Molecular Characterization of Caveolin-1 in Pigs Infected with Haemophilus parasuis

Caveolae are invaginations of the plasma membrane that are 50–100 nm in diameter. Originally identified by Palade (1) and Yamada (2) >50 y ago, caveolae are important organelles in mammalian cells. Caveolins are the major components and protein markers of caveolae (3). Caveolin-1 is the most important gene, both structurally and functionally, in the caveolin gene family, which includes three members in vertebrates: caveolin-1, caveolin-2, and caveolin-3 (4). Caveolin-3 is specific to the muscle and was first identified in the pig as a candidate gene of meat production traits (5). Caveolin-1 and caveolae have been shown to participate in various biological processes (6, 7), such as endocytosis, membrane trafficking, lipid recycling, cholesterol uptake, and various signaling pathways.

The role of caveolin-1 in vertebrate immunity has recently begun to be uncovered. Caveolin-1 has been reported to exist in many immune cell types, including macrophages and mast cells in the mouse and rat, bovine and murine lymphocytes, murine neutrophils, and human and bovine dendritic cells (8). In humans, caveolin-1 is thought to be a tumor suppressor gene (9–12). Caveolin-1 knockout mice exhibit defects in innate immunity, such as in the inflammatory response during infection with Salmonella enterica serovar Typhimurium (13) and macrophage phagocytosis (14). In particular, caveolin-1 plays a pivotal role in regulating the inflammatory responses of neutrophils during lung injury (15, 16). Furthermore, caveolin-1 has been implicated in pathogen–caveolae interactions, which lead to invasion of pathogenic microbes into host cells via caveolae-dependent endocytosis (17, 18). These findings indicate that caveolin-1 has a crucial role in innate immunity.

The pork industry suffers from outbreaks of serious diseases caused by many infectious pathogens, including Salmonella enterica (19), Haemophilus parasuis (20), porcine reproductive and respiratory syndrome virus (PRRSV) (21), porcine circovirus 2 (22), classical swine fever virus (CSFV) (23), and pseudorabies virus (PRV) (24). Thus, identification of key candidate genes that improve disease resistance in pigs is imperative. Although numerous studies on caveolin-1 have been reported in vertebrate animals, little is known about porcine caveolin-1. Thus far, only a partial cDNA sequence of porcine caveolin-1 has been reported in muscle tissue (25). Moreover, the role of caveolin-1 in immune responses against pathogens in pigs has not been investigated. H. parasuis is a Gram-negative bacterium, and virulent strains of H. parasuis can cause Glasser’s disease, which is characterized by fibrinous polyserositis, arthritis, and meningitis (26, 27). H. parasuis has caused huge economic losses each year in the pig industry worldwide (20). However, the pathogenesis of H. parasuis infection remains largely unknown. A recent study demonstrated that caveolin-1 knockout mice were markedly susceptible to Salmonella enterica serovar Typhimurium (13). Whether caveolin-1 is also involved in the pathogenesis of Glasser’s disease in pigs is of great interest.

In this study, we characterized several genomic, genetic, and transcriptional parameters of the porcine caveolin-1 gene. In this study, we also provide the molecular characterization of the caveolin-1 gene in pigs under stimulation with LPS/polyinosinic-polycytidylic acid [poly (I:C)] as well as during infection with H. parasuis.

The official gene name and symbol in this study were obtained from the Gene Ontology (http://www.geneontology.org/). A Basic Local Alignment Search Tool search in GenBank (http://www.ncbi.nlm.nih.gov/nuccore) using a predicted porcine caveolin-1 mRNA sequence (GenBank accession number NM_214438) (28) identified a number of porcine expressed sequence tags (ESTs) with >90% identity. These ESTs were assembled using SeqMan Pro 7.1 (DNASTAR) to obtain two contigs that differed at the 5′ end. The two cDNAs were obtained by assembling the two contigs until no further extension was observed (29). A primer set (Table I) that covered the coding region of the final cDNA contig was designed using Primer Premier 5.0 (PREMIER Biosoft). This primer pair was used to amplify a 667-bp fragment from the porcine cDNA. The PCR amplifications were performed in 10-μl volumes that contained 10 ng of mixed porcine cDNAs, 1× PCR Buffer (plus MgCl₂), 0.25 mM each 2′-deoxynucleoside 5′-triphosphate, 100 pmol each primer, and 0.5 U TaqE DNA polymerase (TaKaRa, Japan). The thermal cycle was 94°C for 3 min followed by 35 cycles of 94°C for 30 s, the annealing temperature (Table I) for 30 s, 72°C for 30 s, and a final extension step at 72°C for 5 min. The PCR product was visualized after electrophoresis on 2% agarose gels with 0.5 μg/ml ethidium bromide, purified using the Biotech purification kit (Biotech), and sequenced by a commercial biotechnology company (Aoke). Two primer pairs were designed to amplify exons 2 and 3 (Table I). The amplification system and protocol were performed as described above. The PCR product of exon 2 was sequenced directly, whereas the product of exon 3 was cloned into the pMD-18T vector (TaKaRa) prior to sequencing (GenBank accession number FJ392024).

Translation of the two cDNA sequences was predicted using ORF-Finder (http://www.ncbi.nlm.nih.gov/projects/gorf/). The deduced protein sequences were analyzed using tools from the ExPASy server (http://expasy.org/). Multiple alignments of amino acid sequences were carried out with DNAman 6.0 (Lynnon). The phylogenetic tree was inferred using the neighbor-joining method and replicated 1000 times in a bootstrap test using MEGA4 (30). Splice sites were identified by the Splign program (31). The conservative promoter and transcription factor binding sites were analyzed by ConSite (32) using ∼1 kb 5′ flanking sequence of the porcine, bovine, and human caveolin-1 genes, with the threshold score and energy set at 100 and −19 kcal/mol, respectively.

The Institut National de la Recherche Agronomique–University of Minnesota porcine radiation hybrid (IMpRH) panel, which consists of 118 hybrid clones (33), was employed to map the porcine caveolin-1 gene. A primer pair for chromosomal mapping that was specific to the pig was designed as shown in Table I. PCR reactions were performed and detected as described above, and the data were analyzed with the IMpRH mapping tool (34). Quantitative trait locus (QTL) analysis was performed by searching for QTLs and markers linked to the genetic position of porcine caveolin-1 in PigQTLdb (35–37). Comparative maps of the caveolin-1 gene and nearby regions of the genomes of pig, cow, human, rat, mouse, chicken, and zebrafish were produced using the source data from the corresponding genes in the National Center for Biotechnology Information-Gene database.

Three electronic single nucleotide polymorphisms (SNPs) were identified from the alignment of the ESTs from UniGene (Ssc.12842). A primer set was designed for verification of the electronic SNPs by PCR sequencing (Table I). PCR reactions were performed to amplify the sequence from three pig breeds, Landrace, Yorkshire, and Meishan (Chinese native pig), and all products were sequenced. After sequencing confirmation (Fig. 8), SNPs were detected using cleaved amplification polymorphisms (38). The PCR products were treated in the reaction mixture directly after amplification with the corresponding restriction enzyme for at least 12 h at 37°C (Table IV), according to the manufacturers’ instructions. The variability in codon usage caused by synonymous and nonsynonymous nucleotide variations in the coding region was assessed by referring to the Codon Usage Database (39).

The HZAU Landrace Immunological Resource Family consists of 355 pure Landrace pigs with 302 F1 offspring and 53 parents (17 sire, 36 dam). The pigs were immunized against CSFV (Qilu Animal Health) at 0 to 1 d and 35 d, PRRSV (DaHuaNong) at 16 d, mycoplasma (RespiSure Mycoplasma; Pfizer), and PRRSV (Ingelvac PRRS MLV; Boehringer Ingelheim) at 7 d and 21 d, respectively. Blood samples were collected by jugular venipuncture on days 1, 17, and 32 after birth and submitted immediately for hematological analysis (Kx-21n; Sysmex). Serum was harvested within 2 h and stored at −70°C for analysis of Ab levels. Ab levels were determined by competitive ELISA using commercial ELISA kits on a Thermo MK3 Microplate Photometer. Abs were detected against PRRSV (PRRS 2XR; IDEXX), CSFV (CSFV-Ab; IDEXX), and PRV (PRV-gpi; IDEXX). Immunological parameters analyzed included RBC and WBC counts, frequencies, and absolute numbers of lymphocytes, monocytes, and neutrophils, hemoglobin concentration, mean cell volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, hematocrit, red cell distribution width, platelet count, platelet distribution width, mean platelet volume, platelet/large cell ratio, and Ab levels against PRRSV, PRV, and CSFV. All animal studies in this paper were approved by the Biological Studies Animal Care and Use Committee of the Hubei Province, People’s Republic of China.

ANOVA was performed to estimate the effects of the caveolin-1 genotypes on each immunological parameter using the Macro statements and Mixed procedure in SAS 8.01 (SAS Institute, Cary, NC) with restricted maximum likelihood estimation methods implemented with a Newton–Raphson algorithm. To improve homoscedasticity, arcsine transformation was performed on the proportional data. The following mixed models were used to estimate the effects of the genotypes on the immunological parameters: Y_ijklmn = μ + Genotype_i + Sex_j + Parity_k + Environment_l + Sire_m + Dam_n (Sire_m) + ε_ijklmn, in which Y is the response vector for the immunological parameters above, and μ is the overall mean of each parameter. Genotype (i = 3: caused by 1 allelic variation), sex (j = 2), parity (k = 2: first versus nonfirst), and environment (L = 2: pigpen 1&2) were considered fixed effects, whereas sire (m = 17) and dam (n = 36) were random effects. In addition, ε represents the residual errors; for each immunological parameter, the least-square means between genotype classes at each time of blood sampling were computed using the lsdmeans statement and compared with an extremely conservative Bonferroni adjustment using pdiff, sderr, and adjust = bon options in SAS (SAS Institute).

Tissue distribution of the alternative transcripts of caveolin-1 was analyzed by SYBR Green Q-PCR. Primer pair “Alpha” (Table I) was specific for caveolin-1-α, whereas primer pair “Beta” (Table I) was specific for caveolin-1-β. In addition, primer set “Both” (Table I) was specific for the consensus region of the two transcripts. The β-actin gene was used as an internal control. Total RNA was extracted from 18 different tissues of two mature Meishan pigs and from two cell types: heart, liver, spleen, lung, kidney, muscle, stomach, intestine, brain, fat, skin, bladder, bone marrow, lymph node, testis, epididymis, uterus, and ovary and PBMCs and porcine kidney (PK) cell lines. Total RNA was isolated using the RNAprep Pure Tissue Kit (Tiangen, Beijing, China), and RNase-free DNase and was reverse-transcribed into cDNA using the TransScript First-Strand cDNA Synthesis Kit (TransGen, Beijing, China), according to the manual. Each 20-μl quantitative PCR (Q-PCR) reaction contained 2× SYBR Green PCR Mixture (ToYoBo), 500 mM each 2′-deoxynucleoside 5′-triphosphate, 0.4 mM primers (Table I), and 0.5 μl normalized template cDNA. The thermal cycle was 94°C for 3 min followed by 35 cycles at 94°C for 30 s, the annealing temperature (Table I) for 30 s, 72°C for 30 s, and a final fluorescence acquisition from 55°C to 95°C for melting curve analysis. The specificity of the Q-PCR was confirmed by melting curve analysis. All PCRs were performed in triplicate, and the cycle threshold values were determined using Gene Expression IQ5 software (Bio-Rad, Richmond, CA). Gene expression was quantified relative to the expression of the housekeeping gene by employing the 2^−ΔΔ cycle threshold method (40).

Blood from three clinically healthy 2-mo-old pigs from each breed was collected from the jugular vein into tubes containing sodium heparin. The breeds used were the Landrace, Yorkshire, Duroc, Meishan, and a three-way cross of Duroc × Landrace × Yorkshire (DLY). Porcine PBMCs were isolated using Lympholyte-H (HaoYang, TianJin, China) according to the manufacturer’s protocol. Total RNA was extracted from PBMCs and purified using the RNAprep Cell Kit and RNase-Free DNaseI (Tiangen). Purified RNA was reverse-transcribed with the TIAN Script Kit (Tiangen) according to the manufacturer’s instructions. The housekeeping gene β-actin was used as an internal control. The Q-PCR assays were performed and analyzed as described above, with primer pair “Both” (Table I). Overall comparison was made using ANOVA with OriginPro 8.07 (OriginLab Corporation). A p value <0.05 between control and treatment groups was considered statistically significant.

Ten groups (three replicates in each group, ∼1 × 10⁵ cells of PK-15 cells with ATCC number CCL-33; American Type Culture Collection) from the same cell-culture flask were grown in culture medium (DMEM) supplemented with 10% heat-inactivated FBS at 37°C with 5% CO₂. Adherent PK-15 cells were obtained by washing off the nonadherent cells with warm culture medium and PBS twice. Adherent cells were cultured further in DMEM (control samples) or treated with 1 μg/ml LPS (Escherichia coli 0127:B8; Sigma-Aldrich) or 10 μg/ml poly (I:C) (Sigma-Aldrich) for 0, 3, 6, 12, 24, or 48 h. The cells were harvested, and total RNA was extracted as described above. The housekeeping gene RPL32 was used as an internal control. The Q-PCR assays were performed with the primer “Both” (Table I) and analyzed as described above.

Eight 1-mo-old male piglets that were confirmed to be uninfected with H. parasuis were divided randomly into two equal groups. Four piglets were challenged intratracheally with 5 × 10⁸ CFU of a virulent strain of H. parasuis, 0165 (serotype 5), whereas the remaining four piglets were maintained without challenge as controls. One week after challenge, all of the piglets were killed, and their tissues were removed and placed immediately in liquid nitrogen. The Q-PCR assays were performed with the primer pair “Both” and internal control 18S (Table I) and analyzed as described above.

Tissues from one H. parasuis-negative pig and two H. parasuis-infected pigs were cut into 5-μm sections and stained with H&E for routine histological examination. Three sections of each tissue were used for immunohistochemistry using polyclonal primary Abs against caveolin-1 (Cell Signaling Technology) and the immunohistochemistry protocol supplied with the commercial Ab. For each tissue, a negative control was done without caveolin-1 Ab. Three sections of each tissue were used for immunoassays. Immunohistochemical images were captured using a Nikon ECLIPSE 80i inverted microscope with a 40× objective lens and NIS-elements D3.00 image software (Nikon). The images were quantitatively analyzed using Image Pro-Plus v. 6.0 (Media Cybernetics). Density means are expressed as mean ± SE:

The Student t test was used to compare data from two groups. The level of significance was set at p < 0.05.

Experimental and in silico cloning identified two isoforms of porcine caveolin-1, caveolin-1-α and caveolin-1-β. The caveolin-1-α transcript is 2494 bp and has an open reading frame of 537 nt flanked by a 60-nt 5′ untranslated region (UTR) and a 1897-nt 3′ UTR. The cDNA of caveolin-1-α encodes a 178-aa protein with a molecular mass of 20.6 kDa and a theoretical isoelectric point of 5.65. The sequence around the caveolin-1-α start codon (ATG) at position 61 shows a partial match with the Kozak consensus (RCCATGG) (41). The 3′ UTR contains two AATAAA polyadenylation signal consensus sequences (42), of which the last is 16–21 nt upstream of the poly(A) tail (Fig. 1A). The caveolin-1-β transcript is 2657 bp and has a 444-nt open reading frame flanked by a 325-nt 5′ UTR and a 1888-nt 3′ UTR. The caveolin-1-β isoform encodes a protein of 147 aa with a molecular mass of 17.2 kDa and a theoretical isoelectric point of 5.27. The 3′ UTR contains three polyadenylation signals, of which the last is 16–21 nt upstream of the poly(A) tail (Fig. 1B). The sequence around the caveolin-1-β start codon at position 326 shows a perfect match with the typical Kozak consensus.

Fig. 2 shows multiple alignments of the predicted porcine caveolin-1 protein with 20 known vertebrate caveolin-1 proteins based on known structural and functional domains (4, 7). All identified domains (underlined), including the N-terminal membrane attachment domain (N-MAD), the caveolin signature, caveolin scaffolding domain (CSD), WW binding domain, C-terminal MAD, and three conserved palmitoyl group cysteine residues, are highly conserved among mammalian, avian, bony fish, and amphibian species. In particular, the N-MAD, caveolin signature, and CSD were almost completely conserved. Phosphorylation at Tyr¹⁴ and Ser⁸⁰ is essential for the function of caveolin-1, and these residues are also conserved in the pig. Although the caveolin-1 proteins are highly conserved, the porcine caveolin-1 molecule has several unique features, including 10 nonconserved aa regions, which are unshaded in Fig. 2, and an N-terminal region containing Tyr¹⁴ that is absent from porcine caveolin-1-β and that of Xenopus laevis (Fig. 2).

To define the molecular evolutionary history of porcine caveolin-1, protein sequences from 26 vertebrates were obtained to construct a phylogenetic tree (Fig. 3). Porcine caveolin-1 clusters with that of Chinese and Indian muntjac in the artiodactyl clade, with extremely high similarity and identity. The phylogeny of caveolin-1 shows a clear divergence among the vertebrates, including fish, amphibians, birds, and mammals. Although the bootstrap values in some nodes are not very high, the main structure of the tree is reasonable and supported by the known phylogeny of the species represented. In addition, the porcine caveolin-1 protein shows the most homology with that of the Indian deer and the lowest homology with that of the platypus (Fig. 3).

A porcine BAC clone (GenBank accession number AC087423) that contains the complete porcine caveolin-1 gene was identified by a Basic Local Alignment Search Tool search in the GenBank database. The porcine caveolin-1 gene spans ∼34.7 kb, which is similar to the length of other mammalian caveolin-1 genes (36.3 kb in humans, 35.5 kb in cows). The transcripts were aligned with their genomic sequences to confirm the genomic structure of porcine caveolin-1 (Fig. 4). As shown in Fig. 4, there are at least two splice patterns of porcine caveolin-1. Three exons (ex1, ex2, and ex3) are transcribed into caveolin-1-α with two introns spliced, whereas two exons (ex1` and ex2`) are transcribed into caveolin-1-β, with one intron spliced. Ex1 is a unique exon, ex2 is partially identical to ex1`, and ex3 is identical to ex2`. All splice sites (Table II) conform to the consensus GT–AG rule (43). Comparison of seven vertebrate caveolin-1-α genes from pig, cattle, rat, mouse, human, chicken, and zebrafish revealed that the gene structure of caveolin-1 is highly conserved among the pig and other mammals, especially in the coding exons (Table III). However, the caveolin-1 gene structures of the pig, fish, and birds are not completely conserved. Furthermore, all of the exons of the vertebrate caveolin-1 genes described above have a coding region, and all of the introns are in phase 0 (44). Along with the results of the multiple alignment and phylogenetic tree analyses, the results described above demonstrate that the gene cloned in this study is that of the caveolin-1 homolog in the pig.

The transcription factor binding sites of porcine caveolin-1 are largely unknown. Promoter analyses of the porcine caveolin-1 gene revealed some potential transcription factor binding sites (Table IV), including sites for NF-κB, c-REL, Sox5, Sox17, c-FOS, TEF-1, and E2F for porcine caveolin-1-α (Table IV) and sites for NF-κB, E2F, the Foxo family (Foxd1, FoxI1, and Foxd3), Sox5, Sox17, Myc-Max, and USF for porcine caveolin-1-β (Table IV). Comparison of porcine, bovine, and human caveolin-1 genes revealed that NF-κB, p50, c-REL, and E2F transcription factor binding sites are present in the caveolin-1-α genes of the three species (Table IV) and that Spz1, Sox-5, AML-1, HAND1, ARNT, USF, and n-MYC transcription factor binding sites are present and identical in the caveolin-1-β gene of the three species (Table IV). Some of the transcription factor binding sites are conserved in the two promoter regions of the alternative transcripts of caveolin-1. No conventional TATA-binding protein binding sites (TATA boxes) were found in the two promoter regions of the porcine, bovine, and human genes, suggesting that transcription of the caveolin-1 gene in these species does not require TATA boxes.

The porcine caveolin-1 gene is located on chromosome 18 and is closely linked to microsatellite marker SW1984 at 37 cR, with a logarithm of odds score of 10.52, and marker SW787 at 36 cR, with a logarithm of odds score of 9.33, respectively (Fig. 5). Marker SW1984 (UniSTS: 252678) has been mapped to SSC18 (45) and updated to SSC18q21 at 29.4 cM, according to MARC Map. Thus, the most probable chromosomal location for the porcine caveolin-1 gene is SSC18q21. This agrees with the comparative map because human caveolin-1 has been mapped to 7q31.1-q31.2 (46), which is a syntenic region of porcine chromosome 18 (47). Further comparative mapping of this genomic region in mammals, birds, and fish revealed a conserved genetic region containing TES, caveolin-2, caveolin-1, MET, and CAPZA2 genes (Fig. 6). Despite differences in intergenic distance, the linkage between these genes and the transcriptional direction are well conserved. Interestingly, the genes in avian species are transcribed in the reverse direction compared with those of mammals and fish. Using microsatellite marker SW1984, QTLs matching with pig caveolin-1 were found in PigQTLdb (37); these QTLs affect mostly traits relating to carcass and meat quality (48).

To identify genetic variation in porcine caveolin-1, the coding regions from three pig breeds were sequenced by a combination of in silico and experimental cloning (Fig. 7). Three SNPs—G387C, A474G, and G512A—were identified (Table V). Among these SNPs, G387C and A474G are synonymous mutations, whereas G512A is a nonsynonymous substitution that changes Arg¹⁷¹ (CGC) into His¹⁷¹ (CAC). This arginine (Arg¹⁷¹) is conserved in all mammalian caveolin-1 genes, suggesting that G512A is a candidate functional mutation in porcine caveolin-1. To further investigate the effects of the mutations on the porcine caveolin-1 protein, the variability of codon usage caused by these mutations was assessed (Fig. 8). The synonymous G387C is present in the codon of Ala¹²⁹ of the porcine caveolin-1-α protein. The codon usage frequencies of Ala (GCG) and Ala (GCC) are 0.89 and 3.17%, respectively (mutations underlined). Unlike GCC, GCG of Ala129 is a low-usage codon. Furthermore, the A474G synonymous mutation is present in the codon of Pro¹⁵⁸ of the porcine caveolin-1-α protein. The codon usage frequencies of Pro (CCA) and Pro (CCG) are 1.43 and 0.84%, respectively. Unlike CCA, CCG of Ala¹²⁹ is a low-usage codon. These findings indicate that the two synonymous mutations, G387C and A474G, can lead to low-usage codons in new translations of porcine caveolin-1 and also suggest that these mutations may influence the function of the porcine caveolin-1 proteins via changes in codon usage. In addition, the distribution of the variant A474G differs between local Chinese breeds and European breeds (Table VI).

To investigate the relationship between porcine caveolin-1 and immunity, we evaluated the effects of two SNPs on immunological parameters in a Landrace population by ANOVA. The G387C polymorphism segregated in the resource family and was consistent with Hardy–Weinberg equilibrium. The Shannon’s Information index of this site is 0.69, which suggests high diversity. Significant associations of G387C with five immunological parameters were found in the resource family (Tables VII, VIII). The A474G polymorphism segregated in the same resource family and was also consistent with Hardy–Weinberg equilibrium; the Shannon’s Information index of this site was 0.61, again suggesting high diversity. Three immunological parameters were found to be significantly associated with the porcine caveolin-1 A474G in the resource family (Tables VII, VIII).

Using transcript-specific and consensus region-specific primers (Table I), the expression of porcine caveolin-1 was detected by Q-PCR in 18 tissues and two cell types and was normalized to the expression of β-actin (Fig. 9). The total mRNA levels of both forms of caveolin-1 were higher than that of the alternative transcripts (α and β) in the 18 tissues and two cell types. This is compatible with the hypothesis (mRNA_Total ≥ mRNA_α + mRNA_β).

Interestingly, the abundance of the α transcript of mature porcine caveolin-1 was notably higher than that of the β transcript in all tissues and cell types examined. Furthermore, caveolin-1 was most highly expressed in the skin, muscle, heart, fat, lung, epididymis, uterus, and ovary, moderately expressed in the spleen, bone marrow, lymph node, liver, intestine, and testis, and lowly expressed in PBMCs, kidney, and brain.

To investigate differences in basal expression of porcine caveolin-1 in different pig breeds, we measured porcine caveolin-1 expression in PBMCs from Landrace, Yorkshire, Duroc, Meishan, and the three-way cross DLY (Fig. 10). Our results indicated that caveolin-1 was highly expressed in PBMCs of Yorkshire pigs, whereas PBMCs of Landrace and DLY pigs expressed relatively low levels of caveolin-1. However, there was no significant difference in the basal expression of caveolin-1 in PBMCs among all pig breeds examined.

The presence of consensus NF-κB binding sites in the promoters of the two transcripts suggests that NF-κB may activate caveolin-1 in pigs. To investigate this possibility, poly (I:C) and LPS were used to activate NF-κB in PK-15 cells. Overnight cultures of PK-15 cells were treated with 1 μg/ml LPS or 10 μg/ml poly (I:C) for 0, 3, 6, 12, 24, and 48 h. LPS stimulation induced expression of porcine caveolin-1 at 3 h and more significantly at 6 h, after which caveolin-1 expression dropped and plateaued for 12–24 h (Fig. 11). In contrast, poly (I:C) stimulation did not induce expression of porcine caveolin-1 until after 24 h, with expression peaking at 48 h (Fig. 12). These findings indicate that both LPS and poly (I:C) can induce the expression of porcine caveolin-1 in vitro.

Decreased expression of caveolin-1 was observed in the brains and lungs of pigs infected with H. parasuis for 1 wk (Fig. 13). To confirm these findings, immunohistochemical staining was performed on paraffin-embedded sections of porcine tissues. Quantitative analyses showed significantly decreased expression of caveolin-1 in the brain, spleen, heart, and kidney (Fig. 14). Positive staining was observed as brown reaction products, mostly in the cytoplasm and on cell membranes of control lung and brain, respectively (Fig. 15A, 15C). Porcine caveolin-1 was widely expressed, particularly in the bronchial epithelial cells surrounding air spaces (Fig. 15A) and brain capillary networks (Fig. 15C). However, caveolin-1 staining in infected porcine lung (Fig. 15B) and brain (Fig. 15D) tissues was much lower than in the control. These results indicate that caveolin-1 protein is downregulated in the brain and lung 1 wk postinfection with H. parasuis. Reduced expression of porcine caveolin-1 postinfection with H. parasuis was also observed in the heart; positive staining for caveolin-1 was detected along the vessels and in the epithelial cells of control heart specimens (Fig. 15E), but the intensity of the positive staining was reduced significantly in similar areas of infected hearts (Fig. 15F). In addition, reduced expression of porcine caveolin-1 postinfection with H. parasuis was observed in the spleen (Fig. 16A, 16B) and kidney (Fig. 16C, 16D) but not in the liver (Fig. 16E, 16F). In summary, the expression of caveolin-1 was reduced 1 wk after H. parasuis infection in the pig brain, heart, kidney, spleen, and lung.

In this study, we identified two alternative transcript variants that encode two isoforms of the porcine caveolin-1 protein, caveolin-1-α and -β. The cloned cDNA sequence of caveolin-1 matched perfectly with the genomic sequence. The porcine caveolin-1 gene exhibits at least two splicing patterns: a three exon/two intron organization for caveolin-1-α and a two exon/one intron organization for caveolin-1-β. The caveolin-1 gene has been cloned in humans (3, 49–52), mice (53), African clawed frog (54), zebrafish (55), and even in the invertebrate Caenorhabditis elegans (56). Studies of the human (57), mouse (53), and zebrafish proteins indicate that caveolin-1 has two isoforms that differ in their N-terminal protein sequences (an additional 31 aa is added to the α-isoform at its N terminus). Our findings show that the two isoforms are also present in the pig.

Two explanations have been proposed for the generation of the two isoforms of caveolin-1, but the molecular mechanisms are still unclear. Mutation analyses in humans suggest that these two isoforms are derived from two alternative start sites of a single transcript (57). However, studies in mice suggest that the two isoforms of caveolin-1 are generated by translation of distinct mRNAs, as a TATA box was found in the first intron, where the promoter of caveolin-1-β lies (53). In contrast, we did not find a TATA box in either of the promoter regions of the porcine caveolin-1 gene. In mammalian genes, the TATA box is located 25–30 bp upstream of the transcription start site and is important for directing accurate transcription initiation via specific interaction with TATA binding protein (58). Some TATA-less promoters appear to direct the initiation of transcription from multiple start sites (59). Therefore, the TATA-less promoters in the porcine caveolin-1 gene are likely responsible for the multiple transcription initiation sites. In addition, porcine caveolin-1-α has two start codons, Met1 and an internal Met32. However, the nucleotide sequence around Met1 (AGAATGT) shows only a partial match with the typical Kozak consensus, whereas Met32 (GCCATGG) shows a perfect match with the typical Kozak consensus. The Kozak consensus is a sequence (RCCATGG) that occurs in eukaryotic mRNA (41) that plays a major role in the process of translation initiation (60). Point mutations close to the AUG initiator codon affect the efficiency of translation (61). Given that the internal Kozak sequence is more typical than the first Kozak sequence in porcine caveolin-1-α, we inferred that the internal Met32 was highly efficient at starting translation; this is also supported by mutation analyses in humans (57). In summary, we hypothesize that the distinct transcript variants generated by the TATA-less promoter and the alternative start sites in caveolin-1-α translation are responsible for generating the two isoforms of porcine caveolin-1. The former may regulate the level of transcription, whereas the latter may regulate the level of translation.

The porcine caveolin-1 gene was initially mapped to SSC18q21, which is syntenic to human Hsc7q31, where the human caveolin-1 gene is located. The porcine caveolin-1 gene is closely linked to the genes TES, caveolin-2, MET, and CAPZA2. Interestingly, all of the genes in this region have been implicated in immunity and disease, such as tumorigenesis, in humans (46). As previously reported in humans (46), porcine caveolin-1 is closely linked to caveolin-2, which suggests that the two genes in the pig are parallel homologs and were likely generated by duplication (52). Therefore, we hypothesize that the genes in this region may be evolutionarily linked and may also be functionally related to immunity in the pig.

The porcine caveolin-1 gene is polymorphic and has three SNPs in its coding region. Rich polymorphisms in the coding region suggest that functional mutations may be present in the caveolin-1 gene in pigs and other higher mammals. Mice deficient in caveolin-1 have reduced serum levels of Abs (62), lower levels of leukocytes, and decreased platelet rolling (63). Our results also showed an association between caveolin-1 and blood Ab levels, lymphocytes, and platelet bioactivity. Synonymous mutations can influence the translation rate, protein folding, and thus biological function of proteins via changes in codon usage (64). We found that the two synonymous mutations of caveolin-1, G387C and A474G, resulted in low-usage codons in new translations of porcine caveolin-1, which suggests that these mutations may influence the function of porcine caveolin-1 proteins. Collectively, our findings indicate that the caveolin-1 gene is polymorphic and that its genetic variants are potentially functional. Given that the caveolin-1 gene is a potential diagnostic and prognostic marker in humans (65), these findings suggest that the caveolin-1 gene can be used to introduce disease resistance in pig breeding programs.

Porcine caveolin-1 was found to be expressed in almost all body systems, including the circulatory (heart, PBMCs), respiratory (lungs), immune (spleen, bone marrow, lymph node, PBMCs, muscles), musculoskeletal (muscles), urinary (bladder, kidneys), digestive (stomach, intestines), nervous (brain), and reproductive (epididymis, testis, uterus, ovary) systems. The human and murine caveolin-1 molecules also show a broad distribution of expression in tissues and cell types (52, 53). Caveolin-1 was moderately expressed in the porcine immune system. Interestingly, the caveolin-1-α transcript was more abundantly expressed than that of caveolin-1-β in a wide range of tissues, consistent with previous findings of caveolin-1 protein expression in different developmental stages of zebrafish (55). These findings indicate that porcine caveolin-1 is a key gene involved in various biological processes, including immunity, and as the main form of caveolin-1, the caveolin-1-α gene may play a predominant role. In addition, the basal expression levels of caveolin-1 in PBMCs from different pig breeds are relatively similar, suggesting that genetic variation has limited effects on basal expression of caveolin-1.

The putative transcription factors that regulate the porcine caveolin-1 gene are involved in a range of signaling pathways, especially the MAPK/ERK pathway. Most of the transcription factor binding sites are also present in the human and bovine genes, suggesting that the regulation of caveolin-1 expression is conserved among mammals. These observations are supported by previous reports showing that activation of protein kinase A downregulates caveolin-1 protein expression (66) and that cytokine production by caveolin-1 also involves the MAPK kinase 3/p38 MAPK pathway (67).

The transcriptional regulatory mechanisms governing the generation of the two alternative transcripts of porcine caveolin-1 may overlap, as some transcription factor binding sites, such as the NF-κB elements, are common in the two promoters. LPS and poly (I:C) can activate NF-κB, and we found that both LPS and poly (I:C) induced the expression of porcine caveolin-1 in PK-15 cells. These findings are consistent with a study showing that LPS treatment upregulated caveolin-1 expression in murine macrophages (68). In addition, LPS has been shown to induce the binding of NF-κB to intronic NF-κB consensus sites of caveolin-1 in microvascular endothelial cells of human lung tissue (69). LPS is a characteristic cell wall component of Gram-negative bacteria, whereas poly (I:C) is a double-stranded viral RNA mimic. Our results show that porcine caveolin-1 can be induced by both Gram-negative bacteria and double-stranded viral RNA, likely through activation of NF-κB. The two alternative transcripts of porcine caveolin-1 also have transcript-specific regulatory factors, which include TEF-1 and c-REL for porcine caveolin-1-α and the Foxo family (Foxd1, FoxI1, and Foxd3), Spz1, USF, and Myc for porcine caveolin-1-β. The FOXO (forkhead box O) protein can bind to the promoter of caveolin-1 and activate its transcription (70). In this article, we found that Foxd1 (forkhead box D1), FoxI1 (forkhead box I1), and Foxd3 (forkhead box D3) could bind to the promoter of caveolin-1-β but not that of caveolin-1-α. Therefore, the forkhead gene family may specifically regulate caveolin-1-β expression.

Glasser’s disease is characterized mainly by meningitis, fibrinous polyserositis, and arthritis (26, 27), all of which are associated with uncontrolled inflammation. Caveolin-1 is a key immunomodulatory molecule in the inflammatory response (13, 16, 67, 71), and decreased expression of caveolin-1 in murine alveolar and peritoneal macrophages has a proinflammatory outcome (67). Fibrinous polyserositis is inflammation of the serous tissues with fibrosis. Caveolin-1 has been reported to be a critical regulator of fibrosis, as knockout studies have shown that caveolin-1 participates in the pathogenesis of systemic sclerosis and idiopathic pulmonary fibrosis (72). Decreased expression of caveolin-1 plays a crucial role in the pathogenesis of tissue fibrosis (73). Therefore, we deduce that decreases in caveolin-1 expression may underlie fibrosis and inflammation in serous tissues during H. parasuis infection in pigs. Meningitis is inflammation of the meninges, and H. parasuis must cross the blood–brain barrier (BBB) to cause meningitis and neurologic disorders. Recent reports show that deficiency in caveolin-1 alters the normal function of the brain and the BBB. Knockdown of caveolin-1 significantly perturbs the normal function of the BBB in humans via regulation of chemokine MCP-1 in astrocytes (74). Caveolin-1 deficiency also increases cerebral ischemic injury and the volume of cerebral infarction, impairs angiogenesis, and increases apoptotic cell death (75). Caveolin-1 knockout mice display several neurologic disorders, such as muscle weakness, reduced activity, and gait abnormalities (76). These neurologic disorders are similar to some of the central nervous clinical signs of pigs infected with H. parasuis. These observations implicate the decrease in caveolin-1 expression during H. parasuis infection in the development of meningitis and other inflammatory processes.

Breeding for disease resistance is one of the most useful methods to improve animal health, welfare, and productivity (77). Although identification of mechanisms of disease resistance is challenging, an essential first step is to investigate the key genes involved in the host response to infectious pathogens (78, 79). In this study, we investigated caveolin-1 as a candidate gene for disease resistance in pigs. We characterized the gene structure and transcriptional regulation of the porcine caveolin-1 gene and provided the in vivo characterization of caveolin-1 in the pathogenesis of systemic H. parasuis infection in pigs. These findings not only provide new support for the use of the caveolin-1 gene as a potential diagnostic and genetic marker for disease resistance in animal breeding programs, but also implicate caveolin-1 in the pathogenesis of Glasser’s disease.

We thank Dr. Martine Yerle (Institut National de la Recherche Agronomique, France) for providing the RH panel. We also thank Prof. Shu-Jun Zhang, Dr. Xue-Ying Hu, Xiang-An Jing, Xi Su, Nu-Nu Sun, and Jian-Bo Cao for technical assistance.

The authors have no financial conflicts of interest.