|Home | About | Journals | Submit | Contact Us | Français|
Porcine reproductive and respiratory syndrome is a major cause of economic loss for the swine industry worldwide. Porcine reproductive and respiratory syndrome virus (PRRSV) triggers weak and atypical innate immune responses, but key genes and mechanisms by which the virus interferes with the host innate immunity have not yet been elucidated. In this study, genes that control the response of the main target of PRRSV, porcine alveolar macrophages (PAMs), were profiled in vitro with a time-course experiment spanning the first round of virus replication. PAMs were obtained from six piglets and challenged with the Lelystad PRRSV strain, and gene expression was investigated using Affymetrix microarrays and real-time PCR. Of the 1409 differentially expressed transcripts identified by analysis of variance, two, five, 25, 16 and 100 differed from controls by a minimum of 1.5-fold at 1, 3, 6, 9 and 12 h post-infection (p.i.), respectively. A PRRSV infection effect was detectable between 3 and 6 h p.i., and was characterized by a consistent downregulation of gene expression, followed by the start of the host innate immune response at 9 h p.i. The expression of beta interferon 1 (IFN-β), but not of IFN-α, was strongly upregulated, whilst few genes commonly expressed in response to viral infections and/or induced by interferons were found to be differentially expressed. A predominance of anti-apoptotic transcripts (e.g. interleukin-10), a shift towards a T-helper cell type 2 response and a weak upregulation of tumour necrosis factor-α expression were observed within 12 h p.i., reinforcing the hypotheses that PRRSV has developed sophisticated mechanisms to escape the host defence.
Porcine reproductive and respiratory syndrome is a major cause of economic loss for the swine industry worldwide (Neumann et al., 2005) and causes high mortality of nursery piglets, reproductive failure in sows, respiratory distress in pigs of all ages and influenza-like symptoms in grow/finish swine (Mengeling & Lager, 2000; Nodelijk, 2002). The aetiological agent is porcine reproductive and respiratory syndrome virus (PRRSV), belonging to the family Arteriviridae with an enveloped, positive-stranded RNA genome of about 14.5 kb (Snijder & Meulenberg, 1998).
A typical hallmark of PRRSV is that it causes an acute viraemic phase (up to 14 days post-inoculation) during which the virus can be detected in serum and all susceptible organs (Beyer et al., 2000; Duan et al., 1997b). This acute phase is followed by virus elimination from serum and most organs, and by persistent replication in tonsils, lungs and some lymph nodes (Allende et al., 2000; Rowland et al., 2003; Wills et al., 2003). This prolonged replication does not represent a true persistent infection, as all animals clear the virus by 6 months after inoculation, thus indirectly showing that the immune system is capable of finally dealing with the virus, although not efficiently. Because of this persistent nature of PRRSV infections, numerous studies have analysed the immune responses that may control PRRSV infections or that may be altered by PRRSV (reviewed by Lopez & Osorio, 2004; Mateu & Diaz, 2007; Murtaugh et al., 2002).
The PRRSV-specific humoral immunity is generally characterized by a strong, non-neutralizing antibody response, which is detected from 5–6 days post-infection (p.i.). In contrast, induction of neutralizing antibodies is severely delayed (starting at 3–4 weeks p.i.) and their levels remain low (Lopez & Osorio, 2004); antibodies were shown to be ineffective in eliminating PRRSV-infected macrophages in combination with complement (Costers et al., 2006). Cellular immune responses against PRRSV infection are characterized by a late onset of lymphocyte proliferative responses (4 weeks p.i.) and the late appearance of gamma interferon (IFN-γ)-secreting cells (Meier et al., 2003). Several studies have also shown weak and atypical innate immune responses, such as weak IFN-α responses and high induction of interleukin (IL)-10. This inadequate recognition of virus infection by the innate defence mechanisms could be responsible for the initially crippled immune response (Albina et al., 1998; Buddaert et al., 1998; Murtaugh et al., 2002; Royaee et al., 2004; Suradhat et al., 2003; van Reeth et al., 1999; Xiao et al., 2004). The mechanism by which PRRSV interferes with innate immune responses has yet to be elucidated.
PRRSV has a highly specific tropism for cells of the monocyte/macrophage lineage, cells that are essential for immune function. In vivo, the virus mainly infects a subpopulation of differentiated macrophages that are present in tonsils, lungs and other lymphoid tissues (Beyer et al., 2000; Duan et al., 1997a, b). Besides macrophages, in vitro analysis of susceptible cells has identified cultivated monocytes and dendritic cells as potential targets, but their role during PRRSV infections in vivo remains to be established (Delputte et al., 2007; Duan et al., 1997a; Loving et al., 2007; Teifke et al., 2001; Voicu et al., 1994; Wang et al., 2007). Lung pathogenesis is another feature of PRRSV infections, and porcine alveolar macrophages (PAMs) are generally considered to be a major target for PRRSV.
The aim of this study was to gain insight into the putative mechanisms by which PRRSV can evade innate immunity, and consequently the adaptive response, using a genome-wide approach. A time-course gene expression profiling of PAMs infected in vitro with a reference strain (Lelystad) was conducted by utilizing an Affymetrix 24K Porcine Chip microarray. Collection of samples at different times during the infection cycle, from 1 h p.i. (virus entry) up to 12 h p.i. (virus release and cell death) allowed us to discriminate between changes in early and late gene expression during infection. Times later than 12 h p.i. were not analysed, as by that time PRRSV infection of macrophages has typically resulted in cell death.
Six 3-week-old hybrid piglets from a PRRSV- and porcine circovirus 2-negative herd of the Rattlerow–Seghers genetic line (a cross-breed between English Landrace, Belgian Landrace, Large White and a synthetic company Landrace) were injected daily with 1 ml enrofloxacin (5% solution) and 1 ml lincospectin/spectinomycin (5 or 10% solution) for 3 days to eliminate eventual bacterial pathogens. Two weeks later, the piglets were sacrificed. PAMs were collected by bronchoalveolar lavage and frozen in liquid nitrogen as described by Wensvoort et al. (1991).
PAMs were thawed and cultured for 48 h before treatment as described previously by Delputte & Nauwynck (2004). One primary culture from each animal was split into two: one was infected at an m.o.i. of 10 with a 13th passage of PRRSV Lelystad virus (kindly provided by G. Wensvoort, Institute for Animal Science and Health, Lelystad, The Netherlands), which was semi-purified as described previously (Delputte & Nauwynck, 2004). The other culture was maintained as a control and was mock inoculated. The percentage of infected cells ranged between 60 and 70% for all batches. Cells were collected at 1, 3, 6, 9 and 12 h p.i. in TRIzol (Invitrogen Life Technologies) for RNA extraction (Fig. 1).
Total RNA extraction from PAMs was performed using TRIzol following standard instructions (Invitrogen) and a clean-up was carried out using RNeasy columns (Qiagen). RNA quality was assessed by microcapillary electrophoresis on an Agilent 2001 Bioanalyser (Agilent Technologies) with RNA 6000 Nanochips. RNA was quantified by spectrophotometry (ND-1000; NanoDrop Technologies). Reverse transcription of 20 μg total RNA and synthesis of biotin-labelled cRNA with one round of amplification were carried out following the standard Affymetrix one-cycle protocol according to the manufacturer's instructions.
Transcriptional profiles were assessed using Affymetrix 24K GeneChip Porcine Genome Arrays (http://www.affymetrix.com/products/arrays/specific/porcine.affx). Based on previous evidence that sample pooling does not significantly affect the results of Affymetrix chip analysis (see, for example, Han et al., 2004), three samples each from control and infected-cell cultures were pooled for each time point (Fig. 1), resulting in two control (pools I− and II−) and two infected pools (pools I+ and II+).
Hybridization and scanning of the arrays were carried out according to standard Affymetrix protocols (Shen et al., 2005) using a GeneChip Scanner 3000 7G.
Signal intensities were evaluated using the GeneChip Operating Software algorithm (gcos version 1.4; Affymetrix). Raw data and statistical analyses were performed with GeneSpring version 7.3.1 software (Agilent). Normalization was performed per chip (normalized to 50th percentile) and per gene (normalized to the median).
A statistical analysis of variance (ANOVA) model was applied to the data and significance was declared accepting a false discovery rate (FDR) of 0.05. Fixed effects of time point and status (infected−non-infected cells) were included in the ANOVA model. A further cut-off threshold was applied based on a fold change of 1.5 between infected and control PAMs. Hierarchical clustering of the conditions was performed using Pearson's correlation coefficient (r) as a measure of similarity and the average linkage method as the clustering algorithm.
In order to test for the presence of outliers in the two pools, the transcriptional profiles of infected animals were analysed separately at the 3 h p.i. time point. A paired t-test (paired across pools by gene) was performed using the range of minimum and maximum corrected expression values within each pool for each gene. The test was applied (i) to the whole set of genes and (ii) to the subset of genes that appeared to be significant for differential expression in the general analysis. No significant difference was observed either for the whole set of genes included in the study or for the subset of differentially expressed genes.
The Database for Annotation, Visualization and Integrated Discovery (DAVID 2006; http://david.abcc.ncifcrf.gov/), an expanded version of the original web-accessible programs described by Dennis et al. (2003), was used to allocate transcripts with similar biological questions into the three gene ontology (GO) categories.
Quantitative real-time PCR analysis was conducted on ten selected swine transcripts and on the ORF7 gene of PRRSV. Hypoxanthine phosphoribosyltransferase (HPRT1) was chosen as the reference gene because the amplifications of all control and infected samples showed very similar threshold cycle (Ct) values (data not shown). The transcript-specific primers were designed using ProbeFinder software (version 2.3) on the Roche website (https://www.roche-applied-science.com/sis/rtpcr/upl/adc.jsp) using standard settings for the human Universal Probe Library Set catalogue (see Supplementary Table S1, available with the online version of this paper).
Two micrograms of total RNA from pools I and II were reverse-transcribed using the Superscript II RT-PCR System (Invitrogen Life Technologies) and standard procedures. The real-time reaction mixture (total 20 μl) included 5 μl cDNA as template (diluted 1:50), 200 nM of each of the two primers (forward and reverse), 100 nM Roche probe and 1× master mix (Applied Biosystems). Real-time PCR was performed in 384-well optical plates using a Tecan Freedom EVO-150 liquid handling workstation (Tecan Trading) and an ABI 7900HT real-time PCR machine (Applied Biosystems) with the GeneAmp 7900HT sequence detection system software (PerkinElmer).
A control cDNA dilution series (1:50, 1:100, 1:500 and 1:5000) was created for each transcript to establish a standard curve for each plate; real-time reactions of the same pools described for the microarray analysis were performed in triplicate. Briefly, the log input amount of the standard curve was plotted against the output Ct values; all amplifications had a slope of between −3.48 and −2.99 and were accepted as quantitative. The log input amount of each sample was then calculated according to the formula (Ct−b)/m, where b is the y-intercept and m is the slope. The log input amount was converted to input amount according to the formula 10log input amount and triplicate input amounts were averaged for each sample. The mean input amount of each gene was normalized to the mean input amount of HPRT1. A t-test (with thresholds for statistical significance set to 0.1 and 0.05) was applied to each gene to verify whether the difference between control and infected macrophages at each time point was significant.
Pearson's correlation coefficient (r) was calculated for each gene on the normalized data to quantify the consistency between microarray experiments and real-time PCR.
The data of the microarray analysis were deposited in the ArrayExpress repository (http://www.ebi.ac.uk/arrayexpress) with ArrayExpress accession number ×10−MEXP-1350, following the guidelines of the rationale of minimum information about a microarray experiment (MIAME) (Brazma et al., 2001).
ANOVA analysis (FDR=0.05) showed that 1409 genes were differentially expressed in macrophages after PRRSV infection. After applying a further filter of 1.5-fold change in expression, two, five, 25, 16 and 100 transcripts were differentially expressed at 1, 3, 6, 9 and 12 h p.i., respectively, compared with the controls at the same time points. Overall, the effect of PRRSV on the host transcription machinery was one of downregulation (115/148 transcripts). The differentially expressed transcripts were annotated based on a previous work (Tsai et al., 2006) and are reported in Table 1. The distribution of signal intensities of the 100 differentially expressed transcripts at 12 h p.i. and the hierarchical clustering of controls and infected replicates for the five time conditions (plus the time 0) are shown in Fig. 2.
At early time points (1 and 3 h p.i.), the profiles of gene expression in the control and infected conditions were very similar and clustered together, i.e. only two (1 h p.i.) and five (3 h p.i.) transcripts were significantly altered. The expression profiles clearly changed between 3 and 6 h p.i., with greater differences detected at the later time points (9 and 12 h p.i.), when PRRSV has been shown to complete its replication (Halbur, 2001; Rossow et al., 1995).
The 6 h p.i. time point was characterized by a consistent downregulation of gene expression in the infected cells. The 24 downregulated transcripts represented genes with functions related to RNA processing (HNRPLL and NXT2), regulation of biological processes (ATF2, PTPRC, HIF1-α, DLC1 and RB1CC1) and signal transduction (PLAA, PTPRC, HIF1-α, DLC1 and GPR160). Only one transcript, an RNA-dependent helicase (DDX17), was upregulated.
The 9 h p.i. time point was the only one at which most transcripts (11/16) were upregulated. These represented genes (IFIT1, GBP1, USP18 and cig5) that encode accessory proteins related to the immune response, and in particular to the pro-inflammatory cytokine IFN-β, but also genes with a known anti-apoptotic function (ADM and TNF-αIP3).
At 12 h p.i., the downregulated transcripts were also largely predominant over the upregulated ones (80 vs 20, respectively). The latter confirmed the main pattern of anti-apoptotic and antiviral response already observed at 9 h p.i., with the addition of two new transcripts representing TNF-α and IL-10. The overall highest fold change (FC) was observed for IFN-β (FC=20.15 at 12 h p.i.), whilst the most downregulated transcript was NP_060114 (FC=0.306 at 12 h p.i.). NP_060114 corresponds to the human DRE1 protein, a member of the kelch-repeat family, which modulates host immune response to viral infection (Prag & Adams, 2003). KHLX belongs to the same family and also showed a consistent downregulation at 12 h (FC=0.654).
The GO analysis assigned the 100 differentially expressed transcripts at 12 h p.i. to 34 biological processes, five molecular functions and three cellular components (Table 2), with the best ranked biological processes (response to stimulus, response to stress and immune response) effectively representing the general pattern of cell response to infection. This was independently confirmed by assigning the same 100 transcripts to regulatory pathways using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. The most represented pathways were all related to the immune response and included mitogen-activated protein kinase (MAPK), JAK–STAT, natural killer cell-mediated cytotoxicity, T-cell receptor signalling, cytokine–cytokine receptor interaction and Toll-like receptor signalling (data not shown).
The genes tested by real-time PCR (see Supplementary Table S1) were selected to validate the microarray results and to confirm the involvement of key biological pathways. The set included the differentially expressed genes IFN-β, cig5, TNF-α, TNF-αIP3, IL-10, USP18 and GRP58, as well as three additional genes selected from recent literature (IFN-α, IFN-αR1 and sialoadhesin), which were represented on the array but had not shown differential expression (Table 3). Pearson's correlation coefficient (r) showed that both the microarray and real-time PCR data were highly correlated for the genes modulated between 6 and 12 h p.i., with a high consistency between pools and with r values ranging from 0.95 (IFN-β) to 0.88 (USP18 and IL-10) (Table 3). Only GRP58, shown by microarrays to be upregulated at 1 h p.i. but downregulated at 6 h p.i., showed inconsistencies between pools and a very poor r value.
Real-time PCR confirmed that IFN-β was the most upregulated gene, whilst IFN-α was not differentially expressed between control and infected cells at 9 and 12 h p.i. Moreover, real-time PCR analysis of PAMs in an independent challenge experiment, with a different viral strain and lower m.o.i., confirmed that, at 24 h p.i., IFN-β was strongly induced whilst IFN-α was only slightly upregulated (data not shown). The expression of IFN-β increased together with the PRRSV titre, as affirmed by PRRSV ORF7 gene expression (Fig. 3). Interestingly, real-time PCR at 3 h p.i. showed a small but significant peak in IFN-β and IFN-α expression, which was not detected by microarrays. IFN-αR1 was confirmed not to be differentially expressed. Statistically significant differences in values of sialoadhesin expression were found between infected and control samples, but this was inconsistent between pools.
The finding that the only gene to be upregulated at 6 h p.i. was a host RNA-dependent helicase (DDX17) indicates that PRRSV does not induce a generalized suppression of host gene transcription. This confirms and reinforces previous observations (Zhang et al., 2000), showing enhanced production of a cellular helicase (RHIV-1) in macrophages in response to PRRSV infection. Other RNA viruses, such as poliovirus and vesicular stomatitis virus, replicating exclusively in the cytoplasm using virus-encoded RNA-dependent RNA polymerases and thus not requiring the host transcriptional apparatus, inhibit nuclear transcription of cellular RNA polymerases (Weidman et al., 2003). This transcription shut-off not only allows the virus to evade cellular responses, but also may favour viral RNA replication by increasing the pool of free ribonucleotides in the cell. The complex expression pattern revealed by the microarrays might suggest that PRRSV blocks specific cellular transcription processes while upregulating others that are potentially beneficial for virus replication. By having a regulatory effect on cellular transcription, viruses may promote their own replication while interfering with the innate and adaptive immune responses that would result in their removal.
The downregulation of four genes encoding mitochondrial proteins (NP_060530, SDHC, MRPL2 and MRPL22) at different time points (6, 9 and/or 12 h p.i.) might add up to the emerging role of mitochondria in antiviral immunity. The mitochondrial antiviral signalling protein MAVS is critical for the IFN-β signalling pathway in response to dsRNA, and is required for both TLR3-mediated and TLR3-independent signalling pathways, such as that triggered by the RNA helicase RIGI (Moore et al., 2008; Xu et al., 2005; Yoneyama et al., 2004). RIGI is the product of DDX58, a member of the DEAD box family of RNA helicases that mediate nucleoside triphosphate-dependent unwinding of dsRNA and are involved in many diverse cellular functions (Lamm et al., 1996). Intriguingly, the only upregulated gene found by microarrays at 6 h p.i. in PAMs (DDX17) belongs to the same family.
The atypical pattern of expression of innate immunity genes indicates that PRRSV has probably developed sophisticated mechanisms to control the antiviral response. Indeed, only a subset (IFIT1, GBP1, USP18 and TNF-αIP3) of genes commonly modulated by pathogens in response to dsRNA and/or stimulated by IFN (Jenner & Young, 2005) were found to be upregulated by PRRSV at 9 and/or 12 h p.i. When the 1.5-fold change threshold was not applied after ANOVA analysis, this subset also included CD44, PML, PRKRA, CCl4, CCl8 and MT2A. Upregulation of USP18 has been observed previously in PAMs following PRRSV infection (Zhang et al., 1999). The same study reported the upregulation of the antiviral gene MX1, but neither MX1 nor MX2 was found to be differentially expressed in the present investigation. Downregulation of NRAMP2 at 12 h p.i. was consistent with the effects observed previously in humans after human immunodeficiency virus infection (reviewed by Jenner & Young, 2005).
Production of IFN-α and IFN-β is a well-known reaction of virus-infected cells; however, only the IFN-β gene was strongly upregulated by PRRSV in PAMs. The induction of IFN-β mRNA, but not IFN-α mRNA, has also been observed in monocyte-derived dendritic cells infected by PRRSV at 12 h p.i. (Loving et al., 2007). Previous studies, both in vitro and in vivo, have also shown that PRRSV is a poor inducer or even a suppressor of IFN-α compared with other respiratory viruses (Albina et al., 1998; Buddaert et al., 1998; Miller et al., 2004; van Reeth et al., 1999). Blocking IFN-α production clearly is beneficial for PRRSV replication, as IFN-α can efficiently block replication when present during infection (Delputte et al., 2007; Loving et al., 2007). IFN-β can also protect macrophages against PRRSV infection (Overend et al., 2007), but it has been suggested that IFN-β alone may be not sufficient to trigger the adaptive immune response (Loving et al., 2007). A recent report has shown that in vitro stimulation of monocytes and macrophages with IFN-α induces expression of sialoadhesin, the main PRRSV receptor in PAMs, and that treatment with IFN-α before inoculation strongly increases PRRSV infection of monocytes (Delputte et al., 2007). In agreement with this, in this study neither the gene encoding sialoadhesin nor that encoding IFN-αR1 (IFN receptor 1) showed consistent differential expression in infected cells.
Despite previous evidence that IFN-β expression by infected cells mediates and potentiates apoptosis (Tanaka et al., 1998), the present study showed a predominance of transcripts leading to prolonged cell survival within 12 h of infection (both upregulation of anti-apoptotic transcripts and downregulation of pro-apoptotic genes). Upregulation was observed for IL-10, ADM and TNF-αIP3. IL-10 has been demonstrated to protect cells against apoptosis (Sieg et al., 1996; Zhou et al., 2001). ADM has been shown (Kubo et al., 1998) to be overproduced by macrophages after inflammation and to modulate cytokine production (specifically TNF-α); several different independent studies support the fact that ADM is an anti-apoptotic peptide on different cell types (Bi et al., 2007; Uzan et al., 2006; Yin et al., 2004). TNF-αIP3 is a cytoplasmic zinc finger protein that inhibits NF-κβ activity and TNF-mediated programmed cell death (Li et al., 2006; Qin et al., 2006). Downregulated genes included those encoding NLK, a stimulator of apoptosis (Yasuda et al., 2003), HIF1-α, which has been suggested to favour apoptosis in the absence of oxygen (Bruick, 2000), and GRM5, known to protect neurons from apoptotic death (Maiese et al., 2000). Taken together, these findings suggest that PRRSV actively induces an anti-apoptotic state in order to complete its virus replication cycle. This is discordant with previous results showing that PRRSV induces infected cells, as well as uninfected bystander cells, to undergo apoptosis (for examples, see Chang et al., 2005; Sirinarumitr et al., 1998), but it should be noted that those data were obtained with in vitro infection treatments much longer than 12 h. On the other hand, the absence of apoptotic induction by PRRSV has been observed in MARC-145 cells (Miller & Fox, 2004) and HeLa cells (Lee et al., 2004). Interestingly, Kim et al. (2002) reported an atypical form of apoptosis that culminates in increased cell membrane permeability and late apoptosis after completion of virus replication.
The upregulation of IL-10 gene expression (FC=1.9) indicates that the IL-10-mediated downregulation of the T-helper cell type 1 (Th1) response may be an important mechanism operated by PRRSV, as well as by other viruses (for reviews, see Fickenscher et al., 2002; Redpath et al., 2001). Upregulation of IL-10 expression was found previously in PRRSV-infected porcine monocytes, macrophages and dendritic cells (Flores-Mendoza et al., 2008; Suradhat et al., 2003) and in vivo in PRRSV-infected pigs (Suradhat & Thanawongnuwech, 2003; Sutherland et al., 2007; Thanawongnuwech & Thacker, 2003; Thanawongnuwech et al., 2004). IL-10 in PRRSV-infected cells seems to be increased concurrent with the onset of viraemia and the development of clinical signs (Díaz et al., 2005). Also, PIK3R1 (upregulated in this study: FC=1.6), is known to positively regulate the production of IL-10 (Saegusa et al., 2007). These findings add to previous studies (Murtaugh et al., 2002; Wang et al., 2007), suggesting that PRRSV causes an imbalanced immune response characterized by an abundance of humoral immunity (Th2-mediated), which is less effective against viral pathogens.
The TNF-α gene was only slightly upregulated at 12 h p.i. (FC=1.5). The role of TNF-α in PRRSV infection is controversial: it has been reported that PRRSV is a potent inducer of TNF-α in PAMs at 18, 36, 54, 72, 90 and 108 h p.i. (Chang et al., 2005) and at 6 and 15 h p.i. (Thanawongnuwech et al., 2004). However, Charerntantanakul et al. (2006) showed that, in porcine monocytes infected by PRRSV, IL-10 gene expression increased, and this response contributed to a reduction in TNF-α production. In fact, crucial anti-inflammatory activities of IL-10 may be due to its inhibitory effects on TNF-α production (Moore et al., 2001). Overall, this suggests that IL-10 may also participate in fine-tuning the production and effects of TNF-α.
Other differentially expressed genes (see Table 1) confirmed that a complex pattern of TNF-α regulation takes place upon PRRSV infection. TNF-αIP3, known to be induced by TNF-α (Dixit et al., 1990; Lee et al., 2000) and suggested to protect against the inflammatory response to influenza virus infection (Onose et al., 2006), was upregulated. STAG2, an enhancer of TNF production (Lara-Pezzi et al., 2004), and the member of the MAPK pathway, ATF2, a transcription activator of both IFN-β and TNF-α in response to virus infection (Biron & Sen, 2001; Tsai et al., 1996), were downregulated. The MAPK pathway was the most highly represented gene network identified in this study, with five differentially expressed genes at 12 h p.i. (ATF2, TNF-α, MAP3K8, MKNK2 and NLK). The MAPK pathway is one of the most important pathways for immune response to infection (Bruder & Kovesdi, 1997; Yang et al., 2007) and has been found to be modulated in PAMs after an antibody-mediated cross-linking treatment of sialoadhesin, the main PRRSV internalization receptor (Genini et al., 2008), although in this case different genes of the pathway were involved.
In conclusion, this work has provided a genome-wide gene expression catalogue of PRRSV pathogenesis and has allowed us to picture how different genes and gene pathways are co-modulated in the physiological context of the PAM. As such, these results provide a large-scale and unbiased basis for further investigations on gene roles and functions. The sequencing of the pig genome currently in progress (www.piggenome.org), besides allowing a complete annotation of all transcripts, will soon give important clues for future genomics studies, for example for the interpretation of genome scan analyses aimed at identifying the genetic components involved in PRRSV resistance/susceptibility in swine populations (Ait-Ali et al., 2007; Lewis et al., 2007).
The authors are very grateful to Dr Joan K. Lunney for critical reading and revision of this manuscript, to Dr John L. Williams and Dr G. Leone for helpful suggestions, and to C. Vanmaercke and L. Sys for technical assistance in the laboratory. This project was supported by a grant from the Italian Ministry of Research (MIUR project, art.10 D.M. 593/00). S.G. is partially supported by the European Network of Excellence EADGENE (www.eadgene.org). P.L.D. is supported by a grant from the Special Research Fund of Ghent University (grant B/06524). The authors declare no competing financial interests.
The genes and primers used for real-time PCR analysis are available with the online version of this paper.