Autoantibodies to IgE and FcεRI and the natural variability of spleen tyrosine kinase expression in basophils

Autoantibodies to IgE and FcεRI and the natural variability of spleen tyrosine kinase expression in basophils Donald MacGlashan, MD, PhD

Baltimore, Md

Background: Secretion from human basophils and mast cells requires spleen tyrosine kinase (SYK) activity, but SYK expression is highly variable in the general population, and this variability predicts the magnitude of IgE-mediated secretion. One known mechanism of modulating SYK expression in human basophils is aggregation of FcεRI. Objective: This study examines the possibility that functional autoantibodies are present in a wide variety of subjects and, in particular, subjects whose basophils poorly express SYK. It also tests whether any found antibodies could modulate SYK expression in maturing basophils and whether interaction with FcgRIIb/CD32b modulates the effect. Methods: An experimental algorithm for classifying the nature of histamine release induced by serum from 3 classes of subjects was developed. Results: The frequency of functional autoantibodies that produce characteristics concordant with FcεRI-mediated secretion was zero in 34 subjects without chronic spontaneous urticaria (CSU). In patients with CSU, the frequency was lower than expected, approximately 7%. For the 5 of 68 unique sera from patients with CSU tested that contained anti-FcεRI or anti-IgE antibodies, these antibodies were found to induce downregulation of SYK in both peripheral blood basophils and basophils developed from CD341 progenitors. Blocking interaction of these antibodies with CD32b did not alter their ability to downregulate SYK expression. Conclusions: This study establishes that functional autoantibodies to IgE/FcεRI do not provide a good explanation for the variability in SYK expression in basophils in the general population. They do show that if antibodies with these characteristics are present, they are capable of modulating SYK expression in developing basophils. (J Allergy Clin Immunol 2018;nnn:nnn-nnn.) Key words: Human, basophil, allergy, Fc receptors

IgE-mediated secretion from basophils and mast cells is a hallmark of the immediate hypersensitivity reaction. This receptor’s activation cascade is initiated by aggregation of cell-surface FcεRI and the aggregation-induced assembly and From the Asthma and Allergy Center, Johns Hopkins University. Supported by National Institutes of Health grants AI100952 and AI116658. Disclosure of potential conflict of interest: D. MacGlashan discloses short-term consultancy for Sixal and Boehringer Ingelheim. Received for publication January 15, 2018; revised April 11, 2018; accepted for publication May 10, 2018. Corresponding author: Donald MacGlashan, MD, PhD, Department of Medicine, Johns Hopkins University Asthma and Allergy Center, 5501 Hopkins Bayview Circle, Unit Office 3, Baltimore, MD 21224. E-mail: [email protected]. 0091-6749/$36.00 Ó 2018 American Academy of Allergy, Asthma & Immunology https://doi.org/10.1016/j.jaci.2018.05.019

Abbreviations used BTK: Bruton tyrosine kinase CIU: Chronic idiopathic urticaria CSU: Chronic spontaneous urticaria HSA: Human serum albumin IC50: Inhibitory concentration of 50% NP: Nitrophenyl nsIgG: Nonspecific IgG PI3K: Phosphoinositide 3-kinase PIPES: Piperazine-N,N9-bis(2-ethanesulfonic acid) SYK: Spleen tyrosine kinase

activation of nonreceptor kinases, such as spleen tyrosine kinase (SYK).1 We have been interested in the observation that expression of SYK in basophils is markedly reduced compared with that of other leukocytes and is highly variable in the general population.2 This observation has functional consequences. First, it provides a reasonable explanation for the high variability in maximum IgE-mediated histamine release from basophils (in vitro).2-4 Second, it appears that starting levels of SYK expression in basophils are predictive of the efficacy of omalizumab treatment in patients with asthma or peanut allergy.5,6 The implication of the latter observation is that basophils play a significant role in expression of food allergy or asthma. For basophils, it is not yet clear whether variability in SYK expression or average low levels of SYK levels result from transcriptional or posttranslational mechanisms. However, one of the few known mechanisms for regulating SYK expression involves aggregation of FcεRI.2,7 It is also known that the aggregation reaction can determine downregulation of SYK, even if the activation cascade is suppressed by coengagement of CD32b on the basophil surface.8 This possibility suggests that posttranslational SYK regulation can occur in vivo without necessarily invoking mediator secretion. Production of autoantibodies as a driver of pathologic states is a common finding in the field of immunologic conditions. This is true for allergic diseases as well. In fact, for 20 years, one dominant explanation for the existence of chronic spontaneous urticaria (CSU; or chronic idiopathic urticaria [CIU]) has been the inappropriate production of antibodies to the high-affinity IgE receptor FcεRI or IgE antibody.9-12 These antibodies could induce secretion from basophils or mast cells by inducing an aggregation reaction involving either IgE (bound to FcεRI) or FcεRI directly. However, there is evidence that these autoantibodies are also found at a high frequency in subjects without CIU/CSU.13 Implicit to the thinking about autoantibodies in patients with CSU is that they are functionally active; that is, they can induce aggregation of FcεRI and activation events that can lead to 1

2 MACGLASHAN

secretion. Postulating that the variability of SYK expression results from autoantibodies would also require that they be able to induce aggregation and processing of SYK. Thus in this study functional antibodies will be the focus, in particular, of autoantibodies to either IgE or FcεRIa. Taken together, the observations that (1) there can be autoantibodies to IgE or FcεRI, (2) that SYK is downregulated by aggregation, and (3) that coengagement of CD32b can allow an aggregation reaction without secretion suggest a hypothesis for the presence of variable SYK expression in basophils: naturally occurring IgE- or FcεRI-specific antibodies interact with basophils to induce downregulation of SYK and that secretion is ablated by simultaneously interacting with surface CD32b. A further qualification for the action of these antibodies is that the reaction occurs during maturation of the basophils in the bone marrow. In other words, the cells might emerge from the marrow having experienced prior aggregation and downregulation of SYK. Whether these antibodies are capable of this interaction requires direct testing of their actions in the presence and absence of CD32b blockade. One prediction from this hypothesis would be that patients without CSU would also express autoantibodies because suppressed SYK expression is found in basophils of both nonatopic and atopic subjects without CSU. Therefore one goal for this study was to determine the frequency of anti-IgE or anti-FcεRI antibodies in healthy subjects and, in particular, subjects whose basophils poorly express SYK. However, as we began the search for natural antibodies to test, it became clear that the ability of sera to directly activate basophils was not always related to aggregation of IgE or FcεRI and that the frequency of sera with properties consistent with activation of FcεRI was much lower than published accounts suggest. This study established a different set of criteria for assessing the presence of functional autoantibodies in sera that can induce secretion from human basophils and then explored whether they can act to modify SYK expression in maturing human basophils.

METHODS General methods The Methods section in this article’s Online Repository at www.jacionline. org includes descriptions of methods in common use or methods previously published.

Sera There were 3 categories of subjects providing serum for this study. Serum from subjects without CSU was obtained generally from subjects whose basophils were well characterized for their releasing properties, including IgE-mediated histamine release. These characteristics will be noted. Serum from patients with CSU was divided into 2 groups, and all sera were provided by the laboratory of Dr Sarbjit Saini. These subjects were clinically well characterized with respect to the expression of CSU. The sera were not always obtained during active disease states. The study began with a general selection of patients with CSU without regard to the presence of basopenia at the time of blood draw (ie, basopenia might or might not have been present). Subsequent to obtaining this group of sera, a second request for sera was made for subjects for whom basopenia was known to be present. Retrospectively, 3 of the sera transmitted to us for the second group were from subjects in the first CSU group.

J ALLERGY CLIN IMMUNOL nnn 2018

Definition of positivity For a serum to have induced histamine release, a threshold was chosen that was based on the general variability of results from a large number of screens. The algorithm used to set a threshold of 6% (above spontaneous release) is discussed in the Methods section in this article’s Online Repository. Because of the technical issues related to the need to precipitate protein before analyzing samples on the autoanalyzer, the maximum concentration of serum tested was 25%. A concentration of 5% was included to get a sense of the serum titer in these screening tests.

Drug sensitivity and similarity index calculation Partially enriched basophils were stimulated in the presence of 25% or 5% serum after a 5-minute incubation with the 4 agents tested: PP1 (PP2 was also tested in pilot experiments but had the same problems with serum found for PP1), NVP-QAB205, PCI-32765, and LY294002 (an Src family kinase inhibitor, a SYK inhibitor, a Bruton tyrosine kinase [BTK] inhibitor, and a phosphoinositide 3-kinase [PI3K] inhibitor, all IgE-mediated release-specific inhibitors). For the final test, the concentrations of the latter 3 drugs were 0.75 mmol/L, 75 nmol/L, and 7.5 mmol/L, respectively. Positive controls included 1 mg/mL of the anti-IgE antibody 6061P in PAGCM or in 25% negative serum (extensive testing of several negative sera established a single serum that was not found to induce release under a variety of circumstances). To calculate a similarity index when determining whether a particular serum behaved toward the drugs tested in a manner similar to 6061P, the ratio of the response of (6061P 1 Drug)/(6061P 2 Drug) was calculated and defined as ADR. We calculated the ratio of (Serum 1 Drug)/(Serum 2 Drug) and defined it as SDR. The similarity index was as follows: (1 2 SDR)/ (1 2 ADR). This value is 1.0 when the drug inhibits serum as effectively as it inhibits 6061P. Indices of greater than 0.9 were considered ‘‘similar’’ to the behavior with 6061P to account for the noise in the assay. A similar metric was calculated for the cross-desensitization experiments.

RESULTS Autoantibodies to FcεRI or IgE The frequency at which one finds sera that induce release from basophils ex vivo appears variable among studies.10-13 On the nature of positivity, our approach was cautious. For the assay itself, the goal was to choose donors whose basophils were sensitive to stimulation through IgE or FcεRI. The Results section in this article’s Online Repository at www.jacionline.org also discusses results on whether ABO incompatibility was a technical issue. Although it was concluded that AB/Rh compatibility was probably not a factor in the ability of serum to induce secretion, for the current purposes, recipient basophils were all O type. The criterion for declaring a serum as causing release was developed from analysis of noise in the assay (see the Methods section). In addition, a positive serum was required to repeat with a different basophil donor. There were 29 single-positive results from 102 distinct sera, resulting in a frequency of 28% (41% in the CSU groups), only 13 of which repeated positive with distinct basophil donors, for a frequency of 13% (or approximately one half of the single-positive results). Because we then engaged in additional testing with each positive serum, replication for true-positive results was a result usually based on multiple experiments. Single-positive results resulted in no further testing. Fig 1 and Fig E1 in this article’s Online Repository at www.jacionline.org summarize the results and extend the summary to 3 categories of serum donor: those without CSU, those with CSU but chosen without regard to the presence of basopenia, and those with CSU selected a priori for coincident

MACGLASHAN 3

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

FIG 1. Criteria for classification of sera as containing functional autoantibodies to FcεRI or IgE. drugs, Three IgE pathway–specific drugs; NR, inability to stimulate basophils of the nonreleasing phenotype; xdesn, cross-desensitization with known aggregating anti-IgE antibody.

basopenia. There were no consistent positive results in the non-CSU group, but consistent positive results were observed in those with CSU. Notably, there was no evidence for histamine-releasing activity in the healthy group without CSU (n 5 34); this group included both allergic (25%) and nonallergic (75%) subjects. For these serum donors, there was information about the release induced by an optimal concentration of anti-IgE antibody/6061P from their basophils as well; this was as variable as previous studies have found (average, 40% 6 28%; coefficient of variation, 0.70), and 6 (17%) of the donors fit the nonreleaser category (2.0% 6 0.4% release vs 48% 6 5% for releasers). Thus the donors who had nonreleasing basophils did not also have sera that induced release (see below as well). The Results section in this article’s Online Repository presents results with IL-3–treated recipient basophils. These experiments show that sera from subjects without CSU can induce release that is not IgE mediated (see Fig E6 in this article’s Online Repository at www.jacionline.org). The chosen positive sera were examined with 4 different assays to determine whether the induced release operated through an

FcεRI-like mechanism. Three of these assays were used on each of the multiply positive sera and the final assay used for the sera that passed the first 3. These results are provided in Figs E2-E5 in this article’s Online Repository at www.jacionline.org. The 4 tests included (1) the ability of 3 known IgE-signaling pathway drugs (SYK, BTK, and PI3K inhibitors) to inhibit serum-induced release, (2) the ability to cross-desensitize serum-induced release with a known anti-IgE antibody, (3) the inability of sera to induce release in basophils with the nonreleasing phenotype, and (4) the ability of IgE- or FcεRIa-Sepharose (and not human serum albumin [HSA]–Sepharose) to adsorb the autoantibody activity. Only 5 sera passed these tests.

Autoantibody sensitivity to CD32b engagement One goal for these experiments was to determine whether human anti-FcεRI or anti-IgE antibodies interacted significantly with CD32b. However, this necessitated first understanding how nonspecific IgG (nsIgG) could interfere with the extrinsic specific antibodies interacting with FcgRIIb/CD32b (or FcgRIIa/CD32a) because serum nsIgG is only avoidable if it can be diluted

4 MACGLASHAN

sufficiently to stop its interference with CD32b binding.14 The Results section in this article’s Online Repository includes experiments addressing this issue (see Fig E7 in this article’s Online Repository at www.jacionline.org). It was determined that if sera were used as stimuli, testing would require sera concentrations to be less than 2%. Of the 5 sera with the expected anti-IgE or anti-FcεRIa characteristics, only 2 were of sufficient titer to allow study with serum concentrations of 2% or less. This was important because it reduces the nsIgG concentrations to less than 150 mg/mL. As noted in the preliminary studies in the Results section in this article’s Online Repository, this concentration might not significantly interfere with FcgRIIb/ CD32b binding, whereas inclusion of Ab10523 (a CD32b selective high-affinity antibody8) would interfere. For these experiments, there were 2 positive controls that were used to insure that a FcgRIIb/CD32b-mediated inhibition of histamine release was observable under the conditions of the experiment. The cells were treated to dissociate a portion of the endogenous IgE and resensitized with a nitrophenyl (NP)–specific IgE. As we have shown previously,14 the mixing of NPspecific IgG2 at an appropriate concentration with NP-BSA before stimulation inhibits histamine release caused by NP-BSA and that this inhibition is partially or fully reversible by blocking FcgRIIb/CD32b with Ab10523 (anti-FcgRIIb/CD32b). A second positive control was that noted in the Results section in this article’s Online Repository, testing for release with a supraoptimal concentration of goat polyclonal anti-IgE antibody with or without Ab10523.8 As above, the inclusion of Ab10523 (antiFcgRIIb/CD32b) should enhance the response to goat anti-IgE antibody. As shown in Fig 2, both of these controls demonstrated the expected behavior for Ab10523 (anti-FcgRIIb/CD32b), reversing NP-specific IgG2 inhibition (with paired analysis, reversal was 62% 6 14%; P 5 .0016) and markedly enhancing the goat polyclonal anti-IgE antibody response (P 5 .0009). It should be noted that these positive control conditions were performed in the presence of a 1% negative serum (to replicate the conditions for the 2 test sera). MM (an anti-IgE–like serum) and AB (an anti-FcεRIa–like serum) were tested at 1% and 0.3% concentrations. There was no effect or possibly some inhibition of release with inclusion of Ab10523 (anti-FcgRIIb/CD32b; Fig 2).

Influence of IgE/Fc-like serum on SYK levels in peripheral blood basophils Stimulation of human basophils through the IgE-mediated pathways induces loss of SYK expression.7 The process is relatively slow, but even low levels of stimulation induce SYK loss when they do not induce histamine release. Therefore this can prove a useful test for low levels of stimulation. The assay was performed with impure cells by using flow cytometry to detect SYK expression or with purified cells by using Western blotting to detect SYK expression. For context, the cells were stimulated with anti-IgE antibody/6061P diluted into a putatively negative serum so that the final serum concentration (2%) was similar to that of the negative control and the other tested sera. Five sera from releasers and 5 sera from nonreleasers were tested. By using either flow cytometry or Western blotting, there was no significant loss of SYK expression (relative to a control containing no serum) for all 10 sera, whereas there was a 6061P-induced loss of 61% 6 5% of SYK. Notably, there was

J ALLERGY CLIN IMMUNOL nnn 2018

FIG 2. Effect of Ab10523 (blocking antibody for FcgRIIb/CD32b) on sera-induced secretion. Two methods were used to demonstrate within the same cell preparations that blocking FcgRIIb/CD32b with Ab10523 (anti-FcgRIIb/CD32b) under known conditions would alter the response to stimuli. These positive controls (left side of plot) are described in the text. By using similar conditions, cells were stimulated with sera MM and AB at concentrations shown in the presence or absence of Ab10523 (anti-FcgRIIb/CD32b). Bars plot the average histamine release for the 4 experiments, and average ratios (response in the presence of Ab10523/ response without antibody) for the 4 experiments are shown above the bars.

no difference between releaser and nonreleaser sera (Fig 3, A and B). In contrast, 3 of the CSU-derived sera that were shown to contain anti-FcεRI– or anti-IgE–like properties induced significant loss of SYK in these cultures. Combining the results in which flow cytometry or Western blotting was used, these 3 sera induced a 56% 6 9% loss of SYK (6061 induced a 85% 6 5% loss). Two of the sera from patients with CSU had a high enough titer that a 1.2% dilution could be used; therefore nsIgG should be approximately 180 mg/mL, low enough that it should not interfere with binding of an anti-IgE– or anti-FcεRI–like antibody (see above). For these 2 sera, the overnight incubation was done with or without Ab10523 (anti-FcgRIIb/CD32b) to block interaction with FcgRIIb/CD32b. Based on previous studies8 showing that SYK loss was not affected by signaling changes downstream of SYK, loss of SYK was not expected to be affected in this experimental design, and for the sera from patients with CSU (average reduction, 0.29 6 0.02 of control) or anti-IgE antibody/6061P or goat anti-IgE diluted into negative serum (average reduction, 0.30 6 0.08 of control), Ab10523 (anti-FcgRIIb/CD32b) did not change the loss of SYK (for sera from patients with CSU, average reduction to 0.30 6 0.05 of control; for anti-IgE antibodies, reduction to 0.30 6 0.02 of control).

Influence of IgE/Fc-like serum on SYK levels in maturing basophils Although there was no evidence that sera from subjects without CSU could induce secretion or alter SYK expression in overnight

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

FIG 3. SYK expression in purified human basophils cultured overnight with sera (see the Methods section) from subjects whose basophils were characterized as nonreleasers or releasers. A, Example Western blot for SYK expression. aIgE, known aggregator 6061P/anti-IgE antibody; NR, serum from nonreleaser phenotype; R, serum from releaser phenotype. The protein p85 (a subunit of PI3K) is used to normalize results (see the Methods section). B, Summary analysis of 2 experiments. Serum effects were calculated relative to media control band intensity. Five releaser and 5 nonreleaser sera were tested in overnight cultures, and ratios were plotted. C, Example Western blot analysis for SYK expression. Sera MM, AB, and AS are known to induce release by the criteria established for FcεRI/IgE-mediated release. Overnight cultures are as described for Fig 3, A. A, Anti-IgE antibody; As, suboptimal anti-IgE antibody; C, media control; Cs, control serum (known not to induce secretion). D, Average results of the conditions aligned in Fig 3, C (n 5 3; 2 experiments using Western analysis and 1 using flow cytometric analysis).

cultures of PBB, the original purpose of this study was to determine whether a serum with demonstrable autoantibodies to FcεRI could modify SYK expression in maturing basophils. Previous studies on the maturation of basophils from CD341 progenitors showed that chronic aggregation could induce a loss of SYK expression without changes in cell-surface FcεRI density or granulation (as measured based on Alcian blue positivity or histamine content).15 A similar experimental design was used to examine one of the better sera from the above studies and specifically a serum that appeared to be an anti-FcεRIa antibody. In addition, Ab10523 (anti-FcgRIIb/CD32b) was added in one set of conditions to test whether blockade of FcgRIIb/CD32b interaction would alter the results. To perform these experiments, the serum needed to have a high titer so that the nsIgG levels would be low. One of the 4 sera studied above, AS, was shown to induce SYK loss in PBB at low enough concentrations to test the effect of Ab10523. This serum was also shown to bind well to unoccupied FcεRI. The positive

MACGLASHAN 5

control in these studies was a concentration of Ab22E7 (antiFcεRI antibody) that was chosen to be suboptimal for induction of histamine release (to better test the low end of stimulation for an effect on SYK expression). For these experiments, conditions included media alone, Ab22E7 (anti-FcεRI), a ‘‘negative’’ serum with or without Ab10523 (anti-FcgRIIb/CD32b), and AS serum with or without Ab10523 (anti-FcgRIIb/CD32b). Both negative serum and AS serum were tested at 1% and 2% concentrations. Not shown in the figure, inclusion of serum (negative or AS) increased the number of Alcian blue–positive cells (1.65 6 0.065, P 5 .009) and total histamine content of the culture cell pellets, with a slight increase in viability in the cultures. Fig 4 shows the 2-dimensional plots of FcεRI versus SYK expression for negative control serum (Fig 4, A) and 1% AS serum (Fig 4, B) for one of 3 experiments, and Fig 4, C, summarizes results for the 3 experiments. Relative to negative serum, AS serum induced a loss of SYK, and the pattern at 1% was similar to the pattern observed in previous studies (and different from the pattern with negative control serum); greater loss occurred at higher densities of FcεRI (thus the negative slope in the dot plot).15 Greater loss occurred at 2% serum, and Ab10523 did not alter the results (with statistical significance). For context, the Ab22E7 (anti-FcεRI) results showed a greater decrease. Also, as observed previously, there was no effect on Alcian blue counts (between negative control serum and AS), total histamine content, or cell-surface FcεRI expression. The apparent increase with AS was not statistically significant (P 5 .095). As a follow-up to the FcgRIIb/CD32b portion of the experiment, expression of FcgRIIb/CD32b in these maturing basophil cultures was examined. Its expression was near zero in the original CD341 cells and increased with considerable variability in rate in different preparations but was nevertheless present by day 9 of culture (data not shown).

DISCUSSION These studies were undertaken to determine whether the sera of subjects whose basophils poorly expressed SYK would contain autoantibodies to FcεRI or IgE, which could explain the suppressed levels of SYK expression. In the same context the question could be expanded to ask whether heterogeneity of SYK expression in the general population could be explained by the presence of functional autoantibodies. This hypothesis was developed because it has been demonstrated that aggregation of FcεRI/IgE is one mechanism for reducing SYK in both mature peripheral blood basophils and in maturing culture-derived basophils.2,7 There is also evidence that suggests that the sera of healthy subjects can contain autoantibodies to FcεRI or IgE.9-13 This hypothesis was predicated on finding autoantibodies that were functionally active for inducing either secretion or, at a minimum, inducing loss of SYK. To provide a context for this search, subjects with clinically demonstrable CSU were used to act as a positive control because frequencies for finding these autoantibodies in subjects with this condition are reported to be as high as 50%. After initial screening, it became clear that a better definition of antibody activity that could induce secretion was needed. It also became clear that the bioassay, basophil activation by a foreign serum, was sensitive to factors not related to FcεRI or IgE cross-linking. By the end of the study, the criteria for a serum containing a functionally active anti-IgE or

6 MACGLASHAN

J ALLERGY CLIN IMMUNOL nnn 2018

FIG 4. SYK expression in 15-day culture–derived basophils treated with AS serum with or without Ab10523 (anti-FcgRIIb/Cd32b). Various stimuli were added at day 3 of culture, and analysis of SYK was done by using flow cytometry (contour lines show isotype controls). A, Flow cytometry of cells treated with control serum at 1% (serum previously determined to not induce release). B, Flow cytometry of cells treated with AS serum at 1% (known to induce release). C, Summary results from 2 to 3 experiments (n 5 3 for 1% serum and n 5 2 for 2% serum) expressed relative to the control serum 22E7 (anti-FcεRI antibody) at a concentration previously determined to induce 5% of maximum release. For the AS-treated cells, results were combined for both 1% and 2% concentrations. Al.Blue, Counts of Alcian blue–positive cells (n 5 3); FceRI, surface FcεRI expression (n 5 3); Tot.Hist., total histamine content (n 5 2).

anti-FcεRI antibody had expanded to 6 conditions. Including the 4 tests noted in the Results section, 2 additional tests included (1) using sera able to be depleted of their activity by specific adsorption with either IgE- or FcεRIa-bearing solid matrix and (2) inducing the loss of SYK expression in overnight basophil cultures. With these considerations, no serum from subjects without CSU was found to contain functional autoantibodies. Notably, this included sera from subjects whose basophils showed near-zero levels of IgE-mediated histamine release (and therefore little SYK expression). There were also no differences between sera from nonallergic and allergic subjects. This result suggests that the variability in SYK expression in the population without CSU (ie, 99% of the population) is not explained readily by the variable presence of functional autoantibodies to FcεRI or IgE. Previous studies on the presence of anti-FcεRI or anti-IgE antibodies in sera from subjects without CSU used various binding assays (eg, an immunoenzymometric-formatted assay and Western blotting) to detect these antibodies.9-13,16 Because the current hypothesis required cross-linking–capable antibodies, the distinction between the current and previous results might reflect the difference in the ability to cross-link IgE versus simple binding. One early study suggested that antibodies with this characteristic exist and are not functionally active.17 This study found that it was possible to induce histamine release from many subjects by simply stimulating the cells with an anti-IgG antibody. Because there are no known activating IgG receptors on basophils,14,18,19 it could be concluded that IgG is bound by other means, possibly to structures like IgE or FcεRI, especially because the characteristics of the release induced by anti-IgG antibody were similar to anti-IgE antibody–induced release.17 It would be surprising to find no examples of autoantibodies in sera from atopic subjects if this interpretation was correct unless these antibodies can bind well but not induce aggregation without the assistance of a secondary cross-linking reagent like anti-IgG antibody. For this previously published study, other than a strong association of these antibodies with allergic diseases, there was no identified consequence to possessing the antibodies. A more

recent study16 identified anti-IgE antibodies in sera by using an ELISA-based assay and determined that some of these antibodies were functionally active, but a consequence of this activity in the subjects from which they were derived was not discussed. In contrast, if autoantibodies are functionally active in patients with CIU/CSU, there would seem to be a pathologic consequence to their presence (although concordance between their presence and disease expression is poor10-12). These studies were limited to a single clinic population, and although the patients were well characterized, a broader study, potentially at other sites, should be used to establish whether this algorithm for identifying positive sera would yield a similar frequency of autoantibody occurrence at other geographic regions. It is also possible that the threshold for positivity used in this study (to make possible the SYK studies) excluded some weakly positive sera. The current study also suggests that the frequency of functionally active autoantibodies in the population with CSU was lower than anticipated. Instead of a frequency of approximately 50%, this study suggests a frequency of 7%. Although this is a far lower frequency than expected, the few sera found pass a stringent set of requirements and clearly possess functional autoantibodies to either IgE or FcεRI. With these sera in hand, it was possible to demonstrate that they can act to induce a reduction in SYK expression in both mature and developing basophils in CD341 cultures. As found previously, an interesting aspect of the reduction in CD34 cultures was the loss of SYK expression while retaining FcεRI expression and the granularity (Alcian blue positivity/histamine content) of the maturing cell.15 In addition, in neither peripheral blood basophils nor CD34-derived basophil cultures did blockade of CD32b modify the ability of the antibodies to induce loss of SYK. This was the behavior expected from previous studies of the signaling requirements for CD32b-mediated inhibition,8 but these results served to strengthen the hypothesis that an autoantibody could induce a change in SYK expression without interference from Fc region binding to CD32b. However, the prediction might be different for SYK downregulation versus histamine release,8

MACGLASHAN 7

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

and the observation that CD32b blockade brought about neither outcome suggests that interaction with CD32b is minimal, regardless of the conditions or outcome metrics of the experiments. Mast cell SYK is also regulated by IgE-mediated stimulation, but past experience with natural tissue mast cells has shown them to be either as sensitive or less sensitive (cell-surface density of IgE required for a response) than basophils. Because this was not examined, it is possible that some sera that were classified as negative for inducing an IgE-mediated response on basophils might behave differently with mast cells, but there is no expectation for this result. Anecdotally, for the 5 sera that showed demonstrable functional autoantibodies to IgE or FcεRIa, the basophil responses of these same donors were a mixture of releasers and nonreleasers. Four of the 5 subjects with serum-induced histamine release classified as having anti-IgE/FcεRI antibodies had nonreleasing basophils. The remaining 8 (8/13) subjects whose sera failed some of the tests had basophils that were evenly split between releasers and nonreleasers. Previous studies13,20 have suggested the frequency of nonreleasers in the population with CSU is 30% to 40%, and therefore the 80% frequency in the autoantibody group (n 5 5) appears high (x2 test against the study by Baker et al21 yields a P value of .054, and x2 test against the study by Rauber et al20 yields a P value of .04) but compatible with having autoantibodies to IgE/FcεRI. A prospective study would be needed to determine whether the titer of autoantibodies validated with this study’s algorithm would predict the releasing phenotype of the subject’s basophils. It was clear that there are technical issues when serum from one subject is used to stimulate leukocytes from another subject. Although there was some concern regarding ABO antibodies, preliminary studies suggested this was not a problem. However, it was also clear that there are unidentified incompatibilities between sera and non-self basophils (see the ABO compatibility section in the Methods section in this article’s Online Repository). This problem was least apparent when using sera from subjects without CSU unless the basophils were first cultured overnight with IL-3. Under these conditions, basophils became responsive to all sera, and histamine release induced by the sera was insensitive to inhibition with a BTK inhibitor. This phenomenon was not explored further but suggests that shifting basophil phenotypes (eg, generated by IL-3 exposure) can complicate studies using sera. The problem was more apparent when using sera from patients with CSU. The frequency of any response with sera from the 2 groups with CSU was approximately 41%. This is very similar to findings from other studies. However, after testing for repeatability and IgE/FcεRI-mediated characteristics, the frequency decreased significantly to 7%. There were a variety of failure points, from nonrepeating results to not being inhibited by signal transduction inhibitors that should alter IgE/FcεRI signaling. The most puzzling results were with 4 sera that passed 2 of the tests for IgE/FcεRI-mediated characteristics but failed the nonreleaser test. Because there was no further study of these unusual sera, there is no conclusion about the mechanisms that underlie the result. However, on a practical level, it appears that there are multiple ways that sera can activate basophils that overlap with IgE/FcεRI-mediated release but are not necessarily activated through IgE/FcεRI. Some of the sera that showed demonstrable autoantibodies were re-examined by using different blood draw dates from the

same patients. This was explored only briefly. In 3 instances in which additional serum samples were obtained 1 year apart or more from the tested serum, the positive results did not repeat. In one instance, a recent patient, serum samples spaced apart by only 2 months were very similar in characteristics. The sampling is too limited to relate these results to clinical status, but the variability has practical implications, as well as highlighting the ephemeral nature of these autoantibodies. In summary, one important conclusion of these studies is that there is no evidence for functional autoantibodies with specificity for IgE or FcεRI in subjects without CSU that could modify SYK expression levels in their circulating basophils. Because SYK expression is highly variable in all subjects, it appears that functional autoantibodies provide a poor explanation for this variability. A second conclusion is that functional autoantibodies can be found in sera of patients with CSU, although if functionality on a FcεRI-bearing cell type is the behavior of interest, the frequency appears to be less than 10%. However, these antibodies do have the capability of modifying SYK expression in a developing basophil and do not appear to interact with FcgRIIb/CD32b. Portions of this study were made possible by generous access to sera from patients with CSU collected by the laboratory of Dr Sarbjit Saini and Ms Kristin Chichester. I thank Valerie Alexander for her excellent technical assistance.

Key messages d

Autoantibodies to either IgE or FcεRIa can induce downregulation of SYK in either mature or maturing human basophils.

d

The frequency of functional autoantibodies in subjects without CSU is near zero, and that in subjects with CSU is only 7%. Therefore autoantibodies to IgE or FcεRIa do not explain the wide variation of SYK expression in the general population.

REFERENCES 1. Vilarino N, MacGlashan D Jr. Transient transfection of human peripheral blood basophils. J Immunol Methods 2005;296:11-8. 2. MacGlashan DW Jr. Relationship between Syk and SHIP expression and secretion from human basophils in the general population. J Allergy Clin Immunol 2007; 119:626-33. 3. Schroeder JT, Bieneman AP, Chichester KL, Keet CA, Hamilton RG, MacGlashan DW Jr, et al. Spontaneous basophil responses in food-allergic children are transferable by plasma and are IgE-dependent. J Allergy Clin Immunol 2013; 132:1428-31. 4. Puan KJ, Andiappan AK, Lee B, Kumar D, Lai TS, Yeo G, et al. Systematic characterization of basophil anergy. Allergy 2017;72:373-84. 5. MacGlashan DW Jr, Savage JH, Wood RA, Saini SS. Suppression of the basophil response to allergen during treatment with omalizumab is dependent on 2 competing factors. J Allergy Clin Immunol 2012;130:1130-5.e5. 6. MacGlashan DW Jr, Saini SS. Syk expression and IgE-mediated histamine release in basophils as biomarkers for predicting the clinical efficacy of omalizumab. J Allergy Clin Immunol 2017;139:1680-2.e10. 7. MacGlashan D, Miura K. Loss of syk kinase during IgE-mediated stimulation of human basophils. J Allergy Clin Immunol 2004;114:1317-24. 8. Macglashan D Jr, Moore G, Muchhal U. Regulation of IgE-mediated signalling in human basophils by CD32b and its role in Syk down-regulation: basic mechanisms in allergic disease. Clin Exp Allergy 2014;44:713-23. 9. Hide M, Francis DM, Grattan CE, Hakimi J, Kochan JP, Greaves MW. Autoantibodies against the high-affinity IgE receptor as a cause of histamine release in chronic urticaria. N Engl J Med 1993;328:1599-604.

8 MACGLASHAN

10. Fiebiger E, Maurer D, Holub H, Reininger B, Hartmann G, Woisetschlager M, et al. Serum IgG autoantibodies directed against the alpha chain of Fc epsilon RI: a selective marker and pathogenetic factor for a distinct subset of chronic urticaria patients? J Clin Invest 1995;96:2606-12. 11. Kikuchi Y, Kaplan AP. Mechanisms of autoimmune activation of basophils in chronic urticaria. J Allergy Clin Immunol 2001;107:1056-62. 12. Soundararajan S, Kikuchi Y, Joseph K, Kaplan AP. Functional assessment of pathogenic IgG subclasses in chronic autoimmune urticaria. J Allergy Clin Immunol 2005;115:815-21. 13. Eckman JA, Hamilton RG, Gober LM, Sterba PM, Saini SS. Basophil phenotypes in chronic idiopathic urticaria in relation to disease activity and autoantibodies. J Invest Dermatol 2008;128:1956-63. 14. MacGlashan D Jr, Hamilton RG. Parameters determining the efficacy of CD32 to inhibit activation of FcepsilonRI in human basophils. J Allergy Clin Immunol 2016;137:1256-8, e1-11. 15. Ishmael SS, MacGlashan DW Jr. Syk expression in peripheral blood leukocytes, CD341 progenitors, and CD34-derived basophils. J Leukoc Biol 2010; 87:291-300.

J ALLERGY CLIN IMMUNOL nnn 2018

16. Chan YC, Ramadani F, Santos AF, Pillai P, Ohm-Laursen L, Harper CE, et al. ‘‘Auto-anti-IgE’’: naturally occurring IgG anti-IgE antibodies may inhibit allergen-induced basophil activation. J Allergy Clin Immunol 2014;134: 1394-401.e4. 17. Lichtenstein LM, Kagey-Sobotka A, White JM, Hamilton RG. Anti-human IgG causes basophil histamine release by acting on IgG-IgE complexes bound to IgE receptors. J Immunol 1992;148:3929-36. 18. Kepley CL, Cambier JC, Morel PA, Lujan D, Ortega E, Wilson BS, et al. Negative regulation of FcεRI signaling by FcgRII costimulation in human blood basophils. J Allergy Clin Immunol 2000;106:337-48. 19. Cassard L, Jonsson F, Arnaud S, Daeron M. Fcgamma receptors inhibit mouse and human basophil activation. J Immunol 2012;189:2995-3006. 20. Rauber MM, Pickert J, Holiangu L, Mobs C, Pfutzner W. Functional and phenotypic analysis of basophils allows determining distinct subtypes in patients with chronic urticaria. Allergy 2017;72:1904-11. 21. Baker R, Vasagar K, Ohameje N, Gober L, Chen SC, Sterba PM, et al. Basophil histamine release activity and disease severity in chronic idiopathic urticaria. Ann Allergy Asthma Immunol 2008;100:244-9.

MACGLASHAN 8.e1

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

METHODS Materials, buffers, and antibodies The following were purchased: piperazine-N,N9-bis(2-ethanesulfonic acid) (PIPES); BSA; ethyleneglycol-bis-(b-aminoethylether)-N,N,N9, N’-tetraacetic acid; EDTA; D-glucose; 2-mercaptoethanol; RPMI 1640 containing 25 mmol/L HEPES and L-glutamine (BioWhittaker, Walkersville, Md); StemCell Pro (Life Technologies, Carlsbad, Calif); Percoll (Pharmacia, Piscataway, NJ); Tris(hydroxymethyl)-aminomethane, Tween-20 (Bio-Rad Laboratories, Hercules, Calif); Syk inhibitor NVP-QAB205 (a gift of GlaxoSmithKline (Research Triangle Park, NC); the Src family kinase inhibitor PP1 (Calbiochem, San Diego, Calif); PCI-32765 (MedChem Express, Monmouth Junction, NJ); LY2940002 (Calbiochem, La Jolla, Calif); mouse anti-human IgE antibody (6061P; Hybridoma Farms, Belair, MD); goat polyclonal anti-hIgE antibody generated and enriched as described previouslyE1; anti-CD32b, Ab10523 (Xencor, Moravia, Calif); anti-CD32a antibody, clone IV.3 (StemCell Technologies, Seattle, Wash); control mouse IgG1 (Zymed, San Francisco, Calif); control mouse IgG2b (eBioscience, San Diego, Calif); anti–IL-3R–phycoerythrin (BD PharMingen, San Jose, Calif); anti-mouse IgG2b–Alexa Fluor 647 (Molecular Probes, Invitrogen, Carlsbad, Calif); anti-FcεRI, 15A5, and 22E7 (gift from Hoffman-LaRoche, Basel, Switzerland); human IgG (Cappel Laboratories/MP Biomedicals, Santa Ana, Calif); anti-Syk (4D10; Santa Cruz Antibodies, Santa Cruz, Calif); and PIPES-albumin-glucose (PAG) buffer consisting of 25 mmol/L PIPES, 110 mmol/L NaCl, 5 mmol/L KCl, 0.1% glucose, and 0.003% HSA. PAGCM was PAG supplemented with 1 mmol/L CaCl2 and 1 mmol/L MgCl2. Labeling with antibodies for flow cytometry was conducted in PAG containing 0.25% BSA in place of 0.003% HSA (EB). ESB is Novex electrophoresis sample buffer containing 5% 2-mercaptoethanol. Lactic acid IgE elution buffer contained 0.01 mol/L lactic acid, 0.14 mol/L NaCl, and 0.005 mol/L KCl at pH 3.9.E2,E3

gentamicin (100 mg/mL) and supplemented to 1 mmol/L CaCl2. When cells were obtained by means of venipuncture and enriched by using 2-step Percoll gradients, the cell-culture density was 2 million total cells/mL of medium. For some experiments, a portion of endogenous IgE was removed by treating the cells with lactic acid buffer (on ice for 6 seconds with buffer before, adding 1.5 mL of EB to neutralize the pH).E3 Histamine release was measured from supernatants of cells stimulated in PAGCM buffer. Histamine was measured by using automated fluorimetry.E8 For desensitization experiments, there were 2 phases to the reaction with or without stimulation with 6061P (anti-IgE antibody) in the absence of extracellular calcium (plus 50 mmol/L EDTA) for 90 minutes, followed by restimulation with anti-IgE antibody or serum for 45 minutes in 1 mmol/L Ca11.

Definition of positivity As the ABO-compatibility issue and the initial screening of sera were explored, a data set of histamine results from the 25% sera challenge of recipient basophils was generated. A simple algorithm for iterating toward a threshold in the presence of sparse true signals was used.E9 All of the results (responses to 25% serum) were first averaged and labeled as A1. This average (A1) was naturally skewed by the presence of eventual positive results. Using A1 as a threshold, the data set was reaveraged by including only results less than A1, and this was labeled as A2. Under the assumption that A2 represented a close approximation of the true background signal, the SD (SD2) of this distribution was also calculated, and a threshold of positivity was set as A2 1 1.96*SD2 (P > .975 of the assumed Gaussian distribution). This value was 6% (note that this is greater than spontaneous release in the histamine release calculation).

Flow cytometry Preparation of FcεRIa-, IgE-, and HSA-Sepharose A Flag-tagged FcεRIa transfection plasmid was a generous gift of Dr Robert Anthony (Harvard, Boston, Mass). The plasmid was used to transfect HEK cells, as described previously.E4 Briefly, 9 million HEK cells were transfected with 60 mg of plasmid in a solution of PEI (Transporter 5; Polysciences, Warrington, Pa)/saline (240 mL of stock in 3 mL of saline) cultured in 2xT75 flasks (10 mL of DMEM medium per flask). Supernatants from the flasks after 72 hours was adsorbed with anti-FLAG antibody beads (300 mL of stock beads, Sigma-Aldrich, St Louis, Mo). The beads were used for adsorption experiments after equilibrating into 13 PIPES buffer. IgE-Sepharose was prepared by coupling 750 mg of JK IgE (Hybridoma Farms) to 80 mg of CNBr-Sepharose 4B fast-flow beads (Sigma-Aldrich). HSA was coupled in a similar manner. For pilot studies, test solutions of either 22E7 diluted to a concentration that was near the optimum for histamine release or goat anti-IgE antibody also titrated to near its optimum for histamine release were treated with different quantities of all 3 beads, and the optimum for specific adsorption was chosen for subsequent experiments.

Basophil purification Basophils were purified from leukopheresis packs or from blood obtained by means of venipuncture. When used at high purity, they were purified to near homogeneity by using sequential application of Percoll gradients and negative selection with a basophil purification kit (STEMCELL Technologies, Vancouver, British Columbia, Canada) and columns from Miltenyi Biotec (Auburn, Calif).E5 The average purity of these basophils, as determined by using Alcian blue staining,E6 was 99%. Starting viability of these cells was typically greater than 97%. For basophils obtained by means of venipuncture (this study was approved by the Johns Hopkins Institutional Review Board), basophils were enriched on a 2-step Percoll gradient.E7

Reaction conditions For experiments requiring longer incubations (functional and expression level studies), culture conditions were RPMI-1640 containing 0.03% HSA and

With some noted exceptions, flow cytometry was performed on cells fixed in 2% paraformaldehyde for 20 minutes at room temperature before blocking with 4% BSA-PBS overnight. If the basophils were highly purified, there was no need for selection antibodies, but if they were only enriched, gating for basophils was based on forward-scatter/side-scatter gates and labeling with anti-FcεRI antibody.E10 The mAbs 22E7 (a gift of Hoffman-LaRoche) or CRA-1 (eBioscience; both antibodies specific for FcεRIa) were used interchangeably (depending on which isotype mouse antibody was needed) to label FcεRI. Isotype control antibodies or specific antibodies were incubated with basophils in EB buffer for 30 minutes before a secondary incubation with labeled secondary antibodies for 30 minutes, followed by washing and analysis. If SYK measurement was the target of the assay, the SYK-selective antibody 4D10 was used in a previously published protocol.E11 If CD32a receptors were the target for labeling, the protocol, as described previously,E12 included 20 mg/mL Ab10523 (anti-FcgRIIb/CD32b) because clone IV.3 (anti-FcgRIIa/CD32a) could bind to FcgRIIb/CD32b through its Fc region. If the FcgRIIb/CD32b receptors were the target, the ‘‘control’’ labeling included 20 mg/mL Ab10523 (anti-FcgRIIb/CD32b, see above), as well as 1 mg/mL Ab10523–Alexa Fluor 647, whereas detection of FcgRIIb/ CD32b included only Ab10523–Alexa Fluor 647. Because Ab10523 (anti-FcgRIIb/CD32b) does not bind to FcgRIIb/CD32b through its Fc region, no blocking of FcgRIIa/CD32a was necessary. In general, mean net fluorescence is reported. Calibration beads were used to ensure the flow cytometer was producing similar results between daily and weekly measurements.

Western blotting Pelleted cells were lysed in 20 mL of hot ESB, the tube was placed in boiling water for 5 minutes, and samples were stored at 2808C until electrophoresis. Samples were run in a 15-well 8% Tris-glycine gel with molecular weight markers. After transblotting to nitrocellulose, membranes were blocked with 4% BSA. Equal lane loading was confirmed by blotting with anti-p85a antibody. We have used this method in a variety of different studies and have found that basophils do not change their p85 levels during IgE-mediated stimulation for short or long periods.E13

8.e2 MACGLASHAN

ABO compatibility All of our donors who supplied basophils for the bioassay of sera-induced histamine release were typed for the AB and Rh antigens. Sera were also obtained from these donors. Donors and sera were tested for histamine release by stimulating a ‘‘recipient’’ (Percoll-separated basophils, see above) basophil with 25% serum from either an AB/Rh-matched donor or a mismatched AB/Rh donor. For a series of 24 donor sera (from subjects without CIU) first tested on a 3 O1 basophils for whom optimal IgE-mediated histamine release averaged 28%, no sera were found to induce histamine release. For a series of 20 sera that were tested on deliberately mismatched ‘‘recipient’’ basophils (eg, A1 on B1), only 1 serum induced modest release (5% to 10%). Further exploration of this example showed that it would induce release only when paired with one particular recipient basophil. Type matching or mismatching could be done without release except for the one particular ‘‘recipient’’ donor. During the course of additional experiments, one additional donor was found with this characteristic. Therefore AB/Rh matching appears irrelevant to whether sera induce release, but in approximately 5% of sera tested, there could be some release induced that could be considered idiosyncratic to specific pairings of ‘‘recipient’’ basophils and serum. Despite the appearance of no problem with AB/Rh matching, for the remainder of the experiments in which serum was tested on ‘‘recipient’’ basophils, these basophils were chosen to be O1 or O2.

RESULTS Difference in positivity between general and basopenic groups There were 3 subjects who appeared in both the general CSU group and in the subsequently screened CSU group with basopenia. Therefore, statistically, the 2 groups are not independent. Removal of the 3 basopenic subjects from the general group (and retaining them in the basopenic group) results in a x2 result of 6.26 (P 5 .011; ie, a difference in frequency between groups), but performing the analysis by removing them from the basopenic group (and retaining them in the general group) leads to no difference in frequency (P 5 .20). Drug sensitivity The first assay tested the effects of 4 pharmacologic inhibitors known to inhibit enzymes involved in IgE-mediated secretion. Previous studies have identified agents that operate on signal transduction steps that are relatively specific for the IgE-mediated signaling cascade but not effective for guanosine triphosphate– binding protein-dependent receptors, such as formyl-met-leu-phe receptor. Four points in the early IgE-mediated cascade were chosen: the Src family kinases, SYK, BTK, and PI3K (likely d in human basophils). The relevant drugs were PP1 or PP2 (Src family kinases, both tested), NVP-QAB205 (SYK inhibitor), PCI32765 (BTK inhibitor), and LY294002 (PI3K inhibitor). Each has received detailed study for their behavior on human basophils.E14E16 Fig E2 shows a schematic of where these drugs are thought to act in the IgE-signaling pathway. For this assay, we discovered that the presence of serum could shift the inhibitory concentration of 50% (IC50) of the drugs being used as indicators of an IgE-mediated reaction. This was most notable for PP1 or PP2, 2 Src family kinase inhibitors. For reasons not explored, for these inhibitors, serum almost completely eliminated the drugs’ effectiveness. Fig E3 summarizes the results for all 4 drugs with and without serum. Therefore results with PP1 or PP2 were not included in subsequent testing. Serum shifted the IC50 of NVP-QAB205 (SYK inhibitor) by 2- to 3-fold,

J ALLERGY CLIN IMMUNOL nnn 2018

LY294002 (PI3K inhibitor) by 6-fold, and PCI-32765 (BTK inhibitor) by only a minor degree. It was not known whether cells were experiencing lower intracellular levels of the drugs (eg, caused by metabolism or binding to sera proteins) or whether a property of the cellular response to the action of the drugs was different in the presence of serum. Because the mechanisms were unclear, for subsequent experiments in the presence of serum, drug concentrations were kept conservatively near the expected IC50 concentrations determined from previous studies. However, this conservative approach led to results in which the ability to inhibit 6061P/anti-IgE antibody–induced release was more variable because of the serum effect, and therefore the method to analyze the results across different experiments was to normalize the extent of inhibition based on how effectively 6061P/anti-IgE antibody–induced release was inhibited in the presence of a nonreleasing serum (see the Methods section). Fig E4, A, plots the similarity index for the 13 multiply positive sera. Four of the sera induced release that was unresponsive to these 3 drugs. A second assay determined whether desensitization of the IgE-mediated pathway also inhibited the ability of the serum to induce release. Parenthetically, we discovered that the inclusion of serum sometimes slowed the rate of desensitization of FcεRI. For this series, basophils were stimulated with 6061P/anti-IgE antibody in the absence of extracellular calcium for 1 hour (or no 6061P/anti-IgE antibody to act as the nondesensitized control), extracellular calcium returned to the reaction, and the cells were stimulated further with 6061P/anti-IgE antibody or serum at a 25% final concentration. The control rechallenge included 6061P/anti-IgE antibody in negative serum or PAGCM. As before, the useful metric was to compare the desensitization inhibition of the serum response with the desensitized inhibition to 6061P/anti-IgE antibody. If 6061P/anti-IgE antibody induced desensitization of serum to an extent similar to 6061P/anti-IgE antibody–induced desensitization to 6061P/anti-IgE antibody, the similarity index was near 1.0. Fig E4, B, plots these results for the 13 multiply positive sera. As found for drug testing, the same 4 sera did not pass the similarity test. The third assay was a natural variation of the NVP-QAB205 (SYK inhibitor) experiments, taking advantage of basophil donors in which the expression level of SYK is so poor that IgE-mediated histamine release is near zero.E3,E17 There is little reason to expect an FcεRI-dependent process or a SYK-dependent process to induce release from these donors’ basophils. Before this experimental series, we noted that some of the 13 multiply positive sera were not found to induce strong release. If release induced with these sera from a nonreleaser was poor, it would be difficult to know whether the release was expected or appropriately suppressed. Therefore the calculation of similarity was based on the expectations for release with each of the 13 sera. This was the last series of studies done, and at this point, there were 4 to 6 independent measurements of these 13 sera to induce release from strongly releasing basophil donors. An average of the multiple measurements for a given serum could be calculated, and the histamine release from the nonreleaser was calculated relative to the expectation. Therefore a value of near zero would be expected for sera that operated through SYK. Several nonreleasing donors were used for this test, and those sera causing release were repeated. With this test, there were only 5 sera that continued to pass the similarity metric (see Fig E4, C).

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

Adsorption of sera Sera were adsorbed with either IgE or FcεRIa beads to verify that these sera likely contained an antibody to IgE or FcεRI. Three sera with sufficient titers were adsorbed with a 12.5% vol/vol quantity of each of these beads for 2 hours at 48C. Bead supernatants were used to stimulate releasing basophils (as above). For positive controls, a solution of anti-FcεRIa (22E7) mAb or anti-IgE (goat) polyclonal antibody (diluted into negative serum) at concentrations that produce optimal and slightly suboptimal histamine release were also adsorbed with the beads. These controls provide a context for the adsorption results with positive sera (not shown in Fig E5, IgE beads did not adsorb 22E7 and a-beads did not adsorb anti-IgE antibody). In all 4 instances tested (1 was excluded because the response was relatively poor for this serum), either IgE (1 instance) or FcεRIa (3 instances) beads inhibited the ability of sera to induce release. Two of the anti-FcεRIa1 sera appeared to also be partially adsorbed with IgE beads and therefore might also be an anti-IgE antibody (Fig E5). In an additional test, basophils were treated with lactic acid buffer to dissociate a significant fraction of their surface IgE antibody or incubated with additional IgE in an attempt to fully occupy any nonoccupied receptors. One set of cells remained untreated. All 3 sets were stimulated with the 5 sera. This test examines whether the anti-FcεRIa antibodies were sensitive to FcεRI occupancy. The anti-IgE–like serum (MM) was inhibited by dissociation. However, each of the anti-FcεRIa–like sera were sensitive to occupancy in that dissociating the endogenous IgE enhanced their ability to induce release and increasing occupancy with sensitization decreased the response to the serum. The one serum that was not examined in the immunoadsorption study was sensitive also to occupancy (ie, serum-induced release was enhanced after IgE dissociation). IL-3 effects on sera-induced histamine release None of the sera from subjects without CIU induced replicable release; this included sera from subjects classified as nonreleasers. An additional test was performed on sera from nonreleasers and releasers. Basophils were cultured for 24 hours with 10 ng/mL IL-3 to markedly enhance their sensitivity to secretagogues. Surprisingly, all sera induced histamine release, but there was no difference between the 2 categories of donor (releaser vs nonreleaser, Fig E6, B; Fig E6, A, shows the release distributions for the basophils stimulated with an optimal concentration of 6061P/anti-IgE antibody of serum source subjects). In addition to the challenge with sera, the effect of the BTK inhibitor PCI-32765 on each response was tested. None of the responses were inhibited (Ratio of response plus inhibitor/ Response 2 Inhibitor 5 1.03 6 0.07; data not shown). In addition, the test was also performed on basophils for which the endogenous IgE was dissociated before culture and challenge. There were no differences in responses to sera and no differences between releasers and nonreleasers (data not shown). Inhibitory effects on IgG The question was whether antibodies that interact with a cell-surface protein (in this case IgE or FcεRI) and are already bound and constrained to the cell surface are inhibited in the interaction of their Fc domains with cell-surface CD32b caused

MACGLASHAN 8.e3

by nsIgG in the extracellular medium. Several known anti-IgE or anti-FcεRI antibodies were examined for the influence of CD32b on their activity and for what concentrations of nsIgG antibody interfered with any interaction. We have demonstrated previously that polyclonal goat anti-IgE antibody interacts with FcgRIIb/ CD32b and that inhibition of this interaction with a selective anti-CD32b antibody (Ab10523) enhances induced histamine release. This interaction is sufficiently strong to observe significant enhancement of release in the presence of Ab10523 (anti-FcgRIIb/CD32b). Fig E7, A, shows that nsIgG can also reverse interaction of goat anti-IgE with FcgRIIb/CD32b with an IC50 of approximately 2.5 mg/mL, with full reversal (relative to Ab10523 [anti-FcgRIIb/CD32b]) not apparent at 10 mg/mL. A mouse IgG2b anti-IgE antibody was examined. There have been indications in previous studies that there is some interaction of mouse antibodies with CD32b, but as is apparent in Fig E7, B, enhancement of release is modest with Ab10523 (anti- FcgRIIb/ CD32b). The inclusion of nsIgG also enhances release to an extent exceeding somewhat or equivalent to Ab10523 (anti-FcgRIIb/CD32b) with an IC50 of 300 mg/mL. We found that 22E7, an anti-FcεRIa antibody is also only weakly enhanced with Ab10523 (anti-FcgRIIb/CD32b; data not shown; enhancement, 1.20 6 0.05-fold). These results suggest that interaction with FcgRIIb/CD32b varies considerably between anti-IgE/Fc antibodies, with goat antibody showing the strongest interaction. These experiments suggested that the range of possible interaction is large but that a typical human IgG interacting with CD32b might be reversible by competing with nsIgG. Therefore these results suggest that (1) nsIgG levels of less than 1.5 mg/mL (equivalent to 10% serum) could be tolerated if a human anti-FcεRI or anti-IgE had Fc-binding characteristics, such as goat anti-IgE, whereas 10-fold lower concentrations would be needed (1% serum) if the characteristics were like those of mouse anti-IgEs. REFERENCES E1. Adkinson NF Jr. Measurement of Total Serum Immunoglobulin E and Allergen-Specific Immunoglobulin E antibody. 2nd ed. Washington (DC): American Society of Microbiology; 1980. E2. Pruzansky JJ, Grammer LC, Patterson R, Roberts M. Dissociation of IgE from receptors on human basophils. I. Enhanced passive sensitization for histamine release. J Immunol 1983;131:1949-53. E3. MacGlashan DW Jr. Relationship between Syk and SHIP expression and secretion from human basophils in the general population. J Allergy Clin Immunol 2007;119:626-33. E4. Shade KT, Platzer B, Washburn N, Mani V, Bartsch YC, Conroy M, et al. A single glycan on IgE is indispensable for initiation of anaphylaxis. J Exp Med 2015;212:457-67. E5. Lavens-Phillips SE, MacGlashan DW Jr. The tyrosine kinases, p53/56lyn and p72syk are differentially expressed at the protein level but not at the mRNA level in non-releasing human basophils. Am J Respir Cell Mol Biol 2000;23:566-71. E6. Gilbert HS, Ornstein L. Basophil counting with a new staining method using Alcian blue. Blood 1975;46:279-86. E7. Warner JA, Yancey KB, MacGlashan DW Jr. The effect of pertussis toxin on mediator release from human basophils. J Immunol 1987;139:161-5. E8. Siraganian RP. An automated continuous-flow system for the extraction and fluorometric analysis of histamine. Anal Biochem 1974;57:383-94. E9. MacGlashan DW Jr. Single-cell analysis of Ca11 changes in human lung mast cells: graded vs. all-or-nothing elevations after IgE-mediated stimulation. J Cell Biol 1989;109:123-34. E10. MacGlashan DW Jr. Endocytosis, re-cycling and degradation of unoccupied FcεRI in human basophils. J Leuk Biol 2007;82:1003-10. E11. MacGlashan DW Jr, Ishmael S, Macdonald SM, Langdon JM, Arm JP, Sloane DE. Induced loss of Syk in human basophils by non-IgE-dependent stimuli. J Immunol 2008;180:4208-17.

8.e4 MACGLASHAN

E12. Macglashan D Jr, Moore G, Muchhal U. Regulation of IgE-mediated signalling in human basophils by CD32b and its role in Syk down-regulation: basic mechanisms in allergic disease. Clin Exp Allergy 2014;44:713-23. E13. MacGlashan D Jr. Subthreshold desensitization of human basophils re-capitulates the loss of Syk and FcepsilonRI expression characterized by other methods of desensitization. Clin Exp Allergy 2012;42:1060-70. E14. Lavens-Phillips SE, Miura K, MacGlashan DW Jr. Pharmacology of IgE-mediated desensitization of human basophils: Effects of protein kinase C and Src-family kinase inhibitors. Biochem Pharmacol 2000;60:1717-27.

J ALLERGY CLIN IMMUNOL nnn 2018

E15. MacGlashan DW Jr, Undem BJ. Inducing an anergic state in mast cells and basophils without secretion. J Allergy Clin Immunol 2008;121: 1500-6. E16. MacGlashan D Jr, Honigberg LA, Smith A, Buggy J, Schroeder JT. Inhibition of IgE-mediated secretion from human basophils with a highly selective Bruton’s tyrosine kinase, Btk, inhibitor. Int Immunopharmacol 2011;11:475-9. E17. Kepley CL, Youssef L, Andrews RP, Wilson BS, Oliver JM. Syk deficiency in nonreleaser basophils. J Allergy Clin Immunol 1999;104:279-84.

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

FIG E1. Serum-induced histamine release distributions for the 3 categories of subject. The final concentration of serum was 25%. The initial screening required that a serum induce release consistently, and a failure in a second screen removed the serum from consideration. Sera that passed the first screening are shown as pairs, with those that failed the second screening shown as open circles and those that passed the second screening as solid gray circles. The average of release of the first test of eventual double-positive results was 43%, whereas the average release for the first test of eventual single-positive results was 19%. The average anti-IgE antibody–induced release of the recipient basophils was 70%.

MACGLASHAN 8.e5

8.e6 MACGLASHAN

FIG E2. Cartoon of the early IgE-mediated signaling reaction to show the relative position in the cascade and the drugs that inhibit each step.

J ALLERGY CLIN IMMUNOL nnn 2018

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

FIG E3. IC50 of 4 drugs with specificity for IgE-mediated secretion with and without 25% serum present. A, Src family kinase inhibitor PP1 (n 5 3). Solid circle, No serum present; solid square, 25% serum present (applies to all panels). B, SYK inhibitor NVP-QAB205 (n 5 1). C, BTK inhibitor PCI-32765 (n 5 1). D, PI3K inhibitor LY294002 (n 5 1). DMSO, Dimethyl sulfoxide.

MACGLASHAN 8.e7

8.e8 MACGLASHAN

J ALLERGY CLIN IMMUNOL nnn 2018

= lightest gray bar, BTK inhibitor PCI-32765 at 75 nmol/L; medium gray bar,

FIG E4. Three tests for sera similarity to an IgE-mediated response. The Methods section describes the metric of similarity as relevant to each test. For some of the sera for which results neared thresholds for exclusion, the tests were repeated (eg, ER). A, Similarity index (>0.90 showing similarity to an IgE-mediated response) for 3 selective inhibitors of the FcεRI-mediated activation pathway. Black bar, SYK inhibitor NVP-QAB205 at 0.75 mmol/L;

PI3K inhibitor LY294002 at 7.5 mmol/L. Asterisks denote sera not passing this test. B, Similarity index for cross-desensitization between a known anti-IgE antibody and sera-induced release (>0.90 showing similarity to an IgE-mediated response). Asterisks denote sera not passing this test. C, Ability of sera to induce release from a known nonreleasing basophil. The metric for this test was the fraction of the expected response for each specific serum based on past performance of the same serum on releasing basophils. To accommodate variation, a value of greater than 0.3 was considered a positive response from a nonreleaser basophil, although a threshold of less than 0.05 would be appropriate for a known aggregating anti-IgE antibody. Asterisks denote sera not passing this test.

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

FIG E5. Adsorption of sera with either HSA-coupled, IgE-coupled, or FcεRIa-coupled Sepharose beads. Basophils were challenged with stimuli that had been adsorbed with the various stimuli shown. The ordinate metric is the response relative to the response to a stimulus after adsorption with HSA-coupled beads. Control responses (absorption with HSA-coupled beads) to anti-IgE antibody (goat anti-IgE antibody), 22E7 (anti-FcεRIa), and AB, MM, AS, and DS sera were 83%, 64%, 18%, 36%, 42%, and 5%, respectively. Data show averages from 3 experiments for the 22E7 (anti-FcεRIa) and anti-IgE antibody (goat anti-IgE antibody), MM, and AS and 1 experiment for DS and AB.

MACGLASHAN 8.e9

8.e10 MACGLASHAN

FIG E6. Effect of overnight IL-3 (10 ng/mL) on the response of basophils to releaser (R) and nonreleaser (NR) sera. A, Histamine release from basophils of serum donors categorized as releasers or nonreleasers. Points indicate the response of each donor’s basophils to stimulation with anti-IgE antibody/6061P at an optimal concentration (0.5 mg/mL). Horizontal bars represent averages. B, Sera from these basophil phenotypes were used to stimulate basophils first cultured for 18 hours with IL-3 at 10 ng/mL. Points are responses to each serum tested, and horizontal bars represent averages 6 SDs.

J ALLERGY CLIN IMMUNOL nnn 2018

J ALLERGY CLIN IMMUNOL VOLUME nnn, NUMBER nn

MACGLASHAN 8.e11

FIG E7. Sensitivity of mouse and goat anti-IgE antibodies to nsIgG to block CD32b. A, The positive control (rightmost bar) in this protocol is inclusion of Ab10523 (anti-FcgRIIb/Cd32b) at 50 mg/mL to block participation of FcgRIIb/CD32b during stimulation with a mouse anti-hIgE mAb. The leftmost bar indicated the response without Ab10523 (anti-FcgRIIb/Cd32b). In between is the concentration dependence of nonspecific hIgG to block participation of FcgRIIb/Cd32b in the response to mouse anti-IgE antibody. B, Similar to Fig E7, A, except the stimulus was goat polyclonal anti-IgE antibody.