61 Noninvasive prediction of pregnancy and birth in cattle by liquid chromatography-high-resolution mass spectrometry analysis of embryo culture medium

Estimation of the embryo competence to reach birth is a major objective in reproductive technology. Discrimination of the metabolic variation induced by fixed factors (FF) (i.e. culture conditions, embryonic stage, bull breed, and embryo cryopreservation) may help refine estimations. We hypothesised that pregnancy and birth of transferred fresh and frozen IVP bovine embryos can be predicted by liquid chromatography-high-resolution mass spectrometry (LC-HRMS) analysis of culture medium (CM). Oocytes fertilised and cultured in groups in modified synthetic oviductal fluid with bovine serum albumin (SOFaaci-BSA) with or without fetal calf serum (FCS), were later individually cultured in protein-free CM for 24h, and transferred to recipients as fresh (28 ETs; 57% birth rate (BR)) or frozen/thawed (58 ETs; BR Day 7 embryos: 46%; BR Day 8 embryos: 12.5% (1/8)) blastocysts. Spent CM was analysed by LC-HRMS. Chromatograms were processed (MZmine v2.53 and R package pRocessomics) to retain 6111 signals, which were analysed as a function of pregnancy at Day 40, Day 62, and birth by sparse partial least squares-discriminant analysis (PLS-DA) and orthogonal PLS-DA. Both within the entire dataset and in sample blocks with any single FF, DA did not yield sample separation. However, some blocks of samples formed by two FFs differed by permutation (e.g. a block of N = 48 frozen (FF1) embryos cultured with BSA (FF2) at birth: Q²: P < 0.03, R²Y: P < 0.01). Thereafter, a univariate study analysed all combinations of FF in blocks under the criteria of fold change (FCh) > |2| (pregnancy/non-pregnancy) and P < 0.05 or tendencies 0.05 > P < 0.10 (ANOVA and Kruskal–Wallis test) and area under the receiver operator characteristic curve (ROC-AUC) >0.700 with FCh > |2|) and t-test P < 0.05 and P < 0.10. At this stage, 34 significantly accumulated metabolites were identified through 198 blocks for birth, 166 for Day 62 and 147 for Day 40. Accumulated metabolites were higher in blocks from non-pregnancy-prone (460) than from pregnancy-prone (51) embryos. Blocks of embryos transferred frozen accounted for 33 metabolites (with 16 of them not represented in fresh embryos); 17 metabolites were independent of embryo cryopreservation; and only one metabolite was exclusive of fresh embryos (5-hydroxy-l-Trp). Classes were amino acids (12) and derivatives (3) (186 blocks), lipids (12; 224 blocks), benzenoids (4; 58 blocks), tri-carboxylic acids (2; 41 blocks) and 5-hydroxy-l-Trp (2 blocks). Some metabolites were predictive as single biomarkers in 95 blocks (ROC-AUC: 0.700–1.000), but more accurate predictions in the largest datasets were given by combinations of 2, 3, and 4 single metabolites in 206 blocks (ROC-AUC = 0.800–1.000). Dimethyl adypate had more blocks on Day 40 (18) and Day 62 (19) than at birth (4), suggesting a candidate to predict late miscarriage. Generally, pregnancy-prone embryos depleted less lipids and cis-aconitic acid, and consumed more amino acids and citric acid. We suggest that competent embryos accumulate fewer lipids in their cells. Our work is the first to identify high numbers of birth biomarkers by LC-HRMS in bovine embryos. Biomarker discovery can be improved upon discrimination of factors that induce metabolic variability in embryos.

This research was supported by Horizon 2020 No 952908 - GLOMICAVE-. IG: MINECO BES-2017-082200. FEDER.