Alterations in the Cellular Immune Compartment of Patients Treated with Third-Party Mesenchymal Stromal Cells Following Allogeneic Hematopoietic Stem Cell Transplantation

Authors

  • Regina Jitschin,

    Corresponding author
    1. Department of Medicine, Karolinska Institutet, Hematology Center, Karolinska University Hospital, Stockholm, Sweden
    • Division of Clinical Immunology, Karolinska University Hospital Huddinge, 14186 Stockholm===

    Search for more papers by this author
    • Telephone: 46-85-858-361; Fax: +46 (0)8-585-836-05

  • Dimitrios Mougiakakos,

    1. Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet, Stockholm, Sweden
    2. Department of Internal Medicine 5, Hematology and Oncology, University of Erlangen-Nuremberg, Erlangen, Germany
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Lena Von Bahr,

    1. Department of Medicine, Karolinska Institutet, Hematology Center, Karolinska University Hospital, Stockholm, Sweden
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Simon Völkl,

    1. Department of Internal Medicine 5, Hematology and Oncology, University of Erlangen-Nuremberg, Erlangen, Germany
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Guido Moll,

    1. Department of Medicine, Karolinska Institutet, Hematology Center, Karolinska University Hospital, Stockholm, Sweden
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Olle Ringden,

    1. Center for Allogeneic Stem Cell Transplantation and Division of Therapeutic Immunology, Karolinska University Hospital, Stockholm, Sweden
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Rolf Kiessling,

    1. Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Stig Linder,

    1. Department of Oncology and Pathology, Cancer Center Karolinska, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.

  • Katarina Le Blanc

    1. Department of Medicine, Karolinska Institutet, Hematology Center, Karolinska University Hospital, Stockholm, Sweden
    Search for more papers by this author
    • Author contributions: R.J. and D.M.: designed the study, performed research, analyzed data, wrote the manuscript; L.V.B.: collected material, provided clinical information; S.V. and G.M.: helped analyzing data; O.R. designed the clinical study, helped writing the manuscript; R.K.: helped designing the experiments; S.L.: performed research; K.L.B.: designed the study, analyzed data, wrote the manuscript. R.J. and D.M. contributed equally to this study.


Abstract

Adoptive transfer of third-party mesenchymal stromal cells (MSCs) has emerged as a promising tool for the treatment of steroid-refractory graft-versus-host disease (GVHD). Despite numerous in vitro studies and preclinical models, little is known about their effects on the patients' immune system. We assessed immune alterations in the T-cell, B-cell, natural killer cell, dendritic cell, and monocytic compartments of steroid-refractory GVHD patients 30, 90, and 180 days after MSC (n = 6) or placebo (n = 5) infusion, respectively. Infused MSCs were bioactive as suggested by the significant reduction in epithelial cell death, which represents a biomarker for acute GVHD. There were several indications that MSCs shift the patients' immune system toward a more tolerogenic profile. Most importantly, infusion of MSCs was associated with increased levels of regulatory (forkhead box P3 (FOXP3)+ and interleukin (IL)-10+) T-cells, reduced pro-inflammatory IL-17+ T(Th17)-cells, and skewing toward type-2 T-helper cell responses. Furthermore, IL-2, which has been recently shown to exert a positive immune modulating effect in GVHD patients, was higher in the MSC patients at all evaluated time points during 6 months after MSC-infusion. Overall, our findings will contribute to the refinement of monitoring tools, for assessing MSC treatment-efficacy and increase our understanding regarding the MSCs' in vivo effects. STEM Cells 2013;31:1715–1725

Introduction

Allogeneic hematopoietic stem-cell transplantation (allo-HSCT) remains the most successful immune therapy and is the only curative option for certain hematological diseases. The donor's immune cells mediate graft-versus-tumor (GVT) effects but also cause graft-versus-host disease (GVHD) [1]. Despite the efforts to improve treatment, GVHD is still the main allo-HSCT complication. Steroids are the first-line treatment with a response rate of 30–50% but the prognosis is dismal in steroid-refractory patients. Thus, development of novel intervention strategies is necessary.

Mesenchymal stromal cells (MSCs) have emerged as an interesting candidate for cellular therapy in GVHD patients. MSCs are multipotent stromal cells that are readily expandable from a bone marrow aspirate [2]. In addition to having a plethora of regenerative and immunoregulatory properties, MSCs are hypoimmunogenic, allowing transfer across human leukocyte antigen barriers, which is a prerequisite for an “off-the-shelf” cellular therapy [2].

Promising observations from the first applications of MSCs in GVHD patients and the relative complication-free procedure [3] have led to the initiation of numerous clinical trials [4]. Despite our vast knowledge of the effects of MSCs on immune cells in vitro and the data from preclinical models [2], few reports have shed light on the effects of MSCs on the patient's immune system [5]. Such knowledge is essential for the treatment of GVHD as it is an immune-mediated reaction involving T-cells [1], natural killer (NK)-cells [6], dendritic cells (DCs) [7], and monocytes [8], which could all be impacted by MSCs [2]. An improved understanding of the effects of MSCs in vivo might help us to design future clinical studies and to establish novel monitoring procedures to reproducibly assess the efficacy of MSC treatment.

Therefore, we decided to investigate alterations in the different immune compartments in patients with GVHD after treatment with third-party MSCs. These patients were compared with a cohort of control GVHD patients who received a placebo infusion.

Materials and Methods

Patients

The study protocol was approved by the ethics committee of Karolinska University Hospital, Stockholm, Sweden. It was performed in accordance with the declaration of Helsinki and all participants treated at the Karolinska University Hospital Huddinge (2008–2010) gave written, informed consent. We analyzed cryopreserved samples from 11 patients with steroid-refractory GVHD (n = 9) and/or hemorrhages (n = 2) who received either a single infusion of third-party MSCs (n = 6) or of placebo (n = 5). Steroid refractoriness was diagnosed if steroid treatment (≥2 mg per kg per day) led to no improvement in clinical symptoms after at least 7 days, or in the case of GVHD progression by at least one grade, within 72h. Peripheral blood samples were retrieved at day (d)+30, d+90, and d+180 after infusion of MSCs. Serum samples were collected immediately before infusion and on d+7, d+30, d+90, and d+180 after infusion. Time-point samples were retrieved relative to the date of allo-HSCT; they were not significantly different between groups (MSC treatment: 220 ± 120 days after allo-HSCT; control: 150 ± 70 days after allo-HSCT; p = .66). Patient characteristics are given in Table 1 and Table 2.

Table 1. Patients characteristics; allogeneic stem cell transplantation
Patient numberAge (years)SexDiagnosisMatchCell sourceConditioning regimen
  1. Abbreviations: AML, acute myeloid leukemia; BM, bone marrow; BU, busulfan; CLL, chronic lymphatic leukemia; CMML; chronic myelo-monocytic leukemia; CY, cyclophoshamide; DCB, double cord blood; F, female; FLU, fludarabine; M, male; MMUD, mismatched unrelated donor; MRD, matched related donor; MUD, matched unrelated donor; PBSC, peripheral blood stem cells.

165FAMLMRDPBSCFLU + BU
244MAMLMMUDDCBBU + CY
366MCMMLMRDPBSCFLU + BU
448MAMLMUDPBSCBU + CY
527MHodgkin lymphomaMRDPBSCFLU + CY
665MMyelofibrosisMUDPBSCFLU + BU
760MMyelofibrosisMUDBMFLU + BU
856MCLLMUDBMFLU + CY
951MAMLMUDPBSCBU + CY
1029FImmune deficiencyMRDBMBU + CY
1147MAMLMRDPBSCBU + CY
Table 2. Patient characteristics; immunosuppression, indications, and response
Patient numberProphylaxisIndicationGroupIS at infusionOther ISResponse/durationcGVHD
  1. Abbreviations: aGVHD, acute graft-versus-host disease; ATG: Antithymocyte globulin; Campath: Alemtuzumab; cGVHD, chronic graft-versus-host disease; CSA, cyclosporin A; ECP, extracorporeal photochemotherapy; GI, gastrointestinal; IS, immunosuppression; MSC, mesenchymal stromal cell; Mtx, methotrexate; other IS, immunosuppression before or within 90 days of MSC/placebo; pred: Prednisolon; Siro, Sirolimus; Tacro, Tacrolimus.

1CSA + MtxaGVHD skin + gut III°placeboSteroid + CSAnoComplete/ContinousYes, severe
2CSA + pred + ATGaGVHD gut II°placeboSteroid + Tacro + SironoComplete/ContinousNo
3CSA + MtxaGVHD liver + skin II°MSCSteroid + CSATacroPartial/cGVHDYes, severe
4CSA + Mtx + ATGaGVHD gut II°placeboSteroid + CSAnoComplete/8 daysNo
5CSA + MtxaGVHD liver II°MSCSteroid + CSA + Mtx + ECPnoNoneYes, severe
6CSA + Mtx + ATGaGVHD gut III°placeboSteroid + CSA + MtxnoPartial/cGVHDYes, moderate
7CSA + Mtx + ATGGI-hemorrhageMSCSteroid + CSAnoComplete/ContinousNo
8Tacro+ Siro + ATGaGVHD gut III°MSCSteroid + TacroSiroComplete/ContinousYes, mild
9CSA + Mtx + CampathaGVHD skin + gut II°MSCSteroid + CSAnoComplete/ContinousYes, moderate
10CSA + Mtx + ATGHemorrhagic cystitisMSCCSAnoComplete/ContinousNo
11CSA + MtxaGVHD gut III°placeboSteroid + CSA + MtxnoNoneYes, severe

Mesenchymal Stromal Cells

MSCs were isolated from the bone marrow of unrelated unmatched donors as described previously in detail [4]. MSCs expressed CD73, CD90, and CD105 but not CD3, CD14, CD31, CD34, CD45, and HLA-DR. They differentiated into cartilage, bone, and fat using special culture conditions and inhibited mixed lymphocyte cultures in vitro. All cells were cultured for three passages and were cryopreserved before infusion. Release criteria for MSCs included viability of >95%, absence of visible clumps, and sterility. In average 2 million cells per kg bodyweight (range: 1.5–2.2) were infused during 15–20 minutes.

Reagents

RPMI 1640 cell medium, fetal bovine serum, and trypan blue (0.4%) were purchased from (Invitrogen, Lifetechnologies, Carlsbad, CA, www.lifetechnologies.com). Dimethyl sulfoxide (DMSO), penicillin-streptomycin, ionomycin, phorbol 12-myristate 13-acetate (PMA), and Brefeldin A were purchased from (Sigma Aldrich, St. Louis, MO, www.sigmaaldrich.com).

Isolation of Cells and Serum Sampling

Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized peripheral blood using density gradient-based methods (Lymphoprep; Medinor AB, Lidingö Sweden, www.medinor.se), subsequently frozen in 10% DMSO, and stored in liquid nitrogen until further analysis. Serum samples were immediately ultracentrifuged and transferred to −80°C until further use.

Antibodies and Flow Cytometry

Cells were stained using fluorochrome-coupled monoclonal antibodies as detailed in Supporting Information Table 1. For intracellular staining, cells were fixed and permeabilized using Cytofix/Cytoperm reagents (BD Biosciences, San Jose, CA, www.bdbiosciences.com). For cytokine staining, cells were stimulated with PMA/ionomycin in the presence of Brefeldin A for 6 hours and then fixed. Acquisition and analysis was carried out using an LSRII flow cytometer (BD Biosciences) and FlowJo Version 9.0.2 software (TreeStar, Ashland, OR, www.flowjo.com).

Cytokine Analysis

Serum concentrations of interleukin (IL)-1α, IL-2, IL-4, IL-6, IL-7, IL-15, IL-12, interferon (IFN)-γ, IFN-α, granulocyte colony-stimulating factor (G-CSF), and tumor necrosis factor (TNF)-α were determined using a multiplex cytokine assay (Millipore, Billerica, MA, www.millipore.com) on a Luminex machine (Luminex Austin, TX, www.luminexcorp.com).

M30/M65 ELISA

Caspase-cleaved K18 was measured using the sandwich enzyme-linked immunosorbent assay (ELISA) M30-Apoptosense ELISA (Peviva AB, Bromma, Sweden), whereas total K18 was measured with the M65 EpiDeath ELISA (Peviva AB). Measurements were performed using 25 μL serum per assay in duplicates. Concentrations were calculated to U/L using a recombinant K18 standard (Peviva).

Statistical Analysis

Differences in means and correlation analyses were evaluated using Wilcoxon–Mann-Whitney and Spearman's rank correlation tests, respectively. All statistical analyses were performed using GraphPad Prism version five (GraphPad Prism Software Inc.) at a significance level of p ≤ .05.

When a great number of tests are performed within one study, there is a risk that some of the tests are significant by chance and the problem of mass-significance has to be dealt with. The study included 60 statistical tests, which means that 0.05 × 60 = 3 tests may be significant by chance at the 5% level. We found that 18 tests were significant at the 5% level and four tests at the 0.01% level. No significance would be expected to be at random for the 0.01% level.

Results

MSC Infusion Reduces Epithelial Cell Death

Recent evidence suggests that progressive epithelial apoptosis is involved in the therapy-refractoriness of GVHD [9]. We therefore measured caspase-cleaved cytokeratin (ccCK) 18, which is indicative of epithelial apoptosis, and uncleaved, K18, which is indicative of total epithelial death (i.e., necrosis and apoptosis)—as GVHD biomarkers in the patients' sera before and both 7 and 30 days after infusion of MSCs or placebo (Fig. 1A). On d+30, levels of both ccK18 (p = .008) and total K18 (p = .039) were significantly reduced (by ∼40%) in patients who were treated with MSCs. In contrast, levels remained stable (ccK18) or even increased (K18) in the placebo (control) group. Relative changes (i.e., increase or decrease) in ccK18 and K18 correlated significantly in the individual samples, suggesting that there was no significant shift between epithelial apoptosis and necrosis (Fig. 1B). This observation was confirmed by analyzing the ccK18/K18 ratios at the different time points in both groups; this did not show any significant alteration (data not shown).

Figure 1.

Epithelial cell death markers in the serum of patients with graft-versus-host disease (GVHD), following treatment with mesenchymal stromal cell (MSCs) or placebo. (A): Alterations in serum levels of caspase-cleaved keratin 18 (ccK18; M30 Apoptosense; a marker of epithelial apoptosis) and total keratin 18 (K18; M65 EpiDeath; a marker of epithelial necrosis and apoptosis) in GVHD patients before and after the infusion of third-party MSCs (green) or placebo (red) (at days +7 and +30, respectively). Values were normalized against the pretreatment levels (set as 100%). (B): Correlation of changes in ccK18 and K18 levels in all samples are relative to pretreatment. *, p < .05; **, p < .01. Abbreviations: ccK18, caspase-cleaved keratin 18; K18, total keratin 18; MSC, mesenchymal stromal cell.

a Higher CD4+/CD8+ T-Cell Ratio in the MSC-Treated Group

T-cells are central for GVHD and GVT effects [10]. The T-cell frequency was not significantly affected by MSC infusion at the time points analyzed (Fig. 2A, 2B). However, CD4/CD8 T-cell ratio was by tendency at d+30 (p = .056) and significantly at d+90 (p = .034) skewed toward CD4+ T-cells in patients receiving MSCs (d+30: 0.64 ± 0.20 vs. 0.12 ± 0.03; d+90: 0.64 ± 0.19 vs. 0.13 ± 0.02) (Fig. 2B). The CD4 and CD8 double negative (DN) T-cells (Fig. 2A) possess regulatory functions beneficial for GVHD [11]. Infusion of MSCs had no impact on the DN T-cell frequency (Supporting Information Figure A).

Figure 2.

T-Cell homeostasis after the infusion of third-party mesenchymal stromal cells (MSCs) or placebo. (A): This representative flow cytometric (fluorescence-activated cell sorting [FACS]) analysis (dot plot) shows our gating strategy for the main T-cell subpopulations: (i) CD3+ T-cells were identified and subclassified (ii) in CD4+, CD8+, and CD4/CD8 double negative (DN) T-cells. Samples from days +30, +90, and +180 after infusion with MSCs (green) or placebo (red) were analyzed. The gray area represents the range (fifth and 95th percentile) of 10 HDs with the mean value depicted as a horizontal line. (B): T-cell frequencies and the CD4+/CD8+ T-cell ratio were assessed in the MSC group and the placebo (control) group as well as healthy donors (HD). (C): Naïve (CD45RA+/NA) and memory (CD45RO+/MEM) cells were defined by FACS in the CD4+ and CD8+ T-cell population, respectively. Memory T-cells were further characterized as CD45RO+CCR7+ central memory (CM) cells or CD45RO+CCR7 effector memory (EM) cells. CD45RA+CCR7 cells were identified as CD45RA+ EM cells, so-called EMRA T-cells. (D, E): The mean values of the naïve (CD45RA+) and memory (CD45RO+) CD4+ and CD8+ T-cell frequencies in MSC group and the control group were calculated and compared for all the time points. Blue bar represents calculated mean values for 10 HD. (F, G): Similarly, CM, EM, and EMRA CD4+ and CD8+ T-cell frequencies were comparatively assessed. (H): Thymic output of CD4+ cells, defined as the proportion of CD31+CD45RA+ recent thymic emigrants (RTEs) among CD4+ T-cells were evaluated by FACS. (I): Peripheral T-cell expansion was quantified by assessing frequencies of Ki-67+ CD4+ and CD8+ T-cells by FACS. (J): The serum levels of interleukin (IL)-2 and IL-7, as quantified in both groups by multiplex ELISA. (The horizontal line represents the mean value of 22 healthy controls, the gray area their range (fifth and 95th percentile). *, p < .05; **, p < .01. Abbreviations: CM, central memory; DN, double negative; EM, effector memory; EMRA, CD45RA+CCR7 effector memory; FSC, forward scatter; HD, healthy donors; IL-2/7, interleukin-2/7; MEM, memory; MSC, mesenchymal stromal cell; NA, naive; PBMCs, peripheral blood mononuclear cells; RTE, recent thymic emigrant.

No significant differences regarding the CD45RA+ naïve and CD45RO+ memory subsets of both CD4+ and CD8+ T-cells were found (Fig. 2C–2E). The distribution of CD45RO+CCR7+ central and CD45RO+CCR7 effector memory T-cells was comparable for CD4+ (Fig. 2F) and CD8+ T-cells (Fig. 2G) in both groups. The most differentiated type of memory T-cell, the CD45RA+CCR7 effector memory (EMRA) cell (Fig. 2F), did not show any differences within the CD4+ T-cell compartment. Notably, MSC-treated patients showed a tendency of increased CD8+ EMRA T-cells on d+30 (26.91 ± 5.4% vs. 17.97 ± 5.09%), which reached statistical significance by d+180 (35.95 ± 5.40% vs. 14.92 ± 5.02%) (Fig. 2G). We noticed a transient decrease in thymic CD4+ T-cell output (2.45 ± 0.38% vs. 9.62 ± 1.20% CD45RA+CD31+ recent thymic emigrants (RTEs) on d+30, p = .029), which subsequently equalized at d+180 (Fig. 2H). The proportion of proliferating Ki67+ CD4+ and CD8+ T-cells was comparable (Fig. 2I). The common γ-chain cytokines are important for T-cell homeostasis. As shown in Figure 2J, IL-2 serum levels were significantly higher in MSC-treated patients at all time points tested, especially after 3 months (9.04 ± 8.89 pg/ml vs. 0.83 ± 1.22 pg/ml, p = .008). IL-7 also appeared to be elevated on d+90 (12.07 ± 3.30 pg/ml vs. 4.96 ± 4.25 pg/ml) but, overall, was characterized by high inter-individual variability (Fig. 2J). There was no correlation between the IL-2 (r = 0.01, p = .96) and IL-7 levels (r = −0.097, p = .54) and the T-cell frequency.

Treatment with MSCs Is Associated with a Lower Late T-Cell Activation and Skewing Toward TH2 Responses

Frequencies of early (CD69+) and late (HLA-DR+) activated CD4+ and CD8+ T-cells were compared in both groups (Fig. 3A). Overall, our data suggested that patients treated with MSCs have lower levels of T-cell activation than the controls during the period between d+30 and d+180. This was reflected by a significantly lower proportion of HLA-DR+ CD4+ T-cells, on d+90 (28.69 ± 5.79% vs. 49.70 ± 9.87%, p = .040) and the same tendency on d+180 (22.50 ± 5.15% vs. 40.87 ± 11.00%) (Fig. 3C). Similarly and although not statistically significant, the frequency of CD69+ CD8+ T-cells appeared to be lower on d+30 (5.84 ± 2.85% vs. 11.10 ± 1.91%) and on d+90 (6.51 ± 1.59% vs. 10.75 ± 2.65%) (Fig. 3D). Furthermore, HLA-DR+ CD8+ T-cells tended to be lower on d+90 (36.65 ± 8.04% vs. 54.45 ± 11.11%), reaching significance (p = .032) by d+180 (13.32 ± 4.89 vs. 37.80 ± 11.38%) (Fig. 3E).

Figure 3.

Levels of T-cell activation and TH1/TH2-typical serum cytokines after infusion with third-party mesenchymal stromal cells (MSCs) or placebo. (A): A representative dot plot analysis of flow cytometric (fluorescence-activated cell sorting) data showing early (CD69+) and late (HLA-DR+) activation of gated CD4+ or CD8+ T-cells. Samples were taken at days +30, +90, and +180 after infusion with MSCs (green) or placebo (red). The gray area represents the value range of healthy donors (n = 10) with the arithmetic mean shown as a horizontal line. (B, C): Frequencies of CD69+ T-cells (panel B) and HLA-DR+ CD4+ T-cells (panel C) were assessed in all groups. Similarly, CD69+ T-cells (panel D) and HLA-DR+ CD8+ T-cells (panel E) were evaluated. (F): The prototypic TH1 and TH2 serum cytokines interferon (IFN)-γ and interleukin (IL)-4, respectively, were quantified by multiplex enzyme-linked immunosorbent assay (ELISA) and the IFN-γ/IL-4 ratio calculated. (G): The serum levels of IL-1α, IL-6, and TNF-α as quantified in both groups by multiplex ELISA. The horizontal line represents the mean value of 22 healthy controls, the gray area the value range (fifth and 95th percentile). *, p < .05. Abbreviations: HLA-DR, human leucocyte antigen-dr; IFN-γ, interferon-γ; IL-1α/6, interleukin-1α/6; MSC, mesenchymal stromal cell; TNFα, tumor necrosis factor α.

The ratio of IFN-γ, a prototypic TH1 cytokine, to IL-4, a prototypic TH2 cytokine, was significantly higher in the control group at all tested time points (p = .029, p = .050, and p = .036, respectively) (Fig. 3F) indicating an MSC-associated skewing toward TH2 responses, which is in line with previous observations [12]. Pro-inflammatory cytokines TNF-α and IL-1α did not differ significantly between both groups (Fig. 3G). IL-6, an important pro-inflammatory cytokine and at the same time effector of several MSC-mediated effects [2], was at d+180 significantly (p = .015) higher in the MSC group (20.85 ± 14.94 pg/ml vs. 1.04 ± 0.92 pg/ml). However, it should be noted that these levels were not increased as compared with the healthy donors (HDs) tested (Fig. 3G).

Increased Regulatory T-Cell Levels in MSC-Treated Patients

Treg-cells have been shown to be involved in the control of alloreactive T-cells during GVHD [13]. In fact, MSCs promote in vitro induction of various Treg-cell subsets including naturally occurring(-like) Treg- and IL-10+ Tr1-cells [13, 14]. First, we analyzed the frequency of naturally occurring(-like) Treg-cells, as shown in Figure 4A. These cells were identified as CD4+CD25med-hiCD127lo T-cells. Treg-cells identified based on this phenotype were confirmed by assessing the intracellular FOXP3-levels (Fig. 4A).

Figure 4.

Induction of regulatory T-cell populations after the infusion of third-party mesenchymal stromal cells (MSCs). (A): Regulatory T-cells (Treg) were defined by flow cytometry (fluorescence-activated cell sorting) as CD25+CD127low CD4+ T-cells. The majority of this population were FOXP3+, as shown in the representative histogram and the cumulative data comparing the frequency of FOXP3+ cells among (CD25+CD127low) Treg and conventional T-cells (Tconv). Cells from days +30, +90, and +180 after infusion of MSCs (green) or placebo (red) were analyzed as well as of 10 healthy donors (range of values (fifth and 95th percentile) shown as a gray area, mean value as a horizontal line). (B): The frequency of circulating Treg-cells was studied in both the groups and at all time points (C): interleukin (IL)-2 levels were correlated with Treg-cells frequencies. (D): Thymic output of Treg cells defined as the proportion of CD31+CD45RA+ recent thymic emigrants (RTEs) among Treg cells. (E): Proportions of CD45RA+ naïve and CD45RO+ memory Treg-cells were assessed in the MSC (green) and placebo (red) group as well as in healthy donors (HD/blue) (n = 10). (F): Immunoregulatory IL10+ CD4+ T-(Tr1-) cells were quantified in all samples and compared between the two groups. (G): The frequency of TH17 cells, defined as IL-17A+ CD4+ T-cells, was investigated in the MSC group and the control group. (H): The Treg/TH17 cell ratio was then calculated. The horizontal line represents the mean value of the according parameter analyzed in healthy donors (n ≥ 10) with the range of values (fifth and 95th percentile) shown in gray. *, p < .05; **, p < .01; ***, p < .001. Abbreviations: FOXP3, forkhead box P3; HD, healthy donors; IL-2, interleukin-2; MEM, memory; MSC, mesenchymal stromal cell; NA, naive; RTE, recent thymic emigrant; Tconv, conventional T-cells; Treg, regulatory T-cell.

The frequency of CD4+CD25med-hiCD127loFOXP3+ Treg-cells was significantly higher in the group of MSC-treated patients on d+30 (9.45 ± 0.68% vs. 5.02 ± 0.33%) (p = .003) and on d+90 (7.59 ± 1.31% vs. 3.62 ± 0.85%) (p = .037). Over time, levels in the MSC group and the control group approached (d+180: 7.26 ± 1.34% vs. 5.21 ± 1.11%) (Fig. 4B). Treg-cells are primarily maintained by IL-2 [15] and we found a positive correlation between Treg-cell and IL-2 levels (Fig. 4C) (r = 0.41, p = .02). The proportions of naïve and memory Treg-cells as well as the RTEs were similar in both groups (Fig. 4D, 4E). Observations from preclinical models have indicated that CD62Lhi Treg-cells are more potent in controlling immune responses in GVHD [16]. However, we could not detect any alterations in CD62L associated with MSC treatment (Supporting Information Figure B). The frequency of Tr1-cells was similarly elevated in the MSC group by tendency on d+30 (1.96 ± 0.78% vs. 0.58 ± 0.26%) (p = .054) and significantly on d+90 (2.19 ± 0.47 vs. 0.89 ± 0.31%) (p = .036) (Fig. 4F).

Recent studies have shown that pro-inflammatory IL-17+ CD4+ T (TH17)-cells are involved in GVHD and that the Treg/TH17 cell ratio might represent a sensitive biomarker of GHVD [17]. MSCs inhibit TH17-cell function and differentiation [2]. In accordance with the already described MSC-mediated effects, TH17-cell frequency was lower and the Treg/TH17-cell ratio higher in MSC-treated patients, both on d+30 (1.60 ± 0.22 vs. 3.86 ± 1.01, p = .032 and 6.48 ± 1.47 vs. 1.81 ± 0.47, p = .008) and on d+90 (2.2 ± 0.50 vs. 4.22 ± 1.57 p = .058 and 4.96 ± 2.35 vs. 0.64 ± 0.25, p = .024) (Fig. 4G, 4H). Notably, TH17-cells were in both patient groups significantly higher as compared with HDs (0.09 ± 0.06%).

MSCs Have No Apparent Effect on the Reconstitution of Myeloid Cells

To evaluate potential alterations in myelopoiesis associated with MSC treatment, we analyzed DCs and monocytes. DCs were defined as lineage-HLA-DR+ cells and further subdivided into CD11c+ myeloid (mDCs) and CD123+ plasmacytoid DCs (pDCs), respectively (Fig. 5A). Low numbers of circulating DCs have been associated with an increased risk of disease relapse and acute GVHD [7]. We found that both mDCs and pDCs tended to be elevated in MSC-treated patients, both on d+30 (2.87 ± 2.07% vs. 1.27 ± 0.68% and 0.26 ± 0.11% vs. 0.03 ± 0.01%, respectively) and on d+90 (3.29 ± 0.94% vs. 1.42 ± 0.78% and 0.38 ± 0.10% vs. 0.15 ± 0.05%) (Fig. 5B). However, ratios of the mDC and pDC frequencies and of their prototypic cytokines—IL-12 (mDCs) and IFN-α (pDCs), respectively—were comparable (Fig. 5C). Surface density of CD83 (a maturation marker) and CD86 (a co-stimulatory molecule) on pDCs and mDCs did not vary significantly between the MSC and control groups, but showed a remarkable individual biological variance (Supporting Information Figure C–D). Next, we assessed monocytes (Supporting Information Figure E). Circulating monocyte levels (Fig. 5D) and their HLA-DR expression density (Supporting Information Fig. 5F), which correlates with monocyte activation, were not significantly different between groups. Serum concentration of G-CSF, one of the main inherent factors promoting myelopoiesis, was also comparable in both groups (Fig. 5E).

Figure 5.

Dendritic cell and monocyte reconstitution following application of mesenchymal stromal cells (MSCs). (A): A representative analysis of the flow cytometric (fluorescence-activated cell sorting) assessment of dendritic cells (DCs). First, lineage-HLA-DR+ cells were defined as DCs and subsequently subdivided into the major subsets of CD11c+ myeloid (mDCs) and CD123+ plasmacytoid DCs (pDCs). Samples taken on days +30, +90, and +180 after infusion with MSCs or placebo were analyzed. Range of values (fifth and 95th percentile) measured in 10 healthy donors is depicted as a gray area, the arithmetic mean as a horizontal line. (B): The frequencies of mDCs and pDCs were evaluated in both groups. (C): The mDC/pDC ratio was calculated and compared, together with the ratio of serum interleukin (IL)-12 (typically released by mDCs) and interferon (IFN)-α (typically released by pDCs). (D): The frequency of CD14+ monocytes was evaluated in the MSC (green) and the control (red) group. The horizontal line represents the mean value for monocytes in healthy donors (n = 36) with the range of values (fifth and 95th percentile) shown in gray. (E): Serum levels of the granulocyte colony-stimulating factor (G-CSF), one of the key intrinsic myelopoietic cytokines, were quantified by multiplex enzyme-linked immunosorbent assay. Range of values (fifth and 95th percentile) measured in healthy donors (n = 19) is depicted as a gray area, the arithmetic mean as a horizontal line. Abbreviations: G-CSF, granulocyte colony-stimulating factor; HLA-DR, human leucocyte antigen-dr; IFN-α, interferon-α; IL-12, interleukin-12; mDC, myeloid dendritic cell; PBMCs, peripheral blood mononuclear cells; pDC, plasmacytoid dendritic cell.

MSCs Have Little Effect on NK Cells and B-Cells

NK-cells play a pivotal role in the GVT reaction following allo-HSCT [18]. Furthermore, there is increasing evidence to suggest that NK-cells are also positively involved in the control of GVHD [19]. NK-cell frequency declined (normalized) in both groups over time and showed a transient peak in the control cohort on d+90 (Fig. 6A). At this time point, MSC-treated patients had by tendency lower NK-cell levels (11.41 ± 4.26% vs. 23.06 ± 5.63%, p = .0507) and a higher serum concentration of IL-15 (19.06 ± 6.59 pg/ml vs. 11.72 ± 7.83 pg/ml, p = .171). Together with IL-2 and IL-7, IL-15 belongs to the common γ-chain cytokines and is of special importance for NK-cell homeostasis [20]. There was no correlation between IL-15 levels and levels of either T-cells or NK-cells (data not shown). When analyzing the main NK-cell subsets (Fig. 6B), we found that the proportion of (activated) CD56bright NK-cells [21] was significantly lower in the MSC-treated patients than in the controls on d+90 (33.22 ± 3.88% vs. 48.92 ± 9.38%, p = .046). We also evaluated NK-cell activation markers (i.e., CD25, CD69, and HLA-DR) and no significant differences were found between the two groups (Supporting Information Figure G).

Figure 6.

Frequency and activation of NK-cells after infusion with third-party mesenchymal stromal cells (MSCs) or placebo. (A): Frequencies of natural killer (NK) cells were assessed at days +30, +90, and +180 after infusion with third-party MSCs (green) or placebo (red). Concordantly, the serum levels of interleukin (IL)-15 (in pg/ml), a key cytokine in NK-cell development, was assessed in both groups. The gray area represents the value range from the fifth to the 95th percentile of 15 healthy controls and the horizontal line the arithmetic mean. (B): The main NK-cell subsets were analyzed by flow cytometry (fluorescence-activated cell sorting) as depicted in the representative analysis. NK-cells were grouped into CD56brightCD16−/dim and CD56dimCD16+ cells. The proportions of both NK-cell subsets were assessed in both patient groups as well as 10 healthy donors (HD/blue). (C): The frequency of CD19+ B-cells was assessed in both groups at day +30, +90, and +180 as well as 10 healthy donors (horizontal line depicts mean value, gray area the range from the fifth to the 95th percentile). Immunoglobulin (Ig) levels for IgA, IgG, and IgM were measured in both groups before (in average 44 ± 14 days prior infusion) and after (in average 56 ± 22 days post infusion) MSC/placebo application. Data is presented as the percental change in Ig levels relative to the pre-infusion baseline (set as 100%). *, p < .05. Abbreviations: HD, healthy donors; Ig, immunoglobulin; IL-15, interleukin-15; MSC, mesenchymal stromal cell; PBMCs, peripheral blood mononuclear cells.

Increasing evidence suggests that B-cells, which can potentially be controlled by MSCs [2], are also involved in the pathogenesis of GVHD [22]. Both MSC- and placebo-treated patients exhibited low B-cell frequencies that steadily increased over time (Fig. 6C). The immunoglobulins (Ig) including IgA, IgG and IgM were similarly reduced in both groups (about ∼50%) as compared to the levels prior MSC/placebo infusion; a reduction that is regularly observed in patients with GVHD in the course after transplantation [23].

Discussion

Here, we present a comprehensive “immunome”-analysis of patients treated with third-party MSC or placebo infusions after allo-HSCT. Most data indicate that transferred MSCs diminish early [24] after eliciting an immediate response (by e.g., releasing cytokines), which initiates a cascade resulting in a longer-lasting immune modulation [2]. We therefore assessed humoral alterations at earlier time points (d+7 and d+30 after MSC infusion) and cellular parameters later on (d+30, d+90, and d+180 after MSC infusion). To objectify the effects elicited by the MSC transfer, we evaluated the epithelial K18 serum levels before and after MSC treatment. Both the cleaved and the uncleaved form of K18 are indicators of epithelial cell death and validated biomarkers for acute GVHD, as they correlate with the extent of organ damage [9]. We observed a significant decline in both molecules in the MSC group, but not in the placebo group, which in our opinion reflects the biological activity of the MSCs (Fig. 1). These data are in line with recent studies evaluating the responsiveness of serum biomarkers of GVHD (e.g., interleukin [IL] two receptor) to an adoptive MSC transfer [3].

After confirming the biological activity of our MSC intervention, we concentrated on the T-cell compartment. T-cells represent the main mediator of GVH immune reactions. At the same time, T-cell reconstitution following allo-HSCT is central for re-establishing an efficient immunity and the desired GVT effect [25]. An ideal immune intervention for ameliorating GVHD would therefore lead to suppression of allo-reactivity while preserving protective immunity. T-cell frequencies were similar in both groups (Fig. 2B, 2D, 2E). In line with the proposed T-cell regulating function of MSCs [2], we observed lower proportions of HLA-DR+ CD4+ and CD8+ T-cells—a marker of late T-cell activation (Fig. 3C, 3E). Acute GVHD is considered to be a TH1-driven disease [1]. MSCs re-establish the balance between TH1 and TH2 responses in models of autoimmunity [12] and GVHD [26]. We evaluated IFN-γ and IL-4 as prototypic TH1 and TH2 cytokines, respectively, and observed a significant skewing toward a TH2 response in the MSC group (Fig. 3F), which could constitute a veto-like function for GVHD [27].

Generally, CD8+ T-cells show a faster recovery than CD4+ T-cells after allo-HSCT. This leads to an inverse CD4+/CD8+ T-cell ratio (in healthy individuals >1), which can last more than 5 years. Interestingly, patients who received MSCs had a significantly higher CD4+/CD8+ T-cell ratio, which declined over time and approached the levels in the placebo group on d+180 (Fig. 2B). These observations coincide with preclinical models showing a normalization of the CD4+/CD8+ T-cell ratio after MSC administration [28]. The observed ratio normalization upon MSC treatment cannot be explained based on in vitro findings in which CD4+ and CD8+ T-cells are equally impaired by MSCs [29]. MSCs might be functionally modified (“licensed”) within the allo-reactive GVHD environment and could therefore differently affect CD4+ and CD8+ T-cells [30]. Alterations in the peripheral proliferation or the thymic output as the underlying mechanism were ruled out. In fact, we observed a trend toward a reduced thymic output of CD4+ T-cells in MSC-treated patients (Fig. 2H). In animal models, infused MSCs home to the thymus [31] and it might be speculated that a reduced thymic output reflects their T-cell-suppressive activity.

Common γ-chain cytokines have an important role in T-cell homeostasis [32]. In agreement with other studies [32], IL-7 and IL-15 levels were elevated following allo-HSCT. None of them correlated with T-cell frequency, which indicates that they were not the primary promoters of homeostatic T-cell proliferation [32]. Levels of IL-15 showed higher variability, but tended to be elevated in the MSC group (on d+90 and d+180) (Fig. 6A). IL-15 may contribute to the observed TEMRA CD8+ T-cell accumulation since TEMRA cells can be generated from central memory cells by IL-15 [33]. MSC-treated patients showed a tendency (at d+90) of higher levels of TEMRA CD8+ T-cells; this reached statistical significance by d+180 (Fig. 2G). TEMRA cells represent the most differentiated type of memory cells. CD8+ TEMRA have a low proliferative capacity and a high susceptibility to apoptosis, but they produce cytokines and have cytotoxic activity [34]. CD8+ TEMRA cells have been associated with the development of chronic GVHD after allo-HSCT [33] (which was not confirmed in our limited patient cohort; data not shown).

Remarkably, treatment with MSCs was associated with increased levels of IL-2 at all evaluated time points, whereas they were below normal in the control group (Fig. 2J) [32]. This is in keeping with a study showing MSCs increased IL-2 in mixed lymphocyte cultures [35]. Recent data indicate that pro-inflammatory mediators such as TNF-α, which are abundant during GVHD, can turn MSCs into a promoter of IL-2 production by T-cells [36]. IL-2 is a key cytokine in Treg-cell homeostasis [15]. GVHD is associated with reduced numbers of circulating suppressive Treg-cells [37], and their reconstitution after allo-HSCT confers protection from GVHD [38]. MSCs induce Treg-cells through multiple mechanisms (e.g., PGE2 or HO-1) [14]. On d+30 and d+90, patients who had received MSCs showed transiently increased levels of naturally occurring CD4+CD25med-hiCD127lo Treg-cells (Fig. 4B) and the proportion of Treg-cells among CD4+ T-cells correlated significantly with IL-2 levels (Fig. 4C). These data fit well with the results of a recent study showing that administration of low-dose IL-2 can promote the induction of Treg-cells, thus alleviating GVHD [39]. It remains to be proven whether MSCs, in addition to directly inducing Treg-cells, (indirectly) elicit phenomena (e.g., enhanced release of IL-2) that abet a prolonged Treg-cell induction and/or survival. In addition to the naturally occurring Treg-cells, MSCs promote the induction of IL-10+ Tr1-cells [14]. In animal models, Tr1-cells control GVHD by promoting long-term tolerance [40]. Patients treated with MSCs had increased levels of Tr1-cells on d+30 and d+90, similarly to the naturally occurring Treg-cell kinetics (Fig. 4F).

TH17-cells are pro-inflammatory cells that are important for the mucosal immunity [41]. In GVHD, TH17-cells expand and correlate with disease severity [42]. MSCs suppress TH17-cell function (i.e., the production of IL-17 and IL-22) and the de novo differentiation of TH17-cells. Moreover, MSCs re-program TH17-cells into Treg-cells [2]. Interestingly, IL-2 signaling also has a role in the reciprocal balance between Treg- and TH17-cells [43] by inhibiting IL-6-mediated STAT3 activation, which is central to TH17-cell development [44]. In line with the current models on MSC function [2] and with the previously mentioned observations regarding the patients' serum IL-2 levels, we observed lower levels of TH17-cells (Fig. 4G) and an increased Treg/TH17 ratio (on d+30 and d+90) in the MSC group (Fig. 4H).

Recovery of DCs [7], monocytes [45], and NK-cells [6] after allo-HSCT has been linked to a better survival. Insufficient myeloid and plasmacytoid DC reconstitution increases the risk of disease relapse and GVHD [7]. On the other hand, GvHD, as well as steroids, impair DC reconstitution [46]. MSCs suppress DC differentiation and function [2]. However, in the MSC group, mDC and pDC frequency appeared to be increased at the early time points (d+30 and d+90) without any shift in the mDC/pDC ratio or phenotypical alterations (Fig. 5A–5C; Supporting Information Figure C-D). Monocytes reconstitute early after allo-HSCT, leading to a transient monocytosis [8]. We could not detect any MSC-associated alterations regarding monocyte frequency or HLA-DR expression, which is a good indicator of the state of activation and of the ability to present antigens (Supporting Information Figure F). However, functional tests (e.g., T-cell suppression)—necessary to evaluate possible qualitative alterations were not performed—due to the limited number of cells.

MSCs suppress in vitro function and proliferation activated NK-cells [2]. NK-cell frequencies in both the groups showed comparable kinetics except for d+90, when the controls had by tendency higher NK-cell levels (Fig. 6B). This discrepancy is most likely explained by a higher proportion of cytokine-producing CD56bright NK-cells in the control group [47] (Fig. 6B). The CD56bright NK-cell expansion is a key feature of NK-cell reconstitution after allo-HSCT [48]. Homeostasis is a cytokine-driven process and CD56bright NK-cells are sensitive to IL-2 and IL-15. Despite increased IL-2 and IL-15 levels in the MSC group (Fig. 2J, 6A), cell frequencies were reduced.

Overall, and despite numerous primarily in vitro studies indicating a potentially negative impact of MSCs on DCs, monocytes and NK-cells [2] we did not detect any striking alterations during the reconstitution process of the MSC group. However, the data presented must be interpreted with caution because of the small cohorts included in the trial and similar response rates in both the MSC and control group. Nevertheless, and despite the limited sample size, this study possess some unique strengths. It was a randomized, placebo-controlled single-center study and all patients were treated similarly regarding supportive care and basic immunosuppression [49]. Most of the significant immunological alterations found in the MSC group are in accordance with our present knowledge regarding MSCs and—in our opinion—only few of the observed differences may have been significant by chance. Clearly, substantially larger patient cohorts will be necessary in the future to enable us to correlate clinical with immunological responses and to evaluate potential synergistic effects of MSCs and immunosuppressive agents.

Conclusion

Taken together, we provide evidence that infusion with third-party MSCs leads to an objective immune response. The immune alterations, especially induction of Treg-cells [50], reduced TH17-cell numbers [51], and a shift toward TH2-cell responses [52], are in line with findings in preclinical models of inflammatory or autoimmune disorders and further corroborate the concept of an overall tolerogenic MSC function. Some other rather unexpected observations such as the MSC-associated promotion of IL-2 production and/or release are of very high interest and warrant further investigation in vitro.

Notably, most MSC-associated effects (e.g., increased Treg- and Tr1-cell levels) were transient. This might be explained by the missing long-term engraftment of transferred MSCs [3]. When prolonged immune suppression is required, our findings could indicate the need for modifying treatment schemes accordingly by for example using repetitive MSC-applications.

Unfortunately, the limited numbers of patients did not allow us to make any conclusions regarding the treatment-related clinical response. Our findings should be validated in larger patient cohorts, which would also better enable us to understand how immune modulation translates into clinical response and help to improve the design of clinical protocols.

Acknowledgements

K.L.B. is supported by grants from the Cancer Society of Stockholm, the Children's Cancer Foundation, Karolinska Institutet, Stockholm City Council, the Swedish Cancer Society, the Swedish Research Council, the Swedish Society of Medicine, the Tobias Foundation, and VINNOVA. D.M. is supported by a Max-Eder research grant of the Deutsche Krebshilfe and an IZKF Erlangen grant.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Ancillary