Balancing speed and accuracy of polyclonal T cell activation: a role for extracellular feedback
© Savir et al.; licensee BioMed Central Ltd. 2012
Received: 7 February 2012
Accepted: 2 August 2012
Published: 27 August 2012
Extracellular feedback is an abundant module of intercellular communication networks, yet a detailed understanding of its role is still lacking. Here, we study interactions between polyclonal activated T cells that are mediated by IL-2 extracellular feedback as a model system.
Using mathematical modeling we show that extracellular feedback can give rise to opposite outcomes: competition or cooperation between interacting T cells, depending on their relative levels of activation. Furthermore, the outcome of the interaction also depends on the relative timing of activation of the cells. A critical time window exists after which a cell that has been more strongly activated nevertheless cannot exclude an inferior competitor.
In a number of experimental studies of polyclonal T-cell systems, outcomes ranging from cooperation to competition as well as time dependent competition were observed. Our model suggests that extracellular feedback can contribute to these observed behaviors as it translates quantitative differences in T cells’ activation strength and in their relative activation time into qualitatively different outcomes. We propose extracellular feedback as a general mechanism that can balance speed and accuracy – choosing the most suitable responders out of a polyclonal population under the clock of an escalating threat.
KeywordsSystems immunology Positive feedback Speed vs. accuracy IL-2 CD25
The immune system offers numerous examples of extracellular positive feedback mediated by cytokines (small secreted proteins). A specific example is provided by the cytokine Interlukin-2 (IL-2) through its roles in the process of T cell activation. T cells are activated in response to binding of an antigen to their T cell receptor (TCR). Upon TCR engagement and receipt of appropriate co-stimulatory signals T cells secrete IL-2, which serves as a major T cell growth factor that promotes their commitment to proliferation [10, 11]. Binding of IL-2 to its receptor (IL-2R) leads to a positive feedback by increasing IL-2R expression levels . Additionally, the IL-2:IL-2R complex undergoes rapid internalization and degradation  leading to consumption of IL-2 from the intercellular space (Figure 1B). Since this system has been extensively studied and many of its kinetic parameters are known, it can be readily modeled mathematically [14–17]. Recently, IL-2 mediated interactions between T regulatory cells (Tregs) and T effector cells has been studied, showing how depletion of the shared cytokine by Tregs contributes to suppression of effector cells [16, 17]. Extracellular feedback driven by IL-2 can also affect polyclonal interactions between effector T cells of different antigen specificities, which interact during their priming within lymphoid organs. These interactions between T cells, together with other factors such as availability of co-stimulatory signals, can shape the repertoire of responding T cells through inter-clonal cooperation and competition .
Experimentally, competition between cells was observed as a possible outcome of polyclonal interactions, where activation of cells with one antigen specificity inhibits the response of cells with a different specificity in their vicinity . However, cooperation between polyclonal cells, in which activation of cells is stronger in the presence of neighboring cells with a different specificity, has also been observed in similar experimental models . In a recent report, both cooperation and competition were observed in the same experimental model, which evaluated interactions between groups of endogenous T cells responding to different mixtures of peptides . Furthermore, time dependency of the outcome has also been observed – a limited window of opportunity of a few hours exists for the inhibition of competing cells, both by regulatory T cells  and by competing effector T cells . The mechanism underlying this time dependency remains obscure.
Here, we present a mathematical model which suggests that extracellular feedback can account for opposite outcomes: competition or cooperation between interacting polyclonal T cells. The outcome in any specific case depends on quantitative parameters of the activation of the interacting cells, such as their relative TCR stimulation strength. Furthermore, time dependency of the interaction outcome emerges from our model. Thus, if interacting cells are activated at different times, their fate will depend on their relative times of activation. We show that these rich behaviors of the system require non-linearity of the feedback. A linear feedback results in competition and exclusion, where the stronger cell wins regardless of activation time. We propose that this time dependency can serve to balance speed and accuracy of the polyclonal T cell response.
We model the behavior of a system comprising two CD4+ T cells that have been activated by their cognate antigens, but before proliferation takes place (within ~30 hours after activation) (Figure 1B). As we aim to understand general design principles, we study a simplified model of the system, which still capture its core features. In the following, we describe the main model we use while various modifications are further discussed later and in the Additional file 1.
Figure 1B schematically presents the components of the model. We assume that IL-2Rs are initially expressed at a low constitutive rate, g c , and are degraded at a rate a U , resulting in a low initial number of IL-2R molecules on the T-cell surface (g c /a U ~ 150, see Additional file 1: Table S1). This provides the cells with an initial responsiveness to IL-2. Upon TCR binding, the T-cells start to secrete IL-2 at a rate s that is proportional to the strength of the TCR signal, which is denoted by the parameter 0 ≤ m ≤ 1. The dependence of IL-2 secretion rate on TCR signal strength can be relaxed: assuming that cells secrete IL-2 in a constant rate, regardless of their TCR strength, does not change the general behavior of the model (Additional file 1: Figure S2A). Signaling induced by binding of IL-2 to its receptor results in increased expression of IL-2Rs on the cell surface, thus closing the positive feedback loop. We assume that the induced expression rate of IL-2R is also proportional to the strength of the TCR signal (see below). Additionally, the IL-2R:IL-2 complex is internalized and degraded at a rate a B .
The TCR signaling strength, m, is parameterized to be between zero and one. This value reflects the effect of TCR signaling on the production rate of IL-2R and IL-2. TCR signaling strength mainly depends on the affinity of the antigen to the TCR, and on the number of bound TCRs on the T cell surface. The mapping of these factors into the value of m may differ between different systems (mode of stimulation, type of antigen) and can be evaluated experimentally for each system, as we discuss below. However, this mapping does not affect the behavior of the model (Eq. 1-4) describing dynamics of the IL-2-IL-2R system. Co-stimulation is modeled effectively within the TCR signaling strength parameter m. Other rate constants that determine the system’s behavior are: k on , k off - binding and unbinding rates, respectively, of IL-2 to its receptor; and d I , IL-2 degradation rate (or its effective removal from the interaction volume due to transport). As noted before (16,17) IL-2 removal is dominated by its endosomal degradation following binding to its receptor and internalization of the complex, which is modeled by the term in Eq. 2.
IL-2R production rate following TCR induction is given by the positive feedback term in Eq. 4, m·g f ·(B2/(B2 + K f 2), with a maximal synthesis rate of m·g f and half saturation at B = K f . We model dependence of IL-2R production rate on both m (TCR signal) and B (IL-2 signal), as both signals are required for efficient activation of the IL-2Rα gene . IL-2 mediated feedback on receptor production is assumed to be in the form of a Hill function, with a Hill coefficient of 2. Cooperation is assumed as the main signaling molecule downstream of IL-2R, STAT5, forms dimers after its phosphorylation, which serve as a transcription factor driving expression of the IL-2Rα gene . The values used for these constants in the following simulations are given in Additional file 1: Table S1. We model interaction between cells that are in a close proximity to each other – intercellular distance of the order of 100 microns (e.g. within the same cluster on a dendritic cell ). We assume a well-mixed microenvironment, hence a volume of 100 microns3 is used in the numerical simulations. Our results are not sensitive to this value (See sensitivity analysis in the Additional file 1).
Modeling the IL2 extracellular feedback system
Extracellular feedback translates quantitative differences in initial conditions into opposite qualitative behaviors: cooperation, coexistence and competition
Next, we solve Eqs. 1,-4, and find the steady state solutions for the case of two interacting T-cells that are activated with TCR signal strengths m1 and m2, respectively. Figure 2B shows the fixed points of B as a function of m2 for two cases: a) Cell2 is activated alone (similar to Figure 2A) and b) the cell is activated alongside a strongly activated cell (m1 = 1). The threshold of commitment for Cell2 (the critical m value at which B increases to its high level), is shifted to higher m values in the presence of the strongly activated cell. Hence, at a range of m values above threshold, the presence of the strong cell excludes Cell2 from committing due to competition on IL-2 between the two cells.
with i = 1 or 2 for the two cells. This interaction index confers the difference between the cell’s bound IL-2R level if activated alongside another cell or if activated alone. The inset in Figure 2B shows the calculated C for the aforementioned scenario. The value of C provides a useful description for the outcome of the two cells interaction: C > 0 implies one cell benefits from the interaction by utilizing the IL-2 secreted by the other cell – we term this “cooperation”, while C < 0, implies one cell suffers from the interaction due to IL-2 consumption by the other cell – we term this “competition”. When C ~ 0, the presence of one cell does not significantly affect the other cell – we term this outcome “coexistence”. We note that these definitions are used here relative to a specific cell and are not exclusive; for example, under some conditions both cells can benefit from the interaction (C1 > 0 and C2 > 0), while under other conditions one cell may benefit on the expanse of the other (C1 > 0, C2 < 0).
A non-linear feedback is essential for obtaining these complex system behaviors. We consider two hypothetical cases: a) no IL-2 mediated feedback on IL-2R synthesis rate and b) a linear feedback on IL-2R synthesis rate. In both cases, the number of bound receptors increases linearly with m for almost all values of m, and no bistability occurs (see Additional file 1 for a detailed analysis). In the first case, receptor synthesis rate depends only on m, and not on the level of bound receptors, and is given by m·g i . In this case, cells always slightly benefit from the interaction (Additional file 1: Figure S1). In the second case, receptor synthesis rate is given by: m·g i ·B. As a result, when two cells interact they do not cooperate but always compete, with the stronger cell excluding the inferior one (Figure 3B).
Our results are in general not sensitive to the exact values of parameters used in the simulations (see sensitivity analysis in the Additional file 1: Figures S3-S9). Changing the rate constants of Additional file 1: Table S1 mainly affects the single cell behavior, by modifying the threshold value for m. Around the modified single-cell threshold, the two cell interaction space does not change significantly, and shows regions of cooperation, competition and coexistence in similar positions relative to the threshold values of the two cells. We also checked the sensitivity of these results to some of the assumption made in our model. One assumption is that IL-2 is made only by the two interacting cell, and is dependent on m (Eq. 1). We modified this assumption in two ways: first, we checked a case in which cells secrete IL-2 at some constant rate, independent of the level of m. Second, we added a constant production term of IL-2, presumably made by other nearby cells and entering the interaction volume of the two modeled cells. Both perturbations didn’t change the qualitative behavior of the model, which still exhibits regions of cooperation, competition and coexistence (Additional file 1: Figure S2A, B).
Interaction outcome depends on the relative activation times of the two cells
The temporal transition point between outcomes of the interaction lies at a critical relative time of activation, Δtc, which depends on the TCR stimulation strength of both cells. In Figure 4B, we show a specific example for the time window for competition, for two interacting cells with m1 = 0.65 and m2 = 1. If the stronger Cell2 is activated within the critical time window after the activation of Cell1 (Δt < Δtc), it competes with and excludes the inferior Cell1. However, if Cell2 is activated after the window has closed (Δt > Δtc), it cannot exclude Cell1 and both cells coexist. For this case, Δtc ≈ 14 hours, which is similar to the critical time window for competition between effector T cells of different TCR specificities that was observed experimentally [21, 22]. Our analysis shows that the length of this critical time window increases as the difference in activation strengths of the two cells is larger (Figure 4C). Hence, the stronger the later activating cell, the more time it has to exclude its inferior competitor.
The overall time dependence of the system over the entire interaction phase-space is shown in Figure 4D, for cases in which cells are activated together (left) or in which Cell1 is activated before Cell2 (middle: Δt = 1250 minutes; right: Δt = 2500 minutes). Activation timing affects the system’s behavior mainly in the region where both cells are above their commitment threshold (upper right quadrant). If m1 > m2, pre-activation of Cell1 leads to its dominance together with inhibition of Cell2, in a range of m values which increases with the activation time difference. Conversely, if m1 < m2, the region of dominance of Cell2 shrinks the later it is activated. However, if both cells are below commitment threshold (lower left quadrant), the later activation of Cell2 allows for its bystander commitment in a somewhat larger parameter range, as more IL-2 accumulates due to the pre-activation of the sub-threshold Cell1. These effects of time of activation do not occur if there is no feedback or if the feedback is linear. In those cases, the cell with higher m value wins and excludes the other cell regardless of the relative activation time of the two cells (see Additional file 1 for detailed analysis).
A specific scenario is the interaction between cells with the same m value. When the cells are activated at the same time, interaction between two cells that are slightly below commitment threshold leads to their commitment (See Additional file 1). In this case, both cells profit from the increased level of IL-2 provided by the larger cell concentration, akin to quorum sensing. A more complex behavior emerges when the two cells are activated at different times (See Additional file 1: Figure S14 for a detailed analysis). If the m value of both cells is slightly below threshold, the later activated T-cell can utilize the IL-2 secreted by the earlier activated T-cell and consequently become committed while the first remains uncommitted. If the cells’ m value is above threshold, there is a competition between the T-cells that could lead to exclusion of either the earlier cell (for m values slightly above threshold) or the later cell (for intermediate m values). This example emphasizes our previous results showing that under non-linear extracellular positive feedback, intercellular interactions can lead to non-trivial behaviors, which depend strongly on the timing of activation.
Time dependence was observed also for suppression of effector T cells by Tregs . We simulate this situation using our model showing explicitly how suppression by Tregs is modulated by relative activation time for all combinations of activation levels of the two cells (Additional file 1: Figure S15).
Multicellular interactions with a larger number of cells
So far, we have considered interaction between two cells. Yet, the features we have demonstrated are more general and not limited to the two cell case. As an example, we study the interaction between an ensemble of 9 cells with uniformly distributed m values in the range 0–1 and a cell with m = 1, denoted as Cell1. We calculate the steady-state level of Cell1’s bound IL-2R for a large number of cases where the m values of the other 9 interacting cells are randomly drawn from the uniform distribution. To study the effects of timing, we examine a case in which all the cells are activated together and a case in which Cell1 is activated with a time delay.
The composition of the competing ensemble affects the response of Cell1. As the mean TCR strength of the competitors is larger (Figure 5B, 5C) the level of IL-2R on Cell1 decreases. When the delay is long, exclusion may occur only if the mean m of the competing group is high enough (above ~ 0.5). Another condition for exclusion is that the strongest competitor is strong enough (above ~ 0.75) (Figure 5D, 5E). Thus, although Cell1 has been activated as strongly as possible, it could be excluded by an ensemble of inferior cells if it is activated too late.
Recent work [16, 17] described IL-2 mediated suppression of effector T cells (Teff) by T regulatory cells (Treg), showing the conditions under which competition for IL-2 serves as an efficient suppression mechanism. Our model examines the effect of IL-2 mediated extracellular feedback on the interaction between effector T cells during their activation. While competition is the prominent case in Teff-Treg interactions, it is only one of the scenarios that Teff-Teff interactions can give rise to. In this polyclonal interactions scenario, both cells secrete IL-2, and thus both cells can also benefit from the IL-2 that their neighbors secrete. While the equations regarding kinetics of IL-2 and its receptor are naturally similar, there are a number of significant differences. In particular, a Treg-Teff interaction is non-symmetric, as Tregs express an initial high level of CD25, and do not secrete IL-2. We study the symmetric case of interaction between two effector cells, where both cells are governed by the same equations and the only difference is in the strength of TCR signal. This situation models the abundant case of polyclonal interactions that occur in vivo between T cells with different TCR specificities. These interactions play an important part in shaping T cell responses through inter-clonal interactions. Our model presents a first mathematical/computational attempt to study the role of IL-2 extracellular feedback in this context.
The significance of examining such a scenario is clear from a biological vantage point, as this type of interaction naturally occurs during T cell activation, when cells with different TCR specificities are activated within the same microenvironment, e.g. surrounding the same antigen presenting cell . In effect, many T cell interactions are polyclonal Teff-Teff interactions – most T cell mediated immune responses start with polyclonal T cell activation. Thus, it is important to understand T cell interactions within this context, as it shapes the repertoire of the immune response.
T cell interactions mediated by IL-2 extracellular feedback can result in opposite outcomes: competition or cooperation
We show that intercellular interactions mediated by IL-2 and amplified by the extracellular positive feedback on IL-2R levels can account for both competition and cooperation between activated T cells. Experimental evidence for competition and cooperation between T cells has been available for several decades [25, 26], and a role for IL-2 in mediating competition has been suggested . There are other factors that influence polyclonal T cell interactions at early stages of activation, such as competition over peptide-MHC (pMHC) complexes, limited co-stimulation by the antigen-presenting cell (APC) or limited physical access of T-cells to the APC . However, several experiments have demonstrated competition and cooperation in setups lacking these factors [20, 22]. Consequently, it has been suggested that secreted factors such as cytokines can play a role in driving intercellular competition and cooperation. Furthermore, both cooperation  and competition  were observed in experiments in which T cells of two TCR transgenic populations, with different specificities, were transferred into recipient mice and were activated by corresponding cognate peptides. Use of two different antigens presented by different MHC molecules ensured that competition was not over pMHC complexes. Though the two systems are different in some of the experimental details, one may still wonder under what conditions either outcome, cooperation or competition, is observed.
An even more direct scenario was described in a recent report which evaluated interactions between endogenous T cells responding to different mixtures of peptides . T cells responding to weakly binding pMHC complexes were inhibited in the presence of competitive T cells, whereas T cells responding to strongly binding complexes gained from the presence of the same competing cells . This demonstrates that even within the same experimental setup, opposite qualitative outcomes can be reached, by changing quantitative parameters of the T cells’ activation. Our model is in line with these experimental observations, as it predicts that in the presence of IL-2 extracellular feedback, strongly activated cells will benefit from interactions with average competitors, while cells that are weakly activated will suffer from such interaction. Thus, our model can resolve the seemingly contradicting experimental outcomes demonstrated for T cell interactions in those experiments, by providing a mathematical description for the complex relationship between relative TCR activation strength of the interacting cells and outcome of the interaction. Such inter-clonal interactions are of importance in many sub-fields of immunology, including vaccine design, cancer immunology, allergy and others.
Recently, multiphoton microscopy experiments have demonstrated that there are three stages in the process of in-vivo T cell activation and commitment to proliferation. In the first stage, T cells that arrive asynchronously into the lymph node scan residing APCs for their target peptide. If a cognate peptide is found the T cells undergo activation, stop at the APC and stay within the lymph node. In the last stage, T cells undergo proliferation and migrate out of the lymph node. Thus, at the second stage T cells arrest for 10–20 hours on the presenting APC [24, 28–30] and secrete IL-2 to the cell’s microenvironment . In addition, T cells form clusters around APCs at this stage, facilitating intercellular interactions among T cells. Thus, this might be the stage in which the interaction we model takes place, between T-cells activated by the same or by a nearby APC.
Our model provides some predictions regarding the roles of IL-2 extracellular feedback in determining the outcome of inter-clonal T cell interactions, which can be tested experimentally. One possibility is to use an in vitro culture system in which T cells of two different TCR specificities will be co-cultured and allowed to interact. For example, T cells from two different strains of TCR transgenic mice can be co-cultured and activated by dendritic cells that present both cognate antigens. Such a system provides a controlled setup for varying parameters of the interaction and measuring their effect. One prediction of the model is that addition of external IL-2 to the growth medium will decrease competition, while depletion of IL-2 can enhance the range of competition between the two cell types. Using such a system it can also be possible to directly test model predictions regarding the effect of time of activation, as one type of cells can be introduced into the culture at later time points. Model predictions can also be tested in vivo, for example by injecting IL-2 to increase its availability during the response in the mouse models used in [19, 20, 22]. A similar approach was recently used in  to reduce suppression by regulatory T cells. Our model predicts that excessive levels of IL-2 will reduce inter-clonal competition in those in vivo systems.
Extracellular feedback can balance speed and accuracy
Our model further predicts that the outcome of interactions depends on the relative activation time of the interacting cells. Thus, a stronger cell will exclude its inferior competitor if it is activated either before or alongside with it, but it will coexist with the latter if it is activated too late. Time dependency of intercellular interactions has been observed experimentally. Competition between T cells of different specificities was observed in-vivo only when the competing cells were introduced to the animal within a time window of several hours post introduction of the first T cell population, otherwise the two populations coexisted .
We suggest that such time dependency may serve to balance two opposite needs of early T cell responses: the need to employ the strongest reaction, and the need to act swiftly against a proliferating pathogen. Employing the strongest reaction in a polyclonal system is a matter of accuracy; that is choosing the clone (or small set of clones) that can best recognize the pathogen. T cells with random TCR specificities continuously scan the body in search of antigens, entering and leaving lymph nodes. A T cell that has engaged its cognate antigen will arrest on the APC (as mentioned above). This offers an opportunity to wait for other T cells, in case a better cell arrives, one that has a higher affinity to the same or related antigens. It can be beneficial to allow the stronger cell to exclude the weaker cell, in a way that will lead to affinity maturation of the T cell response. However, this waiting time should be limited in order to exert the immune response as quickly as possible. Our model shows how extracellular feedback can allow for balancing speed versus accuracy of early T cell responses, by providing a limited window of opportunity for cells to compete. We show that this window of opportunity is shorter the stronger the TCR engagement: thus a more suitable cell allows less time for other cells to compete with it, and it is activated faster than less suitable cells. Conversely, a less suitable cell will wait longer for a better cell to arrive (Figure 4C).
We use a simplified model of the system, aiming to gain understanding of general properties of the extracellular positive feedback loop. Thus, we do not explicitly model STAT5 and its phosphorylation, assuming that the level of pSTAT5 is in quasi-equilibrium with the level of B, such that the receptor synthesis rate directly depends on the level of B. In addition, the IL-2R has three subunits (α,β,γ). Feedback by IL-2 only enhances production of the α subunit. As this subunit is expressed at a much higher level on the surface of activated T cells compared to the other two subunits, it is assumed to govern the cellular response to IL-2. We find that as long as the level of the IL-2Rβ subunit does not change considerably, the system can be accurately described by our simplified model, and the rate constants of a more detailed model, that includes all 3 receptor subunits, are close to those used in our simulations (see Additional file 1: Figures S12 and S13). Our simplified approach allows for a more intuitive understanding of the general properties of the IL-2 feedback system, and for generalization to other systems of extracellular positive feedback.
Other factors can also influence the outcome of the interaction, for example different initial levels of IL-2 receptors expressed by the cells, or other differences in the signaling capacity of the IL-2 and TCR signaling pathways . Such differences can result from stochasticity in gene expression [34, 35]. As a demonstration of these effects, we present in Additional file 1: Figure S18 the results of a simulation of our model for the three cases shown in Figure 3A, but in which the initial number of receptors expressed by the interacting cells is noisy. The two cases showing cooperation (cases I, II) are not sensitive to this change in initial conditions, and the stochastic results are the same as the deterministic ones. However, for the case of competition and exclusion (case III), introduction of noise leads to a bi-modal distribution, where the deterministic results are obtained in ~70% of the cases, while in the rest there is no exclusion. Our current deterministic model can be extended into a stochastic model to account for such differences between cells that result from stochasticity, for example by using the Gillespie algorithm  to model all the reactions.
We show that the non-linearity of the feedback is crucial for obtaining a wide range of possible outcomes for intercellular interactions. A linear feedback would result in a “winner-takes-all” scenario, where the cell with the higher level of TCR signal will out-compete and exclude its inferior neighbor, regardless of the differences in their activation levels or their activation times. This scenario is similar to that studied in population dynamics of two populations competing for a shared resource .
Our results are obtained under the assumption of a well-mixed environment, implying that a cell secreting IL-2 does not have a preference in using it compared with its neighbor. This issue was discussed before , where it is shown that for the specific kinetic values of the IL-2 system, autocrine signaling is not dominant and a well-mixed environment is a good approximation. The actual mode of transport of IL-2 inside lymph nodes has not been established and further experimental investigation is required for gaining a better understanding of the transport properties, which can have important influence on IL-2 mediated intercellular interactions. Our model can be extended to include different modes of IL-2 transport, e.g. diffusion , non-isotropic IL-2 secretion  and also specific cell-cell geometries such as IL-2 secretion into low volume synapses between clustering T cells, as was recently observed in vitro and in vivo .
When relating our model to experimentally measurable parameters, two points should be considered. First, we use a normalized parameter m to describe the level of TCR signaling. In practice, TCR signaling level can be varied, for example, by changing antigen concentration or using signaling inhibitors such as cyclosporine. One can relate the experimental variable to m, by measuring the levels of IL-2Rα (CD25) for varying levels of TCR stimulation, and using the resulting curve for normalization. To avoid effects of extracellular feedback, this calibration would be preferably done with cells lacking IL-2 secretion (e.g. from IL2 knockout mice), in the presence of a constant level of IL-2. Second, we used B (number of bound receptors) as an indicator of cell state. Experimentally, B is difficult to measure while related parameters such as the level of pSTAT5 or total receptor level (B+U) are more accessible. It was recently shown experimentally  that the level of B is linearly related to the level of pSTAT5, which is the main signaling molecule downstream of the IL2R complex. Thus, B is a good estimate for the level of IL2 induced signaling. In Additional file 1: Figure S19 we present the results of our model in terms of the total number of IL2R (B+U). We note that in contrary to B, the total level of receptor does not increase in the region of cooperation (around point I in Figure 3A), due to the different internalization times of the unbound and bound receptor.
Generality of the model
Finally, we note that similar feedback and consumption mechanisms apply for cytokines other than IL-2. For example, the cytokine IL-4 increases expression levels of its own receptor, while cytokine-bound receptors are internalized and degraded . Thus, our model can be applied also to other cytokine mediated intercellular interactions, such as cytokine-driven differentiation of CD4+ cells [40, 41], or differentiation of CD4+ T cells into memory cells, in which competition for IFNγ has been suggested to play a role . Taken together, our model offers a general framework for studying IL-2 induced commitment to proliferation of T cells as a collective process. This can apply not only to the case of interactions between unprimed cells, but also to interactions that involve effector and regulatory cells (see Additional file 1), or interactions which include memory T cells. The latter have been recently shown to drive bystander commitment of CD4+ T cells, potentially involving IL-2 [43, 44].
We suggest a role for extracellular feedback by studying interactions between activated polyclonal T cells as a model system. Using mathematical modeling we find that extracellular feedback can give rise to opposite outcomes of the interaction – either competition or cooperation between the interacting cells, and that this outcome depends on the parameters of the activation. In addition, the interaction outcome depends also on the timing of activation of the cells. As a result, a critical time window exists after which a stronger cell cannot exclude an inferior competitor, as was experimentally observed. We suggest that extracellular feedback can balance a speed vs. accuracy tradeoff in T cell activation – choosing the most suitable responder under the clock of a proliferating threat. These findings can serve to better understand the way the repertoire of an immune response is shaped, in response to pathogens, vaccines and tumor detection.
Co-first authors: Yonatan Savir and Nir Waysbort.
We thank Uri Alon, Benjamin Chain, Ofer Feinerman, Ron Milo, Wilfred Ndifon and Jacob Rimer for helpful comments on the manuscript. This research was supported by The International Human Frontier Science Program Organization and by the Abisch-Frenkel Foundation. TT thanks the Simons Foundation, and Helen and Martin Chooljian for their generous support. TT is the Helen and Martin Chooljian Founders' Circle Member in the Simons Center for Systems Biology at the Institute for Advanced Studies, Princeton. NF is incumbent of the Pauline Recanati Career Development Chair of Immunology.
- Sneppen K, Krishna S, Semsey S: Simplified Models of Biological Networks. Annual review of biophysics. 2010, 39: 43-59. 10.1146/annurev.biophys.093008.131241.View ArticleGoogle Scholar
- Crespi BJ: Vicious circles: positive feedback in major evolutionary and ecological transitions. Trends in ecology & evolution (Personal edition). 2004, 19: 627-633.View ArticleGoogle Scholar
- Murray JD: Biological Oscillators and Switches. Mathematical Biology. Edited by: Antman SS, Marsden JE, Sirovich L, Wiggins S. 2002, Springer, New York, ThirdGoogle Scholar
- Ozbudak EM, Thattai M, Lim HN, Shraiman BI, Van Oudenaarden A: Multistability in the lactose utilization network of Escherichia coli. Nature. 2004, 427: 737-740. 10.1038/nature02298.View ArticleGoogle Scholar
- Brandman O, Meyer T: Feedback loops shape cellular signals in space and time. Science (New York, N.Y.). 2008, 322: 5-View ArticleGoogle Scholar
- Becskei A, Serrano L: Engineering stability in gene networks by autoregulation. Nature. 2000, 405: 590-593. 10.1038/35014651.View ArticleGoogle Scholar
- Hornung G, Barkai N: Noise propagation and signaling sensitivity in biological networks: a role for positive feedback. PLoS computational biology. 2008, 4: e8-10.1371/journal.pcbi.0040008.View ArticleGoogle Scholar
- Legewie S, Herzel H, Westerhoff HV, Blüthgen N: Recurrent design patterns in the feedback regulation of the mammalian signaling network. Mol Syst Biol. 2008, 4: 190-View ArticleGoogle Scholar
- Shvartsman SY, Wiley HS, Deen WM, Lauffenburger DA: Spatial range of autocrine signaling: modeling and computational analysis. Biophys J. 2001, 81: 1854-1867. 10.1016/S0006-3495(01)75837-7.View ArticleGoogle Scholar
- Smith KA: Interleukin-2: inception, impact, and implications. 1988, Science, New York, N.Y, 1169-1176.Google Scholar
- Kim HP, Imbert J, Leonard WJ: Both integrated and differential regulation of components of the IL-2/IL-2 receptor system. Cytokine & growth factor reviews. 2006, 17: 349-366. 10.1016/j.cytogfr.2006.07.003.View ArticleGoogle Scholar
- Burchill MA, Yang J, Vang KB, Farrar MA: Interleukin-2 receptor signaling in regulatory T cell development and homeostasis. Immunol Lett. 2007, 114: 1-8. 10.1016/j.imlet.2007.08.005.View ArticleGoogle Scholar
- Hémar A, Subtil A, Lieb M, Morelon E, Hellio R, Dautry-Varsat A: Endocytosis of interleukin 2 receptors in human T lymphocytes: distinct intracellular localization and fate of the receptor alpha, beta, and gamma chains. The Journal of cell biology. 1995, 129: 55-64. 10.1083/jcb.129.1.55.View ArticleGoogle Scholar
- Fallon EM, Lauffenburger DA: Computational model for effects of ligand/receptor binding properties on interleukin-2 trafficking dynamics and T cell proliferation response. Biotechnol Prog. 2000, 16: 905-916. 10.1021/bp000097t.View ArticleGoogle Scholar
- Forsten KE, Lauffenburger DA: The role of low-affinity interleukin-2 receptors in autocrine ligand binding: alternative mechanisms for enhanced binding effect. Mol Immunol. 1994, 31: 739-751. 10.1016/0161-5890(94)90148-1.View ArticleGoogle Scholar
- Feinerman O, Jentsch G, Tkach KE, Coward JW, Hathorn MM, Sneddon MW, Emonet T, Smith FA, Altan-Bonnet G: Single-cell quantification of IL-2 response by effector and regulatory T cells reveals critical plasticity in immune response. Mol Syst Biol. 2010, 6: 437-View ArticleGoogle Scholar
- Busse D, de La Rosa M, Hobiger K, Thurley K, Flossdorf M, Scheffold A, Höfer T: Competing feedback loops shape IL-2 signaling between helper and regulatory T lymphocytes in cellular microenvironments. Proc Natl Acad Sci U S A. 2010, 107: 3058-3063. 10.1073/pnas.0812851107.View ArticleGoogle Scholar
- Kedl RM, Kappler JW, Marrack P: Epitope dominance, competition and T cell affinity maturation. Curr Opin Immunol. 2003, 15: 120-127. 10.1016/S0952-7915(02)00009-2.View ArticleGoogle Scholar
- Creusot RJ, Thomsen LL, Tite JP, Chain BM: Local cooperation dominates over competition between CD4+ T cells of different antigen/MHC specificity. Journal of immunology (Baltimore, Md.: 1950). 2003, 171: 240-246.View ArticleGoogle Scholar
- Weaver JM, Chaves FA, Sant AJ: Abortive activation of CD4 T cell responses during competitive priming in vivo. Proc Natl Acad Sci U S A. 2009, 106: 8647-8652. 10.1073/pnas.0811584106.View ArticleGoogle Scholar
- Sojka DK, Hughson A, Sukiennicki TL, Fowell DJ: Early kinetic window of target T cell susceptibility to CD25+ regulatory T cell activity. Journal of immunology (Baltimore, Md.: 1950). 2005, 175: 7274-7280.View ArticleGoogle Scholar
- Willis RA, Kappler JW, Marrack PC: CD8 T cell competition for dendritic cells in vivo is an early event in activation. Proc Natl Acad Sci U S A. 2006, 103: 12063-12068. 10.1073/pnas.0605130103.View ArticleGoogle Scholar
- Lin JX, Leonard WJ: The role of Stat5a and Stat5b in signaling by IL-2 family cytokines. Oncogene. 2000, 19: 2566-2576. 10.1038/sj.onc.1203523.View ArticleGoogle Scholar
- Miller MJ, Safrina O, Parker I, Cahalan MD: Imaging the single cell dynamics of CD4+T cell activation by dendritic cells in lymph nodes. The Journal of experimental medicine. 2004, 200: 847-856. 10.1084/jem.20041236.View ArticleGoogle Scholar
- Al-Adra AR, Pilarski LM: Antigen-specific suppression of cytotoxic T cell responses in mice. I. Suppressor T cells are not cytotoxic cells. Eur J Immunol. 1978, 8: 504-511. 10.1002/eji.1830080710.View ArticleGoogle Scholar
- Gerloni M, Xiong S, Mukerjee S, Schoenberger SP, Croft M, Zanetti M: Functional cooperation between T helper cell determinants. Proc Natl Acad Sci U S A. 2000, 97: 13269-13274. 10.1073/pnas.230429197.View ArticleGoogle Scholar
- Günther J, Haas W, Von Boehmer H: Suppression of T cell responses through competition for T cell growth factor (interleukin 2). Eur J Immunol. 1982, 12: 247-249. 10.1002/eji.1830120315.View ArticleGoogle Scholar
- Mempel TR, Henrickson SE, Von Andrian UH: T-cell priming by dendritic cells in lymph nodes occurs in three distinct phases. Nature. 2004, 427: 154-159. 10.1038/nature02238.View ArticleGoogle Scholar
- Beuneu H, Lemaître F, Deguine J, Moreau HD, Bouvier I, Garcia Z, Albert ML, Bousso P: Visualizing the Functional Diversification of CD8(+) T Cell Responses in Lymph Nodes. Immunity. 2010, 33: 412-423. 10.1016/j.immuni.2010.08.016.View ArticleGoogle Scholar
- Shakhar G, Lindquist RL, Skokos D, Dudziak D, Huang JH, Nussenzweig MC, Dustin ML: Stable T cell-dendritic cell interactions precede the development of both tolerance and immunity in vivo. Nat Immunol. 2005, 6: 707-714. 10.1038/ni1210.View ArticleGoogle Scholar
- Henrickson SE, Mempel TR, Mazo IB, Liu B, Artyomov MN, Zheng H, Peixoto A, Flynn MP, Senman B, Junt T, Wong HC, Chakraborty AK, von Andrian UH: T cell sensing of antigen dose governs interactive behavior with dendritic cells and sets a threshold for T cell activation. Nat Immunol. 2008, 9: 282-291. 10.1038/ni1559.View ArticleGoogle Scholar
- Couzin ID: Collective cognition in animal groups. Trends in cognitive sciences. 2009, 13: 36-43. 10.1016/j.tics.2008.10.002.View ArticleGoogle Scholar
- Feinerman O, Veiga J, Dorfman JR, Germain RN, Altan-Bonnet G: Variability and robustness in T cell activation from regulated heterogeneity in protein levels. Science (New York, N.Y.). 2008, 321: 1081-1084. 10.1126/science.1158013.View ArticleGoogle Scholar
- Cai L, Friedman N, Xie XS: Stochastic protein expression in individual cells at the single molecule level. Nature. 2006, 440: 358-362. 10.1038/nature04599.View ArticleGoogle Scholar
- Sigal A, Milo R, Cohen A, Geva-Zatorsky N, Klein Y, Liron Y, Rosenfeld N, Danon T, Perzov N, Alon U: Variability and memory of protein levels in human cells. Nature. 2006, 444: 643-646. 10.1038/nature05316.View ArticleGoogle Scholar
- Gillespie D: Exact stochastic simulation of coupled chemical reactions. J Phys Chem. 1977, 81: 2340-2361. 10.1021/j100540a008.View ArticleGoogle Scholar
- Huse M, Lillemeier BF, Kuhns MS, Chen DS, Davis MM: T cells use two directionally distinct pathways for cytokine secretion. Nat Immunol. 2006, 7: 247-255. 10.1038/ni1304.View ArticleGoogle Scholar
- Sabatos C, Doh J, Chakravarti S, Friedman R, Pandurangi P, Tooley A, Krummel M: A Synaptic Basis for Paracrine Interleukin-2 Signaling during Homotypic T Cell Interaction. Immunity. 2008, 29: 238-248. 10.1016/j.immuni.2008.05.017.View ArticleGoogle Scholar
- Renz H, Domenico J, Gelfand EW: IL-4-dependent up-regulation of IL-4 receptor expression in murine T and B cells. Journal of immunology (Baltimore, Md.: 1950). 1991, 146: 3049-3055.Google Scholar
- Zhu J, Yamane H, Paul WE: Differentiation of effector CD4 T cell populations. Annu Rev Immunol. 2010, 28: 445-489. 10.1146/annurev-immunol-030409-101212.View ArticleGoogle Scholar
- Murphy KM, Reiner SL: The lineage decisions of helper T cells. Nat Rev Immunol. 2002, 2: 933-944. 10.1038/nri954.View ArticleGoogle Scholar
- Whitmire JK, Benning N, Eam B, Whitton JL: Increasing the CD4+ T Cell Precursor Frequency Leads to Competition for IFNγ Thereby Degrading Memory Cell Quantity and Quality. J Immunol. 2008, 180: 6777-View ArticleGoogle Scholar
- Di Genova G, Roddick J, McNicholl F, Stevenson FK: Vaccination of human subjects expands both specific and bystander memory T cells but antibody production remains vaccine specific. Blood. 2006, 107: 2806-2813. 10.1182/blood-2005-08-3255.View ArticleGoogle Scholar
- Di Genova G, Savelyeva N, Suchacki A, Thirdborough SM, Stevenson FK: Bystander stimulation of activated CD4+T cells of unrelated specificity following a booster vaccination with tetanus toxoid. Eur J Immunol. 2010, 40: 976-985. 10.1002/eji.200940017.View ArticleGoogle Scholar