Heterogeneity of magnitude, allergen immunodominance, and cytokine polarization of cockroach allergen‐specific T cell responses in allergic sensitized children

Abstract Background Characterization of allergic responses to cockroach (CR), a common aeroallergen associated with asthma, has focused mainly on IgE reactivity, but little is known about T cell responses, particularly in children. We conducted a functional evaluation of CR allergen‐specific T cell reactivity in a cohort of CR allergic children with asthma. Methods Peripheral blood mononuclear cells (PBMCs) were obtained from 71 children, with mild‐to‐moderate asthma who were enrolled in a CR immunotherapy (IT) clinical trial, prior to treatment initiation. PBMC were stimulated with peptide pools derived from 11 CR allergens, and CD4+ T cell responses assessed by intracellular cytokine staining. Results Highly heterogeneous responses in T cell reactivity were observed among participants, both in terms of the magnitude of cytokine response and allergen immunodominance. Reactivity against Bla g 9 and Bla g 5 was most frequent. The phenotype of the T cell response was dominated by IL‐4 production and a Th2 polarized profile in 54.9% of participants, but IFNγ production and Th1 polarization was observed in 25.3% of the participants. The numbers of regulatory CD4+ T cells were also highly variable and the magnitude of effector responses and Th2 polarization were positively correlated with serum IgE levels specific to a clinical CR extract. Conclusions Our results demonstrate that in children with mild‐to‐moderate asthma, CR‐specific T cell responses display a wide range of magnitude, allergen dominance, and polarization. These results will enable examination of whether any of the variables measured are affected by IT and/or are predictive of clinical outcomes.


| INTRODUCTION
Cockroach is a common allergen in urban and under-resourced areas and a significant source of atopic morbidity worldwide, particularly among children and young adults. [1][2][3] The German cockroach (CR, Blattella germanica) is commonly associated with CR allergies in the United States with CR allergens being detected in 85% of homes in low-income urban communities. 4 CR allergy has a high prevalence, and has long been established as a strong cause of asthma initiation and progression with early exposure leading to increased CR sensitization, asthma severity, and morbidity. [4][5][6][7][8] Several studies defined CR allergens based on IgE reactivity from sensitized individuals and correlated sensitization prevalence with severity of clinical symptoms. 3,[9][10][11][12] T cells, and in particular type 2T helper cells, significantly contribute to the development of allergy and asthma [13][14][15] ; however, CR-specific T cell responses have been characterized in less detail, [16][17][18][19][20][21] and very little information is available particularly for the population most impacted by CR allergies, namely urban and under-resourced children.
In general, allergen-specific T cell responses are often characterized following in vitro expansion steps 17,[22][23][24][25] to account for their low frequency which may also alter the phenotype of responding T cells. 26 We and others previously demonstrated that allergen-specific T cells can be detected ex vivo using a novel assay strategy with the combination of several T cell epitopes into pools. 24,[27][28][29] This technique uses the upregulation of the activation markers such as CD154 (CD40L) as a read-out for T cell reactivity and can be combined with intracellular cytokine staining (ICS) to further identify T cell phenotypes as well as polyfunctionality 24,27,30 and to improve the characterization of CR-specific T cell responses.
Thirteen groups of German CR allergens have been defined based on IgE reactivity and are listed at the official allergen database maintained by the World Health Organization and International Union of Immunological Societies (WHO/IUIS) Allergen Nomenclature Sub-committee (www.allergen.org). 2 As in the case of other allergies (i.e., cat, mite, or mouse 24,27,31,32 ), certain allergens have been described as dominant for CR-specific T cell responses. Those studies however were associated with certain limitations, such as testing a 2 of 12set of candidates derived from predicted epitopes from a more limited set of allergens, reliance on in vitro expansion and restimulation steps, and most importantly they only addressed dominance in sensitized adults. [16][17][18] Here we characterized the patterns of T cell responses to 11 cockroach allergens in a cohort of children with CR sensitization and asthma with a median age of 12 years that were enrolled as potential participants in an IT clinical trial (CRITICAL) and before initiation of treatment. T cell responses for each individual allergen were assessed by conducting ex vivo assays directly from PBMC not prestimulated with CR extract using pools of overlapping peptides spanning the entire protein sequence of the various CR allergens. The basal numbers of regulatory CD4+ T cells (Tregs) was also assessed, and compared with the magnitude of effector T cell responses.

| PBMC isolation
Peripheral blood mononuclear cells (PBMC) were isolated from whole blood by using CPT (cell preparation tubes) tubes according to the manufacturer's instructions (BD Vacutainer CPT tube with sodium heparin BD 362753, BD Biosciences) as detailed in the online supporting information.

| Measurement of IgE, IgG, and IgG4 antibody and peptide synthesis
Cockroach-specific IgE, IgG, and IgG4 antibody levels were measured in sera using a Thermo Fisher Scientific ImmunoCAP system (Phadia 250 Immunoassay Analyzer; Thermo Fisher Scientific) as explained in the online supporting information. Sequences of 11 major cockroach allergens were collected from UniProt. A strategy using 15mer peptides overlapping by 10 amino acids was selected to get the full coverage of all the allergens (Table S1). It is well established that epitopes bind to HLA-class II molecules through a nine residues core, and are on average 14-16 residues in overall length. The peptide selection utilized in this manuscript and commonly utilized in other studies relies on the fact that ensures that each 9 "core" amino acids that bind to the MHC-II groove is contained in at least one peptide.
An alternative strategy would be to generate 15mers overlapping by 14 amino acids. Although more rigorous, this strategy would significantly increase the number of peptides to be generated and tested which was not feasible in light of the limited numbers of cells available to study. Peptides were purchased from A & A as crude material on a small (1 mg) scale.

| Activation Induced Marker (AIM) assay and experimental design
Evaluations of T cell responses were based on previously described Activation Induced Marker (AIM) ex vivo assays, 30,35,36 utilizing the CD154 (CD40L) and CD137 (4-1BB) markers, combined with intracellular cytokine staining (ICS), using the antibodies described in Table S2 and performed as detailed in the online supporting information. The cytokines IL-4, IFNγ, and IL-10 were included as representative of Th2, Th1, and Tr1/T Regulatory (Tregs) CD4 helper T cell subsets, respectively. 24,37 The phenotypic CD127 and CD25 Tregs markers were also included in the cytometry analysis panel 30,38 and the several T cell parameters determined as described in the supporting information.

| Statistical analysis
Comparisons between cytokine responses were performed using the nonparametric two-tailed, paired Wilcoxon test. Correlations between magnitude of response and polarization or between different cytokine production in each participant were performed using the

| Overall magnitude and polarization of Bla g-specific T cell responses
The overall magnitude of CD4+ T cell CR-specific responses is shown in Figure 2a This large variation between participants was not the result of assay-to-assay variability, as demonstrated by specific controls utilized in the assays. More specifically, in each assay, we also included a PBMC aliquot from an adult volunteer with a moderate degree of cockroach allergy, who underwent apheresis to provide numerous cell aliquots available for use. As shown in Figure S1a, limited assayto-assay variability was observed in repeated assays (n = 42) from the lowest to the highest response in the 3.3-fold range, which is far less variable than variability observed within the participant cohort (57-fold). Furthermore, as shown in Figure S1b These results suggest that stronger responses to CR allergens were more IL-4 polarized, and that weaker responses were associated with IFNγ and/or IL-10 production. Correlation between the different cytokine responses in each individual participant was also analyzed.
As shown in 2, IFNγ and IL-10 responses were highly correlated but IL-4 did not correlate with IFNγ or IL-10 responses. Further correlation analysis comparing the polyclonal T cell responses (PMA/Ion stimulation) and CR-specific T cell responses was performed. As shown in Figure S3, CR-specific immune responses in this study do not reflect the general immune responses at the individual donor level. Lastly, in these experiments, we also included a previously described megapool of Bordetella pertussis (BP) epitopes. 41 Children in our cohort, based on their year of birth, are expected to have been vaccinated with the acellular pertussis (aP) vaccine which is associated with a predominantly Th2 response. 42,43 As expected, and shown in Figure S4a, polarization that are antigen/allergen-specific.

| Bla g 9 and Bla g 5 are the most dominant allergens for T cell responses
Previous studies established that certain allergens are immunodominant (i.e., account for a larger fraction of the total response) 17,24,44 at the population level. However, more granular analyses also

| Detection of basal levels of regulatory T cells
In the next series of experiments, we analyzed the number of Tregs at baseline in the absence of antigen stimulation. We defined baseline Tregs as the total number of CD137 + CD154-T cells detected at baseline, without any antigen stimulation. 30 As expected, Tregs defined as CD137+ were correlated with Tregs defined by the alternative markers CD4+CD25 + CD127low 38,45 (Figure 4a and 4b). There was substantial variation in the number of Tregs (total CD137+) among participants (Figure 4c), the number of CD137+ cells varing 90-fold from 925 to 86,000 per million CD4+ T cells.
We hypothesized that the size of the baseline Treg population present in each individual might influence their capacity to mount an allergen-specific T cell response following stimulation with the allergen-derived peptide pools. There was a suggestion of an inverse correlation between the number of Tregs (CD137+ unstimulated cells) and the number of Teff (cytokine + CD154+ allergenstimulated cells), but this was not statistically significant ( Figure 4d). We further observed to be the case in patients with IL-4 but not IFNγ dominant responses ( Figure S6).

| Correlation of T cell responses with sera antibody titers to CR extract, SPT responses, and clinical presentation
We also tested whether T cell responses were associated with IgE, IgG, and IgG4 titers specific for CR as well as with the ratio between IgE/IgG4 responses. The results summarizing these analyses are shown in Table 2 The same results were noted in regards to the correlation with the IgE/IgG4 ratios. In addition, as shown in Table 2, the total magnitude of responses positively correlated with IgE titers or the IgE/IgG4 ratios (p = 0.0082; R = 0.31 and p = 0.0071; R = 0.32, respectively), and no associations were observed for Treg frequencies.
We next examined the correlation of T cell responses with the results of a skin prick test (SPT) specific for extracts of German and American CR, which share homologous allergens (www.allergen.org).
As shown in Table S3, total responses or IL-4 responses were posi- Lastly, we examined correlations of the T cell variables (magnitude, polarization, or dominance of specific allergens) with clinical variables such as those captured by Composite Asthma Severity Index (CASI) score and components. 34 As shown in Table S4, none of the T cell immunological variables correlate with the total CASI score.
Also, no differences were observed in clinical symptoms between patients with IL-4 or IFNγ dominant responses ( Figure S7). These results were not unexpected given the fact that this cohort was designed to encompass relatively homogeneous CASI scores, ranging from low to medium severity (median CASI of 4).

| DISCUSSION
Here we report the direct ex vivo characterization of T cell responses to a panel of 11 previously described CR allergens in a cohort of CR allergic children and adolescents with well-controlled asthma. In particular, we found over 57-fold variation in magnitude of allergenspecific T cell responses. The more vigorous responses were also the most Th2 polarized and while the allergens Bla g 9 and Bla g 5 were most immunodominant, individual subjects exhibited distinctive patterns of allergen dominance. Subjects with higher magnitude of allergen-specific T cell reactivity had lower number of Treg populations. Overall, this study unveils a surprisingly high level of heterogeneity in the pattern of Bla g-specific T cell reactivity, both quantitatively and qualitatively. Since the cohort analyzed is enrolled in a future IT clinical trial, this finding lays the foundation for examining whether this heterogeneity may potentially correlate with IT outcomes.
This study is to the best of our knowledge, the first characterization of ex vivo CR-specific CD4+ T cell responses in CR-sensitized urban and under-resourced children with asthma, and for the majority of the known CR allergens to date, which are also components commonly found in CR extracts used for IT. 2,10 Our results also differ from previous attempts of portrayal of CR-specific responses, 17,19,20 because of the high level of granularity and because the ex vivo

Spearman r IgE Ext (RU a ) IgG Ext (RU a ) IgG4 Ext (RU a ) Ratio IgE/IgG4
Total magnitude of responses (CD154) We also analyzed the functionality of CD4+ T cell responses, which was dominated by IL-4 production and a Th2 polarization profile. Development of early allergic disease seems to be related to sustained Th2-skewed immunity during childhood 48 and previous studies have also demonstrated that levels of Th2 cytokines such as IL-4, IL-5, and IL-13 are associated with pathogenesis of both allergy and asthma. [13][14][15] In our study, Th2 polarization was the dominant phenotype but other patterns were also observed. The accuracy of this characterization was confirmed with a positive control included in each assay. In particular, while IL-10 responses were rare, a profile of IFNγ production or Th1 polarization was observed in one-fourth of the participants and associated with weaker T cell CR-specific responses. Also, since IFNγ is thought to be secreted in higher amounts per cell than IL-4, perhaps the actual net amount of cytokine production might not necessarily correlate with the number of cytokine secreting cells as determined herein, but additional studies would be needed to clarify this conjecture. Interestingly, resolution of allergicrelated immunopathologies is often also attributed to increase of IFNγ levels rather than reduction of Th2-cytokine production, 49 and IL-10 is found to regulate CD4+ Th2 cells during allergic airway inflammation. 50 Therefore, the close monitoring of both IL-10 and IFNγ cytokines would be of interest in the context of IT and could provide new clues for the association of baseline patterns of polarization or dominant allergens and IT outcomes.
Bla g 9 (arginine kinase) and Bla g 5 (glutathione-S-transferase) were the most immunodominant allergens. Interestingly, these enzymes are well conserved among arthropods or insects and have been shown to cross-react with human IgE antibodies 51,52 and are often considered pan-allergens and important players in the clinical manifestation of allergic sensitization. 53 Also, previous studies have reported Bla g 5 as a dominant T cell allergen 10,16,17 but patterns of allergen dominance could be associated with different forms of allergic disease. 17 These observations were made in adult cohorts, and this raises the possibility that Bla g 9 recognition could be a pattern predominantly associated with childhood CR allergy.
This pattern of immunodominance has been observed in subjects from the different clinical sites, irrespective of geographical location, further suggesting that Bla g 5 and Bla g 9 responses to CR are intrinsic to this particular age cohort. Changes in the IgE pattern of reactivity to individual Timothy grass and birch pollen allergens were observed over the course of 20 years for sensitized individuals. 54 Future studies could examine whether changes in allergen immunodominance are apparent as a function of age and exposure history.
The pattern of reactivity as a function of the specific subject considered was also heterogeneous with different individuals recognizing either Bla g 9 or Bla g 5 as dominant, with dominant responses to other allergens or no pattern of dominance also observed. Individual immunodominance patterns were observed in adult cohorts 16 and for IgE reactivity. 55 Different HLA could influence allergen immunodominance. Associations between CR sensitization and HLA class II antigens have been suggested 56,57 and consistently shown in food allergies. 58,59 In addition, we measured basal levels of non-antigen specific regulatory T cells and observed that T cells inversely correlated with the effector CD4+ T cell responses although it did not reach statistical significance. These results were not unexpected and were consistent with previous observations from Bacher et al. 30 . The fact that the population of Treg cells varies amongst participants is also of potential interest, as this will enable monitoring of Treg and addressing whether baseline Treg levels predicts response to IT. A possible outcome is that IT may induce or enhance regulatory T cell function, which can be further tested. Indeed, peanut oral immunotherapy has been shown to increase antigen-induced Treg function. 60 Overall, the current study reveals that in a pediatric cohort of CR-sensitized subjects with mild-to-moderate asthma, subjects exhibit substantial heterogeneity in many key immunological parameters associated with CR-specific T cell responses. We hypothesize that this large dynamic range of T cell reactivity may influence outcomes of allergen-specific IT. These findings will also enable future research examining which of these parameters best predict the trajectory of evolution of allergic disease, and responsiveness to therapeutic intervention.