The high affinity receptor for IgE, FceRI.

The high affinity receptor for IgE, FceRI

Abstract. Progress in our understanding of the structure and function of the receptor with high a⁄nity for IgE (FceRI) is already being applied in attempts to mitigate allergic reactions, speci¢cally by using speci¢c anti-IgEs to prevent sensitization. Investigations of the complex network of intracellular signals generated by FceRI and related receptors are proceeding. Although achieving a ‘complete’ description of the events will be a daunting task, the e¡ort is likely to identify additional targets for therapeutic intervention along the way.

As we learn more about the variety of surface proteins through which cells sense and respond to their environment it becomes increasingly apparent that the simplistic, linear conceptualizations we adopted during our initial experimental probes in which we focused on the cellular consequences of a single type of ligand^receptor interaction, are no longer adequate. We must now confront entire systems of molecules whose interactions with each other vary with time and with changes in the external milieu. In this discussion, I will use the receptor with high affinity for IgE as an example of such a conceptualization, emphasizing not so much the results that have been obtained so far, as the approaches that will be necessary to pursue a more holistic approach to the receptor’s function.

This audience is undoubtedly familiar with many of the basic facts about the receptor (FceRI) and up-to-date reviews have appeared regularly (Kinet 1999, Kawakami & Galli 2002, Metzger et al 1989), so I will sharply limit my discussion of these matters.

Status of structural analyses

X-ray di¡raction structures of IgE, of the IgE-binding ectodomains of the receptor’s a chain, and of the 1:1 complex of these structures have been determined (Wurzburg & Jardetzky 2002). The sites of interaction are sufficiently well de¢ned to assist in the rational development of small molecules that could act as inhibitors of the binding (Pietersz et al 2002). As noted below, the increasing evidence that anti-IgE antibodies that inhibit the interaction are clinically useful (Chang 2000, Metzger 2003), shows that in principle, this approach is worth pursuing.

Regrettably, there has been little progress in structurally de¢ning the cytoplasmic domains of the b and g subunits of the receptor, although there continues to be progress in clarifying the interactions of these domains with other cellular proteins before and after their phosphorylation. Interfering with such interactions is another potential target for low molecular compounds and it would be helpful if structural data comparable to that for the interaction between IgE and the receptor might be available.

Initial events

Aggregation as the initiating event

Considerable experimental data support a model in which it is the aggregation of the receptor induced by the interaction of a multivalent antigen with the receptor- bound IgE that initiates a variety of biochemical cascades (Metzger 1992). The ability of apparently monomeric IgE to likewise trigger similar events is likely a laboratory artefact due to the tendency of certain monoclonal IgE antibodies to aggregate (Asai et al 2001, Kawakami & Galli 2002) either spontaneously or possibly due to some cross reaction with an antigen in the medium or on the cell surface (see also below). Whether the ‘anti-apoptotic’ e¡ect of the binding of IgE to FceRI is similarly based on aggregation remains unclear (Asai et al 2001, Kawakami & Galli 2002).

Molecular consequences of aggregation

A variety of experiments have supported a mechanistic model in which a kinase (Lyn) constitutively associated with the C-terminal cytoplasmic extension of the unphosphorylated b chain transphosphorylates one or more g chains on an alternate receptor that has become approximated by the binding of antigen (Pribluda et al 1994). Experimental data suggest that normally this initial event occurs in specialized lipid ‘micro domains’, but that that is an incidental occurrence rather than an absolute requirement. A somewhat di¡erent model proposes that the aggregation-induced movement of receptors into the micro domains is critical for the phosphorylation of b and g because the micro domains are enriched in the kinase (translocation model) (Baird et al 1999).

Recent new ¢ndings have so-far not resolved the issue. On the one hand, studies using rigid bivalent ligands whose epitopes are thought (but not proven) to be sufficiently distant from each other to make the transphosphorylation sterically impossible, found them capable of stimulating phosphorylation of the receptors, suggesting to the authors that the translocation model is correct (Paar et al 2002). On the other hand, experiments in which normal phosphorylation of the receptors was observed even though Lyn or constructs of Lyn that were incapable of entering the domains (Kovarova et al 2001, Vonakis et al 2001) were used, or the domains were disrupted (Yamashita et al 2000), seriously challenge the proposal that translocation is required for the initiating event. The subtle di¡erence in these proposals is not irrelevant if one were to try to disrupt receptor-initiated responses by interfering with the proximal event following aggregation. On the basis of the transphosphorylation model, one would focus on inhibiting a speci¢c protein^ protein interaction; alternatively one would have to interfere with more non- speci¢c protein^lipid domain interactions. Further exploration of these alternative models is warranted. The recent observation that a related kinase, Fyn, is also required during the initiating events has added a further level of complexity (Parravicini et al 2002).

Later events

Phosphorylation of FceRI attracts a variety of additional proteins to the receptor aggregates. The composition of these signalling complexes, their lifetimes, and their topological movements are being actively pursued (Rivera 2002). That the propagation of signals requires the assembly of a multicomponent macromolecular ‘machine’ explains why at least some of the cellular responses re£ect the lifetime of the receptor aggregates. This leads to a phenomenon called ‘kinetic proofreading’ (Hop¢eld 1974, McKeithan 1995), a mechanistic regime under which the length of the signalling pathway stimulated by di¡erent ligands will be inversely related to their rate of dissociation from their binding component. In the case of FceRI, it is the rate of dissociation of the allergen from the receptor- bound IgE that determines the lifetime of the receptor in the aggregate, and therefore the length of time the aggregate can propagate the sequential downstream events.

We have explored the importance of kinetic proofreading in the IgE^FceRI system and have uncovered two interesting aspects. The ¢rst was the phenomenon of non-competitive ligand inhibition. In this situation one ligand can inhibit the cellular stimulation by another ligand in the absence of direct competition for the ligand-binding sites (Sette et al 1994). The mechanism we uncovered was based upon the inhibitory ligand’s ability to sequester the limited supply of a critical enzyme (Torigoe et al 1998). A second phenomenon, the details of which have just been published (Eglite et al 2003), relates to the ability of a weakly binding (rapidly dissociating) ligand that stimulates a relatively distal response such as degranulation, poorly or not at all, to nevertheless e¡ectively stimulate cellular responses that are even more delayed, e.g. transcription of the gene for the chemoattractant MCP-1. In this instance it appears that relative to its own lifetime, a critical messenger for stimulating the transcription, Ca2+, is generated early enough in the cascade initiated by the receptor aggregates, that adequate concentrations of the ion can be stimulated even by the short-lived aggregates.

Future analyses

Ultimate goals

The goal(s) for those engaged in unravelling the ways particular cells respond to a variety of external signals is to ‘. . . understand as completely as possible the relationships between sets of inputs and outputs that vary both temporally and spatially. How does a cell respond appropriately to one voice when it must listen simultaneously to many, and how does it alter this response in the context of other concurrent or recent signalling events?’(Gilman et al 2002). Ultimately, of course, one wishes to use this information therapeutically, either to correct genetic defects or pathological responses. The more we know about the signalling pathways the better we will be able to pinpoint the appropriate targets for manipulation, but our understanding need not be exhaustive before such interventions can be attempted. A good example, already referred to above, is the use of anti-IgE to prevent sensitization. This approach was initiated even before many of the details about how IgE interacts with FceRI were uncovered. Even now, the impact on the therapeutic e¡ectiveness of the anti-IgE interacting with membrane IgE on B cells (but not with IgE bound to FceRII on lymphocytes, monocytes and platelets) and the presence of IgE:anti-IgE complexes in the serum remain to be determined (Chang 2000).

What will be required and how will we get there?

To achieve the more complete understanding we seek, we must ¢rst identify all the critical molecules engaged in relevant signalling pathways, and characterize the temporal changes in their concentrations, locations, interactions, and modi¢cations normally, and in pathological conditions. Second, these data must be incorporated into mechanistic and quantitative models that can be used to test whether the critical events are understood, and to predict the consequences of interventions.

Those who are following the progress in this area of cell biology are becoming increasingly impressed by the extraordinary complexity of these systems. Although there has been nothing uncovered so far that would justify a nihilistic assessment of the likelihood for achieving any level of understanding we desire, the enormity of the task ahead cannot be minimized. It makes the unravelling of the human genome look relatively simple by comparison. This realization is prompting some novel proposals on the organization of research in this area. Perhaps the most ambitious proposal is the recently organized ‘Alliance for Cellular Signaling’ (AfCS) (Gilman et al 2002).

This consortium currently involves some 50 ‘participating investigators’ from around 20 institutions who will organize their group and individual activities according to an initial organizational network (see Fig. 1 in Gilman et al 2002) that stipulates the interactions for decision-making with respect to research directions, development of new technologies, distribution of experimental ¢ndings and publication of signi¢cant new information.

The AfCS’s initial experimental strategy is presented in Table 1. As can be seen there is nothing particularly novel in the individual components; all of the same elements have been used by individual investigators interested in particular cellular events. The unique feature will be the coordination and comprehensiveness of the e¡orts.

In their inaugural ‘Overview’ the group underscores two aspects of this endeavour. First, that this endeavour is itself an experiment in collaborative science that although not unprecedented in biology (e.g. the human genome project) is unusual in the basic research arena. There are implications to such e¡orts that go well beyond the experiments conceived, planned, executed, interpreted and published by committees. For example, the current reward systems in academic careers in the biological sciences put a premium on individual creativeness, a characteristic that will be di⁄cult to tease out for any particular participant in such group e¡orts.

Second, only a modest contribution can be expected from even the substantial coterie of investigators assembled so far along with its $10 000 000 initial annual budget relative to the size of the problem. The present e¡ort will be directed towards only two cells, mouse B lymphocytes and cardiac myocytes. The

consortium of course hopes that their experiment in ‘socialistic science’ will be sufficiently productive to attract the (¢nancial) interest of pharmaceutical firms, and to encourage additional investigators to use the AfCS approach as a research paradigm.

Quantitative modelling

I shall close this discussion with a brief consideration of the last element of this list on which my own group has focused in a long-term collaboration with a group in the Theoretical Biology and Biophysics Group of the Theoretical Division at Los Alamos. Although using the same overall strategy planned by the AfCS, our e¡ort has of course been much more limited. We have varied only a single type of input (aggregation of FceRI) and quantitatively examined the ‘£ow of information’ through only a selected set of molecules that are engaged very early in the response. A presentation of the detailed mathematical model and its properties has just appeared (Faeder et al 2003).

Many of the molecular mechanisms that are modelled are based on experimental results from ourselves as well as several other groups, but also include a number of assumed aspects. The hope is that such modelling will allow one to test the validity of certain assumptions.

The current version models the phosphorylation of FceRI and Syk following aggregation of FceRI by addition of chemically cross-linked dimers of IgE - a surrogate input for aggregation of receptor-bound IgE by antigen. The model is a network which contains 354 discrete chemical species and the chemical reactions that connect them. Concentrations of the principal components (FceRI, Lyn and Syk) are based on measurements using quantitative Western blotting. Twenty-one rate constants based on direct measurements or other observations are used: four constants relate to binding of the dimers to the receptors, four each to the association of Lyn and Syk kinase, eight to the phosphorylation of tyrosines in the various molecular aggregates and one for the rate of dephosphorylation. Among the simplifying assumptions is that multiple tyrosine residues on the subunits of the receptor and on Syk are treated as single units of phosphorylation. The network is then converted to a predictive model using a set of coupled di¡erential equations. Quantitative outcomes were based on a computer program that generates the states and reaction network from the components and reaction rules. A second program uses the generated network, rate parameters and initial concentrations to generate and solve the set of 354 di¡erential equations as a function of time.

I will not focus on the results of this modelling except to note a few ¢ndings. First, simulations of experiments where RBL cells were exposed to covalently cross-linked dimers of IgE reproduced the observed kinetics of phosphorylation of the receptors’ subunits over a period of an hour, and the predicted dose response curves followed closely those observed experimentally. The model also is consistent with the evidence for a kinetic proofreading constraint with respect to the phosphorylation of Syk kinase as observed experimentally. The model also makes a variety of other mechanistic predictions that can be experimentally tested in future experiments.

There is no reason to think that this approach would not be equally applicable to examine related systems such as those stimulated by other multi-subunit immune response receptors that initiate cellular responses using related molecular mechanisms. But, what is the likelihood that such an approach can be extended to more complete networks involving many more inputs simultaneously triggering multiple receptors thereby activating networks containing possibly hundreds of nodes?

The principal authors of this model at Los Alamos conclude on a pragmatic note: ‘Ultimately we want su⁄cient knowledge for quantitative prediction of larger segments of signalling pathways. To have con¢dence in such predictions, our approach is to proceed step-wise, testing each extension of the model against many di¡erent forms of experimental data’ (Faeder et al 2003).

The leaders of the AfCS believe that this incremental approach will be inadequate to deal with the complexity of cellular networks. ‘For this goal [to model such networks] it will almost certainly be inappropriate to adopt the traditional reductionist approach of simply measuring in vitro all the Kd, Km, and Vmax values, cooperativity constants, rate constants and molar concentrations for all molecules involved’ (Gilman et al 2002). They surmise that it will not be practical to mimic the milieu in which these reactions occur nor the multiple interactions of single molecules and that therefore it will not be possible to estimate accurately many of the avidity and rate constants. Finally, they consider the manipulation of the mass of data and the equations that relate them, ‘daunting’. Certainly no one could argue with the latter assessment.

The alternative approach they propose is ¢rst to express the observed functional interactions in relative terms. For example, fractional activation of a particular molecule would be related to a fractional e¡ect on a downstream molecule in order to arrive at a series of ‘linkage functions’. This approach, albeit still demanding, is much less so than the explicit approach currently used by the Los Alamos team and others. But the strategy proposed by the AfCS also has a potential problem. That is, whereas such linkage functions can be manipulated to give reasonable quantitative predictive values over a given set of inputs, they may not be able to distinguish between substantially di¡erent mechanistic models. The Ptolemaic epicycles and a variety of other mathematical constructs were quite e¡ective in rationalizing the wandering motions of the planets but of course bore no relationship to the reality of our solar system.

Closing comments

The investigation of cellular responses is proceeding from the state where the focus was on identifying individual molecules a¡ected by individual receptors and the linear pathways that connected them, to a more physiologically relevant approach. This will try to describe the multitude of molecular responses that follow variations in the stimuli to the multitude of receptors that cells use to monitor their environment. We already know enough to be able to develop a realistic assessment of the enormity of the task, but despite this enormity there is nothing, so far, that suggests that the task is in principle undoable.

The hope is that such investigations will not only be enlightening but also useful. Speci¢cally, in the networks that are currently known one can distinguish two types of molecular interaction. The members of one class involve ‘nodes’ in which a molecule engages in only a small number of interactions; alternatively other interactions occur in ‘hubs’ and involve a multitude of interactions. Pathological aberrations in nodes may be relatively easy to correct by manipulations that will create detours around them; defects in hubs may be much more challenging to overcome but on the other hand can serve as e¡ective targets for disabling a whole set of responses.

References

Asai K, Kitaura J, Kawakami Y et al 2001 Regulation of mast cell survival by IgE. Immunity 14:791^800

Baird B, Sheets ED, Holowka D 1999 How does the plasma membrane participate in cellular signaling by receptors for immunoglobulin E? Biophys Chem 82:109^119

Chang TW 2000 The pharmacological basis of anti-IgE therapy. Nat Biotechnol 18:157^162 Eglite S, Morin J, Metzger H 2003 Synthesis and secretion of monocyte chemotactic protein-1

stimulated by the high a⁄nity receptor for IgE. J Immunol 170:2680^2687

Faeder JR, Hlavacek WS, Reischl I et al 2003 Investigation of early events in Fc(epsilon)RI- mediated signaling using a detailed mathematical model. J Immunol 170:3769^3781

Gilman AG, Simon MI, Bourne HR, et al 2002 Overview of the Alliance for Cellular Signaling.

Nature 420:703^706

Hop¢eld JJ 1974 Kinetic proofreading: a new mechanism for reducing errors in biosynthetic processes requiring high speci¢city. Proc Natl Acad Sci USA 71:4135^4139

Kawakami T, Galli SJ 2002 Regulation of mast-cell and basophil function and survival by IgE. Nat Rev Immunol 2:773^786

Kinet J-P 1999 The high a⁄nity IgE receptor (FceRI): from physiology to pathology. Annu Rev Immunol 17:931^972

Kovarova M, Tolar P, Arudchandran R, Draberova L, Rivera J, Draber P 2001 Structure^ function analysis of Lyn kinase association with lipid rafts and initiation of early signaling events after Fc epsilon receptor I aggregation. Mol Cell Biol 21:8318^8328

McKeithan TW 1995 Kinetic proofreading in T-cell receptor signal transduction. Proc Natl

Acad Sci USA 92:5042^5046

Metzger H 1992 The receptor with high a⁄nity for IgE. Immunol Rev 125:37^48 Metzger H 2003 Two approaches to peanut allergy. N Engl J Med 348:1046^1048

Metzger H, Blank U, Kinet J-P et al 1989 Emerging picture of the receptor with high a⁄nity for IgE. Intl Arch Allergy Appl Immunol 88:14^17

Paar JM, Harris NT, Holowka D, Baird B 2002 Bivalent ligands with rigid double-stranded DNA spacers reveal structural constraints on signaling by Fc epsilon RI. J Immunol 169:856^864

Parravicini V, Gadina M, Kovarova M et al 2002 Fyn kinase initiates complementary signalsr equired for IgE-dependent mast cell degranulation. Nat Immunol 3:741^748

Pietersz GA, Powell MS, Ramsland PA, Hogarth PM 2002 Fc receptor structure and the design of anti-in£ammatories: new therapeutics for autoimmune disease. Annu Rep Med Chem 37:217^224

Pribluda V, Pribluda C, Metzger H 1994 Transphosphorylation as the mechanism by which the high-a⁄nity receptor for IgE is phosphorylated upon aggregation. Proc Natl Acad Sci USA 91:11246^11250

Rivera J 2002 Molecular adapters in Fc(epsilon)RI signaling and the allergic response. Curr Opin Immunol 14:688^693

Sette A, Alexander J, Ruppert J et al 1994 Antigen analogs/MHC complexes as speci¢c T cell receptor antagonists. Annu Rev Immunol 12:413^431

Torigoe C, Inman JK, Metzger H 1998 An unusual mechanism for ligand antagonism. Science28 1:568^572

Vonakis BM, Haleem-Smith H, Benjamin P, Metzger H 2001 Interaction between the unphosphorylated receptor with high a⁄nity for IgE and Lyn kinase. J Biol Chem 276:1041^1050

Wurzburg BA, Jardetzky TS 2002 Structural insights into the interactions between human IgE and its high a⁄nity receptor Fc epsilon RI. Mol Immunol 38:1063^1072

Yamashita T, Yamaguchi T, Murakami K, Nagasawa S 2000 Detergent-resistant membrane domains are required for mast cell activation but not for tyrosine phosphorylation upon aggregation of the high a⁄nity receptor for IgE. J Biochem (Tokyo) 129:861^868

DISCUSSION

Lasser: I have a rather simplistic question. In what way does the anti-IgE molecule inhibit the binding of IgE? What is the mechanism?

Metzger: The simple answer is that it sterically inhibits: it binds to an area on the molecule that is necessary for it to interact with the receptor. It is a little bit like the princess and the pea: it doesn’t necessarily have to interact directly with those atoms that are involved with the interaction; as long as it prevents the IgE and receptor from forming the van der Waal’s contact, that is enough to prevent the binding.

Lasser: Why does it do that?

Metzger: Because this particular antibody is directed towards a region of the IgE, the binding to which interferes with the interaction between the IgE and the receptor. The critical thing is that those molecules that bind to the IgE and inhibit the binding don’t at the same time bind to molecules of IgE that are already bound. That’s a perfect way of triggering the cell: to use an anti-IgE that aggregates the IgE.

MacGlashan: In the modelling that you and Bryon and his team have done, have you identi¢ed a hub in that signalling network?

Metzger: To some extent Syk would be a hub, as would Lyn. The Lyn phosphorylates Syk which then phosphorylates other molecules.

MacGlashan: I know they have this interesting path analysis where they work out which species are actually dominant in the network.

Metzger: The current concept is that there is some sort of an initial signalling

complex, and that is what that diagram showed. Almost any molecule in that initial complex might be an e¡ective target. We know that in the case of the inhibitor of the intrinsic kinase of the EGF receptor, this is an approach that has some therapeutic possibilities. Whether that would be better than preventing the initial sensitization of the cells, which a small molecule equivalent of anti-IgE could do, I don’t know. I suppose one could say that if we had a way of preventing the interaction with an intracellular hub, we could reverse an allergic reaction, which the anti-IgE can’t do. But this is a much more complex approach.

Galli: There are some recent observations that IgE, prepared in a way to ensure as much as can be done experimentally that one is dealing with monomers, might have e¡ects on mast cells survival and function that are independent of any agent added speci¢cally to induce FceRI aggregation. The two reports (Asai et al 2001, Kalesniko¡ et al 2001) agreed on several points, one of which is that these IgEs would inhibit the development of apoptosis in mouse mast cells from which growth factors have been withdrawn. I will mention two reservations about these two studies. First, they were done entirely using mouse cells, and second, the work was done entirely in vitro. Moreover, there were other observations in the two papers that di¡ered. In the study by Kalesniko¡ et al (2001), the IgE antibody preparations induced the cells to release cytokines, interestingly without evidence that they were releasing histamine. In contrast, in the study by Asai et al (2001), the antibodies did not induce the release of cytokines.

Both groups of course were curious about why they had obtained apparently contradictory results. They discovered that, for the most part, they had been dealing with di¡erent panels of monoclonal IgE antibodies. Most of the IgE antibodies have now been tested in both labs. The simplest way to summarize our results is that when we tested the antibodies used by Kalesniko¡ et al, we got results similar to those that they had reported. In other words, there are some antibodies that can induce an enhanced resistance to apoptosis without inducing detectable cytokine secretion, and there are some that both enhance the resistance to apoptosis and also induce cytokine secretion (Kitaura et al 2003). In our hands, the latter antibodies can also induce histamine release. We performed indirect tests on whether the antibodies can induce FceRI aggregation, such as examining Syk dependence using Syk knockout mast cells, or using monovalent hapten to inhibit the reaction. These experiments showed that the IgE antibodies can induce responses that are like those induced by aggregation of FceRI, in that they are completely Syk dependent and they can be inhibited by monovalent haptens (Kitaura et al 2003). But there is, as yet, no formal proof that FceRI aggregation actually occurs in this setting. If it does, we don’t know how it happens, although it could be related to the speci¢c structural properties of the individual IgE molecules. Moreover, some of the observations have been replicated using serum-free medium (Kalsniko¡ et al 2001, Kitaura et al 2003). So it is unlikely that the phenomenon simply re£ects a reaction of the IgE with an unknown factor present in serum. However, so far, these observations lack a mechanistic explanation. Note: since the conference was held, it has been reported that one of the IgE antibodies that can induce the strongest enhancement of mast cell survival and mediator secretion has an antigen binding site that exists in two di¡erent conformations, one of which binds DNP and the other of which binds an unrelated antigen (James et al 2003). The potential ‘multispeci¢city’ of IgE antibodies, based on the conformational diversity of the antigen binding site, represents an intriguing clue to the mechanism by which certain IgE antibodies might induce FceRI aggregation in the absence of ‘speci¢c’ antigen (James et al 2003, Foote 2003).

Metzger: In some unpublished work (C. Torigoe), the two di¡erent kinds of IgEs behave di¡erently on acrylamide gels as if the stimulatory IgE might be an aggregate.

Galli: Even the antibodies that don’t induce detectable cytokine or histamine release can enhance resistance to apoptosis. This e¡ect can also be blocked by monovalent hapten. One might argue that enhanced resistance to apoptosis on the withdrawal of growth factors is one of the most sensitive ways the cell can indicate that something is happening with the receptor that is very much like aggregation, if not aggregation itself. If that is true, then mostly everything can be explained.

Lasser: Is it not the case that this IgE molecule might elicit anti anti-IgE molecules?

Sampson: As far as I know, no one has reported any antibodies to the antibody when it is given i.v. or subcutaneously.

MacGlashan: There was a report where it was inhaled. It wasn’t really in patients receiving the drug.

Sampson: In the one case I know of it was inhaled, but it hasn’t been seen in subcutaneous or i.v. administered antibody.

Kricek: I wanted to extend this one step further. There exist anti-IgE autobodies, which occur naturally in the serum and which are non-analaphylactogenic. There are not only those which see the receptor binding sites of IgE. The same holds true for anti-a chain antibodies. There are some which are anaphylactogenic, some which are not, and some which become anaphylactogenic if you remove IgE from the receptor. The idea was always that a therapeutic anti-IgE would have to compete with a high a⁄nity interaction. One knows that just a couple of receptors which are still able to be triggered by IgE are su⁄cient to induce anaphylactic reactions. People always said that such a therapeutic approach wouldn’t really work, even using a high-a⁄nity anti-IgE antibody. The antibody that has been developed by Novartis is only moderate. Nevertheless, clinical data show that it works. One of the explanations was the down-regulatory e¡ect on the receptor of removing IgE from the circulation. Another aspect has not yet been discussed: what if you generate anti-idiotypic antibodies? These would then mimic the part of IgE that interacts with the receptor. They might bind to the receptor and potentiate the e¡ect of the passive anti-IgE immunotherapy. I don’t know whether anyone who has developed therapeutic anti-IgE antibodies has already looked into this. The consequence would be that we shouldn’t stu¡ half a gram of this anti-IgE into the patient, but instead we might formulate it as a vaccine and get the same protective e¡ect with much less material.

Gould: I wanted to clarify the statement concerning the comparison of the two IgEs on gel electrophoresis. Did you mean when you said that it acted as an aggregate that it had oligomers? Or could the fact that one is slower be due to di¡erential glycosylation?

Metzger: That still needs to be looked into. It would have to be a very substantial di¡erence in glycosylation.

Marone: We have approached the question of the binding of monomeric IgE from a di¡erent angle. In collaboration with Lars Bj˛rck of the University of Lund. They puri¢ed a protein (protein L) from a bacterium (Peptostreptococcus magnus) which has four binding domains (B1^B4) for k light chains. Each binding domain, B1^B4, was originally described to have a single binding domain for the k light chains. We found that protein L and B1^B4 were potent stimuli for the release of preformed and de novo synthesized mediators from human basophils and mast cells (Genovese et al 2003). Bj˛rck and collaborators recently produced the recombinant B1 that apparently had a single binding domain for k light chains. We found that B1 is also a very important stimulus for the release of mediators from basophils and mast cells. However, Bj˛rck and collaborators went back and did more careful NMR studies. These revealed that there are two binding sites for k light chains on each domain of B1^B4, which turned out to be the explanation for our results.

MacGlashan: The ‘dog in the manger’ idea is applicable in your experiments to two di¡erent antigen speci¢cities, where one draws the Lyn kinase from the other’s reaction, e¡ectively. The lower-a⁄nity one is using up the available Lyn kinase, so that the higher a⁄nity interaction doesn’t seem to take place. Hugh Sampson and I had a conversation earlier about whether or not that would apply when you £ood the cell with a low a⁄nity IgE to the same antigen to which there might be some high a⁄nity interaction. In e¡ect, the same antigen is binding to a low a⁄nity IgE and a high a⁄nity IgE on the cell. If there was enough low a⁄nity IgE it would act in the same way. The lower a⁄nity interaction might not be able to trigger, but it would give you that ‘dog in a manger’ e¡ect, so that it gives you a disproportionately poor response in the cell. Have you ever done an experiment like this, where you had IgEs of di¡ering a⁄nities to the same antigen?

Metzger: No, we have used cross-reacting antigens with di¡erent a⁄nities. The ‘dog in a manger’ e¡ect is done with two antigens, both of which react with the same IgE, but which have very di¡erent a⁄nities.

MacGlashan: So that would be the application of this idea. Part of the anaphylactic story is whether or not, at any given moment, your a⁄nity for the antigen has shifted to one that is dominantly low or dominantly high.

Metzger: As you know, this is an area that those interested in modifying the action of T cells are exploring. By using so-called ‘ligand antagonists’ they hope to prevent T cell stimulation.

Finkelman: There is a very similar phenomenon in the B cell world. This is that if B cells don’t express IgM or some other surface immunoglobulin, they die. This is despite there not being any established ligand for the IgM or IgD on these B cells. The idea is that there is some weak a⁄nity interaction that causes a tiny amount of cross-linking, but it is not really established that cross-linking is critical.

Galli: Trying to prove the presence or absence of a tiny amount of cross-linking is rather di⁄cult.

Finkelman: You mentioned that you were using a univalent ligand as an inhibitor. Which one do you use for FceRI?

Galli: Depending on the speci¢city of the IgE antibody being tested, we used DNP-lysine, or TNP-glutamate (Kitaura et al 2003).

Finkelman: You are inhibiting the aggregation then.

Galli: Presumably. This speaks to Henry’s point about whether some low level of aggregation will also have an anti-apoptotic e¡ect. If the important event in this phenomenon is a low level of FceRI aggregation, then, yes, this can have an anti- apoptotic e¡ect and it is inhibited by the monovalent hapten.

Metzger: The critical point, which is confusing, is that it suggests that spontaneous aggregation is occurring via the antibody combining sites. It is strange. They used a hapten that was speci¢c for the speci¢city of that IgE.

Finkelman: Do you get binding of the IgE to plastic when you do the in vitro culture? One of the ways of promoting T cell signalling is to coat plastic in anti- CD3 antibody. Do you see the same thing with the IgE antibody?

Galli: I don’t think that any of the experiments have involved an attempt to bind the IgE to plastic.

Finkelman: As an accident? Is there anything to prevent it from occurring?

Galli: I can’t answer that. One would have to investigate this possibility directly.

References

Asai K, Kitaura J, Kawakami Y et al 2001 Regulation of mast cell survival by IgE. Immunity 14:791^800

Foote J 2003 Immunology. Isomeric antibodies. Science 299:1327^1328

Genovese A, Borgia G, Bjorck L et al 2003 Immunoglobulin superantigen protein L induces IL- 4 and IL-13 secretion from human Fc epsilon RI+ cells through interaction with the kappa

light chains of IgE. J Immunol 170:1854^1861

James LC, Roversi P, Taw¢k DS 2003 Antibody multispeci¢city mediated by conformational diversity. Science 299:1362^1367

Kalesniko¡ J, Huber M, Lam V et al 2001 Monomeric IgE stimulates signaling pathways in mast cells that lead to cytokine production and cell survival. Immunity 14:801^811

Kitaura J, Song J, Tsai M et al 2003 Evidence that IgE molecules mediate a spectrum of e¡ects on mast cell survival and activation via aggregation of the Fc epsilonRI. Proc Natl Acad Sci USA 100:12911^12916