Expansion of different subpopulations of CD26−/low T cells in allergic and non-allergic asthmatics

CD26 displays variable levels between effector (TH17 ≫ TH1 > TH2 > Treg) and naïve/memory (memory > naïve) CD4+ T lymphocytes. Besides, IL-6/IL−6R is associated with TH17-differentiation and asthma severity. Allergic/atopic asthma (AA) is dominated by TH2 responses, while TH17 immunity might either modulate the TH2-dependent inflammation in AA or be an important mechanism boosting non-allergic asthma (NAA). Therefore, in this work we have compared the expression of CD26 and CD126 (IL-6Rα) in lymphocytes from different groups of donors: allergic (AA) and non-allergic (NAA) asthma, rhinitis, and healthy subjects. For this purpose, flow cytometry, haematological/biochemical, and in vitro proliferation assays were performed. Our results show a strong CD26-CD126 correlation and an over-representation of CD26− subsets with a highly-differentiated effector phenotype in AA (CD4+CD26−/low T cells) and NAA (CD4−CD26− γδ-T cells). In addition, we found that circulating levels of CD26 (sCD26) were reduced in both AA and NAA, while loss of CD126 expression on different leukocytes correlated with higher disease severity. Finally, selective inhibition of CD26-mRNA translation led to enhanced T cell proliferation in vitro. These findings support that CD26 down-modulation could play a role in facilitating the expansion of highly-differentiated effector T cell subsets in asthma.

cells 18 , which explains the presence of CD26 high CD4 + T cells in AA 19,20 . A soluble version of CD26 (sCD26) has been found in the bloodstream as a free or a vesicle-associated protein [http://www.exocarta.org] 21 . Our previous in vitro studies evidenced a positive correlation between soluble DPP4 activity (an indirect measurement of sCD26) and CD26 expression on CD4 + cells 20 . Immune cells also appear to be a source of sCD26 in vivo 15 . However, contrary to our expectations, allergic asthmatics displayed higher membrane expression of CD26 on CD4 + T cells, but decreased levels of sCD26. This finding can be explained through the expansion of a "triple low" (Tlow; CD25 − CD127 − CD26 −/low ) subpopulation of effector T cells (Teff) 20 .
IL-6 is an important cytokine in the differentiation of TH 17 cells that acts via IL-6R (IL-6Rα/CD126 + gp130) 22 . In this sense, this cytokine down-modulates the TGF-β-driven expression of FoxP3 and up-regulates the levels of the transcription factor that controls the development of Th17 cells: RORγt 23,24 . IL-6 signalling is also essential for the generation of functionally active memory CD4 + T cells 25 . Like CD26, CD126 is also found in plasma as a soluble molecule: sIL-6Rα. This circulating protein binds to IL-6 and leads to the activation of CD126 − gp130 + cells, a process known as trans-signalling 26,27 . Indeed, CD4 + T cells down-modulate IL-6R upon inflammatory activation, but these cells retain the IL-6 response capacity through the trans-signalling pathway 28 , a mechanism of paramount importance for the maintenance of inflammatory diseases such as asthma [29][30][31][32] .
Most research in asthma has been focused on the allergic phenotype and CD4 + T cells. However, CD4 − lymphocyte subsets (e.g. CD8 + αβ-T cells or γδ-T lymphocytes) might also be relevant for this disease and its phenotypes/endotypes. Thus, γδ-T lymphocytes become activated by IL-6 trans-signalling 33 , and they are important inducers of allergic asthmatic responses 34 . Moreover, they are major initial producers of IL-17 35 . On the other hand, CD8 + T cells cooperate with CD4 + T cells to promote asthma and have been associated with poor lung fuction and airway obstruction [36][37][38][39] . Interestingly, the expression of both CD26 and CD126 defines diverse stages of differentiation of CD8 + T cells 40,41 , which might be modified in asthma 41 . Given this differential expression of CD26 and CD126 between different lymphocyte subsets, we postulate that both molecules might display coordinated expression levels and help distinguish different asthma phenotypes. Therefore, the aim of our study was to analyse the expression of CD26 and CD126 in CD4 + and CD4 − lymphocytes from healthy subjects and patients with AA/NAA or rhinitis.
Herein, we report a high correlation between the expression of CD126 and CD26 in lymphocytes. These molecules help differentiate between lymphocytes with naïve (CD26 intermediate or CD26 int ), central-memory (CD26 high ), or "terminally-differentiated" (T EMRA )/effector-memory (T EM ) (CD26 −/low ) phenotypes. Moreover, allergic and non-allergic asthmatics display low circulating sCD26 levels, which is consistent with the expansion of CD26 −/low T EM /T EMRA lymphocytes: CD4 + CD26 −/low T cells (previously named Tlow cells) 20 in AA and CD4 − CD26 − γδ-T cells in NAA. Acceleration of the natural course of CD26 down-modulation on T lymphocytes by siRNA leads to higher in vitro proliferation rates, which suggests that CD26 molecules on T lymphocytes could be acting as a "brake mechanism" that prevents their proliferation and the acquisition of an effector phenotype. Finally, a decrease in the number of CD126 molecules on leukocytes correlates with higher asthma severity. Thus, our findings provide new advances in asthma immunophenotyping and on the role of CD26/CD126 in this disease.

Results
Characteristics of study subjects. We performed a case-control study including adult patients with different asthma phenotypes (AA; NAA), rhinitis and healthy controls (HC). The characteristics of the donors in this study are summarized in Table 1. Pulmonary function parameters (FEV1 and FEV1/FVC) were lower in both AA and NAA relative to patients with rhinitis (Table 1). Haematological count revealed an increment of eosinophil numbers in asthma patients (both AA and NAA) compared to HC (Table 1). Furthermore, AA displayed higher blood eosinophil counts than patients with rhinitis ( Table 1). Levels of other leukocyte populations remained unchanged ( Table 1).
As expected, AA subjects exhibited higher levels of IgE than other study groups (Table 1), and the same happened for rhinitis compared to NAA (Table 1). Moreover, our results evidenced a reduction of IgG1 in NAA relative to AA, while IgG4 was higher in AA vs. R (Table 1). We also analysed the erythrocyte sedimentation rate (ESR), which was elevated in NAA vs. remaining groups (Table 1). CD26 and CD126 molecules display a highly correlated expression on lymphocytes, while the expansion of "triple low" (CD25 − CD26 − CD127 − ) CD4 + T cells explains the reduction of sCD26 in allergic asthma. T cells account for ~80% circulating lymphocytes and include CD4 − and CD4 + cells. CD4 + lymphocytes (TH) were subdivided into "conventional" Teff (CD25 low CD127 high ), Treg (CD25 high CD127 low ), and Tlow cells (CD25 low CD127 low ) (Fig. 1a). TH cells displayed the highest percentage of CD26 + cells within leukocytes (~85-90%), and this parameter evidenced a positive association with the percentage of CD126 + cells (Fig. 1b). Amongst TH lymphocytes, "conventional" Teff (CD26 +/high ) and especially Tlow (CD26 −/low ) cells exhibited the same high CD26-CD126 correlation, but Tregs appear to lose this association (Fig. 1c). Furthermore, we found a reinforced expression of CD26 (measured in Antibody Bound per Cell/ ABC; see Material and Methods section) on CD4 + T cells from AA and NAA patients (Fig. 1d) compared to HC. Strikingly, CD126 showed an unaltered expression in CD4 + T cells from the different groups of donors (data not shown) despite the strong correlation between CD26 and CD126 commented above (Fig. 1b).
The higher number of CD26 molecules on CD4 + T cells from AA and NAA patients compared to HC (Fig. 1d) was in line with the expected activated phenotype of these cells, but contrasted with the decreased levels of sCD26 in asthmatics (Fig. 1e). However, as most of sCD26 comes from T cells 42 , the expansion of CD26 −/low T lymphocyte populations may explain the decreased levels of sCD26 in AA. Indeed, even though Teff and Treg proportions remained unchanged, we detected an expanded subpopulation of CD25 − CD127 − CD26 −/low (Tlow) effector T cells in AA patients compared to HC (Fig. 1f). Moreover, contrary to CD4 + and "conventional" Teff cells (CD26 int/high subsets) and despite the CD26 −/low phenotype of these Tlow cells, the percentage of the last www.nature.com/scientificreports www.nature.com/scientificreports/ subset was negatively correlated with the percentage of CD26 + and CD126 + cells within the Tlow compartment (Supplementary Table S1). Interestingly, the percentage of CD127 + Treg cells was augmented in asthmatics compared to rhinitis and HC ( Supplementary Fig. S1).
The CD26 −/low subpopulation of γδ-T lymphocytes is augmented in non-allergic compared to allergic asthmatics. To ascertain the specific lymphocyte subset within the CD4 − compartment showing altered proportions in NAA compared to AA and to limit possible cofounding effects (e.g. age, gender), we analysed blood samples from a second cohort study limited to AA and NAA donors (n = 12/each) with similar age and M/F proportions. The new analysis focused on CD8 + T (CD3 + CD8 + ), NK (CD3 − CD56 + ), NKT (CD3 + CD56 + ), B (CD19 + ), and γδ-T cells (TCRγδ + CD3 + ) (Supplementary Fig. S4). There was no evidence for expansion of these circulating populations in NAA. However, this new study revealed an increased proportion of CD26 −/low γδ-T lymphocytes (and an opposite pattern for CD26 high γδ-T cells) in NAA compared to AA (Fig. 4). Once again, the percentage of γδ-T cells correlated inversely with the percentage of CD26 + cells within this subset (r s = −0.460, P = 0.024). Interestingly, there was also a negative correlation between the percentages of γδ-T cells and B lymphocytes (r s = −0.602, P = 0.002).
CD26 could act as a negative regulator of T-cell proliferation. CD26 down-modulation on T lymphocytes from both AA (CD4 + T cells) and NAA (CD4 − γδ-T cells) patients could be: a) a consequence of different mechanisms that simply reduce the amount of this protein on the cell surface, like, for example, the dilution of CD26 molecules (half-life > 48 h) throughout the successive cytokinesis rounds; or b) a necessary condition to initiate the proliferation/differentiation programme of naïve or memory T cells. To distinguish between both possibilities, we tested the effect of RNA interference (RNAi) on the DPP4 gene during the proliferative response of T lymphocytes to mitogenic triggers. Peripheral blood mononuclear cells (PBMCs) were CFSE-labelled and cultured in vitro with either non-target or CD26/DPP4-specific Accell siRNAs. As CD26 up-regulation during T cell activation was mainly derived from the translocation of this protein from intracellular stores toward the cell www.nature.com/scientificreports www.nature.com/scientificreports/ surface, T-cell division was stimulated with phytohemagglutinin P (PHA) in the presence or absence of IL-12, a cytokine that promotes CD26 mRNA translation. Furthermore, it was required to extend the in vitro culture incubation for 6 days to observe the inhibitory effect of the CD26-specific siRNA on protein levels. As expected, CD26-specific siRNAs down-modulated the expression of CD26, but only in IL-12-stimulated PBMCs (Fig. 5a). After verification of compliance with CD26 down-modulation by RNAi, we estimated the percentage of cells that divided at least once. As Fig. 5b shows, those T cells where DPP4 gene silencing was more intense (i.e. IL-12-costimulated) were the ones showing an increase in the proliferation rate. CD126/IL-6Rα down-modulation in neutrophils, monocytes, and lymphocytes is associated to disease severity and asthma control. Asthma severity could be responsible for changes in pulmonary and inflammatory parameters. To assess this, we segregated asthma patients into moderate-severe (n = 90) and intermittent-mild (n = 102) asthmatics; asthma control degree was also considered ( Table 1). As expected, FEV1 and FEV1/FVC were decreased in moderate-severe and uncontrolled asthmatics (Supplementary Table S2), regardless of phenotype. Eosinophils count was also augmented in moderate-severe and badly-controlled asthmatics (Supplementary Table S2), although the statistical significance is only maintained in AA after segregation according to the phenotype. In contrast, badly-controlled NAA was only characterized by an increased neutrophils count (P = 0.032; data not shown). Interestingly, there was a decrease in the expression of CD126 in many leukocyte subsets (monocytes, neutrophils, CD4 − and CD4 + lymphocytes) as the severity was higher (Fig. 6). This finding extends to Teff, Treg, and Tlow lymphocytes. Similar decreased levels were obtained for CD126 on leukocytes from badly-controlled asthmatics. Furthermore, we did not observe changes in CD26 levels on most leukocyte subpopulations with asthma severity, although the percentage of CD26 + Tregs was higher in moderate-severe NAA patients (Supplementary Fig. S5). Finally, the percentage of CD4 + lymphocytes was higher in badly-controlled AA (P = 0.006; data not shown).
www.nature.com/scientificreports www.nature.com/scientificreports/ CD26 could be relevant to slow the rate of cell division of naïve or central-memory T lymphocytes and, conversely, its down-modulation necessary for rapid proliferation and differentiation into effector cells. Finally, down-regulation of CD126 on leukocytes may be related to asthma severity.
CD26 is an activation marker 18 that identifies TH 1 and especially TH 17 cells 11 . Our results show an increase of "activated" CD4 + cells (CD26 + ) in asthma, in agreement with published work 19 . Despite the IL-13-dependent up-regulation of CD26/DPP4 on human bronchial epithelial cells 43 or the potential role of TH 17 cells in NAA 8,44 , both AA and NAA patients express similar levels of CD26 on CD4 + T cells. A possible explanation for this observation is that following antigen presentation to naïve CD4 + T cells in lymph nodes, these lymphocytes proliferate, differentiate, and move back to the peripheral circulation, where they still have an early differentiation state (CD26 high ; TH 17 ≫ TH 1 > TH 2 ) and an extended half-life compared to innate leukocytes (CD26 −/low ; eosinophils, neutrophils) (Supplementary Table S3). Therefore, these circulating TH cells will still require the down-modulation of CD26 to become T EM /T EMRA cells and migrate into sites of inflammation; a similar event would occur with IL-6Rα/CD126 on T lymphocytes to favour the proinflammatory trans-signalling pathway.
Three major TH subsets coexist in circulation: Treg (CD25 high CD26 low CD127 low ), "conventional" Teff (CD25 low CD26 high CD127 high ), and Tlow cells (CD25 −/low CD26 −/low CD127 −/low ). The latter is a highly differentiated counterpart of "conventional" Teff cells. CD4 + Tlow cells are expanded in AA 20 and their abundance is negatively correlated with CD26/CD126 levels, as it happens for other subpopulations (e.g. CD26 −/low γδ-T lymphocytes). Despite the small expansion of the Tlow cell compartment, this might be relevant for AA pathogenesis considering two facts: (a) total lymphocytes, instead of antigen-specific cells, have been measured; and (b) patients were in stable phase (i.e. absence of exacerbations for 4 weeks before sample collection). On the other hand, even though there is a negative association between the percentage of CD26 + /CD126 + cells and the percentage of Tlow cells, this does not necessarily imply CD26-dependent causation. Therefore, we performed T cell proliferation assays after siRNA-mediated depletion of CD26 mRNA. These results show a negative regulatory role for CD26 in T cell proliferation 15 . Thus, our data agree with the study of Yan et al. working with ovalbumin-induced CD26 −/− C56BL/6 animals 45 , or Stephan et al. showing that oral administration of DPP4-inhibitors aggravates the airway inflammation in a rat model of asthma 46 .
A reduced expression of CD27, CD28, and CCR7 is the hallmark of highly-differentiated effector cells (T EM or T EMRA ). Tlow cells fulfil these criteria; therefore, these lymphocytes might be part of a pool of "conventional" Figure 5. siRNA mediated depletion of CD26 mRNA leads to enhanced T-cell proliferation. PBMCs from healthy subjects were isolated and placed in culture for 6 days in 96 round-well plates. To promote T-cell division, Accell culture medium was supplemented with PHA ± IL-12. Besides, a CD26-specific or a nontargeting Accell siRNAs pool was also used. (a) Expression of CD26 (MFI; mean fluorescence intensity) on PBMCs was assessed by flow cytometry. Three representative assays are shown. 2-way ANOVA with Tukey's multiple comparison test: *P < 0.05, ***P < 0.001, ****P < 0.0001; n.s., non-significant. (b) PBMCs from 2 representative donors were labelled with CFSE and T-cell proliferation induced by PHA ± IL-12 was assessed by CFSE-dilution assays. Responder frequency is the percentage of T lymphocytes that divided at least once. 2-way ANOVA with Tukey's multiple comparison test: *P < 0.05, **P < 0.01, ****P < 0.0001; n.s., non-significant.
Another caveat is if those phenotypic changes on T lymphocytes could be mirrored in serum samples. Different circulating molecules were measured (TGF, TNF, sCD25, sCD26) ( Table 1; Fig. 1), but most of them remained unaltered. We did not quantify "soluble" IL-6R/CD126 (sIL-6R/sCD126), but several authors reported higher levels in stable asthma and especially upon flare-ups due to mechanisms involving sheddases (e.g. ADAM10/17) 53-55 , spliceosomes 53,56 , or vesicles 54  www.nature.com/scientificreports www.nature.com/scientificreports/ of CD4 + T lymphocytes 19 . However, reduced sCD26 levels have been recently described in severe asthmatics 57 or a low eosinophilic TH 2 low severe asthma endotype 58 . Our results are in agreement with these last works and support a rather generalized (AA and NAA) sCD26 down-modulation. The underlying mechanism is likely the expansion of CD26 − T subsets 20 with a T EM or T EMRA phenotype: CD4 + T cells in AA and CD4 − γδ-T cells in NAA. Reduced levels of CD26 on lymphocytes and the extracellular compartment could be concomitant with the loss of caveolin-1 (a CD26 ligand) in bronchial epithelial cells and monocytes from asthmatics 15 . Moreover, the decrease of CD26 levels may be important for the bioavailability of soluble factors (e.g. chemokines, adenosine) and to promote cell functions like proliferation, chemotaxis, and migration toward inflamed tissues 15 .
Treg cell function has been described as impaired in asthma 59 . Although we did not find deregulation of Treg numbers, they showed increased CD26 expression in asthmatic patients. This is relevant because CD39 is an ecto-enzyme expressed by CD26 − Treg lymphocytes 12,60 and involved in adenosine (Ado) production 61 . Ado is an immune-regulatory molecule whose synthesis is counteracted by adenosine deaminase (ADA), an ecto-enzyme anchored to CD26 62,63 . Therefore, a CD26 high phenotype in Tregs could decrease local Ado concentration and exacerbate disease severity 64 . Indeed, the percentage of CD26 + Treg cells in NAA was higher in moderate-severe patients than intermittent-mild subjects. The percentage of CD127 + Treg cells, a phenotype correlated with a diminished suppressive capacity 65 , was also augmented in asthmatics compared to rhinitis and HC. However, future studies including the assessment of Treg function in NAA and AA will be necessary.
Asthma severity is also influencing CD126 levels on CD4 + T cells, neutrophils and monocytes. IL-6 acts via either IL-6R (classic-signalling) or sIL-6R/sCD126 (trans-signalling) 22 . Contrary to the anti-inflammatory role of the first pathway 28 , the trans-signalling route allows CD126 − CD130 + cells to respond to IL-6 27 and is important in asthma through the maintenance of TH 17 cells or the inhibition of T-cell apoptosis 66 . Naïve CD4 + T cells down-modulate IL-6R upon TCR-mediated activation, probably due to protein shedding 28 . This release mechanism has been observed in asthma, and sIL-6R levels have been directly associated with IgE levels, but negatively with lung function 67 . Therefore, reduction of CD126 expression in monocytes, neutrophils, and CD4 + cells from moderate-severe patients highlights the role of IL-6 trans-signalling in asthma severity.
In conclusion, our data provide evidence that both asthma phenotypes share common immune-pathologic mechanisms, with expansion of CD26 −/low subsets in AA (CD4 + Tlow or "highly-differentiated" Teff cells) and NAA (CD4 − T cells; γδ-T lymphocytes) and down-modulation of additional surface molecules (IL-6Rα/CD126, CD27, CD28, IL-7Rα/CD127, CCR7) to produce differentiated effector subsets and extracellular sCD26 reduction. This CD26/sCD26 down-modulation and the potential role in T-cell proliferation should be considered in the light of clinical usage of DPP4 inhibitors and anti-CD26 antibodies.

Material and Methods
Subjects. Adult patients with asthma or allergic rhinitis were recruited from hospital consultations for Pneumology in Galicia (Spain) between 2014 and 2016. The diagnosis of different asthma phenotypes and allergy was confirmed according to Global Strategy for Asthma Management and Prevention (GINA 2006, http://www. seicap.es/documentos/archivos/GINA2006general.pdf) criteria for at least one year prior to study initiation. A positive skin prick test and the presence of allergen-specific IgE were used to confirm sensitization in allergic patients. Lung function parameters (forced expiratory volume in the 1st second (FEV1), forced vital capacity (FVC), and FEV1/FVC ratio) as well as eritro-sedimentation rate (ESR) were also analysed. All asthmatics were in a stable phase of the disease (i.e. absence of exacerbations for at least 4 weeks before sample collection). Healthy donors were subjects without allergy or systemic diseases, who were scheduled for minor surgeries (orthopedic surgery or inguinal hernia). A second cohort of patients was also included with 12 patients with AA (M/F proportion: 6/6; age: 52.75 ± 14.12) and 12 patients with NAA (M/F proportion: 6/6; age: 61.00 ± 10.71), recruited from hospital consultations for Pneumology in Galicia, Spain. The research project was approved by the Ethics Committee of Clinical Research of Galicia (2011/001), Spain, all subjects signed an informed consent, and all research was performed in accordance with the relevant guidelines and regulations.
Flow cytometry assays. Venous peripheral blood from each donor was collected in EDTA tubes (BD Vacutainer K2E). Then, FITC, PE, PE-Cy7, PerCP-Cy5.5 or AlexaFlour-647-labelled mouse IgG1 κ isotype antibodies (BD Bioscience) or specific antibodies were incubated with cells for 30 min in FACS buffer (PBS, pH 7.4, 2% FBS). We used BD FACS TM Lysing Solution (15 min; room temp.) to lyse red blood cells before sample collection. Finally, 200000 events were acquired on a BD FACSort TM flow cytometer and data were analysed using WinMDI 2.9 software (Joseph Trotter, La Jolla, CA. USA). A list of antibodies is shown in Supplementary  Table S4. Isotype antibodies were used to determine the non-specific binding of the antibodies and therefore to set a threshold value to identify negative and positive populations. Single stained lymphocytes were used for fluorescence compensation.
Flow cytometry data are presented as either "percentage of positive cells" or "number of antibodies per cell" (ABC) instead of mean fluorencence intensity (MFI). We used the BD Quantibrite ™ Beads PE kit (Fluorescence Quantitation Kit; BD Bioscience) to estimate ABC according to manufacturer instructions. In brief, we ran a BD Quantibrite PE tube with the same instrument settings as the assay. Therefore, MFI values in FL2 were converted into the number of PE molecules bound per cell. Finally, this number of PE molecules/cell was changed into ABC www.nature.com/scientificreports www.nature.com/scientificreports/ values by using known ratios of PE to antibodies. We took advantage of this transformation to minimise as much as possible the inter-day variation related to the working conditions of the flow cytometer.
In vitro proliferation assays and CD26 mRNA silencing. PBMCs were placed in RPMI 1640 at a cell density of 10 7 cell/mL and incubated with 5 μM CFSE for 8 min at RT in the dark. Then, FBS was added to stop the reaction and cells were thoroughly washed with RPMI 1640 before cell counting. Cell cultures were set up at 0.25 × 10 6 cells/mL in 96-well microplates (round-wells). Accell delivery media (ref. B-005000-500; Dharmacon) was used to culture these cells under non-serum conditions. The Accell delivery medium was supplemented or not with 1 μg/ml PHA (±2 ng/ml IL-12), in the presence of either DPP4-specific or non-targeting Accell siRNAs pools (Dharmacon). To achieve a partial gene silencing we used a commercial Accell SMART pool of 4 short interfering RNA (siRNA) designed to target the mRNA encoded by the human DPP4 gene (ref. E-004181-00-0005; Dharmacon); these siRNAs were designed to minimize the off-target effects. Besides, we also used two non-target siRNAs: a) an Accell green non-targeting siRNA (ref. D-001950-01-05; Dharmacon), which is a fluorescent unspecific siRNA used for assessment of Accell siRNA passive delivery effectiveness; b) a negative control Accell non-targeting siRNA pool of four siRNAs (ref. D-001910-10-05; Dharmacon) to control the background response to siRNA. All these siRNAs were initially resuspended at 100 μM by using a 1X siRNA buffer (ref. B-002000-UB-100; Dharmacon). Strikingly, the working concentration suggested by the manufacturer (1 μM) induced high cell mortality. Therefore, we titrated down the concentration of siRNA by using the Accell green non-targeting siRNA. A final concentration of 0.02 μM was selected to carry out the final experiments. This concentration was enough to label > 98% of cells and allow a high cell viability (>95%).
Upon 6 days of in vitro culture, CFSE fluorescence and the number of cell divisions were measured by flow cytometry (Supplementary Fig. S6), as well as the amount of CD26 protein with both specific anti-CD26 and isotype antibodies. Each condition was tested several times (n = 3-5 technical replicates). Unlabelled cells served as negative controls in cell proliferation assays. The calculated responder frequency (Rf) is the percentage of responder T cells that divided at least once ( Supplementary Fig. S6).
Statistics. Descriptive data are presented as median (interquartile range; IQR1-3). To assess the changes between asthmatic groups, rhinitis, and healthy donors for non-normally distributed variables we used the Kruskal-Wallis test followed by Dunn's multiple comparison test. Spearman's test was used to measure association between these variables. For CFSE proliferation studies, a two-way ANOVA followed by a Tukey's multiple comparison test were used. Finally, t-test was performed with data from Fig. 2c to assess changes in normally distributed variables between moderate-severe allergic asthmatics and healthy subjects. Mann-Whitney U test was used to assess changes in non-normally distributed variables. All analyses were conducted using GraphPad Prism 6.0 (GraphPad Software, Inc., San Jose, California, USA). The statistical significance was defined as P < 0.05.

Data Availability
The datasets analyzed are available from the corresponding author on justifiable request, and not publicly available due to protection of participant confidentiality.