Competing feedback loops shape IL-2 signaling between helper and regulatory T lymphocytes in cellular microenvironments

Abstract

Cytokines are pleiotropic and readily diffusible messenger molecules, raising the question of how their action can be confined to specific target cells. The T cell cytokine interleukin-2 (IL-2) is essential for the homeostasis of regulatory T (Treg) cells that suppress (auto)immunity and stimulates immune responses mediated by conventional T cells. We combined mathematical modeling and experiments to dissect the dynamics of the IL-2 signaling network that links the prototypical IL-2 producers, conventional T helper (Th) cells, and Treg cells. We show how the IL-2-induced upregulation of high-affinity IL-2 receptors (IL-2R) establishes a positive feedback loop of IL-2 signaling. This feedback mediates a digital switch for the proliferation of Th cells and functions as an analog amplifier for the IL-2 uptake capacity of Treg cells. Unlike other positive feedbacks in cell signaling that augment signal propagation, the IL-2/IL-2R loop enhances the capture of the signal molecule and its degradation. Thus Treg and Th cells can compete for IL-2 and restrict its range of action through efficient cellular uptake. Depending on activation status and spatial localization of the cells, IL-2 may be consumed exclusively by Treg or Th cells, or be shared between them. In particular, a Treg cell can deprive a stimulated Th cell of its IL-2, but only when the cells are located in close proximity, within a few tens of micrometers. The present findings explain how IL-2 can play two disctinct roles in immune regulation and point to a hitherto largely unexplored spatiotemporal complexity of cytokine signaling.

Interleukin-2 (IL-2) is a cytokine produced by and acting on T lymphocytes with complex actions (1, 2). On the one hand, CD4⁺CD25⁺FoxP3⁺ regulatory T (Treg) cells depend on IL-2, so that IL-2 is critical for peripheral tolerance (2–5). On the other hand, IL-2 augments immune responses mediated by conventional T cells (6–10). Thus IL-2 serves dichotomous functions in the suppression and enhancement of adaptive immunity (11). The rational design of IL-2 or anti-IL-2 antibody therapies to treat cancer and autoimmune diseases by selectively targeting the desired T cell subpopulations is of great practical interest (12).

IL-2 is produced mainly by conventional (CD4⁺CD25⁻FoxP3⁻) T helper cells (Th cells), whereas Treg cells do not express IL-2 and rely on IL-2-secreting Th cells (3–5, 13–15). Treg cells have been shown to inhibit the activation of Th cells through multiple mechanisms (3–5). IL-2 capture by Treg cells has also been directly linked with their suppressive function (16–19). Expressing high numbers of IL-2 receptors (IL-2R), Treg cells can deprive Th cells of IL-2 and may thus impair Th cell proliferation and survival. By contrast, earlier studies have argued against diffusible factors in the suppressive action of Treg cells because spatial separation of Th and Treg cells in transwell cultures abrogates suppression (3, 14).

IL-2 signaling is highly regulated. The strength of stimulation of a Th cell by antigen and costimuli determines both the probability of secreting IL-2 and the rate of IL-2 secretion (20, 21). Initially, maximal IL-2 secretion requires extracellular IL-2 (17), whereas at later times IL-2 transcription is terminated by negative feedback and the cessation of antigen signaling (2). Regulation occurs also at the IL-2R level. Naïve Th cells do not express the IL-2Rα subunit (CD25); IL-2Rα is induced by antigen stimulation and combines with the constitutive β and γ subunits to the high-affinity IL-2R (22). The expression of IL-2Rα is strongly increased by IL-2 (23), so that IL-2 signaling may be self-enhancing. In Treg cells, the IL-2 gene is silenced and IL-2Rα is constitutively expressed through the action of the Treg-lineage-specifying transcription factor FoxP3 (24, 25). Treg cells also amplify their IL-2Rα expression in response to IL-2 (13, 17, 18).

To rationalize complex interactions in the immune system, mathematical models have been developed for the dynamics of lymphocyte populations and pathogens (26, 27). On a different scale, the molecular interactions in cytokine communication also form regulatory networks whose dynamics are poorly understood. In this paper we develop a mathematical model for the spatiotemporal dynamics of the IL-2 network and test its predictions experimentally on primary T cells (for an earlier modeling study in cell lines see ref. 28). This analysis uncovers critical parameters that govern the targeting of IL-2 to Th or Treg cells and thus helps rationalize the dichotomous role of this cytokine in immune regulation.

Results

Experimentally Based Model of IL-2 Signaling.

We consider the dynamics of IL-2 secretion and autocrine uptake by antigen-stimulated Th cells as well as IL-2 capture by Treg cells (Fig. 1A). IL-2 is bound with high affinity (K_d ≈ 10 pM) by the αβγ heterotrimeric IL-2R. Whereas the β and γ subunits are constitutively present, the α subunit is not expressed in resting Th cells but is induced by antigen and IL-2 signaling. In activated murine Th cells as well as in Treg cells, we observed a strict positive correlation of the IL-2 binding capacity with the cell-surface number of IL-2Rα, demonstrating that IL-2Rα (and not the β or γ subunits) limits IL-2 binding (Fig. S1). Therefore we focus the mathematical modeling on IL-2 dynamics and IL-2Rα expression.

Fig. 1.

Reaction–diffusion model of IL-2 signaling. (A) Antigen-stimulated Th cells secrete IL-2 that can be captured by IL-2Rs constitutively expressed on Treg cells or induced on Th cells. (B) Processes governing the dynamics of extracellular IL-2 (I), unoccupied (R), and occupied (R) IL-2Rα.

The model captures IL-2 signaling in the initial phase after antigen stimulation where the cells are primed for proliferation but have not yet entered cell division (up to 30 h). Indeed, previous experimental work has shown that IL-2 signaling in the first 10 h is critical for the proliferation decision of T cells in culture (17, 29). The IL-2Rα number on Th and Treg cells and of the extracellular IL-2 concentration are governed by the following (Fig. 1B): antigen-induced IL-2 secretion by Th cells (q₁); IL-2 uptake through IL-2R internalization, endosomal degradation, and extracellular decay (k_iC, k_deg, and k_d, respectively); IL-2Rα expression (v₁, v₂), internalization (k_iR), endosomal degradation (k_deg), and partial recycling (k_rec). The interaction scheme translates into coupled reaction–diffusion equations with reactive boundary conditions (SI Text S1: Mathematical Modeling). The kinetic parameters were estimated from previous experiments and FACS measurements of IL-2Rα expression (Table S1).

We solved the reaction–diffusion equations numerically for arrays of ≈150 cells (such as a cell patch in primary culture; RD model). As an approximate model that yields analytic insight, we also considered a Th–Treg cell pair communicating through IL-2. The time-scale separation between IL-2 diffusion (sec–min for L < 100 μm) and IL-2Rα expression (h) allowed us to apply a quasisteady state approximation for diffusion and reduce the model to coupled ordinary differential equations (QSSA model; SI Text S1: Mathematical Modeling).

Feedback Regulation of IL-2 Signaling Causes All-or-Nothing Response in Th Cells.

We first compute the dynamics for a Th cell in the absence of a Treg cell. Antigen stimulation induces both IL-2 and IL-2Rα expression. Secreted IL-2 strongly augments IL-2Rα expression and thus the formation of high-affinity IL-2Rs (Fig. S2). When the IL-2 secretion rate is raised continuously in the model, the expression of IL-2Rα responds in a digital, all-or-nothing fashion (Fig. 2A and Fig. S3). Thus the model predicts that the IL-2 secretion rate must exceed a threshold value θ to switch IL-2Rα expression to the activated state and permit extensive autocrine IL-2 signaling.

Fig. 2.

Digital IL-2Rα expression in Th cells. (A) The stimulus–response curve of IL-2Rα expression shows distinct basal and activated IL-2Rα levels (solid lines) that overlap in a small bi-stable region (dashed line, unstable state) (QSSA model). (B) Digital switching of IL-2Rα expression in a single cell translates into a binary response pattern of the cell population (QSSA model). For 1,000 cells the rates of antigen- and feedback-driven IL-2Rα expression were chosen from log-normal distributions (v₁₀ = 150 ± 23 molecules/cell/h and v₁₁ = 3,000 ± 762 molecules/cell/h) and increasing antigen stimulus modeled by q₀ = 0, 300, 600, 1,800, 2,700, 15,000 molecules/cell/h (Top to Bottom). (C) Simulation of the RD model for 173 Th cells with 25% secreting IL-2 (q₀ = 5,000 molecules/cell/h). The snapshot at 10 h after onset of stimulation shows complete autocrine and rare paracrine activation. (D) Digital distribution of IL-2Rα expression with, Top to Bottom, 10, 20, 30, 40, 50, and 60% IL-2 secreting cells (RD model).

The activation threshold θ will show some random variation from cell to cell. Therefore, the digital response at the single-cell level translates into a bimodal distribution of IL-2Rα expression in the cell population, assuming a generic log-normal distribution of the IL-2Rα expression rate (Fig. 2B). As the rate of antigen-induced IL-2 secretion (q₀) rises, an increasing fraction of cells switches IL-2Rα expression to the activated level. This digital switch is also observed when the full RD model is solved for a patch of cells; according to experimental evidence, only a fraction of cells was taken as (randomly assigned) IL-2 producers (Fig. 2C) (20, 21). Again the fraction of IL-2Rα positive cells increases with rising antigen stimulus, which has been modeled as an increasing fraction of IL-2 producing cells (Fig. 2D). In particular, we found that after upregulation of IL-2Rα the uptake of IL-2 becomes so efficient that IL-2 eventually remains elevated only locally around secreting cells. As a consequence, only a rather small fraction of nonsecreting cells is activated in a paracrine manner.

To examine the prediction of digital IL-2 signaling experimentally, we stimulated murine Th cells in separate experiments with increasing doses of cognate antigen. The cell-surface expression of IL-2Rα showed a bimodal distribution, where an increase in antigen dose raised the fraction of activated cells but not the mean level of IL-2Rα expression (Fig. 3A). Proliferating T cells were found exclusively in the subpopulation that upregulated IL-2Rα (Fig. 3B). The addition of neutralizing anti-IL-2 antibodies abolished both the switching on of IL-2Rα expression (Fig. 3C) and cell proliferation (Fig. 3D). Thus, the regulation of IL-2Rα by IL-2 created a positive feedback loop that caused a digital switch in autocrine IL-2 signaling. Only cells that executed this IL-2 switch proliferated.

Fig. 3.

Th cells exhibit digital IL-2Rα upregulation and proliferation. (A) Digital IL-2Rα expression pattern in Th-cell populations stimulated with increasing antigen doses. OVA-TCR^tg/tg CD4⁺CD25⁻ T cells were labeled with the proliferation marker CFSE and cultured with irradiated antigen-presenting cells and OVA peptide (concentration as indicated). Surface IL-2Rα was measured by flow cytometry at 72 h to correlate it with cell proliferation (1 representative experiment of 3). (B) Only cells with upregulated IL-2Rα proliferated, as seen by loss of CFSE intensity. Addition of blocking anti-IL-2 antibodies (αIL-2) to the cell culture prevented both IL-2Rα upregulation (C) and cell proliferation (D).

Paracrine IL-2 Uptake by Treg Cells Shifts Th-Cell Response Threshold.

To define the conditions for paracrine IL-2 capture by a Treg cell, we analyzed the QSSA model for a Th–Treg cell pair. When the two cells are located in proximity (L = 10 μm), the Treg cell acts as a potent sink for IL-2, and the paracrine IL-2 signal causes further upregulation of IL-2Rα on the Treg cell (Fig. 4A, red curve). At the same time, the activation threshold θ for the Th cell is markedly increased (Fig. 4A, black curve; see θ in Fig. 2A). Thus the presence of a nearby Treg cell is predicted to inhibit autocrine IL-2 signaling of the Th cell.

Fig. 4.

Interaction of Th and Treg cells through IL-2. (A) Bifurcation diagram computed for a proximal Th–Treg cell pair (QSSA model, L = 10 μm). The Th-cell activation threshold θ is increased and bi-stability enhanced (black line) (Fig. 2). By contrast, the upregulation of IL-2Rs on the Treg cell is practically continuous (red line). (B) Accordingly, bimodal expression of IL-2Rα on Th cells and gradual upregulation on Treg cells are predicted (parameter values as in Fig. 2). (C) Simulation of RD model for a Th–Treg coculture (cell ratio 2:1; 20% IL-2 producing Th cells, q₀ = 5,000 molecules/cell/h; snapshot at 10 h after onset of stimulation). None of the Th cells become activated. (D) Upregulation of IL-2Rα on Th and Treg cells (RD model, fractions of IL-2-secreting Th cells were, Top to Bottom, 10, 20, 30, 40, 50, and 60%). Heterogeneity arises due to the random positioning of the different types of cell on a hexagonal array.

The digital nature of the Th-cell response curve is retained in the presence of a Treg cell. The pronounced hysteresis now present in the Th-cell response, as seen by the S-shaped response curve, is consistent with the experimental observation of an early time window for the suppressive action of Treg cells (29) (see following paragraph and Discussion for further explanation). In contrast to the Th-cell dynamics, the Treg-cell response curve does not show a measurable effect of bistability (although a small hysteresis loop is present). Treg cells “escape” the digital feedback switch because high-affinity IL-2Rs are already expressed in the resting state. Therefore, IL-2Rα expression in Treg cells will be a more graded function of Th-cell IL-2 secretion (Fig. 4B). The RD model for a cell patch in Th–Treg coculture confirms that Treg cells can be potent IL-2 sinks and prevent the activation of IL-2-secreting Th cells (Fig. 4C). The pattern of IL-2Rα upregulation on Treg cells is again more graded than on Th cells (Fig. 4D). However, in contrast to the QSSA model (which considers a Treg cell together with an IL-2-secreting Th cell), some Treg cells remain that do not upregulate IL-2Rα as they are not in proximity of IL-2 secreting Th cells.

Distance Dependence of Paracrine IL-2 Uptake.

Over which spatial range can a Treg cell capture IL-2 from a Th cell? The critical distance at which both cells receive an equal share of the IL-2 is given approximately by

(SI Text S2: Calculation of the Critical Distance for Competition L_1/2); r₁ and r₂ denote the IL-2Rα surface densities on Th and Treg cells, respectively, D is the IL-2 diffusion coefficient, and k_on the binding rate constant of IL-2 to the high-affinity IL-2R. For intercellular distances smaller than L_1/2, IL-2 is mainly captured by the Treg cell, whereas autocrine reuptake by the Th cells predominates when the cells are further apart. For reasonable parameter estimates (10⁴ and 10³ IL-2Rs on Treg cell and activated Th cell, respectively, cell diameter 10 μm, D = 10 μm²s⁻¹ and k_on = 111.6 nM⁻¹h⁻¹), the paracrine signaling distance is (because the QSSA model underestimates IL-2 loss, this must be regarded as an upper bound). As is set mainly by the autocrine IL-2 uptake capacity of the Th cell. This explains the strong hysteresis in the Th-cell response curve (Fig. 4A). Before antigenic stimulation, the IL-2 uptake capacity of the Th cell is low and the Treg cell prevents autocrine upregulation of IL-2Rα. However, once sufficient high-affinity IL-2Rs are expressed on the Th cell, the Treg cell cannot interrupt the now established autocrine loop.

In addition to intercellular distance, the IL-2 secretion rate is critical for IL-2 competition. The phase diagram spanned by these two parameters shows three different outcomes of Th-to-Treg IL-2 signaling (Fig. 5). For small intercellular distances and moderate IL-2 secretion, the Treg cell will deprive the Th cell of IL-2 and prevent the autocrine IL-2Rα switch (Fig. 5, red region). When IL-2 secretion rate becomes high enough (corresponding to strong stimulation of the Th cell), paracrine and autocrine IL-2 signaling coexist without competition (gray region). If the Treg cell is located too far away, efficient autocrine IL-2 uptake by an activated Th cell can prevent paracrine IL-2 capture (blue region).

Fig. 5.

Distance dependence of paracrine IL-2 uptake by Treg cells. Behavior of a Th–Treg cell pair in dependence on IL-2 secretion rate by the Th cell and intercellular distance (Left) (QSSA model). The activation threshold for Th cells (θ, thick black line) and the line indicating strong IL-2 capture by Treg cells (thin black line, 10,000 upregulated IL-2Rα per cell) divide the phase diagram into four regions (see text).

IL-2 deprivation of Th cells by Treg cells has been implicated in the immunosuppressive action of the latter (16–19). The model predicts that IL-2 deprivation occurs under conditions of limited IL-2 supply by Th cells and spatial proximity of competing Treg cells.

Adaptive Competition of Treg Cells for IL-2.

To test the predicted Treg response, we cocultured primary Th and Treg cells with different antigen specificity (at ratio 2:1, yielding effective suppression) (17). As in the model, a dose–response to the Th-cell stimulus was measured while keeping the antigen stimulus for the Treg cells constant. The expression patterns of IL-2Rα on Th and Treg cells exhibited the predicted qualitative difference. Th cells displayed digital IL-2Rα expression (Fig. 6A), whereas Treg cells showed gradual upregulation of IL-2Rα (Fig. 6B). However, IL-2Rα appeared to remain at basal level in a small fraction of Treg cells even for strong Th-cell stimulus. This is in agreement with the results of the RD model (Fig. 4D) and can be accounted for by these cells being too distant from IL-2-secreting Th cells, consistent with a short range of paracrine IL-2 signaling.

Fig. 6.

IL-2 uptake by Treg cells. In a coculture of Th and Treg cells, IL-2Rα expression on Th cells is digital (A) whereas the expression of IL-2Rα in Treg cells adapts in a graded manner to the Th-cell stimulus (B). Aggrecan-TCR^tg/tg CD4⁺CD25⁺ Treg cells were cocultured with CFSE-labeled OVA-TCR^tg/tg CD4⁺CD25⁻ Th cells at 1:2 ratio at different concentrations of OVA peptide (as indicated) and constant [aggrecan] = 2 μg/mL. Surface IL-2Rα was determined by flow cytometry (72 h) (1 representative experiment of 3). (C) Mean IL-2Rα expression and (D) fraction of active cells in the activated Th cells (blue lines) and Treg cells (red line) as determined by fitting the experimental IL-2Rα histograms. (E) Divided Th cells have elevated IL-2Rα. At 1 μg/mL OVA some activated Th cells already begin to lose IL-2Rα expression at 72 h so that some divided cells appear IL-2Rα-negative. The Treg cells have not been labeled with CFSE. The onset of IL-2Rα upregulation (F) and cell proliferation (G) are shifted to higher antigen stimuli by the presence of Treg cells, whereas a large stimulus (1 μg/mL OVA) overcomes the inhibitory Treg effect on the proliferation of the Th cells. Cell proliferation was quantified as ΔCFSE = MFI(generation 0) − MFI(generations >0); P-values for equal proliferation of Th cells and Treg cells are (t test): 0.017 (2.5 × 10⁻³ μg/mL OVA), 0.002 (0.01 μg/mL), 0.076 (0.05 μg/mL), and 0.359 (1 μg/mL).

To characterize the IL2Rα expression patterns in Th cells and Treg cells, we fitted the IL-2Rα histograms by the sum of two log-normal distributions, capturing the cells with basal and activated IL-2Rα expression, respectively (SI Text S3: Quantitative Analysis of the IL-2Rα Expression Patterns of the Th and Treg Cells and Fig. S4). The activated Th cells had a constant mean IL-2Rα expression independent of the antigen stimulus (Fig. 6C, blue data), whereas their fraction showed a continuous increase with the antigen stimulus (Fig. 6D, blue data). This behavior defines a digital expression pattern. Conversely, the fraction of activated Treg cells showed no significant change with antigen stimulus (Fig. 6D, red data) whereas their mean IL-2Rα level continuously increased with the Th-cell antigen stimulus (Fig. 6C, red data). Moreover, the decomposition of the Treg cells into basal and activated modes was less well defined than for Th cells (compare the error bars in Fig. 6D), owing to the essentially single-peaked IL-2Rα distribution of the Treg cells. The observed gradual shift of IL-2Rα expression on the Treg cells is matched by the QSSA model (see Figure 4B) whereas the RD model, in addition to intermediate expression values, has more pronounced peaks at high and low IL-2Rα expression (for further analysis, see Fig. S5).

Due to continuous IL-2-dependent upregulation, IL-2Rα expression on the bulk of the Treg cells stayed above the IL-2Rα level reached on activated Th cells, enabling competitive IL-2 uptake by Treg cells (see Eq. 1). Proliferation of Th cells again correlated with the extent of IL-2 signaling as measured by IL-2Rα upregulation (Fig. 6E). Overall, upregulation of IL-2Rα on Th cells (Fig. 6F) and proliferation of Th cells (Fig. 6G) were inhibited by the presence of the Treg cells at intermediate antigen stimuli. Moreover, the mean IL-2Rα level on activated Th cells was somewhat smaller in coculture with Treg cells than in pure Th cell culture, indicating that the Treg cells also exerted a direct negative effect on IL-2Rα expression (Figs. 3A and 6A). Consistent with the model prediction that the IL-2 secretion rate of strongly stimulated Th cells is sufficient to overcome the competitive effect of Treg cells, the inhibition of Th-cell IL-2Rα upregulation and proliferation was abrogated at high antigen stimulus (Fig. 6 F and G). Similarly, addition of IL-2 to a Treg–Th coculture restored proliferation of Th cells (17) (Fig. S6).

In summary, the adaptation of IL-2Rα expression to ambient IL-2 concentration maintained a strong competitive advantage of Treg cells for IL-2 uptake. The dose–response for the suppressive action of Treg cells agreed with the predicted outcome of IL-2 competition, which is effective at moderate but not high rates of IL-2 secretion.

Discussion

The spatiotemporal dynamics of the IL-2 network described here have several functional implications discussed in the following (and summarized in Table S2). It has previously been proposed that a T cell stimulus must exceed a discrete threshold to trigger proliferation (1, 30). The IL-2Rα switch found here provides a mechanistic basis for this activation threshold. Digital regulation in T cells also occurs in antigen signal transduction (31), recently demonstrated to be based on bistable feedback regulation of SOS (32), and in NFAT nuclear translocation (21). The autocrine IL-2 loop may have an integrative function because it requires that the activation signal has already passed the thresholding devices in intracellular signaling. In agreement with this, we observed that cell proliferation correlated with a switch to high IL-2Rα expression. Notably, autocrine positive feedback also occurs with other cytokines (33, 34) or growth factors (35), so that digital-switch mechanisms may be more widespread in cytokine signaling. Other regulatory mechanisms, such as cross-inhibition, can also convert graded input into digital output (36).

The IL-2-mediated activation switch of Th cells is not cell autonomous because IL-2 diffuses. However, the model indicates that paracrine IL-2 signaling (to Treg cells or nonsecreting Th cells) is limited to the neighborhood of IL-2-secreting cells, because IL-2 uptake is very efficient once IL-2Rα becomes upregulated. High IL-2 concentrations (0.1–1 nM) are predicted in microenvironments near secreting cells (compared with the IL-2R K_d of ≈10 pM), whereas IL-2 concentrations in supernatants of T cell cultures are much lower (pM) (18). The directed secretion of cytokines into the immunological synapse between T cells and antigen-presenting cells may also contribute to confine cytokines (37).

Under conditions of limited IL-2 secretion, our results indicate that IL-2 signaling from Th cells to Treg cells takes place in microenvironments (e.g., between cells bound to the same antigen-presenting cell). Proximal Treg cells can then deprive moderately stimulated Th cells efficiently of IL-2 (and possibly other common-γ-chain cytokines), in agreement with a direct suppressive effect of IL-2 competition (15–17). However, when Th cells are strongly stimulated and many cells in a neighborhood produce IL-2, Treg cells cannot deprive them of the cytokine. Therefore, IL-2 competition (and possibly other resource competition mechanisms) can naturally sharpen the border between T cell agonists and antagonists, suppressing bystander activation by self-peptides but allowing activation by strong agonists. The silencing of the IL-2 gene could thus have been a primary event in the evolution of Treg cells.

There are many other mechanisms of Treg cell-mediated immunosuppression that are independent of IL-2 or induced by IL-2 (3–5, 16, 37, 38). However, our study reveals three emerging properties of IL-2 competition that fit salient characteristics of Treg cell action: (i) distance dependence, (ii) hysteresis, and (iii) limitation by strong antigenic stimulation of Th cells. First, transwell experiments with separation of Treg and Th cells by a protein-permeable barrier have demonstrated that Treg-mediated suppression acts only over short distance (3, 14). Whereas these data were taken (and are still so regarded by many authors) as evidence of cell contact-dependent inhibition, our results imply that the localized paracrine IL-2 signaling would equally be interrupted in a transwell experiment. Second, Treg cells added with a delay of 6–10 h to stimulated naïve Th cells (i.e., much before the onset of cell proliferation) no longer have a suppressive effect (29). IL-2 competition is naturally limited to an early time window because Th cells that have not yet switched on the autocrine IL-2 loop can readily be suppressed through IL-2 capture by Treg cells, whereas an already active autocrine loop cannot be interrupted in this way. The reason for this is the intrinsic advantage of autocrine over paracrine cytokine uptake. Third, it has been shown here and previously (17) that Treg-mediated suppression is not effective when the Th cells are acutely and strongly stimulated by cognate antigen. Thus, depending on the strength of the Th-cell stimulus and the localization of Treg cells, Th cells may fully (re)capture the IL-2, they may be deprived of IL-2 by proximal Treg cells, or both subpopulations may share the cytokine (which can trigger delayed IL-2-dependent suppressive mechanisms in the Treg cells) (16, 38). Such dose- and space-dependent responses with discrete thresholds may also exist for other cytokines, and their elucidation will benefit the design of cytokine-based therapeutic approaches.

Materials and Methods

Mice.

BALB/c OVA-TCR^tg/tg DO.11.10 mice purchased from BfR (Berlin); BALB/c aggrecan TCR^tg/wt (5/4E8) (peptid sequence: ATEGRVRVNSAYQDK) obtained from Wilem van Eden (University of Utrecht). All mice were housed in a specific pathogen-free (SPF) environment and used at 8–10 weeks of age.

Antibodies.

Anti-mouse antibodies: FITC- or PE-conjugated anti-CD4 (GK1.5, BD-PharMingen), allophycocyanin-conjugated (APC) anti-CD25 (PC61, BD-PharMingen), biotinylated anti-CD25 (7D4, BD-PharMingen), biotinylated anti-CD25 F(ab)₂ (PC61, DRFZ), anti-CD25-PE (7D4, Miltenyi Biotec), and Cy5-conjugated anti-DO.11.10 OVA-TCR (KJ1.26).

Cell Staining and Purification.

Suspended lymph node and spleen cells were stained with biotinylated anti-CD25 F(ab)₂, incubated with anti-biotin microbeads and sorted by AutoMACS (Miltenyi Biotec) to obtain Treg cells. Subsequently, CD25^– cells were labeled with anti-CD4 microbeads and sorted for CD4 expression. Antigen presenting cells were sorted using anti-MHC class II microbeads. The purity of the various sorted cell populations was higher than 95%.

CFDA-SE Labeling.

CD4⁺CD25^– or CD4⁺CD25⁺ T cells were washed with PBS, resuspended in a 1-mM solution of CFDA-SE (Sigma) at a density of 1 × 10⁷ cells/mL and incubated for 4 min at room temperature and washed with RPMI 1640 culture medium (BioWhittaker) containing 10% FCS.

Proliferation Assays.

A total of 0.33 × 10⁶ irradiated APC and 0.18 × 10⁶ T cells were incubated for 72 h in a 96-well U-bottom plate. CD25⁺ Treg cells and CFDA-SE-labeled CD25^– Th cells were mixed in a 1:2 ratio or cultured alone. Treg cells from aggrecan TCR^tg/wt mice were stimulated with 2 μg/mL aggrecan peptid (Agg_70–84), the CD4⁺CD25- OVA-TCR^tg/tg Th cells with indicated amounts of Ova_323–339-peptide. RPMI-1640 supplemented with 10% heat inactivated FCS, 100 U/mL penicillin plus 100 U/mL streptomycin, 2 mM L-glutamine and 50 lM2-ME (Sigma) used for cell cultures.

FACS Analysis.

Cells were stained for CD4, CD25, and Ova-TCR (anti-Ova-TCR-Cy5), measured with FACS-Calibur (BD), and analyzed with FlowJo (Tree Star). Dead cells were excluded via counter staining with PI (Sigma).

Numerical Simulations.

The RD model and the QSSA for a Th–Treg cell pair have been solved numerically using an ADI finite-difference scheme and the software Xppaut (written by Bard Ermentrout), respectively.