Effects of Dietary Sterol-Rich Yeast Hydrolysate on Serum Lipid Indices, Intestinal Morphology, Digestive Enzyme Activities, Barrier-Related Gene Expression, and Cecal Microbiota in Yellow-Feathered Broiler Breeder Hens

This study evaluated dietary sterol-rich yeast hydrolysate (SYH) at 0, 0.2, 0.5, and 0.8 g/kg in 720 yellow-feathered broiler breeder hens for 6 weeks. The tested SYH, supplied as a powder, contained 9.24% mannan, 3.1% total sterols, 2.0% amino nitrogen, and 3.64 g/kg nucleotides. Serum lipoproteins, intestinal morphology, digestive enzyme activities, barrier-related gene expression, and cecal microbiota were assessed. Low-density lipoprotein cholesterol was higher in J-C than in Control, J-A, and J-B, whereas high-density lipoprotein cholesterol did not differ among treatments. Duodenal lipase showed a significant overall treatment effect, but no Dunn–Holm pairwise comparison remained significant; ileal lipase was higher in J-C than in J-A. For barrier-related transcripts, J-A had higher ZO-1 than the Control and J-B and higher occludin and claudin-1 than all other groups, whereas MUC2 was higher in J-A and J-C than in the Control. These transcript changes are consistent with the modulation of molecular components associated with intestinal barrier integrity, but they do not by themselves demonstrate improved barrier function. Intestinal morphology showed selective, segment-specific differences rather than a consistent dose-related pattern. Alpha diversity did not differ, whereas Bray–Curtis PERMANOVA detected an overall difference among treatments (p = 0.017) despite overlapping PCoA distributions. Pairwise differential-abundance analysis using metagenomeSeq identified six family- or genus-level differences after false-discovery-rate correction, but these were specific to individual pairwise comparisons and showed no monotonic dose pattern. No exploratory microbiota–phenotype correlation remained significant after multiple-testing correction. Overall, SYH elicited inclusion-level- and trait-specific responses rather than a coordinated improvement across endpoints; therefore, the present data do not identify a single optimal inclusion level.