Immunosuppressive properties of cytochalasin B-induced membrane vesicles of mesenchymal stem cells: comparing with extracellular vesicles derived from mesenchymal stem cells

Extracellular vesicles derived from mesenchymal stem cells (MSCs) represent a novel approach for regenerative and immunosuppressive therapy. Recently, cytochalasin B-induced microvesicles (CIMVs) were shown to be effective drug delivery mediators. However, little is known about their immunological properties. We propose that the immunophenotype and molecular composition of these vesicles could contribute to the therapeutic efficacy of CIMVs. To address this issue, CIMVs were generated from murine MSC (CIMVs-MSCs) and their cytokine content and surface marker expression determined. For the first time, we show that CIMVs-MSCs retain parental MSCs phenotype (Sca-1+, CD49e+, CD44+, CD45−). Also, CIMVs-MSCs contained a cytokine repertoire reflective of the parental MSCs, including IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12(p40), IL-13, IL-17, CCL2, CCL3, CCL4, CCL5, CCL11, G-CSF, GM-CSF and TNF-α. Next, we evaluated the immune-modulating properties of CIMVs-MSCs in vivo using standard preclinical tests. MSCs and CIMVs-MSCs reduced serum levels of anti-sheep red blood cell antibody and have limited effects on neutrophil and peritoneal macrophage activity. We compared the immunomodulatory effect of MSCs, CIMVs and EVs. We observed no immunosuppression in mice pretreated with natural EVs, whereas MSCs and CIMVs-MSCs suppressed antibody production in vivo. Additionally, we have investigated the biodistribution of CIMVs-MSCs in vivo and demonstrated that CIMVs-MSCs localized in liver, lung, brain, heart, spleen and kidneys 48 h after intravenous injection and can be detected 14 days after subcutaneous and intramuscular injection. Collectively our data demonstrates immunomodulatory efficacy of CIMVs and supports their further preclinical testing as an effective therapeutic delivery modality.

EVs isolation. Murine MSCs were seeded 1 × 10 6 cells per 10 cm 2 and incubated overnight. Following the overnight incubation, cells were washed in phosphate-buffered saline (PBS) and fresh media containing EVdepleted FBS (Gibco, UK) was applied. EV-depleted FBS was obtained by centrifugation at 120,000×g for 18 h at 4 °C. Cells were incubated for 48 h under standard conditions. After 48 h, the media was collected. The conditioned medium was centrifuged at 400×g for 5 min at 4 °C. The resulting supernatant was sequentially centrifuged at 2,000×g for 20 min at 4 °C and 10,000×g for 45 min. Then the supernatant was transferred to ultracentrifuge tube and centrifuged at 100,000×g for 90 min at 4 °C using SW28Ti rotor (Beckman Coulter, USA) in the BECKMAN L70 ultracentrifuge (Beckman Coulter, USA).

Characterization of the CIMVs and EVs. Scanning electron microscopy. CIMVs-MSCs and MSCs-de-
rived EVs were fixed (10% formalin for 15 min), dehydrated using graded alcohol series and dried at 37 °C. Prior to imaging, samples were coated with gold/palladium in a Quorum T150ES sputter coater (Quorum Technologies Ltd, United Kingdom). Slides were analyzed using Merlin field emission scanning electron microscope (Carl Zeiss, Germany). For size analysis, three independent batches of CIMVs were produced and used to generate at least six electron microscope images for each batch. Data collected was used to determine the CIMVs size.
Leukocyte viability. Viability of leukocytes from spleen, thymus and bone marrow was analyzed. Annexin Statistical analysis. Statistical analysis was done using Wilcoxon signed-rank test (R-Studio) with significance level p < 0.05. Illustrations were generated with "ggplot2" package. Data on cytokine content of CIMVs-MSC and MSC were plotted as a heatmap.
CIMVs-MSCs size. While CIMVs have previously been derived from multiple cell lines, the ability to generate CIMVs from murine MSCs has not previously been reported. Here, we have shown that CIMVs with the size ranging from 100 to 1,300 nm could be produced using primary murine adipose MSCs (Fig. 3A,B). We have characterized EVs derived from the same murine MSCs to compare their size with CIMVs. We found that EVs derived from MSCs are from < 50 to 200 nm in size (Fig. 3C,D). Next, we characterized the EVs and CIMVs based on the presence of the CD9 and CD63 transmembrane tetraspanins proteins. We found that 76.8 ± 2.1% MSCs express CD9 and 15.9 ± 1.2% express CD63, whereas 46.9 ± 5% of CIMVs-MSCs are CD9 positive, 6.8 ± 1.1% of CIMVs-MSCs are CD63 positive, and 30.8 ± 4.8% of EVs are CD9 positive and 27.9 ± 4.1% of EVs are CD63 positive ( Supplementary Fig. S1). In addition, characterization of nucleic acid content in CIMVs was performed using PCR ( Supplementary Fig. S2). We found that MSCs derived CIMVs contain COI DNA indicating the presence of mitochrondia and their components, and but do not contain 18S rRNA DNA (marker of nuclear DNA) ( Supplementary Fig. S2).
Next, we sought to determine the effect of the allogenic MSCs or CIMVs-MSCs on the neutrophil activity. Neutrophils were isolated from mice following intravenous injection of allogenic MSCs (7.5 × 10 4 /mouse) or CIMVs-MSC (15 µg/mouse). Neutrophils were activated with opsonized zymosan (10 mg/ml) in vitro. The release of the reactive oxygen species (ROS) was detected by chemiluminescence and used to determine neutrophil activation (Supplementary Fig. S3A). There was no difference between leukocyte activation in mice treated with MSCs or CIMVs-MSCs as compared to control mice, indicating that neither MSCs or CIMVs-MSCs affected neutrophil activity. The level of spontaneous chemiluminescence and the maximum value of stimulated chemiluminescence are summarized in Supplementary Fig. S3B.  www.nature.com/scientificreports/ Finally the effect of MSCs and CIMVs-MSCs on macrophage activity was quantified. Macrophages were isolated by peritoneal lavage from mice pretreated with MSCs, CIMVs-MSCs or sterile saline (negative control). Macrophage phagocytic activity was determined by neutral-red dye uptake. We found that the phagocytic index (PI) of macrophages from the negative control was 0.52 ± 0.14, while it was lower in cells from mice pretreated with MSCs (0.39 ± 0.07; p = 0.23) or CIMVs MSC (0.34 ± 0.13; p = 0.4) ( Supplementary Fig. S4) this was not statistically significant.
To evaluate the influence of CIMVs on activation of lymphocytes, human PBMCs have been incubated with CIMVs derived from human MSCs for 24 h and then were treated with 10 μg/ml PHA (M021, PanEco, Russia) ( Supplementary Fig. S6). We observed that CIMVs-MSCs do not induce T-cells proliferation compared to control (1.6 ± 0.7% of proliferating cells vs. 1.6 ± 0.4%, respectively). Treatment of human PBMCs with PHA induced T-cells proliferation up to 87.7 ± 2%. Pretreatment of PBMCs with CIMVs led to the inhibition of PHA-activated proliferation of T-cells in 1.4 times (62.8 ± 1% of proliferating cells) (Supplementary Fig. S6).

In vivo analysis of CIMVs-MSCs location.
To analyze in vivo distribution, CIMVs-MSCs were stained with vital membrane dye DiD (Invitrogen, USA) and injected intravenously (50 μg), subcutaneously (25 and 50 µg) or intramuscularly (50 µg). We found that 1 h after intravenous injection of CIMVs-MSCs the fluorescence signal was localized in internal organs presumably in lung (Fig. 8A). To accurately conclude which organ the fluorescent signal originated from the organs were imaged ex vivo (Fig. 8E,F). Mice were perfused prior to organ harvest. In addition, to exclude the risk of detection of free dye per se the supernatant (PBS) after the last washing step was injected in control mice. As shown in Fig. 8E CIMVs-MSCs accumulated mainly in liver and lung, a low signal of CIMVs-MSCs was observed also in spleen and brain. At 48 h, the fluorescent signal remained in liver and lung and started to increase in brain, heart, spleen and kidneys (Fig. 8F).
It appears that the fluorescence signal intensity correlated with the amount of administered CIMVs-MSCs (Fig. 8B,D). When mice received 25 µg CIMVs-MSCs subcutaneous, the fluorescence intensity was 2.7 ± 1.2 RFU, while it was 5.5 ± 1.8 RFU (p = 0.09) when mice were injected with 50 µg of CIMVs-MSCs. Interestingly, the fluorescence signal was detectable 48 h and 14 days after subcutaneous and intramuscular injection respectively www.nature.com/scientificreports/ (Fig. 8C,D). These findings may be explained by CIMVs-MSCs uptake in tissue and artefacts due to the long half-life of the dye 25 .

Discussion
Multiple methods have been used to produce membrane vesicles 26,27 . Treatment with cytochalasin B has been shown to be effective for the generation of vesicles that resemble naturally formed microvesicles in displaying physiological cytoplasmic membrane and in size 14 . EVs are isolated from extracellular medium, such as conditioned cell culture medium or body fluids 28 . Because CIMVs are produced from washed cells and their production protocol does not involve active sorting of molecules within the cells, they are distinct from natural EVs. One of the limitations of the use of CIMVs in preclinical and clinical trials is that current understanding of the immune properties of CIMVs derived from MSCs remains limited.
Recently we have shown that CIMVs can be successfully generated from primary human adipose MSCs 29 . They have similar content, immunophenotype, and angiogenic activity to those of the parental MSCs 29 . Here, for the first time, we report that murine MSCs can be used to generate CIMVs. We found that the majority of murine CIMVs-MSCs (94.75%) have a diameter of 100-700 nm, which is within the range of naturally produced microvesicles 30 . To compare the size of CIMVs and EVs we have applied the most commonly employed protocols for the EVs isolation 31 . We found that EVs are between < 50 and 200 nm in size. EV isolation protocols based on ultracentrifugation leads to the preferential enrichment of exosomes. Next, we compared the expression of CD9, CD63, which are part of the key EV criteria as agreed by the International Society of Extracellular Vesicles, on the surface of EVs and CIMVs. We found that EVs express more CD63-positive (+ 4.1-fold) and less CD9-positive (− 1.5-fold) as compared to CIMVs (Supplementary Fig. S1). It is known that CD63 is a major player in exosomes production. CD63 is often enriched in late endosomal and lysosomal compartments as well as in exosomes 32 . Whereas CD9 has an extracellular domains and is one of the markers of MSCs 33 . Therefore we confirmed the endosomal origin of EVs, whereas CIMVs production protocol involves the budding of membrane vesicles from the cell surface and as a result CIMVs enclosed by a cytoplasmic membrane. www.nature.com/scientificreports/ We detected COI gene DNA in CIMVs which may indicate inclusion of mitochondria and derived components in the process of CIMVs production. It is known that natural EVs of BMSCs also contain mitochondria 34 . The absence of nuclei content in the CIMVs fraction indicate that CIMVs are not able to divide and might be a safer alternative than cell therapy.
It was previously shown that soluble factors play a major role in the immune suppressive effects of MSCs 37 . Several cytokines found in murine MSCs could cause immune suppression. For example, IL-6 could inhibit monocyte differentiation and stimulate T cells 35 . Anti-inflammatory IL-10 could inhibit cytokine secretion, dendritic cell differentiation and antigen presentation 38 . Also, IL-4 and IL-13 could trigger macrophage differentiation into immunosuppressive M2 subset 39 . Interestingly, levels of these cytokines were increased in murine CIMVs-MSCs as compared to MSCs. The mechanism by which cytokines encapsulated in lipid vesicles influence target cells was discussed by Fitzgerald et al. 40 . It is known that cytokine receptors are located on the cell surface 41 . The authors suggested that EVs might release entrapped cytokines during the interaction with the cell surface as a result of leaky membrane formation or in the process of fusion 40 .
Similarities in the surface receptor expression and molecular content suggest that CIMVs-MSCs and MSCs could have similar immune modulatory activity. Therefore we analyzed the effect of CIMVs-MSCs and MSCs on humoral immune response in mice. Our data revealed that CIMVs-MSCs and MSCs suppress antibody production (Fig. 6). The antibody titers to SRBC in mice pretreated with allogenic MSCs or CIMVs-MSCs were lower than the antibody titer in the serum of control mice. The immune inhibitory effect could be explained by the high level of anti-inflammatory cytokine content of CIMVs-MSCs. www.nature.com/scientificreports/ Previously Budoni et al. 42 demonstrated the inhibitory effects of EVs on B-cell proliferation and antibody production. Therefore we compared the immunomodulatory effect of MSCs, CIMVs and EVs. We observed no immunosuppression in mice pretreated with natural EVs, whereas MSCs and CIMVs-MSCs suppressed antibody production (Fig. 6). Immunosuppressive activity of EVs-MSCs has been actively discussed and few authors demonstrated lower/or absence of inhibitory effect compared to parental MSCs. Our findings support the results from Conforti and colleagues showing that MSCs were significantly more capable to inhibit T-cell proliferation and antibody secretion in vitro compared to EVs 43 . Gouveia de Andrade and colleagues observed that EVs derived from bone marrow (BM-MSCs) and adipose tissue (AT-MSCs) failed to suppress lymphocyte proliferation 44 . Trapani and colleagues observed that the immunosuppression effects of EVs was less dramatic compared to the MSCs and was proportional to their uptake by immune cells population 45 . It is believed that observed contrasting action of EVs-MSCs is due to method of isolation, purity and medium size of isolated EVs 44 . In this context CIMVs might be more relevant substitute for MSC administration combining advantages of safety and ease of production with retaining parental MSCs immunomodulatory activity. In addition 36 , we observed a modest macrophage suppressive effect of CIMVs-MSCs. These data suggests that CIMVs-MSCs have inhibitory effect similar to MSCs.
It has been shown that the hemato-lymphoid system is highly sensitive to inhibitory stimuli 46 . Therefore, we analyzed the effect of CIMVs-MSCs and MSCs on the leukocyte count in major lymphoid organs such as spleen, thymus and bone marrow (Supplementary Fig. S5). Leukocyte counts in all lymphoid organs were decreased as compared to control. Although these differences did not reach statistical significance, further studies are required to determine the effect of prolonged CIMV-MSC treatment on leukocyte function and number. The changes in the leukocyte counts were not caused by the cell death, since the viability of leukocytes isolated from the spleen, thymus and bone marrow was not affected by incubation with CIMVs-MSCs (Fig. 7). This data corroborates our previous publication where human CIMVs-MSCs also did not affect the cell viability 14 . Therefore, decreased antibody production in mice receiving MSCs and CIMVs-MSCs is more likely attributable to an immunomodulatory mechanism as opposed to decreased leukocyte count. MSCs and CIMVs linked immunosuppression in mice without effecting leukocyte viability suggests that the hemato-lymphoid cells suppression was due to the inhibition of proliferation and/or differentiation.
Recently Khare et al. demonstrated the inhibitory effect of MSCs derived natural EVs on the proliferation of activated PBMCs and isolated T and B cells 47 . Therefore, we performed T-cell suppression assay in vitro to evaluate the influence of CIMVs on T-cells. We observed the inhibitory effect of allogeneic CIMVs-MSCs on PHA-activated proliferation of T-cells in 1.4 times.
For the first time we have demonstrated the bio-distribution of murine CIMVs-MSCs in vivo following intravenous, subcutaneous and intramuscular injection. We have found that intravenously injected murine CIMVs-MSCs 1 h after injection were localized in the vessel-rich organs such as lung and liver, small amount of CIMVs-MSCs reached spleen and brain (Fig. 8E). After 48 h CIMVs-MSCs were redistributed and localized in liver, lung, as well as in brain, heart, spleen and kidneys (Fig. 8F). These findings may be explained by an uptake of remaining in blood CIMVs-MSCs and their gradual renal excretion. Wiklander and colleagues conducted an extensive biodistribution investigation of EVs and showed that EVs were accumulated mainly in liver, spleen, gastrointestinal tract and lungs 24 h after the systemic injection 25 .
In our study we have showed that CIMVs-MSCs could be detected within the organ up to 14 days following administration via subcutaneous and intramuscular injection, most likely due to incorporation into the tissue and long half-life of the dye. Therefore, we suggest that the subcutaneous and intramuscular injection of CIMVs-MSCs is more suitable for the local therapy.

conclusions
CIMVs can be generated from human and murine MSCs. CIMVs retain the content, immunophenotype, and biological activity of the parental MSCs. CIMVs are similar in size to naturally produced microvesicles and do not contain the nuclei. One limitation to clinical exploitation of CIMVs derived from MSC was an absence of detailed characterization of the immune phenotypes. For this reason, we analyzed the immune modulatory properties of murine MSC and CIMVs-MSCs. The murine CIMVs-MSCs closely resemble parental MSCs and have similar phenotype and content of the biologically active molecules. CIMVs-MSCs also demonstrate suppression of antibody production similar to parental MSCs. There are evidences that the EVs-MSCs immunosuppressive effect is lower compared to parental MSCs. We observed that natural EVs derived from MSCs failed to suppress antibody production in vivo. Our study provides new evidence on the CIMVs as a perspective candidate of a cell-free therapeutics which retain the immunosuppressive activity of parental MSCs.

Data availability
All data generated or analysed during this study are included in this published article (and its Supplementary  Information files). The data that support the fndings of this study are available from the corresponding author upon request.