|Home | About | Journals | Submit | Contact Us | Français|
Crosslinking of IgE-bound FcεRI triggers mast cell degranulation. Previous FRAP and phosphorescent anisotropy studies suggested that FcεRI must immobilize to signal. Here, single quantum dot (QD) tracking and hyperspectral microscopy methods are used to redefine relationships between receptor mobility and signaling. QD-IgE-FcεRI aggregates of at least three receptors remain highly mobile over extended times at low concentrations of antigen that induce Syk kinase activation and near-maximal secretion. Multivalent antigen, presented as DNP-QD, also remains mobile at low doses that support secretion. FcεRI immobilization is marked at intermediate and high antigen concentrations, correlating with increases in cluster size and rates of receptor internalization. The kinase inhibitor PP2 blocks secretion without affecting immobilization or internalization. We propose that immobility is a feature of highly crosslinked immunoreceptor aggregates, is a trigger for receptor internalization, and is not required for tyrosine kinase activation leading to secretion.
The T-cell receptor (TCR), B-cell receptor (BCR) and high affinity IgE receptor (FcεRI) typify the multichain immune recognition receptor (MIRR) family and control key events in the immune response (Sigalov, 2004). The consensus is that these receptors share a common mechanism of activation, where receptor crosslinking initiates Src family-mediated phosphorylation of immunoreceptor tyrosine based activation motifs (ITAMs), recruitment of Syk/ZAP-70 and propagation of signaling (Boniface et al., 1998; Kraft and Kinet, 2007; Thyagarajan et al., 2003). However, additional mechanisms may contribute to signal initiation, as suggested by the persistence of FcεRI signaling in Lyn-deficient mast cells (Hernandez-Hansen et al., 2004; Nishizumi and Yamamoto, 1997; Rivera and Gilfillan, 2006). Crosslink-induced changes in receptor conformation may also be a contributing factor, as suggested by studies in the TCR and BCR systems (Gil et al., 2002; Tolar et al., 2008).
In each of these systems, antigen binding is associated with changes in receptor dynamics and topography (Dustin and Cooper, 2000; Thyagarajan et al., 2003; Wilson et al., 2000). In T cells, it has been shown that contact with surface-associated antigen induces formation of TCR microclusters, which signal actively while undergoing actin-mediated transport through the peripheral supramolecular activation complex (pSMAC), and cease signaling in the central supramolecular activation complex (cSMAC) where they are largely immobile (Campi et al., 2005; Yokosuka et al., 2005). In the FcεRI system, many measurements based on fluorescence recovery after photobleaching (FRAP) and rotational mobility have shown that IgE-FcεRI mobility decreases dramatically upon addition of multivalent antigen (Mao et al., 1991; Menon et al., 1986b; Myers et al., 1992; Pecht et al., 1991; Pyenta et al., 2003; Tamir et al., 1996; Zidovetzki et al., 1986). These studies demonstrate that an activation-associated change in receptor mobility is a common feature of MIRRs.
However, the relationship between mobility changes and signaling is not entirely clear. In the case of FcεRI, most investigators have concluded that receptor immobilization is a prelude to degranulation (Menon et al., 1986a; Menon et al., 1986b; Myers et al., 1992; Pecht et al., 1991; Tamir et al., 1996). One early FRAP study differed from this view, by showing that addition of anti-FITC antibodies to FITC-IgE-labeled FcεRI induced degranulation without causing significant decreases in FcεRI mobility (Schlessinger et al., 1976). Furthermore, previous SEM studies (Seagrave et al., 1991) showed that formation of small oligomers (chains and small clusters) by anti-IgE antibodies was associated with robust degranulation, whereas the formation of large aggregates at higher doses of anti-IgE was associated with reduced secretion. Because larger aggregates of receptors would be expected to diffuse more slowly (Kusumi et al., 2005; Peters and Cherry, 1982), it was suggested that immobilization of receptors would more likely be correlated with signal termination than signal initiation. It is also notable that, despite over a decade of literature linking FcεRI tyrosine phosphorylation to signaling (reviewed (Rivera and Gilfillan, 2006)), roles for ITAM tyrosine phosphorylation in the process of receptor immobilization have not been defined.
To re-examine the relationship between antigen-induced changes in FcεRI mobility and signaling at high resolution, we generated quantum dot (QD)-IgEanti-DNP probes that bind FcεRI without activating the cell (Andrews et al., 2008) and also multivalent 2,6-dinitrophenyl-QD (DNP-QD) probes that can activate cells by crosslinking receptors that are primed with IgEanti-DNP. These novel probes were used to study the mobility of both unstimulated and crosslinked receptors by single particle tracking (SPT). We report that FcεRI immobilization in the RBL-2H3 mucosal mast cell line, as well as in bone marrow-derived murine mast cells (BMMC), is markedly dependent on both antigen dose and valency. Further, we show that at low antigen concentrations or low valency, crosslinked FcεRI aggregates are small, signaling competent and mobile. Our data indicate that receptor immobilization is strongly correlated with receptor internalization but is not required for signal initiation.
Our first goals in this study were to establish the dose-dependency of FcεRI immobilization and to determine if immobilization correlates with important outcomes of FcεRI signaling, such as receptor phosphorylation, increases in intracellular calcium and degranulation. Mast cells were labeled with QD-IgE at low concentrations which permit SPT (see Experimental Procedures). The remaining FcεRI sites were saturated with dark (unlabeled) IgEanti-DNP, such that all FcεRI were capable of binding and responding to multivalent antigen. Single QD tracking was initiated at 20–33 frames/s and cells were stimulated by the addition of varying doses of multivalent antigen, DNPn-BSA (Video 1).
From these time-series images, we calculated the mean instantaneous diffusion coefficient (see Experimental Procedures). This provides a high temporal resolution measure of the average diffusion rate of IgE-FcεRI before, during, and after crosslinking. We then plotted this value as a function of time and antigen dose (Figure 1A, shown for DNP25-BSA).
Our results demonstrate that the rate and extent of QD-IgE-FcεRI immobilization is highly dependent on both the dose (Figure 1A, Table S1) and valency of antigen (Figure 1C). The addition of low concentrations of highly multivalent antigen (0.001 and 0.01 μg/ml DNP25-BSA) caused very little reduction in mobility. In contrast, 0.1 μg/ml DNP25-BSA caused a slow loss of receptor mobility while 1 and 10 μg/ml DNP25-BSA caused a dramatic reduction in mobility within 20 seconds. Differences in receptor mobility are also illustrated in the individual trajectories shown in Figure 1B, representative of receptors treated with low and high doses of DNP25-BSA. Individual FcεRI remained relatively free to diffuse after treatment with 0.001 μg/ml DNP25-BSA, while treatment with 1 μg/ml DNP25-BSA yielded a large proportion of immobile trajectories (defined as a diffusion coefficient <0.01 μm2/s).
We next addressed the influence of antigen valency on receptor mobility in RBL-2H3 cells. Data in Figure 1C reports FcεRI mobility in cells stimulated with 1 μg/ml BSA conjugated to a range of DNP haptens (valencies of 2, 4, 12 or 25). Note that the distribution of diffusion coefficients for receptors stimulated with DNP2-BSA (red solid line) is indistinguishable from the resting state (solid black line). For cells stimulated with DNP4-BSA, there is only a slight shift to the left (green line) indicating a small reduction in diffusion (but not immobilization) that is consistent with a predicted small aggregate size. Stimulation with the intermediate valency antigen (DNP12-BSA; blue line) results a further reduction in diffusion but still does not approach the highly immobile behavior of receptors crosslinked with DNP25-BSA (orange line).
Figures 1D–G report degranulation responses in RBL-2H3 cells stimulated over the full range of doses and DNP-antigen valencies. Importantly, these data show that signaling occurs under conditions where receptors remain mobile. For example, cells stimulated with low doses of highly multivalent antigen (DNP25-BSA) secrete up to 60% of granule contents; under these conditions, receptors are highly mobile (pink line in Figure 1A and traces in Figure 1B). Cells stimulated with DNP4-BSA show peak degranulation response at 1 μg/ml antigen, when diffusion of receptors is marginally affected (green line, Figure 1C). Receptors crosslinked with the same dose of DNP2-BSA remain fully mobile but can initiate a modest but measurable degranulation response (15%).
Since degranulation is measured in RBL-2H3 cells as total β-hexosaminidase release over a 30 minute period of antigen exposure, we considered the possibility that QD-IgE-FcεRI crosslinked at low antigen concentrations were in fact immobilizing at some later time point. We followed the mean instantaneous diffusion coefficient in response to crosslinking with 0.001 μg/ml DNP-BSA out to 15 minutes and noted no decrease in diffusion (Figure S1). We also simultaneously measured the mobility of QD-IgE-FcεRI and calcium responses in individual cells (Figure S2) and observed both mobile receptors and calcium oscillations, as expected for this low dose of antigen (Lee and Oliver, 1995). These data demonstrate unequivocally that intracellular signaling occurs while QD-IgE-FcεRI remain mobile on the RBL cell surface.
Finally, to ensure that these observations extend to primary cells, we performed similar measurements using murine BMMCs. These data show that FcεRI on BMMC remain highly mobile (Figure 1H) under conditions that stimulate maximal degranulation responses (0.01 μg/ml DNP25-BSA or 1 μg/ml DNP4-BSA, Figure 1I).
The results of early FRAP studies suggested that crosslinked FcεRI aggregates rapidly immobilized (Menon et al., 1986b) and even dimerization has been reported to reduce rotational mobility (Myers et al., 1992). To address this directly, we designed an experiment to observe the motion of small antigen-induced aggregates of FcεRI (Figure 2). Cells were labeled with a mixture of five colors of QD-IgE (QD525-, QD565-, QD585-, QD625-, and QD655-IgE) and imaged using a custom-built confocal, hyperspectral microscope (Sinclair et al., 2006). This instrument collects the emission spectra at each sampled point in the image from 490 to 800 nm (512 wavelengths), enabling the identification of each spectrally distinct QD (Lidke et al., 2007). Using this instrument, confocal time series of all five colors of QD-tagged IgE-FcεRI complexes were acquired at a rate of 0.25 frames/s. In this protocol, a sufficient number of receptors are labeled with QD-IgE to initiate a secretory response upon multivalent antigen challenge (Figure S3), avoiding the need to prime additional receptor sites with dark IgE and providing the ability to track every engaged receptor. Based upon our prior work, we expected to observe only transient co-confinement of multiple, resting IgE-FcεRI (Andrews et al., 2008). Consistent with these expectations, we never observed association of two or more colors of QD for more than one frame (<four seconds) in the absence of multivalent antigen (data not shown). However, upon addition of 0.1 μg/ml DNP25-BSA, we frequently observed prolonged co-diffusion of multiple (i.e., dimers and trimers) QD-IgE-FcεRI. Two examples of this behavior are shown in Figure 2, in which three spectrally distinct QD-IgE-FcεRI form aggregates and diffuse together for many seconds to minutes. The spectral signatures were stable over time, identifying each diffusing group as a stable aggregate composed of at least three QD-IgE-FcεRI.
We developed QD-based multivalent analogs of DNP-BSA (DNP-QD) as an additional approach to confirm our conclusions that small, signaling-competent aggregates of antigen-crosslinked FcεRI are mobile. The DNP-QD probes were generated by reacting biotin-DNP with QD-streptavidin conjugates, with optimization for creating a polyvalent reagent. Although the valency of these probes is not precisely known, we estimate that the range of DNPs per QD is between 10–20. We tested the ability of DNP-QD to trigger degranulation and found a robust, dose-dependent secretory response that was essentially identical for stimulation with DNP-QD655 and DNP-QD585 (Figure 3A).
Cells were primed with IgEanti-DNP and stimulated with either a sub-activating dose (1 pM) or an activating dose (500 pM) of DNP-QD. At the low dose, single QD tracking showed that the diffusion coefficient of DNP-QD655 (0.0377 μm2/s) was slower than that of unstimulated QD655-IgE (0.0823 μm2/s) but not immobile (see Table S1). This is consistent with formation of small aggregates of IgE-FcεRI upon binding to DNP-QD. With the addition of the low, activating dose of DNP-QD585, we noted only a slight decrease in the diffusion of DNP-QD655 (0.0321 μm2/s) (Figure 3B and Video 2). These data indicate that, at doses and valency capable of eliciting degranulation, QD-ligands can both stimulate and permit tracking of multimeric receptor complexes. The mobility of DNP-QD bound to surface receptors further supports our conclusion that signaling-competent, small receptor aggregates remain mobile.
To assess the relationship between signal initiation and immobilization, we performed anti-phosphotyrosine western blotting over a range of DNP25-BSA doses and a time course of 0.5 to 30 min. In Figure 4, we focus on the kinetics and extent of tyrosine phosphorylation of FcεRI and its critical binding partner, Syk. At all doses of antigen (DNP25-BSA), Syk phosphorylation was detectable as early as 30 seconds into stimulation and remained elevated over the 30 min time course (Figure 4A and 4B). Together with the observation of calcium oscillations at low antigen doses (Figure S2), these data suggest that low doses of antigen initiate signaling in the absence of receptor immobilization. The observation of Syk phosphorylation at low doses is consistent with the robust secretory response (Figure 1C) and implies that Syk is efficiently activated by a small fraction of crosslinked receptors.
It is generally thought that Syk activation is preceded by phosphorylation of FcεRIβ and FcεRIγ ITAMs. Phosphorylation of the FcεRIβ ITAM is believed to facilitate Lyn recruitment, whereas FcεRIγ ITAM phosphorylation creates a binding site for the tandem SH2 domains of Syk (Rivera and Gilfillan, 2006). In our analysis, FcεRIβ phosphorylation was detectable after 30s of stimulation at both non-immobilizing and immobilizing antigen doses (Figure 4C and 4D). Unexpectedly, FcεRIγ subunit phosphorylation occurred with slower kinetics than Syk or β phosphorylation, requiring several minutes to reach its maximum level and was not detectable at low doses of DNP25-BSA (Figure 4C and 4E). Thus, the slower kinetics of γ subunit phosphorylation do not match closely with that of Syk. Possible explanations for this disparity are discussed below.
To determine if receptor phosphorylation is required for immobilization, we treated RBL-2H3 cells with the Src-family tyrosine kinase inhibitor PP2, prior to antigen stimulation in the continued presence of PP2. At the dose employed (10 μM), PP2 effectively blocks Src family kinase activity and likely also blocks other downstream tyrosine kinases non-selectively. PP2 treatment reduced secretion to background levels over a wide range of antigen doses (Figure S4). PP2 also dramatically blocked antigen-induced tyrosine phosphorylation in response to crosslinking with 1 μg/ml DNP25-BSA (Figure 5A and 5B). Despite this significant inhibition of downstream events, PP2 treatment did not prevent immobilization of QD-IgE-FcεRI following stimulation with 1 μg/ml DNP25-BSA (Figure 5C, Table S1) nor did it prevent antigen-induced, large-scale aggregation of IgE-FcεRI (Figure S5).
We employed a flow cytometry-based assay to measure internalization of IgE-FcεRI in response to treatment with various concentrations of DNP25-BSA. This assay revealed that antigen-induced FcεRI internalization increased with antigen dose (Figure 5D), a relationship that also correlates with the extent of receptor immobilization. Since immobilization did not rely on tyrosine phosphorylation, we investigated the role of phosphorylation in antigen-induced FcεRI internalization by pretreating identical samples with PP2 prior to addition of antigen. PP2 treatment did not appreciably affect the rate or extent of receptor internalization (Figure 5D). Thus, the processes of receptor immobilization and internalization are both independent of receptor phosphorylation.
We used transmission electron microscopy (TEM) of immunogold labeled membrane sheets to compare the dose-dependence of IgE-FcεRI clustering with the kinetics of IgE-FcεRI immobilization. It is important to note that cluster size is not a direct measure of receptor aggregate size, since even uncrosslinked receptors have a non-random distribution (Figure 6A; Wilson et al., 2000). Increases in cluster sizes in response to antigen, as imaged by electron microscopy, likely reflect both the influence of membrane architecture and the degree of crosslink-induced aggregation. After one minute of stimulation, significant dose-dependent differences in IgE-FcεRI cluster size were clearly visible (Figure 6A–C). Quantification of these images confirmed that the extent of antigen-induced IgE-FcεRI clustering closely parallels receptor mobility. Consistent with previous work (Seagrave et al., 1991), we also noted that very large clusters of IgE-FcεRI form at the higher doses of DNP25-BSA (Figure 6D) where the secretory response is modestly diminished (Figure 1C). These results suggest that crosslinking receptors into large complexes is likely the primary mechanism underlying antigen-induced immobilization.
To evaluate the possibility that crosslinked IgE-FcεRI complexes might influence the behavior of non-crosslinked FcεRI, we primed cells with IgE of two different specificities: the DNP-specific IgE (IgEanti-DNP) and a second monoclonal IgE that recognizes dansyl (IgEanti-dansyl). Specificity of each antibody was confirmed by stimulating with their specific antigens. As expected, cells loaded with IgEanti-dansyl and stimulated with dansyl-BSA produced a comparable secretory response to cells loaded with IgEanti-DNP and stimulated with DNP25-BSA (Figure S6). The response was antigen specific over a range of antigen doses (Figure S6).
We next examined whether the two different IgE specificities co-clustered upon stimulation with one or both antigens. We primed cells with a 1:1 mixture of Alexa488-IgEanti-DNP and Alexa633-IgEanti-dansyl. These cells were then exposed to 0.1 μg/ml DNP25-BSA, dansyl-BSA, or both (0.05 μg/ml each) for 1 min, then fixed and imaged by confocal microscopy. Because of the resolution limits of the light microscope, only macroscopic receptor clusters are visible in this assay. As previously suggested (McConnell et al., 1986), IgE-bound receptors only formed large clusters in the presence of their cognate antigens and these clusters had very little overlap when both antigens were applied simultaneously (Figure 7A). This observation suggests that the macroscopic clusters are solely composed of IgE-FcεRI engaged by specific multivalent antigen.
SPT experiments reinforced this conclusion by demonstrating that only crosslinked receptors have altered mobility. Cells were labeled with low amounts of QD655-IgEanti-DNP and then the remaining receptor sites were saturated with either IgEanti-DNP or IgEanti-dansyl. Crosslinking the IgEanti-dansyl primed receptors with 1 μg/ml dansyl-BSA did not significantly affect the diffusion of QD655-IgEanti-DNP primed receptors (Figure 7B, Table S1). As expected, crosslinking the IgEanti-DNP primed receptors with 1 μg/ml DNP-BSA produced the dramatic decrease in the mobility of QD-IgEanti-DNP primed receptors (Figure 7B, Table S1). We also modified this protocol in order to visualize the relationship between antigen-induced aggregates and non-crosslinked FcεRI (Video 3). Cells were labeled with SPT levels of QD655-IgEanti-DNP and then the remaining receptor sites were saturated with Alexa488-IgEanti-dansyl. After labeling the receptors with IgE, we stimulated cells with 1 μg/ml dansyl-BSA and simultaneously imaged clustering of Alexa488-IgEanti-dansyl and diffusion of QD655-IgEanti-DNP. No prolonged interactions of QD655-IgEanti-DNP with the clusters of Alexa488-IgEanti-dansyl were observed. Taken together, these results demonstrate that activation of a subset of receptors does not trap bystander receptors in immobile signaling clusters or result in a global cellular response that affects their mobility.
Previous studies led to speculation that FcεRI immobilization might initiate downstream signaling (Mao et al., 1991; Menon et al., 1986b; Myers et al., 1992; Pecht et al., 1991; Pyenta et al., 2003; Tamir et al., 1996; Zidovetzki et al., 1986). However, technical limitations prevented a detailed examination of this hypothesis, particularly for the study of small receptor aggregates. The signaling competence of small aggregates has been demonstrated in studies using bivalent ligands and anti-receptor antibodies, which have shown that aggregates of at least two FcεRI are sufficient to induce signaling with varying degrees of efficiency (Ortega et al., 1988; Paar et al., 2002). Here, we employ monovalent QD-based probes and multi-color single particle tracking to directly observe the movements of aggregated FcεRI and explore the relationships between antigen valency, receptor aggregate size, mobility, and the initiation and propagation of downstream signaling. By comparing results in both the RBL-2H3 mucosal mast cell line and in murine BMMCs, we demonstrate that receptor behavior is not exclusive to any single model system. This work demonstrates that low valency antigens (average of 2 or 4 haptens per BSA carrier) induce signaling in the absence of receptor immobilization. Even high-valency antigens fail to immobilize receptors when applied at low doses. The use of hyperspectral microscopy allowed us, for the first time, to discriminate between multiple receptors within the same aggregate. We directly observe diffusion of aggregates containing at least three FcεRI. Furthermore, we showed that IgE receptors aggregated directly by multivalent DNP-QD probes exhibited fast mobility. DNP-QD receptor complexes are signaling competent, since they induce degranulation. Therefore, receptor immobilization is not required for signal initiation.
We speculate that receptor orientation, as well as the distance between receptors in an aggregate, are also features that markedly influence signal initiation. This may explain the relatively poor signaling capability of DNP2-BSA, which likely presents a limited number of properly oriented pairs of haptens. We note that, unlike the defined dimerizing reagents described recently by Baird and colleagues (Paar et al., 2002), the reagents used here do not permit evaluation of distances between haptens.
We also evaluated the kinetics of FcεRI β and γ subunit tyrosine phosphorylation over the range of DNP25-BSA doses. Distinct roles for β and γ ITAMs in FcεRI signaling have been established (Donnadieu et al., 2000; Jouvin et al., 1994; Lin et al., 1996; Wilson et al., 1995), where the FcεRI γ-ITAM phosphorylation is essential for Syk recruitment and downstream allergic responses (Sakurai et al., 2004; Takai et al., 1994). Therefore, it is surprising to note that Syk phosphorylation does not correlate well with γ subunit phosphorylation (Figure 4). Similar weak patterns of FcεRI γ phosphorylation have also been reported by others (Lin et al., 1996; Xu et al., 1998). There are several potential explanations for the disparity between Syk and γ phosphorylation. First, at low doses, Syk may be recruited to a small number of phosphorylated receptors that are below the limits of detection by western blotting. In this case, Syk recruitment to receptor tails must be highly transient, permitting many Syk molecules to be activated in succession. This would be a classic example of signal amplification. Second, given recent evidence for conformational changes in the BCR and TCR systems (Aivazian and Stern, 2000; Gil et al., 2002; Tolar et al., 2008), it is possible that minimal receptor crosslinking exposes the γ ITAM and permits Syk recruitment. Under this scenario, γ subunit phosphorylation would strengthen, but not be completely essential for, Syk recruitment. A third possibility is that γ ITAM phosphorylation is difficult to detect due rapid dephosphorylaton; this hypothesis is consistent with the observation that β subunit is robustly phosphorylated and the most likely candidate for anchoring SH2-containing phosphatases (Wilson 1995).
Importantly for the present study, the tyrosine kinase inhibitor PP2 eliminated antigen-induced tyrosine phosphorylation and secretion while having no detectable effect on receptor mobility. Thus, receptor immobilization is not dependent on the Lyn-mediated phosphorylation of receptors and more likely reflects the extent of physical crosslinking by polyvalent antigen. This conclusion is supported by “snapshot” images of IgE receptor cluster size as measured by electron microscopy. In resting cells, most receptors are imaged as singlets and small clusters; statistical analyses confirmed that this non-random distribution is significant (Wilson et al., 2004). Since SPT data have shown that most resting receptors transition rapidly between states of transient confinement and highly diffusive behavior (Andrews et al., 2008), we suppose that fixation stabilizes highly dynamic clusters. As antigen dose increases, receptor clusters progressively enlarge until singlets are a small minority and large clusters are the majority. Since we observed a marked dose-dependent onset of immobilization (between the doses of 0.01 and 0.1 μg/ml of DNP25-BSA), we assume that there is a size threshold at which receptor aggregates become immobile.
Note that we do not claim here that immobilized receptors lack the capacity to signal. However, the duration of signaling from immobilized receptors may be short, since we report here that receptor immobility is markedly correlated with rates of internalization. We found that internalization was minimal at non-immobilizing antigen doses, extensive at immobilizing antigen doses, and insensitive to PP2 treatment. Based on these observations, we hypothesize that extensive crosslinking and subsequent immobilization is sufficient to trigger receptor endocytosis, although the mechanism of endocytosis remains to be defined. Consistent with our observations here, previous mutagenesis studies failed to identify an endocytic motif in any of the FcεRI subunit cytoplasmic tails (Mao et al., 1991). It is clear that at least a portion of crosslinked FcεRI enter cells via coated pits (Pfeiffer et al., 1985). However, endocytosis is not inhibited when clathrin is reduced using siRNA ((Fattakhova et al., 2006), our unpublished results) and receptors bound to antigen immobilized on solid substrates do not recruit clathrin-coated pits or AP-2 (Santini and Keen, 1996). Given these incongruous results, it is clear that this is an area that warrants future investigation.
The signaling competency of small mobile FcεRI aggregates has physiological significance. In man and relevant animal models, the specificity of FcεRI is defined by the repertoire of IgE idiotypes bound to mast cell and basophil surface receptors. It is estimated that less than 10% of receptors on any single cell are likely to be specific for the same allergen (Johansson et al., 2006) and only a fraction of these need to be crosslinked to yield a secretory response (Ortega et al., 1988). Here, we did not find any evidence that antigen-bound receptors engage nearby, non-crosslinked receptors to amplify the signal (Figure 7). This differs from some models of the TCR, where CD4 mediates formation of a “pseudodimer” between two TCRs, enabling T-cells to respond to a single peptide-MHC complex (Irvine et al., 2002). We also conclude that few phosphoryated FcεRI are required to recruit and amplify Syk.
In summary, our QD-based probes and hyperspectral imaging instrumentation allowed us to demonstrate that small antigen-induced oligomers of at least three IgE-FcεRI remain mobile on the cell surface and diffuse together as stable complexes for minutes. Immobilization occurs abruptly at moderate doses of highly multivalent antigen, suggesting a size threshold is reached that strongly limits diffusion of receptor aggregates. These results support the conclusion that signal initiation and receptor immobilization are distinct processes. Internalization is strongly correlated with FcεRI immobilization, through a mechanism that may or may not involve clathrin-coated pits. In the context of the allergic response, where levels of specific IgE and/or antigen may be low, the observation that small, mobile FcεRI aggregates are signaling competent fits well with the estimate that few receptors need to be engaged for maximal degranulation. The observations made here using soluble antigens are also likely relevant to events that occur in the context of the immune synapse, where the contributions of individual immunoreceptors are difficult to determine. Our conclusions regarding the signaling competency of mobile receptors are consistent with observations that newly arrived antigen-induced microclusters of TCR signal actively in the pSMAC (Campi et al., 2005). These microclusters make their way to the cSMAC (DeMond et al., 2008; Kaizuka et al., 2007), where they coalesce, become immobile, and cease signaling prior to being internalized (Campi et al., 2005). Thus, the signaling competency of diffusing, antigen-engaged receptors is likely to be a unifying feature among MIRR family members.
Mouse monoclonal IgEanti-DNP was prepared as described in (Liu et al., 1980). Mouse monoclonal IgEanti-dansyl was from BD Biosciences (San Jose, CA). DNPn-BSA was from Invitrogen (Carlsbad, CA) or Biosearch Technologies (Novato, CA). For unit conversion, 5 nM IgE and 14 nM DNP-BSA are ~1 μg/ml. PP2 was from Calbiochem (La Jolla, CA), anti FcεRIβ antibody was a gift of J. Rivera (NIH, Bethesda, MD) and anti-phosphotyrosine antibodies (PY20/PY99) were from Santa Cruz Biotechnology (Santa Cruz, CA). HRP-conjugated secondary antibodies and SuperSignal® West Pico Chemiluminescent Substrate kits were from Pierce Protein Research Products (Rockford, IL).
Detailed methods are described in (Andrews et al., 2008).
A 1 mM biotin-DNP stock solution was prepared by dissolving 1 mg DNP-X-biocytin-X succinimidyl ester (Molecular Probes) in 100 μl dimethylsulfoxide (Sigma) followed by reaction with ammonium bicarbonate (Sigma) in a 1:10 stoichiometric ratio in water for 30 minutes to quench the succinimidyl ester group. DNP-QD was prepared by reacting biotin-DNP with the indicated Qdot® Streptavidin conjugate (Invitrogen) in a 50:1 ratio in PBS + 1% BSA for four hours at 4°C; the manufacturer estimates each QD is conjugated to 3–5 strepavidins. DNP-QD was then purified by dialysis against PBS and the final concentration determined by absorption at 532 nm. Stock solutions of QD-IgE and DNP-QDs were stored at 4°C and used within four weeks.
Fluorescent IgE was prepared using Alexa Fluor® 488 or Alexa Fluor® 633 Microscale Protein Labeling Kits (Invitrogen, Carlsbad, CA).
This reagent was prepared by modification of the protocol provided with Dansyl-X, SE (Invitrogen, Carlsbad, CA) to yield a dansyl:BSA ratio of 69.3 as measured by absorbance at 335 and 280 nm.
Rat basophilic leukemia (RBL-2H3) cells were grown as adherent monolayers in minimum essential medium with 10% fetal bovine serum (MEM/FBS) (Invitrogen, Carlsbad, CA) as described in (Wilson et al., 2000). BMMC were differentiated from mast cell progenitors in murine bone marrow by culturing in IL-3-containing medium for 4–6 week as in (Kashiwakura et al., 2008).
RBL cell monolayers were grown in 24-well tissue culture plates for 24 hours and primed with 5 nM IgEanti-DNP or IgEanti-dansyl; BMMC were primed in suspension overnight, then harvested, washed and allowed to settle onto 8-well chambers for 2 hrs in phenol red-free RPMI without IL-3. Release of β-hexosaminidase was measured as described in (Ortega et al., 1988). Immunoprecipitations and immunoblotting protocols were performed as in (Wilson et al., 1995) and (Hernandez-Hansen et al., 2004). Measurements of fluorescently-labeled IgE internalization were performed by flow cytometry, as previously described (Barker et al., 1995); where indicated in legend, cells were pretreated with 10 μM PP2 and maintained in the inhibitor throughout the assay.
SPT was performed using an Olympus IX71 inverted microscope equipped with a 60× 1.3 N.A. water objective and an electron multiplying CCD camera (Andor iXon 887) as in (Andrews et al., 2008). Samples were maintained at 34–36 °C by an objective heater (Bioscience Tools, San Diego, CA). For tracking QD-IgE-FcεRI complexes, RBL-2H3 cells were labeled with 500 pM QD655-IgEanti-DNP in HBSS for 10 min at 37°C prior to imaging. Where indicated, cells were incubated with 50 nM IgEanti-DNP or IgEanti-dansyl for 30 min at 37 °C in MEM/FBS, washed, and stimulated with the indicated doses of DNPn-BSA +/− 10 μM PP2 or dansyl-BSA while imaging at 20–33 frames/s at 35 °C. For tracking DNP-QDs, cells were primed by 30 min incubation with 50 nM “dark” IgEanti-DNP at 37 °C, followed by addition of DNP-QD655 or DNP-QD585 at the microscope, as specified in legends. The resulting image series were analyzed and diffusion coefficients obtained using previously described single particle tracking algorithms (Andrews et al., 2008).
RBL-2H3 cells were labeled with a 1:1 mixture of 500 pM QD625- and QD705-IgEanti-DNP for 10 min at 37 °C in HBSS, washed, and incubated at 37 °C for 30 min with 140 nM IgEanti-DNP. Cells were washed and imaged at 20 frames/s at 35 °C in 200 μl HBSS; after ~10s, 100 μl of 3X DNPn-BSA was added at indicated concentrations. Instantaneous diffusion coefficients were calculated as described (Andrews et al., 2008) and the traces from multiple cells (resting, n=51; 10 μg/ml, n=11; 1 μg/ml, n=9; 0.1 μg/ml, n=10; 0.01 μg/ml, n=10; 0.001 μg/ml, n=16) averaged to generate the final plot.
Cells were labeled with 5 nM IgEanti-DNP in suspension overnight, washed and incubated with 1 nM QD655-IgE anti-DNP in 1 ml phenol-red free RPMI for 30 min at 37°C. Cells were resuspended in Hanks-BSA and permitted to settle onto 8-well chambers prior to imaging.
Image processing was performed using Matlab (The MathWorks, Inc., Natick, MA) in conjunction with the image processing library DIPImage (Delft University of Technology). Descriptions of specific analysis routines has been reported previously (Andrews et al., 2008). D1–3, values are reported as medians and interquartile range is provided as a measure of statistical dispersion (Table S1).
Cells on 15 mm round coverslips were incubated for 15 min at RT with a mixture of QD655-, QD625-, QD585-, QD565-, and QD525-IgE at 400 pM each in HBSS, washed, and coverslips mounted onto 25×75 mm glass slides using a 2 mm rubber spacer. Samples were treated +/− 0.1 μg/ml DNP25-BSA and imaged by hyperspectral microscopy as described in (Sinclair et al., 2006). Confocal images were acquired at 0.25 frames/s for 60 frames. The resulting image series were subtracted for dark current and despiked (Jones et al., 2008), then displayed using a custom image analysis program (Haaland et al., 2009). Regions of interest could then be manually selected and the spectra from these regions obtained.
Cells on glass coverslips were primed overnight with 5 nM IgEanti-DNP. Washed cells were stimulated with indicated doses of DNP-BSA for 1 min, followed by fixation (7 min, 0.5% paraformaldehyde) and membrane sheets prepared as previously described (Wilson et al., 2000). Digital images were acquired using a Hitachi H600 transmission electron microscope, followed by image processing and statistical analyses as previously described (Zhang et al., 2006). For plotting cluster size, we employed a cut-off distance of 50 pixels (43 nm), such that receptors within this distance of each other were considered part of the same cluster.
This work was supported by NIH grants R01 AI051575 and GM49814 and the Sandia SURP program. Sandia is a multi-program laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the US Department of Energy under contract DE-AC04-94AL85000. A portion of this work was supported by the Microscale Immune Studies Laboratory Grand Challenge LDRD at Sandia National Laboratories. NLA was supported by NSF IGERT DGE-0549500 and the UNM-SOM MD/PhD Program. We thank Sheli Ryan and Yuko Kawakama for cell culture assistance and Mary Raymond-Stintz, Stanly Steinberg, and Michael Wester for assistance with the quantification of EM data. Support for the UNM Cancer Center Fluorescence Microscopy and Flow Cytometry Facilities was from NIH grants S10 RR14668, S10 RR19287, S10 RR016918, P20 RR11830 and P30 CA118100 and NSF grant MCB9982161; the UNM EM Facility received support from NIH GM067594, S10 RRI5734 and RR022493.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.