Genome-Wide Association Analyses Highlight the Potential for Different Genetic Mechanisms for Litter Size Among Sheep Breeds

gene-09-00118 Ap il 9, 2018 Time: 17:46 # 1 ORIGINAL RESEARCH published: 10 Ap il 2018 doi: 10.3389/ gene.2018.00118 Edi ed by: Jo am Mwashigadi Mwacha o, In e na ional Cen e o Ag icul u al Resea ch in he D y A eas (ICARDA), E hiopia Re iewed by: Shahin Eghbalsaied, Islamic Azad Uni e si y, I an Cla e A. Gill, Texas A&M Uni e si y, Uni ed S a es Da id W agg, The Uni e si y o Edinbu gh, Uni ed Kingdom Mou ad Rekik, In e na ional Cen e o Ag icul u al Resea ch in he D y A eas (ICARDA), Jo dan *Co espondence: Meng-Hua Li [email p o ec ed] †These au ho s ha e con ibu ed equally o his wo k. Special y sec ion: This a icle was submi ed o Li es ock Genomics, a sec ion o he jou nal F on ie s in Gene ics Recei ed: 13 Decembe 2017 Accep ed: 23 Ma ch 2018 Published: 10 Ap il 2018 Ci a ion: Xu S-S, Gao L, Xie X-L, Ren Y-L, Shen Z-Q, Wang F, Shen M, Eyþó sdó i E, Hallsson JH, Kisele a T, Kan anen J and Li M-H (2018) Genome-Wide Associa ion Analyses Highligh he Po en ial o Di e en Gene ic Mechanisms o Li e Size Among Sheep B eeds. F on . Gene . 9:118. doi: 10.3389/ gene.2018.00118 Genome-Wide Associa ion Analyses Highligh he Po en ial o Di e en Gene ic Mechanisms o Li e Size Among Sheep B eeds Song-Song Xu1,2†, Lei Gao3,4†, Xing-Long Xie1,2, Yan-Ling Ren5, Zhi-Qiang Shen5, Feng Wang6, Min Shen3,4, Emma Eyþó sdó i 7, Jón H. Hallsson7, Ta yana Kisele a8, Juha Kan anen9and Meng-Hua Li1,2* 1CAS Key Labo a o y o Animal Ecology and Conse a ion Biology, Ins i u e o Zoology, Chinese Academy o Sciences (CAS), Beijing, China, 2College o Li e Sciences, Uni e si y o Chinese Academy o Sciences, Beijing, China, 3Ins i u e o Animal Husband y and Ve e ina y Medicine, Xinjiang Academy o Ag icul u al and Reclama ion Science, Shihezi, China, 4S a e Key Labo a o y o Sheep Gene ic Imp o emen and Heal hy B eeding, Xinjiang Academy o Ag icul u al and Reclama ion Science, Shihezi, China, 5Shandong Binzhou Academy o Animal Science and Ve e ina y Medicine Academy, Binzhou, China, 6Ins i u e o Sheep and Goa Science, Nanjing Ag icul u al Uni e si y, Nanjing, China, 7Facul y o Na u al Resou ces and En i onmen al Sciences, Ag icul u al Uni e si y o Iceland, Bo ga nes, Iceland, 8All-Russian Resea ch Ins i u e o Gene ics and Fa m Animal B eeding, Russian Academy o Sciences, Moscow, Russia, 9P oduc ion Sys ems, Na u al Resou ces Ins i u e Finland, Jokioinen, Finland Rep oduc ion is an impo an ai in sheep b eeding as well as in o he li es ock. Howe e , despi e i s impo ance he gene ic mechanisms o li e size in domes ic sheep (O is a ies) a e s ill poo ly unde s ood. To explo e gene ic mechanisms unde lying he a ia ion in li e size, we conduc ed mul iple independen genome-wide associa ion s udies in i e sheep b eeds o high p oli icacy (Wadi, Hu, Icelandic, Finnsheep, and Romano ) and one low p oli icacy (Texel) using he O ine In inium HD BeadChip, espec i ely. We iden i ied di e en se s o candida e genes associa ed wi h li e size in di e en b eeds: BMPR1B,FBN1, and MMP2 in Wadi; GRIA2,SMAD1, and CTNNB1 in Hu; NCOA1 in Icelandic; INHBB,NF1,FLT1,PTGS2, and PLCB3 in Finnsheep; ESR2 in Romano and ESR1,GHR,ETS1,MMP15,FLI1, and SPP1 in Texel. Fu he anno a ion o genes and bioin o ma ics analyses e ealed ha di e en biological pa hways could be in ol ed in he a ia ion in li e size o emales: ho mone sec e ion (FSH and LH) in Wadi and Hu, placen a and emb yonic le hali y in Icelandic, olliculogenesis and LH signaling in Finnsheep, o ula ion and p eo ula o y ollicle ma u a ion in Romano , and es ogen and ollicula g ow h in Texel. Taken oge he , ou esul s p o ide new insigh s in o he gene ic mechanisms unde lying he p oli icacy ai in sheep and o he mammals, sugges ing a ge s o selec ion whe e he aim is o inc ease p oli icacy in b eeding p ojec s. Keywo ds: sheep, p oli icacy, genome-wide associa ion s udy, biological pa hways, egula ion INTRODUCTION Rep oduc ion is one o he mos impo an ai s in li es ock p oduc ion pa icula ly o emales. Selec ion o highe p oli icacy in domes ic sheep (O is a ies) has led o a iable li e size (LS) wi hin and among b eeds. Fo example, indi idual li e size o 1 o 8 has been eco ded in he Hu sheep and Finnsheep (Yue, 1996;Da is e al., 2006a). F on ie s in Gene ics | www. on ie sin.o g 1Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 2 Xu e al. Gene ic Basis o Sheep Rep oduc ion P e ious s udies epo ed ha he excep ional p oli icacy o he Boo oola Me ino was a ibu ed o a single majo gene, while a numbe o mu a ions o a majo e ec on li e size ha e been iden i ied in o he sheep b eeds (Table 1; see also Xu and Li, 2017). Vage e al. (2013) de ec ed a mu a ion FecGFin gene GDF9 s ongly associa ed wi h li e size in No wegian Whi e Sheep and Finnish Land ace (Finnsheep) using a genome-wide associa ion analysis. Dema s e al. (2013) epo ed he mu a ions FecXG in G i e e sheep and FecXOin Olkuska sheep associa ed wi h he highly p oli ic pheno ype by a genome-wide associa ion analysis. Cao e al. (2016) ound ha nine candida e genes including he well-known FecB mu a ion played impo an oles in he a iable li e size in Hu and Small- ailed Han sheep h ough me hyla ed DNA-immunop ecipi a ion sequencing da a. Miao e al. (2016) iden i ied a se o di e en ially exp essed genes (e.g., FecB) be ween low- and high-p oli icacy b eeds (Do se s. Small- ailed Han sheep) h ough implemen ing in eg a ed analysis o miRNAs and lncRNAs. Lassoued e al. (2017) ound he mu a ion FecXBa associa ed wi h he p oli icacy in Tunisian Ba ba ine. Despi e i s g ea impo ance he gene ic mechanisms o he high p oli icacy ai in domes ic sheep a e s ill poo ly unde s ood, pa ly due o sho age o s udies conduc ed ac oss mul iple p oli ic sheep b eeds. To da e, nume ous ecundi y-associa ed mu a ions ha e been iden i ied in di e en sheep b eeds, bu e y ew mu a ions ha e been consis en ly de ec ed ac oss he b eeds. Despi e he ep oduc ion o ewes can be a ec ed by he complex in e ac ions o en i onmen al condi ions (i.e., clima e, densi y, and ood abundance) (Wilson e al., 2009), p e ious s udies sugges ed ha gene ic ac o could play impo an oles in he a iable li e size o ewes. In his s udy, we conduc ed mul iple independen genome- wide associa ion s udies (GWAS) on li e size in he sheep b eeds o high (Wadi, Hu, Icelandic, Finnsheep, and Romano ) and low (Texel) p oli icacy wi h a li e size anging om 1 o 6 om di e en geog aphic egions (Figu e 1A) and gene ic o igins (Figu e 1B) o he wo ld, espec i ely. Wadi sheep is a high-p oli icacy na i e b eed om he Shandong P o ince o China (Peng e al., 2017). Hu sheep is amous o ea ly sexual ma u i y and high ecundi y, and a e dis ibu ed in he Taihu Lake a ea o Eas e n China (Yue, 1996). Icelandic and TABLE 1 | Gene ics a ian s associa ed wi h he ecundi y in sheep. Gene Mu a ion Name, allele symbol Founde b eeds Re e ence BMP15 V299D In e dale, FecXI Romney, In e dale Galloway e al., 2000 Q291Te Hanna, FecXH Romney Galloway e al., 2000 S367I Belcla e, FecXB Belcla e Han ahan e al., 2004 Q239R Galway, FecXG Belcla e, Camb idge, Small- ailed Han Han ahan e al., 2004 C321Y Lacaune, FecXL Lacaune Bodin e al., 2007 1P154S159 Rasa A agonesa, FecXR Rasa A agonesa Ma inez-Royo e al., 2008;Mon eagudo e al., 2009 T317I G i e e, FecXG G i e e (F ance) Dema s e al., 2013 N337H Olkuska, FecXO Olkuska (Poland) Dema s e al., 2013 c.301G >T, c.310insC, c.302_304delCTA Ba ba ine, FecXBa Tunisian Ba ba ine Lassoued e al., 2017 Unknown Woodlands, FecXW Woodlands Fea y e al., 2007 BMPR1B Q249R Boo oola, FecBB Boo oola Me ino, Ga ole, Ja anese, Small- ailed Han, Wadi, Hu Mulsan e al., 2001; Souza e al., 2001; Wilson e al., 2001; Chu e al., 2011;Zhang e al., 2011;Cao e al., 2016 GDF9 S395F High Fe ili y, FecGH Belcla e, Camb idge Han ahan e al., 2004 S427R Thoka, FecGT Icelandic Nicol e al., 2009 F345C Emb apa, FecGE San a Ines Sil a e al., 2011 V371M FecGF No wegian Whi e Sheep, Finnsheep Land ace, Belcla e Vage e al., 2013; Mullen and Han ahan, 2014 R315C Vaca ia, FecGV B azilian sheep de Souza e al., 2012 R87H FecGI Baluchi Mo adband e al., 2011 B4GALNT2 Lacaune, FecLL Lacaune D ouilhe e al., 2013 Woodlands Wood-land, FecX2W Coopwo h Da is, 2005 OLKUSKA Olkuska Da is, 2004 BELLE-ILE Belle-Ile Da is, 2005 Unknown FecW Da is e al., 2006b F on ie s in Gene ics | www. on ie sin.o g 2Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 3 Xu e al. Gene ic Basis o Sheep Rep oduc ion FIGURE 1 | (A) Geog aphic loca ions o i e sheep b eeds o high (WAD, Wadi sheep; HUS, Hu sheep; ICE, Icelandic sheep; FIN, Finnsheep; and ROM, Romano sheep) and one low (TEX, Texel sheep) p oli icacy. (B) Neighbo -joining ee o he six sheep b eeds wi h 1000 boo s ap eplica es. Finnsheep (Finnish Land ace) sheep a e no he n Eu opean high- ecundi y b eeds (Mullen and Han ahan, 2014;Ei iksson and Sigu dsson, 2017). Romano sheep om he Volga Valley shows ou s anding ep oduc ion quali ies: ea ly sexual ma u i y, ou -o -season b eeding and ex ao dina y p oli icacy (Denisko a e al., 2017). The Texel sheep is a ela i ely low-p oli icacy b eed o iginally om he island o Texel in he Ne he lands and excels in muscle g ow h and lean ca casses (Casas e al., 2004). Ou esul s will be impo an o u he gene ic imp o emen o he ai and o be e unde s anding he molecula basis o ep oduc ion in sheep as well as o he mammals. MATERIALS AND METHODS Sample Collec ion and Pheno yping A o al o 522 ewes om i e sheep b eeds o high (Wadi, n= 160; Hu, n= 117; Icelandic, n= 54; Finnsheep, n= 54; and Romano , n= 78) and one low (Texel, n= 59) p oli icacy we e collec ed om a ms in China, Iceland, Finland, and Russia (Figu e 1A). Animals included we e as un ela ed as possible based on analysis o pedig ee eco ds and a me s’ knowledge. Da a o he pheno ype o li e size and he o al numbe o li e s collec ed om a m eco ds a e shown in Figu e 2. The li e size anged om 1 o 6 based on pa i y om 1 o 11 in six sheep b eeds. Genomic DNA was ex ac ed om he ea ma ginal issues ollowing a s anda d phenol/chlo o o m me hod and was dilu ed o 50 ng/µl o he SNP BeadChip geno yping (Köchl e al., 2005), excep o he Icelandic samples which we e isola ed om whole-blood using Mas e Pu eTM Comple e DNA Pu i ica ion Ki (Epicen e Bio ech) ollowing he manu ac u e s p o ocol. Geno yping and Quali y Con ol All he samples we e geno yped using he O ine In inium HD BeadChip acco ding o he manu ac u e ’s p o ocol. Geno ypes o FIGURE 2 | Pheno ypic dis ibu ion o li e size in he six sheep b eeds (WAD, Wadi sheep; HUS, Hu sheep; ICE, Icelandic sheep; FIN, Finnsheep; ROM, Romano sheep; and TEX, Texel sheep). a o al o 606,006 SNPs we e ob ained (geno ype and pheno ype da ase s1). We implemen ed quali y con ol o hese SNPs using PLINK 1.07 so wa e (Pu cell e al., 2007). The SNPs o indi iduals we e excluded i hey me any o he c i e ia: (1) no ch omosomal o physical loca ion, (2) call a e <0.95, (3) missing geno ype equency >0.05, and/o (4) mino allele equency (MAF) <0.05. SNPs we e excluded om he analysis i a p- alue o Fishe ’s exac es o Ha dy–Weinbe g equilib ium less han 0.001. Gene ic Rela ionships and Popula ion S uc u e To in es iga e he gene ic ela ionships and popula ion s uc u e among he six domes ic sheep, we pe o med global FST, 1h ps://www.animalgenome.o g/ eposi o y/pub/CAAS2018.0302/ F on ie s in Gene ics | www. on ie sin.o g 3Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 4 Xu e al. Gene ic Basis o Sheep Rep oduc ion neighbo -joining (NJ) ee and p inciple componen analysis (PCA). The global FST alue was calcula ed using GENEPOP 4.2 (Raymond and Rousse , 1995). The gene ic dis ances be ween popula ions we e calcula ed using an iden i y by s a e (IBS) simila i y ma ix (Kang e al., 2010). Then, he dis ances we e used o cons uc a NJ ee wi h 1000 boo s aps using he package PHYLIP .3.695 (Felsens ein, 1989). In addi ion, PCA was conduc ed using he Sma PCA p og am om he EIGENSOFT package e sion 4.2 (Pa e son e al., 2006) based on he geno ypes da a. Genome-Wide Associa ion Analysis To explo e gene ic s uc u e wi hin he b eeds, mul idimensional scaling (MDS) analysis was pe o med based on he independen SNPs using PLINK 1.07. Fi s ly, we implemen ed he op ion o ‘indep-pai wise 50 5 0.05’ in PLINK 1.07, which calcula ed pai wise linkage disequilib ium (LD) in a 50-SNP-window shi ed a a pace o i e SNPs. I he LD es ima e was 2>0.05, one o he pai s o SNPs was emo ed (Pu cell e al., 2007). The independen SNPs e ained by he LD c i e ia we e hen used in he MDS analysis, and he esul s we e plo ed using he GenABEL package in R 3.2.2 (Aulchenko e al., 2007). We pe o med genome-wide associa ion s udies wi hin i e sheep b eeds o high p oli icacy (Wadi, Hu, Icelandic, Finnsheep, and Romano ) and one low p oli icacy (Texel) using he case/con ol design. We anked all indi iduals wi hin he b eeds acco ding o hei li e size om he highes o lowes . Then, we selec ed indi iduals om wo ails o each b eed as ‘case’ and ‘con ol,’ espec i ely. Based on he dis ibu ion o pheno ypes, 114 samples (LS ≥2) in Wadi, 66 samples (LS ≥2) in Hu, 20 samples (LS >2) in Icelandic, 37 samples (LS ≥2.5) in Finnsheep, 40 samples (LS ≥2.5) in Romano and 28 samples (LS ≥1.6) in Texel sheep we e selec ed as ‘cases,’ while 28 samples (LS = 1) in Wadi, 15 samples (LS = 1) in Hu, 15 samples (LS ≤1.75) in Icelandic, 9 samples (LS ≤2) in Finnsheep, 26 samples (LS ≤2) in Romano and 14 samples (LS ≤1.33) in Texel sheep we e selec ed as ‘con ols.’ In he GWAS, we used he unc ion o “q sco e” in he GenABEL package. Associa ed SNPs we e iden i ied a bo h he genome-wide and ch omosome- wise signi icance le els (p<0.05) a e he Bon e oni co ec ion (Bon e oni, 1936). To accoun o sys ema ic biases caused by wi hin-popula ion subs uc u e, he i s and second dimensions om he MDS analyses we e used as he co a ia es (P ice e al., 2006). The co ela ion analysis be ween li e size and pa i y wi hin b eeds showed ha he e we e signi ican e ec s be ween li e size and pa i y in ou b eeds (Wadi, Hu, Icelandic, and Texel), and he e ec o pa i y 1 on li e size was less han ha o pa i ies 2 h ough 10 (Supplemen a y Table S1 and Supplemen a y Figu e S1). Howe e , he pa i y o indi iduals wi hin b eeds was di e en , and we mainly ocused on he mean o li e size o indi idual ( o al li e size/pa i y) in pe b eed. The e o e, we excluded he e ec o pa i y om he model. The Quan ile–Quan ile (Q–Q) plo s we e isualized by plo ing he dis ibu ion o ob ained s. expec ed genome-wide p- alues. Fo geno ype e ec o po en ial SNPs on li e size in each b eed, di e ences be ween means we e analyzed by he S uden ’s - es . The p<0.05 was conside ed s a is ically signi ican . All he esul s we e p esen ed as mean ±s anda d e o (SE). We implemen ed pai wise es s o linkage disequilib ium (LD) be ween he mos signi ican SNPs and hei lanking SNPs wi hin app oxima ely 1 Mb ups eam and downs eam using PLINK 1.07. Regional associa ion plo s we e gene a ed using he R package 3.2.2. Bioin o ma ics Analysis We anno a ed he genes associa ed wi h li e size in each b eed using he O. a ies assembly Oa _ .4.02. Fu he , we submi ed he genes o he DAVID (da abase o anno a ion, isualiza ion and in eg a ed disco e y) da abase3 o gene on ology (GO) en ichmen and pa hways analyses (Huang e al., 2009a,b). The p- alue o 0.1 and a leas wo genes om he inpu gene lis in he en iched ca ego y we e conside ed o he en iched GO e ms. Also, we in es iga ed he p o ein–p o ein in e ac ion ne wo k o he candida e genes using he STRING da abase e sion 10.5 (Szkla czyk e al., 2017). In addi ion, di e en ial exp essions o he candida e genes in a ious issues we e examined using he EMBL-EBI Exp ession A las da abase4(Pe yszak e al., 2016). RESULTS Popula ion Rela ionship and Di e en ia ion Pai wise FST alue a ied om 0.023 o 0.104 among he popula ions wi h he leas gene ic di e en ia ion obse ed be ween Wadi and Hu sheep b eeds (Supplemen a y Table S2). The NJ ee showed ha hese b eeds we e clus e ed in o wo majo g oups acco ding o hei Chinese and Eu opean o igins (Figu e 1B). A simila geog aphic pa e n was seen in he PCA analyses wi h he g ouping o Wadi and Hu sheep sepa a ed om he o he ou Eu opean b eeds (Supplemen a y Figu e S2). Genome-Wide Associa ion Analysis A e he quali y con ol, 508,444 SNPs and 114 indi iduals (91 cases s. 23 con ols) in Wadi, 506,031 SNPs and 80 indi iduals (66 cases s. 14 con ols) in Hu, 443,125 SNPs and 23 indi iduals (8 cases s. 15 con ols) in Icelandic, 492,165 SNPs and 37 indi iduals (28 cases s. 9 con ols) in Finnsheep, 465,794 SNPs and 38 indi iduals (29 cases s. 9 con ols) in Romano , 475,955 SNPs and 39 indi iduals (28 cases s. 11 con ols) in Texel sheep we e e ained in he wo king da ase o he GWAS. We did ind se e al animals ou lying he clus e s o cases, which migh cause biases in he associa ion analyses (Supplemen a y Figu e S3). We ha e epea ed he associa ion analyses wi hou hese animals, and ound he esul s a e e y simila . Thus, we did no exclude hese animals in he associa ion analyses due o he small sample size o he b eeds. The esul ing genomic in la ion ac o s we e equal o 1.07 in Wadi, 1.14 in Hu, 1.12 in Icelandic, 1.14 in Finnsheep, 1.10 in Romano , and 1.05 in 2h p://www.ncbi.nlm.nih.go /genome? e m=o is%20a ies 3h ps://da id.nci c .go / 4h ps://www.ebi.ac.uk/gxa/home/ F on ie s in Gene ics | www. on ie sin.o g 4Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 5 Xu e al. Gene ic Basis o Sheep Rep oduc ion FIGURE 3 | Manha an plo s o GWAS a e shown on (A) Wadi, (B) Hu, (C) Icelandic, (D) Finnsheep, (E) Romano and (F) Texel sheep. The 5% genome-wide signi ican h eshold alue is indica ed by a do ed line. The signi ican SNPs su ounding he genes p e iously epo ed o be associa ed wi h ep oduc ion a e anno a ed a he ch omosome-wise and genome-wide 5% signi icance a e he Bon e oni co ec ion. Texel sheep, sugges ing well-con olled popula ion s a i ica ions (Supplemen a y Figu e S4). In Wadi sheep, we de ec ed 59 and 8 SNPs a he ch omosome- wise and genome-wide (p<1.92 ×10−6) 5% signi icance a e he Bon e oni co ec ion, espec i ely (Figu e 3A and Supplemen a y Tables S3,S4). We obse ed a high le el o LD be ween he op signi ican SNP s416717560 and s421635584 loca ed in gene BMPR1B (Figu e 4A). Fo he SNP s416717560, a e age li e size o indi iduals wi h he G/G geno ype (n= 115, LS = 2.05 ±0.06) was signi ican ly (p<0.01) highe han ha o he ewes wi h he A/G (n= 15, LS = 1.47 ±0.16) geno ype (Figu e 5A). Also, we ound h ee addi ional signi ican F on ie s in Gene ics | www. on ie sin.o g 5Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 6 Xu e al. Gene ic Basis o Sheep Rep oduc ion FIGURE 4 | Plo s o egional associa ion esul s o he op signi ican SNP ( ed squa e) and hei nea SNPs in (A) Wadi, (B) Hu, (C) Icelandic, (D) Finnsheep, (E) Romano , and (F) Texel sheep. Di e en colo s ep esen he 2 alues o pai -wise LD es ima es. SNPs ( s429416173, s402803857, and s160917020) neighbo ing genes BMPR1B,FBN1, and MMP2 (Table 2 and Supplemen a y Table S3). In Hu sheep, we iden i ied 98 and 9 SNPs a he ch omosome- wise and genome-wide (p<2.18 ×10−6) 5% signi icance a e Bon e oni co ec ion (Figu e 3B and Supplemen a y Tables S3,S4). The egional plo showed ha he op signi ican SNPs s429755189 and s420460180 on ch omosome 17 we e in an LD block ha con ained gene GRIA2 (Figu e 4B). Fo he s429755189, a e age li e size o indi iduals wi h he geno ypes G/G (n= 38, LS = 1.99 ±0.07) and A/G (n= 52, LS = 1.94 ±0.06) we e signi ican ly (p<0.001) highe han ha o ewes wi h he geno ype A/A (n= 20, LS = 1.40 ±0.09) in he p esen popula ion (Figu e 5B). Among hese signi ican SNPs, 3 ( s406357666, s427436644 and s412185353) a e loca ed wi hin he genes SMAD1 and CTNNB1 (Table 2 and Supplemen a y Table S3). In Icelandic sheep, we ound 22 SNPs a he ch omosome- wise 5% signi icance a e he Bon e oni co ec ion (Figu e 3C and Supplemen a y Tables S3,S4). The op signi ican SNP s429836421 on ch omosome 3 was loca ed wi hin gene NCOA1 (Figu e 4C). Fo s429836421, a e age li e size o indi iduals wi h he A/G geno ype (n= 19, LS = 2.03 ±0.05) is signi ican ly F on ie s in Gene ics | www. on ie sin.o g 6Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 7 Xu e al. Gene ic Basis o Sheep Rep oduc ion FIGURE 5 | Geno ypic dis ibu ions o he op signi ican SNPs o he li e size (LS) pheno ype in (A) Wadi, (B) Hu, (C) Icelandic, (D) Finnsheep, (E) Romano , and (F) Texel sheep, espec i ely. The means LS we e calcula ed o a ious b eeds. Numbe o ewes pe g oup o geno ype is men ioned. Pai wise s a is ical compa isons be ween means o geno ype’s clades we e pe o med using S uden ’s - es . ∗p<0.05, ∗∗p<0.01, ∗∗∗ p<0.001. (p<0.05) highe han ha o he ewes wi h he geno ype A/A (n= 33, LS = 1.81 ±0.04) (Figu e 5C). In Finnsheep, we de ec ed 102 and 6 SNPs a he ch omosome- wise and genome-wide (p<3.64 ×10−6) 5% signi icance a e he Bon e oni co ec ion, espec i ely (Figu e 3D and Supplemen a y Tables S3,S4). The egional plo e ealed s ong LD be ween he op signi ican SNP s412280524 and i s neighbo ing SNPs s401960737 and s407751830 ha bo ed gene INHBB (Figu e 4D). Fo he SNP s412280524, li e size o ewes wi h he geno ype A/A (n= 40, LS = 2.84 ±0.09) is signi ican ly (p<0.001) highe han ha o he ewes wi h he geno ype A/G (n= 13, LS = 2.08 ±0.16) (Figu e 5D). Also, i e addi ional signi ican SNPs ( s160509574, s417444297, s404890873, s401746929, and s402764237) we e ound o be loca ed nea o genes FLT1,NF1,PTGS2, and PLCB3 (Table 2 and Supplemen a y Table S3). In Romano sheep, we iden i ied 77 and 2 SNPs a he ch omosome-wise and genome-wide (p<4.56 ×10−6) 5% signi icance a e he Bon e oni co ec ion (Figu e 3E and Supplemen a y Tables S3,S4). The op signi ican SNP s423810437 on ch omosome 7 was in he gene ESR2 (Figu e 4E). Fo s423810437, li e size o ewes wi h he geno ype A/A (n= 69, LS = 2.50 ±0.06) is signi ican ly (p<0.001) highe han ha o he ewes wi h he geno ype A/G (n= 8, LS = 1.79 ±0.18) (Figu e 5E). In Texel sheep, we obse ed 133 SNPs a he ch omosome- wise 5% signi icance a e he Bon e oni co ec ion (Figu e 3F and Supplemen a y Tables S3,S4). The egional plo showed ha he op signi ican SNPs s161146164 and s413776054 on ch omosome 16 we e in a s ong LD egion con aining one unc ional gene GHR (Figu e 4F). Fo s161146164, li e size o ewes wi h he geno ype A/A (n= 53, LS = 1.64 ±0.05) F on ie s in Gene ics | www. on ie sin.o g 7Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 8 Xu e al. Gene ic Basis o Sheep Rep oduc ion TABLE 2 | Genome-wide and ch omosome-wise signi ican SNPs and associa ed genes. Popula ion SNP Ch Posi ion (bp) MAF p-unadjus ed p-adjus ed Genes Loca ion Wadi s416717560∗6 29295803 0.07 3.65E-08 8.19E-09 BMPR1B130UTR s421635584∗6 29361782 0.05 4.36E-06 9.78E-07 BMPR1B1In on s429416173 6 29302788 0.2 7.55E-05 2.75E-05 BMPR1B1CDS s402803857 7 58598895 0.1 4.96E-05 2.93E-05 FBN11In on s160917020∗14 23133427 0.19 1.10E-06 3.71E-07 MMP2 Downs eam Hu s429755189∗17 41621298 0.43 1.94E-06 3.21E-07 GRIA21In on s420460180 17 41621269 0.29 8.50E-06 2.43E-06 GRIA21In on s406357666 17 12487861 0.19 1.40E-05 2.66E-05 SMAD11In on s427436644 19 13639996 0.32 7.69E-05 2.14E-05 CTNNB1 Downs eam s412185353 19 13641870 0.33 1.51E-04 4.49E-05 CTNNB1 Downs eam Icelandic s429836421 3 32030054 0.16 4.55E-05 3.63E-05 NCOA11In on Finnsheep s412280524∗2 184578329 0.09 2.62E-05 5.32E-07 INHBB Downs eam s401960737∗2 184579671 0.09 2.62E-05 5.32E-07 INHBB Downs eam s160509574 10 31933001 0.27 1.50E-05 4.71E-05 FLT11In on s417444297 11 18552961 0.11 4.20E-05 5.65E-05 NF1 Downs eam s404890873 12 65662842 0.05 1.87E-04 1.59E-05 PTGS2 Ups eam s401746929 21 41915064 0.08 1.85E-03 1.75E-04 PLCB3 Ups eam s402764237 21 41919836 0.08 1.85E-03 1.75E-04 PLCB3 Ups eam Romano s423810437∗7 73335157 0.07 1.65E-05 3.12E-06 ESR2150 lanking egion Texel s409969387 8 75353388 0.08 1.11E-03 1.21E-04 ESR1 In on s410595930 14 23645021 0.06 1.33E-04 1.46E-04 SPP11In on s401207152 14 25147418 0.06 1.33E-04 1.46E-04 MMP15 Downs eam s161146164 16 31834495 0.06 1.33E-04 9.11E-06 GHR1CDS s413776054 16 31834942 0.06 1.33E-04 9.11E-06 GHR CDS s426666828 16 31882869 0.18 1.88E-04 7.54E-05 GHR1In on s413148060 21 30950537 0.15 1.02E-04 4.17E-05 ETS1 Ups eam s405994606 21 31001548 0.15 1.02E-04 4.17E-05 ETS11In on s161612044 21 31009743 0.14 5.41E-04 1.01E-04 ETS11In on s412251543 21 31178275 0.1 4.01E-03 1.46E-04 ETS1/FLI1 Ups eam/Downs eam Fo genes he bes SNP o which is loca ed ou side o ups eam/downs eam 150 kb egion. Ch ., ch omosome; MAF, Mino Allele F equency. The p-unadjus ed co esponds o exac p o he Fishe ’s es . The p-adjus ed co esponds o he co ec ed signi icance o GWAS a e p inciple componen adjus men . The SNPs wi h symbol (∗) deno e ha bon e oni-co ec ed genome-wide signi ican SNPs. The genes wi h symbol (1) deno e ha he SNPs a e in agenic, o he wise hey a e he nea es genes ups eam and downs eam o he es ed SNPs. is signi ican ly (p<0.01) highe han ha o he ewes wi h he geno ype A/C (n= 6, LS = 1.15 ±0.14) (Figu e 5F). The wo mu a ions ( s161146164, Asn >His; s413776054, P o >Se ) cause he amino acid change in coding egion o he GHR gene. In addi ion, we ound eigh addi ional signi ican SNPs ( s426666828, s409969387, s410595930, s401207152, s413148060, s405994606, s161612044, and s412251543) su ounding genes ESR1,ETS1,FLI1,SPP1, and MMP15 (Table 2 and Supplemen a y Table S3). In addi ion o he sou ce b eed whe e he a ge SNPs ha e been de ec ed, we u he assessed geno ype e ec o he mos signi ican SNPs on li e size in he o he i e sheep b eeds. In gene al, geno ypes o he a ge SNPs did no show signi ican associa ion wi h inc eased li e size in he b eeds o he han he sou ce b eed (Supplemen a y Table S7). Ne e heless, we obse ed some excep ions. Fo example, he geno ype A/G o s429836421, which was iden i ied in Icelandic sheep, showed signi ican associa ions wi h inc eased li e size in bo h Icelandic and Hu sheep b eeds. Howe e , a lack o homozygo es o he SNPs such as he geno ype G/G o s412280524 in Finnsheep, G/G o s423810437 in Romano and C/C o s161146164 in Texel sheep could be because o low equency o he mu a ions and small sample size. Bioin o ma ics Analysis We ound signi ican ly (p<0.1) en iched GO e ms associa ed wi h ep oduc ion o he candida e genes. The GO clus e s we e p ima ily en iched in he ca ego ies o o a ian and oocy e de elopmen (PTGS2,BMPR1B,INHBB,CTNNB1,MMP2, MMP15,FBN1,GHR, and SPP1), phospholipase C ac i i y (FLT1 and ESR1), SMAD p o ein (INHBB and SMAD1) and BMP signaling (SMAD1 and BMPR1B) and posi i e egula ion o ansc ip ion (NCOA1,FLI1,ESR1,ESR2,CTNNB1,ETS1, and BMPR1B), all o which a e in ol ed in he olliculogenesis, ollicle g ow h and g anulosa cell p oli e a ion (Figu e 6 and Supplemen a y Table S5). Ano he ele an GO ca ego y was hindb ain de elopmen (SMAD1 and CTNNB1), which pa icipa ed in egula ing o ula ion (Bai d e al., 2006). In addi ion, we de ec ed 11 genes (i.e., PLCB3,ESR1,ESR2, F on ie s in Gene ics | www. on ie sin.o g 8Ap il 2018 | Volume 9 | A icle 118 gene-09-00118 Ap il 9, 2018 Time: 17:46 # 9 Xu e al. Gene ic Basis o Sheep Rep oduc ion FIGURE 6 | Gene on ology (GO) en ichmen s based on he unc ional genes su ounding he signi ican SNPs a he ch omosome-wise 5% le el. FIGURE 7 | P o ein–p o ein in e ac ion ne wo ks iden i ied by using STRING da abase. Each line indica ed known signaling pa hways and p o ein complexes. MMP2,NCOA1,CTNNB1,INHBB,SMAD1,BMPR1B,PTGS2, and GRIA2) in ol ed in es ogen, hy oid ho mone, TGF-be a, e og ade endocannabinoid and hippo signaling pa hways, and hese pa hways played impo an oles in egula ing ollicle g ow h and o ula ion in li es ock (Supplemen a y Table S5). Howe e , we obse ed di e en GO e ms o he candida e genes in di e en sheep b eeds. Fo example, I-SMAD binding we e en iched in Hu sheep, and ch oma in binding we e en iched in Texel sheep (Supplemen a y Table S6). In he gene ne wo k analysis, we obse ed ha 16 genes (i.e., BMPR1B,FBN1,MMP2, SMAD1,CTNNB1,GRIA2,NCOA1,FLT1,NF1,PTGS2,PLCB3, ESR2,ESR1,ETS1,SPP1, and GHR) showed p o ein–p o ein in e ac ions in he ne wo k (Figu e 7). Exp ession da a u he showed ha he genes BMPR1B,FBN1,MMP2,GRIA2,SMAD1, CTNNB1,NCOA1,NF1,FLT1,PTGS2,PLCB3,ESR2,ESR1,GHR, ETS1,MMP15,FLI1, and SPP1 we e ei he highly o mode a ely exp essed in ep oduc ion- ela ed issues such as o a y, u e ine ce ix, placen a, co pus lu eum, ce ebellum, pi ui a y gland o u e us in sheep (Figu e 8). Also, gene INHBB showed a high exp ession in o a y and u e us o Mus musculus5. DISCUSSION In his s udy, we conduc ed mul iple independen GWAS in di e en sheep b eeds o in es iga e he gene ic mechanisms unde lying he li e size in sheep. Coupled wi h popula ion ela ionship and bioin o ma ics analyses, he GWAS iden i ied di e en genes associa ed wi h he li e size in di e en b eeds and e ealed hei di e en ially gene ic egula ion mechanisms associa ed wi h ollicle g ow h and o ula ion in he ep oduc ion o ewes. The di e se biological pa hways iden i ied om he no el genes anno a ion play an impo an ole in ollicle g ow h and o ula ion o emales in di e en sheep b eeds (Figu e 9). The h ee genes iden i ied in Wadi sheep, BMPR1B,FBN1, and 5h ps://www.ebi.ac.uk/gxa/home/ F on ie s in Gene ics | www. on ie sin.o g 9Ap il 2018 | Volume 9 | A icle 118