Animl Science Pulictions Animl Science 2-2014 Vlidtion nd Further Chrcteriztion of Mjor Quntittive Trit Locus Associted with Host Response to Experimentl Infection with Porcine Reproductive nd Respirtory Syndrome Virus Nichols Jmes Boddicker Iow Stte University Dorin J. Grrick Iow Stte University, dorin@istte.edu Rymond Rowlnd Knss Stte University Jon K. Lunney United Sttes Deprtment of Agriculture Follow this nd dditionl works t: http://li.dr.istte.edu/ns_pus Jmes Prt M. of Reecy the Agriculture Commons, Animl Sciences Commons, Genetics Commons, nd the Iow Veterinry Stte University, Infectious jreecy@istte.edu Diseses Commons See next pge for dditionl uthors The complete iliogrphic informtion for this item cn e found t http://li.dr.istte.edu/ ns_pus/150. For informtion on how to cite this item, plese visit http://li.dr.istte.edu/ howtocite.html. This Article is rought to you for free nd open ccess y the Animl Science t Iow Stte University Digitl Repository. It hs een ccepted for inclusion in Animl Science Pulictions y n uthorized dministrtor of Iow Stte University Digitl Repository. For more informtion, plese contct digirep@istte.edu.
Vlidtion nd Further Chrcteriztion of Mjor Quntittive Trit Locus Associted with Host Response to Experimentl Infection with Porcine Reproductive nd Respirtory Syndrome Virus Astrct Infectious diseses re costly to the swine industry; porcine reproductive nd respirtory syndrome (PRRS) is the most devstting. In erlier work, quntittive trit locus ssocited with resistnce/susceptiility to PRRS virus ws identified on Sus scrof chromosome 4 using pproximtely 560 experimentlly infected nimls from commercil cross. The fvorle genotype ws ssocited with decresed virus lod nd incresed weight gin (WG). The ojective here ws to vlidte nd further chrcterize the ssocition of the chromosome 4 region with PRRS resistnce using dt from two unrelted commercil crossred popultions. The vlidtion popultions consisted of two trils ech of pproximtely 200 pigs sourced from different reeding compnies tht were infected with PRRS virus nd followed for 42 dys post-infection. Across ll five trils, heritility estimtes were 0.39 nd 0.34 for virl lod (VL; re under the curve of logtrnsformed viremi from 0 to 21 dys post-infection) nd WG to 42 dys post-infection respectively. Effect estimtes of SNP WUR10000125 in the chromosome 4 region were in the sme directions nd of similr mgnitudes in the two new trils s hd een oserved in the first three trils. Across ll five trils, the 1-M region on chromosome 4 explined 15 percent of genetic vrince for VL nd 11 percent for WG. The effect of the fvorle minor llele t SNP WUR10000125 ws dominnt. Ordered genotypes for SNP WUR10000125 showed tht the effect ws present irrespective of whether the fvorle llele ws pternlly or mternlly inherited. These results demonstrte tht selection for host response to PRRS virus infection could reduce the economic impct of PRRS. Keywords genome wide ssocition, porcine reproductive nd respirtory syndrome, swine chromosome 4, vlidtion Disciplines Agriculture Animl Sciences Genetics Veterinry Infectious Diseses Comments This rticle is from Animl Genetics 4 5 (2014): 48, doi:10.1111/ge.12079. Rights Works produced y employees of the U.S. Government s prt of their officil duties re not copyrighted within the U.S. The content of this document is not copyrighted. Authors Nichols Jmes Boddicker, Dorin J. Grrick, Rymond Rowlnd, Jon K. Lunney, Jmes M. Reecy, nd Jck C. Dekkers This rticle is ville t Iow Stte University Digitl Repository: http://li.dr.istte.edu/ns_pus/150
Vlidtion nd further chrcteriztion of mjor quntittive trit locus ssocited with host response to experimentl infection with porcine reproductive nd respirtory syndrome virus N. J. Boddicker*, D. J. Grrick*, R. R. R. Rowlnd, J. K. Lunney, J. M. Reecy* nd J. C. M. Dekkers* doi: 10.1111/ge.12079 *Deprtment of Animl Science, Iow Stte University, Ames, IA, 50011, USA. Institute of Veterinry, Animl nd Biomedicl Sciences, Mssey University, Plmerston North 4474, New Zelnd. College of Veterinry Medicine, Knss Stte University, Mnhttn, KS, 66506, USA. United Stte Deprtment of Agriculture, Agriculturl Reserch Services, Beltsville Agriculturl Reserch Center, Beltsville, MD, 20705, USA. Summry Infectious diseses re costly to the swine industry; porcine reproductive nd respirtory syndrome (PRRS) is the most devstting. In erlier work, quntittive trit locus ssocited with resistnce/susceptiility to PRRS virus ws identified on Sus scrof chromosome 4 using pproximtely 560 experimentlly infected nimls from commercil cross. The fvorle genotype ws ssocited with decresed virus lod nd incresed weight gin (WG). The ojective here ws to vlidte nd further chrcterize the ssocition of the chromosome 4 region with PRRS resistnce using dt from two unrelted commercil crossred popultions. The vlidtion popultions consisted of two trils ech of pproximtely 200 pigs sourced from different reeding compnies tht were infected with PRRS virus nd followed for 42 dys post-infection. Across ll five trils, heritility estimtes were 0.39 nd 0.34 for virl lod (VL; re under the curve of logtrnsformed viremi from 0 to 21 dys post-infection) nd WG to 42 dys post-infection respectively. Effect estimtes of SNP WUR10000125 in the chromosome 4 region were in the sme directions nd of similr mgnitudes in the two new trils s hd een oserved in the first three trils. Across ll five trils, the 1-M region on chromosome 4 explined 15 percent of genetic vrince for VL nd 11 percent for WG. The effect of the fvorle minor llele t SNP WUR10000125 ws dominnt. Ordered genotypes for SNP WUR10000125 showed tht the effect ws present irrespective of whether the fvorle llele ws pternlly or mternlly inherited. These results demonstrte tht selection for host response to PRRS virus infection could reduce the economic impct of PRRS. Keywords genome wide ssocition, porcine reproductive nd respirtory syndrome, swine chromosome 4, vlidtion Introduction Porcine reproductive nd respirtory syndrome (PRRS) cuses reproductive filure in reeding femles nd respirtory prolems in growing nimls nd collectively costs the US swine industry pproximtely $664 million nnully, of which $365 million is ttriuted to losses mong growing pigs (Holtkmp et l. 2013). The PRRS virus Address for correspondence J. C. M. Dekkers, Deprtment of Animl Science, Iow Stte University, Ames, IA 50011, USA. E-mil: jdekkers@istte.edu Accepted for puliction 24 June 2013 (PRRSV) is complex, nd mny different strins circulte round the world (Fng et l. 2007). The PRRS Host Genetics Consortium ws estlished to ddress this importnt industry issue y investigting the genetic sis of host response to PRRSV infection in experimentlly infected commercil crossred pigs using PRRSV strin NVSL 97-7985 (Lunney et l. 2011; Rowlnd et l. 2012). Previous work, using dt from the first three infection trils of pproximtely 560 pigs, ll from the sme commercil F1 cross, identified mjor quntittive trit locus (QTL) for PRRS resistnce/susceptiility on Sus scrof chromosome (SSC) 4 (Boddicker et l. 2012). The identified 1-M region ws ssocited with virl lod (VL), mesure of viremi over 21 dys post-infection (dpi), nd WG to 42 dpi (WG42). Pigs with the fvorle SNP WUR10000125 48 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
Vlidtion of PRRS QTL 49 genotypes in the region hd reduced VL nd incresed WG compred with pigs with the unfvorle genotypes. This region explined 15 percent of the vrince in genomic estimted reeding vlues (GEBV) for VL nd 11 percent for WG42. The correltion etween the GEBV for VL nd WG42 for the 1-M region ws perfect t 1, likely reflecting single QTL in the region tht controls oth trits. The ojectives of the current study were to (i) vlidte the ssocition of the SSC4 region with PRRS resistnce in two unrelted commercil crossred popultions nd (ii) further chrcterize the effect of the region in terms of gene ction, prent- nd reed-of-origin effects nd genetic vrince explined. Mterils nd methods The Knss Stte University Institutionl Animl Cre nd Use Committee pproved ll experimentl protocols for this study. Study design A detiled description of the design, dt collection nd moleculr techniques used in PRRS host genetic consortium trils hs een previously pulished (Lunney et l. 2011). A totl of five trils were nlyzed here. Trils 1 through 3 were descried in Boddicker et l. (2012) nd included pigs of the sme cross from single reeding compny. Trils 4 nd 5 re unique to this study nd were sourced from two different reeding compnies, nd ech represents different cross. Pigs were provided from commercil reeding progrms in the United Sttes nd Cnd. Briefly, ech tril of pproximtely 200 commercil pigs involved trnsporting nimls t wening (18 28 dys of ge) to Knss Stte University, where they were sujected to PRRS chllenge. Within tril, pigs were from the sme highhelth frm, except for tril 5, which included pigs from two frms. All frms were free of PRRSV, Mycoplsm hyopneumonie nd swine influenz virus. Upon rrivl, pigs were rndomly plced into pens of 10 15 pigs. After 7-dy cclimtion period, pigs now etween 25 nd 35 dys of ge (dy 0) were experimentlly infected intrmusculrly nd intrnslly with 10 5 (TCID50) of NVSL 97-7985, highly virulent PRRSV isolte (Fng et l. 2007). Blood smples were collected t 6, 0, 4, 7, 11, 14, 21, 28, 35 nd 42 dpi. Body weight (BW) ws mesured t 0, 7, 14, 21, 28, 35 nd 42 dpi. Pigs were euthnized t 42 dpi. Viremi ws mesured using semi-quntittive TqMn PCR ssy for PRRSV RNA, s descried in the study y Boddicker et l. (2012). The ssy results were reported s the log 10 of PRRSV RNA copies per millilitre of serum. Er tissue ws collected from ll pigs for DNA isoltion. Tissue or genomic DNA from the sires of pigs in trils 1 through 3 nd from sires nd dms for trils 4 nd 5 ws supplied y the reeding compnies. Tissues or DNA smples were sent to GeneSeek, Inc. for genotyping with Illumin s Porcine SNP60 BedChip. Dt from pproximtely 960 pigs infected with PRRSV cross the five trils were evluted. Tle 1 provides n overview of the popultion structure y tril. Pigs in trils 1 through 3 were F1 individuls from Lrge White (LW) nd Lndrce (LR) cross (see Boddicker et l. 2012 for further detils). Pigs in tril 4 were from Duroc sires crossed with LW/LR/Yorkshire composite line, nd tril 5 pigs were cross etween Duroc sires nd F1 LR/Yorkshire Tle 1 Popultion structure of the five trils. Tril Breed n WUR10000125 llele frequency 1 Numer offspring per fmily Totl numer of offspring Minimum Men Mximum Brrows Gilts Totl Ded 2 1 3 3,4 Sires LR 33 0.22 1 17.1 114 565 0 565 48 Dms LW 204 0.08 1 2.8 6 4 Sires Duroc 6 0.08 8 32.5 50 109 86 195 2 Dms LW-LR-Y 5 33 0.10 1 5.9 13 5 Sires Duroc 10 0.12 2 19.9 42 109 90 199 14 Dms LR/Y 6 38 0.22 2 5.2 10 LR, Lndrce; LW, Lrge White; Y, Yorkshire. 1 Frequency of the fvorle llele (B) t SNP WUR10000125. For trils 1 3, dm genotypes were not provided, so frequencies were determined using ordered genotypes of the offspring. 2 Numer of piglets tht died prior to 42 dys post-infection. 3 Trils 1 3 consisted of the sme cross from one reeding compny. 4 Trils 1 nd 2 were LR y LW crosses. For tril 3, 121 piglets were from LR sires y LW dms crosses nd 63 piglets were from the reciprocl cross of LW sires y LR dms. 5 LW-LR-Y composite line. 6 LR/Y F1 cross. 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
50 Boddicker et l. dms. A totl of 64 pigs died efore 42 dpi, including two from tril 4 nd 14 from tril 5. Ded pigs were necropsied; susequent gross nd microscopic pthology y ordcertified pthologist identified PRRS-ssocited disese s the mjor source of mortlity. were determined sed on frequency distriutions of genomic reltionships, which clerly showed full- nd hlf-si groupings. Bsed on this, pirs of piglets were identified to e full-sis if their genomic reltionship ws >0.35 nd s hlf-sis if their reltionship ws etween 0.10 nd 0.35. Pedigree Pedigree informtion ws provided y the reeding compnies nd spnned three genertions for the first three trils nd one genertion for trils 4 nd 5. Genotypes on prents nd offspring were used to vlidte nd correct pedigrees, where possile. Monomorphic SNPs were removed, leving pproximtely 58 000 informtive mrkers for ech tril. Prent offspring mismtch frequencies were clculted s the numer of SNPs for which the prent nd offspring hd opposing homozygous genotypes divided y the totl numer of polymorphic SNPs for which the prent nd offspring were oth homozygous. If prent offspring pir hd mismtch frequency <2 percent, then the nmed prent ws ccepted. Otherwise, offspring genotypes were compred with ll possile prents nd the most likely prent ws chosen, provided the mismtch frequency ws <10 percent. Otherwise, the prent ws set to missing. Dm genotypes were not provided for the first three trils, nd therefore, the dms provided y the reeding compny were ssumed correct. Only one individul from the first three trils ws ssigned missing prentl informtion. For tril 4, not ll prent genotypes were provided, nd the pedigree provided y the compny ws not relile. Therefore, fullnd hlf-si fmilies were identified sed on genomic reltionships computed following VnRden (2008). Thresholds for ssigning full- nd hlf-si reltionships Phenotypic trits Detils on the phenotypic trits nlyzed re in Boddicker et l. (2012). Briefly, VL ws quntified s re under the curve for log-trnsformed viremi for 0, 4, 7, 11, 14 nd 21 dpi. WG to 21 or 42 dpi (WG21, WG42) ws clculted s BW t dy 21 or 42 minus BW t dy 0. Edits for trils 1 through 3 re in Boddicker et l. (2012). Edits removed 15 individuls from trils 4 nd 5 for the nlysis of VL, with 13 due to deth prior to 21 dpi, one with missing prentge informtion nd one with missing viremi informtion. For WG42, 18 individuls were removed from trils 4 nd 5, with 17 due to deth prior to 42 dpi nd one with missing prentge informtion. The numer of individuls ville fter edits is listed in Tle 2. Sttisticl nlyses Heritilities nd vrinces due to litter were estimted sed on pedigree reltionships with single-trit niml model using the softwre ASREML (Gilmour et l. 2006). Sex nd the interction of tril nd prity of the sow were included s fixed fctors nd pen within tril, niml nd litter s rndom effects. Piglets were orn from prities rnging from 1 to 7. Prities 3 through 7 were comined into one prity clss. The effect of frm of origin, which ws relevnt only for tril 5, ecuse piglets cme from single Tle 2 Trit mens nd estimtes (SE) of heritility nd litter effects (proportions of phenotypic vrince) for viremi nd weight gin fter infection otined from single-trit pedigree-sed ASREML nlyses. Trils 1 5 Trils 1 3 Tril 4 Tril 5 Trit n Men SD 1 Heritility Litter Heritility Litter Heritility Litter Heritility Litter Viremi 927 5.9 0.54 0.22 (0.11) 0.09 (0.05) 0.39 (0.18) 0.10 (0.07) 0.09 (0.16) 0.04 (0.07) 0 0.16 (0.06) 4 dpi 2 Viremi 928 6.3 0.31 0 0.30 (0.04) 0 0.33 (0.05) 0.11 (0.17) 0.08 (0.08) 0 0.24 (0.08) 7 dpi 2 Viremi 906 6.1 0.41 0.33 (0.14) 0.16 (0.06) 0.39 (0.18) 0.17 (0.08) 0.26 (0.24) 0.04 (0.09) 0.27 (0.34) 0.12 (0.15) 11 dpi 2 Viremi 897 5.6 0.64 0.20 (0.11) 0.20 (0.06) 0.21 (0.14) 0.23 (0.07) 0.04 (0.17) 0.15 (0.10) 0.38 (0.18) 0 14 dpi 2 Viremi 859 3.8 1.0 0.22 (0.08) 0 0.24 (0.12) 0 0.08 (0.16) 0.08 (0.09) 0.14 (0.13) 0 21 dpi 2 VL 909 107 7.6 0.39 (0.15) 0.11 (0.06) 0.49 (0.19) 0.06 (0.08) 0.03 (0.19) 0.38 (0.12) 0.34 (0.17) 0 WG21, kg 909 5.6 2.0 0.41 (0.09) 0 0.42 (0.12) 0 0.48 (0.18) 0 0.22 (0.13) 0 WG42, kg 889 15.1 4.0 0.34 (0.14) 0.09 (0.06) 0.37 (0.18) 0.12 (0.08) 0.40 (0.17) 0 0.20 (0.13) 0 dpi, dy post-infection; WG21, weight gin from 0 to 21 dpi; WG42, weight gin from 0 to 42 dpi; VL, virl lod clculted s re under the curve of log-trnsformed viremi etween 0 nd 21 dpi. 1 Stndrd devition clculted s the squre root of the sum of niml, litter nd residul vrinces from the joint ASREML nlysis of ll five trils. 2 Log 10 templtes/ml (qpcr). 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
Vlidtion of PRRS QTL 51 frm for ech of the other trils, ws not significnt (P > 0.65) nd therefore not included in the nlyses. Genetic correltions etween trits were estimted using ivrite niml models with the sme fixed nd rndom fctors s used in the single-trit models. Genome-wide ssocition nlyses Associtions of SNP genotypes with phenotypes were nlyzed y fitting ll SNPs simultneously using Byesin genomic selection methods (Hier et l. 2011), s implemented in the softwre GENSEL (Fernndo & Grrick 2009), using the following mixed model: y ¼ X þ Xk z i i d i þ e; i¼1 where y = vector of phenotypic oservtions; X = incidence mtrix relting fixed fctors to phenotypes; = vector of fixed fctors of sex, pen within tril nd the interction of tril nd prity clss; z i = vector of the genotype covrite for SNP i sed on the numer of B lleles using Illumin s genotype clling (coded 0, 1, 2 or equl to the tril verge for missing genotypes); i = llele sustitution effect for SNP i; nd d i = indictor for whether SNP i ws included (d i = 1) or excluded (d i = 0) in the model for given itertion of the Monte Crlo Mrkov chin (MCMC). Pen within tril ws included s fixed fctor, s opposed to rndom effect in ASREML, ecuse the current version of GENSEL does not llow dditionl rndom effects. A totl of 50 000 itertions were run for ech phenotypic nlysis, with the first 5000 itertions discrded s urn-in. The proility of d i = 0 ws set equl to p = 0.99. The Byesin model ws implemented using method Byes-B (Hier et l. 2011). Genomic regions ssocited with trits were identified using 1-M non-overlpping windows using uild 10.2 of the swine genome (http://www.nci.nlm.nih.gov/nuccore?term=199 Sus%20scrof%2C%20whole%20genome%20shotgun%20 sequence, ccessed 1 Novemer 2011). The pproch descried y Wolc et l. (2012) nd implemented in version 4.0 of the GENSEL softwre (Fernndo & Grrick 2009) ws used to compute the posterior distriution of the proportion of genetic vrince cptured y ech 1-M window. Vlidtion of the effect of the SSC4 region Vlidtion of the effect of the 1-M region on SSC4, s identified y Boddicker et l. (2012), ws crried out using SNP effect estimtes otined from the first three trils to predict GEBV of nimls in trils 4 nd 5: GEBV i ¼ Xn j¼1 ðz ij ^ j Þ; where GEBV i = genomic estimted reeding vlue for the ith niml of tril 4 or 5, n = numer of SNPs included in the prediction, z ij = genotype covrite of SNP j for niml i (coded 0, 1, 2 or tril verge for missing genotypes) nd ^ j = llele sustitution effect estimte for SNP j sed on the nlysis of trils 1 through 3. GEBV from the wholegenome nlysis were computed either sed on SNPs cross the entire genome (n = 58 277 SNPs fter removing SNPs tht were monomorphic cross ll five trils) or sed on only the 38 SNPs in the 1-M region on SSC4. Accurcy of the resulting GEBV ws evluted y correlting the GEBV with phenotypes in trils 4 nd 5 djusted for fixed effects estimted from the ASREML nlyses of phenotypes. Further chrcteriztion of the QTL on SSC4 Estimtion of SNP effects Detils on single-mrker nlysis of SNP WUR10000125 identified in the genome-wide nlyses for trils 1 through 3 re in Boddicker et l. (2012) nd were repeted here for trils 1 through 5. Briefly, SNP WUR10000125 ws included s n dditionl clss vrile in the ASREML nlysis of trils 4 nd 5 nd ll five trils jointly, ecuse tht SNP contriuted over 99 percent of the vrince explined y the region on SSC4. Non-dditive gene ction ws investigted y fitting contrsts etween genotypic effects. Dominnce ws investigted y testing whether the AB genotype ws significntly different from the verge of the AA nd BB genotypes (llele nmes, A vs. B, were sed on Illumin s A/B genotype clling). The B llele ws fvorle (Boddicker et l. 2012), so the contrst etween genotypes AB nd BB ws used to test for complete dominnce. Dominnce ws investigted using ll five trils jointly. Cuse-nd-effect reltionships of the effects of SNP WUR10000125 on VL nd WG42 were investigted y including VL s fixed liner covrite in the nlysis of WG42 nd WG42 s fixed liner covrite in the nlysis of VL. Breed nd prentl origin effects The effect of SNP WUR10000125 ws evluted for prentnd reed-of-origin effects. Aville genotypes of offspring nd their prents for the 38 SNPs (34 fter removl of monomorphic SNPs) in the 1-M region on SSC4 were phsed using PHASE softwre (Stephens et l. 2001) seprtely for trils 1 3, 4 nd 5. Of the 38 SNPs, four were monomorphic cross ll popultions, resulting in 34 informtive mrkers. Bsed on the identified hplotypes, prentl origins were determined for SNP WUR10000125 lleles of heterozygous nimls. Prentl origin could not e determined for 12 nimls in the first three trils ecuse their genotype for SNP WUR10000125 ws missing. Ordered genotypes (AA, AB, BA, BB) were fitted s clss vrile in the previously descried ASREML univrite niml model, nd significnce of prent- or reed-of-origin 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
52 Boddicker et l. effects ws determined sed on the contrst etween the AB nd BA genotypes. Results Phenotypic prmeters Sex ws not significnt for VL ut ws significnt for WG (P < 0.01), with rrows hving greter WG thn gilts. Rw phenotypic mens for viremi nd WG re shown in Tle 2. Viremi peked, on verge, t 7 dpi nd therefter decresed through 21 dpi. The vrition in viremi ws lowest t 7 dpi nd incresed therefter through 21 dpi. The men WG to 21 dpi ws 5.6 kg nd 15.1 kg to 42 dpi, with n increse in vrition from 21 to 42 dpi. Phenotypic correltions mong trits otined from the ASREML nlyses re shown in Tle 3. Both mesures of WG were phenotypiclly negtively correlted with ll mesures of viremi. The phenotypic correltion etween WG42 nd VL ws moderte. Phenotypic correltions etween VL nd dily mesures of viremi were ll positive, with the highest correltion t 14 dpi. Estimtes of genetic prmeters Estimtes of heritility nd litter components from the single-trit niml model nlyses for ll five trils jointly, nd y tril, re shown in Tle 2. Trils 1, 2 nd 3 were nlyzed together. Heritility estimtes for trils 1 through 3 were greter thn reported for the sme dt y Boddicker et l. (2012), nd litter vrinces were smller, likely due to the pedigree corrections tht were mde. Heritility estimtes for trils 4 nd 5 were low to moderte ut hd lrge SE due to smll smple sizes. The estimte of litter vrince of VL for tril 5 ws 0. These results must e interpreted with cution ecuse estimtes hve lrge SE nd litter nd genetic effects re confounded. For trils 1 through 5 comined, heritility estimtes were ll moderte, with the exception of viremi t 7 dpi, which hd n estimte of 0. Heritility estimtes for VL nd WG42 cross the five trils were 0.39 nd 0.34 respectively. The proportion of phenotypic vrince explined y litter ws 0.11 for VL nd 0.09 for WG42. Estimtes of genetic correltions mong trits sed on joint nlysis of ll five trils re shown in Tle 3. Estimtes of the genetic correltions of WG with ech mesure of viremi were negtive ut hd lrge SE. The genetic correltion etween WG42 nd VL ws 0.31 0.26. Viremi t 11 nd 14 dpi ws highly positively correlted (0.94 0.08). VL ws lso highly positively correlted (>0.9) with viremi t 7, 11 nd 14 dpi. The estimte of the genetic correltion etween WG21 nd WG42 ws perfect t 1. Vlidtion of the effects of the SSC4 region Correltions etween GEBV nd phenotypes djusted for fixed effects for individuls from trils 4 nd 5, where GEBV were predicted sed on estimtes of SNP effects otined from trils 1 through 3, re shown in Tle 4. When dividing y the squre root of heritility, these correltions cn e interpreted s estimtes of the ccurcy of the GEBV. All correltions were positive nd, on verge, the GEBV predicted WG mrginlly etter thn it did VL, nd predictions sed solely on the SSC4 region were mrginlly etter thn were the whole-genome predictions. Results of single-mrker nlyses for the most significnt SNP in the SSC4 region (WUR10000125) tht ws identified y Boddicker et l. (2012) for VL nd WG42 re shown in Fig. 1. This SNP ws significnt for WG42 in tril 4(P < 0.0004); AA nimls gined 3.2 kg less BW over the 42-dy test period compred with AB nimls. Individuls with genotype BB were not significntly different from the other two genotypes, ut the numer of BB nimls ws smll (Fig. 1). The SNP lso ws significnt (P < 0.03) for VL in tril 4; AA individuls hd higher VL compred with AB individuls. For tril 5, the SNP ws significnt for oth WG42 (P < 0.03) nd VL (P < 0.001). Individuls with the AA genotype hd, on verge, lower WG nd higher VL compred with AB individuls. Individuls with genotype Tle 3 Estimtes of phenotypic (ove digonl) nd genetic (elow digonl) correltions sed on ivrite nlyses of viremi nd weight gin following infection for trils 1 through 5. Trit 4 dpi 7 dpi 11 dpi 14 dpi 21 dpi VL WG21, kg WG42, kg Viremi 4 dpi 0.38 0.03 0.09 0.04 0.06 0.04 0.03 0.04 0.34 0.04 0.16 0.04 0.14 0.04 Viremi 7 dpi 0.01 3.7 0.13 0.04 0.09 0.04 0.09 0.04 0.34 0.04 0.16 0.04 0.13 0.04 Viremi 11 dpi 0.28 0.34 0.79 1.0 0.68 0.02 0.30 0.04 0.71 0.02 0.05 0.04 0.14 0.04 Viremi 14 dpi 0.42 0.38 0.96 1.6 0.94 0.08 0.39 0.03 0.78 0.02 0.20 0.04 0.24 0.04 Viremi 21 dpi 0.11 0.44 0.99 0.45 0.78 0.20 0.81 0.18 0.75 0.02 0.17 0.04 0.20 0.04 VL 0.40 0.29 0.98 0.71 0.92 0.07 0.96 0.04 0.86 0.10 0.23 0.04 0.28 0.04 WG21, kg 0.63 0.62 0.73 0.56 0.16 0.28 0.15 0.30 0.04 0.34 0.27 0.24 0.78 0.02 WG42, kg 0.47 0.32 0.61 0.45 0.16 0.30 0.32 0.31 0.21 0.34 0.31 0.26 1.00 0.02 dpi, dy post-infection; WG21, weight gin from 0 to 21 dpi; WG42, weight gin from 0 to 42 dpi; VL, virl lod clculted s re under the curve of log-trnsformed viremi etween 0 nd 21 dpi. 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
Vlidtion of PRRS QTL 53 Tle 4 Percentge of genetic vrince explined y the 1-M region on Sus scrof chromosome 4 for the joint nlysis of trils 1 through 5 using Byes-B nd its significnce sed on the proportion of posterior smples tht hd vrince greter thn expected. Trit % of totl genetic vrince Proportion > expected 1 Viremi 4 dpi 3 2.2 0.860 0.43 Viremi 7 dpi 3 1.3 0.699 0.43 Viremi 11 dpi 3 10.9 0.998 0.58 Viremi 14 dpi 3 6.5 0.984 0.49 Viremi 21 dpi 3 0.77 0.396 0.19 VL 15.1 1.00 0.49 WG21, kg 5.0 1.00 0.70 WG42, kg 11.3 1.00 0.44 Mrker-sed heritility 2 dpi, dy post-infection; WG21, weight gin from 0 to 21 dpi; WG42, weight gin from 0 to 42 dpi; VL, virl lod clculted s re under the curve of log-trnsformed viremi etween 0 nd 21 dpi. 1 Proportion of posterior smples from the Monte Crlo Mrkov chin for which the 1-M region on chromosome 4 explined greter percentge of the genetic vrince thn expected if ll 1-M regions cross the genome explined n equl percentge of the genetic vrince, tht is, 0.039 percent. 2 Heritility estimtes otined from GENSEL softwre using method Byes-B with p = 0.99. 3 Log 10 templtes/ml (qpcr). () 24 22 20 18 16 14 12 Weight gin (kg) 10 8 () 120 Virl lod (re under the curve) 115 110 105 100 95 90 AA AB BB 645 AA AB BB 663 223 All 21 224 22 22 All 21 161 161 28 4 5 Tril 163 28 4 4 122 122 58 3 3 4 4 124 124 57 4 5 Tril 3 BB were not significntly different from the other two genotypes; however, there were few nimls with the BB genotype (Fig. 1). Joint nlysis of ll five trils The posterior men of the percentge of genetic vrince explined y the 1-M region on SSC4 for viremi on individul dys rnged from 0.77 to 10.9 percent (Tle 4), with 21 dpi hving the lowest nd 11 dpi hving the highest percentge. If ll 2592 of the 1-M windows cross the genome contriuted eqully to the genetic vrince, ech window would contriute 0.039 percent. Thus, the SSC4 region ws estimted to contriute sustntilly greter percentge of genetic vrince thn expected under n infinitesiml model for ech dpi (=0.039% sed on 2592 1-M windows evluted). Tle 4 lso shows significnce test for the hypothesis tht the SSC4 region contriutes more vrince thn expected (Wolc et l. 2012) sed on the percentge of smples from the posterior distriution for which the 1-M window on SSC4 contriuted more thn 0.039 percent of the genetic vrince. This percentge ws s high s 99.8 for viremi t 11 dpi nd not lower thn 39.6 for 21 dpi. Tle 4 lso shows the mrker-sed heritility estimtes otined from GENSEL; estimtes were moderte to high nd generlly higher thn pedigree-sed estimtes (Tle 2). The 1-M region on SSC4 ws estimted to explin 15, 5 nd 11.3 percent of the genetic vrince for VL, WG21 nd Figure 1 Lest squres mens y genotype for SNP WUR10000125 cross ll trils 1 through 5 nd within trils 4 nd 5 for weight gin from 0 to 42 dys post-infection () nd virl lod (), clculted s re under the curve of log-trnsformed viremi from 0 to 21 dys post-infection. Within tril, columns with different letter re significntly different t P < 0.05. Numers of individuls within genotype re listed in the rs. WG42 respectively, nd every smple showed greter vrince thn expected for the 1-M region (Tle 4). Figure S1 shows the posterior distriution of the percentge of genetic vrince explined y the 1-M region on SSC4 for every 100th itertion of the MCMC for WG42 (Fig. S2A) nd VL (Fig. S2B). For VL, genetic vrince explined y the region ws t lest 5 percent for ll MCMC smples. For WG42, 94 percent of smples hd vrince >5 percent. Using 2pq 2 to estimte the dditive genetic vrince cptured y SNP WUR10000125 (Flconer & Mcky 1996), this SNP ccounted for 31.8, 5.7 nd 28.6 percent of the dditive genetic vrince (using the estimte otined from joint nlysis of ll trils) for VL for trils 1 3, 4 nd 5 respectively. Similrly for WG42, this SNP ccounted for 22.7, 18.5 nd 8.0 percent of the dditive genetic vrince in trils 1 3, 4, nd 5 respectively. The frequency of the fvorle B llele used to clculte the dditive genetic vrince cptured y SNP WUR10000125 ws 0.16, 0.09 nd 0.17 for trils 1 3, 4 nd 5 respectively. Single-mrker nlysis results for SNP WUR10000125 cross ll trils re presented in Fig. 1. The SNP ws highly 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
54 Boddicker et l. significnt for WG42 (P < 1.5 9 10 12, Fig. 1) nd VL (P < 1.6 9 10 14, Fig. 1). The difference etween the lest squres mens for the AB genotype nd the verge of the AA nd BB genotypes ws 2.8 1.0 units (P < 0.007) for VL nd 1.5 0.6 kg (P < 0.015) for WG42; therefore, dominnce effects were significnt for oth trits. The BB genotype ws not well represented ut ws not significntly different from the AB genotype (P > 0.48) for either trit. The contrst etween AA nd AB genotypes for SNP WUR10000125 were still significnt (P < 0.001) when VL ws included s covrite in the nlysis of WG42 nd of similr mgnitude ( 1.8 vs. 2.3 kg with or without VL included s covrite). Assuming liner effect of VL on WG, this indictes tht the effect of the SSC4 region on WG42 ws not completely cused y the effect of the SSC4 region on VL. Similrly, the contrst etween the AA nd AB genotypes for VL ws still significnt (3.7 vs. 4.7 units with or without WG42 s covrite, P < 0.001) when including WG s covrite in the nlysis of VL, suggesting tht the effect of the SSC4 region on VL is not through its effect on WG. So the effects of the SSC4 region on WG nd VL pper to e comintion of effects tht re common to oth trits nd effects tht re specific to ech trit. Prent- or reed-of-origin effects The effect of SNP WUR10000125 ws tested y tril to determine prent- nd reed-of-origin effects for the SSC4 region. Hplotype proilities were high, with n verge of 0.997 nd SD of 0.03 cross ll five trils, which indictes tht sufficient informtion ws ville to phse genotypes for most individuls. Across ll five trils, 60 unique hplotypes were present. Estimtes of the frequency of the B llele for ech prentl line y tril re shown in Tle 1. Figure 2 shows the lest squres mens for the ordered genotypes for WG42 (Fig. 2) nd VL (Fig. 2). Effects of the heterozygotes ordered y prentl origin (AB vs. BA) were not significntly different (P > 0.30) from ech other for VL or WG42, nd t lest one of the ordered heterozygotes ws significntly different from the AA genotype (P < 0.02) within ech tril. Anlyzed jointly, the AB nd BA genotypes were not significntly different from ech other (P > 0.99), ut the AA genotype ws significntly different from oth AB nd BA genotypes (P < 0.01) for oth VL nd WG42 (Fig. 2). Therefore, there is no sttisticl evidence for prent-of-origin or reed-of-origin effects. Discussion Boddicker et l. (2012) identified region on SSC4 tht ws ssocited with VL nd WG using pigs from single commercil cross etween LW line nd LR line tht were experimentlly infected with PRRSV. In this study, the () Weight gin (kg) () Virl lod (re under the curve) 24 22 20 18 16 14 12 10 8 115 110 105 100 95 AA AB BA BB 645 70 125 21 AA AB BA BB effects of tht region were vlidted in two dditionl, unrelted popultions, nd the fvorle llele ws shown to hve dominnt mode of inheritnce. Furthermore, the effect of the region on SSC4 ws shown to e present in, nd expressed y, oth pternl nd mternl chromosomes from the LR, LW, Yorkshire nd Duroc lines tht were involved in the crosses. In the following, these results will e discussed in further detil nd compred with the literture. It must e emphsized tht comprisons to Boddicker et l. (2012) re confounded ecuse the current nlyses included the dt used in tht study (trils 1 through 3). Genetic prmeters 362 23 97 14 All 1-3 4 5 663 70 126 22 376 23 99 15 Tril Decresed WG is reportedly ssocited with the onset of PRRSV infection (Greiner et l. 2000; Doeschl-Wilson et l. 161 13 9 4 122 34 19 All 1-3 4 5 Tril 163 13 9 4 124 34 18 Figure 2 Lest squres mens y ordered genotype for SNP WUR10000125 cross ll trils 1 through 5 nd within tril for weight gin from 0 to 42 dys post-infection () nd virl lod (), clculted s re under the curve of log-trnsformed viremi from 0 to 21 dys post-infection. Trils 1 through 3 were grouped s nimls were of the sme cross from the sme genetics compny. Within tril, columns with different letter re significntly different t P < 0.05. In the genotype nottion, the first llele is of pternl origin nd the second llele is of mternl origin. 3 3 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
Vlidtion of PRRS QTL 55 2009; Boddicker et l. 2012). Greiner et l. (2000) reported decresed WG t 8 dpi with PRRSV, reltive to WG t 8 dys pre-infection. Boddicker et l. (2012) reported negtive phenotypic correltions etween VL nd WG to 21 dpi nd WG to 42 dpi. In the current study, WG21 nd WG42 lso were found to e phenotypiclly negtively correlted with ll mesures of viremi nd VL, consistent with previously pulished results of decresed WG when pigs re infected with PRRSV. Compred with results of Boddicker et l. (2012), correction of pedigree errors resulted in sustntil increses in heritilities for trils 1 through 3 nd in reduction in the proportion of vrince due to litter (Tle 2). Both mesures of WG were modertely heritle in trils 4 nd 5, which is consistent with results from the first three trils. When ll five trils were nlyzed together, heritility of VL incresed from 0.31 for trils 1 3 in Boddicker et l. (2012) to 0.39 for joint nlysis of ll five trils in the current study. Heritility estimtes for viremi over time for trils 4 nd 5 rnged from 0 to 0.38 ut hd lrge SE due to smll smple sizes within tril. These estimtes my still e underestimted s result of dditionl pedigree errors tht could not e corrected ecuse of missing prentl SNP genotypes. Mrker-sed heritility estimtes were generlly sustntilly higher thn were the estimtes otined from pedigree informtion (Tle 4). This is likely ecuse litter ws not included in the GENSEL nlysis, s the current version of GENSEL does not llow dditionl rndom effects to e included. However, for VL, the two estimtes of heritility were in greement with ech other (pedigree = 0.39, mrker = 0.49). VL is summry sttistic for viremi, which reduces some of the mesurement nd smpling errors tht re present in mesures of viremi on individul dys. Within tril, estimtes of vrince due to litter were not relile due to smll smple sizes nd confounding etween genetic nd litter effects. Joint nlysis of ll trils reveled tht the dm provided mternl component in ddition to her direct genetic contriution tht ided her offspring to respond to the virus, despite the dms eing from PRRS-free herds. Litter explined sustntil proportion of the phenotypic vrince for viremi t 7, 11 nd 14 dpi cross trils (0.31, 0.16 nd 0.18 respectively; i.e. greter for dpi closer to the dy of infection). This mternl component could e due to mternl genetics or due to mternl environmentl effects. Sows tht provide etter mternl environment up to wening result in stronger piglets tht my e le to cope etter with chllenges, such s PRRS. Boddicker et l. (2012) reported lrger litter effect for VL nd WG42 for the first three trils likely due to pedigree errors tht were corrected in the current nlyses. Estimtes of genetic correltions were generlly positive mong mesures of viremi nd mong mesures of WG nd negtive etween mesures of viremi nd WG (Tle 3). However, most of these estimtes were not significntly different from zero (P > 0.10). Similr to wht ws oserved y Boddicker et l. (2012), viremi t 11 nd 14 dpi ws highly nd positively geneticlly correlted with ech other nd with VL. Viremi on these 2 dys hevily influenced the results oserved for VL. Genome-wide ssocition nlysis A multireed dt set ws used for genome-wide ssocition nlysis, which rises concerns out the potentil impct of popultion structure in detecting flse positives (Pritchrd et l. 2000). Breed differences etween trils were ccounted for y including tril s fixed effect in the nlyses. Within tril, reed composition ws homogeneous, with ll pigs eing generted y the sme cross (or its reciprocl in the cse of tril 3). The dms in tril 4 were from three-reed composite, which ws ssumed to e homogeneous. A genome-wide ssocition nlysis ws conducted using the Byesin genomic prediction sttisticl models, which nlyze ll SNPs simultneously ut, other thn tril, did not explicitly include dditionl fctors to ccount for popultion structure. However, our recent reserch (ShrfToosi 2012) hs demonstrted tht these models implicitly ccount for popultion structure nd tht not fitting popultion structure explicitly does not led to n excess of flse positives in these models. One reson for this is tht ny SNPs tht differ in llele frequencies etween supopultions cn cpture the effects of popultion structure, nd thus, ny effects of popultion structure re spred out cross mny SNPs cross the genome. With over 2500 1-M windows, this is expected to increse the vrince ssocited with ech window only y frction of 1/2500 of the vrince contriuted y popultion structure, which is expected to e miniml. Furthermore, if differences due to popultion structure contin genetic informtion, llowing the SNPs to cpture these effects ctully increses the power to detect QTL (ShrfToosi 2012). Vlidtion Boddicker et l. (2012) used Build 10 of the swine genome. The 1-M region on SSC4 tht ws identified y Boddicker et l. (2012) included 37 SNPs in Build 10, of which four were monomorphic. Two of these SNPs were locted upstrem (pproximtely 469 k) from the region in Build 10.2, nd three SNPs tht were unmpped in Build 10 were plced into the region on SSC4 in Build 10.2. Thus, in Build 10.2, the SSC4 region included 38 SNPs, including four monomorphic SNPs. At the whole-genome level, vlidtion correltions etween GEBV nd djusted phenotypes were low to moderte (Tle 5). The correltion estimtes cn e 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
56 Boddicker et l. Tle 5 Correltions (SE) etween phenotypes for trils 4 nd 5 nd genomic estimted reeding vlues t the whole-genome level nd for the 1-M region on Sus scrof chromosome 4 (SSC4) using llele sustitution effect estimtes otined from trils 1 through 3. Trit 1 Tril 4 Tril 5 Genome SSC4 region Genome SSC4 region Virl lod 0.07 0.07 0.14 0.07 0.20 0.07 0.18 0.07 Weight 0.19 0.07 0.12 0.07 0.15 0.07 0.24 0.07 gin, kg 1 Virl lod ws clculted s re under the curve of log-trnsformed viremi etween 0 nd 21 dys post-infection (dpi). Weight gin is from 0 to 42 dpi. converted to estimtes of the ccurcy of GEBV (i.e. the correltion etween true nd estimted reeding vlues) y dividing y the squre root of heritility of the trit. Using heritilities of 0.4, this results in estimtes of ccurcy rnging from 0.11 to 0.32 t the whole-genome level nd 0.19 0.38 for the 1-M region on SSC4. These ccurcies re high, especilly considering the limited size of the trining dt nd the fct tht prediction ws cross reeds. Across-reed predictions re typiclly not ccurte due to dominnce, epistsis nd differences in linkge disequilirium (LD) etween reeds (Grrick 2010). Accurcies of GEBV computed sed on the 1-M region on SSC4 were on verge of similr mgnitude s ccurcies of wholegenome GEBV. This suggests tht the rest of the genome dded s much noise s true effects to the GEBV predictions. The SSC4 region hd very similr LD ptterns in ll five trils (Fig. S2), which my explin the predictive power of effects estimted for this region cross reeds. The effects of SNP WUR10000125 in trils 4 nd 5 were similr to those previously found for the first three trils (Boddicker et l. 2012), in which AA nimls hd incresed VL nd decresed WG42 compred with the AB nimls for oth trits. Therefore, the effect of this region hs now een identified in three unrelted popultions, suggesting the genetic vrint likely existed efore the divergence of the reeds represented in this study. Furthermore, the low frequency of the B llele my indicte tht, historiclly, the B llele hs not hd n dvntge for nturl or rtificil selection. When the SNP ws fitted s fixed fctor in the joint nlysis of ll five trils, the effect of the fvorle llele ws shown to e dominnt. Expressed in terms of phenotypic SD, the effects were 0.54 nd 0.53 for VL nd WG42. The minor llele (frequency = 0.15) ws ssocited with the desired phenotype of reduced VL nd incresed WG. The SSC4 region explined lrge percentge of the genetic vrince for VL nd WG42 (15.1 nd 11.3 percent respectively). These percentges re similr to those reported for the first three trils y Boddicker et l. (2012) ut were clculted in different mnner; percentges in Boddicker et l. (2012) were clculted sed on vrinces of GEBV; here, estimtes were sed on the posterior distriution of the vrinces of true reeding vlues. Nevertheless, ll these results provide convincing evidence tht the previously identified QTL on SSC4 is present in two dditionl, unrelted commercil popultions nd hs lrge effects on host response to experimentl PRRSV infection. Potentil cndidte genes for the QTL identified in the region were discussed in Boddicker et l. (2012) nd include fmily of gunylte-inding proteins (GBP1, GBP2, GBP4, GBP5 nd GBP6) nd generl trnscription fctor 2. As discussed in Boddicker et l. (2012), these proteins hve een reported to hve ntivirl ctivity in humns, nd the GBP fmily of genes is ssocited with the innte immune system. Fine-mpping this QTL nd future identifiction of the custive muttion re clerly hmpered y the extensive LD tht exists in the region cross reeds. Therefore, functionl nlyses will e required to provide ny convincing evidence of cuslity mong cndidte muttions. The frequency of the fvorle B llele is low, which egs the question, why? Interestingly, the frequency is low in oth mternl nd terminl lines, which re selected primrily for reproductive nd production trits respectively. Thus, it is not very likely tht the QTL on SSC4 ssocited with PRRS is in LD with mjor QTL for reproduction or production trits. Bsed on the results ville from PigQTLd (http://nimlgenome.org/ cgi-in/qtld/ss/index), no QTL for production trits, such s growth or feed efficiency, hve een reported within 2 M of the PRRS QTL on SSC4. However, QTL hve een reported for met qulity trits, such s intrmusculr ft content, sher force nd crcss weight within 9 M of the SSC4 region. Non-functionl nipple is the only reproduction-relted QTL tht hs een reported within 2 M of the PRRS QTL. Selection for performnce in the presence of PRRS infection is expected to fvor the B llele. The low frequency of tht fvorle llele is likely due to comintion of the following: (i) PRRSV is reltively new disese nd (ii) selection typiclly occurs in high-helth nucleus frms, which tend to e free of PRRS. Furthermore, rndom drift could cuse low frequency of the fvorle llele nd in differences in llele frequencies etween reeds. Origin of SSC4 effects Prent-of-origin effects re commonly the result of imprinting, nd, s reviewed y Guilmtre & Shrp (2012), there is evidence tht imprinting contriutes to numer of humn disorders such s Alzheimer s disese, utism, rest cncer nd oth type 1 nd 2 dietes. However, to our knowledge, there is no previous evidence of imprinting effects for response to infectious diseses. The prent-of-origin nlysis performed here ws twofold, in the sense tht it tested whether the effect ws specific to pternlly or mternlly 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58
Vlidtion of PRRS QTL 57 inherited lleles nd lso whether the effect ws reed specific. These effects were, however, prtilly confounded ecuse some reeds were used only s pternl or mternl reeds in the crosses nlyzed. The first three trils were sed on cross etween two white reeds (Yorkshire nd LR cross), nd trils 4 nd 5 encompssed the Yorkshire, LR, LW nd Duroc reeds. Duroc pigs typiclly re used s terminl sires in the commercil sector, nd Yorkshire nd LR re typiclly used s mternl reeds. The Yorkshire nd LR lines used in trils 4 nd 5 re, however, expected to e distinct from ech other nd from the Yorkshire nd LR lines used in trils 1 through 3 ecuse of the seprte popultions nd selection strtegies used y ech reeding compny. When SNP WUR10000125 ws fitted s n ordered genotype, the effect ws present irrespective of whether the llele ws pternlly or mternlly inherited. Additionlly, the effect of the fvorle llele ws present for ech of the prentl reeds nd lines represented in the crosses. Conclusions nd implictions A lrge QTL ws previously identified on SSC4 using pproximtely 560 pigs from single commercil cross. The effects of this region hve now een vlidted in two unrelted crosses from two dditionl reeding compnies. The effects were of similr mgnitude nd in the sme direction s previously identified. The QTL cts in dominnt mtter, wherey pigs with the fvorle llele hve reduced viremi nd incresed WG. The effect is present in lines from the mjor reeds involved in commercil pork production, nd there is no indiction of prent-of-origin or reed-of-origin effects, s this study included LR, Yorkshire, LW nd Duroc reeds. Frequencies of the fvorle llele were <17 percent in ech prentl line, suggesting mple room for mrker-ssisted selection on this region. These results re promising for the swine industry nd for swine helth with respect to PRRS. However, they re sed on infection with specific strin of the PRRSV under experimentl conditions. Additionl reserch, such s chllenges with other PRRSV isoltes nd field trils, is required to etter understnd the effects of this genomic region on pig production in the field. Nevertheless, if the effect of this region is common cross reeds nd popultions, the fvorle llele hs low frequency nd the effects oserved here replicte in the field, selection for the fvorle llele could led to sustntil reductions in economic losses due to PRRS. Acknowledgements This project ws supported y the USDA NIFA PRRS CAP Awrd 2008-55620-19132, the Ntionl Pork Bord, the NRSP-8 Swine Genome nd Bioinformtics Coordintion projects, nd the PRRS Host Genetics Consortium consisting of USDA ARS, Knss Stte Univ., Iow Stte Univ., Michign Stte Univ., Wshington Stte Univ., Purdue Univ., University of Nersk-Lincoln, PIC/Genus, Newshm Choice Genetics, Fst Genetics, Genetiporc, Inc., Genesus, Inc., PigGen Cnd, Inc., IDEXX Lortories, Tetrcore, Inc. The uthors cknowledge technicl ssistnce to this project from Jun Pedro Steiel for the lgorithm to compute re under the curve, Mx Rothschild for ssistnce with genotyping, Nder Dee for ssistnce with dt nd nlysis, Eric Fritz for dtse mngement, Ani Wolc nd Dinesh Thekkoot for sttisticl ssistnce, the lortory of Bo Rowlnd, specificlly Becky Eves, Mureen Kerrign, Ben Trile, Jessic Otrdovec, Brooke Bloomerg, Auree Gottlo, Lur O Brien nd Rnjini Chnd for niml cre nd smple collection, nd the lortory of Jon Lunney, specificlly Smuel Arms nd Amer Tietgens for preprtion of ll genomic DNA smples for SNP genotyping. References Boddicker N.J., Wide E.H., Rowlnd R.R.R., Lunney J.K., Grrick D.J., Reecy J.M. & Dekkers J.C.M. (2012) Evidence for mjor QTL ssocited with host response to porcine reproductive nd respirtory syndrome virus chllenge. Journl of Animl Science 90, 1733 46. Doeschl-Wilson A.B., Kyrizkis I., Vincent A., Rothschild M.F., Thcker E. & Glin-Pntoj L. (2009) Clinicl nd pthologicl responses of pigs from two geneticlly diverse commercil lines to porcine reproductive nd respirtory syndrome virus infection. Journl of Animl Science 87, 1638 47. Flconer D.S. & Mcky T.F. (1996) Introduction to Quntittive Genetics, 4th edn. Person, Hrlow. Fng Y., Schneider P., Zhng W.P., Ferg K.S., Nelson E.A. & Rowlnd R.R.R. (2007) Diversity nd evolution of newly emerged North Americn Type 1 porcine rterivirus: nlysis of isoltes collected etween 1999 nd 2004. Archives of Virology 152, 1009 17. Fernndo R. & Grrick D. (2009) User Mnul for Portfolio of Genomic Selection Relted Anlyses, 2nd edn. Iow Stte University. Aville t: http://igs.nsci.istte.edu/igsgui/help.html, ccessed 19 April 2012. Grrick D.J. (2010) Consequences of genomic prediction in cttle. Interull Bulletin 41, 51 8. Gilmour R.A., Gogel B.J., Cullis B.R. & Thompson R. (2006) ASREML User Guide Relese 2.0. VSN Interntionl Ltd, Hemel Hempsted, UK. Greiner L.L., Sthly T.S. & Stel T.J. (2000) Quntittive reltionship of systemic virus concentrtion on growth nd immune response in pigs. Journl of Animl Science 78, 2690 5. Guilmtre A. & Shrp A.J. (2012) Prent of origin effects. Clinicl Genetics 81, 201 9. Hier D., Fernndo R.L., Kizilky K. & Grrick D.J. (2011) Extension of the Byesin lphet for genomic selection. BMC Bioinformtics 12, 186. Holtkmp D.J., Klieenstein J.B., Neumnn E.J., Zimmermn J.J., Rotto H., Yoder T.K., Wng C., Yeske P., Mowrer C. & Hley C. (2013) Assessment of the economic impct of porcine reproduc- 2013 The Authors, Animl Genetics 2013 Stichting Interntionl Foundtion for Animl Genetics, 45, 48 58