Mast Cell Phenotype, Location, and Activation in Severe Asthma

Abstract

Rationale: Severe asthma (SA) remains poorly understood. Mast cells (MC) are implicated in asthma pathogenesis, but it remains unknown how their phenotype, location, and activation relate to asthma severity.

Objectives: To compare MC-related markers measured in bronchoscopically obtained samples with clinically relevant parameters between normal subjects and subjects with asthma to clarify their pathobiologic importance.

Methods: Endobronchial biopsies, epithelial brushings, and bronchoalveolar lavage were obtained from subjects with asthma and normal subjects from the Severe Asthma Research Program (N = 199). Tryptase, chymase, and carboxypeptidase A (CPA)3 were used to identify total MC (MC_Tot) and the MC_TC subset (MCs positive for both tryptase and chymase) using immunostaining and quantitative real-time polymerase chain reaction. Lavage was analyzed for tryptase and prostaglandin D2 (PGD2) by ELISA.

Measurements and Main Results: Submucosal MC_Tot (tryptase-positive by immunostaining) numbers were highest in “mild asthma/no inhaled corticosteroid (ICS) therapy” subjects and decreased with greater asthma severity (P = 0.002). In contrast, MC_TC (chymase-positive by immunostaining) were the predominant (MC_TC/MC_Tot > 50%) MC phenotype in SA (overall P = 0.005). Epithelial MC_Tot were also highest in mild asthma/no ICS, but were not lower in SA. Instead, they persisted and were predominantly MC_TC. Epithelial CPA3 and tryptase mRNA supported the immunostaining data (overall P = 0.008 and P = 0.02, respectively). Lavage PGD2 was higher in SA than in other steroid-treated groups (overall P = 0.02), whereas tryptase did not differentiate the groups. In statistical models, PGD2 and MC_TC/MC_Tot predicted SA.

Conclusions: Severe asthma is associated with a predominance of MC_TC in the airway submucosa and epithelium. Activation of those MC_TC may contribute to the increases in PGD2 levels. The data suggest an altered and active MC population contributes to SA pathology.

AT A GLANCE COMMENTARY.

Scientific Knowledge on the Subject

Severe asthma remains poorly understood. Mast cells (MCs) are implicated in asthma pathogenesis, but it remains unknown how their phenotype, location, and activation relate to asthma severity.

What This Study Adds to the Field

This study reports changes in phenotype profile, distribution, and activity of airway resident MC populations from a large group of subjects with asthma and normal control subjects. The MC population in severe asthma is dominated by the chymase-positive phenotype in submucosa and more uniquely in the epithelium. This is associated with increased bronchoalveolar lavage fluid prostaglandin D2 (PGD2) levels in severe asthma as compared to other steroid-treated subjects with asthma. A greater proportion of chymase-positive MCs and increased PGD2 were identified as important predictors of severe asthma. Targeting PGD2 signaling pathways could be an option for treatment of severe asthma.

Severe asthma represents about 10 to 15% of the asthma population and yet remains poorly understood and treated (1, 2). Although a plethora of studies suggest alterations in mast cells (MCs) may underlie its pathogenesis, their specific role is unclear (3–8).

MCs are evolutionarily ancient cells with complex function, believed to differentiate dependent on the tissue microenvironment in which they reside. In humans, ultrastructural characteristics and differential immunohistochemical staining (IHC) for the MC-specific proteases tryptase and chymase have identified MC_T (tryptase-positive only) and MC_TC (both tryptase- and chymase-positive) phenotypes. In addition to chymase, MC_TC express cathepsin G and carboxypeptidase A3 (CPA3) (9–11). Normally, MC_T predominate in the lung, whereas MC_TC represent less than 20% of lung MCs. The distribution of MC phenotypes within a particular organ/tissue varies as well (3–8). In intestines of mice, the epithelial MC phenotype differs from that of cells in deeper mucosal regions and changes further during a parasitic infection (12). In humans, increases in nasal epithelial MCs in allergic rhinitis and after an allergen challenge have also been reported (13, 14). Although lower airway epithelial MCs have rarely been studied, expression of MC-specific mRNA in human airway epithelial cell brushings was reported to differentiate subjects with asthma from normal subjects (3, 15, 16).

Although general quantitative and phenotype changes in total MCs have been reported with conditions such as asthma and COPD, functional characteristics of IHC-identified MC_TC remain poorly understood (3–8, 17). In vitro studies show that the two MC phenotypes functionally differ. Chymase inhibitors block IgE-mediated histamine release from skin MCs (predominantly MC_TC) but not from “normal” lung MCs (predominantly MC_T) (18). Similar phenotypic differences in response to C5a and compound 48/80 are also seen (19, 20). In addition, generation of prostaglandin (PG) D2 may be more pronounced in MC_TC and not always associated with MC degranulation (21, 22). Finally, corticosteroid (CS) therapy in asthma may decrease MC_T, but not MC_TC (23).

Based on the cited studies, it was hypothesized that there would be a shift in MC phenotype with increasing asthma severity, such that an MC_TC-like phenotype would predominate in the airway submucosa and epithelium of subjects with severe asthma as compared with subjects with milder asthma and normal subjects. This deviation in MC phenotype and distribution would associate with an activation profile more typical of an MC_TC than an MC_T. To investigate this hypothesis, endobronchial tissue samples collected across the National Heart, Lung, and Blood Institute's Severe Asthma Research Program (SARP) from a range of subjects with asthma and normal control subjects were evaluated for: (1) submucosal and epithelial distribution of MC_Tot and MC_TC, (2) mRNA expression of their enzymes, and (3) bronchoalveolar lavage (BAL) fluid levels of tryptase and prostaglandin D2 (PGD2) as a measure of MC-specific activity. Statistical models were applied to determine the relative importance of MCs and their phenotypes and activation in predicting the clinical severity of asthma.

Some of the results have been previously reported in the form of abstracts (24, 25).

METHODS

Subjects

Males and females between 18 and 65 years of age who smoked less than five total pack-years and none in the last year were included in the study. After signing informed consent, subjects completed comprehensive questionnaires, underwent baseline and post-bronchodilator spirometry, allergen skin prick testing, blood draw, and bronchoscopy that included airway mucosa biopsy, epithelium brushing, and BAL. Subjects were recruited from seven SARP sites. SARP is a loose network of site-focused studies that contribute samples and data to a central database. However, the specific interests of the sites remain distinct. The SARP sites in this study include the Brigham and Women's Hospital, Cleveland Clinic, National Jewish Health/University of Pittsburgh (site moved from Denver to Pittsburgh in 2006), University of Virginia, University of Wisconsin, Wake Forest University, and Washington University.

Normal subjects (NC) were healthy atopic or nonatopic individuals with normal lung function and without a history of chronic respiratory conditions, including rhinitis requiring topical corticosteroid treatment. Subjects with asthma were evaluated and classified according to the ATS workshop criteria into subjects with severe or not severe asthma (26, 27). Subjects with severe asthma required high doses of inhaled corticosteroids (ICS) and/or use of oral corticosteroid (OCS) therapy for 50% or more of the previous year. In addition, subjects with severe asthma exhibited two or more of seven minor criteria as outlined in Reference 26. Subjects with not severe asthma were further grouped based on lung function and use of ICS. Those with a prebronchodilator FEV₁ greater than 80% predicted and on nonsteroid therapy only to control their symptoms were termed “Mild asthma/no ICS.” The group termed “Mild asthma/+ICS” had an FEV₁ greater than 80% predicted and were using low-moderate doses of ICS, whereas “Moderate asthma” included subjects with an FEV₁ less than 80% predicted, all of whom were treated with low-moderate dose of ICS, with or without a second controller agent (leukotriene modifier or long-acting β-agonist) (27).

Questionnaires

Information collected through questionnaires included general demographics, medical and smoking history, and frequency of asthma symptoms (see online supplement for details). Subjects also reported the need for three or more steroid bursts, asthma-related urgent care visits, or hospitalizations. Reporting any one or more of those events in the 12 months preceding bronchoscopy identified the subject as having a recent exacerbation.

Pulmonary Function and Atopy Testing

Pulmonary function testing was performed according to American Thoracic Society guidelines and as previously described (27). The parameters measured included prebronchodilator FEV₁ and FVC (as % predicted) and FEV₁/FVC. Serum IgE was measured and atopy was assessed by skin prick testing to 14 common aeroallergens (see online supplement for details).

Bronchoscopy, BAL, and Sample Processing

These procedures were performed across sites according to previously published protocols and the SARP manual of procedures (4, 6, 27, 28). Endobronchial tissue for immunohistochemistry (IHC) was collected at all the SARP sites listed above, whereas tissue and epithelial cell samples for mRNA were collected only at the National Jewish Health and then the University of Pittsburgh site. Epithelial cells were obtained by brushing the proximal airways as previously described (28). The cells were placed in Trizol (Invitrogen, Carlsbad, CA) for extraction of mRNA. Tissue for IHC was fixed in 10% buffered formalin and embedded in paraffin, and tissue for mRNA was placed in Trizol.

BAL cells and fluid were separated by centrifugation at 400 × g. BAL fluid aliquots were stored at −80°C and cytospins processed as previously described (see online supplement for details).

Immunostaining and Tissue Morphometric Analysis

Tissue sections (5 μm) were immunostained at the Cleveland Clinic site using an automated system (see online supplement for details). Positively stained cells were counted in the area between the subepithelial basement membrane and smooth muscle and within an intact epithelial layer, at 400× magnification. Analyzed tissue areas were measured using Image-Pro software (MediaCybernetics, Washington, D.C.) and cell counts reported per square millimeter of submucosa or, for epithelial cell counts, per 10-mm length of subepithelial basement membrane.

Quantitative Real-Time Polymerase Chain Reaction

Tissue and epithelial cell expression of tryptase, chymase and carboxypeptidase A3 (CPA3) mRNA was determined by real-time quantitative polymerase chain reaction (PCR) as described previously (see online supplement) (28).

BAL Fluid Tryptase and PGD2 Measurements

Both mediators were measured using ELISA. BAL fluid tryptase levels were measured in Dr. L. B. Schwartz's laboratory (Virginia Commonwealth University, Richmond, VA) with an in-house total tryptase ELISA after concentrating the BAL fluid 15-fold by lyophilization as described (29). BAL fluid PGD2 was measured at Elisatech (Denver, CO) using commercially available ELISA components (Cayman Chemical, Ann Arbor, MI). The samples were first purified/concentrated using C18 Sep-Pak cartridges (Waters Corp, Milford, MA) and then derivatized per the ELISA manufacturer's instructions. Detection limits for those assays were 20 pg/ml for tryptase and 3 to 5 pg/ml for PGD2.

MC Phenotypes by IHC and Real-Time Quantitative PCR

MC phenotype in tissue samples was assessed by a standard IHC approach, detecting tryptase-positive and, on a consecutive section, chymase-positive cells. As double labeling was not performed, tryptase-positive MCs generally reflect the total number of MCs (MC_Tot), including MCs that are tryptase-only positive (MC_T; their low chymase levels are undetectable by routine IHC) and those that are both tryptase and chymase positive (MC_TC) by routine IHC. Chymase-positive MCs in this study reflected the MC_TC population. Consequently, MC_TC/MC_Tot estimated the proportion of MC_TC in the total MC population.

In addition to IHC, the MC presence was estimated by mRNA levels as measured by quantitative real-time PCR for the phenotype-identifying MC enzymes tryptase and CPA3. Chymase mRNA levels across all groups and compartments were very low. Therefore, the MC_TC phenotype was estimated at the mRNA level through expression of CPA3 mRNA, another enzyme reported to differentiate the MC_TC from a MC_T phenotype (11). A small substudy (n = 14), which showed a significant correlation of chymase-positive MCs with CPA3-positive MCs in tissue by IHC (r_s = 0.61; P = 0.02), supported such an approach.

Final Selection of Study Population

For reasons of the diverse interests of the SARP sites, not all subjects in the bronchoscopy data set (large SARP cohort with IHC and BAL data) had measurements available for all four MC markers (IHC tryptase and chymase, BAL fluid tryptase, and PGD2) analyzed in this report. Therefore, an a priori decision was made to only include subjects with data for at least two of the four MC markers (n = 157). The five groups were not statistically different in fractions of missing data for submucosal tryptase or chymase (overall P = 0.11 and 0.06, respectively), or for epithelial tryptase or chymase (overall P = 0.11 and 0.69, respectively). Details and Figure E1 showing total study population are in the online supplement.

Messenger RNA analysis was performed in subjects from the Pittsburgh site only. Due to the limited number of subjects with moderate asthma (n = 1) with tissue and/or epithelial mRNA, analysis was performed on NC, mild asthma/no ICS, mild asthma/+ICS, and severe asthma subjects (n = 60). Thirty-one subjects had mRNA for MC markers measured in both types of samples. As 18 subjects in the Pittsburgh mRNA cohort were also included in the large SARP cohort (they also had two or more IHC and/or BAL markers measured), a total of 199 subjects were included (see online supplement).

Statistical Analysis

Nonparametric Kruskal-Wallis tests were used to compare medians for the MC markers across the five groups. Analysis of variance was used to compare the means of the log-transformed mRNA data. When an overall significant difference was detected (P < 0.05), individual asthma groups were compared with NC (four comparisons) and the severe asthma group to each of the less severe groups (three comparisons). The Bonferroni procedure was used to conservatively test for the multiple post hoc comparisons between groups using an α = 0.05/7 = 0.0071 (for mRNA data, α = 0.05/5 = 0.01) for each comparison. To determine whether CS use impacted submucosal MC counts, MC_Tot were compared among the four asthma groups treated with increasing doses of CS, whereas MC_Tot and MC_TC were compared in severe asthma treated or not with OCS. Categorical variables were compared by Pearson chi-square tests. Spearman correlations (r_s) measured the association between any two MC markers and between MC markers and continuous measures of lung function among the whole population and, when significant, among the phenotypic groups. The small size of the moderate asthma group (n = 6–12) prevented subgroup analysis on that group. Normally distributed data (log-transformed mRNA data) were correlated using Pearson correlation coefficients (r).

General linear models and contrast coefficients were used to conduct tests for trend on MC markers that could be transformed to more normal distributions. The square root transformation was applied for MC_Tot, MC_TC, and MC_TC/MC_Tot and the natural logarithm transformation for tryptase and CPA3 mRNA levels. The IHC epithelial and BAL fluid markers showed a high proportion of the lowest detectable limit and high positive skew in their distributions. For these markers, the nonparametric Jonckheere-Terpstra test (JT) was used for testing trends across the five groups.

A multiple logistic regression model was used to determine the significance of the associations between the MC markers and severity of asthma (severe versus not severe) among the participants with asthma. Due to the amount of missing data, epithelial MC counts or any mRNA data could not be included in the analysis. Univariate analyses were first conducted for FEV₁% predicted, maximum bronchodilator reversibility, BAL eosinophil percentage, the submucosal markers, BAL fluid PGD2, therapy with leukotriene modifiers, and atopy. Any variable significant at P less than 0.20 was entered into a multiple logistic regression model controlling for race and age. We controlled for age and race because both variables were important potential confounders as they were associated with asthma severity and several of the lung function and MC markers. Due to the high proportion of missing values for submucosal MC_TC/MC_Tot, a sensitivity analysis was conducted using multiple imputation for the final logistic regression model. Results for the models are presented with odds ratios and 95% confidence intervals. The data were analyzed using SAS/STAT software, Version 9.1.3 of the SAS System for Windows.

RESULTS

Demographics

Participants were predominantly women (2:1) across all groups (Table 1). Subjects in the more severe groups were older compared with NC and mild asthma/no ICS. The groups differed by race, with a higher proportion of African Americans in the moderate and severe asthma groups. Atopy was less common in NC, but was similarly present among the asthma groups. Five subjects in the severe asthma group were on anti-IgE therapy (data not shown). Compared with the large SARP cohort, the University of Pittsburgh mRNA cohort's 9 subjects with mild/no ICS asthma were older, whereas the 23 subjects with severe asthma had a lower FEV₁% predicted (median [25th–75th percentile] 49% [35–74] vs. 67% [56–82]; P = 0.005; Table 2). Systemic CS therapy was more common in the Pittsburgh severe asthma group (P = 0.02). The contribution of subjects from each of the participating centers is presented in Table 1.

TABLE 1.

SUBJECT CHARACTERISTIC OF THE LARGE SEVERE ASTHMA RESEARCH PROGRAM COHORT (N = 157)

	Normal Control Subjects (n = 34)	Mild Asthma No ICS (n = 22)	Mild Asthma on ICS (n = 31)	Moderate Asthma (n = 13)	Severe Asthma (n = 57)	Overall Difference (P Value)
Age, yr*	26 (24–34)	23 (20–30)	37 (23–40)	38 (29–41)	44 (32–50)	<0.0001
Sex, % female	62	59	65	77	68	NS
AA/white/other, %	12/70/18	14/77/9	16/81/3	46/54/0	37/58/5	0.01
Atopy, %	32.4	91	77.4	100	77	<0.0001
Serum IgE, kU/l*	23 (13–53)	80 (30–167)	202 (70–352)	150 (88–388)	170 (62–344)	<0.0001
Baseline FEV1% predicted*	99 (93–107)	97 (89–106)	91 (84–104)	69 (64–73)	67 (57–83)	<0.0001
Use of systemic CS therapy, n	N/A	1	1	0	32	<0.0001
Use of LABA, %	N/A	0	81	77	89	<0.0001
Use of LTM, %	N/A	10	17	31	48	0.003
Subjects by site, n (%) of each severity group
BWH	0	0	6 (19)	1 (8)	1 (2)
CLC	8 (23)	3 (14)	1 (3)	2 (15)	6 (11)
PITT	15 (44)	4 (18)	1 (3)	0	20 (35)
UVA	0	0	0	0	1 (2)
WIS	7 (21)	8 (36)	17 (55)	3 (23)	10 (17)
WFU	0	5 (23)	4 (13)	5 (38)	6 (10)
WSL	4 (12)	2 (9)	2 (6)	2 (15)	13 (23)

Definition of abbreviations: AA = African American; BWH = Brigham and Women's Hospital Boston, MA; CLC = Cleveland Clinic, Cleveland, OH; CS = corticosteroids; F = female; ICS = inhaled corticosteroids; LABA = long-acting β-agonists; LTM = leukotriene modifiers; PITT = National Jewish Health, Denver, CO, and University of Pittsburgh, Pittsburgh, PA; UVA = University of Virginia, Charlottesville, VA; WIS = University of Wisconsin, Madison, WI; WFU = Wake Forest University, Winston-Salem, NC; WSL = Washington University, St. Louis, MO.

If not indicated otherwise, analyses were done using Pearson chi-square tests.

^*Data analyzed using Kruskal/Wallis tests and presented as medians (25th–75th percentile).

TABLE 2.

SUBJECT CHARACTERISTICS OF THE UNIVERSITY OF PITTSBURGH COHORT (N = 60)

	Normal Control Subjects (n = 21)	Mild Asthma No ICS (n = 9)	Mild Asthma on ICS (n = 7)	Severe Asthma (n = 23)	Overall Difference (P Value)
Age, yr*	29 (25–49)	38 (27–57)†	27 (21–31)	39 (30–47)	0.05
Sex, % female	43	78	71	61	NS
AA/white/other, %	5/71/24	33/56/11	14/72/14	26/74/0	NS
Atopy, %	52	100	85.7	74	0.05
Serum IgE, kU/l*	25 (14–42)	137 (57–402)	137 (47–341)	154 (43–432)	0.0004
Baseline FEV1% predicted*	100 (94–106)	90 (82–96)	103 (94–107)	49 (35–74)†	<0.0001
Use of systemic CS therapy, n	N/A	0	0	17‡	<0.0001
Use of LABA, %	N/A	14	71	87	0.0005
Use of LTM, %	N/A	14	14	65	0.007

Definition of abbreviations: AA = African American; CS = corticosteroids; F = female; ICS = inhaled corticosteroids; LABA = long-acting β-agonists; LTM = leukotriene modifiers; SARP = Severe Asthma Research Program.

If not indicated otherwise, analyses were done using Pearson chi-square tests.

^*Data analyzed using Kruskal/Wallis tests and presented as medians (25th–75th percentile).

^†P < 0.01 when compared to the SARP cohort.

^‡P = 0.02 when compared to the SARP cohort.

Submucosal Tissue MCs: Severe Asthma is Associated with Reduced MC_Tot Numbers, but a Greater Proportion of MC_TC

Immunohistochemistry for MC_Tot and MC_TC in airway submucosa.

MC_Tot numbers in the proximal airway submucosa differed across the five subject groups (overall P = 0.002; Figure 1A). Submucosal MC_Tot were most numerous in mild asthma/no ICS (median [25th–75th percentile] 285/mm² [135–361]) and were lowest in the severe asthma group (79/mm² [9–188]). There was a significant trend for MC_Tot to lower levels with increasing asthma severity across all groups (P_trend = 0.002).

Figure 1.

(A) Distribution of submucosal total mast cells (MCs) (tryptase-positive MC_Tot; solid circles/blue lines show median and 25th–75th percentiles) and chymase-positive MCs (MC_TC; shaded circles/red lines show median and 25th–75th percentiles) evaluated by immunostaining. Open circles in the severe asthma group represent subjects who are not on systemic corticosteroid therapy. (B) Increase in MC_TC/MC_Tot with increasing asthma severity. The groups were compared using Kruskal/Wallis tests (overall P values) and Jonckheere-Terpstra tests (trend P values). Marked are intergroup differences that remained significant after Bonferroni correction (P ≤ 0.007). ICS = inhaled corticosteroid.

Although MC_TC counts did not differ between the groups (overall P = 0.19), there was a marginal trend for higher MC_TC with increasing asthma severity (P_trend = 0.06; Figure 1A). The overall reduction in MC_Tot and the concomitant development of a predominance of the MC_TC phenotype with greater asthma severity resulted in a higher MC_TC/MC_Tot in severe asthma (overall P = 0.005, P_trend = 0.01; Figure 1B) than in mild asthma/no ICS (P = 0.004) or mild asthma/+ICS (P = 0.001).

Submucosal MC_Tot and MC_TC were modestly correlated across all subject groups (r_s = 0.34; P < 0.001). However, this positive association was only present in mild asthma/+ICS (r_s = 0.78; P < 0.0001) and severe asthma (r_s = 0.38; P = 0.02), suggesting that in these more severe CS-treated groups, a consistent shift in MC_Tot population toward MC_TC phenotype occurs. Table E2.A presents correlations between IHC MC markers in different airway compartments.

Subgroup analysis for tissue mRNA.

Tissue biopsy samples for mRNA analyses were available from 34 subjects. Tissue tryptase and CPA3 mRNA levels did not differentiate NC (n = 14) from mild asthma/no ICS (n = 5), mild asthma/+ICS (n = 4), or severe asthma (n = 11) subjects (overall P = 0.35 and P = 0.36, respectively; data not shown). In contrast to IHC, tissue tryptase and CPA3 mRNA levels were strongly correlated (r = 0.81; P < 0.0001), supporting their common MC origin (Table E2.B). However, there were no significant relationships between submucosal MC numbers (by IHC) and levels of MC-specific markers (by mRNA).

Effect of CS therapy on MC numbers.

As CS therapy has been reported to impact submucosal MC counts, MC numbers were compared between the mild asthma groups on and off ICS, between subjects on low-dose ICS (mild) versus those on higher doses (moderate and severe), as well as between those subjects with severe asthma on and off OCS (23). The mild asthma/+ICS group, which did not differ in FEV₁ compared with the mild asthma/no ICS, had significantly lower submucosal MC_Tot (P = 0.008) supporting an effect of ICS. However, the use of higher doses of ICS (and systemic CS) was not associated with lower MC_Tot numbers in moderate and severe asthma as compared with mild asthma/+ICS (overall P = 0.34). Finally, there were no differences in MC_Tot or MC_TC numbers in subjects with severe asthma treated or not with systemic CS (P values both > 0.8).

Epithelial MCs: Severe Asthma is Associated with Higher Numbers of Epithelial MC_TC Compared with Milder Asthma

Epithelial MC_Tot varied with asthma severity (overall P = 0.02; Figures 2A and 2B). Similar to submucosal MC_Tot, epithelial MC_Tot were highest in mild asthma/no ICS (60/10 mm [2–105]). However, there was no trend for epithelial MC_Tot counts across the groups (JT test for trend P = 0.55). Submucosal and epithelial MC_Tot counts were moderately well correlated (r_s = 0.49; P < 0.0001; Table E2.A). Epithelial MC_Tot and submucosal MC_Tot correlations were consistently high in the presence of CS treatment (mild asthma/+ICS r_s = 0.53; P = 0.02, and severe asthma r_s = 0.65; P < 0.0001). In contrast, the correlations were weak in NC or mild asthma/no ICS (r_s all < 0.2).

Figure 2.

(A) Immunostaining for tryptase (left panels) and chymase (right panels) in the same proximal airway region from a subject with severe asthma. Pictures in upper panels are taken at 200× magnification. Arrows (yellow) in the right upper panel point at chymase-positive epithelial mast cells. Epithelial mast cells within boxes are shown in lower panels at 600× magnification. (B) Distribution of epithelial total mast cells (MCs) (tryptase-positive MC_Tot; black circles/blue lines are median and 25th–75th percentiles) and chymase-positive mast cells (MC_TC; gray circles/red lines are medians and 25th–75th percentiles) evaluated by immunostaining. The groups were compared using Kruskal/Wallis tests (overall P value) and Jonckheere-Terpstra tests (trend P values). (C) Relative mRNA levels for tryptase, and (D) carboxypeptidase A3 (CPA3) in airway epithelial brushings. Y-axis represents a natural log-scale of the original linear values, with the means and SEM transformed from the natural log to the linear values. The groups in C and D were compared using analysis of variance and linear trend analysis. Marked are intergroup differences that remained significant after Bonferroni correction (P ≤ 0.01). ICS = inhaled corticosteroid.

Epithelial MC_TC also differed across groups (P = 0.05; Figures 2A and 2B). Unlike MC_Tot, there was a significant trend to higher numbers from NC to severe asthma (JT test for trend P = 0.01). Epithelial MC_TC were rarely detectable in NC (9% of subjects) but more likely to be seen in milder asthma (∼ 20% of subjects) and severe asthma (42% of subjects). The number of epithelial and submucosal MC_TC marginally correlated overall (r_s = 0.30; P = 0.009), without difference among the individual groups.

Epithelial Cell mRNA subgroup analysis.

Epithelial brushings were available from 19 NC, 9 mild asthma/no ICS, 7 mild asthma/+ICS and 22 severe asthma subjects. Epithelial tryptase and CPA3 mRNA increased in asthma and were highest in severe asthma (overall P = 0.02 and P = 0.008, respectively; P_trend = 0.008 and P_trend = 0.01, respectively; Figures 2C and 2D). The higher levels of mRNA for both markers in severe asthma paralleled the higher numbers of epithelial MC_TC seen in the larger IHC database.

Epithelial tryptase mRNA consistently correlated with CPA3 mRNA (r = 0.74; P < 0.0001) across all subjects and in each individual group (r = 0.65–0.98). Despite being measured in different samples (biopsy tissue vs. brushings), tryptase and CPA3 mRNA correlated between epithelium and tissue (r = 0.73; P < 0.0001 and r = 0.57; P = 0.001, respectively; Table E2.B). There were no statistically significant correlations between MC counts (by IHC) in submucosa and epithelium (n = 18) and matching mRNA levels.

Activity of MCs: Severe Asthma Is Associated with Evidence for MC_TC Activation

MC activity was assessed by comparing BAL fluid levels of tryptase (n = 132) and PGD2 (n = 138). BAL fluid tryptase did not differ among the five groups (overall P = 0.13; Figure 3A). Similar to submucosal and epithelial MC_Tot, BAL fluid tryptase was numerically highest in mild asthma/no ICS and there was no evidence for higher levels with increasing severity (JT test for trend P = 0.56).

Figure 3.

Concentrations of (A) tryptase and (B) prostaglandin D2 (PGD2) measured in bronchoalveolar lavage fluid (BALF) samples. Tryptase levels that were below the limit of detection were assigned a value of 10 pg/ml. PGD2 values that were below detection limit were given value of 2 pg/ml. Y-axis represents a natural log scale of the original linear values. The median and 25th–75th percentiles are transformed from the natural log to the linear values. The groups were compared using Kruskal/Wallis tests (overall P value) and Jonckheere-Terpstra tests (trend P values). Marked are intergroup differences that remained significant after Bonferroni correction (P ≤ 0.007). ICS = inhaled corticosteroid.

In contrast to tryptase, BAL fluid PGD2 differed significantly among the five groups (overall P = 0.02; Figure 3B). Paralleling the increase in submucosal and epithelial MC_TC, PGD2 was higher in severe asthma as compared with other steroid-treated groups (6.0 pg/ml [2.5–11.0 pg/ml] vs. 4.0 pg/ml [2.0–6.0 pg/ml] in mild/moderate groups combined and 5.0 pg/ml [4.0–9.0 pg/ml] in normal controls, respectively; JT test for trend P = 0.28).

Although only BAL fluid PGD2 significantly differed between the groups, BAL fluid PGD2 and tryptase levels were modestly correlated (r_s = 0.33; P < 0.001), suggesting a common MC source. Neither mediator significantly correlated with any IHC marker. In smaller numbers there was a marginal correlation (r = 0.34; P = 0.08) for epithelial tryptase mRNA with BAL fluid tryptase levels, but not with PGD2.

PGD₂ and MC_TC/MC_Tot Are Predictors of Severe Asthma

Univariate analyses.

Relationships of MC markers with hallmarks of asthma, including FEV₁% predicted, frequency of asthma-related symptoms, and history of recent asthma exacerbations, were tested. Recent exacerbations of asthma were reported by 23% of mild asthma/no ICS, 19% of mild asthma/+ICS, 15% of moderate asthma, and 75% of severe asthma subjects (overall P < 0.0001). Higher numbers of submucosal MC_Tot were associated with better FEV₁% predicted (r_s = 0.34; P < 0.001) and fewer nighttime symptoms (P = 0.01), but no relationships were detected with exacerbations or other symptoms. MC_TC/MC_Tot and BAL fluid PGD2 inversely correlated with lung function (r_s = −0.38; P < 0.001 and r_s = −0.27; P = 0.005, respectively). Both markers were associated with higher levels of symptoms (all P values < 0.03; Table E1), whereas higher PGD2 was associated with the history of a recent exacerbation (as defined in Methods) (6.0 pg/ml [2.0–12.0]) compared with those without (4.0 pg/ml [2.0–5.0]; P = 0.001). Other relationships between MC markers and asthma-related outcomes were not significant (data not shown).

Multivariable logistic regression models.

Only FEV₁% predicted, maximum bronchodilator reversibility, MC_TC/MC_Tot, and BAL fluid PGD2 were associated with severe asthma in the initial univariate analyses. The four highest values of MC_TC/MC_Tot were truncated to the fifth highest value and the square root transformation was applied to decrease the influence of extreme data points on the logistic regression coefficients. In addition, the natural log transformation was applied to BAL fluid PGD2 levels before modeling for the same purpose. Model 1 (Table 3) included 62 subjects with asthma with data for all the listed predictors. In that model, both MC_TC/MC_Tot and BAL fluid PGD2 were significant predictors of severe asthma (P = 0.02 for both predictors), when controlling for age and race. In Model 2 (n = 107), with data imputed for the missing MC_TC data to compensate for the large proportion of missing submucosal MC_TC/MC_Tot data, age was the only significant predictor of severe asthma. Both MC_TC/MC_Tot and PGD2 were marginally important (P = 0.095 and 0.097, respectively), but still stronger predictors than FEV₁% predicted (P = 0.2).

TABLE 3.

LOGISTIC REGRESSION MODELS WITH PREDICTORS OF SEVERE ASTHMA (SEVERE VS. NOT SEVERE) AMONG SUBJECTS WITH ASTHMA ONLY

	Model 1 (n = 62)		Model 2 (n = 107)		Model 3 (n = 98)
Predictor	Odds Ratio	Adjusted P Value	Odds Ratio	Adjusted P Value	Odds Ratio	Adjusted P Value
Baseline FEV1% predicted	1.00 (0.96, 1.04)	0.92	0.98 (0.95, 1.01)	0.20	0.98 (0.94, 1.02)	0.26
Maximum reversibility	1.04 (0.99, 1.08)	0.093	1.03 (1.00, 1.07)	0.076	1.05 (1.00, 1.09)	0.0372
BALF PGD2*	3.11 (1.17, 8.26)	0.023	1.78 (0.90, 3.54)	0.097	1.37 (0.53, 3.53)	0.51
Age, yr	1.07 (1.01, 1.14)	0.031	1.08 (1.03, 1.14)	0.0032	1.10 (1.03, 1.17)	0.0039
MCTC/MCTot*	6.16 (1.35, 28.25)	0.019	3.00 (0.82, 11.00)	0.095	5.08 (1.06, 24.24)	0.0418
African American (ref: white)	2.76 (0.71, 10.68)	0.14	2.52 (0.81, 7.85)	0.11	3.66 (0.62, 21.68)	0.145
Clinical center
Overall P value					—–	0.0263
CLC					0.40 (0.03, 4.75)
WFU					0.19 (0.02, 2.09)
WIS					0.19 (0.02, 2.09)
WSL					0.23 (0.01, 3.87)
PITT (reference)					1.00

Definition of abbreviations: BALF = bronchoalveolar lavage fluid; CLC = Cleveland Clinic, Cleveland, OH; MC_TC = both tryptase- and chymase-positive mast cells; MC_Tot = total mast cells; PDG2 = prostaglandin D2; PITT = National Jewish Health, Denver, CO, and University of Pittsburgh, Pittsburgh, PA; WIS = University of Wisconsin, Madison, WI; WFU = Wake Forest University, Winston-Salem, NC; WSL = Washington University, St. Louis, MO.

Model 1: all listed predictors of severe vs. not severe asthma included. Model 2: multiple imputation model with data imputed for the missing MC_TC. Model 3: multiple imputation model controlling for clinical center; excludes University of Virginia and Brigham and Women's Hospital due to too few cases.

^*Log transformation was applied to BALF PGD2 and square root transformation was applied to MC_TC/MC_Tot for modeling purposes.

Controlling for center.

There were highly significant differences among the centers in subjects' severity of asthma, percentages of African Americans, MC_TC/MC_Tot, BAL fluid tryptase, and PGD2 (Table E3). When controlling for the five clinical centers that had more than one subject with severe asthma in the data set (University of Virginia and Brigham and Women's Hospital were thus excluded), age, bronchodilator reversibility, and MC_TC/MC_Tot were significant predictors of severe asthma (P values = 0.004–0.04), whereas FEV₁% predicted and PGD2 were not (P = 0.26 and P = 0.51, respectively) (Table 3, Model 3). However, caution is warranted, as the prohibitive number of predictors (11) in this model limits the interpretation of results.

DISCUSSION

These results provide robust evidence for alterations in distribution, phenotypic profile, and activity of airway MCs over a range of asthma severities. In mild asthma, the submucosal MC population is abundant and maintains a “normal” phenotype profile (low MC_TC). In severe asthma, the total submucosal MC population is low, but dominated by an MC_TC, a phenotype that is also uniquely observed in the epithelium. Despite reduced submucosal MC counts, BAL fluid tryptase levels in severe asthma are not lower than in other groups, perhaps reflecting the reported increased capacity for tryptase activity of an MC_TC-dominant MC population (30). These MC changes are further associated with higher levels of BAL fluid PGD2, a lipid mediator relatively specific to MCs, which was increased in severe asthma as compared with the mild asthma/no ICS group. These alterations in MC phenotype and likely activation correspond to greater airway obstruction, frequency of asthma symptoms, and exacerbations. In toto, these results (from the largest research bronchoscopy study yet reported) strongly suggest a role for MC differentiation, migration, and activation in the pathobiology of severe asthma.

The loss of submucosal MCs in severe asthma appears to be primarily due to the loss of MCs of the MC_T phenotype. Although MC_T predominate in the expanded MC population in mild asthma/no ICS, where only 20% of the MC population are MC_TC, the decrease in MC_T (and hence MC_Tot) is already present in mild asthma treated with ICS. As MC_Tot decrease with greater severity of asthma, the numbers and proportion of MC_TC increase (P = 0.06 and P = 0.01, respectively). In the majority of subjects with severe asthma, airway MCs were predominantly MC_TC. These MC_TC are reported to be both “steroid-resistant” and “T cell–independent,” suggesting a mechanism for their persistence in subjects on CS therapy (23). Alterations in the airway environment in CS-treated asthma may no longer support the MC_T population, but rather support or even induce MC_TC differentiation. Although it is difficult to prove whether loss of MC_Tot is related to CS use, the airway microenvironment associated with severe asthma appears to support this MC_TC phenotype even in subjects on additional leukotriene modifiers (n = 27) or anti-IgE therapy (n = 5; data not shown).

In contrast to the submucosa, epithelial MC numbers in severe asthma are not lower. Rather, epithelial MCs persist, predominantly as an MC_TC phenotype. This epithelial MC_TC phenotype is rarely seen in normal subjects or subjects with milder asthma, suggesting that in severe asthma, the epithelium actively contributes to MC_TC migration (or differentiation), survival, and activity. Supportive of the IHC data, epithelial levels of both tryptase and CPA3 mRNA increase in asthma, and trend analysis supports a further increase with increasing severity. Previous IHC studies have suggested a redistribution of MCs to airway smooth muscle and epithelium in mild asthma (8, 15). This is the first study to report epithelial MC_TC distribution in a range of asthma severities, with MC_TC increase specifically in severe asthma.

The mechanisms for this epithelial migration and differentiation of MCs remain speculative. Murine models of allergic lung inflammation induce relatively limited numbers of epithelial MCs and thus preclude compartment-by-compartment studies of changes in lung MCs or phenotype (31). Therefore, studies to understand the mechanisms behind MC phenotypic changes in association with epithelial migration generally derive from murine gastrointestinal tract parasitic models (12, 32–35). After parasite exposure, MC size, distribution, and phenotype are profoundly orchestrated from initial exposure to parasite expulsion and return to basal conditions. Transforming growth factor (TGF)-β, shown to be increased in asthmatic epithelium, appears to be critical to this homing of MCs to the intestinal epithelium in mice and development of an MC_TC-like phenotype crucial for resolution of Trichinella spiralis infection (32–35). Although airway epithelial-derived stem cell factor has been shown to support an MC_T phenotype in vitro, whether epithelial TGF-β could promote an epithelial MC_TC in human asthma remains unknown (36, 37). However, severe asthma epithelial cells, through enhanced expression of TGF-β, in association with Th2 cytokines and IL-9, could promote development of this MC_TC phenotype (38). Similarly, Th2-induced epithelial eotaxins could also lead to MC recruitment and activation (39, 40).

For MC_TC in severe asthma to play a role in epithelial or other cell function, evidence for phenotype-specific activation is also needed. As noted earlier, BAL fluid tryptase does not decrease as asthma severity increases. These maintained levels despite low MC_Tot may suggest a high level of tryptase release from MC_TC cells, consistent with reports that MC_TC granules contain two to three times more tryptase than MC_T (30). More interestingly, however, PGD2 levels are numerically highest in severe asthma, and significantly higher than in mild asthma/+ICS. Arachidonic acid metabolism in MC_TC is reported to preferentially use the cyclooxygenase pathway (21, 41). Although basal levels of this cyclooxygenase pathway product have been reported to be nearly undetectable in mild asthma, PGD2 dramatically increases after allergen/IgE stimulation (42–44). In this study, no exogenous stimulation was applied, suggesting that in severe asthma, continuous signals (IgE or otherwise) are generating PGD2.

Epithelial MC activation also likely impacts the epithelium through effects on barrier function, epithelial turnover, and even goblet cell formation (45, 46). In addition, the presumed MC expression of PGD2 can act through receptors expressed on Th2 cells (DP1 and DP2/CRTH2 [chemoattractant-homologous receptor expressed on Th2 cells]), which could further amplify Th2 processes at mucosal/epithelial surfaces (47–49). Interest in antagonizing the effects of PGD2 through these receptors is increasing, with a recent abstract reporting efficacy of a CRTH2 antagonist in mild asthma (50). A second proof-of-mechanism study of a CRTH2 antagonist also reported positive data (press release Actelion at http://www1.actelion.com). These cumulative data provide hope that antagonism of this pathway may improve treatment of severe asthma.

In the three models used, MC-associated parameters MC_TC/MC_Tot and BAL fluid PGD2 levels were generally stronger predictors of severe asthma than FEV₁% predicted, one of the most important predictors of asthma phenotype reported in a recent cluster analysis of the SARP population (51). Submucosal eosinophils were not measured in this study due to difficulties in obtaining accurate eosinophil counts by IHC in formalin fixation/paraffin embedded tissue. However, several previous studies have not shown tissue eosinophils to predict asthma severity (4, 52). Intriguingly, MC_TC numbers bore no relation to atopy or BAL measures of eosinophilic inflammation, rather predominating in subjects with symptomatic severe asthma. Whether these MC alterations could tie together some of the asthma phenotypes recently described remains to be determined (51–53).

Although the direction of changes in the epithelial mRNA and BAL fluid data strongly support the IHC data, correlations between MC phenotypes measured by one approach compared with another are modest. The reasons for this are not clear, but could include differences in sensitivity of quantitative real-time PCR versus IHC, differences between stored granule proteins versus transcriptional activation, and others. Alterations in distal lung MCs could also contribute to the lack of correlations between lavage fluid and tissue/epithelial cell measures. We previously reported an increase in MC_TC in the small airways of subjects with severe asthma compared with normal subjects, which could also independently contribute to lavage fluid mediator levels (5).

The large number of subjects from seven SARP centers produced the most comprehensive bronchoscopic study in humans yet reported. This approach is possible because SARP uses standardized protocols for lung function testing, clinical/immunologic evaluation, and bronchoscopy. Thus, SARP allowed comparison of data obtained by invasive procedures from a large number of subjects with a serious medical condition. However, some limitations still exist, including large differences across centers in race, severity, and three of the MC markers of interest, which likely contributed to the loss of significance for PGD2 in a model when center is included (as well as the impact of controlling for five sites). Despite a Manual of Procedures, subtle (but potentially important) differences in processing of tissue and fluids may have contributed to site differences in MC markers.

Additionally, given the nature of SARP, not every MC measurement was performed on every subject. Although automated IHC staining improved consistency, chymase staining was suboptimal in about 20% of subjects. Double staining of MC markers could give a more accurate profile of MC phenotypes. Technically, the lack of availability of primary antibodies from different species or even of different isotypes makes these approaches challenging in large-scale studies. However, a high MC_TC/MC_Tot in severe asthma would be predicted based on the individual markers' data, as chymase-positive MCs were only detected in similar numbers to tryptase-positive cells in severe asthma. Finally, the existence of an MC phenotype expressing only chymase by IHC, which could not be identified without colocalization studies, remains highly controversial. Although the current study also cannot confirm that PGD2 is arising from MCs, the levels produced by other cells are multifold lower (54). The similar patterns of PGD2 and MC_TC profiles support an MC source.

In conclusion, this comprehensive study reports a predominance of an MC_TC phenotype (by both protein and mRNA) in the airway submucosa and epithelium in severe asthma, with evidence for ongoing MC activity (as measured by BAL fluid PGD2) even in the face of high inhaled and systemic CS use. Modeling suggests this shift toward the MC_TC phenotype and increased airway luminal PGD2 are both more predictive of severe asthma than traditional measures. These findings have profound implications for the role that MCs play in human asthma, particularly severe asthma, as well as for identification of new therapeutic targets.