Introduction
Asthma is a heterogeneous disease affecting more than 300 million people worldwide and causing substantial socioeconomic burden [
1]. The disease is characterized by bronchoconstriction and hyper-responsiveness of the airways, but many of the molecular mechanisms causing specific asthma subtypes remains unknown. Currently, most asthmatics are described based on atopy or prevalence of cell type, such as eosinophils or neutrophils. In addition to the clinical phenotypes, there is an overwhelming need for a better understanding of the molecular characteristics and distinct pathophysiological mechanisms making up individual subgroups of asthma, thus defining asthma endotypes [
2]. Due to the heterogeneity of the disease, it is naïve to believe that one treatment fits to all asthma subgroups, therefore, personalized profiling and treatment is needed.
Recently, non-coding RNAs have been shown to be prevalent regulatory molecules involved in a wide array of diseases and cellular processes [
3‐
8]. microRNAs (miRNAs) are small, 18–24 nucleotide, RNAs responsible for the regulation of gene expression mostly through the binding and down-regulation of target mRNAs. Briefly, miRNAs are transcribed in the nucleus before export to the cytoplasm, processing, and loading of the final, mature miRNA into the Argonaute protein. This process has been thoroughly described in detail in several recent reviews [
6,
7,
9]. Although miRNAs have been extensively studied in cancer [
4,
5], their role in asthma remains relatively unknown. Mouse models of airway hyper-responsiveness and inflammation have elucidated potential mechanisms for miRNAs in the airways, but relatively few studies have been conducted in human subjects. Most recently, large scale screens using RNA-sequencing platforms or microarrays have identified potential miRNAs that may be associated with childhood asthma or allergic diseases such as allergic rhinitis [
10‐
14]. Furthermore, miRNAs are ubiquitous throughout the body, both within the cell and in a variety of bodily fluids such as serum, plasma, breast milk and urine [
15‐
19]. Due to the stable nature of miRNAs in fluids that can be acquired non-invasively, they are ideal candidates for biomarkers and potential therapeutic targets.
We set out to examine if a set of miRNAs could be monitored in serum and used as circulating non-invasive biomarkers to distinguish various asthma subgroups. Six miRNAs were chosen for this study based on their previous identification in murine models of asthma as well as their observed involvement in human immune responses [
20‐
25]. Importantly, these miRNAs have mainly been examined in models of allergic asthma [
4,
21,
24‐
33], thus, we wanted to examine if expression would be altered in individuals with non-allergic asthma. We found that the examined miRNAs showed distinct expression patterns in allergic and non-allergic asthma subgroups. Furthermore, the expression of a subset of miRNAs was significantly different depending on inhaled corticosteroid (ICS) usage or blood eosinophil level. Circulating eosinophil levels correlated to levels of eosinophil cationic protein (ECP) in the blood. Several of the circulating miRNAs correlated to a variety of clinical characteristics, including lung function parameters. Finally, in silico pathway analysis revealed that miRNAs increased under ICS usage were observed to target genes involved in leukocyte regulation and glucocorticoid response. Taken together, we believe that these miRNAs form a distinct circulating profile aiding in distinguishing different asthma subgroups. These profiles may be useful in the development of future biomarkers and be used as additional criteria in the development of asthma endotypes.
Discussion
In this study we have shown that different asthma subgroups exhibited altered expression in circulating miRNAs. We found that miR-155 and miR-146a expression in AAs was significantly increased when the examined subjects were using ICS. Furthermore, we found that miR-223 and miR-374a showed significant expression changes in NAAs when blood eosinophil numbers were taken into account. Additionally, several miRNAs correlated to lung function parameters, circulating immune cells, atopy and ICS usage. Finally, we found that miRNAs with increased expression upon ICS usage, miR-155 and miR-146a, shared target genes involved in leukocyte regulation and response to glucocorticoids. Our results suggest that combinations of circulating miRNAs can aide in defining patient groups in asthma.
The finding that miR-155 is increased in serum samples from people with allergic asthma compared to non-allergic asthmatics and healthy individuals is interesting. We recently showed miR-155 to be a critical regulator of type 2 innate lymphoid cells in murine models of allergic airway inflammation [
25] and to be differentially expressed in the airways of allergic asthmatic individuals compared to healthy controls [
24]. A recent study in CD4
+ T-cells demonstrated that miR-155 expression was altered in response to allergic stimuli or glucocorticoid treatment in a set of dust-mite allergic rhinitis, healthy, and asthmatic subjects [
32]. Unlike our data, Daniel et al. observed that miR-155 expression was increased in allergic subjects and was significantly down-regulated upon addition of dexamethasone. Some of the disparity may occur from the sample type used as we used serum from whole blood and the previous study examined isolated CD4
+ T-cells. Furthermore, our subjects were examined at baseline whereas Daniel et al. stimulated cells with the dust mite extract allergen [
32]. miR-155 is likely one of the most studied miRNAs to date, is implicated in a variety of diseases and prevalently studied in asthma [
24‐
26]. As we do not know the cellular origin of our examined circulating miRNAs, we cannot currently identify which cells are responsible for the increased expression observed in ICS treated individuals.
Glucocorticoids (GCs) are often the first and most common line of treatment for most asthmatics and work by dampening or inhibiting type 2 immune responses [
40]. Although studies have noted the usage of ICS in their subjects, few have looked at ICS as a parameter for miRNA expression [
10‐
12,
14,
41]. In our cohort, we observed changes to miRNA expression in AAs with or without ICS usage, but NAA subjects regardless of ICS status exhibited miRNA expression levels similar to healthy subjects. This finding again highlights the differences in AA and NAA subgroups, suggesting a different underlying mechanism for the disease. As in our study, another recent study of childhood asthma also found miR-146a to be increased in children on ICS, thus suggesting that this miRNA may be useful even as an early biomarker for asthma [
42]. Although ICS works very well for most atopic subjects with type 2 asthma, there are other subgroups of asthma that are non-atopic and have varying levels of type 2 response (high or low) and there are subjects that are steroid resistant [
2]. For these individuals, GCs often do not alleviate the symptoms of the disease, which may result in costly trips to the hospital in case of exacerbation. In most cases, the molecular mechanisms controlling these forms of asthma are unknown. Therefore, it is of special interest to identify and define these asthma subgroups from not only a clinical perspective, but to elucidate the molecular pathogenesis of the disease, thus, defining asthma endotypes.
In contrast to the AA group, the miRNA expression in the NAA group did not change in response to ICS usage. However, miR-223 expression was increased in individuals with high circulating eosinophil levels and significantly decreased in individuals with low circulating eosinophil levels as compared to healthy individuals. When examining clinical parameters in our study, we found the strongest significant correlation of miR-223 with eosinophils as well as ECP, suggesting that this miRNA may have a role in eosinophil development in asthmatic subjects. miR-223 is of particular interest in a type 2 inflammation driven-disease such as asthma where increased eosinophil numbers are often used as a clinical biomarker. Interestingly, Lu et al. observed miR-223, one of the most upregulated miRNAs found in eosinophilic esophagitis, was strongly correlated to eosinophil levels in the esophagus [
43]. The group later reported that deficiency of miR-223 led to an increase in eosinophil progenitors in bone marrow of mice [
44]. Although the role of miR-223 in circulating eosinophils is not known, perhaps the increased miR-223 expression we observe in NAAs with high eosinophil counts could lead to lower levels of eosinophil progenitors in the blood. Conversely, the decrease in miR-223 expression even below that of healthy individuals in NAA subjects with low levels of eosinophils (< 0.1 × 10
9 cells /L), may suggest an imbalance in eosinophilopoesis, potentially leading to the manifestation of their asthma subtype. However, in order to determine the relationship and importance of miR-223 and eosinophils, miR-223 must be confirmed to be expressed in human eosinophils as we have examined miRNA expression only in cell-free serum. Furthermore, we observed an interesting opposing relationship between miR-146a and miR-223. In our study, these miRNAs exhibited inverse levels of correlations to clinical parameters (Fig.
4d-f; e.g., ECP, eosinophils, FVC) and exhibited a significant negative correlation to one another (
Supp Fig. 2), perhaps suggesting different cellular origins of these miRNAs. Additionally, Panganiban et al. found that both miR-146a and miR-223 expression were increased in plasma from asthma subjects as opposed to healthy or allergic rhinitis individuals [
14]. Even in murine models of asthma treated with ovalbumin or house dust mite, miR-146a and miR-223 were differentially regulated in plasma [
27]. We also observed that miR-146a and miR-223 showed altered expression patterns between asthma and healthy individuals, but also between our asthma subgroups, with miR-146a exhibiting expression changes in AAs, whereas miR-223 levels differed in NAAs. With these previous studies and our own, both miR-146a and miR-223 appear to be robust miRNAs in blood, able to be identified in both plasma and serum and act as distinguishing markers for asthma subgroups.
miR-374a is a less studied miRNA which has been identified in several screens, but its function still remains elusive [
12,
14,
41]. We found that miR-374a exhibits different expression patterns in AA and NAA asthmatics (Fig.
2c&g and Fig.
3d). We observed that in NAAs, miR-374a appeared to be influenced by circulating eosinophil level rather than ICS status (Fig.
3d), whereas in the AAs, miR-374a clearly showed a strong decrease in expression in ICS naïve individuals (Fig.
2c&g). Further studies to determine the cellular origin of miR-374a and potential gene targets may help to elucidate the functional role this miRNA plays in asthma. In both our study and Kho et al., it was observed that miR-374a correlated to lung function, specifically FEV1/FVC [
12]. Interestingly, in the childhood serum cohort examined by Kho et al., the correlation of FEV1/FVC to miR-374a expression was positive, whereas we observed a negative correlation (Fig.
4d). We did, however, find a positive correlation with miR-374a expression and age. Perhaps this discrepancy in lung function correlation lies in the age of the cohort as we examined an adult cohort (average age 48 years) and Kho and colleagues worked with a childhood asthma cohort [
12]. In order to determine if age and miR-374a expression are associated with lung function, future longitudinal studies would be required.
Through the use of non-invasive biomarkers, we can start to tease apart potential epigenetic markers of interest, predict and study their targeted genes and pathways and propose and develop potentially life altering treatments for numerous individuals. We performed a pathway analysis on common targets of miR-155 and miR-146a, which were found in this study to be increased specifically in AA+ICS individuals (Fig.
5), to gain insight into their potential function. We observed that of the ninety-eight common genes, twenty-six formed a network and the majority of those genes were enriched in biological processes such as leukocyte regulation and response to GCs. As the purpose of ICS are to dampen the type 2 response, it may be that ICS cause an increase in post-transcriptional regulators, such as miR-155 and miR-146a. These regulators may then be involved in the negative regulation of cells involved in type 2 response, thus, alleviating asthma symptoms. We also found an enrichment in genes associated with IL-6 signaling (
Supp Table 1 ). IL-6 signaling has been previously suggested as a biomarker for asthma [
45‐
47] and circulating levels of IL-6 has been shown to be increased in individuals with asthma [
47] and has been associated with high dose ICS in adult onset asthma [
45]. Due to the complexity of IL-6 signaling, it was outside the scope of the current study, but would be a potentially interesting avenue for future research.
Although we have identified miRNAs that may be useful as potential future biomarkers, one limitation of our study is that we have a relatively small number of subjects. Through multivariate logistical regression analysis, we determined the potential biomarker quality of the six candidate miRNAs (
Supp Fig. 3). Although this model appears to be in agreement with our previous statistical analysis, the small subject size (
n = 92) and number of variables (
n = 11) is likely an overestimation of the predictive power of the given miRNAs. Our data suggests that levels of circulating blood eosinophils may influence the relative expression of certain miRNAs, such as miR-223 and miR-374a. Unfortunately, as we have looked at only serum, we cannot be sure of the origin of these miRNAs. In the future, it would be interesting to determine if indeed the aforementioned miRNAs are more abundant in eosinophils compared to others miRNAs that we have examined. Furthermore, in this study we have chosen only six miRNAs to examine based on their previously reported roles in human asthma and inflammation or murine models of asthma [
4,
20‐
33]. In order to gain a better understanding of all the alterations in miRNAs among asthma subgroups, it would be beneficial to use a screening technique such as microarray, Nanostring analysis or RNA-sequencing. Indeed, Rodrigo-Muñoz et al. performed RNA-sequencing in eosinophils and identified several miRNAs other than those which we have examined that were specific in distinguishing asthmatic individuals [
48] .
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.