MATERIALS AND METHODS

Ethical approval

All experimental procedures involving animals were conducted in accordance with the principles and recommendations outlined in the Guide for the Care and Use of Agricultural Animals in Research [17]. The experimental protocol was reviewed and approved by the Research Project Management Committee of the Pedagogical and Technological University of Colombia, Tunja, Colombia, under Resolution Code I-FP-P03-F16 of 2025. Furthermore, the study was designed following the principles of replacement, reduction, and refinement (3Rs) to ensure the ethical use of animals while minimizing unnecessary animal suffering and the number of birds used.

Study period, location, and housing

The experiment was conducted over a 12-week period (September to December 2024), corresponding to 37–48 weeks of age. The study was conducted under controlled commercial conditions at the Tunguavita Experimental Farm in Paipa, Boyacá, Colombia. Birds were housed in a conventional open-sided poultry house equipped with floor pens bedded with rice husk litter at a stocking density of 10 hens/m².

Feed was provided manually using hopper feeders, whereas drinking water obtained from the municipal aqueduct system was supplied ad libitum through automatic bell drinkers. The poultry house was naturally ventilated to maintain bird comfort and adequate air circulation. Environmental conditions during the experiment ranged from 19°C to 24°C, with relative humidity maintained between 60% and 80%. Birds were exposed to the natural photoperiod characteristic of the equatorial region, corresponding to approximately 12 h of daylight per day.

Routine biosecurity and sanitary measures, including regular cleaning and disinfection of facilities and equipment, as well as rodent and insect control, were implemented in accordance with farm management protocols. Birds were managed according to the farm's routine health program, which included vaccination against Newcastle disease every 10 weeks. No therapeutic treatments that could interfere with the experimental objectives were administered during the study.

Study design

A total of 200 Babcock Brown laying hens aged 37 weeks were used in this study. Birds were standardized and randomly assigned to four dietary treatments in a completely randomized design with repeated measurements. Each treatment consisted of five replicates, and each floor pen represented one experimental unit (n = 5 replicates per treatment).

Sample size was calculated in RStudio using the "pwr" package, based on a one-way analysis of variance (ANOVA), to achieve 80% power at α = 0.05. The effect size (Cohen's f = 0.47) was estimated from previous studies conducted by our research group according to current recommendations for sample size determination [18]. The final allocation was consistent with Mead's resource equation and adhered to the 3Rs principle of animal bioethics, thereby minimizing animal use without compromising statistical robustness [19].

The experimental diets were formulated according to the nutritional requirements of the genetic line during the laying phase (Table 1). Four dietary treatments were established. D1-Control consisted of the basal diet without FC and served as the negative control. D2-MOS consisted of the basal diet supplemented with a commercial yeast cell wall product containing 25% MOS and β-glucans at 2,000 mg/kg. D3-FOS consisted of the basal diet supplemented with short-chain FOS (43% purity, including 24% nystose and 19% 1-kestose) at 700 mg/kg. D4-RS consisted of the basal diet supplemented with type III RS (RS3) derived from retrograded potato starch at 700 mg/kg. Each treatment included 50 birds.

All additives were supplied by Promitec® Santander (Promitec Santander SAS, Bucaramanga, Colombia), and supplementation levels were established according to the manufacturer's recommendations.

Productive performance

Egg production was recorded daily throughout the experimental period. Egg weight was measured using an OHAUS® Scout Pro SP 402 electronic balance (OHAUS Corporation, Parsippany, NJ, USA). Based on feed intake and egg production, FCR was calculated as kilograms of feed consumed per dozen eggs produced [20]. Egg production percentage was calculated by dividing the total number of eggs produced by the total number of hens [21].

Ingredients (% as-fed) | Content
Yellow corn | 58.1
Soybean meal | 23.0
Extruded soybean | 4.0
Soybean oil | 2.0
Dicalcium phosphate | 1.2
Limestone | 10.0
Salt | 0.3
DL-methionine | 0.3
L-lysine | 0.2
L-threonine | 0.2
Choline chloride | 0.2
Layer premix* | 0.5
Calculated nutrients | 
Nutrient | Content
Metabolizable energy (kcal/kg) | 2797
Crude protein (%) | 17.0
Crude fiber (%) | 6.0
Calcium (%) | 3.8
Total phosphorus (%) | 0.56
Available phosphorus (%) | 0.32

Evaluation of intestinal permeability and antioxidant activity

Intestinal permeability was evaluated using serum fluorescein isothiocyanate-dextran (FITC-d) concentrations at 48 weeks of age (week 12 of the experimental period). Five hens per treatment were randomly selected, resulting in 20 birds and 40 serum samples. Blood was collected from the brachial vein immediately before and 2 h after oral administration of FITC-d.

FITC-d (Sigma-Aldrich, St. Louis, MO, USA) was administered orally at 2 mL of a 2.2 mg/mL solution, corresponding to a dose of 4.4 mg per bird [22]. Following centrifugation (1,000 × g for 15 min), serum samples were collected and stored for analysis. Fluorescence was measured using a spectrofluorometer, with excitation and emission wavelengths set at 489 and 520 nm, respectively. Intestinal permeability was expressed as ng FITC-d/mL, calculated as the difference between baseline and post-administration values.

Systemic antioxidant capacity was determined using the oxygen radical absorbance capacity (ORAC) assay. Measurements were performed using an FS5-SS spectrofluorometer (Edinburgh Instruments, Livingston, UK). Fluorescence decay was monitored at excitation and emission wavelengths of 485 and 520 nm, respectively, over 20 min. ORAC values were expressed as µmol TE/mL according to a Trolox calibration curve [23].

Serum lipid peroxidation was evaluated using thiobarbituric acid reactive substances (TBARS). Absorbance was measured at 535 nm using a UV-1900 UV-Vis spectrophotometer (Shimadzu Corporation, Kyoto, Japan), and malondialdehyde (MDA) concentrations were calculated from a standard calibration curve [24].

Euthanasia procedures and intestinal sampling

For intestinal sampling, one bird from each experimental unit was randomly selected and euthanized by cervical dislocation at weeks 7 and 12 of the experimental period, resulting in a total of 40 birds. Euthanasia procedures were performed in accordance with the recommendations of the Guide for the Care and Use of Agricultural Animals in Research [17]. Following euthanasia, the coelomic cavity was opened, and the entire small intestine and ceca were removed. Subsequently, 5-cm segments of the duodenum, jejunum, ileum, and cecum were collected and preserved in 10% buffered formalin for histological processing [25].

For microbial community analysis, approximately 3 g of cecal luminal contents were collected and transferred to Falcon tubes containing absolute ethanol (99%). Samples were immediately stored at −70°C until further processing.

Intestinal morphometry

At 48 weeks of age (week 12 of the experimental period), 20 samples (five birds per treatment) from each intestinal segment (duodenum, jejunum, and ileum) were collected for morphometric analysis. Tissue samples were dehydrated through graded ethanol solutions (70%, 80%, 90%, and 100%), cleared in xylene, and embedded in paraffin. Sections of 4 µm thickness were prepared using a rotary microtome and stained with hematoxylin and eosin for microscopic examination.

Histological images were obtained using a Moticam 2300 digital camera (Motic, Hong Kong, China) coupled to a Leica DLMB optical microscope (Meyer Instruments, Houston, TX, USA) at 200× magnification. Quantitative measurements were performed using Motic Images Plus 2.0 software (Motic). Villus height (VH), villus width (VW), and crypt depth (CD) were determined from 10 measurements per field, as described by Nguyen et al. [26]. In addition, the VH-to-CD ratio (V:C ratio) was calculated as described by Nguyen et al. [26].

Goblet cell quantification

Intestinal segments from the duodenum, jejunum, ileum, and cecum obtained from birds selected for morphometric analyses were collected at week 7 and at the end of the experimental period. Paraffin sections (4 µm thick) were prepared for histochemical evaluation of mucins according to the protocols described by the Armed Forces Institute of Pathology of the United States [27]. Alcian blue (pH 1.0) was used to identify strongly sulfated acidic mucins, Alcian blue (pH 2.5) for non-sulfated acidic mucins, and periodic acid-Schiff staining for neutral mucins.

Images were analyzed using ZEN image analysis software (Carl Zeiss, Oberkochen, Germany). A circular area with a diameter of 200 µm was selected, and positively stained cells were quantified. Six measurements were obtained from each mucosal fold, including the apex, lateral regions, and base. The first measurement corresponded to the villus region and the second to the crypt region to evaluate cell distribution. Goblet cell counts were performed according to the method described by Rodríguez et al. [25]. The percentage of goblet cells was calculated as the number of positively stained goblet cells divided by the total number of epithelial cells and multiplied by 100.

16S rRNA gene sequencing and bioinformatic analysis

The composition of cecal microbial communities was investigated using samples collected at 43 and 48 weeks of age (weeks 7 and 12 of the experimental period). The hypervariable V3-V4 regions of the 16S rRNA gene were sequenced.

Total DNA was extracted from ethanol-preserved cecal samples using a combined mechanical and chemical lysis procedure involving silica beads, sodium dodecyl sulfate, urea, and proteinase K, followed by purification with phenol-chloroform-isoamyl alcohol and isopropanol. DNA quality and concentration were evaluated by spectrophotometry and electrophoresis, ensuring 260/280 ratios of 1.8–2.0 and fragment sizes exceeding 800 bp.

Approximately 10 ng of DNA from each sample was used for polymerase chain reaction amplification of the V3-V4 regions using universal primers 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHV GGGTWTCTAAT-3′). Libraries were prepared using the NEBNext Ultra II DNA PCR-free Library Prep Kit (New England Biolabs, Ipswich, MA, USA) and sequenced on the Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA) using a paired-end configuration (250 bp × 2), generating amplicons of approximately 460 bp.

Bioinformatic analyses were performed using the DADA2 package (version 1.26) in R. Sequence filtering was conducted using the following criteria: truncLen = 0, maxN = 0, maxEE = 2, and truncQ = 2, together with PhiX removal. Chimeric sequences were identified and eliminated before generating amplicon sequence variants (ASVs). After quality control, an average of 9,322 high-quality reads per sample was retained, yielding a total of 1,901 ASVs.

Taxonomic classification was performed using the naïve Bayesian classifier trained with the SILVA database (release 138.1; https://www.arb-silva.de/), complemented by nucleotide Basic Local Alignment Search Tool (BLASTn) searches through the National Center for Biotechnology Information (NCBI; https://blast.ncbi.nlm. nih.gov/Blast.cgi) and by consultation of the List of Prokaryotic Names with Standing in Nomenclature (LPSN; https://lpsn.dsmz.de/) database for taxonomic curation of high-frequency ASVs. Sequences originating from eukaryotic, mitochondrial, or chloroplast DNA, as well as those with insufficient statistical support (<98% for species level and <50% for higher taxonomic levels), were excluded. Rarefaction curves were generated to verify that sequencing depth adequately represented cecal microbial diversity.

Microbial community analysis

Microbial community analyses were performed using the phyloseq package in RStudio [28]. A physeq object was generated from the processed datasets for diversity and taxonomic analyses. Alpha diversity, representing species richness, was evaluated using the Shannon diversity index calculated with the microbiome package, followed by comparisons among dietary treatments using ANOVA [29, 30].

Beta diversity was evaluated using principal coordinates analysis based on Bray–Curtis distances calculated with the vegdist function from the vegan package [31]. To assess the effects of dietary treatments on bacterial community composition, analysis of similarity (ANOSIM) and permutational multivariate analysis of variance (PERMANOVA) were conducted using the adonis and ANOSIM functions of the vegan package [32].

Taxonomic composition and relative abundance were determined using functions implemented in the phyloseq package and visualized with ggplot2. Core microbiota were identified using the plot_core function of the microbiome package. Venn diagrams were generated using the VennDiagram package to identify unique and shared taxa among treatment groups. Linear discriminant analysis effect size (LEfSe) was performed using the microbiomeMarker package [33] to identify bacterial taxa significantly enriched by different diets, based on the Kruskal–Wallis rank-sum test (p < 0.05) with an LDA score >3.0.

Statistical analysis

Quantitative variables were analyzed using ANOVA after assessing data normality using the Shapiro-Wilk test. Variables with p ≥ 0.05 were considered normally distributed. When assumptions were met, treatment means were compared using Tukey's multiple comparison test. Equivalent nonparametric procedures were applied to variables that failed to meet the assumptions of normality.

Repeated measurements over time were analyzed using linear mixed models with a first-order autoregressive covariance structure to account for within-subject correlations and random effects for between-subject variation. Differences were considered statistically significant at p < 0.05.

All statistical analyses and graphical representations were performed in RStudio version 2024.04.1 [34], and graphical outputs were generated using the ggplot2 package. Data are presented as mean ± standard error of the mean.