Tumour Cell Generation of Inducible Regulatory T-Cells in Multiple Myeloma Is Contact-Dependent and Antigen-Presenting Cell-Independent

Regulatory T-cells (TReg cells) are increased in patients with multiple myeloma (MM). We investigated whether MM cells could generate and/or expand TReg cells as a method of immuno-surveillance avoidance. In an in vitro model, CD4+CD25- FoxP3 - T-cells co-cultured with malignant plasma cells (primary MM cells and cell lines) induced a significant generation of CD4+CD25+ FoxP3 + inducible TReg cells (tTReg cells; p<0.0001), in a contact-dependent manner. tTReg cells were polyclonal, demonstrated a suppressive phenotype and phenotypically, demonstrated increased FoxP3 (p = 0.0001), increased GITR (p<0.0001), increased PD1 (p = 0.003) and decreased CD62L (p = 0.007) expression compared with naturally occurring TReg cells. FACS-sorted tTReg cells differentiated into FoxP +IL-17+ and FoxP3 -IL-17+ CD4+ cells upon TCR-mediated stimulation. Blocking experiments with anti-ICOS-L MoAb resulted in a significant inhibition of tTReg cell generation whereas both IL-10 & TGFβ blockade did not. MM tumour cells can directly generate functional TReg cells in a contact-dependent manner, mediated by ICOS/ICOS-L. These features suggest that tumour generation of TReg cells may contribute to evasion of immune surveillance by the host.


Introduction
The paradoxical observation of tumor growth despite an attempt by the tumour-bearing host immune system to control and eliminate the malignant cells suggests that the anti-tumour immune response is being attenuated limiting competent immune surveillance (reviewed in [1]). This has been extensively studied by tumour immune-biologists, with results pointing towards soluble factors and altered antigenicity as mechanistic explanations. More recently, with the discovery of a number of different immuneregulatory cell types, focus has shifted towards cellular mediated tumour-induced immune suppression and evasion. Several different subsets of regulatory T-cells have now been identified including naturally occurring T Reg cells (nT Reg cells: CD4 + CD25 + FoxP3 + ), and inducible T r1 and T H3 CD4 + T Reg cells [2] as well as CD8 + T Reg cells [3] and Double Negative T Reg cells.
Originally it was thought that T Reg cells were centrally generated in the thymus, though more recently evidence suggests that peripheral generation is also possible, thereby providing a biological back-drop to investigating their role in the cancerbearing host [4,5]. In fact, several studies have shown that increased levels of T Reg cells can be found in a variety of solid tumours [6,7] and haematological malignancies [8,9,10].
Multiple Myeloma (MM), an incurable malignant plasma cell dyscrasia, is associated with both cellular and humoral immune deficiencies [11]. Many potential mediators of the immunologically hostile microenvironment have been proposed including tumourderived TGFb [12], Prostaglandin E 2 (PGE 2 ) and Interleukin-10 (IL-10) [13]. In addition to soluble mediators, we and others have demonstrated that T Reg cell subsets are functional and increased in the peripheral blood of patients with MM, associated with their disease burden [14]. In particular, we demonstrated a higher level in the ''pre-myelomatous'' condition, MGUS but to a lesser extent than when full disease is present though higher levels of IL-10 were seen in the PB of MGUS compared with patients with MM. In light of this recent evidence, it would now seem that the most promising and synergistic approaches for cancer immunotherapy will be strategies that augment specific anti-tumor immunity whilst simultaneously reducing the effect of tumour-induced immuneregulation. However, in order to perform this later component, a greater understanding of the in vivo mechanism of tumour-induced immune suppression is needed.
In this study, using an in vitro model system, we demonstrate that the tumour cells of MM are not only capable of expanding nT Reg cells but generating T Reg cells de novo, mediated through cell contact. Through our experimentation, we demonstrate that surface ICOS-L on the tumour cells mediates this phenomenon and that the tumour-induced T Reg cells whilst sharing some are phenotypic features also display phenotypic differences but are functionally similar to nT Reg cells. The data presented here provides further evidence of direct tumour manipulation of the immune system to augment immune evasion and propagation of the malignant cell clone.
Primary samples were obtained from patients with myeloma (n = 9) through iliac crest aspirations. The study was approved by the local ethics committee and written informed consent was obtained (NRES Committee Yorkshire & The Humber -Leeds East: Ref 04/Q1206/147). All samples were collected in sterile EDTA containers and mononuclear cells (MNC) were isolated by density gradient centrifugation on Lymphoprep (Axis-Shield, UK) and stored in foetal calf serum with 10% DMSO in the vapour phase of liquid nitrogen at -270uC until the day of analysis.

In vitro Modeling of Tumour Cell and T-cell Interactions
Mononuclear cell (MNC) preparations were made from leukocyte concentrates provided by the National Blood Service. MNC were isolated by density gradient centrifugation on Lymphoprep (Axis-Shield, UK) and washed three times in PBS before use in the culture system. Peripheral blood lymphocyte (PBL) preparations were made by monocyte-depletion of the MNC fraction through plastic adherence by 2-hour incubation of MNC in CM at 37uC. Mitomycin C-treated HMCL were added to MNC/PBL with a responder:HMCL ratio of 2:1 at 1610 6 PBMC/ml with MNC-only controls treated the same way. Supernatants were collected on day 7 of culture and frozen immediately at 280uC for later cytokine assessment. Cells were harvested on the same days for analysis by Flow cytometry.

Cell Sorting and FACS Analysis
Four-colour flow cytometry was performed on a LSRII (BD Biosciences) and analysed with FACS DIVA software. Directly conjugated mAbs against CD4-APC, CD8-APC, CD3-PerCP, CD25-PE (all from BD Biosciences, Oxford, UK) and FoxP3-FITC (eBioscience, San Diego, USA) were used according to the manufacturer's protocol with corresponding isotype-matched controls. 1610 6 cells were stained. The fixation and dead cell discrimination kit (Miltenyi Biotec, Bergisch-Gladbach, Germany) was used to exclude dead cells within the intracellular staining protocol. Using a sequential gating strategy, T Reg cells were identified as CD4 + CD25 + FoxP3 + T-cells and expressed as a percentage of the CD4 + T-cell population. In order to perform functional analysis on the in vitro generated T Reg cells, cells were pre-selected through magnetic cell separation by using the CD4 untouched method (Miltenyi Biotec, Bergisch-Gladbach, Germany) according to the manufacturer's protocol and then FACS sorted using the surface antibodies CD4-APC, CD25-PE and CD127-Pacific Blue. Samples were sorted into a CD4 + CD25 + CD127 -T Reg cells and CD4 + CD25 -CD127 + effector T-cells the MoFlo high performance multi-parameter cell sorter.

Proliferation and Suppression Assays
MNC were sorted into CD4 + CD25effector cells and CD4 + CD25 + T Reg cells as described above. The CD4 + CD25responder cells were plated in 96 well round bottom plates (Nunc plates, Thermo Fisher Scientific, Roskilde, Denmark) in triplicates at a concentration of 1610 5 cells per well in CM. Purified CD4 + CD25 + T Reg cells were added at different concentrations (4:1 and 8:1 responder to suppressor ratio). The suppressive capability of the T Reg cell fraction was determined by 3 H-Thymidine incorporation for 18 hours at 1 mCi per well after 72 hours stimulation with CD3/CD28-Antibiotin MACSIbeads (Miltenyi Biotec, Bergisch-Gladbach, Germany) at a 1 bead: 2 cell concentration. 3 H-thymidine incorporation in the stimulated responder only wells was set as 100% and a stimulation index (SI) calculated. Where indicated, cell populations were stained with 2 mM CFSE before co-culture and analysed by FACS.

Th 17 T-Cell Analysis
To assess if tumour-generated regulatory T-cells have the same capability to produce IL17-producing T-cells (Th 17 cells) as nT Reg cells, both populations were obtained on day 7 of co-culture with HMCL and day 0, respectively. Sorted cells populations were stimulated with anti-CD3 and anti-CD28 coated beads at a cell to bead ratio of 2:1 and cultured for 5 days in CM at 1610 6 cells per ml. Six hours prior to intracellular staining, Brefeldin A (BFA) (10 mg/ml), Ionomycin (1 mM) and phorbol myristate acetate (PMA) (20 ng/ml) were added. Cells were then washed and stained as per intracellular staining protocols using the dead cell discriminator with CD4 Pacific Blue, FoxP3 FITC and IL-17A AlexaFluor 647 (eBioscience, clone eBio64DEC17), using their corresponding isotypes as controls, analysed using the LSRII as described above.

Cytokine Assessment
Capture and detection antibodies were used (BD Biosciences, Oxford, UK) according to the manufacturers protocol. In short, high protein binding 96 well ELISA plates (MaxiSorp, Scientific Laboratory Supplies Ltd., Hessle, UK) were coated at 4uC overnight with IL-10 and TGFb capture antibodies at 1:500 dilution in 1x 1 M NaHCO 3 pH 8.2 at 100 ml per well. For TGFb ELISA, serum samples were diluted 1:5 with PBS and activated with 1N HCL at room temperature for 15 minutes and neutralized with 1N NaOH. After blocking with PBS containing 10% FCS for 2 hrs at room temperature, samples and standards were loaded at 100 ml per well and incubated at 4uC overnight. 100 ml per well detection antibody was then added at 1:1000 dilution for IL-10 and 1:500 dilution for TGFb and incubated for 2 hrs at room temperature followed by Extravidin-Avidin conjugate (100 ml per well at 1:500 dilution in PBS/Tween for 1 hr) and substrate solution (Sigma, Dorset, UK) for approximately 30 minutes for development in the dark. Samples were analysed in triplicate and measured spectrophotometrically at 405 nm. For LUMINEX Extracellular assay, spectrally encoded antibody-conjugated 5.6 mm polystyrene beads were used according to the manufacturer's protocol. Plates were pre-wet and 25 ml antibody coated beads and 200 ml wash solution were added and washed once. Then, 50 ml incubation buffer was added to 100 ml standard or 50 ml sample/50 ml assay diluents. After a 2 hour incubation and washing, the plate was then incubated with 100 ml PBS with the cytokine specific biotinylated detector antibodies. The fluorescent streptavidin-RPE was added and after incubation was analysed with the Luminex IS software.

T-cell Receptor Clonality by PCR
T-cell receptor (TCR) clonality was determined by PCR analyses of TCRG rearrangements as previously described [20]. In brief, DNA was isolated from FACS sorted cells and subjected to PCR performed using the BIOMED-2 multiplex strategy (InVivoScribe Technologies, San Diego, CA). PCR products were labeled in the 6FAM, HEX and NED fluorochromes and Vc usage was identified using ABI Fluorescence detection. Positive controls for clonal T-cell populations were derived from the peripheral blood of patients with T-cell lymphoproliferative disease.

Statistical Analysis
Results were analysed using SPSS version 14.0 for Windows software. Multiple independent variables were analysed with the . Results represent all experiments, expressed as mean 6 SEM and analyzed using a 1-way ANOVA and student t-test. B. Expansion of nT Reg cells when co-culture with HMCL results from cell division, illustrated by a representative flow cytometry plot of CFSE dilution. C. The generation of FoxP3 + CD25 + CD4 + T-cells, expressed as a percentage of CD4 + T-cells, in a co-culture assay with mitomycin C-treated U266B cells (n = 6) with varying starting populations: PB MNC, PBL CD25 depleted and CD4 + CD25 -T-cells. Results demonstrate that increased generation of tumour-induced regulatory T-cells (tT Reg cells) is seen with increasing purity of the seeded population. Results represent all experiments, expressed as mean 6 SEM and analyzed using a 1-way ANOVA and student t-test. D. Representative flow cytometry plots demonstrating the generation of FoxP3 + CD25 + CD4 + T-cells from CD4 + CD25 -T-cells through cell division of de novo generated FoxP3 + T-cells in a 7 day co-culture assay with mitomycin C-treated U266B cells. E. The generation of FoxP3 + CD25 + CD4 + T-cells, expressed as a percentage of CD4 + T-cells, in a co-culture assay of CD4 + CD25 2 T-cells (n = 10) with mitomycin C-treated MM cell lines (U266B, JJN3, JIM3 & KMS11), an erythro-leukaemia cell line (K562) and nonheamatopoietic cell lines (Mel888 & HeLa). Results represent all experiments, expressed as mean 6 SEM and analyzed using a 1-way ANOVA and student t-test (**p,0.001, *p,0.01). F. The generation of FoxP3 + CD25 + CD4 + T-cells, expressed as a percentage of CD4 + T-cells, in a co-culture assay with fresh BM-derived myeloma plasma cells from patient samples (n = 7). Results demonstrate that increased generation of tumour-induced regulatory T-cells (tT Reg cells) is seen with primary myeloma cells. Results represent all experiments, expressed as mean 6 SEM and analyzed using a 1-way ANOVA and student t-test. doi:10.1371/journal.pone.0035981.g001 Kruskal-Wallis test for non-parametric samples and with the Mann-Whitney-U test for 2 independent samples. A p-value of ,0.05 was considered statistically significant. Comparison of patient samples was expressed as median values and co-culture experiments as mean values.

Malignant Plasma Cells Induce Regulatory T-cell Generation
We have previously shown an increase in functional T Reg cells in the peripheral blood of patients with MM, relating to the stage of their disease [14]. To examine the relationship between myeloma tumour cells and T Reg cells, we first determined the effect of co-culturing naturally-occurring T Reg cells (nT Reg cells) with mitomycin-C treated HMCL (U266B HLA-class II Pos ). When nT Reg cells were sorted from the PB of healthy volunteers and cultured in CM alone, a significant reduction in the proportion of nT Reg  Next to determine if this was a MM-specific effect, we cocultured CD4 + CD25 -T-cells with a selection of HMCL (U266, JJN3, JIM3 & KMS11), a myeloid-derived cell line (K562) and non-heamatopoietic cell lines (Mel888 & HeLa). A clear induction of T Reg cells was seen with each of the HMCL and K562, but not the non-haematopoietic cell lines MEL888 or HeLa cell lines (n = 6, 1-way ANOVA p = 0.0015; Figure 1E). When sorted primary bone marrow plasma cells taken from patients with myeloma (n = 7) were co-cultured with CD4 + CD25 -T-cells from healthy donors, a significant generation of T Reg cells was seen (1.2%60.31 vs 12.0264.4, n = 7; p = 0.004), similar to the HMCL, U266B (1.2%60.31 vs 21.9%65.6, n = 7; p,0.0001; Figure 1F).

Tumour-generated Regulatory T-cells are Phenotypically Different to Natural T Reg Cells
Differences in phenotype between naturally occurring and inducible T Reg cells have been reported [21,22]. We therefore sought to characterize the phenotype of tT Reg cells generated in our in vitro assay compared with naturally occurring T Reg cells selected from steady PB of healthy volunteers. Given the potential for heterogeneity of response between the different samples from healthy volunteers, we utilized the one HMCL to provide consistency in the in vitro model, though similar results were generated using other MM cell lines (JIM3, JJN3 & RPMI8226data not shown). When CD4 + CD25 -T-cells were selected as the starting population, the level of FoxP3 expression was significantly greater than naturally occurring T Reg cells either from the PB of healthy controls or patients with MM (15856101 vs 884667,p,0.0001, Kruskal-Wallis test; Figure 2A). Next, using a sequential gating strategy, we examined the expression of key surface markers on CD4 + CD25 + FoxP3 + T-cells. tT Reg cells demonstrated a similar level of CD127 (p = 0.413) and CD4 (p = 0.415) expression but demonstrated significantly higher levels of CD25 (43,49266800 vs 18966137,p,0.0001), GITR (7065 vs 1063,p,0.001) and PD-1 (49.869 vs 5.360.8,p = 0.003), as illustrated in Figure 2B and C. With regards to CD62L, there was an overall lower mean fluorescence intensity (MFI) compared to naturally occurring T Reg cells (88.960.54 vs 97.360.54, p = 0.008; Figure 2C), but a bi-phasic pattern of expression suggests two populations of cells, some of which demonstrated similar expression of CD62L as naturally occurring T Reg cells ( Figure 2C). To determine the clonality of tumour-induced T Reg cells, CD4 + CD25 + CD127 Dim T-cells were FACS sorted after 7 days of co-culture with mitomycin-C treated HMCL and DNA prepared from sorted cell populations. TCRG PCR was performed on genomic DNA derived from the tT Reg cells. The spectrograph indicates multiple ''spikes'' representative of a polyclonal population in respect to the TCRG rearrangements, compared to a single ''spike'' representative of a monoclonal population ( Figure S1).

tT Reg Cells though Functionally Similar to nT Reg Cells Produce Interferonc
It has been reported that T Reg cells from tumour-bearing hosts demonstrate altered suppressive capabilities [9,23] though our studies in myeloma patients demonstrate that T Reg cells are functionally active in suppression of autologous T-cell responses to TCR stimulation [14]. First we sought to determine the proliferative response of tT Reg cells to TCR-mediated stimulation. CD4 + CD25 -T-cells were isolated and co-cultured with HMCL for 7 days then CD4 + CD25 + CD127 Dim T-cells (tT Reg cells) were FACS-sorted. tT Reg cells were stimulated using CD3/CD28coated beads for 5 days, determining their proliferative response by tritiated thymidine incorporation, comparing their response to sorted nT Reg cells from healthy donors and patients with myeloma, similarly stimulated. tT Reg cells demonstrated greater proliferative responses to TCR-mediated stimulation compared with nT Reg cells from normal controls and MM patients, who demonstrated the weakest proliferative responses (1619361860 cpm vs 15106314 cpm vs 605673 cpm, p,0.001; 1 way ANOVA). Next we examined their suppressive capabilities. tT Reg cells generated in a 7 day co-culture were FACS-sorted and co-cultured with autologous T-cells stimulated with CD3/CD28coated beads at the ratios described, for 5 days. The suppressive capacity of tT Reg cells was compared with nT Reg cells from healthy controls. We demonstrate that tT Reg cells were able to suppress anti-CD3/anti-CD28-induced T-cell proliferation in a dose dependent fashion similar to naturally occurring T Reg cells ( Figure 3A). Next we sought to determine the cytokine production by tT Reg cells in this culture system. When the supernatant was analysed for IL-10 on Day 7, the co-culture of T-cells with HMCL generated significantly higher levels of IL-10 compared to HMCLs or CD4 + CD25 -T-cells cultured alone (p,0.001; Figure 3B). However, when the production of IL-10 by tT Reg cells was determined at the single-cell level by FACS, very few tT Reg cells produced IL-10 ( Figure 3C). When the culture supernatant was examined for the level of Interferonc (IFNc), the co-culture of Tcells with HMCL generated significantly higher levels of IFNc compared to either HMCLs or CD4 + CD25 -T-cells cultured alone (p,0.0006; Figure 3D). We sought to determine the cellular origin of IFNc and demonstrated that IFNc-producing tT Reg cells could readily be identified, contributing to the production of IFNc ( Figure 3E). Analysis of nT Reg cells from peripheral blood of healthy age-matched controls and patients with MM demonstrates a subset, albeit small subset, of nT Reg cells that produce IFNc ( Figure 3F).
It is known that the effector T-cell lineage shows great plasticity and that human T Reg cells can differentiate into IL-17-producing cells [24,25]. When tT Reg cells were generated in our in vitro culture model, a significant production of IL-17 was noted in the supernatant after 7 days of co-culture of CD4 + CD25 -T-cells with HMCL (30618 pg/ml vs 0.260.1 pg/ml; p,0.001; Figure 4A). Therefore, we wished to determine if Th17 cells could be generated directly from tT Reg cells and thus, characterizing the plasticity of tT Reg cells generated in our in vitro model, compared to naturally occurring T Reg cells. CD4 + CD25 -T-cells co-cultured with mitomycin-C-treated HMCL for 7 days were FACS-sorted and re-stimulated with CD3/CD28-coated beads with rhIL22 20 U/ml for 5 days. For comparison, naturally occurring T Reg cells were sorted using Miltenyi columns, co-cultured with mitomycin-C-treated HMCL for 7 days, then FACS-sorted and stimulated under identical conditions. After re-stimulation, a subpopulation of IL-17-producing CD4 + T-cells was identified from the FACS-sorted tT Reg Figure 4C).

Myeloma-generated Regulatory T-cells are Induced by Surface ICOS/ICOS-L Interactions not Tumour-derived TGFb
The mechanisms for controlling the induction and expansion of T Reg cells remains to be fully clarified with some investigators demonstrating soluble factors as central to induction whilst others emphasize cell-to-cell contract, especially with dendritic cell contact, as key [26,27,28]. We adapted our antigen presenting cell-free in vitro model to investigate the role of humoral factors versus contact mediation. CD25 -CD4 + T-cells were isolated from PB and co-cultured with mitomycin C-treated HMCL for 7 days with and without transwell separation. The generation of CD4 + CD25 + Foxp3 + T Reg cells through co-culture with HMCL was significantly reduced by abolishing cell-to-cell contact (30.7%65 CD4 + tT Reg cells vs 0.11%60.04 CD4 + tT Reg cells, n = 7, p,0.001; Figure 5A). The inhibition of tumour-generated CD4 + T Reg cells through abolition of cell-to-cell contact was  Results represent all experiments, expressed as mean6SEM and analyzed using student t-test. C. IL-10 production by tT Reg cells after 7 days of co-cultures of CD25 -CD4 + sorted T-cells and HMCL. Results represent all experiments, expressed as mean6SEM (n = 3) and analyzed using student t-test. D. IL-10 production by tT Reg cells after 7 days of co-cultures of CD25 -CD4 + sorted T-cells and HMCL. Results represent all experiments, expressed as mean6SEM (n = 3) and analyzed using student t-test. E. Representative flow cytometry plots demonstrating the generation of IFNc + FoxP3 + CD25 + CD4 + T-cells from CD4 + CD25 -T-cells in a 7 day co-culture assay with mitomycin C-treated U266B cells. F. The proportion of IFNc-producing FoxP3 + CD25 + CD4 + T-cells detectable in the peripheral blood of agematched controls (n = 15), patients with MM (n = 15) and tTReg cells generated in vitro after 7 days of co-cultures of CD25 -CD4 + sorted T-cells and HMCL (n = 3). Histograms represent IFNc production by cells gated on FoxP3/CD25/CD4 positive staing. Results expressed as mean6SEM. doi:10.1371/journal.pone.0035981.g003 associated with a reduction in the production of IL-10 (69.8633.6 pg/ml vs 6.360.01 pg/ml, n = 3, p = 0.079) and IFNc (1215464174 pg/ml vs 0.0460.01 pg/ml, n = 3, p,0.001).
The role that surface TGFb plays in the induction and, in conjunction with IL-10, the propagation of T Reg cells has been extensively studied in both human and murine systems [22]. We therefore wished to determine the role that TGFb may play in the generation of tT Reg cells in our model. Whilst HMCL produce soluble TGFb (data not shown) they express the modulatory cytokine on their surface ( Figure 5B) in addition to HLA class II (DR) and the negative co-stimulatory molecule (second signal) ICOSL (CD275). Therefore, CD4 + CD25 -T-cells were isolated from PB and co-cultured with mitomycin C-treated HMCL for 7 days with and without the specific TGFb antagonist, Latency Associated Peptide (LAP) and an anti-TGFb monoclonal antibody Similarly, the use of anti-IL-10 MoAb failed to demonstrate any significant inhibition of tT Reg cell generation (7.5%62.5 at 10 mM and 4.3%64.1 at 100 mM, n = 3, Figure 5C).
The B7 family members, ICOS/ICOSL have previously been implicated in T Reg cell generation. When examined, HMCL express surface ICOS-L ( Figure 5B). We therefore determined the level of ICOS expression on newly generated tT Reg cells and nT Reg cells. A mean of 65.6%67 tT Reg cells expressed surface ICOS (n = 5) compared with 6.6%61.5 nT Reg cells from agematched controls (n = 14) and 8.1%61.3 nT Reg cells from patients with MM (n = 10; p,0.0001; Figure 5D). Thus we added an anti-ICOS-L MoAb, in increasing concentrations, to CD4 + CD25 -Tcells isolated from PB and co-cultured with mitomycin C-treated HMCL for 7 days. The anti-ICOSL was able to demonstrate a reduction in tT Reg cell generation with an incremental inhibitory effect with increasing concentrations (Inhibition of tT Reg cell  Figure 5E).

Discussion
The efficient generation of an immune response is coupled with a regulatory system to limit that response, preventing the destruction of healthy cells and tissues [29]. However, transformed malignant tissue may adopt one or more mechanisms to interfere with either the effector immune response or the regulatory cell compartment in an attempt to evade immune surveillance [30,31,32,33,34]. In myeloma, both dysfunctional effector responses and augmented regulatory cell compartments have been described [9,12,35,36]. To date, the origin of the expanded regulatory T-cell population has remained elusive. We therefore sought to elucidate a causal relationship between the tumour cell in myeloma and T Reg cell generation. Our model system demonstrates, for the first time, a direct induction of T Reg cells (tT Reg cells) by both fresh myeloma cells and cell lines that demonstrate the phenotype and functionality of T Reg cells whilst inducing IL-10 production in non T Reg cells. However, phenotypic differences between tT Reg cells induced and naturally occurring T Reg cells were noted. In particular tT Reg cells demonstrated a CD25 High FoxP3 High GITR + PD-1 + phenotype, distinct from naturally occurring T Reg cells. IFNc-producing regulatory T cells have been described in the setting of intestinal infection and allograft rejection and evidence suggests a central role of IFNc in inducible T Reg cell generation [37,38]. IFNc cellular effects are mediated through STAT1 phoshorylation and it is known than there is a STAT1 binding site in the proximal region of the FoxP3 gene promoter in humans (but not mice) [39,40]. In addition to which, IFNc has been shown to mediated FoxP3 gene induction in synergy with IL-27 [41]. Here we demonstrate for the first time in a cancer setting, the generation of inducible T Reg cells from CD4+CD25-T cells where a subset produce IFNc in vitro. In addition, similar to naturally occurring T Reg cells, CD4 + CD25 + CD127 Low FACS sorted tT Reg cells demonstrate lineage plasticity by differentiating into IL-17-producing T-cells following further TCR-mediated stimulation [24,42]. Furthermore, data published to date, has demonstrated a central role for antigen presenting cells (APC) in the interactions of tumour cells and T Reg cells. However, in our in vitro system, myeloma tumour cells generate and expand tT Reg cells in an APC-free manner, that is directly inducing CD4 + CD25 -T-cells. The importance of the generation of previously considered pro-inflammatory cytokines in the generation and or propagation of T Reg cells in cancer remain to be elucidated. In patients with cancer, T Reg cells are continuously exposed to tumour antigen (TA), either directly or through the tumour microenvironment which in turn, results in high levels of ICOS expression as has been demonstrated in melanoma and prostate cancer [7,43]. T Reg cells generated in this environment produce high levels of IL-10, which mediates their suppressive capabilities, especially dendritic cell function [44]. We demonstrate with our in vitro model that in myeloma, ICOS-L + tumour cells directly induce tT Reg cell generation mediated in a contact-dependent manner, in the absence of antigen-presenting cells, which is inhibited significantly though not totally using anti-ICOSL monoclonal antibodies. This model of induction tT Reg cell however, does not account for chronic antigen stimulation by the tumour-bearing host, nor does it this culture system take allowances of the effects of immunomoduatory drugs such as steroids and IMiDs (Thalidomide, Lenalidomide, Pomalidomide) which may account for differences in ICOS expression between tT Reg cell and T Reg cells from MM patients, a significant level of IL-10 is produced though this contributes minimally to the generation of tT Reg cells (as evidenced by lack of inhibition through monoclonal antibody blockade). Furthermore, although IL-10 production by ICOS-induced T Reg cells has been documented in both human and murine in vitro systems [45,46,47], the tT Reg cells induced in our system did not produce IL-10.
The role of TGFb in both the generation of T Reg cells and in the mediation of their suppressive effects has been the subject of conflicting reports and may relate to the experimental design of in vitro systems used to study this relationship. In murine model systems, TGFb-mediated FoxP3 induction in naïve T-cells augmented by IL-2, produce T Reg cells with a suppressive phenotype though are rendered hypo-responsive to TCR-mediated stimulation [4,48,49]. In contrast, other investigators have demonstrated TGFb independence in both the generation and mediation of suppression [50,51,52]. Murine prostatic and renal cell cancer cells have been shown in vitro to generate T Reg cells mediated through TGFb. We have previously shown TGFb to have a central role in myeloma-mediated effector cell dysfunction and is detected at high level in peripheral blood and bone marrow [12,36]. However, the data from our in vitro model did not demonstrate a prominent role for TGFb in the induction of tT Reg cell generation, despite the production of TGFb in co-culture supernatant (data not shown) and expressed on the surface of tumour cells.
Recent studies have suggested a close relationship between CD4 + CD25 + FoxP3 + T Reg cells and pro-inflammatory IL-17-producing T helper cells (Th17) [53]. In our studies, we demonstrate that tT Reg cells have a capacity, upon TCR-mediated stimulation to generate IL-17 producing T-cells, both CD4 + FoxP3 + and CD4 + FoxP3cells, indicative of a plasticity of the tT Reg cells, similar to previous reports [53,54,55]. More recently, it has ben shown that different myeloid-derived cellular subsets (CD14 + HLA-DR Dim vs CD14 + HLA-DR + ) can induce both T Reg cell and Th17 cells, with a recognized degree of plasticity [56]. However, our in vitro model system is APC-free and devoid of the proposed myeloid-derived cellular subsets. Though the tT Reg cells were generated by co-culture with HMCL, in the absence of additional TCR-mediated stimulation, the plasticity we observed with these tT Reg cells was purely upon TCR-mediated stimulation in the absence of HMCL and suggests an independent functional plasticity of tT Reg cells.
In summary, our in vitro studies demonstrate that the tumour cells of Myeloma are capable of inducing T-cells with the phenotypic and functional characteristics of T Reg cells, associated with the production of IL-10 and IFNc The induction of T Reg cells is mediated by cell-to-cell contact with the ICOS/ICOS-L system demonstrating a central role in the induction. The data presented here offers a better understanding of the immune evasion adopted by MM tumour cells offering a potential opportunity to manipulate the tumour-bearing host immune micro-environment pharmacologically. The pre-clinical data presented here offers a scientific basis for the development of suitable clinical research protocols to test this in vivo. Figure S1 DNA PCR analyses of TCRG rearrangements of FACS sorted tT Reg cells performed using the BIOMED-2 multiplex strategy. Representative example of 3 experiments. Positive control used was peripheral blood from a patient with T-cell lympho-proliferative disease. ( )