Predicting post-absorptive protein and amino acid metabolism

Abstract

Sustainable production of adequate quantities of food to support a growing human population is a worldwide goal. Under current feeding conditions in the United States, dairy cattle convert dietary nitrogen to milk nitrogen with 25% efficiency. The remaining 75% is excreted, which contributes to air and water quality problems and reduces economic performance of the industry. Efficiency could be improved to 29% if protein was given to just meet current NRC requirements. Additional improvements may be achievable, but only with improved knowledge of amino acid (AA) requirements. The current metabolizable protein requirement model overestimates true requirements due to lack of knowledge of AA supply and requirements and to intrinsic limitations in system data and assumptions. Existing protein supply models based on passage and degradation rates are biased, which undermines predictions of AA supply. The use of an equation driven solely by protein solubility of each ingredient in the diet with no consideration of the effects of passage rate yielded unbiased predictions with significant improvements in precision. However, this still leaves a problem in predicting the AA composition of the ruminally undegraded protein (RUP). Current models generally assume that RUP AA composition equals the parent ingredient composition, but assessments of RUP AA composition indicate that this is false. Thus, bias is being introduced into predictions of the absorbed AA supply, which hampers derivation of estimates of AA digestion and absorption from the small intestine. Emerging isotope-based methods hold promise in allowing assessment of AA availability from individual ingredients in vivo, which will allow construction of a database of true ingredient AA bioavailabilities. These efforts will eventually allow development of more robust predictions of AA supply. On the AA requirement side, numerous data indicate that the efficiency of metabolizable protein use for lactation is variable and maximally 45%, whereas most models assume an efficiency of 65% or greater. The efficiencies of individual AA are centered on the protein efficiency value with those lower in efficiency, likely being provided in large excess. A better representation of the use efficiency of individual AA would allow improvements in overall animal N efficiency. Variable efficiency is driven by regulatory mechanisms that control protein synthesis in response to the supply of energy and individual AA and circulating concentrations of hormones and these drivers act independently and additively. Under this theory, protein synthesis can respond to nutrients other than the one identified as most limiting. Reflecting this regulation in our requirement models will allow better prediction of AA efficiency and enable construction of diets that minimize excess of individual AA by optimizing the energy and hormonal signals to improve N efficiency. Models of such an interacting system have been developed and shown to be superior in performance to models based on current paradigms.

A primer on predictive techniques for food and bioresources transformation processes

“…The field of agricultural management is very much a user of empirical predictions for water management (Adisa et al, 2020; Pereira et al, 2020), energy management (Bersani et al, 2020; Garcia‐Maraver et al, 2017; Shine et al, 2020), and biological control (Haan et al, 2020; Mills & Heimpel, 2018; Mueller‐Schaerer et al, 2020), or to predict direct impacts on productivity. These productivity‐focused models have been employed in a wide range of studies, from predictions looking directly at overall yield (Bekenev, 2019; Horie, 2019; Li et al, 2018; Menzel, 2021) to predictions investigating a specific cause of product loss (Ealy & Seekford, 2019; Zhou et al, 2020), typically by the impact of animal diet (Hanigan et al, 2018; Lyu et al, 2020; Święch, 2017; Trottier & Tedeschi, 2019).…”

“…Progress in expert knowledge and understanding of the phenomenon leads to hybrid empirical–mechanistic prediction (Hanigan et al, 2018; Lyu et al, 2020; Święch, 2017; Trottier & Tedeschi, 2019).…”