Phosphoantigens Overcome Human TCRVγ9⁺ γδ Cell Immunosuppression by TGF-β: Relevance for Cancer Immunotherapy Free

Human γδ cells represent an important subset of lymphoid cell effectors for cancer immunotherapy. The majority of circulating γδ T lymphocytes in the adult (~1% of mononuclear cells) are negative for CD4 or CD8 but express a Vγ9Vδ2 TCR specific for nonpeptide phosphoantigens (PAgs) from tumor cells and microbes (1, 2). Four subsets of Vγ9Vδ2 T lymphocytes corresponding probably to successive maturation stages have been identified on the basis of their phenotypes. The CD45RA⁺CD27⁺ and CD45RA⁻CD27⁺TCRVγ9Vδ2⁺ T cells subsets preferentially express the CD62L and CCR7 lymph node homing receptors; they are defined as naive (N) and central memory (CM) cells, respectively. The CD45RA⁻CD27⁻ and CD45RA⁺CD27⁻ subsets are negative for CD62L and CCR7 but preferentially express receptors for in-flammatory chemokines, such as CCR5 and CXCR3. By analogy with terminally differentiated CD8 memory T cells (3), these γδ T cells are present in inflammatory sites and were referred to as T effector memory helper 1 (TEMh1) and T effector memory RA (TEMRA), respectively (4, 5). A fifth maturation stage of Vγ9Vδ2 T lymphocytes has been further identified on the basis of CXCR5 expression and was defined as follicular helper γδ cells for its phenotypic and functional analogy with conventional follicular helper CD4 lymphocytes (6).

The functions of TCRVγ9⁺ T cells vary with their maturation (7). Upon antigenic stimulation, N and CM cells show vigorous proliferative responses but very low release of cytokines and cytolytic granules. By contrast, activation of TEMh1 and TEMRA TCRVγ9Vδ2⁺ T cells drives a strong release of Th1 cytokines (e.g., IFN-γ and TNF-α) and cytolytic granules but little proliferation (5, 8). In addition, the quantitatively minor subset of follicular helper TCRVγ9Vδ2⁺ cells secretes Th2 cytokines IL-4 and IL-10 and promotes Ig secretion by B lymphocytes when activated (6). Hence, Vγ9Vδ2 T lymphocytes successively mature through the sequence N_{(CD45RA+CD27+)}→CM_{(CD45RA−CD27+)}→TEMh1_{(CD45RA−CD27−)}→TEMRA_{(CD45RA+CD27−)} (4). PAg-activated γδ cells also release proinflammatory chemokines MIP-1α, MIP-1β, and RANTES (9). They also mediate a strong cytolytic activity directed against cancer cells by direct (NK-like) cytolysis. Most TCRVγ9Vδ2⁺ T cells constitutively express the NKG2D coreceptor, which interacts with tumors expressing MICA/B or ULBP1-3 and contributes to cytolytic activation (10). Moreover, activated TCRVγ9Vδ2⁺ T effector memory cells upregulate the expression of CD16 (FcγRIIIA) to perform Ab-dependent cell cytotoxicity (ADCC) (11, 12). The main cytolytic molecules of γδ cells are made up of granzyme A and granzyme B (GzB), perforin, granulysin, Fas ligand, and TRAIL in addition to TNF-α and IFN-γ.

Synthetic PAgs, such as bromohydrin pyrophosphate (BrHPP) (13) and second generation bioisoteres (14–16), have been developed as selective immunostimulating drugs targeting TCRVγ9Vδ2⁺ T lymphocytes, most specifically for cancer immunotherapies. In addition, antiosteolytic drugs, such as aminobisphosphonates, activate the same lymphocytes by inducing endogenous PAgs and could thus have related therapeutic applications (17–19). These drugs have led several groups to demonstrate antitumor functions of human γδ T lymphocytes in different in vivo contexts. Studies in nonhuman primates (20, 21) and in cancer patients (7, 22–27) have illustrated the potential of activated TCRVγ9⁺ T lymphocytes against leukemia, lymphoma, and carcinoma (reviewed in Refs. 28–31). In addition, clinical trials involving PAgs or aminobisphosphonates alone and in combinations are currently assessed by different groups around the world.

Limits to γδ T cell-based cancer immunotherapies are now appearing, however (32). Tumor progression in cancer patients is frequently associated with emergence of malignant cells evading immune surveillance (33). Of the various immunoescape mechanisms used by cancer cells, production of TGF-β is one of the most potent and frequent (reviewed in Refs. 34 and 35). This cytokine has a distinct effect on the main subsets of con-ventional αβ T lymphocytes (36, 37). It inhibits proliferation and functional differentiation of cytolytic CD8 T cells and promotes differentiation of Th17 cells and regulatory T (Treg) cells to reduce anticancer immunity (38, 39).

Nevertheless, whether this cytokine also targets the unconventional human γδ T cells remains unclear. TGF-β inhibits γδ T cell proliferation induced by Mycobacterium tuberculosis-pulsed monocytes (40) and helps induce FOXP3⁺ γδ Treg cells in the presence of IL-15 (41), but little is known about its direct bioactivity on the anticancer functions of human γδ T cells. In this study, we investigated whether TGF-β affects the proliferation, maturation, and cytolytic functions of TCRVγ9⁺ γδ T lymphocytes, either as freshly isolated cells (fics) or as primary cell lines (pcls).

FITC-conjugated Abs to TCRVγ9, PE-Cy5–conjugated anti-CD27 and anti-CD69 or with PE-conjugated Abs to markers TCRVγ9 or with PE-Cy7–conjugated anti-CD16 and anti-NKG2D were from Beckman Coulter (Marseille, France), Pacific Blue-conjugated Abs to CD45RA and CD3 were from Ozyme (St. Quentin en Yvelines, France), the PE-conjugated Ab to TGF-βRII was from R&D Systems Europe (Lille, France), and allophycocyanin-conjugated anti-killer Ig-related receptor (KIR) mAb [clone 1-7F9] (42, 43) was from Innate Pharma (Marseille, France). The respective isotype-matched conjugated controls were from Beckman Coulter, Ozyme, and R&D Systems Europe, respectively. γδ T lymphocytes intracellular cytokine levels and cytotoxic granule production were measured on cells stained at the cell surface and treated with brefeldin A (Sigma-Aldrich, Lyon, France) for 5 h before intracellular staining. Cell surface Ag stainings involved FITC-conjugated anti-TCRVγ9 (Beckman Coulter) and Pacific Blue-conjugated anti-CD3 (Ozyme). After staining, cells were washed twice, fixed, and permeabilized with 4% paraformaldehyde and 0.1% saponine plus 0.5% BSA (Sigma-Aldrich). Cells were then stained with allophycocyanin-conjugated anti–IFN-γ, Ax647-conjugated anti-GzB, or FITC-conjugated anti-perforin Abs (BD Biosciences, Le Pont de Claix, France). Flow cytometry was done with LSR-II and the dedicated software FACSDiva (BD Biosciences) and FlowJo 7.5.5 (Tree Star, Ashland, OR). Quantification of TGF-βRII expression on γδ T cell surface was done by using Quantum beads (Bang Laboratories, Fishers, IN) as described previously (12).

The synthetic PAg BrHPP (Innate Pharma, Marseille, France) has been described previously (13). Recombinant human (rh) IL-2 was provided by Sanofi-Aventis (Toulouse, France), and rituximab was provided by Roche (Neuilly-sur-Seine, France). CFSE was purchased from Molecular Probes (Eugene, OR). rhTGF-β1 was purchased from R&D Systems Europe. Cells were cultured in complete medium containing RPMI 1640 (Invitrogen, Cergy Pontoise, France) supplemented with 2 mM l-glutamine, 100 μg/ml streptomycin, 100 IU/ml penicillin, and 1 mM sodium pyruvate (Cambrex Biosciences, Rockland, ME).

TCRVγ9⁺ T lymphocytes were isolated from PBMCs obtained from human healthy individuals (Etablissement Français du Sang, Toulouse, France) after Ficoll-Hypaque density centrifugation. Distinct TCRVγ9⁺ γδ T cell samples were prepared for this study. On the one hand, freshly isolated γδ T cells (referred below to as γδ fic) were directly purified by immunomagnetic separation using anti-TCRγδ–conjugated magnetic microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). On the other hand, primary TCRVγ9⁺ cell lines (referred below to as γδ pcl) were obtained from PBMCs by in vitro culture for 14 d in complete medium supplemented with 10% Fetal Clone I (HyClone/Thermo Fisher Scientific, Brebières, France), BrHPP (1 μM), and rhIL-2 (300 IU/ml) at day 0, with additional IL-2 (300 IU/ml) renewal every 3 d. After checking for TCR phenotype of the cell samples (>90% TCRVγ9⁺ cells for γδ fics or γδ pcls), the cells were washed and treated or not with TGF-β for 48 h in IL-2–free complete medium supplemented with 10% FBS (Invitrogen). BrHPP was added or not for the last 24 h of γδ fic and pcl culture (these cells referred below to as activated γδ fic or activated γδ pcl).

Freshly isolated PBMCs (5 × 10⁶ cells/ml) were labeled with CFSE (1 μM) for 10 min at 37°C and washed with fresh culture media, according to the manufacturer’s instructions. Labeled cells were then stimulated with BrHPP (100 nM), rhIL-2 (0–200 IU/ml as indicated), and TGF-β (0–10 ng/ml as indicated). Cells were harvested, stained for TCRVγ9, and analyzed for CFSE dilution and cell surface expression of TCRVγ9 by flow cytometry after 7 d of culture. The progressive dynamics of CFSE dilution by dividing TCRVγ9 cells was measured after specified durations following PAg stimulation of PBMCs as above and compared with the theoretical cell growth Cyton model (CPSM software, kindly provided by P.D. Hodgkin) (44). The maturation stages of CFSE-labeled TCRVγ9 cells (1 × 10⁶ cells/ml) activated with BrHPP (100 nM) and rhIL-2 (10 IU/ml) with or without TGF-β (2 ng/ml) were determined by simultaneous immunostaining of TCRVγ9, CD27, and CD45RA surface markers and flow cytometry.

The Daudi and Raji human Burkitt’s lymphoma cell lines used as target cells were cultured at 37°C in complete medium with 10% FBS and 25 mM HEPES (Invitrogen). Specific lysis by Vγ9 T cells was measured by standard 4-h [^51Cr] release assays. The lysis rates were obtained by (experimental release − spontaneous release)/(maximum release − spontaneous release) × 100. Maximum and spontaneous releases were determined, respectively, by adding 0.1% Triton X-100 or complete medium to [^51_Cr]-labeled tumor target cells in the absence of γδ T cells. Data present as the mean of triplicate samples. Specific lysis assays by ADCC involved the same settings as above, except the presence of rituximab (10 μg/ml) added to the 4-h cell coincubation and to the spontaneous release and maximum release controls.

PBMCs were obtained from four human healthy individuals (Etablissement Français du Sang, Toulouse, France) as depicted above. For each donor, TCRVγ9⁺ cells were purified (>99% for each sample) by immunostaining and flow cytometry cell sorting either before activation (γδ fic resting control) or 6 h after PAg plus IL-2 stimulation (γδ fic activated 6 h) or 7 d after PAg stimulation and culture with IL-2 (γδ pcl 7 d). Total RNA was isolated from each of these cell samples using TRIzol reagent (Invitrogen Life Technologies, Paisley, U.K.), according to the manufacturer’s instructions. The quality and integrity of the RNA obtained were assessed by using an Agilent 2100 Bioanalyser (Agilent Technologies, Palo Alto, CA) after denaturation at 70°C for 2 min. cRNAs were then prepared according to one-Cycle Target Labeling protocol (Affymetrix, Santa Clara, CA) starting from 1 μg total RNA. The cRNAs were then fragmented and hybridized to Affymetrix HG-U133 plus 2.0 arrays. The chips were washed and scanned, according to the manufacturer’s instructions. GeneChip Operating Software (version 1.1; Affymetrix) was used for the primary image analysis of the array, for the normalization (global scaling method, target value of 100), and for the different comparisons. Expressions of the selected genes are summarized as means and SDs of the raw expression data from the four donors. Asterisks indicate significant (p < 0.05) changes to the γδ fic resting control, and a full line indicates the absent/present threshold. Lists of significantly changed (up- and downregulated) gene expression levels were built using as criteria p < 0.01 and p > 10-fold changes in expression level for genes in which baseline expression in resting controls was above the cutoff threshold. The resulting sublists of genes were compared with 193 biological signatures defined in Kyoto Encyclopedia of Genes and Genomes (www.genome.jp/kegg) by using an in-house developed program based on our search engine nwCompare (45) and hypergeometrical distribution for p values.

Data shown for transcriptome analysis are means and SDs and means and SEs of the mean (SEM) otherwise. The normality and equal variance of each sample series were evaluated prior to statistical analysis by the specified tests. A one-tailed, paired Student t test was used whenever appropriate or one-way Mann-Whitney rank-sum test was used otherwise using α = 5% for significant differences. All statistical analyses were performed using the SigmaStat 3.0 (SPSS, Chicago, IL) and XL Stat 2008 (AddinSoft, Paris, France) software.

We first monitored the kinetics of γδ cell proliferation by the CFSE dilution assay at several time points after PAg (BrHPP) stimulation to optimize this readout for the further experiments. Cell cultures were supplemented with IL-2, because the PAg-driven expansion of TCRVγ9⁺ γδ cells requires an exogenous supply of this cytokine (46). We observed the CFSE was progressively diluted from days 5 to 7, in good match with a stochastic cell division model (44) involving 70% of responding cells, 80 h for first division and 13 h for subsequent divisions, as reported for CD8⁺ αβ T cells (47) (Fig. 1A). In addition, the transcriptome of γδ cells obtained 7 d after stimulation confirmed gene expression signatures (45) of cell cycle (47 genes; p < 10⁻⁵³), purine and pyrimidine metabolisms (37 genes; p < 10⁻¹⁰), and p53 signaling pathway (19 genes; p < 10⁻¹⁶) (data not shown). These experiments indicated that day 7 of culture in IL-2–containing medium was optimal to analyze γδ cells dividing in response to PAg stimulation.

We then asked whether TGF-β inhibits the proliferation of PAg-activated (100 nM BrHPP) γδ T cells in cultures supplemented with IL-2, and we observed a TGF-β dose-dependent inhibition of TCRVγ9⁺ cell proliferation. The above modelization of cytometry results from cultures at the highest TGF-β/IL-2 ratio suggested ~20% of responding cells (data not shown). The rate of proliferating γδ cells is known to increase with doses of IL-2, as in CD8 T cells (48), and increasing concentrations of IL-2 rescued this inhibition accordingly (Fig. 1B, 1C). Furthermore, increasing dose of PAg also augments the rate of proliferating TCRVγ9⁺ γδ cells (13, 20, 21, 49) and increasing concentrations of BrHPP rescued the proliferating γδ cells from TGF-β inhibition (Fig. 1D, 1E). Thus, TGF-β inhibits proliferation of TCRVγ9⁺ γδ T lymphocytes but higher doses of either IL-2 or PAg bypass this inhibition.

In most adults, the circulating TCRVγ9⁺ T lymphocytes essentially are made up of CM cells plus a few N and effector memory cells (Fig. 2, left) (4, 5). We thus asked whether TGF-β modulated the γδ T cell maturation in the above proliferation assay by analyzing cell surface expression of the CD27 and CD45RA markers on CFSE-labeled γδ cells stimulated with PAg and cultured with IL-2 and with or without TGF-β. The phenotype of undivided parental cells comprised N and CM lymphocytes in both con-ditions. By contrast, the maturation of dividing cells diverged in the presence of TGF-β. In control conditions without TGF-β, the divided cells mostly made up CM (CD27⁺CD45RA⁻) and TEMh1 lymphocytes (CD27⁻CD45RA⁻) (4, 5). By contrast, the few cells that divided in presence of TGF-β not only made up the CM and TEMh1 cells as above but also encompassed a sizeable proportion of N cells (CD27⁺CD45RA⁺) (Fig. 2, right). Recent reports showed IL-17⁺ γδ T cells from individuals with active pulmonary tuberculosis (50) or with HIV infection (51). Although the TCR repertoire of those cells did not contain TCRVγ9⁺ cells, we asked whether TCRVγ9⁺ fics and pcls derived from healthy individuals contained IL-17⁺ γδ T cells. We found no IL-17⁺TCRVγ9⁺ lymphocytes among these cells with or without TGF-β. In addition, we did not detect any FoxP3⁺CD25^brightTCRVγ9⁺ cells in these experiments (data not shown).

So, TGF-β delays the maturation of PAg-activated TCRVγ9⁺ lymphocytes without inducing Th17 or Treg cells.

The above experiments indicated that TGF-β inhibits proliferation and functional maturation of γδ T cells, which normally generate cytolytic effector lymphocytes. Because TGF-β reduces the lytic activity of CD8⁺ T lymphocytes (52) and NK cells (53), we assessed its activity on cytolytic functions of TCRVγ9⁺ cells. Thus, TCRVγ9⁺ fics and pcls were cultured with or without PAg activation for 48 h in complete medium with or without TGF-β, and their intracellular GzB, perforin, and IFN-γ were assessed by immunostaining and flow cytometry. The percentage of cells producing both GzB and perforin among ex vivo TCRVγ9⁺ fics was quite low as previously described (5) (data not shown) and strongly increased upon PAg activation with or without TGF-β. By contrast, resting TCRVγ9⁺ pcls made up a vast majority (~80%) of GzB and perforin-producing cells, in line with their maturation as cytolytic effector cells. These phenotypes were strongly decreased in resting pcls treated with TGF-β as compared with untreated pcl controls: 55% of GzB⁺perforin⁻ and 39% of GzB⁺perforin⁺ cells with TGF-β versus 14% of GzB⁺perforin⁻ and 83% of GzB⁺perforin⁺ cells in control cultures without TGF-β. With secondary PAg activation, however, the intracellular content of TCRVγ9⁺ pcls comprised both GzB⁺ cells and GzB⁺perforin⁺ cell subsets within frequen-cies that were not reduced by TGF-β (Fig. 3A). TGF-β reduced the rates of IFN-γ⁺ cells among activated fics and resting pcls but conversely enhanced this rate among activated pcls. These results were confirmed by dosages of IFN-γ secreted in culture supernatants (Fig. 3B, 3C). The cell surface expression of NKG2D was reduced on resting pcls treated with TGF-β relative to control cells, but TGF-β did not cause this reduction on γδ pcls and γδ fics stimulated with PAg. In the same experiments, the cell surface expression of KIRs and CD16 were unchanged (Fig. 3D). This set of data indicated that the lytic machinery of γδ cells was reduced by TGF-β but rescued by secondary PAg activation of pcls.

We then challenged these conclusions by using functional assays of cytolytic activity, the standard [⁵¹Cr] release assays from Daudi Burkitt’s lymphoma target cells, and specific ADCC of the CD20⁺ Raji target cells with anti-CD20 rituximab. These distinct assays reflect activation of the γδ T cell’s lytic activity for allogenic targets, either driven by TCRVγ9⁺ plus NKG2D (54, 55) or by TCRVγ9⁺ plus FcγRIIIA (12). Both spontaneous lysis and ADCC were higher with γδ pcls than with activated γδ fics; nevertheless, they were both inhibited by TGF-β. By contrast, ADCC mediated by PAg-activated γδ cells, either pcls or fics, was almost unsensitive to TGF-β (Fig. 4). In addition, other functions of PAg-activated γδ cells, such as secretion of chemokines RANTES, MCP-1, MIP-1α, and MIP-1β (56), were not inhibited by TGF-β (data not shown).

Thus, TGF-β inhibits the spontaneous cytolytic activity of γδ T lymphocytes but not ADCC induced by rituximab in presence of PAg.

Because the previous results indicated that TCRVγ9 activation can rescue γδ cells from inhibition by TGF-β, we investigated the mechanism of this recovery. We checked whether PAg activation reduced the expression of high-affinity TGF-βR at the cell surface of TCRVγ9⁺ lymphocytes. Immunolabeling and quantitative flow cytometry of the TGF-βRII protein on γδ T lymphocytes revealed similar expression levels in all culture conditions, however, ex-cluded this possibility (Fig. 5A). Furthermore, comparing the transcriptomes of PAg-activated and resting TCRVγ9⁺ T cells ruled out reduction of the TGF-β response pathway with activation. PAg-activated cells had less TGF-β1 mRNA, as many transcripts for TGF-βRII and TGF-βRIII (in line with the protein expression), and more TGF-βRI mRNA. In addition, activated cells had more of the TGF-β transducer SMAD2 and less inhibitors SKI and SMAD7, corresponding to a fully functional TGF-β pathway (Fig. 5B). Thus, TCRVγ9 activation does not reduce the TGF-β response pathway.

We then attempted to address this point by analyzing TCRVγ9 downmodulation and CD69 upregulation as readouts of PAg signaling in culture conditions with and without TGF-β. PAg induced a weak TCRVγ9 downmodulation and a strong CD69 upregulation on γδ fics (57, 58). Reciprocally with γδ pcls, PAg induced a strong TCRVγ9 downmodulation and a weak CD69 upregulation. Furthermore, in the presence of TGF-β, the activation of γδ fics was unchanged whereas that of γδ pcls was modified. The TCRVγ9 downmodulation by γδ pcls was reduced by TGF-β, and their CD69 upregulation was therefore increased (Fig. 5C, 5D), in line with their above-depicted IFN-γ response (Fig. 3B, 3C).

These findings suggest that TCR can rescue the γδ cells from TGF-β inhibition by triggering additional activation signaling rather than by impairing their TGF-β response pathway.

This study aimed at characterizing the bioactivity of the TGF-β cytokine on the PAg-responsive TCRVγ9⁺ T lymphocytes. We showed in this study that TGF-β inhibits proliferation, matu-ration, and cytolytic functions of these lymphocytes like other T and NK cells (36, 37, 39). In contrast with these latter, however, the TGF-β–mediated immunosuppression of γδ cells can be rescued not only by IL-2 but also by PAg-driven TCR signaling.

TGF-β suppresses proliferation of conventional T cells by blocking expression of their endogenous IL-2 growth factor (59, 60) and of their high-affinity IL-2R, transferring receptor, c-myc transcription factor, and cell cycle regulators (61–63). We showed above that TGF-β also suppressed the proliferation of γδ cells. We propose that TGF-β targets, on the one hand, the endogenous IL-2 production of TCRVγ9⁺ T lymphocytes presumably through blockade of its gene promoter as in αβ T lymphocytes (64, 65) because exogenous IL-2 partially reversed this blockade. On the other hand, TGF-β inhibited the functions induced by TCR signaling, but this was reversed by increasing doses of PAg stimulus. TCRVγ9⁺ γδ lymphocytes usually represent ~1% of the circulating mononuclear cells, so their expansion is critically required to bring a significant contribution to cancer immuno-therapies. By underlining the negative incidence of TGF-β may have on therapeutic protocols involving in vivo γδ cell expansions, this study also provides a rationale for γδ cell-based protocols composed of either in vitro expansions (49) or in vivo activation by high-dose PAg supplemented with exogenous IL-2 (12, 20, 21, 25, 66, 67). In addition, the proliferative response of PAg/IL-2–stimulated γδ lymphocytes is essentially—if not exclusively—mediated by cells from the N and CM compartments of TCRVγ9⁺ lymphocytes (4, 8), suggesting that TGF-β might target more selectively these maturation stages.

TGF-β induces substantial in vitro differentiation of regulatory FOXP3⁺CD25^brightTCRVγ9⁺ cells when combined with IL-2, IL-15, and the weak PAg isopentenyl pyrophosphate (41). Nevertheless, TGF-β alone or combined with the potent agonist BrHPP and IL-2 induced neither γδ Treg cells nor γδ T cells with Th17 phenotype. Both lack of the Treg-promoting IL-15 cytokine (68, 69) and relative strength of the PAg agonist used in our and in most in vivo studies (70) and clinical trials might account for this discrepancy.

We found in this study that TGF-β slows the PAg-induced maturation of TCRVγ9⁺ cells into effector memory cells. TGF-β induces apoptosis of effector memory T cells during infection-induced clonal expansions (71). Low rates of TCRVγ9⁺ cell death because of TGF-β were noticed in our study, arguing against shorter life spans of the most mature γδ cell subsets in this paper. Rather, we hypothesize that in cultures supplemented with TGF-β, IL-2, and PAg, the γδ cells, which have maintained a CD27⁺CD45RA⁺ phenotype after several divisions, had pre-served the N phenotype of their progenitors. This suggests that by analogy with its action on melanocyte or neural stem cells (72, 73), TGF-β might favor the maintenance by self-renewal of an N γδ cell pool. This peculiar bioactivity resembles that of Wnt3a protein (and glycogen synthase kinase-3-β inhibitors) on the maturation of CD8⁺ CTLs (74), but its lymphoproliferative suppression makes it a distinct one. Nevertheless, the maintenance of N T lymphocytes among cells proliferating to Ag in the presence of TGF-β had not been characterized previously and extends the range of the bioactivities of this cytokine.

As discussed above, the differential sensitivity of γδ cell cytolytic functions to TGF-β was inherent to the presence of PAg stimulus; resting γδ pcls were sensitive, whereas PAg-activated γδ pcls were more resistant to TGF-β. In addition to presence of PAg, however, the maturation of γδ cells was also important for res-ponse to TGF-β, with higher sensitivity to this mediator in PAg-activated pcls than in PAg-activated fics, as depicted in conventional T lymphocytes (75). In addition, the most mature γδ cells upregulate various surface coreceptors that confer better responsiveness to PAg (4, 5, 8) by improving the efficiency of stimulus transduction and therefore lowers their sensitivity to TGF-β (76, 77). Along this line, the direct cell–cell binding facilitated by therapeutic rituximab not only provides γδ cells with a physical contact to the Raji cell target to eradicate but also strengthens their stimulatory signaling (12). From the current study, we propose that stimulation with PAg strengthens intracellular signaling for ADCC by TCRVγ9⁺ cells and thus resistance to TGF-β.

This study indicates that signaling pathways for TCR-mediated activation and TGF-β–mediated inhibition coexist in γδ T cells. So, the intracellular dominance of signaling from γδ TCR versus TGF-βR determines the outcome of functional γδ cell responses to their microenvironment, like the γδ TCR versus KIR rela-tionship (78–80). Because increasing doses of stimulating PAg/IL-2 can overcome the inhibition of cytolytic TCRVγ9⁺ T cells by TGF-β, strategies bypassing its bioactivity could be envisaged. Therapeutic protocols based on TCRVγ9⁺ T lymphocytes stimu-lated by BrHPP and IL-2 proved more easily tunable than those based on conventional CTL or NK cells, because these γδ cells do not require tumor-derived peptide Ags presented by HLA molecules. In this aim, however, this work suggest the need to adapt PAg doses delivered to patients in function of their tumor’s ability to produce TGF-β. Furthermore, synergizing this stimulus with rituximab, trastuzumab, or other therapeutic mAbs might also provide effector γδ cells with more robust cytolytic activity (12). Future studies will determine whether such combinations will defeat particularly immunosuppressive tumors of advanced cancers of the breast, prostate and pancreas.

We thank Richard A. Flavell and members of our laboratory for critical suggestions on this work, Innate Pharma for providing BrHPP, laboratoires Sanofi-Aventis for rhIL-2, and laboratoires Roche for gifts of rituximab.

Disclosures The authors have no financial conflicts of interest.