Clinical and molecular phenotyping of asthma and COPD
Boudewijn, Ilse Maria
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Boudewijn, I. M. (2019). Clinical and molecular phenotyping of asthma and COPD. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
9
A novel role of bronchial microRNAs
and long non-coding RNAs
in asthma remission
Ilse M. Boudewijn, Mirjam Roffel, Cornelis J. Vermeulen, Judith M. Vonk, Dirkje S. Postma, Cheng-Jian Xu, Nick H. T. ten Hacken, Wim Timens, Irene H. Heijink, Martijn C. Nawijn, Klaas Kok, M. Martijn Terpstra, Alen Faiz, Antoon J. van Oosterhout, Karen Affleck, Gerard H. Koppelman,
Victor Guryev*, Maarten van den Berge*
*shared last author
Submitted
Chapter 9
ABSTRACT
BackgroundAsthma remission in adulthood occurs in a minority of patients, but knowledge on its fundamental mechanisms is lacking.
Objective
We explored underlying mechanisms of asthma remission by investigating bronchial microRNA expression followed by integration with protein-coding- and long non-coding RNA (lncRNA) expression.
Methods
Bronchial microRNA-seq expression was compared between subjects with complete remission (ComR; n=14; no asthma treatment, symptoms, hyperresponsiveness and airway obstruction), persistent asthma (PersA; n=79) and healthy controls (n=82). We correlated expression of differentially expressed microRNAs between ComR and PersA, with both predicted gene targets and genome wide gene expression. We integrated microRNA-, protein-coding- and lncRNA expression in association with ComR using Bayesian network modeling. In vitro-validation by overexpression of a top-microRNA was performed.
Results
Ten microRNAs were differentially expressed between ComR and PersA and 77 between ComR and Healthy. MicroRNA target genes positively correlating with the 10 differentially expressed microRNAs were enriched in the pathway ‘focal adhesion’. Bayesian network modeling revealed 24 microRNAs, 35 lncRNAs and 20 protein-coding RNAs that associated with ComR. In vitro-validation showed an attenuated inflammatory response in poly-(I:C) stimulated mir-320d-transfected human tracheobronchial cells. Conclusion
Complete asthma remission is associated with a distinct bronchial microRNA expression profile compared to persistent asthma and healthy. We show that many microRNAs and lncRNAs are associated with complete remission in a Bayesian network. Of interest,
mir-320d, one of the microRNAs identified, has potential anti-inflammatory effects. These
findings suggest that future research should focus on the causal role of microRNAs an lncRNAs in asthma remission.
9
INTRODUCTION
Asthma is a prevalent disease, affecting 5-10% of people worldwide, and is associated with significant healthcare costs(1,2). At present, no cure for asthma exists, yet asthma can convert into spontaneous remission, an understudied asthma phenotype. Clinical remission, characterized by the absence of both respiratory symptoms and asthma medication use, has been reported to occur in 30-52% of asthma patients(3). In 2004, Vonk et al proposed the term ‘complete remission’, a definition that additionally includes the absence of airway hyperresponsiveness and airway obstruction. By this stringent definition, the prevalence of complete remission was 22% and 10% in two studies after 30 and 39 years of follow-up respectively(4,5).
We need a better insight into the mechanisms that drive asthma remission, since this may open new avenues for designing novel asthma therapies. The first genome wide association study on asthma remission has recently been published(6), reporting three DNA variants, of which 2 were located in intergenic regions, to be associated with asthma remission. It is becoming increasingly clear that not only the protein-coding transcriptome is important for disease development, progression and therapy response of chronic diseases. Accumulating evidence indicates that microRNAs and long coding RNAs (lncRNAs) might also be highly relevant(7,8). MicroRNAs are short, non-coding RNAs involved in post transcriptional gene repression and are increasingly recognized as important regulators of lung development and growth, as well as disease pathogenesis, including Th2-driven airway inflammation(9). LncRNAs are a relatively underinvestigated transcript type, despite their abundance in the human genome being comparable to that of protein-coding genes. Some of them have extensive regulatory effects on gene expression by blocking or enhancing gene transcription, influencing microRNA functioning and direct binding to proteins(8,10).
We aimed to identify microRNAs associated with asthma remission compared to persistent asthma. We next aimed to understand how microRNAs are interacting with protein-coding RNAs and lncRNAs in relation to asthma remission. We explored also how microRNA expression in asthma remission differs from a healthy situation.
METHODS
Study population
All subjects with persistent asthma, clinical remission or complete remission had a doctor’s diagnosis of asthma and documented airway hyperresponsiveness (AHR) in the past(11). Remission was defined as absence of asthma symptoms in the last 3 years and no asthma medication. Complete remission was further characterized by the absence
of AHR and pre-bronchodilator FEV1>80% predicted, clinical remission by the presence
of either AHR or pre-bronchodilator FEV1≤80% predicted. Healthy subjects had no
respiratory symptoms and normal pulmonary function(12). Sample preparation
Total RNA was isolated from frozen bronchial biopsies, library preparation was done for microRNAs and protein-coding/lncRNAs separately and RNA-sequencing was performed.
Statistical analyses
Differential expression of microRNAs was analyzed with DESeq2 (v1.14.1) (13), adjusting for age, gender, smoking status and library preparation batch. Since inhaled corticosteroids (ICS) influence gene expression(14), we defined two distinct groups: ‘persistent asthma using ICS’ and ‘persistent asthma without ICS’ (i.e. no ICS at all or ICS stopped 6-8 weeks before bronchoscopy). A false discovery rate (FDR) adjusted p-value<0.05 was considered statistically significant. The contrasts of interest were: 1) Complete remission versus persistent asthma without ICS.
2) Complete remission versus healthy.
To assess the possible effect of cell type composition of the biopsies on the analyses, we reran analysis 1 with adjustment for inflammatory cell counts in the biopsies (eosinophils, mast cells, T-cells, macrophages and neutrophils) and checked whether results changed.
To check robustness of analysis 1 and 2, we increased the number of subjects in remission by combining complete and clinical remission into 1 group with the following contrasts: 3) All remission (complete and clinical combined) versus persistent asthma without ICS. 4) All remission (complete and clinical combined) versus healthy.
We performed Spearman correlation analyses of microRNAs differentially expressed between complete remission and persistent asthma without ICS and RNA expression of their predicted targets (identified using Targetscan v7.1), as well as genome wide gene expression. We performed Gene Ontology (GO) enrichment analyses on gene
9
targets that were positively or negatively correlated, using g:Profiler version r1741_ e90_eg37(15). In order to explore the effects of cell type composition of the tissue samples on our results, we assessed co-expression of microRNAs and their correlated RNA transcripts by performing hierarchical clustering of the top-15 strongest correlated RNA transcripts per microRNA, as described previously(16).
We integrated microRNA-, protein-coding RNA- and lncRNA expression in association with complete remission by performing Bayesian network modeling using the CGBayesnets Package in MATLAB(17) with calculation of the Area Under the Receiver Operator Characteristic Curve (AUC) and its statistical significance. To check robustness of the gene type composition in our network, we performed permutation testing (100 iterations in which we swapped the phenotype values among samples) and checked the proportion of protein-coding and non-protein-coding RNAs.
In vitro -validation
In vitro -validation of a differentially expressed microRNA was done by
microRNA-transfection of tracheobronchial and 16HBE cells, followed by stimulation with house dust mite (HDM) or the viral mimic polyionosinic:polycytidylic acid (poly-(I:C)) and measurement of the pro-inflammatory cytokine granulocyte-macrophage colony-stimulating factor (GM-CSF).
RESULTS
Differential microRNA expression analyses
Of 1860 microRNAs detected in the bronchial biopsies, 1342 were filtered out based on low abundance, leading to 518 microRNAs being available for analyses. After exclusion of seven microRNA samples during quality control (due to low coverage and a technical error in one of the batches), data on 206 subjects were available for analyses. Their clinical characteristics are shown in Table 1.
Comparison of complete remission versus persistent asthma without ICS Ten microRNAs were differentially expressed between subjects with complete remission and persistent asthma without ICS: 9 upregulated (miR-320a, miR-193a-5p,
miR-320c, miR-4532, miR-320d, miR-320b, miR-423-3p, miR-133b and miR-3960) and 1
downregulated (miR-126-3p) (Table 2; Figure 1). In a sensitivity analysis including only persistent asthma patients with ‘no ICS at all’, all 10 microRNAs remained significantly differentially expressed. After adjustment for inflammatory cell counts in the biopsies, >90% of results remained FDR-significant (Table E1). When analyzing ‘all remission’
Table 1. C
har
ac
teristics of the study p
opula
tion
Persist
en
t asthma without ICS
Persist en t asthma using ICS Persist en t asthma no ICS a t all Persist en t asthma st opp ed ICS Clinic al remission Complet e remission H ealth y n=33 n=28 n=18 n=31 n=14 n=82 Ag e, y ears * 49 (39-54) 50 (35-54) 55 (39-60) 47 (41-59) 48 (36-53) 42 (23-56) Female gender , n (%) 15 (45) 12 (43) 12 (67) 16 (52) 7 (50) 36 (44) Smok ing sta tus , n (%) †, §, ll § ll ‡ †,‡ Nev er smoker 13 (39) 12 (43) 13 (72) 17 (55) 9 (64) 42 (51) C ur ren t smoker 3 (9) 12 (43) 4 (22) 5 (16) 4 (29) 40 (49) Ex -smoker 17 (52) 4 (14) 1 (6) 9 (29) 1 (7) 0 (0) Pack years *,$ 12 (3-24) 13 (8-26) 9 (6-21) 8 (3-29) 26 (3-27) 16 (4-29) FE V1 , % pr edic ted 81 (±19) ‡,§ 83 (±13) †,ll 84 (±19) **,†† 92 (±14) ‡‡ 103 (±13) §,ll ,†† 101 (±12) †,‡,**,‡‡ FE V1 /FV C, % 68 (±11) ‡,§,ll 71 (±9) † 71 (±12) 75 (±8) ll 79 (±8) § 79 (±6) †,‡ Pr esenc e of a top y, n (%) 26 (81) ‡,1 20 (77) †,2 12 (71) 3 18 (58) 9 (64) 30 (37) †,‡ PC 20 AMP , mg/ml # 59 ‡,§ 31 †,ll ,§§ 29 **,†† 155 ‡‡,§§ 640 §,ll ,†† 543 †,‡,**,‡‡ (0.002-640) (0.02-640) (0.01-640) (0.02-640) (640-640) (33-640) PC 20 H istamine , mg/ml #,## 14.0 § (n=15) 19.4 ll (n=7) 12.1 †† (n=5) 15.9 §§ (n=19) 64 §,ll ,††,§§ (n=14) (0.3-64) (4.7-64) (2.3-64) (1.7-32) (64-64) PC 20 M ethacholine , mg/ml # -31.8 (16-32) ICS dose , µg/da y *,¶¶ 800 -800 ¶ -(475-1000) (500-1000) Time sinc e first asthma a ttack *, y ears 32 (19-44) 45 (35-50) 33 (23-46) 40 (34-43) 45 (40-43) -Values ar e sho wn as mean ( ± standar d devia
tion) unless sta
ted other
wise; *median with in
ter quar tile r ange; †, ‡, §, ll ,**, ††, ‡‡,§§: equal symbols r epr esen t sig nifican t diff er enc es bet w een gr oups (p -v alue <0.05; adjust ed f or multiple t
esting using Holms-B
onf
er
roni); $ only f
or ex
-smokers and cur
ren t smokers; 1 missing v alue f or 1 subjec t; 2 missing v alue f or 2 subjec ts; 3 missing v alue f or 1 subjec t; # geometr ic mean with r
ange (if a 20% fall w
as not r eached , a PC 20 v alue is sta ted as 640, 64 and 32 f or A MP
, histamine and methacholine r
espec tiv ely); ## only pr esen t f or subjec ts with a PC 20 A MP > 320mg/ml; ¶¶ beclomethasone equiv alen t; ¶ ICS w er e st opped 6-8 w eeks pr ior to br onchosc op y; FE V1 =f or ced expir at or y v olume 1 sec ond; FVC=f or
ced vital capacit
y; PC 20 = pr ov oca tiv e c onc en tr
ation inducing a 20% fall in FE
V1
; A
MP = adenosine 5’-monophospha
te; ICS = inhaled c
or tic ost er oids .
9
(complete and clinical combined) versus persistent asthma without ICS, 14 microRNAs were differentially expressed (10 up- and 4 downregulated, supplementary table E2), of which 6 were also identified in the above analysis of complete remission versus persistent asthma without ICS: miR-320a, miR-193a-5p, miR-320c, miR-4532, miR-320b and miR-3960.
Table 2. Ten differentially expressed microRNAs between complete remission and persistent asthma without ICS.
Complete remission vs
persistent asthma without ICS Complete remission vs healthy Mean of
normalized
counts Log
2 FC Adjusted
p value Log2 FC Adjustedp value
miR-320a 3972 0.92 <0.01 0.55 0.04 miR-193a-5p 151 0.86 <0.01 0.31 0.29 miR-320c 30 0.94 0.02 0.32 0.40 miR-4532 69 1.55 0.02 0.16 0.83 miR-320d 21 1.02 0.02 0.33 0.44 miR-320b 113 0.85 0.03 0.38 0.28 miR-133b 1 2.09 0.03 2.05 <0.01 miR-126-3p 3288 -0.54 0.03 -0.55 <0.01 miR-423-3p 1218 0.51 0.03 0.52 <0.01 miR-3960 103 1.15 0.05 -0.07 0.91
miR=microRNA; FC = fold change; microRNAs in bold print are also differentially expressed between complete remission and healthy.
Comparison of complete remission versus healthy
Seventy-seven microRNAs were differentially expressed between subjects with complete remission and healthy controls (62 up- and 15 downregulated) (Figure 1; supplementary Table E3). When analyzing ‘all remission’ (complete and clinical combined) versus healthy, 74 microRNAs were differentially expressed (54 up- and 20 downregulated; supplementary Table E4), of which 57 were also identified in the complete remission versus healthy analysis.
MicroRNAs differentially expressed in both contrasts
Four microRNAs (miR-320a, miR-133b, miR-126-3p and miR 423-3p) were significantly differentially expressed both when comparing complete remission and persistent asthma without ICS, and when comparing complete remission and healthy, with all except mir-126-3p being upregulated in complete remission (Figure 1; Table 2).
Figure 1. Heatmap of 83 differentially expressed microRNAs between complete remission and persistent asthma without ICS and complete remission and healthy. ICS=inhaled corticosteroids; miRNA=microRNA.
Correlation of differentially expressed microRNAs with RNA expression We correlated expression levels of the 10 microRNAs that were differentially expressed between complete remission and persistent asthma without ICS, with expression levels of their predicted gene targets within the same subject (significant correlations shown in Tables E5-E12). Enrichment analysis using Gene ontology (GO) categories revealed 28 pathways that were significantly enriched for positively correlated mRNA transcripts (FDR 0.01, Table 3), of which ‘focal adhesion’ was the most significant. Enrichment analyses of all significantly negatively correlated mRNA transcripts revealed enrichment of genes involved in ‘endocytosis’ and ‘lysine degradation’ (Table 3). Next, we correlated expression levels of the 10 microRNAs with genome wide gene expression available from the same bronchial biopsy. Of 10 microRNAs, 8 had genome-wide significant correlations, the top-100 strongest correlated genes per microRNA are shown in the online supplement (Tables E13-E22). Co-expression of microRNAs and their correlated RNA transcripts was assessed by performing hierarchical clustering of the top-15 strongest correlated genes per microRNA, a.o. to explore effects of cell type composition on our results (Figure 2). The heatmap showed 4 gene clusters associated with 1 or more microRNAs: the first and fourth cluster contained genes correlated with
mir-126-3p expression only, and included genes such as MPRIP and TSPAN1. The second cluster
9
and mir-3960 and included mainly mitochondrial genes, such as MT-ATP8. The third cluster contained genes positively correlated with mir-4532, mir-193a-5p and mir-3960 and included genes such as DCN, PCDH9 and TGFBR3. No specific cell types could be attributed to one of the clusters based on their gene content.
Table 3. Enrichment analysis on mRNA targets that were significantly correlated with expression of the 10 microRNAs differentially expressed between complete remission and persistent asthma without ICS
FDR adjusted p-value POSITIVELY CORRELATED TARGETS
KEGG
Focal adhesion 8.37E-07
Biological process
negative regulation of biological process 2.41E-06 regulation of cell differentiation 3.11E-06 multicellular organism development 8.30E-06 negative regulation of cellular process 1.94E-05
system development 2.17E-05
cell-cell signaling 5.75E-05
regulation of cell communication 6.68E-05 regulation of signal transduction 1.01E-04 regulation of developmental process 1.42E-04 regulation of signaling 1.51E-04 anatomical structure development 3.28E-04 developmental process 8.59E-04 biological regulation 1.43E-03 regulation of cell-substrate adhesion 1.46E-03 nervous system development 1.72E-03
cell adhesion 2.57E-03
biological adhesion 3.04E-03
positive regulation of biological process 3.42E-03 multicellular organismal process 3.71E-03 animal organ development 3.80E-03 regulation of multicellular organismal process 4.57E-03 positive regulation of cellular process 5.92E-03 regulation of multicellular organismal development 6.88E-03 regulation of response to stimulus 8.35E-03 regulation of cell development 8.40E-03 regulation of cellular component organization 8.84E-03
Reactome
Axon guidance 7.57E-04
NEGATIVELY CORRELATED TARGETS KEGG
Endocytosis 1.55E-03
PHLDB1 EFS SHROOM4 MPRIP ZNF618 KLHL29 ARAP3 DUOXA1 DUOX1 KAZN CHD7 MT−ND5 MT−ND4L MT−ND2 MT−CYB MT−ND1 MT−ATP8 MTND2P28 MT−ND4 MTATP6P1 MT−ND6 MT−ATP6 YLPM1 FTH1P23 PRDM2 MTND5P11 MT−CO2 MT−CO1 MT−CO3 VPS53 ARID1A MTND1P23 FAT3 PCDH9 KCNMB4 PGR NELL1 SLC8A3 HOXA5 AGTR1 PCDHB12 TENM3 ASPA GPX3 NCAM2 ADH1B PCDHB7 TRHDE TGFBR3 RGS13 CPS1 QPRT C7 ZFYVE20 SFRP4 HCG11 HRASLS5 C1orf21 GSTM3 CYBRD1 DCN OMD SCN7A CHRDL1 PI16 SERPINB7 RPS27 TSPAN1 CD24P4 JKAMP ENPP4 PFN2 RNF130 TMEM66 CYB5R1
miR−126−3p miR−320c miR−320d miR−320a miR−320b miR−4532 miR−193a−5p miR−3960
−0.6 −0.4 −0.2 0 0.2 0.4 0.6 Figur e 2. H ea
tmap and hier
ar
chic
al clust
ering of r
ho v
alues of genes most signific
an tly c orr ela ted with e xpr ession le
vels of the 8 micr
oRNA
s.
miR = micr
9
Bayesian network modeling
For subjects in complete remission and persistent asthma without ICS, we integrated microRNA-, protein-coding RNA- and lncRNA expression. To this end, expression levels of 518 microRNAs and 22729 protein-coding RNAs, lncRNAs and pseudogenes were used as input in a Bayesian network analysis. The gene network associated with the binary phenotype complete remission (using persistent asthma without ICS as controls) consisted of 24 microRNAs, 20 protein-coding RNAs, 35 lncRNAs and 14 pseudogenes (Figure 2, AUC 0.99, p=0.0027). Of interest, 6 of the 24 microRNAs in the network were also identified in our differential expression analysis: miR-126a-3p, miR-320a, miR-320b,
miR-320c, miR-193a-5p and miR-133b. Only microRNAs and lncRNAs, but not
protein-coding RNAs, were directly connected to complete remission. Permutation analysis showed that our network consistently contained a lower proportion of protein-coding RNAs compared to the permuted networks (p<0.01, Figure E1).
Figure 3. Bayesian network of microRNAs, protein-coding RNAs, lncRNAs and pseudogenes predicting complete remission. AUC=0.99; p=0.0027. Distinct gene types are shown in different colors. Black-lined nodes represent nodes directly connected to complete remission. Diamond-shaped nodes represent microRNAs that were also identified in the differential expression analysis.
In vitro-validation of mir-320d
Mir-320d, one of the top-10 microRNAs upregulated in complete remission compared to
persistent asthma, was previously found to have anti-inflammatory effects in response to cigarette smoke(18). We further explored its role in response to the viral mimic poly-(I:C) and HDM in human tracheobronchial cells and 16HBE cells. In tracheobronchial cells and 16HBE cells stimulated with poly-(I:C), mir-320d transfection significantly decreased GM-CSF production compared to mimic controls (p<0.05, Figure 4C+D). HDM stimulation did not affect GM-CSF production in tracheobronchial cells; neither in
mir-320d transfected cells nor in mimic controls (Figure 4D). In 16HBE cells stimulated
with HDM, no significant difference was observed in GM-CSF production between
mir-320d transfected cells and control cells (Figure 4C).
Figure 4. In vitro-validation of miR-320d in epithelial cells. 16HBE or tracheobronchial epithelial cells were
transfected with a non-targeting microRNA mimic (mimic control) or a miR-320d mimic and stimulated with HDM or poly-(I:C) for 24h. A) miR-320d expression in 16HBE (n=11) and B) tracheobronchial epithelial cells (n=8). C) GM-CSF concentration in 16HBE (n=11) and D) tracheobronchial epithelial cells (n=8) upon stimulation with HDM or poly-(I:C). Values represent median ± interquartile range. A Wilcoxon test was used to test for statistical significance; *=p<0.05, **=p<0.01; ***=p<0.001; HDM=house dust mite; poly-(I:C)=polyionosinic-polycytidylic acid; GM-CSF= granulocyte-macrophage colony-stimulating factor.
9
DISCUSSION
We performed microRNA expression analyses to provide insight into the fundamental mechanisms of asthma remission. Expression levels of 10 microRNAs were significantly different between persistent asthma and complete remission, i.e. objectively diagnosed asthma in the past without current respiratory symptoms, asthma treatment, airway obstruction and hyperresponsiveness. A novel finding is that both microRNAs and lncRNAs are abundantly present in the Bayesian network that, based on RNA profiles, was associated with complete remission of asthma. This suggests that non-coding RNAs, a group consisting of small RNAs such as microRNAs as well as lncRNAs, may be important in complete remission of asthma.
Approximately 2% of the genome codes for transcripts that are translated into proteins. Those transcripts which are not translated into proteins are called non-coding RNAs and are as abundant as protein-coding genes in the genome. Besides their role in translation (rRNA and tRNA), non-coding RNAs play a role in the regulation of gene expression through several mechanisms, including posttranslational histone modification, DNA methylation or blocking translation by hybridization to a protein-coding RNA(19). Of interest, some of these non-coding RNAs are thought to have sponge-like effects on microRNAs and mRNAs leading to inhibition of their function(10). We found a clear signal of 10 microRNAs that differentiate complete remission from persistent asthma. Moreover, upon integration of microRNA-, protein-coding RNA- and lncRNA expression in association with complete remission using Bayesian network modeling, microRNAs and lncRNAs together represented 55 transcripts in the network, while only 20 protein-coding RNAs were present. Permutation analysis corroborated our findings, revealing that this network had a significantly lower proportion of protein-coding RNAs than what could be expected by chance (p<0.01). These findings provide suggestive evidence that especially non-coding RNAs are associated with remission of asthma. Hence, future research should focus on the functional characterization of these non-coding RNAs in asthma remission.
One other group has studied microRNA expression in remission of asthma by performing Bayesian network modeling. In this study, asthma remission was defined by the loss of AHR in 160 children with asthma at age 14 and serum microRNAs were measured at baseline (between 5 and 12 years)(20). The authors identified a remission-associated network of 12 microRNAs that included miR-126-5p, a microRNA that was also present in our Bayesian network in airway wall biopsies of adults with complete asthma remission. Although the study differed in many aspects from ours (e.g. children versus adults, detection in blood versus airway wall, longitudinal- versus cross-sectional design), the
importance of microRNAs exhibiting their function by acting in a network is supported by both studies.
The 10 microRNAs that were differentially expressed between complete remission and persistent asthma were 320a, 193a-5p, 320c, 4532, 320d,
miR-320b, miR-126-3p, miR-423-3p, miR-133b and miR-3960. Of these, miR-320a, miR-126-3p, miR-133b and miR-423-3p are of special interest, since subjects in complete remission
had not only lower 126-3 expression and higher 320a, 133b and
miR-423-3p expression compared to subjects with persistent asthma, but also compared
to healthy controls. Overall, we observed that subjects in complete remission have a distinct bronchial microRNA profile (77 differentially expressed microRNAs) compared to healthy controls. Together, these findings suggest that subjects in complete remission do not resemble healthy subjects but represent a third, separate molecular state in addition to asthma and healthy, which likely reflects the presence of biological mechanisms to actively ‘overcome’ asthma, leading to a normalized pulmonary function and absence of symptoms. Exploring this phenomenon further can aid in developing new asthma therapies.
Of the differentially expressed microRNAs, miR-126-3p has been investigated in relation to Th2-inflammation, an important mechanism underlying asthma(21). In a mouse model of house dust mite (HDM)-induced asthma, miR-126-3p was upregulated in the airway wall after exposure to HDM. Mice treated with a miR-126 antagomir had lower inflammatory cell counts in bronchoalveolar lavage fluid and less hyperresponsiveness than mice treated with a scrambled control. Moreover, Th2 cells in peribronchial lymph nodes of these miR-126 antagomir treated mice showed less IL-5 and IL-13 production upon exposure to HDM than the scrambled control-treated mice. These results suggest that downregulation of miR-126-3p plays a role in suppressing Th2 inflammation. We found miR-126-3p downregulated in complete remission compared to both persistent asthma and healthy controls, suggesting that this microRNA might be important in achieving asthma remission, although causality cannot be established in this study. Four members of the 320 family, namely 320a, 320b, 320c and
miR-320d, were upregulated in complete remission compared to persistent asthma. The miR-320 family has a potential regulatory role in inflammation. The latter is supported
by our in-vitro finding that GM-CSF production is decreased in poly-(I:C)-stimulated and mir-320d-transfected human tracheobronchial cells and 16HBE cells, compared to poly-(I:C)-stimulated mimic controls. Our previous work in bronchial biopsies of COPD-patients has shown that mir-320d is downregulated compared to healthy controls, but
9
again upregulated after ICS treatment(18). Additionally, in vitro -validation showed that overexpression of mir-320d in BEAS2B cells reduced IL-8 production after cigarette smoke exposure(18). It has been shown that inhibition of miR-320 in colonic epithelial cells results in activation of inflammation(22), measured by translocation of NF-kappa-B to the nucleus and upregulation of pro-inflammatory cytokines such as TNF-alpha and IL-8. Of interest, mir-320a and 2 other microRNAs associated with complete remission in our study (mir-193a-5p and mir-423-3p), were found to be downregulated in serum of children with cow’s milk allergy compared to healthy controls(23). We found miR-320a to be upregulated in complete remission versus both persistent asthma and healthy controls. This upregulation of miR-320 in complete remission may thus act to suppress inflammatory processes in asthma, while in healthy subjects, inflammation as a trigger leading to upregulation of miR-320 family members is lacking.
Focal adhesion, a pathway important in cell-matrix adhesion, was the most significant enriched pathway among genes positively correlated with the 10 differentially expressed microRNAs between complete remission and persistent asthma. Genes annotated to ‘Focal adhesion’ that were correlated to our microRNAs of interest were amongst others
LAMC3 (Laminin Subunit Gamma-3, correlated with miR-4532), an important component
of basement membranes, ITGA1 and ITGA6 (Integrin Subunit Alpha 1 and -6, correlated with miR-4532 and miR-126-3p respectively) which are known to contribute to airway inflammation and -remodeling(24) and MYL12A and MYL9 (Myosin Light Chain 12A and -9, correlated with miR-320a and miR-4532 respectively), which are involved in (smooth) muscle contraction.
Hierarchical clustering of the top-15 strongest correlated genes with the differentially expressed microRNAs showed 4 gene clusters. No specific cell type could be attributed to a gene cluster. Of interest, genes annotated to mitochondrial function formed a distinct cluster and were negatively correlated to the miR-320 family, miR-193a-5p and
miR-3960. Mitochondria are essential for a cell’s energy metabolism and asthma has been
associated with increased mitochondrial mass and elevated metabolic activity(25). We found that mitochondrial genes are negatively correlated with upregulated microRNAs in complete remission, which might suggest a lower metabolic need in complete remission.
To date, few studies have investigated underlying mechanisms of asthma remission, and mainly focused on inflammation and airway wall remodeling. Previously, our group showed lower blood eosinophil counts and airway wall eosinophil activation in subjects with complete remission as compared to persistent asthma(11). On the other hand,
Boulet et al observed comparable blood eosinophil counts and regulatory T cell counts between these groups(27). We additionally observed a similar degree of basement membrane thickening between complete remission and persistent asthma (11). Thus, a switch from a pro-inflammatory profile to resolution of inflammation may play a role in complete remission of asthma, resulting in different subsets of hematopoietic cells present in the airway wall, while airway wall remodeling persists to at least some degree. Therefore, we need to take into account that part of the differences observed in our transcriptome analysis, reflecting the remission-specific molecular state, might be driven by altered cell type composition of the biopsies. Hence, we corrected for cell type composition in our analysis, which did not reveal an effect of cell type composition on our results.
A strength of this study is the careful characterization of the subjects at baseline and at follow-up with standardized questionnaires and extensive pulmonary function tests, such as bronchial provocation tests. This enabled us to accurately classify the study subjects into ‘complete remission’, ‘clinical remission’, ‘persistent asthma’ and ‘healthy’, allowing us to study underlying mechanisms of these distinct physiological states. A limitation is that the samples of healthy controls were obtained between 2009 and 2012 and the other samples between 2002 and 2006. Technical variation of expression profiles is a common source of bias in RNA experiments, so we cannot fully exclude variation between samples from healthy controls and the other groups due to differences in storage time. As fortunately samples of subjects with persistent asthma, clinical remission and complete remission were obtained within the same period, technical variation due to storage time is less likely. Another limitation is the lack of a replication cohort for our findings. Unfortunately, only few studies have been published on complete remission, especially at adult age, and to our knowledge no data on bronchial microRNA/RNA expression are available in these cohorts. It will be of great interest to replicate the findings presented in this paper in an independent cohort. In conclusion, we show that subjects with persistent asthma, subjects in complete remission of asthma and healthy controls differ in their bronchial microRNA expression profile. Of interest, when integrating microRNA-, protein-coding RNA- and lncRNA expression by performing Bayesian network modeling, we identified a network characteristic of complete remission of asthma in which microRNAs and lncRNAs are abundantly present. These findings suggest that future research should further explore the functional role of microRNAs and lncRNAs in remission of asthma.
9
ACKNOWLEDGEMENTS
We thank the UMCG Genomics Coordination center, the UG Center for Information Technology and their sponsors BBMRI-NL & TarGet for storage and compute infrastructure.
SUPPORT
This work was supported through a scientific research collaborative agreement with GlaxoSmithKline.
REFERENCES
(1) Martinez FD, Vercelli D. Asthma. Lancet. 2013 Oct 19;382(9901):1360–72.
(2) Bahadori K, Doyle-Waters MM, Marra C, Lynd L, Alasaly K, Swiston J, FitzGerald JM. Economic burden of asthma: a systematic review. BMC Pulm Med. 2009 Dec 19;9(1):24.
(3) Upham JW, James AL. Pharmacology & Therapeutics Remission of asthma : The next therapeutic frontier ? Pharmacol Ther. 2011;130(1):38–45.
(4) Vonk JM, Postma DS, Boezen HM, Grol MH, Schouten JP, Koëter GH, Gerritsen J. Childhood factors associated with asthma remission after 30 year follow up. Thorax. 2004 Nov 1;59(11):925–9.
(5) Carpaij OA, Nieuwenhuis MAE, Koppelman GH, van den Berge M, Postma DS, Vonk JM. Childhood factors associated with complete and clinical asthma remission at 25 and 49 years. Eur Respir J. 2017;49(6):1601974.
(6) Vonk JM, Nieuwenhuis MAE, Dijk FN, Boudier A, Siroux V, Bouzigon E, Probst-Hensch N, Imboden M, Keidel D, Sin D, Bossé Y, Hao K, van den Berge M, Faiz A, Koppelman GH, Postma DS. Novel genes and insights in complete asthma remission. Clin Exp Allergy. 2018 May 22;
(7) Esteller M. Non-coding RNAs in human disease. Nat Rev Genet. 2011;12(12):861–74.
(8) Salviano-Silva A, Lobo-Alves S, Almeida R, Malheiros D, Petzl-Erler M. Besides Pathology: Long Non-Coding RNA in Cell and Tissue Homeostasis. Non-Non-Coding RNA. 2018;4(1):3.
(9) Heffler E, Allegra A, Pioggia G, Picardi G, Musolino C, Gangemi S. MicroRnas Profiling in Asthma: Potential Biomarkers and Therapeutic Targets. Am J Respir Cell Mol Biol. 2017;57(6):642–50.
(10) Schmitz SU, Grote P, Herrmann BG. Mechanisms of long noncoding RNA function in development and disease. Cell Mol Life Sci. 2016;73(13):2491–509.
(11) Broekema M, Timens W, Vonk JM, Volbeda F, Lodewijk ME, Hylkema MN, Ten Hacken NHT, Postma DS. Persisting remodeling and less airway wall eosinophil activation in complete remission of asthma. Am J Respir Crit Care Med. 2011;183(3):310–6.
(12) Telenga ED, Oudkerk M, van Ooijen PMA, Vliegenthart R, ten Hacken NHT, Postma DS, van den Berge M. Airway wall thickness on HRCT scans decreases with age and increases with smoking. BMC Pulm Med. 2017;17(1):1–8.
(13) Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014 Dec 5;15(12):550.
(14) van den Berge M, Steiling K, Timens W, Hiemstra PS, Sterk PJ, Heijink IH, Liu G, Alekseyev YO, Lenburg ME, Spira A, Postma DS. Airway gene expression in COPD is dynamic with inhaled corticosteroid treatment and reflects biological pathways associated with disease activity. Thorax. 2014 Jan 7;69(1):14–23. (15) Reimand J, Arak T, Adler P, Kolberg L, Reisberg S, Peterson H, Vilo J. g:Profiler-a web server for functional
interpretation of gene lists (2016 update). Nucleic Acids Res. 2016;44:W83–9.
(16) Xu CJ, Söderhäll C, Bustamante M, Baïz N, Gruzieva O, Gehring U, Mason D, Chatzi L, Basterrechea M, Llop S, Torrent M, Forastiere F, Fantini MP, Carlsen KCL, Haahtela T, Morin A, Kerkhof M, Merid SK, van Rijkom B, Jankipersadsing SA, Bonder MJ, Ballereau S, Vermeulen CJ, Aguirre-Gamboa R, de Jongste JC, Smit HA, Kumar A, Pershagen G, Guerra S, Garcia-Aymerich J, Greco D, Reinius L, McEachan RRC, Azad R, Hovland V, Mowinckel P, Alenius H, Fyhrquist N, Lemonnier N, Pellet J, Auffray C, van der Vlies P, van Diemen CC, Li Y, Wijmenga C, Netea MG, Moffatt MF, Cookson WOCM, Anto JM, Bousquet J, Laatikainen T, Laprise C, Carlsen KH, Gori D, Porta D, Iñiguez C, Bilbao JR, Kogevinas M, Wright J, Brunekreef B, Kere J, Nawijn MC, Annesi-Maesano I, Sunyer J, Melén E, Koppelman GH. DNA methylation in childhood asthma: An epigenome-wide meta-analysis. Lancet Respir Med. 2018;6(May):379–88.
(17) McGeachie MJ, Chang H-H, Weiss ST. CGBayesNets: Conditional Gaussian Bayesian Network Learning and Inference with Mixed Discrete and Continuous Data. Murphy RF, editor. PLoS Comput Biol. 2014 Jun 12;10(6):e1003676.
(18) Faiz A, Borggrewe M, Postma D, Steiling K, Spira A, Lenburg M, Jonker M, Koppelman G, Hiemstra P, Sterk P, Timens W, Brandsma C-A, Van den Berge M, Heijink I. MiR-320d: A novel anti-inflammatory miRNA up regulated by corticosteroids. Eur Respir J. 2015;46(59).
9
(19) Peschansky VJ, Wahlestedt C. Non-coding RNAs as direct and indirect modulators of epigenetic regulation. Epigenetics. 2014;9(1):3–12.
(20) McGeachie MJ, Davis JS, Kho AT, Dahlin A, Sordillo JE, Sun M, Lu Q, Weiss ST, Tantisira KG. Asthma remission: Predicting future airways responsiveness using an miRNA network. J Allergy Clin Immunol. 2017;140(2):598–600.e8.
(21) Mattes J, Collison A, Plank M, Phipps S, Foster PS. Antagonism of microRNA-126 suppresses the effector function of TH2 cells and the development of allergic airways disease. Proc Natl Acad Sci U S A. 2009;106(44):18704–9.
(22) Pierdomenico M, Cesi V, Cucchiara S, Vitali R, Prete E, Costanzo M, Aloi M, Oliva S, Stronati L. NOD2 Is Regulated by MIR-320 in Physiological Conditions but this Control Is Altered in Inflamed Tissues of Patients with Inflammatory Bowel Disease. Inflamm Bowel Dis. 2016;22(2):315–26.
(23) D’Argenio V, Del Monaco V, Paparo L, De Palma FDE, Nocerino R, D’Alessio F, Visconte F, Discepolo V, Del Vecchio L, Salvatore F, Berni Canani R. Altered miR-193a-5p expression in children with cow’s milk allergy. Allergy Eur J Allergy Clin Immunol. 2018;73(2):379–86.
(24) Wright DB, Meurs H, Dekkers BGJ. Integrins: Therapeutic targets in airway hyperresponsiveness and remodelling? Trends Pharmacol Sci. 2014;35(11):567–74.
(25) Prakash YS, Pabelick CM, Sieck GC. Mitochondrial Dysfunction in Airway Disease. Chest. 2017;152(3):618– 26.
(26) Veltri KL, Espiritu M, Singh G. Distinct genomic copy number in mitochondria of different mammalian organs. J Cell Physiol. 1990;143(1):160–4.
(27) Boulet LP, Turcotte H, Plante S, Chakir J. Airway function, inflammation and regulatory T cell function in subjects in asthma remission. Can Respir J. 2012;19(1):19–25.
SUPPLEMENTAL MATERIAL
Supplemental material is online available by scanning the following QR-code or using the URL: www.publicatie-online.nl/publicaties/i-boudewijn/chapter-9-supplemental/
