Inhibition of TGF-β1 Signaling Promotes Central Memory T Cell Differentiation

Abstract

The present study affirmed that isolated CD8⁺ T cells express mRNA and produce TGF-β following cognate peptide recognition. Blockage of endogenous TGF-β with either a TGF-β-blocking antibody or a small molecule inhibitor of TGF-βRI enhances the generation of CD62L^high/CD44^high central memory CD8⁺ T cells accompanied with a robust recall response. Interestingly, the augmentation within the central memory T cell pool occurs in lieu of cellular proliferation or activation, but with the expected increase in the ratio of the Eomes/T-bet transcriptional factors. Yet, the signal transduction pathway(s) seems to be non-canonical, independent of SMAD or mTOR signaling. Enhancement of central memory generation by TGF-β blockade is also confirmed in human PBMCs. The findings underscore the role(s) that autocrine TGF-β plays in T cell homeostasis and, in particular, the balance of effector/memory and central/memory T cells. These results may provide a rationale to targeting TGF-β signaling to enhance antigen-specific CD8⁺ T cell memory against a lethal infection or cancer.

Introduction

Recent progress in immunology has emphasized the importance of generating better quality memory T cells. Central to those efforts has been an expectation that the development of a robust long-term T cell memory would fortify vaccines and enhance host protection against infectious diseases and cancer immunotherapy. Indeed, it has been reported that, in both mice and non-human primates, central memory CD8⁺ T cells are superior to effector memory CD8⁺ T cells as mediators of host immune-based protection against viruses and cancer (1–3). In mice, central and effector memory CD8⁺ T cells can be separated into two distinct populations according to their respective CD44 and CD62L expression levels (1, 3–6). A CD44^highCD62^low CD8⁺ T cell population that is found mainly in peripheral tissues and rapidly acquires effector functions constitutes the effector memory, while CD8⁺ T cells expressing a CD44^highCD62L^high population, which typically reside in secondary lymph nodes where they acquire profound proliferative capacities upon antigen recognition, represent the central memory T cells. Investigators are beginning to unravel the molecular pathways that regulated the differentiation of long-lived central memory T cells. Along with those phenotypic markers, certain cytokines such as IL-2 and IL-15, and selective intracellular signal transduction molecules, such as AMPK-α and mTOR, have been implicated in the differentiation of effector to central memory CD8⁺ T cells (6–10).

TGF-β is a well-known immune suppressive cytokine that affects multiple cell types within the immune system. For example, TGF-β controls T cell homeostasis by directly inhibiting both proliferation and activation of naïve CD4⁺ and CD8⁺ T cells (11). Disruption of TGF signaling in naïve T cells results in the emergence of autoimmune diseases in mice (11, 12). The inhibitory effects of TGF-β are not limited to the activation of naïve T cells. CD8⁺ T cells activated in the presence of exogenous TGF-β do not acquire cytotoxic T lymphocyte (CTL) function (13), and CD4⁺ T cells fail to become TH1 or TH2 cells (14, 15). However, the fact that CD8⁺ T cells produce endogenous TGF-β upon activation (16) and despite the presence of the cytokine, naïve T cells still differentiate into effector cells in a vaccine setting. These observations underscore the intriguing differential effects of endogenous and exogenous levels of TGF-β on T cell activation and differentiation. Since exogenous TGF-β effects on T cells present a seemingly different set of parameters, such as the complexities of the activation mechanism of latent TGF-β (17), the present study focused on the physiological changes associated with CD8⁺ T cell generation and differentiation and autocrine TGF-β.

Splenocytes from H-2Db-restricted NP68-specific CD8⁺ TCR transgenic mice (F5 mice) offer a possible in vitro model to directly examine the role(s) that TGF-β and small molecule TGF-β receptor inhibitors play in CD8⁺ T cell differentiation (18). Upon stimulation with cognate peptide, the CD8⁺ F5 T cells acquire both phenotypic changes and immune effector functions that are reminiscent of those described during the generation of an in vivo antigen-specific T cell response, i.e., priming, expansion, contraction, and memory. The present study affirmed that isolated CD8⁺ T cells express mRNA and produce TGF-β following cognate peptide recognition. In addition, blockage of endogenous TGF-β with either a TGF-β-blocking antibody or a small molecule inhibitor of TGF-βRI enhances the generation of central memory T cells. Interestingly, the augmentation within the central memory T cell pool occurs in lieu of cellular activation and seems to be mediated via a pathway independent of SMAD. The findings underscore the role(s) that autocrine TGF-β plays in T cell homeostasis and, in particular, the balance between effector and central memory T cells.

Materials and Methods

Animals

Female C57BL/6 mice (8–12 weeks old) were obtained from the National Cancer Institute, Frederick Cancer Research Facility (Frederick, MD). F5 mice that are transgenic (Tg) for nucleoprotein of influenza virus A/NT/60/68 (³⁶⁶ASNENMDAM³⁷⁴; NP68)-specific, H-2Db-restricted T cell receptor (19, 20) were obtained from Taconic Farms (Hudson, NY). All mice were housed and maintained in microisolator cages under specific pathogen-free conditions and in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care guidelines. All experimental studies were carried out under the approval of the Intramural Animal Care and Use Committee. Splenocytes from SMAD-2 conditional knockout mouse were kindly provided by Dr. Kang of the University of Massachusetts Medical School (21).

Reagents

TGF-β mAb (clone #1D11) was purchased from R&D Systems (Minneapolis, MN). SD208 was purchased from Sigma and dissolved in dimethylsulfoxide (DMSO), and diluted in culture media; maximum concentration was 0.1%. Recombinant human TGF-β1 was purchased from PeproTech. H-2D^b-restricted influenza virus A/NT/60/68 peptide (³⁶⁶ASNENMDAM³⁷⁴; NP68 peptide) was synthesized by CPC Scientific.

Poxvirus Constructs

Recombinant fowlpox viruses containing murine B7-1, ICAM-1, and LFA-3 genes in combination with nucleoprotein of influenza virus A/NT/60/68 (³⁶⁶ASNENMDAM³⁷⁴;NP68) (rV/F-NP68-TRICOM) have been described previously (22). The recombinant fowlpox virus containing the gene for murine GM-CSF has also been described previously (23).

In Vitro T cell Assay

Primary splenocytes were dispersed into single-cell suspensions, the red blood cells were removed by lysis, and the remaining cells seeded into 6-well plates at 6×10⁵ cells/ml in complete RPMI (RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 units/mL penicillin and 100 µg/mL streptomycin) media. Splenocytes from C57BL/6 mice were seeded into 6-well plates at 3×10⁶ cells/ml in complete RPMI and stimulated with soluble anti-CD3e and anti-CD28. Splenocytes from F5 mice were purified by CD8⁺ T isolation Kit II (Miltenyi Biotec) and stimulated with 10⁻⁴ µg/ml of NP68 peptide, soluble diametric mouse H-2Db (Dimmer X) and anti-CD28 then used in the appropriate experiments.

PBMCs were obtained from heparinized blood of healthy donors and separated using lymphocyte separation medium gradient (MP Biomedicals) according to the manufacturer’s guidelines. CD45RO⁺ memory T cells were depleted from the PBMCs using anti-CD45RO FITC Ab and anti-FITC beads (Miltenyi Biotec). CD45RO⁺ cellsdepleted PBMCs were seeded into 24-well plates at 3×10⁵ cells/ml in RPMI 1640 medium (Mediatech), supplemented with 10% human AB serum (Gemini Bio-Products) and stimulated with plate-coated mouse anti-human CD3 (1 µg/ml) (eBioscience); various doses of SD208 were added to the culture media 36 hr after stimulation.

Real Time PCR

Total RNA prepared by using the RNeasy kit (QIAGEN) was reverse transcribed with Advantage RT-for-PCR (Clontech). cDNA (20 ng) was amplified in triplicate using Gene Expression Master Mix and the following TaqMan gene expression assays (Applied Biosystems, Foster City, CA): Tgfb1 (Mm01178820), Tcf7 (Mm00493445), Bmi1 (Mm03053308), Bcl6 (Msm00477633), Prdm1 (Mm00476128), Eomes (Mm01351985), Tbx21 (Mm00450960) and mGAPDH 4352339E). Mean Ct values for target genes were normalized to mean Ct values for the endogenous control GAPDH [−ΔCt=Ct(GAPDH)-Ct(target gene)]. The ratio of mRNA expression of target gene vs. GAPDH was defined as 2(−ΔCt).

Cytokine Assays

Mouse IFN-γ, IL-2 and TGF-β1 ELISAs were performed using Quantikine® ELISA kits (R&D Systems, Minneapolis, MN). The TGF-β3 ELISA was performed using a kit from MyBiosource.com. Each ELISA protocol was carried out according to the manufacturers’ instructions. Human IFN-γ ELISA was performed using human IFN-γ ELISA kit (Invitrogen), according to the manufacturer’s protocol.

Flow Cytometry Analysis

Mouse splenocytes or human PBMCs were stained with antibodies to the following cellsurface markers: mouse-CD8a, CD19, CD44, CD62L and human-CD8, CD4, CD62L, CD45RA and CD45RO, which were purchased from BD Biosciences (Mountain View, CA). Antibodies that recognize mouse-IL-7R (CD127), T-bet and EOMES were purchased from eBioscience and the human-CCR7 antibody was purchased from R&D Systems. Annexin V staining was performed using annexin V staining kit (BD Biosciences). Cells were also stained with appropriate isotype-matched controls. To identify influenza A NP68-specific cells, splenocytes were stained with NP68 dextramer (ASNENMDAM) (Immudex) or LCMV dextramer (³⁹⁶FQPQNGQFI⁴⁰⁴) (Immudex). Intracellular protein staining was performed using Foxp3 staining buffer set (eBioscience) and protocol. Stained cells were acquired on a FACSCalibur or LSRII flow cytometer (BD Biosciences). Dead cells were excluded from the analysis based on scatter profile.

Western Blot Analysis

Western blot was conducted as previously described (18) using phospho-specific antibodies against SMAD2 (Ser465/467) (#3101), SMAD3 (Ser423/425) (#9520), p44/42MAPK (Thr202/Try204) (#4370), p38MAPK (Thr180/182) (#4511), SAPK/JNK (Thr183/185) (#4671), p70S6 Kinase (Ser371) (#9208), p70S6 Kinase (Thr389) (#9234) and S6 Ribosomal Protein (Ser240/244) (#5364). All of the phospho-specific antibodies were purchased from Cell Signaling Technology, Inc. (Danvers, MA). Detection was performed with the Odyssey Infrared imaging system (LI-COR Biotechnology).

Adoptive Transfer

Splenocytes from F5 mice were pre-incubated for 1 hr with 3 µM of SD208 or vehicle, and then stimulated with 10⁻⁵ µg/ml of cognate peptide for 96 hr. Splenic CD8⁺ T cells were purified with magnetic beads and 1×10⁷ cells/ml were labeled with 1 µM of CFSE, incubated for 10 min at 37°C and washed twice with PBS. Purified CD8⁺ T cells (1×10⁷) were adoptively transferred into naïve C57B6 mice on day 0 and the mice were vaccinated with rF-NP68-TRICOM on day 3 to induce a recall response.

Statistical Analysis

Statistical significance was calculated using GraphPad Prism statistical software (Version 5.0c) (GraphPad Software, Inc.). Where not specified, results of tests of significance are reported as p values, derived from either the two-tailed unpaired Student’s t-test or using one-way ANOVA followed by Tukey's multiple comparison test to compare between the groups. In the graphic representations of data, y-axis error bars indicate the SEM for each point on the graph.

Results

TGF-β Production by Isolated CD8⁺ T Cells Following Cognate Peptide Stimulation

Two to 4 hrs after recognition of the cognate peptide-bound MHC in the presence of anti-CD28, CD8⁺ T cells isolated from F5 TCR.Tg mice produced a spike in TGF-β mRNA. The CD8⁺ T cells also acquired an effector memory phenotype (i.e., CD44^high/CD62L^low) in the absence of measurable proliferative activity (Figure S1) and this initial 24-hr period has been referred to as the T cell priming phase (18). From 24 hr to 72 hr post-cognate peptide stimulation, the proliferating T cells endogenously produce substantial amounts of IFN-γ, IL-2 and TGF-β, which are secreted into the culture supernatant (Figure 1). This time interval has been referred to as the expansion phase (18), during which the percentage of CD8⁺ T cells expressing surface markers indicating a central memory phenotype, CD62L^high/CD44^high, increased from 11% at 24 hr to >65% at 72–96 hr, respectively (Figure S1B). T cell production of IFN-γ and IL-2 following cognate peptide stimulation has been tied to the acquisition of effector functions and proliferation, respectively. Indeed, beginning at 96 hrs post-peptide stimulation, if those CD8⁺ T cells were rested for an additional 48 hr in the presence of a low dose of IL-2, two distinct cell populations emerged which were isolated by flow cytometry based on their CD62L expression levels (Figure S1B; day 6 panel). When each CD8+ T cell population was restimulated with the cognate peptide, the CD62L^high/CD44^high CD8⁺ T cells produced higher levels of IL-2 (597 vs. 224 pg IL-2/ml; p<0.0001) (Figure S1C), and had a higher proliferative response (Figure S1F) than the corresponding CD62L^low/CD44^high CD8⁺ T cells. Those functional differences are consistent with identifying those cells as central and effector memory T cells (1, 3–6), respectively. Of interest were the T cell function(s), particularly those of T cell differentiation, which may be associated with endogenous TGF-β production, which became the focus of subsequent study.

Figure 1.

Temporal-dependent in vitro production of TGF-β, IFN-γ and IL-2 by isolated CD8⁺ T cells following cognate peptide stimulation. Splenic CD8⁺ T cells from TCR transgenic mice for the nucleoprotein of influenza virus NP68 (F5 Tg-mice) were isolated by magnetic beads, and stimulated with 10⁻⁴ µg/ml of cognate peptide (NP68 peptide), 1.0 µg/ml of H2Db-dimer X and 2.0 µg/ml of anti-CD28. (A) TGF-β mRNA as measured by quantitative PCR at the indicated time points. (B) TGF-β1, (C) IFN-γ and (D) IL-2 production in the T cell culture supernatant as measured using appropriate cytokine ELISA assays. Data represent the mean ± SE of triplicate samples from three independent experiments.

Blockade of TGF-β signaling increases central memory phenotype in isolated T cells from F5 TCR.Tg and wild-type mice

These studies utilized two TGF-β blocking reagents – an anti-TGF-β monoclonal antibody and a TGF-βR1 kinase inhibitor, SD208 – to examine what consequences might occur in the transition of peptide-stimulated T cells through the memory differentiation program with the interruption of biological contributions of TGF-β. As previously described, 72 hr after post-cognate peptide stimulation, 65–70% of the CD8⁺ T cells from F5.TCR.Tg mice expressed high levels of CD62L and CD44 (i.e., CD62L^high and CD44^high), indicative of a central memory phenotype. The addition of the anti-TGF-β-specific mAb to the F5 CD8⁺ T cell culture media resulted in a dose-dependent increase in the percentage of cells expressing that phenotype (i.e., 85%: 10 µg/ml of anti-TGF-β mAb, Figure 2A). Other studies have used the expression of the homeostatic cytokine receptor, CD127, as another phenotypic marker for central memory T cells. Flow cytometric analysis confirmed that TGF-β blockade also increased CD62L^high/CD127^high central memory phenotype from 40% to 58% (10 µg/ml of TGF-β mAb, Figure 2A, lower panel). The acquisition of the phenotypic markers indicative of memory T cell differentiation occurred in the absence of any measurable changes in T cell proliferation as determined by CFSE (data not shown). As a result, there was an overall increase in the absolute number of central memory T cells (Figure 2B, solid bars) with a commensurate reduction in the number of CD62L^low/CD44^high effector memory T cells (Figure 2B, open bars). Blockade of endogenous TGF-β slightly reduced IFN-γ production, an indicator of CD8⁺ T cell activation (Figure 2C). In addition, no TGF-β3 was found in these culture supernatants (data not shown). To examine whether the observed increase in memory T cells coincided with the interruption of TGF-β signaling, isolated CD8⁺ T cells were stimulated with peptide-bound MHC and anti-CD28 and simultaneously treated with SD208, a TGF-βR1 kinase inhibitor. That also increased the percentage of T cells with the CD62L^high/CD44^high central memory phenotype in a dose-dependent manner (i.e., 66% vehicle vs 89% at 3 µM, Figure 2D). The interruption of TGF-β signaling by SD208 treatment also increased the total number of central memory T cells (Figure 2E, solid bars) while reducing CD62L^low/CD44^high effector memory T cells (Figure 2E, open bars).

Figure 2.

Blockade of TGF-β signaling increased central memory T cell phenotype. Splenic CD8⁺ T cells isolated from F5 mice were pretreated with either the anti-TGF-β mAb (0.1–10 µg/ml) or TGF-β receptor I kinase inhibitor, SD208 (0.3–3.0 µM), for 1 hr prior to stimulation with 10⁻⁴ µg/ml of the cognate peptide (NP68 peptide), 1.0 µg/ml of H2Db-dimer X and 2.0 µg/ml of anti-CD28 and all analyses were performed 72 h later. (A) Flow cytometric analyses of F5-CD8⁺ T cells, as determined by anti-CD62L and either anti-CD44 (upper panel) or anti-CD127 (lower panel) are shown. (B) Total number of CD62L^high/CD44^high central memory (solid bars) versus CD62L^low/CD44^high effector memory (open bars) CD8⁺ T cells recovered and assessed by trypan-blue exclusion. *, p<0.05 (0.1 µg/ml vs 1.0 and 10 µg/ml anti-TGF-β mAb); ***, p<0.001 (untreated vs. 0.1 µg/ml vs 1.0 and 10 µg/ml anti-TGF-β mAb) (C) IFN-γ production as measured by ELISA. **p<0.01 untreated vs. 0.1 µg/ml anti-TGF-β mAb); ***, p<0.001 (untreated vs. 1.0 and 10 µg/ml anti-TGF-β mAb) (D) SD208 effects on the phenotypic changes of F5-CD8⁺ T cells, as determined by anti-CD62L and anti-CD44. (E) Total number of CD62L^high/CD44^high central memory (solid bars) versus CD62L^low/CD44^high effector memory (open bars) CD8⁺ T cells recovered and assessed by trypan-blue exclusion. *, p<0.05 (untreated vs 0.3 µM SD208); **, p<0.05 (untreated vs 1 µM SD208); ***, p<0.001 (untreated vs. 3 µM SD208). (F–H) Splenocytes from C57BL/6 mice were pre-incubated with various doses of SD208 (0.3–3.0 µM) for 1 hr, and stimulated with 2.5 µg/ml of anti-CD3 and 1.25 µg/ml of anti-CD28 and all analyses were done 72 hr later. (F) Flow cytometric analyses of CD8⁺ and CD4⁺ T cells stained with anti-CD62L and anti-CD44 antibodies. Numbers in each upper right quadrant denote percentage of cells. (G and H) Total number of (G) CD8⁺ or (H) CD4⁺ CD62L^high/CD44^high central memory (solid bars) versus CD62L^low/CD44^high effector memory (open bars) T cells recovered following stimulation with cognate peptide. Cell numbers were determined by trypan-blue exclusion and the total number of CD8⁺, CD4⁺, CD62L^low/CD44^high, CD62L^high/CD44^high cells were determined based on flow cytometry data. (G) CD8⁺ - *, p<0.05 (0.3 vs 1.0 µM & 1.0 vs. 3.0 µM SD208); ***, p<0.001 (untreated vs. 0.3, 1.0 and 3.0 µM; 0.3 vs. 3.0 µM SD208). (H) CD4⁺ - ***, p<0.001 (untreated vs. 0.3, 1.0 and 3.0 µM SD208). Data are the mean ± SE of triplicate samples of at least three independent experiments. Statistical significance was measured by the one-way ANOVA followed by Tukey's multiple comparison test.

When CD4⁺ or CD8⁺ T cells isolated from naïve, wild-type B6 mice were stimulated in vitro with anti-CD3 and CD28, they proceeded through the same priming, expansion and contraction phases as previously described for the F5 CD8⁺ cells. In vitro treatment of either CD8⁺ or CD4⁺ T cells isolated from naïve B6 mice with SD208, followed by anti-CD3 and CD28 stimulation, resulted in a significant dose-dependent increase in the percentage of cells expressing the CD62L^high/CD44^high central memory phenotype (Figure 2F). For example, the percentage of CD8⁺ T cells that expressed high CD62L and CD44 levels was increased from 27% in untreated cells to 41%, 45% and 57% following treatment with 0.3, 1.0 and 3.0 µM of SD208, respectively. SD208 treatment up to 1 µM had no detrimental effect on the growth of either CD8⁺ or CD4⁺ T cells (data not shown), thereby increasing the total number of CD62L^high/CD44^high central memory cells in both T cell compartments (Figures 2G and 2H, solid bars). In the culture supernatants containing splenic T cells in the presence of 3 µM of SD208, there was an increase of IL-2 levels (data not shown) which might account for the reduction of CD4⁺ T cells via AICD (Figure 2H). In any case, the findings argue that the differentiation of central memory T cells by either CD8⁺ cells from F5 TCR.Tg mice or CD8⁺ and CD4⁺ T cells from wild-type B6 mice can be controlled, in part, through TGF-β receptor signaling.

Exogenous TGF-β inhibits central memory CD8⁺ T cells differentiation

The complexities of TGF-β interactions with T cells were underscored when exogenous TGF-β was added in vitro to F5 CD8⁺ T cells prior to stimulation with the cognate peptide. Despite endogenous TGF-β production, pre-incubation of those CD8⁺ T cells with 0.1–5 ng/ml exogenous TGF-β significantly reduced the percentage of CD8⁺ T cells expressing the CD62L^high/CD44^high central memory phenotype from 72% to 30% (5 ng/ml of rh-TGF-β) (Figure 3A, upper panel) as well as CD62L^high/CD127^high central memory phenotype (55% vs 8%) (Figure 3A, lower panel). Commensurate with the phenotypic changes was a substantial reduction in the amount of IFN-γ by the peptide-stimulated CD8⁺ T cells, which has been reported in previous studies (Figure 3B, 11). Interestingly, the effect of exogenous TGF-β on total cell numbers was not dose-dependent (Figure 3C; bell shape curve). While the lowest dose of exogenous TGF-β (0.1 ng/ml) slightly reduced the number of CD62L^high/CD44^high central memory T cells (Figure 3C, solid bars), the number of T cells expressing the CD62L^low/CD44^high effector memory phenotype rose (Figure 3C, open bars), which was consistent with reduced apoptosis at the lowest dose of exogenous TGF-β (Figure 3D; U-shape curve). The lowest dose of exogenous TGF-β did not affect cell proliferation in the CFSE assay (data not shown). In contrast, higher doses of exogenous TGF-β preferentially reduced the number of CD62L^high/CD44^high central memory T cells (Figure 3A and 3C, solid bars) with an accompanying increase in Annexin V staining (Figure 3D).

Figure 3.

Exogenous TGF-β inhibits central memory CD8⁺ T cell differentiation. Splenocytes from F5 mice were pre-incubated with 0.1–5.0 ng/ml recombinant human TGF-β1 for 1 hr and then stimulated with 10⁻⁴ µg/ml cognate peptide for 72 hr. (A) Flow cytometric analysis of the phenotypic changes 72 hr after peptide stimulation. Cells were stained with anti-CD8a, CD62L and CD44 (upper panel in A) or CD127 (lower panel in A). Numbers in each quadrant denote percentage of cells. (B) IFN-γ production in the culture supernatant measured by ELISA. ***, p<0.001 (untreated vs. 0.1, 0.5, 1.0 and 5.0 ng/ml rhTGF-β). (C) Number of CD62L^high/CD44^high central memory (solid bars) versus CD62L^low/CD44^high effector memory (open bars) CD8⁺ T cells recovered following stimulation with cognate peptide. Total viable cells were determined by trypan-blue exclusion and the total number of CD8⁺, CD62L^low/CD44^high/CD8⁺, CD62L^high/CD44^high CD8⁺ cells were determined based on flow cytometry data. Data are the mean ± SEM of triplicate samples from three independent experiments. (D) Apoptotic cells, as determined by staining with anti-CD8a and Annexin-V antibodies, after exogenous TGF-β pre-incubation and peptide stimulation. **, p<0.01 (untreated vs. 1.0 ng/ml rhTGF-β). ***, p<0.001 (untreated vs. 0.1, 0.5 and 5.0 ng/ml rhTGF-β). Data are the mean ± SE of triplicate samples from three independent experiments. Statistical significance was measured by the one-way ANOVA followed by Tukey's multiple comparison test.

TGF-β blockade of central memory T cell differentiation is not mediated through SMAD or MAPK super family signaling pathways

SMAD-2, not SMAD-3, was phosphorylated 15–60 minutes after the in vitro addition of 5 ng/ml rh-TGF-β to isolated non-activated CD8⁺ T cells (Figure S2). SD208 addition suppressed the TGF-β-induced SMAD-2 phosphorylation in a dose-dependent manner, starting at 0.3 µM (Figure 4A). Other well-known SMAD-independent pathways such as Erk, p38MAPK and JNK (24) were not activated by exogenous TGF-β stimulation in non-activated CD8⁺ T cells and SD208 addition was inconsequential (Figure 4A). These results implicated SMAD2 signaling in central memory CD8⁺ T cell differentiation. Next, splenocytes from SMAD2 conditional knockout mice were stimulated with anti-CD3 and CD28 in the presence of various doses of the anti-TGF-β mAb. Contrary to our expectations, TGF-β blockade still increased the percentage of T cells expressing the CD62L^high/CD44^high central memory phenotype in both CD4⁺ and CD8⁺ T cells (Figure 4B). These observations suggested that even though SMAD2 is the signaling pathway for TGF-β, it does not seem to be involved in the differentiation pathway for central memory T cells. The possibility of crosstalk between TGF-β signal and mTORC1 pathway, which is known to be involved in central memory differentiation, was subsequently evaluated. TGF-β blockade had no effect on mTOC1 pathway (phosphorylation of p70S6 Kinase and S6 ribosome protein) of peptide activated CD8⁺ F5 T cells (Figure 4C), suggesting no crosstalk between TGF-β signaling and mTORC1 pathways. These findings suggest that the TGF-β signaling interruption that facilitates central memory T cell differentiation does not occur via the SMAD2, MAPK super family or mTOR signal transduction pathways, thus suggesting the involvement of a yet to be defined non-traditional pathway(s) that mediates those changes.

Figure 4.

TGF-β blockade of central memory T cell differentiation is not mediated through SMAD or MAPK super family signaling pathways. (A) CD8⁺ T cells were pre-incubated with SD208 (0.3–3.0 µM) for 1 hr, then treated with 5 ng/ml of recombinant human TGF-β1. Cells were harvested 1 hr later and the phosphorylation of the Erk, P38 MAPK and JNK signal transduction pathways was examined using Western immunoblots. (B) Splenocytes from SMAD2 conditional knockout (CKO) mice were pre-incubated with 1–10 µg/ml TGF-β mAb for 1 hr, stimulated with anti-CD3 and anti-CD28 for 72 hr and then stained with anti-CD8a, CD4, CD44 and CD62L antibodies and analyzed by flow cytometry. Numbers in the upper right quadrant denote percentage of cells. (C) Splenocytes from F5 mice were pre-incubated with 3 µM of SD208 for 1 hr, stimulated with cognate peptide, and p70 and S6 phosphorylation in the TORC1 pathway was examined by Western immunoblot at the indicated time points.

Change of transcriptional factors, eomes and T-bet, following TGF-β blockade

Investigators have reported that both Eomesoderm (Eomes) and T-bet (Tbx21) are two T-box containing transcriptional factors that are regulated by mTOR activity and consequently control effector and memory functional decisions in CD8⁺ T cells (25, 26). Even though the present data does not implicate mTOR signaling during TGF-β blockage contributing to central memory T cell differentiation, the fate of those transcriptional factors during this treatment was of interest. Blocking of TGF-β signaling with the addition of the anti-TGF-β mAb to CD8⁺ F5 T cells had no demonstrable effect on T-bet mRNA levels (Figure 5A), while Eomes mRNA expression, was significantly upregulated at 72 hr following peptide stimulation (0.0042 vehicle vs 0.0115 anti-TGF-β mAb; p<0.001) (Figure 5B). These changes led to an overall increase in the Eomes/T-bet mRNA ratio at 72 hrs. (1.05 vs 3.52; p<0.01) (Figure 5C). Messenger mRNA expression levels of other transcriptional factors, such as Tcf-7, Bmi-1, Bcl-6, Blimp-1 (Prdm1) (25, 27), were unchanged following the addition of the anti-TGF-β mAb (Supplemental Figure 3). When Eomes and T-bet protein levels were measured by intracellular FACS-based staining, the addition of the anti-TGF-β antibody to F5 CD8⁺ T cells had a dose-dependent increase in Eomes (38.1% vs 48.9% at 10 µg/ml TGF-β mAb), while T-bet levels were decreased (43.1% vs 30.4% at 10 µg/ml TGF-β mAb) (Figure 5D). Those changes resulted in an increased Eomes/T-bet protein ratio (Figure 5E) which is consistent with differentiation of memory T cells (25).

Figure 5.

Change of transcriptional factors T-bet and Eomes following TGF-β blockade. Isolated splenic CD8⁺ T cells from F5 mice were pre-incubated with 1 µg/ml of TGF-β mAb for 1 hr and then stimulated with 10⁻⁴ µg/ml of cognate peptide (NP68 peptide), 1.0 µg/ml of H2Db-dimer X and 2.0 µg/ml of anti-CD28. (A) T-bet mRNA levels, (B) Eomes *, p<0.05 (untreated; open circle vs. 1.0 µg/ml anti-TGF-β MAb; filled circle @ 72 hrs) (C) Eomes/ T-bet mRNA calculated ratios. ***, p<0.001 (untreated vs. 1.0 µg/ml anti-TGF-β mAb @ 48 and 72 hrs). (D) T-bet and Eomes protein expressions measured by intracellular FACS and calculated ratio at 96 hr post-peptide. (E) Eomes/ T-bet protein ratios calculated ratios. **, p<0.01, ***, p<0.001 (untreated vs. 0.1, 1 and 10 µg/ml anti-TGF-β mAb) Statistical significance was determined using the two-tailed unpaired Student’s t-test (A, B and C) or the one-way ANOVA followed by Tukey's multiple comparison test (E).

Increased CD8⁺ central memory phenotype alters in vivo proliferation upon recall response

The functional characteristics of SD208-treated CD8⁺ T cells that had an increased central memory phenotype were evaluated using an adoptive T cell transfer protocol. As before, splenic CD8⁺ T cells from F5 mice were stimulated in vitro in the presence of cognate peptide alone or combined with 3 µM of SD208. After 96 hr, consistent with previous results (Figure 2A), 62% of the untreated CD8⁺ T cells expressed CD62L^high/CD44^high central memory markers, while a higher percentage (89%) of SD208-treated T cells expressed the central memory phenotype (Figure 6A). At that time, T cells were CFSE labeled and adoptively transferred into naïve B6 mice. Three days post-transfer, the distribution of the transferred cells in the peripheral blood and spleen were similar (Figure 6B, left and center graph), while a higher percentage of SD208-treated T cells were found in the lymph nodes (3.4% vs 2.6%; p<0.0016) (Figure 6B, right graph). Three days after adoptive transfer, the transferred cells had also maintained their respective CD62L^high/CD44^high central memory phenotype (60% vs 84%, untreated vs SD208 treatment) (Figure 6C) without any sign of proliferation (Figure 6D). At that time the mice were vaccinated with rF-NP68-TRICOM, the cognate peptide engineered in a recombinant avipox vector, to evaluate the ability of the transferred T cells to mount a peptide-specific recall response. Three days post-vaccination, mice that were adoptively transferred with the SD208-treated T cells had higher numbers of NP-68 specific CD8⁺ T cells (1.73%, untreated vs. 6.31%, SD208-treatment, p<0.05) (Figure 6E and F). The NP-68-tetramer⁺, SD208-treated adoptively transferred CD8+ T cells had a higher proliferative index as determined by changes in the CSFE dye dilution assay than the untreated cells (Figure 6G, p<0.01), a characteristic of central memory T cells. A low frequency of NP-68 specific CD8⁺ T cells (0.28%) was found after rF-NP68-TRICOM vaccination of naïve, non-adoptively transplanted mice (Figure 6E). These data collectively suggest that SD208-treated cells showed increased central memory T cells phenotypically were capable of eliciting a strong recall response following encounter in vivo with the cognate peptide.

Figure 6.

Increased central memory phenotype alters in vivo proliferation upon recall response. Splenocytes from F5 mice were pre-incubated with 3 µM of SD208 or vehicle for 1 hr, stimulated with 10⁻⁴ µg/ml of cognate peptide for 96 hr and then stained with anti-CD8, CD62L and CD44. (A) CD62L and CD44 expression on CD8 T cells 96 hr after in vitro stimulation. (B) In vivo distribution of adoptively transferred F5 memory CD8⁺ T cells after 3 days and prior to rF-NP68-TRICOM vaccination. Mice (3/group) were euthanized and the peripheral blood, spleen and inguinal lymph nodes were analyzed for NP68 dextramer staining. Each dot represents a single mouse and the horizontal line indicates the mean. NS: no statistical significance. (C) A representative FACS plot for each group. The numbers in the top right quadrant denote the percentage of CD62L^high/CD44^high cells among CD8⁺/dextramer⁺ T cells from splenocyte. (D) A representative CFSE dilution for each group of CD8⁺/dextramer⁺ T cells from splenocyte. (E–G) Recall response 3 days after rF-NP68-TRICOM challenge. Peripheral blood from rF-NP68-TRICOM vaccinated mice (n=3/group) (no adoptive transfer, adoptive transfer with vehicle treated-cells and adoptive transfer with SD208-treated-cells) was collected and analyzed for NP68 dextramer staining. (E) A representative FACS plot for each group. The numbers in the top right quadrant denote the percentage of dextramer⁺ cells among CD8⁺ T cells. (F) Each dot represents a single mouse and the horizontal line indicates the mean. (G) A representative CFSE dye dilution for each group of CD8⁺/dextramer⁺ T cells from a single experiment. Three separate experiments were carried out and the increased proliferation in the SD208-treated T cells was statistically significant (p<0.01) as measured using the one-way ANOVA followed by Tukey's multiple comparison test.

Increased central memory T cells in human PBMCs following TGF-β blockade

As previously stated, like their murine counterparts, isolated human T cells also produce TGF-β following in vitro stimulation (16). Therefore, it was of interest whether TGF-β blockage resulted in similar changes in cellular differentiation in human CD8⁺ T cells. To address that question, CD45RO⁻/CD8⁺ T cells that expressed CD45RA (98%), CCR7 (85%) and CD62L (70%) were isolated from human peripheral blood mononuclear cells (PBMCs) (Figure S4). CD45RO was chosen because of its expression profile on activated or a memory human T cells is similar to that of CD44 on mouse T cells. Six days after in vitro stimulation in the presence of anti-CD3, approximately 90% of the isolated CD8⁺ T cells had CD45RO expression, suggesting memory differentiation (Figure S4B). Furthermore, 45% of the CD45RO⁺ cells expressed high levels of CD62L (CD62L^high), indicative of central memory cells. SD208 addition to the PBMC culture media resulted in a dose-dependent increase in the percentage of cells expressing that phenotype in CD8⁺ T cells (i.e., 61%: 3 µM of SD208, Figure 7A) as well as an increase in the MFI (i.e., 9,440 with vehicle vs 15,800 with 3 µM of SD208). The acquisition of the phenotypic markers indicative of memory T cell differentiation occurred in the absence of any measurable changes in viable cell number as determined by trypan blue (Figure 7B) and was CD8⁺ T cell specific (i.e., no commensurate change in CD4⁺ T cells; data not shown). As a result, there was an overall increase in the absolute number of central memory T cells (Figure 7B, solid bar) in CD8⁺ T cells and, similar to the murine cells, a dose-dependent decrease in the number of effector memory T cells (Figure 7B, open bars). Blockade of endogenous TGF-β slightly increased IFN-γ production, an indicator of CD8⁺ T cell activation (Figure 7C). These findings indicate that increased central memory cell generation by TGF-β blockade is a characteristic shared by both murine and human CD8⁺ T cells. Differential changes in IFN-γ production in the murine and human T cells may be explained by the use of different modalities to inhibit TGF-β signaling: the anti-TGF-β antibody for mouse T cells and SD208, a TGF-βR1 kinase inhibitor, for human T cells.

Figure 7.

Increased central memory T cells in human PBMCs following TGF-β blockade. (A) CD62L expression on gated CD8⁺/CD45RO⁺ cells 6 days after anti-CD3 stimulation. Numbers in the upper and lower quadrants denote percentage of CD62L^high and MFI of CD62L expression on CD8⁺ T cells, respectively. (B) Total number of CD62L^high/CD45RO⁺ central memory (solid bars) versus CD62L^low/CD45RO⁺ effector memory (open bars) CD8⁺ T cells recovered following stimulation with anti-human CD3. The number of viable cells was determined by trypan-blue exclusion and the total number of CD8⁺, CD62L^low/CD45RO⁺, CD62L^high/ CD45RO⁺ cells was determined based on flow cytometry data. *, p<0.05 (untreated vs. 0.3 µM SD208); **, p<0.01 (untreated vs. 1.0 µM SD208); ***, p<0.001 (untreated vs. 3.0 µM SD208). (C) IFN-γ production in the culture supernatant as measured by ELISA. Data are the mean ± SE of triplicate samples of at least three independent experiments. Statistical significance was measured using the one-way ANOVA followed by Tukey's multiple comparison test.

Discussion

In the present study, CD8⁺ splenic T cells from H-2Db-restricted NP68-specific TCR transgenic mice (F5 mice) were used as an in vitro model to study TGF-β modulation of intrinsic T cell metabolic pathways and their role(s) in immune T cell differentiation. Upon stimulation with cognate peptide, the CD8⁺ F5 T cells acquire both phenotypic changes and immune effector functions that are reminiscent of the in vivo phases described for an antigen-specific primary immune response, i.e., expansion, contraction, and memory. During the initial 72 hr after cognate peptide recognition, CD8⁺ F5 T cells enter an expansion phase characterized by CD44 acquisition that distinguishes effector and memory T cells from their naïve counterparts. This phenotypic change was immediately followed by CD62L^high expression, indicating a shift to central memory T cell differentiation. During that same time interval, there was an early (2–4 hr post-peptide recognition) spike in TGF-β mRNA prior to a significant increase in TGF-β production commensurate with IL-2 and IFN-γ (Figure 1). Endogenous TGF-β production by isolated human CD8⁺ T cells stimulated in vitro with phytohemaggutinin (PHA) has been reported, indicating a shared characteristic within the T cell compartments of the two species (16). The present study also showed that blockage of TGF-β actions using either a blocking antibody or a TGF-β RI kinase inhibitor increased CD62L^high central memory cells independent of changes in either proliferation or cytokine secretion (Figure 2). Whereas in prior studies the addition of IL-15 during the expansion and memory phase induced CD62L^high central memory T cell differentiation (8), the present findings offer a new mechanism by which T cells control their differentiation through changes in the autocrine secretion of TGF-β.

CD62L plays an essential role as a homing receptor to facilitate central memory T cells’ entry into secondary lymphoid or inflamed tissues via high endothelial venules (28). Therefore, its loss could negatively impact recruitment of memory cells to appropriate immune tissues during the generation of an immune response. CD62L expression is regulated by TACE/Adam17-mediated shedding and transcriptionally by Klf2 (6, 29, 30). In the present study, the presence of TGF-β either by endogenous production or exogenous addition reduced the number of CD62L^high central memory cells. Those changes occurred with dramatic, paradoxical effects on T cell proliferation, apoptosis and IFN-γ production. Exogenous TGF-β addition also significantly reduced CD62L expression, but with an accompanying increase in apoptosis and loss of IFN-γ production. Loss of CD62L expression on resting memory cells began approximately 24 hr post-TGF-β addition (data not shown), suggesting that the effect might be due to changes in transcriptional regulation by Klf2, not shedding by TACE activation. Additional study is needed to address those and other possible explanations.

It was of subsequent interest to examine those intrinsic cellular pathways through which blockade of endogenously produced TGF-β enhanced immunologic memory. Recently, it was reported that inhibition of signal transduction through PI(3)K or mTOR induces CD62L^high central memory CD8⁺ T cells (6, 7). Our data indicated that the signaling pathway(s) that regulates central memory differentiation by TGF-β seems to be non-canonical, independent of SMAD, MAPK super family as well as the mTOR pathway (Figure 5). It is intriguing to contrast the effects of TGF-β with those of IL-2 on memory T cell differentiation. However, the differences seem to reside in IL-2 mediating its changes via the Akt-mTOR signaling pathway, whereas TGF-β seems to signal through another pathway, not Akt-mTOR (Figure 6). So it leads one to speculate what signals constitute whether the transition to memory T cells proceeds independent of SMAD, MAPK and mTOR or via the Akt-mTOR pathway. Several differences, including cell type (T cell versus tumor cell) and the characteristics and context within which each cell type encounters TGF-β, might govern the selection of a particular signaling pathway. Within the first 24–36 hrs after recognition of its cognate peptide, endogenous TGF-β production by the CD8⁺ T cells becomes measurable which continues through the activation phase a time-dependent fashion. Interruption of the TGF-β signaling pathway either by normal means during the T cell contraction phase or by blockage of TGF-β signaling leads to memory T cell differentiation. On the other hand, when the CD8⁺ T cells encounter a bolus amount of TGF-β as an external signal prior to entering the activation phase, perhaps these circumstances enact a completely different set of signaling. mTOR becomes activate when it encounters such an external signal and since this is such a strong signal, it requires a specific mTOR inhibitor, such as rapamycin, to inhibit the pathway leading to the transition to memory T cells. Contrary to the change of Eomes and T-Bet, other transcriptional factors that have been related to the central memory differentiation such as Tcf7, Bmi-1 and Bcl-6 (25–27) remained unchanged following TGF-β blockade (Figure S3). Whether there is a common molecule in Eomes/T-bet expression that is shared by TGF-β and IL-2 or it is controlled by an independent mechanism should be a focus for future study.

Subsequent functional analyses of the F5-TCR.Tg CD62L^high memory CD8⁺ cells became critical to determine whether the observed phenotypic changes truly represented central memory T cells. CD62L^high memory CD8⁺ cells were generated in vitro and following their adoptive transfer into naïve mice were found to preferentially migrate to secondary lymph nodes (Figure 6B). Those findings were in agreement with a previous report using LCMV gp33-specific CD62L^high memory CD8⁺ T cells (3). Migration to the secondary lymph nodes by the F5-TCR.Tg CD62L^high memory CD8⁺ cells seemed to occur passively due to the absence of any measurable change in T cell phenotype or proliferation. Upon re-exposure to the antigen via the rF-NP34-TRICOM vaccine, there was a more robust recall response as measured by a higher number of Ag-specific CD8⁺ T cells in mice that were adoptively transferred with SD208-treated F5-TCR.Tg CD62L^high memory CD8⁺ cells (Figure 6E–G). These results also agree with previous reports that adoptively transferred CD62L^high memory CD8⁺ T cells possessed superior proliferative capacity, augmented viral clearance and were more protective against tumor challenge (1, 3). Thus, CD62L^high cells generated in vitro by TGF-β blockade of F5-TCR.Tg memory CD8⁺ cells exhibit both phenotypic and functional characteristics of central memory T cells capable of fortifying the effectiveness of a vaccine. That hypothesis requires additional study since human CD8⁺ T cells seem to have a similar response to blockage of TGF-β action, an increase in central memory differentiation.

Recent reports have shown that the energy sensitive kinase mammalian target of rapamycin (mTOR) responds to extrinsic factors that impact cellular metabolic states (ATP-AMP) and can alter differentiation (7, 10). The present results indicate that intrinsically the ongoing autocrine TGF-β production seems to counteract CD62L expression and differentiation of CD8⁺ T cell memory. These observations raise the question - why? As stated previously, CD62L allows memory T cells to extravasate to lymph nodes and become a resident of the memory pool. It has been suggested that a balance exists between the sizes of the memory T cell pools and necessary space for the deposition of future pathogen-specific memory cells. One possible explanation is that endogenous TGF-β production may play a homeostatic role in establishing the size of memory pool. If that indeed is the case then targeting TGF-β signaling may be a viable approach when one needs to enhance antigen-specific CD8⁺ T cell memory against a lethal infection or cancer.