Machine learning approach to integrated endometrial transcriptomic datasets reveals biomarkers predicting uterine receptivity in cattle at seven days after estrous

Artificial intelligence and porcine breeding

Livestock management is evolving into a new era, characterized by the analysis of vast quantities of data (Big Data) collected from both traditional breeding methods and new technologies such as sensors, automated monitoring system, and advanced analytics. Artificial intelligence (A-In), which refers to the capability of machines to mimic human intelligence, including subfields like machine learning and deep learning, is playing a pivotal role in this transformation. A wide array of A-In techniques, successfully employed in various industrial and scientific contexts, are now being integrated into mainstream livestock management practices. In the case of swine breeding, while traditional methods have yielded considerable success, the increasing amount of information requires the adoption of new technologies such as A-In to drive productivity, enhance animal welfare, and reduce environmental impact. Current findings suggest that these techniques have the potential to match or exceed the performance of traditional methods, often being more scalable in terms of efficiency and sustainability within the breeding industry. This review provides insights into the application of A-In in porcine breeding, from the perspectives of both sows (including welfare and reproductive management) and boars (including semen quality and health), and explores new approaches which are already being applied in other species.

Transcriptomics analysis of the bovine endometrium during the perioestrus period

During the oestrous cycle, the bovine endometrium undergoes morphological and functional changes, which are regulated by alterations in the levels of oestrogen and progesterone and consequent changes in gene expression. To clarify these changes before and after oestrus, RNA-seq was used to profile the transcriptome of oestrus-synchronized beef heifers. Endometrial samples were collected from 29 animals, which were slaughtered in six groups beginning 12 h after the withdrawal of intravaginal progesterone releasing devices until seven days post-oestrus onset (luteal phase). The groups represented proestrus, early oestrus, metoestrus and early dioestrus (luteal phase). Changes in gene expression were estimated relative to gene expression at oestrus. Ingenuity Pathway Analysis (IPA) was used to identify canonical pathways and functional processes of biological importance. A total of 5,845 differentially expressed genes (DEGs) were identified. The lowest number of DEGs was observed at the 12 h post-oestrus time point, whereas the greatest number was observed at Day 7 post-oestrus onset (luteal phase). A total of 2,748 DEGs at this time point did not overlap with any other time points. Prior to oestrus, Neurological disease and Organismal injury and abnormalities appeared among the top IPA diseases and functions categories, with upregulation of genes involved in neurogenesis. Lipid metabolism was upregulated before oestrus and downregulated at 48h post-oestrus, at which point an upregulation of immune-related pathways was observed. In contrast, in the luteal phase the Lipid metabolism and Small molecule biochemistry pathways were upregulated.

BioDiscViz: A visualization support and consensus signature selector for BioDiscML results

Machine learning (ML) algorithms are powerful tools to find complex patterns and biomarker signatures when conventional statistical methods fail to identify them. While the ML field made significant progress, state of the art methodologies to build efficient and non-overfitting models are not always applied in the literature. To this purpose, automatic programs, such as BioDiscML, were designed to identify biomarker signatures and correlated features while escaping overfitting using multiple evaluation strategies, such as cross validation, bootstrapping and repeated holdout. To further improve BioDiscML and reach a broader audience, better visualization support and flexibility in choosing the best models and signatures are needed. Thus, to provide researchers with an easily accessible and usable tool for in depth investigation of the results from BioDiscML outputs, we developed a visual interaction tool called BioDiscViz. This tool provides summaries, tables and graphics, in the form of Principal Component Analysis (PCA) plots, UMAP, t-SNE, heatmaps and boxplots for the best model and the correlated features. Furthermore, this tool also provides visual support to extract a consensus signature from BioDiscML models using a combination of filters. BioDiscViz will be a great visual support for research using ML, hence new opportunities in this field by opening it to a broader community.

miRNA expression profiles of peripheral white blood cells from beef heifers with varying reproductive potential

Reproductive performance is the most critical factor affecting production efficiency in the cow-calf industry. Heifers with low reproductive efficiency may fail to become pregnant during the breeding season or maintain a pregnancy. The cause of reproductive failure often remains unknown, and the non-pregnant heifers are not identified until several weeks after the breeding season. Therefore, improving heifer fertility utilizing genomic information has become increasingly important. One approach is using microRNAs (miRNA) in the maternal blood that play an important role in regulating the target genes underlying pregnancy success and thereby in selecting reproductively efficient heifers. Therefore, the current study hypothesized that miRNA expression profiles from peripheral white blood cells (PWBC) at weaning could predict the future reproductive outcome of beef heifers. To this end, we measured the miRNA profiles using small RNA-sequencing in Angus-Simmental crossbred heifers sampled at weaning and retrospectively classified as fertile (FH, n = 7) or subfertile (SFH, n = 7). In addition to differentially expressed miRNAs (DEMIs), their target genes were predicted from TargetScan. The PWBC gene expression from the same heifers were retrieved and co-expression networks were constructed between DEMIs and their target genes. We identified 16 differentially expressed miRNAs between the groups (p-value ≤0.05 and absolute (log2 fold change ≥0.05)). Interestingly, based on a strong negative correlation identified from miRNA-gene network analysis with PCIT (partial correlation and information theory), we identified miRNA-target genes in the SFH group. Additionally, TargetScan predictions and differential expression analysis identified bta-miR-1839 with ESR1 , bta-miR-92b with KLF4 and KAT2B, bta-miR-2419-5p with LILRA4, bta-miR-1260b with UBE2E1, SKAP2 and CLEC4D, and bta-let-7a-5p with GATM, MXD1 as miRNA-gene targets. The miRNA-target gene pairs in the FH group are over-represented for MAPK, ErbB, HIF-1, FoxO, p53, mTOR, T-cell receptor, insulin and GnRH signaling pathways, while those in the SFH group include cell cycle, p53 signaling pathway and apoptosis. Some miRNAs, miRNA-target genes and regulated pathways identified in this study have a potential role in fertility; other targets are identified as novel and need to be validated in a bigger cohort that could help to predict the future reproductive outcomes of beef heifers.

Machine Learning-Based Co-Expression Network Analysis Unravels Potential Fertility-Related Genes in Beef Cows

Simple Summary

Female reproductive failure is still a challenge for the beef industry. Several biological processes that underlie fertility-related traits, such as the establishment of pregnancy and embryo survival, are still unclear. Increased availability of transcriptomic data has allowed a deep investigation of the potential mechanisms involved in fertility. This study investigated candidate gene biomarkers predictive of pregnancy status and underlying fertility-related networks. To this end, we integrated gene expression profiles through supervised machine learning and gene network modeling. We identified nine biologically relevant endometrial gene biomarkers that could discriminate against pregnancy status in cows. These biomarkers were co-expressed with genes critical for uterine receptivity, including endometrial tissue remodeling, focal adhesion, and embryo development. This study outlined key pathways involved with pregnancy success and provided predictive candidate biomarkers for pregnancy outcome in cows.

Abstract

Reproductive failure is still a challenge for beef producers and a significant cause of economic loss. The increased availability of transcriptomic data has shed light on the mechanisms modulating pregnancy success. Furthermore, new analytical tools, such as machine learning (ML), provide opportunities for data mining and uncovering new biological events that explain or predict reproductive outcomes. Herein, we identified potential biomarkers underlying pregnancy status and fertility-related networks by integrating gene expression profiles through ML and gene network modeling. We used public transcriptomic data from uterine luminal epithelial cells of cows retrospectively classified as pregnant (P, n = 25) and non-pregnant (NP, n = 18). First, we used a feature selection function from BioDiscML and identified SERPINE3, PDCD1, FNDC1, MRTFA, ARHGEF7, MEF2B, NAA16, ENSBTAG00000019474, and ENSBTAG00000054585 as candidate biomarker predictors of pregnancy status. Then, based on co-expression networks, we identified seven genes significantly rewired (gaining or losing connections) between the P and NP networks. These biomarkers were co-expressed with genes critical for uterine receptivity, including endometrial tissue remodeling, focal adhesion, and embryo development. We provided insights into the regulatory networks of fertility-related processes and demonstrated the potential of combining different analytical tools to prioritize candidate genes.

Identification of differentially expressed long noncoding RNAs in the ovarian tissue of ewes Shal and Sangsari using RNA-seq

Background

The ovary has an important role in reproductive function. Animal reproduction is dominated by numerous coding genes and noncoding elements. Although long noncoding RNAs (LncRNAs) are important in biological activity, little is known about their role in the ovary and fertility.

Methods

Three adult Shal ewes and three adult Sangsari ewes were used in this investigation. LncRNAs in ovarian tissue from two breeds were identified using bioinformatics analyses, and then target genes of LncRNAs were discovered. Target genes were annotated using the DAVID database, and their interactions were examined using the STRING database and Cytoscape software. The expression levels of seven LncRNAs with their target genes were assessed by real‐time PCR to confirm the RNA‐seq.

Results

Among all the identified LncRNAs, 124 LncRNAs were detected with different expression levels between the two breeds (FDR < 0.05). According to the DAVID database, target genes were discovered to be engaged in one biological process, one cellular component, and 21 KEGG pathways (FDR < 0.05). The PES1, RPS9, EF‐1, Plectin, SURF6, CYC1, PRKACA MAPK1, ITGB2 and BRD2 genes were some of the most crucial target genes (hub genes) in the ovary.

Conclusion

These results could pave the way for future efforts to address sheep prolificacy barriers.

Progesterone-dependent and progesterone-independent modulation of luminal epithelial transcription to support pregnancy in cattle

In cattle, starting 4-5 days after estrus, preimplantation embryonic development occurs in the confinement of the uterine lumen. Cells in the endometrial epithelial layer control the molecular traffic to and from the lumen and, thereby determine luminal composition. Starting early postestrus, endometrial function is regulated by sex steroids, but the effects of progesterone on luminal cells transcription have not been measured in vivo. The first objective was to determine the extent to which progesterone controls transcription in luminal epithelial cells 4 days (D4) after estrus. The second objective was to discover luminal transcripts that predict pregnancy outcomes when the effect of progesterone is controlled. Endometrial luminal epithelial cells were collected from embryo transfer recipients on D4 using a cytological brush and their transcriptome was determined by RNASeq. Pregnancy by embryo transfer was measured on D30 (25 pregnant and 18 nonpregnant). Progesterone concentration on D4 was associated positively (n = 182) and negatively (n = 58) with gene expression. Progesterone-modulated transcription indicated an increase in oxidative phosphorylation, biosynthetic activity, and proliferation of epithelial cells. When these effects of progesterone were controlled, different genes affected positively (n = 22) and negatively (n = 292) odds of pregnancy. These set of genes indicated that a receptive uterine environment was characterized by the inhibition of phosphoinositide signaling and innate immune system responses. A panel of 25 genes predicted the pregnancy outcome with sensitivity and specificity ranging from 64%-96% and 44%-83%, respectively. In conclusion, in the early diestrus, both progesterone-dependent and progesterone-independent mechanisms regulate luminal epithelial transcription associated with pregnancy outcomes in cattle.

Genes and regulatory mechanisms associated with experimentally-induced bovine respiratory disease identified using supervised machine learning methodology

Bovine respiratory disease (BRD) is a multifactorial disease involving complex host immune interactions shaped by pathogenic agents and environmental factors. Advancements in RNA sequencing and associated analytical methods are improving our understanding of host response related to BRD pathophysiology. Supervised machine learning (ML) approaches present one such method for analyzing new and previously published transcriptome data to identify novel disease-associated genes and mechanisms. Our objective was to apply ML models to lung and immunological tissue datasets acquired from previous clinical BRD experiments to identify genes that classify disease with high accuracy. Raw mRNA sequencing reads from 151 bovine datasets (n = 123 BRD, n = 28 control) were downloaded from NCBI-GEO. Quality filtered reads were assembled in a HISAT2/Stringtie2 pipeline. Raw gene counts for ML analysis were normalized, transformed, and analyzed with MLSeq, utilizing six ML models. Cross-validation parameters (fivefold, repeated 10 times) were applied to 70% of the compiled datasets for ML model training and parameter tuning; optimized ML models were tested with the remaining 30%. Downstream analysis of significant genes identified by the top ML models, based on classification accuracy for each etiological association, was performed within WebGestalt and Reactome (FDR ≤ 0.05). Nearest shrunken centroid and Poisson linear discriminant analysis with power transformation models identified 154 and 195 significant genes for IBR and BRSV, respectively; from these genes, the two ML models discriminated IBR and BRSV with 100% accuracy compared to sham controls. Significant genes classified by the top ML models in IBR (154) and BRSV (195), but not BVDV (74), were related to type I interferon production and IL-8 secretion, specifically in lymphoid tissue and not homogenized lung tissue. Genes identified in Mannheimia haemolytica infections (97) were involved in activating classical and alternative pathways of complement. Novel findings, including expression of genes related to reduced mitochondrial oxygenation and ATP synthesis in consolidated lung tissue, were discovered. Genes identified in each analysis represent distinct genomic events relevant to understanding and predicting clinical BRD. Our analysis demonstrates the utility of ML with published datasets for discovering functional information to support the prediction and understanding of clinical BRD.

Transcriptomic Analysis of STAT1/3 in the Goat Endometrium During Embryo Implantation

Interferon tau (IFNT), a pregnancy recognition signal in ruminants, promotes the establishment of embryo implantation by inducing the expression of interferon-stimulated genes (ISGs) via the Janus kinase/signal transducer and activator of transcription (JAK/STAT) signaling pathway. However, the precise regulatory mechanism of IFNT in goat embryo implantation remains largely unknown. In this study, we performed RNA sequencing of goat endometrial epithelial cells (gEECs) with or without 20 ng/mL IFNT treatment. Differential comparison showed that there were 442 upregulated differentially expressed genes (DEGs) and 510 downregulated DEGs. Bioinformatic analyses revealed that DEGs were significantly enriched in immune-related functions or pathways. The qRT-PCR validation results showed that the expression levels of STAT family members (STAT1, STAT2, and STAT3) were significantly upregulated in gEECs after IFNT treatment, which is in agreement with the RNA-seq data. Meanwhile, the protein levels of p-STAT1 and p-STAT3 increased significantly in gEECs after 6 and 24 h of IFNT treatment, respectively. Further in vivo experiments also confirmed that both mRNA and protein phosphorylation levels of STAT1 and STAT3 in the uterus on day 18 of pregnancy (P18) were significantly increased compared to those on day 5 (P5) and day 15 of pregnancy (P15). On P5, STAT1 and STAT3 proteins were primarily located in the uterine luminal epithelium (LE) and glandular epithelium (GE), and were also detected in the stromal cells. The intense immunostaining of STAT1 and STAT3 proteins were decreased on P15 and then increased on P18, especially in the superficial GE and subepithelial stromal cells. Moreover, p-STAT1 and p-STAT3 were highly expressed in the deep GE on P18. Collectively, these results highlight the role of IFNT in regulating endometrial receptivity in gEECs and uncover the temporal and spatial changes in the expression of STAT1/3 during embryo implantation in the goat endometrium.