Expression of SMARCD1 interacts with age in association with asthma control on inhaled corticosteroid therapy

Asthma affects over 300 million persons globally and costs more than $50 billion annually in the U.S. [1] Despite effective treatment options, exacerbations from asthma account for substantial preventable morbidity [2]. Most individuals with asthma respond to inhaled corticosteroids (ICS), the most effective asthma controller medication, with significant symptom improvement; however, approximately one third of individuals respond minimally or not at all [3]. Furthermore, age appears to modify ICS response, with increasing treatment failures for each year over the age of 30 [4]. While genetics explains substantial variability in asthma drug disposition and effects [5, 6], the field of pharmacogenomics has not addressed the role of age in modulating response to asthma medications.

Asthma pharmacogenomic studies to date have shed light on biological mechanisms. For example, a genomic study characterizing transcriptomes identified multiple genes involved in the inflammatory pathway that influence ICS response including Cysteine Rich Secretory Protein LCCL Domain Containing 2 (CRISPLD2) [7]. CRISPLD2 mRNA has been shown to strongly vary with age [8], and protein levels were shown to increase in response to treatment with a known pro-inflammatory cytokine, interleukin 1 beta (IL1β). Moreover, a transcriptomic study of ORMDL Sphingolipid Biosynthesis Regulator 3 (ORMDL3) found that a variant in that gene might influence the route of anti-inflammatory action of glucocorticoids by modifying the transcriptional activation of ORMDL3 in subjects with asthma [9]. As ORMDL3 is associated with childhood-onset asthma, ORMDL3’s influence on the inflammatory pathway may be age-dependent; however, no published studies have examined this [9].

Global gene expression levels are known to be highly dependent upon gross demographic features including age [10, 11], yet identification of age-related genomic indicators has yet to be comprehensively undertaken in an asthma-specific context. The objective of this study was to discover genomic indicators specific to response to ICS in individuals with asthma by accounting for age-dependent genomic interactions.

There were 118 subjects with gene expression data passing quality control from CD4+ lymphocytes in Asthma BRIDGE who reported taking inhaled corticosteroids as a controller medication in the previous year. These subjects had mean Chronic Asthma Control Scores (CACS) of 10.8 (+/− 6.9), shown in Fig. 1. A general trend emerged showing increasing CACS with increasing age (increase of 0.15/year, 95% confidence interval [0.079, 0.22], p = 3.4 × 10^− 5), indicating worsening asthma control with advancing age. This trend remained significant when stratified by sex, and no significant difference was observed in males (increase of 0.17/year, 95% CI [0.065, 0.28], p = 0.0012) versus females (increase of 0.12/year, 95% CI [0.011, 0.22], p = 0.03).

In Asthma BRIDGE, the SMARCD1 gene (SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily D member 1) strongly interacted with age to indicate higher CACS on inhaled corticosteroids, with a doubling of expression associated with an increase of 1.3 units of CACS per year increase in age (95% CI [0.86, 1.74], p = 6 × 10^− 9), indicating increasing importance of SMARCD1 with advancing age (Fig. 2). In the replication population, GALA II, the age-by-SMARCD1 interaction was associated with poorly-controlled asthma compared to subjects with controlled asthma, with a ratio of odds ratios relative to control of 1.72 (95% CI [1.42, 2.10], p = 3.8 × 10^− 8), indicating increasing probability of being in the poorly-controlled group with age.

We observed that both the specific study within Asthma BRIDGE and self-identified race were strongly split by age (Table 1): GRAAD and CAG are adult studies and CARE is a study of children. Similarly, CARE is predominantly non-Hispanic white with a minority population of Hispanic white; while GRAAD and CAG are mostly African American. Since we wanted to focus on the interaction between gene expression and age in asthma control, we did not include either of these demographic variables in our main analysis. However, there was no significant association of race or study with SMARCD1 expression after adjustment for age (ANOVA, p = 0.29 and p = 0.3, respectively).

For completeness, we also considered SMARCD1-age interaction in other cell types available in ABRIDGE. These included (1) Alveolar Macrophages (n = 33, p = .75), although these samples did not span a similar age range (ages 8 to 22 years). (2) Bronchial Epithelium (n = 16, p = .2), Bronchoalveolar cells (n = 12, p = .22), and whole blood (n = 4, p not computed), although these had low sample sizes. Finally (5) CD4+ lymphocytes stimulated with hemagglutinin (n = 139, p = .01), although this stimulation may not be representative of asthma biology.

To investigate whether SMARCD1 plays a role in the inflammatory response as well as in the anti-inflammatory action of corticosteroids in vitro, we knocked down SMARCD1 in a human lung epithelial cell line (similar to the replication samples which were enriched for airway epithelium) stably expressing nuclear factor-kappa B (NF-κB) luciferase reporter [24]. We transfected this reporter cell line with two different SMARCD1 siRNAs. One of these resulted in a large (greater than 80%) reduction of SMARCD1 expression relative to scramble-control siRNA transfection (Fig. 3). In this successful knockdown of SMARCD1, we observed a significant increase in luciferase activity relative to scramble-control in response to both IL-1β stimulation (p < 0.01) and IL-1β plus dexamethasone (p < 0.01) (Fig. 4). This verified that, at least in vitro, SMARCD1 is an important part of the inflammatory response and its absence increases the anti-inflammatory action of corticosteroid treatment.

In addition to performing pathway enrichment studies using our own results (see Additional file 1), we also investigated connections of the SMARCD1 gene and its protein product (also labeled SMARCD1) in NDEX, a searchable collection of gene expression and protein-protein interaction networks from multiple network and pathway databases [23]. A search on SMARCD1 revealed the following biologically relevant network for our phenotype of interest, ICS response, in the current analysis: The glucocorticoid receptor regulatory network (a protein-protein interaction network) was derived from the latest BioPAX3 version of the Pathway Interaction Database. A portion of the larger network including our gene of interest (the one-step adjacent network surrounding SMARCD1) is shown in Fig. 5. This diagram indicates that cortisol is a controller of SMARCD1, and that expression of SMARCD1 is related to SGK1 (glucocorticoid regulated kinase 1) the glucocorticoid receptor Nuclear Receptor Subfamily 3 Group C Member 1 (NR3C1).

Our study has three key findings. First, age appears to modulate the association of SMARCD1 with chronic asthma control. Secondly, focusing on age-dependent factors is likely to yield important indicators of asthma medication response. Third, interactions between age and expression of SMARCD1 in CD4+ lymphocytes were replicated in nasal samples enriched for respiratory epithelial cells, suggesting that the potential age-dependent effects of SMARCD1 on asthma control could be mediated by systemic as well as locally induced changes in the airways. That SMARCD1 interacts with age to affect asthma control on ICS, across two different cell types, in two different racial ancestry groups, and using two different measures of asthma control, indicates a robust effect.

SMARCD1, a member of the SWI/SNF chromatin remodeling complex family, regulates gene transcription by binding specific transcriptional factors and altering local chromatin structure [25]. To our knowledge, SMARCD1 has not been reported as associated with ICS response in prior studies, although it has been found to be associated with asthma in a prior analysis that used GEO data from adult cohorts of the Unbiased Biomarkers for the Predictions of Respiratory Disease Outcomes (U-BIOPRED) research study [26]. Protein-protein interaction network data support the role of SMARCD1 in modulating asthma control while on ICS. Synthetic glucocorticoid medications (such as inhaled budesonide and dexamethasone) are potent inducers of glucocorticoid signaling, exerting their anti-inflammatory effects through glucocorticoid receptor binding [27]. A search of NDEX [23] shows SMARCD1 as a key effector and regulatory protein within the glucocorticoid receptor regulatory network. In protein-protein interaction studies, SMARCD1 complexes with NRC31 (nuclear receptor subfamily 3 group C member 1), a glucocorticoid receptor [28]. We further validated the in vitro role of SMARCD1 in the inflammatory process and the action of dexamethasone on that process.

In addition to its direct involvement in glucocorticoid signaling, SMARCD1 is also associated with apoptosis pathways [29, 30]. Interestingly, in vitro studies of ICS response have identified apoptosis as a potential protective mechanism associated with resolution of asthmatic inflammation [31‐34]. An analysis of gene regulatory networks for ICS response in mid-childhood identified enrichment of pro-apoptosis pathways in good ICS responders, and anti-apoptosis pathway enrichment in poor ICS responders. While cellular apoptosis is a natural process occurring at all stages of human development, changes in expression of apoptosis genes occur with age [35, 36]. In our study, alterations in SMARCD1 expression with age may relate to its involvement in apoptosis pathways. Future studies will be required to understand whether age-associated changes in apoptosis pathways underlie the age-related effects of SMARCD1 on ICS response in asthmatics.

Our study has several strengths. Our gene expression analysis was appropriately timed to recent inhaled corticosteroid use. In our discovery analysis, asthma control while on ICS was assessed using a questionnaire based on the ACT criteria, which accounts for multiple aspects of control, including symptoms, quality of life metrics (missed school/work), medication use and health care utilization (emergency room/doctor visits for asthma). In addition to replication across tissue types, we found replication of the SMARCD1 by age interaction across multiple racial ethnic groups, which contributes to the generalizability of our findings. Our study also had some limitations. Assessment of asthma control was somewhat different in the discovery and replication cohorts, which may have affected our ability to replicate additional age-by-gene expression interactions in models of asthma control while on ICS. Although our in vitro validation demonstrated a biological relationship between SMARCD1, NF-kB, and dexamethasone, the nature of the experiment with cell lines precludes the assessment of age; thus this result does not directly support a differential effect of SMARCD1 with increasing age. Furthermore, this effect was in the opposite direction as one might expect: SMARCD1 knockdown increased NF-kB, which generally increases inflammation, which generally increases asthma symptoms, and would lead to higher CAC. We speculate that this difference in effect direction could be due to differences between in vitro and in vivo conditions, the inability to include age in the cell lines, or differences between airway epithelial cells and CD4+ lymphocytes. However, our strongest evidence indicates that increasing SMARCD1 is associated with worsening asthma control.