Evidence of mesenchymal stromal cell adaptation to local microenvironment following subcutaneous transplantation

Abstract Subcutaneous transplantation of mesenchymal stromal cells (MSC) emerged as an alternative to intravenous administration because it avoids the pulmonary embolism and prolongs post‐transplantation lifetime. The goal of this study was to investigate the mechanisms by which these cells could affect remote organs. To this aim, murine bone marrow–derived MSC were subcutaneously transplanted in different anatomical regions and the survival and behaviour have been followed. The results showed that upon subcutaneous transplantation in mice, MSC formed multicellular aggregates and did not migrate significantly from the site of injection. Our data suggest an important role of hypoxia‐inducible signalling pathways in stimulating local angiogenesis and the ensuing modulation of the kinetics of circulating cytokines with putative protective effects at distant sites. These data expand the current understanding of cell behaviour after subcutaneous transplantation and contribute to the development of a non‐invasive cell‐based therapy for distant organ protection.

transplantation, the majority of administered cells are trapped in the lung capillaries. 13 Nonetheless, iv infusion of MSC was reported to reduce the inflammatory response and promote tissue repair 14,15 in many experimental settings, which indicated an important role of the secretome (MSC-secreted molecules) in modulating the innate and adaptive immune responses. 6,14,16 Based on the reported therapeutic effects of the MSC secretome, we and others have proposed the subcutaneous transplantation procedure as an alternative to iv administration of MSC, the benefit of which is to overcome the risk of pulmonary embolism and prolong the lifetime of cells post-transplantation. [17][18][19][20] Here, we provide evidence that after subcutaneous transplantation, MSC shape into multicellular aggregates that activate hypoxia signalling pathways and the ensuing local angiogenesis. This is followed by the transient modulation of a large panel of circulating cytokines with putative protective effects at distant sites. These data sustain the existence of a blood-borne-mediated pathway activated by MSC after subcutaneous transplantation, with no need of homing to the site of injury.

| Animals
All animal experiments were conducted in accordance with the

| Isolation and characterization of MSC
The cells were isolated from mouse bone marrow as previously described. 4 Briefly, bone marrow was obtained from male C57BL/6 mice of 6-8 weeks of age by flushing the medullary cavity of femurs and tibias with complete medium, consisting in low-glucose DMEM, supplemented with 10% MSC-qualified FBS and 1% antibiotic-antimycotic (all reagents were purchased from Thermo Fisher Scientific). Then, the cell suspension was passed through needles of decreasing size from 18 to 25 gauge to obtain a single cell suspension. Collected cells were centrifuged at 400 g for 5 minutes, resuspended in complete medium and seeded at 10 6 cells/cm 2 . At 24 hours, the non-adherent cells were removed by changing the medium. After 1 week, the cells were detached with 0.25% trypsin and gently scraped with a rubber policeman, followed by seeding at a density of 5000 cells/cm 2 in complete medium. The next 5-6 passages were done at 90% confluency, until the culture was totally free of CD45 + cells (starting at passage no 7).
The presence of MSC characteristic markers (Sca-1, CD105, CD44), the absence of haematopoietic markers CD45 and CD11b, and the in vitro differentiation potential of cells into osteogenic, adipogenic and chondrogenic lineages were evaluated to confirm the MSC attributes. 4 These attributes were retained for at least 10 passages after completing the selection process. 21 Cells were used between the 8th and 13th passages. The 3D aggregates were obtained by assembling various number of cells (from 10 4 to 3 × 10 5 ) for 3 days using the hanging-drop method as previously described. 22 The aggregate diameter was determined under a Nikon Eclipse Ti-E inverted microscope using a Ds-Fi1 camera (Nikon) and NIS-Elements AR 3.0 software.
For in vivo imaging of hypoxia, the cells were transfected with HRE-luciferase plasmid (Addgene # 26731, a gift from Navdeep Chandel) 23

| MSC transplantation procedure
Mice were anaesthetized with a mixture of ketamine-xylazine (100-5 mg/kg bodyweight), and the hair around the site of injection was removed with an electrical clipper. To elect the most favourable place for subcutaneous MSC injection, 50 µL of cell suspension (containing 10 6 or 3.5 × 10 6 cells) was slowly injected subcutaneously in different anatomical regions (interscapular, inguinal and abdominal).
This way, the cell spreading was avoided and the formation of 3D aggregate was certified by the presence of the subcutaneous swelling after transplantation.

| Assessment of serum cytokines after MSC subcutaneous transplantation
The serum profile of sham-and MSC-treated mice was analysed using Analyser, and the pixel density was quantified by TotalLab CLIQS software. Results were analysed using the online software Morpheus (https://softw are.broad insti tute.org/morpheus), in order to compact bulky information into clusters and find patterns in the data.

| In vitro assays for visualization of hypoxia
Cells were cultured in hanging drops in culture medium supplemented with 100 nmol/L HypoxiSense 680 (HS680) from PerkinElmer.
Various sizes of cellular aggregates were obtained by seeding different cell numbers (10 3 , 5 × 10 3 , 10 4 , 2 × 10 4 and 5 × 10 4 ) per hanging drop. After 3 days, cellular aggregates were briefly washed three times with PBS and imaged with IVIS Spectrum system and Living  To determine the local inflammatory response in vivo after MSC transplantation, animals were injected i.p. with lucigenin (12.5 mg/kg body weight) 10 minutes before imaging, as previously described. 25 Surface images were analysed using Living Image 4.5 software (PerkinElmer) and quantification of bioluminescence was done by manually defining the regions of interest. The bioluminescent signal was calculated as average radiance.

| Statistics
The results were expressed as the mean ± SEM (standard error of the mean). Statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, Inc). Comparisons between groups were done using unpaired t tests with two-tailed distribution or two-way ANOVA using a Bonferroni post hoc test where appropriate. Difference was considered statistically significant when P-value was <.05.

| Establishing the most favourable transplantation site and dose for the MSC graft stability
The first issue addressed was to select the site of MSC subcutaneous transplantation that offers the maximum benefits, in terms of local cell survival and engraftment. To this purpose, CMTPX-labelled MSC were subcutaneously transplanted in three different anatomic regions, each of them receiving different inputs from the nearby adipose tissue: interscapular (in the proximity of brown adipose tissue), inguinal (in the proximity of white adipose tissue) and abdominal (with no adipose tissue nearby) ( Figure S1a). The animals were imaged at 30 minutes and at 7 days after cell transplantation. The fluorescent signal indicated no significant differences between the three groups ( Figure S1b); moreover, the intensity determined at 7 days after transplantation was with one order of magnitude lower than that of the signal quantified 30 minutes after the procedure.
The data indicated that similar number of cells was engrafted and survived locally when transplanted in either region, and there was no influence of the nearby adipose tissue.
To evaluate whether and to which extent MSC migrated from the site of transplantation, analysis of the major organs involved in cell migration was performed at the time of harvest by ex vivo fluorescence imaging. Of the analysed organs (spleen, lymph nodes, adipose tissue, lung, liver and heart), only local lymph nodes and adjacent adipose tissue showed detectable, although very low, fluorescent signals in all groups, which were around two orders of magnitude lower than that of the locally transplanted cells ( Figure S1c). For instance, at 7 days after subcutaneous transplantation of MSC in the interscapular region, 4.4 ± 2.4% and 0.8 ± 0.4% of the graft signal were detected in the interscapular adipose tissue and lymph nodes, respectively, suggesting that MSC did not significantly migrate from the injection site ( Figure 1A-C). Based on these data, the interscapular site was selected for the subsequent experiments of MSC transplantation, a route that was also used in our previous work. 17 The next issue we addressed was the optimal cell number to be this was followed by a gradual reduction of the signal ( Figure S2). It is worth mentioning that the total BLI signal is influenced by multiple factors, such as luciferin penetrance into the aggregate, cellular status, oxygen availability, cell proliferation and cell death. However, the decrease in survival sloped faster in high-dose cell aggregate than in low-dose cell aggregate ( Figure 1D and S2); at 7 days after transplantation, the cell survival rates were 57% ± 12% in low dose and 17% ± 6% in high dose. In terms of estimated cell number, the two aggregates contained the same number of viable cells at 7 days after transplantation, as the bioluminescent signals from the two groups were similar, irrespective of the initial cell dose ( Figure 1D).
These results demonstrated that the cell viability is a constraining factor in establishing the transplantation dose, being likely governed by the competition for microenvironmental nutrients. Moreover, low-dose cell aggregates showed a greater vessel ingrowth at day 7 ( Figure 1E), indicating that cell survival depends on the presence of a stable host-derived vascular network to support the biological functions of the grafted cells.
To further assess whether the grafted cells elicit local inflammation, animals were injected with lucigenin (12.5 mg/kg body weight, i.p.) at 7 days after transplantation and the presence of macrophages within the grafts was assessed based on the bioluminescence signal produced by the direct interaction between phagocyte NADPH oxidase and lucigenin. The results revealed that the bioluminescent signal produced in low-dose aggregates was close to background level, which indicated low inflammation associated with the MSC grafts; however, an increased local inflammatory bioluminescent signal was noted in high-dose cell aggregates ( Figure 1F). Together, the above data revealed that the transplantation dose, but not the transplantation site, is a limiting factor of graft stability. The subcutaneously transplanted MSC are not affected by the local adipose tissue and do not considerably migrate from the injection site, and the graft survival is enhanced when using low number of cells, which facilitates the growth of vascular network and the ensuing access of nutrients, while limiting local inflammation. Our data showed that, in contrast to U87-MG tumour cell line, which reportedly exhibit a strong constitutive expression of CA9 protein, 27 MSC did not bind HS680 in normoxic 2D culture conditions ( Figure 2A). In contrast, HS680 specifically and strongly bound to MSC aggregates (formed by hanging-drop assay) at a comparable level to U87-MG cells ( Figure 2B). Besides, CA9 protein level was significantly increased in 2D-cultured MSC exposed to hypoxic conditions ( Figure S3), thus confirming the capacity of HS680 to label hypoxic MSC.

| Hypoxia is activated in MSC aggregates both in vitro and in vivo
Assembling of different numbers of MSC in the hanging-drop assay in the presence of HS680 revealed a linear correlation between the cell number and the aggregate diameter size (R 2 = 0.9825).
Yet, a monotonic correlation was detected between the cell number and hypoxia level, namely higher hypoxia signals in larger aggregates ( Figure 2C). These data sustain the existence of a threshold value for the aggregate size above which the hypoxia signalling is activated within the aggregate, an observation that correlates well with other reports. 28,29 Thus, aggregates formed by less than 10 5 cells exhibit only mild levels of hypoxia, while in aggregates formed by more than 2 × 10 5 cells, the increase in hypoxia signal rate was significantly higher.
These results were further investigated by in vivo experiments, in which mice were transplanted with either a high dose (3.5 × 10 6 ) or a low dose (1 × 10 6 ) of MSC ( Figure 2D). After 2 and 7 days, the animals were iv injected with HS680 and the fluorescent signal of the aggregates was assessed, first in vivo, and then ex vivo, by imaging the inner face of the skin containing the aggregate. While the signal could not be detected in living animals (data not shown), specific HS680 signal was detected in cell aggregates ex vivo, at 7 days after transplantation ( Figure 2D). Although both doses resulted in hypoxia activation, higher signals were measured in high-dose cell aggregates as compared to low-dose cell aggregates. Importantly, no specific HS680 signal was detected at 2 days after transplantation, either in vivo or ex vivo (data not shown), suggesting that HS680 could only detect prolonged and/or severe hypoxia in MSC aggregates.

| Time-course activation of hypoxia signalling pathways in vivo
Hypoxia was assessed in MSC grafts by following the time-course activation of HIF-1α signalling pathway in transplanted cells. To this aim, MSC were transiently transfected with a HRE-luciferase construct before being subcutaneously transplanted. Transfection efficiency, evaluated using pEGFP-N1 vector (Clontech) and flow cytometry analysis, was more than 60% at 24 hours post-transfection ( Figure S4). In vivo bioluminescence analysis of transplanted mice showed that hypoxia signalling was activated 1 day after cell transplantation and persisted for at least four days ( Figure 3A), thus suggesting a time window for hypoxia signalling in MSC of several days following subcutaneous transplantation.
To assess whether hypoxia activation is an intrinsic mechanism and not a high-dose-induced effect, the same cell number The cytokines with first-day peak level are highlighted in magenta, while cytokines with third-day peak level are highlighted in violet. B, Graphs illustrating the highly expressed cytokines clustered according to their max peak expression in the serum. At right, the "low expression level" denotes the cytokines whose maximum level of expression at any time-point was higher than the half but below two thirds of the median value of all spots (highest pixel density between 2000 and 2500). The y-axis on the left represents the relative expression level as fold change as compared to day 0. The colour-coded right y-axis shows the abundance of each cytokine in the serum of mice before transplant (day 0) (10 6 cells) was administered as five distinct injections forming rosette-shaped aggregates, so that each of the five cell aggregates consisted of one fifth of the total cell number. The results were very much the same as those with single injections, with activation of HIF-1α signalling pathway starting on day 1 post-transplant and maintained for 4 days ( Figure 3B). As hypoxia-induced HIF activation is reportedly followed by miR-210 induction, as a common feature of the hypoxic response in cancer and normal cells, 24,30 the promoter activation of miR-210 was also evaluated in MSC aggregates in vivo. By using a similar luciferase assay, the results showed the activation of miR-210 between days 1 and 4 after subcutaneous transplantation of MSC ( Figure S5). Together, these data suggested that hypoxia signalling is important in grafted cell aggregates for vascular network infiltration and graft stabilization. and G-CSF, had peak levels (higher than 4.5-fold change) at 3 days after transplant ( Figure 4B). Among these cytokines, IL-1ra and IL-4 have also been previously reported as being highly secreted by MSC in culture, in both normoxic and hypoxic conditions. 4 Moreover, IL-1ra was also identified in the secretome of MSC, but not of dermal fibroblasts, aggregated in hanging-drop culture, which thus qualified it as a putative MSC-derived cytokine responsible for the effector properties in vivo ( Figure S6). It is also worth mentioning that pentraxin-3, previously identified as increased in MSC aggregates in vivo and in the serum of transplanted mice, 17 was identified as having a threefold increase in serum collected at 1 and 3 days after transplantation ( Figure 4A). These results strengthen the hypothesis of MSC-released circulating mediators to induce tissue regeneration in remote diseased organs. More experiments are needed to find the most important biomolecules involved in this process.

| D ISCUSS I ON
The results of our experiments revealed that upon subcutaneous transplantation, MSC do not significantly migrate from the graft and organize into multicellular aggregates. These aggregates activate hypoxia-inducible signalling pathways, which in turn stimulate local angiogenesis and a broad release of therapeutic biomolecules into the circulation, a mechanism that can contribute to their therapeutic effects on remote organs. Taking advantage of the ability of MSC to secrete molecules that operate in an endocrine-like manner, we put forward the subcutaneous (remote) transplantation as a therapeutic approach that offers several advantages as compared to other routes of cell delivery.
First, the subcutaneous approach is minimally invasive, almost painless, does not require general anaesthesia and does not imply blood loss. Secondly, this approach can be repeated and many doses can be administered periodically with no risk for the patients. Thirdly, the clinical use of the remote therapy may be applied as a stand-alone therapy (alternative to conventional therapies) for patients with increased surgical risk or may represent an adjuvant therapeutic option in combination with conventional therapies for further benefits.
Our experiments bring evidence that subcutaneously transplanted MSC activate hypoxia-inducible signalling pathways and stimulate local angiogenesis, which are mandatory both for graft survival and for the release of protective molecules at distant sites.
It is particularly noteworthy that inflammatory and immune signatures often accompany hypoxia programmes in vivo, 34 with a significant cross talk between transcription factors that respond to either hypoxia or inflammation. 35 Thus, a more prudent interpretation of hypoxia as being the only mediator of this process could be warranted. Nevertheless, considering the low level of inflammation produced by small-dose aggregates and the syngeneic scenario of the transplant procedure (that is associated with no immune responses), we assumed that inflammation had minimal contributions to the local mechanisms activated after MSC transplantation. Still, the increased inflammation associated with high-dose transplantation may significantly impact the outcome of cell therapy, thus emphasizing the importance of accurately determining the number of transplanted cells in order to achieve positive outcomes.
The extensive lines of evidence that MSC produce a broad repertoire of trophic and immunomodulatory cytokines have highlighted the importance of the MSC secretome in the field of stem cell biology. 16,[34][35][36] Although the secretome composition is varying depending on cell type and tissue origin sources, the common feature is the enrichment in molecules that are associated with the cell survival, angiogenesis process and immune regulatory functions. 16

ACK N OWLED G EM ENTS
The authors are grateful to Roxana Vladulescu and Nae Safta for excellent technical assistance. This work was supported by a project co-financed from the European Regional Development

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

R E FE R E N C E S S U PP O RTI N G I N FO R M ATI O N
Additional supporting information may be found online in the Supporting Information section.