Cross-reactive carbohydrate determinant-specific IgE obscures true atopy and exhibits ⍺-1,3-fucose epitope-specific inverse associations with asthma

2.1 Study design and population

The three studies comprised rural and urban cross-sectional surveys on allergy outcomes and a case-control study on asthma risk factors among schoolchildren in Uganda.

The rural survey (September 2015-August 2016) was conducted in Sm-endemic fishing villages of Koome islands, Lake Victoria, Uganda (Figure S1). It was the three-year outcome survey of the Lake Victoria Island Intervention Study on Worms and Allergy-related diseases [LaVIISWA; ISRCTN47196031³⁹], a cluster-randomized trial of community-wide standard versus intensive anthelminthic treatment.^{39, 40} The urban survey (September 2016-September 2017) was conducted in Entebbe municipality, a lower helminth exposure, urban setting³⁷ situated on northern shores of Lake Victoria (Figure S1).

The asthma study (May 2015-July 2017)³⁸ enrolled children with doctor-diagnosed asthma and controls from schools in Entebbe and surrounding areas. All schoolchildren were screened using the International Study on Allergy and Asthma in Children (ISAAC) questionnaire.⁴¹ Children with a history of wheeze in the last year underwent a detailed clinical assessment by study clinicians, including a medical and treatment history, lung function tests, and were assessed for asthma control using the Asthma Control Test.^{38, 42} Among cases, 55.5% had well-controlled asthma, 29.5% partly controlled and 15% poorly controlled asthma. Only three cases had abnormal lung function tests. All cases were seen at one time point, started on the recommended treatment and were referred for additional management.⁴² Children in the same class as cases, with no history of wheezing, were eligible as controls. A Stata program (College Station, Texas, USA) was used to randomly select participants from each class register such that the number of controls was twice the number of cases.

Recent wheeze was also assessed in the rural and urban surveys using ISAAC questionnaires. Other procedures were identical in all three studies, including assessment of skin prick test (SPT) reactivity to allergens common in our setting⁴³: Dermatophagoides mix (D pteronyssinus and D farinae), Blomia tropicalis and Blattella germanica (ALK-Abelló; supplied by Laboratory Specialities [Pty] Ltd., South Africa). Visible flexural dermatitis [evaluated following Williams’ online manual⁴⁴], and questionnaire-determined recent rhinitis and urticarial rash were also assessed.

2.2 Parasitological examinations

Infection with Sm and intestinal helminths was investigated using the Kato-Katz technique⁴⁵ conducted on one stool sample per participant (two slides, read by different technicians). In the urban and rural surveys, stool was further examined for Sm, Necator americanus and Strongyloides stercoralis infections using PCR,^{46, 47} and urine for Sm circulating cathodic antigen (CCA, Rapid Medical Diagnostics, South Africa).

2.3 Measurement of allergen- and glycan-specific IgE

The ImmunoCAP^® specific IgE test [Thermo Fisher Scientific, Uppsala, Sweden] (hereinafter “ImmunoCAP^®”) was used to measure plasma IgE levels to whole allergen extracts of house dust mite (D pteronyssinus, HDM), peanut (Arachis hypogaea) and German cockroach (B germanica), with ≥ 0.35 kU/L defining allergic sensitization.⁴⁸ The ImmunoCAP^® method is described in this article's Supplementary information.

To assess IgE reactivity to established major allergenic components of these and other extracts, the ISAC (Immuno Solid-phase Allergen Chip) microarray [Thermo Fisher Scientific]^49-51 was used. The ISAC microarray comprised112 allergen components from 51 sources. The binding assay is described in this article's Supplementary information. The array comprises 67 recombinant components [produced in E coli, hence nonglycosylated⁵²]; and 45 components purified from natural extracts (denoted by prefix “n”, eg nMUXF3 for the glycan epitope from bromelain). Previous work has shown that nMUXF3 and glycoproteins from pollen (nPhl p 4, nCyn d 1, nPla a 2, nCry j 1, nCup a 1) and food (nJug r 2) are recognized by anti-CCD IgE, while other natural allergens on the ISAC are not.⁵³ Herein, “CCD-bearing components” denotes ISAC components confirmed to carry the classical core β-1,2-xylose and/or ⍺-1,3-fucose substituted N-glycans. We also use the term “established major allergenic components” to refer to specific proteins that bind IgE to their protein backbone and are well-recognized as the molecular drivers of allergenic potency, in specific allergen sources.^{54, 55} Participants with an ISAC standardized unit (ISU) measurement of ≥ 0.3 were considered sensitized to that allergen component.^{56, 57} Measurements were also reported as detectable versus undetectable (lower detection limit: 0.06 ISU).

A noncommercial microarray of 135 chemo-enzymatically synthesized glycans with and without β-1,2-xylosylation and/or ⍺-1,3-fucosylation (Figure S2) was employed to measure plasma anti-glycan IgE. Microarray construction procedures have been published.^{58, 59} The IgE-binding assay and the microarray image processing procedures were adapted from previous studies^{27, 60-62} and are detailed in the Supplementary information. Median fluorescence intensities (MFIs) reported herein represent anti-glycan IgE concentrations, because fluorescence-labelled anti-human IgE was used to detect plasma IgE bound to individual glycans on the array.

Samples were also assessed for Schistosoma egg [SEA]- and adult worm [SWA] antigen-specific IgE, IgG and IgG4, by ELISA [described elsewhere⁶³].

2.4 Statistical analysis

Statistical analyses were done in Stata 13.1, GraphPad Prism 7.0a (Fay Avenue, CA, USA) and R via the RStudio interface (version 1.1.383; Boston, USA). Initial analyses in the rural survey investigated the impact of trial intervention on IgE profiles using a cluster-level approach, as previously described.³⁶ Differences in characteristics between rural and urban survey participants and between asthmatic and nonasthmatic children were assessed using logistic or linear regression, allowing for the survey designs (clustering and weighting). Unadjusted analysis of differences in individual N-glycan-/ allergen-specific IgE levels between rural and urban participants, and between asthmatics and controls was done using the Mann-Whitney U test, correcting for multiple testing using a Monte Carlo simulation approach⁶⁴ with 1000 permutations, to generate empirical p values. For analyses comparing prevalences of ISAC-determined IgE sensitization between these groups, chi-squared tests (or Fisher's exact test, for expected cell counts < 5) were used.

Since anti-glycan IgE responses were strongly correlated, they were further analysed using principal component analysis (PCA). Data for participants from all studies were pooled and principal component (PC) scores generated. Unadjusted and age- and sex-adjusted associations between PC scores and atopic sensitization, asthma status, Sm infection and survey setting were assessed using linear regression.

3.1 Participants’ characteristics

Allergen- and glycan-specific IgE measurements were conducted in a subset of randomly selected samples per study (Figure 1). Data on ImmunoCAP^®-determined IgE sensitization were available for 780, 345 and 400 participants of the rural survey, urban survey and asthma study, respectively. For the rural survey, we present data combined from both anthelminthic treatment arms, because there was no effect of trial arm on ImmunoCAP^®- or ISAC-determined IgE sensitization or on anti-glycan IgE reactivity (Table S1).³⁶

Table 1 shows participants’ characteristics. Rural participants were on average older than urban participants (P = .001) and more likely to be male. SPT reactivity to Dermatophagoides mix (P = .003) and B tropicalis (P = .025) was higher among urban participants, while total IgE (P < .001) and allergen extract-specific IgE sensitization (ImmunoCAP^® concentration ≥ 0.35 kU/L), particularly to cockroach (P < .001), was higher among rural participants. Urticarial rash was more common in rural participants (P < .001), while wheeze, rhinitis and dermatitis were rare in both settings. Rural, compared to urban participants, was more frequently infected with Sm (P < .001), T trichiura (P = .002), hookworm (P = .016) and S stercoralis (P = .001) and had higher median levels of SEA- and SWA-specific IgE, IgG and IgG₄ (P < .001). Adjusting for age and sex did not impact these differences.

Asthmatics, compared to nonasthmatic controls, had higher prevalence of SPT reactivity, ImmunoCAP^® IgE sensitization, dermatitis and rhinitis, and higher total IgE levels (Table 1). ImmunoCAP^® IgE sensitization was linked to exposure-related asthma symptoms. For example, of the children who reported dust as a trigger for their asthma symptoms, 65.6% were sensitized (IgE ≥ 0.35 kU/L) to D pteronyssinus, compared to 46.7% who did not report dust as a trigger (P = .009). This difference was not statistically significant for B germanica and A hypogaea-specific IgE sensitization.

Prevalence of helminth infections in the asthma study was low, but concentrations of SEA- and SWA-specific antibodies were moderate, and similar between asthmatics and nonasthmatics (Table 1).

3.2 Overall responses to structures on the ISAC and N-glycan microarray

Figure S3 shows prevalence of sensitization (IgE ≥ 0.3 ISU) to allergen components on the ISAC microarray. Among rural participants, sensitization to components on the array was dominated by reactivity to natural components bearing classical CCDs and to insect venom proteins (rPol d 5, rVes v 5, rApi m 1). Among urban participants and nonasthmatics, sensitization patterns were more varied: natural and/or recombinant HDM, venom, food and CCD-bearing components contributed most to sensitization. However, sensitization among asthmatics was dominated by reactivity to major recombinant and natural HDM allergens.

Figure S4 shows average MFIs for structures on the glycan microarray. Although the highest responses were raised predominantly against core β-1,2-xylose and/or α-1,3-fucose substituted N-glycans in all three studies, responses to other carbohydrate structural elements represented in the array, such as tri-mannose (G99), fucosylated GlcNAc (G111) and galactose-alpha-1,3-galactose (α-1,3-gal) [G112, G113], were also as high.

3.3 Sensitization to allergen extracts does not reflect sensitization to their established major allergenic components, except among asthmatics

In all study settings, sensitization to HDM, German cockroach and peanut extracts (ImmunoCAP^® IgE ≥ 0.35 kU/L) was high (Table 1). Fifty five per cent and 43% of rural and urban participants and 73% and 54% of asthmatic schoolchildren and their controls, respectively, were sensitized to at least one of the three extracts. Prevalence of SPT reactivity to each extract was lower than prevalence of IgE sensitization to the extracts of the same allergens (Table 1). The majority of SPT reactive individuals (88% rural, 90% urban, 95% asthmatic and 93% nonasthmatic controls) were also IgE-sensitized to at least one of the three extracts on ImmunoCAP (data not shown).

However, ISAC microarray showed that sensitization to both natural (non-CCD-bearing) and recombinant forms of established major allergenic components of these extracts was very low among rural and urban survey participants, and among nonasthmatic schoolchildren (Figure 2): 0%-3% (rural), 0%-10% (urban) and 0%-12% (nonasthmatics) of tested participants were sensitized (IgE ≥ 0.30 ISU) to HDM, cockroach and peanut components on ISAC. The picture was strikingly different among asthmatics: 18%-36% were sensitized to HDM allergens (42% to at least one) and 0%-10% to cockroach, but none to peanut components.

3.4 Reactivity to CCDs dominates ISAC-determined IgE profiles

Despite the low prevalence of ISAC-determined IgE sensitization to major natural and recombinant components of common allergens in our study settings, many participants mounted responses to components bearing classical CCDs (Figure 3A and B). A higher proportion of rural, compared to urban participants, recognized CCDs (Table S2 and Figure 3A). However, this was statistically significant only for the CCD-bearing components nPhl p 4 and nCry j 1. Conversely, a higher proportion of urban, compared to rural participants, recognized recombinant established major allergens of HDM (Table S2 and Figure 2A), the food components rApi g 1 and rMal d 1 and the cat allergen rFel d 4 (Table S2).

Although reactivity to HDM components was higher among asthmatics compared to controls, reactivity to components bearing classical CCDs and most other allergen components on the array was similar between the two groups (Figure 2 and Table S2).

Many participants also mounted responses to venom proteins (Figure S5). This may represent insect venom sensitization; however, higher reactivity was observed among rural compared to urban participants (Table S2 and Figure S5) and among Sm infected compared to uninfected participants (Table S3), possibly reflecting sensitization to Venom-Allergen-Like (VAL) proteins expressed by schistosomes⁶⁵ and other helminths.^{66, 67}

3.5 Reactivity to core β-1,2-xylose/α-1,3-fucose is positively associated with Sm infection and sensitization to allergen extracts, but not sensitization to their established major allergenic components

Cross-reactive carbohydrate determinants expressed by several insect and plant glycoproteins, and by some nematodes and trematodes (such as schistosomes), are typified by presence of α-1,3-linked core fucose and β-1,2-linked core xylose motifs.^{11, 18} We combined the three studies and further explored associations between CCD-specific IgE and Sm infection and atopic sensitization using microarray binding studies of core β-1,2-xylose and α-1,3-fucose substituted N-glycans.

Responses to individual core substituted N-glycans were strongly correlated and were combined using PCA. Figure 4A shows scatterplots of PC1 and PC2 loadings (all participants). PC1 was characterized by responses to core β-1,2-xylose and/or α-1,3-fucose substituted N-glycans while PC2 was characterized by responses to nonxylosylated and nonfucosylated glycans. There were positive associations between PC1 scores and (a) Sm infection (Figure 4B) [but not other helminths, data not shown], (b) the rural environment (Figure 4C), (c) HDM, cockroach and peanut extract-sensitization (on ImmunoCAP^®) and (d) CCD sensitization [on ISAC] (Figure 4D-G). However, no associations were observed between PC1 scores and sensitization to any of the major natural and recombinant HDM, cockroach and peanut components on ISAC (Figure 4D-F).

3.6 Inverse associations between reactivity to a subset of core α-1,3-fucosylated N-glycans and asthma, but not for CCD-specific IgE in general

No associations were observed between glycan microarray-assessed IgE response PC scores and asthma or SPT reactivity (Figure 4H-I). Assessment of associations between asthma and responses to individual glycans on the array showed that asthmatics, compared to controls, mounted significantly lower responses to N-glycans carrying core α-1,3-fucose only or in combination with α-1,6-fucose, but not to any other structures on the array (Figure 5).

We did not group asthma cases into “allergic” and “nonallergic” based on ImmunoCAP^® IgE sensitization, as our data showed that IgE to allergen extracts reflected sensitization to environmental exposures such as CCDs, and could not accurately define allergic asthma. However, we grouped asthma cases based on SPT reactivity to at least one allergen extract, to assess associations with responses to N-glycans carrying core α-1,3-fucose (Figure S6A). Compared to nonasthmatic controls, both “allergic” (SPT+) and “nonallergic” (SPT−) asthmatics mounted lower responses to core α-1,3-fucose substituted N-glycans. There were no differences in responses between SPT- and SPT + asthma cases. Similar results were observed when we based on sensitization to at least one recombinant allergen component on the ISAC microarray to group asthma cases into “allergic” (IgE ≥ 0.3 ISU) and “nonallergic” (IgE < 0.3 ISU) [Figure S6B].