Animals
Male 6- to 8-wk-old BALB/c mice were obtained from Experimental Animal Centre (Beijing, China). Performance of all experimental procedures was done with approval from the Animal Care and Use Committee of Capital Medical University.
AR mouse model
BALB/c mice were randomly assigned to three groups of six mice: control, OVA, and 4w-after group. Mice in control and OVA groups were sensitized and challenged with saline or ovalbumin (OVA), respectively, and euthanized 2 h after the last challenge. Mice in 4w-after group were sensitized and challenged with OVA and euthanized 4 weeks after the last challenge.
Animals were sensitized by intraperitoneally (i.p.) injecting saline or 50 μg OVA (grade V; Sigma, St Louis, Mo, USA) emulsified in 5 mg Al(OH)3 on days 1 and 7. From day 14, the animals were challenged by nasal instillation of saline or 50 μg OVA in 20 μL of 0.9% saline three successive days a week for three consecutive weeks.
At 2 h or 4 weeks after the last challenge, the blood was harvested under anesthesia with pentobarbital sodium (80 mg/kg, i.p.) and nasal mucosa was harvested after euthanizing by cervical dislocation. Nasal mucosa samples were used for RNA isolation and histological analysis, while serum was stored at − 70 °C and used for OVA-specific IgE detection.
Measurement of nasal symptoms
The numbers of nasal rubbings and sneezes were counted for 20 min before the animals were euthanized, by three observers blinded to the experimental treatments given to the mice.
Measurement of serum OVA-specific IgE
Serum OVA-specific IgE levels were determined by enzyme-linked immunosorbent assay (ELISA) as previously described [10]. Plates were coated with OVA and horseradish peroxidase-conjugated anti-mouse IgE (1:4,000) (Southern Biotech) was used. A standard curve was prepared from serial dilutions of an arbitrary standard, and the levels of OVA-specific IgE were expressed as arbitrary units (AU).
Histological analysis
Nasal samples were embedded in paraffin and cut into 5 µm thick sections. The sample sections were stained with hematoxylin and eosin (H&E) to assess eosinophils. The sample sections were also assessed by means of immunohistochemistry staining for NIMP-R14 (Abcam, 2557, neutrophil marker), BCL3 (Santa Cruz, sc-185), NFKB2 ((Proteintech, 10409-2-AP), SOCS3 (Abcam, 280884), CD14 (HuaBio, ET1610-85) and TLR4 (HuaBio, ER1706-43). Horseradish peroxidase-conjugated secondary antibodies (Zhongshanjinqiao, Beijing, China) and substrate 3, 3′-diaminobenzidine were used, which rendered positive staining cells brown. The numbers of eosinophils, neutrophils, and positive staining cells for BCL3, NFKB2, SOCS3, CD14 and TLR4 were counted at 400X magnification by two observers who were blinded to the treatment.
Microarray
Total RNA was isolated from the nasal mucosa samples, three mice each group, using the RNeasy mini kit (Qiagen, Valencia, CA, USA). Total cRNA was generated using a WT Expression Kit (Ambion; Austin, TX, USA) and labeled using a GeneChipWT Terminal Labeling Kit (Affymetrix; Santa Clara, CA, USA). The labeled cRNA was hybridized to GeneChip Mouse Gene 1.0 ST arrays (Affymetrix, Santa Clara, CA, USA) at 45 °C for 16 h and at the end of hybridization, the arrays were washed using the Fluidics station 450, prior to being scanned with a GeneChip Scanner 3000 7G (Affymetrix). The images of all arrays were transformed into digital data using the Command Console Software 4.0 (Affymetrix), and the data for each array were normalized by RMA + DABG normalization using the expression console.
Bioinformatics analysis
Differentially expressed genes (DEG)
The random-variance model (RVM) F-test was applied to filter the DEG across the three groups because this test can raise the degrees of freedom effectively in small sample size cases [11]. After analysis for statistical significance and false discovery rate (FDR) analysis, the differentially expressed genes were selected (both P value and FDR less than 0.05).
Pathway enrichment analysis
Pathway analysis was performed to determine the significant pathway of the differential genes according to KEGG [12,13,14], Biocarta, and Reatome. Fisher's exact test and χ2 tests were used to select significant pathways with a threshold of P < 0.05.
Series tests of cluster (STC)
We also used STC method to analyze the expression dynamics of DEG by STEM software. To define a set of model profiles independent of the data, the amount of change a gene could exhibit among the three groups was set as ‘one unit’; thus, a changed gene could go up either one unit, stay the same, or go down one unit. Since this method relies on correlation, ‘one unit’ may be defined differently for different genes. For the three groups, this strategy results in 8 distinct profiles, and we, therefore, used a set of 8 unique model profiles to represent any expression changes that might occur. The raw expression values of DEG were converted into log2 ratios, where the ratios were determined based on the expression of the first group. The value of the first group after transformation was thus always 0. Passing the data normalization, each differential gene expression was assigned to the model profile that most closely matched the gene's expression profile as determined by correlation coefficient analysis [15, 16]. Significant profiles of STC analysis were determined by Fisher's exact test, which have significantly more genes assigned under the true ordering of time points compared to the average number assigned to the model profile in the permutation runs. And a P value of less than 0.05 is considered significant.
STC-gene ontology (GO) analysis
GO analysis was applied to the genes in each profile to identify the main function of the genes having the same expression trend according to the Gene Ontology, the key functional classification of NCBI [17]. Fisher’s exact test and X2 test were used to classify the GO category, with the P-values of less than 0.05 considered statistically significant. Enrichment is related to the specificity of the function. The higher the enrichment, the more specific the corresponding function.
Gene co-expression network
The co-expression network was generated using the Matlab 7.1-java software. The gene adjacency matrix M between two genes, i and j, which reflected the correlations between genes, was computed and presented as the network map.
In this map, each node describes a gene, and the relation between them is represented with a line segment. The most elementary characteristic of a node is its degree (the number of genes interacting with it). The greater degree of a gene indicated a more central role of the gene within the network. The network’s hub is the most important central gene, affecting the structure of the whole network and other associated genes. Clustering coefficient calculates the density in this neighborhood, with a larger clustering coefficient indicating that the actor is coupled with a stronger degree of "clique-like" local neighborhoods.
Real-time PCR validation
The 21 DEG with higher degree identified by gene co-expression analysis were validated for their expression by real-time PCR. The RNA samples were same as those used in the microarray (see Additional file 1 for Primers). The gene expression levels were quantified relative to the expression of GAPDH or β-actin by Comparative Ct (delta delta Ct) method.
Statistical analysis
Statistical analyses were performed using the SPSS 11.0 (SPSS, Inc, Chicago, IL, USA). Normally distributed data were expressed as mean ± SD, and the significance of any difference between the treatment groups was assessed using an unpaired Student’s t-test. Values of P < 0.05 were considered to be statistically significant.
