Research ArticleAsthma

TH2 and TH17 inflammatory pathways are reciprocally regulated in asthma

See allHide authors and affiliations

Science Translational Medicine  19 Aug 2015:
Vol. 7, Issue 301, pp. 301ra129
DOI: 10.1126/scitranslmed.aab3142

A tale of two asthmas

Classifying diseases according to symptoms is rapidly becoming a thing of the past. Targeted therapeutics have shown us that sets of symptoms can be caused by different pathogenic mechanisms. Now, Choy et al. demonstrate that asthma can be divided into three immunological clusters: TH2-high, TH17-high, and TH2/17-low. The TH2-high and TH17-high clusters were inversely correlated in patients. Moreover, neutralizing one signature promoted the other in a mouse model of asthma. These data suggest that combination therapies targeting both pathways may better treat asthmatic individuals.


Increasing evidence suggests that asthma is a heterogeneous disorder regulated by distinct molecular mechanisms. In a cross-sectional study of asthmatics of varying severity (n = 51), endobronchial tissue gene expression analysis revealed three major patient clusters: TH2-high, TH17-high, and TH2/17-low. TH2-high and TH17-high patterns were mutually exclusive in individual patient samples, and their gene signatures were inversely correlated and differentially regulated by interleukin-13 (IL-13) and IL-17A. To understand this dichotomous pattern of T helper 2 (TH2) and TH17 signatures, we investigated the potential of type 2 cytokine suppression in promoting TH17 responses in a preclinical model of allergen-induced asthma. Neutralization of IL-4 and/or IL-13 resulted in increased TH17 cells and neutrophilic inflammation in the lung. However, neutralization of IL-13 and IL-17 protected mice from eosinophilia, mucus hyperplasia, and airway hyperreactivity and abolished the neutrophilic inflammation, suggesting that combination therapies targeting both pathways may maximize therapeutic efficacy across a patient population comprising both TH2 and TH17 endotypes.


Asthma is a chronic disorder, characterized by episodic airway hyperresponsiveness (AHR) and remodeling with variable degrees of eosinophilic and neutrophilic inflammation. Asthma causes significant morbidity and mortality (14), and about 10% of patients have disease that is resistant to current therapies (1, 4). This group consumes 50 to 60% of health care costs attributed to asthma, underscoring the necessity to discover new therapies (5).

The clinical expression of asthma is heterogeneous with several distinct phenotypes identified (6, 7). Identifying the molecular mechanisms driving subtypes of asthma has the potential to reveal drug targets, biomarkers to predict treatment response, and appropriately target therapy, as evidenced by recent clinical studies of T helper 2 (TH2) cytokine antagonists (8, 9).

In addition to the TH2 pathway, attention has focused on TH17 cytokines as candidate alternative drivers of severe asthma pathophysiology (10). Interleukin-17A (IL-17A) and IL-17F can amplify selected nuclear factor κB (NF-κB)–dependent signaling pathways such as those induced by tumor necrosis factor–α (TNF-α), a cytokine up-regulated in asthmatic airways, and are further up-regulated after allergen challenge and experimental rhinovirus infection (1117). In particular, IL-17A may contribute to neutrophilic airway inflammation via up-regulation of CSF3 and CXCL chemokines (1820), mucus gland hyperplasia, AHR (18, 21), and corticosteroid resistance (18, 22).

Therapeutic strategies targeting TH2 and TH17 inflammatory pathways are currently under active investigation in asthma. However, the nature and extent of the activity of these two pathways in individual patients are unclear. TH2 cytokines can negatively regulate TH17 cytokine expression, and inhibiting TH2 cytokines in vitro or in vivo has the potential to increase IL-17A production and IL-17A–dependent airway inflammation (23, 24). The crosstalk between TH2 and TH17 pathways is therefore complex, and it has been proposed that targeting TH2 cytokines might promote corticosteroid-resistant IL-17–dependent neutrophilic airway inflammation (10, 24). Here, we show that TH2- and TH17-related gene expression signatures are mutually exclusive in the airways of asthma patients, but both are associated with eosinophilic inflammation. In an in vivo preclinical model, we show that therapeutic targeting of TH2 and TH17 cytokines can lead to amplification of activity of the opposing pathway.


IL-13 and IL-17A induce distinct gene expression patterns in bronchial epithelial cells

To define core sets of IL-13– and IL-17–inducible transcripts for analysis in bronchial tissue, we stimulated normal human bronchial epithelial (NHBE) cells cultured at the air-liquid interface (ALI) with IL-13 (10 ng/ml) or IL-17A (10 ng/ml) ± TNF-α (10 ng/ml), and assessed the expression levels of transcripts associated with IL-13 and IL-17A signaling (19, 25, 26). We confirmed the IL-13–dependent expression of POSTN, CLCA1, and SERPINB2 (Fig. 1), which have been previously described as highly correlated with IL13 and IL5 expression and eosinophilic airway inflammation (25, 26) in asthmatic bronchial epithelial brushings. IL-17A and IL-17F induce the expression of the neutrophil hematopoietic factor CSF3 and the neutrophil chemoattractants CXCL1, CXCL2, CXCL3, and IL8 (19). The expression of CSF3, CXCL1, CXCL2, CXCL3, and IL8 was modestly inducible by IL-17A alone or more robustly in combination with TNF-α (Fig. 1). This is in keeping with existing literature and relevant to asthma, where TNF-α is expressed at elevated levels (1117). Relative to unstimulated cells, IL-13 stimulation suppressed the expression of IL-17A–inducible transcripts. To a lesser degree, IL-17A suppressed the expression of IL-13–inducible transcripts.

Fig. 1. IL-13 and IL-17A induce TH2 and TH17 signature genes in NHBE cells.

NHBE cells were grown and differentiated at ALI and then stimulated for 24 hours with IL-13 (10 ng/ml), IL-17A (10 ng/ml), TNF-α (10 ng/ml), or IL-17A (10 ng/ml) and TNF-α (10 ng/ml) (n = 3 technical replicates). Gene expression was assessed by quantitative polymerase chain reaction (qPCR). Differential expression is represented by heatmap of averaged replicates of untreated control zero-centered and scaled values.

Transcriptional analysis of bronchial biopsies based on TH2 and TH17 signatures identifies three distinct subgroups of asthma patients

To assess the relative levels and relationships between transcripts inducible by TH2 or TH17 cytokines, we analyzed previously reported gene expression microarrays of endobronchial biopsies from 51 asthma patients spanning a range of severity and corticosteroid use (27). Consistent with the in vitro stimulation experiments, the three IL-13–inducible genes comprising the TH2 signature, and the five IL-17–inducible genes comprising the TH17 signature were highly intercorrelated (fig. S1). TH2 and TH17 gene expression scores were calculated on the basis of average values of zero-centered gene expression values. The expression levels of canonical TH2 and TH17 cytokines, IL13 and IL17A, measured by qPCR were generally low, as 93% (39 of 42) and 76% (32 of 42) of available samples, respectively, had undetectable levels (<LLOQ, table S1). However, TH2 and TH17 scores were significantly elevated (Fig. 2, A and B) in samples in which IL13 or IL17A mRNA could be detected by qPCR.

Fig. 2. Three distinct asthma subgroups are defined by mutually exclusive TH2 and TH17 signature expression.

(A and B) The expression of endobronchial biopsy (A) TH2 and (B) TH17 signature scores is significantly elevated (P < 0.05, Kruskal-Wallis) in samples with detectable IL13 and IL17A, respectively. (C) TH2 and TH17 scores from 51 asthmatic bronchial biopsy microarrays are plotted by scatterplot. TH2 and TH17 signature scores are inversely correlated (Spearman’s ρ = −0.35, P = 0.011). Clustering TH2 and TH17 scores was used to classify subjects as TH2 score dominant (TH2-high, red), TH17 score dominant (TH17-high, blue), and TH2/17 score low (TH2/17-low, gray).

There was a significant negative correlation between TH2 and TH17 gene expression scores (Fig. 2C, Spearman’s ρ = −0.35, P = 0.011). In keeping with the modest reciprocal suppression of TH2 signature transcripts by IL-17A + TNF-α and TH17 signature transcripts by IL-13 in NHBE cells (Fig. 1), TH2-high and TH17-high gene signatures in bronchial biopsies were mutually exclusive: there were no patients who were simultaneously TH2-high and TH17-high. Similarly, subjects for whom IL13 mRNA could be detected had undetectable levels of IL-17A mRNA and vice versa (table S1).

Next, we classified subjects on the basis of their TH2 and TH17 scores. We used a two-step cluster analysis using the log likelihood method based on their factor loading score for TH2 or TH17 scores. Fifteen models were evaluated using different numbers of clusters between 1 and 15. The best fitting model, and thus the number of clusters, was determined by first calculating the Bayesian information criterion. On the basis of this approach, we identified three clusters as optimally fitting the data structure that appeared to capture patients with either TH2 score dominant (“TH2-high” asthma), TH17 score dominant (“TH17-high” asthma), or TH2/TH17 score low (“TH2/17-low” asthma) (Fig. 2C).

TH2-high and TH17-high molecular phenotypes are both associated with eosinophilic inflammation

TH2-high steroid-naïve asthma is associated with eosinophilic inflammation (25, 28), whereas TH17 signature genes include potent neutrophil chemoattractants (29, 30). We therefore hypothesized that molecular phenotypes of TH2 and TH17 inflammation would relate to physiologic measures of eosinophilic and neutrophilic inflammation, respectively. Unexpectedly, we observed evidence of eosinophilic inflammation in both TH2-high and TH17-high asthma.

Fractional exhaled nitric oxide (FeNO) and sputum eosinophils were significantly increased in TH2-high asthma compared to TH17-high and TH2/17-low asthma (Fig. 3, A and B). Median blood eosinophil counts were significantly elevated in TH2-high versus TH2/17-low asthma, but were also elevated in TH17-high asthma, though not statistically significant versus TH2/17-low asthma (Fig. 3C). Lamina propria eosinophil counts in bronchial biopsies were comparable in both TH2-high and TH17-high asthma, and both were significantly elevated versus TH2/17-low asthma (Fig. 3D). We assessed serum periostin, a systemic biomarker of eosinophilic airway inflammation (31), and IL-13 activity (32) in subjects for whom serum and bronchial biopsies were collected contemporaneously. Unexpectedly, serum periostin concentrations were highest in TH17-high asthma as compared to TH2-high and TH2/17-low asthma, but not statistically significant. There were no significant differences or associations between measures of neutrophilic inflammation in peripheral or airway compartments and molecular phenotypes of TH2 or TH17 inflammation (fig. S2).

Fig. 3. TH2-high and TH17-high asthma are both associated with elevated levels of physiologic TH2/eosinophilic measures of inflammation.

Pairwise comparisons (Mann-Whitney test) were made among TH2/17-low (Low), TH2-high (TH2), and TH17-high (TH17) asthmatics subjects. Red font indicates P < 0.05. (A and B) TH2-high subjects were associated with elevated (A) FeNO and (B) sputum eosinophil percentage as compared to low-inflammatory and TH17-high asthma. (C) Blood eosinophils were elevated in TH2-high versus low-inflammatory asthma. (D) Lamina propria eosinophil counts were elevated in both TH2-high and TH17-high versus low-inflammatory asthma. (E) Median levels of serum periostin were highest in TH17-high versus TH2-high and TH2/17-low asthma.

The TH17-high molecular phenotype is associated with corticosteroid-dependent moderate to severe asthma

The clinical characteristics of the patients separated by molecular phenotypes of TH2/17 inflammation are shown in table S2. We did not detect significant differences in patient characteristics or clinical measures including FEV1 (forced expiratory volume in 1 s) % predicted, FEV1/FVC (forced vital capacity) ratio, AHR, or exacerbation frequency. However, mean values of FEV1 % predicted and FEV1/FVC ratio were lowest in the TH17-high group.

Although there was a range of doses of inhaled (ICS) or oral (OCS) corticosteroids across all three TH2/17 molecular phenotypes, the TH17 signature was not observed in steroid-naïve patients (table S2). Integrating this observation with the associations between TH2-high and TH17-high asthma and eosinophilic inflammation (Fig. 3), we have delineated subsets where molecular phenotypes of TH2-high and TH17-high asthma are observed (Fig. 4). TH2-high asthma was observed only in subjects with evidence of eosinophilic asthma defined as blood eosinophils ≥300/μl or sputum eosinophils ≥3% or biopsy lamina propria eosinophils ≥10/mm2. TH2-high, TH17-high, and TH2/17-low phenotypes were observed among eosinophilic moderate-severe asthma in approximately equal proportions. The molecular phenotype of TH2/17-low asthma was observed among all subsets and was predominant among moderate-severe asthmatics using ICS who did not display any evidence of eosinophilia. Finally, TH17-high asthma was observed only in moderate to severe asthma patients, primarily in subjects who had evidence of eosinophilia and were taking ICS or ICS + OCS.

Fig. 4. Proportions of TH2/17 molecular phenotypes of asthma by clinical evidence of eosinophilia and severity.

Asthmatics were classified on the basis of evidence of eosinophilic asthma, defined as blood eosinophil count ≥300/μl or sputum eosinophil percentage ≥3, or biopsy lamina propria eosinophil count ≥10/mm2 and clinical severity (British Thoracic Society/Scottish Intercollegiate Guidelines Network treatment step). Pie charts represent the proportion of TH2/17-low, TH2-high (TH2), and TH17-high (TH17) asthmatic subjects per eosinophilic/severity category. The area of each pie chart is proportional to the number of subjects in that category.

Therapeutic blockade of TH2 cytokines during experimental allergic asthma induces TH17 inflammation

Considering the mutually exclusive TH2 and TH17 signatures identified in human asthmatic airways, and the potential for IL-13 suppression to promote IL-17 production in vitro (23), we investigated the effects of IL-4 and IL-13 blockade, singly or in combination, in a murine model of allergic asthma. These studies consisted of 3 weeks of intranasal sensitization against house dust mite (HDM) extract and 4 weeks of biweekly intranasal HDM challenges with antibody therapy targeting IL-4, IL-13, or both cytokines on the preceding day (Fig. 5A). Anti–IL-4, anti–IL-13, and anti–IL-4/13 reduced airway inflammation (Fig. 5B). Mice receiving anti–IL-13 and anti–IL-4/13 were significantly protected from arteriolar hypertrophy and fibrosis, whereas mice treated with anti–IL-4 were less protected from fibrosis and still had evidence of arteriolar hypertrophy (Fig. 5, B and C). The mucus response assessed by histological analysis and lung Muc5ac expression was significantly abrogated by anti–IL-13 or anti–IL-4/13 treatment, but less so by anti–IL-4 alone (Fig. 5D and fig. S3A). Anti–IL-4 and anti–IL-4/13 reduced lung IL13 expression and CD4+IL-13+ cells in lung and lymph node (Fig. 5E and fig. S3B). All anti-TH2 interventions resulted in significantly increased tissue expression of IL17 and CD4+IL-17+ cells in lung and lymph node, particularly anti–IL-4 (Fig. 5F and fig. S3B). Furthermore, downstream eosinophilic and neutrophilic infiltration reflected this shift in TH2/TH17 effector cytokines. Anti–IL-4– or anti–IL-4/13-treated mice had significantly reduced eosinophil numbers in tissue, whereas anti–IL-13 alone produced a significant but less marked reduction (Fig. 5E and fig. S3C). Meanwhile, we found marked increases in neutrophil infiltration in lungs of all treated mice, most pronounced with anti–IL-4 (Fig. 5F). Furthermore, expression of TH2-driven genes, Clca3 (ortholog for human CLCA1) and Ccl11, was substantially reduced with anti–IL-13 or anti–IL-4/13 treatment (Fig. 5E). Despite effectively reducing markers of the TH2 response, anti–IL-13 did not affect tissue expression or CD4 staining for IL-13 and may have induced greater IL-13 expression. These observations are consistent with previous reports that IL-13 serves as an important regulator of downstream TH2 effector functions rather than affecting TH2 cell differentiation, and may be explained in part by an altered balance in receptor complex availability for IL-4 in the context of IL-13 blockade (3336). IL-17–responsive genes Cxcl1, Cxcl3, and Csf3 were increased in anti–IL-4–treated animals, and Csf3 was increased by both anti–IL-13 and anti–IL-4/13 (Fig. 5F). Notably, the increases in neutrophils, IL-17, and IL-17–responsive gene expression were modest or nonexistent in untreated mice compared to saline controls, suggesting that a robust TH17 response only occurred with perturbation of the TH2 pathway.

Fig. 5. Therapeutic blockade of TH2 cytokines during experimental allergic asthma induces compensatory TH17 inflammation.

(A) Mice were intranasally (i.n.) sensitized against HDM extract for 3 weeks at 200, 100, and 100 μg each week and subsequently challenged with biweekly intranasal exposures for 4 weeks with or without targeted antibody therapy each day before challenge against IL-4, IL-13, or both cytokines [250 μg, intraperitoneally (I.P)]. (B and C) Lung pathology was assessed by scoring for gross inflammation in Giemsa-stained histological sections (B), assaying for hydroxyproline content as a surrogate for fibrosis (B), and Masson’s trichrome staining of airways (C). (D) Airway mucus production was assessed by periodic acid–Schiff (PAS) staining. (E and F) IL-13 (E) and IL-17 (F) production was measured by gene expression and in restimulated CD4 T cells by flow cytometry. The percent of eosinophils and neutrophils was quantified from Giemsa-stained histological lung sections. (F) Whole lung tissue expression of Clca3 and Ccl11 TH2 markers, and Cxcl3, Cxcl1, and Csf3 TH17 markers were assessed by qPCR (P values shown, two-tailed t tests, n = 4 to 10).

Dexamethasone suppresses TH2 pathways but enhances TH17 pathways in the mouse asthma model

Given the association of corticosteroid use with patients exhibiting a TH17-high signature, we interrogated the murine HDM model to determine whether corticosteroid therapy alone might contribute to features of TH17-high asthma. Dexamethasone treatment reduced pathophysiologic features including AHR, tissue fibrosis, eosinophil infiltration, and IL-4/13 expression. Dexamethasone treatment partially suppressed IL-17 mRNA expression but increased the frequency of neutrophils in the lung concomitant with marked up-regulation in tissue expression of the IL-17–inducible chemokines Cxcl3 and Cxcl1 (fig. S4). In addition to recapitulating observations from published studies that corticosteroid treatment effectively suppresses TH2-driven inflammation, but fails to significantly reduce IL-17–driven inflammation (18), these data add that corticosteroid use may specifically contribute to creating an environment permissive to an enhanced neutrophilic response.

Combined blockade of IL-13 and IL-17 attenuates the TH17 signature observed during anti–IL-13 inhibition of the TH2 response

The reciprocal regulation of TH2 and TH17 responses, both in vivo and in epithelial cells in vitro, provides an explanation for the mutually exclusive TH2 and TH17 signatures in a large proportion of patients with asthma (Figs. 2 and 4). This could be important clinically when targeting these cytokine pathways, because efficacy-limiting compensation may occur if only a single pathway is blocked. Therefore, we tested whether combined IL-13 and IL-17 antibody blockade would reduce TH2-driven disease and concomitantly inhibit the TH17 response. Mice treated with either anti–IL-13 alone or anti–IL-13 + anti–IL-17 had significantly reduced AHR and mucus response relative to control antibody–treated animals or anti–IL-17 alone (Fig. 6, A and B). The increased IL-17 production and neutrophil infiltration associated with anti–IL-13 treatment alone were partially and completely abrogated, respectively, in mice receiving dual anti–IL-13/anti–IL-17 therapy (Fig. 6C). Furthermore, we detected increased IL-4– and IL-13–producing CD4+ T cells in the lungs of anti–IL-17–treated animals (fig. S5), supporting dual cross-regulation between the TH2 and TH17 cytokine networks.

Fig. 6. Dual therapeutic blockade of IL-13 and IL-17 prevents emergence of a TH17 signature induced by antibody inhibition of the TH2 response.

Mice on the chronic HDM model of allergic asthma were treated a day before each challenge with anti–IL-17, anti–IL-13, or both (150 μg). (A and B) Efficacy of IL-13/17 dual blockade was assessed by airway hyperreactivity [two-way analysis of variance (ANOVA), SEM shown, n = 5 to 10] (A) and mucus response gene expression (B). (C) Immune response was measured by gene expression analysis of IL-13 and IL-17, and flow cytometric analysis of the frequency of Siglec-F+ eosinophils and Ly6G+ neutrophils in bronchoalveolar lavage (BAL) (P values shown, two-tailed t tests, n = 5 to 28).


Until recently, asthma was considered as a single disease entity associated with eosinophilic airway inflammation driven by type 2 inflammatory cytokines (IL-4/5/13) (37). However, asthma can occur in the absence of significant TH2/eosinophilic inflammation across the spectrum of severity (25, 3842). There is increasing interest in the role of TH17-dependent pathways in asthma, but it is not known how TH2 and TH17 pathways interact in asthmatic airways. With the potential for counterregulation already identified, it is important to understand the consequences of singly inhibiting TH2 and TH17 pathways.

Here, we used TH2- and TH17-related gene signatures and individual gene expression data to measure the activity of TH2 and TH17 pathways in human asthmatic airways, and HDM-sensitized murine airways. The analysis of TH2/17-dependent gene transcription adds important value by providing evidence of relevant cytokine activity. This is exemplified by the previously described TH2 gene signature, which is robustly expressed in a subset of asthmatic subjects (25, 26), correlates with airway eosinophilia, and has precipitated the development of periostin as a biomarker for predicting the response to TH2-targeted therapy (9, 31). Using this approach, we have validated a TH17-dependent gene signature (CXCL1, CXCL2, and CXCL3, IL8, and CSF3) and have shown that the induction of these genes is further amplified by TNF-α, an important mediator associated with IL-17 inflammation, that has been described to be up-regulated in the airways in many studies of asthma (1117, 4346). We also confirmed in NHBE cells that IL-13 and IL-17 reciprocally regulate each other’s respective signatures. Using this approach, we have shown that IL-17–related signaling is evident in the airway tissue of a subset of patients with moderate-severe asthma using corticosteroids. Intriguingly, TH2 and TH17 activity was inversely correlated, and clustering patients demonstrated that TH2-high and TH17-high disease were mutually exclusive. This suggests that there is reciprocal regulation of these two pathways in vivo in human asthma, which is consistent with the reciprocal relationship we found for TH2 and TH17 signature genes induced by IL-13 and IL-17A in vitro, and the ability for IL-13 to attenuate IL-17A production in human TH17 cells (23). However, this interpretation is limited by the inherent restrictions of studying human subjects cross-sectionally.

To examine this potential counterregulation in more detail, we turned to an in vivo HDM-driven murine asthma model. This demonstrated that with a strong TH2 stimulus, there is also IL-17 induction. However, the downstream consequences of IL-17 induction appeared to be relatively limited because there were no changes in airway neutrophils and the IL-17–dependent cytokines Cxcl1 and Cxcl3 did not increase significantly. However, although IL-13 and IL-4/13 blockade in combination mitigated a wide array of pathological consequences to HDM exposure, these same interventions enhanced IL-17 expression, IL-17–dependent chemokine/cytokine expression, and lung neutrophilia. This confirms that TH2 cytokines are powerful suppressors of IL-17–driven inflammation, and also raises the possibility that TH2-targeted treatment in asthma may contribute to the emergence of an adverse TH17-permissive environment, limiting therapeutic efficacy over time. Given recent interest in IL-17 as an alternative driver of asthma, and the proposed idea that targeting TH2 cytokines might promote corticosteroid-resistant IL-17–dependent neutrophilic airway inflammation (10, 24), these observations highlight the importance of understanding the regulation and interplay of TH2 and TH17 responses for developing and optimizing therapeutic intervention strategies.

Our identification of mutually exclusive TH2 and TH17 expression may seem contradictory with reports of dual TH2 and TH17 cytokine–expressing CD4+ T cells (47, 48). However, the demonstration that IL-13 strongly represses IL-17A–dependent genes in epithelial cells suggests that in the context of coexpression, IL-17–dependent transcription will be attenuated. Our data from human bronchial biopsies support this interpretation.

Because the TH17 signature genes include potent neutrophil chemoattractants (29, 30), we hypothesized that TH17 signature expression in patients would relate to measures of neutrophilic airway inflammation, as we described previously for inflammatory dermatoses (49). As observed previously, TH2-high asthma was associated with eosinophilic inflammation (25, 28). Unexpectedly, TH17-high asthma was also associated with elevated numbers of lamina propria eosinophils, and there were no relationships in blood, sputum, or lamina propria neutrophils among TH2 or TH17 molecular phenotypes. However, tissue neutrophil counts are reported to be similar between health and asthma (27, 39, 5052), and their activity may be more important than their number. Nevertheless, these observations have important implications for the selection of patients in clinical trials of anti–IL-17 therapy. Indeed, anti–IL-17RA therapy in symptomatic moderate-severe asthmatics using ICS failed to demonstrate efficacy in a recent study, which may not have been appropriately stratified to assess subsets of patients with activity of the IL-17 pathway (53). An ongoing study of another anti–IL-17 therapy in moderate-severe asthma excludes patients with elevated blood eosinophils and, hence, may exclude patients with the potential for benefit (54).

An important question arises as to the stability of the TH2 and TH17 phenotypes and whether patients can move from one to the other. Indirect evidence from the expression of periostin suggests that this is a likely scenario. Periostin gene expression is directly inducible by IL-13 but not IL-17A (Fig. 1) (42), elevated serum periostin levels predict clinical benefit from IL-13 inhibition (9), and serum periostin concentrations in anti-IL-13–treated moderate-severe asthma patients fall to levels observed in healthy nonasthmatic subjects (32). Furthermore, IL-13 induces NOS2 expression, contributing to elevated FeNO in asthma patients (55), which also predicts benefit from and is decreased by anti–IL-13 treatment (9, 32). We observed elevated FeNO in TH2-high but not in TH17-high asthma, suggesting that FeNO reflects airway IL-13 activity at a given point in time, whereas the turnover times for serum periostin and tissue eosinophils may be longer. Together, the elevated serum periostin and bronchial tissue eosinophilia but low FeNO in TH17-high asthma suggest that moderate-severe asthma may alternate between TH2-high, TH17-high, and TH2/17-low states depending on recent exposures to immunostimulatory factors. This cannot be determined formally due to the cross-sectional nature of the present study, but future studies should examine these patterns over multiple longitudinal samples and exposures.

We observed TH17-high asthma exclusively in corticosteroid-treated moderate-severe asthma (Fig. 4), consistent with the demonstration that corticosteroids may promote IL-17 production in some patients (56, 57). In addition, dexamethasone intervention in the HDM model potently inhibited disease and the TH2 response, but resulted in enhanced markers of TH17 inflammation despite reducing IL-17. This could be the result of experimental timing or potentiated signaling from residual IL-17 due to deregulation of the TH2 inhibitory pathway. Together, our data suggest that inhibition of TH2 responses and the concomitant loss of IL-17 regulation resulting from corticosteroid exposure or selective TH2 inhibition in asthmatic airways may create a TH17-permissive environment. Subsequent exposure to frequently encountered exogenous stimuli such as allergen, infection, pollution, and perhaps corticosteroids themselves may then enhance IL-17 expression. The same environment would exist in true TH2-low patients who should also be susceptible to IL-17 up-regulation (summarized in fig. S6). In our preclinical mouse model, combined treatment with anti–IL-13/17 ameliorated AHR, lung pathology, and the TH2 response, and markedly reduced the IL-17–dependent chemokine and neutrophil response observed with anti–IL-13 treatment alone. Thus, optimal therapeutic benefit in asthma may be achieved through simultaneous TH2 and TH17 pathway–directed therapy. However, although therapies targeting TH2 cytokines have shown efficacy in subsets of patients with moderate-severe asthma (58), formal assessments of the impact of IL-4, IL-5, and/or IL-13 inhibition on TH17 activity have not been described in those studies and represent an area for future investigation.

There were limitations to this study that may serve as subjects of further investigation or alternative interpretations. This post hoc analysis was based on a cross-sectional study and is unable to assess the longitudinal intrapatient variability of TH2/17 signature classification. The number of asthmatics (n = 51) and the relative paucity of subjects who are “non-eosinophilic” limit our ability to evaluate this population where current and emerging therapies are unlikely to provide meaningful clinical benefit. Although our designated “TH17 signature” is consistent with the biological activity of IL-17A and/or IL-17F, we cannot exclude the possibility that other cytokines are contributing to it in vivo. Although TH2 and TH17 signature expression was statistically significantly up-regulated (Fig. 2, A and B) in subjects in whom IL13 or IL17, respectively, was expressed, caution must be made in the interpretation of these data due to the high percentage of subjects with undetectable cytokine levels. Furthermore, assessing this gene expression pattern in endobronchial biopsies depends on invasive procedures and precludes routine use, highlighting the need to identify and develop noninvasive biomarkers of this TH17 pathway activity.

In summary, we have identified TH2-high, TH17-high, and TH2-17-low clusters of patients with asthma. We propose that with suppression of TH2 activity by targeted therapy or corticosteroids, or absent TH2 activity (true TH2-low), a TH17-permissive environment exists. Combined with data from a murine model of allergic asthma, we suggest that in a subset of patients, a direct relationship between TH17-high and TH2-high disease exists, whereby, through mutual cross-regulation, TH17-high asthma may represent a transition or switch away from TH2-mediated disease. Our studies therefore suggest that combined targeting of IL-13 and IL-17 in patients expressing either a TH2 or a TH17 signature could provide additional efficacy over single TH2 or TH17 inhibition.


Study design

The overall goals of this study were to identify gene signatures associated with IL-13– and IL-17–driven inflammation and then use these signatures to characterize the regulation and interaction of these cytokines in asthma and their association with commonly studied biomarkers of asthmatic disease. Candidate signature associated genes were identified via cytokine stimulation of human bronchial epithelial cells in vitro, and then a post hoc analysis was conducted in patient bronchial biopsies. To evaluate these pathways in the context of therapeutic blockade, we used a chronic HDM model of murine asthma with therapeutic anti-cytokine antibody treatment. In the murine studies, sample sizes were determined on the basis of previous experience and previous statistical analyses of the model. In the murine studies, standard measurements were used to assess inflammation, gene expression, and airway hyperreactivity. Murine histology was scored by blinded observers. Sample size and replicates for all mouse studies are included in the figures and legends.

NHBE cell culture and stimulation

Primary NHBE cells were purchased from Lonza. Transwell plates (6.5-mm diameter, 0.4-μm pore density; Corning Life Sciences) were collagen-coated using PureCol (100 μg/ml) from Advanced BioMatrix. NHBE cells were seeded in Transwells and maintained in serum-free bronchial epithelial cell growth medium (Lonza) for 96 hours or until confluent. Thereafter, the apical medium was removed, and cells were fed basolaterally with PneumaCult-ALI complete medium (Stemcell) and differentiated for a period of 21 days. Differentiated NHBE cells were cultured alone in ALI culture medium or stimulated for 24 hours with IL-13 (10 ng/ml), IL-17A (10 ng/ml), or TNF-α (10 ng/ml), alone or in combinations thereof (n = 3 technical replicates). Total RNA was extracted from NHBE cells using the Qiagen RNeasy Kit.

Gene expression analyses

RNA was isolated from homogenized bronchial biopsies, and real-time qPCR was performed as described previously (28). Whole-genome expression microarrays from asthmatics (n = 51) (27) were analyzed. Microarray-based TH2 and TH17 score was calculated by case-wise averaging of zero-centered gene expression data after annotation-based independent filtering (59), that is, if multiple probes correspond to an Entrez gene, the probe with the highest interquartile range was selected. SERPINB2, CLCA1, and POSTN were used as IL-13–responsive TH2 signature genes as described previously (25, 26). CXCL1, CXCL2, CXCL3, IL8, and CSF3 were used as IL-17–responsive TH17 signature genes.

Gene expression analyses from in vitro NHBE cell stimulation experiments and bronchial biopsy samples were conducted using TaqMan Gene Expression Assays that were purchased and conducted per the manufacturer’s instructions for CXCL1 (Hs00236937_m1), CXCL2 (Hs00601975_m1), CXCL3 (Hs00171061_m1), IL8 (Hs00174103_m1), CSF3 (Hs00738432_g1), SERPINB2 (Hs01010736_m1), CLCA1 (Hs00976287_m1), POSTN (Hs00170815_m1), IL13 (Hs99999038_m1), and IL17A (Hs99999082_m1). Target gene expression was normalized by housekeeping genes GAPDH (4333764F) and HPRT1 (Hs02800695_m1).

Patients and assessments

Patients were recruited from two centers, Leicester and Belfast, and their clinical data and tissues have been used in previously published studies (27, 60). Asthma severity was defined by British Guideline on the Management of Asthma treatment steps (61). Of the 26 severe patients at step 4/5, 21 met the American Thoracic Society criteria for refractory asthma (1).

The study was approved by the Research Ethics Committees of both institutions (04/Q2502/74, 06/NIR02/114). Written informed consent was gained from all participants before their involvement.

Physiologic measures are reported only for those derived during the bronchoscopic study. Consequently, FeNO, sputum eosinophil percentage, and serum periostin measurements are reported for Leicester subjects only.

Fiber-optic bronchoscopy

Subjects underwent bronchoscopy conducted according to the British Thoracic Society guidelines (62). The bronchial mucosal biopsy specimens were taken from the right middle lobe and lower lobe carinae, fixed in acetone, and embedded in glycol methacrylate (GMA), as described previously (63). The biopsies were also placed immediately in RNA preservative (RNAlater, Ambion) and before processing gene expression analysis.


The GMA-embedded tissue was cut and immunostained as described previously (63). Primary antibodies were used against the following antigens: eosinophil major basic protein and neutrophil elastase. Isotype controls were also performed.

Assessment and quantification of immunohistochemical staining

Epithelial and submucosal areas in sections were identified and measured using a computer analysis system (AnalySIS, Olympus). Numbers of positively stained nucleated cells in each compartment were counted blind. Cells staining in sequential sections were colocalized using computer analysis.

Serum periostin

Serum periostin levels were measured with a proprietary sandwich enzyme-linked immunosorbent assay (ELISA) with two monoclonal antibodies capable of detecting all known splice variants of human periostin as previously described (31).

Murine HDM model

Six- to 8-week-old female BALB/c mice obtained from Taconic Farms Inc. were sensitized by intranasal inhalation with 200 μg of HDM extract in 30 μl of sterile saline on day 0, and 100 μg on days 7 and 14. Challenges were repeated twice a week with 50 μg intranasally for 4 weeks starting on day 21. For therapies, 250 or 150 μg each of anti–IL-4 (11b11), anti–IL-13 (262A-5-1), anti–IL-17 (16H4.4F3), or control (10E7.1D2) antibody was injected intraperitoneally the day before intranasal challenges starting on day 20. Mice were terminally anesthetized on day 46 with sodium pentobarbital. All animals were housed under specific pathogen–free conditions at the National Institutes of Health (NIH) in an American Association for the Accreditation of Laboratory Animal Care–approved facility.

Ethics statement

The National Institute of Allergy and Infectious Diseases (NIAID) Division of Intramural Research Animal Care and Use Program, as part of the NIH Intramural Research Program, approved all of the experimental procedures (protocol LPD 16E). The Program complies with all applicable provisions of the Animal Welfare Act ( and other federal statutes and regulations relating to animals.


Murine lung lobes were harvested and inflated/fixed with a Bouin’s-Hollande solution. Fixed tissue was embedded in paraffin for sectioning and stained with Wright’s Giemsa for airway inflammation and cellular infiltration analysis and PAS for assessing airway mucus. Inflammation, PAS staining, and cellular infiltrate were scored by a blinded observer.

Fibrosis quantification

Hydroxyproline was measured to determine collagen content by hydrolyzing a weighed lung lobe in 6 N HCl at 110°C overnight before neutralization with 10 N NaOH. Quantification was performed by colorization compared to a standard curve of hydroxyproline.

Cell isolation and flow cytometry

About 75 mg of lung tissue was diced and incubated in collagenase (100 U/ml) (Sigma) at 37°C for an hour with rocking. Tissue was then passed through a 100-μm nylon filter to obtain a single-cell suspension. Leukocytes were isolated on a 40% Percoll (Sigma) gradient and treated with ACK buffer to remove erythrocytes. Lymph nodes were passed over a 100-μm filter and treated with ACK. For BAL samples, the trachea of anaesthetized mice was cannulated and lungs were lavaged with 1 ml of 5 mM EDTA sterile phosphate-buffered saline. BAL samples were centrifuged and treated with ACK before fixation. Isolated cells from lung, lymph node, and BAL were either immediately fixed for cellular analysis or stimulated with phorbol 12-myristate 13-acetate (10 ng/ml) and ionomycin (1 μg/ml) in the presence of brefeldin A (10 μg/ml) for 3 hours and fixed. Cells were permeabilized (Cytofix/Cytoperm buffer; BD Biosciences) and stained for 30 min with antibodies for CD16/32, CD45, Siglec-F, Ly6G, Ly6C, CD11b, CD4, IL-4, IL-13, TNF-α, IFN-γ (interferon-γ), and IL-17.

Murine gene expression analyses

Lung tissue was homogenized in TRIzol Reagent (Life Technologies) with a Precellys 24 (Bertin Technologies). Total RNA was extracted with chloroform using a MagMax-96 Total RNA Isolation Kit (Qiagen) and reverse-transcribed to complementary DNA using SuperScript II Reverse Transcriptase (Life Technologies). Gene expression was quantified using Power SYBR Green PCR Master Mix (Applied Biosystems) by real-time PCR on an ABI Prism 7900HT Sequence Detection System (Applied Biosystems). Gene expression is described relative to RPLP2 mRNA levels in saline-challenged lung tissue.

Murine primer sequences
View this table:

Murine airway hyperreactivity measurement

Mice were analyzed for AHR on day 46 of the HDM exposure model by total body plethysmography (Data Sciences International–Buxco Research Systems). Mice were challenged with increasing doses of methacholine inhalation (3 to 50 mg/ml), and measurements collected over the following 5 min were used to calculate Penh readouts. Data were analyzed using FinePointe software (Buxco).

Statistical analysis

Prism (version 6; GraphPad), R Project software (version 2.15.1; (64), and SPSS statistics (version 20, IBM) were used for statistical analysis and graphing. Mann-Whitney or Kruskal-Wallis test was used for testing the dependence of continuous versus categorical variables. Spearman’s rank correlation was used to test for dependence between two numeric variables. Fisher’s exact test was used to test for dependence between categorical variables. Missing data were not imputed and treated as “missing completely at random.” Murine data were compared with a two-tailed t-test, with Welch’s correction when an F test comparing variances had a P value of <0.05, or a two-way ANOVA.


Fig. S1. Intercorrelation between the three IL-13–inducible genes and the five IL-17–inducible genes in endobronchial biopsies.

Fig. S2. Neutrophilic inflammation in peripheral or airway compartments is not associated with molecular phenotypes of TH2 or TH17 inflammation.

Fig. S3. Differential effects of TH2 blockade on mucous and inflammatory responses in allergic asthma.

Fig. S4. Dexamethasone suppresses TH2 pathways but enhances TH17 pathways in the mouse asthma model.

Fig. S5. Evidence of dual cross-regulation between the TH2 and TH17 cytokine networks.

Fig. S6. A summary of the potential interplay between TH2 (IL-4/13)– and IL-17–dependent signaling in asthmatic airways.

Table S1. Mutual exclusivity of IL-17 and IL-13 expression in endobronchial biopsies.

Table S2. Clinical characteristics of TH2/17-low, TH2-high, and TH17-high asthma.


  1. Funding: Supported in part by the Intramural Research Program of the NIH, NIAID. Work in Leicester was supported by grants from the Asthma UK project grant AUK-PG-2013-208 and a grant-in-aid from Genentech Inc., and was supported by the National Institute for Health Research (NIHR) Leicester Respiratory Biomedical Research Unit. The views expressed are those of the author(s) and not necessarily those of the National Health Service (NHS), the NIHR, or the Department of Health. Author contributions: T.A.W., K.M.H., P.B., L.C.W., D.F.C., and J.R.A. conceived the study; D.F.C., K.M.H., L.A.B., T.R.R., J.G.E., J.R.A., and P.B. designed the experiments; D.F.C., K.M.H., L.A.B., K.M.V., J.C.S., R.L.G., R.W.T., S.W., A.S., E.D., C.A.B., D.R.N., G.J., and J.J. performed the experiments; D.F.C., K.M.H., L.A.B., T.R.R., P.B., A.R.A., J.R.A., A.S., and S.S. analyzed the data; C.M.O., P.B., and L.G.H. performed bronchoscopy; B.H. recruited patients and performed the clinical assessments; D.F.C., K.M.H., T.A.W., P.B., J.R.A., and T.R.R. wrote the paper. All authors reviewed the manuscript for intellectual content and approved the final version. Competing interests: D.F.C., D.R.N., G.J., E.D., A.R.A., J.J., L.C.W., J.G.E., and J.R.A. are currently or were employees of Genentech Inc. during the execution of this study. P.B. and L.G.H. received funding from Genentech Inc. toward this work. The other authors declare no competing interests.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article