Article Sialylation of immunoglobulin E is a determinant of allergic pathogenicity

https://doi.org/10.1038/s41586-020-2311-z Kai-Ting C. Shade1,5, Michelle E. Conroy1,5, Nathaniel Washburn2, Maya Kitaoka1, Daniel J. Huynh1, Emma Laprise1, Sarita U. Patil1,3,4, Wayne G. Shreffler1,3,4 & Received: 13 May 2019 Robert M. Anthony1 ✉ Accepted: 11 March 2020

Published online: 20 May 2020 Approximately one-third of the world’s population sufers from allergies1. Exposure to Check for updates allergens crosslinks immunoglobulin E (IgE) antibodies that are bound to mast cells and basophils, triggering the release of infammatory mediators, including histamine2. Although IgE is absolutely required for allergies, it is not understood why total and allergen-specifc IgE concentrations do not reproducibly correlate with allergic disease3–5. It is well-established that glycosylation of IgG dictates its efector function and has disease-specifc patterns. However, whether IgE glycans difer in disease states or afect biological activity is completely unknown6. Here we perform an unbiased examination of glycosylation patterns of total IgE from individuals with a peanut allergy and from non-atopic individuals without allergies. Our analysis reveals an increase in sialic acid content on total IgE from individuals with a peanut allergy compared with non-atopic individuals. Removal of sialic acid from IgE attenuates efector-cell degranulation and anaphylaxis in several functional models of allergic disease. Therapeutic interventions—including removing sialic acid from cell-bound IgE with a neuraminidase enzyme targeted towards the IgE receptor FcεRI, and administering asialylated IgE—markedly reduce anaphylaxis. Together, these results establish IgE glycosylation, and specifcally sialylation, as an important regulator of allergic disease.

IgE antibodies bind to the surface of mast cells or basophils that express class in circulation, and, as such, analysis of hIgE glycosylation has the high-affinity IgE receptor FcεRI (ref. 2). Subsequent exposure to been restricted to samples from individuals with myelomas, hyper-IgE allergens crosslinks cell-bound IgE, leading to cellular activation and syndromes, hyperimmune syndromes pooled from multiple donors, the release of allergic mediators, including histamine, prostaglandins or recombinant IgE15–18. These studies revealed that there is a single and leukotrienes2. This cascade culminates in the canonical symptoms N-linked oligomannose glycan at N394 on IgE, that N383 is unoccupied, of allergic diseases, the most severe of which is anaphylaxis. Although and that the remaining five sites are occupied by complex antennary IgE that recognizes otherwise innocuous allergens is well-established glycans (Fig. 1a). Previously, the importance of glycans to IgE biology as the causative agent of most allergic diseases1,2, clinical allergy diag- has been examined through treatment with glycosidases (enzymes nostics remain relatively inaccurate3–5, and curative therapies, including that hydrolyse glycosidic bonds)17,19 and through mutation of glyco- oral immunotherapy, are cumbersome and only partially effective7,8. sylation sites17,20. This revealed that the N394-linked oligomannose was Further, allergen-specific IgE is detected in many people who do not required for appropriate IgE folding and FcεRI binding17,20 to initiate experience allergic symptoms3,5. Thus, while IgE is absolutely neces- effector functions. sary for triggering the allergic cascade, it is not clear how IgE causes Here we investigated whether allergic-disease-specific glycosylation allergic disease in some circumstances and not others. patterns exist for IgE, and, if so, whether those patterns influence the The composition of the glycans attached (via a single asparagine biological activity of IgE. We characterized individuals as non-atopic if (N) residue) to immunoglobulin G markedly influences its biologi- they reported no history of atopy, with low total IgE titres and little IgE cal activity, and affects the outcome of many diseases, including reactivity to peanut allergen (Ara h 2), birch tree pollen allergen (Bet v 1), Dengue haemorrhagic fever9, Mycobacterium tuberculosis latency10, house dust mite allergen (Der p 1) or cat allergen (Fel d 1) (Fig. 1b, c, influenza vaccination11, rheumatoid arthritis6,12, and granulomato- Extended Data Fig. 1a and Extended Data Table 1). Individuals with a sis with polyangiitis13,14. There are seven N-linked glycosylation sites peanut allergy reported multiple atopies, had approximately twofold distributed across the heavy chains of human IgE (hIgE)6,15. However, higher total IgE titres and reactivity to peanut allergen (Ara h 2) but not to whether particular IgE glycans are associated with allergic diseases, other tested allergens, and were confirmed by clinician-supervised oral or affect IgE function, is unknown. IgE is the least abundant antibody challenge (Fig. 1b, c, Extended Data Fig. 1a and Extended Data Table 1)8.

1Center for Immunology and Inflammatory Diseases, Division of Rheumatology, Allergy and Immunology, Department of Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. 2Momenta Pharmaceuticals Inc, Cambridge, MA, USA. 3Division of Pediatric Allergy and the MGH Food Allergy Center, Department of Pediatrics, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA. 4Food Allergy Science Initiative at the Broad Institute of the Massachusetts Institute of Technology and Harvard, Cambridge, MA, USA. 5These authors contributed equally: Kai-Ting C. Shade, Michelle E. Conroy. ✉e-mail: [email protected]

Nature | Vol 582 | 11 June 2020 | 265 Article

acb P = 0.0099 900 80 P = 0.0186 Human IgE ) Non-atopic –1 Fab Allergic N140 Cε1 N168 60 N218 ) N265 Cε2 –1 600

N371 40 Fc N383 x Cε3 x N394 tal IgE (kU l 300

Cε4 To 20

Complex N gen speci c IgE levels (kU l

0 Aller 0 Oligomannose Ara h 2 Der p 1 Fel d 1 Bet v 1 (peanut) (dust mite) (cat) (birch pollen)

d P = 0.9645 e P = 0.1266 f P = 0.0491 g P < 0.0001 h P = 0.0009 16 9.0 5 3 15

15 4 esidues esidues esidues esidues

14 esidues 8.5 2 12 3 13 2

12 8.0 ) per IgE molecule 1 9 ) per IgE molecule ( ) per IgE molecule ( ( ( ) per IgE molecule ( ) per IgE molecule 1 Number of fucose r

Number of mannose r 11 Number of galactose r Number of biGlcNAc r Number of sialic acid r

10 7.5 0 0 6 P < 0.0001 S1, P = 0.0020 Trisialylated i j (S1, S2) S2, P < 0.0001 100 (S3) 100 Disialylated (S2) 80 80 Monosialylated } (S1) 60 60 Digalactosylated (G2) Monogalactosylated 40 40 rue positive AUC } (G1) T Sialic acid 0.970 Glycans (%) Man-5 Man-8 Galactose 0.970 20 Fucose 0.691 20 biGlcNAc 0.560 Man-6 Man-9 Oligomannose 0.524 0 0 020406080 100 N A NA NA NA N A NA NA Man-7 Other False positive Unoccupied N140 N168 N218 N265 N371 N383 N394 N = Non-atopic IgE glycan sites A = Allergic

Fig. 1 | Glycan composition of non-atopic and allergic IgE. a, Human IgE with allergies versus non-atopic individuals: sialic acid (non-atopic, n = 9; allergic, the Fab and Fc domains, as well as N-linked glycosylation sites, shown. Filled n = 11); galactose (non-atopic, n = 14; allergic, n = 19); fucose (non-atopic, n = 14; circles, complex biantennary glycans; hatched circles, oligomannose; X, allergic, n = 15); biGlcNAc (non-atopic, n = 14; allergic, n = 19); and oligomannose unoccupied; blue squares, GlcNAc; green circles, mannose; red triangle, (non-atopic, n = 15; allergic, n = 14). AUC, area under curve. j, gMS analysis of fucose; yellow circles, galactose; maroon diamonds, sialic acid. b, c, Total IgE site-specific N-glycan structures on total IgE from non-atopic individuals titres (b) and allergen-specific IgE levels (c) in non-atopic individuals (blue, (N140, n = 11; N168, n = 13; N218, n = 11; N265, N371, N394, n = 12) and individuals n = 17) and individuals with allergies (red, n = 13). d–h, Glycan moieties per IgE with allergies (N140, n = 11; N168, n = 15; N218, n = 17; N265, n = 18; N371, n = 14; molecule, quantified by glycopeptide mass spectrometry (gMS), in N394, n = 12). Representative glycan structures per group are detailed in non-atopic individuals (blue) and individuals with a peanut allergy (red): Extended Data Fig. 1h. Data shown are means ± s.e.m. (b, c, j), medians (solid mannose (d; non-atopic, n = 15; allergic, n = 14); fucose (e; non-atopic, n = 10; lines) and interquartile ranges (dotted lines) (d–h); two-tailed unpaired t-test allergic, n = 11); biGlcNAc (f; non-atopic, n = 10; allergic, n = 11); galactose (g; (b, d–h), two-way analysis of variance (ANOVA) with Sidak’s (c) or Tukey’s non-atopic, n = 14; allergic, n = 19); and sialic acid (h; non-atopic, n = 9; allergic, multiple comparison test (j). n = 11). i, ROC for total IgE glycan moieties isolated from individuals with

We sensitized human LAD2 mast cells with similar amounts of total Next, we analysed the N-glycans on total IgE enriched from these IgE enriched from the sera of these cohorts and activated the cells by two cohorts by mass spectrometry16,18,21 (Extended Data Figs. 2, 3). The anti-IgE crosslinking. We observed less degranulation—as measured mannose content on oligomannose moieties was similar between total by β-hexosaminidase release—in mast cells sensitized with IgE isolated non-atopic and allergic IgE (Fig. 1d). Fucose, N-acetyl glucosamine from sera of non-atopic individuals compared with patients with a pea- (GlcNAc), galactose and sialic acid can be attached to complex gly- nut allergy (Extended Data Fig. 1c), despite similar surface IgE loading cans (Fig. 1a). Although the total fucose content was similar between (Extended Data Fig. 1d, e). This suggested intrinsic functional differences non-atopic and allergic IgE (Fig. 1e), substantially increased levels of between non-atopic and allergic IgE, independent of allergen specificity. bisecting GlcNAc (biGlcNAc) and terminal galactose were found on

266 | Nature | Vol 582 | 11 June 2020 a Anti-OVA–mIgE b c PBS Sia As Sia Sia As P < 0.0001 mIgE versus mIgE mIgE mIgE mIgE ) 1.0 kDa P < 0.0001 2.5 AsmIgE SNA: 205 620 80 α2,6- 0.8 2.0 sialic acid 150 AsmIgE versus Re-siamIgE 60 0.6 1.5 205 Coomassie 40 0.4 1.0

150 Count PBS 0.2 0.5 20 ascular leakage (OD V 0 0 0 Fold change in anti-mIgE MFI –103 0 103 104 105 PBS Re-sia PBS mIgE mIgE a mIgE mIgE mIgE–FITC Sia As mIgE Si As Sia As dePBS SiamIgE AsmIgE PBS mIgE mIgE f SiamIgE, t1/2 = 11.92 h g PBS SiamIgE AsmIgE AsmIgE, t1/2 = 10.85 h 1 15 P < 0.0001 1 P < 0.0001 0 0 )

–1 10 10 –1 –1 e change (˚C) e change (˚C) –2 –2 5 5 mIgE (ng µl –3 Histamine (µM) –3 emperatur emperatur –4 T P < 0.0001 T –4 0 0 –5 0102030405060 0102030405060 0612 18 24 0102030405060 Time (min) Time (min) Time (h) Time (min) + 10 µg ml–1 OVA h i P < 0.0001 PBS j k SiahIgE P < 0.0001 PBS Anti-OVA–hIgE SiahIgE 40 P < 0.0001 3 AshIgE SiahIgE 40 AshIgE 0.56 kDa SiahIgE AshIgE AshIgE P < 0.0001 SNA: 205 P < 0.0001 30 PBS α2,6- 30 2 150 P = 0.0017

sialic acid (%) 20 P < 0.0001 + 2.65 205 20 P < 0.0001 Coomassie SiahIgE 150 CD63 1 105 10 4 degranulation (%) 10 degranulation (%) β -Hexosaminidase 10 β -Hexosaminidase 103 102 1.07 1 0 0 0 10 As 0 10 2 0.4 0 10 2 0.4 0 10 2.5 0 CD123–PE 10 hIgE OVA (µg ml–1) OVA (µg ml–1) OVA (µg ml–1) 102103104105106107 CD63–FITC

Sia Sia l hIgE to hFcεRIα; KD = 3.57 nM m hIgE to OVA; KD = 0.697 nM As As hIgE to hFcεRIα; KD = 4.35 nM hIgE to OVA; KD = 0.701 nM 0.5 0.5 0.4 0.4 0.15 0.15 0.3 0.3 0.10 0.10 0.2 0.2 0.05 0.05

Binding (nm ) 0.1 0.1 Binding (nm) 0 0 0 0 0200 400600 800 1,000 1,200 0200 400 600800 1,000 1,200 0 200 400 600 800 1,000 1,200 0 200 400 600 800 1,000 1,200 Time (s) Time (s) Time (s) Time (s)

Fig. 2 | Removal of sialic acid attenuates IgE. a, Blot for Sambucus nigra (SNA) Temperature change following passive food anaphylaxis elicited by oral TNP– lectin (top) and Coomassie-stained gel (bottom) of OVA-specific SiamIgE and OVA administration in mice sensitized with PBS, TNP-specific SiamIgE or AsmIgE AsmIgE. b, Left, quantification of ear blue colouration, and right, representative (n = 2, 4 and 4, respectively). h, SNA lectin blot and Coomassie-stained gel of ear images following OVA-induced PCA by PBS, OVA-specific SiamIgE, AsmIgE or OVA-specific SiahIgE and AshIgE. i–k, OVA-induced degranulation in LAD2 mast Re-Sia mIgE (n = 2, 8, 12 and 4 mouse ears respectively). OD620, optical density at cells (i), peripheral blood mononuclear cell derived human mast cells (j) or 620 nm. c, Left, mean fluorescence intensity (MFI), and right, representative basophils (k) sensitized with PBS (i, n = 3; k, n = 1), OVA-specific SiahIgE or AshIgE histograms of anti-mIgE determined by fluorescence-activated cell sorting (i, n = 3; j, n = 3; k, n = 4). n is the number of technical replicates and is (FACS) on dermal mouse ear mast cells following sensitization with PBS, representative of three biologically independent experiments. PE, OVA-specific SiamIgE or AsmIgE (n = 5, 6 and 6 mouse ears, respectively, from phycoerythrin. l, m, Binding kinetics of OVA-specific SiahIgE or AshIgE to two independent experiments). FITC, fluorescein isothiocyanate. hFcεRIα (l) or OVA (m). KD, equilibrium dissociation constant. Data shown are d, e, Temperature change (d) and serum histamine levels (e) following means ± s.e.m. (b–g, i–k) and are representative of two (f) or three DNP-induced PSA in mice sensitized with PBS, DNP-specific SiamIgE or AsmIgE (a, b, d, e, g–m) independent experiments. One-way ANOVA with Tukey’s (b), or (n = 3, 5 and 5 mice, respectively). f, Serum levels of DNP-specific SiamIgE (n = 4 two-way ANOVA with Tukey’s (d, e, i, k) or Sidak’s (g, j) multiple comparison mice) or AsmIgE (n = 5 mice) after intraperitoneal administration. g, test. For gel source data, see Supplementary Fig. 1.

non-atopic IgE (Fig. 1f, g), whereas increased terminal sialylation was IgE were uniquely strong predictors of allergic disease. Of note, differ- detected on allergic IgE (Fig. 1h). ences in IgE sialylation were not sex- or age-dependent (Extended Data To determine whether the glycan differences on total IgE were pre- Fig. 1f, g). Glycosylation-site analysis showed that N140, N168, N265 and dictive of allergic disease, we assessed the variable glycan content on N394 of IgE were fully occupied by N-linked glycans, N218 and N371 were non-atopic and allergic IgE using receiver operating characteristics (ROC) partially occupied (75% and 30% respectively), and N383 was completely curves (Fig. 1i). We found that the galactose and sialic acid contents of unoccupied (Fig. 1j and Extended Data Fig. 4a), consistent with previous

Nature | Vol 582 | 11 June 2020 | 267 Article

Anti-OVA–SiahIgE + SiaFetuin Sia As c b PBS hIgE hIgE Anti-OVA–SiahIgE + AsFetuin a PBS SiahIgE AshIgE + 10 µg ml–1 OVA P = 0.0346 0530 0530 0530 Time (min) 4 1.2 + OVA

pSyk 2 3 1.0 ) 0 (Tyr 352) 0 /F

F / 2 0.8

< 0.0001 Syk 1 ma x

P

∆(F 1 0.6 β-Actin 0 0 0.4 0 100 200 300 400 Fold change in degranulation Time (s) PBSa hIgEhIgE d Si As 0.2 Day 0: i.d. Day 1: i.v. challenge e sensitization Day 0: i.v. Day 1: i.v. Day 2: i.v. challenge 0 sensitization treatment 5 0 5 0. 1.0 PBS or OVA Mock –1. –1.0–0.5 Sia Sia DNP Measure Anti-OVA– mIgE (20 ng) Anti-DNP– mIgE + PBS –HSA temperature Log glycoprotein (µg) Anti-OVA–SiamIgE (20 ng) + anti-OVA–AsmIgE (200 ng) Anti-DNP–SiamIgE + SiamIgE isotype control Anti-OVA–SiamIgE (20 ng) + SiamIgE isotype control (200 ng) Anti-DNP–SiamIgE + AsmIgE isotype control

1.2 1 Sia + DNP-HSA

) L: mIgE

620 Sia As R: mIgE + mIgE 0 0.8 –1 P = 0.0478 P = 0.0321 e change (˚C) P = 0.0211 L: SiamIgE –2 SiamIgE 0.4 Sia R: mIgE + isotype control

mperatur –3 ascular leakage (OD Te V P = 0.0001 0 –4 0102030405060 PBS + ––– Time (min) SiamIgE – +++ As ––+ – gh i mIgE Anti-OVA–SiahIgE + NEUFcε Day 0: i.v. Day 1: i.v. Day 2: i.v. challenge Sia Fc Isotype ctrl – ––+ Anti-OVA–SiahIgE + Isotype control Peanut-allergic hIgE + NEU ε sensitization treatment Sia Sia Fc Peanut-allergic hIgE + Isotype control Anti-OVA– hIgE + H-I NEU ε Sia Measure Fcε Anti-OVA– mIgE + PBS OVA No IgE + NEUFcε No IgE + NEU temperature Anti-OVA–SiamIgE + NEUFcε f NEUFcε + OVA + Crude peanut extract Anti-OVA–SiamIgE + Isotype control 1.2 1.4 NEU + OVA 1.2 1.0 0 = 0.0184 = 0.0003P 1.0 0.8 P < 0.0001 P = 0.0008

P –1 C 2 0.8 ε P < 0.0001 0.6 0.6 e change (˚C) 0.4 –2 Cε3 0.4 0.2 Cε4 0.2 mperatur –3 Te Fold change in degranulation Fold change in degranulation 0 0 0 0102030405060 –4 –4 –3.0–2.5–2.0–1.5–1.0–0.5 –3.0–2.5–2.0–1.5–1.0–0.5 Time (min) Log (NEUFcε or Log (NEUFcε or mIgE isotype µg ml–1) mIgE isotype µg ml–1)

Fig. 3 | Modulating IgE sialylation and anaphylaxis. a, Immunoblots of intravenous. e, Temperature change following DNP-induced PSA in mice phosphorylated (p)Syk, total Syk and β-actin in LAD2 mast cells sensitized with receiving DNP-specific SiamIgE on day 0 and PBS, OVA-specific SiamIgE or AsmIgE PBS, OVA-specific SiahIgE or AshIgE after OVA stimulation. b, OVA-induced Ca2+ on day 1 (n = 6, 7 and 7, respectively, from two independent experiments). flux traces (left) and maximum values (right) in LAD2 cells sensitized with PBS f, Illustration of NEUFcε. g, OVA-induced degranulation in LAD2 cells sensitized (black), OVA-specific SiahIgE (maroon) or AshIgE (gold). n = 5 biologically with OVA-specific SiahIgE or PBS and treated with NEUFcε, heat-inactivated independent samples from three independent experiments. F, fluorescence ; NEUFcε (H-I NEUFcε) or IgE isotype control. h, Peanut-induced degranulation in Sia Fcε F0, fluorescence at time 0; Fmax, maximum fluorescence. c, OVA-elicited LAD2 cells sensitized with peanut-allergic hIgE or PBS and treated with NEU degranulation in LAD2 cells sensitized with OVA-specific SiahIgE and treated or IgE isotype control. g, h, n = 3 technical replicates. i, Temperature changes with SiaFetuin (maroon) or AsFetuin (gold). n = 3 technical replicates. following OVA-induced PSA in mice receiving OVA-specific SiamIgE on day 0 and d, Quantification of ear blue colouration (bottom left) and representative PBS, NEUFcε or IgE isotype control on day 1. n = 4 mice per group. Data are images (bottom right) following OVA-induced PCA in mice sensitized with PBS, means ± s.e.m. (b–e, g–i) and are representative of two (a, d) and three (c, g–i) OVA-specific SiamIgE, OVA-specific SiamIgE plus OVA-specific AsmIgE, or independent experiments. Two-tailed paired t-test (b), one-way ANOVA with OVA-specific SiamIgE plus SiamIgE isotype control (n = 2, 6, 3 and 3 mouse ears, Tukey’s (d), or two-way ANOVA with Sidak’s (c) or Tukey’s multiple comparison respectively). Top, the experimental protocol; i.d., intradermal; i.v, test (e, g–i). For gel source data, see Supplementary Fig. 1.

results15,16,18. N394-linked oligomannose structures (Fig. 1j and Extended nephropathy and influenza neutralization23,24, and IgM-induced Data Fig. 4b) and the N140-, N168-, N265- and N371-linked fucose and inhibitory signalling on B and T cells25,26. We found that sialic acid is biGlcNAc contents (Fig. 1j and Extended Data Fig. 4c, d) were similar attached in α2,6 linkages on hIgE and mouse IgE (mIgE), as determined between samples. However, N140- and N265-linked complex glycans by neuraminidase (NEU) digestion assays and lectin blotting (Fig. 2a terminating in galactose were enriched on non-atopic IgE, while termi- and Extended Data Fig. 5a–c), consistent with previous studies6,15,17. nal sialic acids, particularly disialylated glycans, were greatly enriched Thus, we treated mIgE with NEU and digestion buffer or with digestion at N168 and N265 on allergic IgE (Fig. 1j and Extended Data Figs. 1h, 4e, buffer alone to generate mIgEs of identical allergen specificity that f). Together, these results reveal that specific glycosylation patterns differed only in sialic acid content (Fig. 2a). In a model of passive cuta- distinguish allergic from non-atopic total IgE. neous anaphylaxis (PCA), we sensitized mice with phosphate-buffered Sialylation has been implicated in regulating several antibody classes, saline (PBS), ovalbumin (OVA)-specific sialylated mIgE (SiamIgE), or including the anti-inflammatory activity of IgG1 (ref.22), IgA-induced OVA-specific asialylated IgE (AsmIgE) intradermally in the ears. The

268 | Nature | Vol 582 | 11 June 2020 next day, we challenged the mice with allergen (OVA) in Evan’s blue stimulated with allergen, and cellular lysates were collected at defined dye intravenously. Forty minutes after challenge, we quantified the intervals. Western blotting of mast-cell lysates for the tyrosine kinase amount of blue dye in the ear as a surrogate of histamine-mediated Syk revealed reduced phosphorylation in cells sensitized with AshIgE vascular leakage. PBS injection elicited little blue dye accumulation in at 5 min and 30 min after stimulation (Fig. 3a). Similarly, calcium flux the ear injection site, while substantial blue colouration was observed was reduced in AshIgE-sensitized LAD2 mast cells following allergen in SiamIgE-sensitized ears (Fig. 2b and Extended Data Fig. 5d). Notably, stimulation compared with SiahIgE-sensitized cells (Fig. 3b). We then AsmIgE-sensitized ears exhibited markedly reduced blue colouration, asked whether a surrogate asialylated glycoprotein could attenuate indicative of attenuated anaphylaxis (Fig. 2b and Extended Data Fig. 5d). anaphylaxis similarly to asialylated IgE. LAD2 mast cells were sensitized To confirm that removal of sialic acid was responsible for reduced with SiahIgE, and supplemented with either sialylated fetuin (SiaFetuin) PCA reactivity, we reattached sialic acid to AsmIgE in vitro (Re-siamIgE). or asialylated fetuin (AsFetuin) (Extended Data Fig. 5b). Quantifying Re-siamIgE triggered a robust PCA reaction (Fig. 2b), demonstrating that allergen-specific degranulation revealed that addition of sialylated IgE sialylation affects the magnitude of anaphylaxis. Flow cytometry fetuin had no effect, while asialylated fetuin inhibited allergen-induced analysis of mast cells recovered from the mouse ears revealed no dif- mast-cell degranulation (Fig. 3c). Together, these results suggest that ferences in IgE loading following sensitization with SiamIgE or AsmIgE removal of sialic acid exposes an inhibitory glycan that dampens FcεRI (Fig. 2c and Extended Data Fig. 5e), and that SiamIgE and AsmIgE bound signalling. allergen similarly, as determined by enzyme-linked immunosorbent These observations indicated that AsIgE can actively inhibit ana- assay (ELISA) (Extended Data Fig. 5f). Thus, the attenuation of cutane- phylaxis in vivo. We therefore sensitized mice intradermally in the ous anaphylaxis by AsmIgE was independent of IgE loading on mast cells ears with PBS, OVA-specific SiamIgE, a combination of OVA-specific in vivo or allergen recognition. SiamIgE and tenfold more OVA-specific AsmIgE, or a combination of Next, we systemically sensitized mice with SiamIgE, AsmIgE or PBS OVA-specific SiamIgE and tenfold more trinitrophenyl (TNP)-specific and challenged them with allergen the following day in a model of SiamIgE isotype control. The next day we challenged mice with OVA, passive systemic anaphylaxis (PSA). SiamIgE-sensitized mice elicited and quantified blue colouration of the ears. Extensive vascular a robust anaphylactic response underscored by a marked tempera- leakage occurred in ears sensitized with OVA-specific SiamIgE alone ture loss 20 min after allergen challenge (Fig. 2d and Extended Data (Fig. 3d). However, co-sensitization of OVA-specific SiamIgE with either Fig. 5g, h). However, minimal temperature drop was observed in OVA-specific AsmIgE or TNP-specific SiamIgE both resulted in greatly AsmIgE- or PBS-sensitized mice (Fig. 2d and Extended Data Fig. 5g, h). reduced vascular leakage (Fig. 3d). Next, we systematically sensitized Consistent with this, we detected a systemic increase in histamine in mice with 2,4-dinitrophenyl (DNP)-specific SiamIgE on day 0, and then SiamIgE-sensitized animals following challenge, but not in AsmIgE- or PBS, OVA-specific SiamIgE or OVA-specific AsmIgE on day 1, finally chal- PBS-treated mice (Fig. 2e). Asialylated glycoproteins have a decreased lenging them with DNP conjugated to human serum albumin (HSA) on serum half-life27, and we therefore compared the levels of SiamIgE and day 2. Mice that were sensitized with DNP-specific SiamIgE on day 0 and AsmIgE in circulation following systemic administration. However, sialic PBS or OVA-specific SiamIgE on day 1 exhibited a robust temperature acid removal had little effect on IgE half-life (Fig. 2f and Extended Data loss after allergen challenge. However, DNP-specific SiamIgE-sensitized Fig. 5i). To extend these findings to a model of passive food allergy, we mice that received OVA-specific AsmIgE on day 1 had a greatly attenuated sensitized mice systemically with PBS, SiamIgE or AsmIgE, and challenged temperature loss upon allergen challenge (Fig. 3e). Systemic challenge them with allergen orally the following day. Sensitization to SiamIgE, but of these treatment groups with OVA revealed that only sensitization not to AsmIgE or PBS, resulted in a marked temperature loss following with OVA-specific SiamIgE resulted in a temperature drop, while all other oral allergen challenge (Fig. 2g). groups were unaffected (Extended Data Fig. 7a). These results suggest We next asked whether sialylation similarly regulated hIgE, and sen- that AsmIgE attenuates anaphylaxis by occupying FcεRI, but can actively sitized human LAD2 mast cells with PBS or with sialylated or asialylated dampen systemic anaphylaxis. hIgE (SiahIgE or AshIgE, respectively; Fig. 2h). The cells were stimulated with As removal of sialic acid attenuates IgE effector functions, we allergen, and degranulation was quantified in β-hexosaminidase-release explored whether targeting sialic acid on IgE-bearing cells represents a assays. AshIgE-sensitized cells showed markedly reduced degranula- viable strategy for attenuating allergic inflammation. Thus, we geneti- tion following allergen challenge, compared with SiahIgE-sensitized cally fused a neuraminidase to the N terminus of IgE Fc Cε2–4 domains cells (Fig. 2i). We examined LAD2 mast cells after sensitization by (NEUFcε; Fig. 3f and Extended Data Fig. 7b) to direct sialic acid removal flow cytometry, finding comparable hIgE loading following SiahIgE specifically to IgE-bearing cells. This fusion protein retained binding to or AshIgE sensitization (Extended Data Fig. 6a). Similar findings were FcεRI (Extended Data Fig. 7c), could be loaded on mast cells (Extended observed in human mast cells derived from primary peripheral blood Data Fig. 7d), and had neuraminidase activity (Extended Data Fig. 7e–h). CD34+ cell culture, where AshIgE-sensitized cells had markedly reduced We sensitized LAD2 mast cells with OVA-specific SiahIgE, incubated allergen-specific degranulation compared with SiahIgE-sensitized cells them briefly with increasing concentrations of NEUFcε, heat-inactivated (Fig. 2j and Extended Data Fig. 6b). In parallel, we sensitized primary NEUFcε, or an IgE isotype to control for FcεRI occupancy, and stimulated basophils with PBS, SiahIgE and AshIgE and stimulated them with allergen them with OVA. Treatment with NEUFcε, but not with heat-inactivated (Extended Data Fig. 6c). AshIgE-sensitized basophils elicited reduced NEUFcε or the isotype control, attenuated OVA-induced degranula- degranulation after allergen stimulation—as measured by surface stain- tion in a dose-dependent manner (Fig. 3g). To extend our findings to ing of the granule marker CD63—compared with basophils sensitized allergic hIgE from patients with a peanut allergy, we sensitized LAD2 with SiahIgE (Fig. 2k). Although mast-cell loading was similar between mast cells with peanut-allergic SiahIgE and treated them with NEUFcε or mouse and human SiaIgE and AsIgE (Fig. 2c and Extended Data Fig. 6a), with an IgE isotype control. Consistently, allergen-induced degranula- we investigated whether sialylation altered the kinetics of hIgE binding tion was markedly attenuated by NEUFcε treatment of peanut-allergic to its receptor, FcεRI. Biolayer interferometry (BLI) assays revealed no SiahIgE-sensitized cells compared with IgE isotype control treatment difference in SiahIgE and AshIgE interactions with FcεRI (Fig. 2l). Sialylation (Fig. 3h). Unsensitized LAD2 mast cells treated with NEUFcε did not also did not alter IgE binding to the allergen (Fig. 2m). Thus, removing degranulate (‘No IgE plus NEUFcε’; Fig. 3g, h), indicating that NEUFcε sialic acid from IgE attenuates its effector functions in vivo and in vitro, treatment does not stimulate mast cells. We next explored the thera- while binding to allergen, mast cells and FcεRI remains intact. peutic potential of modulating sialic acid content in vivo. Mice were Because sialylation does not alter IgE interactions with allergen sensitized systemically with SiamIgE on day 0, and treated with PBS, and receptor, we examined whether signalling downstream of FcεRI NEUFcε or IgE isotype control on day 1. The following day, the mice were is affected. LAD2 mast cells sensitized with SiahIgE or AshIgE were challenged systemically with allergen, and core body temperature

Nature | Vol 582 | 11 June 2020 | 269 Article was measured. SiamIgE-sensitized mice that received PBS or isotype 7. Bégin, P. et al. Phase 1 results of safety and tolerability in a rush oral immunotherapy Fcε protocol to multiple foods using omalizumab. Allergy Asthma Clin. Immunol. 10, 7 (2014). control exhibited robust drops in temperature (Fig. 3i). NEU treat- 8. Patil, S. U. et al. Peanut oral immunotherapy transiently expands circulating Ara h ment markedly attenuated the allergen-induced temperature drop 2-specific B cells with a homologous repertoire in unrelated subjects. J. Allergy Clin. (Fig. 3i), providing evidence of the therapeutic potential of targeting Immunol. 136, 125–134 (2015). 9. Wang, T. T. et al. IgG antibodies to dengue enhanced for FcγRIIIA binding determine sialic acid on IgE-bearing cells. disease severity. Science 355, 395–398 (2017). IgE-mediated allergic diseases are multifactorial, with a broad range 10. Lu, L. L. et al. A functional role for antibodies in tuberculosis. Cell 167, 433–443 (2016). of clinical presentations, and paradoxically many individuals produce 11. Wang, T. T. et al. Anti-HA glycoforms drive B cell affinity selection and determine influenza vaccine efficacy. Cell 162, 160–169 (2015). allergen-specific IgE without manifestation of disease. Further, there 12. Parekh, R. B. et al. Association of rheumatoid arthritis and primary osteoarthritis with 3,5,8,28 is a high rate of false-positive test results for food allergens . Many changes in the glycosylation pattern of total serum IgG. Nature 316, 452–457 (1985). non-mutually exclusive mechanisms for this discrepancy exist, includ- 13. Espy, C. et al. Sialylation levels of anti-proteinase 3 antibodies are associated with the activity of granulomatosis with polyangiitis (Wegener’s). Arthritis Rheum. 63, 2105–2115 ing differences in IgE affinity or epitope diversity for allergens, mast-cell (2011). numbers, FcεRI expression levels, Syk signalling, allergen-specific IgG 14. Wang, T. T. & Ravetch, J. V. Functional diversification of IgGs through Fc glycosylation. antibodies, anti-IgE antibodies, and numbers of regulatory T cells29. J. Clin. Invest. 129, 3492–3498 (2019). 15. Arnold, J. N. et al. The glycosylation of human serum IgD and IgE and the accessibility of Here we have shown that the sialic acid content on total IgE distinguishes identified oligomannose structures for interaction with mannan-binding lectin. peanut-allergic and non-atopic IgE. Further, we have found that allergic J. Immunol. 173, 6831–6840 (2004). reactions are attenuated by removing sialic acid from IgE or administer- 16. Plomp, R. et al. Site-specific N-glycosylation analysis of human immunoglobulin E. J. Proteome Res. 13, 536–546 (2014). ing asialylated glycoproteins. Although the sialic acid content of IgE and 17. Shade, K. T. et al. A single glycan on IgE is indispensable for initiation of anaphylaxis. its role in other contexts is unknown, we propose that sialylation is an J. Exp. Med. 212, 457–467 (2015). additional factor that regulates its biology. Thus, exploitation of the IgE– 18. Wu, G. et al. Glycoproteomic studies of IgE from a novel hyper IgE syndrome linked to PGM3 mutation. Glycoconj. J. 33, 447–456 (2016). sialylation axis presents a compelling diagnostic and therapeutic strategy. 19. Yamazaki, T. et al. Receptor-destroying enzyme (RDE) from Vibrio cholerae modulates IgE activity and reduces the initiation of anaphylaxis. J. Biol. Chem. 294, 6659–6669 (2019). 20. Jabs, F. et al. Trapping IgE in a closed conformation by mimicking CD23 binding prevents Online content and disrupts FcεRI interaction. Nat. Commun. 9, 7 (2018). 21. Wuhrer, M. et al. Glycosylation profiling of immunoglobulin G (IgG) subclasses from Any methods, additional references, Nature Research reporting sum- human serum. Proteomics 7, 4070–4081 (2007). maries, source data, extended data, supplementary information, 22. Nimmerjahn, F. & Ravetch, J. V. Anti-inflammatory actions of intravenous immunoglobulin. Annu. Rev. Immunol. 26, 513–533 (2008). acknowledgements, peer review information; details of author con- 23. Ding, J. X. et al. Aberrant sialylation of serum IgA1 was associated with prognosis of tributions and competing interests; and statements of data and code patients with IgA nephropathy. Clin. Immunol. 125, 268–274 (2007). availability are available at https://doi.org/10.1038/s41586-020-2311-z. 24. Maurer, M. A. et al. Glycosylation of human IgA directly inhibits influenza A and other sialic-acid-binding viruses. Cell Rep. 23, 90–99 (2018). 25. Colucci, M. et al. Sialylation of N-linked glycans influences the immunomodulatory effects of IgM on T cells. J. Immunol. 194, 151–157 (2015). 1. Gupta, R. S. et al. Prevalence and severity of food allergies among US adults. JAMA Netw. 26. Collins, B. E., Smith, B. A., Bengtson, P. & Paulson, J. C. Ablation of CD22 in ligand-deficient Open 2, e185630 (2019). mice restores B cell receptor signaling. Nat. Immunol. 7, 199–206 (2006). 2. Gould, H. J. & Sutton, B. J. IgE in allergy and asthma today. Nat. Rev. Immunol. 8, 205–217 27. Abe, K., Yamamoto, K. & Sinohara, H. Clearance of desialylated mouse (2008). alpha-macroglobulin and murinoglobulin in the mouse. J. Biochem. 108, 726–729 (1990). 3. Bird, J. A., Crain, M. & Varshney, P. Food allergen panel testing often results in 28. Fleischer, D. M. et al. Oral food challenges in children with a diagnosis of food allergy. misdiagnosis of food allergy. J. Pediatr. 166, 97–100 (2015). J. Pediatr. 158, 578–583 (2011). 4. Du Toit, G. et al. Randomized trial of peanut consumption in infants at risk for peanut 29. Sampath, V. et al. Deciphering the black box of food allergy mechanisms. Ann. Allergy allergy. N. Engl. J. Med. 372, 803–813 (2015). Asthma Immunol. 118, 21–27 (2017). 5. Peters, R. L., Gurrin, L. C. & Allen, K. J. The predictive value of skin prick testing for challenge-proven food allergy: a systematic review. Pediatr. Allergy Immunol. 23, Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 347–352 (2012). published maps and institutional affiliations. 6. Arnold, J. N., Wormald, M. R., Sim, R. B., Rudd, P. M. & Dwek, R. A. The impact of glycosylation on the biological function and structure of human immunoglobulins. Annu. Rev. Immunol. 25, 21–50 (2007). © The Author(s), under exclusive licence to Springer Nature Limited 2020

270 | Nature | Vol 582 | 11 June 2020 Methods A linear gradient from 1% to 35% mobile phase B was run over 75 min. Mass spectra were recorded on a Thermo Q Exactive mass spectrom- IgE antibodies eter operated in positive mode using data-independent acquisition All human samples were collected under Institutional Review Board (DIA), targeting the masses shown in Extended Data Table 2. Glyco- (IRB)-approved protocols by Massachusetts General Hospital (MGH) peptides were quantified on the basis of the extracted ion area of the and Research Blood Components, including informed consent obtained Y1 ion (Extended Data Fig. 2). Relative abundances were calculated for in accordance with relevant ethical regulations. Serum samples were all identified glycan species for each site. Myeloma IgE (Sigma Aldrich obtained from individuals with a peanut allergy before treatment. catalogue number AG30P) was run before paired sample sets to moni- Peanut allergy was confirmed by clinical history, allergen-specific tor retention-time shifts and to ensure consistency in the analytical IgE screening and double-blind placebo-controlled oral challenge results across the sample set. The number of sugar residues per site (trial number NCT01750879/PNOIT2)8 (Extended Data Table 1). per IgE molecule was calculated using the relative abundance of each Non-atopic adults were recruited on the basis of self-identification glycan. For example, if we determined a particular site to have 60% as non-allergic donors. Non-atopy was confirmed by clinical history monosialylated and fucosylated glycans (A1F) and 40% of disialylated and allergen-specific IgE screening (Extended Data Table 1). Total and fucosylated glycans (A2F), then the number of sialic acids at one IgE, Ara h 2-specific IgE, Fel d 1-specific IgE, Der p 1-specific IgE, and site would be 1.4 ((0.6 × 1) + (0.4 × 2)), with a total of 2.8 sialic acids per Bet v 1-specific IgE were determined by ImmunoCap Assay (Phalleon, molecule (accounting for two sites). Thermo Scientific) according to the manufacturer’s protocols. Primary IgE was enriched from serum samples by serially depleting IgG using Generation of NEUFcε protein G agarose (GE Healthcare) followed by anti-IgE conjugated We designed the neuraminidase fusion protein by fusing a κ light chain N-hydroxysuccinimide (NHS) beads (GE Healthcare). IgE purity was secretion signal sequence to the sialidase gene from Arthrobacter urefa- confirmed by protein electrophoresis and Coomassie gel staining. ciens (EC 3.2.1.18, gene AU104)30. The stop codon of AU104 was omitted; Recombinant OVA-specific IgE was generated as described17. In brief, instead, a short flexible linker peptide (GGGGGG), the mouse IgE Cε2, complementary DNA sequences for generating OVA-specific heavy (ε) Cε3 and Cε4 domains, and a 6× histidine tag were inserted into the and light (κ) chains of mouse and human IgE17 were cloned into plasmid carboxy terminus of the sialidase. The gene was codon-optimized for pcDNA3.4 using restriction-enzyme sites XbaI and AgeI. To generate humans and synthesized by GenScript. The 288-kDa protein was then recombinant OVA-specific mouse or human IgE, plasmids containing produced by WuXi biologics. Sialidase activity of NEUFcε was determined OVA-specific heavy and light chains were transiently co-transfected by the level of p-nitrophenol released from 250 μM 2-O-(p-nitrophenyl)- at a 1:1 ratio using an Expi293 Expression System Kit (Life Technolo- α-d-N-acetylneuraminic acid (Sigma) in 100 mM sodium phosphate gies) according to the manufacturer’s protocol. Cells expressing IgE (pH 5.5) for 10 min at 37 °C. The reaction was terminated by adding were selected by adding 400 μg ml−1 of the antibiotic G418 to the cul- 0.5 M sodium carbonate and absorbance quantified at 405 nm. ture medium for two weeks, and maintained before expanding to a larger-scale production. OVA-specific IgE was purified from cell culture Mice supernatant using OVA-coupled agarose beads17. Five- to six-week-old female BALB/c mice were purchased from the Jackson Laboratory and used here. All mice were housed in ELISA specific-pathogen-free conditions according to the National Institutes Sandwich ELISA for quantifying mIgE and OVA-specific binding was of Health (NIH), and all animal experiments were conducted under conducted as described17. In brief, 96-well Nunc plates were coated protocols approved by the MGH Institutional Animal Care and Use with goat polyclonal anti-mouse IgE (Bethyl Laboratories) or OVA, Committee (IACUC), in compliance with appropriate ethical regula- and blocked with bovine serum albumin (BSA) in PBS (1% BSA for mIgE tions. For all experiments, the allocation of age- and sex-matched mice and 2% for OVA) before sample incubation. Samples were probed with to experimental groups was randomized, with four to five mice per goat polyclonal anti-mouse IgE conjugated to horseradish peroxidase group, each experiment being repeated three independent times. No (HRP; 2 ng ml−1; Bethyl Laboratories). The reactions were detected using statistical method was used to determine sample size. 3,3,5,5-tetramethylbenzidine (TMB; Thermo Fisher Scientific) and PCA was conducted as described17. In brief, monoclonal SiamIgE or stopped with 2 M sulfuric acid; absorbance was measured at 450 nm. AsmIgE specific for OVA or DNP (clone SPE-7, Sigma Aldrich) was injected intradermally into mouse ears. For experiments in which OVA-specific gMS and glycopeptide analysis AsmIgE was added to OVA-specific SiamIgE, an mIgE isotype control We quantified the site-specific glycosylation of IgE isolated from (clone MEA-36, Biolegend) was included. The next day, mice were non-allergic donors and from donors with a peanut allergy by using intravenously challenged with 125 μg OVA or DNP–HSA (both from nanoscale liquid chromatography with tandem mass spectrometry Sigma Aldrich) and 2% Evans blue dye in PBS. Forty-five minutes after (LC-MS/MS) following enzymatic digestion of the proteins as described challenge, the ears were excised and minced before incubation in previously, with minor modifications16–18 (Extended Data Table 2). Iso- N,N-dimethyl-formamide (EMD Millipore) at 55 °C for 3 h. The degree lated polyclonal primary hIgE and myeloma hIgE (Sigma Aldrich cata- of blue dye in the ears was quantified by absorbance at 595 nm. logue number AG30P) were prepared for proteolysis by denaturing the PSA was elicited as described, with minor modifications31,32. In brief, protein in 6 M guanidine HCl, followed by reduction with dithiothreitol, mice were injected intravenously with monoclonal mIgE specific for alkylation with iodoacetamide, and dialysis into 25 mM ammonium OVA or DNP (clone SPE-7, Sigma Aldrich) in PBS, and challenged the bicarbonate (pH 7.8). Proteolysis was carried out with either trypsin next day intravenously with PBS containing 1 mg OVA or DNP–HSA to quantify N218, N371 and N394, or chymotrypsin to quantify N140, (both from Sigma Aldrich). To examine the therapeutic potential of N168 and N265. For the tryptic digest, IgE was incubated with trypsin AsmIgE, on the first day we injected mice intravenously with 10 μg (Trypsin Gold Promega) at a 1:50 enzyme-to-substrate ratio overnight DNP-specific mIgE (clone SPE-7, Sigma Aldrich); on the second day at 37 °C. For the chymotryptic digest, IgE was incubated with chymo- we injected them intravenously with PBS, 20 μg OVA-specific SiamIgE trypsin (Sequencing Grade Promega) at a 1:100 enzyme-to-substrate or 20 μg OVA-specific AsmIgE; and on the third day we challenged them ratio for 4 h at 25 °C. Both enzymes were quenched with formic acid intravenously with 1 mg DNP–HSA or OVA (Sigma Aldrich). To test the added to 2% w/w. The separation was performed on a Thermo EasySpray therapeutic potential of NEUFcε, mice injected intravenously with 10 μg C18 nLC column (0.75 μm × 50 cm) using water and acetonitrile with OVA-specific mIgE on the first day were further injected intravenously 0.1% formic acid for mobile phase A and mobile phase B, respectively. with PBS, 100 μg NEUFcε or 100 μg mIgE isotype control (clone MEA-36, Article Biolegend) the next day and challenged intravenously with 1 mg OVA Degranulation assays were performed as described17. LAD2 or (Sigma-Aldrich) the third day. Core temperature was recorded at the peripheral blood derived human mast cells were sensitized over- baseline and every 10 min after allergen challenge in a blinded man- night with 1 μg ml−1 OVA-specific hIgE or 50 ng ml−1 peanut-allergic ner with a rectal microprobe thermometer (Physitemp). Histamine in hIgE. The following day, the cells were pelleted by centrifugation, the blood was quantified using a histamine enzyme immunoassay kit resuspended in HEPES buffer, plated in 96-well plates, and stimulated (SPI-Bio) according to the manufacturer’s protocol. In brief, histamine with allergen OVA or crude peanut extract at defined concentrations. in the blood was derivatized and incubated with plate precoated with Upon allergen challenge, mast cell degranulation was determined monoclonal anti-histamine antibodies and histamine–acetylcholinest- by the amount of substrate p-nitrophenyl N-acetyl-β-d-glucosamide erase tracer at 4 °C for 24 h. The plate was washed and developed with digested by β-hexosaminidase release from mast cell granules at Ellman’s reagent and absorbance measured at 405 nm. absorbance of 405 nm. To assess the effect of sialic acid removal on Passive food anaphylaxis was elicited by adapting the PSA proto- IgE-bound mast cells, we treated IgE-sensitized LAD2 cells with NEUFcε, col described above. In brief, mice injected intravenously with 20 μg heat-inactivated NEUFcε or mIgE isotype control (clone MEA-36, Bio- monoclonal mIgE specific for TNP (clone MEA-36, Biolegend) in PBS legend) for 20 min before allergen challenge. To inactivate NEUFcε, we on the first day were administered with 20 mg TNP–OVA in PBS (Bio- heated the enzyme at 95 °C for 10 min. To determine whether addi- search Technologies) by oral gavage the next day. Core temperature tion of a surrogate asialylated glycoprotein could recapitulate the was recorded at the baseline and every 10 min after the challenge using phenotype of sialic acid removal from IgE, LAD2 cells sensitized with a rectal microprobe thermometer (Physitemp) in a blinded manner. OVA-specific SiahIgE were incubated with sialylated fetuin (SiaFetuin) To determine in vivo half-lives of SiamIgE or AsmIgE, we injected mice or asialylated fetuin (AsFetuin) at defined amounts for 20 min before intraperitoneally with 30 μg DNP-specific SiamIgE or AsmIgE, and col- allergen challenge. lected blood at the indicated times after injection into a Microtainer blood-collection tube with clot activator/serum separator tube (SST) Preparation of crude peanut extract gel (BD Diagnostics). We quantified the level of mIgE by mIgE ELISA Unsalted dry-roasted peanuts (blanched jumbo runner cultivar, Plant- as above. ers) were ground to a smooth paste, then washed with 20 volumes of cold acetone, filtered using Whatman paper, and dried as described17. Basophil activation tests Protein was extracted by agitating the peanut flour overnight with Basophil activation was performed as described33. Buffy coats of PBS containing protease-inhibitor cocktail without EDTA (Roche). human blood from healthy, de-identified, consenting donors were The peanut protein extracts were collected as the supernatant after obtained from the MGH Blood Transfusion Service. Peripheral blood centrifugation at 24,000g for 30 min. mononuclear cells (PBMCs) were separated from buffy coats by density gradient centrifugation using Ficoll Paque Plus (GE Healthcare), and IgE glycosylation engineering resuspended in 0.5% BSA in RPMI 1640 medium (GE Healthcare). PBMCs To remove sialic acids on IgE, we digested IgE with glyko sialidase were incubated for 2 min with ice-cold lactic acid buffer (13.4 mM lac- A (recombinant from A. urefaciens expressed in Escherichia coli, tate, 140 mM NaCl, 5 mM KCl, pH 3.9) to remove endogenous human IgE Prozyme) at 37 °C for 72 h according to the manufacturer’s instruc- on the cell surface before neutralization with 12% Tris (pH 8). Cells were tions. To resialylate AsmIgE by in vitro sialylation, we incubated AsmIgE then washed and incubated 1 h at 37 °C with 1 μg OVA-specific SiahIgE with human α-2,6 sialyltransferase 1 (ST6GAL1, provided by H. Meade, or AshIgE per 1 × 106 cells in basophil activation buffer (0.5% BSA, 2 mM LFB-USA) at a ratio of 20 μg AsmIgE per μg of ST6GAL1 and 5 mM

CaCl2 and 2 mM MgCl2 in RPMI 1640 medium). Sensitized cells were cytidine-5′-monophospho-N-acetylneuraminic acid (CMP-Neu5Ac2, washed and resuspended in basophil activation buffer supplemented Nacalai USA) in the sialylation buffer (150 mM NaCl, 20 mM HEPES, with 10 ng ml−1 human interleukin-3 (PeproTech) before activation with pH 7.4) overnight at room temperature as described36. Following OVA for 30 min. Activation was stopped by addition of ice-cold 0.2 M reactions, OVA-specific SiaIgE or AsIgE was purified with OVA-coupled EDTA in FACS buffer. Cells were washed and resuspended in FACS buffer beads to remove glycosylation modifying enzymes as described17. All before antibody staining (Extended Data Table 3) for activation markers digestion or sialylation reactions were verified by lectin blotting or (LAMP-3 or CD63+) on basophils (CD123+ HLADR−). high-performance liquid chromatography (HPLC).

Culture and degranulation of human mast cells Protein gel stain and lectin blotting The human LAD2 mast-cell line was a gift from D. D. Metcalfe (National Equal amounts of SiaIgE or AsIgE were resolved on 3–8% Tris-acetate Institute of Allergy and Infectious Diseases (NIAID), NIH) and was main- protein gels (Life Technologies) in SDS–PAGE under nonreducing tained as described17,34. In brief, LAD2 cells were cultured in StemPro-34 conditions. For protein staining, gels were incubated in AcquaStain SFM medium (Life Technologies) supplemented with 2 mM l-glutamine, protein gel stain (Bulldog Bio) for 1 h at room temperature and 100 U ml−1 penicillin, 100 μg ml−1 streptomycin and 100 ng ml−1 recombi- destained in distilled water. For lectin blotting, the protocol was nant human stem cell factor (PeproTech). The cells were hemi-depleted conducted as described17. Briefly, after resolved proteins on the each week with fresh medium and maintained at 2 × 105 to 5 × 105 cells gel were transferred to Immobilon-PSQ polyvinylidene difluoride per millilitre at 37 °C and 5% CO2. membranes (Millipore Sigma), the membranes were blocked Primary human mast cells were generated as described35. In brief, we with 0.2% BSA in Tris-buffered saline (TBS) for 1 h at room tem- separated PBMCs from buffy coats as above, before isolating CD34+ perature, washed in TBS, and then incubated with biotinylated pluripotent haematopoietic cells using the EasySep human whole blood S. nigra (SNA) lectin (0.4 μg ml−1, Vector Laboratories) in TBS with CD34-positive selection kit II (Stemcell Technologies). We cultured 0.1 M Ca2+ and 0.1 M Mg2+ for 1 h at room temperature to deter- CD34+ cells in StemPro-34 SFM medium (Life Technologies) supple- mine the level of terminal α2,6-sialic acids on N-linked glycans of mented with 2 mM l-glutamine, 100 U ml−1 penicillin, 100 μg ml−1 strep- proteins. The membrane was then washed in TBS and incubated tomycin, 50 ng ml−1 recombinant human stem cell factor (PeproTech) with alkaline-phosphatase-conjugated goat anti-biotin (1:5,000 and 50 ng ml−1 human interleukin-6 (PeproTech), and with 10 ng ml−1 dilution; Vector Laboratories) in TBS for 1 h at room temperature. human interleukin-3 (PeproTech) in the first week. After the first week Sialylated proteins on membranes were visualized by incubation we matured the cells in similar culture medium but without human with one-step nitro-blue tetrazolium chloride (NBT)/5-bromo- interleukin-3 for ten weeks. Cultured mast cells were confirmed by 4-chloro-3′-indolyphosphate p-toluidine (BCIP) substrate solution FACS staining of CD45+, KIT+ and FcεRI+. (Thermo Scientific). anti-phospho-Syk (Tyr352) antibody (Cell Signaling Technology) in Flow cytometry 5% BSA in TBST overnight at 4 °C. The membrane was washed in TBST, Details of antibodies used for surface staining are listed in Extended incubated with anti-rabbit-HRP for 1 h at room temperature, and washed Data Table 3. To stain mouse cells, we incubated suspended cells with again in TBST; this was followed by chemiluminescent detection using anti-mouse CD16/CD32 (clone 2.4G2, BD Biosciences) before antibody Immobilon western chemiluminescent HRP substrate (Millipore staining. Cells were incubated in FACS buffer with desired staining Sigma). To detect total Syk on the membrane, after chemiluminescent antibodies for 20 min at 4 °C. Cells were then washed in FACS buffer detection using autoradiography film, the membrane was stripped by before being acquired by an LSRII flow cytometer (BD Biosciences) incubation in stripping buffer (2% SDS and 0.1 M β-mercaptoethanol or CytoFLEX (Beckman Coulter). Data were analysed using FlowJo in Tris buffer) at 50 °C for 30 min. The stripped membranes were then software version 10.4 software (Tree Star). To quantify OVA-specific blocked, washed as above, incubated with 1:2,000 rabbit anti-Syk anti- hIgE loading following sensitization, we incubated PBS or 1 μg ml−1 body (Cell Signaling Technology) for 2 h in 5% BSA in TBST at room OVA-specific SiahIgE or AshIgE with 2.5 × 105 LAD2 cells per millilitre temperature, and washed again before being incubated with 1:30,000 overnight, before washing with FACS buffer and staining with anti-Kit anti-rabbit-HRP for 1 h at room temperature. To probe for β-actin, the antibody and OVA-A647. To quantify native hIgE loading on LAD2 mast membranes were incubated with 1:150,000 anti-β-actin HRP (Santa cells, we sensitized cells with 32 ng total non-atopic or allergic hIgE Cruz Biotechnology) for 1 h at room temperature and then washed; overnight before washing in FACS buffer and staining with anti-Kit signal was determined by chemiluminescent detection. and anti-hIgE antibodies. To quantify dermal mast cell IgE loading, we generated single-cell suspensions from mouse ears as described17. Calcium flux Ears were intradermally injected with 40 ng OVA-specific SiamIgE or We sensitized 5 × 105 LAD2 cells overnight with PBS or 500 ng ml−1 AsmIgE. The following day, ears were removed, separated into dorsal OVA-specific SiahIgE or AshIgE. The next day, sensitized cells were washed and ventral halves, and minced before incubation in Dulbecco’s modi- before loading with 2 μM Fluo-4-AM (Invitrogen) at 37 °C in HEPES fied Eagle medium (DMEM) containing 2% fetal calf serum (FCS), 1% buffer for 20 min. After loading, the cells were washed and resuspended HEPES, 500 units ml−1 collagenase type 4 (Worthington), 0.5 mg ml−1 in HEPES buffer. Fluorescence was filtered through the 530/30 bandpass hyaluronidase (Sigma) and DNase I (Roche) at 37 °C for 1 h at 180 rpm. filter and collected in FL-1/FITC. Baseline Ca2+ fluorescence levels were The digested sample was then subjected to disruption by gentleMACS recorded for 1 min on the Accuri C6 (BD Biosciences) before adding and filtered through a 70-μm cell strainer followed by a 40-μm cell the indicated allergen or buffer to each sample. At the end of allergen strainer in FACS buffer (2 mM EDTA and 0.5% BSA in PBS). mIgE loading stimulation, 2 μM Ca2+ ionophore A23187 (Sigma) was added to cells was detected by FACS using anti-mIgE antibodies on dermal mast cells as a positive control. (CD45+ CD11b− CD11c− Gr1− Kit+). Statistical analyses Biolayer interferometric assays for binding Results are shown as means ± s.e.m., except in the case of gMS-quantified We studied binding kinetics and affinity of protein interaction using glycan residues per IgE molecule (Fig. 1d–h), where results are pre- the Octet K2 system (Molecular Devices) with Octet buffer (PBS sented as medians and interquartile ranges. The number of mice used with 0.025% Tween and 1% BSA). To measure hFcεRIα interactions, in each experiment is indicated in the figure legends. The investiga- 0.25 μg ml−1 histidine-tagged hFcεRIα (Acro Biosystems) was loaded onto tors were not blinded to allocation during experiments and outcome anti-penta-histidine (HIS1K) biosensors (Molecular Devices). To analyse assessment, except during the measurement of rectal temperature in OVA interactions, we immobilized OVA (100 μg ml−1) onto amine-reactive mice upon antigen challenge in passive systemic and food anaphylaxis. second-generation (AR2G) biosensors in 10 mM sodium acetate, pH 5, Visual examination of the data distribution as well as normality testing using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride demonstrated that all variables appeared to be normally distributed. (EDC)/sulfo-NHS based chemistry. We determined association of analyte Statistical analyses were performed using Prism 8 (GraphPad software) OVA-specific SiahIgE or AshIgE in threefold serial dilution from 90 nM to with unpaired and paired Student’s t-test for assessing two unmatched 1 nM, and of NEUFcε in threefold serial dilution from 24 nM to 0.3 nM in and matched groups, respectively; two-way ANOVA with Sidak’s multi- Octet buffer. Analyte dissociation was measured in Octet buffer. Analysis ple comparison test for comparing two groups of multiple conditions; of binding kinetic parameters was performed by Octet data analysis and one-way or two-way ANOVA with Tukey’s multiple comparison test software v.10.0, using interaction of ligand-loaded biosensor with no for three or more groups. The P values noted throughout highlight bio- analyte during the association phase as the reference sensor. logically relevant comparisons. The accuracy of distinguishing allergic from non-atopic IgE by means of individual IgE glycan moieties was Immunoblotting for Syk signalling analysed by ROC curves using Prism 8 (GraphPad software). The area We sensitized 1.5 × 106 LAD2 cells with PBS or 1 μg ml−1 OVA-specific under each ROC curve (AUC) was calculated for each glycan moiety. SiahIgE or AshIgE. Sensitized cells were washed and resuspended in AUCs are interpreted as follows: the maximum AUC (1) indicates that a HEPES buffer the next day, and then stimulated with 10 μg ml−1 OVA particular glycan moiety can distinguish allergic IgE from non-atopic at 37 °C for the indicated times. Cells were immediately centrifuged IgE; an AUC of 0.5 indicates that the differentiation capacity of a specific after OVA stimulation and the cell pellets lysed in ice-cold lysis buffer glycan moiety is poor. (RIPA buffer (Boston BioProducts), 1× Halt protease-inhibitor cocktail (Thermo Scientific), 1× Halt phosphatase-inhibitor cocktail (Thermo Reporting summary Scientific) and 2.5 mM EDTA) for 30 min on ice. After incubation on Further information on research design is available in the Nature ice, lysed pellets were passed rapidly through a 27G needle on ice and Research Reporting Summary linked to this paper. centrifuged at 15,000 rpm at 4 °C for 15 min to clear the membrane and nuclei. The protein concentration was quantified using a Pierce BCA protein assay kit (Thermo Scientific) and 20 μg of protein lysate Data availability was loaded per well on 4–12% Bis-Tris protein gels (Life Technologies) Source Data are provided for Figs. 1–3 and Extended Data Figs. 1, 4–7. in SDS–PAGE under denaturing and reducing conditions. Briefly, after Full scans of all uncropped protein gel stains, lectin blots and west- protein transferred to PVDF membranes as above, the membranes ern blots with size marker indications presented here can be found in were blocked with 5% milk in TBS with 0.1% Tween (TBST) for 1 h at the Supplementary Information. All other data supporting the findings room temperature, washed in TBST, and incubated with 1:2,000 rabbit of this study are available from the corresponding author upon request.

Article

30. Christensen, S. & Egebjerg, J. Cloning, expression and characterization of a sialidase supported by grants from the NIH (DP2AR068272 and R01AI139669) and FARE to R.M.A.; NIH gene from Arthrobacter ureafaciens. Biotechnol. Appl. Biochem. 41, 225–231 (2005). T32 and K12 fellowships (T32AR007258 and K12HL141953) to K.C.S.; an NIH NIAID K23 grant 31. Dombrowicz, D. et al. Anaphylaxis mediated through a humanized high affinity IgE (K23AI121491) to S.U.P.; an NIH NIAID U19 grant (5U19 AI095261-03) to W.G.S.; and a Sanofi receptor. J. Immunol. 157, 1645–1651 (1996). iAward to M.E.C. 32. Dombrowicz, D., Flamand, V., Brigman, K. K., Koller, B. H. & Kinet, J. P. Abolition of anaphylaxis by targeted disruption of the high affinity immunoglobulin E receptor alpha Author contributions K.C.S. conducted mouse experiments, performed functional assays, chain gene. Cell 75, 969–976 (1993). developed methods, and generated resources, along with D.J.H.; M.K. cloned, generated and 33. MacGlashan, D. Jr, Lavens-Phillips, S. & Katsushi, M. IgE-mediated desensitization in characterized NEUFcε; M.E.C. purified samples and analysed data, along with E.L.; N.W. human basophils and mast cells. Front. Biosci. 3, d746–d756 (1998). performed mass spectrometry on purified samples; S.U.P. and W.G.S. acquired patient sample 34. Kirshenbaum, A. S. et al. Characterization of novel stem cell factor responsive human resources; S.U.P. and W.G.S. provided collected clinical samples; and R.M.A. conceived and mast cell lines LAD 1 and 2 established from a patient with mast cell sarcoma/leukemia; supervised the study, and wrote the manuscript along with K.C.S., M.E.C., S.U.P. and W.G.S. activation following aggregation of FcεRI or FcγRI. Leuk. Res. 27, 677–682 (2003). 35. Bandara, G., Metcalfe, D. D. & Kirshenbaum, A. S. Growth of human mast cells from bone Competing interests The authors declare no competing interests. marrow and peripheral blood-derived CD34+ pluripotent hematopoietic cells. Methods Mol. Biol. 1220, 155–162 (2015). Additional information 36. Anthony, R. M. et al. Recapitulation of IVIG anti-inflammatory activity with a recombinant Supplementary information is available for this paper at https://doi.org/10.1038/s41586-020- IgG Fc. Science 320, 373–376 (2008). 2311-z. Correspondence and requests for materials should be addressed to R.M.A. Acknowledgements We thank K. Jeffrey, F. Wermeling and A. Luster for constructive Peer review information Nature thanks Steve Galli, Kari Nadeau and Jim Paulson for their comments; T. Stanley and D. Hayden (Harvard Catalyst) for guidance in biostatistical analysis; contribution to the peer review of this work. N. Smith for technical assistance; and the patients participating in our studies. This work was Reprints and permissions information is available at http://www.nature.com/reprints. abc d

) P = 0.0259 P = 0.0086 Serum sample Non-atopic 50 PBS 1800 20 ) Allergic Non-atopic PBS Protein G Allergic 40 Non-atopic 15 (anti-IgG) beads Allergic 1200 30 P = 0.0077 300 10 Total IgG Flow through 20 hIgE MF I 200 600 oun t 5 ß-hex degranulation (% P = 0.0403 10 C 100 anti-IgE beads 0 0 0 llergen specific fraction of total IgE (% 0 1 2 3 4 5

A Ara h 2 Der p 1 Fel d 1 Bet v 1 10 10 10 10 10 1 1 0 (Peanut) (Dust mite) (Cat)(Birch pollen) 0. PBS Total IgE Flow 0.01 hIgE-APC through Allergic α-hIgE (µg/mL) Non-atopic

IgE by age 0-9 20-29 40-49 60-69 e f Female Male g ELISA: anti-hIgE (years) 10-19 30-39 50-59 SiahIgE Sialic acid AshIgE 1.5 10 Galactose 1.0

biGlcNAc 5 OD450 nm

0.5 per IgE molecule Total # glycan residues Fucose 0.0 0 0.5 1.0 1.5 2.0 2.5 3.0 0510 15 Fucose biGlcNAc GalactoseSialic acid Total # glycan residues per IgE molecule Glycan residues Log10 OVA-specific hIgE (ng/mL) h

Extended Data Fig. 1 | Characterization of non-atopic and allergic human APC, allophycocyanin. e, Anti-hIgE from c, d binds similarly to SiahIgE and IgE. a, Allergen-specific IgE levels for Ara h 2 (peanut; non-atopic, n = 11; AshIgE, as determined by hIgE ELISA assays. n = 2 technical replicates per group, allergic, n = 30), Der p 1 (dust mite; n = 5, 3), Fel d 1 (cat; n = 5, 3) and Bet v 1 (birch representative of three experiments. f, g, IgE glycan distribution by sex (f; n = 9 pollen; n = 4, 3). b, Strategy for enriching IgE from human sera. c, Quantified males; n = 12 females) and age (g; 0–9 years, n = 2; 10–19, n = 2; 20–29, n = 6; 30– degranulation of human LAD2 mast cells sensitized with PBS, non-atopic IgE or 39, n = 7; 40–49, n = 1; 50–59, n = 2; 60–69, n = 1). h, Representative structures of allergic IgE and stimulated with anti (α)-human IgE (PBS, n = 1; non-atopic, n = 4; complex N-linked glycans from Fig. 1j. Data shown are means ± s.e.m. allergic, n = 4). d, Quantified MFI (left) and representative histograms (right) of (a, c, d, f, g). P-values were determined by two-tailed unpaired t-test (d, f) or anti-hIgE FACS staining on human LAD2 mast cells sensitized with PBS, two-way ANOVA with Sidak’s multiple comparison test (a, c, g). n represents non-atopic hIgE or allergic hIgE (PBS, n = 3; non-atopic, n = 4; allergic, n = 3). biologically independent serum samples (a, c, d, f, g). Article

a

b

AllergicAllergic

Non-atopic

Allergic

Non-atopic

Allergic

Non-atopic

Allergic

Non-atopic

c

Allergic

Non-atopic

Allergic

Non-atopic

Allergic

Non-atopic

Allergic

Non-atopic

Extended Data Fig. 2 | N-linked glycans observed on native human IgE. biologically independent samples). b, Extracted ion chromatograms for IgE a, Representative MS/MS spectrum for N265-linked A2F glycopeptide, showing N265-linked sialylation variants from a patient with an allergy and a non-atopic the B ions (non-reducing end fragments) and Y ions (reducing end fragments donor. c, Extracted ion chromatograms for IgE N168-linked sialylation variants containing the intact peptide) resulting from glycosidic-bond cleavage, and from a patient with an allergy and a non-atopic donor. AV, average (referring to the b ions containing the peptide N terminus resulting from peptide-bond the number of spectra averaged); BP, base peak; MA, manual integration; RT, cleavage. The Y1 ion used for quantification of glycopeptides is circled (n = 18 retention time. a

b

Extended Data Fig. 3 | N-linked glycans observed on IgE myeloma standard. a, b, Extracted ion chromatograms for site-specific N-linked glycosylation from chymotryptic (a) or tryptic (b) digest of the IgE myeloma sample used as a standard. Article

a b 100 80 Non-atopic IgE Allergic IgE

80 60 d 60 40

% occupie 40

20 20 % of oligomannose glycans at N394 0 0 N140 N168 N218 N265 N371 N383 N394 Man-5Man-6 Man-7Man-8 Man-9 IgE glycosylation site c d

2.0 2.0 ( )

1.5 1.5 residues ( ) 1.0 1.0 per IgE molecule of fucose per IgE molecule 0.5 0.5 l# l# of biGlcNAc residues ta ta To To 0.0 0.0 N140 N168 N218 N265 N371 N140 N168 N218 N265 N371 IgE glycosylation site IgE glycosylation site

e P = 0.0344 1.5 P = 0.0413 f P = 0.0375 P = 0.0132

4

1.0 3

2 0.5

1 Total # of galactose ( ) residue s Total # of sialic acid ( ) residues 0.0 0 N140 N168 N218 N265 N371 N140 N168 N218 N265 N371 IgE glycosylation site IgE glycosylation site

Extended Data Fig. 4 | Site-specific characterization of resolved IgE glycans (n = 15, 13), N168 (n = 16, 17), N218 (n = 15, 19), N265 (n = 16, 20) and N371 (n = 16, from non-atopic individuals and individuals with allergies. a, Occupancy of 17). e, Number of galactose residues: N140 (n = 15, 14), N168 (n = 15, 17), N218 N-linked glycosylation sites: N140 (non-atopic, n = 15; allergic, n = 13), N168 (n = 15, 19), N265 (n = 12, 19) and N371 (n = 15, 17). f, Number of sialic acid (n = 16, 14), N218 (n = 15, 15), N265 (n = 12, 15), N371 (n = 15, 15), N383 (n = 16, 15) residues: N140 (n = 14, 13), N168 (n = 15, 13), N218 (n = 15, 17), N265 (n = 12, 19) and and N394 (n = 13, 16). b, Percentage of oligomannose moieties at N394 (n = 23, N371 (n = 15, 17). Data plotted are means ± s.e.m. P values were determined by 18). c, Number of fucose residues: N140 (n = 15, 13), N168 (n = 15, 17), N218 (n = 15, two-way ANOVA with Sidak’s multiple comparison test. n represents 19), N265 (n = 12, 18) and N371 (n = 15, 17). d, Number of biGlcNAc residues: N140 biologically independent serum samples (a–f). a b Allergic hIgE Fetuin α2-3 or α2-6

Utx Allergic IVIG hIgE Fetuin 185 Coo α2-3, 6, 8, 9 massie NEU-tx 115 50

185 α2-3 NEU-tx

SNA: Fluorescenc e α2,6- 115 50 sialic acid Retention Time (min) c MALI: 185 αOVA-mIgE Agalactosylated α2,3- Galactosylated sialic acid 115 50 Sialylated

Utx

α2-3, 6, 8, 9 NEU-tx Fluorescenc e

Retention Time (min) d αDNP-mIgE P = 0.0161 ID injection: AsmIgE (Left) vs SiamIgE (Right) 1.5

1.0 nm ) 595

D ID injection: PBS on both ears

(O 0.5 scular leakage Va 0.0 S

PB a mIgEmIgE Si As e CD45+CD11b- Mast cells (CD45+CD11b- Live cells CD45+ cells CD45+CD11b- cells CD11c-Gr1-cells CD11c-Gr1-c-Kit+) 250K 250K 5 5 5 10 10 10 40 200K 200K 4 4 4 10 10 10 30 150K 150K 3 3 3 20 100K 100K 10 10 10

C 50K C 50K 0 0 0 10 3 3 3

SS SS 0 0 CD45- APC/Cy7 -10 CD45- APC/Cy7 -10 CD45- APC/Cy7 -10 0 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 3 3 4 5 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 -10 0 10 10 10 Viability Dye CD45 CD11b CD11c/Gr1 c-Kit-PE mIgE-FITC -eFluor450 -APC/Cy7 -PerCP/Cy5.5 -APC

g h D0: i.v. D1: i.v. D0: i.p. D1: i.v. sensitization challenge sensitization challenge SiamIgE or AsmIgE OVA SiamIgE or AsmIgE OVA or PBS or PBS PBS PBS f OVA-specific AsmIgE OVA-specific AsmIgE i OVA-specific SiamIgE OVA-specific SiamIgE 1 1.2 SiamIgE 1 As 0 1 1.0 mIgE 0 -1 ) ˚C) ˚C) ( 0.8 ( nm

0 -1 -2

45 0.6 0.5

∆Temp -3 ∆Temp

OD -2 0.4 mIgE (ng/µl -4 P = 0.0018 0.2 -3 P = 0.0012 P < 0.0001 -5 P < 0.0001 0.0 -4 0 0123 0102030405060 0102030405060 0123 Log10 αOVA-mIgE (ng/mL) Time (min) Time (min) Time (h)

Extended Data Fig. 5 | See next page for caption. Article Extended Data Fig. 5 | Removal of sialic acid from IgE. a, Protein gel stain and mast cells. Shown are representative FACS plots used to identify mast cells in lectin blots of IVIG, native human IgE purified from patients with allergies, and mouse ears and to determine IgE levels on mouse ear mast cells. SSC, side fetuin. MALI, Maackia amurensis lectin I. b, HPLC glycan traces of undigested scatter. f, Binding of OVA-specific SiamIgE and AsmIgE to OVA, as determined by (Utx) allergic hIgE and fetuin; of allergic hIgE and fetuin digested with sialidase ELISA. n = 2 technical replicates per group, representative of three biologically from A. ureafaciens to release α2,3-, α2,6-, α2,8- and α2,9-linked sialic acids independent experiments. g, h, OVA-elicited systemic anaphylaxis as (α2-3, 6, 8, 9 NEU-tx); and allergic hIgE and fetuin digested with sialidase from measured by temperature drop in mice sensitized with PBS, OVA-specific Streptococcus pneumoniae to release α2,3-linked sialic acids (α2-3 NEU-tx). SiamIgE (g, n = 4; h, n = 6) or OVA-specific AsmIgE (g, n = 5; h, n = 6) by intravenous c, HPLC glycan traces of undigested or recombinant OVA-specific mIgE (g) or intraperitoneal (h) injection. i, Serum levels of DNP-specific SiamIgE digested with sialidase from A. ureafaciens. d, Left, quantification of vascular (n = 4) and AsmIgE (n = 3) in mice at defined times after systemic administration, leakage by Evan’s blue dye (n = 6 mouse ears per group) after PCA in mice determined by ELISA. Data are means ± s.e.m. and are representative of three sensitized with PBS or with SiamIgE or AsmIgE specific for DNP; right, experiments. P values determined by two-tailed paired t-test (d) or two-way representative ear images. e, Gating strategy for IgE loading on mouse skin ear ANOVA with Tukey’s multiple comparison test (g, h). a PBS 3000 P = 0.0134 SiahIgE AshIgE

2000 600

A-A647 MF I 400 Count

OV 1000 200

0 0 1 2 3 4 5 S 10 10 10 10 10 PB a hIgEhIgE Si As OVA-A647 b Unstained control Stained with indicated antibodies

Cells Count Count Count SSC FSC CD45-FITC c-Kit-APC FcɛRIɑ-PE

c Singlets CD123+HLADR- cells A FSC- CD123-PE CD123-PE

FSC-H HLADR-PE/Cy7 CD63-FITC

Extended Data Fig. 6 | FACS analysis of loading of human mast cells with experiments; one-way ANOVA with Tukey’s multiple comparison test. SiahIgE or AshIgE, phenotypic staining of PBMC-derived mast cells, and b, Representative phenotypic staining by FACS of primary human mast cells activation in primary basophils. a, MFI (left) and representative histogram from peripheral blood derived CD34+ pluripotent haematopoietic cells (n = 2 (right) of surface-bound hIgE on LAD2 mast cells following sensitization with technical replicates per group). c, Gating strategy for basophil activation assay, PBS, OVA-specific SiahIgE or OVA-specific AshIgE (n = 3 technical replicates per showing representative FACS plots used to determine basophil activation from group). Data are means ± s.e.m. and are representative of three independent PBMCs. Article

a D0: i.v. D1: i.v. D2: i.v. challenge sensitization treatment αDNP-SiamIgE (10 µg) +PBS OVA Measure αDNP-SiamIgE (10 µg) +αOVA-SiamIgE (20 µg) Temp αDNP-SiamIgE (10 µg) +αOVA-AsmIgE (20 µg)

+ OVA 0

-1 (˚C ) -2

∆Temp -3

-4 P < 0.0001 0102030405060 Time (min)

b CoomassieIB: αmIgE c d NEUFcε to hFcεRIα; Denatured Denatured -12 Fcε Fcε Fcε Fcε KD = < 1 x10 M

NEU NEU NEU NEU

0.7 I) 12 0.6 270 MF 10 270 0.5 5 205 205 0.4 8 150 150 0.3 6 inding (nm )

B 0.2 4 120 120 0.1 2 0.0 PE-mIgE (x1 0 85 85 0 0 200 400 600 800 10001200 1 1 65 0. 10 65 0.01 100 Time (s) 0.001 NEUFcε (µg/mL)

efCoomassie SNA: α2,6-sialic acid g ECL: β1,4-galactose h Fetuin mIgE FetuinmmIgE Fetuin IgE ) ε ε ε ε ε ε 1.2 Fcε Fc Fc Fc Fc Fc Fc NEU er U er U er U U U U er U U er U nm

NE NE NE 405 1.0 + Buff + NEU+ + Buff + NEU+ + Buff + NE + NE + Buffer+ NE + NE + Buff + NE + + Buff + NEU+ NE D 0.8 185 185 185 0.6 115 115 115 0.4 80 80 80 0.2 65 65 65

Enzymatic activity (O 0.0 50 50 50 0.11 10 100 NEUFcε (ng)

Extended Data Fig. 7 | PSA for IgE isotype controls and characterization of of sensitization at 37 °C. n = 3 technical replicates per group, representative of NEUFcε. a, Temperature change following OVA-induced PSA in mice receiving two independent experiments. e–h, Neuraminidase activity of NEUFcε DNP-specific SiamIgE on day 0, and then PBS, OVA-specific SiamIgE or determined by digestion of mIgE or fetuin overnight (e–g), and detection of OVA-specific AsmIgE isotype controls (from Fig. 3e) on day 1. n = 4 mice for all protein loading by Coomassie blue (e), terminal α2,6-sialic acid by SNA (f), and groups; two-way ANOVA with Tukey’s multiple comparison test. b, Protein gel terminal galactose by Erythrina cristagalli lectin (ECL) (g) or by the amount of stain (left) and immunoblot (IB) for mIgE (right) of native and denatured NEUFcε. substrate 2-O-(p-nitrophenyl)-α-d-N-acetylneuraminic acid digested by NEUFcε c, Binding kinetics of analyte NEUFcε to ligand hFcεRIα on biosensor. Kinetics of in a colorimetric assay (h; n = 3 technical replicates per group, representative of analytes of threefold serial dilution from 26.2 nM to 0.32 nM were analysed. two independent experiments). Data are means ± s.e.m. d, FACS analysis of surface-bound NEUFcε on LAD2 mast cells following 30 min Extended Data Table 1 | Patient demographic data Article Extended Data Table 2 | Targeted mass list for IgE glycopetides

The mass [m/z] listed identifies N140-, N168- and N265-linked glycopeptides from a chymotryptic digest and N218-, N371- and N394-linked glycopeptides from a tryptic digest. CS, charge state; (N)CE, normalized collision energy. Extended Data Table 3 | Pertinent information regarding commercial antibody reagents      

()00123)4564789@A)0B2CD H)F10@gXb4@A)4G    E82@9358@15FG89@A)0B2CD  `1Fnoanpnp  

H13)0@647I9PP80G    

Q8@901H12180RAS62A12@)6P30)T1@A10130)59R6F6U6@G)V@A1S)0W@A8@S139FU62AXYA62V)0P30)T65122@09R@901V)0R)4262@14RG845@084238014RG   

64013)0@647X`)0V90@A1064V)0P8@6)4)4Q8@901H12180RA3)U6R612a211b9@A)02cH1V10112845@A1d56@)068Ue)U6RG(A1RWU62@X     

I@8@62@6R2   `)08UU2@8@62@6R8U848UG212aR)4V60P@A8@@A1V)UU)S6476@1P2801301214@64@A1V67901U17145a@8FU1U17145aP864@1f@a)0g1@A)5221R@6)4X

4h8 ()4V60P15

q YA11f8R@28P3U126i1BpCV)018RA1f3106P14@8U70)93hR)456@6)4a76T14828562R01@149PF10845946@)VP182901P14@

q b2@8@1P14@)4SA1@A10P182901P14@2S101@8W14V0)P562@64R@28P3U12)0SA1@A10@A128P128P3U1S82P182901501318@15UG YA12@8@62@6R8U@12@B2C9215bQqSA1@A10@A1G801)41r)0@S)r26515 q sptuvwxyyxpv€‚€‚v‚ƒx„t v†v ‚w‡ˆ† v‚xttuv†uvp‰yv ‚w‡ˆ†vyx‡vwxy‘t’v€wƒpˆ“„‚vˆpv€ƒv”€ƒx ‚v‚w€ˆxp•

q b512R063@6)4)V8UUR)T8068@12@12@15

q b512R063@6)4)V84G8229P3@6)42)0R)001R@6)42a29RA82@12@2)V4)0P8U6@G84585–92@P14@V)0P9U@63U1R)P38062)42 bV9UU512R063@6)4)V@A12@8@62@6R8U3808P1@10264RU95647R14@08U@14514RGB1X7XP1842C)0)@A10F826R12@6P8@12B1X7X01701226)4R)1VV6R614@C q bQqT8068@6)4B1X7X2@84580551T68@6)4C)0822)R68@1512@6P8@12)V94R10@864@GB1X7XR)4V6514R164@10T8U2C

`)049UUAG3)@A1262@12@647a@A1@12@2@8@62@6RB1X7X—a€a‡CS6@AR)4V6514R164@10T8U2a1VV1R@26i12a5170112)VV0115)P845˜T8U914)@15 q ™ˆdv˜vd‰t„‚v‰‚v’‰w€vd‰t„‚veƒpd‡v‚„ˆ€‰†t•

q `)0f8G12684848UG262a64V)0P8@6)4)4@A1RA)6R1)V306)02845g80W)TRA864g)4@1(80U)21@@6472

q `)0A61080RA6R8U845R)P3U1f5126742a6514@6V6R8@6)4)V@A18330)3068@1U1T1UV)0@12@2845V9UU013)0@647)V)9@R)P12

q d2@6P8@12)V1VV1R@26i12B1X7X()A14g2 ae1802)4g2‡Ca6456R8@647A)S@A1GS101R8UR9U8@15

s„‡ve†vwxttw€ˆxpvxpv‚€‰€ˆ‚€ˆw‚vhx‡v†ˆxtxiˆ‚€‚vwxp€‰ˆp‚v‰‡€ˆwt‚vxpvy‰puvxhv€ƒv‘xˆp€‚v‰†xd• I)V@S801845R)51 e)U6RG64V)0P8@6)48F)9@8T86U8F6U6@G)VR)P39@10R)51 q8@8R)UU1R@6)4 `b(I28P3U12S1018Rm96015FG84EIHrrVU)SRG@)P1@10Bfqf6)2R614R12C)0(G@)`EdsBf1RWP84()9U@10CXYA18F2)0F84R18@top4PV)0 dErIbatpo4PV)0P82@R1UU5170849U8@6)4845268U658218R@6T6@G8228G2845ouo4PV)0FU915G1)VP)9211802S101P1829015FG g)U1R9U80q1T6R12I31R@08g8feU92vwtP6R0)3U8@1018510XI6@1r231R6V6R7UGR)2GU8@6)4S8251@10P6415FG484)E(rgIhgI926478YA10P) d82GI308G(xw4E(V)UU)S15FG8YA10P)ydf8R@6T1P822231R@0)P1@10XH1R@8U@1P3108@901S82P1829015FG801R@8UP6R0)30)F1 @A10P)P1@10BeAG26@1P3CXf64564764@108R@6)4F1@S114848UG@1845U67845S828Rm96015FGzR@1@IG2@1Pq8@8bRm9626@6)4I)V@S801X

q8@8848UG262 q8@8S82848UG215FGk083Ae85e062PT1026)4w845`U)S{)T1026)4xpX`)0F645647W641@6R3808P1@102a848UG212S101310V)0P15FG zR@1@58@8848UG2622)V@S801xpXpX `)0P8492R063@29@6U6i647R92@)P8U7)06@AP2)02)V@S801@A8@801R14@08U@)@A1012180RAF9@4)@G1@512R06F156439FU62A15U6@108@901a2)V@S801P92@F1P8518T86U8FU1@)156@)02h01T61S102X j12@0)47UG14R)90871R)51513)26@6)4648R)PP946@G013)26@)0GB1X7Xk6@l9FCXI11@A1Q8@901H12180RA79651U6412V)029FP6@@647R)51c2)V@S801V)0V90@A1064V)0P8@6)4X q8@8 e)U6RG64V)0P8@6)48F)9@8T86U8F6U6@G)V58@8 bUUP8492R063@2P92@64RU951858@88T86U8F6U6@G2@8@1P14@XYA622@8@1P14@2A)9U530)T651@A1V)UU)S64764V)0P8@6)4aSA101833U6R8FU1D 

rbRR1226)4R)512a946m916514@6V6102a)0S1FU64W2V)039FU6RUG8T86U8FU158@821@2  

rbU62@)VV679012@A8@A8T1822)R68@1508S58@8 ! "

rb512R063@6)4)V84G012@06R@6)42)458@88T86U8F6U6@G # $ % & I)90R158@8@8FU1280130)T6515V)0`672Xx|v845df@14515q8@8`672Xxat|}X`9UU2R842)V8UU94R0)331530)@16471U2@8642aU1R@64)0S12@104FU)@2S6@A26i1P80W10 ' 6456R8@6)4301214@1564@A62P8492R063@8012A)S464@A1I933U1P14@80Gr4V)0P8@6)4XbUU)@A1058@82933)0@647@A1V645647264@A622@95G8018T86U8FU1V0)P@A1 R)00123)4564789@A)093)401m912@X

   `61U5r231R6V6R013)0@647     eU182121U1R@@A1)41F1U)S@A8@62@A1F12@V6@V)0G)90012180RAXrVG)98014)@2901a0185@A18330)3068@121R@6)42F1V)01P8W647G)9021U1R@6)4X   

q E6V12R614R12 f1A8T6)908Uc2)R68U2R614R12 dR)U)76R8Ua1T)U9@6)480Gc14T60)4P14@8U2R614R12  

`)0801V1014R1R)3G)V@A15)R9P14@S6@A8UU21R@6)42a21148@901XR)Ph5)R9P14@2h40r013)0@647r29PP80GrVU8@X35V       

E6V12R614R122@95G512674    bUU2@95612P92@562RU)21)4@A1213)64@21T14SA14@A1562RU)2901624178@6T1X   I8P3U126i1 Q)2@8@62@6R8UP1@A)5S829215@)51@10P64128P3U126i1X`)0P906411f3106P14@2aS1714108UUG921troP6R13101f3106P14@8U70)93845 A8T1V)945@A8@310V)0P647@A1@G312)V1f3106P14@2512R06F1564@A622@95GS6@AtroP6R13101f3106P14@8U70)9362P84871P14@845 

30)T6510)F92@58@8S6@A2@8@62@6R8UUG26746V6R84@56VV1014R12F1@S114@A170)932X 

q8@81fRU926)42 Q)58@8S821fRU9515V)08UU1f3106P14@2X

H13U6R8@6)4 bUUV645647264@A622@95GS10101U68FUG0130)59R15826456R8@1564@A1V67901U171452X`)064T6@0)2@95612a@A128P3U12S10164593U6R8@1)0 @063U6R8@1845310V)0P158@U182@v6451314514@@6P12@)R)4V60P@A1V645647X`)0P906412@95612atroP6R13101f3106P14@8U70)9362714108UUG 921584501318@158@U182@nrv6451314514@@6P12XjA143)226FU1a@A1V645647S82T106V61592647r7d)V856VV1014@84@6714231R6V6R6@GX

H845)P6i8@6)4 `)08UUP906411f3106P14@2aP6R1S1010845)PUG8226741564@)8R)4@0)U)0@018@P14@70)93X

fU645647 YA1P182901P14@)V01R@8U@1P3108@90164P6R193)484@6714RA8UU14716438226T12G2@1P6R845V))58483AGU8f62S82310V)0P15648FU64515 P84410X

H13)0@647V)0231R6V6RP8@1068U2a2G2@1P2845P1@A)52 j101m960164V)0P8@6)4V0)P89@A)028F)9@2)P1@G312)VP8@1068U2a1f3106P14@8U2G2@1P2845P1@A)52921564P84G2@95612Xl101a6456R8@1SA1@A1018RAP8@1068Ua 2G2@1P)0P1@A)5U62@156201U1T84@@)G)902@95GXrVG)98014)@29016V8U62@6@1P833U612@)G)90012180RAa0185@A18330)3068@121R@6)4F1V)0121U1R@64780123)421X g8@1068U2c1f3106P14@8U2G2@1P2 g1@A)52 4h8 r4T)UT1564@A12@95G 4h8 r4T)UT1564@A12@95G q b4@6F)5612 q (Arer21m

q d9W80G)@6RR1UUU6412 q `U)SRG@)P1@0G

q e8U81)4@)U)7G q gHrrF82154190)6P87647

q b46P8U2845)@A10)078462P2

q l9P84012180RA380@6R6384@2

q (U646R8U58@8 b4@6F)5612 b4@6F)56129215 `)084@6F)56129215@)51@1R@r7dU)85647)4P)9212W64P82@R1UU2S1018UU390RA8215V0)Pf6)U17145D b4@6rP)921(qtoBRU)41Dvpr`xxaVU9)0)RA0)P1Dbe(h(G84641}a(8@Dxpvxx€aU)@49PF10Dfnuppn€a56U9@6)4DxDtppC b4@6rP)921(qxxFBRU)41Dgxh}paVU9)0)RA0)P1De10(eh(G84641oXoa(8@DxpxnnwaU)@49PF10Dfnwvptua56U9@6)4DxDxppC b4@6rP)921(qxxRBRU)41DQtxwaVU9)0)RA0)P1D`rY(a(8@Dxx}vp€aU)@49PF10Dfn}}pvpa56U9@6)4DxDxppC b4@6rP)921EGr€khEGr€(Bk0rxCBRU)41DHf€rw(oaVU9)0)RA0)P1D`rY(a(8@Dxpwtp€aU)@49PF10Dfnwxpx}a56U9@6)4DxDxppC b4@6rP)921(qxx}BRr6@CBRU)41DnfwaVU9)0)RA0)P1Dbe(a(8@DxpowxnaU)@49PF10Dfnx}wooa56U9@6)4DxDxppC b4@6rP)921r7dBRU)41DHgdrxaVU9)0)RA0)P1Deda(8@Dtp€upwaU)@49PF10Dfn€n€t}a56U9@6)4DxDxppC

`)084@6F)5612845018714@29215@)51@1R@r7dU)85647)4A9P84EbqnR1UU2D b4@6rA9P84(qxx}BRr6@CBRU)41DxptqnaVU9)0)RA0)P1Dedaf6)U17145a(8@DvxvnptaU)@49PF10Dfnxvnvoa56U9@6)4DxDtppC

b4@6rA9P84r7dBRU)41DgldrxwaVU9)0)RA0)P1Dbe(af6)U17145a(8@DvnoopwaU)@49PF10Dfxwuvtua56U9@6)4DxDtppC    zT8UF9P64rb€t}Br4T6@0)714a(8@Dzvt}wtaU)@49PF10Dxo}upvna56U9@6)4DxDtpp)VnP7hPE2@)RWC ! " # $ %

bUU84@6F)5612921564F82)3A6U8R@6T8@6)4@12@2S101390RA8215V0)Pf6)U17145D & ' b4@6rA9P84lEbrqHBRU)41DEntvaVU9)0)RA0)P1Dedh(G}a(8@Dvp}€x€aU)@49PF10Dfnxu€}ta56U9@6)4DxDvoC b4@6rA9P84(qxnvBRU)41D€l€aVU9)0)RA0)P1Deda(8@Dvp€pp€aU)@49PF10Dfntpuw}a56U9@6)4DxDvpC b4@6rA9P84(q€vBRU)41Dlo(€aVU9)0)RA0)P1D`rY(a(8@DvovppoaU)@49PF10Dfxuxpnva56U9@6)4DxDvpC

~ bUU84@6F)56129215V)03A14)@G36R2@864647)VA9P84P82@R1UU2S101390RA8215V0)Pf6)U17145D

b4@6rA9P84(qxx}BRr6@CBRU)41DxptqnaVU9)0)RA0)P1Dbe(a(8@Dvxvnp€aU)@49PF10Dfx}u}toa56U9@6)4DxDtppC  

b4@6rA9P84(qtoBRU)41DlrvpaVU9)0)RA0)P1D`rY(a(8@Dvptpp€aU)@49PF10Dfnvvxtpa56U9@6)4DxDxppC   

b4@6rA9P84`R Hr†BRU)41DbdHrv}B(HbrxCaVU9)0)RA0)P1Deda(8@Dvvt€puaU)@49PF10Dfxup€€€a56U9@6)4DxDtppC    

Y)51@1R@@A1R1UUT68F6U6@GD`6f8FU1ƒ68F6U6@GqG11`U9)0‡topB1f6)2R614R1a(8@D€orpw€vrxtaU)@49PF10Dnxtvtwwa56U9@6)4D  

xDoppC 



`)084@6F)56129215@)51@1R@r7d64dErIbD  

k)8@3)UGRU)48U84@6|P)921r7db4@6F)5GBf1@AGUE8F)08@)0612a(8@8U)7buprxxobaU)@49PF10Dbuprxxobrn€a56U9@6)4DxDnppC  

k)8@3)UGRU)48U84@6|P)921r7db4@6F)5GalHe()4–978@15Bf1@AGUE8F)08@)0612a(8@8U)7buprxxoeaU)@49PF10Dbuprxxoervna   

56U9@6)4DxDvpapppC  k)8@3)UGRU)48U84@6|A9P84r7db4@6F)5GBf1@AGUE8F)08@)0612a(8@8U)7bwprxpwbaU)@49PF10Dbwprxpwbrx€a56U9@6)4DxDnppC 

k)8@3)UGRU)48U84@6|A9P84r7db4@6F)5GalHe()4–978@15Bf1@AGUE8F)08@)0612a(8@8U)7bwprxpweaU)@49PF10Dbwprxpwernua   56U9@6)4DxDvpapppC  `)084@6F)56129215646PP94)FU)@@647D  g)92184@6rˆrbR@64b4@6F)5GB(tCalHe()4–978@15BI84@8(09if6)@1RA4)U)7Ga(8@8U)72Rrt}}}wlHeaU)@49PF10D{xuxna 56U9@6)4DxDopapppC b4@6rH8FF6@r7kBl‰ECalHe()4–978@15Be0)P178a(8@8U)7jtpxxaU)@49PF10Dpppppv}ovna56U9@6)4DxDvpapppC H8FF6@84@6reA)23A)rIGWBYG0vonCb4@6F)5GB(1UUI6748U647Y1RA4)U)7Gae0)59R@n}pxaU)@49PF10Dxpa56U9@6)4DxDnpppC H8FF6@84@6rIGWb4@6F)5GB(1UUI6748U647Y1RA4)U)7Gae0)59R@n}xnaU)@49PF10Dx€a56U9@6)4DxDnpppC

`)0Pr7d84@6F)561292156438226T18483AGU8f622@95612D g)4)RU)48UPr7d231R6V6RV)05646@0)3A14GUBqQeaRU)41Iedr}„I67P8rbU506RAa(8@Dqwtp€aU)@49PF10DpxwgtwpoƒC g)4)RU)48UPr7d231R6V6RV)0@0646@0)3A14GUBYQeaRU)41gdbrv€„f6)U17145a(8@Dtpx}pxaU)@49PF10Dfn€upv}C

ƒ8U658@6)4 bUU84@6F)5612S101S1UUr12@8FU62A15RU)412845R)PP10R68UUG8T86U8FU1V0)Pf6)U17145aI67P8rbU506RAa(1UUI6748U647Y1RA4)U)7Ga e0)P178aI84@8(09if6)@1RA4)U)7Gaf1@AGUE8F)08@)0612a1f6)2R614R1XŠ4U122)@A10S6214)@15a8UU84@6F)5612S10192158@ R)4R14@08@6)4201R)PP14515FG@A1P849V8R@9010V)0@A101U1T84@8228G2XYA1T8U658@6)458@8V)0@A184@6F)561284530)59R@ R6@8@6)4280130)T6515)4@A1P849V8R@9010‹2S1F26@1X d9W80G)@6RR1UUU6412 e)U6RG64V)0P8@6)48F)9@R1UUU6412 (1UUU6412)90R1B2C l9P84EbqnP82@R1UU2S101)F@86415V0)Pq0Xg1@R8UV1BQrbrqaQrlC@A8@714108@15@A1R1UUU641845df36nuv`(1UU2S101 390RA8215V0)PYA10P)`62A10IR614@6V6RX

b9@A14@6R8@6)4 (1UU2U6412921564@A622@95GS1014)@89@A14@6R8@15X

gGR)3U82P8R)4@8P648@6)4 (1UUU6412921564@A622@95GS1014)@@12@15V)0PGR)3U82P8R)4@8P648@6)4X

()PP)4UGP626514@6V615U6412 (1UUU6412921564@A622@95GS1014)@U62@1564YA1r(Eb(H1762@10)Vg626514@6V615(1UUE6412X BI11r(Eb(01762@10C b46P8U2845)@A10)078462P2 e)U6RG64V)0P8@6)48F)9@2@9561264T)UT647846P8U2„bHHrƒd79651U641201R)PP14515V)0013)0@647846P8U012180RA E8F)08@)0G846P8U2 `6T1r@)26frS11Wr)U5V1P8U1fbEfhRP6R1S101390RA8215V0)P@A1{8RW2)4E8F)08@)0G845921564@A12@95612X

j6U5846P8U2 YA12@95G5654)@64T)UT1S6U5846P8U2X

`61U5rR)UU1R@1528P3U12 YA12@95G5654)@64T)UT128P3U12R)UU1R@15V0)P@A1V61U5X

d@A6R2)T10267A@ bUUP6R1S101A)921564231R6V6R38@A)714rV011R)456@6)428RR)05647@)@A1Q8@6)48Ur42@6@9@12)Vl18U@ABQrlCa8458UU846P8U 1f3106P14@2S101R)459R@159451030)@)R)U28330)T15FG@A1gklr42@6@9@6)48Ub46P8U(801845Š21()PP6@@11X

Q)@1@A8@V9UU64V)0P8@6)4)4@A18330)T8U)V@A12@95G30)@)R)UP92@8U2)F130)T651564@A1P8492R063@X   

! " # $ l9P84012180RA380@6R6384@2 % & ' e)U6RG64V)0P8@6)48F)9@2@9561264T)UT647A9P84012180RA380@6R6384@2 e)39U8@6)4RA808R@1062@6R2 xwP8U12anxV1P8U12a8715xor€pB3U1821211df@14515q8@8Y8FU1xCX

‚ H1R096@P14@ e1849@8UU1076R29F–1R@2S10101R096@15V0)P@A1gklbUU107G(U646R2XYA101V1006473AG26R684214@3)@14@68U29F–1R@2)0@A160 

38014@2801R096@P14@U1@@10XYA1U1@@1064RU9515R)4@8R@64V)0P8@6)4V)02@95G2@8VV2)@A129F–1R@R)9U5)3@64@)380@6R638@6)4XrV  

4)R)4@8R@S8201R16T15a@A12@95G2@8VVR)4@8R@15@A129F–1R@a562R92215@A12@95Ga845R)459R@158@1U13A)412R0114X   e8@614@2S101R)4V60P15@)A8T131849@8UU107GFG@A1301214R1)V231R6V6Rr7d@)31849@2B2109P31849@r231R6V6Rr7d(ber`drb  

ŽxpWŠhEC845801R14@A62@)0GBS6@A64xnP)4@A2C)V)F–1R@6T1RU646R8U2GP3@)P2S6@A64€pP649@128V@10@A164712@6)4)V  

31849@2X   

YA138014@284538@614@S101159R8@158F)9@@A12@95G84564V)0P15R)4214@S82)F@86415648RR)0584R1S6@ArHf3)U6RGFG  01V1006478UU10762@2)0FG@A12@95G64T12@678@)0aS6@A8@U182@twA)902306)0@)140)UUP14@8450845)P6i8@6)4Xg64)02F1@S114  @A18712)V}845xvS10176T14@A1)9@Ab2214@`)0P@)2674„P64)02V0)P8712xt@)x}267415@A138014@8UR)4214@V)0PX  

I9F–1R@2140)UU1582P64)02S10130)T651584859U@R)4214@V)0PV)026748@901SA14@A1G018RA15871xw845S10101r  

R)4214@15XBdf@14515q8@8Y8FU1xCX     rHfr8330)T154)4r8@)36R859U@2S10101R096@15BH12180RAfU))5()P3)414@2af)2@)4agbC)4@A1F8262)V21UVr6514@6V6R8@6)482 

4)4r8UU1076R5)4)02XQ)4r8@)3GS82R)4V60P15FGRU646R8UA62@)0Ga8458UU10714r231R6V6Rr7d2R0114647X 

gklrHfl9P84I9F–1R@2b229084R1Q9PF10ppppvxv€„e1849@z08UrPP94)@A1083GDeQzrYnD(U646R8UY068U2rq49PF10 d@A6R2)T10267A@  Q(Ypx}opw}u  Q)@1@A8@V9UU64V)0P8@6)4)4@A18330)T8U)V@A12@95G30)@)R)UP92@8U2)F130)T651564@A1P8492R063@X

`U)S(G@)P1@0G eU)@2 ()4V60P@A8@D

q YA18f62U8F1U22@8@1@A1P80W10845VU9)0)RA0)P19215B1X7X(qtr`rY(CX

q YA18f622R8U12801RU180UGT626FU1Xr4RU95149PF1028U)478f12)4UGV)0F)@@)PU1V@3U)@)V70)93B8g70)93g6284848UG262)V6514@6R8UP80W102CX

q bUU3U)@2801R)4@)903U)@2S6@A)9@U6102)032195)R)U)03U)@2X

q b49P106R8UT8U91V)049PF10)VR1UU2)0310R14@871BS6@A2@8@62@6R2C6230)T6515X g1@A)5)U)7G

I8P3U13013808@6)4 Y)30138012647U1R1UU29231426)42V0)PP)9211802aP)9211802S10101P)T15a213808@1564@)5)028U845T14@08UA8UT12a845 P64R15F1V)0164R9F8@6)464F9VV10R)4@864647R)UU87148218@v}‘(V)0xA8@xwpHegXYA156712@1528P3U1S82@A1429F–1R@15 @)562093@6)4FGk14@U1gb(I845V6U@1015@A0)97AR1UU2@086410X

f9VVGR)8@2)VA9P84FU))5V0)PA18U@AGa51r6514@6V615aR)4214@6475)4)02S101)F@86415V0)P@A1gklfU))5Y0842V926)4 I10T6R1Xe1063A108UFU))5P)4)49RU180R1UU2Befg(2CS101213808@15V0)PF9VVGR)8@2FG851426@G7085614@R14@06V978@6)492647 `6R)UUe8m91eU92Bkdl18U@AR801C8450129231451564pXo’fIb64Hegrx€tpg1568Bkdl18U@AR801CXefg(2S10164R9F8@15V)0 nP64S6@A6R1rR)U5U8R@6R8R65F9VV10BxvXtPg U8R@8@1axtpPg Q8(UaoPg (Ua3lvXuC@)01P)T1145)714)92A9P84r7d)4 @A1R1UU290V8R1306)0@)419@08U6i8@6)4FGxn’Y062B3lwCX(1UU2S101@A14S82A1584564R9F8@15xA)908@v}‘(S6@Ax“7zƒbr 231R6V6RI68Ar7d)0b2Ar7d310xfxp€R1UU264F82)3A6U8R@6T8@6)4F9VV10BpXo’fIbanPg (8(Un845nPg g7(Un64Hegrx€tp g1568CXI1426@6i15R1UU2S101S82A158450129231451564F82)3A6U8R@6T8@6)4F9VV102933U1P14@15S6@Axp47hPEA9P84rEv Be130)@1RAC306)0@)vpP64zƒb8R@6T8@6)4XbR@6T8@6)4S822@)3315FG8556@6)4)V6R1rR)U5pXngdqYb64`b(IF9VV10X(1UU2S101 S82A158450129231451564`b(IF9VV10306)0@)84@6F)5G2@864647Bdf@14515q8@8Y8FU1vCV)08R@6T8@6)4P80W10BEbgerv„(q€v ‰C)4F82)3A6U2B(qxnv‰lEbqHrCX

e06P80GA9P84P82@R1UU2S101714108@1582301T6)92UG512R06F15Xf061VUGaefg(S101213808@15V0)PF9VVGR)8@282512R06F15 8F)T1F1V)0162)U8@6)4)V(qvt‰3U9063)@14@A1P8@)3)61@6RR1UU2FGd82GI13Ygl9P84jA)U1fU))5(qvte)26@6T1I1U1R@6)46@ rrBIYdg(dEEY1RA4)U)7612CX(qvt‰R1UU2S101R9U@901564I@1Pe0)rvtI`gP1569PBE6V1Y1RA4)U)7612C2933U1P14@15S6@AnPg Er7U9@8P641axppŠhPU3146R6UU64axpp”7hPU2@013@)PGR64aop47hPE01R)PF6484@A9P842@1PR1UUV8R@)0Be130)@1RACaop47hPE A9P84rEr€Be130)@1RACa845S6@Axp47hPEA9P84rErvBe130)@1RAC64@A1V602@S11WXYA1R1UU2S101R9U@90156426P6U80R9U@901 P1568F9@S6@A)9@A9P84rErv8V@10@A1V602@S11W@)P8@901V)0xpS11W2X(9U@9015P82@R1UU2S101R)4V60P15FG`b(I2@864647 )V(qto‰aRrrY‰a`R Hr‰X

Y)1406RAV)048@6T1A9P84r7dV0)P2109Pa2109PS82@018@15S6@A30)@164k@)513U1@1r7kF1V)0164R9F8@647S6@A84@6rA9P84 r7dR)4–978@158780)21F1852X

Y)714108@101R)PF6484@zƒbr231R6V6RP)921)0A9P84r7da3U82P652R)4@864647zƒbr231R6V6RA18TG845U67A@RA864S101

@0842614@UGR)r@0842V1R@1564df36nuv`R1UU292647df36nuvdf301226)4IG2@1P6@XYA1R1UU21f30122647r7dS10121U1R@15FGtpp    “7hPEktxw8451f384515@)8U807102R8U130)59R@6)4Xzƒbr231R6V6Rr7dS823906V615V0)PR1UUR9U@9012931048@84@FGzƒbr ! "

R)93U158780)21F1852X # $ % & r42@09P14@ `b(I28P3U12S1018Rm96015FG84EIHrrVU)SRG@)P1@10Bfqf6)2R614R12C)0(G@)`EdsBf1RWP84()9U@10CX '

I)V@S801 bUU58@8S101848UG215FGk083Ae85e062PT1026)4wX`b(I58@8S101848UG215FG`U)S{)T1026)4xpX

(1UU3)39U8@6)48F94584R1 YA1R1UU2S1014)@2)0@1564@A622@95GX

Œ k8@6472@08@17G Y)21U1R@P)9212W64P82@R1UU2Bdf@14515q8@8`67Xo1Ca@A1R1UU2S101V602@78@15)46@2T68F6U6@GV)UU)S15FG(qto‰1f301226)4X 

j6@A64(qto‰3)39U8@6)4aR1UU2S6@AU)S1f301226)4)V(qxxFa(qxxRa845kHxB(qxxFr(qxxRrkHxrCS10121U1R@15V)UU)S15FG  

78@647)4@A13)39U8@6)4S6@AA67ARr6@BRr6@‰C1f301226)4X     Y)78@1F82)3A6U2Bdf@14515q8@8`67X€RCa2647U1@R1UU221U1R@15F8215)4`I(rbh`I(rlS10178@15FG@A1A67A1f301226)4)V  

(qxnvB(qxnv‰C845U)S1f301226)4)VlEbqHBlEbqHrCX   

q Y6RW@A62F)f@)R)4V60P@A8@8V679011f1P3U6VG647@A178@6472@08@17G6230)T651564@A1I933U1P14@80Gr4V)0P8@6)4X                  

! " # $ % & '

•