The online version of this article (doi:10.1186/s12890-015-0088-x) contains supplementary material, which is available to authorized users.
The authors declare that they have no competing interests.
DNG and SQY conceived of the study. SC participated in its design, DIC, DPH, LQZ carried out detailed model evaluation and sample selection from the publicly available data. DNG and SQY did statistical analyses of all data and drafted the manuscript. All authors read and approved the final manuscript.
Accumulated to-date gene microarray data on Acute Respiratory Distress Syndrome (ARDS) in the Gene Expression Omnibus (GEO) represent a rich source for identifying new unsuspected targets and mechanisms of ARDS. The recently developed expression-based genome-wide association study (eGWAS) for analysis of GEO data was successfully used for analysis of gene expression of comparatively noncomplex adipose tissue, 75 % of which is represented by adipocytes. Although lung tissue is more heterogenic and does not possess a prevalent cell type for driving gene expression patterns, we hypothesized that eGWAS of ARDS samples will generate biologically meaningful results.
The eGWAS was conducted according to (Proc Natl Acad Sci U S A 109:7049-7054, 2012) and genes were ranked according to p values of chi-square test.
The search of GEO retrieved 487 ARDS related entries. These entries were filtered for multiple qualitative and quantitative conditions and 219 samples were selected: mouse nsham/ARDS = 67/92, rat n = 13/13, human cells n = 11/11, canine n = 6/6 with the following ARDS model distributions: mechanical ventilation (MV)/cyclic stretch n = 11; endotoxin (LPS) treatment n = 8; MV + LPS n = 3; distant organ injury induced ARDS n = 3; chemically induced ARDS n = 2; Staphylococcus aureus induced ARDS n = 2; and one experiment each for radiation and shock induced ARDS. The eGWAS of this dataset identified 42 significant (Bonferroni threshold P < 1.55 × 10−6) genes. 66.6 % of these genes, were associated previously with lung injury and include the well known ARDS genes such as IL1R2 (P = 4.42 × 10−19), IL1β (P = 3.38 × 10−17), PAI1 (P = 9.59 × 10−14), IL6 (P = 3.57 × 10−12), SOCS3 (P = 1.05 × 10−10), and THBS1 (P = 2.01 × 10−9). The remaining genes were new ARDS candidates. Expression of the most prominently upregulated genes, CLEC4E (P = 4.46 × 10−14) and CD300LF (P = 2.31 × 10−16), was confirmed by real time PCR. The former was also validated by in silico pathway analysis and the latter by Western blot analysis.
Our first in the field application of eGWAS in ARDS and utilization of more than 120 publicly available microarray samples of ARDS not only justified applicability of eGWAS to complex lung tissue, but also discovered 14 new candidate genes which associated with ARDS. Detailed studies of these new candidates might lead to identification of unsuspected evolutionarily conserved mechanisms triggered by ARDS.
Additional file 1: Table S1. Detailed representation of data obtained from GEO. ARDS gene expression submissions were retrieved from GEO using two terms “Acute lung injury” and “Lung injury”, which resulted in 23 and 25 data sets, respectively. These 48 entries were filtered down to 31 entries according to conditions described in Methods. The reason for filtering out an experiment is provided. (XLSX 16 kb)12890_2015_88_MOESM1_ESM.xlsx
Additional file 2: Table S2. Chi-square values of cross-referenced genes. Given that different microarray platforms have multiple probes for a single gene, the cross-referencing of such platforms generates numerous combinations of expression values for a given gene (32160 in this study). Chi-square value was calculated for each of these combinations and combination with the best chi-square value for a given gene was retained, which resulted in 3152 unique gene entries. These genes were linked to human genome and plotted against their location (Fig. 1). (XLS 185 kb)
Additional file 3: Table S3. Contribution of each ARDS model to the overall gene expression signal of an ARDS gene candidate. Each dataset obtained from GEO was reanalyzed using SAM 2.0. The d-score and fold change values for 42 gene candidates were extracted from SAM 2.0 outputs and reported according to the ARDS model. (XLSX 45 kb)12890_2015_88_MOESM3_ESM.xls
Crapo JD, Barry BE, Gehr P, Bachofen M, Weibel ER. Cell number and cell characteristics of the normal human lung. Am Rev Respir Dis. 1982;126(2):332–7. PubMed
Toscano MG, Ganea D, Gamero AM. Cecal ligation puncture procedure. J Vis Exp. 2011;51
Can I, Tahara-Hanaoka S, Shibuya A. Expression of a splicing isoform of MAIR-V (CD300LF), an inhibitory immunoglobulin-like receptor on myeloid cells. Hybridoma (Larchmt). 2008;27(1):59–61. CrossRef
- Identification of new biomarkers for Acute Respiratory Distress Syndrome by expression-based genome-wide association study
Dmitry N. Grigoryev
Dilyara I. Cheranova
Daniel P. Heruth
Li Qin Zhang
Shui Q. Ye
- BioMed Central
Neu im Fachgebiet Innere Medizin
Meistgelesene Bücher aus der Inneren Medizin
Mail Icon II