The influence of progesterone on bovine uterine fluid energy, nucleotide, vitamin, cofactor, peptide, and xenobiotic composition during the conceptus elongation-initiation window

Conceptus elongation coincides with one of the periods of greatest pregnancy loss in cattle and is characterized by rapid trophectoderm expansion, commencing ~ Day 13 of pregnancy, i.e. before maternal pregnancy recognition. The process has yet to be recapitulated in vitro and does not occur in the absence of uterine gland secretions in vivo. Moreover, conceptus elongation rates are positively correlated to systemic progesterone in maternal circulation. It is, therefore, a maternally-driven and progesterone-correlated developmental phenomenon. This study aimed to comprehensively characterize the biochemical composition of the uterine luminal fluid on Days 12–14 – the elongation-initiation window – in heifers with normal vs. high progesterone, to identify molecules potentially involved in conceptus elongation initiation. Specifically, nucleotide, vitamin, cofactor, xenobiotic, peptide, and energy metabolite profiles of uterine luminal fluid were examined. A total of 59 metabolites were identified, of which 6 and 3 displayed a respective progesterone and day effect, whereas 16 exhibited a day by progesterone interaction, of which 8 were nucleotide metabolites. Corresponding pathway enrichment analysis revealed that pyridoxal, ascorbate, tricarboxylic acid, purine, and pyrimidine metabolism are of likely importance to to conceptus elongation initiation. Moreover, progesterone reduced total metabolite abundance on Day 12 and may alter the uterine microbiome.


Results
P4-releasing intravaginal device (PRID) insertion on Day 3 elevated (P ≤ 0.05) serum P4 on Day 5 (3.17 ± 0.341 vs. 1.53 ± 0.163 ng/ml), as determined by a two-way ANOVA coupled with a Holm-Sidak non-parametric post hoc. This difference was not apparent by . Moreover, no difference in systemic P4 was observed between Days 12-14 within the normal and high P4 groups, as described in Simintiras et al. 20 .
A total of 59 metabolites were identified in this study, spanning 19 pathways; 11 metabolites (18.6%) related to energy metabolism, 19 (32.2%) to nucleotide, 12 (20.3%) to cofactor and vitamin, 15 (25.4%) to xenobiotic, and 2 metabolites (3.4%) related to peptide metabolism (Table 1). Five metabolites (8.5%) displayed a P4 main effect, i.e. differed in abundance between high vs. normal P4 heifers, irrespective of day, whilst 2 (3.4%) exhibited a day main effect, i.e. were temporally dynamic independently of P4, and 1 metabolite (1.7%) -phosphate -was affected by both day and P4. Sixteen metabolites (27.1%) displayed a day by P4 interaction, of which half are implicated in nucleotide metabolism (Fig. 1).  Networks on the left-hand side show biochemicals which displayed (A) a progesterone (P4) main effect, (B) a day main effect, and/or (C) a day by P4 interaction -i.e. concentrations differed between groups (normal vs. high P4) at different times (Days 12 vs. 13 vs. 14). Significance is represented by node colour and diameter combined: a large dark red node indicates an (p ≤ 0.05) effect/interaction (node border thickness is inversely proportional to the magnitude of the p-value), whereas a medium light red node depicts a trend (0.05 < p < 0.10) towards an effect/interaction, and a small black nodes depict a lack of significance. Networks on the right-hand side compare metabolite flux magnitude in the uterine luminal fluid of normal vs. high P4 heifers on Days (D) 12, (E) 13, and (F) 14.
Here, node diameter is proportional to the fold change observed, and node colour represents the significance of the change: dark red depicting a significant (p ≤ 0.05) increase, light red highlighting an increasing trend (0.05 < p < 0.10), dark blue denoting a significant (p ≤ 0.05) decrease, and light blue depicting a decreasing trend (0.05 < p < 0.10). Black nodes depict a lack of a statistically significant flux. In addition to node colour, node border thickness is inversely proportional to the magnitude of the p-value. The contribution (%) of each sub-pathway to total (a) progesterone (P4) effects -i.e. the proportion of corresponding metabolites which were hormonally responsive, (b) day effects -i.e. the proportion of corresponding metabolites which were temporally dynamic, (c) day by P4 interactions -i.e. the proportion of corresponding metabolites whose concentrations differed between groups (normal vs. high P4) at different times (Days 12 vs. 13 vs. 14), and (d) abundance -i.e. the size of the sub-pathway. Abbreviations: progesterone (P4), super-pathways: energy metabolite (E), nucleotide metabolite (N), vitamin and cofactor metabolite (VC), xenobiotic (X), and peptide (P).  Biochemical node colour corresponds to day by progesterone (P4) interaction -dark red indicates a significant (p ≤ 0.05) interaction (node border thickness is inversely proportional to the magnitude of the p-value). Within these statistically significant nodes, node diameter is correlated to metabolic hierarchy (cofactor and intermediate metabolite nodes are smallest, followed by by-products, and central metabolites). Light red depicts an increasing trend (0.05 < p < 0.10), black depicts an identified amino acid which did not exhibit a day by progesterone interaction, and grey represents a biochemical present in the metabolic library but not detected in this study. www.nature.com/scientificreports www.nature.com/scientificreports/ Considering main effect or interaction proportions, which are less susceptible to 'single-metabolite' pathway bias than enrichment analyses, pathways representing most P4 effects were: tricarboxylic acid cycle (33.3%), oxidative phosphorylation (16.7%), purine [(hypo)xanthine/inosine] (16.7%), vitamin A (16.7%), and chemical (16.7%) metabolism. Similarly, pathways exhibiting most day effects were: oxidative phosphorylation (33.3%), nicotinate and nicotinamide (33.3%), and benzoate (33.3%) metabolism; whereas pathways displaying the greatest proportion of day by P4 interactions were: purine [(hypo)xanthine/inosine] (18.8%), pyrimidine [uracil] (12.5%), nicotinate and nicotinamide (12.5%), and vitamin B6 (12.5%) (Fig. 3). Also provided in Fig. 3 is the relative abundance of each pathway (i.e. the number of metabolites constituting the pathway as a proportion of all identified metabolites).
P4 supplementation, moreover, reduced (p < 0.05) the concentrations of 21 metabolites on Day 12 (Fig. 2D). In contrast, the concentrations of just 2 metabolites differed on Day 13 in high vs. normal P4 heifers (Fig. 2E), including the greatest individual metabolite flux observed -a 14.34-fold increase in triethanolamine (Table 1D). Biochemical node colour corresponds to day by progesterone (P4) interaction -dark red indicates a significant (p ≤ 0.05) interaction (node border thickness is inversely proportional to the magnitude of the p-value). Within these statistically significant nodes, node diameter is correlated to metabolic hierarchy (cofactor and intermediate metabolite nodes are smallest, followed by by-products, and central metabolites). Light red depicts an increasing trend (0.05 < p < 0.10), black depicts an identified amino acid which did not exhibit a day by progesterone interaction, and grey represents a biochemical present in the metabolic library but not detected in this study. Box plots: The y-axes are the relative metabolite concentrations, with the central horizontal line representing the median value with outer boundaries depicting the upper and lower quartile limits. Error bars depict the minimum and maximum distributions, with + representing the mean value and ○ the extreme data point.  Biochemical node colour corresponds to day by progesterone (P4) interaction -dark red indicates a significant (p ≤ 0.05) interaction (node border thickness is inversely proportional to the magnitude of the p-value). Within these statistically significant nodes, node diameter is correlated to metabolic hierarchy (cofactor and intermediate metabolite nodes are smallest, followed by by-products, and central metabolites). Light red depicts an increasing trend (0.05 < p < 0.10), black depicts an identified amino acid which did not exhibit a day by progesterone interaction, and grey represents a biochemical present in the metabolic library but not detected in this study. Box plots: the central horizontal line represents the median value with outer boundaries depicting the upper and lower quartile limits. Error bars depict the minimum and maximum distributions, with + representing the mean value and ○ the extreme data point. www.nature.com/scientificreports www.nature.com/scientificreports/ Nucleotides account for approximately 15-20% of non-aqueous cellular biomass 26 and fulfil a plethora of central biological processes, including in DNA replication, RNA production, and protein synthesis, rendering them of likely importance to trophoblast proliferation and consequent conceptus elongation. Moreover, accelerated cellular proliferation requires rapid protein synthesis, for which cells require more ribosomes, in turn necessitating more ribosomal RNA (rRNA), and therefore, an increased accumulation rate of ribonucleoside triphosphates (rNTPs) 26 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Whilst all cells have an inherent capacity for de novo nucleotide biosynthesis -a highly endergonic processthey can also uptake nucleotides 35,36 . Interestingly, all four ribonucleosides (adenosine, cytidine, guanosine, and uridine) were present in ULF, as were three of the corresponding ribonucleobases (adenine, guanine, and uridine). This finding, in conjunction with previous identifications of proteins relating to nucleotide transport and metabolism in ULF -e.g. polypyrimidine tract-binding protein 1 and bifunctional purine biosynthesis protein 37suggests that uterine nucleotide release into ULF during the period of elongation initiation, is conducive to successful conceptus elongation, presumably for RNA production (Fig. 5); the RNA-exclusive nucleoside, uridine, displayed the second greatest day by P4 interaction, increasing in luminal abundance on Day 14 vs. 12 and 13 in the high P4 group, whilst thymine and thymidine, the respective DNA-exclusive nucleobase and nucleoside, were not identified.
Cofactors and vitamins was the only metabolite group displaying a significant flux when comparing all metabolites by treatment and day (Fig. 4). Regarding day by P4 interactions, the enrichment values for nicotinate and nicotinamide, vitamin B6, and ascorbate and aldarate metabolism were 2.5, 2.5, and 1.2, respectively. Nicotinamide is a key intermediate in NAD synthesis, which is crucial for energy metabolism (discussed below). Concentrations of both nicotinamide and nicotinamide riboside were lower in the ULF from high vs. normal P4 heifers on Day 12. Their concentrations, moreover, decreased in the normal P4 group from Day 12 to 14, suggesting decreased uterine nicotinamide salvage and metabolism in the high P4 heifers, likely as a result of altered epithelial energy metabolism. This may be important as a recent study conducted in humans and mice found the nicotinamide pathway plays an important role in miscarriage prevention 38 .
Retinol (vitamin A) presence was also interesting as it is involved in uterine epithelial cell differentiation regulation and matrix remodelling 39 . Retinol exhibited a P4 main effect by trending higher in the ULF of high P4 heifers on Days 12 and 14 relative to control. This finding is broadly in line with data from the porcine, wherein P4 increases uterine retinol binding protein (RBP) abundance 40 . Whilst RBP4 has been identified in bovine ULF 37 , further research is required to elucidate the roles of retinol and RBPs in ruminant maternal-embryo communication; not least as retinol inhibits activator protein 1 binding to the cyclic AMP response element of the COX-2 promoter, in turn, inhibiting COX-2 transcription, and reducing prostaglandin synthesis 41 . Moreover, it has been proposed that retinol contributes to uterine quiescence in the human myometrium during pregnancy 42 .
Similarly to nicotinamide and nicotinamide riboside, the concentration of all detected vitamin B6 metabolites (pyridoxamine, pyridoxal, and pyridoxate) was lower in high vs. normal P4 heifers on Day 12, and generally decreased over time in the normal P4 cohorts (Fig. 6). Interestingly, bacteria of the mammalian intestinal microbiome have the capacity to synthesise vitamins B3, B5, B6, and B12 43 . Given the day by P4 interaction displayed by pyridoxamine and pyridoxate, phenol sulfate 20 , and trend towards a day by P4 interaction by hippurate -all of which are microbiome-associated and unrelated metabolites -it is tempting to suggest that P4 may alter the bacterial profile of the bovine uterus 44 and, in turn, ULF composition.
Further supporting this notion that ULF composition may be influenced by symbiotes is the identification of microbiome-associated 45,46 xenobiotics ergothioneine and mannonate in ULF. Moreover, the presence of ascorbate (vitamin C), in addition to uric acid -both potent antioxidants 47 -hints at the capacity of ULF to regulate redox homeostasis. These findings, taken together, highlight the potential role, and need for research into the impact, of factors beyond endometrial secretions on maternal-embryo communication.
Even though just two peptide metabolites were identified, the most pronounced day by P4 interaction observed in this study was that of γ-glutamylglutamine -a dipeptide of amino acids glutamate and glutamine, and proteolytic breakdown product 48 . The increase in γ-glutamylglutamine levels on Days 13 and 14 vs. 12 in high P4 heifers may be indicative of elevated uterine luminal protease 37 activity. However, whether γ-glutamylglutamine, and phenylacetylglycine for that matter, are short-lived intermediates en route to further proteolysis, or have physiological cell-signalling effects, is unknown. It is worth noting the murine hatching embryo secretes a trypsin-like protease in vitro 49 , and, in the equine, peptidase D and the proteasome 26S subunit were in greater abundance in the ULF of pregnant vs. cyclic mares on Day 13 50 . Thus, uterine luminal protein degradation may be a free amino acid generation mechanism.
The TCA cycle is generally considered the central metabolic hub of the cell as it is the final common pathway for fuel molecule oxidation, including carbohydrates, fatty acids, and amino acids 51 . Forde et al. 37 previously identified 5 of the 8 enzymes involved in the TCA cycle in bovine ULF: aconitase, isocitrate dehydrogenase, fumarate hydratase (fumarase), malate dehydrogenase, and citrate synthase. Here we identified 6 of the 10 key corresponding metabolites: citrate, aconitate, isocitrate, 2-oxoglutarate, succinate, and malate; raising the question of whether parts of the TCA cycle (Fig. 7) are active in ULF.
Of particular intrigue is α-ketoglutarate (a.k.a. 2-oxoglutarate), which exhibited a P4 main effect owing to a decrease in high vs. normal P4 heifers on Day 12 in addition to a decline in normal P4 heifers on Day 14 vs. 12 -i.e. 2-oxoglutarate levels inversely correlated with an advanced elongation environment. Whilst this intuitively implies that extracellular 2-oxoglutarate is not important for the elongating conceptus, it is telling that ULF is enzymatically equipped to produce 2-oxoglutarate from any one of malate, citrate, or isocitrate 37 . Aside from its role as a TCA cycle intermediate, 2-oxoglutarate is a common source of amino acids glutamate and glutamine 52 ; though such a breakdown in ULF is unlikely as the catalyst, glutamate dehydrogenase, has not been identified in ULF. Nonetheless, 2-oxoglutarate can scavenge and sequester hydrogen peroxide (H 2 O 2 ) in vitro prior to succinate conversion 53 . As H 2 O 2 is a main source of reactive oxygen species, known to increase embryonic DNA damage 54 , further investigation into this phenomenon within an in vivo reproductive context is warranted.
Succinate followed an identical profile to 2-oxoglutarate and trended towards a day by P4 interaction. It is important to highlight that succinate synthesis would be the terminal step in a ULF TCA cycle, as succinate dehydrogenase [a.k.a. succinate-coenzyme Q reductase (SQR) or respiratory complex II], which catalyses the reversible oxidation of succinate to fumarate, is solely localised at inner mitochondrial membranes 55 . Regardless, succinate plays numerous signalling roles, such as stabilizing the hypoxia-inducible factor-1 alpha (HIF-1α) transcription factor 56 , whose knockout in the mouse is embryo-lethal owing to vascular defects in the embryonic and extraembryonic vasculature 57 . Moreover, G-protein coupled receptor 91 (GPCR91), previously considered an orphan receptor 58 , has been recently shown to sense extracellular succinate 59 , and in utero upregulation of GPCR91 in the rat foetus predisposes the offspring to hypertension 60 .
Given that all primary constituents of the first TCA cycle reactions have been identified in ULF, it is plausible to assume a role for energy metabolites in conceptus elongation, perhaps beyond energy production. Moreover, the presence of tricarballylate -an aconitase (the enzyme catalysing the formation of isocitrate from citrate) inhibitor 61 of aerobic bacterial origin 62 -further exemplifies the need to investigate the impact of the microbiome in uterine metabolism and maternal-embryo communication.
Furthermore, whilst this work on maternal-embryo communication using cyclic heifers inherently focuses on a unidirectional monologue (maternally secreted metabolites), future work includes similarly analysing ULF from pregnant heifers -in addition to corresponding conceptus-conditioned culture medium -to tease out the reciprocity of bovine pregnancy establishment. Additional future work includes further investigating P4-induced changes in the ruminant ULF proteome. For example, uterine serpins (or milk proteins) are secreted by the ovine [63][64][65] and bovine 66 uterus, in a hormonally responsive manner, presumably as a mechanism of immunomodulation, despite their classification as proteinase inhibitors 67 . Similarly, elucidating the precise role of conceptus-derived proteins, such as trophoblast Kunitz domain proteins -also categorised as proteinase inhibitors 68 -is essential to furthering our understanding of maternal-embryo communication and pregnancy establishment in ruminants.
Lastly, it is important to note that elevated P4 on Day 5, but not thereafter, in the high P4 group is consistent with previous observations of accelerated conceptus elongation on Days 12-14 14 . More specifically, elevated P4 resulting from PRID introduction on Day 3 temporally advances the endometrial transcriptome by ~48-72 h during the conceptus elongation window 69 despite systemic P4 concentrations returning to normal 12 .
In conclusion, this manuscript presents and describes the dynamic presence of energy substrates, nucleotides, cofactors, vitamins, and xenobiotics in the bovine uterine luminal fluid during the conceptus elongation-initiation window. The data show that elevated P4 suppresses ULF metabolite concentrations on Day 12 and may alter the uterine microbiome. Moreover, ULF has the capacity to act as a redox buffer, and biochemicals in ULF involved in purine, pyrimidine, pyridoxal, ascorbate, and tricarboxylic acid cycle metabolism likely facilitate conceptus elongation initiation.

Methods
Sample collection details are described in detail in Simintiras et al. 20 , a summary of which is provided below.
Uterine luminal fluid recovery. The uterine horn ipsilateral to the corpus luteum was excised and flushed with 10 ml phosphate buffered saline (PBS; Sigma Aldrich) within 60 mins of slaughter, prior to centrifugation for 15 min at 1000 × g. The supernatant was aliquoted, snap-frozen in liquid nitrogen, and stored at −80 °C until analysis.
Progesterone analysis. Blood was collected from all heifers by coccygeal venepuncture on Days 3 and 5 in addition to the morning of slaughter, stored at 4 °C for 24 h, and centrifuged at 1500 × g for 20 min at 4 °C. The supernatant was stored at −20 °C until analysis. P4 was quantified by solid-phase radioimmunoassay (PROG-RIA-CT, DIAsource) in line with manufacturer guidelines, and as described in Simintiras et al. 20 .
Metabolomics and data extraction. Metabolon Inc. performed the sample preparation and analysis by ultrahigh performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS). Briefly, all samples were analysed by four reverse-phase (RP/UPLC)-MS/MS methods involving positive and negative ion mode electrospray ionization, in addition to hydrophilic interaction chromatography UPLC-MS/MS. Metabolites were identified by retention time and a m/z within ± 10 ppm, and quantified against known internal and recovery standards, which were run in parallel at random intervals. Data were corrected for inter-day instrument tuning variations -median peak areas for each metabolite were registered as 1.00 prior to the proportional normalisation of each data point. Data were then logarithmically transformed and quantified by relative abundance using MetaboLync pathway analysis (MPA) software (portal.metabolon.com) and visualized using Java Cytoscape 3.6.1. Relative mean metabolite concentrations were calculated by averaging the median scaled imputed data of all metabolites for each aforementioned experimental group. Data analysis and statistics. Semi-quantitative metabolomic data (either expressed as main effects, interactions, fold-changes, or relative concentrations), were statistically analysed by two-way ANOVA with a p ≤ 0.05