Profiling neutrophil heterogeneity in asthma with single cell multiparameter CyTOF

Abstract

Asthma is a chronic inflammatory disease of the airways that afflicts over 30 million individuals in the United States and over 300 million individuals worldwide. The inflammatory response in the airways is often characterized by the analysis of sputum which contains multiple types of cells including neutrophils, macrophages, lymphocytes, and rare bronchial epithelial cells. Subtyping patients using microscopy of the sputum has identified both neutrophilic and eosinophilic infiltrates in airway inflammation. However, with the extensive heterogeneity among these cell types, a higher resolution understanding of the inflammatory cell types present in the sputum is needed to dissect the heterogeneity of disease. Improved recognition of the distinct phenotypes and sources of inflammation in asthmatic granulocytes, may identify relevant pathways for clinical management or investigation of novel therapeutic mediators. Here, we employed mass cytometry or CyTOF (Cytometry by Time-Of-Flight) to quantify frequency and define functional status of sputum derived airway cells in asthmatic patients and healthy controls. This in-depth single cell analysis identified multiple distinct subtypes of airway immune cells between healthy controls and asthmatic patients, and within an asthmatic patient cohort, especially in neutrophils. Significance was discovered by statistical analysis as well as a data-driven unbiased clustering approach. Our multidimensional assessment identifies differences in cellular function that may contribute to diverse clinical responses and provide meaningful insights to advance our knowledge of asthmatic inflammation.

Summary Sentence:

An in-depth single cell analysis identifies multiple distinct subtypes of airway immune cells and provides insights of cellular function relevant for pathogenesis of asthma.

Introduction

Asthma is a chronic inflammatory disease of the airways that afflicts over 30 million individuals in the United States and over 300 million individuals worldwide [1]. It is characterized by bronchial hyperreactivity and variable, chronic respiratory symptoms that are non-specific in the clinic--asthma is now recognized to be clinically and biologically heterogenous [2]. While a majority of individuals have mild disease and mild symptoms that are easily controlled, 10–15% of individuals have severe asthma and require higher doses of inhaled corticosteroids (ICS), are more likely to require hospitalization or oral corticosteroids (OCS), and have worse lung function [3]. Biologically, most patients have airway inflammation that is characterized by T2 inflammation, a specific IgE production, mucus production, and infiltration of eosinophils, T cells, and monocytes [4]. A minority of patients have what is termed non-T2 disease and lack allergies. The inflammatory response in these patients is poorly understood but is typically associated with neutrophilic inflammation and no specific IgE response [5, 6].

Heterogeneity is a hallmark of asthma symptomatology and response to therapy [7]. There is increasing appreciation that asthma is driven by both adaptive and innate immune responses to varying degrees in every individual. This underscores the need for a higher resolution understanding of the inflammatory responses associated with the disease. Neutrophils are generally the most abundant immune cells recovered from sputum, however, the heterogeneity of these cells and their function in the airway in asthma remains incompletely understood [8]. While the cytokines and effector cells associated with a T2 inflammatory response have been extensively studied, our understanding of granulocyte populations including eosinophils and neutrophils remains limited. A more in-depth understanding of granulocyte subsets present in the airway and their function in both T2 and non-T2 disease will improve our understanding of asthma heterogeneity and move us towards more precise approaches to treating asthmatic patients [9].

Powerful technical advances now allow phenotypic profiling of protein and gene expression from primary airway cells with only a few thousand cells. These single cell technologies have provided techniques that can resolve disease in humans at an unprecedented level of detail, capturing the clinical and biological heterogeneity of disease. In addition, definition of the microbiome—from airway and gut--may provide a mechanistic link to disease severity [10]. Thus, we now have the potential to identify inflammatory pathways within a single patient, paving the way for more targeted therapies to treat asthmatics who remain refractory to medication even with current biologics [11].

We have previously taken advantage of the novel cell profiling platform, CyTOF, to detect cells in induced sputum and numerous studies have validated this platform for multiparameter profiling of single cells from heterogeneous populations [12–15]. Here we describe the methods that are currently in use to non-invasively collect and analyze cells from sputum samples from patients enrolled in the Yale Center for Asthma and Airway Diseases (YCAAD) asthma phenotyping protocol [16, 17]. We focus on the profiling of neutrophils and eosinophils using standard gating approaches and computational profiling and a hybrid of both approaches. This in-depth profiling can define functional cell subsets relevant both for investigation of disease mechanisms and for the potential integration of single-cell technology into clinical trial research.

Methods

Human Subjects

Asthmatic and healthy control participants were enrolled with written informed consent under the guidelines of the Institutional Ethics Committee of Yale University which approved this study. Samples were obtained as part of the Yale Center for Asthma and Airway Diseases (YCAAD) phenotyping protocol which collects clinical characteristics and induced sputum samples according to established protocols [18–20]. Asthma diagnosis was adopted from the Global Initiative for Asthma (GINA) guidelines including historical evidence of variable airflow obstruction determined by an improvement in FEV₁ > 12% and 200 ml compared to baseline after a short-acting bronchodilator, or treatment with corticosteroids, 20% diurnal variation of peak expiratory flow rates on 2 days over a 2–3 week period, or methacholine reactivity causing a 20% decrease in FEV1 (PC20) of <8 mg/ml [18, 21]. Healthy controls were non-smokers (former smokers had not smoked in 1 year or more and smoked ≤ 10 pack years) without fever or antibiotics at the time of sputum induction and confirmed to have no history of lung disease per patient report and/or medical record review.

Sample collection and processing

Airway cell samples were acquired by sputum induction with hypertonic saline as described previously with minimal processing and gentle treatment of cells to optimize viability [7, 12]. Mucus plugs were isolated from sputum sample on the day of collection, washed, dissolved in DTT (Calbiochem/EMD Millipore, Billerica, MA), and centrifuged to generate cell pellet and supernatant as described. Cell viability was assessed microscopically using trypan blue staining. Airway cells were resuspended in RPMI (0.5 – 2.0 × 10⁶ cells/ml) and treated with 25 U/mL DNAse (Sigma-Aldrich, St. Louis, MO) and 150U/mL collagenase IV (Worthington Chemical, Lakewood, NJ) for 15 minutes at 37°C as described previously [7, 12, 18–20].

Sample Labelling and Mass Cytometry Acquisition

Surface markers on fresh airway cells were labeled prior to fixation for optimal labeling as described previously [12]. Metal-conjugated antibodies (Table S1) for labeling cells were purchased (Fluidigm, Markham, ON), or carrier-free antibodies were conjugated in house using MaxPar X8 labeling kits according to manufacturer’s instructions (Fluidigm). EPX antibody was the generous gift of Elizabeth Jacobsen, PhD from the Mayo Clinic, Scottsdale, AZ [22]. Briefly, cells were washed with CyFACS buffer (PBS with 0.1% BSA, 2 nM EDTA, 0.05% Na azide; Maxpar® Cell Staining Buffer; Fluidigm, South San Francisco, CA), resuspended in 5 μM cisplatin (Enzo Life Sciences) for 5 minutes, and quenched with 100% FBS. Samples were incubated with 11% Fc blocking buffer (Biolegend, Dedham, MA), surface labeled for 30 min on ice, fixed (BD FACS Lyse), and frozen at -80°C. Intracellular labeling was conducted on batches of samples (10–20/day). Fixed cells were permeabilized (BD FACS Perm II) for labeling with intracellular antibodies for 45 min on ice. Cells were suspended overnight in iridium interchelator (125 nM; Fluidigm) in 2% paraformaldehyde in PBS and washed 1X in PBS and 2X in H2O immediately before acquisition. Samples were run on Helios 2 CyTOF instrument at a flow rate of 30 μL/min and a minimum of 100,000 events was collected.

CyTOF data processing and analysis

Raw FCS files were bead normalized [23] and imported to Cytobank for gating. Live cells were determined following exclusion of debris (DNA^lo), dead cells (cisplatin^hi), and doublets (event count >40) using the default data transformation of a hyperbolic arcsine with cofactor 5 (Figure S1). Individual subsets from the data were furthered compared using CyTOFKit, an R package for analyzing mass cytometry data (github.com/JinmiaoChenLab/cytofkit). Cells were visualized using a tSNE plot and clustered using the FlowSOM algorithm [24] on functional markers HLA-DR, IL-5, CD69, IL-5R, eotaxin, MIP-1β, IL-6, Siglec-8, IL-8, CD62L, CCR3, and CD16 with k = 9 clusters and random seed 42. Statistical analysis was performed using GraphPad Prism 8.

Results

Profile of airway cell phenotypes

To provide more in-depth definition of neutrophils in asthma and heterogeneity of functional status that may be relevant clinically or therapeutically, we quantified immune cell subsets of untreated sputum samples in a cohort of well-characterized adult asthmatic patients compared to healthy volunteers (Table 1). We employed a 42-marker antibody panel focused on defining neutrophil phenotypes and functional status (Table S1). Cell counts and frequencies of immune cell subsets were determined after manual gating according to a standard strategy (Fig. S1) with exclusion of debris (DNA^lo), doublets (event count >40), and dead cells (cisplatin^hi). When we compared samples of sputum from asthmatic subjects and healthy controls, we found that the total number of live cells recovered from a sputum sample was not different between asthmatic (AS) and healthy control (HC) subjects (Fig 1A; AS 29,217 ± 5,108 vs HC 62,144 ± 17,509, mean ± sem; Mann Whitney NS). The viability of cells from both groups was high (71.4%) and was not different between asthmatic patients and healthy controls (AS 70.6 ± 2.7%; HC 73.3 ± 6.2%, NS). Similarly, the percentage of live singlet cells detected was equal for both groups (28% of total events). Cytokeratin+ squamous epithelial cells were present at <5% of total and were not different between groups (AS 4.1 ± 1.3; HC 2.0 ± 0.6; NS). This demonstrates that the total yield of viable cells is sufficient for in depth characterization of sputum as we have previously shown statistically equivalent detection of major cell subjects in peripheral blood mononuclear cells from as few as 10,000 cells [25].

Table 1.

Study Subject Demographics and Clinical Status

	Severe Asthmatics (n = 7)	Moderate/Mild Asthmatics (n = 11)	Healthy Controls (n = 9)
Demographics
Age mean ± SD,	48.14 ± 14.22	50.36 ± 15.70	43.89 ± 16.82
Sex Female, (n), %	(5/7) 71.43%	(7/11) 63.64%	(8/9) 88.89%
Race/Ethnicity (n)
Caucasian	3	6	5
African American/Black	4	5	1
Asian	0	0	2
Other	0	0	1
Hispanic/Latino	1	1	0
BMI mean ± SD	38.38 ± 5.31	28.44 ± 6.33	25.39 ± 7.10
Clinical Characteristics
FEV1 % (mean ± SD)	65.5 ± 18.22	74.36 ± 16.43
L (mean ± SD)	1.98 ± 1.09	2.11 ± 0.68
BD Reversibility % (mean ± SD)	4.5 ± 6.3	8.2 ± 5.5
ICS total/day mcg (mean ± SD)	934 ± 206	268 ± 203
Biologic use	(2/7) 28.57%	(1/7) 14.29%

Figure 1. Frequency of cells from sputum.

Induced sputum samples from asthmatic (AS, n=18) and healthy control (HC, n= 9) subjects were labeled fresh on the day of collection with metal-conjugated antibodies and analyzed by mass cytometry. (A) Cell yield mean ± sem, NS (B) Frequency of cell subsets (C) Frequency of macrophage subsets. Data shown is median with interquartile range; Wilcoxon signed rank test, * P<0.01.

Altered frequency of immune cell subsets in asthmatic patients

To provide in depth profiles of airway cells in asthma, sputum samples were labeled with antibodies for cell lineage specific surface markers (n=28). We quantified differences in 13 distinct cell subsets simultaneously using a manual gating strategy to define airway cell types (Fig. S1). As expected, the most abundant live cell population in sputum in both asthmatics and healthy controls was neutrophils, which were higher in frequency in asthmatic patients consistent with increased inflammation in asthma (Fig. 1B; AS 48.2; HC 40.9; p< 0.01). Macrophages, the second most abundant cell type in sputum, were lower in asthmatic patients compared to healthy controls (Fig. 1B; AS 16.8%; HC 35.6%; P<0.01). Macrophages in sputum have historically been defined morphologically. Here we used expression levels of macrophage markers to define distinct subsets of “classical” CD14⁺⁺CD16^-, “intermediate” CD14⁺⁺CD16⁺ and “non-classical” CD14⁺/lowCD16⁺ monocytes [26] which have been shown to have distinct functional roles in tissue [27, 28]. Macrophages from sputum were predominantly CD16⁺ (non-classical) and were lower in asthmatic patients (Fig. 1C; AS 12.2% vs HC 25.8%). CD16⁺ cells are defined as more anti-inflammatory [27, 28] which may be relevant to the heterogeneity of asthma. Additional cells types present in lower frequencies are readily detected by CyTOF (Fig. 1B) and defined by expression of surface markers as T lymphocytes (CD4⁺ and CD8⁺), B lymphocytes, eosinophils, NK cells, dendritic cells, and cytokine-producing innate lymphoid cells (ILCs) [29]. No significant differences in frequencies between asthmatic patients and healthy controls were detected for these less abundant cell types.

Heterogeneity and functional status of neutrophils in asthmatic patients

Sputum cell composition across asthmatic patients was strikingly heterogenous which is a known characteristic of asthma [30]. While overall the asthmatic patients had more neutrophils compared to the healthy controls, individual asthmatics showed dramatic differences from one another in the proportion of neutrophils (Fig. 2A). Indeed, the proportion of neutrophils ranges from 6.7% to 83.3%. Other cells types were also highly variable including macrophages (1.5%53.9%) and eosinophils (range 0.1–25.9% of total). Lower levels of eosinophils may also be due to being well controlled with medication (i.e. ICS or mepolizumab) at the time of the visit [31]. Differences detected in the frequencies of 13 distinct immune cell subsets highlights significant differences expected between patients and controls (Fig. 2B) and may have implications for pathogenesis and subtyping of asthma [3, 30, 32].

Figure 2. Frequencies of sputum cell sub-populations in asthmatics and healthy controls.

Induced sputum samples from asthmatic (AS, n=18) and healthy control (HC, n= 9) subjects were labeled fresh on the day of collection with metal-conjugated antibodies and analyzed by mass cytometry. Frequency of immune cell subsets displayed using Boolean analysis (A) and median values in each population by group (B).

To assess the implications of these differences in cell frequencies for pathogenesis, we assessed the functional status of the airway cells. Differences in cellular activation status between asthmatic patients and healthy controls can be readily detected ex vivo by quantifying levels of key immune markers in untreated sputum. CyTOF supports analysis of 18 functional markers (Table S1) in each of the 13 distinct subsets simultaneously. While this in-depth multidimensional assessment provides valuable insights for cellular function, the scale and complexity of CyTOF data present challenges for routine analytic approaches. To accelerate analysis and identify additional structure in the CyTOF data, we focused on granulocyte subsets (CD66b⁺, CD15⁺; Fig S1) as these cells are the most abundant in sputum and have a known role in asthma pathogenesis. To gain functional information about subpopulations, we employed a data-driven technique for clustering and dimensionality reduction, FlowSOM [33], to generate an unsupervised self-organizing map. The FlowSOM heatmap analysis revealed 9 distinct clusters of granulocytes with different levels of expression of functional markers which can be interpreted to define functional cell subsets (Fig. 3). IL-8, a key marker and an important neutrophil chemotactic factor, is present at very high levels in cluster 1 and 9 and very low levels in cluster 3, 6 and 8. Similarly, pro-inflammatory cytokines IL-6 and MIP-1β, which identify more activated phenotypes, are very highly expressed in cluster 3 and absent in clusters 5 and 6.

Figure 3. Functional analysis of asthmatic and healthy control granulocyte populations.

Induced sputum samples were labeled fresh on the day of collection with metal-conjugated antibodies and analyzed by mass cytometry. Analysis by FlowSOM clustering heatmap of the median expression levels of functional markers of granulocyte subsets for AS (n=4) and HC (n=3).

The functional differences in the clusters were more readily visualized in a two-dimensional tSNE plot of all samples with the FlowSOM clusters overlaid (Fig. 4A). The largest number of cells are found in clusters 1 and 4 and the fewest in clusters 6 and 8. The majority of the neutrophils were in cluster 1, which includes expression of IL-17A and IL-8, both pro-inflammatory cytokines associated with neutrophil recruitment and activation. Further, striking differences in neutrophil status are apparent when the clusters are separated between the asthmatic patients (Fig. 4B) and healthy controls (Fig. 4C). In-depth examination of clusters being expressed in one but not the other group shows clusters 2, 3, and 4 were highly expressed in asthmatics and absent in the healthy controls. Markers in these clusters reflect high expression of IL-6, MIP-1β, CCR3, Siglec-8, IL-13, IL-17 which are characteristic of an inflammatory activity associated with cell activation [34]. In contrast, clusters 5 and 6 are largely present only in healthy controls and express little to no cytokines, and are negative for CD69, a marker of cell activation, and would be expected to contribute less to inflammatory processes. Both clusters 5 and 6 contain cells expressing HLA-DR, while cluster 5 was positive for the IL-4Rα and cluster 6 had increased IL-5R and IL-13R1 expression. This clustering suggests two specialized subsets of neutrophils in the airway of healthy donors which, in an individual with asthma, could become activated when there is T2 inflammation and increased levels of IL-4, -5, and -13 being generated by activated bronchial epithelial cells and T cells. Cluster 4, which is also highly expressed in asthma and largely absent in healthy controls, includes neutrophils which are positive for eotaxin expression, a recruitment factor for eosinophils, and IL-7R, which plays an important role in the activation of Type 2 innate lymphoid cells [35]. Neutrophil subsets in cluster 4 also express low levels of CD62L (CD62L^dim), a marker of an activated neutrophil subset reported to play a role in inflammation and airway cell responses [36, 37]. Expression of CCR3 and Siglec-8 in clusters 2 and 3 suggest these clusters include eosinophils, which are critical mediators of allergic asthma. Indeed, detailed examination of cluster 3, with high expression of IL-3R, CD69, IL-5, combined with low levels of CD62L, suggests that this population may represent inflammatory eosinophils, a recently described subset in asthmatic sputum [38]. Future studies will benefit from further definition of eosinophil populations with use of the eosinophil peroxidase marker (EPX) (Fig S2).

Figure 4. Functional clustering of asthmatic and healthy control granulocyte populations.

Induced sputum samples were labeled fresh on the day of collection with metal-conjugated antibodies and analyzed by mass cytometry. Visualization of FlowSOM clusters by Tsne for neutrophil clusters from all samples (A) and separated for AS (B) andHC (C).

Discussion

Here we have employed the high parameter cellular data obtained using CyTOF to provide detailed functional status of airway cells. We have identified multiple subtypes of airway immune cells which show distinct profiles both between healthy controls and asthmatic patients, and within an asthmatic patient cohort. This demonstrates the power that single cell profiling technologies bring to understanding phenotypes of asthma that have broad therapeutic and scientific implications. CyTOF can overcome the historical obstacles of tremendous heterogeneity in a relatively small cell sample size of sputum. The enhanced marker detection (>45 markers) combined with advanced clustering techniques dramatically improves our ability to analyze live heterogenous cell subsets. Improved recognition of distinct cellular phenotypes and functions provide insights into cell functions such as recently described low density neutrophils which correlate with disease progression in lung cancer patients [39]. In concert with single cell RNASeq, these advances have recently identified a new cell type, the ionophore, that contributes to the pathogenesis of cystic fibrosis [40], and changes in alveolar macrophages and epithelial cells during the progression of pulmonary fibrosis [41]. These approaches may identify relevant pathways that significantly advance the personalization of clinical asthma management and identification of novel therapeutic mediators.

As these in-depth methods gain a foothold in translational studies, it is a natural progression to expand the use of CyTOF into clinical trials. Use of single-cell technologies in clinical trials can augment findings with important mechanistic insights derived from collected samples. High dimensional datasets can give investigators a comprehensive view of disease pathogenesis within individual patients and across cohorts, shed light on heterogeneous cellular responses, and may help reveal mechanisms relevant to understanding responders and non-responders. To this end, our group has established relevant parameters to support integrating single cell technology into a multi-site clinical trial. In collaboration with the Inner-City Asthma Consortium (ICAC), we have introduced an immune sub-study to the ongoing MUPPITS-2 trial, a phase III study evaluating the efficacy of mepolizumab in pediatric asthmatics (ClinicalTrials.gov Identifier: NCT03292588) [42, 43]. We have adapted our methods to assess sputum samples collected as part of the trial and hope that the addition of CyTOF to this study will identify meaningful differences that contribute to diverse clinical responses and advance our knowledge of asthmatic inflammation.