Cutting Edge: Abortive Proliferation of CD46-Induced Tr1-Like Cells due to a Defective Akt/Survivin Signaling Pathway¹

Regulatory T cells are potent cells that inhibit self-Ag and innocuous foreign Ag-specific adaptive immune effectors (1). Disrupting their functional activity often results in clinical pathologies. Moreover, pathogens and tumor cells can subvert immune responses by deviating regulatory T cells to their benefit (1, 2). Therefore, identifying what triggers the generation of regulatory T cells and understanding their biology appear to be required to struggle certain diseases and/or to efficiently use them as therapeutic tools (3).

Among regulatory T cells, the peripherally induced T regulatory cell 1 (Tr1)³ cells were shown to control inflammation and allergy, to induce transplantation tolerance, and to suppress immune responses to pathogens both in mice and humans (4, 5, 6, 7). Tr1 cells are characterized by their production of IL-10 and IFN-γ, their inability to secrete IL-2 and IL-4, and their poor proliferative potential (4). Tr1 cells were differentiated in vivo from mice injected with Ag-pulsed IL-10-treated dendritic cells (8) and in vitro from CD4⁺ T cells cultured with their cognate Ag in the presence of IL-10 both in mice and humans (4). Recently, it has been reported that the minimal engagement of the TCR and CD46 induces human CD4⁺ T cells with a Tr1-like phenotype, able to suppress T cell proliferation through the production of IL-10 (9). CD46 is a complement regulator that we initially described for its potential to costimulate T cells (10, 11). Nevertheless, unlike Tr1 cells, CD3/CD46-induced Tr1-like cells were described to exhibit a strong and prolonged proliferation (9).

Using low concentration of anti-CD3 to better assess for the costimulatory role of CD46, we differentiated weakly proliferating human Tr1 cells. We report here that the poor proliferation of CD3/CD46-induced Tr1 cells results from a defect in cell cycle progression and an increased sensitivity to cell death, both mechanisms being explained by a defective Akt pathway.

Human CD4⁺ T cells (>95% purity) were purified from blood donors (Etablissment Francais du Sang, Lyon, France) as described (10) and cultured in complete RPMI 1640. For cell culture, the mAbs used were: anti-CD3 (OKT3, 1 μg/ml, otherwise indicated), anti-CD46 (20.6, 10 μg/ml), anti-CD28 (CD28.2, 10 μg/ml) (all Abs purified in our laboratory), and anti-CD95 (ZB4, 1 μg/ml) and isotype control (10 μg/ml) (both from Beckman Coulter). For Western blot, the Abs used were: anti-Survivin (D-8), and the rabbit polyclonal Abs specific for Akt, p27/Kip1, cyclin A, cyclin E, and cyclin-dependent kinase 2 (Cdk2) (all obtained from Santa Cruz Biotechnology); anti-cyclin D1 (DCS6) and anti-phospho-Akt (193H12) (both purchased from Cell Signaling Technologies); anti-phospho-Erk1/2 (12D4) and the rabbit polyclonal Abs specific for Erk1/2 (both obtained from Upstate Biotechnology), and anti-actin (Sigma-Aldrich).

Naive CD4⁺ T cells were cultured on plates coated with the indicated mAbs, at a concentration of 1 × 10⁶ T cells/ml. When indicated, 10 U/ml recombinant human IL-2 was added (a gift from Dr. Menetrier-Caux, Institut National de la Santé et de la Recherche Médicale U590, Lyon France). [³H]Thymidine incorporation assays were as described (10). Cytokines were quantified from the supernatant of 3 days stimulated T cells by ELISA according to the manufacturer’s instructions: IL-2, TNF-α, and IL-8 (R&D Systems), IL-4, IL-5, IFN-γ, and IL-13 (Pierce Perbio), IL-10 and TGF-β (Bender MedSystems).

For cell death assessments, T cells were labeled with 67 nM of ToPro-3 (Molecular Probes). To determine cell division numbers, 2 × 10⁷ T cells/ml (RPMI 1640 plus 2% FCS) were kept at 37°C for 13 min after the addition of 0.5 M CFSE (Molecular Probes), washed three times in cold RPMI 1640 plus 10% FCS and cultured for various periods of time before being analyzed by FACS. For cell cycle analysis, 0.5 × 10⁶ CD4⁺ T cells were labeled 30 min at room temperature with 20 μM 7-aminoactinomycin D (7-AAD) in 500 μl of 0.03% saponin/NASS buffer (100 mM phosphate-citrate buffer (pH 6), 150 mM NaCl, 5 mM EDTA, and 0.5% BSA). Cells were kept on ice 5 min and 5 μM pyronin Y (PY) (Sigma-Aldrich) was added before analysis.

T cells were lysed in ice-cold buffer, the protein concentration from each lysate determined (MicroBCA kit; Pierce Perbio) and 50 μg of proteins were analyzed by SDS-PAGE and Western blotting as described (11).

A total of 1 μg of mRNA was extracted from cultured T cells with TRIzol (Invitrogen Life Technologies) and reverse transcribed into cDNA using Superscript II reverse transcriptase (Invitrogen Life Technologies), 80 pg of random hexamer primers (Promega) and 150 ng of oligo(dT)_12–18 (Invitrogen Life Technologies). Real-time PCR was performed with Platinium SYBR Green qPCR Supermix UDG (Invitrogen Life Technologies) on an Applied Biosystems GeneAmp 7600 thermocycler. The TATA box-binding protein (TBP) was used as a housekeeping gene for mRNA normalization. The specificity of amplification was checked after each run and for each sample with a melting curve. The TBP-specific primers were: forward 5′-TGCTCATACCGTGCTGCTATCTG-3′ and reverse 5′-TTCTCCCTCAAACCAACTTGTCAAC-3′. Bcl-2 and Bcl-xL primers were a gift from Dr. N. Bonnefoy-Berard (Institut National de la Santé et de la Recherche Médicale Unité 503, Lyon France). Primers amplification efficiencies were 1.90 for TBP, 1.94 for Bcl-2, and 1.96 for Bcl-x_L.

We previously reported that CD46 provides a costimulatory signal for human T cells (10). It was subsequently described that, in addition, CD46 favors the differentiation of human Tr1-like regulatory cells (9). To allow for an efficient assessment of the role of CD46 co-stimulation in Tr1 cell differentiation, we stimulated freshly purified naive human CD4⁺ T cells with an immobilized nonsaturating concentration of anti-CD3 (1 μg/ml) in combination with anti-CD46 (Fig. 1,A, left panel). Such CD3/CD46-stimulated T cells displayed a weak proliferation that was reminiscent of the Tr1 cell phenotype. On the contrary, as previously reported (9), higher concentrations of anti-CD3 did allow for a more sustained proliferation of CD3/CD46-stimulated T cells compared with CD3/CD28-stimulated (control effector) T cells (Fig. 1 A).

We then ensured that using a limiting concentration of anti-CD3 induced functional Tr1 cells. As shown in Fig. 1 B, low proliferating CD3/CD46-stimulated CD4⁺ T cells produced indeed the described phenotype of Tr1 cells (4, 9, 12). They produced IL-10, IFN-γ, and IL-5, but neither IL-2 nor IL-4. However, they did not produce TGF-β, a cytokine produced at low levels when saturating concentrations of anti-CD3 were used (9). Moreover, low proliferating CD3/CD46-stimulated CD4⁺ T cells produced higher concentrations of IL-8 than CD3-stimulated CD4⁺ T cells (0.45 ± 0.07 ng/ml vs 0.19 ± 0.03 ng/ml, respectively), but equivalent IL-13 (0.4 ng/ml vs 0,38 ng/ml, respectively) and no significant levels of IL-6 and TNF-α were produced (data not shown).

Finally, the supernatant of weakly proliferating CD3/CD46-stimulated T cells efficiently suppressed the proliferation of CD3/CD28-stimulated CD4⁺ T cells in a dose-dependent manner, whereas the supernatant of CD3-stimulated T cells did not (Fig. 1 C), confirming the regulatory phenotype of these cells. Altogether, these results clearly demonstrate that CD46 is a sufficient cosignal for the differentiation of weakly proliferating human suppressive Tr1 cells.

We reasoned as if the poor proliferation of Tr1-like cells could result from their inability to sustain cell division. A FACS analysis of CFSE-labeled stimulated naive CD4⁺ T cells revealed that both CD46 and CD28 costimulated T cells started to divide by day 3 with a similar kinetic (Fig. 2,A). However, by day 5 of stimulation Tr1-like cells hardly performed >4 cycles of division (5% divided ≥5 times) compared with control effector T cells (21% divided ≥5 times) (Fig. 2 A).

The incompetence of Tr1-like cells to sustain cell division might result from a cell-intrinsic deficiency in signal(s) controlling the progression of the cell cycle. Therefore, we analyzed the expression/down-regulation of proteins associated with cell cycle in CD3/CD46-stimulated T cells. Three days after stimulation, CD3/CD46-differentiated Tr1-like cells up-regulated the expression of the cyclins A, D1, and E and of Cdk2 to an equivalent level than CD3/CD28-stimulated CD4⁺ T cells. However, by day 5, CD3/CD46-stimulated T cells failed to sustain the expression of the cyclins (Fig. 2 B).

We then examined the down-regulation of the inhibitor of cell cycle progression p27/Kip1. Whereas CD28 costimulated T cells degraded p27/Kip1 as soon as day 1 after activation and maintained such degradation up to day 5, CD46 costimulated T cells only partially down-regulated p27/Kip1 mainly on day 3 (Fig. 2,B). Since the degradation of p27/Kip1 is required for cells to progress from the G₀/G₁ phase to the S phase of the cell cycle (13), we wondered whether CD3/CD46-stimulated T cells might have a defect in cell cycle progression. Staining cells with 7-AAD and PY revealed that by day 3 of culture, CD3/CD46-stimulated T cells accumulated in the G₀/G₁ phase more abundantly than CD3/CD28-stimulated T cells (Fig. 2 C). This result indicates that the lack of sustained proliferation of CD3/CD46-generated Tr1-like cells is due, at least partially, to a G₀/G₁ blockage in their cell cycle progression.

In addition to a cell-intrinsic defect in cell cycle progression, the lack of sustained Tr1-like cell proliferation could result from an increased sensitivity of these cells to death. Therefore, we examined the survival of proliferating CD3/CD46-stimulated T cells by determining the rate of dead cells by ToPro-3 staining after 3, 4, and 5 days of culture. Until day 4, CD4⁺ T cells displayed comparable levels of apoptosis whatever the nature of the stimulation, CD3, CD3/CD28, or CD3/CD46 (Fig. 3,A). However, by day 5 >30% of the CD3/CD46-stimulated cells were apoptotic whereas <20% of cells in all other conditions of culture were ToPro-3⁺. We then asked if the sensitivity to death of Tr1-like cells correlated with their proliferation because Tr1 are low-proliferating cells. Indeed, the more 5-day CD3/CD46-stimulated T cells divided, the more they underwent apoptosis with >40% of the cells that divided 5 times that were ToPro-3⁺ (Fig. 3 B). On the contrary, whatever the number of division considered, we constantly observed ∼20% of CD3/CD28-stimulated T cells that were ToPro-3⁺.

Since Tr1-like cells do not produce IL-2, it was possible that CD3/CD46-stimulated T cells died because of the absence of a survival signal provided by IL-2. However, adding IL-2 to the CD3/CD46-stimulated T cell culture only marginally rescued differentiated Tr1-like cells from death (Fig. 3 C). Therefore, the defective proliferation of Tr1-like cells is due not only to a limited potential of proliferation but also to an increased sensitivity to cell death.

Likewise, proliferating T cells may become sensitive to death through a Fas/FasL-mediated mechanism (14). However, preventing the interaction between Fas and FasL with a blocking anti-Fas mAb did not rescue CD3/CD46-stimulated CD4⁺ T cells from death (Fig. 3,D). Finally, the death of CD3/CD46-stimulated T cells could result from a defect of Bcl-2 and/or Bcl-xL expression, two anti-apoptotic genes crucial for T cell survival (15). However, as revealed by real-time RT-PCR analysis from 3 days stimulated T cells, CD46 and CD28 costimulation up-regulated the expression of Bcl-xL mRNA to an equivalent extend, above the one observed from CD3-alone-stimulated T cells (Fig. 3 E). Moreover, the up-regulation of the Bcl-2 mRNA induced in CD3-stimulated T cells was neither impaired nor increased when cells were costimulated with either CD46 or CD28. Thus, proliferating Tr1-like cell death is not related to a defect of expression of either Bcl-2 or Bcl-xL. Altogether our results strongly suggest that the CD3/CD46-differentiated Tr1-like cells die because of an abortive proliferation mechanism.

The differences in terms of proliferation and survival between Tr1-like cells and effector T cells, evoked that an important intracellular signaling pathway might be impaired in CD3/CD46-stimulated naive T cells. The PI3K/Akt pathway appeared to be a privileged target since Akt activation is crucial in CD28 costimulation-dependent T cell proliferation and survival (16, 17). We looked for Akt activation by Western blot using an activated Akt-specific mAb. Very interestingly, only marginal activation of Akt was detected from lysate of T cells stimulated for 3 days with CD3 and CD46 (Fig. 4,A). On the contrary, as expected Akt was heavily activated in 3 days CD3/CD28-stimulated T cells, and such activation was strongly maintained until day 5. The defective Akt phosphorylation pathway in CD46-induced Tr1 cells was specific because the Erk1/2 MAPK pathway was equivalently phosphorylated from 3 day CD3/CD46- and CD3/CD28-stimulated naive CD4⁺ T cells (Fig. 4,A). Moreover, although the PI3K/Akt pathway was required for CD3/CD46-activated T cell proliferation, the sensitivity of such cells to PI3K-specific inhibitors, wortmannin and LY2940002, was much more exacerbated compared with CD3/CD28-stimulated T cells and close to CD3-activated T cells, confirming the weak Akt activation in Tr1-like cells (Fig. 4,B and data not shown). On the contrary, CD3/CD46-stimulated CD4⁺ T cells were as sensitive as CD3/CD28-stimulated CD4⁺ T cells toward the Mek/Erk1/2-specific inhibitor U0126 (Fig. 4 C).

In murine T cells, the activation of the Akt kinase regulates the expression of survivin, a gene recently described to be crucial to maintain murine CD3/CD28-stimulated CD4⁺ T cell division over time and to antagonize T cell apoptosis (18). We therefore assessed whether an absence of Survivin expression might explain the sensitivity of the Tr1-like cells to death. First, we found that, as in mouse T cells, the costimulation of human CD4⁺ T cells with CD3/CD28 did induce the expression of the Survivin protein by day 3 (Fig. 4 D). Second, CD3/CD46-stimulated T cells did not up-regulate the expression of survivin over the level observed from CD3-stimulated T cells. Therefore, the impairment of CD46 to allow the induction of expression of Survivin probably explains why Trl-like cells have a limited proliferative potential and exhibit an increased potency to death.

The present work demonstrates that human CD3/CD46-induced Tr1 cells undergo abortive proliferation caused by a defect of cell cycle progression associated with a death of proliferating cells. The identification of a defective Akt pathway in human Tr1-like cells explains this particular phenotype and highlights several biological characteristics of these regulatory T cells. First, it explains why Tr1-like cells are incompetent at producing IL-2. This cytokine is produced by T cells when both Akt and Erk signaling pathways are activated (19); however, we and others (11, 20) reported that the Erk pathway is fully activated in CD3/CD46-stimulated T cells (Fig. 4 A). Second, the Akt pathway is required to optimally degrade p27/Kip1 (16), an inefficient event of Tr1-like cells, that is necessary for an efficient T cell proliferation. Third, the Akt pathway is crucial for the Ag-induced T cell survival and regulates the Survivin expression (17, 18), two processes that we show here to be defective in Tr1-like cells. Because Bcl-xL induction requires Akt activation it is possible that, in CD3/CD46-stimulated naive CD4⁺ T cells, the weak Akt activation is sufficient to induce Bcl-xL expression but not to sustain proliferation and Survivin expression.

The nature of the signals involved in Tr1 differentiation in vivo still remains elusive and might be due to the circumstances. By engaging CD46, at least one self (complement C3b factor) and one nonself (streptococcal M protein) molecules have been shown to induce Tr1 cells (9, 21). Whatever the nature of the ligand engaging CD46 is, our results describe a mechanism that might be crucial to control the functional activity of the peripheral CD46-induced Tr1 cells to limit bystander incongruous immunosuppression. Indeed, this control could be achieved by the elimination of activated Tr1 cells through a process of abortive proliferation.

We thank Dr. C. Viret for comments on the manuscript, Dr. N. Bonnefoy-Berard, Dr. L. Perrin-Cocon, and Dr. Menetrier-Caux for gift of reagents and B. Vanbervliet for technical help. G. Meiffren is recipient of a fellowship from the French Ministry of National Education, Research and Technology.

The authors have no financial conflict of interest.