WANG Kai,WU Ping-xian,CHEN De-juan,ZHOU Jie,YANG Xi-di,JlANG An-an,MA Ji-deng,TANG Qian-zi,XlAO Wei-hang,JlANG Yan-zhi,ZHU Li,QlU Xiao-tian,Ll Ming-zhou,Ll Xue-wei,TANG Guo-qing

1 Farm Animal Genetic Resources Exploration and Innovation Key Laboratory of Sichuan Province,Sichuan Agricultural University,Chengdu 611130,P.R.China

2 College of Life Science,Sichuan Agricultural University,Ya’an 625014,P.R.China

3 National Animal Husbandry Service,Beijing 100125,P.R.China

Abstract We performed a genome-wide scan to detect selection signatures that showed evidence of positive selection in the domestication process by re-sequencing the whole genomes of Landrace and Yorkshire pigs.Fifteen annotated elements with 13 associated genes were identified using the Z-transformed FST (Z(FST)) method,and 208 annotated elements with 140 associated genes were identified using the Z-transformed heterozygosity (ZHp) method.The functional analysis and the results of previous studies showed that most of the candidate genes were associated with basic metabolism,disease resistance,cellular processes,and biochemical signals,and several were related to body morphology and organs.They included PPP3CA,which plays an essential role in the transduction of intracellular Ca2+-mediated signals,and WWTR1,which plays a pivotal role in organ size control and tumor suppression.These results suggest that genes associated with body morphology were subject to selection pressure during domestication,whereas genes involved in basic metabolism and disease resistance were subject to selection during artificial breeding.Our findings provide new insights into the potential genetic variation of phenotypic diversity in different pig breeds and will help to better understand the selection effects of modern breeding in Landrace and Yorkshire pigs.

Keywords:pig,variation,whole-genome sequence,selection signature,phenotypic trait

1.lntroduction

Animal domestication is important in human evolutional history because it was a key factor that allowed the transition from gathering and hunting to a settled lifestyle.Pig is an important food source for humans,so their domestication is of great interest.Pigs originated from local wild boar,Sus scrofa,mainly in Asia and Europe,approximately 10 000 years ago (Giuffraet al.2000;Larsonet al.2005,2007;Groenenet al.2012).Since then,pigs have experienced the combined effects of natural selection and intense artificial selection,which have resulted in significant phenotypic diversity in appearance,growth,reproduction,local fitness,and behavior (Anderssonand Georges 2004;Darwin 2010).Genome-wide scanning to identify selection signatures is a widely used method that has been applied to explore the genetic variation underlying the phenotypic diversity of pig breeds.Selection signatures have been detected for candidate genes related to morphological variations such as body size (Rubinet al.2012),ear shape (Wilkinsonet al.2013),and coat color (Maet al.2015),and production traits such as fertility (Leeet al.2017),and growth traits such as growth rate (Wanget al.2018) and development of tissues and organs (Yanget al.2014).Several methods for genomic signature detection have been used,including ZHp,Z(FST) values,and high-FST outliers.The FST value is a measure of population differentiation that provides an estimate of genetic differentiation between populations.A locus with significantly high FST values compared with other loci indicates that the locus may have undergone positive selection (Nielsen 2005).

Landrace and Yorkshire pigs are important commercial pig breeds.Long term breeding and artificial selection have shaped the specific genetic variation of these two breeds.As a result,they have different phenotypic traits,including morphological,growth,and reproduction traits.We hypothesized that Landrace and Yorkshire pigs would have different selected footprints on the pig genome that reflects their specific phenotypic characteristics due to the strong artificial selection with different breeding goals between two breeds in long term breeding process.Thus,the objective of this study was to detect the differential selected footprints between Landrace and Yorkshire pigs by genome-wide scan of selection signatures using the Z(FST) and ZHp methods.

2.Materials and methods

2.1.DNA extraction and whole-genome re-sequencing

The pigs used in this study were purchased from the Fujian Yichun Agricultural Development Co.,Ltd.(Nanping,China).Genomic DNA was extracted from ear tissue of 20 Landrace and 40 Yorkshire pigs (Table 1).After the ear tissue was collected,the pigs remained on the farm.Genomic DNA from the 60 animals was prepared from the ear tissue for a single individual in this study.We performed 150-bp pairedend sequencing using an Illumina HiSeq 4000 System according to the manufacturer’s protocols.Paired-end reads were aligned against the Sscrofa 11.1 reference genome using BWA v0.7.17 (http://bio-bwa.sourceforge.net/) with the default parameters (Liand Durbin 2009a).SAMtools v1.7 (http://www.htslib.org/) was used to convertaligned reads into BAM files and to remove duplicate reads(Liand Durbin 2009b).GATK v4.0.2.1 (https://software.broadinstitute.org/gatk/) (Van der Auweraet al.2013) was used for downstream processing and variant calling for single-nucleotide polymorphisms (SNPs).The detected SNPs were annotated with SnpEff v4.3T (http://snpeff.sourceforge.net/) (Cingolaniet al.2012).Quality control was determined using VCFtools v0.1.17 (http://vcftools.sourceforge.net/) from a VCF file for subsequent analysis(Daneceket al.2011).Strict criteria were applied to obtain high-quality data as follows:(1) remove SNPs with minor allele frequency ＜0.05;(2) discard SNPs with call rate ＜0.9;(3) remove SNPs with Hardy-Weinberg disequilibrium test＜1e−6;(4) remove SNPs with unknown location and SNPs on sex chromosomes;and (5) remove SNPs with mean depth ＜6.The SNP data in this study is available in Figshare with doi:10.6084/m9.figshare.11429046.

Table 1 The information of 60 pigs in this study

2.2.Population genetic analysis

Principal component analysis was performed with GCTA Software using the whole-genome SNPs from each pig(Yanget al.2011).A phylogenetic neighbor-joining tree was constructed using PHYLIP v3.697 (http://evolution.genetics.washington.edu/phylip.html) (Felsenstein 2004)and Figtree v1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/)(Bandeltet al.1999) based on a pairwise identity-by-state distance matrix calculated from the genome-wide SNPs using PLINK v1.9 (http://www.cog-genomics.org/plink2)(Purcellet al.2007).

2.3.Selection signature statistical analysis

Identification of selective sweeps was performed using the approach of Rubinet al.(2012).Briefly,a slidingwindow (40-kb window with 20-kb step size) was applied to determine FST and heterozygosity.Sliding windows with ＜10 SNPs were removed.First,the FST value for each SNP was calculated using VCFtools (v 0.1.17).Next,the expected heterozygosity (Hp) within each window was calculated using an in-house R script.For each SNP,the numbers of reads corresponding to the most (nMAJ) and least(nMIN) abundant alleles for each window in each breed were used to calculate a pooled heterozygosity (Hp) score as:

where,∑nMAJand ∑nMINare the sums ofnMAJandnMINfor all the SNPs in the windows.IndividualHpvalues were then Z-transformed as follows:

Then,we identified candidate genes on the basis of the 40-kb windows using Ensemble BioMart (http://www.biomart.org) (Kinsellaet al.2011).Genes that partially or completely spanned the window regions were regarded as putative candidate genes under positive selection.

2.4.Candidate gene annotation

Function annotation of the candidate genes identified by FST was performed using the Gene Ontology (GO) (http://geneontology.org/) and KEGG pathway (Kyoto encyclopedia of genes and genomes) (http://www.genome.jp/kegg/pathway.html) analysis tools in DAVID (The Database for Annotation,Visualization and Integrated Discovery) (http://david.ncifcrf.gov/).

3.Results

3.1.Whole-genome sequencing of Landrace and Yorkshire pigs

We selected 20 Landrace and 40 Yorkshire pigs and performed whole-genome sequencing.We obtained a total of 20.17 billion 150-bp paired-end reads with average sequencing depth of 17.11× (Appendix A).A total of 14 445 656 SNPs was identified in the 60 pig genomes following stringent quality control criteria (Table 2).Among them,9 356 051 SNPs were intergenic,4 551 864 were intronic,and 111 830 were exonic.

Table 2 Summary of variants identified from the breed-pair used in this study

3.2.Population genetic structure

The whole-genome sequencing data were used to obtain a wide view of the genetic relationships among the pigs.The first two axes of the principal component analysis(PC1 and PC2) showed the genetic diversity in the two breeds (Fig.1-A).PC1 separated the Landrace pigs from the Yorkshire pigs explaining 11.7% of the total variation.Besides,PC2 separated the Yorkshire pigs into two clusters,implying there may be other lineages in the Yorkshire pigs.Analyses based on pairwise identity-by-state distance using the neighbor-joining tree also identified the two distinct groups (Fig.1-B).As expected,the results showed clear population clusters with respect to each breed.

Fig.1 The principal component analysis results and neighbor-joining tree of Landrace pigs and Yorkshire pigs based on genomewide single-nucleotide polymorphisms (SNP).A,principal component analysis results based on genome-wide SNP.The red represents the Landrace pigs and the blue represents the Yorkshire pigs.B,neighbor-joining tree of Landrace pigs and Yorkshire pigs based on pairwise identity-by-state (IBS) distance calculated from genome-wide SNPs.The red line represents the Landrace pigs and the blue line represents the Yorkshire pigs.

3.3.Selection sweep analysis

To identify selection signatures in the pig genomes,we first performed an FST analysis using the SNPs on autosomes and calculated the Hp values within the two populations.We then Z-transformed the Hp (ZHp) and FST (Z(FST)) values and defined potential selective regions as having an ZHp value ＜−2.8 and Z(FST) value ＞5 following a similar principle set in a previous study (Zhuet al.2017).Using this strategy,we identified 17 and 363 windows as outliers according to the empirical distribution for Z(FST) (Fig.2) and ZHp (Fig.3),respectively.However,the genomic regions identified by the two methods did not coincide.Besides,15 annotated elements were found based on Z(FST) outliers,including 13 genes with associated names,and 208 annotated elements with 140 genes were found based on ZHp outliers (Appendix B).The distribution of the 40-kb window on the autosomal genomes is shown in Fig.4.

Fig.2 Genome-wide distributions of Z-transformed FST (Z(FST)) value alone pig autosomes 1–18.The blue line indicated the cut-off threshold (Z(FST)＞5).

Fig.3 Genome-wide distributions of Z-transformed heterozygosity (ZHp) value alone pig autosomes 1–18.The blue line indicated the cut-off threshold (ZHp＜–2.8).

Fig.4 The bubble diagram of Gene Ontology (GO) terms.The X-axis represents the P-value of genes enriched in the corresponding GO terms.The Y-axis represents the GO term to which the genes enriched.The shape of bubble represents the classification of GO terms.The color of bubble represents the log transformation of P-value.

3.4.Candidate genes and functional analysis

We then identified candidate genes on the basis of the 40-kb windows using Ensemble BioMart.The candidate genes identified by Z(FST) outliers are listed in Table 3.The G protein subunit geneGNAI3was found in Chr4:110.54–110.58 (Z(FST)=5.448945).Two candidate genes,UGP2andVPS54,were found in Chr3:77.94–77.98(Z(FST)=5.332162).UGP2encodes a UDP-glucose pyrophosphorylase that plays a crucial role in cellular metabolic pathways as a glucosyl donor (Pengand Chang 1993) and also was shown to be associated with muscle formation in pigs (Yuet al.2014).KCNMA1,which encodes a channel protein that plays a key role in controlling excitability in many systems,was found in Chr14:79.32–79.36 (Z(FST)=5.310708).Four candidate genes,BZW1,ENSSSCG00000016096,NIF3L1,andORC2,were found in Chr15:104.58–104.62 (Z(FST)=5.251026).RAB3C,which is associated with protein transport,was found in Chr16:37.64–37.68 (Z(FST)=5.182907).TLL1,whichwas found in Chr8:42.94–42.98 (Z(FST)=5.161398),encodes a protein that plays a central role in formation of the mammalian heart and interventricular septum in mammals (Clarket al.1999).

Table 3 Summary of candidate genes which locate in the significant 40 kb-windows by Z-transformed FST (Z(FST))

A total of 208 annotated elements with 140 genes were identified by ZHp outliers.In this study,we focused on the top 10 candidate genes (Table 4).We found they were distributed mainly on Chr1,Chr7,Chr8,Chr15,and Chr18,and were related to a variety of cellular and biochemical processes.Among them,PPP3CAon Chr8 encodes a protein that plays an essential role in the transduction of intracellular Ca2+-mediated signals.A genome-wide association study of intramuscular fat traits in the Italian Large White pig breed identifiedPPP3CAas the gene involved in the oxidative metabolism of skeletal muscle(Davoliet al.2015).The authors suggested thatPPP3CAmay be an interesting candidate gene for intramuscular fat content in pigs.Two other genes on Chr8,PTPN13andMAPK10,were found in the Chr8:132.24–132.28 region.ThePTPN13is involved in signaling pathways that regulate a variety of cellular processes,MAPK10acts as integration points for multiple biochemical signals.

We performed GO (Fig.4) and KEGG pathway (Fig.5)analyses to functionally annotate the candidate genes.A total of 13 GO terms and 11 KEGG pathways were found to be enriched in multiple signaling and signal transduction pathways and a wide variety of other cellular and biochemical processes.

4.Discussion

Over the past 300 years,intense artificial selection for desirable traits have led to the development of commercial pig breeds with high production performance,especially the Landrace and Yorkshire pigs.In this study,we implemented a genome wide scan of selection signatures for Landrace and Yorkshire pigs using Z(FST) and ZHp methods.We identified 17 differential outliers between two breeds according to the empirical distribution for Z(FST) and then we found 15 annotated elements with 13 genes based on these outliers,such as theUGP2,RAB3C,andTLL1genes.Besides,we identified 363 differential outliers between two breeds using ZHp method.Based on these outliers,we found 208 annotated elements with 140 genes,including thePPP3CA,PTPN13,andMAPK10genes.These results confirmed our hypothesis that Landrace and Yorkshire pigs have distinctly different selected footprints on the pig genome due to long term artificial selection.These identifieddifferential genomic regions or candidate genes will be given more attentions in the genetic analysis of complex traits for Landrace and Yorkshire pigs.

Table 4 Summary of top 10 candidate genes which locate in the significant 40 kb-windows by Z-transformed heterozygosity (ZHp)value

Fig.5 The bubble diagram of the Kyoto encyclopedia of genes and genomes (KEGG) pathway terms.The X-axis represents the P-value of genes enriched in the corresponding KEGG pathway terms.The Y-axis represents the KEGG pathway term to which the genes enriched.The color of bubble represents the log transformation of P-value.

In this study,we conducted genome-wide scan for selection signatures using 20 Landrace (14 males and 6 females) and 40 Yorkshire pigs (13 males and 27 females).Since sex chromosomes were not considered in this study,the unbalanced sex ratio was not the factor in the results.However,the small sample size was our limitation,which might affect the effect of our detection of selection signatures.Although we had increased the number of Yorkshire pigs to some extent,we acknowledged that the unbalanced population ratio between Landrace and Yorkshire pigs may still have an impact on our results.The relatively small number of Landrace pigs would affect the detection of the selection signatures between the two breeds.

However,the genomic regions identified by Z(FST) and ZHp did not overlap.Several possible explanations could account for this result.These two methods are based on different principles;FST is a detection method based on population differentiation,whereas Hp is a detection method based on site frequency spectrum.Further,strict screening criteria were used to obtain high-quality data,so less strict criteria may give better results.Despite these limitations,the results partly explained the selection pressures that Landrace and Yorkshire pigs underwent in the progress of evolution.

The FST method is a classical approach for detecting selection signatures based on population differentiation.Under the condition of neutral evolution,the FST value is affected mainly by migration and genetic drift.However,when a mutation is subjected to artificial selection or natural selection,its increased frequency will increase the level of population differentiation,and the FST value will increase accordingly.Thus,the FST method is sensitive to selection signatures between very different populations.Conversely,the Hp test is a common approach that is based on the site frequency spectrum.Compared with the FST method,the Hp test is more effective for some artificial selected signatures considering the change of site frequency spectrum;however,the Hp test cannot exclude the effect of linkage disequilibrium.Because Landrace and Yorkshire pigs are commercial breeds with similar high production traits,the Hp test is more suitable than the FST method for detecting the selection signatures of these two breeds.

The skeletal morphology of domestic pig breeds differs from that of wild boar.Following domestication,significant changes occurred in body and cranial dimensions (Mason and Mason 1984),leading to big differences between the different breeds.Wilkinsonet al.(2013) found a region on Chr1 associated with cranial dimension in comparisons between pig breeds and wild boar.Landrace and Yorkshire pigs have fast growth and high lean meat rates,which makes them commercially important.Because of the strong effect of artificial selection,these two breeds show diversity in multiple phenotypic traits,including body morphology,indicating that genes associated with the development of body form had been subject to different selection.For example,Landrace pigs are longer and more symmetrical than Yorkshire pigs,and Landrace pigs are flat-eared whereas Yorkshire pigs are prick-eared.However,we found that most of the candidate genes were involved in basic metabolism,disease resistance,cellular processes,and biochemical signals,which may suggest that genes and regions associated with body morphology were subject to selection pressure during the process of domestication,rather than during artificial breeding.This idea is consistent with many previous results (Evinet al.2015,2017).

The genomic regions identified in this study did not overlap with those identified by Rubin and colleagues (Rubinet al.2012) using ZHp and Z(FST) approaches.In their study,multiple breeds were used to detect selection signatures,so more potential selective regions were identified.In addition,Rubin and colleagues identified candidates more likely to be associated with the domestication process and post-domestication pig development.By contrast,we used Landrace and Yorkshire pigs,which are quite similar,to detect selection signatures by ZHp and Z(FST) approaches.Because there is a lack of genetic information for wild boar and other breeds,it is difficult to detect selection signatures during domestication using a genetic differentiation method.

Selection signatures in Landrace and Yorkshire pigs have been studied by other researchers.Moonet al.(2015)used two complementary methods,population branch statistic and integrated haplotype score tests,to detect recent selective sweeps of Landrace and Yorkshire pigs,whereas Wanget al.(2014) used the relative extended haplotype homozygosity test to detect selection signatures of Landrace and Yorkshires pigs.The different statistical methods used in these and ours studies might explain the differences in the results.Nevertheless,these previous studies provide some insights into our research.We found that 37 of the genes identified in our study overlapped with genes identified in the study of Moonet al.(2015);four(WWTR1,KCNMA1,RAB3C,andTLL1) were identified by Z(FST),and the others were identified by ZHp.Additionally,eight genes (e.g.,LHX4,QSOX1,XRCC2) identified by ZHp in our study overlapped with genes identified in the study of Wanget al.(2014) by the relative extended haplotype homozygosity test.

Compared with the Rubinet al.(2012) study,we used different populations and the same methods (Z(FST)and ZHp) but with stricter criteria to detected selection signatures,which resulted in fewer outliers.Compared with the Moonet al.(2015) and Wanget al.(2014) studies,we used the same breeds (Landrace and Yorkshire) and different methods,which resulted in many different genes and a few overlapping genes.These results showed that the FST method based on population differentiation detects fewer selection signatures for breeds (Landrace and Yorkshire in this study) with small differences,while methods based on site frequency spectrum and linkage disequilibrium detected more selection signatures.

5.Conclusion

We performed a genome-wide scan to detect the differential footprints between Landrace and Yorkshire pigs based on whole-genome re-sequencing data.Fifteen annotated elements with 13 associated genes were identified using the Z(FST) method,and 208 annotated elements with 140 associated genes were identified using the ZHp method.Our results indicated that Landrace and Yorkshire pigs have distinctly different selected footprints on the pig genome due to long term artificial selection.These identified differential genomic regions or candidate genes will be given more attentions in the genetic analysis of complex traits for Landrace and Yorkshire pigs.

Acknowledgements

We appreciate the research materials provided by Fujian Yichun Agricultural Development Co.,Ltd.,China.All experimental procedures were performed in accordance with the Institutional Review Board (IRB14044) and the Institutional Animal Care and Use Committee of the Sichuan Agricultural University of China under permit number DKY-B20140302.The study was supported by the grants from the Sichuan Science and Technology Program,China (2020YFN0024),the earmarked fund for the China Agriculture Research System (CARS-35-01A),the National Key R&D Program of China (2018YFD0501204),the National Natural Science Foundation of China (C170102),and the Sichuan Innovation Team of Pig,China (sccxtd-2021-08).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Ethical approval

All applicable internaltional,national and institutional guidelines for the care and use of animals were followed.

Appendicesassociated with this paper are available on http://www.ChinaAgriSci.com/V2/En/appendix.htm