Is the Control Diet Really a Control? Multiplatform Chemical Profiling Reveals Distinct AIN-93G, Conventional Control, and High-Sugar-Fat Rodent Feeds

Abstract: Experimental rodent diets are often treated as background conditions and defined primarily by nominal macronutrient composition or labels such as “standard”, “control”, or “high-fat”. However, these descriptors may not capture the chemical complexity of the final feed matrix, potentially shifting the dietary baseline against which researchers interpret experimental phenotypes [1,2]. This study investigated whether an American Institute of Nutrition-93 Growth diet (AIN-93G), an in-house conventional normocaloric Control, and the solid-feed component of an in-house high-sugar-fat (HSF) model represent chemically interchangeable dietary backgrounds or distinct feed matrices. A multiplatform analytical strategy combined proton nuclear magnetic resonance (¹H NMR) metabolite profiling (Varian/Agilent DD2, 499.72 MHz), gas chromatography with flame ionization detection (GC-FID; Shimadzu GC-2014) for fatty-acid methyl ester profiling, ultraviolet-visible (UV-Vis) spectroscopy, and Fourier-transform near-infrared (FT-NIR) spectroscopy. NMR quantified 38 metabolites and revealed markedly different extractable small-molecule architectures: 11 metabolites were quantifiable in AIN-93G versus 36 in Control and 35 in HSF. AIN-93G contained 68.045 mg/g sucrose, whereas HSF showed a characteristic fructose-sucrose signature (64.989 and 42.923 mg/g, respectively); Control contained substantially lower fructose and sucrose concentrations (2.134 and 3.987 mg/g). GC-FID resolved 33 fatty acid methyl esters, of which 29 remained significantly different among diets after false-discovery-rate correction. Principal component analysis separated all three feeds, with the first two components explaining 95.1% of the variance, while permutational multivariate analysis of variance (PERMANOVA) confirmed a strong global diet effect (R² = 0.958, p = 0.005). AIN-93G showed the highest polyunsaturated fatty acid proportion (44.86%) and linoleic acid (42.09%), Control the highest palmitic acid (32.11%), and HSF the highest oleic (36.72%) and stearic acid (13.98%) proportions. UV-Vis likewise distinguished the diets across 220–320 nm (PERMANOVA R² = 0.895, p = 0.034), whereas FT-NIR showed lower Control absorbance across much of 1900–2500 nm while AIN-93G and HSF largely overlapped. Importantly, no single pattern of similarity held across analytical platforms, showing that feeds resembling one another in one chemical domain may differ substantially in another. This multidimensional divergence matters because AIN-93G and conventional Control diets may both function as non-obesogenic comparators, yet previous in vivo work showed they can establish different physiological baselines [3]. Thus, nominal macronutrient composition and experimental labels alone do not define dietary equivalence. “Control diet” should therefore be understood operationally: its suitability depends on the experimental contrast that must be isolated, and feed composition should be treated as an intrinsic component of preclinical model design and reproducibility.

Acknowledgments: FAPESP (2021/13050-2, 2022/01925-7, and 2023/03886-1), CNPq (306640/2023-6), and CAPES, Brazil.