Successful Milk Oral Immunotherapy Promotes Generation of Casein-Specific CD137+ FOXP3+ Regulatory T Cells Detectable in Peripheral Blood

Background Oral immunotherapy (OIT) is an emerging treatment for cow’s milk protein (CMP) allergy in children. The mechanisms driving tolerance following OIT are not well understood. Regulatory T cells (TREG) cells are key inhibitors of allergic responses and promoters of allergen-specific tolerance. In an exploratory study, we sought to detect induction of allergen-specific TREG in a cohort of subjects undergoing OIT. Methods Pediatric patients with a history of allergic reaction to cow’s milk and a positive Skin Pick Test (SPT) and/or CMP-specific IgE >0.35 kU, as well as a positive oral challenge to CMP underwent OIT with escalating doses of milk and were followed for up to 6 months. At specific milestones during the dose escalation and maintenance phases, casein-specific CD4+ T cells were expanded from patient blood by culturing unfractionated PBMCs with casein in vitro. The CD4+ T cell phenotypes were quantified by flow cytometry. Results Our culture system induced activated casein-specific FOXP3+Helios+ TREG cells and FOXP3- TEFF cells, discriminated by expression of CD137 (4-1BB) and CD154 (CD40L) respectively. The frequency of casein-specific TREG cells increased significantly with escalating doses of milk during OIT while casein-specific TEFF cell frequencies remained constant. Moreover, expanded casein-specific TREG cells expressed higher levels of FOXP3 compared to polyclonal TREG cells, suggesting a more robust TREG phenotype. The induction of casein-specific TREG cells increased with successful CMP desensitization and correlated with increased frequencies of casein-specific Th1 cells among OIT subjects. The level of casein-specific TREG cells negatively correlated with the time required to reach the maintenance phase of desensitization. Conclusions Overall, effective CMP-OIT successfully promoted the expansion of casein-specific, functionally-stable FOXP3+ TREG cells while mitigating Th2 responses in children receiving OIT. Our exploratory study proposes that an in vitro TREG response to casein may correlate with the time to reach maintenance in CMP-OIT.

Background: Oral immunotherapy (OIT) is an emerging treatment for cow's milk protein (CMP) allergy in children. The mechanisms driving tolerance following OIT are not well understood. Regulatory T cells (T REG ) cells are key inhibitors of allergic responses and promoters of allergen-specific tolerance. In an exploratory study, we sought to detect induction of allergen-specific T REG in a cohort of subjects undergoing OIT.
Methods: Pediatric patients with a history of allergic reaction to cow's milk and a positive Skin Pick Test (SPT) and/or CMP-specific IgE >0.35 kU, as well as a positive oral challenge to CMP underwent OIT with escalating doses of milk and were followed for up to 6 months. At specific milestones during the dose escalation and maintenance phases, casein-specific CD4 + T cells were expanded from patient blood by culturing unfractionated PBMCs with casein in vitro. The CD4 + T cell phenotypes were quantified by flow cytometry.
Results: Our culture system induced activated casein-specific FOXP3 + Helios + T REG cells and FOXP3 -T EFF cells, discriminated by expression of CD137 (4-1BB) and CD154 (CD40L) respectively. The frequency of casein-specific T REG cells increased significantly with escalating doses of milk during OIT while casein-specific T EFF cell frequencies remained constant. Moreover, expanded casein-specific T REG cells expressed higher levels of FOXP3 compared to polyclonal T REG cells, suggesting a more robust T REG phenotype. The induction of casein-specific T REG cells increased with successful CMP desensitization and correlated with increased frequencies of casein-specific Th1 cells

INTRODUCTION
Cow's milk allergy (CMA) affects close to 0.6% of children under 2-years of age (1,2). Up to 80% of children are expected to outgrow CMA by adulthood (3), but persistent CMA is a major risk factor for anaphylaxis due to accidental milk ingestion in school age-children (4). Cow's milk oral immunotherapy (CM-OIT) is emerging as an effective experimental approach to induce tolerance to milk protein, with up to 75% of patients successfully achieving desensitization (4)(5)(6)(7).
However, there are still a number of patients who fail to achieve sustained unresponsiveness to CMP, lose their state of desensitization to CMP during the maintenance period or discontinue treatment despite the demonstrated clinical efficacy of CM-OIT (8). Furthermore, successful CM-OIT requires rigorous patient compliance, any deviation in protocol may prolong the length of time required to reach maintenance or increase the risk of developing an allergic reaction the scheduled CMP doses (9). Undoubtedly, individual differences in immunity can also contribute to the variable clinical outcomes observed in CM-OIT studies. Many efforts have been made to identify clinically relevant biomarkers that predict individual CM-OIT outcomes, none of which have been successful thus far (10,11). Since the clinical response to CM-OIT is highly variable, developing biomarkers that successfully predict ability to achieve desensitization, time to reach maintenance or risk of developing adverse events during therapy would enable the individualization of CM-OIT and increase safety of the procedure.
Recently, investigators have focused on examining the upstream cellular mechanisms implicated in oral tolerance to food. Regulatory T cells (T REG ), a class of CD4 + T cells expressing the transcription factor Forkhead box P3 (FOXP3), have been of particular interest given their key roles in induction and maintenance of peripheral tolerance to a plethora of self and non-self antigens (12). Allergen-specific T REG cells can suppress both innate and adaptive arms of an allergic response, preventing mast cell activation, IL-4 production, Th2 cell development and IgE production by B cells (13).
T REG cells can be readily measured in the peripheral blood and defects in their abundance and function have been implicated in the pathophysiology of food allergy (14). Indeed, mutations within the FOXP3 locus are associated with the development of severe food allergies due to a widespread loss of tolerance to innocuous antigens (15). Children with IgE-mediated food allergy have significantly lower FOXP3 expression compared to healthy controls (16,17), and decreased frequencies in circulating T REG cells after allergen exposure (18)(19)(20). In patients with peanut allergy, OIT increases both the abundance and suppressive function of T REG cells as well as induces epigenetic changes such as hypomethylation of the FOXP3 locus required for maintenance of a stable suppressive T REG cell phenotype (21). In children with milk allergy, those who tolerate baked milk have a higher frequency of peripheral blood casein-specific suppressive FOXP3 + CD25 + CD127 -T REG cells compared to children who do not, and this correlates with a higher likelihood of achieving milk tolerance (14). Similarly, children who outgrow their milk allergy have higher levels of peripheral CD4 + CD25 + T REG cells and lower in vitro T-cell proliferative responses to ßlactoglobulin than those who do not (22). However, while the frequencies of antigen-specific T REG cells and their secreted cytokines (IL-10, TGFb) increase during OIT (23), neither successfully predict OIT outcomes (10).
In addition to potential disease heterogeneity and methodological variations that may have contributed to failed prediction of OIT outcomes in these studies, lack of reliable human T REG cell markers is a significant limitation. T REG cells are a functionally heterogenous population (24,25) and traditional markers like CD25, CD127 and FOXP3 do not adequately discriminate between T REG from T EFF cells particularly in settings of T cell activation like allergy (25,26). Most commonly used T REG markers are also inducible on effector T cells (T EFF ) upon TCR-mediated activation, blurring the distinction between human T REG and activated T EFF cells, increasing the functional heterogeneity of the population and confounding the interpretation of results (25). Importantly, we have previously shown that expression of the transcription factor Helios alongside FOXP3, can reliably discriminate stablysuppressive T REG cells from T EFF cells in activated immune settings (25). Moreover, the differential expression of CD137 (4-1BB), a direct target of FOXP3, and CD154 (CD40 ligand) can further discriminate recently activated, functionally suppressive T REG from activated T EFF cells in human peripheral blood (27).
In this pilot CM-OIT clinical study, we performed in-depth, phenotypic characterization of CD4 + T cell subsets specific to casein, the major protein allergens in cow's milk. We aimed to evaluate whether CM-OIT induced casein-specific, stablysuppressive FOXP3 + Helios + T REG cells and whether this cellular response correlated with successful OIT. Here, we characterized casein-specific T REG and T EFF cell phenotypes, based on differential CD137 (4-1BB) and CD154 (CD40L) expression, respectively, at several time-points during CM-OIT in 7 pediatric patients that successfully achieved CMP desensitization. We hypothesized that successful CM-OIT would require the expansion of casein-specific CD137 + T REG cells rather than the polyclonal expansion of total peripheral blood T REG . Here, we propose that peripheral casein-specific CD137 + T REG responses during CM-OIT can be used to identify patients likely to achieve successful CMP desensitization and may correlate with CM-OIT time to reach maintenance.

Human Subjects
Seven patients were recruited from a prospective randomizedcontrolled trial aiming to compare adverse events in patients undergoing CM-OIT to patients that continued to avoid CMP. This study was conducted at the Pediatric Allergy and Clinical Immunology Department of the Montreal Children's Hospital (MCH) in Montreal, Quebec, Canada (4). Informed consent was obtained for every patient and the study was approved by the Research Ethics Board of the McGill University Health Center (PED-12-090).
Whole blood samples were obtained from 7 children who successfully completed CM-OIT (defined as successful challenge to 200 ml milk or 8000 mg milk protein) and from one healthy non-allergic control for comparison (26-year-old male), depicted in Figure 4. Briefly, for each study patient, IgE-mediated CMA was diagnosed by compatible clinical history and positive skin prick testing (SPT) with commercial CMP extract (≥3 mm over saline control) or positive serum casein-specific IgE levels (>0.35k U/L). Placebo-controlled single-blinded oral challenge to CM was used to confirm CMP allergy, and patients were assigned in a 1:1 ratio to either CM-OIT or CM avoidance for 1 year with crossover at the end of this period. The CM-OIT protocol started with rush desensitization and was followed by an early escalation phase (E; dose escalation from 6 ml to 25 ml of CM), a late escalation phase (L; dose escalation from 125 ml to 200 ml of CM) and a maintenance phase (M; maintained 200 ml of CM) (illustrated in Figure 1A). Blood samples were taken before OIT (baseline or B), during the E phase, the L phase, and 6 months after reaching the M phase (4).

Peripheral Blood Mononuclear Cells and Lymphocyte Isolation
Whole blood samples were collected at B, E, L, M phase timepoints as well as from the healthy non-allergic control, as described above. PBMC were isolated from heparinized blood using Ficoll-based density gradient centrifugation. Isolated lymphocytes were labelled with CTV (Cell Trace Violet) or CFSE (carboxyfluorescein diacetate succinimidyl ester) and distributed into 96-well flat-bottom plates at a concentration of 5 × 10 5 cells/well. Casein was dissolved in sodium hydroxide for 12 hours and adjusted to a pH of 7.3-7.4 with HCl before use.
Lymphocytes were incubated with prepared casein protein (500mg/ml) or medium alone (RPMI 1640 supplemented with 10% Nu-serum) and cultured at 37°C in a 5% CO 2 humidified incubator for 10 days, fresh media was replenished twice daily.

IgE and IgG Detection
Milk/casein-specific serum immunoglobulins were measured by ELISA. The 96-well polystyrene plates were coated with casein or capture antibodies for IgE or IgG4. Casein was dissolved using 1M NaOH for 4 hours. The protein concentration was adjusted with coating buffer to 20 ug/ml. Capture antibodies were diluted 1:3000 with coating buffer (pH 9.6). The coated plates were incubated overnight at 4°C. Coated plates were washed twice with PBS-T containing PBS (pH 6.8) and 0.05% Tween 20. The plates were blocked with 1% bovine serum albumin (BSA) in PBS-T for 2 hours at room temperature (RT), washed, and 50 ul of milk OIT participant serum diluted in blocking buffer was added to the plates and incubated for 2 hours at RT. Each participant serum sample was added in duplicate.
Serial dilutions of known concentrations of IgE or IgG4 standard were added to wells coated with IgE or IgG4 capture antibodies. Blank wells, wells containing only blocking buffer, and well containing serum from non-milk allergic healthy volunteers were used as negative controls. Following four washes with PBS-T, the plates were incubated for one hour at RT with biotinylated goat anti-human IgE antibody diluted 1:3000 or biotinylated mouse anti-human IgG4 antibody diluted 1:250 in blocking buffer. The plates were then washed twice with PBS-T then incubated for one hour at RT with Streptavidin-HRP. After four washes with PBS-T, 50ul of tetramethylbenzidine (TMB) was added to each well then incubated for 15 minutes at RT. The reaction was stopped with 50ul of 1M phosphoric acid. The optical density was measured at 450nm with a reference wavelength of 570nm. Values were converted from ng/mL to kU/L by dividing by a factor of 2.4.

Statistical Analysis
A non-parametric one-way ANOVA followed by a Dunn's Multiple Comparison post-test was used for longitudinal comparisons of parameters across more than two phases of the study (SPT wheal size, casein-specific sIgE and sIgG levels, changes in the proportions of peripheral T REG subsets), while a Wilcoxon Signed Rank test was used for longitudinal comparisons across two phases only (frequencies of peripheral Th1 and Th2 cells). To determine correlations between CD137 + T REG cells and cytokine-producing T EFF cells or number of escalation days, we conducted a Pearson correlation. For comparisons of cell proportions or protein expression (MFI) between two or more T cell populations within a single phase of our study, a Wilcoxon Signed Rank was employed. Parametric unpaired student's t-test or two-way ANOVA with Tukey's post-testing were used to determine significance in in vitro experiments completed in triplicates from a single individual. A two-sided p-value of <0.05 was considered statistically significant. Statistical analyses were performed using Prism 7 Software (GraphPad, San Diego, CA).

Successful OIT Patients Show Decreased Cow's Milk SPT and Increased Casein-Specific IgG4 Responses
The details of the global trial design were recently published and is depicted in Figure 1A (4). Seven children from this cohort who successfully achieved CMP-OIT maintenance dosing were randomly selected for this study. Baseline demographics and clinical characteristics of all subjects are outlined in Table 1. The mean age was 12 years and 4/7 were female (57%). All patients reached the target maintenance dose of 200 ml with an average escalation period of 266 days (range: 168-504, IQR=98). The mean cow's milk SPT was 10.5 mm (range: 8-15, IQR=1.75) at study entry and 4.79 mm (range: 0.5-9, IQR=4) after 6 months of CM-OIT maintenance, representing a significant decrease from baseline (p=0.03) ( Figure 1B). Casein-specific sIgE were available in all 7 patients but sIgG4 levels were only available for 6/7 patients. No significant changes in casein-specific IgE levels were detected during the study period (p=0.15) ( Figure 1C), whereas casein-specific IgG4 increased in all patients by the M phase (p=0.0071) ( Figure 1D). There was no

Desensitization Is Associated With Casein-Specific T EFF Cells With Altered Cytokine-Secreting Potentials
PBMC from each study subject was cultured with casein or Tetanus Toxoid (TT) for 10 days before T cell profiles were evaluated by flow cytometry. CM-OIT dose escalation was associated with the increased expansion of IFN-g-producing Th1 (CD4 + Foxp3 -) cells following in vitro casein challenge (Figures 2A, C, P=0.0625). In contrast, IL-4-producing Th2 cell expansion following casein challenge tended to decrease during CM-OIT dose escalation ( Figures 2B, D, P=0.0625). Correspondingly, the ratio of Th1 to Th2 cells increased between E and L phases ( Figure 2E, P=0.0625), albeit not significant. Analysis of Th1 and Th2 cells were only completed on 5 patients during E and L phases due to sample availability. Our data demonstrates a deviation in circulating Th2 responses towards Th1 immunity over the course of CM-OIT.

Casein-Specific Expansion of Stably-Suppressive FOXP3 + Helios + T REG Cells
To evaluate a potential increase in immunoregulation with CM-OIT, we aimed to characterize T REG cells both ex vivo and in our in vitro casein re-stimulation system. We compared the phenotypic definition of T REG cells using traditional markers (CD25 High CD127 Low ) to T REG cells defined by FOXP3 and Helios co-expression in a representative CMA patient before and after reaching maintenance dosing ( Figures 3A, B). Indeed, we have previously shown that FOXP3 + Helios + T REG cells represent a stably suppressive population of T REG in healthy individuals (24,25). Ex vivo and following in vitro stimulation with TT (antigenspecific T cell activation), the CD25 High CD127 Low gating excluded more than half of the FOXP3 + Helios + T REG cells (Figures 3C, D).
In contrast, after aCD3 stimulation (strong polyclonal T cell activation), the FOXP3 + Helios + gating was more stringent than CD25 High CD127 Low gating with the latter definition also including FOXP3 -T EFF cells and FOXP3 + Helios -T REG cells alongside FOXP3 + Helios + T REG cells ( Figures 3C, D). Thus, we decided to define T REG cells as FOXP3 + Helios + in both CM-OIT and our in vitro culture systems.
In healthy, non-allergic control conditions, casein stimulation elicited a weak FOXP3 + Helios + T REG proliferative response compared to stimulation with TT ( Figures 4A, B). However, in subjects with CMA, stimulation with casein elicited a robust proliferative response in FOXP3 + Helios + T REG cells ( Figure 4C), suggesting the presence of casein-specific T REG cells circulating in these patients.

Differential Expression of CD137 and CD154 Distinguish Casein-Specific T REG Cells and T EFF Cells, Respectively
Recently, it was suggested that CD137 and CD154 differential expression can identify antigen-specific T REG and T EFF cells in human PBMC, respectively (27,28). Hence, to evaluate the presence of casein-specific T cells in our in vitro culture system, we utilized these markers. Proliferating T REG cells were characterized by a significantly higher expression of CD137 than their non-proliferating counterparts ( Figures 4C, D); similarly, proliferating T EFF expressed higher levels of CD154 than nonproliferating T EFF cells (Figures 4C, E). These results show that within all casein-specific T cells, CD137 expression is confined to proliferating T REG cells whereas CD154 expression is confined to expanding T EFF cells. CD137 + is a marker of proliferating caseinspecific T REG cells, whereas CD154 + is a marker of proliferating casein-specific T EFF cells. We then evaluated the difference between CD137 + T REG and CD137 -T REG in terms of FOXP3 and Helios expression levels ( Figure 5). While CD137 + T REG cells expressed higher levels of FOXP3 at each timepoint (E, L, M) ( Figures 5B, C), Helios was differentially expressed between CD137 + T REG and CD137 -T REG at the L and M phase ( Figures 5D, E).

Induction of Casein-Specific CD137 + T REG Cells Correlates With Milk Sensitization, an Attenuated Th2 Response and Predicts the Length to Maintenance Phase
Since all patients successfully achieved the target CM-OIT maintenance dose, we sought to determine whether T REG or T EFF responses could be used as a marker of milk desensitization.
Using the T REG cell markers FOXP3 and Helios alone was insufficient to identify any differences in T REG responses to in vitro casein challenge from PBMC isolated during E, L and M phases ( Figures 6A, B). However, when stratifying T REG cell responses based on CD137 expression, we observe that proliferating FOXP3 + Helios + CD137 + T REG cells steadily increased during successful CM-OIT ( Figure 6C). The proportion of FOXP3 -Helios -CD154 + T EFF cells remained constant throughout the E, L and M phases ( Figure 6D), suggesting that in vitro CD137 + T REG cell induction rather than a reduction in antigen specific CD154 + T EFF cell is associated with casein desensitization. Moreover, we found patients who reached maintenance phase under 36 weeks had highest frequency of FOXP3 + Helios + CD137 + T REG than patients with more than 36 weeks to maintenance phase at M ( Figure 6C), suggesting higher frequency of FOXP3 + Helios + CD137 + T REG may be related to patients reaching M earlier. In early and late phases, the induction of FOXP3 + Helios + CD137 + T REG cells correlated with an increase in the frequency of T EFF cells with a Th1 phenotype and Th1/Th2 ratio in vitro ( Figures 6E, G). There was also a modest negative correlation between FOXP3 + Helios + CD137 + T REG and the frequency of T EFF cells with a Th2 phenotype, albeit not significant ( Figure 6F). Lastly, there is a negative correlation between the proportion of FOXP3 + Helios + CD137 + T REG and the number of escalation days required to reach maintenance at E ( Figure 6H), this is also observed for L and M, albeit non-significant ( Figures 6I-J). This suggests that FOXP3 + Helios + CD137 + T REG at E may correlate with individual time to reach maintenance.

DISCUSSION
Cow's milk OIT is an effective treatment for inducing oral tolerance in milk-sensitized individuals. However, its clinical applicability is limited by the inability to predict the probability of achieving successful desensitization or sustained unresponsiveness. In this exploratory proof-of-concept study, we suggest that stably-suppressive, casein-specific CD137 + FOXP3 + Helios + T REG may be a good candidate biomarker for identifying patients most likely to achieve successful CMP desensitization and be useful to predict time to reach maintenance in patients undergoing CM-OIT. We characterized the immune parameters of 7 children with successful CM-OIT at several timepoints during treatment. We began by evaluating the standard published biomarkers, namely SPT to cow's milk, casein-specific sIgE levels, casein-specific sIgG4 levels, as well as peripheral casein-specific Th1 and Th2 cells. As expected, casein-specific sIgE levels remained relatively stable during the study period, cow's milk SPT size decreased and casein-specific sIgG4 levels increased with successful desensitization. Most patients maintained a positive SPT to cow's milk and casein-specific sIgE levels in the maintenance phase, demonstrating an ongoing potential for reactivity to CMP despite clinical induction of desensitization.
Since allergen-specific T cell subsets are emerging as a potential prognostic indicator of OIT outcomes, we then examined at casein-specific T EFF and T REG subsets at each phase of our study. To identify casein-specific T cells, we labelled PBMC with either CTV or CFSE proliferation dyes to identify expanding (CTV low or CFSE low ) subsets upon exposure to casein. We observed an expansion of IFN-g-producing T EFF (Th1) cells from culture with casein, with a modest corresponding decrease in IL-4-producing T EFF (Th2) cells between E and L phases, but this was not seen across the entire study period. This observation is in keeping with previous reports that CM-OIT induces a shift away from the predominant Th2 response to milk protein early during the desensitization process (3). Mechanisms of tolerance likely differ between dose escalation and maintenance phase which may explain why Th1 prominence only increased significantly during dose escalation in our study. Although T EFF subsets may change during OIT, predictive thresholds, appropriate timing of sampling and robust correlations with clinical phenotypes are lacking, and further studies are required to validate their clinical usefulness (10). Of note, we did not find any correlation between T EFF subtypes and the time to reach maintenance.
Induction of allergen-specific T REG cells has classically been shown to be a later effect of OIT, and product of local differentiation of conventional T cells into allergen-specific T REG cells following allergen exposure. These induced T REG cells (iT REG ) are less stable than their thymic-derived natural T REG (tT REG ) counterparts and have the potential to lose their suppressive phenotype under specific inflammatory contexts (29). Although the mechanisms of OIT mediating allergen tolerance have not been completely elucidated, stable T REG induction seems to be central for the achievement and maintenance of CMP desensitization and loss of suppressive function or possible conversion of these cells to a Th2 cell phenotype could be associated with OIT failure (30). Previous studies have routinely evaluated T REG in the clinic to predict OIT responses, but have been limited by the availability and choice of relevant surface markers to identify functional T REG phenotypes (10). While both iT REG and tT REG cell subsets may be engaged in milk OIT, our results indicate that the emerging casein-specific T REG cells express Helios, a transcription factor more frequently  associated with T REG cells of thymic origin (tT REG ). Recently, however, Helios expression has also been shown to reflect T REG stability and suppressive function, rather than mere T REG lineage, as Helios acts to maintain the chromatin structure required for the induction and maintenance of the T REG developmental program (31). Therefore, we interpret enhanced Helios expression as a marker of functionally suppressive T REG . CD4 + T REG cells have classically been defined by their expression of intracellular FOXP3, high cell surface expression of CD25 and low surface expression of CD127. However, CD25 and CD127 can be transiently modulated on CD4 + T EFF cells upon immune activation and FOXP3 can be transiently expressed in T EFF cells upon T cell receptor (TCR) ligation (32,33). Furthermore, although FOXP3 reliably identifies T REG in their resting, non-activated state, not all CD25 + CD127 low FOXP3 + T REG clones are functionally suppressive (24). Thus, traditional markers of T REG cells are not sufficient to identify functional and dysfunctional T REG phenotypes.
Differential expression of a transcription factor of the Ikaros family, Helios, has been shown to reliably distinguish suppressive Helios + FOXP3 + T REG from non-suppressive Helios -FoxP3 + T REG clones (25). However, CTV low CD4 + FOXP3 + Helios + T REG did not vary significantly during early, late and maintenance phases of CM-OIT in our study indicating that Helios may not be sufficient to identify allergen-specific T REG . Next, we sought to evaluate CD137 (4-1BB), a T REG co-stimulatory receptor and a direct target of FOXP3 which has lately been identified as a robust marker of recently activated, antigen-specific, functionally suppressive iT REG (27). Since effective T REG suppression is antigen-specific, we hypothesized that successful CM-OIT would correlate with the expansion of casein-specific FOXP3 + Helios + CD137 + T REG cells (CD137 + T REG ) rather than polyclonal T REG activation or decrease in allergen-specific T EFF . In keeping with this hypothesis, we did observe that proliferating CD137 + T REG significantly increase during early, late and maintenance phases of CM-OIT. Moreover, we found that the induction of CD137 + T REG correlated with an increase in the frequency of T EFF cells with a Th1 phenotype and a modest Th1/ Th2 ratio suggesting that CD137 + T REG suppress Th2 immune responses in CM-OIT. The negative correlation between frequencies of CD137 + T REG cells and number of escalation days, and the finding that individuals with higher frequencies of CD137 + T REG cells during the M phase needed less time to reach maintenance suggests that CD137 + T REG may be useful for predicting time to reach maintenance during CM-OIT. To ensure that casein tolerance was possibly driven by CD137 + T REG induction rather than a decrease in antigen-specific T EFF cells, we compared proliferative T EFF responses at each CM-OIT timepoint. Using CD154 as a marker of recently activated, antigen-specific T EFF cells (27,28), we found no significant difference in terms of proliferating CD4 + FOXP3 -Helios -CD154 + T EFF cells (CD154 T EFF ) throughout the study period. Since a higher level of FOXP3 and Helios expression has been associated with increased suppressive potency and stability of the T REG phenotype (25), we sought to determine differential expression of these two markers on CD137 + and CD137 -T REG cells. Indeed, casein-specific CD137 + T REG cells exhibited a higher level of FOXP3 expression than their CD137counterparts at each timepoint, whereas Helios was only differentially expressed between CD137 + T REG and CD137 -T REG at the M phase. These observations suggest that the circulating casein-specific CD137 + T REG cells acquire a stable and more suppressive phenotype throughout CM-OIT, and that Helios expression, thus far not described in the OIT literature, may be utilized as a marker of successful OIT.
In summary, we have performed an exploratory CM-OIT study and identified a potential clinically useful biomarker to identify patients most likely to achieve successful CMP tolerance and sustained unresponsiveness during CM-OIT. This remains a pilot study and our conclusions will be The induction of CD137 + proliferative T REG correlated with an increase in the CD4 + IFN-g + T EFF cells from culture with casein and the ratio of CD4 + IFN-g + T EFF to CD4 + IL-4 + T EFF during Early and Late phase. (F) There was also a trend of correlation between CD137 + proliferative T REG and CD4 + IL-4 + T EFF cells from culture with casein, although there is a no significance. (H) There is a negative correlation between the proportions of CD137 + proliferative T REG at (E) and escalation days to maintenance. (I, J) There was also a trend of correlation between the proportions of CD137 + proliferative T REG at Late and Maintenance phase. and escalation days to maintenance, albeit no significance. Each symbol represents 1 subject. Of 7 patients, 5 patients from E and L phase are involved in analysis/figure (E-G). Yellow symbols represent data at Early phase Blue symbols represent data at Late phase. Red symbols represent data at Maintenance phase. P-values in (A-D) were determined using a one-way ANOVA with Dunn's multiple comparisons and in (E-J) with a Pearson correlation (*p < 0.05, n.s, not significant). Bars represent the mean ± s.d.
validated in larger cohorts of patients which will include additional age appropriate non-allergic controls and patients having failed CM-OIT. The clinical utility of CD137 + T REG quantification during CM-OIT merits further investigation and validation in larger cohorts.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by IRB of the McGill University Health Centre.
Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
YZ, LL, GG, DK, SB, NP, DL, T-AA-A, and BT: sample processing, experimental design, assay development and execution, data analysis/reporting, and/or figure/manuscript preparation MB, BM, and CP: trial design, experimental design, data analysis and reporting, figure preparation, and manuscript preparation. All authors contributed to the article and approved the submitted version.