Transcriptional Profiling of Swine Lung Tissue after Experimental Infection with Actinobacillus pleuropneumoniae

Porcine pleuropneumonia is a highly contagious respiratory disease that causes great economic losses worldwide. In this study, we aimed to explore the underlying relationship between infection and injury by investigation of the whole porcine genome expression profiles of swine lung tissues post-inoculated with experimentally Actinobacillus pleuropneumoniae. Expression profiling experiments of the control group and the treatment group were conducted using a commercially available Agilent Porcine Genechip including 43,603 probe sets. Microarray analysis was conducted on profiles of lung from challenged versus non-challenged swine. We found 11,929 transcripts, identified as differentially expressed at the p ≤0.01 level. There were 1188 genes annotated as swine genes in the GenBank Data Base. GO term analysis identified a total of 89 biological process categories, 82 cellular components and 182 molecular functions that were significantly affected, and at least 27 biological process categories that were related to the host immune response. Gene set enrichment analysis identified 13 pathways that were significantly associated with host response. Many proinflammatory-inflammatory cytokines were activated and involved in the regulation of the host defense response at the site of inflammation; while the cytokines involved in regulation of the host immune response were suppressed. All changes of genes and pathways of induced or repressed expression not only led to a decrease in antigenic peptides presented to T lymphocytes by APCs via the MHC and alleviated immune response injury induced by infection, but also stimulated stem cells to produce granulocytes (neutrophils, eosinophils, and basophils) and monocyte, and promote neutrophils and macrophages to phagocytose bacterial and foreign antigen at the site of inflammation. The defense function of swine infection with Actinobacillus pleuropneumoniae was improved, while its immune function was decreased.

cytokines were activated and involved in the regulation of the host defense response at the site of inflammation; while the cytokines involved in regulation of the host immune response were suppressed. All changes of genes and pathways of induced or repressed

Introduction
Porcine pleuropneumonia (PP) is a highly contagious respiratory disease that causes great economic losses worldwide [1]. The disease, which occurs in swine of all ages, is highly infectious, often fatal, and characterized by necrotizing, hemorrhagic bronchopneumonia and serofibrinous pleuritis [1]. Actinobacillus pleuropneumoniae (APP) is the causative agent of PP and can spread quickly by air-borne particles and/or touching a contaminated surface, and often kills infected animals in the acute phase when extensive lung hemorrhage and necrosis occur. Swine that survive often develop pleurisy, the sequelaes of local necrosis of the pleura, or became healthy carriers of APP.
The porcing lung infected with APP has previously been reported to result in local production of proinflammatory proteins or to mRNA encoding the cytokines interleukin (IL)-1α, IL-1α, IL-6 and the chemokine IL-8 [2]. Likewise, bioactive protein and /or mRNA code IL10, IL12p35, TNF-α and INF-α have shown to be up-regulated after infection with APP in vivo or in vitro [2][3][4]. Using cDNA microarrays, Moser and co-workers found 307 anonymous transcripts in blood leukocytes from swine that were significantly affected by experimental infection with APP [5]. Hedegaard et al. investigated the molecular characterization of the early response in pigs to experimental infection with APP serotype 5B, using cDNA microarrays [6]. In this study, two-colour microarray analysis was conducted to identify genes being significantly differently expressed in non-inflamed lung tissue compared with inflamed lung tissue sampled from the same animal [6]. The samples of lung tissue were studied by manual hybridization to the pig array DIAS_PIG_27K2 that contains 5375 PCR products amplified from unique cDNA clones [6]. Hedegaard and co-workers found three subsets of genes consistently expressed at different levels depending upon the infection status, and a total of 357 genes differed significantly in their expression levels between infected and non-infected lung tissue from infected versus non-infected animals [6]. Mortensen et al. studied the local transcriptional response in different locations of lung from pigs experimentally infected with the respiratory pathogen APP 5B, using porcine cDNA microarrays (DJF Pig 55 K v1) representing approximately 20,000 porcine genes printed in duplicate [7]. Within the lung, Mortensen and co-workers found a clear division of induced genes as, in unaffected areas a large part of differently expressed genes were involved in systemic reaction to infections, while differently expressed genes in necrotic areas were mainly concerned with homeostasis regulation [7]. However, a limited number of genes relative to the whole Porcine Genome have been studied in previous documents by using cDNA microarrays [5][6][7]. Thus, transcriptional profiling of whole porcine genome in lung tissue sampled from inoculated versus non-inoculated swine would lead to greater knowledge of the host response dynamics to bacterial infection in the lung. This knowledge is important to obtain a more complete picture of the lung-specific host reactions in the pathogenesis of respiratory infection.
In the present study, the Agilent Whole Porcine Genome Oligo (4 × 44 K) Microarrays (one-color platform), which is a commercially available Agilent Porcine Genechip that included 43,603 probe sets, were used to detect the changes in gene expression of infected pigs' lungs from non-inoculated animals. Ten transcripts (top six up-regulated and top four down-regulated in microarray data) were selected to verify the accuracy and reproducibility of the microarray data by real-time qRT-PCR.

Clinical Symptoms and Necropsy Findings
The symptoms of lung lesions in the TG were typical after swine infected with APP. Swine showed hyperthermia (40.6-42.0 °C), dyspnea and anorexia after inoculation with APP 24-48 h. Two swine died with respiratory distress at post-inoculation 36-48 h. In the autopsy, the lungs were found to be severely damaged by acute, multifocal, fibrino-necrotizing and hemorrhagic pneumonia complicated by acute diffuse fibrinous pleuritis. The tracheobronchial lymphoid nodes were enlarged and congested.
No lesions were observed in CG lung ( Figures 1A and 2). Lung showed swelling, bleeding and fibrinous exudate sticking to the lung surface in TG ( Figure 1B). The histopathologic changes were characterized by hemorrhage, lymphocyte infiltration, fibrinous exudation vascular thrombosis, necrotic focus and edema in TG ( Figure 3).

Microarray Profiling
Expression profiling was conducted using a commercially available Agilent Porcine Genechip that included 43,603 probe sets. The transcriptome of the lung was determined. Expression was detected for 30,574 probes (70.12% of all probe sets) of the CG. A total of 31,957 probes (73.29% of all probe sets) were expressed in TG. When probe set intensities were normalized and filtered, there were still 26353 probes used to significantly identify DE genes. There were 11,929 genes identified as DE at the p ≤ 0.01 level by comparing the log2 (normalized signal) of the two groups using T-test analysis.
Hierarchical clustering was applied to the mean log-ratio of the replicated spots from the DE genes by the average linkage and using euclidean distance as the similarity metric ( Figure 4). The expression profiles of samples were divided into two groups-one from the non-inoculated swine (M-P-1, M-P-2, M-P-3) and the other group from the inoculated swine (M-P-4, M-P-5, M-P-6). The principal component map to three-dimensional space, also found that the distance of CG (samples M-P-1, M-P-2, M-P-3) is short, and the gene expression pattern is more consistent. Also the distance of TG (samples M-P-4, M-P-5, M-P-6) is relatively discrete because of the differences in the degree of lesion, and the gene expression pattern is similar ( Figure 5). The six samples were set as variables, the principal component analysis (PCA) of the co-expressed differentially genes (CG vs TG) showed the contribution rate of the first principal component which reached 96.95%, the first three principal components of the total contribution rate reach 99.754% (Table 1).

Verification of Gene Expression Pattern from Microarray Data Using Real-Time QRT-PCR
Ten genes (i.e., RETN, ADAM17, GPNMB, CHRM1, ALDH2, IL6, KLRK1, DUOX2, OAS2 and KCNAB1) were selected to confirm expression patterns using real-time qRT-PCR. The results indicate that the expression patterns of all the genes were consistent with the microarray data (r = 0.905 ± 0.125, Figure 7).

Figure 7.
Validation of the microarray data by the real-time qRT-PCR analyses of ten representative genes. The x-axis represents the genes and the y-axis shows their relative expression levels (−ΔCt) values for quantitative real-time RT-PCR; Log (Sample signal, 10) for microarray. Three biological replicates were conducted for both assays. R represents the Pearson correlation coefficient. The significance of differences for gene expression between the CG and the TG was calculated using a two-tailed T-test.

Discussion
In the present study, we revealed 11,929 DE genes using Agilent Whole Porcine Genome Oligo Microarrays (one-color platform) that contain 43,603 probes. There were 1188 genes annotated as swine genes in the GenBank Data Base (DB). GO term analysis identified that a total of 89 BP categories, 82 cellular components and 182 molecular functions were significantly affected and at least 27 BP categories were related to the host immune response.
The NOD-like receptor signaling pathway, Fc epsilon RI signaling pathway, acute myeloid leukemia, melanoma, progesterone-mediated oocyte maturation, spliceosome, type II diabetes mellitus and steroid hormone biosynthesis etc. were significantly enriched in inflamed lung. Five pathways as type I diabetes mellitus, autoimmune thyroid disease, allograft rejection, tight junction and cardiac muscle contraction were significantly enriched in non-inflamed lung. The NOD-like receptor signaling pathway is one of the most important pathways associated with microbial recognition and host defense [8,9]. The innate immune system comprises several classes of pattern recognition receptors, including Toll-like receptors (TLRs), NOD-like receptors (NLRs) and RIG-1-like receptors. Two NLRs, NOD1 and NOD2, sense the cytosolic presence of the peptidoglycan fragments, meso-DAP and muramyl dipeptide, respectively, and drive the activation of MAPK and the transcription factor NF-kappaB (NF-κB). A different set of NLRs induces caspase-1 activation through the assembly of large protein complexes named inflammasomes [9]. Inflammasomes are critical for generating mature proinflammatory cytokines in concert with Toll-like receptor signaling pathways [10]. Nod proteins fight off bacterial infections by stimulating proinflammatory signaling and cytokine networks and by inducing antimicrobial effectors, such as nitric oxide and antimicrobial peptides [11]. Fc epsilon RI-mediated signaling pathways in mast cells are initiated by the interaction of antigen with IgE bound to the extracellular domain of the alpha chain of the Fc epsilon RI [12][13][14][15][16]. The activation pathways are regulated by mast cells that release histamines and proteoglycans (especially heparin), lipid mediators such as leukotrienes (LTC4, LTD4 and LTE4) and prostaglandins (especially PDG2), and cytokines such as TNF-alpha, IL-4 and IL-5. These mediators and cytokines contribute to inflammatory response [14].
The NOD-like receptor signaling pathway, Fc epsilon RI signaling pathway, acute myeloid leukemia pathway, progesterone-mediated oocyte maturation pathway were strongly linked to the MAPK signaling pathway, which are involved in environmental information processing and signal transduction [41][42][43]; the acute myeloid leukemia pathway progesterone-mediated oocyte maturation pathway, melanoma pathway were all strongly linked to the cell cycle pathway, which plays an important role in the regulation of cell growth and death [44][45][46]; and the NOD-like receptor signaling pathway, acute myeloid leukemia pathway, type II diabetes mellitus pathway, melanoma pathway were all directly or indirectly linked to the apoptosis pathway, which plays an important role in the regulation of apoptosis (programmed cell death) [47][48][49]. Hence, the pathways linked to the cell function regulation, including the MAPK signaling pathway, apoptosis and Cell cycle pathway, were also affected directly or indirectly by the process of the body's resistance to infection.
Many cytokines as shown by leading edge analysis were activated at the site of inflammation, including IL-6, IL-8, IL-18, TNF, GM-CSF, CCL2, p101 protein, HLA-B associated transcript 1, Fc fragment of IgE, MAPK14, MAP2K1, IGF-1, STAT3 and STAT5B, etc. IL-6 is responsible for stimulating acute phase protein synthesis, as well as the production of neutrophils in the bone marrow.
IL-8 is synthesized by macrophages, endothelial cells and epithelial cells as host defenses against severe infection [50,51]. It serves as a chemical signal that attracts neutrophils to the site of any inflammation. Significant increases in IL-8 and IL6-mRNA after infection with APP have previously been observed in lung lavage as well as lung tissue using northern blotting and in situ hybridization [52,53]. IL-18 plays multiple roles in chronic inflammation and in a number of infections and enhances both Th-1-and Th-2-mediated immune response [54]. IL-18 is able to induce IFN gamma, GM-CSF, TNF-α and IL-1 in immunocompetent cells to activate killing by lymphocytes and to up-regulate the expression of certain chemokine receptors. GM-CSF stimulates stem cells to produce granulocytes (neutrophils, eosinophils, and basophils) and monocytes. Monocytes exit the circulation and migrate into tissues, whereupon they mature into macrophages. Thus, these cells play a part in the immune-inflammatory cascade, by which activation of a small number of macrophages can rapidly lead to an increase in their number, a process crucial for fighting infection. CCL2 recruits monocytes, memory T cells and dendritic cells to the sites of tissue injury, infection, and inflammation [55,56]. TNF-α can promote inflammatory response by inducing the production of other proinflammatory cytokines at the vicinity of the infection [57], and increase the expression of endothelial surface HLA-B by activation of the nuclear transcription factor NF-κB [58,59]. P101 protein is a single regulatory subunit of the phosphoinositide 3-kinase gamma (PI3Kα), which plays a crucial role in inflammatory and allergic processes [60,61], including neutrophil chemotaxis, mast cell degranulation, and cardiac function [62,63].
IL-2 is a type of cytokine immune system signaling molecule which is a leukocytotrophic hormone made in response to microbial infection that can identify the difference between self and non-self [75,76]. When environmental substances (molecules or microbes) gain access to the body, these substances (termed antigens) are recognized as foreign by antigen receptors that are expressed on the surface of lymphocytes. MHC can present antigenic peptides to T lymphocytes, which are responsible for a specific immune response that can destroy the pathogen producing those antigens [77]. CD40 is a co-stimulatory protein found on antigen presenting cells (APC) and is essential in mediating a broad variety of immune and inflammatory responses including T cell-dependent immunoglobulin class switching, memory B cell development, and germinal center formation [78]. The macrophage can express more CD40 and TNF receptors on its surface, which can increase the level of activation culminating in the induction of potent microbicidal substances in the macrophage; these include reactive oxygen species and nitric oxide, leading to the destruction of the ingested microbe [79][80][81][82]. IL-12 is an essential inducer of Th1 cell development, and has an important role in sustaining a sufficient number of memory/effector Th1 cells to mediate long-term protection against an intracellular pathogen [83]. The suppression of these immunomodulatory cytokines leads to a decrease in antigenic peptides presented to T lymphocytes by APC via the MHC, as well as to alleviate immune response injury induced by infection at the site of inflammation.
Many genes encoding metabolism as well as ribosomal proteins were affected at the site of inflammation. Genes related to metabolism and ribosomal proteins synthesis were induced in the inflamed lung, including adiponectin, cell division cycle 25 homolog C, cytochrome P450 (CYP 3A29, CYP 3A39 and CYP 3A46), FYN, PTEN, PDK, Snrpa, Syk and SPI1, among others. The repressed genes comprised those encoding MYH1, MYH2, tropomyosin (alpha, beta), troponin I type 3 (cardiac), CACNB4, CACNA2D1, CPE and FXYD2; while the CYP2E1 and the CYP3A29 were known to be down-regulated during inflammation in another study [84].
SOCS3 and CISH, both found to be up-regulated in the present study, are members of the suppressor of cytokine signaling (SOCS) family of proteins whose members regulate protein turnover by targeting proteins for degradation [42]. Expression of the members of the SOCS family is induced by cytokines such as IL-6 and IL-10, both found to be up-regulated in this study; both function as negative feed-back regulators of cytokine signaling [85,86]. The statistically significant increase in mRNA coding for the anti-inflammatory cytokine IL-10, found in inflamed areas of the lung, is probably due to the function of IL-10 in counteracting the host mediated tissue damage caused by proinflammatory and chemotactic cytokines [87]. The lower expression levels observed for genes encoding ribosomal proteins could be due to a general downregulation of ribosomal biogenesis in the necrotic areas of the lung. Previous studies have shown that 41 out of 54 genes encoding ribosomal proteins were down-regulated in Pseudomonas aeruginosa after treatment with H 2 O 2 induced oxidative stress [88].
As described above, we found that: (1) A total of 89 biological process categories, 82 cellular components and 182 molecular functions were significantly affected, and more than 27 biological process were involved in the host immune response; (2) At the site of inflammation, 13 pathways associated with host responses were affected significantly; many proinflammatory-inflammatory cytokines were activated and several immunomodulatory cytokines were suppressed at the gene expression level reflecting the complex machinery at work during an infection; (3) Many genes which were involved in a variety of cellular functions-proliferation, differentiation, growth arrest or apoptosis of normal cells that activated, could stimulate stem cells to produce granulocyte (neutrophil, eosinophil, and basophil) and monocyte. All changes of the genes and pathways which induced or repressed expression, not only led to decrease in antigenic peptides presented to T lymphocytes by APC via the MHC and alleviated immune response injury induced by infection, but also stimulated stem cells to produce granulocyte (neutrophil, eosinophil, and basophil ) and monocyte, and promote neutrophil and macrophages to phagocytose bacterial and foreign antigen at the site of inflammation. Additional work including more animals and time points is clearly needed to further delineate the host response to APP infection and will contribute to a more detailed description of the dynamics of host responses in general.

Animals, Bacterial Inoculation and Samples
All animal procedures were performed according to protocols approved by the Biological Studies Animal Care and Use Committee of Sichuan Province, China. Twenty 12-week-old male castrated Danish Landrace/Yorkshire/Duroc crossbred swine from a healthy herd free from APP were divided equally into a control group (CG) and the treatment group (TG). APP serotype I (Strain provided by the Animal Biotechnology Center, Laboratory of Animal Disease and Human Health, Sichuan Agricultural University) was cultivated overnight at 37 °C in air on trypticase soy broth (TSB) (Hangwei, Hangzhou, China). Bacterial counts of the suspensions were performed at the same time as the start of the inoculation. The inoculation was performed by holding the pigs (1-10) from the TG in an upright sitting position and spraying 0.25mL diluent containing (3.5-4) × 10 7 CFU/mL APP per kilogram weight into the nostrils during inspiration. Swine from the CG (swines [11][12][13][14][15][16][17][18][19][20] were inoculated with physiological saline (0.9% wt/vol NaCl) by the same means. In the TG, lung tissue was collected from three swine (swines 1, 2 and 3) after abattage and used for total RNA extraction and pathological analysis. Another three swine (swines 11, 12 and 13) from the CG were sacrificed 48 h post-inoculation and their lung tissues were collected. The remaining swine were used for other trials.

Microarray Hybridizations and Data Analysis
Total RNA was extracted from tissues using Trizol reagent (Invitrogen, Carlsbad, CA, USA). RNA was purified and DNase treated using the RNeasy QIAGEN RNeasy ® Mini Kit. cDNA was synthesized from 2 μg of total-RNA using the direct cDNA Labeling System. Aminoallyl-cRNA was synthesized from cDNA using the Superscript Indirect cDNA Labeling System. The cRNA was purified and DNase treated using RNeasy QIAGEN RNeasy ® Mini Kit. RNA integrity was confirmed with a bioanalyzer (model 2100; Agilent Technologies, Palo Alto, CA, USA) according to the manufacturer's protocol. Labeling and hybridization of the cRNA was performed with Agilent Whole Porcine Genome Oligo (4 × 44 K) Microarrays (one-color platform) at the National Engineering Center for Biochip at Shanghai, according to the manufacturer's protocols. The slides were scanned and analyzed using the histogram method with default settings in an Agilent G2565AA and Agilent G2565BA Microarray Scanner System with SureScan Technology. The array data were submitted to GEO [89].
Comparisons between the CG and TG were carried out using three biological replicates for each group. CG samples and TG samples were used for microarray analysis. The six Microarray data were normalized using the quantile normalization method [90] with WebarrayDB (http://www.webarraydb.org/webarray/) [91] and were filtered and assessed by the MIDAW online analysis program (http://www.webarraydb.org/webarray/) [92] using the method of weighted K-nearest neighbor [93]. T-tests and hierarchical cluster analyses of the significantly differentially expressed (DE) genes (clustering method: complete linkage; similarity measure: Pearson product momentum correlation; ordering function: average value) for microarray data were carried out by a MultiExperiment Viewer (MeV) software package (Version 4.5, Dana-Farber Cancer Institute, Boston, MA, USA, 2009) [94].
Tests for statistical significance (p < 0.05), overrepresentation of Gene ontology (GO) terms [74,95], and pathway in Kyoto Encyclopedia of Genes and Genomes (KEGG) DB [10,96] (http://www.genome.jp/kegg/) both induced and repressed genes were conducted using the ErmineJ [97] and the Database for Annotation, Visualization and Integrated Discovery (DAVID) Online platform (http://david.abcc.ncifcrf.gov/) with a threshold of a minimum three genes annotated at each node. The leading edge analysis for the pathway of differential expression in microarray data with a threshold of a minimum of two genes and maximum of 20 genes annotated at each node was conducted using the GSEA V2.06 package [98,99]. More detailed descriptions of the microarray experiments are available at the NCBI Gene Expression Omnibus [100][101][102].

Real-Time QRT-PCR
In order to confirm the reliability the expression profile in the microarray analyses, the expression level 10 gene (six up-regulated and four down-regulated) were performed by real-time qRT-PCR. Sequences for primers were obtained from Genbank and NCBI. Primers were designed using Primer 5 and synthesized at Invitrogen (Shanghai, China) ( Table 1). Extracted RNA was converted into cDNA by reverse transcription of 1 μL total RNA using SYBR® PrimeScriptTM RT-PCR Kit (TaKaRa, Japan) according to the manufacturer's protocol and then cDNA was stored at −20 °C until use. Quantitative PCR was performed in a 25 μL reaction volume (2 μL cDNA, 12.5 μL of SYBR ® Premix Ex TaqTM (2×) TaKaRa, Japan), 0.5 μL of 10 μM upstream and downstream primers respectively, and added ddH 2 O to 25 μL) on the BIO-RAD IQ5 System (BIO-RAD, Hercules, CA, USA). Real-time PCR conditions were as follows: 30 s at 95.0 °C, 40 cycles of denaturation at 95 °C for 5 s followed by 30 s annealing and elongation at 51.2-60 °C ( Table 8). Efficiency of primer pairs is reported in Table  1. Melting curves were obtained at the end of each run to confirm a single PCR product. All samples were run in triplicate. Non-template controls were included in each run to exclude contamination and nonspecific amplification. Expression levels of samples were normalised by using a normalisation factor calculated by the program geNorm. This normalisation factor was calculated based on RT-qPCR results for three selected reference genes, ACTB, TOP2B and TBP.
This allowed quantification of the target gene in one sample relative to that in another (the calibrator) using the "2 −ΔΔCt method" of calculating fold changes in gene expression [103]. Correlation analysis between qRT-PCR and microarray was conducted.

Conclusions
We have generated reliable mRNA transcriptomes of swine lung tissues from APP-infected and negative control pigs. We have identified a set of differentially expressed (DE) genes in our current case-control study, and a functional enrichment analysis indicated that these DE genes mainly related to "host immune response" and "host response". In addition, we also found that, in the APP-infected lung tissues, many proinflammatory-inflammatory cytokines were activated and involved in the regulation of the host defense response at the site of inflammation, while the cytokines involved in regulation of the host immune response were suppressed. The current study provides data that can be used in future studies to decipher the molecular mechanism of the systematic influences from porcine pleuropneumonia. Our findings will also help promote the further development of therapy for porcine pleuropneumonia.