WO2016097996A1 - Use of fourier transform infrared spectroscopy analysis of extracellular vesicles isolated from body fluids for diagnosing, prognosing and monitoring pathophysiological states and method therfor - Google Patents

Abstract

The method analyzes biological fluids collected from patients, and uses attenuated total reflecting (ATR) sensors and spectral analysis by Fourier transform infrared spectroscopy (FTIR) to form a spectral fingerprint of extracellular vesicles (EVs) isolated from said biological fluids.

Description

USE OF FOURIER TRANSFORM INFRARED SPECTROSCOPY ANALYSIS OF EXTRACELLULAR VESICLES ISOLATED FROM BODY FLUIDS FOR DIAGNOSING, PROGNOSING AND MONITORING PATHOPHYSIOLOGICAL STATES AND METHOD THERFOR

Corresponding applications

The present application claims priority to earlier European Patent Application N° 14198358.5 filed on December 16, 2014 in the name of Ecole Polytechnique Federate de Lausanne (EPFL), the content of this earlier application being incorporated in its entirety by reference in the present application.

Field of the invention The present invention concerns the field of analysis of extracellular vesicles (EV) as a help for disease diagnosis such as cancer. More particularly, the invention relates to the use of infrared spectroscopy to analyze body fluids derived EVs for identifying pathophysiological states of the donor. Background

Early detection of cancer has proven to drastically decrease mortality rates. To confirm the presence of the disease, current diagnostic procedures includes numerous steps such as tissue biopsies, blood/urine tests, imaging and genetic analysis. None of these techniques are self-sufficient and can suffer from invasiveness (patient suffering), low sensitivity (late diagnosis) and low specificity (false-positive diagnosis).

A striking example is breast cancer, where an early diagnosis is critical and consequently requires a systematic screening of the population via mammography, with final confirmation through tissue biopsies (see reference 21 ).

In that context, molecular diagnostics has emerged as a promising tool to assess the pathophysiological conditions of the patient by primarily targeting both genetic materials and proteins (see reference 22). Alterations in protein expression or function together with nucleic acid mutations or copy number changes are the principle molecular indicators of a disease. Identifying these biomarkers in cancer patients can additionally provide precious information on both the prognosis of the disease and the evolution of a given therapy. However, their identification is a challenging task with highly uncertain outcomes, which limits the development of reliable diagnostic tests.

Moreover, a unique biomarker will not likely be sufficient to draw conclusions as cancer is a highly heterogeneous disease with multiple stages and subtypes, reflected by multiple molecular abnormalities. Hence, an appealing alternative would be an overall recognition response describing both the genome and proteome of the tumor serving as a fingerprint of the disease.

To address these issues, extracellular vesicles (EVs) appear as highly promising diagnostic candidates since they are secreted by cancer cells, already at the very early stage of the disease, into their microenvironment and ultimately into biological fluids such as blood, saliva and urine. These nanometer-sized containers are cell-derived vesicles enclosed by a lipid bilayer, which transport the molecular identity of their mother cells by inheriting a functionally active cargo composed of lipids, proteins and nucleic acids. EVs play a central role in the intercellular communication {see reference 30) as well as in the propagation of diverse pathologies, in primis cancer (see reference 31 ), inflammatory (see reference 32) and neurodegenerative (see reference 33) diseases, by activating signaling cascades in or delivering bioactive molecules such as lipids, proteins, or RNAs to the target cells (see reference 34). Numerous studies have shown that most cancer cells produce elevated levels of EVs and highlighted their role in stimulating the tumor progression via the following processes (see reference 6): (i) inducing cell proliferation, (ii) stimulating angiogenesis, (iii) promoting the extracellular matrix remodeling, (iv) facilitating the formation of pre-metastatic niche and (v) promoting immune escape of tumors. Beside cancer cells, EVs secreted by other cells, such as endothelial cells, fibroblasts and leukocytes, under tumoral stress also influence and modulate cancer progression and thus contribute to the overall population of anomalous EVs in body fluids during the disease (see reference 44). The analysis of blood of cancer patients have revealed the presence of elevated levels of EVs, which number were found to increase with progressive stages of cancer development (see reference 45). Hence, cancer associated EVs can be regarded as early indicators of the presence and propagation of the disease as they receive from the tumor and its microenvironment specific molecules. These biomarkers are thus enriched within the EVs, which would otherwise constitute only a small fraction of the whole blood. In this framework, active research is undertaken to isolate EVs and to identify specific surface receptors and lumen embedded microRNAs (miRNAs) related to the type and stage of the cancer (see reference 35). The identification of such markers requires a combination of high-resolution mass spectrometry with bioinformatic tools which ends up with a list of tens to hundreds proteins overexpressed in cancer EVs. The candidates are further screened by immuno or enzymatic assays, which are time-consuming techniques and require large amount of EVs, to isolate one or few targets having high predictive value. However, despite considerable effort, the successes of this approach have been relatively scarce, since the potential biomarkers often fail to stand the test of clinical trial when they are confronted to a large cohort of samples.

On the other hand, an interesting optical method to analyze biological entities is the Fourier transform infrared spectroscopy (FTIR), which measures molecular vibrational spectrum of the particular sample irradiated by infrared light (see reference 23). Upon excitation, chemical bonds absorb energy and vibrate at specific frequencies. The FTIR spectrum displays these specific absorption bands and represents the molecular signature of the sample in terms of chemical composition and molecular structures.

The sensitivity of this spectroscopic analysis is enhanced when spectra are recorded in the attenuated total-reflecting (ATR) mode, where an internal reflection element of high refractive index concentrates the IR light of the probe beam to the first micrometers of the sample. Hence, the contribution of bulk water is dramatically reduced and the signal-to-noise ratio of the spectroscopic signal is increased.

FTIR and ATR-FT1R have enabled the non-invasive and label-free identification of the principal constituents of complex biological specimens, such as tissues and cells (see reference 24). In particular, spectral analysis has been used to delineate cellular or tissue properties on the basis of protein, lipid, nucleic acid and carbohydrate composition (see reference 25).

For example, ATR-FTIR has helped the histopathologicai characterization of diverse cancer tissues, such as prostate (see reference 26), brain (see reference 27), breast (see reference 28) or colon (see reference 29). However, these studies required access to biopsy samples, which implies the presence of a visible tumor mass, often associated with an advanced cancer. Therefore, according to the present invention FTIR spectra appears to offer a unique possibility to decipher the EV complexity, which enables the classification of samples of different origins according to distinct spectral characteristics. Indeed, a number of studies have evidenced the presence of specific molecular features on blood circulating EVs of cancer patients: (i) changes in the glycosylation patterns (see reference 46), (ii) increase concentration of specific lipids such as sphingomyelin or cholesterol (see reference 47). (iii) tumor-specific content of nucleic acids such as miRNA, mRNA and double stranded DNA (see reference 48) and (iv) increase expression of specific proteins such as cell adhesion proteins or growth factors (see reference 45). Taken together, these molecular alterations will consequently modify the infrared bands in terms of amplitude, position and shape.

Thus, the combination of FTIR analysis and EVs offers an innovative remote and painless biopsy of tumor mass undetectable with classical methods. This methodology appears highly suitable for the early diagnosis, prognosis and monitoring of cancers currently lacking identified biomarkers.

Description of the Invention

The methodology used by the present invention is based on three main components which are combined to achieve the desired result:

(i) extracellular vesicles named "EVs",

(ii) Fourier transform infrared spectroscopy named "FTIR",

(iii) comparative analysis such as computer-aided multivariate analysis.

• Extracellular vesicles (EVs)

EVs are complex biological entities, which carry the molecular identity of the cell (e.g. tumor cell) from which they are released.

EVs are nano- to micrometer-sized containers secreted by cells as membrane enclosed volumes comprising cellular components like proteins, lipids and nucleic acids. There are three main EV families:

(i) Exosomes are unilamellar vesicles of 30 to 100 nm diameter released from a cell into the extracellular environment by the fusion of intracellular, multivesicular bodies (MVBs) with the cell's plasma membrane. (ii) Microvesicles are directly derived by an exocytotic process from the cell's plasma membrane and vary in diameters from hundreds to thousands of nanometers.

(iit) Apoptotic bodies are large vesicles (>1 pm) released by apoptotic cells, exhibiting phosphatidylserine on the outer leaflet of their membrane (see reference 2).

EVs are naturally present in body fluids and carry chemical information for intercellular communication by activating signaling cascades in or delivering bioactive molecules such as lipids, proteins or RNA to the target cells (see references 1 and 2). There is increasing evidence that EVs play a central role in the development of diverse pathologies such as cancer (see reference 3), inflammatory (see reference 4) and neurodegenerative diseases (see reference 5), with high potential to be used as biomarkers (see reference 3) and therapeutic agents (see reference 6). The identification of a unique cancer EV biomarker is a challenging task with highly uncertain outcome, which limits the development of reliable diagnostic tests.

Hence, the use of FTIR analysis, or preferably ATR-FTIR, offers a unique chance to decipher the EV complexity. In addition, thanks to a multivariate analysis, it is possible to extract key features from infrared spectra to further classify these EVs and finally diagnose a disease.

Moreover, EVs are easily accessible in biological fluids and appear at an early stage of the disease, making them early indicators of the development of the disease. The combination of EVs and ATR-FTIR thus offers a unique, minimally invasive tool highly suitable for the diagnosis, prognosis and monitoring of diseases, which would benefit from early stage detection and currently lacking of identified biomarkers. In particular, cancer and neurodegenerative diseases are clearly targeted by our technology. · ATR-FTIR

In the method of the present invention, FTIR spectroscopy measures molecular vibrations in a sample irradiated by infrared light. Upon excitation, chemical bonds absorb energy and vibrate at specific frequencies.

The FTIR spectrum displays these specific absorptions bands and represents the molecular fingerprint of the sample in terms of molecular composition and structural features of the molecules. The sensitivity of this measurement is enhanced when recorded in ATR mode which is a preferred mode in the frame of the present invention; in this configuration, the sample contacts an internal reflection element of high refractive index, which concentrates the IR light to the first micrometers of the sample. Hence, sample consumption is minimized, the contribution of water is dramatically reduced and the signal-to-noise ratio is increased.

The use of FTIR or ATR-FTIR permits the non-invasive and label-free identification of the principal constituents of complex biological specimens (see reference 7), such as tissues, cells and in our case EVs. For instance, lipids, which compose the plasma membrane, have strong absorption bands around 2800 - 3000 cm~\ while proteins exhibit a highly conformation-specific bands in the 1600-1700 cm^"1 region. The low- wavenumber part of the ATR-FTIR spectrum (900-1200 cm ¹) is representative of nucleic acids and carbohydrates. Consequently, spectral analysis can be used to delineate cellular or tissue hierarchy on the basis of protein, lipid, nucleic acid and carbohydrate composition (see reference 8).

Interpretation of these complex data sets may be facilitated by multivariate analysis, which can be performed through machine learning. · Multivariate analysis

The aim of multivariate analysis in the present context is the extraction of most important features in the FTIR spectra of samples of different origin for further clustering and classifications according to distinct compositional characteristics. It requires algorithms that reduce the dimensionality of the spectral information into a few key components allowing interpretation with minimal loss of information (see references 9,10).

In the present invention, the data set is represented by FTIR spectra taken from EVs isolated from body fluids of both cancer patients and healthy donors, whose medical related characteristics are known (e.g. age, sex, type, stage and prognosis of the disease).

The multivariate analysis identifies the similarities and differences between classes (categories within the data set, in our case type and stage of the cancer). Among the existing algorithms, we use the combination of principal component analysis (PCA, see reference 11) and linear discriminant analysis (LDA). PCA transforms the original set of variables (in the case of the invention FTIR spectra in form of absorption intensities versus wavenumbers) to a few so-called principal components (PCs), which are linear combinations of the original variables. These PCs are the main spatial directions along which the data set exhibits the largest variations and are used for the subsequent LDA, which maximizes the variance between the classes, while minimizing the intraclass variability. This initial phase allows the production of a model for the classification of the EVs donors and is used to detect and diagnose cancer disease from unknown donors.

Accordingly, the invention relates to a generic method to detect pathophysiological changes and abnormalities through extracellular vesicle (EV) immobilization on attenuated total reflecting (ATR) sensor surfaces and spectral analysis by Fourier transform infrared spectroscopy (FTIR).

The unique spectral fingerprint of the EVs, isolated from biological fluids (e.g., blood, urine, saliva, breast milk, lymph, amniotic fluid), enables a remote, non-invasive diagnosis and prognosis of cancer and ultimately any other proliferative diseases.

Combined with multivariate analysis, ATR-FTIR spectroscopy/imaging has shown great potential in cancer diagnosis. Present clinical practice analyzes samples comprising tissue sections (see reference 7), or cytology specimens from biopsies or biological fluids such as whole blood, urine or saliva (see reference 18). There are two publications concerning the use of FTIR spectroscopy for cancer diagnosis, based on whole blood (see reference 19) and sputum analysis (see reference 20). However, such samples are very complex fluids containing a bewildering number of components, leading to high sample variances and thus to questionable reliability and specificity. The present invention answers these issues as EV samples are highly purified, well characterized and carry the molecular entities of the diseased cells.

In one embodiment, the invention concerns the use of the Fourier transform infrared spectroscopy (FTIR) to analyze purified extracellular vesicles (EVs) collected from a sample of a fluid.

In an embodiment, a multivariate analysis is applied to the FTIR spectrum. in an embodiment, the spectra obtained by FTIR are recorded in an attenuated total- reflecting (ATR) mode to enhance the sensitivity of the spectroscopic analysis. in an embodiment of the use, the fluid to be analyzed is a biological fluid comprising blood, urine, saliva, breast milk, lymph, amniotic fluid, bile, blood serum, cerebrospinal fluid, semen, sweat, tears, mucus, vaginal secretion or another biological fluid.

In an embodiment, the invention concerns a method of analyzing biological fluids, wherein said method uses a spectral analysis of extracellular vesicles (EVs) isolated from said biological fluids by Fourier transform infrared spectroscopy (FTIR) to form a spectral fingerprint of said extracellular vesicles.

In an embodiment, the method comprises at least the following steps:

(A) collecting a biological fluid sample from a subject;

(B) purifying the extracellular vesicles from the collected sample of biological fluid; (C) immobilizing the purified extracellular vesicles;

(D) acquiring an FTIR spectrum of the extracellular vesicles thus forming said spectral fingerprint.

In an embodiment of the method, purified extracellular vesicles are preferably immobilized on an attenuated total-reflecting (ATR) crystal.

In an embodiment of the method, the spectra are acquired in the attenuated total- reflecting mode (ATR) to enhance sensitivity. In an embodiment, the method further comprises an analysis of the acquired spectrum for clustering and classifying features of the acquired spectrum. in an embodiment, the analysis comprises the comparison of the acquired spectrum with a reference spectrum.

In an embodiment, the analysis comprises a multivariable analysis or another equivalent analysis.

In an embodiment, the multivariable analysis comprises the comparison of the acquired spectrum with spectra taken from EVs isolated from body fluids of cancer patients and healthy donors whose medicai-reiated characteristics such as age, sex, type, stage and prognosis of the disease are known.

In an embodiment of the method, the biological fluid is blood, urine, saliva, breast milk, lymph, amniotic fluid, bile, blood serum, cerebrospinal fluid, semen, sweat, tears, mucus, vaginal secretion or another biological fluid.

The use and method as defined herein may be used for the analysis of the state of a subject (for example a mammal) from which the biological fluid has been taken of.

In an embodiment, the invention concerns a use or the method that serve for an early detection of a disease in patients or subjects. in an embodiment, the disease comprises cancer. Other possible diseases comprise proliferative diseases such as neurodegenerative diseases.

Detailed description of the invention

The invention will be better understood from the following detailed description and from the drawings which show

Figures 1A to 1E are schematicai iliustrations of the steps of a method according to the present invention; Figures 2A to 2C illustrate embodiments of purification and analysis steps according to the present invention of an example;

Figures 3A and 3B illustrate comparisons of ATR-FTIR spectra in the 1800-900 cm^-1 range of examples of cells and EVs;

Figures 4A and 48 illustrate an example of a multivariate analysis of spectra derived from breast cancer cell lines;

Figures 5A to 5D illustrate an example of characterization and analysis of EVs derived from human piasma.

The invention provides a method for analyzing EVs based on FTIR spectroscopy as illustrated in Figures 1A to 1E. More precisely, the first phase of the method according to the invention consists in a collection and purification of the EVs from body fluids see figures 1 A and 1 B. Examples of methods to carry out these steps are given below in the discussion of the current results. Then, in a second phase, the purified EVs are immobilized on an ATR crystal for the acquisition of a FTIR spectrum in this case in the ATR-FTIR mode. This step is illustrated in Figure 1C.

For example, as an exemplary embodiment, EVs which have been isolated and purified from minute quantities of body fluids from patients, are immobilized on a high refractive index crystal (e.g. zinc selenide, germanium, silicon), enabling a measurement in ATR mode.

To this effect, the sample is illuminated by a beam of infrared light (figure 1C) and the recorded absorption spectrum (Figure 1D) is subsequently analyzed by a diagnostic computer algorithm (Figure 1 E).

The software is preferably previously trained by correlating spectral data to pathophysiological features of the disease with the help of EVs collected from patients with well-defined type and stage of the disease to provide a comparison basis. This allows the construction of reference data that may be used later for comparison purposes and determination of the status of patients being examined with a protocol including the method according to the present invention. The last phase of the process (Figure 1 E) analyses the vibrational bands and either classes or attributes the corresponding sample with the help of multivariate analysis according to the principles discussed in the present application.

Current results of methods using the present invention are discussed herebelow.

Example 1

The use and method according to the present invention and described in the present application has been validated with EVs derived from three cultured cell lines of breast cancer, namely MCF-7, BT-474 and MDA-MB-231 ; each of these ceil lines represents a different class of breast carcinoma, Luminal A, Luminal Band Claudin-low, respectively (see reference 12) as described hereunder.

Materials and methods of example 1

Materials: Dulbecco's phosphate-buffered saline (D-PBS, Life Technologies), DMEM/F-12+ GlutaMAX (Life Technologies), Newborn Calf Serum (NBCS, Life Technologies). Cell culture: MCF-7, BT-474 and MDA-MB-231 human breast cancer cells were grown in DMEM/F-12 + GlutaMAX medium supplemented with 10% NBCS (unless otherwise stated) in a humidified 5% CO2 atmosphere at 37°C.

Isolation of EVs methodology: EVs from cultured breast cancer cell lines were isolated using our previously published protocol (see reference 36) with slight modifications. The conditioned medium (CM) of ~40*10^δ cultured cells, maintained in serum-free medium during the last 48 h, was first centrifuged at 300 g for 4 min and then filtered through a 0.45 pm pore-sized filter. EVs were concentrated by ultrafiltration (UF) using a 100 kDa molecular weight cut-off (MWCO) Amicon Ultra-15 centrifugal filter unit (Millipore), resuspended in D-PBS and concentrated again by UF to a volume of approximately 250 μΙ. EVs were further purified by size exclusion chromatography

(SEC) using a Sephacryl 500 10/40 GL column (GE Healthcare) equilibrated in D-PBS with an AKTA-Purifier system (GE Healthcare). EV-containing fractions were identified by absorption at λ = 280 nm and concentrated to the desired volume with a 100 kDa MWCO Amicon Ultra-4 centrifugal filter unit (Millipore). Connection of a RF-20A Prominence spectrofluorimetric detector (Shimadzu) allowed monitoring of fluorescence during elution (Aex = 485 nm, Aem = 515 nm). Cryo-EM: EVs were processed for visualization by cryo-EM as previously described in reference 36. The samples were prepared in a controlled environment using a vitrification system Vitrobot (FEI, Mark IV). The chamber temperature was held at 22- 23°C, and the relative humidity was kept close to saturation (100%) to prevent water evaporation from the sample. A 5 pi sample drop was placed on a Lacey carbon film grid (Agar Scientific) and blotted with filter paper to obtain a thin liquid film. The grid was then rapidly plunged into iiquid ethane at -180*C. The grids containing the vitrified specimens were stored in Iiquid nitrogen and transferred into a Tecnai F20 microscope using a Gatan 626 cryo-holder and its workstation. The acceleration voltage was 200 kV and the working temperature was kept below - 170°C. The images were recorded digitally with a Eagle camera 4k x 4k (FEI) under low-dose conditions with an underfocus of approximately -2 pm and 25'000x-50'000x of magnification.

ATR-FTIR spectroscopy: 20 μΙ of EV samples were spread and dried on a ZnSe ATR crystal. The spectra were recorded with an FTIR spectrometer equipped with an MCT detector (IFS 28, Bruker Optics). For both background and EV samples, 500 scans were co-added at a resolution of 4 cnrv¹ between 4000 cnr¹ and 800 cm'¹. Spectra were background corrected and atmospheric water vapor bands were subtracted using OPUS 4.2 software (Bruker Optics). Prior to multivariate analysis, the ATR-FTIR spectra were cut between 1800-900 cnr¹, rubberband baseline corrected and vector normalized with an OPUS macro (Bruker Optics).

Multivariate analysis: The aim of multivariate analysis in the present context is the automated extraction of similarities and differences in FTIR spectra of EVs of different origin for further clustering and classifications according to distinct compositional characteristics. An FTIR or ATR-FTIR spectrum comprises n-wavenumbers with different absorbance and can be considered as a n-dimensional object. The analysis of all EV preparations and their FTIR spectrum was thus simplified through dimensionality reduction with minimal loss of information using open-source multivariate analysis package available on R software (R Development Core Team, 2010).

First, Principal Component Analysis (PCA) algorithm transforms the original set of variables to a few principal components (PCs), which are linear combinations of the original variables. These PCs are the main spatial directions along which the data set exhibits the largest variations.

Then, Linear Discriminant Analysis (LDA) was performed on these PCs to maximize the variance between the classes, represented by EV origins, while minimizing the intraclass variability. This initial phase produces a reference model for the classification of the EVs originating from the three cell lines and is further used to identify EVs of unknown origin.

Results and Discussion of Example 1 EVs secreted from MCF-7, BT-474 and MDA-MB-231 breast cancer cell lines were purified from conditioned medium and enriched by a combination of ultrafiltration and size-exclusion chromatography as depicted on Figure 2A. Specifically, the conditioned medium was first centrifuged and filtered with a 0.22 pm pore size filter in order to remove cell debris and big particles. Large-size proteins were discarded through UF while SEC further purified EVs from residual soluble proteins and organic molecules.

Cryo-electron microscopy demonstrated the preservation of EV morphological integrity as well as the absence of soluble proteins and nucleic acids (see Figure 2B which illustrates a cryo-electron micrograph of EVs secreted from MCF-7 cells, scale bar 50 nm). These EV preparations cleared from small soluble contaminants were spread and dried on the ATR crystal and FTIR spectra were recorded.

An example of a typical FTIR spectrum obtained for EVs derived from MCF-7 is illustrated on Figure 2C. Ail the collected spectra was composed of three distinct regions as expected from the principal EV constituents:

(1 ) the principal absorbance bands in the 3100-2900 cm^"1 area were attributed to the stretching vibrations of Chb and CH2 groups of the lipids,

(2) the weak absorbance between 1700-1500 cm"¹ was assigned to the amide I and amide II from the peptide bonds of the proteins

while

(3) the strong bands at 1200-800 cm^"1 referred to the stretching vibrations of phosphate groups from nucleic acids and phospholipids and finally to the stretching vibrations of ester bonds of polysaccharides. The FTIR spectrum of pure phospholipids confirmed the prominent role of the lipids in the absorbance bands in region 1 while minimizing the influence of the lipid phosphate group in region 3, as shown elsewhere (see reference 37).

Interestingly, FTIR spectra of nucleic acids extracted and purified from EVs displayed strong absorbance bands in region 3 with negligible bands in region 1 , demonstrating that EVs carry in their lumen non-negligible amount of nucleic acids able to largely contribute to the FTIR spectrum. Concomitantly, the presence of glycosylated proteins on the surface of EVs also participates to the absorption bands in region 3 (see reference 46). Together, these regions provide a biochemical fingerprint of EVs based on chemical bond vibrations, tightly reflecting their molecular composition and structure. Differences in the absolute or relative intensity and position of peaks can thus be attributed to biochemical changes. A glaring example is the variation between the FTIR spectrum of EVs from the spectrum of their mother cells which is illustrated as an example in Figure 3A. Specifically, this figure 3A shows EVs secreted by MCF- 7 cells (solid line) vs MCF-7 cells (dashed line). Although most of the molecular components are present in both specimen and observable in their respective spectra, their number and proportion are different leading to distinct absorption responses. For instance, EV spectra exhibit weaker bands in the amide l/ll region than the cell spectra, which is partly explained by the lower overall density of proteins in EVs compared to the cells. Conversely, specific absorption by nucleic acid and/or carbohydrates (1200- 800 cm"¹) is leading in EV spectra, which is consistent considering their enrichment in nucleic acids and more specifically miRNAs (see reference 38), together with the presence of glycoproteins (see references 39,40). More surprising, the comparison between FTIR spectra of EVs derived from different breast cancer lines afso comprise spectra! dissimilarities as illustrated in Figure 3B showing a mean spectra of EVs derived from BT-474 (n=12), MCF-7 (n=18) and MDA- MB-231 (n=11 ), the shaded areas corresponding to the standard deviation of the mean, which implies distinct compositional or structural features that may reflect the cellular origin and, by extension to cancer cells, the stage of the tumor. While the lipid region is constant in all EV spectra, major spectral divergences are located in the regions 2 and 3. Taking together, these results suggest that the specific pattern in the FTIR spectrum of EVs is tightly associated with nucleic acids and proteins composition and post-translational modifications, providing a specific bioiogical signature of the cancer type and stage.

Interpretation of these complex data sets is facilitated by multivariate analysis. The aim of this analysis is the extraction of most important features in FTIR spectra of EVs for further clustering and classifications according to their distinct compositional characteristics. This analysis requires algorithms that reduce the dimensionality of the spectrum into few key parameters allowing interpretation with minimal loss of information (see references 41 ,42). in the particular case, the data set is represented by FTIR spectra taken from a large number of EV preparations and the analysis identifies the similarities and differences between classes (categories within the data set, in our case the breast cancer cell lines). Among the existing algorithms, one may use the combination of principal components analysis (see reference 43) (PCA) and linear discriminant analysis (LDA). PCA transforms the original set of variables (in the present case FTIR spectra in form of absorption intensities versus wavenumbers) to a few so-called principal components (PCs), which are linear combinations of the original variables. These PCs are the main spatial directions along which the data set exhibits the largest variations and are used for the subsequent LDA, which maximizes the variance between the classes, while minimizing the intraclass variability. This procedure enabled to cluster and classify the three cultured cell lines as in Figures 4A and 4B which illustrates a multivariate analysis of ATR-FTIR spectra of EV derived from the three breast cancer lines mentioned above. Specificalfy, Figure 4A illustrates a 2D scatter plot of PCA-LDA analysis where each dot represents the result of a single EV purification. The ellipses show the 95% confident region. The model was verified through leave-one-out cross-validation, which results are presented in a confusion matrix represented in Figure 4B. With the current data set, this procedure identifies unknown samples with accuracy up to 91% and can be periodically refined and retrained as new EV preparations become available. Example 2

In another example and application of the present invention, the clinical applicability and usability of the methodology according to the present invention was assessed by analyzing extracellular vesicles (EVs) derived from plasma samples obtained from healthy donors.

Materials and methods of example 2

Isolation of EVs methodology: For EV isolation from clinical samples, 500 μΙ_ of plasma was centrifuged at 300 g for 4 min, filtered through a 0.22-μηι pore-sized filter, and then purified by SEC using a Sephacryl 500 10/40 GL column (GE Healthcare) equilibrated in NaCI 0.9% with an AKTA-Purifier system (GE Healthcare).

Cryo-EM: EVs were processed for visualization by cryo-EM as previously described in Examples 1.

Dynamic light scattering (DLS): Experiments were carried out on a Zetasi2er Nano ZS (Malvern) equipped with a He-Ne laser of 633 nm (4 mW), and the light scattering from the samples was measured at an angle of 175° to the incident laser beam. A typical experiment comprised a sequence of 12 * 10 s recordings which were repeated three times.

ATR-FTIR spectroscopy: 1 ml of EV samples were spread and dried on a Germanium ATR crystal. The spectra were recorded with an FTIR spectrometer equipped with an MCT detector (IFS 28, Bruker Optics). For both background and EV samples, 500 scans were co-added at a resolution of 4 cm ¹ between 4000 cnr¹ and 800 cm ¹. Spectra were background corrected and atmospheric water vapor bands were subtracted using OPUS 4.2 software (Bruker Optics).

Results and Discussion of Example 2

As an example, 500 pL of plasma was first centrifuged and filtered to remove cells, cell debris, and large particles and then injected into a size-exclusion chromatography (SEC) column. A typical elution profile is depicted in Figure 5A: a SEC elution profile of a centrifuged and filtered human plasma sample. The absorbance was measured at 280 nm, reflecting the amount of protein content. The EV fraction appeared between 12 and 15 ml elution volume. This figure shows a small but well-resolved peak around 12.5 ml elution volume originating from the optical absorption at 280 nm of EV proteins. Soluble, non-EV proteins are eluted at larger elution volumes as broader peaks.

Compared to classical procedures using differential ultracentrifugation, this methodology offers several advantages for the isolation of EVs from biofluids, especially from blood derived samples:

(i) the purified EVs retain their morphological integrity without aggregating as demonstrated by cryo-electron microscopy (cryo-EM, Figure 5B illustrates Four cryo- EM micrographs of purified EVs, with Scale bar, 50 nm) and dynamic light scattering (DLS, Figure 5C: Size distribution of purified plasma EVs obtained by DLS) with a size distribution ranging from 30 to 200 nm,

(ii) they are separated from lipoproteins and protein aggregates, and moreover, (iii) the overall procedure is performed within less than 30 min.

Thus, from 500 pL of plasma, we can retrieve 3 mL of highly purified EVs at a concentration of ~10¹⁰ particles per mL. About 1 mi of this EV preparation, cleared from small soluble contaminants, was spread and dried on the ATR crystal and FTIR spectra were recorded.

An example of a typical FTIR spectrum is depicted on Figure 5D which illustrates a typical infrared spectra of the purified plasma EVs. Al! the collected spectra were composed of three distinct regions as expected from the principal EV constituents:

(i) the principal absorbance bands in the 3000-2800 crrr¹ area were attributed to the stretching vibrations of CH3 and CH2 groups of the lipids,

(it) the absorbance between 1700-1500 cm^*1 was assigned to the amide I and II vibrations from the peptide bonds of the proteins while

(Hi) the bands at 1200-1000 cnrr¹ referred to the stretching vibrations of phosphate groups from nucleic acids and phospholipids and finally to the stretching vibrations of ester bonds of polysaccharides.

Together, these regions provide a biochemical fingerprint of EVs based on chemical bond vibrations, tightly reflecting their molecular composition and structure. Differences in the absolute or relative intensity and position of peaks can thus be attributed to biochemical changes.

Applications

Presently, cancer diagnosis includes numerous steps to confirm the presence of the disease such as tissue biopsies, blood/urine tests, imaging and genetic analysis. None of these techniques are self-sufficient and can suffer from invasiveness (patient suffering), sensitivity (late diagnosis) and specificity (faise-positive diagnosis), ideally, a cancer diagnostic test should reliably detect the appearance of the disease at a very early stage and provide continuous non-invasive monitoring of its progress. The method of the present invention may accordingly be used in diagnosing, prognosing and monitoring normal and diseased state in a subject.

The methodology according to the present invention needs only minute quantities of body fluids as a source of EVs and thus can be considered as minimally invasive. The specificity and sensitivity is brought by the utmost detection limit of ATR-FTIR spectroscopy combined with the early appearance of cancer EVs during the disease progression.

Typically, the biological fluids used in the present method may comprise blood, urine, saliva, breast milk, lymph, amniotic fluid, bile, blood serum, cerebrospinal fluid, semen, sweat, tears, mucus, vaginal secretion.

ATR-FTIR spectroscopy enables thus to decipher EV complexity, to spotlight the principal compositional characteristics specific to each EVs and thus predict their cellular origin with utmost reliability. Since EVs are easily accessible in body fluids such as blood, urine or spinal fluids, the approach would offer a unique, minimally invasive tool highly suitable for the diagnosis, prognosis and continuous monitoring of cancers, which would benefit from early stage detection and currently lacking of identified biomarkers.

The described technology provides easy, fast, minimally invasive and continuous access to the state of tumors and therefore offers a novel strategy in cancer diagnosis and personalized medicine without injuries of the patients. Hence, the presented approach is highly beneficial for the diagnosis, prognosis and progression monitoring of the disease. Moreover, this novel strategy would be helpful in the orientation of the aftercare of cured patients in detecting cancer recurrence. The reference data set would be built with FTIR spectra taken from EVs isolated from body fluids of both cancer patients and healthy donors, whose medical-related characteristics are known (e.g. age, sex, type, stage and prognosis of the disease), and periodically refined and retrained as new EV preparations become available.

The present method can be integrated to the already established systematic diagnostic screening campaigns, such as ones performed for breast, and to the opportunistic ones, such as for prostate and colon cancer. Moreover, it can be implemented as a point-of-care testing product to support clinical decision-making in doctors' offices, medical laboratory and other primary and specialty care settings.

The embodiments of the invention described herein are of course illustrative examples and should not be considered in a limiting way. Variants are possible using equivalent means and methods within the spirit and scope of the present invention. The embodiments described herein may also be combined as desired.

For example, the spectral analysis may comprise the comparison of the acquired spectrum with a reference spectrum.

The analysis may be a multivariate analysis or another equivalent analysis.

The analysis may the comparison of the acquired spectrum with spectra taken from EVs isolated from body fluids of cancer patients and healthy donors whose medical- related characteristics such as age, sex, type, stage and prognosis of the disease are known.

References

1. Raposo, G. & Stoorvogel, W. Extracellular vesicles: exosomes, microvesic!es, and friends. J Cell Bio 200, 373-383 (2013).

2. Mittelbrunn, M. & Sanchez-Madrid, F. Intercellular communication: diverse structures for exchange of genetic Information. Nat Rev Mol Cell Bio 13, 328-335

(2012).

3. Rak, J. Extracellular vesicles - biomarkers and effectors of the cellular Interactome In cancer. Front Pharmacol 4, 21 (2013).

4. Buzas, E. I,, Gydrgy, B., Nagy, G., Falus, A. & Gay, S. Emerging role of extracellular vesicles In Inflammatory diseases. Nat Rev Rheumatol 10, 356-364 (2014).

5. Be!lingham, S. A., Guo, B. B., Coleman, B. M. & Hill, A. F. Exosomes: vehicles for the transfer of toxic proteins associated with neurodegenerative diseases? Front Physiol 3, 124 (2012).

6. EI-Andaioussl_: S., Mager, I., Breakefield, X. O. & Wood, M. J. A. Extracellular vesicles: biology and emerging therapeutic opportunities. Nat Rev Drug Discov 12,

347-357 (2013).

7. Baker, M. J. et al. Using Fourier transform IR spectroscopy to analyze biological materials, Nature Protocols 9, 1771-1791 (2014).

8. Martin, F. L. et al. Distinguishing cell types or populations based on the computational analysis of their Infrared spectra. Nature Protocols S, 1748-1760 (2010).

9. Trevisan, J., Angelov, P. P., Carmichael, P. 1.., Scott, A. D. & Martin, F. L. Extracting biological Information with computational analysis of Fourier-transform Infrared (FTIR) biospectroscopy datasets: current practices to future perspectives. Analyst 137. 3202 (2012).

10. Goodacre, R. Explanatory analysis of spectroscopic data using machine learning of simple, Interpretable rules. Vibrational Spectroscopy 32, 33-45 (2003).

11. Ringner, M. What Is principal component analysis? Nat Biotechnol 26, 303-304 (2008).

12. Holliday, D. L. & Speirs, V. Choosing the right cell line for breast cancer research. Breast Cancer Res. 13,215 (2011 ).

13. Wyss, R. et al. Molecular and dimensional profiling of highly purified extracellular vesicles by fluorescence fluctuation spectroscopy. Anal Chem 86, 7229-7233 (2014).

14. Noninvasive Cancer Diagnostics and Technologies Market - Global Share, Industry Overview, Analysis, Growth, Trends Opportunities and Forecast 2012 - 2020. Available:

Accessed 20l4 October 27

15. Breast Cancer Diagnostic and Drug Technologies: Global Markets. Available at: http://www.bccresearch.com/market-research/healthcare/breast-cancer- dlagnosticdrug-technologtes-hlc106a.html. Accessed 2014 October 27.

16. Office federal de la statistique. Available at:

http.7/www.bfs.admin.ch/bfs/portal/fr/index/themen/14/02/05/key/01701.html.

Accessed 2014 October 27.

17. Interpharma. Sante publique en Suisse. 1-89 (2014).

18. Ollesch, J. et ai. FTIR spectroscopy of biofluids revisited: an automated approach to spectral biomarker identification. Analyst 138, 4092 (2013).

19. Baker, M. J., Abel, P. & Lea, R. W. Diagnosing and/or prognosing a proliferative disorder in a subject, involves performing spectroscopic analysis upon a blood sample of the subject to produce a spectroscopic signature characteristic of the blood sample. WO2014076480 20. Lewis, P. D. & Mur, L A. J. Diagnosing lung cancer involves performing Fourier transform infrared spectroscopy analysis of sputum sample, and comparing spectrum with that of normal control, where difference between them at specific wavenumber(s) indicates lung cancer. International Publication No. WO 2012/001370 A!

21. Veronesi, U., Boyle, P., Goldhirsch, A., Orecchia, R. & Viale, G. Breast cancer.

Lancet 365, 1727-1741 (2005).

22. Gonzalez de Castro, D., Clarke, P. A., Al-Lazikani, B. & Workman, P. Personalized Cancer Medicine: Molecular Diagnostics, Predictive biomarkers, and Drug Resistance.

Clin Pharmacol Ther 93, 252-259 (2012).

23. Baker, M. J. et al. Using Fourier transform IR spectroscopy to analyze biological materials. Nature Protocols 9, 1771-1791 (2014).

24. Diem, M., Romeo, M., Boydston-White, S., Miljkovic, M. & Matthaus, C. A decade of vibrational micro-spectroscopy of human cells and tissue (1994-2004). Analyst 129,

880-885 (2004).

25. Movasaghi, Z., Rehman, S. & ur Rehman, D. I. Fourier Transform Infrared (FTIR) Spectroscopy of Biological Tissues. Applied Spectroscopy Reviews 43, 134-179 (2008).

26. Fernandez, D. C, Bhargava, R., Hewitt, S. M. & Levin, I. W. Infrared spectroscopic imaging for histopathologic recognition. Nat Biotechnoi 23, 469-474 (2005).

27. Krafft, C, Thummler, K., Sobottka, S. B., Schackert, G. & Salzer, R. Classification of malignant gliomas by infrared spectroscopy and linear discriminant analysis.

Biopolymers 82, 301-305 (2006).

28. Walsh, M. J., Holton, S. E., Kajdacsy-Balla, A. & Bhargava, R. Vibrational Spectroscopy .Vibrational Spectroscopy 60, 23-28 (2012).

29. Kailenbach-Thieitges, A. et al. Immunohistochemistry, histopathology and infrared spectral histopathology of colon cancer tissue sections. J. Biophoton. 6, 88-100 (2012).

30. Colombo, M., Raposo, G. & Thery, C. Biogenesis, Secretion, and Intercellular Interactions of Exosomes and Other Extracellular Vesicles. Annu Rev Cell Dev Biol 30, 255-289 (2014).

31. Rak, J. Extracellular vesicles - biomarkers and effectors of the cellular interactome in cancer. Front Pharmacol 4, 21 (2013).

32. Buzas, E. I., Gyorgy, B., Nagy, G., Falus, A. & Gay, S. Emerging role of extracellular vesicles in inflammatory diseases. Nat Rev Rheumatol 10, 356-364 (2014).

33. Bellingham, S. A., Guo, B. B., Coleman, B. M. & Hill, A. F. Exosomes: vehicles for the transfer of toxic proteins associated with neurodegenerative diseases? Front Physiol 3, 124 (2012).

34. El-Andalousst, S., Mager, I., Breakefield, X. O. & Wood, M. J, A. Extracellular vesicles: biology and emerging therapeutic opportunities. Nat Rev Drug Discov 12,

347-357 (2013).

35. Katsuda, T., Kosaka, N. & Ochiya, T. The roles of extracellular vesicles in cancer biology: Toward the development of novel cancer biomarkers. Proteomics 14, 412- 425 (2014). 36. Wyss, R. et al. Molecular and dimensional profiling of highly purified extracellular vesicles by fluorescence fluctuation spectroscopy. Anal Chem 86, 7229-7233 (2014).

37. Pohle, W. et al. FTIR-spectroscopic characterization of phosphocholine-headgroup model compounds. J Mol Struct 563, 463-467 (2001).

38. Squadrito, M. L. et al. Endogenous RNAs Modulate MicroRNA Sorting to Exosomes and Transfer to Acceptor Cells. CellReports 1-15 (2014).

39. Escrevente, C. et al. Sia!oglycoproteins and N-Glycans from Secreted Exosomes of Ovarian Carcinoma Cells. PLoS ONE 8, e78631 (2013).

40. Sanchez-Madrid, F. Post-translational modifications of exosomal proteins. 1-7 (2014). doi:10.3389/fimmu.2014.00383/abstract

41. Trevisan, J., Angelov, P. P., Carmichae!, P. L, Scott, A. D. & Martin, F. L. Extracting biological information with computational analysis of Fourier-transform infrared (FTIR) biospectroscopy datasets: current practices to future perspectives. Analyst 137, 3202 (2012).

42. Goodacre, R. Explanatory analysis of spectroscopic data using machine learning of simple, interpretab!e rules. Vibrational Spectroscopy 32, 33-45 (2003).

43. Ringner, M. What is principal component analysis? Nat Biotechnol 26, 303-304 (2008).

44. Ge, R.₍ Tan, E., Sharghi-Namini, S., & Asada, H. H. Exosomes in Cancer Microenvironment and Beyond: have we Overlooked these Extracellular Messengers?

Cancer Microenvironment : Official Journal of the international Cancer Microenvironment Society, 5, 323-332 (2012).

45. Milane, L, Singh, A., Mattheolabakis, G., Suresh, M., & Amiji, M. M. Exosome mediated communication within the tumor microenvironment. Journal of Controlled Release : Official Journal of the Controlled Release Society, 219, 278-294 (2015).

46. Pinho, S. S., & Reis, C. A. Glycosylation in cancer: mechanisms and clinical implications. Nature Publishing Group, 15, 540-555 (2015).

47. Llorente, A., Skotland, T., Sylvanne, T., Kauhanen, D., Rog, T., Ortowski, A., et al. Molecular Jipidomics of exosomes released by PC-3 prostate cancer cells. Biochimica Et Biophysica Acta, 1831 , 1302-1309 (2013).

48. Yanez-Mo, M., Siljander, P. R. M., Andreu, Z., Zavec, A. B., Borras, F. E., Buzas, E. I., et al. Biological properties of extracellular vesicles and their physiological functions. Journal of Extracellular Vesicles, 4, 27066 (2015).

Claims

1. Use of the Fourier transform infrared spectroscopy (FTIR) to analyze purified extracellular vesicles (EVs) collected from a sample of a fluid.

2. The use as defined in claim 1 , in which a multivariate analysis is applied to the FTIR spectrum.

3. The use as defined in one of the preceding claims, in which the spectra obtained by FTIR are recorded in an attenuated total-reflecting (ATR) mode to enhance the sensitivity of the spectroscopic analysis.

4. The use as defined in one of claims 1 to 3, wherein said fluid is a biological fluid comprising blood, urine, saliva, breast milk, lymph, amniotic fluid, bile, blood serum, cerebrospinal fluid, semen, sweat, tears, mucus, vaginal secretion.

5. A method of analyzing biological fluids, wherein said method uses a spectral analysis of extracellular vesicles (EVs) isolated from said biological fluids by Fourier transform infrared spectroscopy (FTIR) to form a spectral fingerprint of said extracellular vesicles.

6. The method as defined in claim 5, comprising at ieast the following steps:

(A) collecting a biological fluid sample from a subject;

(B) purifying the extracellular vesicles from the collected sample of biological fluid; (C) immobilizing the purified extracellular vesicles;

(D) acquiring an FTIR spectrum of the extracellular vesicles thus forming said spectral fingerprint.

7. The method as defined in claim 5 or 6 wherein the purified extracellular vesicles are immobilized on an attenuated total-reflecting (ATR) crystal.

8. The method as defined in one of claims 5 to 7, wherein the spectra are acquired in the attenuated total-reflecting mode (ATR) to enhance sensitivity.

9. The method as defined in one of the preceding claims 5 to 8, wherein it further comprises an analysis of the acquired spectrum for clustering and classifying features of the acquired spectrum.

10. The method as defined in claim 9, wherein the analysis comprises the comparison of the acquired spectrum with a reference spectrum.

11, The method as defined in claim 9, wherein the analysis is a multivariable analysis.

12. The method as defined in the preceding claim 5 to 11 , wherein the multivariable analysis comprises the comparison of the acquired spectrum with spectra taken from EVs isolated from body fluids of cancer patients and healthy donors whose medical- related characteristics such as age, sex, type, stage and prognosis of the disease are known.

13. The method as defined in one of the preceding claims 5 to 12, wherein said biological fluid is blood, urine, saliva, breast milk, lymph, amniotic fluid, bile, blood serum, cerebrospinal fluid, semen, sweat, tears, mucus, vaginal secretion.

14. The use as defined in claims 1 to 4 or the method as defined in one of claims 5 to 13 for use in an analysis of the state of a subject from which the biological fluid has been taken of.

15. The use as defined in one of claims 1 to 4 or the method as defined in one of claims 5 to 14, wherein said use or method serve for an early detection of a disease in patients.

16. The use or method as defined in the preceding claim, wherein said disease comprises cancer

17. The use or method as defined in claim 15, wherein said disease comprises proliferative diseases such as neurodegenerative diseases.