Report

An Unusual Mechanism for Ligand Antagonism

See allHide authors and affiliations

Science  24 Jul 1998:
Vol. 281, Issue 5376, pp. 568-572
DOI: 10.1126/science.281.5376.568

Abstract

The ratio of late to early events stimulated by the mast cell receptor for immunoglobulin E (IgE) correlated with the affinity of a ligand for the receptor-bound IgE. Because excess receptors clustered by a weakly binding ligand could hoard a critical initiating kinase, they prevented the outnumbered clusters engendered by the high-affinity ligands from launching the more complete cascade. A similar mechanism could explain the antagonistic action of some peptides on the activation of T cells.

Binding of ligand to a cell surface receptor often stimulates an elaborate biochemical cascade. If one of the initiating interactions must be preserved during the course of subsequent time-dependent, energy-consuming steps, the fidelity of the response can be considerably greater than would be predicted simply from the free energy released by the initial interaction with ligand. That is, ligands with lower affinity—which generally means those forming complexes with shorter lifetimes—would be less likely to stimulate responses that went to completion, a process that in related multistep systems has been dubbed “kinetic proofreading” (1). This formulation has been applied to explain the discriminatory prowess of the antigen receptors of T cells (2) and possibly to account for the action of variant peptides that can act as partial agonists or antagonists (3). Some experimental evidence supports this formulation (4, 5), but specific molecular explanations have not yet been described.

The clonotypic antigen receptor on T cells is one of a family of receptors called “multichain immune recognition receptors” that share numerous structural and functional attributes (6). The high-affinity receptor for the Fc portion of immunoglobulin E (IgE), FcɛRI, is a member of this family. We examined the kinetic proofreading formulation in the context of FcɛRI and explored whether ligands of differing affinity could act as mutual antagonists under conditions at which simple displacement could not occur.

We loaded the FcɛRIs on rat mucosal-type mast cells (line RBL-2H3) (7) with a monoclonal IgE specific for the 2,4-dinitrophenyl (DNP) hapten (8). The IgE's affinity for several nitrophenyl hapten analogs relative to DNP was ascertained (9), selected multivalent hapten-protein conjugates were prepared (10), and the cellular responses to high- or low-affinity ligands were monitored. The phosphorylation of tyrosines on the receptor and of several proteins in response to aggregation of FcɛRI was quantitated (Fig. 1, A and B). The high-affinity DNP antigen stimulated phosphorylation of tyrosines on the β and γ subunits of FcɛRI with comparable kinetics, and the data shown for the receptor represent the combined values for the two types of subunits (Fig. 1A). The phosphorylation of the kinase Syk and the adaptor protein Nck reached a maximum at virtually the same time as that of FcɛRI, but the phosphorylation of the kinase Erk2 was delayed. Higher and lower doses of the antigen correspondingly accelerated or slowed the kinetics of phosphorylation, but the order in which the proteins were modified was unaltered.

Figure 1

Time course of phosphorylation of tyrosines on several proteins in mast cells loaded with anti-DNP IgE after stimulation with (A) a high-affinity DNP-conjugated antigen (50 ng/ml) or (B) a low-affinity 2NP-conjugated antigen (375 ng/ml) (36). In (A), the data were normalized for each component analyzed by dividing the absorbance at each time point by the absorbance of the sample having the maximum during the course of the experiment. In (B), the value at any time point was divided by the maximum absorbance achieved for that component in the cells stimulated by the DNP conjugate. •, FcɛRI; ▴, Syk; □, Nck; ◊, Erk2. (C) Relative phosphorylation of receptors and release of hexosaminidase stimulated by different antigens. Release was measured on adherent cells as described (24, 37,38). The absorbance related to the phosphotyrosine on receptors in the sample of cells stimulated with the 2NP-conjugated antigen was designated as “1.00,” and the values for the cells stimulated with the alternate ligands were related to this value (open bars). The same procedure was used to compare the release of hexosaminidase (hatched bars), except that the data were normalized to the release stimulated by the DNP ligand. The net release in the duplicate samples stimulated with the DNP antigen was 26.07 ± 1.7%. The error bars show the range of duplicate samples.

For methodological and other reasons, the absolute amounts of phosphotyrosine on different proteins are not directly comparable. Nevertheless, we estimated that the maximum amount of phosphotyrosines on the FcɛRI, Syk, Nck, and Erk2 was about in the ratio of 3:1:0.2:0.4.

The IgE antibody to DNP (anti-DNP) bound the 2-nitrophenyl (2NP) moiety with an intrinsic affinity less than 0.001 of that for DNP (9). At comparable doses, the 2NP hapten conjugate induced less vigorous phosphorylation of the FcɛRI than the DNP antigen, but this could be compensated for by the use of somewhat higher doses (Fig. 1B). Here the values shown correspond to the value at each time point divided by the maximum value achieved for that protein when stimulated as in Fig. 1A with the high-affinity ligand. Despite any ambiguity in comparing the extent of phosphorylation of different proteins directly (above), a comparison between the phosphorylation of the same components by the two stimuli is straightforward. The low-affinity ligand was progressively less effective in activating “downstream” components per unit of phosphorylation of the receptor (11). Thus, when the maximum phosphorylation of the receptor stimulated by the low-affinity ligand was more than twofold higher, the maximum phosphorylation of Syk was less than one-third of that achieved by the high-affinity ligand, and that for Erk2, whose activation is known to be “downstream” of Syk (12), was only one-tenth as much. These findings are consistent with a kinetic proofreading regime. [Our data place the activation of Nck temporally between Syk and Erk2 (Fig. 1A), but its location in one or another pathway is still unknown.]

Secretion of granular contents from the cells was also measured (Fig. 1C). The weakly bound 2NP ligand also stimulated release of hexosaminidase poorly, even at doses sufficient to stimulate phosphorylation of the FcɛRI severalfold greater than that stimulated by the high-affinity DNP ligand. On the other hand, a conjugate made from the 2,4-dinitro-6-carboxyphenyl moiety (oDNCP), a hapten whose relative affinity for the anti-DNP IgE is 3% that of DNP (9), was almost as effective in stimulating degranulation as the DNP antigen in our initial experiments with cells loaded with bivalent anti-DNP IgE. However, we observed a clear difference when the cells were sensitized with a bispecific IgE antibody that is monovalent with respect to its DNP-binding site [(10) and below] (Fig. 1C). Here, the maximum size of the aggregates that are generated can be no greater than the valence of the antigen (13).

Variant peptides can antagonize the stimulation of T lymphocytes induced by wild-type peptides under nondisplacing conditions (3). We examined whether a similar effect would be observed in responses mediated by FcɛRI. RBL-2H3 cells were loaded with a mixture of two monoclonal mouse IgEs, one specific for DNP and the other for the non–cross-reacting dansyl moiety (DNS) (9). The cells were then exposed to the DNS-protein conjugate, the low-affinity 2NP conjugate, or both simultaneously, and immunoprecipitates of several proteins were assayed for phosphotyrosine (Fig. 2A).

Figure 2

Antagonistic action of weakly binding ligand. (A) RBL-2H3 cells were loaded with IgE as described for Fig. 1, except that a mixture of two IgEs, one specific for DNP and the other for DNS, was used such that about one-half of the cells' receptors would be occupied by each of the specific IgEs. The cells were then exposed at 25°C to DNS-protein conjugate (2.7 ng/ml) (lane 1), the low-affinity 2NP conjugate (100 ng/ml) (lane 3), or both simultaneously (lane 2). Portions were removed for immunoprecipitation of solubilized proteins at the times of maximum phosphorylation of each component and then separated by gel electrophoresis and assayed for phosphotyrosine. The far left panel shows phosphorylation of the β and γ chains in the receptors from 106 cell equivalents. Because the immunoprecipitations of the receptors were through their bound IgE with anti-IgE, the precipitates contained receptors bound to both the DNP- and DNS-specific IgE. The other panels show immunoprecipitations with anti-Syk (106 cell equivalents), anti-Pyk2 (5 × 106 cell equivalents), and anti-Erk2 (4 × 106cell equivalents). (B) The experiment is like that shown in (A), except that separate portions of the cells were sensitized alternatively with anti-DNS and anti-DNP IgE and then mixed after first washing the cells well to remove unbound IgE. The cells were then stimulated with the DNS-protein conjugate alone (lane 4), the low-affinity 2NP conjugate alone (lane 6), or both simultaneously (lane 5). Only the phosphorylation of the receptor and of Erk2 is shown.

At the dose of the DNS conjugate used, the phosphorylation of FcɛRI, although modest, was sufficient to stimulate phosphorylation of the three downstream components examined: Syk, Pyk2, and Erk2. Pyk2, a member of the family of “focal adhesion” kinases, is phosphorylated subsequent to the activation of Syk in RBL-2H3 cells (14). As before, the FcɛRI was phosphorylated when the cells were stimulated with only the low-affinity 2NP ligand (15), but phosphorylation of the downstream components was diminished. When the cells were stimulated with the mixture of non–cross-reacting ligands, the phosphotyrosine in the total receptor subunits was undiminished and phosphorylation of Syk and Pyk2 was minimally reduced, but phosphorylation of Erk2 was substantially decreased (16).

To test whether the inhibitory effect of the low-affinity ligand was mediated by a released soluble factor, separate portions of cells were loaded with anti-DNS IgE and anti-DNP IgE alternatively and, after being washed to remove unbound IgE, were mixed. They were then stimulated under the three conditions as in Fig. 2A (Fig. 2B). The phosphorylation of the receptor proceeded as before, as did the phosphorylation of Erk2 by the DNS conjugate or the lack thereof by the 2NP conjugate. However, phosphorylation of Erk2 was no longer reduced in response to the mixture of ligands. Thus, the low-affinity ligand is only inhibitory when bound to the same cells as the high-affinity ligand.

The time dependence of the inhibitory effect was explored (Fig. 3A). Again, the cells were doubly sensitized with anti-DNP and anti-DNS IgE and exposed to the high-affinity DNS conjugate either alone or admixed with an excess of the low-affinity 2NP conjugate. The inhibition was not transient, and the temporally more distal components of the signaling cascade (Pyk2 and Erk2) were more profoundly affected. In six such experiments in which the dose of the low-affinity ligand was sufficient to stimulate the phosphorylation of the receptor severalfold relative to the high-affinity ligand, the inhibition of phosphorylation of Erk2 averaged almost 70% (17). Increasing doses of the low-affinity ligand also progressively inhibited secretion of hexosaminidase (Fig. 3B). Because of the ineffectiveness of the 2NP antigen in stimulating even the phosphorylation of the receptors at 37°C (see above), this and three additional experiments, which gave virtually identical results, were conducted at 25°C. As a control, release from cells loaded only with anti-DNS IgE was not inhibited by the added 2NP ligand, ruling out the trivial explanation of a “toxic” effect of 2NP.

Figure 3

Inhibition of cellular response by low-affinity antigen. (A) Time dependence of the inhibitory effect. Cells were loaded with a mixture of anti-DNP and anti-DNS IgE and stimulated with DNS conjugate (2.7 ng/ml) (•) or DNS conjugate (2.7 ng/ml) plus 2NP conjugate (100 ng/ml) (○). The indicated proteins were analyzed for phosphotyrosine. The dimensions of the ordinate scales for the four panels have been roughly normalized. (B) Inhibition of hexosaminidase release by low-affinity ligand. The experiment was performed at 25°C. Cells were loaded either with both anti-DNS and anti-DNP (grey bars) or with only anti-DNS (black bar). Samples were reacted for 30 min with DNS conjugate (2.7 ng/ml), in addition to 2NP conjugate (0, 100, or 500 ng/ml). The bars indicate the absolute percentage of release after subtracting the percentage of hexosaminidase released spontaneously (2.13 ± 0.11%). The error bars show the range for the duplicate incubations studied. Three additional experiments gave virtually identical results.

In this system, it has been proposed that the initiating event is an aggregation-induced transphosphorylation of the receptor by the small amount of lyn constitutively associated with the FcɛRI (18) and that the amount of weakly associated kinase can limit the intensity of the response (19). We therefore hypothesized that clusters of receptors induced by an excess of low-affinity ligand could sequester the kinase, so that the smaller number of clusters associated with the high-affinity ligand would be deprived of lyn and, therefore, of the means to initiate the signaling cascade. If so, then in an experiment such as that shown in Fig. 2A, phosphorylation of those receptors specifically aggregated by the high-affinity ligand would be expected to be reduced when isolated from the cells stimulated with the mixture of high- and low-affinity ligands, even though the receptor population as a whole showed a higher amount of phosphorylation. We had technical difficulty in efficiently immunoprecipitating the high-affinity DNS ligand, so for these experiments we used preformed oligomers (dimers) of rat IgE as a surrogate high-affinity ligand (20).

Cells were first partially loaded with mouse anti-DNP IgE and then stimulated with rat IgE dimers alone, the low-affinity 2NP conjugate alone, or a mixture of both (Fig. 4). The dimer-clustered receptors stimulated phosphorylation of Erk2, whereas, as before, the low-affinity 2NP ligand was not only deficient in stimulating phosphorylation of Erk2 by itself but also inhibited the action of the high-affinity stimulant. Under these conditions, the receptors clustered by the low-affinity antigen were phosphorylated (Fig. 4B). However, the phosphotyrosine on the receptors that were clustered by the high-affinity ligand was reduced by about two-thirds (Fig. 4C).

Figure 4

Molecular mechanism of antagonism. Cells were first partially loaded with mouse anti-DNP IgE. At time zero, they were then reacted with rat IgE dimer alone (100 ng/ml) (lane 1), a mixture of dimer (100 ng/ml) and low-affinity 2NP conjugate (200 ng/ml) (lane 2), or the low-affinity 2NP conjugate alone (200 ng/ml) (lane 3). Cell extracts were specifically immunoprecipitated with (A) anti-Erk2, (B) anti-mouse IgE, or (C) anti-rat IgE, and the precipitated material was analyzed on gels for protein-bound phosphotyrosine as before.

In the previous experiments, we deliberately used a protocol in which the receptors were clustered in two distinct pools—one for each ligand. When we instead used IgE of a single specificity in combination with limited amounts of low- and high-affinity ligands, the receptor clusters would contain both high- and low-affinity ligands. Depending on a number of variables, addition of low-affinity ligand either inhibited or augmented downstream signaling stimulated by high-affinity ligand. However, when we used the bispecific antibody, where the receptors binding the low- or high-affinity ligand cannot co-cluster, the low-affinity ligand reproducibly inhibited signaling from receptors binding the high-affinity ligand, as in the protocols with two distinct ligands (21).

The essential feature of systems subject to kinetic proofreading is that the cascade of events that follows interaction with a ligand persists only as long as the initiating interaction is maintained (22). We have determined that the FcɛRI-initiated cascade of cellular responses behaves like a system subject to kinetic proofreading. The molecular explanation for which we have obtained direct evidence relates to findings on how receptors that are not themselves kinases may need to compete for the limited extrinsic kinase that initiates the transphosphorylation of approximated receptors (19). In such a system, weakly clustered receptors act like the “dog in the manger” (23): Like the dog, they impede access to a necessity (the kinase) in spite of their inability to use it productively. The pathophysiological sequelae produced by such activators are likely to differ from those capable of stimulating the full cellular response. As already noted (22), it is the lifetime of individual receptors in a cluster that determines their likelihood of initiating a “complete” cascade. Therefore, the ratio of late to early signals stimulated by a receptor, such as an FcγR, that uses a similar signaling cascade as FcɛRI but whose interaction with its cognate Ig is much weaker might be different (lower) than that stimulated by FcɛRI (all other things taken equal), with corresponding differences in the cellular response. Some relevant although limited data on this matter [Table I in (24)] show no such tendency, but a kinetic analysis or an analysis of later events (for example, gene transcription) might uncover such a trend.

In principle, the molecular mechanism we propose can relate to other receptors that require recruitment of an extrinsic component in limited supply and that are subject to a kinetic proofreading regimen. In the immune system, the family of multichain immune recognition receptors and the cytokine receptors are obvious candidates.

REFERENCES AND NOTES

View Abstract

Navigate This Article