IFNγ Regulates Activated Vδ2+ T Cells through a Feedback Mechanism Mediated by Mesenchymal Stem Cells

γδ T cells play a role in a wide range of diseases such as autoimmunity and cancer. The majority of circulating human γδ T lymphocytes express a Vγ9Vδ2+ (Vδ2+) T cell receptor (TCR) and following activation release pro-inflammatory cytokines. In this study, we show that IFNγ, produced by Vδ2+ cells, activates mesenchymal stem cell (MSC)-mediated immunosupression, which in turn exerts a negative feedback mechanism on γδ T cell function ranging from cytokine production to proliferation. Importantly, this modulatory effect is limited to a short period of time (<24 hours) post-T cell activation, after which MSCs can no longer exert their immunoregulatory capacity. Using genetically modified MSCs with the IFNγ receptor 1 constitutively silenced, we demonstrate that IFNγ is essential to this process. Activated γδ T cells induce expression of several factors by MSCs that participate in the depletion of amino acids. In particular, we show that indolamine 2,3-dioxygenase (IDO), an enzyme involved in L-tryptophan degradation, is responsible for MSC-mediated immunosuppression of Vδ2+ T cells. Thus, our data demonstrate that γδ T cell responses can be immuno-modulated by different signals derived from MSC.


Introduction
Mesenchymal stem cells (MSCs) are multipotent non-hematopoietic precursors that can be isolated from various tissues and are capable of differentiation into multiple lineages, among them chondrocytes, adipocytes and osteocytes [1]. This notwithstanding, recent interest has focused on their potential clinical application based on their profound immunosuppressive properties. These studies have largely reported the capacity of MSCs to suppress proliferation and/or cytotoxic effector functions of distinct cells types of the innate and adaptive immune systems, such as T cells, Natural Killer (NK) cells, B cells and dendritic cells [2][3][4][5][6][7][8]. These properties are already being tested in numerous clinical trials worldwide. So far, none have reported significant side effects related to the transplantation of MSCs, which has encouraged the initiation of trials to treat practically any disease with links to autoimmunity (e.g. graft versus host a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 disease, pulmonary disease, solid organ transplant, rheumatoid arthritis or systemic lupus erythematosus) [5,[8][9][10][11].
MSCs home specifically to injured tissues, attracted by pro-inflammatory cytokines [3,12]. The immunosuppressive capacity of MSCs is not constitutive, but rather induced by crosstalk with cells of the immune system; thus, the inflammatory environment, and in particular the immune cells involved in each phase of an immune response, are likely to be critical triggers of this regulatory process. In recent years, several reports have demonstrated the role of interleukin-1 (IL-1), IFNγ and TNFα as main factors in this process [5,[13][14][15][16]. Thus, it is likely that induction of immunosuppression is not dependent on a single factor, but instead results from multiple regulatory mechanisms without an obvious hierarchy of importance. These molecules are clearly able to activate molecular pathways that increase production of soluble immunomodulatory factors such as indoleamine 2,3-deoxigenase (IDO) [3,17], prostaglandin E2 [18], iNOS (the murine counterpart of IDO) [13], transforming growth factor β (TGFβ), hepatocyte growth factor [4], human lymphocyte Ag molecule 5, and IL-10 [19]. The influence of these MSC-secreted factors on the immune system has been recently reviewed [20].
Regarding the targets of MSC-mediated immunoregulation, most work in the field has focused on conventional T cells (αβ T cells). By contrast, the effects of MSCs on γδ T cells have not been elucidated. γδ T cells express both the γδ TCR and natural killer receptors (e.g. NKG2D), and represent a link between innate and adaptive immunity [21,22]. In humans, γδ T cells are usually sub-divided based on use of one of two variable regions of the TCRδ-chain; Vδ1+ γδ T cells are largely found in epithelial layers such as skin and intestine, while Vδ2+ γδ T cells are mainly present in peripheral blood [23]. Most circulating Vδ2+ cells also use a Vγ9-containing TCRγ-chain, and are potently activated by low molecular weight non-peptidic phosphoantigens such a (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP), an intermediate metabolite from microbial isoprenoid biosynthesis. Vδ2+ cells have the ability to produce a variety of cytokines that regulate inflammation, eliminate pathogens, and maintain tissue homeostasis [21,24]. However, despite their beneficial roles, they have been implicated, like their αβ T cell counterparts, in the pathogenesis of a number of inflammatory diseases such as lupus erythematosus, rheumatoid arthritis, and psoriasis [25][26][27][28][29].
Several reports have demonstrated the inhibitory function of human bone marrow MSCs on Vδ2+ cells, mainly through PGE2 [30][31][32][33][34]. All of these studies used chemical inhibitors to identify and discriminate between different effector molecules secreted by MSCs. Since activated Vδ2+ cells produce pro-inflammatory cytokines upon activation, we aimed to elucidate to what degree other pathways were involved in MSC-mediated immunoregulation. Here, we report that the IFNγ/IDO pathway is a key factor for MSC-induced immunoregulation of Vδ2+ cells.

MSC culture
Human bone marrow-derived MSCs were obtained from the Inbiobank Stem Cell Bank (http://www.inbiomed.org/Index.php/servicios_externos/inbiobank) as described previously [35]. In short, cadaveric marrow was obtained from brain-dead donors after informed consent and under the Spanish National Organization of Transplant supervision (ONT). MSCs were positive for CD29, CD73, CD90, CD105, CD166 and CD146 but negative for markers of the hematopoietic lineage; CD34, CD45, CD14, CD19 and CD31. Moreover, they displayed a fibroblast-like phenotype and showed at least a tri-lineage potential differentiating into osteocytes, chondrocytes and adipocytes. MSCs were cultured in DMEM low-glucose medium supplemented with 10% FBS (Lonza, Walkersville, MD, USA), 2 mM glutamine, 100 U/ml penicillin and 0.1 mg/ml streptomycin (Sigma, St. Louis, MO, USA). Upon reaching confluence MSCs were treated with 0.25% Trypsin-EDTA solution (Sigma) and seeded at a density of 1000-1500 MSC/cm 2 . Cells were obtained from the Inbiobank Stem Cell Bank at passage three and all experiments were carried out with cells from low passages (passage number 4-8).

PBMC
Peripheral blood from healthy donors was obtained after informed consent from the Basque Biobank for Research OEHUN (http://www.biobancovasco.org). Mononuclear cells (MNCs) were prepared using Ficoll Paque (Lymphprep, Axis-Shield, Oslo, Norway) according to the manufacturer's instructions. Cells were cultured in RPMI 1640 Dutch modification (Gibco, Grand Island, NY, USA) supplemented with 10% FBS (Lonza), 2 mM glutamine, 100 U/ml penicillin, 0.1 mg/ml streptomycin (Sigma), 10 ng/ml recombinant human Interleukin-2 (IL-2, R&D, Minneapolis, MN, USA), and 1 mM (E)-4-hydroxy-dimethylallyl pyrophosphate (HDMAPP, Cayman Chemical) for the indicated times. In co-culture MNCs were activated and cultured in the presence of MSCs for the indicated times and ratios under the conditions described above. Co-culture was performed either in direct cell-to-cell contact or MNC were separated from MSCs by a transwell system (Corning, NY, USA). The inhibitor 1-Methyl-DLtryptophan (1mM) (Sigma) was added at initiation of the co-culture.

Lentiviral transduction of MSCs
Oligonucleotide sequences were validated at the RNAi Consortium and were purchased from Sigma. Primer sequences were as follows: IFNγRi: Fwd 5'-CATGAACCCTATCGTATATTG and Rev 5'-CATGAACCCTATCGTATATTG;IDOi: Fwd 5'-ACTGGAACTGCCTCCTATT and Rev 5'-AATGGAACTGCCTCCTATT. After annealing, the respective primer pairs were first cloned into the pSUPER plasmid and subsequently sub-cloned into the pLVTHM vector (Addgene, Cambridge, MA, USA). Viral particles were produced using the Viral Vector Platform at Inbiomed Foundation (http://www.inbiomed.org) and MSCs were transfected at a multiplicity of infection (MOI) of 10 in order to obtain a transduction efficiency of 100%.

Real-Time quantitative PCR
Total RNA extraction from MSCs and DNAse treatment was done using the RNAqueous1-Micro Total RNA Isolation Kit (Ambion, Carlsbad, CA, USA). Reverse transcription of RNA to cDNA was performed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA, USA) following the manufacturer's instructions. Quantitative Real-Time PCR was carried out on a Thermocycler Step One Plus (Applied Biosystems) using 5x PyroTaq PROBE qPCR Mix Plus (ROX) from CMB. Data were normalized to MSC-PLVTHM using the ΔΔCt method and GAPDH as housekeeping gene. Primer sequences were as follows: IFNR Fwd 5'-TCCAGGCATGCATACCGAAGACAA and Rev 5'-ATGCTGCCAGGTTCAGAC TGGTTA; IDO Fwd 5'-CTACCATCTGCAAATCGTGACTAAGT and Rev 5'-GAAGGGTCTTC AGAGGTCTTATTCTC; GAPDH Fwd 5'-TGCACCACCAACTGCTTAGC and Rev 5'-GGCAT GGACTGTGGTCATGAG.

Statistical analysis
Data were summarized by mean and standard deviation. Statistical analyses were conducted using the paired t-test. p-values less than 0.05 were considered statistically significant.

MSCs inhibit expansion of Vδ2+ cells mainly by soluble mediators
Vγ9Vδ2+ (Vδ2+) cells were activated from total PBMCs using HDMAPP and rh-IL2, and cocultured with increasing numbers of MSCs. After 7-days the mononuclear fraction enriched in Vδ2+ cells was collected and analysed by flow cytometry; gating out CD4+ and CD8+ T cells, and focusing on CD3+Vδ2+ cells. As shown in Fig 1A and 1B, MSCs inhibited the expansion of Vδ2+ cells in a dose-dependent manner. As an MSC:MNC ratio of 1:25 completely abolished Vδ2+ cell proliferation, we decided to use this condition thereafter. As expected, CFSE-labeled activated Vδ2+ cells were unable to expand in the presence of MSCs, in contrast to activated Vδ2+ cells cultured without MSCs (Fig 1C), suggesting that this was not related to cell death. Proliferation of Vδ2+ cells without MSCs gave a proliferation index of 3.35 ± 0.39 as shown by decline of CFSE fluorescence. By contrast, activation of Vδ2+ cells in the presence of MSCs led to a proliferation index three times lower than control (1.59 ± 0.34) (Fig 1C, right  panel). Next, we determined whether soluble factors secreted by activated Vδ2+ cells or cellto-cell contact were responsible for the observed effect. Activated Vδ2+ cells were co-cultured with MSCs either in direct contact or in transwell dishes. As shown in Fig 1E, MSCs inhibited the expansion of Vδ2+ cells in both types of culture demonstrating that soluble molecules were responsible for the induction of immunoregulation.

IFNγ signalling is necessary for induction of immunosuppression by MSC
The induction of immunosuppression by MSCs on Vδ2+ cells can be attributed to various factors. As activation in vitro of Vδ2+ cells rapidly induces a production of pro-inflammatory cytokines such as IFNγ [37][38][39][40], we decided to investigate the role of this cytokine in the induction of MSC-mediated immunomodulation. As shown in Fig 2A and 2B, MSCs caused a significant reduction in IFNγ production by activated Vδ2+ cells even after short periods of time (4, 6 and 12 hours). Interestingly, when Vδ2+ cells were pre-activated for shorter than 12 hours (data not shown) or longer than 24 hours, MSCs failed to show any inhibitory effect ( Fig  2C). Indeed, pre-activation with HDMAPP and IL-2 for longer than 48 hours gave rise to an opposite effect when in the presence of MSCs, as higher percentages of Vδ2+ cells were observed, probably due to production of pro-survival factors by the MSCs (e.g. IL-6) [41]. These data indicate that MSCs exert their inhibitory effect during a critical time window after which they cannot develop their regulatory capacity.
We next analysed whether inhibition of IFNγ receptor expression correlated with decreased capacity of MSCs to regulate Vδ2+ cell proliferation. We used a GFP-expressing lentiviral vector (pLVTHM) to transduce MSCs with a shRNA that targeted the IFNγ receptor. Fig 3A shows a representative FACS plot of GFP expression in MSCs 4 days after shRNA transduction. A multiplicity of infection (MOI) of 10 resulted in~90% of cells being transduced with either empty vector (MSC-pLV) or pLVTHM-IFNγR (MSC-IFNγRi). Stable transduction resulted in a significant reduction (to less than 5% of control) of IFNγ receptor expression in MSCs (Fig 3B). Transduction did not lead to increased cell death as shown by AnnexinV staining (Fig 3C). More importantly, Vδ2+ cells expanded by~2-fold more in co-cultures with MSCs transduced with shRNA against the IFNγ receptor, compared to co-cultures with MSCs transduced by empty vector MSC-pLV (Fig 4A and 4B). This result also correlated with data obtained by CFSE staining ( Fig  4C); the proliferation index of Vδ2+ cells being increased 2-fold when the IFNγ receptor was silenced in MSCs (2.23 ±0.43 in MSC-IFNγRi vs 1.59 ±0.34 in MSC-pLV). Interestingly, cytokine production by Vδ2+ cells was partially restored by inhibition of IFNγ receptor expression in MSC, demonstrating statistical significance when we compared MSC-pVL vs. MSC-IFNγRi at 12 hours post-activation (Fig 4D). Taken together, our data demonstrate a feedback loop in which IFNγ produced by activated Vδ2+ cells can induce immunosuppressive capacity in MSCs, which in turn, can inhibit both proliferation and cytokine production of activated Vδ2+ cells.

IDO expression by MSC induces immunosuppression of Vδ2+ cells
To further investigate the importance of IFNγ for induction of MSC-mediated immunosuppression we modulated signalling downstream of its receptor. The expression of indoleamine 2,3-deoxigenase (IDO), an enzyme involved in tryptophan catabolism, is able to modulate αβ T cell activity in response to IFNγ [3,5,17]. Thus, we asked whether IDO is also a key player in the immunoregulation of Vδ2+ cells by MSCs by applying the same experimental approach as described above; IDO expression was silenced using shRNA delivered by a lentiviral vector. As IDO is not constitutively expressed in MSC [5,42], transcription was assessed in the presence of activated Vδ2+ cells. After 24-hour co-culture, quantitative Real-Time PCR for IDO clearly showed that the expression was significantly reduced in MSC transduced with specific shRNA (MSC-IDOi) compared to cells transduced with empty vector (Fig 5A). To demonstrate that IDO expression was induced in the presence of IFNγ produced by activated Vδ2+ cells, we included "knock-down" MSCs for IFNγ receptor (MSC-IFNγRi) in the same analysis. Data shown in Fig 5A confirmed that IDO expression was also reduced in MSC-IFNγRi cells.
We next evaluated the effects that MSC-IDOi cells have on Vδ2+ cells with respect to proliferation and intracellular IFNγ production. First, we checked the expansion of Vδ2+ cells cultured in the presence MSC-IDOi. As shown in Fig 5B and 5C, interfering with the IFNγ pathway at the intracellular level augmented the percentage of Vδ2+ cells in culture. Whereas Vδ2+ cells were 5.67% ±2.44 of total cells in the presence of MSCs transduced with empty vector, they constituted 11.08% ±1.24 in the presence of MSC-IDOi (11.08%), a level comparable tot that seen in the presence of MSC-IFNγRi. Consistent with this, activated CFSE-labeled Vδ2+ cells cultured with MSC-IDOi proliferated more than those cultured with MSC-pLV ( Fig 5D). Importantly, the absence of IDO enzyme in MSC-IDOi allowed Vδ2+ cells to maintain IFNγ expression levels 12-hours post-activation when compared to MSCs transduced with the empty vector (MSC-pLV) alone. Notably, silencing of IDO resulted in a slightly lower percentage of Vδ2+ cells and a lower proliferation index compared with silencing of the IFNγR, which may indicate that IFNγ production is delayed in MSC-IDOi compared to MSC-IFNγRi (Fig 5E). To summarize, our data clearly show that IFNγ via the IFNγR-IDO pathway plays an important role in the immunoregulation of Vδ2+ cells by MSCs.

Discussion
In the past decade MSC-mediated immunoregulation of T cells has attracted increasing interest due to its potential clinical application in autoimmune pathologies. Many studies have identified molecular mechanisms that underpin the immunoregulatory properties of MSCs [43]. In this study, we focused on MSC-mediated immunoregulation on Vδ2+ γδ T cells, a diverse subset that bridges innate and adaptive immunity in terms of their activation and effector functions. We demonstrate that MSCs inhibit the proliferation of Vδ2+ cells in a dosedependent manner. Notably, this inhibitory effect is independent of cell-to-cell contact. We did not observe an increase in Vδ2+ cells after physical separation from MSCs, indicating that soluble mediators likely drive the immunoregulatory effects. Likewise, the immunomodulatory properties of MSCs are induced by cytokines secreted in their environment and/or molecules expressed by target cells such as Vδ2+ cells. Vδ2+ cells produce significant amounts of proinflammatory cytokines like TNFα and IFNγ in order to counteract bacterial infections or tumour development [44]. Although other studies have demonstrated MSC-mediated inhibition of both cytokines [32], we here demonstrate that IFNγ produced by Vδ2+ cells induces MSC-mediated immunosuppression of Vδ2+ cells, as judged by cytokine production and proliferation, in a negative feedback loop. Our experimental approach interfered directly with the IFNγ pathway in MSCs by silencing IFNγR1, one of the subunit of the IFNγR involved in IFNγ signalling. We showed that in the presence of MSC that lacked IFNγR1, we partially restored Vδ2+ cell expansion and intracellular IFNγ production, in contrast to the when MSC were transduced with empty vector (MSC-pLV) alone. These results indicate that the inhibitory effect exerted by MSCs on Vδ2+ cells can affect different aspects of activation and effector function, ranging from cytokine production to proliferation. Interestingly, this mechanism does not appear to be specific for Vδ2+ cells; other studies have reported that MSCs inhibit other cells (such as NK and dendritic cells) by limiting early production of IFNγ and TNFα [18,45]. Whether this is a general anti-inflammatory mechanism employed by MSCs requires further investigation. However, we speculate that low concentrations of IFNγ are probably sufficient to promote MSC-mediated immunoregulation, which in turn, rapidly inhibits the production of this pro-inflammatory cytokine, leading to its elimination from the inflammatory environment and its capacity to active other immune cells. In this regard, it was shown that an anti-IFNγ neutralizing antibody significantly reduced the ability of HMBPP-activated Vδ2+ cells to antagonize regulatory T cell expansion [46]. Importantly, the MSC-mediated modulatory effects were limited to a short time window (<24 hours), after which MSCs appear to have no effect on the expansion of Vδ2+ cells. It is noteworthy that we have observed the same effect on activated αβ T cells (data not shown). This correlates with the antigen-presenting properties of MSCs that also occurs during a narrow window at low levels of IFN-γ [47]. Further investigations are required to determine if these observations in vitro are related, and how they translate to disease settings in vivo.
IFNγ can, through IFNγR, activate several genes that contain interferon-response-elements (IRE) in their promoter regions [48,49]. Among these is IDO, which catalyzes the degradation of L-tryptophan to N-formylkynurenine, and has been reported to suppress αβ T cell responses [50][51][52]. In our hands, IFNγR and IDO gene knock-down in MSCs gave similar results. These data contrast those obtained by Martinet et al., that showed that a specific chemical inhibitor of IDO (1-methyl-DL-tryptophan (1-MT)) had no effect on MSC-mediated inhibition of Vδ2+ cells [31]. This may be due to differences in experimental design; one explanation could be that different phosphoantigens have been used for Vδ2+ cell activation (BrHPP vs. HDMPP in our study). Second and maybe more importantly, we have observed that 1-MT by itself has a profound effect on proliferation of Vδ2+ cells (S1 Fig). However, it is clear from both studies that several mechanisms are operating in Vδ2+ cells immunosuppression: activated Vδ2+ cells produce IFNγ and TNFα, which are able to induce IDO (mainly by IFNγ) and COX2 (IFNγ and TNFα [31]) in MSCs. Most studies that aim to elucidate the molecules responsible for MSC-mediated immunoregulation have been performed on αβ T cells. Similar to αβ T cells, human Vδ2+ cells display remarkable functional plasticity, with reports describing production of a range of effector molecules that depends on the conditions of activation [23]. How MSCs effect this functional differentiation in vivo, and the implications for health and disease, are still unknown. γδ T cells were previously implicated in pathogenesis in animal models of autoimmune diseases such as rheumatoid arthritis and multiple sclerosis in which Th1 cytokines are thought to play a central role [53,54]. Moreover, they contribute to other autoimmune diseases such as psoriasis in which epidermal CCR6+ Vδ2+ cells express high levels of IL-17 and IL-22 [55]. IL-17+ Vδ2 + cells are known to express low levels of IFNγ [56]. Hence, it remains to be determined how these different subtypes of Vδ2+ cells would interact with MSCs, which molecules would be involved, and to what extent the role of MSCs might be beneficial in the immunotherapy of Vδ2+ cell-mediated diseases.
Finally, our findings also contrast previous reports suggesting that MSC-associated immunosuppression of Vδ2+ cells is exclusively mediated by PGE2. As shown in this study, IDO is also involved, suggesting that several pathways underpin the immunomodulatory capacities of MSCs.