization (Jagadeesan and Ekstrom, 2017; Rathore et al., 2008) . It has been used to identify bacterial pathogens by mass fingerprinting for clinical usage (Varadi et al., 2017; Yi et al., 2018 Yi et al., , 2019 and determine treatment benefit from epidermal growth factor receptor inhibitors in patients with non-small cell lung cancer as a laboratorydeveloped clinical test (Butts, 2014) . It has also shown great potential in screening of Alzheimer's disease (Nakamura et al., 2018) , rapid diagnosis and treatment monitoring of active Mycobacterium tuberculosis (TB) disease (Liu et al., 2017) , and detection of COVID-19 (Yan et al., 2021) . Recently, it has been further utilized for the detection and analysis of exosomes. Stü biger et al. used MALDI-TOF MS to demonstrate the differential expression of proteins in extracellular vesicles (EVs) of different parental cells with an increasing chemoresistance (Stü biger et al., 2018) . Singhto et al. applied MALDI-TOF MS to discriminate the urinary exosomes from microvesicles (Singhto et al., 2019) . Lin et al. demonstrated that MALDI-TOF MS can identify urinary EVs proteins, i.e., alpha 1-antitrypsin and histone H2B1K, for the diagnosis and prognosis of urothelial carcinoma (Lin et al., 2016) . Nguyen et al. proved the platelet factor 4 as a biomarker of exosomes from human plasma by MALDI-TOF MS (Nguyen et al., 2019) . Zhu et al. used MALDI-TOF MS to diagnose melanoma and explore the disease progression (Zhu et al., 2019b) .

In this study, we demonstrate that MALDI-TOF MS profiling of plasma exosomes can be used to evaluate osteosarcoma lung metastasis (Scheme 1). We first analyzed the exosomes from normal osteoblast cells and osteosarcoma cells and found that the mass fingerprints of exosomes can better resolve different cells than the mass fingerprints of the cells themselves. Then, mass fingerprints of plasma-derived exosomes from patients with osteosarcoma with/without lung metastasis (n = 20 each) as well as healthy controls (n = 12) were collected. The plasma-derived exosomes from patients with osteosarcoma can be well distinguished from healthy controls by principal component analysis (PCA) and hierarchical clustering analysis of their MALDI-TOF mass spectra. The method can further discriminate osteosarcomas with lung metastasis from those without lung metastasis. Combining machine learning methods and LC-MS/MS-based proteomic analysis, we identified seven exosomal proteins from the MALDI-TOF mass spectra of osteosarcoma lung metastasis as potential biomarkers. These results demonstrated that the MALDI-TOF MS profiling of exosomes can be used for the diagnosis, treatment, and prognosis of osteosarcoma, with the advantages of rapid assay, high-throughput, low sample consumption, label-free, and low cost. 

Exosomes isolated from 143B osteosarcoma cells were used as model samples to illustrate the proposed method. First, exosomes were separated from 143B cell culture medium through differential centrifugation and then characterized by transmission electron microscopy (TEM). As shown in Figure 1A , typical cupshaped exosomes can be observed in accordance with previous report (Jerez et al., 2017) . The isolated exosomes were further characterized by Western blot for exosome-associated biomarkers, as well as by NTA for size distribution. As shown in Figure 1B , the exosomes were positive for CD9 and TSG101 biomarkers, while calnexin, a negative biomarker, only existed in cell lysate. As shown in Figure 1C , the exosomes had a homogeneous size distribution with an average diameter of 120 G 11 nm.

Subsequently, MALDI-TOF MS methods were optimized for profiling the exosomes. Sinapic acid (SA) was a more suitable MALDI matrix for exosome characterization than 2,5-dihydroxybenzoic acid (DHB) and a-cyano-4-hydroxycinnamic acid (HCCA), as illustrated in Figure S1 . MALDI-TOF mass spectrum of 143B osteosarcoma-cells-derived exosomes was acquired in the mass-to-charge (m/z) range 2000-15,000, as shown in Figure 1D . As shown in Figure S2 , the mass spectra across batches were highly consistent, demonstrating the sound stability of MALDI-TOF MS platform.

Then, MALDI-TOF mass spectra of exosomes isolated from the normal osteoblast cells (3T3E1) and the osteosarcoma cells (143B, HOS, MG63 and U2OS) are presented in Figure 2A . It can be observed that each type of exosomes has its own characteristic mass fingerprints. For example, the mass spectrum of 3T3E1 exosomes had four strong specific peaks at m/z = 4964, 9969, 10,958, and 11,314. The mass spectrum of 143B exosomes had eight strong specific peaks at m/z = 3805, 4153, 5468, 8102, 8230, 9422, 9715, and iScience Article osteosarcoma exosomes. HOS exosomes, MG63 exosomes, and U2OS exosomes had identical and differential peaks. For comparison, MALDI-TOF mass spectra of different cells were also collected and are shown in Figure S3A . It was observed that all the osteosarcoma cells shared similar mass fingerprints.

The mass fingerprints of different exosomes and cells were analyzed by PCA and hierarchical clustering, as shown in Figures 2B, S3B , 2C, and 2D. As shown in Figure 2B , the mass spectra of different exosomes can be clearly grouped into five sets and completely distinguished by PCA. In the Figure S3B , the mass spectra of different cells cannot be well distinguished. The hierarchical clustering heatmaps are displayed in Figures  2C and 2D . The mass spectra of exosomes derived from same cells were classified into same clusters. However, on the cell level, the data of 143B cells were wrongly clustered. All of these results demonstrated that MALDI-TOF mass spectra of exosomes can better discriminate different cells than those of the whole cell.

We have also performed the hierarchical clustering combining the MALDI-TOF mass spectra of exosomes and cells. As shown in Figure S4 , all the cells were clustered together. The exosomes from HOS, 143B and MG63 were clustered together. The exosomes from 3T3E1 and U2OS were clustered together and closer to the cells than the other exosomes. Nevertheless, even for 3T3E1 and U2OS, the mass fingerprints of the cells were more similar to the other cells than the corresponding exosomes. The results suggest that the mass spectra differences between biological entities are more significant than those among the cells of origin.

Based on the success in the cell model, the MALDI-TOF profiling of exosomes was applied to real clinical samples (Table S2) . Plasma-derived exosomes were isolated from patients with osteosarcoma and healthy controls and then characterized by NTA, TEM, and Western Blot. As shown in Figure S5 , the exosomes isolated from patients with osteosarcoma and healthy controls were typical cup-shaped, positive for CD63, CD9, and CD81, and had a homogeneous size distribution with an average diameter of 120 G 10 nm. Figure 3A shows the representative exosome mass fingerprints of a patient and a healthy volunteer. Their fingerprint spectra were significantly different in the mass range of m/z 2000 to 20,000. PCA was performed to the MALDI-TOF mass spectra of plasma-derived exosomes from 12 patients with osteosarcoma and 12 healthy controls. The peak intensity matrix of all the samples is listed in the Data S1, which is derived from the raw data after data preprocessing and peak alignment by MALDIquant (Gibb and Strimmer, 2012) . As shown in Figure 3B , healthy volunteers and patients with osteosarcoma can be clearly classified into two sets in score plots of PCA based on the mass spectra of exosomes. Hierarchical clustering was also used to analyze the mass fingerprints of the exosomes. As shown in Figure S6 , hierarchical clustering indicated that the patients with osteosarcoma can be well distinguished from the healthy controls. Then, partial least squares-discriminant analysis (PLS-DA), a supervised classification method, was applied to find out the distinctive features on the exosomal mass fingerprints between patients with osteosarcoma and healthy volunteers ( Figure 3C ). Thirty-two peaks with variable importance in the projection (VIP) value >1.5 were considered as potential biomarkers.

To demonstrate whether the MALDI-TOF profile difference between patients with osteosarcoma and healthy controls is osteosarcoma-specific or cancer-specific, we have collected the plasma of five patients with lung cancer. Exosomes were isolated from the plasma samples and then characterized by MALDI-TOF MS. Figure S7A shows a representative mass spectrum of plasma-derived exosomes from a patient with lung cancer. PCA was performed on the MALDI-TOF mass spectra of the plasma-derived exosomes from 12 patients with osteosarcoma, 12 healthy controls, and 5 patients with lung cancer. As shown in Figure S7B , the mass spectra of exosomes can be clearly classified into three sets, i.e., healthy controls, patients with osteosarcoma, and patients with lung cancer. Hierarchical clustering was also used to analyze the mass spectra of the exosomes. As shown in Figure S7C , hierarchical clustering indicated that the patients with osteosarcoma or patients with lung cancer can be well distinguished from the healthy controls, and patients with osteosarcoma can also be distinguished from patients with lung cancer. The distance between the different types of cancer is significantly smaller than the distance between patients with cancer and healthy controls. Therefore, the exosomes from patients with cancer were significantly different in MALDI-TOF mass spectra compared with those from healthy volunteers, while the exosomes from patients with different kinds of cancer were also different in MALDI-TOF mass spectra.

iScience 24, 102906, August 20, 2021 5 iScience Article

With the success on the classification of patients with osteosarcoma from healthy controls, we further studied the possibility of using MALDI-TOF MS profiling of plasma-derived exosomes for the evaluation of osteosarcoma lung metastasis. Plasma-derived exosomes isolated from 10 patients with osteosarcoma non-lung metastasis and 10 patients with lung metastasis (Table S2) were measured by NTA, TEM, Western blot, and MALDI-TOF MS. As shown in Figure S5 , the exosomes were typical cup-shaped, positive for CD63, CD9, and CD81, and had a homogeneous size distribution with an average diameter of around 120 G 6 nm. Figure 4A shows the representative MALDI-TOF MS fingerprints of plasma-derived exosomes from a patient with osteosarcoma lung metastasis and a patient with non-lung metastasis. The fingerprint spectra are clearly different in the range of m/z 2000 to 20,000. By PCA classification, the mass spectra of plasma-derived exosomes from the patients with osteosarcoma non-lung metastasis and patients with lung metastasis can be separated into two discriminative groups, shown in Figure 4B . As shown in Figure S8 , the hierarchical clustering can also classify the patients with osteosarcoma lung metastasis and patients with non-lung metastasis based on the mass spectra. The results demonstrated the potential of MALDI-TOF profiling of plasma-derived exosomes for the diagnosis of osteosarcoma lung metastasis.

To find out the potential biomarkers of osteosarcoma lung metastasis, PLS-DA was performed on the MALDI-TOF mass spectra of plasma-derived exosomes from the patients with 10 osteosarcoma nonlung metastasis and the 10 patients with lung metastasis. Figure 4C lists the VIP of top 15 features, representing the contribution of each feature to the PLS-DA model, and there were 33 peaks with VIP value >1.5, iScience Article which can be considered as candidate biomarkers. To identify the MALDI-TOF features, proteomic analysis was executed on the pooled plasma-derived exosomes isolated from the patients with osteosarcoma lung metastasis and patients with non-lung metastasis. A total of 353 proteins were identified from the plasmaderived exosomes (Data S2). The MALDI-TOF mass spectra peaks were matched to the proteomic analysis results under the criteria that MALDI-TOF must detect at least two continuous charge states from 1+, 2+, and 3+ of a protein with an m/z tolerance of 2000 ppm. Finally, 7 proteins matched with the MALDI-TOF significant features of osteosarcoma lung metastasis selected by the PLS-DA analysis as candidate biomarkers (Table 1) , including immunoglobulin lambda variable 2-23 (IGLV2-23), immunoglobulin lambda variable 4-3 (IGLV4-3), immunoglobulin lambda variable 1-51 (IGLV1-51), immunoglobulin kappa variable 3-15 (IGKV3-15), immunoglobulin heavy variable 4-4 (IGHV4-4), immunoglobulin lambda variable 4-60 (IGLV4-60), and hemoglobin subunit alpha (HBA1). To further evaluate the discriminatory power of the 7 potential biomarkers, cross-validation receiver operating characteristic (ROC) analysis was performed.

The ROC result showed an area under the curve value of 1.0 ( Figure 4D ), demonstrating that the 7 proteins can serve as potential biomarkers of osteosarcoma lung metastasis. To further validate the 7 protein biomarkers for osteosarcoma lung metastasis, another 10 patients with osteosarcoma lung metastasis and 10 patients with non-lung metastasis were recruited as pseudo-blind samples for model test. PCA model was built using the 7 protein biomarkers on the MALDI-TOF data of the training set shown in Figure 4 . As shown in Figures 5A and 5B , the test samples were correctly classified, revealing the 7 protein biomarkers robust for the identification of osteosarcoma lung metastasis. The profiling of gene ontology (GO) analysis in Figure 5C showed the biological functions of the 7 protein biomarkers.

The predilection age of osteosarcoma is 10-14 years in line with the age of bone growth. The current treatments can achieve a 70% cure rate of patients with osteosarcoma (Kansara et al., 2014). However, metastasis is the leading cause of death of patients with osteosarcoma (Kohama et al., 2019) . Metastatic sites of osteosarcoma are detected primarily in the lungs. Therefore, there is an urgent need to develop a robust and straightforward method to diagnose lung metastasis of osteosarcoma, in order to improve the survival of patients with osteosarcoma.

Exosomes are small vesicles with lipid bilayers with the diameter of 30-150 nm, containing nucleic acids, proteins, and lipids . They are considered to be the means by which cells transmit information, and their delivery cargos can vary greatly depending on the function of the original cell type and its current state (Colombo et al., 2014) . It has been reported that exosomes are associated with many human diseases and can serve as surrogate biomarkers for the diagnosis of tumor progression and metastasis instead of invasive clinical procedures (Le et al., 2014) . iScience Article identify specific subpopulation of exosomes, resulting in the missing information of other extracellular subpopulations, and cannot provide biomolecular composition of the whole exosomes. MALDI-TOF MS is a powerful approach for the profiling of exosomes with advantages in several aspects: 1) label-free for the analysis of exosomal composition; 2) fast detection; and 3) high-throughput analysis. In MALDI-TOF MS analysis, matrix is used to isolate analyte molecules, absorb laser energy, and assist the ionization of analyte molecules. Therefore, it is critical to select an appropriate matrix for MALDI-TOF MS analysis. The commonly used MALDI matrices in protein and peptide analysis include SA, DHB, and HCCA. DHB has strong polarity, generates large crystals, leads to poor uniformity, and is normally used for peptides, glycopeptides, glycoproteins, and proteins (Kaletas x et al., 2009) . HCCA and SA matrices can generate small crystals and lead to good homogeneity. HCCA is mainly utilized in polypeptide analysis (Cohen and Chait, 1996; Wiangnon and Cramer, 2015) , whereas SA is more suitable for proteins (Marvin et al., 2003) . Moreover, it was demonstrated that SA can provide good reproducibility in the detection of biological samples in a wide mass range (Zhu et al., 2018 (Zhu et al., , 2019b , which is in accordance with our results. Hence, SA was chosen in this work for the analysis of exosomes. We collected the MALDI-TOF mass spectra of plasma-derived exosomes secreted by patients with osteosarcoma with or without lung metastasis, patients with lung cancer, and healthy volunteers. By unsupervised statistical analysis (PCA and hierarchical clustering), it was clear that the mass spectra of plasmaderived exosomes were different between patients with cancer and healthy controls, between patients with different kinds of cancer, and between patients with osteosarcoma with and without lung metastasis. Both males and females were included in the same type of samples, and we did not observe any effect from sex on the results of the study. The significant MALDI-TOF features between the patients with and without lung metastasis were identified by supervised statistical analysis (PLS-DA) and LC-MS/MS-based proteomic analysis. The identified significant features were further validated by using another group of pseudo-blind samples that was independent from the ones used for model training. Seven protein biomarkers were identified for the osteosarcoma lung metastasis, mostly immunoglobulin-related proteins. Immunoglobulins are found at high frequency in plasma-derived exosomes samples and are often associated with various cancer types (Ndede et al., 2019; Shuai et al., 2019; Son et al., 2020; Zhang et al., 2012) . It was reported that immunoglobulins in exosomes are closely related to the proliferation, invasion, and cancer metastasis (Jerez et al., 2017) . GO analysis in Figure 5C shows that the identified significant proteins participate in the regulation of Fc-gamma receptor signaling pathway and Fc-epsilon receptor signaling pathway, which are involved in the adjustment of the tumor microenvironment (Milane et al., 2015) and are essential to activate immune response to tumor cells. IGLV2-23, IGLV1-51, and IGKV3-15 are components of extracellular exosomes and participate in all the biological processes listed in Figure 5C . IGLV2-23 has been identified in EVs from lung transplant recipients and reported to play a significant role in the transport of small molecules, signal transduction, and extracellular matrix organization (Bansal et al., 2020) . IGLV4-3 with increased L-phytohemagglutinin-reactive glycoproteins is involved in the response of breast carcinoma tumor cells to environmental stress (Abbott et al., 2008; Andaluz Aguilar et al., 2020) . IGLV1-51 serves as the validating candidate recurrent fusion gene in patients with hepatocellular carcinoma (Zhu et al., 2019a) . IGKV3-15 was used for the development of cancer vaccines in B cell non-Hodgkin lymphomas (Martorelli et al., 2012) . IGHV4-4 is involved in phagocytosis, innate immune response, and complement activation (Gaudet et al., 2011) and presents in patients with chronic lymphocytic leukemia (CLL) as an unmutated form (Karan-Djurasevic et al., 2012) . The exosomal protein, IGLV4-60, participates in antigen binding and is related with immune activation, including immunoglobulin synthesis, programmed cell death, and DNA repair (Bansal et al., 2020) . IGLV4-60 is expressed at higher frequency in the cohort of 76 l + CLL cases than that in the 97 l + CLL samples (Widhopf et al., 2008) . The exosomal protein, HBA1, is involved in receptor-mediated endocytosis, oxygen transport, positive regulation of cell death, and other biological processes (Keerthikumar et al., 2016; Mathivanan et al., 2012; Mathivanan and Simpson, 2009) . HBA1 is utilized for the monitor of glycemic control and can serve as a biomarker in the regulation of prostate cancer (Wu et al., 2020) . Furthermore, these proteins in exosomes are related to complement activation with an immune evasion mechanism (Brady et al., 2018) and complement activation within tumor microenvironment to increase tumor growth and disease progression (Gadwa and Karam, 2020) . The receptor-mediated endocytosis is beneficial for targeted cancer therapy (Jia et al., 2020) . Immune response and its regulation is closely related to cancer development and tumor progression (Becker et al., 2016; Hernandez et al., 2016) . Based on the biological roles of the identified significant proteins, the rationality of the biomarkers and the MALDI-TOFbased exosomes profiling method for the detection of osteosarcoma lung metastasis are further demonstrated.

In summary, we show that MALDI-TOF mass fingerprinting of exosomes has great potential in clinical application. The detection and analysis of exosomes will contribute to the diagnosis and intervention of osteosarcoma, thereby being conducive to early prevention. The method can be further applied to find stable biological differences between early cancer metastasis and premetastatic groups, healthy and premetastatic groups, chemotherapy-resistant and nonresistant groups, as well as different subtypes of cancer.

Although MALDI-TOF MS profiling of plasma-derived exosomes can distinguish patients with osteosarcoma from healthy controls and distinguish osteosarcoma lung metastasis from non-lung metastasis, the exosomes were collected by ultracentrifugation, which is inefficient and time-consuming. It is preferred to develop efficient methods, e.g., based on microfluidics, for high-throughput purification of exosomes from plasma, and to intergrade the exosomes purification methods with the MALDI-TOF analysis in the follow-up work. 

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Liang Qiao (liang_qiao@fudan.edu.cn).

All reagents generated in this study are available without restriction. This study did not generate new unique reagents.

The data of this study are available within the article and supplementary information. The peak intensity matrix of all the samples is listed in Sheet AlignmentMatrix (Data S1), which is derived from the raw MALDI-TOF data after data preprocessing and peak alignment by MALDIquant. Machine learning data are listed in Sheet Osteosarcoma Discrimination, Sheet Osteosarcoma Specificity, Sheet TrainingCohort, and Sheet TestCohort (Data S1), which are derived from Sheet AlignmentMatrix after log2 transformation and quantile normalization.

The raw proteome data have been deposited to ProteomeXchange via the PRIDE (Deutsch et al., 2020; Perez-Riverol et al., 2016 partner repository with the data set identifier PXD024072 and Mendeley Data (https://doi.org/10.17632/r5snnfc89s.1), which are publicly available. Accession numbers are listed in the key resources table.

There is no original code reported in this work.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

The high aggressive osteosarcoma cell lines (143B and HOS cells) (ATCC, Manassas, VA) and low-invasive osteosarcoma cell lines (MG63 and U2OS cells) (ATCC, Manassas, VA) were cultured in Dulbecco's modified eagle medium (DMEM) with 10% EV-depleted fetal bovine serum (FBS) and 1% penicillin-streptomycin at 37 C in a humid atmosphere with 5% CO 2 . When the cells reached 40% confluency (Keklikoglou et al., 2019) , they were washed twice by 0.01 M phosphate buffered saline (PBS) and then cultured for 24 h in the DMEM with 10% exosomes-depleted FBS and 1% penicillin-streptomycin to reach a confluency of 80%-90%. The normal osteoblast cells, 3T3E1 cells (ATCC, Manassas, VA), were cultured by the same method except that DMEM was replaced by MEM Alpha. The thawed FBS was ultracentrifuged at 120,000 g for 14 h and then filtered with a 0.22-um filter to obtain the exosomes-depleted FBS.

Forty patients with osteosarcoma with (n = 20) or without (n = 20) lung metastasis and 5 patients with lung cancer as well as 12 heathy controls were randomly recruited from our longitudinal observational clinical research project. Twenty patients with osteosarcoma lung metastasis consisted of 14 males aged 13-62 years and 6 females aged 13-46 years. Twenty patients with osteosarcoma non-lung metastasis consisted of 13 males aged 5-38 years and 7 females aged 11-69 years. Five patients with lung cancer consisted of 4 males aged 51-67 years and 1 female aged 60 years. Twelve healthy controls consisted of 8 males aged 10-57 years and 4 females aged 30-39 years. The detailed gender, age, and developmental stage of subjects can be found in Table S2 . Five milliliters of fresh blood was collected from each of subjects. Blood samples collected in disposable vacuum blood vessel collection tubes were kept for 10 min at room temperature, followed by centrifugation at 2000 g for 10 min and 3500 g for 10 min to eliminate cells and debris. Finally, the plasma samples were collected and stored at À80 C for further usage.

iScience Article and statistics-sensitive nonlinear iterative peak-clipping (SNIP) baseline correction. The warpMassSpectra command was performed for the mass value alignment. The signal-to-noise ratio of peak detection was set to 3, and the half window size was set to 20. Peaks were removed with the binPeaks command with a tolerance of 0.002. Finally, a peak intensity matrix was obtained including the m/z values and intensity of peaks normalized by the sum intensity of the corresponding mass spectrum. Multivariate statistical analyses, including data transformation (log2 transformation and autoscaling), PCA, PLS-DA, hierarchical clustering, and multivariate ROC curve analysis, were performed on the obtained peak intensity matrix using Metaboanalyst 4.0 (McGill University, Montreal, Canada, https://www.metaboanalyst.ca/) (Chong et al., 2018) and BacteriaMS (Fudan University, Shanghai, China, http://bacteriams.com/) (Yang et al., 2017b) . Briefly, the statistical analysis was performed on the MetaboAnalyst 4.0 web server, where the peak intensity matrix was uploaded, followed by normalization by sum, log2 transformation, and autoscaling. Then, analysis paths, such as PCA, PLS-DA, etc, were chosen to explore the results. Finally, the figures of multivariate statistical analyses can be directly obtained using Metaboanalyst 4.0 analysis, and other figures of mass spectra were plotted by Origin software (OriginLab, Northampton, MA, USA). CorelDraw Graphics Suite software (Corel, Ottawa, Canada) was used for typesetting the figures.

Eighty microliters of exosomes isolated from human plasma was ground with shock for 33 400 s, lysed in ice-water bath for 30 min using a lysis solution (1% SDS, 8 M urea in water, and 1x Protease Inhibitor Cocktail), and centrifuged at 15,000 rpm for 15 min at 4 C to obtain the supernatant. One hundred micrograms of the extracted protein was adjusted to 100 mL of 8 M urea. Two microliters of 0.5 M tris(2-carboxyethyl) phosphine (TCEP) was added, followed by the addition of 4 mL of 1 M iodoacetamide. The reaction lasted 40 min at room temperature in dark. Then, 400 mL of precooled acetone was added for protein precipitation overnight at À20 C followed by centrifugation at 12,000 g for 20 min at 4 C. One milliliter of precooled acetone was added, vortexed, and then centrifuged as described earlier. After drying at room temperature, the precipitate was resuspended in 100 mL of 100 mM triethylamonium bicarbonate (TEAB). Trypsin (Promega, Madison, Wisconsin, United States) was added according to the mass ratio of enzyme to protein = 1:50 and hydrolyzed overnight at 37 C. Then, the concentration of the digested peptide was determined by the Pierce quantitative peptide assay kit (Thermo Scientific, San Jose, USA). The final peptides were desalted using a MonoSpin TM C18 column (GL Science Inc., Tokyo, Japan).

The peptides were redissolved in solvent A (A: 0.1% formic acid in water) and analyzed by an on-line nanoelectrospray LC-MS/MS using Orbitrap Fusionä Lumosä Tribridä (Thermo Fisher Scientific, MA, USA) coupled to EASY-nLC 1200 system (Thermo Fisher Scientific, MA, USA). Two microliters of peptide was loaded to an analytical column (Acclaim PepMap C18, 75 mm x 15 cm) and separated with a 60-min gradient, from 5% to 32% B (B: 0.1% formic acid in ACN) . The column flow rate was maintained at 300 nL/min with the column temperature of 40 C. The electrospray voltage of 2 kV versus the inlet of the mass spectrometer was used. The mass spectrometer was run under data-dependent acquisition mode and automatically switched between MS and MS/MS mode. The parameters were as follows: (1) MS: scan range (m/z) = 350-1400; resolution = 120,000; AGC target = 5e5; maximum injection time = 50 ms; include charge states = 2-6; dynamic exclusion = 30 s; (2) HCD-MS/MS: resolution = 15,000; isolation window = 1.6; AGC target = 5e4; maximum injection time = 50 ms; collision energy = 28.

LC-MS/MS data were processed using PEAKS Studio version X+ (Bioinformatics Solutions Inc., Waterloo, Canada). PEAKS DB was set up to search Homo_sapiens_sp_201907 databases (20,414 entries) assuming trypsin as the digestion enzyme. PEAKS DB was searched with a fragment ion mass tolerance of 0.02 Da and a parent ion tolerance of 7 ppm. Carbamidomethylation (C) was specified as the fixed modifications. Oxidation (M) and acetylation (Protein N-term) were specified as the variable modifications. Peptides were filter by 1% false discovery rate (FDR). Proteins were identified with at least 1 unique peptide with 1% FDR.

For an in-depth understanding of the identified significant proteins, GO enrichment was performed using DAVID (https://david.ncifcrf.gov/summary.jsp), an online bioinformatics tool for annotating the function of genes or proteins (Huang da et al., 2009) . Briefly, the identified significant proteins were imported into the DAVID online database for reannotation. The parameters were as follows: 1) List: protein-coding gene names; 2) Identifier: official gene symbol; 3) Species: Homo sapiens. Then, the enrichment results of three ll OPEN ACCESS

Targeted glycoproteomic identification of biomarkers for human breast carcinoma

Sequential phosphoproteomics and N-glycoproteomics of plasma-derived extracellular vesicles

Extracellular vesicles in cancer: cell-to-cell mediators of metastasis

Liquid biopsy for cancer screening, patient stratification and monitoring

VeriStrat validated in patients with non-small-cell lung cancer

Exosomal PD-L1 contributes to immunosuppression and is associated with anti-PD-1 response

Exosomes in bone sarcomas: key players in metastasis

Mapping subpopulations of cancer cell-derived extracellular vesicles and particles by nano-flow cytometry

0: towards more transparent and integrative metabolomics analysis

Influence of matrix solution conditions on the MALDI-MS analysis of peptides and proteins

Biogenesis, secretion, and intercellular interactions of exosomes and other extracellular vesicles

Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver

The ProteomeXchange consortium in 2020: enabling 'big data' approaches in proteomics

General approach to engineering extracellular vesicles for biomedical analysis

Deciphering the intricate roles of radiation therapy and complement activation in cancer

Exosomal miRNAs in pediatric cancers

Extracellular vesicle sizing and enumeration by nanoparticle tracking analysis

Phylogenetic-based propagation of functional annotations within the Gene Ontology consortium

Intratumor heterogeneity and branched evolution revealed by multiregion sequencing

MALDIquant: a versatile R package for the analysis of mass spectrometry data

Damage-associated molecular patterns in cancer: a double-edged sword

Extracellular vesicle and particle biomarkers define multiple human cancers

Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources

Emerging technologies for profiling extracellular vesicle heterogeneity

MALDIViz: a comprehensive informatics tool for MALDI-MS data visualization and analysis

Exosomal secretion of cytoplasmic prostate cancer xenograft-derived proteins

Proteomic analysis of exosomes and exosome-free conditioned media from human osteosarcoma cell lines reveals secretion of proteins related to tumor progression

Functionalized graphene@gold nanostar/lipid for pancreatic cancer gene and photothermal synergistic therapy under photoacoustic/photothermal imaging dual-modal guidance

An insight into the roles of microRNAs and exosomes in sarcoma

MiR-200-containing extracellular vesicles promote breast cancer cell metastasis

Long non-coding RNAs regulate drug resistance in cancer

Osteosarcoma treatment -where do we stand? A state of the art review

IGKV3 proteins as candidate "off-the-shelf" vaccines for kappa-light chain-restricted B-cell non-Hodgkin lymphomas

Matrix-assisted laser desorption/ionization timeof-flight mass spectrometry in clinical chemistry

ExoCarta 2012: database of exosomal proteins, RNA and lipids

ExoCarta: a compendium of exosomal proteins and RNA

Addition of pamidronate to chemotherapy for the treatment of osteosarcoma

Exosome mediated communication within the tumor microenvironment

High performance plasma amyloid-beta biomarkers for Alzheimer's disease

Immunoglobulin heavy variable (IgHV) gene mutation and micro-RNA expression in Burkitt's lymphoma at Moi Teaching and Referral Hospital in Western Kenya

Platelet factor 4 as a novel exosome marker in MALDI-MS analysis of exosomes from human serum

Dynamic light scattering for the characterization and counting of extracellular vesicles: a powerful noninvasive tool

The PRIDE database and related tools and resources in 2019: improving support for quantification data

PRIDE inspector toolsuite: moving toward a universal visualization tool for proteomics data standard formats and quality assessment of ProteomeXchange datasets

Detergent binding explains anomalous SDS-PAGE migration of membrane proteins

Development of an inhibitor screening platform via mass spectrometry

Functional and transcriptomic analysis of extracellular vesicles identifies calprotectin as a new prognostic marker in peripheral arterial disease (PAD)

The U1 spliceosomal RNA is recurrently mutated in multiple cancers

Discrimination of urinary exosomes from microvesicles by lipidomics using thin layer liquid chromatography (TLC) coupled with MALDI-TOF mass spectrometry

Quantitative proteomic analysis of bile in extrahepatic cholangiocarcinoma patients

MALDI-MS protein profiling of chemoresistance in extracellular vesicles of cancer cells

Isolation and characterization of exosomes from cell culture supernatants and biological fluids

An analysis of mesenchymal stem cell-derived extracellular vesicles for preclinical use

Methods for the detection and identification of pathogenic bacteria: past, present, and future

A standardized method to determine the concentration of extracellular vesicles using tunable resistive pulse sensing

Exosomal PD-L1 and N-cadherin predict pulmonary metastasis progression for osteosarcoma patients

A nanodrug consisting of doxorubicin and exosome derived from mesenchymal stem cells for osteosarcoma treatment in vitro

MIFlowCyt-EV: a framework for standardized reporting of extracellular vesicle flow cytometry experiments

Macrophage-derived exosomes mediate osteosarcoma cell behavior by activating AKT signaling

Rapid detection of COVID-19 using MALDI-TOFbased serum peptidome profiling

A next generation sequencing based approach to identify extracellular vesicle mediated mRNA transfers between cells

Bacterial whole cell typing by mass spectra pattern matching with bootstrapping assessment

Multifunctional detection of extracellular vesicles with surface plasmon resonance microscopy

Identification of pathogenic bacteria in human blood using IgGmodified Fe 3 O 4 magnetic beads as a sorbent and MALDI-TOF MS for profiling

Plasmonic colloidosome-based multifunctional platform for bacterial identification and antimicrobial resistance detection

Expression of immunoglobulin G in esophageal squamous cell carcinomas and its association with tumor grade and Ki67

Detection of antimicrobial resistance-associated proteins by titanium dioxide-facilitated intact bacteria mass spectrometry

The fusion landscape of hepatocellular carcinoma

MALDI detection of exosomes: a potential tool for cancer studies

Briefly, 45 mL of culture medium was centrifuged at 500 g for 5 min, 2000 g for 15 min, and 10,000 g for 30 min under 4 C and then filtered through a 0.45-mm filter to remove dead cells, debris, and microvesicles. Subsequently, the supernatant was ultracentrifuged at 110,000 g for 11 h, and then the pellets were resuspended in 13PBS followed by ultracentrifugation at 110,000 g for 90 min under 4 C. Finally, the collected exosomes were resuspended in 100 mL of deionized (DI) water and kept under À80 C. To isolate exosomes from plasma sample, 8 mL of 83 diluted plasma by PBS was centrifuged at 10,000 g for 30 min and then

Twenty microliters of exosome pellets were loaded on a 200-mesh carbon-coated copper grid, negatively stained with 2% phosphotungstic acid for 10 min, dried under an infrared lamp, and then observed under a working voltage of 200 kV. Exosomes were also analyzed using a ZetaViewâ BASIC Nanoparticle Tracking Video Microscope

Total proteins (30 mg) were loaded and separated on 10% sodium dodecyl sulfate polyacrylamide gel electrophores (SDS-PAGE) Bis-Tris Gels (Biotech Well, Hong Kong, China) and then transferred to a 0.45-mm polyvinylidene fluoride (PVDF) membrane (Millipore

mouse anti-CD81 antibody

mouse anti-CD9 antibody

), rabbit anti-TSG101 antibody

and mouse anti-Calnexin antibody

To investigate the expression of CD63, CD81, CD9, TSG101, and Calnexin, the separation was performed under nonreducing conditions. For the other proteins, the separation was performed under reducing conditions. Then, the corresponding secondary horseradish peroxidase (HRP) conjugated antibodies were used to incubate with the membrane, and then the immunoreactive bands were detected with a chemiluminescent substrate

) equipped with a pulsed 337-nm nitrogen laser using SA (20 mg/mL in 50% acetonitrile

dried under ambient condition, and then overlayered with 1 mL of matrix. When the matrix was dried, MALDI-TOF MS analysis was performed under linear positive mode. The optimized parameters were as follows: 70% laser intensity, laser attenuator with 35% offset and 40% range, accumulation of 500 laser shots, 10.33 detector gain, and 150 ns delayed extraction time

Briefly, the mass range was set to 2 to 20 kDa. Then, the square root transformation was performed, followed by Savitzky-Golay smoothing ll OPEN ACCESS iScience 24

Z.H. performed the experiments and data analysis and wrote the original draft. C.P. performed the experiments. J.Y., Q.L., and Y.Y. helped with the data processing. S.L. collected the clinical samples. L.Q. revised the original draft of the manuscript. Y.W. modified the manuscript. L.Q. and Y.S. acquired the funding for this research and supervised all aspects of the work. All authors were involved in the design of the study. zThese authors contribute equally to this work.

The authors declare no competing interests.

We worked to ensure ethnic or other types of diversity in the recruitment of human subjects. We worked to ensure that the study questionnaires were prepared in an inclusive way. The author list of this paper includes contributors from the location where the research was conducted who participated in the data collection, design, analysis, and/or interpretation of the work.