Author contributions: G.M., I.R.D., K.N.E., O.K., and K.L.B.: designed the study and wrote the manuscript; O.R. and K.L.B.: led the clinical study; G.M., I.R.D., L.V.B., A.M.C.A., and J.S.: performed the research and analyzed the data; P.M., G.E., L.F., L.L., O.H., and Y.T.: assisted various experiments.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS April 20, 2012.
Multipotent mesenchymal stromal cells (MSCs) are tested in numerous clinical trials. Questions have been raised concerning fate and function of these therapeutic cells after systemic infusion. We therefore asked whether culture-expanded human MSCs elicit an innate immune attack, termed instant blood-mediated inflammatory reaction (IBMIR), which has previously been shown to compromise the survival and function of systemically infused islet cells and hepatocytes. We found that MSCs expressed hemostatic regulators similar to those produced by endothelial cells but displayed higher amounts of prothrombotic tissue/stromal factors on their surface, which triggered the IBMIR after blood exposure, as characterized by formation of blood activation markers. This process was dependent on the cell dose, the choice of MSC donor, and particularly the cell-passage number. Short-term expanded MSCs triggered only weak blood responses in vitro, whereas extended culture and coculture with activated lymphocytes increased their prothrombotic properties. After systemic infusion to patients, we found increased formation of blood activation markers, but no formation of hyperfibrinolysis marker D-dimer or acute-phase reactants with the currently applied dose of 1.0–3.0 × 106 cells per kilogram. Culture-expanded MSCs trigger the IBMIR in vitro and in vivo. Induction of IBMIR is dose-dependent and increases after prolonged ex vivo expansion. Currently applied doses of low-passage clinical-grade MSCs elicit only minor systemic effects, but higher cell doses and particularly higher passage cells should be handled with care. This deleterious reaction can compromise the survival, engraftment, and function of these therapeutic cells. Stem Cells2012;30:1565–1574
Based on their immunomodulatory and tissue reparative properties, multipotent mesenchymal stromal cells (MSCs) have been thought to offer a novel therapeutic approach for treatment of various inflammatory diseases [1–3]. At present, MSCs are being evaluated in clinical trials in cartilage repair, stroke, spinal cord injury, graft-versus-host disease (GvHD), heart disease, Crohn's disease, and several other diseases . Intravenous infusion of MSCs appears safe,  and no acute toxicity has been reported, at the currently applied cell dose. However, many basic questions concerning the hemocompatibility of MSCs  and their fate after systemic infusion remain unanswered [4, 6].
Early investigation into the biodistribution of MSCs after systemic delivery indicated that MSCs are rapidly cleared from the circulation in humans . MSCs infused intravenously into rodents are rapidly entrapped in the lung [8–14] and then cleared from the circulation . The microemboli that form in the pulmonary circulation are potentially lethal to the animals if sufficient numbers of cells are infused [11, 12]. Similarly, MSCs injected on the arterial side arrest at the precapillary level . Studies in large animals have corroborated the formation of arterial microinfarcts as well as lung entrapment after infusion on the venous side [16, 17]. Rapid trapping in the lung has been attributed to the combination of high concentrations of strongly adhesive MSCs entering lung capillaries of a low diameter during the first passage . However, the fraction of MSCs trapped in the lung is not significantly reduced by infusing MSCs at lower concentrations or by pretreating the cells with antibodies against integrins in order to reduce their adhesiveness . Thus, other mechanisms are likely to contribute to the embolization and cell loss that occurs in the microvasculature.
Pancreatic islets and hepatocytes trigger an innate immune attack after exposure to blood in vitro and in vivo [5, 19-21]. This reaction is characterized by activation of complement/coagulation cascades, binding of activated platelets to the cells, and consecutive clot infiltration by neutrophil granulocytes and monocytes, eventually leading to cell destruction. This so-called instant blood-mediated inflammatory reaction (IBMIR) results in a rapid loss of up to 80% of the infused cells shortly after cell infusion . Consequently, cells to be used for therapeutic purposes need to be designed to either avoid, or at least resist, the negative effects of IBMIR if cell persistence is desired . As yet, the surface properties of MSCs have not been well studied. It is not clear whether trypsin-detached MSCs promote a prothrombotic or antithrombotic state after blood exposure and how cell expansion affects these properties.
Several factors such as nitric oxide, prostacyclin, tissue factor pathway inhibitor (TFPI), and heparin-bound antithrombin mediate the strong antithrombotic properties on the luminal surface of endothelium , while prothrombotic tissue factor (TF) and collagens are found in the subendothelial/perivascular compartment to ensure hemostasis by rapidly activating the coagulation cascade after vascular damage . Collagens, which are typically found on MSCs and other stromal cells, [23–25] are known to act as platelet agonists. Furthermore, negatively charged collagen residues trigger the contact activation system by autoactivation of factor XII, which then initiates the coagulation cascade (intrinsic pathway) . Another potent initiator of coagulation is TF, which is located in nonvascular tissues such as the stroma and adventitia of larger blood vessels and on malignant cells (extrinsic pathway) [20, 26]. Expression of TF within the endocrine tissue of islet cell grafts triggers the IBMIR .
The aim of this study was to characterize the intrinsic blood compatibility of culture-expanded MSCs on a molecular and functional level. We used an in vitro whole-blood model to study different clinical parameters such as the MSC donor, cell dose, and cell-passage number on the blood's response to MSCs. We furthermore compared our in vitro observations with measurements of blood activation markers in MSC-treated patients.
MATERIALS AND METHODS
Forty-four MSC recipients who underwent hematopoietic stem cell transplantation (HSCT) at the Karolinska University Hospital, Huddinge, Sweden and received treatment between 2003 and 2010 were included in this analysis. Patients received myeloablative (n = 28) or reduced-intensity conditioning (n = 16) and GvHD prophylaxis, according to previously published procedures . The indications for MSC administration were: failure of standard treatment approaches for acute GvHD refractory to standard therapy (30 patients) and tissue injury after HSCT (hemorrhagic cystitis and pneumomediastinum) (14 patients). All the participating patients have been included in previous reports [1, 3]. Patients received MSC infusions from third-party unrelated donors (n = 53), from haploidentical related donors (n = 11), or from human leukocyte antigen (HLA)-identical siblings (n = 5). Patients received up to five infusions of MSCs at passage 1-4 in doses of approximately 1.0–3.0 × 106 cells per kilogram. Plasma samples were collected preinfusion and at 1, 3, and 24 hours after MSC infusion. Recipients gave informed consent, and the study was approved by the Regional Ethics Review Board.
Clinical Expansion of MSCs
Human MSCs were obtained from bone marrow aspirates following protocol approval by the ethics committee and review board at Karolinska University Hospital, Huddinge, Sweden and were isolated and characterized as described previously [3, 28]. All MSC donors (n = 35) were considered healthy after assessment of medical history, physical examination, and serological screening for HIV and hepatitis viruses. To isolate MSCs, bone marrow aspirates of approximately 50 ml were taken from the iliac crest of healthy donors (n = 35; median age, 37; range, 1–66 years). The expansion and characterization of the MSCs were performed according to the guidelines of the MSC Consortium of the European blood and marrow transplantation group and approved by the Swedish National Board of Health and Welfare, as previously described in detail [2, 3]. The clinical MSCs were cultured in medium containing 10% fetal calf serum (FCS), and a median dose of 1.6 × 106 cells per kilogram was given at a low cell-passage number (P1-4). For rapid availability, most of the cells were stored in liquid nitrogen and freshly thawed for IV infusion. Flow cytometry analysis indicated that the MSCs were positive for CD73, CD90, and CD105 but negative for CD14, CD31, CD34, and CD45. Adipogenic and osteogenic differentiation after induction were evaluated as previously described . The MSC suspensions were culture negative for bacteria and fungi and polymerase chain reaction (PCR) negative for Mycoplasma pneumoniae [2, 3].
Isolation and Culture of Cells for Experiments
To isolate MSCs, bone marrow mononuclear cells were separated over a gradient of Redigrad (GE Health Care, Uppsala, Sweden), washed, and resuspended in Dulbecco's modified Eagle's medium (DMEM) low-glucose medium (Invitrogen, Grand Island, NY); supplemented with 100 IU/ml penicillin, 0.1 mg/ml streptomycin, and 10% FCS (Hyclone, Logan, UT); and plated at 1.6 × 105 cells per square centimeter. When the cultures neared confluence (>80%), the cells were detached by treatment with trypsin and EDTA (Invitrogen) and replated/passaged at a density of 4.0 × 103 cells per square centimeter for up to eight passages. Human umbilical vein endothelial cells (ECs) (Promocell, Heidelberg, Germany) were grown in EC growth medium (Promocell), supplemented with 100 IU/ml penicillin and 0.1 mg/ml streptomycin, and replated at 10,000 cells per square centimeter. Cells for experiments were obtained from cell layers that had been washed twice with sterile phosphate buffered saline (PBS) to remove nonadherent or dead cells. The cell viability was assessed by trypan blue exclusion (generally >95%), and the cell suspensions were adjusted to 1–2 × 106 cells per milliliter in PBS.
Gene Expression Analysis
Mixed Lymphocyte Reactions
Mixed lymphocyte reactions (MLRs) were performed as described elsewhere . MSCs were stimulated by coculture with activated peripheral blood mononuclear cell (PBMCs) for 5 days in transwell MLRs and compared to unstimulated MSCs or ECs. Responder PBMCs were stimulated with a pool of allogeneic donors (n = 5), which were placed together in the top well of the culture plate; third-party MSCs were placed in the bottom well, at a ratio of 1:10 to PBMCs.
RNA Isolation, cDNA Preparation, and Quantitative Real-Time PCR Analysis
Cell lysates were harvested with lysis buffer (Qiagen, Hilden, Germany), RNA was extracted using a Qiagen RNeasy minikit, and then stored in RNase-free water at −70°C. The concentration and purity of the RNA were estimated by reading absorbance at 260 and 280 nm with a spectrophotometer (Nanodrop; Thermo Fisher Scientific, Wilmington, DE). The cDNA samples used for PCR analysis were obtained using the high-capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) or, alternatively, the RT2 PCR Array First Strand kit (SAB; SuperArray Bioscience, Frederick, MD). Quantitative real-time PCR (qRT-PCR) assays were performed using the human EC biology RT2 Profiler PCR Array (SAB) on an ABI PRISM 7900 HT Fast Block (Applied Biosystems). Data analysis is available at the company website: http://www.superarray.com/pcr/arrayanalysis.php. Additional qRT-PCR analysis was performed with the Applied Biosystems 7900 HT sequence detection system. Specific primers were designed for TF, collagen type-1 subunit A1 (COL1A1), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Supporting Information Table S1). The expression level of GAPDH was used as an internal standard: Ct = Ct(Gene) − Ct(GAPDH), with Ct being the cycle threshold of GAPDH or the gene of interest. Results were calibrated against a negative control and further analyzed by the 2 method .
Confocal Microscopy Analysis
Subconfluent cells were detached with trypsin and allowed to adhere to microscope slides, fixed with 70% ethanol containing 30% acetone, and labeled with antibodies reconstituted in PBS with 2% goat serum and 1% bovine serum albumin (BSA). The cells were first labeled with mouse control IgG (Dako) or mouse anti-human monoclonal antibodies (Supporting Information Table S2) directed against endoglin (CD105, BD), TF (Calbiochem), collagen type-1 (COL1, Sigma), or fibronectin 1 (FN1, Sigma), then visualized with a secondary AlexaFluor488-conjugated goat anti-mouse antibody (Molecular Probes, Eugene, OR). Hoechst 33342 (Sigma) was used to detect cell nuclei. Images were acquired at ×63 magnification with a confocal microscope (Zeiss LSM 510 Meta; Carl Zeiss, Jena, Germany). To obtain three-dimensional (3D) projections, several 20-μM Z-stacks were acquired and visualized with IMARIS imaging software (Bitplane AG, Zurich, Switzerland); 3D reconstructions of surfaces and a central two-dimensional slice view are shown.
Whole-Blood Chandler Loop Experiments
ECs and resting or MLR-stimulated MSCs were exposed to human blood using the Chandler loop system, consisting of tubing with a heparinized inner surface (Corline Systems) [19, 21, 30]. Fresh non-anticoagulated ABO-compatible human blood was obtained from healthy volunteers who had given informed consent and had received no medication for at least 10 days. Pieces of tubing containing 7 ml of human blood were prepared and supplemented with 100 μl PBS ± different cell types. In indicated experiments, TF-pathway antagonists were added to a final concentration in blood of 1 nM for active site-inactivated activated factor VII (FVIIai) and 7.5 μg/ml for anti-TF blocking antibody (#4509) or respective control antibody (#4503) without blocking activity (Amercian Diagnostica, Greenwich, CT) . The tubing was closed with a heparinized connector and placed in a 37°C heated rocking apparatus to simulate blood flow. To determine the time course of the reaction between the blood and the cells, 1-ml samples from each blood tube were collected before cell addition and at 5, 15, 30, and 60 minutes after cell addition. Reactions were stopped by addition of 10 mM EDTA (pH 7.4). Platelet and cell counts were obtained for each sample using a cell counter (Beckman Coulter). The remaining sample volume was centrifuged at 3,000g for 20 minutes at 4°C. Plasma samples were collected and stored at −80°C for analysis by ELISA.
Measurement of Blood Activation Markers and Cell Counts in Patients
Formation of the blood activation markers thrombin-antithrombin complex (TAT), activated FVII antithrombin complex (FVIIa-AT), FXI-AT, FXII-AT, and complement activation makers C3a and sC5b-9 in plasma were measured by ELISA, according to previously described methods [31–33]. Quantification of patient blood counts, D-dimer, fibrinogen, albumin and creatinine was performed by diagnostic routine methods. Patients with hemorrhage and/or those who received any type of transfusion within 48 hours of the MSC infusion were excluded from the blood count analysis.
Statistical analyses were performed using analysis of variance (ANOVA) and Student's t test. When the data did not fit a normal distribution, the Mann-Whitney U test or the Wilcoxon matched-pairs test was used (two-tailed confidence intervals, 95%; p < .05 was considered statistically significant; Prism 5.0; Graphpad Software).
EC Profiling of MSCs and ECs
ECs are prototypic blood compatible cells ; it is not yet clear whether MSCs have similar properties. Accordingly, we performed EC biology transcript analysis on low-passage MSCs from three individual donors, comparing resting or MLR-activated MSCs to ECs (Supporting Information Table S3). We found that both MSCs and ECs expressed typical hemostatic regulators (Fig. 1A), including TFPI, tissue- and urokinase-type plasminogen (PLG) activator (tPA and uPA), and prostacyclin synthase, the source of prostacyclin 2 (PGI2), a strong inhibitor of platelet activation. Endothelial NO-synthase was more prominent in ECs, whereas its inducible form predominated in MSCs. ECs showed a strong expression of von Willebrand factor, which was absent from MSCs (p < .05). The expression levels of PLG, thrombomodulin, PLG activator inhibitor 1, and procoagulant platelet factor 4 (PF4) were similar in both cell types (p > .05). Exposure of MSCs to proinflammatory stimuli led to upregulation of PF4, integrin ICAM1, and immunomodulatory and trophic factors, such as chemokines (CX3CL1 and CCL2/5), cytokines (interleukin-1-beta [IL1B], IL6, IL7, and IL11), and growth factors (placental growth factor, fibroblast growth factor, GM colony stimulating factor, and particularly vascular endothelial growth factor) (Supporting Information Table S3). ECs displayed high levels of the VEGF receptor transcripts FLT1, KDR, and KIT (p < .05), which were absent from MSCs. In contrast, MSCs expressed higher levels of platelet-derived growth factor receptor-alpha and angiopoietin 1 (p < .01). Expression of FN1 and its corresponding receptor subunits (ITGB1 and ITGA5) was high in both cell types. The gelatinases matrix metalloproteinase 2 (MMP2) and MMP9 were expressed to similar levels, whereas the expression of interstitial collagenase MMP1 was lower in MSCs, but its expression increased in response to proinflammatory stimuli (p < .001).
Differential Expression of Procoagulant Molecules by MSCs and ECs
The expression of procoagulant factors was analyzed with qRT-PCR and confocal microscopy. TF transcripts were detected at low levels in resting ECs and MSCs and its expression increased in MSCs upon challenge with proinflammatory stimuli in transwell MLRs (p < .01; Fig. 1B). We found high constitutive expression of COL1-specific transcript COL1A1 in MSCs but not in ECs (p < .001; Fig. 1B), which was downregulated in response to MLRs (p < .001). Both ECs and MSCs expressed moderate to high levels of FN1, respectively, and its expression was not significantly changed upon stimulation in MLRs. Confocal microscopy analysis confirmed weak expression of TF and strong expression of COL1 and FN1 in MSCs (Fig. 1C, 1D), but no TF or COL1 and only moderate amounts of FN in ECs. Both cell types expressed similar amounts of characteristic surface marker endoglin (CD105). 3D modeling revealed that TF, COL1, and FN1 were mainly localized in vesicular structures close to or directly on the cell surface (Supporting Information Fig. S1A). Furthermore, expression of TF and COL1 increased when MSCs were cultured to higher passages (p < .001, Fig. 1C, 1D and Supporting Information S1B).
In Vitro Exposure of MSCs to Blood Triggers the IBMIR
The ability of ECs and MSCs to trigger the IBMIR was assessed in the Chandler blood loop model, which is shown in Supporting Information Figure S2A. ECs did not trigger the IBMIR. Blood exposure of MSCs led to initiation of the coagulation cascade, as indicated by increased formation of thrombin (TAT, p < .001; Fig. 2A), and other clotting factors, including activated FVII (FVIIa), FXIa, and FXIIa (Supporting Information Fig. S2B–S2D). The number of free platelets strongly decreased when compared with control blood (p < .001; Fig. 2B). Visible clots are formed in MSC-treated blood, with a concomitant decrease in granulocytes and monocytes (p < .05 and p < .001, Fig. 2C), but not lymphocytes. Only minute amounts of complement activation marker C3a and sC5b-9 could be detected at the currently applied cell dose (Fig. S2E, S2F), restricted mostly to higher passage MSCs (P5-8; p < .05 and p = .1, respectively; Fig. S2G).
We next investigated the level of MSC donor variability and the impact of cell dose and cell-passage number, which may affect the quality of the cell product [34–36]. We found that MSCs from different donors triggered the IBMIR to varying degrees (Fig. 3A). In repeated testing, ECs generally antagonized platelet activation, as opposed to MSCs, which elicited an average 20%–95% drop in free platelets after 30 minutes of blood exposure, compared to PBS-treated control blood (p < .01; Fig. 3A). The blood-activating properties of MSCs were dose-dependent (Fig. 3B). Low doses of MSCs (3,000 cells per milliliter) elicited significantly weaker platelet activation and thrombin formation than did cell doses used clinically (15,000 cells per milliliter) (p < .05 to p < .01, at 5–30 minutes). However, low-passage MSCs (P1-4) elicited significantly less platelet activation and thrombin formation than did MSCs harvested at passage 5-8 (p < .05 to p < .01, at 5–30 minutes; Fig. 3C).
We then wondered how blood would respond to transwell MLR-stimulated MSCs? Our initial screening revealed upregulation of TF and downregulation of ECM-component COL1 after coculture of MSCs with activated lymphocytes (Fig. 1B), which could also be confirmed on proteomic level (p < .05, Fig. 4A), while CD105 and FN1 were unaffected. Blood treated with MLR-stimulated MSCs showed stronger clot formation than blood treated with resting cells (Fig. 4B). Triggering of IBMIR correlated with TF expression on MSCs (p = .01, R2 = 0.57, Supporting Information Fig. S2H). Visible clot formation was prevented by addition of TF-pathway blocking agents FVIIai and monoclonal anti-TF-4509, but not by a similar concentration of anti-TF-4503, which recognizes a nonfunctional epitope (Fig. 4B). MLR-stimulated MSCs mediated a significantly stronger reduction in free platelets and increase in TAT after 30 minutes of blood exposure (p < .05, Fig. 4B, 4C), which was abrogated in the presence of FVIIai or anti-TF-4509 but not by anti-TF-4503.
Intravenous Infusion of MSCs Elicits a Weak Triggering of IBMIR
Forty-four patients received 69 infusions of MSCs for treatment of life-threatening complications to HSCT. Patient characteristics are shown in Table 1. No adverse events were observed either during or after MSC infusion. We performed a retrospective analysis of the patients' charts and found a weak drop in platelet counts after MSC infusion (∼ 15% drop, p < .01; Table 2), but counts of circulating leukocytes, neutrophils, lymphocytes, and monocytes did not show any acute changes within 24 hours. An analysis of soluble blood activation markers in patients' plasma samples revealed a mean fivefold increase in coagulation marker TAT (from 6.5 to 37 mg/l, p < .01) and complement activation maker C3a (from 195 to 1,049 mg/l, p < .01), whereas hemoglobin and hyperfibrinolysis marker D-dimer were not significantly increased after MSC infusion. No significant changes in fibrinogen, albumin, or creatinine were seen.
Table 1. Characteristics of MSC donors and recipients
Table 2. Measurement of blood activation markers in patients receiving MSCs
Statistical significance: mean ± SD. Exclusion criteria: (a) hemoglobin: excludes patients with macroscopic hemorrhage, erythrocyte transfusions, or apheresis within 48 hours; (b) leukocytes: excludes patients receiving granulocyte transfusions or apheresis within 48 hours; (c) platelets: excludes patients with macroscopic hemorrhage, platelet transfusions, or apheresis within 48 hours; and (d) coagulation factors: excludes patients with macroscopic hemorrhage, receiving fresh frozen plasma, or apheresis within 48 hours.
Abbreviations: C3a, complement component 3 fragment a; MSC, mesenchymal stromal cell; n, number of measurements; N.S., not significant; TAT, thrombin-antithrombin complex.
Within the last three decades, MSCs have been extensively studied, and their therapeutic potential has been recognized . After infusion, long-term engraftment is low or absent in MSC recipients . To better understand the early events after intravenous injection, we studied the interactions of MSCs with human blood. We found that MSCs trigger the IBMIR after blood exposure, a reaction initiated by prothrombotic tissue/stromal factors expressed on their surface. Triggering of IBMIR increased after activation of MSCs in a proinflammatory milieu (transwell MLRs), due to upregulation of TF, but could be antagonized by blocking the TF-pathway. The expression of prothrombotic factors furthermore increased after extended culture, as did the cells' prothrombotic effect. Low-passage MSCs, as typically used in clinical applications, showed only a weak triggering of IBMIR at the currently applied cell dose, but our results indicate that higher cell doses, and particularly higher passage MSCs, should be handled with care.
MSCs are believed to reside within the perivascular niche in vivo [37–40], and it can be assumed that these cells have at least a certain degree of intrinsic blood compatibility. The prototypic blood compatible stromal cells, however, are ECs. They are strongly adapted to establish hemostasis at the interior of the vessel wall by balancing prothrombotic and antithrombotic factors . We therefore compared the hemostatic properties of ECs and MSCs, to better understand how MSCs interact with blood. Both cell types expressed typical hemostatic regulators such as TFPI, tPA, and uPA as well as NO- and PGI2-producing enzymes. As compared to ECs, MSCs expressed increased amounts of prothrombotic TF and platelet agonist COL1 on their surface, typically found within the stromal compartment [20, 23, 25, 39].
We found donor-to-donor variability in the expression of procoagulant molecules, which was furthermore strongly modulated in an proinflammatory environment, and a generally higher level of expression in late-passage MSCs (P5-8). TF is the more potent initiator of clotting; already very small amounts of TF elicit thrombotic effects, while collagen is rather associated with an amplification of IBMIR by recruitment of activated platelets. Our data indicate that although there is a weak downregulation of the abundant COL1 in MSCs after MLR-stimulation, expression of TF increases. In addition, its antagonist TFPI is not upregulated, thus further promoting coagulation. In agreement with the literature, not only the hemostatic profile of MSCs but also their immunomodulatory, angiogenic, and ECM remodelling properties were changed upon encounter of proinflammatory mediators [39–41].
Intravenous or arterial infusion in animal models leads to embolization of MSCs in the lung or arrest at the precapillary level [11, 12, 15]. Intravital microscopy has indicated that this blockage is a mechanical phenomenon rather than active adhesion to the endothelium, since it occurred in microvessels of small caliber but was not observed in arterioles, as has been described for cancer cells [10, 15, 16]. Our in vitro results indicate, that prior to these events, a weak activation of the coagulation and complement cascade occurs as soon as MSCs are exposed to blood. We observed formation of thrombin and other clotting factors, a reduction in free platelets, granulocytes, and particularly monocytes, and eventually fibrin clot formation. Indeed, in keeping with previous studies of pancreatic islets and hepatocytes, where the destructive effect of IBMIR is well documented, MSCs expressed TF and collagen, which initiate this process [5, 19-21].
Only weak formation of complement activation products was observed in vitro, at the currently applied cell dose, which was most evident with higher passage MSCs. We previously reported on MSCs resistance to direct complement lysis . The cells were found to weakly activate the complement system, owing to their relative lack in surface-bound complement regulatory proteins,  leading to deposition of complement activation fragments on their cell surface, and consecutive effector cell activation in blood. Of interest, MSCs may indeed inhibit complement activation within their native tissue environment by locally secreting factor H . However, due to the large blood volume that the cells encounter upon systemic infusion, a protective effect of factor H may be more difficult to imagine. This is, unless secreted factor H concentrates its inhibitory activity directly on the surface of MSCs, for example, after binding to its corresponding cell surface ligands, such as heparin sulfates, as typically observed on endothelium.
To date, no thrombotic events have been reported after systemic delivery of MSCs, although reports of potentially lethal microvascular plugging in experimental animals have warned clinicians of possible risks associated with this type of delivery [11, 12, 14, 16]. In patients infused with MSCs through a central venous catheter, we observed a weak drop in platelet counts within 24 hours after infusion, which was accompanied by fivefold increase in thrombin and C3a formation, possibly indicating the activation of coagulation and complement cascade. However, no systemic increase in D-dimer was detectable, indicating that the occurrence of thrombotic events was rather limited on a systemic level, at the currently applied cell dose. Triggering of IBMIR did not appear to be related to the degree of HLA-disparity between the patient and MSC donor, although larger studies are necessary to clarify this aspect.
MSCs used in experimental settings are often cultured to higher passages. We found that higher passage MSCs (P5-8) initiated significantly more IBMIR in the Chandler loops. Furthermore, clotting occurred faster and was more prominent with higher doses of MSCs: 3,000 as compared to 15,000 MSCs per milliliter, a dose corresponding to the dose of 1–3 × 106/kg recipient weight used in many clinical protocols . The combination of late-passage MSCs infused at high doses may potentially contribute to the lethal thrombotic complications reported in animal studies. The MSCs used in our patients were all harvested in low passage, not exceeding passage 4. Furthermore, very low-passage MSCs appeared to yield a higher therapeutic benefit . Early-passage MSCs, with limited prothrombotic effects, may therefore generally be recommended for most clinical applications. An exception is the treatment for hemorrhagic cystitis or major hemorrhages, where a localized prothrombotic activity of MSCs could be desirable, to stop the bleeding at sites of vascular damage [1, 45]. Of particular interest is the fact that MSCs upregulate prothrombotic factors such as TF and PF4 after encounter of activated immune cells in MLRs. Thus, triggering of IBMIR is potentially augmented at inflammatory sites and after embolization in the microvasculature, where MSCs may encounter activated immune cells such as macrophages/monocytes.
Several rodent studies have indicated that MSCs, which are embolized in small vessels [11, 13-15], are activated to release biologically active substances, which mediate tissue repair by limiting stress responses and apoptosis [11, 14, 46, 47]. Upon embolization, MSCs were shown to modulate immune responses by recruiting and educating immune and reparative cells, which lead to a systemic shift in cytokine production [40, 48]. Most interestingly, chemotaxis and immune-modulatory properties of MSCs can also be triggered by complement activation products [40, 42, 49], which were found to be formed in blood, after systemic infusion of these cells. After contact with activated platelets, MSCs secrete fibrinolytic enzymes and exert ECM remodeling activity , which may contribute to repair of tissue damage. Thus, triggering of IBMIR after systemic infusion of MSCs may potentially contribute to both wound healing and immune modulation.
CONCLUSIONS AND SUMMARY
We conclude that MSCs trigger the IBMIR after blood exposure in vitro and in vivo. Particularly, long-term expansion of MSCs promotes their prothrombotic profile and may compromise survival, engraftment, and long-term function of these therapeutic cells in vivo. At this stage, we do not want to exclude the possibility that triggering of the IBMIR, and the resulting interactions with other immune effector cells, may be needed for their paracrine effects to occur in vivo. Our clinical observations indicate that infusion of low-passage MSCs, at the currently applied cell dose, does not pose an immediate thrombotic risk. However, for optimal patient safety and to avoid confounding results, we recommend that further studies should be conducted with low-passage clinical-grade MSCs, together with dose-escalation studies in suitable animal models. Future studies will need to clarify how the negative effects of IBMIR can be overcome.
We would like to thank Dirk Pacholsky and Jan Grawé from the BioVis Core Facility at Ruedbeck Laboratory for their support with confocal microscopy and data analysis. This study was supported by grants from the Swedish Cancer Society (11 0315), the Children's Cancer Foundation (PROJ11/034), the Swedish Medical Research Council (K2011-??X-20742-04-6), VINNOVA (2010-00501), the Stockholm County Council (ALF) (20110152), the Cancer Society in Stockholm, the Swedish Society of Medicine, the Tobias Foundation, and Karolinska Institutet (to K.L.B); the Swedish Medical Research Council (K2007-65X-05647-28-3 950, 2009-4462, 2009-4675, and K2011-65X-12219-15-6), Swedish Research Council/VINNOVA/Swedish Foundation Strategic Research (6076170), and the Juvenile Diabetes Foundation International (to O.K. and B.N.); O.K. and B.N. positions are in part supported by NIH 2U01AI065192-06.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.