Kinetics and Extent of T Cell Activation as Measured with the Calcium Signal

Peptides were synthesized with standard FMOC chemistry and purified by reverse phase HPLC. Anti-CD4 antibody GK1.5 was purchased from Becton-Dickinson (Mountain View, CA).

The moth cytochrome c (MCC)-specific 5C.C7 T cells were derived from the 5C.C7 TCR transgenic mouse (37, 38). Spleen cells were primed with 10 μM MCC peptide and expanded for 8 d in IL-2 containing medium. The T cells (1 × 10⁶ cells/ml) were restimulated with irradiated (3,000 rads) B10.BR spleen cells and 10 μM MCC peptide. After 10 d of culture and expansion in the presence of IL-2, CD4⁺ cells were sorted by flow cytometry and restimulated in 10-d intervals.

For calcium imaging, 5C.C7 T cells were loaded for 30 min at room temperature with 2 μM Fura-2 in 10% FCS in PBS with 1 mM CaCl₂ and 0.5 mM MgCl₂. Chinese hamster ovary (CHO) cells transfected with I-E^k were grown in 8-well coverslips (Nunc, Naperville, IL) and peptide loaded with 25 μM peptide for 4–16 h. Peptide loading efficiency throughout this time range was almost constant. After washing unbound peptide away with 5% FCS in PBS with 1 mM CaCl₂ and 0.5 mM MgCl₂, Fura-2–loaded T cells were added onto a patch of confluent CHO cells on the microscope stage to start the experiment. Intracellular calcium in T cells was determined using C-Imaging1280 System hardware and the Simca Quantitative Fluorescence Analysis software package; both from Compix, Inc., Imaging Systems (Mars, PA). The imaging system was coupled to a Nikon Diaphot 300 inverted microscope, which was equipped with the epifluorescence attachment and a 75W Xenon arc lamp. Alternate excitation of Fura-2–loaded T cells at 340 and 380 nm was achieved using a Ludl high speed dual filter wheel (Ludl Electronic Products, Hawthorne, NY) controlled by the Simca software. Images were collected with a CCD camera (Dage-MTI CCD72) in combination with a SuperGenII intensifier, (DageMTI, Michigan City, IN) to amplify fluorescence. Analysis of intracellular calcium and the generation of figures was achieved with the Simca software package. Per run, 30 to 40 cells were picked at random and analyzed. Data of different runs were pooled. For the figures, data of different experiments in which the topic of the figure was studied in direct comparison of the conditions under question was pooled. Only the 102S titration series (part of Figs. 2 and 3) shows a single representative experiment. The fluorescence of the Fura-2–loaded T cell is only visible when the T cell is right on top of the CHO layer. Because the T cells are added to the chamber from the top, their appearance in the fluorescence image is taken as the time of first contact.

One signaling component, A, will be converted into another, B, with the rate constant k₁, triggered by the MHC–peptide–TCR contact. Both the concentration of peptide and the type of peptide, as described by the half-life of its MHC– peptide–TCR complex (28–30), influence k₁. Because partial agonist and antagonist peptides give a qualitatively different signal the effect on k₁ of changing the peptide type shall be much more drastic than changing the peptide concentration. Using a first order kinetics, and assuming that the concentration of B before the MHC–peptide–TCR contact is 0, the concentration of A is A₀, B would form over time with B(t) = A₀ × (1 − e^(−k1t)). We now suggest that a minimal amount of B has to form to propagate the signal, the trigger concentration B_tr. The time t_tr that is needed to accumulate B_tr is given by: t_tr = −k₁⁻¹ × ln(1 − B_tr/A₀). From this formula it can be seen that the smaller k₁ is, the longer it will take to reach the trigger concentration. However, one problem with this simple description is that even very weak signals would trigger activation, given enough time. As we do not observe this we suggest that, either directly or indirectly, B reverts back to A with the rate constant k₋₁ that for reasons of simplicity is assumed to be independent of the MHC–peptide–TCR contact. The rate of the formation of B then becomes B(t) = A₀ × (1 + k₋₁ / k₁) × (1 − e^(k1+k−1)t). Using this model, the maximum amount of B that is formed depends on the ratio of k₋₁ to k₁. The smaller k₁ is relative to k₋₁, the smaller this maximum level will be. Using weak activation conditions, as very low concentrations of peptide or antagonist peptides by themselves, the trigger level is never reached.

A number of TCRs have been characterized that recognize the MCC peptide (88–103) restricted by I-E^k. Different variants of the peptide have been studied in several respects: (a) peptide binding to I-E^k (39), (b) kinetic rate constants for the binding of the peptide–MHC to the 2B4 and 5C.C7 TCR (30, 33, 40), (c) ability to stimulate IL-3 production, proliferation, and acid release in a 5C.C7 line (36, 39). Taking advantage of this extensive data set, we chose one of the best studied of these models, 5C.C7, to characterize the changes in the concentration of intracellular calcium after T cell activation in response to different stimuli. The particular 5C.C7 line we chose to study is derived from the CD4⁺ cells of the spleen of 5C.C7 TCR transgenic mice (37; Fazekas de St. Groth, B., and M.M. Davis, unpublished results). It is maintained by periodic restimulation with antigen (36). As an APC, we have used CHO cells transfected with I-Ek, which can effectively stimulate a variety of T cells (41, 42).

The calcium response to well-characterized peptides in the MCC family was studied (Fig. 1). For all of them the anchor residues to the MHC are unchanged, and thus binding to I-E^k is very similar (39). Two of the three major TCR contact residues, positions 99 and 102 (39, 43), were changed more or less drastically in different peptides.

The changes in the level of intracellular calcium during T cell stimulation were recorded from single cells for 16– 18 min, using the calcium-sensitive fluorescent dye, Fura-2 (45). In a single cell assay differences in the response within a population of responders can be resolved, allowing more extensive analysis. This proved to be crucial in particular for the characterization of different types of calcium signals. The experiments were done at 25°C. Relative differences in type, frequency, and timing of the calcium signal are the same as at 37°C, but the movement of the cells after activation is considerably less (Wülfing, C., and M.M. Davis, data not shown). This reduction in cell movement at 25°C allows a more precise determination of the form of the calcium signal. In our assays, we first recorded the time between APC–T cell contact and the onset of the calcium signal. Second, the form of the calcium signal after the onset, as judged by the emission of the dye Fura-2 when subsequently excited at 340 nm and 380 nm, as well as the ratio of these emissions, was classified. Analyzing not only the ratio, but also the single emission traces, allows us to distinguish between specific elevation of intracellular calcium and ratio changes caused by the movement of the cells before activation.

Under different activation conditions, we find four types of calcium signals. Each starts with a rapid rise in intracellular calcium. However, the calcium levels after this initial rise vary. Using the MCC peptide, more than 95% of the cells respond with full signal, as shown in Fig. 2 and described in Table 1. With the altered peptides, we find three qualitatively different variants of the full signal, a reduced, partial, and transient signal (Fig. 2; Table 1). These describe the increasingly weaker calcium responses that we see and have predictive value.

Fig. 3 summarizes the T cell calcium signaling triggered by the peptides used in this report. Using the MCC peptide the response is full, as discussed above. For the partial agonist peptide 102S, the frequency of signaling is the same as for the MCC peptide, but for almost a third of the cells the signal is not complete. For the antagonist 102G peptide, both the frequency of signaling and the signal type are significantly different. The antagonist peptide 99R gives no signaling at all above background. The background level is defined using no peptide or the neutral MCC93E99T102A variant, which shows no biological response in acid release and proliferation assays and no measurable affinity to the 2B4 TCR. In both of these cases, we see no full signals and reduced or partial signals in less than 10% of the T cells (Fig. 3).

To determine the effect of low concentrations of stimulating peptide in a defined manner, the MCC and 102S peptides were each serially diluted into the neutral peptide 93E99T102A, such that the total peptide concentration during loading of the APCs is kept constant at 25 μM. As shown in Fig. 3, diluting the MCC peptide 1:1,000, resulting in a 25 nM MCC concentration, has no effect on the frequency or the type of the calcium signal. At a 1:10,000 dilution of the MCC (2.5 nM), more than 85% of the cells still respond with a full signal (see Fig. 5,A). When 102S is diluted into MCC93E99T102A in a dilution series ranging from 25 μM to 25 nM, the signaling frequency throughout the series is still almost 100% (Fig. 3). However, at a 25 nM (1:1,000 dilution), 70% of the cells respond with a reduced signal, as opposed to 30% at 25 μM.

The altered or absent calcium response for the peptides 102S, 102G, and 99R correlates well with the other biological effects of these peptides (Table 2). The loss of extended elevation in the calcium signal in the 102S dilution series correlates with a profound reduction in the ability of 102S to induce proliferation (36). The strongly reduced or absent calcium signal of the 102G and 99R antagonist peptides correlates with their lack of ability to induce any proliferation. As shown by Rabinowitz et al. (36) the acidification response is an early T cell activation event with both MCC and 102S giving identical dose-response curves. This is reflected in a high calcium signal responder frequency for both. On the other hand, the strongly reduced or absent calcium signal for 102G and 99R correlates with the lack of an acidification response. Because acidification is measured in a bulk assay, the small degree of activation measured in a single cell calcium assay, e.g., for the peptide 102G, might not be detectable above background in the acid release assay. Thus, only a full calcium signal correlates with a proliferation response. In contrast, acid release correlates with a high frequency response, irrespective of signal type.

The design of our calcium assay allows us to determine the precise time between first APC–T cell contact and the rise in intracellular calcium. We refer to this as the onset delay. This onset delay is composed of two components. Even under the strongest activation conditions, there is a minimal time delay of ∼1 min for the activation of the first T cells. Using weaker stimuli, we find that the onset delay increases as the stimulus gets weaker (Fig. 4). This delay is seen both in the time it takes to activate the first cell, as well as in the average onset time of the T cells in a given experiment.

For the MCC peptide, the onset delay is the shortest with an average of about 3 min and stays unchanged even at 25 nM MCC (a dilution of 1:1,000 into the neutral peptide 93E99T102A) (Fig. 4). 25 μM 102S or 102S diluted only 1:10 into the neutral peptide (2.5 μM) shows an onset delay that is similar to MCC. However, at lower concentrations there is a significant increase in the onset delay with a shift from about 3 min to 6 min for 25 nM 102S (1:1,000 dilution into the neutral peptide) (Fig. 4). Furthermore, the antagonist peptide, 102G, gives an onset delay of 10 min, in combination with a strongly reduced responder frequency. In general, a full calcium signal shows a rapid onset, a slightly reduced calcium signal shows a small onset delay, and a strongly reduced calcium signal displays a large onset delay. Thus, this delay in T cell activation seems to be an integral part of the response to different ligands.

We also studied the effect of blocking CD4 coreceptor engagement on the 102S and MCC calcium response. In the presence of 100 ng/ml of the anti-CD4 antibody GK1.5, the calcium signal induced by 25 nM MCC or 102S is reduced with respect to the frequency and type of signal (Fig. 5,A). The onset is not significantly delayed (Fig. 5 B). This suggests that CD4 plays a role predominantly after the onset of the calcium signal. The inhibition of the calcium signal by CD4 correlates with the data of Rabinowitz et al. (36), who have shown that GK1.5 blocks proliferation in response to MCC and 102S.

The peptides 99R and 102G have been shown to act as antagonists in cytokine secretion and proliferation assays (Reay et al., in preparation; 36). Furthermore, the partial agonist 102S can be changed into an antagonist in the presence of 100 ng/ml of the anti-CD4 antibody (36). To assess the effect of antagonists on the calcium response, the MCC peptide was serially diluted into the three different antagonist conditions. For a 1:10,000 dilution of MCC into 102S/GK1.5, resulting in 2.5 nM MCC concentration, 10% of the cells do not respond at all, and 30% respond with a reduced calcium signal (Fig. 6,A). This is essentially the same signal seen with the 102S peptide and GK1.5 alone without any MCC (see Fig. 5). A 1:10,000 dilution of the MCC into 99R and 102G leads to a reduction in the frequency of responding cells to less than 40%, and almost all calcium signals are partial (Fig. 6,B). Thus, the normal MCC calcium response can be efficiently antagonized under limiting activation conditions. In all three cases, the antagonist effect depends strongly on the concentration of the agonist, as MCC diluted to 1:1,000 shows only a small antagonistic effect, as judged by the slight reduction in type and frequency of the calcium signal (Fig. 6,A). Interestingly, however, all three antagonist conditions delay the onset of the calcium signal, ranging from slightly (about 1 min larger) to more than twice as long (Fig. 6 B).

We show here that the form of the intracellular calcium traces can vary widely depending on the activation conditions. Under fully activating conditions, the calcium trace shows a rapid rise followed by a smooth plateau. Under suboptimal activation conditions, the intracellular calcium concentration fluctuates considerably more after the initial rise (Fig. 2). Other investigators have described single cell calcium traces for peptide-dependent T cell activation at high concentrations of agonist peptide or with cross-linking antiCD3 antibody. They found either oscillations (9, 11, 14, 46) or smooth calcium traces (47), or a combination of both forms (12). While these different results may be due to different ligands used, an alternative explanation relates to the degree of transformation of the T cells. When we compare calcium traces from 5C.C7 naive cells, T cells primed once, multiply primed T cells, i.e., the line used in this publication, and an restimulation-independent 5C.C7 T cell line, we find that the incidence of oscillations increases with the degree of transformation of the T cells (Wülfing, C., and M.M. Davis, in preparation). Consistent with this, the studies that show the most oscillatory T cell calcium traces (14, 46) both used hybridomas.

In the system used here, we find that the elevation of intracellular calcium varies from the full response to various reduced or oscillatory patterns and that this correlates with other bioassays as summarized in Table 2. A positive result in the acidification assay is always accompanied by some calcium signaling, whereas a proliferative response correlates with a full calcium signal in most of the cells. Even a small disturbance of the full signal, such as a delayed onset or a response from less than 100% of the T cells, correlates with an inability of the stimulus to induce proliferation of the whole population. This suggests that the need for paracrine help may require the proliferation of the large majority of the individual cells in a population. Furthermore, other intracellular signaling pathways that are necessary for full T cell activation may be activated independent of the calcium signal to some extent. Nevertheless, the importance of a sustained calcium signal has been shown by Timmerman et al. (48), who have demonstrated that only a sustained calcium elevation signal is able to keep NF-ATc in the nucleus. Similarly, Negulescu et al. (46) have shown that a sustained elevation of the intracellular calcium concentration is necessary for the activation of an IL-2 promoter reporter construct. From the data of Karttunen and Shastri (49) and Valitutti et al. (50), it also appears that a full calcium stimulus must be maintained for at least 30–60 min to induce an irreversible commitment to T cell activation, particularly at lower peptide concentrations. Given the importance of the sustained phase of the calcium signal, a small reduction might contribute significantly to a severe reduction in proliferation.

During activation, the T cell receives a wide variety of signals through a large number of individual receptors that have to be integrated within the signaling pathway to allow the cell to decide whether to proliferate or not. Our data allows us to estimate to what extent the calcium signal is already integrated. The pattern of elevation of the intracellular calcium elicited by different peptides is not only quantitatively different, but qualitatively as well. Antagonism is more effective at a higher dilution of the agonist peptides into the antagonist. A blocking anti-CD4 antibody reduces the calcium signal. These three observations suggest that a T cell has counted and qualified the different signals through TCR and CD4 before the elevation of the intracellular calcium concentration. In the process of signal integration, the calcium signal thus seems to be much closer to the final proliferation decision than to the events elicited directly by the single engaged receptors.

With respect to TCR antagonist ligands, recently Sloan-Lancaster et al. (47) have shown that several antagonist ligands can induce weak calcium fluxes and suggest that this might be part of an antagonist signal. Our own data shows that whereas two different antagonist stimuli (102G alone or 102S + αCD4 antibody) generate detectable, but reduced calcium responses, one (99R) does not. This indicates that the mobilization of intracellular calcium is not necessary for antagonist peptide–MHC ligands to block a response. Thus, their primary effect may lie upstream in the T cell signaling process. Similarly, Windhagen et al. (26) have described an antagonistic peptide that does not induce a calcium signal. Also consistent with a more TCR proximal effect of antagonism are the data of Sloan-Lancaster et al. (32) and Madrenas et al. (31) showing alterations in CD3-γ phosphorylation and subsequent ZAP-70 association as a consequence of antagonist ligand engagement.

In this report, we determined the time between initial T cell–APC contact and onset of the calcium response. Comparing all activation conditions, we find that a minimal delay of 45 s is seen for the MCC peptide at high as well as low concentrations. Therefore, this initial delay must reflect events that are independent of the stimulus strength. Likely factors in this time frame are the time required for the establishment of cell–cell adhesion and for the transmission of intracellular signals from the membrane to the calcium stores.

Weak stimuli, such as partial agonist or antagonist peptides, result in an additional onset delay of up to 10 min. Recent studies have demonstrated that, in most cases, partial agonists and antagonists dissociate more rapidly from a given TCR than agonist ligands (28–30). Therefore, most altered peptide ligands should reach steady-state with respect to TCR binding more rapidly than agonist ligands. Furthermore, one characteristic of the onset delay is that large changes in the peptide dose result in only small changes in the activation time (Fig. 4). For example, going from 250 nM to 25 nM of the peptide 102S causes only a 40% increase in the onset time. If the formation of individual TCR–peptide–MHC pairs was the rate-limiting step in the induction of the calcium signal there would be, for concentrations of free MHC–peptide smaller than the concentration of free TCR, a linear increase in the onset time as the peptide concentration decreases. If the formation of TCR–peptide–MHC oligomers was critical there would be an even more dramatic increase in the onset time as a function of peptide concentration. Even more sophisticated models in which TCR–peptide–MHC mono- or oligomers need to interact for some minimal time in order for signaling to occur do not fit the observed peptide dependency of the onset delay (Kantor, R., C. Wülfing, and M.M. Davis, unpublished results). Thus, the increased onset time observed for weak stimuli does not appear to depend on some threshold level of TCR ligation or oligomerization. This is not to say that oligomerization does not occur or that the half-life of the TCR–peptide–MHC complex is not important, as other studies have shown that it is (30, 33), but that these factors may not be rate-limiting at the cellular level.

Instead, we suggest that the rate-limiting factor in this delay reflects intracellular reactions leading to the accumulation of signaling molecules. These reactions could involve a continuous competition between phosphorylation and dephosphorylation, or between production or breakdown of other parts of the signaling machinery. We hypothesize that T cell activation shifts the balance of this equilibrium towards the accumulation of signaling molecules. A simple kinetic model based on a reaction scheme of this type agrees well with the onset delay that we observe (Fig. 7). The crucial feature of this scheme is that the turnover of the rate-limiting signaling intermediate is determined by an activation strength-dependent first-order forward and an activation strength independent first-order reverse reaction:

The concentration over time of this signaling intermediate B is given by B(t) = A₀ × (1 + k₋₁ / k₁)⁻¹ × (1 − e^{(k1 + k−1)t}) (see Materials and Methods for details).

The half-life of these reactions can be derived from our kinetic data. Under the very weak activation conditions of the antagonist 102G most calcium responses occur between 8–12 min. Because only a small percentage of the cells respond, the reaction leading to signaling must be nearly complete (i.e., at the new steady-state) before any responses occur. This is depicted in Fig. 7,C. Because the time to reach 95% of a new steady-state level of signaling intermediates is four times the half-time, 2 min is a good approximation for the half-time of the reaction towards the steady-state. At the same time, this is the lower limit for the half-times of the individual forward and reverse reactions of this steadystate. The half-times of this stimulus-independent reverse reaction and this forward reaction for 102G are a few minutes under the additional assumption that to trigger the calcium response the accumulation of the signaling intermediate has to be substantial but not complete (Fig. 7 C). The half-times of the forward reaction for stronger activation conditions would be shorter. As these measurements were done at 25°C, one must also allow for a two to four times acceleration in this value at 37°C. In a broader biological picture, we thus suggest that the accumulation of a signaling intermediate triggers the calcium signal. This intermediate is continuously degraded with a half-time of a few minutes. Therefore, every T cell stimulus that results in calcium signaling has to produce this intermediate with a halftime of less than a few minutes.

Other early processes in T cell activation occur within a similar time frame. One example is ζ chain phosphorylation (31, 32), which reaches steady-state within a few minutes for agonist signaling and within 2–10 min for antagonist signaling (32). Such a reaction could produce the postulated critical signaling intermediate.

We thank S. Palmieri from the Stanford Cell Science Imaging Facility for help and advice.