A regulatory cross-talk between Vγ9Vδ2 T lymphocytes and mesenchymal stem cells



The physiological functions of human TCRVγ9Vδ2+ γδ lymphocytes reactive to non-peptide phosphoantigens contribute to cancer immunosurveillance and immunotherapy. However, their regulation by mesenchymal stem cells (MSC), multipotent and immunomodulatory progenitor cells able to infiltrate tumors, has not been investigated so far. By analyzing freshly isolated TCRVγ9Vδ2+ lymphocytes and primary cell lines stimulated with synthetic phosphoantigen or B-cell lymphoma cell lines in the presence of MSC, we demonstrated that MSC were potent suppressors of γδ-cell proliferation, cytokine production and cytolytic responses in vitro. This inhibition was mediated by the COX-2-dependent production of prostaglandin E2 (PGE2) and by MSC through EP2 and EP4 inhibitory receptors expressed by Vγ9Vδ2 T lymphocytes. COX-2 expression and PGE2 production by MSC were not constitutive, but were induced by IFN-γ and TNF-α secreted by activated Vγ9Vδ2 T cells. This regulatory cross-talk between MSC and Vγ9Vδ2 T lymphocytes involving PGE2 could be of importance for the antitumor and antimicrobial activities of γδ T cells.


Most circulating γδ T lymphocytes (1–5% of PBMC) from humans or primates express Vγ9Vδ2+ T-cell receptor and have an unconventional pattern of reactivity and physiological functions that link innate and adaptive immunity to infections and cancer. Vγ9Vδ2 T lymphocytes are specific for HLA-unrestricted, non-peptide antigens referred to as phosphoantigens, small phosphorylated metabolites from microbial pathogens, human tumor cells or phosphorylated analogs produced by chemical synthesis (reviewed in 1, 2). Activated Vγ9Vδ2 T lymphocytes participate in the early phase of immune responses through the release of pro-inflammatory chemokines and cytokines, a cross-talk with DC 3–5 and their cytotoxic activity 6, 7. They also provide help to drive adaptive immunity by favoring Ig production 8, immunological memory 9, 10 and protection against viral infections 11. In monkeys and humans treated with the synthetic phosphoantigen bromohydrin pyrophosphate (BrHPP) 12, the strong amplification of blood Vγ9Vδ2 T lymphocytes 13, 14, together with their promising anti-tumor activity 15–18, make Vγ9Vδ2 cells highly attractive to develop innovative immunotherapy against solid and lymphoid malignancies.

However, human tumors develop in a favorable microenvironment, which may comprise various cell types with immunomodulatory properties 19. Similarly, the stroma of mammary and glioma tumors may recruit large numbers of bone marrow-derived mesenchymal stem cells (MSC) 20, 21. MSC are multipotent and uncommitted cells able to differentiate into a wide variety of cell types: osteoblasts, adipocytes, chondrocytes, tenocytes or skeletal myocytes 22. Their ability to migrate to injured tissues after systemic injection 23 makes MSC attractive for tissue engineering and regenerative cell therapies 24. MSC escape immune recognition and inhibit αβ T-cell proliferation induced by mitogens or cognate peptides 25–27. MSC also inhibit proliferation of NK cells 28, 29 and the differentiation and maturation of DC 30, 31. The underlying immunosuppressive bioactivity of MSC involves multiple mechanisms mediated by soluble factors and cell contact 32. These encompass release of prostaglandin E2 (PGE2), IDO, NO, HGF as well as TGF-β1 and IL-10 cytokines 32.

Along this line, the suppressive properties of MSC could possibly be deleterious to anti-tumor immunity. Accordingly, MSC were able to promote the growth of an allogeneic tumor in immunocompetent mice 33. Despite this piece of evidence, it remains unclear whether MSC regulate the function of unconventional HLA-unrestricted T-cells such as TCRVγ9Vδ2+ lymphocytes. In addition, quite a few compounds with negative regulatory activity for TCRVγ9Vδ2+ cells have already been described 34, 35. In this study, we demonstrate that MSC exert a strong regulation of Vγ9Vδ2 T-cell responses, mediated by secretion of inhibitory PGE2. The MSC-up-regulated PGE2 production in response to the γδ cytokines uncovers a regulatory cross-talk between MSC and γδ lymphocytes.


MSC inhibit the antigen-induced Vγ9Vδ2 T-cell responses

PBMC or freshly purified γδ T-cells were cultured with BrHPP, IL-2 and with or without allogeneic irradiated MSC at cell ratios of 0.2:10, 1:10 and 2:10 for MSC:PBMC or MSC:γδ. CFSE dilution assay showed that MSC inhibition of Vγ9Vδ2 T-cell proliferation was dose dependent (Fig. 1A). The rate of dividing Vγ9Vδ2 T-cells among PBMC or among purified γδ T-cells was significantly lower at the 1:10 and 2:10 ratios (p<0.01) (Fig. 1B). Accordingly, the [3H]-thymidine incorporation by phosphoantigen-stimulated Vγ9Vδ2 T-cells was reduced when MSC were added to the culture (data not shown). Moreover, after 6 days of co-culture, the presence of MSC resulted in less Vγ9Vδ2 T-cells in PBMC cultures (Fig. 1C). CFSE and 7-aminoactinomycin D (7-AAD) staining showed that the percentage of apoptotic divided and undivided Vγ9Vδ2 T lymphocytes was not increased when irradiated MSC were added to the culture (Fig. 1D–E). Altogether, these data indicate that MSC inhibit the phosphoantigen-induced Vγ9Vδ2 T-cell proliferation without increasing T-cell death.

Figure 1.

MSC inhibit antigen-induced Vγ9Vδ2 T-cell responses. (A–E) CFSE-labeled PBMC or purified γδ T-cells stimulated with BrHPP+IL-2 were co-cultured for 6 days with irradiated allogeneic MSC. (A) Representative CFSE dilution by dividing Vγ9+ cells from PBMC (upper panel) or purified γδ T-cells (lower panel). (B) Rates of dividing Vγ9+ cells from the above experiments (means±SD from n=5 independent experiments; **p<0.01 (Mann–Whitney test)). (C) Absolute Vγ9+ cell counts recovered from PBMC+MSC co-cultures (means±SD from n=4 independent experiments, **p<0.01 (Mann–Whitney test)). (D) Dividing and dying Vγ9+ cells from purified γδ T-cells co-cultured with MSC (representative experiment out of four). (E) Rates of apoptotic Vγ9+ cells among divided (black bars) or undivided (white bars) Vγ9+ cells in the above experiments (means±SD, n=4 independent experiments, N.S. p>0.05 (Student's t-test)). (F and G) BrHPP-stimulated γδ T-cells producing IFN-γ and TNF-α in the presence of MSC. (F) Representative experiment; (G) means±SD from n=3 independent experiments, ***p<0.001 (Student's t-test)). (H) Inhibition of Daudi target cell lysis by Vγ9Vδ2 T-cells in the presence of MSC as follows: □: no MSC added, ♦: 0.2 MSC per 10 γδ ▴: 1 MSC per 10 γδ •: 2 MSC per 10 γδ ▪: 5 MSC per 10 γδ (representative from n=3 independent experiments).

Because BrHPP-activated Vγ9Vδ2 T-cells produce the pro-inflammatory cytokines IFN-γ and TNF-α we analyzed the effect of MSC on this production. When added to purified γδ T-cell cultures at 0.2:10, 1:10 and 2:10 cell ratios, MSC strongly inhibited the production of IFN-γ and TNF-α as shown by intracellular staining of Vγ9Vδ2 T-cells (Fig. 1F). The percentages of both IFN-γ-producing and TNF-α-producing Vγ9Vδ2 T-cells were significantly decreased (p<0.01, n=3 donors) at 1:10 and 2:10 MSC/γδ T-cell ratios (Fig. 1G). This effect was not confined to freshly isolated γδ T-cells, since the production of IFN-γ and TNF-α by PBMC-derived Vγ9Vδ2 T-cell lines in response to BrHPP was also severely decreased by MSC (data not shown).

Upon recognition of endogenous phosphoantigen metabolites, Vγ9Vδ2 T-cells show spontaneous lytic activity towards tumoral cell lines such as the Daudi Burkitt's lymphoma 36. Therefore, we tested whether the MSC affected this function. When added to standard (4 h) 51Cr-release assays, MSC induced a dose-dependent reduction of the Vγ9Vδ2 T-cell line cytotoxicity, total abrogation being seen at 5 MSC:10 Vγ9Vδ2 T-cell ratio (Fig. 1H). Although activated NK cells are able to lyse autologous and allogeneic MSC 28, 29, neither freshly purified γδ T-cells nor activated Vγ9Vδ2 T-cell lines killed allogeneic MSC (data not shown).

COX activity is necessary for Vγ9Vδ2 T-cell inhibition by MSC

To determine how MSC inhibited Vγ9Vδ2 T-cell biological responses, the above experiments were repeated using Transwell settings. When MSC were plated in the lower compartment of the Transwell device, freshly purified γδ T-cells did not proliferate in response to BrHPP (Fig. 2A). In these conditions, MSC also inhibited the Vγ9Vδ2 T-cell production of IFN-γ (Fig. 2B) and TNF-α (Fig. 2C).

Figure 2.

COX activity from MSC inhibits Vγ9Vδ2 T-cell responses. (A–C and E) Purified γδ T-cells co-cultured with BrHPP+IL-2 and allogeneic MSC in conditions allowing (black) or not allowing (white) cell contact. (A) Rates of dividing Vγ9+ cells after 6 days in the co-cultures (means±SD from five independent experiments). (B and C) Rates of IFN-γ+ and TNF-α+ Vγ9+ cells 24 h after stimulation by BrHPP (means±SD from three independent experiments). (D) Standard 51Cr release assay from Daudi target cells lysed by Vγ9Vδ2 T-cells in complete medium alone (□), in medium with MSC culture supernatant (♦) or in medium with MSC and Vγ9Vδ2 T-cell line co-culture supernatant (▴) (representative of three independent experiments). (E) Rates of dividing Vγ9+ PBMC after 6 days in co-cultures with MSC supplemented with the specified drugs (means±SD from five independent experiments). N.S. p>0.05; ***p<0.001 (Student's t-test).

The Transwell co-culture settings were not appropriate to assay cytotoxicity. Thus, we performed standard (4 h) 51Cr-release assays in medium supplemented with the supernatants from MSC cultivated for 3 days, either alone or with Vγ9Vδ2 T-cell lines. Only the supernatant of MSC co-cultured with Vγ9Vδ2 T-cell lines decreased the Daudi lysis by γδ lymphocytes (Fig. 2D). Thus, MSC produce a soluble mediator able to inhibit the Vγ9Vδ2 T-cell lytic activity. Interestingly, this mediator was not produced by MSC cultured alone but only when MSC were co-cultured with Vγ9Vδ2 T-cells.

To identify the soluble mediator responsible for MSC suppression of Vγ9Vδ2 T-cell responses, we tested the following specific inhibitors: antibodies against soluble or membrane-bound TGF-β, IL-10 and HGF, and chemical inhibitors of either COX (indomethacin), iNO-synthase (N-nitro-L-arginine methyl ester (L-NAME)) or IDO (1-methyl-dl-tryptophan (1-MT)) enzyme activities. This screening was applied to a BrHPP-induced Vγ9Vδ2 T-cell proliferation assay from PBMC co-cultured with MSC at different T-cell ratios, in the presence of the inhibitors. Only the COX1 and 2 inhibitor indomethacin significantly blocked the suppressive activity of MSC (p<0.01)Fig. 2E).

Soluble PGE2 inhibits Vγ9Vδ2 T-cell activation and biological responses

The above data suggested that PGE2, the final metabolite of the COX pathway, could play a role in the suppressive activity of MSC. Therefore, we tested the effect of exogenous PGE2 addition in Vγ9Vδ2 cell activation experiments devoid of MSC. Figure 3A shows that soluble PGE2 inhibited the calcium flux induced in Vγ9Vδ2 T-cell line by TCRVγ9Vδ2 cross-linking with anti-CD3 mAb. Cytokine production and cytotoxicity from either freshly purified γδ T-cells or Vγ9Vδ2 T-cell lines were strongly inhibited by soluble PGE2 (Fig. 3B and C). Moreover, the exogenous PGE2 concentration able to abrogate both target cell lysis (Fig. 3C) and IFN-γ/TNF-α secretion (Fig. 3C) by Vγ9Vδ2 T-cell lines (100 ng/mL) corresponded to the PGE2 concentration secreted by MSC co-cultured with activated Vγ9Vδ2 T-cells (see the Activated Vγ9Vδ2 T-cells induce COX2 expression and PGE2 production by MSC section). Exogenous PGE2, added from the beginning of the culture or 24 h after activation, significantly inhibited Vγ9Vδ2 T-cell proliferation in PBMC (data not shown) or in purified γδ T-cells culture (Fig. 3E) only at 1000 ng/mL (p<0.01 Student's t-test). This discrepancy can be due to the length of the proliferation assay (6 days). When the inhibition of proliferation was achieved with MSC, they produced PGE2 throughout the assay. Conversely, the exogenous PGE2 was added once at day 0 and might undergo degradation by serum enzymes 37. CFSE and 7-AAD staining indicate that the percentage of apoptotic-divided and -undivided Vγ9Vδ2 T lymphocytes were not statistically increased even when 10 000 ng/mL of PGE2 was added to the culture (Fig. 3D and F). Altogether, these results indicate that similar to MSC, PGE2 inhibits phosphoantigen-induced Vγ9Vδ2 T-cell responses without increasing cell death.

Figure 3.

Soluble PGE2 inhibits Vγ9Vδ2 T-cell activation and responses. (A) Ca++ flux induced by anti-CD3 stimulation of Vγ9Vδ2 T-cell lines preincubated (gray histogram) or not (black line) with 1 μg/mL of PGE2; GAM: goat anti-mouse cross-linking (representative of three independent experiments). (B) IFN-γ+ and TNF-α+ Vγ9+ cells 5 h after stimulation of Vγ9Vδ2 T-cell lines by BrHPP in the presence of PGE2 (representative of five independent experiments). (C) PGE2 inhibits the specific lysis of 51Cr-pulsed Daudi target cells by Vγ9Vδ2 T-cell lines □: no PGE2 added, ♦: 10 ng/mL PGE2, ▴: 100 ng/mL PGE2, •: 1000 ng/mL PGE2, ▪: 10 000 ng/mL PGE2 (representative of five independent experiments). (D) Representative CFSE dilution and 7-AAD staining of purified γδ T-cells stimulated by BrHPP+IL-2 in the presence of PGE2 (representative of four independent experiments). (E) Rates of dividing Vγ9+ cells from purified γδ T-cell cultures stimulated by BrHPP+IL-2 in the presence of PGE2 added from the beginning of the culture (black bars) or 24 h after BrHPP+IL-2 stimulation (white bars) (means±SD from n=4 independent experiments). (F) Rates of apoptotic Vγ9+ cells among divided (black bars) or undivided (white bars) purified γδ T-cells stimulated by BrHPP+IL-2 in the presence of PGE2 (means±SD, n=4 independent experiments). N.D. non detectable, N.S. p>0.05; **p<0.01; ***p<0.001 (Student's t-test).

Expression of PGE2 receptors by Vγ9Vδ2 T lymphocytes

PGE2 is a ligand for the G-protein-coupled receptors EP1–4, the expression level of which in γδ cells was unknown. Using RT-PCR, we then evaluated the level of mRNA encoding the four PGE2 receptor subtypes. Only EP2, EP3 and EP4 were expressed by Vγ9Vδ2 T-cell lines (Fig. 4A). A strong cell-surface expression of the EP2, EP3 and EP4 receptor proteins was next confirmed by flow cytometry analysis on Vγ9Vδ2 T-cell lines and on resting purified Vγ9Vδ2 T-cells either unstimulated or 2 days following activation with BrHPP+IL-2 (Fig. 4B). EP2 and EP4 are coupled to Gs protein, activate adenylate cyclase and increase cyclic adenosine monophosphate (cAMP), whereas the major signaling pathway of EP3 inhibits adenylate cyclise via Gi protein and decreases cAMP 38. Using forskolin, the chemical activator of adenylate cyclase that mimics the EP2 and EP4 signaling pathways, we reproduced the inhibitory effect of soluble PGE2 on IFN-γ production by Vγ9Vδ2 T-cell lines (Fig. 4C).

Figure 4.

Expression of PGE2 receptors by Vγ9Vδ2 T-cells. (A) RT-PCR for EP1, EP2, EP3 and EP4-mRNA from Vγ9Vδ2 T-cell lines; HPRT: mRNA for hypoxantine phosphoribosyl transferase (representative from four independent cell lines). (B) Cell-surface expression of EP2, EP3 and EP4 receptors on purified Vγ9Vδ2 T-cells stimulated or not by BrHPP+IL-2 or on Vγ9Vδ2 T-cell lines; gray line, isotype control; black line, EP2 receptor; dotted line, EP3 receptor; dashed line, EP4 receptor. (C and D) Rates of IFN-γ+ Vγ9+ cells 5 h after stimulation of Vγ9Vδ2 T-cell lines by BrHPP and increasing doses of (C) forskolin (▪) or control medium alone (□); or of (D) increasing doses of PGE2 (□), EP1–3 agonist sulprostone (♦), EP2 agonist butaprost (▴) or EP 3–4 agonist 1-Hydroxy-PGE1 (•) (means±SD from four independent experiments).

In addition, EP-specific agonists reproduced the effect of soluble PGE2 on IFN-γ production by Vγ9Vδ2 T-cell lines (Fig. 4D). Butaprost, a structural analog of PGE2 with good selectivity for the EP2 receptor, was the most effective inhibitor among these agonists; its activity was virtually the same as that of PGE2. The EP3 and EP4 agonist 1-Hydroxy PGE1 was also inhibitory of Vγ9Vδ2 T-cells although at much higher concentrations. In this case, however, inhibition by 1-Hydroxy-PGE1 was not mediated through EP3 since sulprostone, an agonist of EP1 and EP3 receptors, was inactive. Together, these results demonstrate that PGE2 inhibits Vγ9Vδ2 T-cells through their EP2 and EP4 receptor-dependent activation of adenylate cyclase.

Activated Vγ9Vδ2 T-cells induce COX2 expression and PGE2 production by MSC

Since MSC inhibition of Vγ9Vδ2 T-cell responses depends on COX activity and PGE2, we checked the expression of this enzyme and the production of its PGE2 metabolite by MSC. We analyzed COX1 and COX2, the two major isoforms of the COX enzyme, by flow cytometry in MSC either cultured alone or co-cultured with resting and phosphoantigen-activated Vγ9Vδ2 T-cell lines (cell ratio of 2:10). We observed a low level of COX1 in MSC cultured in all conditions (data not shown). COX2 expression was also low in MSC cultured alone, but it was greatly increased in MSC co-cultured with activated Vγ9Vδ2 T-cell lines (Fig. 5A). A dose-dependent up-regulation of COX2 could also be achieved with supernatants of the co-cultures, suggesting that COX2 expression by MSC is enhanced by soluble factors released by activated Vγ9Vδ2 T-cell lines (Fig. 5B). The presence of PGE2 was titrated by ELISA in the supernatant of MSC and Vγ9Vδ2 T-cells co-cultures. Although MSC secreted only low levels of PGE2 when cultured alone or co-cultured with resting Vγ9Vδ2 T-cell lines (cell ratio 2:10), MSC co-cultured with activated Vγ9Vδ2 T-cell lines produced up to 100 ng/mL of PGE2 (Fig. 5C). The effect of Vγ9Vδ2 T-cell lines (activated or not) was totally mimicked by their culture supernatants. It is noteworthy that PGE2 production by MSC was significantly inhibited by adding indomethacin at the beginning of the co-cultures or in the supernatant tests (data not shown).

Figure 5.

Activated Vγ9Vδ2 T-cells induce COX2 expression and PGE2 production by MSC. (A) Intracellular COX2 expression in MSC cultured for 24 h either alone (medium) or with Vγ9Vδ2 cells activated or not by BrHPP. (B) Rates of COX2+ MSC after 24 h of culture alone, with Vγ9Vδ2 T-cells, or with Vγ9Vδ2 T-cell line supernatant (SN) (means±SD from five independent experiments, ***p<0.001 (Student's t-test)). (C) PGE2 concentrations in the specified culture supernatants (representative of four independent experiments).

IFN-γ and TNF-α from activated Vγ9Vδ2 T-cells induce PGE2 production by MSC

Soluble IFN-γ and TNF-α are able to induce COX2 expression 39, 40 and PGE2 production by MSC 30. Since Vγ9Vδ2 T-cells are potent producers of IFN-γ and TNF-α, we wondered whether these cytokines were inducing COX2 expression in MSC. IFN-γ and TNF-α were depleted from the Vγ9Vδ2 T-cell lines supernatants by immunoprecipitation with specific antibodies (average depletion: 90%, data not shown) and led to supernatants totally unable to induce COX2 expression (Fig. 6A) and PGE2 production (Fig. 6B). We found that 100 ng/mL of soluble IFN-γ or TNF-α alone could induce only low level of intracellular COX2 in MSC as checked by flow cytometry. However, when both cytokines were added together at 100 ng/mL each in MSC cultures, over 40% of MSC were COX2+, indicating a cumulative effect on COX2 induction (Fig. 6C). Altogether, these results demonstrate that both IFN-γ and TNF-α produced by activated Vγ9Vδ2 T-cells are necessary to induce COX2 expression and PGE2 synthesis in MSC. Interestingly, an average of 10 ng/mL of IFN-γ and TNF-α were detected in the supernatants of BrHPP-activated Vγ9Vδ2 T-cell lines whereas 100 ng/mL of commercial recombinant cytokines were necessary to induce COX2 expression by MSC. The obvious explanation is that IFN-γ and TNF-α produced by Vγ9Vδ2 T-cells might be more bioactive than commercial recombinant cytokines used in our experiments. However, we cannot exclude that other cytokines produced by activated Vγ9Vδ2 T-cells might also potentiate IFN-γ and TNF-α ability to stimulate COX2 expression by MSC.

Figure 6.

IFN-γ and TNF-α from activated Vγ9Vδ2 T-cells induce PGE2 production by MSC. (A and B) Depletion of both IFN-γ and TNF-α from supernatants of activated Vγ9Vδ2 cell cultures reduces (A) COX2 induction in MSC (means±SD from four independent experiments; N.S. p>0.05, **p<0.01 (Student's t-test) versus activated γ9δ2 cells SN-treated MSC) and (B) PGE2 secretion (representative of four independent experiments). (C) Rates of COX2+ MSC after 24 h of culture in medium alone, or supplemented with recombinant IFN-γ and/or recombinant TNF-α (10 or 100 ng/mL) (means±SD from three independent experiments; N.S. p>0.05, **p<0.01 (Student's t-test) versus medium-treated MSC).


This study uncovered a regulatory cross-talk between the unconventional human γδ T lymphocytes and MSC, in which IFN-γ and TNF-α from activated Vγ9Vδ2 T-cells induce PGE2 production by MSC. The cell-surface expression of PGE2 receptors EP2 and EP4 by γδ T lymphocytes reported here enables soluble PGE2 to transduce a strong cAMP-mediated inhibitory signaling that efficiently blocks the physiological functions of Vγ9Vδ2 T lymphocytes.

Vγ9Vδ2 T-cells actively participate in anti-infectious immunity through the recognition of bacteria-derived phophoantigens 2. PGE2 release triggered by microbial products or inflammatory mediators secreted by macrophages 39, 41 could represent an important regulatory pathway to prevent deleterious excessive Vγ9Vδ2 T-cells' activation and to turn off their responses at the end of infection. This observation is also of importance with regard to the prospect of developing γδ T-cell-based cancer immunotherapy in humans, in which either treatment with BrHPP+IL-2 or autologous Vγ9Vδ2 T-cell therapy is expected to restore strong antitumor immunity 2, 17. Most if not all tumoral cells produce phosphoantigens triggering Vγ9Vδ2 T-cell proliferation, cytokine production and cytotoxicity 36. In addition, tumors frequently show progressive loss of cell-surface expression of HLA class I molecules 19, which thereby facilitates recognition by cytolytic NK cells and Vγ9Vδ2 T lymphocytes. Unfortunately, many tumors, especially solid ones, secrete high amounts of PGE2 in their microenvironment 41 and might therefore escape Vγ9Vδ2 T-cell cytotoxicity. Neutralizing the PGE2 pathway with COX2 inhibitors already aims at improving αβ T-cell-based cancer immunotherapy 42. Thus, this report suggests that such approaches could also benefit Vγ9Vδ2 T-cell activity. However, other cAMP-triggering soluble factors from the tumor microenvironment such as adenosine 43 might also be detrimental to the Vγ9Vδ2 T-cell functions as demonstrated in experiments conducted with forskolin.

Recent studies indicated that MSC are recruited to the stroma of developing tumor 20, 21. The inhibitory effects of MSC on phosphoantigen-activated Vγ9Vδ2 T-cells described in vitro, where the two cell types are in close vicinity, raise the question of MSC being able to temper their antitumor effect in vivo as well. Ren et al. 44 demonstrated that pro-inflammatory cytokine-activated mouse MSC attract αβ T-cells and inhibit their responses, both in vitro and in vivo, through the production of a wide array of chemokines including CXCL9, CXCL10 and CXCL11. Thus, activated Vγ9Vδ2 T-cells that express CXCR3 receptor for these chemokines 6 might also migrate toward activated MSC present within the tumor stroma and be targeted by the multiple inhibitory factors produced by these cells.

Based on the therapeutic benefit of MSC to control acute graft-versus-host diseases 45, MSC immunosuppressive properties are of substantial interest for several chronic inflammatory pathologies 23, 32. Indeed, in experimental models of rheumatoid arthritis and MS, MSC infusion reduced the extent of autoimmune disorders 46, 47. γδ T-cells have been implicated in the pathogenesis of these autoimmune diseases in which Th1 cytokines and cytotoxic factors are thought to play a critical role 48, 49. For example, in MS, γδ are present in the inflammatory lesions, and show evidence of clonal expansion 50–52. The regulatory cross-talk between MSC and γδ T-cells described in this report could contribute to the bioactivity of MSC in ongoing MS clinical trials.

MSC, as part of the bone marrow stromal microenvironment, secrete a number of cytokines, growth factors and extracellular matrix molecules that play a pivotal role in the proliferation, migration and differentiation of HSC 53, 54. Assuming that MSC in vitro reflect their physiological function, their suppressive activity in response to pro-inflammatory cytokines suggests that within the bone marrow, MSC could also protect normal hematopoiesis from harmful inflammation triggered by activated γδ lymphocytes 55.

Materials and methods


CFSE, indomethacin, 7-AAD, 1-MT, L-NAME, PGE2, purified mouse mAb against IFN-γ (clone 25718.11) and TNF-α (clone 28401.111) and forskolin were obtained from Sigma-Aldrich (St. Louis, MO, USA). Butaprost, sulprostone and 1-Hydroxy-PGE1, purified rabbit polyclonal antibodies directed against EP2, EP3 and EP4 were from Cayman Chemical (Ann Arbor, MI, USA). Indo-1 AM was from Molecular Probes (San Diego, CA, USA). Purified mAb against TGF-β (clone AF-101), HGF (clone 24612) and IL-10 (clone 25209) were obtained from R&D Systems (Minneapolis, MN, USA). FITC-conjugated anti-COX1, PE-conjugated mAb against TCRVγ9, IFN-γ, TNF-α and COX2 were purchased from BD Bioscience (San Jose, CA, USA). FITC-conjugated mAb anti-TCRVγ9 and FITC-conjugated goat F(ab′)2 anti-rabbit IgG were from Beckman-Coulter (Fullerton, CA, USA). Purified mouse anti-CD3 mAb was purchased from eBioscience (San Diego, CA, USA). Recombinant human IFN-γ and TNF-α were obtained from Peprotech (Rocky Hill, NJ, USA).

Isolation and culture of human MSC

Human MSC were isolated from PBS-washed filters used during bone marrow graft processing for allogeneic bone marrow transplantation. Cells were cultured at a density of 5×104 cells/cm2 in MEM-α medium (Invitrogen, San Diego, CA, USA) supplemented with 10% of FCS (Hyclone, Logan, UT, USA) and ciprofloxacin (10 μg/mL; Bayer Schering Pharma, Germany). After 72 h at 37°C in 5% CO2, non-adherent cells were removed and the medium was changed. Cultures were fed every 3–4 days for 21 days or until confluency. Adherent cells were then trypsinized, harvested and cultured at a density of 103/cm2 for 1–3 wk. MSC able to differentiate into osteoblasts, chondroblasts and adipocytes were CD34, CD45 and CD73+, CD90+, CD105+ (data not shown)

TCR Vγ9Vδ2 cells

Primary Vγ9Vδ2 T-cell lines were generated from PBMC as described previously 56; these comprised more than 95% TCRVγ9Vδ2+ cells. γδ T-cells were purified (≥90%) from freshly isolated PBMC using magnetically activated cell sorting (Miltenyi Biotec, Auburn, CA, USA).

Proliferation assays

The specified cells were labeled with 1 μM CFSE for 10 min at 37°C; 3×105 CFSE-labeled cells were cultured in 96-well plates with or without irradiated MSC or increasing doses of soluble PGE2. For Transwell cultures, 2×105 allogeneic MSC were seeded in the lower chamber of a 0.4 μm pore size membrane Transwell plate (BD Bioscience) and 1×106 CFSE-labeled PBMC were added in the upper chamber of the Transwell. After 6 days of culture in the presence of 300 U/mL IL-2 and 200 nM BrHPP, CFSE dilution and 7-AAD staining were evaluated by flow cytometry analysis. When specified, indomethacin (5 μM), 1-MT (1 mM), L-NAME (1 mM), anti-TGF-β, anti-IL-10 or anti-HGF mAb (all at 5 μg/mL) were added at the beginning of the culture.

Cytokine production

A total of 5×105 Vγ9Vδ2 T-cells or purified γδ T-cells were stimulated by 20 nM BrHPP with allogeneic MSC in 48-well plates for 24 h, with adding 10 μM Brefeldin A (Sigma-Aldrich) for the last 5 h of culture. Cells were stained with anti-TCRγ9-FITC, fixed with PBS 2% paraformaldehyde, stained for 30 min in PBS 1% saponin with the specified PE-conjugated mAb and analyzed by flow cytometry. In experiments using soluble PGE2, butaprost, sulprostone, 1-Hydroxy-PGE1 or forskolin, Vγ9Vδ2 T-cell lines were restimulated by 20 nM BrHPP for 5 h in the presence of 10 μM Brefeldin A.

MSC and Vγ9Vδ2 T-cell supernatant collection

MSC culture supernatants were obtained from a 3 day culture of 1×106 MSC plated alone or with 5×106 allogeneic Vγ9Vδ2 T-cells in 6-well plates. Vγ9Vδ2 T-cell line culture supernatants were obtained with 30×106 Vγ9Vδ2 T-cells cultured in 30 mL of complete medium with or without restimulation of 200 nM BrHPP. All supernatants were filtered with a 0.25 μm pore size filter and frozen at −20°C before use. IFN-γ and TNF-α depletions of Vγ9Vδ2 T-cell line culture supernatants were conducted by overnight incubation at 4°C with 20 μg/mL of either anti-IFN-γ, anti-TNF-α or IgG1-control mAb. 100 μg/mL of ProteinA-agarose were then added for 1 h and supernatants were recovered after centrifugation. IFN-γ and TNF-α levels, before and after depletion, were analyzed by flow cytometry using the “CBA beads Th1/ Th2” kit (BD Bioscience).

COX2 expression and PGE2 titration

A total of 2×105 MSC were seeded alone or in the presence of 1×106 allogeneic Vγ9Vδ2 T-cells with or without 200 nM BrHPP. After 24 h, intracellular staining using FITC-anti-COX1 and PE-anti-COX2 mAb was performed as described in the Cytokine production section, while PGE2 levels were measured by ELISA (“Prostaglandin E2 EIA Kit-monoclonal”, Cayman Chemical).

Cytotoxic assays

Standard 4 h 51Cr-release assays with 51Cr-labeled Daudi target cells were modified by adding either MSC culture supernatants or soluble PGE2 into the co-incubation, or by co-incubating Vγ9Vδ2 T-cells with MSC for 24 h at the specified cell ratios before the experiment.

Calcium flux assessment

Vγ9Vδ2 T-cells were loaded with 5 μM Indo-1 AM (Molecular Probes) for 45 min at 37°C and coated with 5 μg/mL anti-CD3 mAb for 30 min at 4°C in the presence or absence of 1000 ng/mL PGE2, and cross-linked with 15 μg/mL of goat anti-mouse mAb. Ca++ flux was based on Indo-1 AM 410/480 nm fluorescence analyzed by flow cytometry.

PGE2 receptor expression

Total RNA was isolated from Vγ9Vδ2 T-cell lines by TRIzol reagent (Invitrogen). RNA was reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Invitrogen) and cDNA encoding human EP1, EP2, EP3, EP4 and hypoxantine phosphoribosyl transferase (HPRT) were amplified by PCR using the following forward and reverse primers respectively:

5′-GGCCGCTATGAGCTGCAGT-3′ and 5′-GAGGTAGAGCTCCAGCCGC-3′ (EP1); 5′-GACCGCTTACCTGCAGCTGTA-3′ and 5′-GACTGAACGCATTAGTCTCAGAACA-3′ (EP2); 5′-AACGGGACTAGCTCTTCGCAT-3′ and 5′-AGTTGTTTGTGTGGTACCTGATCTG-3′ (EP3); 5′-GGTCATCTTACTCATTGCCACCT-3′ and 5′-GGAGGGTCTGAGATGTACTGCTG-3′ (EP4); 5′-ACTGAACGTCTTGCTCGAGATGT-3′ and 5′-GGTCCTTTTCACCAGCAAGCT-3′ (HPRT). The fragment sizes for EP1, EP2, EP3, EP4 and HPRT were 455, 409, 451, 377 and 369 bp, respectively. The corresponding receptor proteins were detected by flow cytometry using polyclonal rabbit primary antibodies directed against EP2, EP3 and EP4 (Cayman chemical) and FITC-conjugated goat F(ab′)2 anti-rabbit IgG secondary antibody.

Statistical analysis

Significant differences were assessed by Student's t-test (for normal distributions) or Mann–Whitney rank sum (for other distributions) two-tailed tests with a=0.01 with the SigmaStat software (Systat Software, San Jose, CA, USA).


This work is supported by institutional grants from the Institut National de la Santé Et de la Recherche Médicale (INSERM), the University Paul Sabatier, the Etablissement Français du Sang (EFS) and grants from the Association pour la Recherche sur le Cancer (ARC contract 3757, Mevalonate pathway and tumor immunity) and the Institut National du Cancer (INCa, contract 07/3D1616/IABC-23-8/NC-NG).

Conflict of interest: The authors declare no financial or commercial conflict of interest.