Mechanisms and associated biomarkers of early embryo mortality in Holstein-Friesian cows

Abstract

Intensive genetic selection for milk yield has declined fertility trait levels that has led to infertility for over 50 years in Holstein-Friesian cows. A substantial contributor to infertility is embryo mortality, but the exact mechanism of how a pregnancy is poorly understood. Recent use of candidate single nucleotide polymorphisms (SNPs) in genetic panels has shown to improve genomic estimates of predicted transmitting abilities for fertility traits that are known to be of low heritability. The objectives of this dissertation were then divided into two chapters. The first chapter had as an objective to elucidate the transcriptomic responses of reproductive tissues (endometrium, peripheral blood mononuclear cells [PBMC] and corpus luteum [CL]) in normal compared to pregnancies with embryo mortality in lactating Holstein- Friesian cows based upon two follow up experiments (E1 and E2). At day 16, after artificial insemination (pregnant group) or of estrous cycle (non-pregnant group), reproductive tissues (endometrium, PBMC and CL [only for E2]), serum (E1) or plasma (E2) and uterine flushings were collected. Cows from pregnant group were re-classified based on embryo morphology and appearance [embryo mortality (EM) pregnancies had pink, red, restricted in elongation and (or) opaque conceptuses or normal (N) pregnancies had translucent and elongated conceptuses]. The main findings for this chapter were that: N conceptuses were longer compared to EM conceptuses. Interferon-tau (IFNT) protein concentrations in uterine flushings were greater in N compared to EM and NP in E1 but not for E2. Western blot analyses for interferon stimulated genes (ISG) 15 protein in endometrium from N cows were greater (E1 and E2) than EM and NP and EM tended to be greater than non-pregnant in E2. The RTqPCR for ISG15 mRNA levels in N endometrium were greater (E1 and E2) when compared to EM and NP endometrium but only tended to be greater in EM compared to NP endometrium in E2. Concentrations of progesterone in the radioimmunoassay were only significant on days 7 and 16 in E2. In RNA sequencing, IFNT mRNA were greater (E1 and E2) in N conceptuses compared to EM conceptuses. For E1 and E2, IPA revealed for EM compared to N conceptuses the key canonical pathways of T helper 1 (Th1) and Th2 to be up-regulated and are adaptive immune response that activated pro-inflammatory cytokines. Within EM compared to N endometrium, E1 had Th1 and Th2 pathways up-regulated while E2 identified differentially expressed genes (DEGs) that were up-regulated and associated with estradiol-mediated luteolytic action. Comparison of EM compared to N in PBMC were only significantly in E1 and had down-regulated DEGs associated with tissue growth, remodeling and/or development, cell cycle, conceptus implantation, essential mineral transporters, innate immune system and Th1 activation. The CL of E2 for embryo mortality compared to normal exhibited up-regulation of DEGs associated with inflammation, calcium sequestration/delivery, glucose- and estradiol-metabolism that may be involved in the luteolysis pathway. The second chapter focused on the identification and validation of candidate SNPs within pregnancies with early embryo mortality that were associated with inferior fertility traits. The RNA sequencing of conceptuses (normal and embryo mortality) in Holstein-Friesian (n=15) cows from the first chapter were used to conduct the SNP discovery phase. Validation of candidate SNPs and genotype to phenotype analysis were conducted in a different cohort of Holstein-Friesian cows (n=500) by collecting blood samples to be genotyped via a genotyping assay panel and collecting cow farms records. Further filtering of candidate SNPs involved removing those that were monomorphic and not in minor allele frequency and a quality control pipeline via pLink software. The main findings for this chapter were: a total of sixty-nine candidate SNPs were initially discovered but only twenty-three passed the quality control pipeline in pLink software. All candidate SNPs were found explain a higher amount of the R2 variation of each of the models and were in close proximity to SNP that were associated with quantitative trait loci of fertility traits. Out of the twenty-three candidate SNPs, seven (DSC2: age of cows at 1st calving were older with A allele; SREBF1 and UBD: cows took longer to conceive with T or G allele, respectively; UMPS and SREBF1: required longer time to their 1st artificial insemination with C allele; DECR1 and FASN: cows were less likely to become pregnant at 1st artificial insemination with C allele; SREBF1 and BOLA-DMB: cows were less likely to become pregnant at 150 days in milk with T allele) were significantly associated to fertility traits. It was also found that two candidate SNPs (DSC2 gene: 4 SNPs and 2 SNPs] were considered as TAG SNPs. Only two of the seven candidate SNPs had significant allele substitution effects where DSC2 in cows with G allele decreased in the age at 1st calving by 10 days and SREBF1 [rs41912290] in cows with the C allele decreased days to 1st artificial insemination by 5 days and more probability of becoming pregnant at 150 days in milk by 6%. In summary, the data and results described in this thesis describe mechanisms of why most pregnancies fail, while others succeed in purebred Holstein-Friesian cows. The EM embryos undergo a massive T helper response either as part of or a consequence of dying. These studies may lead to the development of future technologies to improve reproductive efficiency. In addition, the identified candidate SNPs could then be used to genetically screen young heifers to identify the most fertile females while also making progress in milk production. These genetic tools will aid farmers in making decisions of culling reproductively inefficient heifers and cows within a herd.