An early reaction of CD4+ T lymphocytes to Ag is the production of cytokines, notably IL-2. To detect cytokine-dependent responses, naive Ag-specific T cells were stimulated in vivo and the presence of phosphorylated STAT5 molecules was used to identify the cell populations responding to IL-2. Within hours of T cell priming, IL-2-dependent STAT5 phosphorylation occurred primarily in Foxp3+ regulatory T cells. In contrast, the Ag-specific T cells received STAT5 signals only after repeated Ag exposure or memory differentiation. Regulatory T cells receiving IL-2 signals proliferated and developed enhanced suppressive activity. These results indicate that one of the earliest events in a T cell response is the activation of endogenous regulatory cells, potentially to prevent autoimmunity.
Immune responses to foreign Ags have to be precisely regulated to prevent inadvertent activation of self-reactive lymphocytes or collateral damage to normal tissues. Regulatory T cells (Tregs)5 are one of the major mechanisms that limit responses to self- and foreign Ags (1, 2, 3, 4). The conventional view is that most Tregs are generated by self-Ag recognition in the thymus and work in peripheral tissues to prevent reactions against these self-Ags (5, 6). Some Tregs may be generated in peripheral tissues as immune responses develop, thus stopping these responses from becoming harmful (7). Although the Treg population as a whole appears chronically activated, expressing activation markers such as CD25 and CTLA-4 (1), a fraction can be identified that is in a more proliferative state, presumably responding to tissue self-Ags (8). Tregs that are controlling T cell responses to microbial infections may be activated by self-Ags as well, released by tissue damage or by direct recognition of microbial Ags (9). Identifying when and how Tregs are activated during the course of a physiological immune response is of great interest to better understand the way immune responses develop and are controlled. Given the well-established capacity of Tregs to limit unrelated Ag responses through bystander suppression (10), we hypothesized that communication between Ag-activated T cells and Tregs of different specificity is required to control the initiated response and limit potential damaging side effects. One interesting candidate to deliver a signal from Ag-activated T cells to Tregs is IL-2, which is produced early after T cell activation and is known to be essential for the generation and maintenance of Tregs. The functions of IL-2 during a T cell response are seemingly antagonistic, since the cytokine not only supports the development of effector and memory cells (11, 12, 13) but is also critical for the maintenance of Foxp3+ Tregs (14, 15). Since Tregs do not produce IL-2, it is likely that they receive IL-2 signals from other T cells that are responding to Ag. To address the question of what the cellular targets of IL-2 are during a conventional immune response, we measured IL-2-dependent biochemical signals in different cell populations in vivo. Surprisingly, we found that Tregs, and not the actual IL-2-producing cells, are the first responders to IL-2, leading to their activation and proliferation.
Materials and Methods
BALB/c mice were purchased from Charles River Laboratories and used at 6–8 wk of age. DO.11.10 transgenic mice were a gift from K. Murphy (Washington University, St. Louis, MO) and were crossed onto Rag2−/− mice. DO11.10 × Il-2−/− mice were generated in our laboratory by crossing DO11.10 mice with Il-2−/− mice, all on the Rag−/− background. Foxp3-GFP mice (16) were backcrossed 10 generations to BALB/c. All mice were bred and maintained in accordance with the guidelines of the Laboratory Animal Resource Center of the University of California, San Francisco and approved by the Stanford Administrative Panel on Laboratory Animal Care/Institutional Animal Care and Use Committee protocols.
Adoptive transfer and immunization
Adoptive transfer of Il-2+/+ and Il-2−/− DO11.10 Rag−/− T cells was performed as previously described (11, 17). DO11 Rag−/− TCR transgenic T cells were adoptively transferred into wild-type BALB/c mice and then immunized with 100 μg of OVA peptide with or without LPS, or with mature OVA-loaded bone marrow-derived dendritic cells (BMDCs). Serum-free medium (Cellgro/Mediatech) was used for the last 48–72 h of culture, since BMDCs loaded with FCS-derived peptides were capable of stimulating a nonspecific T cell response (data not shown). In some experiments, mice were initially exposed to Ag and then restimulated 60 h or 5 wk after initial Ag priming.
Preparation of splenic single-cell suspensions
Following stimulation in vivo, the spleens of recipient mice were harvested as previously described (18). In brief, mice were killed at various time points by a brief (30 s) CO2 administration followed by cervical dislocation. Spleens were excised and immediately dissociated into a 10-ml PBS solution containing 1.6% paraformaldehyde (Electron Microscopy Sciences). Following a 15-min fixation period, cell suspensions were transferred through a 70-μm pore size mesh and ice-cold methanol was directly added to a final concentration of 80% methanol.
Staining and flow cytometry
Normalization to compare intracellular stains of naive and activated T cells
Twenty-four hours following 100-μg OVA peptide stimulation, DO11 T cells enter the mitotic “blast phase” of activation that is characterized by increased median values of forward light scatter and side light scatter (SSC). Early experiments determined that every mouse and rabbit Ab tested for intracellular staining had increased fluorescence in T cell blasts even when specific for an Ag not present in mammalian cells. To normalize median fluorescence between naive and activated T cells, the following equation was used as a derived parameter in FlowJo: [(median phospho x at time y value)/(median SSC value at time y/median SSC value of PBS control)]/10. Derived parameters are shown in FlowJo on standard “analog” style scaling from 1 to 104 U, while digital data display negative numbers to 2 × 105 U. Because there is a 10-fold difference in the maximum values of these two scaling types, we divided the digital data by 10 to place them at a similar position on the histogram plot when displayed on an analog scaling. Normalization using control Ab specific for keyhole limpet hemocyanin in place of SSC yielded similar results. The biexponential transformation could not be applied to derived parameters (FlowJo limitation), and thus normalized data are shown with 4-log scales.
IL-2 capture assay
To follow the kinetics of IL-2 production by DO11 T cells primed in vivo, lymph nodes and spleen were harvested at various times after immunization and IL-2 secretion was measured during 45 min of incubation at 37°C without restimulation, using an IL-2 capture assay (Miltenyi Biotec).
Naive CD4+ T cells and Tregs were sorted from BALB/c Foxp3-GFP reporter mice using a MoFlo high-speed cell sorter (Dako). Cells were lysed using a nondenaturing cell lysis buffer containing 1% Triton X-100. Protein extracts of 5 × 105 cells were analyzed by Western blotting for suppressor of cytokine signaling-3 (SOCS-3) presence using a polyclonal anti-SOCS-3 Ab (CST) at a 1/1000 dilution, followed by an anti-rabbit/HRP Ab (Zymed Laboratories). β-Actin was probed to ensure equal loading of protein.
IL-2 treatment and in vivo suppression assay
IL-2/anti-IL-2 mAb complexes with Treg-specific effects were prepared as described (21). Forty-eight hours after i.p. injection of IL-2 complexes, BALB/c splenocytes were harvested and prepared for flow cytometry. For suppression assays, BALB/c mice were injected with IL-2 complexes every other day for 1 wk. Two days following initial IL-2 administration, CFSE-labeled DO11 T cells were transferred into IL-2-treated and untreated mice and stimulated with OVA. Spleens were harvested for analysis 3 days later.
Vaccinia virus (wild-type strain, Western Reserve) was obtained from American Type Culture Collection. Various amounts of virus based on PFU counts were injected via the tail vein into BALB/c mice.
Early IL-2 production by Ag-stimulated T cells induces STAT5 phosphorylation in endogenous Tregs
To define the cells that respond to IL-2 during an immune response, we used single cell biochemical analysis of distinct lymphocyte populations (22, 23, 24). Such assays exploit phospho-specific Abs as sensitive indicators of activated signaling pathways in complex cell populations, coupled with the visualization of phenotypic markers to allow the identification of individual cell types. IL-2 production was induced during an Ag-induced T cell reaction by transferring CD4+ T cells specific for chicken OVA, from the DO11.10 TCR transgenic mouse, into normal BALB/c recipients and immunizing with OVA peptide. At various times after immunization, splenocytes were harvested, fixed, and stained with a 9-color panel of Abs that could detect phosphorylated STAT5 (pSTAT5), the canonical IL-2-induced transcription factor, and various immune cell populations, including the transferred DO11 T cells. Although other signaling pathways (PI3K, Shc) are induced by IL-2 as well, many IL-2-induced functions in T cells are mediated by STAT5 (25), and measuring pSTAT5 is thus a powerful read-out to compare qualitative differences in IL-2 signaling between cell populations. We found that 6 h after OVA immunization, increased STAT5 phosphorylation was totally absent in endogenous naive T and B cells and minimally induced in a fraction of NK and dendritic cells. In contrast, the majority of Foxp3-expressing CD4+ T cells showed high levels of sustained STAT5 activation (Fig. 1⇓). The induction of pSTAT5 in Foxp3+ cells was rapid and was seen with different forms of Ag administration, including OVA plus the soluble adjuvant LPS and injection of mature BMDCs incubated with OVA (Fig. 2⇓, A and B). Although in most experiments we followed pSTAT5 induction in endogenous Foxp3+ T cells in the spleen, the same phenomenon was seen in lymph nodes (data not shown). Furthermore, the magnitude of pSTAT5 induction was proportional to the frequency of responding DO11 cells (Fig. 2⇓C). To determine if the induction of pSTAT5 in Tregs followed the kinetics of IL-2 production by the Ag-responsive T cells, we measured IL-2 secretion by DO11 T cells in lymph nodes and spleen at various times after Ag administration. In confirmation of published data (26, 27), DO11 T cells started producing IL-2 as early as 2 h after Ag administration and stopped producing the cytokine by 24 h. Thus, emergence of pSTAT5 in Tregs correlates with the timing of IL-2 secretion (Fig. 2⇓D).
To formally show that IL-2 produced by DO11 T cells was required to induce pSTAT5 in Tregs, we transferred either normal (Il-2+/+) or Il-2−/− DO11 T cells, immunized the recipients with OVA or control buffer, and stained for pSTAT5 in different cell populations after 9 h. Only when the DO11 T cells were capable of producing IL-2 was pSTAT5 induced in endogenous Foxp3+ T cells (Fig. 2⇑E). These results established that IL-2 produced during an Ag-specific T cell response acts rapidly on a large number of endogenous Foxp3+ regulatory T cells. It is noteworthy that 10–15% of Tregs are pSTAT5-positive in the absence of immunization (Figs. 1⇑ and 2⇑). However, the reduced population of Tregs present in Il-2−/− mice (15) does not show significant pSTAT5 signaling (data not shown), supporting previous reports demonstrating that homeostasis of Tregs in the steady-state depends on low levels of IL-2 present in the environment (28, 29).
Ag-stimulated T cells only acquire responsiveness to IL-2 after repeated encounter with Ag
Because the best established function of IL-2 is as an autocrine growth and differentiation factor, it is predicted to act on the Ag-responsive, IL-2-producing T cells. Surprisingly, at the times that Foxp3+ cells showed clear induction of pSTAT5, the OVA-stimulated DO11 cells remained negative (Fig. 3⇓A). This was not attributable to a failure to express the high-affinity IL-2 receptor, because after Ag exposure a large fraction of the DO11 cells expressed high levels of CD25 (comparable to the expression of CD25 on endogenous Foxp3+ cells) (Fig. 3⇓B). Upon activation, the T cells also expressed other components of the IL-2R, the IL-2/15Rβ and common γ-chains (Fig. 3⇓C). Early induction of pSTAT5 in Tregs but not in the Ag-responding DO11 cells was also seen after immunization with Ag and CFA (data not shown). It has been suggested that induction of an effective immune response requires repeated or prolonged exposure to Ag (30, 31). To determine if repeated stimulation with Ag was required to induce pSTAT5 in T cells, naive DO11 T cells were primed in vivo with Ag-pulsed dendritic cells and reexposed to Ag either 60 h or 5 wk later, with the latter reflecting a recall response of previously primed memory T cells (11). Repeated stimulation of IL-2-producing DO11 cells or DO11 memory cells induced a rapid autocrine pSTAT5 response in the Ag-responsive T cells (Fig. 3⇓, D and E). Furthermore, bystander memory T cells were also capable of inducing pSTAT5 in response to naive DO11 T cell activation (Fig. 3⇓F). Thus, the ability to initiate STAT5 signaling is acquired only after repeated antigenic stimulation and is preserved in memory T cells, but it is absent in naive T cells and is an inherent feature of regulatory T cells. The failure to activate IL-2-dependent STAT5 phosphorylation in the Ag-specific T cells may be because the secreted IL-2 is rapidly consumed by Tregs or because biochemical pathways relaying IL-2 signals are actively inhibited in naive T cells and repeated stimulation or memory cell differentiation is required to remove this inhibition and acquire responsiveness to IL-2 signals. When naive DO11 T cells were primed in Treg-deficient hosts, they still failed to activate autocrine STAT5 signals, indicating that consumption of IL-2 by Foxp3+ Tregs (32) does not explain the absence of pSTAT5 in DO11 T cells (Fig. 4⇓A). Furthermore, administration of potent IL-2/anti-IL-2 Ab complexes (21) during priming also failed to induce pSTAT5 in DO11 cells (Fig. 4⇓B). These results suggest the possibility of an inhibitory mechanism in naive T cells that is released upon repeated Ag encounter. SOCS proteins inhibit STAT signaling, and expression of SOCS-3 mRNA and protein is known to decrease after activation of helper T cells (33). We confirmed that SOCS-3 mRNA expression is reduced in activated DO11 T cells (data not shown), suggesting that this protein may regulate STAT5 signaling in Ag-stimulated T cells. Western blot analysis revealed diminished presence of the protein in CD4+Foxp3+ Tregs compared with naive T cells (Fig. 4⇓C), confirming a previous report showing translational regulation of SOCS-3 in Tregs (34). These data potentially explain the enhanced responsiveness of Tregs to IL-2 and are in agreement with other recent studies suggesting that diminished levels of SOCS-1 and SOCS-3 are required for robust STAT5 signaling in Tregs (34, 35).
IL-2 induces enhanced suppressive function in endogenous Tregs
The activation of STAT5 indicates that Foxp3+ cells are responding to IL-2, but it does not reveal the functional outcome of this response. To address this question, we assayed cycling and functional activity of Foxp3+ T cells exposed to IL-2. Following Ag-induced activation of DO11 T cells, the endogenous Foxp3+CD25+ cells showed increased expression of both Foxp3 and CD25 (Fig. 5⇓A). Furthermore, 48 h following Ag exposure, up to 50% of these cells stained positively for Ki-67, a marker of cell division, and the frequency of Foxp3+ cells almost doubled over time (Fig. 5⇓B). To determine if exposure to IL-2 also altered the functional activity of the Foxp3+ regulatory T cells, BALB/c mice were treated with IL-2/anti-IL-2 Ab complexes. This treatment induced high levels of pSTAT5 specifically in Foxp3+ T cells, as well as expansion of these cells (Fig. 5⇓C). In these treated mice, Ag-induced proliferation of transferred DO11 cells, as measured by CFSE dye dilution, was markedly inhibited (Fig. 5⇓D). Thus, IL-2 acts as a signal for Foxp3+ T cells to become potent suppressors. It is noteworthy that Il-2−/− T cells show enhanced responses to Ag under certain conditions (26), presumably because they are unable to activate Tregs.
Viral infection induces STAT5 phosphorylation and proliferation in endogenous Tregs
In most of the experiments described thus far, IL-2 was induced in vivo from transferred TCR transgenic cells exposed to their cognate Ag. The reasons for using this approach are that we could formally establish the physiologic requirement for IL-2 production from cells responding to the Ag (by using cells from the same TCR transgenic crossed with Il-2−/− mice), and we could readily distinguish the effects of IL-2 on the Ag-specific cells from those on endogenous (“bystander”) cells. However, a concern with studies using cells from TCR transgenic mice is that the high frequency of Ag-specific lymphocytes may not accurately reflect the normal immune repertoire. To determine if activation of endogenous Foxp3+ cells also accompanies a conventional immune response, normal mice were infected with vaccinia virus and STAT5 phosphorylation was assayed by flow cytometry. Viral infection led to a rapid dose- and time-dependent induction of pSTAT5 in Foxp3+CD4+ T cells (Fig. 6⇓A) (36). Mice that had been previously infected showed a more rapid response of Foxp3+ T cells upon rechallenge with the virus, presumably reflecting the accelerated kinetics of IL-2 production in memory T cells (Fig. 6⇓, B and C). The Foxp3+ T cells also stained for Ki-67, indicating commitment to cell division (Fig. 6⇓D). Thus, in a conventional immune response to a virus, as we observed using a TCR transgenic approach, the endogenous Foxp3+ T cells are early responders to IL-2 produced by virus-specific T cells.
Our experiments looking at the single cell biochemistry of lymphocytes in an immune response in vivo have shown that endogenous Foxp3+ Tregs receive signals from Ag-stimulated cells early in the response. Surprisingly, IL-2, generally thought to function as the key growth factor for Ag-reactive T cells, was identified as the “messenger” cytokine instructing Foxp3+ T cells to proliferate, increase expression of their essential transcription factor Foxp3, and enhance their suppressive activity. The autocrine action of IL-2 on the Ag-responding T cells, as measured by pSTAT5 induction, is not seen in the early phase of the response, despite high levels of CD25 expression. Since the absence of STAT5 signaling was also observed in cells stimulated in the absence of Tregs and competing endogenous T cell populations (Fig. 4⇑A), a cell-intrinsic inhibitory mechanism likely regulates IL-2 signal transduction. SOCS proteins play important roles in innate and adaptive immune responses by regulating cytokine responses (37). SOCS-1 and SOCS-3 have been shown to inhibit STAT5 signals downstream of IL-2 (38, 39). SOCS-1-deficient T cells show enhanced responsiveness to γc cytokines, and such uncontrolled cytokine responses may result in autoimmunity (37). Thus, the presence of SOCS-1 and SOCS-3 in naive T cells may explain their initial inability to transduce STAT5 signals in response to IL-2. Consequently, it would be predicted that T cells signaling through STAT5, such as Tregs and memory cells, express low levels of these inhibitory molecules. In support of this hypothesis, several recent studies have demonstrated a reduced presence of SOCS proteins in Tregs and activated T cells (33, 34, 35), resulting in increased responsiveness to IL-2. Our results thus suggest that there are important biochemical differences in how Tregs respond to their environment as compared with the naive T cells they are meant to regulate. On the other hand, memory cells responded to IL-2 in a similar manner as Tregs, phosphorylating STAT5 early after antigenic stimulation. Although the functional significance of these early STAT5 signals in memory cells is unclear at this time, some intriguing hypotheses can be made. For example, the generation of fast recall responses to Ag by memory cells may be dependent on their competence to receive early IL-2 signals. Additionally, the enhanced sensitivity to cytokines of memory cells may render these cells less susceptible to regulation by Tregs, which potentially also contributes to their increased Ag reactivity.
What might be the functional consequences of the IL-2-dependent activation of endogenous Tregs? It is likely that most endogenous Tregs are thymus-derived, “natural” Tregs, presumably specific for self-Ags (5, 40). It may be that early activation of these Tregs is a mechanism to minimize the risk of autoimmunity during an immune response to foreign Ags. Early activation of Tregs may also provide a constraint, or buffer, on immune responses to foreign Ags, thus preventing pathologic side-effects of physiologic immunity. Only after this “safety net” of regulatory cells is established would the Ag-specific response develop. Although rather high precursor frequencies of TCR transgenic T cells were necessary to elicit STAT5 signaling in the majority of Tregs residing in the lymphoid organs, our experiments with vaccinia virus infection demonstrate that a broad activation of Tregs is feasible under physiological circumstances. Also, while systemic Treg activation is unlikely after initiation of an immune response by a limited number of Ag-specific T cells, it is likely that Tregs localized in the vicinity of the responding T cells will receive IL-2 signals and control the environment where the response takes place. The strong activation of the regulatory compartment after the initiation of an immune response also raises the obvious question of how protective immune responses develop in the presence of a large number of activated Tregs. One possibility is that in the event of an infection, specific signals, for example, through the Toll pathway, are overriding the suppressors, thereby creating a permissive environment for the response to develop, as has been suggested by Pasare and Medzhitov (41). In a prolonged or secondary response, effector/memory T cells may directly inhibit Treg generation (42).
Thus, IL-2 may be the prototype of a signaling molecule that serves to first establish controls on immune responses and then to promote the development of such responses. Such a scenario also suggests that limiting the transient IL-2-dependent activation of regulatory cells may serve to maximize the effectiveness of vaccines.
Technologies associated with phospho-flow are licensed in part to BD Biosciences, and Dr. Garry P. Nolan is a consultant for BD Biosciences, a supplier of the reagents used in this report.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work was supported by National Institutes of Health Grants RO1 AI073656 and PO1 AI35297 (to A.K.A.) and RO1 AI065824, P01 AI36535, and National Heart, Lung, and Blood Institute Contract N01-HV-28183 (to G.N.).
↵2 W.E.O. and H.D. contributed equally to this work.
↵3 G.P.N. and A.K.A. share senior authorship for this paper.
↵4 Address correspondence and reprint requests to Dr. Hans Dooms, University of California at San Francisco School of Medicine, HSW-518, 513 Parnassus Avenue, San Francisco, CA 94143. E-mail address: or Dr. Abul K. Abbas, University of California at San Francisco School of Medicine, M-590, 505 Parnassus Avenue, San Francisco, CA 94143. E-mail address:
↵5 Abbreviations used in this paper: Treg, regulatory T cell; BMDC, bone marrow-derived dendritic cell; SSC, side light scatter; SOCS, suppressor of cytokine signaling.
- Received March 3, 2009.
- Accepted May 3, 2009.
- Copyright © 2009 by The American Association of Immunologists, Inc.