Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 https://doi.org/10.1186/s40104-021-00636-5 RESEARCH Open Access Metabolomic fingerprinting of pig seminal plasma identifies in vivo fertility biomarkers Yentel Mateo-Otero1,2, Pol Fernández-López3, Ariadna Delgado-Bermúdez1,2, Pau Nolis4, Jordi Roca5, Jordi Miró6, Isabel Barranco1,2,7*† and Marc Yeste1,2*† Abstract Background: Metabolomic approaches, which include the study of low molecular weight molecules, are an emerging -omics technology useful for identification of biomarkers. In this field, nuclear magnetic resonance (NMR) spectroscopy has already been used to uncover (in) fertility biomarkers in the seminal plasma (SP) of several mammalian species. However, NMR studies profiling the porcine SP metabolome to uncover in vivo fertility biomarkers are yet to be carried out. Thus, this study aimed to evaluate the putative relationship between SP- metabolites and in vivo fertility outcomes. To this end, 24 entire ejaculates (three ejaculates per boar) were collected from artificial insemination (AI)-boars throughout a year (one ejaculate every 4 months). Immediately after collection, ejaculates were centrifuged to obtain SP-samples, which were stored for subsequent metabolomic analysis by NMR spectroscopy. Fertility outcomes from 1525 inseminations were recorded over a year, including farrowing rate, litter size, stillbirths per litter and the duration of pregnancy. Results: A total of 24 metabolites were identified and quantified in all SP-samples. Receiver operating characteristic (ROC) curve analysis showed that lactate levels in SP had discriminative capacity for farrowing rate (area under the curve [AUC] = 0.764) while carnitine (AUC = 0.847), hypotaurine (AUC = 0.819), sn-glycero-3-phosphocholine (AUC = 0.833), glutamate (AUC = 0.799) and glucose (AUC = 0.750) showed it for litter size. Similarly, citrate (AUC = 0.743), creatine (AUC = 0.812), phenylalanine (AUC = 0.750), tyrosine (AUC = 0.753) and malonate (AUC = 0.868) levels had discriminative capacity for stillbirths per litter; and malonate (AUC = 0.767) and fumarate (AUC = 0.868) levels for gestation length. Conclusions: The assessment of selected SP-metabolites in ejaculates through NMR spectroscopy could be considered as a promising non-invasive tool to predict in vivo fertility outcomes in pigs. Moreover, supplementing AI-doses with specific metabolites should also be envisaged as a way to improve their fertility potential. Keywords: Artificial insemination, in vivo fertility, Metabolomics, NMR, Pregnancy outcomes, Seminal plasma Background is mainly based on the use of artificial insemination (AI), Predicting the reproductive potential of sires remains a an essential tool applied globally to improve reproduct- pending challenge for the livestock industry. This is of ive efficiency [1]. Over the last few years, the enhance- particular relevance for the swine sector, whose breeding ment of AI-procedures in this species has led to (1) a decrease in the sperm numbers per AI-dose, and (2) a * Correspondence: isabel.barranco@unibo.it; marc.yeste@udg.edu reduction in the number of AI performed per sow, † Isabel Barranco and Marc Yeste contributed equally to this work. without modifying in vivo fertility outcomes [2]. This 1 Biotechnology of Animal and Human Reproduction (TechnoSperm), Institute situation entails that a higher number of AI-doses are of Food and Agricultural Technology, University of Girona, ES-17003 Girona, Spain elaborated from a single AI-boar and a higher number Full list of author information is available at the end of the article of sows are inseminated with a single AI-boar, which © The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 2 of 15 leads to an increase in the reproductive and economic from 1,525 inseminations were recorded over a year repercussion of AI-boars on swine farms [2]. Although (> 100 sows inseminated per boar). Using this approach, AI-boars are selected on the basis of their genetic merit the present study was able to identify several SP- and the results obtained by routine sperm analyses metabolites able to potentially predict AI outcomes. (which include sperm concentration, morphology and motility), differences among AI-boars on in vivo fertility Methods outcomes are still notable [3, 4]. For this reason, many Experimental design efforts have been made to uncover biomarkers capable A total of 24 entire ejaculates were collected from eight to predict in vivo fertility outcomes. AI-boars (three ejaculates per boar) throughout a year During the past few years, special emphasis has been (one ejaculate every 4 months). Immediately after collec- paid to identify these biomarkers in seminal plasma (SP), tion, ejaculates were centrifuged to obtain SP-samples, a heterogenous fluid secreted by the epididymis and which were stored (− 80 °C) for subsequent metabolomic accessory sex glands [5]. This fluid has been poised as a analysis. Seminal AI-doses (2,400 × 106 spermatozoa in potential source of biomarkers, due to its complex com- 80 mL) were prepared from these AI-boars and used to position and its ability to interact with sperm and the fe- inseminate (cervically; two times per estrus) a total of male genital tract, playing a key role in sperm physiology 1,525 weaned multiparous sows (1–7 litters produced) and maternal environment modulation [6–8]. In this throughout a year. These sows (Landrace and Large sense, high-throughput technologies (including genom- White) were housed in different farms in Spain with ics, lipidomics, proteomics, metabolomics and tran- comparable management conditions. Each boar serviced scriptomics) may hold the key for uncovering reliable more than 100 sows. fertility biomarkers in SP, since they provide a more in- Fertility outcomes were recorded from each AI-boar depth understanding of reproductive processes [9]. In included in the study during the same year that SP- the last decade, many studies conducted in mammalian samples were collected and AI were performed. Re- SP have employed these novel technologies to collect corded fertility variables were: (1) farrowing rate (per- large amounts of data to discover novel fertility bio- centage of inseminated sows that farrowed), (2) litter markers [10–13]. size (total number of piglets born per litter), (3) number Metabolomics is the last emerging -omics technology of stillbirths per litter, and (4) duration of pregnancy that has become a promising tool to identify biomarkers (days). These fertility records were corrected for farm- of (in) fertility [11, 14]. This high-throughput method al- related parameters and sows using the multivariate stat- lows for the study of cells, tissues and biological fluids istical model described by Broekhuijse et al. [23]. This by evaluating metabolic products, which are the finished model allows isolating the direct boar effect on each outputs of cellular processes [11, 15]. The identification in vivo fertility parameter. of (in) fertility biomarkers in SP through metabolomics approaches has been extensively reported in several Boars and ejaculates mammalian species, including human [16–19], porcine All ejaculates were collected from AI-boars housed in a [20] and bovine [21, 22]. In pigs, Zhang et al. (2021) Spanish AI-Center (AIM Iberica, Topigs Norsvin Spain compared the SP metabolome obtained by ultra-high SLU, Calasparra, Murcia, Spain). This center fulfilled the performance liquid chromatography-quadrupole time- Spanish (ES300130640127, August 2006) and European of-flight mass spectrometry between boars with high and (ES13RS04P, July 2012) rules in matters of animal low conception rates after AI (< 70 sows inseminated health, collection of boar ejaculates and per boar), identifying some SP-metabolites (such as commercialization of AI-doses. As no animal was ma- Pro-Asn, Ile-Tyr, and D-Biotin) as potential fertility nipulated by the authors but rather the AI-Center pro- biomarkers [20]. However, neither the concentration vided AI-doses and fertility data, no permission from an of SP-metabolites, nor the putative relationship between Ethics Committee was required. SP-metabolites and other in vivo fertility outcomes The entire ejaculates used in this study were collected (e.g. litter size, stillbirths per litter or gestation duration) from healthy, mature (12 to 36 months), fertile boars was reported by these authors. from different breeds (Landrace and Large White) using The aim of this study was to evaluate the relationship a semi-automatic collection system (Collectis®, IMV between the presence/concentration of SP-metabolites Technologies, L’Aigle, France). These boars were in- and reproductive performance (including farrowing rate, cluded in an AI-program and subjected to regular ejacu- litter size, stillbirths per litter and duration of pregnancy) late collection (twice per week) for producing seminal of liquid-stored pig semen using Nuclear Magnetic Res- AI-doses. The entire ejaculates included in this study onance (NMR) spectroscopy. To achieve this goal, a total satisfied the semen quality limits required to produce of eight AI-boars were included in the study and data commercial AI-doses (sperm concentration > 200 × 106 Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 3 of 15 sperm/mL; sperm motility > 70%; sperm with normal Statistical analysis morphology > 75%). All analyses were carried out using R software (version Boars were housed in individual pens with controlled 4.0.2; https://www.r-project.org/). For all analysis, the temperature (15–25 °C) and light (16 h; natural and arti- level of significance was set at P ≤ 0.05. Statistical ana- ficial). Animals had free access to water and were fed lysis of NMR data was performed in two steps: a) nu- with agricultural feedstuff in agreement with the nutri- meric (fertility) vs. numeric (SP-metabolite tional requirements of AI-boars. concentration) variables, and b) categoric (fertility par- ameter) vs. numeric (SP-metabolite concentration) variables. Seminal plasma processing and storage First, numerical analysis, namely Pearson correlations, For SP-harvesting, the entire ejaculates were centrifuged were used to preliminary evaluate the potential linear re- (1,500 × g for 10 min at room temperature [Rotofix 32A; lationship between SP-metabolite concentration and Hettich Centrifuge UK, Newport Pagnell, Buckingham- each fertility parameter. shire, England, UK]) twice immediately after ejaculate Onwards, data were split into two different groups for collection. The resulting second supernatants (SP-sam- each reproductive parameter (farrowing rates, litter size, ples) were subsequently analyzed (Eclipse E400; Nikon, stillbirths per litter and pregnancy length). Samples with Tokyo, Japan) to warrant the absence of sperm. Finally, values lower than the median were classified as negative SP-samples were stored in 2-mL cryotubes at − 80 °C farrowing rate, decreased litter size and stillbirths per lit- (Ultra Low Freezer; Haier Inc., Qingdao, China) until ter, and shorter pregnancy duration; samples with values metabolomic profiling was carried out. higher than the median were classified as positive far- rowing rate, increased litter size and stillbirths per litter, 1 H NMR analysis and longer pregnancy duration. This process was exe- The SP-samples were thawed on ice and one of the ali- cuted for each individual fertility parameter, yielding a quots (500 μL) used. Each aliquot was vortexed and cen- specific categorization for each one. trifuged through 0.5 mL Amicon® Ultra Centrifugal To evaluate potential differences in SP-metabolite con- Filters (14,000 × g at 4 °C for 90 min) for discarding pro- centrations between fertility groups, a Wilcoxon rank teins and cell debris. Then, 100 μL of PBS containing sum test (equivalent to Mann-Whitney U test) was per- 10% D2O with 0.33% of DSS (Merck KgaA, Darmstadt, formed. As opposed to t-test, Wilcoxon does not assume Germany; pH 7.4) were added to the eluted fractions and normal distribution of samples, which did not occur in transferred into a 5-mm Wilmad® NMR tube (Merck some cases. A multivariate analysis was also carried out KgaA), where 100 μL of D2O was added. Finally, the 1H to evaluate putative inter-metabolite relationships and NMR profile was acquired. patterns that could predict fertility outcomes. In this sense, a Bayesian logistic regression was used, treating 1 the groups above the median as success (1) and those H NMR spectra below the median as failure (0). An individual model A Bruker 600-MHz AVANCE III NMR spectrometer was run for each of the fertility parameters considering (Bruker Biospin, Rheinstetten, Germany) operating at a 1 all the metabolites as potential predictors, using the R H frequency of 600.13 MHz and 300 K with a previous package ‘rstanarm’ (R package version 2.21.1; [24]), with equilibration time (10 min) was used to obtain NMR non-informative prior distributions, high resolution sam- spectra. The 1D-1H-nuclear Overhauser effect spectros- pling of the posterior distribution (adapt_delta = 0.99) copy (1D-NOESY) pulse sequence from the Bruker li- and 4,000 iterations. The remaining parameters of the brary was used. The parameters applied were: (1) mixing models were left by default. The Bayesian framework time: 100 ms (d8); (2) recovery delay: 2 s (d1); (3) 90° was selected over the classical frequentist one because of pulse: 10.39 μs (p1); (4) spectral width: 7211.539 Hz; (5) the structure of data. Usually, with a higher number of spectral size: 32 k; (6) number of scans: 128; and (7) ac- features (or predictors) than samples, as in the present quisition time: 2.27 s. dataset, models tend to get overfitted. While classical re- gression models rely on confidence intervals to estimate Data processing and analysis their reliability, the Bayesian framework estimates the The Chenomx 8.0 profiler software was used for pro- whole posterior (the approximately ‘real’) distribution cessing and analyzing spectra. This software delivers and allows quantifying the uncertainty of coefficients tools for automatic phase, baseline correction, reference and predictions accordingly. This methodology is par- calibration and libraries of metabolites for profiling. The ticularly of interest not only for having a good sense of concentration of each metabolite identified in SP was how accurate predictions are, but also for improving the calculated based on DSS concentration (0.216 mmol/L). models as more information about data becomes Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 4 of 15 available (e.g., knowledge about the mean or the range metabolites were categorized in: i) amino acids (n = 7; of the ‘true’ distribution of the predictors). alanine, glutamate, isoleucine, leucine, phenylalanine, As a last step to assess the predictability of the differ- tyrosine and valine); ii) alcohols (n = 2; ethanol and ent fertility parameters, two additional analyses were methanol); iii) saccharides (n = 1; glucose); iv) salts (n = performed. A sparse partial least square discriminant 7; acetate, benzoate, citrate, formate, fumarate, lactate analysis (sPLS-DA) model was run separately for each of and malonate); and v) other organic compounds (n = 5; the parameters, using the ‘mixOmics’ R package [25]. carnitine, creatine, creatine-phosphate, hypotaurine, Similar to principal components analysis (PCA), this myo-inositol, sn-glycero-3-phosphocholine and tri- method is useful for identifying key features in the data- methylamine N-oxide). set. However, while PCA relies on maximizing the vari- ance of the features in the principal components, sPLS- Association between SP-metabolites and AI outcomes DA maximizes their covariance. Metabolites that were Correlations between the concentration of SP- relevant in the sPLS-DA analysis and/or exhibited differ- metabolites and in vivo fertility parameters were calcu- ences between groups were further tested in a Receiver lated (Fig. 1). Five SP-metabolites were found to be cor- Operating Characteristic (ROC) curve, using the ‘pROC’ related (P < 0.05) with farrowing rate: lactate (R = − 0.62), package for R [26]. This method allows for further valid- leucine (R = 0.55), phenylalanine (R = 0.45), tyrosine (R = ation of the predictive performance of metabolites and 0.49) and valine (R = 0.53). Moreover, two SP- provides a ‘cut-off’ or threshold value to discriminate (or metabolites were positively correlated (P < 0.05) with lit- predict) sample fertility (‘high’ or ‘low’). Results are ter size: carnitine (R = 0.42) and hypotaurine (R = 0.51). expressed as the area under the curve (AUC). The dis- Additionally, the number of stillbirths per litter was criminant relevance was measured by the following AUC negatively correlated (P < 0.05) with nine SP-metabolites: ranges: 0.0–0.5 = no discriminant value, 0.5–0.6 fail dis- citrate (R = − 0.42), creatine (R = − 0.51), creatine phosphate criminant value, 0.6–0.7 poor discriminant value, 0.7– (R = − 0.46), isoleucine (R = − 0.47), leucine (R = − 0.46), 0.8 fair discriminant value, 0.8–0.9 good discriminant methanol (R = − 0.53), phenylalanine (R = − 0.54), tyro- value, and 0.9–1 excellent discriminant value. sine (R = − 0.52) and valine (R = 0.57). Finally, the duration of gestation was negatively correlated (P < 0.05) Results with four SP-metabolites: citrate (R = − 0.51), creatine Metabolite profile of pig SP (R = − 0.45), methanol (R = − 0.63) and myo-inositol The 1H-NMR profile allowed the identification and (R = − 0.59). quantification of a total of 24 metabolites in pig SP- Bayesian multiple logistic regression analyses were car- samples (see Supplementary Fig. 1). The identified ried out with the aim (i) to develop a potential predictive Fig. 1 Correlations between pig seminal plasma-metabolites and in vivo fertility outcomes (farrowing rates, litter size, stillbirths per litter and gestion duration). Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1,525 inseminations. The colour saturation of red to blue represents the correlation coefficients (R) between metabolites, from 1 to − 1, respectively. Significant correlations (P < 0.05) are marked with * Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 5 of 15 model, (ii) to quantify the relative contribution of each differed (P < 0.05) between groups, showing the opposite SP-metabolite to each in vivo fertility parameter, and (iii) pattern to the aforementioned metabolites; indeed, SP- to identify specific metabolite patterns that may have an samples from boars with increased litter size deviation influence on in vivo fertility parameters. However, no (median ± SD; 0.24 mmol/L ± 0.273) exhibited lower glu- model showed a clear relationship with any of the repro- cose concentration than those with decreased litter size ductive outcomes (Supplementary Fig. 2A-D). deviation (median ± SD; 0.79 mmol/L ± 0.245). The sPLS-DA analysis (Fig. 3A) showed that the Relationship between SP-metabolites and farrowing rate first two components explained 53.9% of the total Boars were classified into two groups based on their far- variance of the sample. The plot showed two different rowing rate deviation from the median: negative farrow- groups; SP-samples from boars classified with in- ing rate deviation (ranging from − 2.80 to − 1.60; n = 4) creased litter size deviation (red) were mainly influ- and positive farrowing rate deviation (ranging from 2.82 enced by the second component, whereas SP-samples to 7.54; n = 4). Only the concentration of lactate in SP from boars classified with decreased litter size devi- differed (P < 0.05) between groups, showing higher levels ation (blue) were mainly affected by the first compo- in SP-samples from boars with negative farrowing rate nent. The loadings plot revealed that while carnitine, deviation (median ± SD; 1.90 mmol/L ± 0.508) compared hypotaurine, sn-glycero-3-phosphocholine and glucose to those with positive farrowing rate deviation (median ± strongly influenced the first component, glutamate SD; 1.22 mmol/L ± 0.585). and methanol had that effect on the second compo- The sPLS-DA analysis was carried out to select the most nent (Fig. 3B). ROC curve analysis showed that all predictive or discriminant features in the dataset to clas- SP-metabolites identified as relevant by the loadings sify samples [27]. The sPLS-DA analysis for farrowing plot had a significant AUC (P < 0.05; Fig. 3C). Specif- rates deviation using the first two components explained ically, carnitine showed the highest AUC of 0.840, 49.9% of the total variance of the sample (Fig. 2A). The hypotaurine displayed an AUC of 0.819, sn-glycero-3- resulting plot showed two different groups: SP-samples phosphocholine showed an AUC of 0.833, glucose from boars exhibiting negative farrowing rate deviation exhibited an AUC of 0.750, and glutamate had an (blue) were mainly discriminated by the second compo- AUC of 0.799. Thus, the ROC curve showed that nent, whereas SP-samples from boars classified as positive while carnitine, hypotaurine, sn-glycero-3-phospho- farrowing rate deviation (red) were separated by the first choline and glutamate exhibited a good discriminant component. The loadings plot, which shows the most value for predicting litter size due to their high AUC relevant variable for a given component, revealed that (ranging 0.8–0.9), glucose had a fair discriminant whereas lactate and formate were the most important var- predictive value (ranging 0.6–0.7) for litter size. iables for the first component, trimethylamine N-oxide and alanine were the most relevant for the second one (Fig. 2B). ROC curve analysis indicated that only lactate Relationship between SP-metabolites and the number of was able to predict farrowing rate deviation (P < 0.05; Fig. stillbirths per litter 2C). Specifically, lactate showed a fair discriminant value Boars were categorized into two groups depending on with an AUC of 0.764. their stillbirths per litter deviation from the median: decreased stillbirths per litter deviation (ranging from Relationship between SP-metabolites and litter size − 0.10 to 0.02; n = 4) and increased stillbirths per litter devi- Boars were classified into two groups depending on their ation (ranging from 0.05 to 0.14; n = 4). Concentrations of litter size deviation from the median: reduced litter size citrate, creatine, phenylalanine and tyrosine in SP differed (ranging from − 0.40 to 0.02; n = 4) and increased litter (P < 0.05) between groups: the SP-samples from boars with size (ranging from 0.11 to 0.52; n = 4) deviation. Concen- decreased stillbirths per litter deviation showing higher con- trations of carnitine, hypotaurine, sn-glycero-3-phospho- centrations than those from boars with increased stillbirths choline and glutamate in SP differed (P < 0.05) between per litter (median ± SD; for citrate: 7.10 mmol/L ± 1.738 vs. groups, displaying higher levels in SP-samples from 4.95 mmol/L ± 2.192; for creatine: 0.54 mmol/L ± 0.158 vs. boars with increased litter size deviation than in SP- 0.31 mmol/L ± 0.156; for phenylalanine: 0.03 mmol/L ± 0.011 samples from boars with reduced litter size deviation vs. 0.02 mmol/L ± 0.012; for tyrosine: 0.03 mmol/L ± 0.017 (median ± SD; for carnitine: 0.82 mmol/L ± 0.223 vs. vs. 0.02 mmol/L ± 0.016, respectively). In an opposite 0.43 mmol/L ± 0.244; for glutamate: 1.71 mmol/L ± 0.437 manner, malonate was found to be higher (P < 0.05) vs. 1.33 mmol/L ± 0.607; for hypotaurine: 2.85 mmol/L ± in SP-samples from boars with increased stillbirths 0.604 vs. 1.77 mmol/L ± 0.813; for sn-glycero-3-phospho- per litter (median ± SD; 0.16 mmol/L ± 0.064) than in choline: 6.45 mmol/L ± 1.373 vs. 4.69 mmol/L ± 1.932, those from boars with decreased stillbirths per litter respectively). Concentration of glucose in SP also (median ± SD; 0.09 mmol/L ± 0.054). Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 6 of 15 Fig. 2 A sPLS-DA analysis for farrowing rates deviation showing sample distribution in component 1 and component 2. Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1525 inseminations. The colored areas represent the 95% confidence interval. Boars classified with positive farrowing rate deviation (ranging from 2.82 to 7.54; n = 4) are represented in red and those with negative farrowing rate deviation (ranging from − 2.80 to − 1.60; n = 4) are shown in blue. Each dot symbolizes an ejaculate. B Loading plot for components 1 and 2 for the sPLS-DA model. Variables are ranked by the absolute values of their loadings. C Receiver operating characteristic (ROC) curve analysis for lactate concentration in seminal plasma and farrowing rate deviation. Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1,525 inseminations. The plot shows the ability of a given metabolite to discriminate farrowing rate of semen doses. AUC: area under the curve; CI: confidence interval Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 7 of 15 Fig. 3 A sPLS-DA analysis for litter size deviation showing sample distribution in component 1 and component 2. Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1525 inseminations. The colored areas represent the 95% confidence interval. Boars with decreased litter size deviation (ranging from − 0.40 to 0.02; n = 4) are represented in blue and those with increased litter size deviation (ranging from 0.11 to 0.52; n = 4) are shown in red. Each dot symbolizes an ejaculate. B Loading plot for components 1 and 2 for the sPLS-DA model. Variables are ranked by the absolute values of their loadings. C Receiver operating characteristic (ROC) curve analysis for carnitine, glucose, sn-glycero-3-phosphocholine, glutamate and hypotaurine concentrations in seminal plasma and litter size deviation. Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1,525 inseminations. The plot shows the ability of the metabolites to discriminate litter size of semen doses. AUC: area under the curve; CI: confidence interval Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 8 of 15 Regarding the sPLS-DA analysis (Fig. 4A), the first two fumarate showed an AUC of 0.767. Considering these components were found to explain 51.8% of the total results, the ROC curve revealed that while malonate had variance. Moreover, two different groups were observed: a good discriminant value for predicting gestation dur- while SP-samples from boars with decreased deviation in ation, fumarate showed a fair discriminant predictive the number of stillbirths per litter (blue) were mainly in- value for this parameter. fluenced by both components, SP-samples from boars with increased deviation (red) were influenced by the Discussion second component. The loadings plot revealed that the In the last years, metabolite identification and quantifi- first component was mainly influenced by creatine and cation for male infertility assessment has become an malonate, and the second component by malonate, emerging area of research [9, 10, 28]. In this field, NMR benzoate and formate (Fig. 4B). ROC curve analysis spectroscopy is one of the three most common analytical showed that the AUC was significant (P < 0.05; Fig. 4C) methods for metabolite profiling [28]. The present re- for citrate, creatine, malonate, phenylalanine and tyro- port evaluated the potential relationship between SP- sine. Specifically, citrate exhibited an AUC of 0.743, cre- metabolite concentrations and AI outcomes in pigs atine displayed an AUC of 0.812, malonate showed an using NMR approaches, figuring out which SP- AUC of 0.868, phenylalanine displayed an AUC of 0.750, metabolites could be used as in vivo fertility biomarkers. and tyrosine showed an AUC of 0.753. Therefore, the Specifically, this study demonstrated that: i) the concen- ROC curve revealed that while creatine and malonate tration of SP-lactate was related to farrowing rate; ii) had a good discriminant value for predicting stillbirths concentrations of carnitine, hypotaurine, sn-glycero-3- per litter (as their AUC ranged from 0.8 to 0.9), citrate, phosphocholine glutamate and glucose in SP were asso- creatine and tyrosine displayed a fair discriminant ciated with litter size; iii) concentrations of citrate, creat- strength (as their AUC ranged from 0.7 to 0.8). ine, malonate, phenylalanine and tyrosine in SP were related to the number of stillbirths per litter; and iv) Relationship between SP-metabolites and duration of concentrations of malonate and fumarate in SP were as- gestation sociated to gestation duration. Boars were classified into two groups depending on the In accordance with our previous report [29], this study deviation of gestation duration from the median, i.e. identified and quantified a total of 24 SP-metabolites. In shorter gestation duration (ranging from − 0.85 to 0.03; addition, the results of the present work showed that n = 4) and longer gestation duration (ranging from 0.10 several SP-metabolites were related to specific in vivo to 0.52; n = 4) deviation. Concentration of malonate in fertility parameters. However, since all the relationships SP differed (P < 0.05) between groups, showing higher were found to be moderate (as Pearson correlation coef- levels in SP-samples from boars with longer gestation ficients were lower than 0.6) and no SP-metabolite pat- duration deviation (median ± SD; 0.16 mmol/L ± 0.072) tern for specific fertility parameters was observed using compared to those from boars with shorter gestation Bayesian multiple logistic models, sPLS-DA and ROC duration deviation (median ± SD; 0.09 mmol/L ± 0.030). analysis were run. Using these statistic tools, potential On the contrary, fumarate exhibited higher levels (P < biomarkers for all the assessed reproductive performance 0.05) in SP-samples from boars with shorter gestation variables were identified. duration (median ± SD; 0.01 mmol/L ± 0.002) than in The results of the present study evidenced that lactate those from boars with longer gestation duration (me- was the only SP-metabolite related to farrowing rate. dian ± SD; 0.004 mmol/L ± 0.002). These results differ from those reported by Zhang et al. sPLS-DA analysis for gestation duration showed that who found that several amino acids and D-biotin in SP the first two components explained 33.6% of the total were related with conception rates in pigs [20]. Differ- variance (Fig. 5A). Again, two different groups were ences in (1) the analytical method (mass spectrometry identified: while SP-samples from boars classified with a vs. NMR) and in (2) the analysis of fertility records, since shorter deviation in the gestation duration (blue) were the work of Zhang et al. did not take other parameters mainly affected by the second component, SP-samples that could influence conception rates (sows, farms …) from boars with longer deviation (red) were equally in- into consideration [20], may contribute to explain the di- fluenced by both components. The loadings plot re- vergent results between both studies. In the present vealed that the first component was strongly influenced work, the highest lactate concentrations were found in by malonate, and the second component by glutamate, SP-samples from boars classified with negative farrowing sn-glycero-3-phosphocholine and carnitine (Fig. 5B). For rate deviation. It is well known that lactate is one of the these SP-metabolites, both malonate and fumarate main non-monosaccharide substrates for sperm in bulls showed a significant ROC curve (P < 0.05; Fig. 5C). Spe- [30], stallions [31], men [32] and boars [32, 33]. Non- cifically, malonate exhibited an AUC of 0.868 and oxidative metabolism of pig sperm has been shown to Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 9 of 15 Fig. 4 A sPLS-DA analysis for the number stillbirths per litter deviation showing sample distribution in component 1 and component 2. Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1,525 inseminations. The colored areas represent the 95% confidence interval. Boars with decreased stillbirths per litter deviation (ranging from − 0.10 to 0.02; n = 4) are represented in blue and those with increased stillbirths per litter deviation (ranging from 0.05 to 0.14; n = 4) are shown in red. Each dot symbolizes an ejaculate. B Loading plot for components 1 and 2 for the sPLS-DA model. Variables are ranked by the absolute values of their loadings. C Receiver operating characteristic (ROC) curve analysis for citrate, creatine, phenylalanine, tyrosine and malonate and stillbirths per litter. They show the ability of metabolites to discriminate the number of stillbirths per litter after artificial insemination with semen doses. AUC: area under the curve; CI: confidence interval Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 10 of 15 Fig. 5 A sPLS-DA analysis for gestation duration deviation showing sample distribution in component 1 and component 2. Seminal plasma samples from entire ejaculates (24 ejaculates) of eight artificial insemination-boars (three ejaculates per boar) were used. Fertility parameters were recorded from 1,525 inseminations. The colored areas represent the 95% confidence interval. Boars with longer gestation duration deviation (ranging from 0.10 to 0.52; n = 4) are represented in red and those with shorter gestation duration deviation (ranging from − 0.85 to 0.03; n = 4) are shown in blue. Each dot symbolizes an ejaculate. B Loading plot for components 1 and 2 for the sPLS-DA model. Variables are ranked by the absolute values of their loadings. C Receiver operating characteristic (ROC) curve analysis for malonate and fumarate and pregnancy duration. They show the ability of metabolites to discriminate the number of stillbirths per litter after artificial insemination with semen doses. AUC: area under the curve; CI: confidence interval consume lactate [34], which is transformed into pyruvate production, thereby leading to lower SP-lactate concen- through lactate dehydrogenase to produce ATP [35]. tration, as confirmed by the present study. However, it is Considering these findings, one could assume that sperm worth mentioning that these results are not in agree- from boars classified with positive farrowing rate devi- ment with previous findings reported in cattle, in which ation could better metabolize lactate for energy the highest lactate levels were found in the SP of high Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 11 of 15 fertility bulls [30], and in men, as infertile patients had Hypotaurine is an antioxidant present in human SP lower SP-lactate levels than fertile controls [16, 18, 36, and sperm [50, 51]. The present study found a positive 37]. These differences could be attributed to: (1) differ- influence of SP-hypotaurine on litter size. This relation- ent metabolic sperm strategies, as while pig sperm can ship could be driven by both an effect on sperm and/or use lactate as an energy source, the rate between oxida- oocyte fertilization. With regard to sperm, the addition tive phosphorylation and glycolysis is higher in bovine of hypotaurine to cryopreservation media has been re- sperm [38]; and/or (2) differences in SP composition as ported to exert a positive effect on sperm quality and a result of differences in mating strategies between functionality parameters in sheep [52] and humans [51, species [39, 40]. 53]. In addition, sperm from bulls with high fertility Regarding the litter size, the present study showed that records also have high hypotaurine levels [54]. While, concentrations of glucose, carnitine, hypotaurine, sn- considering all this evidence, one could surmise that SP- glycero-3-phosphocholine and glutamate were related hypotaurine has a positive impact on pig sperm physi- with this fertility parameter. Interestingly, the ROC ology, further studies are required to confirm this hy- curve revealed that these SP-metabolites had a discrim- pothesis. On the other hand, supplementation of in vitro inating ability to predict the litter size, so that all the culture media with hypotaurine increases embryo cleav- four could be considered as promising biomarkers for age and, in consequence, embryo development in bovine this AI outcome. [55] and improves the intracellular oxidative status of It is well known that glucose is one of the main mono- pre-implantational porcine embryos [56]. Thus, SP- saccharides used by mammalian sperm to produce en- hypotaurine could also affect early embryo development ergy [32, 41]. The present study reported that boars stages, thus increasing litter size. classified with an increased litter size deviation exhibited Glutamate is an amino acid involved in cellular energy lower glucose concentrations in their SP than those with production and in the synthesis of many other amino a decreased litter size deviation. Similar findings were re- acids and nucleotides [18]. Low levels of SP-glutamate ported in humans, in which men with idiopathic infertil- have been related to several forms of infertility in ity had higher glucose levels in their SP compared to humans [18, 37]. In agreement with these results, the healthy individuals [17]. The most feasible explanation present study found that higher levels of SP-glutamate for such findings would be that sperm from boars with a were related to increased litter size deviation. Based on decreased litter size would consume less glucose from these data, the effect of glutamate on AI outcomes could SP; thus, glucose would be extracellularly accumulated. be driven by its repercussion on sperm, as equine intra- This hypothesis would be in agreement with the existing cellular glutamate has been proposed: i) to contribute to literature, as the supplementation of semen extenders sperm functionality through its metabolization via non- with glucose has been reported to increase sperm motil- canonical pathways; and ii) to be exchanged for extracel- ity and ATP concentration in humans [42]. Taken these lular cysteine to produce reduced glutathione [57]. data together, it could be suggested that low levels of Nonetheless, before could glutamate be used as a litter glucose in SP are beneficial for both sperm physiology size biomarker, the aforementioned hypothesis should and reproductive performance. be tested in the pig. Carnitine is an antioxidant that has been widely Finally, sn-glycero-3-phosphocholine, which is in- demonstrated to be involved in mammalian sperm volved in glycerophospholipid metabolism, has been re- motility [43, 44]. Moreover, a protective role of this ported to play a vital role in sperm capacitation and antioxidant on DNA and plasma membrane oxidation acrosome reaction in rats [58]. The results of the present damage in humans [44, 45] and pigs [46] has also study showed a positive relationship between sn-glycero- been reported. In addition, dietary carnitine supple- 3-phosphocholine concentration in SP and high litter mentation in boars has been proved to improve size. These results seem to agree with previous studies sperm quality parameters [47, 48]. In agreement with performed in other species, in which infertile men were these studies, the results reported herein indicate that observed to exhibit lower sn-glycero-3-phosphocholine boars with an increased litter size deviation exhibit levels in their SP compared to their fertile counterparts the highest SP-concentration of this metabolite. These [59]. In addition, it has been reported that rat sperm results are in accordance with the study of Zöpfgen head accumulates lipid metabolites as a result of sn- et al., who found that infertile men had lower SP- glycero-3-phosphocholine metabolism during acrosome levels of carnitine than their fertile counterparts [49]. reaction, which could have an involvement in sperm- Nevertheless, this result, together with the aforemen- oocyte interaction and even in gamete fusion [58]. Con- tioned findings, open the possibility of using the sidering all these findings, further research addressing measurement of carnitine in SP as a potential litter the specific role played by SP-sn-glycero-3-phosphocho- size biomarker. line in pig fertility is warranted. Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 12 of 15 The present study also evaluated the relationship be- fumarate exhibited lower levels in that group. However, tween SP-metabolites and stillbirths per litter. Citrate, further studies for fumarate validation should be con- creatine, phenylalanine and tyrosine were observed to be ducted, as no information regarding the effect of this promising biomarkers for stillbirths per litter due to metabolite on sperm physiology or fertilizing ability has their ROC curves. In this sense, citrate is involved in the been published. On the other hand, malonate may have Krebs cycle, which is the most relevant metabolic path- a double predictive value for both stillbirths per litter way for energy production [60]. The results of the size and the estimation of gestation duration. Malonate present study revealed that high SP concentrations of is an intermediate metabolite of the Krebs cycle that in- this metabolite were related to a low number of still- hibits ROS production via competition for succinate de- births per litter. These results came as a surprise consid- hydrogenase [68, 69]. The present study identified a ering that low levels of SP-citrate have been observed in positive relationship between SP-malonate levels and high-fertility bulls [21] and SP-citrate has been widely both stillbirths per litter size and gestation duration. A proposed as a biomarker for different human infertility similar negative influence of SP-malonate has also been forms [18, 36, 61]. Considering the opposite trend of the found in humans, as infertile patients exhibited higher results presented herein, the exact mechanism through malonate levels than fertile controls [70]. On the other which SP-citrate could positively influence AI outcomes hand, malonate can act as protein post-translational needs to be clarified in future studies. modification [71]. Based on these findings, while no Creatine is involved in the regulation of ATP and both studies have been conducted to evaluate the influence of the supplementation of in vitro fertilization medium malonate on sperm physiology, it could be posited that a with creatine [62] and the presence of this metabolite in high SP-malonate concentration could: i) modify key SP [63] have been reported to influence sperm physi- proteins involved in gamete interaction or even embryo ology in terms of motility and viability in humans [62, development, or ii) inhibit the Krebs cycle. In any case, 63]. Interestingly, the present study found that high SP- the fact that malonate relates to two distinct in vivo fer- creatine concentration was associated with decreased tility parameters reinforces its potential value as a pre- stillbirths per litter deviation. This result may be ex- dictor of AI outcomes in pig SP. plained by the fact that creatine has been found to en- As aforementioned, the differences found between the hance fertilization and promote blastocyst and normal results of the present research and those reported in embryo development [62]. Consequently, although this other metabolomic studies conducted in pigs or in other should be further confirmed, it could be posited that species may be due to several factors: i) differences in high SP-creatine has a positive effect on both gametes, the sensitivity of the metabolomic approaches; ii) varia- thus improving AI outcomes and decreasing the number tions in the preparation of samples; iii) the species- of stillbirths per litter. specific role of seminal metabolites in fertility; and iv) Phenylalanine and tyrosine, amino acids involved in the use of non-comparable fertility parameters. For this the same metabolic pathway [64], were found to be reason, although -omics approaches are powerful tools, higher in SP-samples from boars with decreased still- they should be used as a first steppingstone in the re- births per litter deviation. While, to the best of our search of (in-)fertility biomarkers [10, 29, 72]. In effect, knowledge, no information about the effect of tyrosine while the main strength of the present work is that a set on sperm physiology has been reported, phenylalanine is of SP-metabolites has been proposed to predict AI out- known to stimulate the ability of human sperm to cap- comes, they should all be further validated using a acitate and undergo acrosomal exocytosis [65]. In cattle, higher number of individuals and other approaches to phenylalanine levels in SP are positively related to post- overcome the intrinsic limitations of-omics approaches. thaw sperm viability, suggesting that this amino acid Following this, the measurement of metabolites in SP could be involved in oxidoreductase and oxidant reac- could be potentially used as an accurate fertility test to tions [66]. Interestingly, SP-tyrosine has also been found select boars before they are included in an AI-program. to contribute to the total antioxidant capacity of SP [67]. Moreover, future research needs to be conducted to as- If these findings were confirmed in pigs, concentrations sess i) the specific role of each SP-metabolite in male of phenylalanine and tyrosine in SP would also appear as fertility, and ii) whether supplementing AI-extenders exerting a beneficial effect upon sperm through regula- with specific metabolites can improve the fertility poten- tion of reactive oxygen species (ROS) and could be used tial of semen doses. to predict fertility outcomes in porcine. Finally, the relationship between gestation duration Conclusions and concentration of SP-metabolites was also investi- The metabolite profiling of pig SP using NMR spectros- gated, and whereas malonate showed higher levels in copy allowed the identification and quantification of 24 SP-samples from boars with longer gestation duration, metabolites. The results evidenced that 13 of these Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 13 of 15 metabolites were related with AI outcomes, pointing out Author details 1 to putative in vivo fertility biomarkers. Specifically, Biotechnology of Animal and Human Reproduction (TechnoSperm), Institute of Food and Agricultural Technology, University of Girona, ES-17003 Girona, lactate could be used as a farrowing rate indicator; Spain. 2Department of Biology, Unit of Cell Biology, Faculty of Sciences, carnitine, hypotaurine, sn-glycero-3-phosphocholine, University of Girona, ES-17003 Girona, Spain. 3Centre d’Estudis Avançats de glutamate and glucose could predict litter size; citrate, Blanes (CEAB), Spanish Research Council (CSIC), ES-17300 Girona, Spain. 4 Magnetic Nuclear Resonance Facility, Autonomous University of Barcelona, creatine, phenylalanine, tyrosine and malonate would be Bellaterra, ES-08193 Cerdanyola del Vallès, Spain. 5Department of Animal biomarkers for the number of stillbirths per litter; and, Medicine and Surgery, Faculty of Veterinary Medicine, University of Murcia, finally, malonate and fumarate would anticipate the ES-30100 Murcia, Spain. 6Equine Reproduction Service, Department of Animal Medicine and Surgery, Faculty of Veterinary Medicine, Autonomous duration of gestation. University of Barcelona, Bellaterra, ES-08193 Cerdanyola del Vallès, Spain. 7 Department of Veterinary Medical Sciences, University of Bologna, IT-40064 Abbreviations Ozzano dell’Emilia, Bologna, Italy. AI: Artificial insemination; AUC: Area under the curve; NMR: Nuclear magnetic resonance; PCA: Principal components analysis; ROC: Receiver operating Received: 9 May 2021 Accepted: 5 September 2021 characteristic; ROS: Reactive oxygen species; SP: Seminal plasma; sPLS- DA: Sparse partial least square discriminant analysis References Supplementary Information 1. Waberski D, Riesenbeck A, Schulze M, Weitze KF, Johnson L. Application of The online version contains supplementary material available at https://doi. preserved boar semen for artificial insemination: past, present and future org/10.1186/s40104-021-00636-5. challenges. Theriogenology. 2019;137:2–7. https://doi.org/10.1016/j. theriogenology.2019.05.030. Additional file 1 Supplementary Fig. 1 H-NMR (noesygppr1d) profile 2. Roca J, Parrilla I, Bolarin A, Martinez EA, Rodriguez-Martinez H. Will AI in pigs (600 MHz) from 0 to 8 ppm of pig seminal plasma. become more efficient? Theriogenology Elsevier Ltd. 2016;86(1):187–93. https://doi.org/10.1016/j.theriogenology.2015.11.026. Additional file 2 Supplementary Fig. 2 (A-D). Bayesian multiple 3. Broekhuijse MLWJ, Feitsma H, Gadella BM. Field data analysis of boar semen logistic regression models for all the in vivo fertility parameters. The quality. Reprod Domest Anim. 2011;46(Suppl 2):59–63. https://doi.org/1 distribution of the coefficients (X axis) is depicted for each metabolite (Y 0.1111/j.1439-0531.2011.01861.x. axis). The coefficient distributions depict their effect on the model, as 4. Roca J, Broekhuijse MLWJ, Parrilla I, Rodriguez-Martinez H, Martinez EA, well as their associated uncertainties (credible intervals). Thus, changes in Bolarin A. Boar differences in artificial insemination outcomes: can they be one unit on the coefficient value has a multiplicative effect on the log- minimized? Reprod Domest Anim. 2015;50(Suppl 2):48–55. https://doi.org/1 odds of the prediction, equal to the value of the coefficient. Blue lines 0.1111/rda.12530. represent the 95% credible intervals, boxes show the 50% credible inter- 5. Rodriguez-Martinez H, Kvist U, Saravia F, Wallgren M, Johannisson A, Sanz L, vals, and dots are the distribution median. et al. The physiological roles of the boar ejaculate. Soc Reprod Fertil Suppl. 2009;66:1–21. Acknowledgements 6. Rodríguez-Martínez H, Kvist U, Ernerudh J, Sanz L, Calvete JJ. Seminal The authors of the manuscript would like to thank AIM Iberica (Topigs plasma proteins: what role do they play? Am J Reprod Immunol. 2011;66: Norsvin Iberica, Spain) for boar ejaculates. 11–22. https://doi.org/10.1111/j.1600-0897.2011.01033.x. 7. Recuero S, Fernandez-Fuertes B, Bonet S, Barranco I, Yeste M. Potential of seminal plasma to improve the fertility of frozen-thawed boar spermatozoa. Theriogenology. Authors’ contributions 2019;137:36–42. https://doi.org/10.1016/j.theriogenology.2019.05.035. Conceptualization: YM-O, MY and IB; Methodology: YM-O, PF-L, PN and IB; 8. Morgan HL, Watkins AJ. The influence of seminal plasma on offspring Formal analysis and investigation: YM-O, PF-L, AD-B, JR-M, MY and IB; Writing development and health. Semin. Cell Dev. Biol. 2020;97:131–7. - original draft preparation: YM-O; Writing - review and editing: PN, JR-M, JM, 9. Deepinder F, Chowdary HT, Agarwal A. Role of metabolomic analysis of JR, MY and IB; Funding acquisition: JM, JR and MY; Supervision: IB and MY. biomarkers in the management of male infertility. Expert Rev Mol Diagn. All authors have read and agreed to the published version of the manuscript. 2007;7(4):351–8. https://doi.org/10.1586/14737159.7.4.351. The author(s) read and approved the final manuscript. 10. Kovac JR, Pastuszak AW, Lamb DJ. The use of genomics, proteomics, and metabolomics in identifying biomarkers of male infertility. Fertil Steril. 2013; Funding 99(4):998–1007. https://doi.org/10.1016/j.fertnstert.2013.01.111. The present study was funded by the Ministry of Science and Innovation, 11. Mehrparavar B, Minai-Tehrani A, Arjmand B, Gilany K. Metabolomics of male Spain (Grants: RYC-2014-15581, AGL2017–88329-R and FJCI-2017-31689), the infertility: a new tool for diagnostic tests. J Reprod Infertil. 2019;20(2):64–9. Seneca Foundation Murcia, Spain (19892/GERM-15) and the Regional Govern- 12. Long JA. The “omics” revolution: use of genomic, transcriptomic, proteomic ment of Catalonia (Grants: 2017-SGR-1229 and 2020-FI-B-00412). and metabolomic tools to predict male reproductive traits that impact fertility in livestock and poultry. Anim Reprod Sci. 2020;220:106354. https:// Availability of data and materials doi.org/10.1016/j.anireprosci.2020.106354. All data generated or analyzed during this study are included in this 13. Panner Selvam MK, Finelli R, Agarwal A, Henkel R. Proteomics and published article (and its supplementary information files). metabolomics - current and future perspectives in clinical andrology. Andrologia. 2021;53(2):e13711. https://doi.org/10.1111/and.13711. Declarations 14. Minai-Tehrani A, Jafarzadeh N, Gilany K. Metabolomics: a state-of-the-art technology for better understanding of male infertility. Andrologia. 2016; Ethics approval and consent to participate 48(6):609–16. https://doi.org/10.1111/and.12496. Ethical review and approval for animal participants was not required as 15. Shulaev V. Metabolomics technology and bioinformatics. Brief Bioinform. samples were provided by the AI-Centre and authors did not manipulate 2006;7(2):128–39. https://doi.org/10.1093/bib/bbl012. any animals. 16. Hamamah S, Seguin F, Barthelemy C, Akoka S, Le Pape A, Lansac J, et al. 1H nuclear magnetic resonance studies of seminal plasma from fertile and Consent for publication infertile men. J Reprod Fertil. 1993;97(1):51–5. https://doi.org/10.1530/jrf.0. Not applicable. 0970051. 17. Qiao S, Wu W, Chen M, Tang Q, Xia Y, Jia W, et al. Seminal plasma Competing interests metabolomics approach for the diagnosis of unexplained male infertility. The authors declare that they have no competing interests. PLoS One. 2017;12(8):1–13. https://doi.org/10.1371/journal.pone.0181115. Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 14 of 15 18. Mumcu A, Karaer A, Dogan B, Tuncay G. Metabolomics analysis of seminal 39. Hunter RHF. Sperm transport and reservoirs in the pig oviduct in relation to plasma in patients with idiopathic Oligoasthenoteratozoospermia using the time of ovulation. J Reprod Fertil. 1981;63(1):109–17. https://doi.org/10.1 high-resolution NMR spectroscopy. Andrology. 2020;8(2):450–6. https://doi. 530/jrf.0.0630109. org/10.1111/andr.12707. 40. Druart X, Rickard JP, Mactier S, Kohnke PL, Kershaw-Young CM, Bathgate R, 19. Xu Y, Lu H, Wang Y, Zhang Z, Wu Q. Comprehensive metabolic profiles of et al. Proteomic characterization and cross species comparison of seminal plasma with different forms of male infertility and their correlation mammalian seminal plasma. J Proteome. 2013;91:13–22. https://doi.org/10.1 with sperm parameters. J Pharm Biomed Anal. 2020;177:112888. https://doi. 016/j.jprot.2013.05.029. org/10.1016/j.jpba.2019.112888. 41. Albarracín JL, Fernández-Novell JM, Ballester J, Rauch MC, Quintero-Moreno 20. Ting ZY, Liu Y, Lin LH, Qian XQ, Hua LZ, Gang WX. Metabolomic differences A, Peña A, et al. Gluconeogenesis-linked glycogen metabolism is important of seminal plasma between boars with high and low average conception in the achievement of in vitro capacitation of dog spermatozoa in a rates after artificial insemination. Reprod Domest Anim. 2021;56(1):161–71. medium without glucose. Biol Reprod. 2004;71(5):1437–45. https://doi.org/1 https://doi.org/10.1111/rda.13861. 0.1095/biolreprod.104.029041. 21. Kumar A, Kroetsch T, Blondin P, Anzar M. Fertility-associated metabolites in bull 42. Williams AC, Ford WCL. The role of glucose in supporting motility and seminal plasma and blood serum: 1H nuclear magnetic resonance analysis. capacitation in human spermatozoa. J Androl. 2001;22(4):680–95. Mol Reprod Dev. 2015;82(2):123–31. https://doi.org/10.1002/mrd.22450. 43. Jeulin C, Lewin LM. Role of free L-carnitine and acetyl-L-carnitine in post- 22. Velho ALC, Menezes E, Dinh T, Kaya A, Topper E, Moura AA, et al. gonadal maturation of mammalian spermatozoa. Hum Reprod Update. Metabolomic markers of fertility in bull seminal plasma. PLoS One. 2018; 1996;2(2):87–102. https://doi.org/10.1093/humupd/2.2.87. 13(4):e0195279. https://doi.org/10.1371/journal.pone.0195279. 44. Chavoshi Nezhad N, Vahabzadeh Z, Allahveisie A, Rahmani K, Raoofi A, 23. Broekhuijse MLWJ, Šoštarić E, Feitsma H, Gadella BM. Relationship of flow Rezaie MJ, et al. The effect of L-carnitine and coenzyme Q10 on the sperm cytometric sperm integrity assessments with boar fertility performance motility, DNA fragmentation, chromatin structure and oxygen free radicals under optimized field conditions. J Anim Sci. 2012;90(12):4327–36. https:// during, before and after freezing in Oligospermia men. Urol J. 2021;18(3):330–6. doi.org/10.2527/jas.2012-5040. 45. Naderi Noreini S, Malmir M, Ghafarizadeh A, Faraji T, Bayat R. Protective 24. Goodrich B, Gabry J, Ali I, Brilleman S. Rstanarm: Bayesian applied regression effect of L-carnitine on apoptosis, DNA fragmentation, membrane integrity modeling via Stan. 2020. Available from: https://mc-stan.org/rstanarm/ and lipid peroxidation of spermatozoa in the asthenoteratospermic men. 25. Rohart F, Gautier B, Singh A, Lê Cao K-A. mixOmics: An R package for ‘omics feature Andrologia. 2020;53(2):e13932. https://doi.org/10.1111/and.13932. selection and multiple data integration. PLOS Comput Biol. 2017;13:e1005752. 46. Yang K, Wang N, Guo HT, Wang JR, Sun HH, Sun LZ, et al. Effect of L-carnitine 26. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an on sperm quality during liquid storage of boar semen. Asian-Australasian open-source package for R and S+ to analyze and compare ROC curves. J Anim Sci. 2020;33(11):1763–9. https://doi.org/10.5713/ajas.19.0455. BMC Bioinformatics. 2011;12(1):77. https://doi.org/10.1186/1471-2105-12-77. 47. Yeste M, Sancho S, Briz M, Pinart E, Bussalleu E, Bonet S. A diet 27. Lê Cao KA, Boitard S, Besse P. Sparse PLS discriminant analysis: biologically supplemented with l-carnitine improves the sperm quality of Piétrain but relevant feature selection and graphical displays for multiclass problems. not of Duroc and large white boars when photoperiod and temperature BMC Bioinformatics. 2011;12(1):253. https://doi.org/10.1186/1471-2105-12-253. increase. Theriogenology. 2010;73(5):577–86. https://doi.org/10.1016/j. 28. Emwas AH, Roy R, McKay RT, Tenori L, Saccenti E, Nagana Gowda GA, et al. theriogenology.2009.10.013. NMR spectroscopy for metabolomics research. Metabolites. 2019;9(7):123. 48. Kozink DM, Estienne MJ, Harper AF, Knight JW. Effects of dietary L-carnitine 29. Mateo-Otero Y, Fernández-López P, Ribas-Maynou J, Roca J, Miró J, Yeste M, supplementation on semen characteristics in boars. Theriogenology. 2004; et al. Metabolite profiling of pig seminal plasma identifies potential 61(7-8):1247–58. https://doi.org/10.1016/j.theriogenology.2003.07.022. biomarkers for sperm resilience to liquid preservation. Front cell Dev Biol. 49. Zöpfgen A, Priem F, Sudhoff F, Jung K, Lenk S, Loening SA, et al. 2021;9:669974. https://doi.org/10.3389/fcell.2021.669974. Relationship between semen quality and the seminal plasma components 30. Menezes EB, Velho ALC, Santos F, Dinh T, Kaya A, Topper E, et al. carnitine, alpha-glucosidase, fructose, citrate and granulocyte elastase in Uncovering sperm metabolome to discover biomarkers for bull fertility. infertile men compared with a normal population. Hum Reprod. 2000;15(4): BMC Genomics. 2019;20(1):1–16. https://doi.org/10.1186/s12864-019-6074-6. 840–5. https://doi.org/10.1093/humrep/15.4.840. 31. Darr CR, Varner DD, Teague S, Cortopassi GA, Datta S, Meyers SA. Lactate 50. Holmes RP, Goodman HO, Shihabi ZK, Jarow JP. The taurine and and pyruvate are major sources of energy for stallion sperm with dose Hypotaurine content of human semen. J Androl. 1992;13(3):289–92. effects on mitochondrial function, motility, and ROS production. Biol 51. Seify M, Zarabadipour M, Ghaleno LR, Alizadeh AR, Rezazadeh VM. The anti- Reprod. 2016;95(2):34. https://doi.org/10.1095/biolreprod.116.140707. oxidant roles of taurine and Hypotaurine on acrosome integrity, HBA and 32. Rodriguez-Gil J. Mammalian sperm energy resources management and HSPA2 of the human sperm during vitrification and post warming in two survival during conservation in refrigeration. Reprod Domest Anim. 2006; different temperature. Cryobiology. 2019;90:89–95. https://doi.org/10.1016/j. 41(s2):11–20. https://doi.org/10.1111/j.1439-0531.2006.00765.x. cryobiol.2019.07.004. 33. Paventi G, Lessard C, Bailey JL, Passarella S. In boar sperm capacitation L- 52. Bucak MN, Ateşşahin A, Varişli Ö, Yüce A, Tekin N, Akçay A. The influence of lactate and succinate, but not pyruvate and citrate, contribute to the trehalose, taurine, cysteamine and hyaluronan on ram semen. Microscopic mitochondrial membrane potential increase as monitored via safranine O and oxidative stress parameters after freeze-thawing process. fluorescence. Biochem Biophys Res Commun. 2015;462(3):257–62. https:// Theriogenology. 2007;67(5):1060–7. https://doi.org/10.1016/j. doi.org/10.1016/j.bbrc.2015.04.128. theriogenology.2006.12.004. 34. Aalbers JG, Mann T, Polge C. Metabolism of boar semen in relation to 53. Brugnon F, Ouchchane L, Pons-Rejraji H, Artonne C, Farigoule M, Janny L. sperm motility and survival. J Reprod Fertil. 1961;2(1):42–53. https://doi.org/1 Density gradient centrifugation prior to cryopreservation and hypotaurine 0.1530/jrf.0.0020042. supplementation improve post-thaw quality of sperm from infertile men 35. Medrano A, Fernández-Novell JM, Ramió L, Alvarez J, Goldberg E, Rivera with oligoasthenoteratozoospermia. Hum Reprod. 2013;28(8):2045–57. MM, et al. Utilization of citrate and lactate through a lactate dehydrogenase https://doi.org/10.1093/humrep/det253. and ATP-regulated pathway in boar spermatozoa. Mol Reprod Dev. 2006; 54. Saraf KK, Kumaresan A, Dasgupta M, Karthikkeyan G, Prasad TSK, Modi PK, 73(3):369–78. https://doi.org/10.1002/mrd.20414. et al. Metabolomic fingerprinting of bull spermatozoa for identification of 36. Gupta A, Mahdi AA, Ahmad MK, Shukla KK, Jaiswer SP, Shankhwar SN. 1H fertility signature metabolites. Mol Reprod Dev. 2020;87(6):692–703. https:// NMR spectroscopic studies on human seminal plasma: a probative doi.org/10.1002/mrd.23354. discriminant function analysis classification model. J Pharm Biomed Anal. 55. Miller GF, Gliedt DW, Rakes JM, Rorie RW. Addition of penicillamine, 2011;54(1):106–13. https://doi.org/10.1016/j.jpba.2010.07.021. hypotaurine and epinephrine (PHE) or bovine oviductal epithelial cells 37. Gupta A, Mahdi AA, Shukla KK, Ahmad MK, Bansal N, Sankhwar P, et al. (BOEC) alone or in combination to bovine in vitro fertilization medium Efficacy of Withania somnifera on seminal plasma metabolites of infertile increases the subsequent embryo cleavage rate. Theriogenology. 1994;41(3): males: a proton NMR study at 800 MHz. J Ethnopharmacol. 2013;149(1):208– 689–96. https://doi.org/10.1016/0093-691X(94)90178-L. 14. https://doi.org/10.1016/j.jep.2013.06.024. 56. Suzuki C, Yoshioka K, Sakatani M, Takahashi M. Glutamine and hypotaurine 38. Storey BT. Mammalian sperm metabolism: Oxygen and sugar, friend and improves intracellular oxidative status and in vitro development of porcine foe. Int J Dev Biol. 2008;52(5-6):427–37. https://doi.org/10.1387/ijdb.072522 preimplantation embryos. Zygote. 2007;15(4):317–24. https://doi.org/10.101 bs. 7/S0967199407004273. Mateo-Otero et al. Journal of Animal Science and Biotechnology (2021) 12:113 Page 15 of 15 57. Ortiz-Rodríguez JM, Martín-Cano FE, Gaitskell-Phillips G, Silva A, Tapia JA, Gil MC, et al. The SLC7A11: sperm mitochondrial function and non-canonical glutamate metabolism. Reproduction. 2020;160(6):803–18. https://doi.org/1 0.1530/REP-20-0181. 58. Zanetti SR, Monclus MDLÁ, Rensetti DE, Fornés MW, Aveldaño MI. Differential involvement of rat sperm choline glycerophospholipids and sphingomyelin in capacitation and the acrosomal reaction. Enferm Infecc Microbiol Clin. 2010;28(12):1886–94. https://doi.org/10.1016/j.biochi.2010.08.015. 59. Mieusset R, Bujan L, Mansat AFP, Grandjean H, chap H. Glycerophosphocholine in seminal plasma of fertile and infertile men. Int J Androl. 1988;11(5):405–13. https://doi.org/10.1111/j.1365-2605.1988.tb01013.x. 60. Akram M. Citric acid cycle and role of its Intermediates in metabolism. Cell Biochem Biophys. 2014;68:475–8. 61. Hamamah S, Seguin F, Bujan L, Barthelemy C, Mieusset R, Lansac J. Quantification by magnetic resonance spectroscopy of metabolites in seminal plasma able to differentiate different forms of azoospermia. Hum Reprod. 1998;13(1):132–5. https://doi.org/10.1093/humrep/13.1.132. 62. Umehara T, Kawai T, Goto M, Richards JS, Shimada M. Creatine enhances the duration of sperm capacitation: a novel factor for improving in vitro fertilization with small numbers of sperm. Hum Reprod. 2018;33(6):1117–29. https://doi.org/10.1093/humrep/dey081. 63. Nasrallah F, Hammami MB, Omar S, Aribia H, Sanhaji H, Feki M. Semen creatine and creatine kinase activity as an indicator of sperm quality. Clin Lab. 2020;66(09/2020):1751–7. https://doi.org/10.7754/Clin.Lab.2020.191248. 64. Matthews DE. An overview of phenylalanine and tyrosine kinetics in humans. J Nutr. 2007;137(6):1549S–55S. https://doi.org/10.1093/jn/137.6.1549S. 65. Houston B, Curry B, Aitken RJ. Human spermatozoa possess an IL4I1 L- amino acid oxidase with a potential role in sperm function. Reproduction. 2015;149(6):587–96. https://doi.org/10.1530/REP-14-0621. 66. Ugur MR, Dinh T, Hitit M, Kaya A, Topper E, Didion B, et al. Amino acids of seminal plasma associated with freezability of bull sperm. Front Cell Dev Biol. 2020;7:1–14. https://doi.org/10.3389/fcell.2019.00347. 67. Van Overveld FWPC, Haenen GRMM, Rhemrev J, Vermeiden JPW, Bast A. Tyrosine as important contributor to the antioxidant capacity of seminal plasma. Chem Biol Interact. 2000;127(2):151–61. https://doi.org/10.1016/ S0009-2797(00)00179-4. 68. Tretter L, Patocs A, Chinopoulos C. Succinate, an intermediate in metabolism, signal transduction, ROS, hypoxia, and tumorigenesis. Biochim Biophys Acta Bioenerg. 1857;2016(8):1086–101. https://doi.org/10.1016/j.bba bio.2016.03.012. 69. Bowman CE, Wolfgang MJ. Role of the malonyl-CoA synthetase ACSF3 in mitochondrial metabolism. Adv. Biol. Regul. 2019;71:34–40. 70. Jayaraman V, Ghosh S, Sengupta A, Srivastava S, Sonawat HM, Narayan PK. Identification of biochemical differences between different forms of male infertility by nuclear magnetic resonance (NMR) spectroscopy. J Assist Reprod Genet. 2014;31(9):1195–204. https://doi.org/10.1007/s10815-014-0282-4. 71. Peng C, Lu Z, Xie Z, Cheng Z, Chen Y, Tan M, et al. The first identification of lysine malonylation substrates and its regulatory enzyme. Mol Cell Proteomics. 2011;10(12):1–12. https://doi.org/10.1074/mcp.M111.012658. 72. Goodacre R, Vaidyanathan S, Dunn WB, Harrigan GG, Kell DB. Metabolomics by numbers: acquiring and understanding global metabolite data. Trends Biotechnol. 2004;22(5):245–52. https://doi.org/10.1016/j.tibtech.2004.03.007.