Impact of selection for uterine capacity on the placental transcriptome.

Abstract

Direct single trait selection for 11 generations resulted in a 1.6 pig advantage for uterine capacity (UC) while average birth and placental weights at term remained unchanged. Uterine capacity was defined as the total number of fully-formed pigs produced to term when ovulation rate was not limiting, using a unilateral hysterectomy-ovariectomy model. A serial slaughter experiment conducted throughout gestation determined the critical time period for the line difference in litter size was already established between d 25 and 45 of gestation and generated direct evidence of differential relative growth rates for placental tissues at these times. Timing of line differences in fetal survival as well as anecdotal evidence of tissue structural differences pointed to the developing placental tissue as a target of particular interest. Our objective was to gain insight into placental transcriptional changes during this critical stage of gestation and identify genetic loci impacted by quantitative selection for uterine capacity. Thirty gilts each from the UC and control (CO) lines were subjected to unilateral hysterectomyovariectomy at approximately 160 d of age and mated within line at approximately 280 d. Gilts were slaughtered at d 25, 30, or 40 of gestation. Fetal and placental tissues were obtained from each live embryo. Fetal liver samples were used to extract DNA and determine sex of each fetus by PCR. Two male and two female embryos closest to the litter mean for placental weight were chosen to represent each litter sampled (n 3 litters per line and time point combination). Placental tissues were pooled within litter and total RNA was extracted. Samples were labeled and hybridized to Affymetrix porcine array chips (n — 18) using the manufacturers suggested protocols. Signal intensities were normalized using GC content Robust Multi-array Average (GCRMA) on the probe level data. Filtering was based on perfect match intensities as implemented for Affymetrix Human arrays. Two-way ANOVA (two lines and three stages) was performed. Threshold values were set at a minimum of 1.5 fold difference and the false discovery rate was set to P < 0.05 (Benjamini and Hochberg algorithm). Lessstringent two-way comparisons (t-tests)were also conducted between lines within each gestation stage. GeneSifter® software web tools were utilized to conduct the analyses and generate the gene lists. An additional analysis was conducted to examine the potential for a bioinformatics method of identifying single feature polymorphism (SNP) targets between the two lines associated with the behavior of the expression data. A gene by gene linear mixed model of probe intensity variation from 11 probes per gene target was investigated to identify line x probe interactions. Log2-transformed perfect-match intensities for all observations were fit to a linear mixed model that broadly corrected for effects of breed and array. A gene-specific mixed model was fit to the normalized intensities (residuals from first model) accounting for fixed breed, probe, and breed-by-probe interaction effects, and a random array effect. Probes that produced statistical evidence (qvalue s 0.05 to account for multiple testing and Ifold change I a 2) for the breedby-probe interaction were identified as candidates containing SNP. The interaction was declared