Induction of granzyme B expression in T-cell receptor/CD28-stimulated human regulatory T cells is suppressed by inhibitors of the PI3K-mTOR pathway

The following antibodies were used for the flow cytometric evaluation of T cell subsets in this study: anti-CD4 (clone RPA-T4), anti-CD25 (clone BC96), anti-CD127 (clone HCD127), anti-FOXP3 (clone 259D), all from Biolegend Inc. (San Diego, CA). Anti-granzyme B (clone GB11) was from eBioscience (San Diego, CA). For the carboxy-fluorescein diacetate, succinimidyl ester (CFSE)-based proliferation studies only, anti-FOXP3 (clone 259D/C7) was purchased from BD Biosciences (San Jose, CA). Cells were permeabilized using a commercially available fixation and permeabilization kit (FOXP3 fix/perm kit; Biolegend) per the manufacturer's instructions. Propidium iodide (PI) and annexin V-PE were from Pharmingen (San Diego, CA). CFSE (Invitrogen, Carlsbad, CA) was used at a final concentration of 1 μM. Cells were labeled for 5 minutes in phosphate buffered saline (PBS) containing CFSE at 37°C then placed in pre-warmed media for 10 minutes, washed two times and plated as indicated in the figures. Flow cytometry was performed on a FACSCanto II flow cytometer (BD Biosciences) using FACSDiva software (BD Biosciences) except for cell sorting which was performed on a FACSVantage flow cytometer (BD Biosciences). Final data analysis was performed using FloJo software (Tree Star, Inc, Ashland, OR). The statistical significance of differences in granzyme B expression and Foxp3 expression between subsets was assessed using the Student's t-test.

Institutional review board (IRB) approval was obtained from the University of Utah Health Sciences Center to obtain peripheral blood from normal adult volunteers. All blood donors provided informed consent prior to blood collection. Blood was collected by peripheral venipuncture in ethylenediaminetetra-acetic acid (EDTA)-coated vacutainer tubes (BD, Franklin Lakes, NJ). Peripheral blood mononuclear cells were isolated from normal donors by density gradient centrifugation using Ficol-Paque density medium (GE Healthcare, Uppsala, Sweden) according to the manufacturers recommendations. Subsequently, CD4+, CD25+ Tregs were enriched using a magnetic bead-based kit (Miltenyi Biotec, Auburn CA) according to the manufacturers instructions. This technique routinely resulted in samples that were enriched in regulatory T cells to a purity of approximately 60-70% or more based on FOXP3 staining by flow cytometry. However, higher purity Tregs were required for the cytotoxicity assays. For the 6-hour cytotoxicity assays, peripheral blood nTregs were isolated based on flow cytometric sorting of the CD4+, CD25^bright (brightest 1/3 of CD25 positive cells) T cell fraction. For the 48-hour cytotoxicity assays, isolation based on the CD4+, CD25+, CD127-dim/negative expression pattern characteristic of Tregs was performed using a magnetic bead based kit (Miltenyi Biotec). Both methods resulted in a similar pre-expansion purity of >90% Tregs, based on flow cytometic evaluation of FOXP3 expression.

Human peripheral blood T cells were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum (both from Hyclone, Logan, UT) at 37°C in 5% CO₂ for the indicated time. Where indicated, cells were expanded with anti-CD3/anti-CD28 coated beads (Invitrogen) or anti-CD3 coated beads (Invitrogen), both at a bead:cell ratio of 4:1. Where indicated, recombinant human IL-2 (rhIL-2) at 250 IU/mL (R&D Systems, Minneapolis, MN) was also added. In some experiments, inhibitors were added to the culture 30 minutes prior to the addition of the above stimulants. These included rapamycin (Calbiochem, San Diego, CA) at 10 ng/mL or 100 ng/mL, or the PI3K inhibitor LY294002 (Cell Signaling Technology, Danvers, MA) at 1 μM or10 μM, all in dimethyl sulfoxide (DMSO; Sigma, St. Louis, MO). In some cases, freshly enriched CD4+, CD25+ T cell samples were frozen in the vapor phase of liquid nitrogen in freezing medium (10% DMSO in RPMI 1640 medium with 20% fetal calf serum) then thawed and analyzed by flow cytometry at the same time as their cultured (stimulated) counterparts. These samples are designated as "unstimulated" in the figures. Freezing in this way did not have an effect on antigen expression patterns or viability. The human Hodgkin lymphoma cell line, L428 was obtained from the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany) and maintained in RPMI medium supplemented with 10% fetal calf serum. For the cytoxocity assays, purified peripheral blood nTregs were expanded as indicated then co-cultured at a 3:1 effector:target ratio with L428 target cells, pre-labeled with CFSE in media without added stimuli. Controls consisted of L428 target cells alone or L428 target cells co-cultured with non-activated conventional CD4+ T cells (Tconv). An additional positive control for the 6-hour samples consisted of L428 cells alone cultured with 10 μM staurosporine (Sigma) to induce apoptosis. Samples were then stained with annexin-V-PE to assess early apoptosis in the 6-hour assays or for propidium iodide to assess late apoptosis in the 48-hour assays. Annexin-V binding or PI staining were then assessed in the CFSE labeled target cell population to quantify apoptosis. Significance was assessed with the Student's t test.

Tregs and Tconv were isolated from the peripheral blood of normal adults. Freshly isolated cells were evaluated for expression of CD4, CD25, FOXP3 and granzyme B by flow cytometry. FOXP3 expression was only seen in the CD25+ subset. Granzyme B expression was not detected in peripheral blood CD4+ T cells in any of the freshly isolated samples from the four normal volunteers tested. Representative dot plots showing expression of the above markers are shown in Figure 1 for the fraction depleted of Tregs (Figure 1A) and the fraction enriched in Tregs (Figure 1B).

Tregs were enriched from peripheral blood then subjected to stimulation for 4 days in vitro using IL-2 alone (250 IU/mL), CD3 stimulation alone with anti-CD3 coated beads or combined CD3/CD28 stimulation with anti-CD3 and anti-CD28 coated beads with IL-2 (250 IU/mL). On day 4, samples were simultaneously stained for CD4, FOXP3 and granzyme B (or an equal concentration of isotype control antibody instead of anti-granzyme B) and evaluated by flow cytometry (Figure 2). Granzyme B was assessed in the distinct FOXP3 bright population representing Tregs (gate shown) and granzyme B staining was expressed as mean fluorescence intensity (MFI) (Figure 2). Neither IL-2 alone or anti-CD3 alone promoted an increase in granzyme B expression. However, anti-CD3/anti-CD28 coated beads plus IL-2 induced a marked increase in the level of granzyme B staining. Staining of a duplicate CD3/CD28 and IL-2 stimulated sample with anti-CD4, anti-FOXP3 and a granzyme B isotype control antibody confirms the specificity of anti-granzyme B antibody staining (Figure 2).

Enriched peripheral blood nTregs were subjected to in vitro expansion for 4 days using CD3/CD28 beads and IL-2 (250 IU/mL) in the presence of the indicated inhibitor. In the case of rapamycin, similar conditions have previously been shown to lead to selective expansion of CD4+, CD25^bright, FOXP3+ Tregs with retained suppressive capabilities [21]. Using flow cytometric evaluation of CD4, FOXP3 and granzyme B and gating on FOXP3-bright cells (Figure 3A), we found that PI3K inhibition resulted in a dose-dependent suppression of granzyme B expression that was complete at 10 μM LY294002 (Figure 3B). Rapamycin treatment at only 10 ng/mL also resulted in marked suppression of granzyme B expression (Figure 3B). To confirm that these findings were not simply due to the induction of cell death by the inhibitors, the cultures were evaluated for PI staining by flow cytometry (Figure 3C). There was no increase in the number of PI positive cells in any of the inhibitor treated samples as compared to samples without inhibitor treatment. Flow cytometric evaluation of FOXP3-bright cells demonstrated an activation induced increase in the level of FOXP3 expression (Figure 3A and 3D), versus that seen in freshly isolated peripheral blood Tregs in all stimulated samples, regardless of inhibitor. Evaluation of triplicate samples showed no significant suppression (or augmentation) of activation-induced enhancement of FOXP3 expression in the rapamycin or low dose LY294002 treated samples (Figure 3D). The sample treated with high dose LY294002 (10 μM) showed a slight but significant decrease in activation induced FOXP3 enhancement (Figure 3D). Given this finding, we conclude that there is no correlation between granzyme B expression and FOXP3 expression.

The proliferation of Tregs and Tconv in response to the strong stimuli of CD3/CD28 beads and IL-2 was assessed in the presence of mTOR or PI3K inhibitors. Enriched peripheral blood Tregs (approximately 80% Tregs, 20% Tconv), pre-labeled with CFSE, were cultured with rapamycin (10 and 100 ng/mL) or LY294002 (1 μM) plus CD3/CD28 beads and IL-2 (250 IU/mL) for 4 days. To quantify the initial number of Tregs and Tconv in the culture, CD4, FOXP3 and CD127 levels were assessed by flow cytometry prior to expansion (Figure 4A). After expansion, single cell evaluation of CFSE staining and FOXP3 expression was performed by flow cytometry in order to separately quantify proliferation in the FOXP3+ and FOXP3- subsets. This revealed preferential expansion of Tregs in the presence of rapamycin at 10 ng/mL and 100 ng/mL but similar proliferation in the sample expanded without inhibitors and in the sample expanded in the presence of LY294002 (Figure 4B). To more rigorously compare the proliferative response of the FOXP3+ and FOXP3- fractions in the context of mTOR inhibition by rapamycin the percentage of cells in each generation was quantified (Table 1). In the setting of 10 ng/mL rapamycin, which corresponds to therapeutic human plasma levels [23], six total Treg generations and five Tconv generations could be defined by CFSE staining with 42.3% of Tregs in generations 5 and 6 versus 10.4% of Tconv. In the samples subjected to the highest rapamyin concentration (100 ng/mL) similar results were obtained although there was suppression of proliferation in both Treg and Tconv fractions (22% of Tregs in generations 4 and 5 vs 6.7% of Tconv). These findings show that under identical conditions, proliferation of Tregs is highly favored by rapamycin treatment.

The IL-7 receptor (CD127) has been shown to be expressed at lower levels in Foxp3+ Tregs and to inversely correlate with their suppressive capacity [24]. In agreement, we found that prior to expansion, CD127 levels were indeed lower in FOXP3+ Tregs (Figure 4A/Figure 5). We wondered whether expansion with CD3/CD28 stimulation and IL-2 resulted in a retained pattern of low CD127 expression in Tregs. Interestingly, we found essentially identical CD127 levels in expanded Tregs and Tconv (Figure 5) suggesting a limited utility for CD127-based purification techniques for separating in vitro expanded Tregs from Tconv.

Multiple previous studies have demonstrated that Tregs expanded in rapamycin maintain their suppressive function [20, 21, 25]. However, such assays require co-culture of Tregs with target cells, usually CD4+ or CD8+ conventional T cells, in the presence of TCR/CD28 stimulation either via antibodies or irradiated antigen presenting cells, the very conditions that we have shown to induce granzyme B expression. In order to avoid inducing de novo granzyme B expression during suppression assays in Tregs previously expanded in rapamycin, and to separate the suppressive and cytotoxic function of Tregs, we developed a flow cytometry based cytotoxicity assay using a Hodgkin lymphoma cell line (L428) that we had previously observed to be susceptible to in vitro Treg-mediated cytotoxicity (unpublished observation). We used this observation to test the cytotoxic function of Tregs expanded in vitro with and without rapamycin. Peripheral blood nTregs were isolated by flow cytometry based on CD4+/CD25^bright characteristics (brightest 1/3 of CD25+ cells; Figure 6A/B) or by a bead-based method designed to isolate based on a CD4+, CD25+, CD127- surface antigen expression profile (Figure 6C/D). Both methods gave similar pre-expansion nTreg purities of >90%, based on FOXP3 expression. nTregs were then expanded with CD3/CD28 beads plus IL-2, with and without rapamycin (10 ng/mL) as indicated. Post-expansion Treg purities are shown in Figure 6A/C. Samples were subsequently co-cultured at a 3:1 ratio of FOXP3+ Tregs (or Treg-depleted nonactivated Tconv as a control) to CFSE-labeled L428 target cells for 6 or 48 hours as indicated. Apoptosis was measured by flow cytometric analysis of the CFSE labeled target cell fraction. Early apoptosis was measured by annexin-V binding to target cells in the 6 hour cytotoxicity assays. For the 48-hour assays, late apoptosis was measured by PI staining of the target cell population. Representative dot plots from a 6-hour cytotoxicity assay are shown in Figures 6A/B. Statistical analysis of multiple 6-hour replicates showed a significant increase in annexin-V-positive apoptotic target cells co-cultured with nTregs expanded without rapamycin (versus target cells cultured alone; P = .04). The increase in apoptotic target cells co-cultured with nTregs expanded in rapamycin was not statistically significant (P = .07). Staurosporine treated cells served as a positive apoptosis control. For the 48-hour cytotoxicity assays, representative data, including statistical significance, are shown in Figure 6C/D. Although cytotoxicity was present for both Treg subsets (+/- rapamycin), it was significantly higher for Tregs expanded without rapamycin. Thus, in this system cytotoxicity at both 6 and 48 hours correlates with granzyme B expression, and inversely with rapamycin exposure, in expanded nTregs.