Crosstalk of Endothelial and Mesenchymal Stromal Cells under Tissue-Related O2

Mesenchymal stromal cells (MSCs) are considered a valuable tool for cell therapy. After systemic administration, the outcome of MSCs and endothelial cells (ECs) interactions strongly depend on the local microenvironment and tissue O2 levels in particular. In vitro analysis of EC effects on MSC regenerative potential in co-culture was performed after short-term interaction at “physiological” hypoxia (5% O2) and acute hypoxic stress (0.1% O2). At 5% O2, MSCs retained stromal phenotype and CFU-f numbers, osteogenic RUNX2 was upregulated. A shift in the expression of adhesion molecules, and an increase in transcription/synthesis of IL-6, IL-8 contributed to facilitation of directed migration of MSCs. In the presence of MSCs, manifestations of oxidative stress in ECs were attenuated, and a decrease in adhesion of PBMCs to TNF-α-activated ECs was observed. Under 0.1% O2, reciprocal effects of ECs and MSCs were similar to those at 5% O2. Meanwhile, upregulation of RUNX2 was canceled, IL-6 decreased, and IL-8 significantly increased. “Protective” effects of MSCs on TNF-α-ECs were less pronounced, manifested as NOS3 downregulation and intracellular NO elevation. Therefore, interaction with ECs at “physiological” hypoxia enhanced pro-regenerative capacities of MSCs including migration and anti-inflammatory modulation of ECs. Under acute hypoxic stress, the stimulating effects of ECs on MSCs and the “protective” potential of MSCs towards TNF-α-ECs were attenuated.


Introduction
Multipotent mesenchymal stromal cells (MSCs) are localized in a number of tissues and can be isolated and cultured in vitro. MSCs demonstrate a "trophic activity" due to the secretion of growth factors, cytokines, and other soluble mediators, including those with immunomodulatory properties, and have multilineage potential as well [1][2][3][4]. The above features determine the interest in MSCs as a perspective tool for additional therapy of various diseases [5][6][7][8]. Due to their immune evasiveness, not only autologous but also allogeneic MSCs can be used in clinical practice after in vitro processing. Currently, systemic administration is one of the most commonly used methods for the MSC delivery to target tissues [9,10]. It is assumed that upon entering the circulation, MSCs can follow the chemoattractant gradients and, reaching the area of damage, transmigrate through the vascular wall endothelium to the periendothelial space. It is this phenomenon that underlies the intravascular delivery of MSCs in cell therapy protocols [11][12][13]. It is obvious that interaction with endothelial cells (ECs) is one of the important factors that determine the fate of injected MSCs.
Upon systemic administration into circulation, the time of MSC interaction with the endothelial monolayer was shown to be several hours [9,14,15]. The data on the effects of ECs on MSCs during this time is limited. After short-term (24-72 h) interaction with ECs, it was demonstrated the formation of EC clusters surrounded by MSCs, upregulation of genes encoding proteins of myogenic differentiation (SMA-actin, calponin, myocardin) [16,17], and elevation of IL-6 production [18]. These findings indicate changes in the functional activity of MSCs, which may affect their subsequent involvement in reparation in vivo.
The outcome of MSC-EC communications significantly depends on the local microenvironment, including those caused by low O 2 (tissue-related levels). The terms "physioxia" [19] or "physiological" hypoxia [20] were proposed to refer to such O 2 conditions. In vitro, the different features of MSCs at tissue-related O 2 ("physiological" hypoxia) vs. standard laboratory level (20% O 2 ) were clearly demonstrated. At 5% O 2 , MSCs proliferated more actively, had more CFU-f, and susceptibility to osteo-and adipogenic differentiation stimuli was attenuated, which contributed to the maintenance of their low committed state [21][22][23][24][25][26][27]. Accordingly, interplay with ECs may affect the above MSC properties.
MSC administration was successfully applied in pre-clinical and clinical studies to cure ischemic disorders with impaired tissue oxygenation [28][29][30][31]. In this case, transplanted MSCs were affected by a number of factors: deprivation of growth factors, elevated inflammatory cytokines, and an acute decrease in O 2 -hypoxic stress [32,33]. Short-term hypoxic exposures (1-72 h, <1% O 2 ) are known to induce certain functional changes in MSCs, such as an increased contribution of glycolysis to the ATP metabolism [24,[34][35][36][37]. The above has an effect on the MSC regenerative potential, resulting in cytoskeletal reorganization and alteration of their migratory activity [38,39]. In vitro studies have shown that short-term hypoxic stress activates HIF-1α-dependent signaling pathways responsible for MSC adaptation and survival [35,36,[40][41][42]. Activation of signaling cascades stimulates the secretion of various paracrine factors that have a "trophic" effect on damaged tissues [43][44][45]. Therefore, acute O 2 deprivation, typical of damaged tissues, can have an ambiguous effect on MSCs and their communication with ECs as well.
The purpose of this study was to determine how a short-term interaction under "physiological" hypoxia (5% O 2 ) and short-term hypoxic stress (0.1% O 2 ) affect the functional activity of ECs and MSCs, in particular, the MSC functions requested for their reparative potential implementation, as well as on the MSC ability to regulate inflammatory response of ECs.

Human Peripheral Blood Mononuclear Cells' (PBMCs) Isolation
Human PBMCs were isolated from blood samples that were collected from healthy volunteers with written informed consent. A density gradient centrifugation with Histopaque ® -1077 (Sigma-Aldrich, St. Louis, MO, USA) was used for separation of PBMCs [48]. Cells were maintained in RPMI 1640 (Gibco, Life Technologies, Carlsbad, CA, USA) supplemented with heat-inactivated 5% FBS, 50 U/mL penicillin, and 50 µg/mL streptomycin.

Experimental Design
The experimental procedures are illustrated in Figure 1. MSCs and ECs were cultured to a confluence of 80-90%. For co-culture experiments, MSCs were trypsinized and added to the EC monolayer in a 1:1 ratio. The culture medium was replaced in all cell cultures with a fresh one containing α-MEM and 199 (1:1 ratio). Then, EC or MSC monocultures and their co-cultures were exposed to hypoxic conditions for 24 h. Hypoxic chambers (StemCell Technology, Seattle, WA, USA), with the O 2 level controlled by an O 2 sensor, were used to maintain cells at 5% or 0.1% O 2 . After hypoxic exposure, the samples of conditioned medium (CM) were collected, centrifuged at 2500× g to remove cell debris, and stored at −80 • C (low-temperature freezer, Sanyo, Osaka, Japan) for ELISA. MSCs and ECs were grown as monocultures till the confluence of 80-90%. For co-culture experiments, MSCs were harvested by trypsinization and added to the EC monolayer in a 1:1 ratio. EC or MSC monocultures and their co-cultures were exposed to hypoxic conditions (5% or 0.1% O 2 ). After 24 h, the samples of conditioned medium (CM) were collected for ELISA, cells were separated using an immunomagnetic protocol and analyzed with flow cytometry, RT-PCR, and MSCs were seeded in limiting density for CFU-f evaluation. To examine the effects of MSCs on inflammatory activation of ECs, ECs were treated with TNF-α for 24 h, and then a suspension of MSCs was added to the EC monolayer. After 24 h of hypoxic exposures at 5% or 0.1% O 2 , conditioned medium was collected; cells were immunoseparated and then processed as above. Besides, a part of TNF-α primed ECs in monoculture and co-culture with MSCs was used in the PBMC adhesion assay.
Cells were washed with PBS (PanEco, Moscow, Russia), trypsinized, and separated by magnetic immunoseparation (Miltenyi Biotec, Auburn, CA, USA). For this, cell suspensions were mixed with magnetic microparticles bearing antibodies against endothelial marker PECAM (CD31) (Miltenyi Biotec, Auburn, CA, USA), and incubated for 15 min at +4 • C. After incubation, the cell suspensions were washed from unbound particles, resuspended in fresh culture medium without FBS, and placed on a column fixed in a magnet. Microparticles were retained in a magnetic field, while a suspension of unbound cells (MSCs, CD31-) came out through the pores of the column. The column was removed from the magnet and washed with 199 medium to collect ECs (ECs, CD31+). After magnetic separation, cells were harvested by centrifugation at 1500× g and analyzed by flow cytometry, real-time (RT) PCR, and MSCs were seeded in limiting density for CFU-f evaluation.
To examine the effects of MSCs on the TNF-α-induced activation of ECs, the experiments were executed as follows ( Figure 1). ECs in a monolayer were exposed to TNF-α for 24 h. The ECs were washed with PBS and a suspension of MSCs was added as described above. After 24 h of hypoxic exposure at 5% or 0.1% O 2 , the conditioned medium was collected; cells were immunoseparated and then processed as above. Besides, a part of TNF-α primed ECs in monoculture and co-culture with MSCs was used in the PBMC adhesion assay.
NO was detected using DAF-FM diacetate (Molecular Probes Inc., Eugene, OR, USA) according to the manufacturer's instructions. Briefly, mono-and co-cultures were washed with PBS and incubated with 5 µM DAF-FM diacetate for 20 min at 37 • C. After washing with PBS, fresh medium was added; cells were incubated for an additional 30 min. After this, cells were trypsinized, stained with anti-CD90-APC antibodies, and analyzed by flow cytometry (Ex/Em = 490/530 nm for NO). The NO level in ECs was estimated as the mean fluorescence intensity (MFI) of CD90 negative cells.

CFU-f Assay
For CFU-f evaluation, MSCs were plated at a limiting density-100 cells per 35 mm cell culture dish (Corning, NY, USA). After 14 days, the dishes were stained with 0.5% crystal violet solution for 5 min (Sigma-Aldrich, Saint Louis, MO, USA), and colonies with ≥50 cells were counted. The data are presented as an average number of colonies per 1000 seeded cells.

In Vitro Cell Migration Assays
Non-targeted MSC migration was evaluated using the in vitro "scratch" assay [50]. A confluent monolayer was scratched with a sterile pipette tip to create a "wound" approximately 0.8-1.0 mm wide. Then, the culture medium was changed with CM from EC monocultures, diluted with α-MEM (supplemented with 10% FBS) in a 1:1 ratio. All scratch assays were performed in five replicates. To estimate the wound closure, digital images of five randomly selected view fields along the scratch were captured using Nikon Eclipse Ti-U microscope (Nikon, Tokyo, Japan) before and 24 h after the scratch. The images were analyzed using NIS-Elements software (Nikon, Tokyo, Japan) to measure the square of the scratch at marked points. The migration area was calculated as the difference between the initial and final wound squares (um 2 ).

In Vitro Assay to Measure PBMC Adhesion to Activated Human ECs
PBMCs (10 6 cells/mL) in complete RPMI 1640 medium were added to monocultures of quiescent or TNF-α-activated ECs, as well as to co-cultures of MSCs and TNF-α-activated ECs. These co-cultures were exposed under "physiological" hypoxia and acute hypoxic stress for 30 min. Non-adherent PBMCs were removed, the cells were washed twice for 5 min with PBS, and stained with Giemsa azure-eosin 0.02% aqueous solution (Sigma-Aldrich, St. Louis, MO, USA) for 20 min at room temperature. Stained cells were washed twice with PBS followed by deionized water, air-dried, and photographed using a Nikon Eclipse Ti-U microscope. The number of adhered PBMCs was counted in at least 10 randomly selected view fields.

Enzyme Linked Immunosorbent Assay (ELISA)
Samples of conditioned medium (CM) from mono-and co-cultures were collected in accordance with the experimental protocols. Deep frozen probes were thawed at room temperature, centrifuged, and an ELISA was performed according to the manufacturer's instructions using kits for determining IL-6 (Human IL-6 ELISA set) and IL-8 (Human IL-8 ELISA set), (BD OptEIA, San Diego, CA, USA). Optical density was measured using a BioRad PR2100 microplate reader (BioRad, Redmond, WA, USA) at a wavelength of 450/620 nm. The concentration of soluble mediators was evaluated using a calibration curve. IL-6 and IL-8 secretion is presented as ng/10 5 cells.

Quantitative RT-PCR Analysis
To evaluate gene expression, total RNA was extracted using the QIAzol reagent (Qiagen, Hilden, Germany) and purified by the phenol/chloroform technique. The quality and concentration of RNA samples were estimated using a Nanodrop ND-2000c (Thermo Fisher Scientific, Waltham, MA, USA). Ambion DNase I (RNase-free) (Thermo Fisher Scientific, Waltham, MA, USA) was used for genomic DNA degradation. Reverse transcription was performed using the MMLV RT Kit (Eurogene, Moscow, Russia) according to the manufacturer's protocol. The cDNA was mixed with qPCR mix-HS SYBR (Eurogene, Moscow, Russia) and added to 96-well plates. Expression of the genes (OCT4, NANOG, SOX2, RUNX2, PPARγ, SOX9, IL6, IL8) was analyzed using Qiagen primers (Qiagen, Hilden, Germany). The expression levels of three housekeeping genes (ACTB, GAPDH, and HPRT) were used as a reference. The polymerase chain reaction was performed using the Mx300P system (Stratagene, San Diego, CA, USA). Normalized gene expression was calculated with the 2 −∆∆ Ct method [51].

Statistical Analysis
All data were calculated from 3-5 independent experiments and expressed as the mean ± standard deviation. Analysis of group differences was performed by the Mann-Whitney U-test for independent samples using STATISTICA 10 software (Statsoft, Tulsa, OK, USA). Statistical significance was considered at p < 0.05.

Stromal Phenotype and Progenitor State of MSCs upon Interaction with ECs
MSCs, as stromal precursors, exhibit a number of activities that determine their relevance in regenerative medicine. There is, first of all, a low-committed state that determines the ability to multilineage differentiation, high proliferative and paracrine potential, and the ability to migrate.
The stromal phenotype of MSCs implies the expression of CD90, CD73, and CD105 surface molecules [52]. Adipose tissue-derived MSCs used in the experiments met the minimal criteria of the International Society for Cellular Therapy: they rapidly attached to plastic, had a fibroblast-like shape, and proliferated for a long time in the culture, differentiated into adipo-and osteo-directions, expressed stromal markers CD90, CD73, and CD105 and did not bear hematopoietic pan-leukocyte CD45 antigen (Figure 2a). It was demonstrated that a high CD90 expression was typical of all MSCs, regardless of the source of their isolation, and correlated with the undifferentiated cell state, while the proportion of CD90+ cells decreased with an increase in commitment [53,54]. In our study, MSCs showed a similar level of stromal marker expression in monoculture and in co-culture with ECs, regardless of the O 2 levels (Figure 2a,b).
After 24 h of co-culture, specific patterns of MSC and EC distribution were observed. The ECs were organized in clusters surrounded by MSCs. The modes of MSC and EC distribution were similar under both O 2 levels (Figure 3a).
Short-term interaction with ECs also had no effect on the number of CFU-f among MSCs (Figure 3b), which may indicate their preserved proliferative ability. (c) Heatmap analysis of the expression levels of "stemness" genes and "commitment" master-genes in MSCs. Genes with higher expression are depicted in red, genes with lower expression are depicted in green. The data are presented as fold changes of transcription levels of co-cultured MSCs vs. monocultured MSCs, n = 3. p-values indicate the significance of differences between MSCs in monocultures vs. MSCs from co-cultures with ECs. MSCs in monocultures were exposed to the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly). To assess the potential of self-maintenance/differentiation, the expression of genes responsible for the "stemness" properties (OCT4, SOX2, NANOG) and the expression of genes that regulate the osteogenic (RUNX2), adipogenic (PPARγ), and chondrogenic (SOX9) differentiation were analyzed in MSCs (Figure 3c).
After co-culture with ECs at 5% O 2 , the level of NANOG was 2-fold downregulated, and OCT4 was 1,6-fold upregulated. The transcription of osteodifferentiation master gene RUNX2 was more than 2-fold elevated, while chondrodifferentiation master gene SOX9almost 2-fold decreased. Under hypoxic stress (0.1% O 2 ), interaction with ECs had no significant effect on the transcription of "stemness" OCT4 and NANOG, the transcription of SOX2 was significantly reduced. SOX9 downregulation was more pronounced, while the expression of RUNX2 had no significant changes. Therefore, the profile of genes whose products are involved in MSC self-maintenance/differentiation changed differentially after short-term contact with ECs under hypoxic stress or "physiological" hypoxia.

Expression of Cell-to-Cell and Cell-Matrix Adhesion Molecules
After a short-term interaction with ECs, the proportions of MSCs expressing integrin α1 and integrin αVβ3 were increased, whereas the percentages of MSCs carrying integrin α4 and N-cadherin were decreased both at 5% and 0.1% O 2 (Figure 4a).
These changes were accompanied by corresponding alterations in the number of adhesion molecules calculated as mean fluorescence intensity (MFI) (Figure 4b). MFI of integrin α1 and integrin αVβ3 were elevated, while integrin α4 and N-cadherin were stained less intensively. The interaction had no effect on the proportions and MFI of ICAM-1-positive MSCs. Upon short-term hypoxic stress, the MSC integrin α5 MFI increased significantly in both monoculture and co-culture (p = 0.047) (Figure 4b). MSCs in monocultures were exposed at the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly).
Thus, the interaction with ECs at different O 2 (5%, 0.1% O 2 ) resulted in a changed expression of MSC adhesion molecules involved in the formation of hetero-and homotypic intercellular contacts as well as contacts with the extracellular matrix.  p-values indicate the significance of differences between MSCs in monocultures vs. MSCs from co-cultures with ECs. MSCs in monocultures were exposed to the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly).

Interleukin's Profiles upon EC-MSC Interactions
After short-term hypoxic exposures (5% O 2 vs. 0.1% O 2 ), the levels of IL-6 or IL-8 were quite close in MSC monocultures. Similar effects were observed in EC monocultures (Figure 5a).
At "physiological" hypoxia, The MSC-EC interactions resulted in a substantial increase in IL-6 and IL-8, significantly exceeding the levels of these cytokines in MSC and EC monocultures, accordingly. At 0.1% O 2 , the concentration of IL-6 after the MSC-EC cocultures was notably lower, while the IL-8 level was remarkably higher compared to that at 5% O 2 (Figure 5a). RT-PCR data were consistent with the results of the ELISA. At 5% O 2 , IL6 and IL8 were upregulated in ECs and MSCs after cocultivation. Under hypoxic stress, this effect was less pronounced for IL6 in contrast to IL8 which was significantly upregulated in both cell types (Figure 5b). p-values indicate the significance of differences between MSCs in monocultures vs. MSCs from co-cultures with ECs. MSCs in monocultures were exposed to the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly).

EC Paracrine Influence on MSC Migratory Potential
The paracrine effects of ECs on MSC migration were assessed in non-targeted ("scratch") ( Figure 6a) and targeted (Transwell ® culture plates) (Figure 6b) assays.
After "scratch", the motility of MSCs in MSC culture medium or EC conditioned medium was similar at 5% and 0.1% O 2 (Figure 6c). Meanwhile, the number of migrating MSCs in Transwell ® culture plates was significantly higher in EC conditioned medium under both O 2 conditions. EC paracrine regulation significantly stimulated targeted migration of MSCs enhancing its response to chemotactic stimuli (Figure 6d).

Effect of MSCs on Inflammatory-Activated ECs
The above data have shown that a short-term interaction with EC led to an alteration in the functional activity of MSCs. It is reasonable to suppose that during the contact, MSCs can affect the EC properties in physiological and pathological conditions. Inflammatory mediators in the sites of injury are known to provoke endothelial dysfunction negatively affecting regenerative processes [55][56][57][58]. Increased adhesion of circulating immune cells to inflammatory-activated ECs and elevated oxidative stress in ECs are the most marked manifestations of endothelial dysfunction [59][60][61][62].
We analyzed the effects of MSCs on quiescent and TNF-α activated ECs at different O 2 levels.
Both quiescent and TNF-α primed ECs in monoculture expressed a high level of endothelial markers PECAM and VE-cadherin (Figure 7a). After interaction with MSCs, no changes were detected in the surface presentation of the above adhesion molecules (Figure 7b,c). p-values indicate the significance of differences between ECs in monocultures vs. ECs from co-cultures with MSCs. ECs in monocultures were exposed to the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly).
The expression of molecules involved in the adhesion of immune cells to the ECs VCAM-1 (CD106), E-selectin (CD62E), ICAM-1 (CD54)), and integrin α1 (CD49a) were analyzed in EC monocultures. As a result of TNF-α activation, MFI of ICAM-1 and integrin α1 expression was increased. Foreseen elevation of these molecules was retained in coculture with MSCs at both O 2 levels.
The increased adhesion of PBMCs to inflammatory-activated ECs is one of the marked manifestations of endothelial dysfunction. To find out how MSCs can govern this process, we examined the attachment of PBMCs to quiescent and TNF-α primed ECs (Figure 8a). The efficacy of PBMCs' adhesion to inflammatory-activated ECs in monoculture and co-culture with MSCs. Column chart. The data are presented as the number of adhered PBMCs per mm 2 , n = 3. Dotted line-the number of PBMCs adhered to quiescent ECs. p-values indicate the significance of differences between the number of PBMCs adhered to TNF-α-activated ECs in monocultures or co-cultures with MSCs. ECs in monocultures were exposed to the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly).
The surface of quiescent ECs was non-adhesive for PBMCs both in monoculture and in co-culture with MSCs (Figure 8b, dotted line). TNF-α primed ECs in monoculture supported a significantly higher level of PBMC adhesion at both O 2 levels (Figure 8b). Compared to "physiological" hypoxia, acute hypoxic stress potentiated the adhesion of immune cells to TNF-α ECs (Figure 8b). In the presence of MSCs, a substantial reduction in the adhesion of PBMCs was observed. The effect was more pronounced at 5% O 2 : 3-fold elevation vs. 2.3-fold at 0.1% O 2 . These results may indicate the negative effect of hypoxic stress on the MSC's ability to support the EC barrier function (Figure 8b).
Noteworthy, the decreased number of adhered PBMCs in EC-MSC co-cultures was not associated with a change in the expression of adhesion molecules integrin α1, VE-cadherin, ICAM-1, VCAM-1, and E-selectin on ECs (Figure 7). This suggests the existence of some additional mechanisms that reduce TNF-α EC-adhesivity in the presence of MSCs.
In addition to the adhesion molecules, NO synthase that regulates the production of nitric oxide (NO) plays an important role in the EC physiological maintenance. The activity of NO-synthase may be changed under stress conditions, which will be reflected on the level of its metabolite-NO [63]. NO levels in quiescent and TNF-α ECs in monocultures and co-cultures with MSCs were determined using flow cytometry (Figure 9a). (b) Differential expression of NO synthase gene. The data are presented as fold changes vs. monocultured ECs. Column chart, n = 3. p values indicate the significance of differences between ECs or TNF-α-activated ECs in monocultures and in co-cultures with MSCs. ECs or TNF-α-activated ECs in monocultures were exposed to the same O 2 levels as co-cultures (at 5% or 0.1%, accordingly).
Under hypoxic stress (0.1% O 2 ), NO level in monocultured ECs was 6-7 times higher vs. 5% O 2 . Inflammatory activation resulted in a slight but significant increase in NO in ECs at 5% O 2 but had no effect at 0.1% O 2 . In the co-culture with MSCs, NO levels were partially decreased both in quiescent and TNF-α ECs under hypoxic stress only (Figure 9a).
Using RT-PCR, the transcription of NOS3 (endothelial NO synthase) in ECs before and after interaction with MSCs was characterized. NOS3 expression did not differ in quiescent and TNF-α ECs in monoculture and co-culture with MSCs. Hypoxic stress (0.1% O 2 ) provoked substantial upregulation of NOS3 in ECs vs. 5% O 2 . In the presence of MSCs, a partial downregulation of NOS3 under hypoxic stress was observed. These data are in good agreement with the cytometry data on NO levels (Figure 9b). Therefore, during short-term interaction under "physiological" hypoxia, MSCs attenuated the ability of inflammatory-activated ECs to support PBMC adhesion. Under hypoxic stress, this effect was attenuated, which may be associated with an increased NO production and inhibition of NO-synthase transcription in ECs.

Discussion
The aim of this study was to determine how the functional activity of MSCs changes after a short-term interaction with ECs under "physiological" hypoxia (tissue-related O 2 ) and under a critical O 2 reduction (acute hypoxic stress).
In addition, we were interested in how MSCs can affect the properties of inflammatoryactivated ECs at different levels of hypoxia. For these goals, we used the model of shortterm EC-MSC contact co-cultures at 5% and 0.1% O 2 , respectively.
Firstly, we will consider changes in MSCs after interaction with ECs under "physiological" hypoxia.
The progenitor state of MSCs before and after interaction with ECs was characterized by expression of stromal markers, the number of CFU-f, transcription of differentiation master genes (osteogenic-RUNX2, adipogenic-PPARγ, and chondrogenic-SOX9), as well as of the "stemness" genes (OCT4, NANOG, SOX2). The short-term contact with ECs was found to be sufficient to induce changes in MSCs at the transcriptional level. The observed SOX9 downregulation and RUNX2 upregulation may be related to the reciprocal regulation of these transcription factors during osteodifferentiation [64] and indicate the osteogenic commitment of MSCs. Meanwhile, earlier it was demonstrated that the upregulation of differentiation-related genes in MSCs after contact with ECs was not sufficient for initiation of functionally competent progeny [17,65]. Moreover, even after a long-term interaction with ECs, MSCs retained multilineage potential despite the changes in the transcriptomic profile [17]. Our data on the preserved stromal phenotype and clonally active stromal progenitor pool under tissue-related O 2 clearly support the above assumptions.
It is well established that the migration, proliferation, and differentiation of MSCs are regulated by communications with various cell types and extracellular matrix through intercellular adhesion molecules and integrin receptors [66,67]. In our study, the expression of homotypic intercellular contact protein (N-cadherin) and integrin α4 on MSCs was decreased after interaction with ECs. Simultaneously, the expression of proteins involved in migration and the formation of heterocellular MSC-EC contacts (integrin α1, integrin αVβ3) was increased.
Cadherin-cadherin contacts are known to provide a connection between the cytoskeleton components of neighboring cells [68][69][70][71]. A high level of N-cadherin expression is assumed to be characteristic of low committed stromal progenitors [72][73][74]. A decrease of N-cadherin expression was detected in osteodifferentiated MSCs [75,76]. Attenuation of Ncadherin expression observed here can be interpreted as a sign of MSC commitment, which is consistent with our data on downregulation of chondroregulator SOX9 and upregulation of osteoregulator RUNX2.
Integrin α1 is involved in the MSC adhesion to ECs [77]. Its elevation in our experiments confirms the formation of heterocellular contacts during an interaction. Along with other receptors, integrin αVβ3 mediates homing and migration of stromal cells [77]. Elevation of integrin αVβ3 expression in myofibroblasts was observed as a consequence of the increase in fibronectin synthesis [78]. In our case, this path of integrin αVβ3 regulation apparently can be involved, since both ECs and MSCs produce fibronectin. On the whole, alteration of adhesion molecule profile after interaction with ECs may indicate the formation of the migratory phenotype of MSCs. This assumption is supported by our findings on the enhanced targeted migration of MSCs in vitro.
MSCs are able to secrete a number of soluble mediators that are important for reparative process. Transcriptional changes of certain cytokines and growth factors in MSCs following interaction with ECs have been described earlier [79,80]. For the first time, in the present study, we not only analyzed the amount of IL-6 and IL-8 in a conditioned medium but estimated the transcriptional activity of these interleukins in both cell types. We managed to demonstrate that the significant increase in IL-6 and IL-8 after co-culture can be a result of the up-regulation of relevant genes in ECs and MSCs. As a potential explanation of the above effects, the VEGF production by MSCs can be considered. VEGF has been shown to upregulate IL6 and IL8 in ECs through binding to the VEGFR2 receptor and activation of interleukin gene' transcription via protein kinase D1 [81]. On the other hand, ECs can synthesize IL-1α, and its secretion was shown to increase under hypoxia [82]. IL-1α possesses IL-1β-like activity resulted in elevated production of IL-6 and IL-8 [83]. Based on the above, it can be supposed that VEGF and IL-1α-mediated signaling pathways-favored upregulation of IL6 and IL8 genes in communicated MSCs and ECs supporting increased production of these interleukins. Therefore, at "physiological" hypoxia, interaction with ECs can positively modulate the MSC activities requested in cell therapy. The "stemness"/differentiation gene shift, which consists of the increased activity of the latter, was found in MSCs. Changes in the expression of surface molecules responsible for MSC migration and homo-and hetero-cellular contacts, as well as an increase in the transcription/translation of the main proinflammatory cytokines IL-6 and IL-8, may facilitate MSC migration to target tissues.
Another issue raised in this study was how a critical decrease in O 2 , as a component of ischemia in tissue damage, can affect the EC-MSC interaction.
The effects of ECs on MSCs under hypoxic stress (0.1% O 2 , 24 h) were similar to those at "physiological" hypoxia (5% O 2 ). MSCs retained their stromal phenotype and clonogenic activity and showed signs of commitment at the transcriptional level. Under hypoxic stress, "physiological" hypoxia-induced upregulation of RUNX2 was canceled and an additional SOX2 downregulation occurred in MSCs. Co-culture caused similar changes in the expression of adhesion molecules, interleukins' production, and MSC migratory activity. At the same time, hypoxic stress partially eliminated the increase in IL-6 production and stimulated a significant increase in IL-8 in the MSC-EC co-cultures. Similar patterns of IL-6 and IL-8 expression were described in co-cultures of MSC and hematopoietic cells at 1% O 2 [84]. Although the mechanisms of hypoxic effects on the paracrine activity of MSCs in co-culture with other cells are not fully understood, the involvement of the HIF, Notch, Wnt/β-catenin, and Hedgehog signaling pathways is assumed [84,85]. As mentioned above, VEGF and IL-1α may be involved in the regulation of cytokine production in MSCs and ECs. Considering the stimulating effect of hypoxia on the production of IL-1α in ECs [82], its role in the induction of IL-8 expression under hypoxic stress may be supposed. Integrin α5 elevation was observed in MSCs in monoculture as well as in co-culture with ECs, which may indicate a potentiating effect of oxygen deprivation on the expression of this molecule, as suggested by Saller et al. [86] and Choi et al. [87].
It is obvious that upon MSC-EC interaction the EC activity could be altered as well. In this study, we evaluated the effect of MSCs on functions of inflammatory-activated ECs.
In the body, ECs form a barrier-the main role of which is the regulation of permeability of the vascular wall for different substances and cells including immune ones [61]. The surface of quiescent ECs is non-adhesive for PBMCs. Increased endothelial permeability as a result of inflammatory activation is an important mechanism in the response to damage. Various inflammatory cytokines (TNF-α, IL-1β, IFN-γ) activate ECs provoking the expression of adhesion molecules mediating further interaction with PBMCs [14,55,88]. Activated ECs do not restrict but enhance transendothelial migration of immune cells into the tissues, which indicates an important role of EC activation in the immune response [58]. All the above factors, along with the inflammatory induction of oxidative stress in ECs, interfere with successful tissue regeneration [56,57].
In animal models of inflammatory diseases, a decreased systemic inflammatory response was found after MSC administration, but the exact mechanisms remain unknown [89][90][91]. Apparently, MSCs may be involved in the regulation of the endothelial monolayer permeability [92]. To understand the possible mechanisms of MSC effects on EC activation under "physiological" hypoxia (5% O 2 ) and acute O 2 deprivation (0.1%), we analyzed the effect of co-culture with MSCs on the expression of adhesion molecules, PBMCs' adhesion, and on the oxidative stress level (NO production and NOS3 activity) in ECs.
Under "physiological" hypoxia in monoculture, inflammatory activation of ECs resulted in integrin α1 and ICAM-1 elevation, increased NO production, and enhanced ability to adhere PBMCs. Interaction with MSCs had no effect on the expression of adhesion molecules and partially eliminated the manifestations of oxidative stress in ECs. Meanwhile, a significant decrease in the adhesiveness of activated ECs for PBMCs was observed, which confirms the ability of MSCs to attenuate the inflammatory state of ECs. Actually, in the animal hemorrhagic shock model, the therapeutic effect of MSCs was shown to reduce leukocyte infiltration [61]. A decreased EC ability to attract and support the adhesion of immune cells was demonstrated in the presence of MSCs in vitro, which authors associated with MSC-produced cytokines, IL-6 and TGFβ [79,93]. As shown in the present study, EC-MSC interaction under "physiological" hypoxia was followed by a multifold increase in IL-6 level in the conditioned medium, which was associated with an increase in IL-6 transcription in both cell types. Apparently, this had a positive effect on the reduction of EC adhesiveness for PBMCs. It can be concluded that under "physiological" hypoxia, MSCs had a protective effect on activated ECs, thus reducing the manifestation of inflammatory activation.
Under hypoxic stress, the protective effects of MSCs were less pronounced. The above effect may be related to the fact that under hypoxic conditions (1% O 2 ), ECs produce the HAL1/13 adhesion molecule, which contributes to an increase in their adhesiveness for PBMCs [94], and also are characterized by ICAM-1 elevation [95]. Apparently, a higher basal expression of ICAM-1 and other molecules involved in the EC-PBMC interaction under hypoxic stress may cause the attenuated MSC effects on the adhesiveness of ECs.
Therefore, as a result of short-term interaction with ECs under "physiological" hypoxia, the functional activity of MSCs was modified, which may be relevant for the process of reparative tissue remodeling. In the presence of MSCs, signs of EC activation, such as NO production and increased PBMC adhesion, are partially eliminated. Compared to "physiological" hypoxia, acute hypoxic stress may attenuate the stimulating effect of ECs on MSCs as well as partially cancel the "protective" effect of MSCs on activated ECs.  Data Availability Statement: Data are available to colleagues upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

MSCs
Mesenchymal stem cells IFATS International Federation for Adipose Therapeutics and Science ISCT International Society for Cell and Gene Therapy CD Cluster of differentiation CM Conditioned medium ECs Endothelial cells FBS Fetal bovine serum PBMC Peripheral blood mononuclear cells