Raman spectroscopy uncovers biochemical tissue-related features of extracellular vesicles from mesenchymal stromal cells

Abstract

Extracellular vesicles (EVs) from mesenchymal stromal cells (MSC) are emerging as valuable therapeutic agents for tissue regeneration and immunomodulation, but their clinical applications have so far been limited by the technical restraints of current isolation and characterisation procedures. This study shows for the first time the successful application of Raman spectroscopy as label-free, sensitive and reproducible means of carrying out the routine bulk characterisation of MSC-derived vesicles before their use in vitro or in vivo, thus promoting the translation of EV research to clinical practice. The Raman spectra of the EVs of bone marrow and adipose tissue-derived MSCs were compared with human dermal fibroblast EVs in order to demonstrate the ability of the method to distinguish the vesicles of the three cytotypes automatically with an accuracy of 93.7%. Our data attribute a Raman fingerprint to EVs from undifferentiated and differentiated cells of diverse tissue origin, and provide insights into the biochemical characteristics of EVs from different sources and into the differential contribution of sphingomyelin, gangliosides and phosphatidilcholine to the Raman spectra themselves.

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Figure 1

EV characterisation by means of Western blotting and transmission electron microscopy. (A) Western blot analysis of extracellular vesicles-enriched fractions (EVs) produced by BM-MSC, ASCs and DFs using the indicated antibodies. Flotillin-1, CD63 and CD9 are positive markers for EVs and Calnexin is considered a negative one. Corresponding cell lysate (CL) was used as control and depicts the specificity of the three antibodies. Western blots were cropped to improve clarity. All bands within the range of the molecular markers were retained and processing of the film was applied equally across the entire image. (B–D) Representative transmission electron photomicrographs of ultracentrifuged EVs from BM-MSCs (B), ASCs (C) and DFs (D) conditioned medium. The TEM images were used for size measurements. Bars = 100 nm.

Figure 2

Raman fingerprint of BM-MSCs, ASCs, and DFs. Average Raman spectra obtained using an excitation wavelength of 532 nm and 10 seconds of exposure for 2 accumulations for each spectrum. The solid black line indicates the average of 40–50 spectra ± 1 standard deviation (shaded grey areas). The Raman bands corresponding to lipids are highlighted in yellow (500–540 cm−1; the band centred at 700 cm−1; 1700–1740 cm−1; 2850–2950 cm−1), those corresponding to proteins are in blue (1200–1300 cm−1; the band centred at 1450 cm−1), and those corresponding to nucleic acids are in red (720–820 cm−1; the band centred at 915 cm−1; 1060–1100 cm−1; the band centred at 1360 cm−1). The arrows indicate the 1003 cm−1 peak of Phenylalanine.

Figure 3

Multivariate analysis of the Raman spectra. (A) Raman spectra of reference lipid molecules, PC1 and PC2 that were considered for CLS fitting. The grey lines highlight the correspondences between the peaks of the standard lipids and the PC loadings. (B) Scatter plot of the PCA results showing the PC1 and PC2 scores assigned to each spectrum recorded from the EVs of BM-MSCs (pink), ASCs (blue) and DFs (green). Each square represents one spectrum. The scatter plot shows that a positive PC1 score described ASC-derived EVs better than those derived from DFs and BM-MSCs, whereas a positive PC2 score better described EVs derived from BM-MSC. (C) CLS scores obtained after fitting the spectra of Cer (light green), Chol (green), SM (light blue), GM1 (blue), PCh (pink), PE (red) and PA (violet) with the PC1 and PC2 loadings. The positive and negative scores obtained after CLS fitting are visualized as a bar graph and respectively indicate the contribution of each standard molecules to the positive or negative peaks visible in the PC1 and PC2 loadings. Cer: ceramide; Chol: cholesterol; SM: sphingomyelin; GM1: monosialotetrahexosylganglioside; PCh: phosphatidylcholine; PE: phosphatidylethanolamine; PA: phosphatidic acid.

Figure 4

Linear discriminant analysis. The first 25 PC loadings calculated by means of PCA were used for the LDA. The scatter plot shows the LDA scores obtained for EVs from BM-MSCs (pink), ASCs (blue), and DFs (green). Each square represents a single spectrum. The crosses indicate the mean canonical observation score obtained for each group.