Background Cytoplasmic and nuclear maturation of oocytes, as well as interaction with the surrounding cumulus cells, are important features relevant to the acquisition of developmental competence. Methods Here, we utilized Brilliant cresyl blue (BCB) to distinguish cattle oocytes with low activity of the enzyme Glucose-6-Phosphate Dehydrogenase, and thus separated fully grown (BCB positive) oocytes from those in the growing phase (BCB negative). We then analyzed the developmental potential of these oocytes, mitochondrial DNA (mtDNA) copy number in single oocytes, and investigated the transcriptome of single oocytes and their surrounding cumulus cells of BCB positive versus BCB negative oocytes. Results The BCB positive oocytes were twice as likely to produce a blastocyst in vitro compared to BCB- oocytes (P < 0.01). We determined that BCB negative oocytes have 1.3-fold more mtDNA copies than BCB positive oocytes (P = 0.004). There was no differential transcript abundance of genes expressed in oocytes, however, 172 genes were identified in cumulus cells with differential transcript abundance (FDR < 0.05) based on the BCB staining of their oocyte. Co-expression analysis between oocytes and their surrounding cumulus cells revealed a subset of genes whose co-expression in BCB positive oocytes (n = 75) and their surrounding cumulus cells (n = 108) compose a unique profile of the cumulus-oocyte complex. Conclusions If oocytes transition from BCB negative to BCB positive, there is a greater likelihood of producing a blastocyst, and a reduction of mtDNA copies, but there is no systematic variation of transcript abundance. Cumulus cells present changes in transcript abundance, which reflects in a dynamic co-expression between the oocyte and cumulus cells.

Background: Infertility or subfertility is a critical barrier to sustainable cattle production, including in heifers. The development of heifers that do not produce a calf within an optimum window of time is a critical factor for the profitability and sustainability of the cattle industry. The early identification of heifers with optimum fertility using molecular phenotyping is a promising approach to improving sustainability in beef and dairy cattle production. Results: Using a high-density single nucleotide polymorphism (SNP) chip, we collected genotypic data from 575,053 SNPs. We also produced quantitative transcriptome data for 12,445 genes (12,105 protein-coding genes, 228 long non-coding RNAs, and 112 pseudogenes) and proteome data for 213 proteins. We identified two SNPs significantly associated with heifer fertility (rs110918927, chr12: 85648422, P = 6.7x10-7; and rs109366560, chr11:37666527, P = 2.6x10-5). We identified two genes with differential transcript abundance (eFDR ≤ 0.002) between the two groups (Fertile and Sub-Fertile): Adipocyte Plasma Membrane Associated Protein (APMAP, 1.16 greater abundance in the Fertile group) and Dynein Axonemal Intermediate Chain 7 (DNAI7, 1.23 greater abundance in the Sub-Fertile group). Our analysis revealed that the protein Alpha-ketoglutarate-dependent dioxygenase FTO was more abundant in the plasma collected from Fertile heifers relative to their Sub-Fertile counterparts (FDR < 0.05). Interestingly, two proteins did not reach the significance threshold in the model accounting for all samples (Apolipoprotein C-II, APOC2 (FDRglmm = 0.06) and Lymphocyte cytosolic protein 1, LCP1 (FDRglmm = 0.06)), but both proteins were less abundant in the plasma of Fertile Holstein heifers (P < 0.05). Lastly, an integrative analysis of the three datasets identified a series of features (SNPs, gene transcripts, and proteins) that can be useful for the discrimination of heifers based on their fertility. When all features were utilized together, 21 out of 22 heifers were classified correctly based on their fertility category. Conclusions: Our multi-omics analyses confirm the complex nature of female fertility. Very importantly, our results also highlight differences in the molecular profile of heifers associated with fertility that transcend the constraints of breed-specific genetic background.

Infertility or subfertility is a critical barrier to sustainable cattle production, including in heifers. The development of heifers that do not produce a calf within an optimum window of time is a critical factor for the profitability and sustainability of the cattle industry. In parallel, heifers are an excellent biomedical model for understanding the underlying etiology of infertility because well-nourished heifers can still be infertile, mostly because of inherent physiological and genetic causes. Using a high-density single nucleotide polymorphism (SNP) chip, we collected genotypic data, which were analyzed using an association analysis in PLINK with Fisher’s exact test. We also produced quantitative transcriptome data and proteome data. Transcriptome data were analyzed using the quasi-likelihood test followed by the Wald’s test, and the likelihood test and proteome data were analyzed using a generalized mixed model and Student’s t-test. We identified two SNPs significantly associated with heifer fertility (rs110918927, chr12: 85648422, P = 6.7 × 10−7; and rs109366560, chr11:37666527, P = 2.6 × 10−5). We identified two genes with differential transcript abundance (eFDR ≤ 0.002) between the two groups (Fertile and Sub-Fertile): Adipocyte Plasma Membrane Associated Protein (APMAP, 1.16 greater abundance in the Fertile group) and Dynein Axonemal Intermediate Chain 7 (DNAI7, 1.23 greater abundance in the Sub-Fertile group). Our analysis revealed that the protein Alpha-ketoglutarate-dependent dioxygenase FTO was more abundant in the plasma collected from Fertile heifers relative to their Sub-Fertile counterparts (FDR < 0.05). Lastly, an integrative analysis of the three datasets identified a series of molecular features (SNPs, gene transcripts, and proteins) that discriminated 21 out of 22 heifers correctly based on their fertility category. Our multi-omics analyses confirm the complex nature of female fertility. Very importantly, our results also highlight differences in the molecular profile of heifers associated with fertility that transcend the constraints of breed-specific genetic background.

No abstract

The objective of this study was to assess differences in reproductive performance of natural service and artificial insemination (AI) sired beef females based on pregnancy outcomes, age at first calving, and calving interval. Data were sourced from 8,938 cows sired by AI bulls and 3,320 cows sired by natural service bulls between 2010 and 2017. All cows were in a commercial Angus herd with 17 management units located throughout Virginia and represented spring and fall calving seasons. All calves were born to dams managed with estrus synchronization. Pregnancy was analyzed with generalized linear mixed models and other reproductive measures with linear mixed models in R. Six models were evaluated with the dependent variables of pregnancy status at the first diagnosis, pregnancy status at the second diagnosis, pregnancy type (AI or natural service) at the first diagnosis, pregnancy type at the second diagnosis, calving interval, and age at first calving. Independent variables differed by model but included sire type of the female (AI or natural service), pre-breeding measures of age, weight, and body condition score, postpartum interval, sex of the calf nursing the cow, and management group. No differences were observed between AI- and natural service-sired females based on pregnancy status at first and second pregnancy diagnosis (P &gt; 0.05). Sire type was only found to be significant for age at first calving (P &lt; 0.05) with AI-sired females being 26.6 ± 1.6 days older at their first calving, which was expected because AI-sired females were born early in the calving season making them older at breeding. Surprisingly, age and body condition score were not significant predictors of pregnancy (P &gt; 0.05). Body weight at breeding was not significant for pregnancy (P &gt; 0.05) but was significant for age at first calving (P &lt; 0.05). These data suggested that lighter heifers calved earlier which contradicts our original hypothesis. Overall, commercial Angus females sired by AI or natural service bulls had similar reproductive performance. Factors that were commonly associated with reproductive success were not significant in this commercial Angus herd managed with estrus synchronization. Given the size of these data, the importance of body condition, age, and weight should be reassessed in modern genetics and management practices.

Background: A gap currently exists between genetic variants and the underlying cell and tissue biology of a trait, and expression quantitative trait loci (eQTL) studies provide important information to help close that gap. However, two concerns that arise with eQTL analyses using RNA-sequencing data are the normalization of data across samples and the data not following a normal distribution. Multiple pipelines have been suggested to address this. For instance, the most recent analysis of the human and farm Genotype-Tissue Expression (GTEx) project proposes using trimmed means of M-values (TMM) to normalize the data followed by an inverse normal transformation. Results: In this study, we reasoned that eQTL analysis could be carried out using the same framework used for differential gene expression (DGE), which uses a negative binomial model, a statistical test feasible for count data. Using the GTEx framework, we identified 38 significant eQTLs (P<5x10-8) following the ANOVA model and 15 significant eQTLs (P<5x10-8) following the additive model. Using a differential gene expression framework, we identified 2,471 and nine significant eQTLs (P<5x10-8) following an analytical framework equivalent to the ANOVA and additive model, respectively. When we compared the two approaches, there was no overlap of significant eQTLs between the two frameworks. Because we defined specific contrasts, we identified trans eQTLs that more closely resembled what we expect from genetic variants showing complete dominance between alleles. Yet, these were not identified by the GTEx framework. Conclusions: Our results show that transforming RNA-sequencing data to fit a normal distribution prior to eQTL analysis is not required when the DGE framework is employed, thus this may be more suitable for finding genes whose expression are impacted by genetic variants. Our approach detected biologically relevant variants that otherwise would not have been identified due to data transformation to fit a normal distribution.

Background A gap currently exists between genetic variants and the underlying cell and tissue biology of a trait, and expression quantitative trait loci (eQTL) studies provide important information to help close that gap. However, two concerns that arise with eQTL analyses using RNA-sequencing data are normalization of data across samples and the data not following a normal distribution. Multiple pipelines have been suggested to address this. For instance, the most recent analysis of the human and farm Genotype-Tissue Expression (GTEx) project proposes using trimmed means of M-values (TMM) to normalize the data followed by an inverse normal transformation. Results In this study, we reasoned that eQTL analysis could be carried out using the same framework used for differential gene expression (DGE), which uses a negative binomial model, a statistical test feasible for count data. Using the GTEx framework, we identified 35 significant eQTLs (P < 5 × 10–8) following the ANOVA model and 39 significant eQTLs (P < 5 × 10–8) following the additive model. Using a differential gene expression framework, we identified 930 and six significant eQTLs (P < 5 × 10–8) following an analytical framework equivalent to the ANOVA and additive model, respectively. When we compared the two approaches, there was no overlap of significant eQTLs between the two frameworks. Because we defined specific contrasts, we identified trans eQTLs that more closely resembled what we expect from genetic variants showing complete dominance between alleles. Yet, these were not identified by the GTEx framework. Conclusions Our results show that transforming RNA-sequencing data to fit a normal distribution prior to eQTL analysis is not required when the DGE framework is employed. Our proposed approach detected biologically relevant variants that otherwise would not have been identified due to data transformation to fit a normal distribution.