Cutting Edge: An Alternative Pathway of CD4⁺ T Cell Differentiation Is Induced Following Activation in the Absence of γ-Chain-Dependent Cytokine Signals¹

In addition to conventional Th1 and Th2 effector cells, alternate CD4⁺ T cell subsets have been the recent focus of intense study. Of particular interest are regulatory T cell subsets that play a role in peripheral tolerance (1, 2). Whether CD4⁺ T cells develop helper or regulatory effector function following stimulation depends on many factors, including cytokines. Consequently, many well-characterized cytokine receptor signaling pathways have been targeted for the development of inhibitors that could be used to modulate immune responses. A primary target of these efforts is the Janus family tyrosine kinase Jak3 (3). Jak3 is exclusively expressed in hemopoietic cells and is essential for signaling through cytokine receptors that use the common γ-chain (γc)³ (i.e., IL-2R, IL-4R, IL-7R, IL-9R, IL-15R, and IL-21R). As such, Jak3 plays a crucial role in immune system development and function, a fact that is underscored by the finding that deficiencies in Jak3 are associated with human SCID conditions that are indistinguishable from deficiencies in γc (4).

Mice deficient in Jak3 have several immunological abnormalities, including a block in B, NK, and γδ T cell development (5). In these mice, the cellularity of the thymus is greatly reduced because of a deficit in the number of thymic progenitor cells. Despite this reduced thymocyte cellularity, αβ T cell development progresses normally in Jak3^−/− mice, and adult Jak3-deficient mice accumulate normal numbers of peripheral T cells. Interestingly, these T cells are predominantly CD4⁺ and have a surface phenotype resembling that of memory T cells. Functionally, Jak3^−/− T cells fail to proliferate or secrete IL-2 after in vitro stimulation and therefore appear anergic. Consistent with these data, one previous study showed that T cell anergy induced by TCR stimulation in the absence of costimulation could be prevented by the addition of IL-2, IL-4, or IL-7; furthermore, the prevention of anergy was specifically dependent on the phosphorylation of Jak3 (6). Therefore, we were surprised to find that Jak3^−/− CD4⁺ T cells are proliferating in vivo (7), and, rather than being globally unresponsive to TCR stimulation, these cells are capable of producing transcripts for cytokines other than IL-2 (5).

To pursue these findings, we investigated the underlying molecular profile of Jak3^−/− CD4⁺ T cells and determined the overall functional consequences of CD4⁺ T cell activation in the absence of all γc-dependent cytokine signals. Microarray analysis (S. R. Mayack and L. J. Berg, manuscript in preparation) revealed that Jak3^−/− CD4⁺CD44^high T cells, as compared with Jak3^+/− CD4⁺CD44^high cells, express increased levels of programmed death-1 (PD-1) and lymphocyte activation gene-3 (LAG-3) mRNA following ex vivo isolation, indicating an overall increase in inhibitory cell surface receptor gene expression. Jak3^−/− CD4⁺CD44^high T cells also had increased levels of mRNA for the immunomodulatory cytokines IFN-γ, IL-10, and TGF-β. These findings prompted us to examine whether Jak3^−/− CD4⁺ T cells have functional similarities to regulatory T cells and, further, whether this phenotype is spontaneously acquired by naive Jak3^−/− T cells following activation in vivo.

Jak3^−/− (B6; 129S4-Jak3^tm1Ljb/J) and Jak3^+/− mice (8) have been back-crossed to C57BL/6 for more than eight generations. All mice were 8–10 wk of age and maintained in a specific pathogen-free facility following review and approval by the Institutional Animal Care and Use Committee. RAG2^−/− (B6. 129S6-Rag2^tm1FwaN12) mice were used as recipients for adoptive transfer (AT) experiments.

CD4⁺ T cells were purified on AutoMACS LS⁺ columns (Miltenyi Biotec). CD4⁺CD44^high T cells were purified by sorting on a BD Biosciences FACStar. CD4⁺ single positive (SP) thymocytes were enriched by depleting CD8⁺ cells. CD4⁺CD44^high peripheral T cells were routinely >95% pure, and CD4⁺ thymocytes were ∼70% pure.

Cells were stained with the indicated Abs and analyzed on a BD Biosciences FACSCalibur. Data were analyzed using CellQuest software (BD Immunocytometry Systems). The Abs and flow cytometry reagents used were α-CD4-FITC and α-CD44-CyChrome (BD Pharmingen) and α-PD-1-PE, α-Armenian hamster IgG-PE, α-rat IgG2a, λ-PE, α-mouse IFN-γ, and α-mouse IL-10 (eBioscience).

RNA and cDNA were prepared as described previously (9). Real-time quantitative PCR was performed on an i-Cycler (Bio-Rad). Primer sequences are available upon request.

A total of 1 × 10⁶ CD4 SP thymocytes purified from Jak3^+/− or Jak3^−/− mice were injected i.v. into RAG2^−/− mice. At 2, 4, 6, and 8 wk following AT, splenocytes from injected or uninjected control mice were harvested and analyzed.

A total of 5 × 10⁵ purified CD4⁺ T cells were plated with 10 μg/ml immobilized α-CD3 Ab (clone 145–2C11; BD Pharmingen). For coculture experiments, target cells were mitomycin C-treated and stimulated at a 1:1 target/responder ratio at a total cell density of 5 × 10⁵ per 0.2 ml. Proliferation was measured by [³H]thymidine (PerkinElmer) incorporation. For cytokine analysis, supernatants were assayed at 18–22 h for the presence of IL-2, IFN-γ, IL-4, IL-5, and IL-10 (BD Pharmingen) and TGF-β (Emax immunoassay system from Promega) by ELISA.

Cell lysates were prepared as described previously (10) and probed with Abs to FOXP3 (BioLegend) and ERK1/2 (BD Biosciences).

T regulatory cells (Tregs) have been classified into two basic categories. “Natural” Treg cells are those that originate in the thymus, constitutively express CD25, and have a contact-dependent, cytokine-independent mechanism of suppression that is regulated by FOXP3 expression. These natural Treg cells are nonproliferative and produce IL-10 and TGF-β following stimulation (1).

A second category of regulatory T cells are “adaptive” Tregs that arise from activated peripheral T cells (2). Adaptive Treg cells are also nonproliferative in vitro and function via both cytokine- and cell contact-dependent mechanisms. T regulatory type-1 (Tr1) cells are adaptive Tregs that produce high levels of IL-10 and varying amounts of IFN-γ and TGF-β. Tr1 cells do not express FOXP3 and are instead induced by activation in the presence of immunosuppressive factors such as IL-10, vitamin D3 dexamethasone, or tolerogenic dendritic cells.

As a first step to addressing whether Jak3^−/− T cells had differentiated into T cells with regulatory characteristics, we examined the expression of FOXP3, a transcription factor strongly associated with Treg activity (1). FOXP3 has been detected only in CD25⁺ natural Tregs and is not expressed by the adaptive Tr1 subset of Tregs (11). For this experiment, CD4⁺ T cells were isolated from either Jak3^−/− mice or Jak3^+/− controls, and FOXP3 protein levels were determined by immunoblotting. As shown in Fig. 1,A, FOXP3 is undetectable in Jak3^−/− CD4⁺ T cells but is easily detected in Jak3^+/− CD4⁺ T cells. Because FOXP3 expression is generally associated with CD4⁺CD25⁺ T cells, this prompted us to investigate whether this subset of cells is present in Jak3^−/− mice. Typically, CD4⁺CD25⁺ cells constitute ∼5% of total splenocytes (1), as can be seen in the Jak3^+/− controls (Fig. 1 B). In contrast, only 0.2% of splenocytes are CD4⁺CD25⁺ in Jak3^−/− mice. The virtual absence of the CD4⁺CD25⁺ T cell subset in Jak3^−/− mice is consistent with studies showing that IL-2R signaling is required for the development of natural CD4⁺CD25⁺ T cells in the thymus (1).

PD-1 and LAG-3 expression have been associated with anergic and IL-10-positive regulatory T cell subsets and Treg activity (12, 13, 14). To confirm the differential expression of PD-1 and LAG–3 between Jak3^+/− and Jak3^−/− T cells, we examined PD-1 and LAG-3 mRNA levels by real-time quantitative PCR (Fig. 2,A) and protein levels by flow cytometry (Fig. 2 B). Together, these findings indicate that Jak3^−/− CD4⁺ T cells have characteristics associated with adaptive regulatory T cells that arise in the peripheral immune organs and are consistent with our previous data demonstrating that Jak3^−/− T cells acquire their functional defects following emigration from the thymus into the periphery (15).

Adaptive Treg cells typically have a cytokine profile distinct from those of Th0, Th1, or Th2 effector cells (16). For example, Tr1 cells produce IL-10 and IFN-γ, with or without TGF-β, and some IL-5, but little or no IL-2 and IL-4 (17). Cytokines such as IL-10, IFN–γ, and TGF-β promote immune suppression and participate in the suppressive activity mediated by subsets of regulatory T cells in vivo (17, 18, 19). To determine the pattern of cytokines secreted by Jak3^−/− T cells following stimulation, CD4⁺CD44^high T cells from Jak3^+/− or Jak3^−/− mice were stimulated in vitro, and supernatants were collected and assayed for IL-2, IL-4, IL-5, IL-10, IFN-γ, and TGF-β by ELISA. Fig. 3 shows that Jak3^−/− T cells secrete substantially higher levels of IL-10, IFN-γ, and TGF-β compared with Jak3^+/− controls, but no detectable IL-2 and low levels of IL-5 and IL-4. These findings demonstrate that Jak3^−/− CD4⁺CD44^high T cells are not conventional Th1 or Th2 cells but instead secrete a panel of cytokines consistent with their differentiation into a subset of regulatory CD4⁺ T cells.

The hallmark attribute of a regulatory T cell is the ability to suppress naive T cell responses. To test whether Jak3^−/− T cells could suppress the proliferation of wild-type CD4⁺ T cells in a coculture assay, total CD4⁺ T cells were isolated from either Jak3^+/− or Jak3^−/− mice and stimulated with a plate-bound α-CD3 Ab. At 24 or 48 h of culture, equivalent numbers of mitomycin C-treated Jak3^+/− or Jak3^−/− CD4⁺ T cells were added to the responder cells, and proliferation was assessed on day 3. In addition, Jak3^+/− or Jak3^−/− CD4⁺ T cells were stimulated separately as controls. Mitomycin C treatment was included to ensure that the proliferation being measured was only that of the responding T cell population. Experiments performed in the absence of mitomycin C treatment gave comparable results (data not shown).

When Jak3^−/− T cells were added to Jak3^+/− responder T cells at the initiation of culture or 24 h later, there was no apparent effect on responder T cell proliferation (Fig. 3, and data not shown). However, when Jak3^−/− T cells were added on day 2 of culture, Jak3^+/− T cell proliferation was inhibited ∼50%. The fact that Jak3^−/− T cells rapidly die following in vitro culture (7) may account for the weak suppressive activity of these cells and their inability to inhibit when they are added at earlier days of culture. Finally, the inhibitory function of Jak3^−/− T cells is not mediated solely by PD-1 or IL-10, because the addition of blocking Abs directed against these molecules had no effect on the suppressive activity of Jak3^−/− T cells (data not shown). Neutralizing IFN-γ resulted in a very modest inhibition in the ability of Jak3^−/− CD4⁺ T cells to suppress the proliferation of control CD4⁺ T cells, which was enhanced when blocking Abs to both IFN-γ and IL-10 were added (data not shown). These results suggest that the suppressor activity of Jak3^−/− CD4⁺ T cells is complex and may be mediated by a combination of cytokines plus additional cell contact-dependent or soluble components. Interestingly, we also observed that the simultaneous neutralization of IFN-γ and IL-10 led to a mild but detectable enhancement of the proliferative response of Jak3^−/− CD4⁺ T cells (data not shown), suggesting that the inhibitory cytokines produced by these cells may contribute to their poor in vitro proliferative responses.

To determine whether naive Jak3^−/− CD4⁺ T cells would spontaneously acquire Treg characteristics following activation in vivo, we performed AT studies using Jak3^−/− CD4⁺ thymocytes. These cells are developmentally normal, able to proliferate normally in response to antigenic stimulation in vivo, and do not secrete IL-10 or exhibit increased expression of PD-1 (Fig. 4, B and C).

To promote T cell activation and proliferation, CFSE-labeled Jak3^+/− or Jak3^−/− thymocytes were adoptively transferred into Rag2^−/− mice. At 2, 4, 6, and 8 wk post-AT, splenocytes were harvested from Rag2^−/− recipient mice that had received Jak3^+/− (Jak3^+/− AT) or Jak3^−/− (Jak3^−/− AT) thymocytes. At all time points the loss of CFSE was assessed, and absolute T cell numbers were calculated. Over the first four weeks following AT, the population of Jak3^−/− T cells expanded comparably to that of wild-type T cells in Rag2^−/− hosts (Fig. 4,A, and data not shown), demonstrating that homeostatic proliferation of CD4⁺ T cells is not dependent on signaling via γc-dependent cytokine receptors. At later time points of 6–8 wk, Jak3^−/− T cell numbers began to decline in the AT recipients, strongly suggesting impairment in the long-term survival of cells lacking all γc-dependent cytokine signals (Fig. 4 A). This latter notion is consistent with previous studies suggesting that γc-dependent cytokines such as IL-7 play a more critical role in T cell survival than proliferation (20, 21).

To assess cytokine production, purified splenocytes from AT recipients were stimulated in vitro, and IL-2 and IL-10 production were determined by ELISA and intracellular cytokine staining (Fig. 4,B, and data not shown). Surprisingly, when CD4⁺ T cells were stimulated two weeks post-AT, comparable IL-2 secretion was observed from both Jak3^+/−AT and Jak3^−/−AT T cells (Fig. 4,B). This finding is in stark contrast to the undetectable levels of IL-2 production typically seen from peripheral Jak3^−/− CD4⁺ T cells isolated from adult Jak3^−/− mice and stimulated immediately ex vivo (Fig. 3). At 4 and 6 wk post AT, IL-2 production by in vitro stimulated Jak3^−/−AT T cells declines (Fig. 4B), and it is undetectable by intracellular cytokine staining at 8 wk post-AT (data not shown). Furthermore, in contrast with freshly isolated Jak3^−/− thymocytes, which produce no IL-10, Jak3^−/−AT cells produce IL-10 in response to in vitro stimulation at all of the time points analyzed (Fig. 4,B). In addition, the levels of IL-10 produced by Jak3^−/−AT cells increase over time, from 100 pg/ml at two weeks post-AT to 600 pg/ml by 6 wk post-AT (Fig. 4B). By 8 wk posttransfer, Jak3^−/− AT CD4⁺ T cells also secrete increased levels of IFN-γ and TGF- β similar to that seen for freshly isolated peripheral Jak3^−/− CD4⁺ T cells (data not shown).

We also examined PD-1 expression on the adoptively transferred T cells. At 4 wk posttransfer, PD-1 expression is increased on both Jak3^+/− AT and Jak3^−/− AT CD4⁺ T cells. However, by 8 wk PD-1 expression has declined on Jak3^+/− AT CD4⁺ T cells but remains at high levels on Jak3^−/−AT CD4⁺ T cells (Fig. 4 C). These data demonstrate that the differentiation of naive Jak3^−/− CD4⁺ T cells into cells that produce immunosuppressive cytokines and express high levels of PD-1 is a T cell-intrinsic property.

In this study, we show that Jak3^−/− CD4⁺ T cells express high levels of PD-1 and LAG-3, secrete a Tr1-type cytokine profile following direct ex vivo stimulation, and modestly suppress the proliferation of wild-type T cells in vitro. We also show that these traits are spontaneously acquired by naive Jak3^−/− CD4⁺ cells as a result of their activation and proliferation in vivo. In the absence of Jak3-dependent signaling, T cells differentiate down an alternative pathway, leading to the establishment of an immunosuppressive phenotype. We conclude that CD4⁺ Jak3^−/− T cells share properties with regulatory T cells that have an important role in peripheral tolerance and the prevention of autoimmunity.

The finding that Jak3-mediated cytokine receptor signals are distinct in effector vs regulatory T cell populations is consistent with previous studies (Ref.22 and www.immuno2004.org/onlineabstracts/5098.html). This earlier work showed that several molecules directly downstream of Jak3/γc signaling, including STAT5 and PI3K/Akt (22), have reduced activity in CD4⁺CD25⁺ regulatory T cells and, in addition, that stimulated regulatory T cells exhibit decreased phosphorylation and down-regulated expression levels of Jak1, Jak3, STAT5a, and STAT5b (www.immuno2004.org/onlineabstracts/5098.html). Recent efforts to generate pharmacological Jak3 inhibitors have yielded compounds with efficacy in blocking graft rejection in both murine and primate transplantation models (23). Our data suggest the intriguing possibility that, with the appropriate timing, inhibition of Jak3 might also be capable of inducing long-term Ag-specific tolerance. Overall, these findings implicate Jak3 as a central component of signaling pathways that regulate T cell differentiation into effector vs regulatory lineages.

We thank Min Shi, Yoko Kosaka, and Sara Gazalo for critical reading of the manuscript, Amanda Prince, Joseph Masciaszek, Hai Nguyen, and Sung-Kwon Kim for technical assistance, and Joonsoo Kang and Cynthia Chambers for helpful discussions.

The authors have no financial conflict of interest.