Genomic diversity and differentiation of a managed island wild boar population.

The evolution of island populations in natural systems is driven by local adaptation and genetic drift. However, evolutionary pathways may be altered by humans in several ways. The wild boar (WB) (Sus scrofa) is an iconic game species occurring in several islands, where it has been strongly managed since prehistoric times. We examined genomic diversity at 49 803 single-nucleotide polymorphisms in 99 Sardinian WBs and compared them with 196 wild specimens from mainland Europe and 105 domestic pigs (DP; 11 breeds). High levels of genetic variation were observed in Sardinia (80.9% of the total number of polymorphisms), which can be only in part associated to recent genetic introgression. Both Principal Component Analysis and Bayesian clustering approach revealed that the Sardinian WB population is highly differentiated from the other European populations (FST=0.126-0.138), and from DP (FST=0.169). Such evidences were mostly unaffected by an uneven sample size, although clustering results in reference populations changed when the number of individuals was standardized. Runs of homozygosity (ROHs) pattern and distribution in Sardinian WB are consistent with a past expansion following a bottleneck (small ROHs) and recent population substructuring (highly homozygous individuals). The observed effect of a non-random selection of Sardinian individuals on diversity, FST and ROH estimates, stressed the importance of sampling design in the study of structured or introgressed populations. Our results support the heterogeneity and distinctiveness of the Sardinian population and prompt further investigations on its origins and conservation status.

Porcine colonization of the Americas: a 60k SNP story.

The pig, Sus scrofa, is a foreign species to the American continent. Although pigs originally introduced in the Americas should be related to those from the Iberian Peninsula and Canary islands, the phylogeny of current creole pigs that now populate the continent is likely to be very complex. Because of the extreme climates that America harbors, these populations also provide a unique example of a fast evolutionary phenomenon of adaptation. Here, we provide a genome wide study of these issues by genotyping, with a 60k SNP chip, 206 village pigs sampled across 14 countries and 183 pigs from outgroup breeds that are potential founders of the American populations, including wild boar, Iberian, international and Chinese breeds. Results show that American village pigs are primarily of European ancestry, although the observed genetic landscape is that of a complex conglomerate. There was no correlation between genetic and geographical distances, neither continent wide nor when analyzing specific areas. Most populations showed a clear admixed structure where the Iberian pig was not necessarily the main component, illustrating how international breeds, but also Chinese pigs, have contributed to extant genetic composition of American village pigs. We also observe that many genes related to the cardiovascular system show an increased differentiation between altiplano and genetically related pigs living near sea level.

Comprehensive and high-resolution typing of swine leukocyte antigen DQA from genomic DNA and determination of 25 new SLA class II haplotypes.

We previously reported the development of genomic-DNA-based high-resolution genotyping methods for SLA-DQB1 and DRB1. Here, we report the successful typing of SLA-DQA using similar methodological principles. We designed a method for comprehensive genotyping of SLA-DQA using intronic sequence information of SLA-DQA exon 2 that we had obtained from 12 animals with different SLA-DQB1 genotypes. We expanded our typing to 76 selected animals with diverse DQB1 and DRB1 genotypes, 140 random animals from 7 pig breeds, and 3 wild boars. This resulted in the identification of 17 DQA alleles with 49 genotypes. Two new alleles were identified from wild boars. Combine with SLA-DQB1, and DRB1 typing results, we identified 34 SLA class II haplotypes including 25 that were previously unreported.

Identification of high utility SNPs for population assignment and traceability purposes in the pig using high-throughput sequencing.

The objectives of this study were to develop breed-specific single nucleotide polymorphisms (SNPs) in five pig breeds sequenced with Illumina's Genome Analyzer and to investigate their usefulness for breed assignment purposes. DNA pools were prepared for Duroc, Landrace, Large White, Pietrain and Wild Boar. The total number of animals used for sequencing was 153. SNP discovery was performed by aligning the filtered reads against Build 7 of the pig genome. A total of 313,964 high confidence SNPs were identified and analysed for the presence of breed-specific SNPs (defined in this context as SNPs for which one of the alleles was detected in only one breed). There were 29,146 putative breed-specific SNPs identified, of which 4441 were included in the PorcineSNP60 beadchip. Upon re-examining the genotypes obtained using the beadchip, 193 SNPs were confirmed as being breed specific. These 193 SNPs were subsequently used to assign an additional 490 individuals from the same breeds, using the sequenced individuals as reference populations. In total, four breed assignment tests were performed. Results showed that for all methods tested 99% of the animals were correctly assigned, with an average probability of assignment of at least 99.2%, indicating the high utility of breed-specific markers for breed assignment and traceability. This study provides a blueprint for the way next-generation sequencing technologies can be used for the identification of breed-specific SNPs, as well as evidence that these SNPs may be a powerful tool for breed assignment and traceability of animal products to their breeds of origin.

A high-resolution comparative map of porcine chromosome 4 (SSC4).

We used the IMNpRH2(12,000-rad) RH and IMpRH(7,000-rad) panels to integrate 2019 transcriptome (RNA-seq)-generated contigs with markers from the porcine genetic and radiation hybrid (RH) maps and bacterial artificial chromosome finger-printed contigs, into 1) parallel framework maps (LOD ≥ 10) on both panels for swine chromosome (SSC) 4, and 2) a high-resolution comparative map of SSC4, thus and human chromosomes (HSA) 1 and 8. A total of 573 loci were anchored and ordered on SSC4 closing gaps identified in the porcine sequence assembly Sscrofa9. Alignment of the SSC4 RH with the genetic map identified five microsatellites incorrectly mapped around the centromeric region in the genetic map. Further alignment of the RH and comparative maps with the genome sequence identified four additional regions of discrepancy that are also suggestive of errors in assembly, three of which were resolved through conserved synteny with blocks on HSA1 and HSA8.

Nomenclature for factors of the SLA system, update 2008.

This report summarizes the new swine leukocyte antigen (SLA) allele sequences and haplotypes designated by the SLA Nomenclature Committee of the International Society for Animal Genetics. There have been 74 new SLA alleles, comprising 18 SLA-1 alleles, 11 SLA-2 alleles, six SLA-3 alleles, two SLA-6 alleles, one SLA-DRA allele, 20 SLA-DRB1 alleles, three SLA-DQA alleles and 13 SLA-DQB1 alleles. Twelve new SLA class I and four new class II haplotypes have also been designated. This is the first official update since the 2005 reports on the nomenclature for factors of the SLA class I and II systems. This report also summarizes recent updates to the Immunopolymorphism Database-Major Histocompatibility Complex (IPD-MHC) website (http://www.ebi.ac.uk/ipd/mhc/sla/). All information has now been integrated to the SLA section of the IPD-MHC database, which serves as the repository for maintaining a list of all recognized SLA genes and their allelic sequences.

DNA-based animal models of human disease: from genotype to phenotype.

Biomedical research utilizes animal models to elucidate human disease processes at the cellular and molecular level and for the development of new therapies. Traditionally, mammalian models have been limited to the mouse, primarily because of well characterized genetic lines and the ability to manipulate the genome to directly test hypotheses regarding causal mutations and disease phenotypes. The emerging availability of genome sequences of other mammals (bovine, canine, equine, feline, and porcine) now permits utilization of the mammal in which the phenotype best approximates the human condition. Equally important is the use of somatic cell nuclear cloning (SCNT) coupled with targeted germline manipulation to create animals to resolve the molecular mechanisms of the disease state. Our efforts have focused on the pig, which has emerged as an important biomedical mammalian model due to its closer physiology to humans. The utility of porcine genetically-defined tumour, cardiovascular and neurological disease models is described.

High-resolution comprehensive radiation hybrid maps of the porcine chromosomes 2p and 9p compared with the human chromosome 11.

We are constructing high-resolution, chromosomal 'test' maps for the entire pig genome using a 12,000-rad WG-RH panel (IMNpRH2(12,000-rad))to provide a scaffold for the rapid assembly of the porcine genome sequence. Here we present an initial, comparative map of human chromosome (HSA) 11 with pig chromosomes (SSC) 2p and 9p. Two sets of RH mapping vectors were used to construct the RH framework (FW) maps for SSC2p and SSC9p. One set of 590 markers, including 131 microsatellites (MSs), 364 genes/ESTs, and 95 BAC end sequences (BESs) was typed on the IMNpRH2(12,000-rad) panel. A second set of 271 markers (28 MSs, 138 genes/ESTs, and 105 BESs) was typed on the IMpRH(7,000-rad) panel. The two data sets were merged into a single data-set of 655 markers of which 206 markers were typed on both panels. Two large linkage groups of 72 and 194 markers were assigned to SSC2p, and two linkage groups of 84 and 168 markers to SSC9p at a two-point LOD score of 10. A total of 126 and 114 FW markers were ordered with a likelihood ratio of 1000:1 to the SSC2p and SSC9p RH(12,000-rad) FW maps, respectively, with an accumulated map distance of 4046.5 cR(12,000 )and 1355.2 cR(7,000 )for SSC2p, and 4244.1 cR(12,000) and 1802.5 cR(7,000) for SSC9p. The kb/cR ratio in the IMNpRH2(12,000-rad) FW maps was 15.8 for SSC2p, and 15.4 for SSC9p, while the ratio in the IMpRH(7,000-rad) FW maps was 47.1 and 36.3, respectively, or an approximately 3.0-fold increase in map resolution in the IMNpRH(12,000-rad) panel over the IMpRH(7,000-rad) panel. The integrated IMNpRH(12,000-rad) andIMpRH(7,000-rad) maps as well as the genetic and BAC FPC maps provide an inclusive comparative map between SSC2p, SSC9p and HSA11 to close potential gaps between contigs prior to sequencing, and to identify regions where potential problems may arise in sequence assembly.

Genetic induction of tumorigenesis in swine.

The transition from basic to clinical cancer research for a number of experimental therapeutics is hampered by the lack of a genetically malleable, large animal model. To this end, we genetically engineered primary porcine cells to be tumorigenic by expression of proteins known to perturb pathways commonly corrupted in human cancer. Akin to human cells, these porcine cells were quite resistant to transformation, requiring multiple genetic changes. Moreover, the transformed porcine cells produced tumors when returned to the isogenic host animal. The ability to now rapidly and reproducibly genetically induce tumors of sizes similar to those treated clinically in a large mammal similar to humans in many respects will provide a robust cancer model for preclinical studies dependent on generating large tumors.

Diet affects nutrient digestibility, hematology, and serum chemistry of senior and weanling dogs.

The objective of this experiment was to determine the effects of age and diet on serum chemistry, hematology, and nutrient digestibility in healthy dogs. Twelve senior (11 yr old; six males and six females) and 12 weanling (age = 8 wk old; six males and six females) beagles were randomly assigned to one of two dietary treatments: 1) an animal product-based (APB) diet or 2) a plant product-based (PPB) diet. The APB diet was primarily composed of brewer's rice, chicken by-product meal, and poultry fat, whereas the primary ingredients of the PPB diet included corn, soybean meal, wheat middlings, and meat and bone meal. Dogs remained on experiment for 12 mo. A 4-d total fecal collection was performed to determine apparent macronutrient digestibilities after 3 and 10 mo. Blood samples were collected at baseline and after 3, 6, 9, and 12 mo on study. After 3 mo, dogs fed the APB diet had greater (P < 0.001) DM (6 percentage units) and OM (7 percentage units) digestibilities than dogs fed the PPB diet. Senior dogs had greater DM (2.5 percentage units; P = 0.07) and OM (3 percentage units; P < 0.01) digestibilities than young dogs. Dogs fed the PPB diet had a lower (P < 0.001) fecal DM percentage (7.5 percentage units) and greater (P < 0.001) fecal output (253 vs. 97 g/d, as-is basis). After 10 mo, age did not affect nutrient digestibility or fecal characteristics. However, the effect of diet after 10 mo was similar to that observed after 3 mo, as dogs fed the PPB diet had a lower (P < 0.001) fecal DM percentage (7 percentage units), lower OM (4 percentage units; P = 0.09) and fat (6 percentage units; P < 0.001) digestibilities, and greater (P < 0.005) fecal output (235 vs. 108 g/d, as-is basis). At baseline, most serum metabolites were different between age groups, with weanlings having several metabolite concentrations outside the reference ranges for adult dogs. Blood cholesterol, red blood cells, hemoglobin, hematocrit, creatinine, total protein, albumin, bilirubin, sodium, chloride, and alanine transaminase were present in greater (P < 0.05) concentrations in senior dogs, but weanling dogs had greater (P < 0.05) concentrations of glucose, platelets, Ca, P, K, and alkaline phosphatase. Over time, blood cholesterol concentrations were affected by age (P < 0.05) and diet (P < 0.01). Senior dogs had greater (P < 0.05) cholesterol concentrations than weanling dogs. Moreover, dogs fed the APB diet had greater (P < 0.05) cholesterol concentrations than dogs fed the PPB diet. Overall, although serum metabolite concentrations of weanlings were different from senior dogs at baseline, as weanlings matured into young adults, metabolite concentrations were similar to those of senior dogs. Diet had the largest effects on nutrient digestibilities and fecal characteristics. Canine age and diet must be considered when interpreting experimental and clinical data.

Generation and exploration of a dense genetic map in a region of a QTL affecting corpora lutea in a Meishan x Yorkshire cross.

Previously genomic scans revealed quantitative trait loci (QTL) on porcine Chromosome 8 (SSC8) as significantly affecting the number of corpora lutea (CL) in swine. In one study, statistical evidence for the putative QTL was found in the chromosomal region defined by the microsatellites (MS) SW205, SW444, SW206, and SW29. A Yeast Artificial Chromosome library was screened by using the corresponding primers for clones containing these MS by PCR. From five positive YAC clones, 10 additional MS were isolated and mapped to SSC8 with the INRA-University of Minnesota porcine Radiation Hybrid (IMpRH) panel. The genetic map position of the QTL has been refined by addition of these 10 markers. The QTL evaluation included pedigrees of F2-intercross Meishan x Yorkshire design, with phenotypic data of 108 F2 female offspring and genotypic data for 29 MS markers on SSC8. The analysis was performed by using the least squares regression method. The calculated QTL effect for CL obtained by the multilocus least squares method showed a maximum test statistic (F value = 13.98) at position 99 cM between three MS derived from YACs containing SW205 and SW1843 spanning an interval of 7.1 cM. The point-wise (nominal) P-value was 5.21 x 10-6 corresponding to a genome-wide P-value of 0.009. The additive QTL effect explained 17.4% of the phenotypic variance.

Directed isolation and mapping of microsatellites from swine Chromosome 1q telomeric region through microdissection and RH mapping.

Several quantitative trait loci (QTLs) (vertebrate number, birth weight, age at puberty, growth rate, gestation length, and backfat depth) have been independently mapped to the distal region of swine Chromosome (SSC) 1q in several resource populations. In order to improve the map resolution and refine these QTLs more precisely on SSC1q, we have isolated and mapped additional microsatellites (ms), using chromosome microdissection and radiation hybrid (RH) mapping. Five copies of the telomeric region of SSC1q were microdissected from metaphase spreads and pooled. The chromosomal fragment DNA was randomly amplified by using degenerate oligonucleotide primed polymerase chain reaction (DOP-PCR), enriched for ms, and subcloned into a PCR vector. Screening of subsequent clones with ms probes identified 23 unique ms sequences. Fifteen of these (65%) were subjected to radiation hybrid (RH) mapping by using the INRA-University of Minnesota porcine RH panel (IMpRH); and the remaining eight were not suited for the RH mapping. Twelve microsatellites were assigned to SSC1q telomeric region of IMpRH map (LOD >6), and three remain unlinked (LOD <6). Out of the 15 microsatellite markers, 9 were polymorphic in NIAI reference population based on the Meishan and Göttingen miniature pig. In summary, we have used microdissection and radiation hybrid mapping to clone and map 12 new microsatellites to the swine gene map to increase the resolution of SSC1q in the region of known QTLs.

Interval mapping of carcass and meat quality traits in a divergent swine cross.

An autosomal scan of the swine genome with 119 polymorphic microsatellite (ms) markers and data from 116 F2 barrows of the University of Illinois Meishan x Yorkshire Swine Resource Families identified genomic regions with effects on variance in carcass composition and meat quality at nominal significance (p-value <0.05). Marker intervals on chromosomes 1, 6, 7, 8 and 12 (SSC1, SSC6, SSC7, SSC8, SSC12) with phenotypic effects on carcass length, 10th rib backfat thickness, average backfat thickness, leaf fat, loin eye area and intramuscular fat content confirm QTL effects identified previously based on genome wide significance (p-value <0.05). Several marker intervals included nominally significant (p-value <0.05) dominance effects on leaf fat, 10th rib backfat thickness, loin eye area, muscle pH and intramuscular fat content.

Altered hepatic mRNA expression of apoptotic genes during dimethylnitrosamine exposure.

The role of TNFalpha in regulating apoptotic signaling was investigated during subacute, low-dose (5.0 mg/kg) dimethylnitrosamine (DMN)-induced hepatotoxicity. In TNFalpha receptor (TNFR) intact (wild-type, WT) mice following 4 and 7 DMN exposures, hepatic transcripts for TNFalpha and TNFR-1 were elevated as compared to vehicle controls. DMN hepatotoxicity in WT and TNFR-1/TNFR-2 double knockout (DKO) mice were then compared over a 7-d exposure period. Liver RNA was isolated to measure hepatic expression of TNFalpha/Fas-related genes and the Bcl-2 family of genes that impact apoptosis. Hepatic mRNA levels for Fas, the apoptosis-promoting gene Bax, and the anti-apoptotic gene, Bcl-X(L), were up regulated following 4 and 7 DMN exposures in both WT and TNFR DKO mice as compared to vehicle controls. Notably, hepatic transcript levels for Bax were higher in TNFR DKO mice treated with DMN compared to identically treated WT mice. However, we detected approximately equal DMN-induced apoptotic degradation of liver DNA following 1, 4, and 7 exposures in WT and TNFR DKO mice. Taken together, these data show DMN-induced hepatic TNFalpha expression and suggest that TNFR-1 signaling may be up regulated following 4 and 7 daily DMN exposures. However, TNFalpha is not required for apoptotic signaling at the mRNA transcript level within the liver and instead may actually decrease Bax production.

An RNA helicase, RHIV -1, induced by porcine reproductive and respiratory syndrome virus (PRRSV) is mapped on porcine chromosome 10q13.

The impact of porcine reproductive and respiratory syndrome virus (PRRSV) infection on porcine alveolar macrophages (Mo) was examined by differential display reverse transcription PCR (DDRT-PCR). A PRRSV-induced expressed gene tag (EST) was used to isolate and identify a single cDNA clone from a library prepared from porcine peripheral blood. Rapid amplification of cDNA ends (RACE) was employed to clone a 1.5 kb fragment at the 5' end of the mRNA. DNA sequencing identified an open reading frame (ORF) of 2820 bp. Deduced amino acid sequence revealed the eight conserved domains characteristic of the DEAD/H box protein superfamily. The putative porcine RNA helicase induced by virus (RHIV -1) showed 84% amino acid similarity to human retinoic acid-induced gene (RIG-I). Porcine RHIV -1 transcripts were ubiquitously expressed in various pig tissues, while in PRRSV-infected pigs, higher expression was observed in several tissues persistent for PRRSV. These data indicate the association of PRRSV genome replication with enhaced host cell RNA helicase gene expression. Finally, the RHIV -1 gene was localized on porcine chromosome 10q13 between markers SSC25A02 and SWR334 via somatic cell panel and radiation hybrid (RH) mapping strategies.

Acute hepatotoxicant exposure induces TNFR-mediated hepatic injury and cytokine/apoptotic gene expression.

Tumor necrosis factor receptor knockout (TNFR KO) mice were used to examine the role of tumor necrosis factor-alpha (TNFalpha) signaling during acute hepatotoxicant exposure. Mice were exposed intraperitoneally (ip) to either vehicle, phosphate-buffered saline (PBS), or dimethylnitrosamine (DMN, 100 mg/kg) for 24 h. Histological evaluation showed that DMN-treated TNFR-2 KO mice had increased liver damage compared to wild type (WT), TNFR-1 KO, or TNFR double KO (DKO) mice. Also, 3 of 8 TNFR-2 KO mice died following DMN treatment, suggesting that hepatic TNFR-2 signaling produces protective responses that counteract TNFR-1-mediated damage. DMN-induced cellular infiltration was absent in TNFR-1-deficient mice, indicating that infiltrating cells do not exacerbate acute hepatotoxic events. In separate experiments, mice were exposed ip to either DMN (5.0 or 100 mg/kg), carbon tetrachloride (CCl4, 0.3 or 1.0 ml/kg), or corresponding PBS/corn oil controls for 6 or 24 h to compare the hepatic mRNA expression of cytokine- and apoptotic-associated genes. Following 24 h of DMN (100 mg/kg) or 6-24 h of CCl4 treatment, hepatic transcripts for TNFalpha, interferon (IFN)-gamma, IL (interleukin)-1RI, and transforming growth factor (TGF)-betaRII were induced. Hepatotoxicant-treated WT and TNFR DKO mice induced liver transcripts for the pro- and anti-apoptotic genes, Bax and Bcl-X(L), respectively, indicating TNF-independent gene activation. The anti-apoptotic gene, Bfl-1, was highly expressed in CCl4-treated, TNFR-positive strains, but minimally expressed in TNFR DKO mice, suggesting that hepatic Bfl-1 is TNF-regulated. Taken together, these data show that acute hepatotoxicant exposure is followed by upregulation of liver cytokine, cytokine receptor, and apoptotic transcripts, and that TNFalpha regulates various aspects of liver inflammation and injury in a TNFR-specific fashion.

Detection and parameter estimation for quantitative trait loci using regression models and multiple markers.

A strategy of multi-step minimal conditional regression analysis has been developed to determine the existence of statistical testing and parameter estimation for a quantitative trait locus (QTL) that are unaffected by linked QTLs. The estimation of marker-QTL recombination frequency needs to consider only three cases: 1) the chromosome has only one QTL, 2) one side of the target QTL has one or more QTLs, and 3) either side of the target QTL has one or more QTLs. Analytical formula was derived to estimate marker-QTL recombination frequency for each of the three cases. The formula involves two flanking markers for case 1), two flanking markers plus a conditional marker for case 2), and two flanking markers plus two conditional markers for case 3). Each QTL variance and effect, and the total QTL variance were also estimated using analytical formulae. Simulation data show that the formulae for estimating marker-QTL recombination frequency could be a useful statistical tool for fine QTL mapping. With 1,000 observations, a QTL could be mapped to a narrow chromosome region of 1.5 cM if no linked QTL is present, and to a 2.8 cM chromosome region if either side of the target QTL has at least one linked QTL.

Microsatellite markers from a microdissected swine chromosome 6 genomic library.

To develop additional microsatellite (MS) markers in the region of the porcine skeletal muscle ryanodine receptor gene (RYR1), a microdissected genomic library was generated from the proximal half of the q arm of swine chromosome 6. Purified DNA was restriction enzyme-digested, ligated to oligonucleotide adaptors and amplified by PCR using primers complementary to the adaptor sequences. The purity of the amplified products and boundaries of the microdissected chromosomal region were verified by fluorescence in situ hybridization. (CA)n-containing sequences were then identified in a small insert genomic library generated from the PCR-amplified microdissected DNA. Oligonucleotide primers were developed for the PCR amplification of 30 of the 46 (CA)n repeat-containing clones, which were subsequently used to amplify DNA isolated from unrelated pigs of different breeds to determine the informativeness of these MS markers. Twenty-two of these MS markers were genotyped on the University of Illinois Yorkshire x Meishan swine reference population. These 22 markers were all assigned within a 50.7-CM region of the swine chromosome 6 linkage map, indicating the specificity of the microdissected library.

Molecular responses of macrophages to porcine reproductive and respiratory syndrome virus infection.

The detailed mechanism(s) by which porcine reproductive and respiratory syndrome virus (PRRSV) impairs alveolar Mo homeostasis and function remains to be elucidated. We used differential display reverse-transcription PCR (DDRT-PCR) to identify molecular genetic changes within PRRSV-infected Mo over a 24 h post infection period. From over 4000 DDRT-PCR amplicons examined, 19 porcine-derived DDRT-PCR products induced by PRRSV were identified and cloned. Northern blot analysis confirmed that four gene transcripts were induced during PRRSV infection. PRRSV attachment and penetration alone did not induce these gene transcripts. DNA sequence revealed that one PRRSV-induced expressed sequence tag (EST) encoded porcine Mx1, while the remaining 3 clones represented novel ESTs. A full-length cDNA clone for EST G3V16 was obtained from a porcine blood cDNA library. Sequence data suggests that it encodes an ubiquitin-specific protease (UBP) that regulates protein trafficking and degradation. In pigs infected in vivo, upregulated transcript levels were observed for Mx1 and Ubp in lung and tonsils, and for Mx1 in tracheobronchial lymph node (TBLN). These tissues correspond to sites for PRRSV persistence, suggesting that the Mx1 and Ubp genes may play important roles in clinical disease during PRRSV infection.