Long‐term selection for litter size in swine results in shifts in allelic frequency in regions involved in reproductive processes

Functional genomics of reproduction in pigs: Are we there yet?

Reproductive failure is the main reason for culling females in swine herds and is both a financial and sustainability issue. Because reproductive traits are complex and lowly to moderately heritable, genomic selection within populations can achieve substantial genetic gain in reproductive efficiency. A better understanding of the physiological components affecting the expression of these traits will facilitate greater understanding of the genes affecting reproductive traits and is necessary to improve and optimize management strategies to maximize reproductive success of gilts and sows. Large-scale genotyping with single-nucleotide polymorphism (SNP) arrays are used for genome-wide association studies (GWAS) and have facilitated identification of positional candidate genes. Transcriptomic data can be used to weight SNP for GWAS and could lead to previously unidentified candidate genes. Resequencing and fine mapping of candidate genes are necessary to identify putative functional variants and some of these have been incorporated into new genotyping arrays. Sequence imputation and genotype by sequence are newer strategies that could reveal novel functional mutations. In this study, these approaches are discussed. Advantages and limitations are highlighted where additional research is needed.

Identifying selection signatures for litter size in Guangxi Bama Xiang pigs

Litter size is an important economic trait in pig production. However, the genetic mechanisms underlying varying litter size in Guangxi Bama Xiang pigs remain unknown. To identify selection signatures for litter size in Guangxi Bama Xiang pigs, we obtained 297 Illumina PorcineSNP50 BeadChip array data and the average born number (ABN) from parity one to nine in Guangxi Bama Xiang pigs. Fixation index (Fst) methods were used to identify the selection signature of the litter size, and three phenotypic gradient differential population pairs (according to the ABN) in individuals were used to reduce the false positives of signature selections. Single nucleotide polymorphisms (SNPs) were identified in the VEGFA promoter and exons. The general linear model was used to analyse the differences in distinct genotypes after they were typed using three-round multiplex PCR technology. Finally, the transcriptome factor and CpG island in the VEGFA promoter were predicted. A total of 328, 328 and 317 significant loci were identified in the 1st, 2nd and 3rd population pairs, respectively. After removing the false positives, 25 SNPs were defined as the selection signatures in relation to litter size. Ten (VEGFA, USP49, USP25, SRPK1, SLC26A8, RPL10A, PPARD, MAPK14, HMGA1 and CHRDL2) out of 52 genes in the selection regions were annotated as the candidate genes of litter size, respectively, VEGFA. There were no SNPs in the VEGFA exon region, but we obtained three SNPs (rs786889605, rs343769603 and rs323942424) in the VEGFA promoter regions. The ABN in CC was significantly higher than that in TT in rs786889605, and the ABN in TT was significantly lower than that in GG in rs323942424. Meanwhile, the mutation of the VEGFA promoter result in the loss of Sp1 and NF-1 and the formation of Oct-1. In summary, we obtained ten candidate genes, and two mutations in the VEGFA promoter that could be important potential molecular biomarkers for litter size in Bama Xiang pigs.

Effects of restricting energy during the gilt development period on growth and reproduction of lines differing in lean growth rate: Responses in reproductive performance and longevity

Longevity and reproductive performance are economically important traits in the swine industry that are largely influenced by nutrition and other environmental factors. Reproductive performance and longevity through 4 parities was assessed in gilts of 2 genetic lines developed on ad libitum access to feed or restricted to 75% of ad libitum intake. A total of 661 gilts were used in a 2 x 2 factorial with half of the gilts allocated to an ad libitum diet (AL; n = 330), while the other half were energy restricted by 25% (R; n = 331) from 123 to 235 d of age. All gilts were sired by an industry maternal line. Dams of the gilts were from either a Large White by Landrace industry maternal line or Nebraska Selection Line 45X, producing gilts designated as W x L (n = 355) and L45X (n = 306), respectively. Daily estrus detection began at 140 d of age to obtain age at puberty (AP). Gilts (n = 510) were mated on their second or later estrus, beginning at 240 d of age. Sow weight and backfat were recorded at 110 d of gestation and weaning of each parity. Number of live-born, stillborn, and mummified pigs per litter and piglet birth and weaning weights were recorded through 4 parities. More L45X than W x L and more AL than R gilts reached puberty by 230 d of age (P < 0.01). Dietary treatment did not affect probability to produce parities 1 to 4 or any litter trait analyzed. The L45X females tended to be more likely to produce parities 1 (P < 0.08) and 3 (P < 0.06), while W x L had heavier litters at birth (P < 0.01) and weaning (P = 0.01). Treatment by parity interactions (P < 0.01) existed for weight and backfat prior to farrowing and backfat at weaning, and weight at weaning exhibited a line by treatment by parity interaction (P = 0.04) as R sows had lower weights and backfats in earlier parities, but caught up to AL sows in later parities. A treatment by parity interaction (P < 0.01) was also present for backfat loss from farrowing to weaning as R gilts lost less backfat than AL in parities 1 and 2, but more in parities 3 and 4. No significant differences were detected between lines or treatments for lifetime production traits. The populations of pigs and data presented here provide a framework for a diverse array of further studies. Alternative approaches to restrict energy have been assessed in addition to methods of marker-assisted and genomic selection for improvement of litter size and sow longevity.

Potential functional variants in AHR signaling pathways are associated with age at puberty in swine

Puberty in female pigs is defined as age at first estrus and gilts that have an earlier age at puberty are more likely to have greater lifetime productivity. Because age at puberty is predictive for sow longevity and lifetime productivity, but not routinely measured in commercial herds, it would be beneficial to use genomic or marker-assisted selection to improve these traits. A GWAS at the US Meat Animal Research Center (USMARC) identified several loci associated with age at puberty in pigs. Candidate genes in these regions were scanned for potential functional variants using sequence information from the USMARC swine population founder animals and public databases. In total, 135 variants (SNP and insertion/deletions) in 39 genes were genotyped in 1284 phenotyped animals from a validation population sired by Landrace and Yorkshire industry semen using the Agena MassArray system. Twelve variants in eight genes were associated with age at puberty (P < 0.005) with estimated additive SNP effects ranging from 1.6 to 5.3 days. Nine of these variants were non-synonymous coding changes in AHR, CYP1A2, OR2M4, SDCCAG8, TBC1D1 and ZNF608, two variants were deletions of one and four codons in aryl hydrocarbon receptor, AHR, and the most significant SNP was near an acceptor splice site in the acetyl-CoA carboxylase alpha, ACACA. Several of the loci identified have a physiological and a genetic role in sexual maturation in humans and other animals and are involved in AHR-mediated pathways. Further functional validation of these variants could identify causative mutations that influence age at puberty in gilts and possibly sow lifetime productivity.

Fine mapping genetic variants associated with age at puberty and sow fertility using SowPro90 genotyping array

Sow fertility traits, such as litter size and the number of lifetime parities produced (reproductive longevity), are economically important. Selection for these traits is difficult because they are lowly heritable and expressed late in life. Age at puberty (AP) is an early indicator of reproductive longevity. Here, we utilized a custom Affymetrix single-nucleotide polymorphisms (SNPs) array (SowPro90) enriched with positional candidate genetic variants for AP and a haplotype-based genome-wide association study to fine map the genetic sources associated with AP and other fertility traits in research (University of Nebraska-Lincoln [UNL]) and commercial sow populations. Five major quantitative trait loci (QTL) located on four Sus scrofa chromosomes (SSC2, SSC7, SSC14, and SSC18) were discovered for AP in the UNL population. Negative correlations (r = -0.96 to -0.10; P < 0.0001) were observed at each QTL between genomic estimated breeding values for AP and reproductive longevity measured as lifetime number of parities (LTNP). Some of the SNPs discovered in the major QTL regions for AP were located in candidate genes with fertility-associated gene ontologies (e.g., P2RX3, NR2F2, OAS1, and PTPN11). These SNPs showed significant (P < 0.05) or suggestive (P < 0.15) associations with AP, reproductive longevity, and litter size traits in the UNL population and litter size traits in the commercial sows. For example, in the UNL population, when the number of favorable alleles of an SNP located in the 3' untranslated region of PTPN11 (SSC14) increased, AP decreased (P < 0.0001), while LTNP increased (P < 0.10). Additionally, a suggestive difference in the observed NR2F2 isoforms usage was hypothesized to be the source of the QTL for puberty onset mapped on SSC7. It will be beneficial to further characterize these candidate SNPs and genes to understand their impact on protein sequence and function, gene expression, splicing process, and how these changes affect the phenotypic variation of fertility traits.

Association between single-nucleotide polymorphisms within candidate genes and fertility in Landrace and Duroc pigs

Finding effective predictors of traits related to boar fertility is essential for increasing the efficiency of artificial insemination systems in pig breeding. The objective of this study was to find associations between single-nucleotide polymorphisms (SNPs) within candidate genes and fertility in the breeds Landrace and Duroc. Animals with breeding values for total number of piglets born, were re-sequenced for exonic regions of 14 candidate genes related to male and female fertility using samples from 16 Landrace boars and 16 Duroc boars (four with high and four with low breeding value of total number of piglets born for each breed for male fertility, and the same for female fertility) to detect genetic variants. Genotyping for the detected SNPs was done in 619 Landrace boars and 513 Duroc boars. Two SNPs in BMPR1 and one SNP in COX-2 were found significantly associated with the total number of piglets born in Landrace. In Duroc, two SNPs in PLC_z, one SNP in VWF and one SNP in ZP3 were found significantly associated with total number of piglets born. These SNPs explained between 0.27% and 1.18% of the genetic variance. These effects are too low for being used directly for selection purposes but can be of interest in SNP-panels used for genomic selection.

Evaluation of genotype quality parameters for SowPro90, a new genotyping array for swine1

Understanding early predictors of sow fertility has the potential to improve genomic predictions. A custom SNP array (SowPro90 produced by Affymetrix) was developed to include genetic variants overlapping quantitative trait loci for age at puberty, one of the earliest indicators of sow fertility, as well as variants related to innate and adaptive immunity. The polymorphisms included in the custom genotyping array were identified using multiple genomic approaches including deep genomic and transcriptomic sequencing and genome-wide associations. Animals from research and commercial populations (n = 2,586) were genotyped for 103,476 SNPs included in SowPro90. To assess the quality of data generated, genotype concordance was evaluated between the SowPro90 and Porcine SNP60 BeadArray using a subset of common SNP (n = 44,708) and animals (n = 277). The mean genotype concordance rate per SNP was 98.4%. Differences in distribution of data quality were observed between the platforms indicating the need for platform specific thresholds for quality parameters. The optimal thresholds for SowPro90 (≥97% SNP and ≥93% sample call rate) were obtained by analyzing the data quality distribution and genotype concordance per SNP across platforms. At ≥97% SNP call rate, there were 42,151 SNPs (94.3%) retained with a mean genotype concordance of 98.6% across platforms. Similarly, ≥94% SNPs and ≥85% sample call rates were established as thresholds for Porcine SNP60 BeadArray. At ≥94% SNPs call rate, there were 41,043 SNPs (91.8%) retained with a mean genotype concordance of 98.6% across platforms. Final evaluation of SowPro90 array content (n = 103,476) at ≥97% SNPs and ≥93% sample call rates allowed retention of 89,040 SNPs (86%) for downstream analysis. The findings and strategy for quality control could be helpful in identifying consistent, high-quality genotypes for genomic evaluations, especially when integrating genotype data from different platforms.

A genome-wide association study for gestation length in swine

Selection for increased litter size in swine has potentially resulted in a correlated increase in preweaning mortality. Additional selection criteria should be considered when selecting for increased litter size to account for associated decreases in piglet quality, specifically piglet survival, initial weight and growth. Traits such as gestation length (GL), which have been associated with piglet performance, could be utilized to improve piglet development and survivability. The objective of this study was to conduct a genome-wide association study to identify genomic regions associated with GL in differing parities in swine (n = 831) from the University of Nebraska-Lincoln reproductive longevity project. Gestation length was calculated as the number of days between last insemination administered and farrowing. Sows were genotyped with the Illumina SNP60 BeadArray, and the data were analyzed using Bayesian mixture models for GL at parity 1, 2, 3 and 4 (GL1, GL2, GL3 and GL4 respectively). Means (SD) for GL1-GL4 were 113 (1.4), 114 (1.2), 114 (1.3) and 115 (1.2) respectively. Posterior mean heritability estimates (PSD) for GL1, GL2, GL3 and GL4 were 0.33 (0.06), 0.34 (0.07), 0.32 (0.08) and 0.20 (0.08) respectively. Rank correlations between genomic estimated breeding values between GL1 and GL2, GL3 and GL4 respectively were moderate: 0.67, 0.65 and 0.60. The top SNP (ASGA0017859, SSC4, 7.8 Mb), located in the top common genomic region associated with GL1, GL2 and GL3, was associated with a difference of 1.1 days in GL1 between homozygote genotypes (P < 0.0001). The results of this study suggest that GL is a largely polygenic trait with relatively minor contributions from multiple genomic regions.

Longitudinal analysis of weight showed little relationship with age at puberty in gilts

Continued selection for increased gilt body weight could negatively impact selection for age at puberty (AP) in gilts. The purpose of this study was to compare the genetic potential for growth to that for reducing age of puberty in swine. Females utilized (n = 1,079) were produced over a 6-yr period from a population developed to determine the impact of energy restrictions and genetic influences on sow development and longevity. From 120 to 235 d of age, BW was collected every 14 d and attainment of puberty tested. Age at puberty was defined as the first observed standing estrus in the presence of a mature boar. All females were genotyped with the Porcine SNP60K BeadChip and genotypes were used to construct a genomic relationship matrix. Univariate (AP), repeatability (BW), and random regression (BW; RR) models were fitted. Univariate analysis included the fixed effects of contemporary group (CG) and age at first boar exposure, and random direct additive and common litter effects. Repeatability analysis included the fixed effects of CG and random effects of direct additive, common litter, and permanent environmental (PE) effects. Random regression analysis included fixed effects of CG, and random direct additive, common litter, and PE effects. Proportion of phenotypic variation due to direct additive and common litter variance for AP were 0.33 and 0.06, respectively. Proportion of phenotypic variation due to direct additive, common litter, and PE variance estimated from the repeatability model for BW were 0.26, 0.11, and 0.40, respectively. Proportion of phenotypic variation due to direct additive, common litter, and PE variance estimated from the RR for BW ranged from (mean) 0.19 to 0.30 (0.27), 0.08 to 0.31 (0.19), and 0.42 to 0.62 (0.50), respectively. Direct additive correlations between test days for BW from RR ranged from 0.30 to 0.99. Rank correlations between estimated breeding values (EBV) for AP and BW from the RR were near zero across all age points ranging from -0.03 to 0.09. Rank correlations were higher (0.63) when BW was considered at the age at which puberty was reached. Correlations between AP and BW EBV from the repeatability model were low (-0.11). Growth appears to be less related to AP than previously reported, suggesting the need to either directly measure AP or investigate alternative indicator traits. Selecting gilts with most desirable BW EBV alone would not result in improvement in AP, at least in the current population.