Untargeted urine metabolomics suggests that ascorbic acid may serve as a promising biomarker for reduced feed intake in rabbits Authors: Pablo Jesús Marín-García, Miriam Piles, Juan Pablo Sánchez, Mariam Pascual, Lola Llobat, Juan José Pascual, Mette Skou Hedemann Published: 2024 ABSTRACT AbstractFeed restriction is a common nutritional practice in rabbit farming; however, decreased feed intake can also signal potential digestive disorders at an early stage. This study endeavors to investigate the impact of feed restriction on selected productive traits and the urinary metabolome of juvenile rabbits across diverse genetic backgrounds. Our objective is to identify potential biomarkers capable of detecting periods of fasting. A total of 48 growing rabbits were used from two genetic types: Prat line (selected for litter size at weaning, n = 24) and Caldes line (selected for post-weaning growth rate, n = 24). At 60 days of age, a digestibility trial was carried out. Changes in productive traits (through bioelectrical impedance analysis, live weight control, average daily gain, energy, and protein retention) were evaluated when the animals were fed ad libitum from 60 to 64 days of age and when the same animals were subjected to feed restriction (50% of maintenance energy requirements) from 70 to 74 days of age, in a split-plot trial. In addition, untargeted urine metabolomics analysis was performed at both periods (ad libitum vs. restricted). Although some differences between genetic lines were observed in the animals’ performance traits (average daily gain and retention of energy and protein), no differences in the urine metabolome were found between genetic types. However, feed restriction caused notable changes in the metabolome. When the animals were subjected to feed restriction, they had higher levels of ascorbic acid (P = 0.001) and p-cresol sulphate (P = 0.058) and lower levels of pyrocatechol sulphate/hydroquinone sulphate (P < 0.001), resorcinol sulphate (P = 0.002), enterolactone sulphate (P < 0.001), enterolactone (P < 0.001), kynurenic acid (P = 0.0002), proline betaine (P < 0.001), pipecolic acid betaine (P < 0.001), xanthurenic acid (P < 0.001) and quinaldic acid (P < 0.001) than the same animals when they were fed ad libitum. This study proposes urine ascorbic acid as potential biomarker for fasting events in rabbits. As urine ascorbic acid is the sole metabolite that significantly increases in the restricted group, it offers promising indicator for early detection and targeted management of digestive disorders in rabbits.Keywords: Ascorbic acid, p-cresol sulphate, Rabbit, Reduced intake, MetabolitesSubject terms: Metabolomics, Animal physiology Introduction In pursuit of minimizing feeding expenses, selection for growth rate has been implemented in paternal rabbit lines, aiming to indirectly enhance their feed efficiency1. Genetic progress in the paternal rabbit lines for feed efficiency should be due to different improvements in animal’s life weight, growth rate and nutrients utilisation, in comparison with maternal lines, selected by reproductive traits2. Nutrients, mainly energy and protein, acquisition could be measured at different levels: digestibility or nutrient retention. Traditionally, the reference method for determining body composition and nutrient retention has been comparative slaughter. However, recent advancements have introduced alternative methods, such as bioelectrical impedance analysis (BIA), for assessing animal body reserves and nutrient retention. BIA, based on the body’s resistivity to an alternating electrical current, has recently undergone investigation in rabbits3,4. BIA has demonstrated a good correlation with nutrient retention and is a good alternative measurement procedure without the need of slaughtering the animals, even in rabbit production. Resource allocation refers to the distribution of available energy and nutrients among vital life processes, as well as the formation of body structures and tissues5. Nutrient partitioning and also resource allocation between functions must be genetically driven6, so it is therefore interesting to study the responses, with respect to nutrient partitioning and resource allocation, of different genetic types to certain nutritional challenges. In production systems, balancing animal nutritional requirements with performance enhancement poses a significant management challenge. This could be done by adjusting the ration to the nutrient requirements, or managing the fattening and meat quality7. Some of these interventions are used in rabbit production with different aims; rearing females aiming to avoid excessive fattening8; controlling body and meat quality9; or even techniques to minimise the incidence of digestive disorders, especially in weaning and post-weaning periods10. Epizootic Rabbit Enteropathy (ERE) stands out as a predominant digestive syndrome, serving as the primary cause of mortality in rabbit farming11. The first detectable sign of ERE syndrome is a sudden onset of reduced feed intake12 and a decrease in water consumption, followed by depression and abdominal distension13. However, a reduction of feed intake is not an exclusive clinical sign of this disease, it has different probable causes such as pain, stress, chronic renal failure and different digestive disorders, as in the case of ERE14. It will therefore be especially interesting to quickly detect signs that are directly related to a reduction of feed intake. In this context, it is particularly intriguing to identify biomarkers that enable rapid determination of periods of reduced ingestion in rabbits, along with exploring potential metabolic pathways involved, given their association with ERE incidence. It has been shown that some changes in metabolism are seen more quickly than in other productive parameters in different species as it has been observed in pigs15,16 and birds17. In this context, untargeted metabolomics emerges as a pivotal tool for discovering biomarkers18. This advancement facilitates the development of early detection systems for targeted treatment of these disorders, potentially reducing reliance on antibiotics and advancing precision farming in livestock management. The use of some metabolites for the adjustment of nutritional requirements has been previously used in growing rabbits19–21, and recently metabolomics also has been implemented to address this objective22. Feed restriction could change the metabolome of rabbits as has been shown for other animals such as rats23, pigs24 or chicks25. For this reason, techniques such as untargeted metabolomics can help to determine the effects (cause/consequences) of periods with reduced ingestion, while generating biomarkers to be used as an on-site detection system for a specific metabolite related to this lack of intake. This study aims to elucidate the impact of feed restriction on resource acquisition and the urine metabolome of rabbits across two genetic lines. The overarching goal is to identify potential biomarkers that could serve as predictive indicators for periods of reduced ingestion associated with ERE—mainly validating in a system of ERE. Results The evolution of live weight (LW) throughout the experimental period is summarised in Fig. 1. The Caldes line consistently exhibited higher LW (P < 0.05) than the Prat line, with no significant differences observed within each line and feeding system throughout the trial. Relationship between the live weight (g) and day of age as function of genetic type during ad libitum and restricted feeding. The colours and shapes correspond to the four experimental groups and are explained at the bottom. The restrictions were calculated individually, estimating 50% of the maintenance needs of each animal; a,b were significantly different within feed management at P < 0.05. LS means and standard errors (error bars). A similar trend was shown for the estimated energy content of the whole body (Fig. 2). However, the energy content of the Caldes line (P < 0.05) at 60 days of life was significantly lower than that observed at 71 and 74 days of life. However, Caldes animals gained more weight and body energy during the ad libitum period but lost more weight and body energy and protein during the restricted period than Prat animals (P < 0.05, Table 1). These differences were more pronounced when the periods were compared (Table A1) showing that ad libitum animals had higher ADG (P < 0.0001), energy (P < 0.0001), and protein retention (P < 0.0001) than restricted animals. Relationship between the energy content of the whole body (MJ) and the day of age as function of genetic type during ad libitum and restricted feeding. The colours and shapes correspond to the four experimental groups and are explained at the bottom. The restrictions were calculated individually, estimating 50% of the maintenance needs of each animal; a,b,c were significantly different within feed management (P < 0.05). LS means and standard errors (error bars). Table 1Effect of the genetic type (Caldes and Prat line) and the feeding system (ad libitum and restricted) on performance traits. LS means and standard errors (±).Caldes linePrat lineP-valueAverage daily gain (g/d) Δ Ad libitum48.98 ± 3.3538.27 ± 3.350.0324 Δ Restricted− 43.12 ± 7.36− 15.31 ± 7.640.0147Body energy retention (MJ)1 Δ Ad libitum1.87 ± 0.151.25 ± 0.150.0072 Δ Restricted− 1.33 ± 0.24− 0.53 ± 0.240.0257Body protein retention (g)1 Δ Ad libitum16.38 ± 2.0315.01 ± 2.030.6371 Δ Restricted− 16.53 ± 2–75− 4.72 ± 02.860.00611Estimated through bioimpedance in all periods; Δ Ad libitum: Calculated as the difference between day 60 and day 64 and Δ Restricted: Calculated as the difference between day 71 and day 74. Effect of the genetic type (Caldes and Prat line) and the feeding system (ad libitum and restricted) on performance traits. LS means and standard errors (±). 1Estimated through bioimpedance in all periods; Δ Ad libitum: Calculated as the difference between day 60 and day 64 and Δ Restricted: Calculated as the difference between day 71 and day 74. Figure 3 depicts the first two principal components derived from PLSDA analysis of untargeted metabolomics data. Notably, these components, which account for 54% of the total variance, do not allow for differentiation between genetic lines. However, a discernible effect of feeding systems (ad libitum vs. restricted) on the urine metabolic profile is evident, as reflected in the distinct distribution of animals in the figure according to their feeding regimen. Urine metabolites responsible for the discrimination between the feeding systems are summarised in Table 2. Animals subjected to feed restriction showed higher levels of ascorbic acid (0.17 of fold; P = 0.001) and p-cresol sulphate (0.451 of fold; P = 0.058) in urine. However, ad libitum animals showed higher levels of enterolactone (6.5 times higher; P < 0.001), hydroquinone/resorcinol sulphate (5.2 times higher; P = 0.002), quinaldic acid (5 times higher; P < 0.001), enterolactone sulphate (4 times higher; P < 0.001), pyrocatechol sulphate (4 times higher; P < 0.001), pipecolic acid betaine (3.5 times higher; P < 0.001), kynurenic acid (3.2 times higher; P = 0.0002), xanthurenic acid (3.1 times higher; P < 0.001), and proline betaine (2.5 times higher; P < 0.001) than restricted animals. Partial Least-Squares Discriminant Analysis (PLS-DA) of the urine metabolome acquired in positive mode of growing rabbits fed ad libitum or restricted and being from the Caldes or Prat line. The colours and shapes correspond to the four experimental groups and are explained at the bottom. The restrictions were calculated individually, estimating 50% of the maintenance needs of each animal. Table 2Urine metabolites discriminating among feed systems (ad libitum vs. restricted).m/z-RTIONMetabolitePotential source/functionP_valueFold1Annotation level89175.0247/RT0.812[M−H]−Ascorbic acidDecreased anti-oxidant potential0.0010.1651187.0069:RT4.062[M−H]−p-Cresol sulphateGut transit time0.0580.5411188.9862:RT1.623[M−H]−Pyrocatechol sulphateHydroquinone sulphate ORResorcinol sulphate Microbial metabolism< 0.0013.998 1188.9862:RT2.293[M−H]−Microbial metabolism0.0025.2224377.0696:RT5.422[M−H]−Enterolactone sulphateMicrobial metabolism< 0.0014.0234297.1130:RT6.442[M−H]−EnterolactoneMicrobial metabolism< 0.0016.4731188.0352:RT2.962[M−H]−Kynurenic acidTryptophan metabolism0.00023.2281114.102/RT0.732[M−H]+Proline betaineAntioxidant< 0.0012.4901158.112/RT0.832[M−H]+Pipecolic acid betaineAntioxidant< 0.0013.5211206.045/RT2.792[M−H]+Xanthurenic acidTryptophan metabolism< 0.0013.1361174.055/RT3.932[M−H]+Quinaldic acidTryptophan metabolism< 0.0015.00211Fold change was calculated by dividing the mean intensity of plasma metabolites in ad libitum animals by the mean intensity of the corresponding metabolite in the restricted animals, italics and bold represent a greater amount of this metabolite in restricted and ad libitum animals, respectively. 2Identified using commercial standards. 3Tentatively identified using MS/MS and METLIN/HMDB databases. Urine metabolites discriminating among feed systems (ad libitum vs. restricted). Pyrocatechol sulphate Hydroquinone sulphate OR Resorcinol sulphate 1Fold change was calculated by dividing the mean intensity of plasma metabolites in ad libitum animals by the mean intensity of the corresponding metabolite in the restricted animals, italics and bold represent a greater amount of this metabolite in restricted and ad libitum animals, respectively. 2Identified using commercial standards. 3Tentatively identified using MS/MS and METLIN/HMDB databases. In terms of validation, the average R2 and Q2 obtained from the PLS data were 0.98 and 0.63, respectively. Regarding validation of the metabolomics analysis, R2 assesses how well the model fits the observed data, while Q2 evaluates its classification power. As outlined in the SIMCA users’ guide, a Q2 value exceeding 0.5 is admitted for good classification26. Based on our obtained average results for R2 and Q2 (0.98 and 0.63, respectively), we can conclude that the models utilized fit well and can be considered as good classification. Discussion The purpose of this work was to address how feed restriction affects some body traits and the urine metabolome of growing rabbits from two genetic lines subject to selection for very different, even antagonist, criteria (litter size or growth rate). In addition, an attempt was made to find possible biomarkers for periods of reduced ingestion. However, it is crucial to acknowledge the absence of a direct comparison between the urinary metabolites of rabbits under ad libitum and restricted feeding conditions simultaneously. This introduces the possibility that certain metabolites may also be influenced by the age of the animals, despite a mere difference of 3 days. Weight and body trends followed the expected pattern, increasing under ad libitum feeding and decreasing during feed restriction.. However, this trend was more marked in body energy and protein retention in both periods than in other measured traits. Animals from Caldes line showed greater body loss in the fasting period than those from Prat Line. This data could be indicating that during the ad libitum period Caldes animals showed higher body weight, energy, and protein retention higher than Prat ones, respectively. In relation to that observed by Marín-García27, paternal lines fed with commercial diets exhibited lower growth rates than expected, possibly due to insufficient protein availability. This suggests a need for diets with higher protein content to meet the nutritional requirements of these animals. For this reason, it was observed a higher accretion of the body energy (fat) and lower in protein (no significant). During the restricted period, Caldes animals experienced a greater decline in body weight and energy retention compared to Prat animals, with losses being 2.5–2.8 times higher. Additionally, protein retention losses in Caldes animals were 3.5 times higher than those observed in Prat animals. This fact could be indicating the dependence of the parental lines on protein level. Despite observing differences between genetic types in some of the body traits, the urine metabolome was not sensitive to the effect of the genetic type, but it did show a strong effect of the feeding strategy (Fig. 3). It is expected that during a restriction most of the metabolites in the urine are reduced. Following three days of strict caloric restriction, there was an observed increase in urine levels of ascorbic acid and p-cresol sulphate. Conversely, levels of several other metabolites, particularly enterolactone, hydroquinone sulphate, and quinaldic acid, decreased during this period. Some differences in different physiological states have been observed in this same species in the wild28. Ascorbic acid levels were higher in restricted animals than in ad libitum animals (with a fold of 0.165, as specified in Table 2, the folds were calculated by dividing the mean of the ad libitum-fed animals by the mean of the restricted animals. Ascorbic acid, has been used as biomarker of physiological response to diets29. This metabolite plays a key role in several metabolic pathways related with B vitamin, folic acid, cholesterol, amino acids, among other30. In this study, an increase in urinary ascorbic acid levels was observed in restricted animals; therefore, the elevated urine levels of this metabolite cannot be attributed to excessive feed intake31. Rassmunsen et al., has also observed that a decrease of the protein intake was correlated with an increase of the ascorbic acid content in the urine32. In fact, Lillefosse et al. observed that mice showing a lower body and fat accretion had also a clear reduction of main of the ascorbic acid cycle (TCA) intermediate metabolites, as a direct loss of energy through drainage of carbon-containing molecules from the body33. When there is an increase in fatty acid oxidation, pyruvate decarboxylation, amino acid catabolism and ketone body breakdown because of a lack of nutrients (restriction), an increase in acetyl-CoA occurs. Acetyl-CoA serves as an anaplerotic molecule in the TCA cycle, contributing additional molecules to this metabolic pathway. Consequently, during restriction, there should be heightened production of acetyl-CoA, leading to an increase in intermediate metabolites of the citric acid cycle, including ascorbic acid. This mechanism could explain the elevated presence of ascorbic acid in urine during periods of restriction. As early as 1939, Smith and Meyer observed that ascorbic acid losses may be more related to protein than energy restriction34. In humans, ascorbic acid has been shown to lower the pH of urine, making it an inhospitable environment for bacteria, decreasing the incidence of urinary tract infection35. This could indicate a physiological response to adaptation to reduced nutritional intake and the possibility of becoming ill more quickly. The concentration of ascorbic acid was found in clinical investigations aiming to assess the exact amount of oxidative stress, which was taken as an indicator of the level of illness and thus, with different illness. This is why new ascorbic acid biosensing methods are being studied36 and could be very useful in livestock precision farming. Enterolactone sulphate (with a fold of 4.023), and enterolactone (with a fold of 6.473) were higher in ad libitum animals than in restricted animals. Low levels of enterolactone are related to changes in gut bacteria, reducing several species37 and associated with decreasing liver enzyme functions38, and related with obesity and metabolic alterations38. The detection of these metabolites by LC-tandem MS in urine clearly confirms the conversion of plant lignan into mammalian lignans (as these metabolites), this has been previously demonstrated in rats39. Enterolactone is formed by the action of intestinal bacteria on plant lignan precursors present in the diet. The diet of rabbit is rich on lignan and a reduction of the enterolactone could be a marker of a low lignan intake. p-Cresol sulphate levels tended to be higher (1.8 times higher) in restricted animals than in ad libitum animals. Urine and plasmatic p-cresol sulphate levels have been related to an increase in intestinal transit time in some species40. It is known that a feed restriction increases transit time in young rabbits. Gidenne and Feugier, observed that the mean retention time of particles increased by 50% for restricted animals compared to the ad libitum ones41. Longer colonic transit time results in an elevated bacterial fermentation of amino acids42,43 and may subsequently induce overgrowth of proteolytic bacteria in the colon. A higher p-cresol concentration has also been related with longer transit time in an in vitro test44 and with suppression of the gut microbiota activity45. Both, ascorbic acid and p-cresol sulphate are related to microbial metabolism46–48, and could be indicating an alteration in gut microbiota activity49–51 when animals are subjected to an strong feed restriction. However, it is likely that a significant change in gut microbial composition may not occur after only 4 days of altered feeding strategies, with initial effects likely reflecting the response of basal metabolism followed by changes in microbiota. Xanthurenic acid, kynurenic acid, and quinaldic acid content in urine were higher in ad libitum animals than in restricted animals (3.1, 3.2 and 5 times higher, respectively). Kynurenic acid and its related metabolites (quinaldic acid and xanthurenic acid) are related to tryptophan metabolism52. The tryptophan metabolism produces an increase in these three metabolites via the kynurenine pathway, and this pathway is directly associated with serotonin/melatonin metabolism and glutamate activity53. The observed differences may be because the animals fed ad libitum are, obviously, subject to a higher dietary concentration of tryptophan and this could indicate a greater action of the metabolites involved in its pathway. This was also seen in weaned pigs, where it was observed that animals fed with high levels of dietary tryptophan (+ 21%) showed higher plasma tryptophan and its related metabolites (+ 27%, and + 5.7%, for tryptophan and kynurenine, respectively)54. An increase in urine metabolites related to tryptophan metabolism has also been observed in mice, when they were fed with higher dietary protein concentrations55. On the other hand, different metabolites of this pathway cause damage in some body systems, including immune, endocrine and others related to the ageing process56. The tryptophan metabolism has also been related to gut microbiota and intestinal homeostasis57. Hydroquinone sulphate, resorcinol sulphate or pyrocatechol sulphate content in urine were higher in ad libitum animals than in restricted animals (5.2, 4, and 4 times higher, respectively), as we have 2 features that might be one of these potential three metabolites. All of them are considered phenolic fats58 and are related to several diseases. Pyrocatechol (or catechol) sulphate is a derivative of catechol, a prevalent phenolic compound in plants known for its various physiological and medicinal properties. These properties include scavenging free radicals, inhibiting lipid peroxidation, and exerting anti-inflammatory and anticancer effects59–61. In this case, higher plant intake could increase the levels of these metabolites. Catechol is the organic compound of catecholamines62, which are related to different nervous system functions63. Several studies show that the catechol administered in the diet is excreted mainly as catechol sulphate64,65, and high urinary levels of this metabolite are inversely associated with dietary acid load66, or in this case, could be related to higher feed intake. In addition to what it has been explained above, some lignan vegetable compounds are metabolised to form catechol metabolites by intestinal microbiota of mammals67, another indication that supports changes in microbiota activity. Finally, Proline betaine and pipecolic acid betaine content in the urine were also higher in ad libitum animals than in restricted animals (2.5 and 3.5 times higher, respectively). Betaine is a methyl donor in one-carbon metabolism, with antioxidant and anti-inflammatory activities68, and also modulate gut microbiome69. There are not many works that analyse these metabolites in production animals. However, Bertram et al., using untargeted metabolomics in pigs, found a relationship between betaines and resorcinol (Alkylresorcinols). In this work, the metabolic effect of dietary wholegrain content was studied, increases of plasmatic and urine betaine and creatinine were observed when a higher level of wholegrain content (also related to the level of intake) was offered70. In the present study, the higher urinary levels of these metabolites could be explained by a higher dietary intake of cereal for the ad libitum animals. Furthermore, Pekkien et al., concluded that urinary excretion of various betaines was greatly increased in mice fed rye brans, this could be indicating another hint of markers related to intake71. Eleven metabolites have demonstrated significant effects on the differences in the metabolome between restricted and ad libitum-fed animals. When selecting potential biomarkers, one is proposed: Ascorbic acid Ascorbic acid is the only metabolite showing significantly higher levels in restricted animals compared to ad libitum-fed animals. This provides assurance that the increase is not related to elevated intake of any nutrient/raw material, but rather directly linked to reduced feeding. Although further validation is still necessary. Restricted animals showed higher urine ascorbic acid and p-cresol sulphate levels. An increase in urine ascorbic acid could be explained by a decrease of the protein intake. In the case of p-cresol sulphate, it could have a relationship with the transit time, and changes in the activity of gut microbiota due to reduced feed ingestion.). The relationship between gut microbiota activity and the metabolome of mammals has already been observed72,73. Our data suggest a fast modification of gut microbiota activity (assessed based on the metabolites that produces) when rabbits are subjected to a feed intake reduction. This relationship has been previously observed in rats, where it was observed that 6-days restriction changed gut microbiota composition74, in humans, where a fasting period of 16 h/d changed the microbiota composition75 and in pigs76. Before concluding, it is important to note that ascorbic acid hold promise as potential indicator of periods of reduced feeding in rabbits, which could signal the onset of enteric or other pathological processes. However, further studies are necessary to establish a clear relationship between these biomarker levels and the incidence of various pathological conditions, including enteropathy. This work aimed to determine the effect of feed restriction on the acquisition of resources, as well as on the urinary metabolome of the animals, looking for possible biomarkers that detect periods of fasting. In this sense, the main conclusions drawn from the work are: (i) Untargeted metabolomics is more sensitive and faster in detecting dietary restrictions than weight and bioimpedance measurements. (ii) Feed restriction causes notable changes in the metabolome of individuals, increasing the levels of some metabolites (ascorbic acid and p-cresol sulphate) and reducing those of others (xanthurenic acid, kynurenic acid, quinaldic acid, hydroquinone sulphate, resorcinol sulphate, pyrocatechol sulphate, proline betaine, pipecolic acid betaine, enterolactone sulphate and enterolactone) with respect to animals fed ad libitum. (iii) It is suggested that urine ascorbic acid can be used as potential early biomarker of periods of reduced feed intake in rabbits. Further studies mainly validating in a system of ERE, will be necessary to translate these findings into practical applications for animal production. Material and methods This study was carried out in compliance with the ARRIVE guidelines. This study was carried out in accordance with the relevant guidelines and regulations of the animal care and use committee of the Institute for Food and Agriculture Research and Technology (IRTA) which adopts “The European Code of Conduct for Research Integrity”. The experimental protocols were approved by the Animal Welfare Ethics Committee of the IRTA and carried out following the Spanish Royal Decree 53/2013 on the protection of animals used for scientific purposes77. Animal diet and experimental design All animals involved in the study were raised at the rabbit facilities of the Institute of Agrifood Research and Technology (IRTA). A total of 48 growing rabbits from two genetic types belonging to IRTA were used. Sex distribution was balanced—with 50% females and 50% males—in each experimental group. The first genetic type was the Caldes line (n = 24), selected for litter weight and individual growth rate until 1992 and for growth rate until 2011. From 2011 to 2016 no selection was performed on these animals, and selection is currently being carried out for feed efficiency (see Piles et al. for more details). The second was the Prat line (n = 24), which has been selected for litter size at weaning since 199278. The experimental farm where animals were housed has an insulated roof and walls, controlled lighting and ventilation, and a cooling system to avoid high temperatures in summer. After weaning, the animals were randomly housed in groups of 6–8 animals in the same room of the experimental rabbit farm under commercial production conditions. When animals reached 53 days of age, they were housed in individual cages (26 × 50 × 31 cm) kept at 15 to 22 °C throughout the experimental period, under a photoperiod of 12 h of light (06:00 to 18:00) and 12 h of darkness and fed ad libitum with a commercial diet (Table 3). No antibiotics in feed or water were used during the experiment. The feed intake was monitored weekly throughout the trial (Fig. 4). Table 3Chemical composition of the experimental diet and the faecal digestibility of energy and protein during ad libitum feeding.TraitUnitsLevelDry matter (DM)g/kg874Ashesg/kg DM82Organic matter (OM)g/kg DM917Gross energyMJ/kg DM18.4Crude proteing/kg DM190Neutral detergent fibreg/kg DM365Acid detergent fibreg/kg DM199Ligning/kg DM58Faecal digestibility Digestibility of energy1MJ/Kg DM10.3 Digestibility of protein1g/kg DM142 Digestible protein/digestible energyg/MJ DM13.81Faecal digestibility determined experimentally according to Perez et al.16. Using 28 healthy growing rabbits of 60 d of live per diet weaned at 28 d. Faeces were collected during 4 d under ad libitum feeding. Chemical composition of the experimental diet and the faecal digestibility of energy and protein during ad libitum feeding. 1Faecal digestibility determined experimentally according to Perez et al.16. Using 28 healthy growing rabbits of 60 d of live per diet weaned at 28 d. Faeces were collected during 4 d under ad libitum feeding. Representation of the experimental design. From 60 to 64 days of age growing rabbits were fed ad libitum. Faeces were collected and weighed during this period according to Pérez et al.79 to enable determination of faecal digestibility, feed consumption was also controlled. Faeces were stored in identified plastic bags and frozen at − 20 °C until analysis. Apparent faecal digestibility coefficients for dry matter (DM), crude protein (CP) and gross energy (GE) were determined for each animal. At 64 days of age, a urine sample from 28 randomly selected individuals (14 for each genetic line) was collected to be analysed by untargeted metabolomics. After this period, the animals were fed ad libitum the same diet until 71 of age, when a feed restriction was imposed until 74 days of age, when the second urine sample was recorded. This period was selected based on findings from prior experiments involving the same cohort of animals under consistent conditions, demonstrating a stabilization in feed intake during this temporal window, while the rabbits remained in the growth phase80. Dietary restriction was carried out calculating 50% of the maintenance requirements of each animal (215 kJ/day kg LW0.75; Xiccato and Trocino, 2020), offering a much lower average intake than that of the previous week (on av. 25% of ad libitum intake) This restriction was conducted with the aim of observing rapid changes in the metabolome due to a reduction in ingestion. Live Weight and body composition by BIA of all animals were monitored at 60, 64, 71 and 74 days of age (Fig. 1). BIAs and weight At 11:00 h, the body weight and feed intake of all animals were registered (n = 48), and a four-terminal body composition analyser (Model BIA-101; RJL Systems, Detroit, MI, USA) was used to register resistance (Rs, Ω) and reactance (Xc, Ω) values. Animals were placed on a non-slip and plain surface (polystyrene) located above the cages to allow animals to be quiet and calm. Then, each pair of electrodes was inserted subcutaneously using a disposable needle (21 gauge × 1.5″ (0.8 mm × 40 mm)), at 4 cm from the base of the ears and at 4 cm from the base of the tail, respectively. In each animal, BIA measurement was performed twice. The distance (D) between the internal electrodes (red ones) and the body length (L) of the animals (from the base of the ears to the tail) was also measured. The impedance (Z) value was calculated as Z = (Rs2 + Xc2)1/2. Finally, energy and protein content of the whole body was calculated according to Saiz del Barrio et al.3. Energy and protein retention were calculated in both periods (ad libitum and restricted) by difference. Chemical analysis Chemical analyses of diet and faeces were performed following the Association of Official Agricultural Chemists’ methods81: 934.01 for DM, 942.05 for ash and 976.06 for CP. Neutral detergent fibre, acid detergent fibre and acid detergent lignin were analysed sequentially82 by method 973.1881 and83, respectively, with a thermostable α-amylase pre-treatment and expressed exclusive of residual ash, using a nylon filter bag system (Ankom, Macedon, NY, USA). GE was determined using a calorimetric bomb. Statistical analysis Live Weight, Energy content of the whole body weight, Average daily gain, Body energy and protein retentions, and digestibility coefficients were statistically analysed using SAS software84. Data were analyzed using the GLM model of SAS, including the feed management and the genetic type as fixed effect. To validate the obtained partial least-squares discriminant analysis (PLS-DA) models, a cross-validation was performed using 5 as the maximum number of components to search, along with the fivefold CV method. For each generated PLS model, both R2 and Q2 were calculated. Orthogonal contrasts between experimental groups were also performed (t-test). Chemical solvents and standards for metabolomics analysis High-performance liquid chromatography (HPLC)-grade solvents and eluents were used for the untargeted metabolomics analysis as follows: HPLC-grade acetonitrile (VWR, West Chester, PA, USA), formic acid (FA, Fluka, Merck KGaA, Darmstadt, Germany) and MilliQ grade water (MilliporeSigma, Burlington, MA, USA). Internal standards included during the sample preparation were glycocholic acid (Glycine-1-13C) and 4-chloro-DL-phenylalanine (Sigma, Merck KGaA, Darmstadt, Germany) and all other standards used for compound identification were purchased from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany) and Cayman Chemical (Ann Arbor, MI, USA). Untargeted liquid chromatography–mass spectrometry (LC–MS) metabolomics analysis of urine Urine samples were prepared by deproteinisation of a 150 µL sample with 450 µL ice-cold acetonitrile (100% ACN) containing an internal standard mix of glycocholic acid (glycine-1-13C) and p-chlorophenylalanine at a final concentration of 0.01 mg/mL. Samples were prepared in 96-well plates with 1 mL wells. Plates were mixed for 1 min, incubated at 4 °C for 10 min, and centrifuged for 25 min at 2250×g and 4 °C. Approximatively 400 µL supernatant was transferred to Sirocco Protein precipitation plates (Waters Corporation, Milford, MA, USA). The filtered supernatant was transferred to two 200 µL 96-well plates (65 µL per well), and plates were vacuum centrifuged to dryness (ca. 2.5 h, 805 g and 30 °C). Resuspension of the samples was done in a mix of H2O:ACN:FA (95:5:0.1) using the same volume before evaporation. A protective film was welded onto the plate using a heat sealer and the plates were centrifuged at 3700 rpm at 4 °C for 25 min before the LC–MS analysis. UPLC-MS analysis was performed using a Dionex UltiMate 3000 ultra-high pressure liquid chromatography system (UHPLC; Dionex, Sunnyvale, CA, USA) coupled with an Impact HD Quadrupole Time-of-Flight (QTOF) mass spectrometer (Bruker Daltonics GmbH, Bremen, Germany) operating in positive electrospray ionisation mode (ESI+) and negative electrospray ionisation mode (ESI−) using the instrumental parameters described by Curtasu et al.85. The chromatographic separation of compounds was performed on an Acquity HSS T3 1.7 µm 100 × 2.1 mm (Waters Corporation, Milford, MA, USA) column equipped with a VanGuard Pre-column, 100 Å, 1.8 µm, 2.1 mm × 5 mm (Waters Corporation, Milford, MA, USA). The column temperature was set to 30 °C, samples were kept in the autosampler at 10 °C throughout the analysis and the injection volume was set to 3 µL for urine. The chromatographic analysis was performed under a gradient of water/formic acid (100:0.1, v/v, mobile phase A) and acetonitrile/formic acid (100:0.1, v/v, mobile phase B). For the urine analysis, a linear gradient was used from 5% B to 100% B over 12 min, and a 1 min hold at 100%. Sample quality control and metabolomics data pre-processing The quality of the chromatographic runs, the UPLC system stability and the accuracy of sample preparation were monitored using quality control samples (QCs). Urine QCs were prepared by pooling an aliquot number of samples and subjecting them to the same sample preparation protocol as the samples. The QCs were injected multiple times throughout the analysis as well as at the beginning and end of the analysis and used in the data pre-processing for signal drift correction. Blanks were injected during the chromatographic analysis to monitor any external contaminants from solvents, eluents and carry-over effects. The sample order was randomised for the chromatographic analysis to eliminate biases in the results and to ensure that each sample group was affected equally. Mass spectra were calibrated and converted into the mzXML file format. The R-based XCMS package86 was used for extraction of the mass features; peak peaking was performed using ‘centWave’ method and retention time aligned using ‘Obiwarp’; missing values were substituted using the ‘fillPeaks’ method; adducts, fragments and isotopes were annotated using CAMERA. Exported data tables were filtered to eliminate features present in blanks, retention times were truncated to contain only portions with chromatographic peaks, and masses higher than 700 m/z were discarded. Urine metabolomics data were normalised using the Van der Kloet87 procedure based on the quality control samples. Multivariate data analysis of metabolomics data An initial principal component analysis was performed using LatentiX 2.12 (LATENTIX Aps., Gilleleje, Denmark) to check the quality of the data set and eliminate potential outliers. Prior to any analysis, all metabolomic datasets were pre-treated by Pareto scaling, and preliminary data mining for an overview was performed using consensus principal component analysis (CPCA)88. PLS-DA models were constructed to determine the metabolites responsible for the differences between experimental groups (Caldes ad libitum, Caldes restricted, Prat ad libitum and Prat restricted) of rabbits. Model validation was performed using repeated random subsampling validation. Models were evaluated using the explained variation in Y, plots showing actual and predicted values, and the proportion of variation explained (R2). Variables for identification were selected using VIP scores. Metabolite identification Accurate mass and mass fragmentation patterns of discriminating features were used for compound identification based on queries in the following databases: The Human Metabolome Database (http://www.hmdb.ca/), METLIN (http://metlin.scripps.edu/) and MetFrag database (https://msbi.ipb-halle.de/MetFragBeta/). Annotated features were classified on different levels of identification according to Sumner et al.89. Supplementary Information Supplementary Information. Author contributions M.P, J.P.S, M.P, J.P conducted the experiment(s), P.J.M.G and L.L wrote the manuscript. P.J.M.G and M.S. H provided statistical analysis of results. P.J.M.G, M.P, J.P.S, M.Pa and M.S.H provided funding for the project. P.J.M.G and M.S.H prepared figures, and conceived the experiment(s). All authors reviewed the manuscript. Funding This study was supported by the Regional Ministry of Innovation, Universities, Science and Digital Society of the Generalitat Valenciana (CIGE/2023/12), and the EU project Feed-a-Gene funded by The European Union’s H2020 Program under grant agreement no 633531. Data availability Data is provided within the manuscript or supplementary information files. Pablo Jesús Marín García should be contacted if someone wants to request the data from this study. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-024-80701-x. References Associated Data Supplementary Materials Supplementary Information. Data Availability Statement Data is provided within the manuscript or supplementary information files. Pablo Jesús Marín García should be contacted if someone wants to request the data from this study.