Author contributions: J.N.: conception and design (main contributor), collection and assembly of data (main contributor), data analysis and interpretation (main contributor), manuscript writing (main contributor), and final approval of manuscript; H.A.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; J.T., M.P., T.H., R.T., A.H., T.S., S.N., S.I.L., and T.H.: collection and assembly of data, data analysis and interpretation, and manuscript writing; M.K.: administrative support, data interpretation, and manuscript writing; S.L.: conception and design, administrative support, and provision of study material; L.V.: conception and design, financial support, administrative support, data analysis and interpretation, and manuscript writing; P.L.: conception and design, financial support, provision of study material, collection and assembly of data, data analysis and interpretation, and manuscript writing. H.A., J.T., L.V., and P.L. contributed equally to this article.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS November 6, 2012.
The promising clinical effects of mesenchymal stromal/stem cells (MSCs) rely especially on paracrine and nonimmunogenic mechanisms. Delivery routes are essential for the efficacy of cell therapy and systemic delivery by infusion is the obvious goal for many forms of MSC therapy. Lung adhesion of MSCs might, however, be a major obstacle yet to overcome. Current knowledge does not allow us to make sound conclusions whether MSC lung entrapment is harmful or beneficial, and thus we wanted to explore MSC lung adhesion in greater detail. We found a striking difference in the lung clearance rate of systemically infused MSCs derived from two different clinical sources, namely bone marrow (BM-MSCs) and umbilical cord blood (UCB-MSCs). The BM-MSCs and UCB-MSCs used in this study differed in cell size, but our results also indicated other mechanisms behind the lung adherence. A detailed analysis of the cell surface profiles revealed differences in the expression of relevant adhesion molecules. The UCB-MSCs had higher expression levels of α4 integrin (CD49d, VLA-4), α6 integrin (CD49f, VLA-6), and the hepatocyte growth factor receptor (c-Met) and a higher general fucosylation level. Strikingly, the level of CD49d and CD49f expression could be functionally linked with the lung clearance rate. Additionally, we saw a possible link between MSC lung adherence and higher fibronectin expression and we show that the expression of fibronectin increases with MSC culture confluence. Future studies should aim at developing methods of transiently modifying the cell surface structures in order to improve the delivery of therapeutic cells. STEM CELLS2013;31:317–326
Multipotent mesenchymal stromal/stem cells (MSCs) are the subject of intense investigation due to their promising therapeutic potential based on their immunomodulatory properties and their ability to secrete favorable endothelial and epithelial growth factors. By now, MSCs have been explored in novel cell therapeutic regimens to treat different inflammatory conditions, central nervous system injury, cartilage and bone injury, stroke, Crohn's disease, and as immunosuppressants in graft-versus-host disease (GvHD) (reviewed in, e.g., [1, 2]). Optimal cell delivery is crucial for efficacious MSC therapy and the route of cell administration will be indication-dependent. Local transplantation is the primary choice for regenerative tissue treatments. Systemic infusion of MSCs is, however, the only feasible method in many situations, especially in immunosuppressive treatments. A number of studies have explored the in vivo biodistribution in animal models of intravenously (i.v.) introduced MSCs isolated from various sources such as bone marrow (BM), placenta, and adipose tissue [3–11]. All studies reveal, without exception, cell entrapment in the lung as an inevitable consequence of systemic MSC infusion. Entrapment of cells in the lung has been observed despite a xenogeneic or homogenous in vivo setting and interestingly also after i.v. infusions of both MSCs and mononuclear cells in humans [12, 13].
Various reasons for MSC entrapment to the lung capillary endothelium have been presented by others, and both cell size and cell surface adhesion molecules have been presented as causative agents [2, 3, 9, 11, 14, 15]. MSC lung entrapment could thus involve an interaction between available receptors and ligands on the MSC cell surface versus lung endothelia. Cell size also certainly plays a role in lung adhesion due to the small diameter of the pulmonary capillaries, but the fact that also smaller mononuclear cells are extensively entrapped in the lung after systemic infusion, as we and others have shown, highlights the existence of more specific mechanisms behind this phenomenon [12, 16–19].
Current knowledge does not allow us to make definite conclusions whether entrapment of MSCs or any other cell in the lung is (a) harmful, (b) unwanted, or (c) beneficial in novel cell therapy regimens. Additionally, we do not have sufficient knowledge about the cellular effects of lung entrapment. In many pathological inflammatory conditions, such as acute respiratory distress syndrome (ARDS), the lung is the major site of clinical manifestation of excess inflammation. Until we can fully understand the role of the lung in the various indications where MSC therapy will be used, lung entrapment needs to be explored in more detail and taken into consideration in every novel cell therapy based on systemic infusion.
In this study, we explored causatives for the biodistribution behavior of systemically infused MSCs beyond cell size. As cellular models we used MSCs derived from two clinically relevant sources, namely adult human BM and human umbilical cord blood (UCB). We combined data of in vivo biodistribution behavior with thorough analysis of cell surface structures, both proteins and glycan structures. This enabled us to develop a unique setup to identify molecules with a potential role in cell lung entrapment. We found a striking difference in the subacute biodistribution pattern between the BM-MSCs and the UCB-MSCs, with a much faster dislodgement from the lung for the UCB-MSCs. We can show that although the BM-MSCs and UCB-MSCs differ in cell size, the cells have different levels of known adhesion molecules and glycans. We thus identified a crucial target, the cell surface composition, interacting in the clearance of cells from the lungs. Technologies for controlled modifications of the cell surface composition might be future tools to enhance cell targeting.
MATERIALS AND METHODS
BM-MSC: BM was collected from an unaffected bone site in patients who were operated for osteoarthrosis (>50 years of age) or scoliosis (<20 years of age). All patients gave their written informed consent according to the Declaration of Helsinki and the ethical committee of Oulu University Hospital had approved the study protocol. The isolation and establishment of BM-MSCs were done essentially as described previously . The following established BM-MSC lines were used in this study: the animal experiments with lines 283, 303, 320 (derived from donors >50 years of age), the N-glycome profile with line 320, the flow cytometric analysis presented in Figure 4 with lines 320, 425, 412 (derived from donors >50 years of age) and in Figure 5 with lines 397, 381, 384 (from donors of >50 years of age) and lines 358, M2, and 363 (from donors <20 years of age). UCB-MSC: The isolation and establishment of UCB-MSCs were done as described . All donors gave their informed consent and the study protocol was approved by the ethical review board of Helsinki University Central Hospital. The animal experiments were performed with UCB-MSC lines 391P, 454T (6), and 454T (7) and the N-glycome profile with 454T (6). The flow cytometric analysis presented in Figure 4 was done with UCB-MSC lines 454T (6), 609, and 618. The flow cytometric analysis presented in Figure 5 was done with UCB-MSC lines 454T (7), 391P, and 588P. All MSCs were used in passage numbers between 2 and 5 and with comparable confluence and passage number in each individual experiment. The MSCs were characterized as described in the following chapter. The differentiation capacity of the UCB-MSCs into adipogenic, osteogenic, and chondrogenic lineages was also confirmed (data not shown). Cell counting was performed with a Bürker chamber.
Cell Surface Analysis with Flow Cytometry
The MSCs were characterized according to suggested minimal criteria . Additionally, the expression of podocalyxin-like protein (PODXL), a6-integrin CD49f, a4-integrin CD49d, c-Met, CXCR4, CX3CR1, and GD-2 were analyzed according to  and cell surface fibronectin with an antibody recognizing IST-1. Cell surface glycan epitopes were analyzed with glycan-specific antibodies. Additional information about the used antibodies and isotype controls are included in supporting information. Flow cytometric analysis was performed on FACSAria (BD Biosciences, San Diego, CA, www.bdbiosciences.com) with a 488-nm blue laser for phycoerythrin (PE) and fluorescein isothiocyanate (FITC) and a 633-nm red laser for allophycocyanin (APC). Fluorescence was measured using 530/30-nm (FITC), 585/42-nm (PE), and 660/20-nm (APC) band-pass filters. Data were analyzed using FACSDiva version 5.0.2 software (BD Biosciences).
N-Glycan Isolation and Mass Spectrometry
The cells used for the mass spectrometric analysis were detached by gentle scraping from the cell culture vessel and not by trypsinization. Since the size ratio of adherent BM-MSCs and UCBMSCs with the used culture conditions is approximately 3:1 (Fig. 1A), mass spectrometric samples of 1 × 106 BM-MSCs and 3 × 106 UCB-MSCs were prepared in order to obtain similar amounts of glycoprotein material. The cell pellets were frozen immediately after preparation and stored at −70°C before analysis. N-Glycans were detached from cellular glycoproteins by Elizabethkingia meningosepticumN-glycosidase F digestion (Calbiochem, LA Jolla, CA, www.millipore.com/calbiochem) in a reaction volume of 50 μl (1 × 106 cells) or 100 μl (3 × 106 cells). The released asparagine-linked glycans were purified for MALDI-TOF analysis by organic extraction-precipitation and miniaturized solid-phase extractions steps as described . The samples were divided into three fractions; one fraction containing both neutral and acidic N-glycans, and another divided into neutral and acidic fractions. All samples were analyzed in parallel to ensure optimal comparability between the N-glycan profiles.
Radioactive Labeling of Cells
Cells were labeled with 1,100 MBq of 99mTc hydroxymethylpropylene amine oxime (Tc-HMPAO, Ceretec, GE Healthcare, Buckinghamshire, UK, www.gehealthcare.com) for 15 minutes at room temperature. The supernatant was discarded after centrifugation and the cells were resuspended in 4 ml saline. The cell amounts were rechecked after labeling and cell amount-specific standard curves were prepared to establish amounts of radioactivity per cell. The labeling efficiency and specific radioactivity of each labeled cell batch were established by gamma counting (Wallac Wizard 1480, Perkin Elmer, Gaithersburg, MD, www.perkinelmer.com) of the cell standard curves. Measured radioactivity was normalized by taking label half-life into consideration.
Iron Nanoparticle Labeling of Cells
The labeling of MSCs was performed as described in . Briefly, the MSCs were labeled overnight by adding 0.2 mg Fe/ml (3.9 mg) ferucarbotran (Resqvist, Schering, Berlin, Germany, www.bayer-pharma.com) into the culture medium. After incubation, the cells were washed three times with phosphate buffered saline (PBS), detached from the culture vessels, and resuspended in saline.
All animal experiments were approved and authorized by the National Animal Experiment Board in Finland (Eläinkoelautakunta, ELLA). Ninety-five female immunodeficient Athymic Foxn 1 nude mice aged 7–8 weeks were used. Anesthesia was induced with a subcutaneous injection of 100 μl/10 g solution mixture of one part of Hypnorm (fentanylsitrate 0.315 mg/ml and fluanisoni 10 mg/ml), one part Dormicum (midatsolam 5 mg/ml), and two parts of water. All animals were injected i.v. with 500,000 cells in 100 μl saline in the tail vein. The mice were sacrificed by CO2 after 1, 12, or 24 hours.
The animals were sacrificed before imaging and imaged with a Siemens Orbiter gamma camera (Siemens Gammasonics Inc., Des Plaines, IL, www.siemens.com) equipped with a pin-hole collimator. Images were acquired at 1 or 12–13 hours after injections with Tc-HMPAO-labeled cells. An acquisition time of 10 minutes or 25 minutes was used, respectively. The matrix size was 128 × 128. The images were evaluated semiquantitatively by calculating counts in regions-of-interests drawn over lungs, liver, tail, and the whole animal.
Quantitative Measurement of Radioactivity from Tissues
Lungs, liver, heart, spleen, kidneys, bladder, bones (one femur per animal), BM (out-flushed in 1 ml saline from the other femur per animal), brain, subcutaneous adipose tissue, and intestines (GI) were dissected from the sacrificed animals and placed into scintillation tubes. Radioactivity was measured using a gamma counter (Wallac Wizard 1480, Perkin Elmer). The presented data were normalized to the known labeling efficiency of the used cells and half-life of the label.
Tissue samples were collected from sacrificed mice 1 or 24 hours after i.v. injections of iron-labeled MSCs. Dissected samples except femurs were fixed in 10% phosphate buffered formaldehyde and subsequently embedded in paraffin. The samples were cut into 5-μm thick sections, mounted on glass slides, and deparaffinized. The slides were stained with Prussian blue with or without hematoxylin and eosin (H&E) counter staining or with a modified Prussian blue method without Kernechtrot staining. Femurs were fixed with 70% ethanol and embedded in methylmetacrylate. The plastic-embedded, demineralized bone samples were cut into 5-μm thick sections and mounted on glass slides. Bone samples were stained with Prussian blue or Prussian blue with H&E counter staining.
GraphPad Prism was used for all statistical analysis (GraphPad Software, Inc., LA Jolla, CA, www.graphpad.com). Multiple group comparison analysis was done with one-way analysis of variance (ANOVA) and Bonferroni post-test with a significance levels set at 0.05 (95% confidence interval). Two-tailed unpaired t test with 95% confidence interval was used when comparing two groups. A p-value <.05 was determined as significant (*), a p-value <.01 as very significant (**) and a p-value of <.001 as highly significant (***).
UCB-MSCs Exhibit a Faster Lung Clearance Rate than BM-MSCs
The biodistribution of i.v. infused Tc-HMPAO-labeled BM-MSCs and UCB-MSCs was assessed and compared by planar whole-body imaging (pinhole scintigraphy) and quantitative measurements of radioactivity in dissected organs. The in vivo biodistribution study was repeated three times (n = 3) with different MSC lines used each time. One separate animal experiment was additionally done with iron-labeled cells and MRI imaging (data not shown) with subsequent histologic postanalysis. Figure 1A shows a representative summary of the biodistribution of the cells 1 and 12 hours postinjection. It is evident that there is strong lung adhesion 1 hour after injection for both cell types, but that the dislodgement from the lung is significantly faster for the UCB-MSCs during the subacute phase (12 hours after injection). Quantitative results from organ radioactivity measurements are presented in Figure 1B and 1C. The difference in lung adhesion 12 hours after i.v. injection between BM-MSCs and UCB-MSCs is statistically significant (p = .0006, Fig. 1C). Interestingly, significantly more UCB-MSCs are transported to the liver 12 hours after infusion (Fig. 1C). All the other studied organs exhibited very equal levels of radioactivity (Fig. 1B, 1C). We have previously obtained similar results with intracardially introduced MSCs (data not shown).
The Infused MSCs Are Localized in the Vicinity of the Endothelium and Without Evident Signs of Microemboli
To exclude the possibility that significant cell occlusion of the microvasculature is the sole reason for the observed biodistributional pattern, we conducted a histological subcellular analysis of iron-labeled cells 1 and 24 hours postinjection. The labeled cells were found in the vicinity of the alveolar endothelium in the lung sections (Fig. 2A, 2B). In the liver, the labeled cells were localized in the vicinity of hepatocytes and sinusoids (Fig. 2C, 2D). Although the homing of MSCs is minimal to the BM, some extravasated cells were clearly also localized in the BM (Fig. 2E, 2F). Background staining was evaluated in control animals injected with the iron label only and a clear difference was concluded in the staining intensity of control animal samples and animals injected with iron-labeled cells (data not shown). To confirm the cell-specific staining and subcellular localization of the cells, lung sections without H&E counterstaining were also examined (Fig. 3). At 1 hour postinjection, all cells were located in capillaries as single cells or in very small clusters (Fig. 3A, 3C, 3F) and with no evident difference between BM-MSCs and UCB-MSCs. It was however difficult to distinguish between capillary embolism and cells passing through capillaries at this time point. At 24 hours postinjection, the BM-MSCs were still found in significant amounts in the lung capillaries and mostly as small clusters (Fig. 3B, 3D), whereas the UCB-MSC were found only in small amounts (3G). At least some of the BM-MSCs had traversed the vascular wall and were located inside the thin lung stroma (Fig. 3E). Significant signs of microemboli were not detected at this time point.
Significant Differences in Cell Surface Adhesion Molecules Are Seen Between the Two Cell Types
The animal data together with the subcellular localization of the cells lead us to believe that the differences in the dislodgement rates from the lung could be caused by cell surface adhesion molecules. The cell surface profile was studied using flow cytometry and a panel of antibodies against both protein and glycan molecules relevant in adhesion (Fig. 4). Interestingly, UCB-MSCs expressed significantly higher levels (p = .0029 and p = .0096, respectively) of α6 (CD49f) and α4 (CD49d) integrins when compared with BM-MSCs (Fig. 4A). UCB-MSCs also expressed significantly higher amounts (p = .0013) of the hepatocyte growth factor receptor c-Met (Fig. 4A). Fibronectin expression was concluded to be higher in the BM-MSCs than UCB-MSCs, but the difference was not statistically significant (p = .3361, Fig. 4A). The expression of the PODXL was at an anticipated and equal level in both cell types and verifies the use of a comparable confluence when collecting the cell material . The expression levels of the stromal cell-derived factor-1 (CXCL12) receptor CXCR4 and the CX3CL1 receptor CX3CR1 were very low in both cell types (Fig. 4A).
As all the above-mentioned adhesion proteins are known to be glycosylated, we next wanted to explore important glycan epitopes with established roles in cell migration and adhesion using glycan-specific antibodies. It is noteworthy that some glycan-specific antibodies may have lower affinity to their ligands as compared to antiprotein antibodies. As presented in Figure 4B, there are significantly more of the relevant glycolipid carbohydrate epitopes GD2, a disialoganglioside, and SSEA4, the stage-specific embryonic antigen-4, in BM-MSCs when compared with UCB-MSCs (Fig. 4B). In line with previous published data, the UCB-MSCs also express SSEA4 , but significantly less than the BM-MSCs (p = .0004). As we expected to see a difference in the fucosylation level, the expression of the sialylated Lewis x epitope (sLex) was studied with a variety of antibodies. Surprisingly, a dramatic difference was seen between these supposed sLex-binders which could be explained by their slightly different core structure specificity (www.functionalgenomics.org), but no statistical differences between BM-MSCs and UCB-MSCs were seen (Fig. 4B). The highest expression level was observed with a sLex antibody reported to bind core-2 O-glycan (CHO-131) (Fig. 4B), but we have lately discovered that this antibody also binds core-2 O-glycan without fucosylation (data not shown). A trend toward a higher α1,3-fucosylation-dependent sLex expression in the UCB-MSCs was observed with the CD15s (CSLEX) sLex antibody, which has been reported to bind sLex both in N- and O-glycans (Fig. 4B). The expression level of the nonsialylated Lewis x epitope (Lex) was higher in the UCB-MSCs, but not statistically significant (p = .5662, Fig. 4B). Together, the data suggest that UCB-MSCs have an overall higher degree of cell surface α1,3-fucosylation, namely in the antennae of the oligosaccharides.
A Higher Fucosylation Level in the N-Glycan Profile of UCB-MSCs
The results from the flow cytometry study with glycan-specific antibodies gave us reason to believe that there could be relevant differences in glycosylation between BM-MSCs and UCB-MSCs. Subsequently, an in-depth analysis of N-glycan structures was performed by MALDI-TOF mass spectrometry (Table 1). The most common glycan structures in both cell types were unprocessed high-mannose-containing neutral N-glycans and complex, fully processed acidic N-glycans. Remarkably, the UCB-MSC neutral N-glycan profile contained approximately 20% more high-mannose N-glycans than in the BM-MSC profile. In line with the flow cytometry results, BM-MSCs and UCB-MSCs contained equal amounts of sialylated complex-type N-glycans. Furthermore, the N-glycan profiling revealed that the proportion of unsialylated N-glycans (nHexNAc = 3 and nHex >1 and nHexNAc >3 and nHex >2) was at least the double in BM-MSCs than in UCB-MSCs, whereas larger unsialylated multiantennary N-glycans (5 ≤ nHexNAc = nHex – 1) were more abundant in UCB-MSCs. The most striking N-glycosylation difference between BM-MSCs and UCB-MSCs was the level of fucosylation. UCB-MSCs had more sialylated N-glycans with at least one fucose, indicating higher core α1,6-fucosylation  and more sialylated and unsialylated glycans with additional fucosylation (N-glycan structures containing ≥2 fucose), indicating either α1,2-, α1,3-, or α1,4-fucoslyation. For example, the percentage of glycan structures with ≥2 fucose was 41.7% of neutral N-glycans for UCB-MSCs and only 27.5% for BM-MSCs (Table 1). Considering the minor increase in the expression of the Lex epitope in UCBMCSs, indicating α1,3-linked fucosylation (Fig. 4B), the more dominant fucosylation increase seen in the mass spectrometric analysis may indicate increased α1,2-fucosylation such as Lewis b, Lewis y, H type 1 or H type 2 antigens, and/or α1,4-fucosylation (such as Lewis a or sialyl Lewis a). The sialylation level was concluded to be similar between the cell types (Table 1).
We identified a significant difference in the α6 (CD49f) and α4 (CD49d) integrin expression levels between BM-MSCs and UCB-MSCs (Fig. 4A). To explore the downregulated expression levels of these integrins in the BM-MSCs further, we studied BM-MSCs isolated from donors of both higher age (>50-year old) and young age (<20-year old) and compared them to UCB-MSCs. Only BM-MSCs isolated from donors of higher age (>50-year old) had been used in the biodistribution studies (Figs. 1, 2, 3) and the flow cytometry study presented in Figure 4. Strikingly, the expression levels of CD49f and CD49d were comparable to UCB-MSCs when the BM-MSCs were derived from young donors, but yet a significant difference between UCB-MSCs and BM-MSCs from older donors was observed (Fig. 5A, 5B). Subsequently, the biodistribution of BM-MSCs from younger donors was studied in vivo and compared to UCB-MSCs (Fig. 4C). It was evident that the lung adhesion at 12 hours postinjection was less abundant for the BM-MSCs from younger donors than for the BM-MSCs from older donors (Fig. 1) and more comparable to UCB-MSCs. Taken together, the results indicate that the CD49f and CD49d might have crucial roles in MSC lung adhesion and subsequent dislodgement.
To rule out that the observed differences in the cell surface expression of CD49f, CD49d, and c-Met could be due to differences in the level of confluence at the time of cell collection, we explored the expression of the most relevant molecules by flow cytometry in UCB-MSCs collected in either at 40%, 70%, or 100% confluence (Fig. 6). There is an obvious impact of confluence on PODXL, CD49f, CD49d, and c-Met expression in UCB-MSCs and cell surface expression gradually decreases with higher confluence, but the expression levels never fall below the mean levels observed in Figure 4A even at a very high confluence. These results thus verify the observed significant difference in CD49f and CD49d cell surface expression between BM-MSCs and UCB-MSCs (Figs. 4A, 5A, 5B). The cell surface PODXL1 expression is dramatically downregulated when confluence rises above 70%, which is in line with previously published results  and also validates our methodology. Strikingly, fibronectin expression/secretion is gradually upregulated with cell confluence (Fig. 6), which is highly interesting and should be explored further. The results also implicate that the expression of sLex is not affected by confluence.
Cell therapies are expected to provide cures to a multitude of disease and disorders currently without adequate treatments and it is envisioned that cell therapy will compose the fourth and final therapeutic pillar in modern healthcare . MSCs currently hold the highest expectations as future cell therapy products for a variety of indications. The most promising early clinical results with MSCs have been observed in difficult immunological conditions such as steroid-resistant GvHD . Delivery routes are essential for therapy efficiency and systemic delivery by infusion is the obvious goal for many forms of cell therapy. However, there are some obstacles in this delivery route, one of the main being the observed massive lung adhesion after systemic infusion observed by us and others (see Introduction). In addition to the negative impact lung adhesion poses on the possibility to reach clinically relevant numbers to target organs, lung entrapment of MSCs has been observed to cause severe lung damage in mice models [14, 23]. Paradoxically, lung-entrapped MSCs improved myocardial infarction in mice through secreting the anti-inflammatory protein TSG-6 . There are also promising initial results for therapeutic effects of MSCs in several models of lung disease, especially in acute lung injury ALI/ARDS reviewed in [31, 32]. Beneficial clinical effects on pulmonary pressure have also been reported after intracoronary infusion of BM-derived mononuclear cells after myocardial infarction .
MSCs initially entrap in the lung after systemic infusion, but we unexpectedly found a striking difference in the subacute biodistribution pattern between BM-MSCs and UCB-MSCs with a much faster lung dislodgement rate for the UCB-MSCs. Previous studies have explored different strategies to minimize lung adhesion and improve homing of systemically infused cells: use of vasodilator , prebolus injection of MSCs , reducing the number of cells to be introduced , heparin saturation of MSCs , or preincubation of cells with white blood cells . Some beneficial effects on the lung adhesion have been concluded, but the major mechanism behind this profound phenomenon is still unsolved, which indicates that the explanatory mechanism can be multivariate. Previous studies have presented speculations for receptor-mediated mechanisms behind MSC lung entrapment [2, 11]. However, some studies have claimed cell size as the major reason behind lung entrapment [9, 11, 15, 36]. We also clearly observed a cell size difference between the BM-MSCs and UCB-MSCs used in this study (Fig. 1), probably reflecting the different culture conditions used. The histology, however, indicated that there could exist more specific mechanisms behind the lung adherence and cell dislodgement, because we did not see significant microemboli and the lung-entrapped cells adhered to the lung vascular endothelium. A more detailed analysis of the cell surface profiles revealed differences in the expression levels of relevant adhesion molecules. The UCB-MSCs, with the observed faster lung clearance rate, had significantly higher expression levels of α6 (CD49f) and α4 (CD49d) integrins and c-Met. We ruled out that the higher expression of these adhesion molecules in UCB-MSC could be due to confluence differences as has been shown before . The hepatocyte growth factor receptor c-Met has a reported role in MSC mobilization, tissue repair, and wound healing , and thus makes it an intriguing molecule for future studies concerning MSC delivery. CD49f and CD49d have well-established roles in homing and migration of hematopoietic stem cells (HSCs) [38–40] and it has been previously reported that BM-MSCs express only low levels of these integrins , which is in line with our data. The possible role for these integrins in the faster cell dislodgement from the lung is somewhat contradictory to earlier HSC studies, where blockade of functional CD49f or CD49d has been found to favor engraftment to a certain degree [11, 38–40]. It is challenging to compare the MSC results to HSC engraftment studies since it is obvious that MSCs, as adherent in vitro cultured cells with full anchorage-dependent behavior, display a larger variety of cell surface adhesion molecules and use different adhesion molecules and activation mechanisms than, for example, HSCs and committed immune cells. It is known that MSCs have receptors for numerous extracellular matrix components including collagen, laminin, fibronectin, and vitronectin and have a distinct expression pattern of integrins [41–44]. Interestingly, the expression level of fibronectin, recognized by at least 10 cell surface receptors of the integrin family, was higher in the more lung adhesive BM-MSCs (Fig. 4A). We also show for the first time that the expression of fibronectin increases with confluence, which highlights the connection between MSC culture stage and expression level of adhesion molecules also shown by others . This should be explored further, since remodeling of the extracellular matrix composition might be a relevant target in clinical cell production. In addition to confluence, we also observed that CD49f and CD49d expression was affected by BM donor age and that BM-MSCs from donors younger than 20 years have a more comparable cell surface profile to UCB-MSCs. The influence of donor age on the BM-MSCs is an active research topic in several laboratories and we have previously shown that BM donor age affects the energy metabolism of the BM-MSCs . Our results in this study further highlight the impact the raw material might have on MSC production.
We believe that there is a strong rationale behind adhesion molecule involvement in MSC lung entrapment and dislodgement. Size certainly matters, but in that respect that a larger cell does have a larger cell surface area with possibly more adhesion molecules expressed per cell and a higher probability for cell-cell contact through adhesion molecules and their cognate ligands on the alveolar endothelium. It is worthy to keep in mind that the initial 80% of cell entrapment for the BM-MSCs is decreased to 40% after 12 hours (Fig. 1B, 1C), that is, half of the cells are easily mobilized albeit the large cell size. Hence, we hypothesized that the cell surface composition is of additional importance. Many glycans also have established roles in adhesion and migration and the most remarkable glycan studied is the sLex epitope. In line with the initial hypothesis, we found relevant differences in the fucosylation levels between BM-MSCs and UCB-MSCs. The UCB-MSCs exhibited a higher percentage of fucosylated structures which goes well together with an observed more migratory phenotype [46, 47]. Our results thus strengthen the concept of modifying cell surface fucosylation levels to improve cell targeting [46, 47]. The glycan profiling additionally revealed interesting targets from a MSC marker point of view: the UCB-MSCs contained more high-mannose N-glycans than BM-MSCs, and we have previously shown that elevated number of such unprocessed N-glycans relates to more primitive cell stages in HSCs, BM-MSCs, and embryonic stem cells [24, 27, 48, 49]. In line with earlier observations , the expression of the neural ganglioside GD2 was only expressed in BM-MSCs. The totally absent GD2 expression in UCB-MSCs was striking and GD2 could thus serve as a marker to distinguish BM-derived MSC from MSC of other origin.
This study used a randomized experimental setup to explore the mechanisms behind the MSC lung entrapment using two different MSC cell types. The different culturing conditions for the cell types were intentional, and the UCB-MSC culturing protocol also contained some added cytokines to increase the probability of expression differences in cell surface adhesion molecules. This is obviously also a methodological weakness in this study, because the conditions as such cause an effect. Differences in culture conditions are however the reality in the cell therapy modality where parameters such as plating density, confluence, passage number, use of animal-derived reagents, cytokine supplements, and freezing makes it challenging to compare different cell products. Taken together, our results show that a variety of molecules may have a role in the lung entrapment of systemically infused cells and the composition of the cell surface can affect the migratory behavior of therapeutic cells. Our early findings especially suggest a relevance of cell surface α4 and α6 integrins in MSC lung adherence and dislodgement. We present a hypothesis that rather than focusing on single adhesion molecule analysis, novel cell therapy products gain from more complex cell surface profiling to predict efficacy and safety. Technologies for controlled modifications of the cell surface composition might be future tools to increase the delivery success of therapeutic cells.
We thank Sirkka Hirschovits and Minna Savilampi for expert technical assistance. We also thank Timo Ruuska and Roberto Blanco for expertise and help in cell labeling methodology and in vivo imaging. This study was supported by Tekes—the Finnish Funding Agency for Technology and Innovation.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
Tero Satomaa is a shareholder in Glykos Finland Ltd. The other authors report no conflicts of interest.