Compartmentalized activation of the high affinity immunoglobulin E receptor within membrane domains.

The earliest known step in the activation of the high affinity IgE receptor, FcεRI, is the tyrosine phosphorylation of its β and γ subunits by the Src family tyrosine kinase, Lyn. We report here that aggregation-dependent association of FcεRI with specialized regions of the plasma membrane precedes its tyrosine phosphorylation and appears necessary for this event. Tyrosine phosphorylation of β and γ occurs in intact cells only for FcεRI that associate with these detergent-resistant membrane domains, which are enriched in active Lyn. Furthermore, efficient in vitro tyrosine phosphorylation of FcεRI subunits occurs only for those associated with isolated domains. This association and in vitro phosphorylation are highly sensitive to low concentrations of detergent, suggesting that lipid-mediated interactions with Lyn are important in FcεRI activation. Participation of membrane domains accounts for previously unexplained aspects of FcεRI-mediated signaling and may be relevant to signaling by other multichain immune receptors.

The earliest known step in the activation of the high affinity IgE receptor, Fc⑀RI, is the tyrosine phosphorylation of its ␤ and ␥ subunits by the Src family tyrosine kinase, Lyn. We report here that aggregation-dependent association of Fc⑀RI with specialized regions of the plasma membrane precedes its tyrosine phosphorylation and appears necessary for this event. Tyrosine phosphorylation of ␤ and ␥ occurs in intact cells only for Fc⑀RI that associate with these detergent-resistant membrane domains, which are enriched in active Lyn. Furthermore, efficient in vitro tyrosine phosphorylation of Fc⑀RI subunits occurs only for those associated with isolated domains. This association and in vitro phosphorylation are highly sensitive to low concentrations of detergent, suggesting that lipid-mediated interactions with Lyn are important in Fc⑀RI activation. Participation of membrane domains accounts for previously unexplained aspects of Fc⑀RI-mediated signaling and may be relevant to signaling by other multichain immune receptors.
The plasma membrane contains specialized regions that have distinct compositions and can serve unique functions in the regulation of cell surface receptor activation. For example, caveolae have been shown to associate with certain signaling proteins (1,2) and have been implicated in receptor activation (3)(4)(5)(6), vesicular transport (7,8), and the uptake of small molecules (9). Compositionally related membrane domains, which lack the invaginated morphology of caveolae as well as the membrane protein caveolin, have also been identified and biochemically separated from caveolae (10). These membrane domains, like caveolae, are resistant to solubilization in nonionic detergents such as Triton X-100, are enriched in sphingolipids and glycosylphosphatidylinositol-linked proteins, and are associated with palmitoyl-anchored signaling molecules including Src family tyrosine kinases (10 -14). Detergent-resistant membrane domains isolated from rat basophilic leukemia (RBL) 1 cells, a mast cell line, contain at least 30% of the cellular Lyn, a Src family tyrosine kinase, and no detectable caveolin (15).
Aggregation of Fc⑀RI on mast cells and basophils by multivalent antigens leads to phosphorylation of immunoreceptor tyrosine-based activation motifs within the ␤ and ␥ receptor subunits by Lyn (16 -19). This initiates a signaling cascade culminating in secretion of inflammatory mediators and cytokines that play an important role in the allergic response (20,21). The molecular mechanism by which aggregation of Fc⑀RI initiates its phosphorylation by Lyn is incompletely understood. Selective binding of Lyn directly to unphosphorylated Fc⑀RI ␤ (22) has been proposed to mediate an initial transphosphorylation of aggregated Fc⑀RI (23), but this does not account for the capacity of Fc⑀RI lacking the ␤ subunit (24,25) or chimeric receptors containing only the ␥ cytoplasmic tail (26 -28) to become tyrosine-phosphorylated upon aggregation. The involvement of detergent-resistant membrane domains in Fc⑀RI signaling was recently suggested by the observation that aggregation of Fc⑀RI on RBL cells significantly increased the amount of active Lyn associated with these structures (15). Furthermore, fluorescence microscopy studies showed that aggregation of Fc⑀RI at the surface of intact cells co-redistributes ganglioside-enriched membrane patches that are related to the isolated membrane domains (29,30). The aggregation-dependent association of Fc⑀RI with these less fluid regions of the membrane (30,31) is also consistent with decreased lateral and rotational mobility of aggregated Fc⑀RI (reviewed in Ref. 32). In the present study, we establish conditions for preserving the interaction of aggregated Fc⑀RI with these membrane domains following cell lysis, and we demonstrate the importance of this interaction to the initial step in signaling, the tyrosine phosphorylation of Fc⑀RI.
Immunoblotting-Electrophoresis of samples was carried out on 12.5% acrylamide SDS gels under nonreducing conditions, and semidry transfer to Immobilon P (Millipore, Bedford, MA) was performed as described (15). Anti-phosphotyrosine immunoblots were performed using 0.1 g/ml monoclonal antibody 4G10 conjugated to horseradish peroxidase (UBI, Lake Placid, NY) and Supersignal ECL substrate (Pierce, Rockford, IL). For the results in Fig. 2, tyrosine phosphorylation of the Fc⑀RI ␤ subunit was quantified from anti-phosphotyrosine immunoblots of post-nuclear supernatants of 10 6 RBL cells lysed in 0.2% Triton X-100. The prominent 34-kDa band detected in these blots after Fc⑀RI stimulation was identified as ␤ based on selective immunodepletion by IgE-specific agarose beads. 2 This band was quantified with a 256 gray-scale scanner (Umax Vista-S6E) and NIH Image software.
Immunoprecipitations-After adjusting the sucrose fractions to 0.2% Triton X-100 to extract Fc⑀RI from the membrane domains, Fc⑀RI was immunoprecipitated for 90 min with trinitrophenyl-conjugated Sepharose 4B (which efficiently binds anti-DNP IgE). The immunoprecipitates were washed twice with 0.2% Triton X-100 and once in lysis buffer lacking detergent prior to elution by boiling in nonreducing SDS sample buffer.
In Vitro Kinase Assays-Kinase assays were performed by adding kinase buffer (20 mM Tris, pH 7.6, 10 mM MgCl 2 , 1 mM ATP, and 1 mM Na 3 VO 4 ) to the sucrose fractions and incubating at 37°C for 15 min. The reaction was quenched either with 5 ϫ nonreducing SDS sample buffer or by adding 50 mM EDTA followed by immunoprecipitating Fc⑀RI.

RESULTS AND DISCUSSION
In order to determine if the interaction of Fc⑀RI with membrane domains is involved in the activation of this immunoreceptor, we developed conditions that preserve this association during the isolation of these complexes by equilibrium sucrose density ultracentrifugation. As shown in Fig. 1A, limiting amounts of Triton X-100 used for cell lysis preserve the association of aggregated Fc⑀RI (q, f) with the detergent-resistant membrane domains which migrate as low density vesicles (fractions 3-7). In 29 separate experiments, 54 Ϯ 7% of biotin-IgE Fc⑀RI complexes aggregated with streptavidin associate with the membrane domains (q). Significant but lesser amounts of antigen-aggregated receptors associate (f; 11 Ϯ 1%, n ϭ 6), most likely reduced by the partial reversal of IgE-antigen binding during the overnight ultracentrifugation. In contrast, monomeric Fc⑀RI (E) is nearly absent from the membrane domains (3 Ϯ 1%, n ϭ 26) and found almost entirely in the 40% sucrose fractions containing solubilized proteins (fractions 10 -16). The association of Fc⑀RI with isolated membrane domains depends on its aggregation at the cell surface, as less than 5% association is seen for Fc⑀RI aggregated after cell lysis or for Fc⑀RI aggregated with antigen on cells and then dissociated with monovalent hapten after lysis. 2 The interaction between aggregated Fc⑀RI and the membrane domains is very sensitive to the detergent:cell lipid ratio during solubilization and ultracentrifugation, as indicated by its disruption when concentrations of Triton X-100 greater than 0.05% are used (15). This sensitivity is similar to that observed by Pribluda et al. (23) for Fc⑀RI coupling to Lyn in cell lysates, and it contrasts with cytoskeleton-mediated detergent insolubility of aggregated Fc⑀RI (33)(34)(35), which is not disrupted by high Triton X-100 concentrations.
Although this reduction in Triton X-100 used for cell lysis dramatically increases the amount of aggregated Fc⑀RI that remains associated with detergent-resistant membrane domains, these domains are otherwise very similar to those isolated after lysis in high Triton X-100 (Ն0.2%). When directly compared, domains from low and high detergent lysis conditions contain the same fraction of cellular Lyn, and neither has detectable amounts of Src. 2 In addition, both preparations contain a similar spectrum of tyrosine kinase substrates as revealed in in vitro tyrosine kinase assays (Ref. 15 and as described below), and both contain similar amounts of cellular protein (Ͻ2% of the total). 2 By these criteria, the domains obtained using 0.05% Triton X-100 for cell lysis appear to be identical to other membrane domains previously described that do not contain caveolin (10,(13)(14)(15). Furthermore, the aggregation-dependent association of Fc⑀RI with membrane domains shows selectivity among transmembrane cell surface receptors, as Fc⑀RI but not Type I interleukin-1 receptors, both expressed on Chinese hamster ovary cells, associate with membrane domains following aggregation. 3 Association of Fc⑀RI with membrane domains does not require tyrosine phosphorylation of the receptor subunits. As shown in Fig. 1B, RBL cells permeabilized with Streptolysin O in the presence of excess EDTA to inhibit kinase activity show a similar amount of aggregation-dependent association of Fc⑀RI with domains as intact cells (Fig. 1A). As previously shown with broken cells (36), stimulated tyrosine phosphorylation of Fc⑀RI ␤ and other substrates is prevented by EDTA in these permeabilized cells. 2 The presence of Lyn and aggregated Fc⑀RI within the same subregions of the plasma membrane suggests that domainassociated Lyn could be responsible for the initial phosphorylation of the immunoreceptor tyrosine-based activation motifs. Fc⑀RI associates with membrane domains very rapidly at 37°C (E, Fig. 2A) and is more than 50% complete within 30 s, whereas substantially less than 50% of the maximal tyrosine phosphorylation of Fc⑀RI ␤ occurs during this time (q, Fig. 2A). The amount of ␤ tyrosine phosphorylation declines after 2 min at 37°C, and domain-associated receptor also decreases between 5 and 30 min in parallel with its internalization. 2 At 4°C, the association of Fc⑀RI with domains occurs more slowly (E, Fig. 2B), but is clearly more rapid than the ␤ tyrosine phosphorylation during the first 5 min (q, Fig. 2B). Fc⑀RI internalization and downstream signaling such as Ca 2ϩ mobilization and phosphatidylinositol hydrolysis do not occur at 4°C (37), indicating that they are not required for domain association. These results demonstrate that association of Fc⑀RI with membrane domains on cells is an early, aggregationdependent event that is sufficiently rapid to mediate receptor tyrosine phosphorylation.
Evidence for Fc⑀RI tyrosine phosphorylation occurring within membrane domains of intact cells is shown in Fig. 3. Stimulation of biotin-IgE-sensitized RBL cells with streptavidin dramatically increases the tyrosine phosphorylation of many proteins. When lysates of these cells are analyzed by sucrose gradient ultracentrifugation, most of the proteins with 3 K. A. Field, D. Holowka, and B. Baird, manuscript in preparation.  (15) were stimulated for 5 min at 37°C with 500 ng/ml antigen (f, DNP-bovine serum albumin), 10 nM streptavidin (q), or left unstimulated (E) prior to lysis in 0.025% Triton X-100 at 4 ϫ 10 6 cells/ml. Lysates were then diluted 1:1 with 80% sucrose containing 0.025% Triton X-100 and loaded into sucrose step gradients (right axis) followed by ultracentrifugation at 250,000 ϫ g overnight at 4°C. After fractionating the gradients in 0.2-ml aliquots, the distribution of 125 I-IgE-Fc⑀RI was determined and is expressed as the fraction of total 125 I present in the gradient, including the pellet. Error bars show the range of duplicate gradients run on the same day. B, RBL cells sensitized as above and suspended in 20 mM phosphate, pH 7.5, 150 mM NaCl, and 5 mM EDTA were permeabilized with 0.4 units/ml of Streptolysin O (Burroughs-Wellcome, Research Triangle Park, NC) for 15 min at 37°C. The permeabilized cells were then stimulated with streptavidin (q) or not (E) followed by lysis and ultracentrifugation as in A. Staining with Trypan blue confirmed that nearly 100% of the cells were permeabilized by Streptolysin O, and immunoblotting showed no stimulated phosphorylation of Fc⑀RI ␤ under these conditions. 2 enhanced tyrosine phosphorylation are found with the solubilized proteins at 40% sucrose (fractions 11-16), as expected (15). Associated with the membrane domains (fractions 3-8) after stimulation are tyrosine-phosphorylated proteins of approximately 90, 53/56, 45, 34, and 25-30 kDa. The 53/56-kDa doublet was identified as Lyn by reprobing the blot with rabbit anti-Lyn (UBI). 2 Significantly, the 45-, 34-, and 25-30-kDa bands appear only with stimulation and are markedly enriched in membrane domains relative to the other fractions. The domain-associated proteins of 34 and 25-30 kDa correspond to phosphorylated ␤ and ␥ 2 Fc⑀RI subunits, respectively, as identified by immunoprecipitating Fc⑀RI from the sucrose gradient fractions (Fig. 3B). Fig. 3 clearly shows that the tyrosine-phosphorylated ␤ and ␥ subunits are almost entirely associated with membrane domains. The majority of other tyrosine kinase substrates phosphorylated as the result of Fc⑀RI aggregation are located in the solubilized protein fractions, presumably because they are cytosolic or associated with membranes that are solubilized in 0.05% Triton X-100. Syk, the ZAP-70-related tyrosine kinase responsible for phosphorylating the majority of substrates downstream of Fc⑀RI (19,38,39), is also found exclusively in these soluble fractions, 2 as expected because activated Syk is localized primarily in the cytosol after receptor stimulation (40,41). 2 Thus, following Fc⑀RI aggregation on cells, Lyn phosphorylates the ␤ and ␥ subunits of domainassociated receptors. This apparently leads to a transient association and the activation of Syk, followed by Syk-mediated phosphorylation of downstream substrates, most of which are not stably associated with membrane domains.
Additional support for the involvement of these domains in initiating Fc⑀RI activation comes from in vitro tyrosine kinase assays performed on sucrose fractions, followed by immunoprecipitation of Fc⑀RI in the presence of 0.2% Triton X-100 (which releases Fc⑀RI from membrane domains). Fig. 4A shows that aggregated Fc⑀RI associated with membrane domains isolated after cell lysis in 0.05% Triton X-100 (MDϩ) are efficiently tyrosine-phosphorylated in vitro, whereas receptors in the sucrose fractions containing solubilized proteins (40ϩ and 40Ϫ) are not phosphorylated, and membrane domains from unstimu-lated cells (MDϪ) also show no phosphorylated Fc⑀RI. Streptavidin-aggregated Fc⑀RI from cells lysed in 0.2% Triton X-100 migrate at a high density (50 -70% sucrose) in these gradients (15). When in vitro tyrosine kinase assays are performed on these high density sucrose fractions (HDϩ), a relatively small amount of ␤ subunit phosphorylation is seen. This fraction does contain a small amount of Lyn that may be responsible for the phosphorylation detected, 2 but it is not known whether this represents Lyn directly associated with Fc⑀RI, Lyn contaminating this fraction from the 40% sucrose fraction, or fragments of membrane domains which remain receptor-associated in 0.2% Triton X-100. Consistent with the last possibility, the in vitro phosphorylation in the HDϩ fraction shows a Triton X-100 sensitivity similar to that of the MDϩ fraction (see below).
When in vitro tyrosine kinase assays are performed on membrane domains, stimulated Fc⑀RI phosphorylation is highly sensitive to the concentration of Triton X-100 present. As shown in Fig. 4B, addition of submicellar (0.01%) Triton X-100 to the membrane domains causes a slight enhancement of Fc⑀RI ␤ and ␥ tyrosine phosphorylation, but higher Triton X-100 concentrations dramatically reduce this phosphorylation. The association of Lyn with membrane domains, as well as its activity toward the exogenous substrate, enolase, is not significantly affected by Triton X-100, 2 and neither is phosphorylation of Lyn itself or the 45-kDa substrate (Fig. 4B). The exquisite sensitivity of Fc⑀RI in vitro phosphorylation to Triton X-100 indicates that lipid-mediated association of these receptors with the membrane domains is required for this activation step. Fig. 4C supports this conclusion, demonstrating that treatment of isolated membrane domains with 0.05% Triton X-100 (f) dissociates the aggregated receptors from these membrane domains as indicated by reanalysis on a second sucrose gradient. Treatment of the same domains with 0.01% Triton X-100 (q) does not dissociate the receptor, although it does cause a slight change in the distribution of the receptor within the gradient relative to untreated membrane domains (E). Thus, Fc⑀RI associated with membrane domains is functionally coupled to Lyn, and this association is easily disrupted by detergent.
The involvement of membrane domains in this early step of Fc⑀RI activation provides a new model in which the initial phosphorylation of the ␤ and ␥ subunits by Lyn is mediated by lipid-protein interactions. Although previous results have explained how the phosphorylation of Fc⑀RI ␤ and ␥ and subsequent events proceed after the association of active Lyn with a receptor cluster (23,25,38,39), the structural basis for the initial interaction between Lyn and Fc⑀RI in the activation process has remained poorly defined. Several studies that detected association of Lyn with unstimulated Fc⑀RI utilized methods that could stabilize the association of these receptors with membrane domains, including chemical cross-linking (18) or low detergent:cell lipid ratios (23). We find that unstimulated receptors do not co-isolate with the Lyn-containing domains to a large extent (Fig. 1), although weak and/or transient interactions could occur on intact cells. The size and stability of the domains on the surface of intact, unstimulated cells are unknown. These domains are likely to be small and dynamic in composition, but appear to coalesce together with aggregated Fc⑀RI (29,30). Thus, localization of Lyn within membrane domains could serve to sequester this kinase away from Fc⑀RI prior to receptor aggregation and, in turn, provide a pool of active or readily stimulated Lyn for aggregated Fc⑀RI that stably associate with the domains. Support for this aspect of the model comes from our observation that isolated membrane domains contain abundant tyrosine kinase activity even in the absence of Fc⑀RI activation (15), as well as from other studies on Src family members that associate with detergent-resistant membrane domains. These other investigations have found that Fyn, Lck, and Fgr associated with isolated membrane domains show higher specific activity in vitro than soluble forms of these kinases (42,43), possibly because of the capability for kinases concentrated within domains to trans-autophosphorylate readily. 4 Our results demonstrate that aggregation of Fc⑀RI causes its rapid and efficient association with specialized domains in the plasma membrane that are enriched in the tyrosine kinase, Lyn. Fc⑀RI associated with membrane domains are rapidly tyrosine-phosphorylated in intact cells, and this phosphorylation is also observed in vitro preferentially for receptors associated with membrane domains. The interaction of Fc⑀RI with these specialized membrane domains does not depend on the ␤ subunit, 3 and thus can account for the initiation of receptor signaling independent of specific protein-protein interactions between the ␤ subunit and Lyn (24 -28). In addition, this receptor-membrane domain association may facilitate coupling to processes such as Ca 2ϩ mobilization, lipid metabolism, and exocytic vesicle fusion. Recent evidence indicates that specialized membrane domains, including caveolae, are involved in the signaling of other cell surface receptors such as certain growth factor receptors (3,5) and glycosylphosphatidylinositollinked mitogenic receptors (12,44,45). Receptor-domain interactions also may be important for other multichain immune recognition receptors that utilize Src family kinases during their initial signaling steps (46,47). FIG. 4. In vitro tyrosine phosphorylation of Fc⑀RI associated with membrane domains. A, preferential phosphorylation of Fc⑀RI associated with membrane domains from stimulated cells. Kinase assays were performed on fractions from sucrose gradients without Na 3 VO 4 containing either detergent-resistant membrane domains (MD), the 40% sucrose fraction (40), or immune complexes in high density sucrose (HD), from either unstimulated RBL cells (Ϫ) or cells stimulated with 10 nM streptavidin for 5 min at 37°C (ϩ). Fc⑀RI was then immunoprecipitated and subjected to anti-phosphotyrosine immunoblotting as in Fig. 3B. The relative amount of 125 I-IgE loaded in each lane is (left to right) 0.04, 1.00, 0.76, 0.34, and 0.42. Experiments where ATP and Mg 2ϩ were omitted from the kinase assay, or where cells without IgE were used, showed no detectable phosphorylation of ␤ and ␥ under these conditions. 2 B, detergent sensitivity of in vitro Fc⑀RI phosphorylation. Membrane domains isolated as in Fig. 1 from streptavidin-stimulated cells were incubated with the indicated concentration of Triton X-100 prior to performing kinase assays on the sucrose fractions. The samples were then boiled with SDS, electrophoresed, and immunoblotted with anti-phosphotyrosine. C, extraction of Fc⑀RI from membrane domains with Triton X-100. Membrane domain fractions isolated from streptavidin-stimulated cells as for B were treated with no Triton X-100 (E), with 0.01% Triton X-100 (q), or with 0.05% Triton X-100 (f), readjusted to 40% sucrose, and ultracentrifuged overnight within a sucrose gradient (right axis) as in Fig. 1.