One lyn molecule is sufficient to initiate phosphorylation of aggregated high-affinity IgE receptors

Abstract

In response to antigenic stimuli, the multisubunit immune recognition receptors become aggregated and then phosphorylated on their cytoplasmic tyrosines. For the clonotypic receptors of B and T cells and for Fc receptors such as the high-affinity receptor for IgE (FcepsilonRI), a Src family kinase initiates this phosphorylation. We ask whether aggregation of the initiating kinase itself is required for signal transduction or whether, alternatively, a single associated kinase molecule can phosphorylate the receptors in an aggregate. We formulate the alternative molecular mechanisms mathematically and compare predictions with experimental findings on FcepsilonRI-bearing cells expressing varying amounts of the transfected Src family kinase Lyn. The data are consistent with the requirement of only a single Lyn molecule per FcepsilonRI aggregate to initiate signaling and are inconsistent with a mechanism requiring more than one Lyn molecule.

Figure 1

The key phosphorylation steps for model A, in which one Lyn can initiate phosphorylation of aggregated FcɛRI, and model B, in which two Lyns per aggregate are required for phosphorylation to proceed. FcɛRI on mast cells and basophils consists of four chains, α (an IgE-binding chain), β, and two γs. The β and γ chains each contain one immunoreceptor tyrosine-based activation motif (ITAM). Each ITAM contains two essential tyrosines, although within the β chain ITAM there is an additional tyrosine. In modeling the receptor we do not keep track of distinct ITAMS but simply take receptors to be phosphorylated or not phosphorylated. Additional reactions in the models include weak constitutive association of Lyn with unphosphorylated FcɛRI and stronger association of Lyn with phosphorylated receptors. For model A, the rate constants p₊₁ and p₊₂ govern, respectively, the phosphorylation of FcɛRI when Lyn is bound constitutively and when Lyn is bound to a phosphorylated ITAM. Although the model does not include phosphatases explicitly, their effects are included implicitly in the dephosphorylation rates p₋₁ and p₋₂. This formulation assumes that the availability of phosphatases remains essentially constant for the duration of the experiments under consideration. In model B, phosphorylation of the receptors explicitly requires juxtaposition of two Lyns. Possibly this would be required to induce a change in the specific activity of Lyn, as might be evidenced by enhanced activity toward an exogenous test substrate. Alternatively, it might be required to enhance the susceptibility of the receptor to modification, through some other mechanism. In the figure, we depict this phenomenon as a change in the configuration of Lyn, without thereby implying a particular mechanism. The forward and reverse rate constants that govern this transition are denoted by k_+a and k_−a. In both models, we have made the simplifying assumption that a receptor cannot be in the phosphorylated state and still have Lyn bound constitutively. The difference we will demonstrate between predictions of the two models does not depend on this assumption. (For additional detail on the equations, states, and parameters of the models, see the supplemental material on the PNAS web site at www.pnas.org.)

Figure 2

Relative phosphorylation of FcɛRI after stimulation, as a function of expressed Lyn. This figure is illustrative of the analyses that form the basis of this study. Each clone was reacted with 0.5 μg/ml dimeric IgE for 15 min before solubilization of the cells, reaction of the centrifuged lysate with anti-IgE, and analysis of the immunoprecipitates on gels. (A) Line 2 reports the relative amount of Lyn per FcɛRI for each of the seven clones. (Upper) Western blot with anti-PY. (Lower) Western blot with anti-β after stripping of the gel shown in Upper. For reasons that have not been investigated, the immunoprecipitates from clone A9 regularly showed an abnormally low yield of β chains relative to recovery of radiolabeled IgE, suggesting that the receptors were unusually unstable. (B) log–log plot of the phosphorylation of the β chain ([PY]_rel) vs. the relative expression of Lyn per receptor ([Lyn]_rel/FcɛRI), for the experiment shown in A. The line that gives the best fit to the data is shown. Its slope is 1.1. The correlation coefficient is 0.99. The data from clone A9 were not included because the small yield of β chains (see above) precluded accurate quantification. In 26 similar plots, reflecting measurements of tyrosine phosphorylation on the β and γ chains at distinct doses of ligand and distinct times, the mean slope is 0.96 ± 0.05.

Figure 3

Results of simulations of the experiments, based on alternative models. Markers indicate the fraction of receptors phosphorylated, as predicted by models A (○, Fig. 1A), B (⋄, Fig. 1B), and an analogue of model A in which receptors are aggregated by trimers of IgE (●), for the experiments in which cells were exposed to 0.3 μg/ml dimeric IgE for 15 min. For each of the three curves, we determined the line that gave the best fit to predicted values of the fraction of FcɛRI phosphorylated, for Lyn/FcɛRI in the range 10⁻⁵ to 10⁻² (some of the values used lie outside the range shown in this plot). The line giving the best fit to the simulated data, in this linear range, has a slope of 1.0 for model A, in which one Lyn can initiate phosphorylation, and 2.1 for model B, in which two juxtaposed Lyn molecules are required. In the model where aggregation is induced by trimers of IgE but only one Lyn is needed for phosphorylation of receptor tyrosines, the predicted slope is 1.0. Predictions of the slopes are robust. Large changes in the rates of phosphorylation and dephosphorylation, the rates of association and dissociation of the receptor with Lyn, and the times at which phosphorylation is measured result in almost no change in the predicted slopes. Even when we chose the rate of dephosphorylation to be slow compared with the rate of dissociation of Lyn, so that by 15 min a significant fraction of the receptor aggregates that are phosphorylated are not associated with Lyn, the slopes for the two models remain essentially equal to 1 and 2.