Membrane particles generated from mesenchymal stromal cells modulate immune responses by selective targeting of pro-inflammatory monocytes

Mesenchymal stromal cells (MSC) are a promising therapy for immunological disorders. However, culture expanded MSC are large and get trapped in the capillary networks of the lungs after intravenous infusion, where they have a short survival time. Hypothetically, living cells are a risk for tumor formation. To reduce risks associated with MSC infusion and improve the distribution in the body, we generated membrane particles (MP) of MSC and MSC stimulated with IFN-γ (MPγ). Tracking analysis and electron microscopy indicated that the average size of MP was 120 nm, and they showed a round shape. MP exhibited ATPase, nucleotidase and esterase activity, indicating they are enzymatically active. MP and MPγ did not physically interact with T cells and had no effect on CD4+ and CD8+ T cells proliferation. However, MP and MPγ selectively bound to monocytes and decreased the frequency of pro-inflammatory CD14+CD16+ monocytes by induction of selective apoptosis. MP and MPγ increased the percentage of CD90 positive monocytes, and MPγ but not MP increased the percentage of anti-inflammatory PD-L1 monocytes. MPγ increased mRNA expression of PD-L1 in monocytes. These data demonstrate that MP have immunomodulatory properties and have potential as a novel cell-free therapy for treatment of immunological disorders.

Generation and characterization of Membrane Particles (MP). MP were generated from unstimulated and IFN-γ stimulated AT-MSC. The number of cells used for each analysis was between 1 × 10 6 -1.5 × 10 6 cells (80% confluency). The size distribution of the obtained MP was studied using Nanoparticle tracking analysis (NTA). The size of the particles ranged from 63 to 700 nm (Fig. 2a), and the mode size of the samples was 121.7 ± 35.5 nm for MP and 138.3 ± 62.1 nm for MPγ (Fig. 2b). The percentage of particles with a size larger than 200 nm was lower than 5% in every sample.
Based on the particle concentration per ml, the average number of particles generated from each MSC was 1.2 × 10 5 ± 2.7 × 10 4 for MP and 1.1 × 10 5 ± 2.8 × 10 4 for MPγ (Fig. 2c). There was no significant difference in size distribution or concentration (MP/MSC) between MP and MPγ.
The transmission electron microscopy images illustrate that MP consist of a population of particles heterogeneous in size with most of the particles showing a size of less than 200 nm (Fig. 2d) but some showing larger sizes. This result confirms the NTA analysis. It can be clearly observed that both the larger and smaller MP have a round shape.
Membrane Particles from AT-MSC possess enzyme activity. To analyze whether MP have enzyme activity, we examined the ability of MP and MPγ to convert ATP to ADP by ATPase activity, and AMP to adenosine by the nucleotidase activity of CD73. The last product of these two reactions is free phosphate, so the samples for these assays were prepared in milliQ water to avoid contamination with free phosphates from saline buffers. Before measurement of enzyme activities, MP (diluted in milliQ water) were analyzed by NTA for determination of their concentration. Figure 3a shows the ATPase activity (units/l) calculated from the standard curve generated with known free phosphate concentrations. MP and MPγ were able to convert ATP to free phosphate and the level of free phosphate was dependent on the concentration of MP. There was no statistical difference between MP and MPγ.
To examine whether MP and MPγ possess CD73 activity, the production of free phosphates by 2, and 1 ng of purified CD73 was compared with different concentrations of MP, and MPγ. Both types of MP were able to produce free phosphates after adding the substrate (AMP). The detection of free phosphate was dependent on concentration of MP and the amount of CD73 present in MP was calculated through the CD73 controls (Fig. 3b).
Esterase activity was measured by the conversion of non-fluorescent CFDA-SE to fluorescent CFSE by MP using flow cytometry based on a FITC fluorescence triggering strategy (Fig. 3c). This fluorescent-based flow cytometry protocol allows detection of particles based on positive fluorescence signals, not on size, as the average MP size of 120 nm is too small to be detected by most flow cytometers. Controls used for this flow cytometry protocol were PBS + CFDA-SE, and non-labeled MP (top 2 graphs). As expected, these controls were negative as no CFSE fluorescence can be expected. When MP were incubated with CFDA-SE, they converted CFDA to fluorescent CFSE, as demonstrated by the detection of fluorescent events (lower 4 graphs) showing that MP have esterase activity. As an additional control, MP were diluted before CFSE staining. The results shown indicate the recording of all samples during 1 min. The number of detected particles decreased for more diluted samples, but the MFI of the CFSE staining of the particles remained the same. This means that single MP can be detected with the used flow cytometry strategy. Fluorescent-based flow cytometry protocols were recently described in literature 29,30 . Effects of Membrane Particles on T cell proliferation. CFSE loaded human peripheral blood mononuclear cells (PBMC) stimulated with anti-CD3/antiCD28 antibodies were cultured with different ratios of MP for 4 days (1:5,000, 1:10,000, 1:40,000, 1:80,000). To analyze lymphocyte proliferation, CFSE dilution was measured in CD4 + and CD8 + T cells. Addition of increasing concentrations of MP or MPγ did not affect the proliferation of CD4 + and CD8 + T cells ( Fig. 4a and b).
Membrane Particles decrease the proportion of CD16 + monocytes and increase CD90+ and PD-L1+ monocyte subsets. Monocytes were cultured with different ratios of MP for 24 h (1:10,000, 1:40,000, 1:80,000) to determine whether MP could affect monocyte cell surface markers expression and immune function. Monocytes were cultured in polypropylene tubes to avoid the adherence of the cells and differentiation into macrophages. Culture of monocytes in the presence of MP or MPγ treatment decreased the frequency of pro-inflammatory CD14 + CD16 + cells at ratios of 1:40,000 (by 45% and 49%, respectively) and 1:80,000 (by 48% and 35%, respectively) (Fig. 5a).
Monocytes treated with MP at ratios of 1:40,000 and 1:80,000 furthermore increased the expression of CD90 by 17% and 25%, respectively. Meanwhile, the MPγ group showed an increase in CD90 expression at ratios of 1:10,000 by 8%, 1:40,000 by 16% and 1:80,000 by 20% (Fig. 5b). Moreover, MPγ treatment induced anti-inflammatory PD-L1 expression in monocytic cells by 16% at a 1:10,000 ratio, 43% at a 1:40,000 ratio and Membrane Particles affect the expression of pro-and anti-inflammatory genes in monocytes. In order to examine the effect of MP on monocyte immune function, and to examine whether the immunophenotypic changes observed were a result of protein transfer or of gene expression regulation, mRNA expression of a number of genes with pro-and anti-inflammatory function was analyzed in monocytes by qPCR after 24 h of stimulation with MP. Upregulation of CD90 gene expression as a result of particles stimulation was observed in MP and MPγ treated monocytes (p < 0.05) (Fig. 5d). Moreover, expression of the anti-inflammatory factors IDO and PD-L1 was increased in monocytes treated with MPγ, but not MP (p < 0.05) (Fig. 5d). There was a trend for increased expression of IL-6 after MP and MPγ treatment, but this was not significant. Significant changes in gene expression were also not observed for the pro-inflammatory cytokines TNF-α and anti-inflammatory cytokine IL-10.
Monocytes but not lymphocytes are able to take up Membrane Particles. Since the previous results showed that MP had immunomodulatory properties on monocytes but not on lymphocytes, we analyzed the interaction of MP with both types of immune cells. For that purpose, MP labeled with PKH membrane dye were added to PBMC (ratio 1:40,000) and incubated during 1 h and 24 h at 37 °C. As a control the cells were incubated at 4 °C, at which temperature no active uptake of MP is expected. A representative flow cytometry analysis is showed in Fig. 7a and b.
1 h after the addition of MP, a small percentage of CD3-lymphocytes (1.3 ± 0.2%) were positive for PKH-MP ( Fig. 7c) while 20 ± 5.3% of CD14-monocytes was able to uptake MP (p < 0.05) (Fig. 7d). The difference between the MP uptake by monocytes and lymphocytes was higher after 24 h (lymphocytes: 5.2 ± 1.4%, monocytes: 93 ± 4.3%; p < 0.05). The 4 °C control for uptake was always below 3% for monocytes and lymphocytes in all the time points. This result indicated that MP uptake was mediated in an energy-dependent process. The nucleotidase activity of the MSC marker CD73 was measured for three concentrations of MP (1 × 10 12 , 1 × 10 11 and 1 × 10 10 particles/ml). MP and MPγ were able to produce free phosphates after adding AMP substrate in a dose-dependent fashion. CD73 enzyme (2 and 1 ng) was used to calculate the concentration of CD73 in the MP. There was no statistical difference in enzyme activity between MP and MPγ. (c) Esterase activity of three concentrations of MP (1 × 10 9 , 1 × 10 8 and 1 × 10 7 particles/ml) was measured by the conversion of CFDA-SE to CFSE by flow cytometry. Fluorescent events were observed in MP labeled with CFSE (CFSE-MP), and the number of CFSE-MP detected was dependent on the concentration of MP. There was no statistical difference between MP and MPγ in esterase activity. Controls (PBS + CFSE and non-labeled MP) were negative. Data are presented as mean ± SD. Enzyme activities were detected in MP generated from 5 different MSC donors. To examine whether MP could be internalized by monocytes, confocal immunofluorescence microscopy was performed with isolated CD14 + cells from PBMC. The membrane of the monocytes was labeled with PKH-67 and cultured with PKH-MP (1:40,000). Time-lapse recordings showed that MP bound to the plasma membrane of the monocytes but they were not internalized. To look in detail at the localization of MP on the monocytes, z-stack images were analyzed by confocal microscopy (Fig. 8). These images confirmed that MP remained localized to the cell surface of the monocytes.

Discussion
The immunomodulatory capacity of MSC is often attributed to the secretion of soluble factors 11 . We recently demonstrated that inactivated MSC without the capacity to secrete factors can modulate immune responses in vitro and in vivo 9 . Inactivated MSC showed similar bio-distribution as living MSC as both are trapped in the lungs following intravenous administration. Here, we went one step further and generated nanoparticles from the membranes of adipose tissue MSC with diverse immunomodulatory properties by induction of regulatory proteins on the plasma membrane after treating the MSC with IFN-γ.
To generate MP, supernatants of MSC cultures were discarded and the cells were washed several times with PBS. Hereby the inclusion of soluble proteins in the MP preparations is avoided, which is a major challenge in the field of natural extracellular vesicles (EV) and causing misinterpretation of results 31 . The isolation methods for obtaining EV allow the co-precipitation of proteins, and RNA associated to lipoproteins secreted by the cells 32 . These contaminations mask the functional properties of EV and hamper their therapeutic application. With our novel protocol, we avoid the inclusion of artefacts from soluble molecules, and make MP a good alternative to EV.
Nanosight technology and electron microscopy were used for the characterization of MP. Most of the MP showed a size below 200 nm, and a round shape. Both characteristics make the MP an attractive therapeutic tool. Firstly, because their small size MP can easily maneuver through the capillary network of the lungs and reach sites of action beyond the lungs. Secondly, their morphology (closed circular structures) would allow loading of MP with compounds of interest and use MP as a delivery vehicle for future applications. Sun et al. provided evidence that anti-inflammatory drugs can be loaded into EV from myeloid cells and thereby enhance the delivery of the drug to activated monocytes in a LPS-induced septic shock model 33 . The use of MP from MSC as a natural delivery vehicle would have the advantage that the vehicle per se show immunomodulatory properties, which gives the carrier additional value. It is also important to consider that the production and manipulation of MP is easier and cheaper than the methodology used for the collection of EV, as it is possible to generate about 1.5 × 10 5 particles/ cell.
In addition to their morphological characteristics, MP were shown to possess enzyme activity. It has been reported that extracellular vesicles from MSC have a cargo rich in enzymatically active glycolytic enzymes, ATPases, and ATP-generating enzymes, such as adenylate kinase and nucleoside-diphosphate kinase 34 . Enzymatic activity has been demonstrated to be important for modulating the conditions in the vesicle nano-environment by consuming or generating metabolic energy. Katsuda et al. demonstrated the unique potential of extracellular vesicles from adipose tissue derived MSC for treatment of Alzheimer´s disease. These authors found that these extracellular vesicles carry Neprilysin, a metalloprotease, which ameliorates the disease's symptoms 35 . We showed that MP possess nucleotidase and esterase activity, which are major enzymes regulating immunity and inflammation 21,22 .
Lymphocyte proliferation is the most commonly used assay to demonstrate the immunomodulatory capacity of MSC and it has been used as a standard assay to compare the immunosuppressive effect of MSC from different tissue sources. Comparative studies have sometimes however produced conflicting results. Puissant et al. have reported similar inhibition of T cell proliferation by bone marrow and adipose tissue MSC 36 , whereas Ribeiro et al. found that adipose tissue MSC to have stronger suppressive effects than bone marrow and umbilical cord MSC 8 . In pilot experiments, we generated MP from bone marrow derived MSC. These MP demonstrated similar properties as MP from adipose tissue MSC.
The mechanisms through which MSC suppress lymphocyte proliferation are largely dependent on soluble mediators. In our study, we found no effect of MP on lymphocyte proliferation. This can be explained by the fact that MP cannot secrete soluble factors, but also because lymphocytes were shown to be unable to bind or uptake MP. However, MP induced modulation of monocyte cell surface markers expression and changed their immune function. Furthermore, MP and MPγ induced the selective apoptosis of proinflammatory CD14 + CD16 + monocytes.
CD16 + monocytes are major producers of inflammatory cytokines such as TNF-α and IL-12 37,38 , and high numbers of CD16 + monocytes are associated with acute and chronic inflammatory conditions 39 . Our results therefore suggest that MP act as immunomodulators that eliminate pro-inflammatory monocytes. Importantly, we also found that the immunomodulation induced by MP and MPγ is different. MPγ but not MP increased PD-L1 in the membrane of the monocytes and the mRNA expression of the anti-inflammatory factor IDO. Thus, the modification of the membrane protein composition of MSC by treatment of the cells with various stimuli provides us the opportunity to generate MP adapted for treatment of a specific immunological disorder. For example, MPγ with their enhanced capacity to induce PD-L1 and IDO by monocytes may be suitable for treatment of more severe immune responses involving inflammatory monocytes, while MP derived from MSC pre-treated with factors that induce proteins with regenerative function may be useful for inducing regenerative processes after resolving inflammation. As there is a lot of knowledge about modulation of MSC properties by cytokine treatment, there are tools in hand to control the make-up of MP. Thus, the potential therapeutic applications of MP are far reaching.
We demonstrated that the interaction of MP with monocytes is by binding and fusion with the plasma membrane of the monocytes. This is an active and specific mechanism for monocytes because at low temperatures MP were unable to fuse with the monocyte membranes. It is furthermore specific because MP do not bind to lymphocytes. The confocal microscopy images showed that there is no internalization of MP into monocytes, indicating phagocytosis plays no role in the uptake of MP. The mechanism of binding and fusion of MP with monocyte membranes supports the idea that MP can be a natural delivery vehicle for monocyte-targeting drugs.
In conclusion, MP represent a therapeutic strategy that combines the potential of MSC therapy with reduced risks associated with the use of living cells and improved ability to reach sites beyond the lungs. Our data demonstrates that MP target monocytes, via which they may have a broad immunomodulatory effect (Fig. 9). These data suggest that MP can serve as a novel cell-free therapeutic for treating immunological disorders. Additional studies, both in vitro and in vivo, are needed to improve our understanding the mechanisms of action of this potential immunosuppressive tool.

Materials and Methods
Ethics statement and human tissue samples. The MSC were provided by Internal Medicine Department, Transplantation laboratory of the Erasmus MC (The Netherlands). The cells were isolated from subcutaneous adipose tissue from healthy donors that became available during the kidney donation procedure. The tissues were not procured from prisoners, and were collected after obtaining written informed consent for all patients, as approved by the Medical Ethical Committee of the Erasmus University Medical Centre Rotterdam (protocol no. MEC-2006-190). All experiments were performed in accordance with the approved guidelines.
Isolation and culture of MSC from adipose tissue. Subcutaneous adipose tissue from five healthy human kidney donors became available during the donation procedure. The adipose tissue was collected in minimum essential medium-α (MEM-α) (Sigma-Aldrich, St. Louis, MO) supplemented with 100 IU/ml penicillin, 100 mg/ml streptomycin (P/S) (Lonza, Verviers, Belgium), and 2 mM L-glutamine (Lonza).
The tissue was mechanically disrupted and enzymatically digested with 0.5 mg/ml collagenase type IV in RPMI for 30 min at 37 °C under continuous shaking. Thereafter, the cells were resuspended in MEM-α with 15% fetal bovine serum (FBS; Lonza), 2 mM L-glutamine and 1% P/S, filtered through a 100 µm cell strainer, and transferred to a 175 cm 2 culture flasks (Greiner Bio-one, Essen, Germany).
Two MSC culture conditions were used for the experiments: unstimulated MSC, and pretreated with IFN-γ (50 ng/ml, Sigma-Aldrich). After incubation for 3 days, MSC were collected to generate cell membrane particles. MSC were used for experiments between passages 2 and 6.   measurements per sample (30 s/measurement), temperature 23.61 ± 0.8 °C; viscosity 0.92 ± 0.02 cP, frames per second 25. Each video was analyzed to give the mean, mode, median and estimated concentration for each particle size. The samples were diluted to obtain the right number of particles (1 × 10 8 particles/ml) in accordance with the manufacturer's recommendations.
Transmission electron microscopy examination of MP. After fixation with paraformaldehyde (2%), all the samples were adsorbed for 20 min to glow-discharged carbon coated grids by floating the grids on 10 μL drops on parafilm. Grids with adhered MP were washed with water, stained with 2% uranyl acetate in water and examined in the electron microscope Tecnai T12 Spirit equipped with an Eagle CCD camera 4kx4k (FEI Company, Eindhoven, The Netherlands).
ATPase assay. ATPase activity from MP and MPγ was measured using an ATPase assay kit according to the manufacturer's instructions (Sigma-Aldrich). A phosphate standard was used for creating a standard curve. MP (1 × 10 12 , 1 × 10 11 , 1 × 10 10 and 1 × 10 9 particles/ml) were incubated with 4 mM ATP for 30 min at room temperature in assay buffer with malachite green reagent. The formation of the colorimetric product that formed in the presence of free phosphates was measured with a spectrophotometer at 620 nm.
As a control for possible phosphate contamination, the four MP concentrations were incubated in assay buffer without ATP. The signal from these samples was subtracted from the samples incubated with ATP. CD73 activity assay. A modified protocol of CD73 inhibitor screening assay kit (BPS Bioscience) was used to determine whether MP were able to degrade AMP into adenosine plus phosphate. MP and MPγ (1 × 10 12 , 1 × 10 11 and 1 × 10 10 particles/ml) were incubated with AMP (500 μM) during 25 min at 37 °C. Then, colorimetric detection reagent was added to measure the free phosphate from the CD73 reaction. Samples without AMP were measured as a control for free phosphate contamination. CD73 enzyme (2 and 1 ng) was used to calculate the concentration of CD73 in the MP, and MPγ.
Esterase activity by CFSE. CFDA-SE, which is non-fluorescent, enters the cytoplasm of cells where intracellular esterases remove the acetate groups and convert the molecule to a fluorescent ester (CFSE). This conversion was used to detect whether MP have esterase activity. After MP generation, 1 × 10 10 , 1 × 10 9 , 1 × 10 8 and 1 × 10 7 particles/ml were labeled with 50 μM of CFDA-SE and incubated at 37 °C for 30 min. Dilution of the MP was performed to obtain a proper stoichiometry of the CFSE staining. PBS + CFDA-SE and non-stained MP were used as controls.
CFSE fluorescence was measured by flow cytometry (FACS Canto II, BD Biosciences). Due to the small size of the MP, reliable FSC and SSC measurements could not be obtained. Instead, MP were identified by setting a fluorescence threshold triggering on the CFSE fluorescence so that events above the threshold could be identified as CFSE-loaded MP.
CD3/CD28 T cell proliferation assay. To evaluate the immunomodulatory capacity of MP, PBMC were labeled with 1 μM of CFSE and plated in round bottom 96-well culture plates at a density of 5 × 10 4 cells/well. T cell proliferation was stimulated by adding human anti-CD3/anti-CD28 antibodies (1 µl/ml each) with a linker antibody Ig (2 µl/ml) (BD Biosciences). PBMC were incubated with different ratios of MP, or MPγ (1:5,000, 1:10,000, 1:40,000, 1:80,000) for 4 days. On the fourth day, non-adherent PBMC were removed from the plate, washed with FACS Flow and incubated with monoclonal antibodies against CD4-PerCP and CD8-PE-Cy7 (antibodies were purchased from BD Biosciences) at room temperature for 30 min. When a CFSE-labeled cell divides, its progeny are endowed with half the number of CFSE-tagged molecules and thus each cell division can be assessed by measuring the corresponding decrease in cell fluorescence by flow cytometry.
Apoptosis of monocyte subsets. Monocytes were cultured with MP, or MPγ at different ratios (1:10,000, 1:40,000 and 1:80,000) in polypropylene tubes overnight. Then, cells were incubated with monoclonal antibodies against CD14-Pacific Blue and CD16-FITC (antibodies were purchased from BD Biosciences) at room temperature for 30 min. After washing step, cells were stained with fluorochrome-conjugated Annexin-V for 15 min at RT to assess the apoptotic cells. All data were measured on a FACSCanto II flow cytometer (BD) and analyzed using FACSDiva software.