Clinical evaluation of infected pigs

Overview of differential expressed genes in PAMs

To investigate the dynamic gene transcriptional profiles of PAMs in response to H. parasuis infection, six microarrays were used in this experiment, corresponding to the RNAs from PAMs of three H. parasuis infected piglets and three controls. The total RNA samples were hybridized with Affymetrix GeneChip Porcine Genome Array, and the microarray data were analyzed using Significance Analysis of Microarrays (SAM) [17]. Hybridization results indicated that 14,228 and 13,813 probes sets, corresponding to 58.9% and 57.3% of all probe sets, were detected in H. parasuis serovar 5 and mock-infected PAMs (Additional file 1). After quantile normalization and statistical analysis, 623 transcripts were identified at SAM |Score(d)| ≥ 2. Furthermore, Genes whose relative transcription levels showed a fold change FC ≥ 1.33 and SAM |Score(d)| ≥ 2 were considered to be up-regulated, and those with an FC ≤ 0.75 and SAM |Score(d)| ≥ 2 were considered to be down-regulated. In this study, 575 transcripts showed a level of expression that differed significantly from that of the control group with H. parasuis serovar 5 SH0165 strain infected group. These included 428 genes annotated with DAVID or by searching against the GenBank database (Additional file 2). Among these, 338 genes were up-regulated and 90 genes were down-regulated (Figure 1).

Validation of microarray data by quantitative real-time PCR (qPCR)

STRING analysis of the relationships between DE genes

STRING is a web-based interface that could predict protein associations which can mean direct physical binding and can also mean indirect interaction such as participation in the same metabolic pathway or cellular process on the basis of genomic context, high-throughput experiments, co-expression and data from the literature http://string.embl.de [21, 22]. DE genes were analyzed using STRING for predicting network of proteins encoded by DE genes. Among the 428 annotated DE genes, 236 genes containing 181 up-regulated genes and 55 down-regulated genes were eligible to STRING analyses when the Sus Scrofa database was chosen. In order to seek the possibility of the associations between DE genes, the combined score of 0.15 was chosen. The network of predicted associations for all of the DE genes encoded proteins are shown in Additional file 5. Some molecules are the key molecules that link to other proteins according to the STRING analysis. However, many proteins do not link to others, indicating that their functions are unrelated or unknown. As shown in Figure 2A, a total of 12 DE genes encoded proteins are associated with IL-1β according to the textmining evidence, and they form the IL-1β network. Furthermore, the CD14 and SOD2 are associated with IL-1β according to the co-expression evidence. Two of phagocytosis-related genes (cd14, fcgr2β) are associated with a total of 11 DE genes encoded proteins (Figure 2B) according to the textmining evidence, and they form the phagocytosis network. Furthermore, the IL-1β, FGL-2, CCL2 and FCGR2β are associated with CD14 according to the co-expression evidence.

Identification of novel infection-related DE genes

In order to identify novel candidates for disease-related DE genes, we evaluated the DE genes that were not highlighted in the KEGG or STRING analysis. Although the s100a4, s100a6 and coronin 1a were not highlighted in the KEGG or STRING analysis, they were found to play roles in the immune response [23–34]. These observations suggest that the three genes may be novel candidates for disease-related DE genes.

Among the three genes, coronin 1a has not been identified in pigs before. So we cloned and sequenced the porcine coronin 1a gene according to description of Liu et al [35] and the sequence was submitted to the GenBank [GenBank: JN092377]. The full-length cDNA of porcine coronin 1a contains 1386 bp and 461 amino acid residues. Multiple sequence alignment with the identified coronin 1a of cattle, human, mouse, rat and the predicted coronin 1a of other species showed that the nucleotide sequence of the poCORONIN 1A ORF is 93.58%, 93.58%, 92.42%, 92.35%, 92.28%, 91.70%, 87.81% and 86.87% identical to that of panda, cattle, human, chimpanzee, northern white-cheeked gibbon, common marmoset, mouse and rat coronin 1a, respectively. At the amino acid level, the corresponding identities were 96.96%, 97.61%, 95.66%, 95.66%, 95.88%, 96.10%, 93.71% and 92.84%, respectively. To define the molecular evolutionary history of poCORONIN 1A, protein sequences from 9 vertebrates were obtained to construct a phylogenetic tree. Phylogenetic analysis showed that poCORONIN 1A belongs to the group containing the Bos taurus sequence (Additional file 6). Structural analysis with the ExPASy server http://expasy.org/ indicated that the poCORONIN 1A contains putative domains of Trp-Asp (WD) repeats signature, Trp-Asp (WD) repeats profile and Trp-Asp (WD) repeats circular profile at the N-terminus (Additional file 7).

Expression analyses of S100A4, S100A6 in PK15 cells stimulated with LPS and Poly (I:C)

In order to investigate the expression patterns of s100a4 and s100a6 under general conditions that mimic bacterial and viral infection, the immunostimulation assay was carried out in PK-15 cells by using the LPS and Poly (I:C) as the stimulators.

Overnight cultures of PK-15 cells were treated with 1 μg/ml LPS or 10 μg/ml Poly (I:C) for 0, 2, 6, 12, 24 and 48 h. LPS and Poly (I:C) stimulation did not induce expression of porcine s100a4 until 48 h (Figure 3A, C). LPS stimulation induced expression of s100a6 at 2 h and 12 h, after which s100a6 expression dropped and plateaued for 24-48 h (Figure 3B). After Poly (I:C) stimulation, the expression of s100a6 reached the peak at 12 h, after which s100a6 expression dropped at 24 h, and the up-regulation of s100a6 was again observed at 48 h (Figure 3D). These observations indicate that both LPS and Poly (I:C) can induce the expression of porcine s100a4 and s100a6 in vitro.

In vivo expression of s100a4 and s100a6 in pigs with systemic infection of H. parasuis

In order to understand the expression of the s100a4 and s100a6 in pigs with systemic infection of H. parasuis, the different tissues obtained from the H. parasuis infected pigs and the controls were selected for the qPCR analysis. Our qPCR examination demonstrated that the increasing expression of s100a4 was observed in the lungs, spleen and lymph nodes of pigs infected with H. parasuis for 6 days (Figure 4A). The expression of s100a6 in the lungs, spleen and lymph nodes had the same expression tendencies (Figure 4B). However, in brain and heart of H. parasuis infected pigs, the expression of s100a4 and s100a6 did not show significant changes compared to the controls.

Animals for Microarray experiment and porcine alveolar macrophages isolation

All animals' tissue collection procedures were performed according to protocols approved by the Hubei Province PR China for Biological Studies Animal Care and Use Committee. Six piglets which were obtained from a commercial herd free of Glässer's disease were weaned at 27 days, shipped to the Animal Disease Center of Huazhong Agricultural University, and raised with isolation facilities. Three piglets were randomly allocated to the non infected group and three to the infected group. The three piglets were intratracheally challenged with H. parasuis strain 0165 (serovar 5) at a dose of 6 × 10⁹ colony-forming units (CFU). The noninfected group piglets were treated similarly with identical volume of PBS served as control. All piglets were determined to the HPS-free by serum indirect haemagglutination (IHA) test before artificial bacterial challenges. Clinical signs and lesions of Glässer's disease were apparent in the challenged group at 6 days post-infection (dpi). All piglets were slaughtered at 6 dpi. Bacterial isolation, nested PCR and LAMP were performed after the piglets were killed at 6 dpi. PAMs were isolated according to Olvera's description [13]. Briefly, Bronchoalveolar lavage of the lungs was performed with 100 mL aliquots of sterile PBS containing gentamicin at 70 μg/mL (Sigma-Aldrich). To collect the porcine alveolar macrophages (PAM), lavage fluids were centrifuged at 230 g for 15 min, and then cells were washed twice with Dulbecco's Modified Eagle's Medium (DMEM) with gentamicin (50 μg/mL). PAM isolation was confirmed by detection of macrophage markers (SWC3, CD169 and SLAII) in the cells by flow cytometry.

RNA preparation for Microarray experiment

Total RNA were extracted from PAM of each group with Trizol (Invitrogen) then quantified using the NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific Inc., USA). The quality of the RNA was checked by formaldehyde denaturing gel electrophoresis in 1.2% agarose gels, which showed dispersed bands (28S and 18S) without any obvious smearing patterns that would indicate degradation.

Microarray hybridization and data analyses

Affymetrix GeneChip Porcine Genome Array, which contains 24,123 probe sets to interrogate 23,256 transcripts in pig, represents 20,201 genes, was used in microarray analysis. Hybridization, data capture and analysis were performed by CapitalBio Corporation (Beijing, China), a service provider authorized by Affymetrix Inc. (Santa Clara, CA). Briefly, a total of 1 μg RNA was used for cDNA synthesis and to produce biotin-tagged cRNA with GeneChip IVT Labeling kit (Affymetrix). A total of 15 μg fragmented cRNA, with contol oligo B2 and eukaryotic hybridization controls (bioB, bioC, bioD, cre) was hybridized to each GeneChip array at 45°C for 16 hours (Affymetrix Gene Chip Hybridization Oven 640) according to manufacturer's instructions. After hybridization, the GeneChip arrays were washed and stained with streptavidin phycoerythrin onan (SAPE) with Affymetrix Fluidics Station 450 followed by scanning with the Affymetrix GeneChip Scanner 3000. Six microarrays were used in the experiment, corresponding to the RNAs from PAMs of three H. parasuis infected piglets and three controls.

The hybridization data were analyzed using GeneChip Operating Software (GCOS, version 1.4), which uses statistical criteria to generate a 'present' or 'absent' call for genes represented by each probe set on the array. The scanned images were first assessed by visual inspection and then analyzed to generate raw data files saved as CEL files using the default setting of GCOS 1.4. Microarray data were normalized using the robust multi-array average (RMA) method [83], which consists of three steps: background correction, quantile normalization (each performed at the individual probe level), and robust linear model fit using log-transformed intensities (at the probe set level). Significance Analysis of Microarrays (SAM) add-in to Microsoft Excel was used for comparisons of replicate array experiments. SAM identifies genes with statistically significant changes in expression by assimilating a set of gene-specific t-tests, and provides an estimate of the false discovery rate (FDR) from randomly generated data. Genes with scores higher than a threshold value or genes with FDR value lower than the threshold value were deemed potentially significant. Furthermore, fold-change analysis which calculates the ratios of geometric means of expression intensities of H. parasuis-infected PAMs relative to controls was performed. These ratios were reported as the up-or down-fold change. In this study, genes were considered statistically significant if they had SAM |Score(d)| ≥ 2 [84, 85] and exhibited a fold change ≥ 1.33 and ≤ 0.75. DE genes performed for hierarchical cluster (Ver.3.0) and TreeView (Ver.1.1.1) analyses. Genes with significant similarities to the transcripts in nr database based on BLASTX searches were selected for GO analysis, performed by MAS 3.0 software which was based on DAVID database (CapitalBio, Beijing, China) [16]. Annotation results were obtained by inputting the list of gene symbol as identifier [18]. The Pathway analysis was done using the MAS 3.0 software which was based on the Kyoto Encyclopedia of Genes and Genomes database (KEGG) (CapitalBio, Beijing, China). All microarray results from this study were deposited in NCBI'S Gene Expression Omnibus (GEO) database, accession numbers are: Platform, GPL 3533, Samples, GSM 747145, GSM 747146, GSM 747147, GSM, 747148, GSM 747149, GSM 747150 with the series accession number GSE 30172.

QPCR analysis

Total RNA were extracted from the PAMs of each group with Trizol (Invitrogen) and 5 μg of total RNA were used for first strand cDNA synthesis by using Superscript II cDNA amplification System (Invitrogen) following manufacturer's instructions. Real-time PCR was performed using LightCycler 480 (Roche Applied Science) and Quantitect SYBR Green PCR kit (Roche) following the companies' instructions. Briefly, PCR assay was performed under the following conditions: 95°C for 15 sec, 55°C for 15 sec and 72°C for 15 sec. Real-time PCR primers for each gene were indicated in Table 3. All the primers were originally designed using Primer 3 software (Rozen & Skaletsky, 2000) or according to the published papers. Results were calculated by minus delta delta threshold cycle (-ddCt) method. Briefly, the threshold cycle Ct₁ of each sample reaction were deducted with the threshold cycle Ct₂ of GAPDH reaction for normalization, then deducted from the threshold cycle Ct₃ of calibration control (40 Cycles in this experiment); thus, the final result was represented by the formula: Ct₃-(Ct₁-Ct₂).

Expression of S100A4, S100A6 in PK-15 cells stimulated with LPS and Poly (I:C)

PK-15 cells have been shown especially useful for the study of infectious disease processes in swine [86, 87]. In this study, 12 groups (with three repeats in each group, ~1 × 10⁵ cells/samples) of PK-15 cells were grown in culture medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum at 37°C with 5% CO₂. Adherent PK-15 cells were obtained by washing off nonadherent cells with warm culture medium and PBS twice, respectively. Adherent cells were further cultured in DMEM (control samples) or treated with 1 μg/mL LPS (Sigma-Aldrich, E.coli 0127:B8) or 10 μg/mL Poly (I:C) (Sigma-Aldrich) respectively (stimulated samples) for 0 h, 2 h, 6 h, 12 h, 24 h and 48 h. Cells were harvested and total RNA were extracted as described above.

DNA preparation from bacterial isolates and clinical samples

Bacterial cultures were harvested from trypticase soy agar (TSA) using an inoculation loop and were placed into a 1.5 mL tube to which was added with 500 μL of phosphate buffered saline (PBS). One milliliter of the fluid and 0.5 g of the tissue samples were respectively placed in sterile tubes containing 5 mL of trypticase soy broth (TSB), 5 μL nicotinamide adenine dinucleotide (NAD) and 500 μL sterilized fetal bovine serum and then incubated for 8 h at 37°C with agitation. Five hundred microliters of the suspension was removed to a new 1.5 mL tube. Tubes containing bacteria, tissue and fluid suspensions were centrifuged at 13,400 g for 5 min. After centrifugation, the supernatant was discarded and the remaining pellet was suspended in 200 μL PBS, boiled for 10 min. After boiling, tubes were centrifuged at 13,400 g for 5 min. Fifty microliters of supernatant from each sample containing extracted DNA were mixed with 50 μL of Tris-EDTA buffer and stored at 4°C. This final solution was used as DNA template in nested PCR and LAMP reaction. The primers for nested PCR and LAMP were listed in Additional file 9. The procedure of bacterial isolation, nested PCR and LAMP were carried out according to description of Wang et al [6].

Detection of s100a4 and s100a6 expression in different tissues

Three pigs in H. parasuis infection group and control group were selected for the analysis of s100a4 and s100a6 expression in different tissues. Total RNA from 5 porcine organs (inguinal lymph node, heart, spleen, lung, brain) was isolated with RNAprep pure Tissue Kit (TianGen Biotech (Beijing) Co., Ltd). Total RNA was then quantified by NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific Inc., USA). The quality of the RNA was checked by formaldehyde denaturing gel electrophoresis in 1.2% agarose gels, which showed dispersed bands (28S and 18S) without any obvious smearing patterns that would indicate degradation. Two microgram of total RNA was used for reverse transcription polymerase chain reaction, using the TransSript First Strand cDNA Synthesis SuperMix according to the manufacturer's instructions (TianGen Biotech (Beijing) Co., Ltd). The qPCR assays were performed and analyzed as described above, with primers listed in Table 3.

Data for STRING and IPA analysis

Differentially expressed (DE) genes were analyzed using STRING http://string.embl.de, a database of known and predicted protein interaction for DE gene encoded proteins. The results were obtained by inputting the list of gene symbol as identifier (organism = sus scrofa, combined score = 0.15). Ppp1r13l gene was selected for network exploration using Ingenuity Pathway Analysis (Ingenuity^® Systems, http://www.ingenuity.com). The data set containing gene identifier and corresponding expression value was uploaded into in the application. The identifier was mapped to its corresponding object in Ingenuity's Knowledge Base; Network Eligible molecules were then overlaid onto a global molecular network so that network of Network Eligible Molecules could be algorithmically generated based on their connectivity.