Detection of tumor-derived extracellular vesicles interactions with immune cells is dependent on EV-labelling methods

Cell-cell communication within the complex tumor microenvironment is critical to cancer progression. Tumor-derived extracellular vesicles (TD-EVs) are key players in this process. They can interact with immune cells and modulate their activity, either suppressing or activating the immune system. Understanding the interactions between TD-EVs and immune cells is essential for understanding immune modulation by cancer cells. Fluorescent labelling of TD-EVs is a method of choice to study such interaction. This work aims to determine the impact of EV labelling methods on the detection of EV interaction and capture by the different immune cell types within human Peripheral Blood Mononuclear Cells (PBMCs), analyzed by imaging flow cytometry and multicolor spectral flow cytometry. EVs released by the triple-negative breast carcinoma cell line MDA-MB-231 were labeled either with the lipophilic dye MemGlow-488 (MG-488), with Carboxyfluorescein diacetate, succinimidyl ester (CFDA-SE), or through expression of a MyrPalm-superFolder GFP (sfGFP) that incorporates into EVs during their biogenesis using a genetically engineered cell line. Our results showed that these different labeling strategies, although analyzed with the same techniques, led to diverging results. While MG-488-labelled EVs incorporate in all cell types, CFSE-labelled EVs are restricted to a minor subset of cells and sfGFP-labelled EVs are mainly detected in CD14+ monocytes which are the main uptakers of EVs and other particles, regardless of the labeling method. Moreover, MG-488-labeled liposomes behaved similarly to MG-488 EVs, highlighting the predominant role of the labelling strategy on the visualization and analysis of TD-EVs uptake by immune cell types. Consequently, the use of different EV labeling methods has to be considered as they can provide complementary information on various types of EV-cell interaction and EV fate.

Phenotypic changes triggered by EVs in the recipient cells require uptake of EVs. The mechanism of uptake comprises several sequential steps that are not always concurrent: i) interaction with the recipient cells, ii) entrance in the cells or cell uptake and iii) delivery of EV content to the recipient cell (9). Interaction with and entrance into recipient cells could occur either through specific molecular interactions or through unspecific mechanisms as macropinocytosis (10). EVs can enter into the recipient through endocytosis of the intact vesicle or alternatively, can fuse with the plasma membrane (9). The internalization of EVs has been shown on a wide range of cells such as dendritic cells (11), macrophages (12), dermal fibroblast (13), endothelial and myocardial cells (14). In some cases, cargo delivery has also been demonstrated, but this process could be inefficient depending on the type of recipient cells (15,16).
In order to study TD-EV interaction with immune cells many different EV labelling strategies have been developed. One of the most common method for the study of EV uptake consists in the labelling of EV membrane with fluorescent lipophilic membrane dyes (10, 17). Other EV components such as proteins can be targeted using permeable chemical compounds that enter the EV lumen or with the expression of fluorescentlytagged reporters (17). Here, we have compared different approaches of labelling, in order to evaluate the best approach to track EVs capture by recipient cells, monitored by flow cytometry and imaging flow cytometry. Tracking the fate of EVs into immune cells will provide significant insights into immune modulation by TD-EVs in tumor microenvironment and give clues to new therapeutic strategies.

The uptake of EVs by PBMCs is time-and temperature-dependent
We used tumor-derived EVs (TD-EVs) from the triple-negative breast carcinoma cell line MDA-MB-231 to test different labelling methods. TD-EVs were obtained from tumor cell conditioned medium (CCM) using size exclusion chromatography (SEC) and collected side-by-side with pools of intermediate and soluble protein fractions as previously described (18). Pooled EVs (fractions 7-11), intermediate (fractions [12][13][14][15][16] and soluble (fractions 17-21) SEC fractions were compared by Western Blot following opposed to the one of CFSE-EV+-PBMCs and sfGFP-EV+-PBMCS which remained very low at this temperature. As most energy-dependent uptake processes are inhibited at 4°C, non EV uptake-mediated MG-488 labeling of target cells might occur.
To exclude unspecific labelling of PBMCs by residual free dye present in the EV sample, MG-488 diluted in PBS was processed as MG-EV samples and the same volume was added to PBMCs. No fluorescence was detected in PBMCs incubated with MG-PBS at both temperatures ( Figure 1D) ruling out the contamination by free dye after the labelling procedure of EVs. To further investigate the temperature sensitivity of membrane-bound MG-488 dye incorporation, we labelled liposomes with MG-488.
Efficient capture by PBMCs was observed with MG-488-labelled liposomes both at 37 and 4 °C, demonstrating that temperature independence was mainly due to the use of MG-488. It suggests that the lipophilic dye MG-488 might diffuse between EVs or liposomes and cells after short/transient contacts.

Detection of the capture and fate of the EVs by PBMCs differs depending on the labelling technique.
Next, PBMCs were analyzed by Imaging Flow Cytometry following one-hour incubation with the different fluorescently-labeled particles to analyze their capture and distribution. Cells were defined as focused events that were also singlets, circular and live ( Figure S2A). Gating of EV+/beads+-PBMCs was done based on fluorescence ( Figure S2B). Consistently to our previous observation ( Figure 1C), a large percentage of PBMCs incorporated the MG-EV signal (15 to 100%). The CFSE-EV signal was detected in a smaller but significant percentage of EVs (5 to 60%). By contrast, the percentage of PBMCs that incorporated beads was low (around 5%), while hardly any cells with incorporated sfGFP-EVs were detected ( Figure S2B).
We took advantage of the imaging technology combined to flow cytometry to analyze the intracellular distribution of fluorescence in each cell. The texture parameter "Homogeneity" was used to differentiate between cells with dotted (low homogeneity) and diffused (high homogeneity) distributions, to distinguish the retention of fluorescence within intact EV from its release into cells respectively (Figure 2A,2B).
We verified that cells incubated with FBs that cannot deliver their content or be degraded were all of low homogeneity values. By contrast, the majority of the cells incubated with MG-and CFSE-EVs presented high homogeneity values, suggestive of a delivery of the EVs content, although dye transfer from EV to the cell membrane in the case of MG could also occur. For both types of labelled EVs, we also detected the presence of cells with low homogeneity values, likely corresponding to intact EVs at the surface of or inside the cells ( Figure 2C). By contrast, the small percentage of fluorescent cells detected upon incubation with sfGFP-EVs showed only low homogeneity values. We also noticed that cells presenting low homogeneity values were mainly cells with small area, characteristic of lymphocytes among PBMCs ( Figure   2D, left). However, when comparing the percentages of cells that had incorporated EVs/beads ( Figure 2D EVs. The membrane/cytosol ratio was close to 1 for CFSE+PBMCS indicating a uniform distribution of EV content in the recipient cell. By contrast, this ratio was greater than 1 in cells that incorporated the MG-EVs signal, indicating that this signal was mostly incorporated at the plasma membrane level.

CD14+ monocytes are the major cells capturing EVs and beads within PBMCs.
Heterogeneous fluorescence distribution between small and large area cells suggested some selectivity of EV cell uptake among PBMC types. To confirm this hypothesis and identify the major immune cells capturing EVs, we used a panel of specific antibody markers to identify major immune subsets within PBMCs using multicolor spectral flow cytometry. After gating single live cells ( Figure 1B To evaluate the subsets of PBMCs which preferentially captured labelled EVs or beads after 3 hours of incubation, we first gated the EV+-PBMCs or FB+-PBMCs and then identified the different immune cells subtypes ( Figure 4A). We compared the frequency of each immune subtype within the positive cells and within the total PBMCS ( Figure  4B, Figure S5). Although T cells were the most abundant immune cells among PBMCs ( Figure 4B, Figure S5), classical monocytes (CD14+ CD16-) were the most efficient to capture beads or EVs, independently of the labelling method used ( Figure 4A).
However, classical monocytes represented around 40% of the cells capturing CFSE-EVs and sfGFP-EVs but only 20% for MG-EVs, close to the percentage of classical monocytes found in the non-labelled PBMCs ( Figure 4B). Lymphoid cells, such as CD4+T, CD8+T, B cells and NKT cells were also labelled upon MG-EVs as well as MG-liposomes incubation ( Figure 4C). These similarities in immune subtype composition of MG+-PBMCs after incubation with MG-labelled EVs or liposomes, which significantly differ from the ones observed with CFSE/sfGFP EVs, further attests the prevalence of the labeling method in the fluorescence-based analysis of EV uptake.

DISCUSSION
In this work we have compared three different approaches to label EVs for their capacity to monitor TD-EVs uptake and intracellular fate in PBMCs by spectral flow cytometry and imaging flow cytometry: the lipophilic dye MG-488, the luminal-labelling dye CFSE and the genetically encoded MyrPalm-sfGFP. Regardless of the labelling method used, classical monocytes (CD14+CD16-) are the best uptakers among PBMCS.
Most importantly, our work shows that the detection of TD-EV-associated fluorescence in the recipient cells mainly depends on the labeling method used. Indeed, the three labeling strategies tested in this study are expected to result in different cellular staining patterns after uptake (10). For example, MG-488 is incorporated into the membrane of TD-EVs and, after cell uptake or possibly cell contact, it can be transferred to cell membranes. On the other hand, CFSE and the MyrPalm-sfGFP are inside the TD-EVs that have to be internalized into cells before being able to deliver their content. Importantly, by imaging flow cytometry, we were able to differentiate and quantify cells with likely intact EVs (Low Homogeneity) from those in which the EV content or the lipophilic dye has been transferred (High Homogeneity). Our results show that all labeling strategies allow the detection of intact EVs in a low percentage of EV+-PBMCs. EV content delivery or dye transfer (by membrane contact or fusion) predominates in cells incubated with CFSE-EVs or MG-EVs, respectively. We also observed a strong impact of the EV labelling method on the distribution of fluorescence among the different immune cell types, MG-488 being incorporated in all cell types, CFSE in a minor subset of cells (including lymphoid cells; NK and BC) and sfGFP mainly restricted to CD14+ cells.
The fluorescent properties of MemGlow makes this dye more useful for EV labeling than traditional lipid dyes (21,23). Compared to the other fluorescent dyes used in this study, MG-EV fluorescence is detected in the highest percentage of EV+-PBMCs.
However, MG-EV fluorescence is also detected in PBMCs when incubation is performed at 4°C and is not due to free dye contamination. Dye transfer may thus actually occur following simple interaction between the EV and the cell membrane, contrary to CSFE or sfGFP fluorescence. This could lead, at least in part, to a pattern of fluorescence due to redistribution of the dye by normal membrane recycling rather than EV uptake as previously proposed (10,17) Our experiments indicate that stable interactions/incorporation of EVs occur mainly in CD14+ cells in accordance with previous publications (8) and consistent with their phagocytic capacity and contribution to particle clearance. However, the uptake of sfGFP+EVs is detected is detected only in a small percentage of PBMCs. Since sfGFP signal is mainly present in cells with low homogeneity values, we hypothesize that we are only able to detect it during the first steps of uptake because after internalization the signal is lost due to dilution, quenching or degradation. Accordingly, our results strongly indicate that CFSE labeling of EVs appears to be the best labeling method to study EV uptake in vitro. CFSE allows the detection of intact EVs and their content delivery in a subset of immune cells previously described to incorporate EVs. Importantly, the strict temperature dependence of fluorescence accumulation in PBMCs treated with CSFE-EV clearly supports the requirement for an actual uptake step in these recipient cells. CFSE labelling has been used previously to label EVs (22,24) and does not seem to perturb the size of EVs nor their biodistribution (25). In a recent publication, however, it was described that CFDA-SE was not able to label EVs using protocols inspired by cell labeling (26). In our work we used longer incubation time, in addition to adding a step to remove the free dye after labeling using SEC, as previously described (22).
In conclusion, we have demonstrated in this work that the method used for EV labeling influences the detection of the different types of EV interactions with the recipient cell, including transient EV-PM interaction, EV content delivery and uptake of intact EVs.
All these interactions likely occur differently in the various immune cell types and could lead to different functional modifications relevant for communication in the tumor microenvironment.

Plasmids.
For the generation of the MyrPalm-sfGFP-encoding plasmid pTCP-MPsfGFP, a synthetic construct encoding sfGFP (accession: ASL68970) was fused in frame at its N-terminus with the acylation sequence of mouse LCK protein (aa 1-10) and at its Cterminus with a P2A-puromycin cassette. The construct was inserted into pTRIP-SFFV at the SrfI-KpnI sites. The SFFV promoter was replaced by the CMV promoter of pCDNA5 vector.

Extracellular Vesicle Isolation.
To obtain cell conditioned medium (CCM), 3x10 6 of MDA-MB-231 or MyrPalm-sfGFP MDA-MB-231 cells were plated per T150 flask with DMEM with 10% FCS. After 48hrs, the cells were washed with PBS and the medium was replaced with DMEM without FCS for 24 hrs. CCM was recovered after 24hrs and centrifuged at 300 g for 10 min at 4°C to pellet cells. Cells were counted and viability was measured. After 300 g centrifugation, supernatant was transferred to new tubes and centrifuged at 2,000 g for 20 min at 4°C to discard 2K pellet and then concentrated on a Centricon Plus-70 Centrifugal Filter (Millipore; MWCO 10kDa) by centrifugation at 2,000 g at 4°C until the volume was lower than 500ul. EVs were then isolated by Size Exclusion In order to wash the excess of dye from the solution, EVs were diluted with filtered PBS and concentrated using 10Kda filter (Millipore Amicon Ultra 0.5mL Ultracel 10K Centrifugal Filters) at 9000 g, until the volume was reduced to 50ul. NTA was performed after labeling to determine the concentration of particles and their percentage of fluorescence.

CFSE Staining.
EVs isolated by SEC were incubated with 20µM CFDA-SE (Thermofisher) for 2hrs at 37° as previously described (22). To remove the excess of dye, SEC using 35 nm qEVoriginal columns was performed. The CFSE+EVs were collected on fractions 7-10 (2 mL) and the concentration of particles and the % of fluorescence was evaluated by NTA.

Spectrometer analysis.
The fluorescence in each sample was quantified using a spectrophotometer (iD3 SpectraMax microplate reader. Molecular Devices, California, USA). Triplicates containing 3x10 8 particles from each sample were measured in 96-well flat-bottom black plates.

PBMCs isolation.
Fresh blood pockets from healthy donors were processed the same day of arrival on a and Ratio was generated. Gating strategy was also used to evaluated "Small" and "Large" followed by "Ch02+" events plus "Intern GFP" and then "Low H" and "High H".