Interactions Between Human Mesenchymal Stem Cells and Natural Killer Cells



Mesenchymal stem cells (MSCs) are multipotent progenitor cells representing an attractive therapeutic tool for regenerative medicine. They possess unique immunomodulatory properties, being capable of suppressing T-cell responses and modifying dendritic cell differentiation, maturation, and function, whereas they are not inherently immunogenic, failing to induce alloreactivity to T cells and freshly isolated natural killer (NK) cells. To clarify the generation of host immune responses to implanted MSCs in tissue engineering and their potential use as immunosuppressive elements, the effect of MSCs on NK cells was investigated. We demonstrate that at low NK-to-MSC ratios, MSCs alter the phenotype of NK cells and suppress proliferation, cytokine secretion, and cyto-toxicity against HLA-class I– expressing targets. Some of these effects require cell-to-cell contact, whereas others are mediated by soluble factors, including transforming growth factor–β1 and prostaglandin E2, suggesting the existence of diverse mechanisms for MSC-mediated NK-cell suppression. On the other hand, MSCs are susceptible to lysis by activated NK cells. Overall, these data improve our knowledge of interactions between MSCs and NK cells and consequently of their effect on innate immune responses and their contribution to the regulation of adaptive immunity, graft rejection, and cancer immunotherapy.


Mesenchymal stem cells or marrow stromal cells (MSCs) represent a multipotent adult cellular population, able to contribute to the maintenance and regeneration of multiple tissues [1]. MSCs reside mainly in bone marrow (BM) [2, 3] but have also been isolated from umbilical cord blood [4], adipose [5], and other tissues [3]. Injury models have shown that systemic administration of MSCs is followed by migration and engraftment into the damaged sites [3, 6], whereas in the absence of injury they engraft into a plethora of nonhematopoietic tissues [7].

Besides their regeneration abilities, MSCs possess immunomodulatory functions, being able to suppress immune reactions both in vitro and in vivo in a non–major histocompatibility complex (MHC)—restricted manner. They inhibit several functions of naïve and memory T cells [810], suppress the development of monocyte-derived dendritic cells (DCs) [11], and prolong histoincompatible graft survival in murine [12] and baboon skin allotransplantation models [13]. MSCs are not inherently immunogenic, being unable to elicit allogeneic T-cell responses [14]. They express negligible levels of MHC-class II, low levels of MHC-class I molecules, and no co-stimulatory molecules [15]. Interferon-γ (IFN-γ) induction of MHC-class II does not result in alloreactivity stimulation [16]. Because of these properties, MSCs have been used for the effective treatment of graft-versus-host disease [17].

Natural killer (NK) cells are large granular lymphocytes with innate immune function, exhibiting a critical role in early host defense against infections and cancer. They are characterized by the surface expression of the CD56 antigen and the lack of CD3 [18]. NK cells exert their effector function by release of immunoregulatory cytokines, such as IFN-γ, tumor necrosis factor–α (TNF-α), interleukin (IL)-10, and GM-CSF, as well as numerous chemokines that generate immediate immune responses [19], whereas they exhibit spontaneous cytolytic activity and mediate antibody-dependent cellular cytotoxicity [20]. NK cell killing is regulated by a balance of signals transmitted by activating and inhibitory receptors interacting with specific HLA molecules on the target cells [18]. Besides being able to effectively lyse infected or tumor targets, NK cells possess a prominent immunoregulatory role, affecting both innate and adaptive immunity through the production of a wide range of cytokines and by crosstalk with DCs [21]. Peripheral blood NK cells can be divided into two functionally distinct subpopulations, the CD56dim, which represents the majority of circulating NK cells (>90%), and a minority subset of CD56bright cells. The first is responsible for cellular cytotoxicity, whereas the latter for cytokine production [22].

The interactions between MSCs and cell populations of the immune system other than T cells and DCs, such as NK cells, have been only moderately studied. To this end, Le Blanc's group tested the ability of freshly isolated NK cells to lyse killer immunoglobulin-like receptor (KIR)–ligand mismatched MSCs, and NK-sensitive tumor targets. Both experimental settings showed no interactive effect between the two populations [14]. Furthermore, in a most recent study, Pittenger's group showed that MSCs cause a decrease in IFN-γ secretion by NK cells in coculture experiments [23].

In the present study, we sought for the first time to delineate the effect of MSCs on NK cells, with respect to their phenotype, proliferation, cytotoxic potential, and cytokine secretion during activation and to analyze the underlying mechanisms, thereby extending our knowledge on the immunomodulatory properties of MSCs. Furthermore, we demonstrate that MSCs are not entirely ignored by the immune system, because they represent potential targets for activated NK cells.

Materials and Methods

Isolation of Human MSCs

BM aspirates (3–5 ml) were obtained under local anesthesia from the iliac crests of normal adult donors already typed for their HLA alleles. The study was approved by the Saint Savas Hospital Institutional Review Board, and informed consent was provided according to the Declaration of Helsinki. BM samples were diluted 1:2 in Hanks' balanced salt solution (HBSS; Life Technologies Ltd., Paisley, Scotland,, and mononuclear cells were isolated by Ficoll-Hypaque (Biochrom AG, Berlin, centrifugation, washed thoroughly in HBSS, resuspended, and cultured at 10,000 per cm2 in Minimum Essential Medium alpha (aMEM) with Glutamax (Life Technologies Ltd.) supplemented with 10% fetal calf serum (FCS; Biochrom AG), lot selected for optimal growth of human MSCs, and 50 μg/ml gentamicin (Life Technologies Ltd.). On day 7, nonadherent cells were discarded. Adherent cells were detached with 5–10 minutes' incubation in 0.05% trypsin/EDTA (Life Technologies Ltd.), harvested, counted, and passaged at 100 cells per cm2. Thereafter, cells were passaged at weekly intervals. After three passages, phenotypical and differentiation assays were performed and cells were used for further experiments. MSCs used were positive for SH2, SH3, STRO-1, CD29, CD44, CD71, and CD105 and negative for the hematopoietic markers CD34 and CD45. They expressed low levels of HLA-ABC and no HLA-DR molecules and were able to differentiate into osteocytes, adipocytes, cardiomyocytes, and neural cells in vitro, as previously described [1, 24, 25].

For the production of MSC-conditioned medium, passage-3 MSCs were cultured for 4 days at 100,000 per well, in 96-well flat-bottomed plates in 200 μl per well aMEM supplemented with 10% FCS, with or without 20 ng/ml IL-15 (R&D Systems, Abington, U.K., On day 4, supernatants were collected by centrifugation and used immediately.

NK Cell Isolation

Peripheral blood was obtained from normal volunteers, already typed for their HLA alleles. Mononuclear cells were isolated by Ficoll-Hypaque centrifugation. CD3+ cells were eliminated by negative selection with anti-CD3 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany, according to the manufacturer's instructions. CD56+ cells were isolated from the CD3 fraction using anti-CD56 microbeads (Miltenyi Biotec). To obtain highly purified CD56+ cells, the CD3 population was passed through a depletion column (Miltenyi Biotec) prior to incubation with anti-CD56 microbeads, whereas CD56+ cells were passaged through two sequential LS columns (Miltenyi Biotec). CD56dim and CD56bright subpopulations were separated by cell-sorting from the purified CD56+ cell fraction, with a Coulter Epics Altra cell-sorter (Beckman Coulter, Hialeah, FL, Purity was confirmed by fluorescence-activated cell sorting (FACS) analysis and was greater than 99% in all experiments performed.

Cell Culture

Irradiated MSCs (30 Gy) were plated at 100,000, 10,000, or 1,000 cells per well, in 96-well flat-bottomed plates or in the lower section of transwell systems (Nunc A/S, Roskilde, Denmark, in aMEM supplemented with 10% FCS. The following day, NK cells were added to the cultures (over the MSC layer or in the transwell inserts) at 100,000 per well in aMEM with 10% FSC, supplemented with IL-15 to reach a final concentration of 20 ng/ml. NK cells were also cultured under the same conditions without the addition of MSCs, as well as in the presence of MSC-conditioned medium. In the latter experimental setting, culture medium consisted of equal volumes of MSC-conditioned and fresh media. When indicated, anti–TGF-β1 (at 1μg/ml; R&D Systems) and/or prostaglandin-E2 (PGE2) synthesis-inhibitor NS-398 (at 5 μM; Cayman Chemicals, Irvine, CA, was added to the culture medium for the entire incubation period.

Cell Lines

The human cell lines K562 (erythroleukemia), Daudi (Burkitt's lymphoma), and SK-BR-3 and MCF-7 (both breast carcinoma) were obtained from the American Type Culture Collection (Manassas, VA, The melanoma cell line FM3 was a kind gift of Dr. Zeuthen, Department of Tumor Cell Biology, Danish Cancer Society Research Center, Copenhagen. K562, Daudi, and FM3 cell lines were cultured in RPMI-1640 medium (Life Technologies Ltd.) supplemented with 10% FCS, 2mM l-glutamine (Life Technologies Ltd.), and 50 μg/ml gentamicin. SK-BR-3 was cultured in McCoy's 5a medium (Life Technologies Ltd.) supplemented with 10% FCS, 2mM l-glutamine, and 50 μg/ml gentamicin, whereas the MCF-7 cell line was cultured in Dulbecco's modified Eagle's medium (Life Technologies Ltd.) supplemented with 10% FCS, 1% insulin (Humulin Regular, Lilly, France,, and 50 μg/ml gentamicin.

Monoclonal Antibodies and Immunophenotyping

Monoclonal antibodies against human CD44 conjugated with phycoerythrin (PE) and CD105 conjugated with fluorescein isothiocyanate (FITC) were purchased from Serotec (Oxford, U.K., Anti-CD34 and anti-NKG2D conjugated with PE were obtained from Becton, Dickinson and Company (Mountain View, CA, Anti-CD29 conjugated with allophycocyanin (APC), anti-CD3, anti-CD94, anti-CD158a and anti-2B4 conjugated with FITC, and CD132 conjugated with PE were purchased from Pharmingen (San Diego, PEcy5-conjugated anti-CD16, anti-CD56, anti-CD45 and anti–HLA-DR, PE-conjugated anti-CD71, anti-CD158b and anti-CD161, and FITC-conjugated anti–HLA-ABC were obtained from Immunotech (Beckman Coulter). For the determination of SH2, SH3, and STRO-1 expression, culture supernatants from the respective hybridoma cell lines (obtained from American Type Culture Collection for SH2 and SH3, and from Developmental Studies Hybridoma Bank, Iowa City, IA,∼dshbwww for STRO-1) were used. Staining with unlabeled antibodies was followed by a second cycle of immunostaining with goat anti-mouse Ig antibody conjugated with PE (DAKO A/S, Glostrup, Denmark, For the estimation of dead cells or cells undergoing apoptosis, the nucleic acid dye 7-amino-actinomycin D (7AAD) and Annexin V-PE (both from Pharmingen), respectively, were used according to the manufacturer's protocol. Samples were analyzed using FACSCalibur and CellQuest analysis software (both from Becton, Dickinson and Company).

Proliferative Assay

Cultures were set up as described above in the Cell Culture section. On day 3, 1 μCi per well [3H]TdR (Amer-sham Pharmacia Biotech, Cardiff, U.K., was added for the last 18 hours of culture. Cells were then harvested, and [3H]TdR uptake was measured in a microbeta counter (Wallac, PerkinElmer, Inc., Wellesley, MA, All cultures and controls were performed in triplicates.

Cytotoxicity Assay

NK cells were cultured for 4 days as described in the Cell Culture section, in the indicated combinations of 500 IU/ml IL-2 (Proleukin; Chiron, Emeryville, CA,, 2 ng/ml IL-12, 100 ng/ml IL-18 (both from R&D Systems), or 20 ng/ml IL-15. Cytotoxic activity was determined in a standard 4-hour 51 Cr-release assay. In brief, target cells were labeled with 100 μ Ci sodium [51Cr] chromate (Amersham Pharmacia Biotech) per 106 target cells for 1 hour. Effector cells were incubated with target cells at the indicated ratios. When NK cells from MSC/NK contact cultures were used as effectors, NK cells were recovered through removing MSCs by negative immunoselec-tion, using anti-CD105 antibody and M-450 sheep anti-mouse IgG magnetic beads (Dynal A.S., Oslo, Norway, according to the manufacturer's protocol. Spontaneous 51Cr release was measured by incubating target cells in the absence of effector cells. Maximum 51Cr release was determined by adding 1% Triton X-100 (Sigma, St. Louis, Spontaneous lysis did not exceed 10% of the maximum release. The amount of 51Cr released was measured in a microbeta counter, and the percent lysis was calculated as follows: percent specific lysis = (experimental 51Cr release − spontaneous 51Cr release)/(maximum 51Cr release − spontaneous 51Cr release) × 100.

Quantitation of Cytokines in Culture Supernatants

For cytokine production determinations, the protocol reported by Perez et al. [26] was applied. In brief, MSC/NK-cell cultures were set as described in the Cell Culture section. On day 2, IL-12 and IL-18 were added to the cultures at 2 ng/ml and 100 ng/ml, respectively. After 2 additional days, viable cell numbers were determined, and supernatants were collected by centrifugation and stored at −70°C until use. Cytokines were quantitated using commercially available ELISA (enzyme-linked immunosorbent assay) kits (Diaclone Research, Besançon, France, according to the manufacturer's instructions. Results are expressed as pg or ng per 106 NK-cell input, as numbers after the 4-day culture period did not statistically significantly differ.

Statistical Analysis

Statistically significant differences in the parameters tested were assessed by applying Student's t-test statistics to the experimental data. The cutoff value of significance was .05.


MSCs Inhibit IL-15–Induced NK Cell Proliferation Both in Contact and in Transwell Systems Without Inducing Cell Death

CD56+CD3 cells were cultured in the presence or absence of irradiated MSCs at MSC:NK ratios of 1:1, 1:10, and 1:100 in IL-15–supplemented medium, in contact or in transwell systems, as well as with the addition of MSC-conditioned medium. As presented in Figures 1A and 1B, MSCs inhibited the [3H]TdR incorporation of NK cells in a dose-dependent manner. The reduction in the [3H]TdR incorporation was significant at 1:1 and 1:10 ratios (p < .05), whereas no statistically significant differences were detected at the 1:100 ratio (p < .20). This effect was evident both in cell-to-cell contact and in transwell systems, suggesting the involvement of soluble factors within this process. To exclude a mass effect mediating this outcome, we performed the same experiments using irradiated NK cells instead of MSCs, at the ratio of 1:1, in contact cultures and transwell systems. No inhibitory effect was observed in these settings (data not shown). Addition of MSC-conditioned medium had no effect on the [3H]TdR incorporation by the NK cells, irrespective of IL-15 addition throughout the MSC-culture period (Fig. 1C), thus demonstrating that the factors secreted by the MSCs are induced upon MSC/NK-cell coculture, as this was also reported for MSC-induced, T-cell suppression [23].

Figure Figure 1..

MSCs exert a dose-dependent inhibitory effect on IL-15–activated NK cell–[3H]TdR incorporation without inducing cell death. CD56+CD3 cells were cultured in IL-15–supplemented medium for 4 days in the presence or absence of MSCs, at MSC:NK ratios of 1:1, 1:10, and 1:100, in contact (A) and in transwell systems (B), or with the addition of conditioned medium from MSCs cultured with or without IL-15 (C). The data are expressed as mean ± SD of triplicates of four independently performed experiments. (D): Cell death was determined by FACS analysis using 7AAD (dead cells) and Annexin V (apoptotic or undergoing apoptosis cells). One representative experiment out of three independently performed is presented. *Statistically significant (p < .05) versus the cultures performed without MSCs. Abbreviations: 7AAD, 7-amino-actinomycin D; FACS, fluorescence-activated cell sorting; IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

Nevertheless, statistically significant differences in NK-cell numbers growing in cultures set up as above were not observed using a hemocytometer (data not shown). Moreover, the number of NK cells on day-4 cultures never exceeded a 1.5-fold increase compared with the cell input, thus indicating that high [3H]TdR incorporation counts reflect the cells' entry into the cell cycle, however without yet reaching division.

To delineate whether the reduced [3H]TdR incorporation was due to cell death, NK cells were stained with 7AAD/Annexin V-PE. As shown in Figure 1D, there was no significant difference in dead or apoptotic cells among NK cells cultured alone or with MSCs, either in contact or in transwell systems (viable cells, stained negative for 7AAD and Annexin V, were 92 ± 5, 94 ± 2, and 92 ± 4, respectively).

Coculture with MSCs Differentially Affects IL-15–Stimulated NK Cell Subpopulations

CD56+CD3− cells stimulated for 4 days with IL-15, in the presence or absence of MSCs, both in contact and in transwell systems, were tested for CD56 expression by flow cytometry. As shown in Figure 2A, NK cells cultured with MSCs in transwell systems exhibited high mean fluorescence intensity (MFI) for CD56, as this was the case with NK cells in cultures in the absence of MSCs (1,009 ± 114 and 963 ± 103, respectively; p = .83); in contrast, NK cells in contact with MSCs had significantly lower CD56 MFI (566 ± 65, p = .03). However, the percentage of gated CD56bright and CD56dim populations remained unchanged within total NK cells cultured in the presence or absence of MSCs, suggesting that the decreased CD56 MFI levels did not result from selective expansion of CD56dim cells. (CD56bright cells represented 57% ± 17% [range 35%–83%] for control cultures, versus 52% ± 15% [range 26%–75%] for contact cultures with MSCs and 60% ± 15% [range 40%–82%] for transwell systems [p > .10].) However, whereas CD56 MFI of CD56dim-gated cells did not differ significantly among groups, CD56 surface expression was significantly lower in CD56bright-gated cells cultured in contact with MSCs (Fig. 2B). In all cases, CD56 MFI was higher compared with day 0 (129 ± 31 for CD56dim-gated and 900 ± 107 for CD56bright-gated cells).

Figure Figure 2..

MSCs differentially affect IL-15–activated NK-cell subpopulations. Total CD56+CD3(A, B) and CD56dim and CD56bright-sorted cells (B, D) were cultured in IL-15–supplemented medium for 4 days in the presence or absence of MSCs in contact or in transwell systems. The number of CD56 molecules present on the NK-cell surface, as expressed by the mean fluorescence intensity, was assessed by flow cytometry on total NK cells (A), CD56dim and CD56bright-gated cells (B), and CD56dim and CD56bright-sorted cells (D). For the sorted subpopulations, [3H]TdR incorporation was also assessed (C). Results are presented for a 1:1 MSC:NK-cell ratio. The data are expressed as mean ± SD of triplicates of four (A, B) or three (C, D) independently performed experiments. *Statistically significant (p < .05) when compared with the cultures performed only with NK cells. Abbreviations: IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

To further investigate the effect of MSCs on CD56 expression, we sorted NK cells into CD56bright and CD56dim and applied the same experimental settings. As previously reported, CD56bright cells show high proliferative responses induced with IL-15 or IL-2, whereas CD56dim cells maintain relatively low proliferation rates [27]. Moreover, IL-15–stimulated NK-cell cultures consist mainly of CD56bright cells [28]. When NK-cell subpopulations were cultured in cell-to-cell contact with MSCs, [3H]TdR incorporation was highly inhibited in CD56bright cells (p = .04), whereas [3H]TdR uptake in CD56dim cells remained unaffected (p = .22) (Fig. 2C). FACS analysis of the sorted cells after 4 days in culture revealed that CD56 MFI of both CD56bright and CD56dim cells was increased compared with day 0 in all experimental settings applied (data not shown). However, as observed in cultures with total NK cells, CD56bright cell MFI was decreased upon cell-to-cell contact with MSCs. In contrast, CD56dim cell MFI was unaffected by the presence of MSCs (Fig. 2D).

Expression of CD16, the C-type lectins CD94 and CD161, as well as the KIRs detected by anti-CD158a and anti-CD158b, was not statistically significantly altered upon MSC coculture (data not shown). On the contrary, the activating receptors 2B4 and NKG2D were downregulated in NK cells upon cell-to-cell contact with MSCs. As shown in Figure 3, 2B4 MFI was 37 ± 11 for control cultures, 20 ± 6 for contact cultures, and 32 ± 12 for transwell systems; p = .04 and .31, respectively (MFI of the isotype control was 7.5 ± 1.1), whereas MFI of NKG2D was 319 ± 109, 135 ± 75, and 276 ± 134; p = .04 and .34, respectively (MFI of the isotype control was 3.2 ± 0.9). Furthermore, expression of the common γ-chain (γc-chain) of IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21 receptors, detected by anti-CD132 antibody, was decreased to almost undetectable levels in all MSC/NK-cell cultures.

Figure Figure 3..

MSCs induce downregulation of surface molecules of NK cells. CD56+CD3 cells were cultured in IL-15–supplemented medium for 4 days in the presence or absence of MSCs in contact and in transwell systems, and stained for the surface expression of the activating receptors 2B4 and NKG2D, as well as the γc-chain. The data show the results from 1:10 MSC:NK-cell ratio coculture of one representative experiment out of three independently performed. Abbreviations: IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

MSCs Inhibit IL-15–Induced Cytokine Secretion by NK Cells in a Dose-Dependent Manner

Cytokine secretion represents a major effector and immunoregulatory means of NK cells. To test the impact of MSCs on NK-cell function, CD56+CD3 cells, cultured alone or in the presence of MSCs, were assayed for IFN-γ, IL-10, and TNF-α production. As shown in Figure 4A, MSCs inhibited cytokine secretion by NK cells in a dose-dependent manner, both in contact and in transwell systems. This inhibition was evident in all ratios tested for IFN-γ and IL-10 production, whereas for TNF-α it was statistically significant only at a MSC:NK ratio of 1:1. MSC-conditioned medium (irrespective of IL-15 presence during MSC culture) inhibited only TNF-α and IL-10 secretion (Fig. 4A), whereas the MSC-inhibitory effect on the cytokine profile of sorted CD56dim and CD56bright cells was observed mainly in the CD56bright cell subset (Fig. 4B), thus demonstrating the diversity of the mechanisms regulating NK-cell functions.

Figure Figure 4..

Inhibitory effect of MSCs on cytokine production by IL-15–activated NK cells. Total CD56+CD3 cells (A) or sorted CD56bright and CD56dim cells (B) were cultured for 4 days in IL-15–supplemented medium in the presence or absence of MSCs, as well as with the addition of conditioned medium from MSCs cultured in or without IL-15. The last 2 days, IL-12 and IL-18 were added to the cultures. Cytokine concentrations in the supernatants were estimated by ELISA and expressed as the quantity of cytokine produced by 106 NK-cell input. Bars represent the mean values ± SD of three independently performed experiments for total CD56+CD3d− cells and of two independently performed experiments for sorted cells. For sorted cells, the results of MSC/NK-cell cocultures at a ratio of 1:1 are presented. *Statistically significant (p < .05) versus the cultures performed without MSCs. Abbreviations: ELISA, enzyme-linked immunosorbent assay; IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

MSCs Reduce Cytotoxic Potential of IL-15–Induced NK Cells Against HLA Class I–Expressing Targets

IL-15–stimulated NK cells from 4-day coculture with MSCs, in contact or in transwell systems, were tested in cytotoxicity assays against the HLA-class I negative cell lines K562 and Daudi (representing NK-sensitive and NK-resistant targets, respectively). Our results show no alteration in the cytotoxic potential of NK cells against these targets (Figs. 5A, 5B). However, when HLA-class I positive tumor targets, such as SK-BR-3, MCF-7, and FM3, were used, NK cytotoxicity was significantly reduced (69%, 42%, and 44% decrease against SK-BR-3, MCF-7, and FM3, respectively, at a 10:1 effector: target cell ratio, when NK cells were cultured with MSCs at a ratio of 1:1; p < .05). MSC-mediated cytotoxicity inhibition was evident up to an MSC:NK-cell ratio of 1:10 (data not shown). NK cells derived from transwell culture systems did not present any statistically significant alteration in their cytolytic activity against the same targets (Figs. 5C, 5D). This was equally evident in both CD56dim and CD56bright sorted cells, whereas MSC-conditioned medium had no effect on the cyto-toxic potential of NK cells (data not shown).

Figure Figure 5..

MSCs affect the cytotoxic responses of IL-15–activated NK cells against HLA-class I–expressing targets. Cytotoxic activity of CD56+3 cells activated for 4 days with IL-15 in the presence or absence of MSCs, in contact and in transwell systems, was estimated regarding (A) the HLA-class I NK-sensitive cell line K562, (B) the HLA-class I NK-resistant cell line Daudi, and (C) the HLA-class I+ cancer cell lines SK-BR-3 [(also in (D)], MCF-7, and FM3. NK:target cell ratios used are as indicated in (A, B, D), whereas in (C) it is shown at the ratio of 10:1. Values represent the mean ± SD of triplicates of three independently performed experiments. Abbreviations: IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

Mechanisms of MSC-Induced NK Cell–Suppression

In an attempt to identify the mechanisms underlying the MSC effect on NK cells, we set up MSC/NK-cell cocultures adding, alone and in combination, neutralizing factors for TGF-β1 and PGE2, which are known to have the potential to suppress NK-cell proliferation and function. The data depicted in Figure 6A demonstrate that PGE2-synthesis inhibitor NS-398 partially restored [3H]TdR-uptake by NK cells (1.2-fold decrease compared with 2.4-fold in the absence of NS-398; p = .01); this was also the case, although at a lesser extent, for anti–TGF-β1 (1.6-fold decrease; p = .05). The combination of NS-398 and anti–TGF-β1 fully restored the ability of NK cells to incorporate [3H]TdR.

Figure Figure 6..

Modulation of NK-cell properties by MSC-produced factors. CD56+CD3 cells were cultured in IL-15–supplemented medium for 4 days in the presence or absence of MSCs, in contact or in transwell systems, supplemented with 1 μg/mL anti–TGF-β and/or 5μM NS-398. Their proliferation (A), cytotoxic potential against the FM3 cell line at a 10:1 effector:target cell-ratio (B), and surface expression of CD56 and γc-chain (C) are depicted. Values represent the means ± SD of triplicates of three independently performed experiments, and are presented for MSC/NK-cell coculture ratio of 1:10. *Statistically significant (p < .05) and **statistically significant (p < .01) versus the cultures performed without MSCs. Abbreviations: IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

The unchanged CD56 expression and cytolytic activity of NK cells in transwell systems precluded any involvement of soluble factors in MSC-mediated effects. However, both parameters were restored upon PGE2-synthesis blocking (Figs. 6B, 6C). This could suggest that in cell-to-cell contact cultures, the locally produced levels of PGE2 are high enough to mediate NK-cell suppression. Addition of PGE2-synthesis inhibitor restored γc-chain expression to normal levels (Fig. 6D), whereas addition of neutralizing factors for TGF-β1 and PGE2 had no apparent effect on the expression of 2B4 and NKG2D or on cytokine production (data not shown).

MSCs Are Efficiently Lysed by Activated NK Cells

MSCs were tested as targets in cytotoxicity assays involving freshly isolated and cytokine-activated CD56+CD3 cells. In line with previous reports, our results verify that MSCs cannot be lysed from freshly isolated NK cells [14] (Fig. 7A). Nonetheless, NK cells cultured for 4 days in IL-15–supplemented medium could effectively lyse MSCs, not only from HLA-B and -C mismatched donors, but also autologous to the activated NK cells (Fig. 7B).

Figure Figure 7..

MSCs represent targets for activated NK cells. CD56+CD3 cells were tested in 4-hour cytotoxicity assays against MSCs, using the NK-sensitive K562 cell line as control. (A) MSCs from HLA-mismatched individuals and K562 were tested as targets of freshly isolated NK cells. (B): MSCs from HLA-mismatched individuals, MSCs autologous to the NK cells used as effectors, as well as K562, were used as targets of IL-15–activated NK cells. The data presented are the results from MSC/NK coculture at a ratio of 1:10. (C): MSCs from HLA-mismatched donors and K562 cells were used as targets of NK cells activated with the indicated cytokines, at an NK:MSC ratio of 25:1. The data are expressed as mean ± SD of triplicates of three independently performed experiments. Abbreviations: IL, interleukin; MSC, mesenchymal stem cell; NK, natural killer.

To test the ability of different cytokines to induce NK cytotoxicity against MSCs, IL-2 and the combinations of IL-12/IL-15 and IL-12/IL-18 were also used. All stimulating protocols resulted in induction of lytic activity against MSCs, which did not differ significantly among the various groups (p > .05; Fig. 7C).


MSCs are the focus of regenerative and immunologic studies due to their potential usage in tissue engineering and immunotherapeutical approaches. Previous reports demonstrated the immunosuppressive effect of MSCs on several cell types of the immune system. Their mediation in T-cell stimulation-inhibition has been shown in multiple in vitro [811, 14, 16] and in vivo [13, 17, 29] studies and has been attributed mainly to soluble factors secreted by MSCs, including hepatocyte growth factor, TGF-β1 [8], and PGE2 [23], to l-tryptophan catabolites generated by indoleamine 2,3-dioxygenase (IDO) activity [10], whereas in the system used by Krampera et al. [9] cell-to-cell contact was essential for their inhibitory function. Moreover, MSCs escape recognition by cytotoxic T lymphocytes and freshly isolated NK cells [14, 15], whereas their interaction with DCs, directly or via soluble factors, results in phenotype alteration and maturation inhibition of the latter and subsequently in the decreased capacity of antigen presentation and consequently T-cell activation [11, 23].

In this report, the effects of MSCs on the activation of NK cells with respect to their phenotype, proliferation, and function were thoroughly examined. Our data clearly illustrate that MSCs suppress NK-cell proliferation, cytokine secretion, and cytotoxicity. To evaluate the effects of MSCs on NK-cell activation, we used as inducing factor IL-15, a cytokine known to promote NK-cell proliferation, survival, and effector functions [30]. When NK cells were stimulated with IL-15 in the presence of MSCs, both in contact and in transwell systems, there was a dose-dependent reduction in their [3H]TdR incorporation, apparent for MSC:NK ratios up to 1:10. This suggested that MSCs mediate their effects on NK cells via soluble factors and these are enhanced upon interaction with NK cells, because MSC-conditioned medium (irrespective of IL-15 addition during MSC culture) failed to affect NK cell–[3H]TdR incorporation. Similar observations have also been published for MSC/T-cell interactions [10, 23]. The MSC-mediated inhibition of [3H]TdR incorporation cannot be attributed to cell death because necrosis or apoptosis was excluded.

IL-15 stimulation of NK cells results in cultures consisting mainly of CD56bright cells [28]. Nevertheless, in contact cultures with MSCs, MFI of CD56 is lower than in control cultures, without, however, significant alteration in the CD56dim/CD56bright cell ratio. Sorting experiments showed that CD56 expression in contact cultures, as well as [3H]TdR incorporation and cytokine secretion both in cell-to-cell contact and in transwell systems, is observed only in CD56bright cells, whereas MSCs in contact cultures suppress both subpopulations with respect to their cytolytic activity and activating receptor expression.

NK-cell function is regulated by a balance of activating and inhibitory signals transduced via surface receptors. The receptors tested herein that appeared to be affected by MSC/NK-cell coculture are 2B4, which results in enhanced cytotoxicity and IFN-γ production [31], and NKG2D, which has a major role in NK-mediated cytotoxicity and recognizes defined antigens on stressed and cancer cells [32]. Both receptors were downregulated when NK cells were cultured in cell-to-cell contact with MSCs. Downregulation of 2B4, NKG2D, and possibly other surface receptors not included in our study could account for the MSC-mediated inhibition of NK-cell function. Moreover, because these molecules trigger different signaling pathways [33], their modulation may reflect the multiple ways by which MSCs are able to regulate NK-cell function.

The modification in cytokine profile of NK cells activated by IL-2, in the presence of MSCs, has recently been reported with respect to IFN-γ secretion [23]. In this culture system, the MSC-mediated effect was much more evident, because there was an 80% decrease in IFN-γ secretion (at 1:1 NK:MSC ratio) compared with a 43% decrease observed in our study. To this end, we should mention that IL-2–activated NK cells are able to produce IFN-γ [34], whereas IL-15–activated NK cells produce cytokines only after coculture with IL-12 and/or IL-18 [35]. This difference among the protocols may account for the observed quantitative discrepancy. Additionally, in our approach there was a remarkable reduction in IL-10 and TNF-γ production, the only NK-cell properties also inhibited by MSC-conditioned medium. Cytokines secreted by NK cells mediate the eradication of pathogens and infected cells and furthermore modulate the development of adaptive immune responses, providing the means for a dynamic interaction between innate and adaptive immunity [36]. Thus, modulation of this function by MSCs could primarily result in the downregulation of the potency of innate immune responses, but may also be considered as an additional mechanism capable of suppressing the function of cells of the adaptive immune system.

One major effector function of NK cells is their lytic activity against infected and cancer cells. Cytolytic potential is tightly regulated through equilibrium of activating and inhibitory receptors with ligand specificity for HLA-class I or HLA-class I–related molecules on target cells. These receptors exist as pairs of opposite function with highly homogeneous extracellular ligand-binding domains [20]. Thus, HLA-class I or HLA-class I–mismatched cells represent potential targets of NK cells. However, NK cells, by involving certain activating receptors, are able to lyse autologous tumor and stressed cells in vivo [32]. In our system, NK-cell cytotoxicity against the HLA-class I cell lines K562 and Daudi is not affected by short-term MSC coculture, which is in agreement with previous experiments by Rasmusson et al. [14]. Interestingly, when we tested the HLA-class I–expressing cancer cell lines SK-BR-3, MCF-7, and FM3 (all of them expressing different MHC-class I alleles) as targets, we detected a significant decrease in NK-cell cytotoxicity in cell-to-cell contact cultures, possibly indicating an MSC-induced downregulation of MHC-class I–specific triggering receptors on NK cells [37].

NK-cell function is controlled by a variety of mechanisms, some of which are used by MSCs to mediate NK-cell inhibition. Data reported herein suggest that the modulation of different NK-cell parameters is based on distinct mechanisms, some of which require cell-to-cell contact, whereas others involve soluble factors, secreted in MSC cultures or induced by MSC/NK-cell coculture. With respect to soluble factors, in vitro studies have shown that MSCs, without or after stimulation, secrete a wide range of regulating molecules [2], some of which (such as IL-15, TGF-β1, and PGE2) have the potential to affect NK-cell proliferation and function, whereas others have a role during the early phases of differentiation or influence only their homing capacity [38, 39]. PGE2 suppresses IL-2–and IL-15–mediated NK-cell cytotoxicity and cytokine production [40, 41]. This mediator is secreted into tumors and accounts for NK-cell suppression via downregulation of the γc-chain [41]. Indeed, in our system, γc-chain expression was reduced to negligible levels when NK cells were cultured with MSCs and was restored after addition of PGE2-synthesis inhibitor to culture medium. PGE2 appears to interfere with MSC inhibition of NK-cell proliferation, CD56 expression, and cytotoxicity but to have no impact on cytokine production or 2B4 and NKG2D expression.

TGF-β1 is an immunosuppressive factor capable of inhibiting NK-cell expansion, cytotoxicity, and cytokine production [42] and of affecting the expression of activatory receptors [43]. Because of its suppressive activities, TGF-β1 is involved in the enhancement of tumor invasion and metastasis [44]. TGF-β1 secretion by both cancer cells and tumor stroma [44], together with the fact that tumor stroma consists mainly of MSC-derived cells [45], suggests a new role for MSCs in cancer progression and tumor-cell escape of immune surveillance. In our system, TGF-β1 was found to be responsible for MSC-mediated inhibition of NK-cell proliferation, without however being able to affect cytotoxic potential. PGE2 and TGF-β1 had an additive inhibitory effect on NK-cell proliferation, as neutralization of both factors resulted in complete restoration of NK cell[3H]TdR uptake capacity, whereas when added separately proliferative capacity was only partially restored. This is actually in agreement with previous data, which show that PGE2 and TGF-β1 suppress NK-cell activity by different mechanisms [46].

IDO catalyzes l-tryptophan degradation, thus exerting a major inhibitory effect on T-cell and NK-cell proliferation [47]. IDO activity is induced by IFN-γ in many cell types, including MSCs, and was found responsible for MSC-mediated inhibition of allogeneic T-cell responses [10]. In our system, IDO activity could not account for the MSC-mediated effects on NK cells, because NK cells stimulated with IL-15 alone produce negligible amounts of IFN-γ, thus are not able to induce IDO production by MSCs.

A major obstacle to the use of NK cells as immunotherapeutical tools against cancer is that they are incapable of infiltrating the tumor mass, with a large proportion residing within the tumor stroma [48], tissue consisting mainly of MSC-derived cells [45]. Rasmusson et al. [14] have already shown that freshly isolated NK cells cannot lyse MSCs, which is in agreement with our results. However, our data clearly illustrate that NK cells activated by different stimulating factors are capable of effectively lysing MSCs, thus providing a means to enter the tumor mass and thereby eliminate cancer cells. Therefore, cancer immunotherapeutic protocols involving NK cells should employ appropriately activated NK cells, to allow them not only to pass through the tumor stroma, but also to preserve their killing abilities against cancer cells.

Our results, besides extending our knowledge of MSC interactions with cells of the immune system, clearly possess potential therapeutic significance. NK cells are involved in the early phase of innate immune responses, the regulation of adaptive immunity, cancer cell eradication, and graft rejection. Thus, the suppression induced by MSCs could be beneficial for transplantation and autoimmune diseases, while it could negatively interfere in cancer immunotherapy involving NK cells. More importantly, the fact that MSCs are not totally ignored by the immune system, but represent targets for activated NK cells, should be considered in the planning of studies for tissue regeneration and gene therapy. The complete mechanisms underlying these interactions are, however, yet to be clarified.


The authors would like to thank the Division of Nephrology and Transplantation of Laiko General Hospital, Athens, Greece, and the Department of Haematology and Lymphomas of Evangelismos Hospital, Athens, Greece, for kindly providing the bone marrow samples. This work was supported by a grant from the Regional Operational Program Attika No. 20, MIS code 59605GR to M.P.


The authors indicate no potential conflicts of interest.