Introduction: The development of precision feeding strategies for growing beef cattle is critical to address the sustainability challenges facing the beef sector by formulating diets that closely align with the nutritional requirements of each individual within a herd. The INRA model (Hoch and Agabriel, 2004) predicts the daily processes of body protein and lipid synthesis and degradation of growing beef cattle, crucial factors for estimating nutrient requirements and formulating diets. A previous attempt of model calibration for individual Charolais bulls showed significant correlations between its metabolic coefficients and in vivo feed efficiency metrics (Cantalapiedra-Hijar and Lerch, 2023). The model could therefore support precision feeding tool by predicting individual’s potential. However, estimating model’s coefficients requires daily feed intake and kinetics of body weight and composition, which are difficult to record in practice. As the first approach, this study aimed to test the potential of using various metabolites analyzed on muscle tissue, as proxies for predicting the model’s coefficients related to protein and lipid synthesis and degradation. Material and methods: Cantalapiedra-Hijar and Lerch (2023) previously calibrated the INRA beef growth model at the individual level for 32 Charolais growing bulls by separately adjusting coefficients for rates of body protein and lipid synthesis (α and β), or degradation (γ and δ) based on daily metabolizable energy intake and three estimations of body compositions (d0, d84 and d200). The present study is based on the same experimental setup. A total of 285 muscle metabolites directly quantified by mass spectrometry with the MxP Quant 500 Assay from Biocrates, were obtained after slaughter and used to explore relationships with adjusted metabolic coefficients of INRA model using Pearson correlation (r) and partial least squares (PLS) regression. Results and discussion: For protein synthesis and degradation, only Lysophosphatidylcholine 18:1 showed a correlation greater than 0.5 with α (r = 0.51, P < 0.05), while both 3-Methylhistidine and Lysophosphatidylcholine 18:1 showed correlations less than -0.5 with γ (r = −0.54 and −0.50, respectively, P < 0.05). Conversely, lipid synthesis and degradation were positively (r > 0.5) and negatively (r < −0.5) correlated (P < 0.05) with 18 metabolites including 3-Methylhistidine, Diacylglyceride 17:0_18:1, three Lysophosphatidylcholines, ten Phosphatidylcholines and three Triacylglycerides. The PLS regression model indicated that the body lipid growth dynamics could be moderately predicted (Q2 ≥ 0.47) using only a subset of Phosphatidylcholines isomers, such as 32:3, 34:3 and 42:1 for synthesis and 32:3, 42:1, 42:2 and 44:3 for degradation (Table 1). Conclusion and implications: In this study, modelling the dynamics of body lipid deposition in individual Charolais growing bulls could be achieved by only a small subset of muscle metabolites. This highlights the potential for developing precision feeding strategies based on the use of metabolite biomarkers for adjusting mechanistic model coefficients at the level of individual cattle. In the next step, plasma metabolites measured at various growth time-points could be evaluated to enhance the early identification of individual cattle’s requirements.
