Metabolic differences between bronchial

epithelium from healthy individuals and patients

with asthma and the effect of bronchial
thermoplasty
Abilash Ravi, PhD,a,b* Annika W. M. Goorsenberg, MD, PhD,a* Annemiek Dijkhuis, MSc,b Barbara S. Dierdorp, BSc,b
Tamara Dekker, BSc,b Michel van Weeghel, PhD,c Yanaika S. Sabogal Pin~ eros, PhD,a,b Pallav L. Shah, MD, FRCP,d,e,f
Nick H. T. ten Hacken, MD, PhD, Jouke T. Annema, MD, PhD, Peter J. Sterk, MD, PhD,a Fre
g a
deric M. Vaz, PhD,c
Peter I. Bonta, MD, PhD, à and Rene
a
 Lutter, PhD à
a,b
Amsterdam and Groningen, The Netherlands, and London, United Kingdom

GRAPHICAL ABSTRACT

From athe Department of Respiratory Medicine, bthe Department of Experimental Immu- Disclosure of potential conflict of interest: The authors declare that they have no relevant
nology, Amsterdam Infection and Immunity Institute, and cthe Laboratory Genetic conflicts of interest.
Metabolic Diseases, Core Facility Metabolomics, Department of Clinical Chemistry, Received for publication July 7, 2020; revised November 23, 2020; accepted for publi-
Amsterdam University Medical Center, University of Amsterdam; dRoyal Brompton cation December 1, 2020.
Hospital, London; eNational Heart and Lung Institute, Imperial College, London; Available online February 6, 2021.
f
Chelsea and Westminster Hospital, London; and gDepartment of Pulmonology, Uni- Corresponding authors: Abilash Ravi, PhD, B02-075, Department of Pulmonology, Lei-
versity Medical Center Groningen, University of Groningen. den University Medical Center, Albinusdreef 2, 2333 ZA, Leiden, The Netherlands.
*These authors contributed equally to this work. E-mail: a.ravi@lumc.nl. Or: Rene Lutter, PhD, K0-150 Department of Respiratory
àThese authors contributed equally to this work. Medicine and Experimental Immunology, Amsterdam University Medical Center,
Supported by the Netherlands Asthma Foundation (currently the Lung Foundation [pro- Meibergdreef 9, 1105 AZ, Amsterdam, The Netherlands. E-mail: r.lutter@
jects 3.2.10.069, 3.2.07.012, and 3.2.06.031]), GSK (CRT 114696), and Stichting amsterdamumc.nl.
Astma Bestrijding (project 2013_009). The explorative trial RILCO (Role of Innate The CrossMark symbol notifies online readers when updates have been made to the
Lymphoid Cells in COPD) was supported in part by Medimmune (Gaithersburg, article such as errata or minor corrections
Md). The TASMA (Unravelling Targets of Therapy in Bronchial Thermoplasty in Se- 0091-6749
vere Asthma) trial is funded by the Netherlands Lung Foundation (grant Ó 2021 The Authors. Published by Elsevier Inc. on behalf of the American Academy of
5.2.13.064JO), The Netherlands Organization for Health Research and Development Allergy, Asthma & Immunology. This is an open access article under the CC BY li-
(ZonMw grant 90713477), and Boston Scientific Corporation. cense (http://creativecommons.org/licenses/by/4.0/).
https://doi.org/10.1016/j.jaci.2020.12.653

1236
J ALLERGY CLIN IMMUNOL RAVI ET AL 1237
VOLUME 148, NUMBER 5

Background: Asthma is a heterogeneous disease with

differences in onset, severity, and inflammation. Bronchial Abbreviations used
epithelial cells (BECs) contribute to asthma pathophysiology. BEC: Bronchial epithelial cell
Objective: We determined whether transcriptomes of BECs BMP: Bis(monoacylglycero)phosphate
reflect heterogeneity in inflammation and severity in asthma, BT: Bronchial thermoplasty
COPD: Chronic obstructive pulmonary disease
and whether this was affected in BECs from patients with severe
CXCL-8: C-X-C motif chemokine ligand 8/IL-8
asthma after their regeneration by bronchial thermoplasty. GSEA: Gene set enrichment analysis
Methods: RNA sequencing was performed on BECs obtained by ICS: Inhaled corticosteroid
bronchoscopy from healthy controls (n 5 16), patients with mild IPA: Ingenuity Pathway Analysis
asthma (n 5 17), patients with moderate asthma (n 5 5), and LPC: Lysophosphatidylcholine
patients with severe asthma (n 5 17), as well as on BECs from OXPHOS: Oxidative phosphorylation
treated and untreated airways of the latter (also 6 months after PC: Phosphatidylcholine
bronchial thermoplasty) (n 5 23). Lipidome and metabolome TASMA: Unravelling Targets of Therapy in Bronchial Thermo-
analyses were performed on cultured BECs from healthy plasty in Severe Asthma
controls (n 5 7); patients with severe asthma (n 5 9); and, for TIA-1: T-cell internal antigen-1
TiAR: T-cell internal antigen-1 related protein
comparison, patients with chronic obstructive pulmonary
disease (n 5 7).
Results: Transcriptome analysis of BECs from patients showed
a reduced expression of oxidative phosphorylation (OXPHOS)
genes, most profoundly in patients with severe asthma but less likely that the underlying innate immune mechanisms, especially
profoundly and more heterogeneously in patients with mild those directed by bronchial epithelium, may at least in part be
asthma. Genes related to fatty acid metabolism were shared.
significantly upregulated in asthma. Lipidomics revealed Previous studies of the differences between epithelial cells
enhanced levels of lipid species (phosphatidylcholines, from patients with asthma in particular have focused on unbiased
lysophosphatidylcholines. and bis(monoacylglycerol)phosphate), stratification of cohorts of patients with asthma by grouping into
whereas levels of OXPHOS metabolites were reduced in BECs different clusters and on distinguishing asthma phenotypes by
from patients with severe asthma. BECs from patients with mild using transcriptomics9 and proteomics.10 Omics approaches
asthma characterized by hyperresponsive production of comparing patients and healthy control subjects, however, are
mediators implicated in neutrophilic inflammation had virtually lacking. The aim of this study was to compare the tran-
decreased expression of OXPHOS genes compared with that in scriptome of bronchial epithelial cells (BECs) from patients with
BECs from patients with mild asthma with normoresponsive mild to moderate to severe asthma with transcriptome of BECs
production. BECs obtained after thermoplasty had significantly from healthy controls, substantiate major differences between
increased expression of OXPHOS genes and decreased them, and relate these differences to the previously described
expression of fatty acid metabolism genes compared with BECs translational defect. To put these findings into further perspective,
obtained from untreated airways. we performed additional analysis in patients with asthma who un-
Conclusion: BECs in patients with asthma are metabolically derwent bronchial thermoplasty (BT). BT is a nonpharmacologic
different from those in healthy individuals. These differences bronchoscopic treatment for severe asthma that delivers radiofre-
are linked with inflammation and asthma severity, and they can quency energy to heat the airway wall and thereby target airway
be reversed by bronchial thermoplasty. (J Allergy Clin Immunol smooth muscle cells. Furthermore, the bronchial epithelium is the
2021;148:1236-48.) most superficial airway wall layer that encounters radiofrequency
heating energy, which has been shown to result in epithelial
Key words: Bronchial epithelium, metabolism, thermoplasty sloughing as detected by optical coherence tomography directly
after BT.11,12 Therefore, we postulated that BT may subsequently
lead to restoration of bronchial epithelium, which would be re-
The airways of patients with obstructive airway diseases such flected in the bronchial epithelial transcriptome.
as asthma and chronic obstructive pulmonary disease (COPD) are
chronically inflamed.1 Airway epithelial cells together with mac-
rophages constitute the first line of defense to interact and respond METHODS
to external stimuli, thus driving immune responses. This, together Subjects and design
with their abundance, makes airway epithelial cells potential ma- This study comprised a cross-sectional design using bronchial epithelial
jor contributors to the inflammation and pathogenesis of asthma2 brushes from 3 trials and 2 explorative studies. The RESOLVE trial
and COPD.3 Recently, we identified a defect in translational con- (NCT1677)13 involved patients with mild asthma and healthy controls. The
trol in bronchial epithelial cells from patients with asthma that MATERIAL trial (NTR01520051)14 enrolled patients with mild asthma,
correlated well with neutrophilic responses.4 Several earlier re- and the RILCA (Role of Innate Lymphoid Cells in Asthma) study
(NL48912.018.14) enrolled patients with moderate asthma and patients with
ports have shown that both genetic and acquired defects5 in
severe asthma from the TASMA (Unravelling Targets of Therapy in Bronchial
airway epithelial cells6 can lead to impaired mucociliary clear-
Thermoplasty in Severe Asthma) (NCT02225392) trial.15 The patients with
ance7 and barrier function.8 Apart from genetic defects that COPD were patients included from the RILCO (Role of Innate Lymphoid
may underlie such a defect, airway epithelial cells are continu- Cells in COPD) study (NL53354.018.15). All study protocols were reviewed
ously exposed to an inflammatory milieu, as a consequence of and approved by the ethical review committee and were in accordance with the
which epithelial cell biology may change. Even though asthma Declaration of Helsinki. All study participants provided written informed con-
and COPD are different in etiology and pathophysiology, it is sent. The collection of BECs in the aforementioned trials and studies was
1238 RAVI ET AL J ALLERGY CLIN IMMUNOL
NOVEMBER 2021

conducted by 1 center, namely, the Department of Respiratory Medicine of the were sampled and mapped to annotated genomic references. Mapping
Amsterdam University Medical Center, Amsterdam, The Netherlands. The positions were classified as intragenic, exonic, intergenic, intronic, and
TASMA trial was conducted at the Department of Pulmonology, University rRNA. Clustering and DNA sequencing were performed according to the
Medical Center Groningen, Groningen, The Netherlands, and Royal Bromp- manufacture’s protocol using the Illumina NextSeq 500. A concentration of
ton and Imperial College Hospital London, United Kingdom. The baseline 1.6 pM was used as the input. The reads were trimmed for adapter
characteristics of the patients with asthma and healthy controls who were sequences by using Trimmomatic, version 0.30, before the alignment.
involved in this study are provided in Table I. Presumed adapter sequences were removed from the read when the bases
The inclusion and exclusion criteria for patients with mild asthma (from the matched a sequence in the adapter sequence set (TruSeq adapters) with 2 or
RESOLVE and MATERIAL trials) are described elaborately16 in the Methods fewer mismatches and an alignment score of at least 12. The reference
section of the Online Repository (available at www.jacionline.org). Patients Homo_sapiens.GRCh37.75 was used to align the reads. The mapping to the
with severe asthma (from the TASMA trial) fulfilling the World Health Orga- reference sequence was done by using a short reader aligner based on
nization or modified innovative medicines initiative criteria of severe refrac- Burrows-Wheeler transform. The default of mismatch rate of 2% (3
tory asthma were included.17,18 The RILCA (Role of Innate Lymphoid Cells mismatches in a read of 150 bases) was used. The frequency of the reads
in Asthma) study and RILCO (Role of Innate Lymphoid Cells in COPD) mapped on the transcript was determined in terms of counts, which were
were 2 explorative studies involving the impact of rhinovirus-16 challenge used as an input for the downstream analysis.
in patients with moderate asthma and patients with COPD, respectively, and Additionally, reads per kilobase of exon per million reads mapped and
exploring mechanisms driven by innate lymphoid cells. The inclusion and fragments per kilobase of exon per million reads mapped values were
exclusion criteria are mentioned in detail in the Methods section of the Online calculated. Image analysis, base calling, and quality check were performed
Repository. with Illumina data analysis pipeline RTA, version 2.4.11, and Bcl2fastq,
version 17. The read counts were loaded into the DESeq2 software package,
which is a statistical package within the R/BioConductor platform that is
Corticosteriod use designed to determine the differentially expressed genes and in which an
The patients with mild asthma had not received inhaled or systemic adjusted P value less than .05 was considered statistically significant. The DE-
corticosteroids or any other treatment other than inhaled short-acting b2-ago- seq2 method uses Benjamini-Hochberg correction as an adjustment for false
nists within 2 weeks before the start of the study. The patients with moderate discovery rates.
asthma used a dose equivalent of no more than 500 mg of fluticasone propio- For the heat maps, log2 gene expression values were normalized with z
nate per day. The patients with severe asthma had been using an inhaled corti- scores calculated by using the equation X – (m/s), where X is the value of
costeroid (ICS) at a dosage of at least 500 mg of fluticasone equivalent per day the individual sample, m is the average of the row, and s is the SD of the
and a long-acting ß2-agonist at a dosage of at least 100 mg of salmeterol per row. The clustering of genes represented in the heat maps was based on
day or equivalent for the past 6 months, as well as a systemic corticosteroid k-means clustering. The combination of z scores for all genes is represented
_20 mg of prednisone equivalent per day). The patients with COPD were al-
(< below the heat maps in red and blue, with red indicating high expression
lowed specific medication, namely, long-acting ß2-agonists and long-acting and blue indicting low expression.
muscarinic agonists, but no ICSs.

GSEA
Sampling and RNA isolation The gene sets were identified by using the online platform accessed via the
Bronchoscopy was performed according to a standardized method.14 website www.gsea-msigdb.org. All differentially expressed genes with a q
Mucosal brushes collected by brushing the left lower lobe consisted pre- value less than 0.05 compared between 2 groups were used for gene set enrich-
dominantly (>95%) of bronchial epithelial cells. Two brushes were ob- ment analysis (GSEA). The gene identifiers were uploaded in the molecular
tained from each participant and then pooled and pelleted by signature database of the website, and the overlapping gene sets were
centrifugation at 1240 rpm (using a Rotanta 460S centrifuge) for 10 mi- computed. The P values were derived on the basis of hypergeometric distribu-
nutes at 48C. The pellet was dissolved in 1 mL of TRIzol (Thermofischer tion for k – 1, K, N – K, and n, where k is the number of genes in the intersection
Scientific, Paisley, United Kingdom) and stored at –808C until the RNA of the query set with a set from the Molecular Signature Database, K is the
was isolated. After all the samples had been obtained, they were thawed number of genes in the set from the Molecular Signature Database), N is the
at room temperature, and after 200 mL of chloroform was added, they total number of all human genes, and n is the number of genes in the query
were shaken vigorously for 30 seconds. The samples were kept at room set). The false discovery rate (q value) analog of the hypergeometric P value
temperature for 10 minutes, after which the phases were separated by was determined by correction for multiple hypothesis testing according to
centrifugation at 16,000 g for 15 minutes at 48C. The aqueous phase was Benjamini and Hochberg. The gene sets thus identified were plotted in heat
concentrated with protocol 5.3 using the NucleoSpin system and an maps to visualize gene expression of all individuals separately.
RNA XS extraction kit (Macherey-Nagel, Duren, Germany). The quality
and concentration of the samples were assessed by using a fragment
analyzer (Advanced Analytical Technologies, Inc, Ankeny, Iowa). IPA
Ingenuity Pathway Analysis (IPA) is used to quickly visualize and
understand complex omics data and perform insightful data analysis and
RNA sequencing, analysis, and heat maps interpretation. The IPA output represented in Fig 1, E is based on all genes
The NEBNext Ultra II Directional RNA Library Prep Kit for Illumina differentially expressed in patients with severe asthma versus those expressed
(Ipswich, Mass) was used to process the samples. The sample preparation in healthy controls with a q value less than 0.05 (see Table E1 in this article’s
was performed according to the protocol for the NEBNextUltra Directional Online Repository at www.jacionline.org). These genes were uploaded in IPA,
RNA Library Prep Kit for Illumina (NEB, catalog no. 7420S/L). Briefly, which displays the most significantly affected top canonic pathways. The
oligo-dT magnetic beads were used to isolate mRNA from total RNA. output of graphical illustration shows all of the genes involved in the pathway,
cDNA synthesis was performed after fragmentation of the mRNA. This was with the genes differentially expressed being highlighted.
used for ligation with sequencing adapters followed by PCR amplification of
the resulting product. The quality and yield after sample preparation were
measured with the fragment analyzer. The size of the resulting products was Lipidomics and metabolome analysis
consistent with the expected size distribution of between 300 and 500 bp. To The BECs for lipidomics and metabolomics were obtained by brush during
evaluate the quality of the library preparation and kits used, the raw data bronchoscopy (P0) from healthy individuals, patients with severe asthma, and
J ALLERGY CLIN IMMUNOL RAVI ET AL 1239
VOLUME 148, NUMBER 5

TABLE I. Baseline characteristics of patients with mild, moderate, and severe asthma and healthy controls for transcriptome
analyses
Patients with Patients with Patients with
Characteristic Healthy controls mild asthma moderate asthma severe asthma

Subjects (no.) 16 17 5 17
Age (y), mean (min-max) 22.42 (19-31) 22.82 (18-38) 37 (23-52) 44.82 (25-66)
Sex ratio (male:female), no./no. 4:12 6:11 3:2 5:12
FEV1 before bronchodilator (L), mean (SD; min-max) 4.05 (0.72; 3.44-4.93) 3.94 (0.85; 2.56-5.16) 3.27 (0.88; 2.11-4.53) 2.79 (0.76; 1.7-4.37)
FEV1 after bronchodilator (L), mean (SD; min-max) 4.12 (0.72; 3.52-5.05) 4.23 (0.84; 3.06-6.14) 3.41 (0.91; 2.21-4.64) 3.07 (0.76; 2.11-4.49)
FEV1 before bronchodilator (% predicted), mean (SD; 105.5 (10.92; 90-121) 97.43 (6.64; 86-109) 93.5 (18.76; 93.5-122) 85.35 (26; 38-129)
min-max)
FEV1 after bronchodilator (% predicted), mean (SD; 107.5 (11.15; 98-124) 103.83 (8.63; 90-120) 97 (18.49; 73-125) 100.88 (20.91; 67-137)
min-max)
FEV1 reversibility (%),median (SD; min-max) 3 (2.86; –2 to 7) 7.5 (4.36; –4 to 12) 3 (2.87; 0-8) 10 (17.23; 1-64)
PC20 (mg/mL), median (SD; min-max) >16.0 1.49 (0.03-7.35) na 0.28 (0.0075-5.53)
FENO (ppb), mean (SD; min-max) 21.9 (11.4) 58.03 (37.14) 21 (10.04) 52 (63.97)
FENO, Fraction of exhaled nitric oxide represented in terms of ppb; max, maximum; min, minimum; na, not available; PC20, histamine provocative concentration causing a 20%
drop in FEV1.

patients with COPD. The baseline characteristics for participants subjected to RESULTS
these analyses are provided in Table II.
Dysregulation of metabolic genes in bronchial
The BECs were plated on 6-well plates precoated with PurCol (Advanced
Biomatrix, Carlsbad, Calif) and grown until confluence in bronchial epithelial epithelium from patients with severe asthma
basal medium (Lonza, Basel, Switzerland) supplemented with growth factors Despite heterogeneity within each group, there were 472 genes
(Lonza) and ciprofloxacin (Sigma, St Louis, Mo) at a concentration of 2 mg/ differentially expressed in the BECs from patients with mild
mL. The cells were then passaged into T25 flasks (passage 1) until grown asthma (see Table E2 in this article’s Online Repository at www.
confluent. Subsequently, the cells were detached by using trypsin/EDTA jacionline.org) and 640 genes in those BECs from patients with se-
(Lonza), and after the addition of trypsin-neutralizing solution, they were vere asthma (see Table E1) compared with the numbers in the
pelleted by centrifugation for 7 minutes at 1240 rpm (with a Rotanta 460S BECs from healthy controls (q < 0.05). There were no genes
centrifuge) at 88C. The pellet was washed twice with PBS and stored at –808C differentially expressed between the BECs from patients with
until all samples were analyzed in parallel.
mild asthma and the BECs from patients with severe asthma,
Lipidomics and metabolome analysis was performed on a high-
with the q value cutoff set at less than 0.05. Because of the small
performance liquid chromatography system (Ultimate 3000 binary high-
performance liquid chromatography system, Thermo Scientific, Waltham, number of patients with moderate asthma used for analysis, there
Mass), as described previously.19 The extract of the cell pellet was injected were was no comparison of differentially expressed genes in BECs
onto a normal phase column (LiChrospher 2 3 250-mm silica-60 column) between bronchial epithelium from patients with moderate
and a reverse-phase column (Acquity UPLC HSS T3, Milford, Mass). A Q Ex- asthma and bronchial epithelium from healthy controls. GSEA
active plus Orbitrap (Thermo Scientific) mass spectrometer was used in the of differentially expressed genes in BECs from patients with
negative and positive electrospray ionization modes. In both ionization modes, mild asthma versus in BECs from healthy controls showed gene
mass spectra of the lipid species were obtained by continuous scanning from sets belonging to mitotic spindle, IL-2–signal transducer and acti-
mass-to-charge ratio 150 to mass-to-charge ratio 2000 with a resolution of vator of transcription 5 (STAT5) signaling, UV response, oxida-
280,000 full width at half maximum. Detailed analysis of the data is provided
tive phosphorylation (OXPHOS), and inflammatory response
in the Methods section of the Online Repository.
(Fig 1, A). Similar comparisons for patients with severe asthma
and healthy controls revealed gene sets belonging to OXPHOS,
BT IFN-a response, mitotic spindle, apical junction, and late estrogen
The TASMA trial was conducted by using a protocol approved by the response (Fig 1, B).
Medical Ethics Committee (NL45394.018.13).20 All subjects provided prior Both the mitotic spindle and OXPHOS gene sets stood out in
written, informed consent. The design of the TASMA trial has been elabo- terms of BECs from patients with mild asthma and patients with
rately described elsewhere.12,20 A total of 23 patients21 (6 patients with severe severe asthma, with the mitotic spindle highest in the BECs of
asthma were included in addition to those in the severe asthma group patients with mild asthma and OXPHOS highest in the BECs from
mentioned in Table I) were treated with BT by using the Alair System (Boston patients with severe asthma. We further investigated the OXPHOS
Scientific, Boston, Mass) according to the current standard22 and sedated by
gene set. Heat maps of 15 OXPHOS genes were reduced (shown in
using remifentanil/propofol23 or general anaesthesia. Patients were treated
green) (Fig 1, C) in BECs from the patients with severe and mod-
with 50 mg of prednisolone 3 days before treatment, on the day of the proced-
ure itself, and 1 day thereafter. During the first procedure, the right lower lobe erate asthma in comparison with those from the controls (shown in
was treated; during the second procedure, the left lower lobe was treated; and red) (Fig 1, C). In the patients with mild asthma there was a hetero-
finally, both upper lobes were treated. The right middle lobe remained un- geneous gene expression pattern, with 5 of 17 patients showing
treated; therefore, this region could be used to compare the treatment effect. high expression (red) and the remainder showing low expression
Six months after BT, bronchoscopy was performed. During the bronchoscopy (green) of OXPHOS genes (Fig 1, C). Apart from these metabolic
procedure, endobronchial brushes were obtained from the untreated middle OXPHOS genes, fatty acid metabolism–related genes (fatty acid
lobe and treated left lower lobe airways. The baseline characteristics of the pa- synthase [FASN], arachidonate 15-lipoxygenase [ALOX15], phos-
tients with severe asthma used to analyze the effects of BTon bronchial epithe- pholipase C eta 1 [PLCH1], pyruvate dehydrogenase phosphatase
lium are provided in Table III; they are no different from those reported
regulatory subunit [PDPR], acyl-CoA oxidase 3 [ACOX3], and
originally for all patients.21
1240 RAVI ET AL J ALLERGY CLIN IMMUNOL
NOVEMBER 2021

FIG 1. Metabolic genes in BECs from patients with asthma versus in BECs from healthy controls. GSEA
showed differentially expressed gene sets in BECs from patients with mild (A) and severe (B) asthma, with
green indicating highly significant and black indicating less significant q values (< 0.05). OXPHOS gene
expression in BECs from patients with severe (n 5 17), moderate (n 5 5), and mild asthma (n 5 17) versus
in those from controls (n 5 16) (C). D, As in (C), but for genes related to fatty acid metabolism (in red and
green, with red indicating high expression and green indicating low expression). The combined z scores
for the genes shown are represented below the heat maps (in red and blue, with red indicating high expres-
sion and blue indicating low expression). IPA of differentially expressed genes of the respiratory chain in
BECs from patients with severe asthma versus in BECs from healthy controls (n 5 17). E, Significantly down-
regulated genes (adjusted P value of <.05) are shown in pink.
J ALLERGY CLIN IMMUNOL RAVI ET AL 1241
VOLUME 148, NUMBER 5

FIG 1. (Continued).

TABLE II. Baseline characteristics of patients with severe asthma, patients with COPD, and healthy controls for lipidomic and
metabolomic analyses
Characteristic Healthy controls Patients with severe asthma Patients with COPD

Subjects (no.) 7 9 6
Age (y), mean (min-max) 37 (24-56) 52.8 (40-63) 63.8 (52-74)
Sex ratio (male:female), no./no. 3:4 4:5 3:3
FEV1 pre-bronchodilator (L), mean (SD; min-max) 3.94 (0.98; 2.6-5.09) 2.66 (0.33; 2.18-3.1) 2.228 (0.49; 1.49-2.99)
FEV1 post-bronchodilator (L), mean (SD; min-max) 4.15 (1; 2.77-5.37) 2.61 (0.5; 1.58-3.14) 2.41 (0.48; 1.61-3.06)
FEV1 pre-bronchodilator (% predicted), mean (SD; min-max) 105.75 (0.82; 105-107) 84.33 (23.13; 41-107) 70.8 (4.95; 65-77)
FEV1 post-bronchodilator (% predicted), mean (SD; min-max) 111.5 (2.17; 108-114) 90.66 (21.15; 52-115) 77.2 (4.35; 72-85)
FEV1 reversibility, median (SD; min-max) 6.5 (1.63; 3-7) 8 (4.4; 0-11) 6 (3; 2-11)
max, Maximum; min, minimum.

acetyl-CoA carboxylase-b [ACACB]) were also shown by heat BECs and depicted the extent of differences in lipid profiles
maps. Interestingly, in the BECs from all patients with asthma between the 2 groups. Phosphatidylcholines (PCs), lysophospha-
(mild, moderate, or severe), these genes were upregulated (red) tidylcholines (LPCs), lysophosphatidylethanolamines, and bis(-
compared with those in the healthy controls (Fig 1, D). In addition monoacylglycero)phosphates [BMPs]) (shown in green) (Fig 2,
to GSEA, IPA also showed that the OXPHOS pathway was the top A) were significantly upregulated (P < .05 and log2 fold
dysregulated pathway in BECs from patients with severe asthma change > 1) in the BECs from patients with severe asthma
compared with in BECs from healthy subjects. The downregu- compared with in the BECs from healthy controls. To visualize
lated genes were from complexes I, III, IV, and V of the electron the 28 lipid mediators that were significantly upregulated in
transport chain (Fig 1, E). Heat maps of 7 mitotic spindle genes BECs from patients with severe asthma, we plotted these in
displayed significantly enhanced expression, particularly in heat maps. The levels of lipid species (PCs, LPCs, lysophospha-
BECs from patients with mild and moderate asthma, but not in tidylethanolamines, and BMPs) were increased in the BECs from
those from patients with severe asthma (see Fig E1 in the Online the patients with severe asthma (red) compared with in those from
Repository at www.jacionline.org). The RNA sequencing data the healthy controls (blue) (Fig 2, B). The metabolome analyses in
sets comparing bronchial epithelium from patients with mild, BECs from patients with severe asthma and healthy controls did
moderate, and severe asthma with that from healthy controls can not show a clear distinction between these 2 groups (Fig 2, C). The
be accessed by using the link https://www.ncbi.nlm.nih.gov/geo/ metabolomic analysis measured a total of 75 metabolites, 16 of
query/acc.cgi?acc5GSE161245. which are depicted in the heat maps in Fig 2, C. The remaining
59 metabolites that were not significantly different between pa-
tients with asthma and healthy controls are shown in Table E3
Elevated lipid profiles in BECs from patients with (in this article’s Online Repository at www.jacionline.org).
severe asthma versus in BECs from controls
To substantiate the findings for metabolic genes, lipid profiles
and metabolome were analyzed after expansion of the bronchial Distinct lipid species in BECs from patients with
epithelium ex vivo. As there was a clear and distinct reduction of COPD versus in BECs from healthy controls and
OXPHOS gene expression in BECs from patients with severe patients with severe asthma
asthma compared with in BECs from the controls, we compared Previous reports showed altered metabolites and glycerophos-
these cohorts. Volcano plots showed all lipid species quantified in pholipids in PBMCs and airway smooth muscle cells24 from
1242 RAVI ET AL J ALLERGY CLIN IMMUNOL
NOVEMBER 2021

TABLE III. Baseline characteristics of patients with severe asthma who underwent the BT procedure
Characteristic Value in patients with severe asthma

Subjects (no.) 23
Age (y), mean (min-max) 47.69 (25-64)
Sex ratio (male:female), no./no. 7:16
FEV1 before bronchodilator (L), mean (SD; min-max) 2.67 (0.57; 1.59-3.66)
FEV1 after bronchodilator (L), mean (SD; min-max) 3.03 (0.54; 2.11-4.26)
FEV1 before bronchodilator (% predicted), mean (SD; min-max) 89.3 (19.89; 57-129)
FEV1 after bronchodilator (% predicted), mean (SD; min-max) 101.13 (18.66; 67-137)
FEV1 reversibility, mean (SD; min-max) 10 (11.65; 0-57)
PC20 (mg/mL), mean (SD; min-max) 0.41 (0.0075-32)
ACQ score before BT, mean (SD; min-max) 2.62 (0.57; 1.5-3.66)
AQLQ score before BT, mean (SD; min-max) 3.98 (0.81; 2.65-6)

ACQ, Asthma Control Questionnaire; AQLQ, Asthma Quality of Life Questionnaire; max, maximum; min, minimum; PC20, histamine provocative concentration causing a 20%
drop in FEV1 value.

patients with COPD.25 Hence, we analyzed whether BECs from Hyperresponsive BECs have reduced expression of
patients with COPD also displayed a similar difference in meta- OXPHOS genes
bolic status after ex vivo expansion. The heat maps of lipidomics Earlier, we identified BECs from patients with mild asthma
analysis of BECs from patients with COPD compared with BECs with a defective translational control as hyperresponsive (exag-
from healthy controls showed a prominent upregulation of mainly gerated and corticosteroid-unresponsive production of C-X-C
PC lipid species. In addition, lipid species (triglycerides and motif chemokine ligand 8/IL-8 [CXCL-8], and other cytokines)
LPCs) were also enhanced in BECs from patients with COPD as opposed to normoresponsive (lower and corticosteroid-
(Fig 3, A). Strikingly, when BECs from COPD were compared responsive production of CXCL-8 and other cytokines).4 This
with those from patients with severe asthma, there was also a sig- epithelial hyperresponsiveness was correlated with neutrophilic
nificant increase in triglycerides in the patients with COPD (Fig 3, inflammation. The transcriptome of hyperresponsive BECs
B). With respect to the metabolites analyzed, the levels of uridine from patients with mild asthma (subgroup from Table I and Fig
diphosphate (UDP)-hexose, 2-dehydrogluconate-6P, deoxyade- 1) displayed a prominent reduced expression of OXPHOS genes
nosine monophosphate (dAMP), oxidized nicotinamide adenine compared with that in their normoresponsive counterparts (Fig
dinucleotide (NAD1), fructose 1,6 diphosphate, creatine-P, gluta- 4, E).
mine, a-ketoglutarate, nicotinamide adenine dinucleotide
hydrogen (NADH), adenine, and creatinine were significantly
lower than in the BECs from patients with COPD than in the DISCUSSION
BECs from the healthy controls. Heat maps representing these Unbiased transcriptome analysis of BECs from patients with
different metabolites in the 2 groups are provided in Fig 3, C. severe asthma compared with those from healthy controls showed
The other 64 metabolites with levels that were not significantly a profound reduction in OXPHOS genes belonging to complexes
different between bronchial epithelium from patients with I, III, IV, and V of the electron transport chain, whereas for BECs
COPD and bronchial epithelium from healthy controls are from patients with mild asthma, this reduction was heteroge-
shown in Table E4 (in this article’s Online Repository at www. neous. Genes related to fatty acid metabolism, however, were
jacionline.org). significantly upregulated in BECs from all patients with asthma,
thus differentiating patients with asthma from healthy controls.
This differential expression in bronchial epithelium was validated
BT treatment–induced metabolic shift in BECs by lipidomics with enhanced levels of lipid species (PCs, LPCs,
BECs obtained from treated and, as controls, untreated and BMPs). The reduction in metabolites in bronchial epithelium
airways (ie, the right middle lobe) of patients with severe of patients with severe asthma was observed trendwise only, with
asthma 6 months after initiation of the treatment were subjected no clear differences. Most interestingly, in BECs from patients
to transcriptomics to analyze differentially expressed genes. with severe asthma who had received BT, a metabolic shift toward
Interestingly, the heat maps (Fig 4, A) and z scores (Fig 4, B) that in BECs from healthy individuals was observed. In BECs
showed that the BECs obtained from the treated airways had from patients with COPD also, there was a marked upregulation
significantly upregulated OXPHOS genes compared with the of lipid profiles compared with that in BECs from healthy
BECs from untreated airways. In addition, reduced glycolysis controls, and the upregulation was even more pronounced than
gene expression in the BECs from treated airways was observed that in BECs from patients with severe asthma.
in the heat maps (Fig 4, C) and z scores (Fig 4, D) when Together, these data indicate that bronchial epithelial cells are
compared with the glycolysis expression in the BECs from un- subject to metabolic adaptations in asthma. Studies have shown a
treated airways. Thus, thermoplasty induced a metabolic shift in metabolic shift from OXPHOS to glycolysis, particularly for
bronchial epithelium toward that observed in healthy controls. macrophages26 and dendritic cells,27 which has been linked to
The RNA sequencing data sets comparing bronchial epithelium inflammation. Another study showed that enhanced glycolysis
from untreated and BT-treated regions can be accessed by using in lung epithelium is required for IL-1a/b-induced proinflamma-
the link https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc5 tory responses.28 There could be several reasons for this attenu-
GSE161453. ated OXPHOS gene expression. For example, inflammatory
J ALLERGY CLIN IMMUNOL RAVI ET AL 1243
VOLUME 148, NUMBER 5

FIG 2. Functional analysis of lipid profiles and metabolites in BECs cultured ex vivo. Volcano plots (signif-
icance [–log10(P value)] vs effect size [log2(fold change)]) on the y-axis and x-axis, respectively) of lipid pro-
files in BECs from healthy controls (n 5 7) compared with in BECs from patients with severe asthma (n 5 9)
cultured ex vivo. A, The levels of lipids (dots labeled in green, red, or yellow) are significantly (P < .05)
reduced in control subjects with a log fold change greater than 1 or with a fold change less than 1, or
reduced with a fold change of greater than 1 but nonsignificant, respectively. B, Heat maps of lipid profiles
in BECs cultured ex vivo from healthy controls and patients with severe asthma (P < .05 with a log fold
change greater than 1). The asterisks in the heat maps indicate patients with severe asthma with low levels
of lipid species. C, Heat maps of metabolome in BECs cultured ex vivo from healthy controls (n 5 6) and
patients with severe asthma (n 5 8). (P < .05 and log fold change >1), with red indicating upregulation
and blue indicating downregulation.

mediators such as TNF-a, can inhibit COX1 by tyrosine phos- (ATP synthase) of the OXPHOS pathway.31 Therefore, a possible
phorylation, thereby switching from aerobic metabolism to scenario is that a defective TiAR and/or TIA-1 in BECs from pa-
glycolysis.29 Interestingly, here we have shown that the attenuated tients with asthma leads to the observed metabolic shift and the re-
OXPHOS gene expression is observed in BECs that are hyperre- sulting lipid mediators may contribute to activation of the
sponsive, which results in an increased production of proinflam- bronchial epithelium,32 which can lead to chronic inflammation
matory mediators, prominently, the neutrophilic chemoattractant in asthma.33 Alternatively, but not mutually exclusively, enhanced
CXCL-8, owing to a defective translocation of the translational reactive oxygen species and mitochondrial damage in bronchial
repressor T-cell internal antigen-1–related protein (TiAR).8 Inter- epithelium were found to parallel high levels of CXCL-8, IL-6,
estingly, TiAR and the closely related T-cell internal antigen-1 and IL-1b production.34 Further support for a link between mito-
(TIA-1) have also been implicated in mitochondrial biogenesis.30 chondrial dysfunction and neutrophilic inflammation comes from
In addition, TiAR and TIA-1 knockdown attenuates complex V clustering of genes in sputum from patients with asthma, in which
1244 RAVI ET AL J ALLERGY CLIN IMMUNOL
NOVEMBER 2021

FIG 2. (Continued).
J ALLERGY CLIN IMMUNOL RAVI ET AL 1245
VOLUME 148, NUMBER 5

FIG 3. Lipidomics and metabolome analysis of expanded BECs from patients with COPD versus BECs from
controls and patients with severe asthma. A, Heat maps of lipid profiles in ex vivo–cultured BECs from pa-
tients with COPD (n 5 6) and healthy controls (n 5 7) (P < .05 with log fold change >1). B, Heat maps of lipid
profiles in ex vivo–cultured BECs from patients with COPD (n 5 6) and patients with severe asthma (n 5 9)
(P < .05 with log fold change > 1). C, Heat maps of metabolome in ex vivo–cultured BECs from patients with
COPD (n 5 6) and healthy controls (n 5 7) (P < .05 with log fold change > 1), with red indicating upregulation
and blue indicating downregulation.
1246 RAVI ET AL J ALLERGY CLIN IMMUNOL
NOVEMBER 2021

FIG 4. BT treatment alters metabolic gene expression in BECs from patients with severe asthma. A and B,
Upregulation of OXPHOS genes shown in heat map (A) and line graph plotting z scores (B) in a pairwise
comparison of BECs obtained 6 months after thermoplasty from the treated (left lower lobe) and untreated
region (middle lobe). C and D, Also, similar pairwise analysis in heat maps (C) and plotted z scores (D) show
reduced expression of glycolysis-related genes in BECs from treated versus untreated airways in patients
with severe asthma (n 5 23). E, Hyperresponsive (see Ravi et al4) BECs (n 5 4) from patients with mild
asthma displayed enhanced expression of OXPHOS genes (the same genes mentioned in Fig 1, A), as
compared with that in normoresponsive BECs (n 5 4).

case high neutrophilic inflammation was strongly associated with The differences in metabolites between BECs from patients
a significant reduction in OXPHOS genes in transcriptome- with severe asthma and those from healthy individuals appear to
associated cluster 2 (TAC2),35 and ether lipids from neutrophils36 affect purine metabolism, amino acid biosynthesis, and glycol-
downregulate mitochondrial activity.37 Finally, BECs from pa- ysis. In BECs from patients with COPD, the pathways that may be
tients with COPD, which is associated with neutrophilic inflam- affected are creatine metabolism and the tricarboxylic acid cycle.
mation,38 also display a very distinct increase in lipid species, This translation of metabolites to metabolic pathways, however,
even when compared with that from patients with asthma. is based merely on the presence of 3 or more significantly
Together, these findings provide strong evidence for a prominent different metabolites and should thus be considered with caution.
reduction in mitochondrial activity in BECs that is linked to The consequences of these different metabolic pathways on the
neutrophilic inflammation in asthma, and possibly in COPD. functioning of BECs remain to be determined, but they likely
J ALLERGY CLIN IMMUNOL RAVI ET AL 1247
VOLUME 148, NUMBER 5

affect biosynthesis, energy housekeeping, and ultimately inflam- that cultured bronchial epithelial cells maintain their phenotype
matory responses and possibly lung function. In our analysis, 2 ex vivo4 and depict asthma severity.48 This indicates that bron-
patients with severe asthma displayed markedly lower levels of chial epithelium can retain intrinsic features, even after culturing
lipid species and had lower FEV1 % reversibility than other the ex vivo.4 Third, because of the high variability of lipid mediators
patients did. Despite the low numbers, it would be interesting to measured in lung epithelial lining fluids,49 it is possible that more
explore whether higher lipid metabolism in bronchial epithelium patients should have been included for functional validation by
results in higher FEV1 % reversibility in patients with severe metabolome analysis. The transcriptome data, however, were
asthma. substantiated by lipidomics. Fourth, OXPHOS is known to be
Altered mitochondrial metabolism in airway epithelial cells attenuated by aging,50 and both those patients with severe asthma
may affect various epithelial functions, such as that of the and those with COPD were older than those in the healthy cohort.
mucociliary escalator. Epithelial cells in healthy lungs require However, the reduced expression of OXPHOS genes was also
energy for proper hydration of the mucus layer and mucociliary observed in younger patients included in the cohort of patients
clearance by ciliary beat.39 Consequently, mitochondrial damage with mild asthma, and thus, it is unlikely that aging underlies
in bronchial epithelium of patients with asthma leads to ciliary the observed differences between patients and healthy controls.
dysfunction, resulting in poor mucus clearance,40 and in COPD In addition, BT apparently resets the metabolic changes in
this has been linked to exacerbations.41 In the current study, the BECs, suggesting that the reduced expression of OXPHOS genes
BECs from some patients with asthma from the steroid-naive is acquired rather than due to aging. With respect to the latter, we
cohort with mild asthma displayed reduced OXPHOS gene cannot exclude the possibility that regional differences in the air-
expression. Therefore, at least for this cohort, the reduced expres- ways underlie the observed different bronchial epithelial tran-
sion of OXPHOS genes is independent of corticosteroid use. scriptomes after BT. Finally, for our study we used submerged
BT is a treatment option for patients with severe asthma who cultures, which typically contain undifferentiated BECs as
are uncontrolled despite optimal medical therapy, including a opposed to, for example, BECs grown at air-liquid interface cul-
high dose of inhaled bronchodilators and inhaled or oral tures, which are considered more representative of airway epithe-
corticosteroids. The results from biopsy and imaging studies are lial cells. Submerged cultures were chosen because the cells
conflicting as to whether the untreated parts of the airways are could be analyzed after a shorter culture time span (2 weeks),
also altered by BT.12,42 However, our results show a significantly as opposed to the 6 weeks required for air-liquid interface cul-
increased expression of OXPHOS genes in BECs obtained from tures, and were therefore less likely to lose their phenotype. As
treated parts compared with in BECs from untreated parts (middle our control BECs were also cultured submerged, our findings
lobe). Interestingly, imatinib, a protein tyrosine kinase inhibitor, are genuine, but the consequences of this for differentiated cells
is shown to improve airway hyperresponsiveness in severe refrac- need to be established.
tory asthma,43 and it also inhibits the platelet-derived growth fac- In summary, we have shown metabolic differences between
tor that induces airway smooth muscle cell proliferation.44 In BECs from patients with asthma and patients with COPD and
addition, imatinib induces an increase in OXPHOS gene expres- BECs from healthy individuals, with the differences worsening
sion in bronchial epithelium, specifically, in the responders to the with the severity of asthma. As these metabolic defects link to the
treatment.45 This suggests that reversing metabolic defects in earlier reported epithelial hyperresponsiveness,4 these metabolic
bronchial epithelium from patients with asthma could be benefi- effects appear to underlie airway inflammation. In this context, it
cial for patients with asthma. is of interest that BT partially normalizes the metabolic differ-
For the first time, a persistent and altered metabolism in ences, thereby resetting the bronchial epithelium, which may
bronchial epithelium has been demonstrated in vivo in patients contribute to the response to BT treatment.
with mild asthma and, more prominently, in patients with severe
asthma. The strengths of our study comprise the inclusion of pa- Key messages
tients with asthma who differ in severity of their asthma, healthy
controls, and patients with COPD. For severe asthma, we have d Bronchial epithelium from patients with severe asthma is
also shown that BT affects this altered metabolism in BECs. There metabolically altered.
was an earlier report comparing BECs obtained from sequential d Levels of lipid species in bronchial epithelium from pa-
BT sessions,46 but as the samples were collected on different tients with asthma or COPD bronchial epithelium are
days, those analyses may have been biased. In the BT trial we enhanced.
collected and analyzed BECs from treated versus nontreated air-
d BT treatment reverses these metabolic alterations.
ways in parallel. However, there are some potential limitations to
the current study. First, the patients with severe asthma were tak-
ing both oral corticosteroids and ICSs, which may have influ-
enced the alteration of metabolic genes in their bronchial
epithelium. In some patients with mild steroid-naive asthma,
however, there was a reduced expression of OXPHOS and
REFERENCES
enhanced fatty acid metabolism genes. Second, functional valida- 1. Hoogsteden HC, Verhoeven GT, Lambrecht BN, Prins JB. Airway inflammation in
tion by lipidomics and metabolome analysis was done after asthma and chronic obstructive pulmonary disease with special emphasis on the
culturing and expanding the bronchial epithelium. This ex vivo antigen-presenting dendritic cell: influence of treatment with fluticasone propio-
expansion of BECs may affect gene expression47 and we cannot nate. Clin Exp Allergy 1999;29:116-24.
2. Holgate ST. The airway epithelium is central to the pathogenesis of asthma. Aller-
exclude the possibility that mitochondrial dysfunction in bron- gol Int 2008;57:1-10.
chial epithelium of patients with severe asthma is reversible by 3. Hiemstra PS, McCray PB, Bals R. The innate immune function of airway epithelial
culturing BECs ex vivo. However, it also has been demonstrated cells in inflammatory lung disease. Eur Respir J 2015;45:1150-62.
1248 RAVI ET AL J ALLERGY CLIN IMMUNOL
NOVEMBER 2021

4. Ravi A, Chowdhury S, Dijkhuis A, Bonta PI, Sterk PJ, Lutter R. Neutrophilic metabolic modules that regulate macrophage polarization. Immunity 2015;42:
inflammation in asthma and defective epithelial translational control. Eur Respir 419-30.
J 2019;54:1900547. 27. Everts B, Amiel E, van der Windt GJW, Freitas TC, Chott R, Yarasheski KE, et al.
5. Moffatt MF, Gut IG, Demenais F, Strachan DP, Bouzigon E, Heath S, et al. A Commitment to glycolysis sustains survival of NO-producing inflammatory den-
Large-Scale, Consortium-Based Genomewide Association Study of Asthma. N dritic cells. Blood 2012;120:1422-31.
Engl J Med 2010;363:1211-21. 28. Qian X, Aboushousha R, van de Wetering C, Chia SB, Amiel E, Schneider RW,
6. Moheimani F, Hsu ACY, Reid AT, Williams T, Kicic A, Stick SM, et al. The ge- et al. IL-1/inhibitory kappa B kinase epsilon-induced glycolysis augment epithelial
netic and epigenetic landscapes of the epithelium in asthma. Resp Res 2016;17: effector function and promote allergic airways disease. J Allergy Clin Immunol
119. 2018;142:435-50.
7. Bonser LR, Zlock L, Finkbeiner W, Erle DJ. Epithelial tethering of MUC5AC-rich 29. Samavati L, Lee I, Mathes I, Lottspeich F, Huttemann M. Tumor necrosis factor
mucus impairs mucociliary transport in asthma. J Clin Invest 2016;126:2367-71. alpha inhibits oxidative phosphorylation through tyrosine phosphorylation at sub-
8. Sweerus K, Lachowicz-Scroggins M, Gordon E, LaFemina M, Huang XZ, Parikh unit I of cytochrome c oxidase. J Biol Chem 2008;283:21134-44.
M, et al. Claudin-18 deficiency is associated with airway epithelial barrier dysfunc- 30. Carrascoso I, Alcalde J, Sanchez-Jimenez C, Gonzalez-Sanchez P, Izquierdo JM.
tion and asthma. J Allergy Clin Immunol 2017;139:72-81. T-cell intracellular antigens and hu antigen r antagonistically modulate mitochon-
9. Kuo CS, Pavlidis S, Loza M, Baribaud F, Rowe A, Pandis I, et al. A transcriptome- drial activity and dynamics by regulating optic atrophy 1 gene expression. Mol Cell
driven analysis of epithelial brushings and bronchial biopsies to define asthma phe- Biol 2017;37:e00174-17.
notypes in U-BIOPRED. Am J Respir Crit Care Med 2017;195:443-55. 31. Isquierdo JM. Control of the ATP synthase beta subunit expression by RNA-
10. Schofield JPR, Burg D, Nicholas B, Strazzeri F, Brandsma J, Staykova D, et al. binding proteins TIA-1, TIAR, and HuR. Biochem Bioph Res Co 2006;348:
Stratification of asthma phenotypes by airway proteomic signatures. J Allergy 703-11.
Clin Immunol 2019;144:70-82. 32. Holtzman MJ. Arachidonic acid metabolism in airway epithelial cells. Annu Rev
11. d’Hooghe JNS, Ten Hacken NHT, Weersink EJM, Sterk PJ, Annema JT, Bonta PI. Physiol 1992;54:303-29.
Emerging understanding of the mechanism of action of bronchial thermoplasty in 33. Wenzel SE. Arachidonic acid metabolites: mediators of inflammation in asthma.
asthma. Pharmacol Ther 2018;181:101-7. Pharmacotherapy 1997;17:S3-12.
12. Goorsenberg AWM, d’Hooghe JNS, de Bruin DM, van den Berk IAH, Annema JT, 34. Hoffmann RF, Zarrintan S, Brandenburg SM, Kol A, de Bruin HG, Jafari S, et al.
Bonta PI. Bronchial thermoplasty-induced acute airway effects assessed with opti- Prolonged cigarette smoke exposure alters mitochondrial structure and function in
cal coherence tomography in severe asthma. Respiration 2018;96:564-70. airway epithelial cells. Resp Res 2013;14.
13. van der Sluijs KF, van de Pol MA, Kulik W, Dijkhuis A, Smids BS, van Eijk HW, 35. Kuo CHS, Pavlidis S, Loza M, Baribaud F, Rowe A, Pandis I, et al. T-helper cell
et al. Systemic tryptophan and kynurenine catabolite levels relate to severity of type 2 (Th2) and non-Th2 molecular phenotypes of asthma using sputum transcrip-
rhinovirus-induced asthma exacerbation: a prospective study with a parallel- tomics in U-BIOPRED. Eur Respir J 2017;49:1602135.
group design. Thorax 2013;68:1122-30. 36. Lodhi IJ, Wei XC, Yin L, Feng C, Adak S, Abou-Ezzi G, et al. Peroxisomal lipid
14. Sabogal Pineros YS, Bal SM, van de Pol MA, Dierdorp BS, Dekker T, Dijkhuis A, synthesis regulates inflammation by sustaining neutrophil membrane phospholipid
et al. Anti-IL-5 in mild asthma alters rhinovirus-induced macrophage, B-cell, and composition and viability. Cell Metab 2015;21:51-64.
neutrophil responses (MATERIAL). A placebo-controlled, double-blind study. Am 37. Kuerschner L, Richter D, Hannibal-Bach HK, Gaebler A, Shevchenko A, Ejsing
J Respir Crit Care Med 2019;199:508-17. CS, et al. Exogenous ether lipids predominantly target mitochondria. PLoS One
15. D’Hooghe JNS, ten Hacken NHT, JJTH Roelofs, Annema JT, Bonta PI. Airway 2012;7:e31342.
smooth muscle mass reduction after bronchial thermoplasty; the TASMA random- 38. Pilette C, Colinet B, Kiss R, Andre S, Kaltner H, Gabius HJ, et al. Increased
ized controlled trial. Eur Respir J 2017;50:PA3027. galectin-3 expression and intra-epithelial neutrophils in small airways in severe
16. Ravi A, Koster J, Dijkhuis A, Bal SM, Sabogal Pineros YS, Bonta PI, et al. Inter- COPD. Eur Respir J 2007;29:914-22.
feron-induced epithelial response to rhinovirus 16 in asthma relates to inflamma- 39. Button B, Okada SF, Frederick CB, Thelin WR, Boucher RC. Mechanosensitive
tion and FEV1. J Allergy Clin Immunol 2019;143:442-7.e10. ATP release maintains proper mucus hydration of airways. Sci Signal 2013;6:ra46.
17. Bousquet J, Mantzouranis E, Cruz AA, Ait-Khaled N, Baena-Cagnani CE, 40. Thomas B, Rutman A, Hirst RA, Haldar P, Wardlaw AJ, Bankart J, et al. Ciliary
Bleecker ER, et al. Uniform definition of asthma severity, control, and exacerba- dysfunction and ultrastructural abnormalities are features of severe asthma.
tions: document presented for the World Health Organization Consultation on Se- J Allergy Clin Immunol 2010;126:722-9.
vere Asthma. J Allergy Clin Immunol 2010;126:926-38. 41. Lapperre TS, Koh MS, Low SY, Ong TH, Loo CM, Lim WT, et al. Ciliary dysfunc-
18. Bel EH, Sousa A, Fleming L, Bush A, Chung KF, Versnel J, et al. Diagnosis and tion and respiratory epithelial ultrastructural abnormalities are features of copd ir-
definition of severe refractory asthma: an international consensus statement from respective of exacerbator phenotype. Respirology 2017;22:31.
the Innovative Medicine Initiative (IMI). Thorax 2011;66:910-7. 42. Pretolani M, Dombret MC, Thabut G, Knap D, Hamidi F, Debray MP, et al. Reduc-
19. Herzog K, Pras-Raves ML, Ferdinandusse S, Vervaart MAT, Luyf ACM, van tion of airway smooth muscle mass by bronchial thermoplasty in patients with se-
Kampen AHC, et al. Functional characterisation of peroxisomal beta-oxidation dis- vere asthma. Am J Resp Crit Care 2014;190:1452-4.
orders in fibroblasts using lipidomics. J Inherit Metab Dis 2018;41:479-87. 43. Cahill KN, Katz HR, Cui J, Lai J, Kazani S, Crosby-Thompson A, et al. KIT inhi-
20. d’Hooghe JNS, Goorsenberg AWM, ten Hacken NHT, Weersink EJM, Roelofs bition by imatinib in patients with severe refractory asthma. N Engl J Med 2017;
JJTH, Mauad T, et al. Airway smooth muscle reduction after bronchial thermo- 376:1911-20.
plasty in severe asthma correlates with FEV1. Clin Exp Allergy 2019;49:541-4. 44. Ingram JL, Bonner JC. EGF and PDGF receptor tyrosine kinases as therapeutic tar-
21. Goorsenberg AWM, d’Hooghe JNS, Slats AM, van den Aardweg JG, Annema JT, gets for chronic lung diseases. Curr Mol Med 2006;6:409-21.
Bonta PI. Resistance of the respiratory system measured with forced oscillation 45. Foer D, Baek S, Cahill KN, Israel E, Cui J, Weiss ST, et al. Individual gene expres-
technique (FOT) correlates with bronchial thermoplasty response. Respir Res sion signatures in severe asthmatics identify responders and non-responders to im-
2020;21:52. atinib. Am J Resp Crit Care 2019;199.
22. Bonta PI, Chanez P, Annema JT, Shah PL, Niven R. Bronchial thermoplasty in se- 46. Liao SY, Linderholm AL, Yoneda KY, Kenyon NJ, Harper RW. Airway transcrip-
vere asthma: best practice recommendations from an expert panel. Respiration tomic profiling after bronchial thermoplasty. ERJ Open Res 2019;5:00123.
2018;95:289-300. 47. Schamberger A, Verhamme F, Lindner M, Behr J, Eickelberg O. Transcriptome
23. d’Hooghe JNS, Eberl S, Annema JT, Bonta PI. Propofol and remifentanil sedation analysis of the human airway epithelium during in vitro differentiation. Eur Respir
for bronchial thermoplasty: a prospective cohort trial. Respiration 2017;93:58-64. J 2016;48:e18378.
24. Wiegman CH, Michaeloudes C, Haji G, Narang P, Clarke CJ, Russell KE, et al. 48. Gras D, Bourdin A, Vachier I, de Senneville L, Bonnans C, Chanez P. An ex vivo
Oxidative stress-induced mitochondrial dysfunction drives inflammation and model of severe asthma using reconstituted human bronchial epithelium. J Allergy
airway smooth muscle remodeling in patients with chronic obstructive pulmonary Clin Immunol 2012;129:1259-66.e1.
disease. J Allergy Clin Immunol 2015;136:769-80. 49. Brandsma J, Goss VM, Yang X, Bakke PS, Caruso M, Chanez P, et al. Lipid phe-
25. Cruickshank-Quinn CI, Jacobson S, Hughes G, Powell RL, Petrache I, Kechris K, notyping of lung epithelial lining fluid in healthy human volunteers. Metabolomics
et al. Metabolomics and transcriptomics pathway approach reveals outcome- 2018;14:123.
specific perturbations in COPD. Sci Rep 2018;8:17132. 50. Greco M, Villani G, Mazzucchelli F, Bresolin N, Papa S, Attardi G. Marked aging-
26. Jha AK, Huang SCC, Sergushichev A, Lampropoulou V, Ivanova Y, Loginicheva related decline in efficiency of oxidative phosphorylation in human skin fibroblasts.
E, et al. Network integration of parallel metabolic and transcriptional data reveals FASEB J 2003;17:1706-8.