Body weight and feed intake
Animals were weighed weekly using a digital electronic scale (Toledo, prix III, São Bernardo do Campo, Brazil). Feed intake was evaluated three times per week and expressed as the difference between offered feed and residual feed. The feed conversion rate was computed dividing the feed intake by the weight gain in a given period.
Euthanasia and tissue preparation for analysis
The animals were euthanised at 113 days of age, after being fed the experimental diet for 52 days. The animals had been fasted for 12 h and then weighed and anaesthetised with ketamine hydrochloride and xylazine hydrochloride administered via intraperitoneal. The rats were then euthanised by section of the aorta arteries. Whole organs and tissues (heart, liver and adipose tissue) were removed, cleaned and weighed on an analytical balance. The organs and tissues of six animals from each group were kept at -80ºC until oxidative stress, lipid profile and fatty acid analyses could be performed. The organs and tissues of the remaining three animals per group were washed in saline solution (0.9% NaCl) and fixed in 10% buffered formalin until histological analysis could be carried out.
Blood serum biochemical parameters
Blood samples, after euthanasia, were centrifuged (58 136 g for 15 min at 25°C) to obtain serum samples that were kept at -80 ºC until biochemical analyses were performed. Total cholesterol and fraction of high density (HDL), low density (LDL) and very low density lipoproteins (VLDL), triglycerides, glucose, alanine aminotransferase (ALT) and aspartate aminotransferase (AST) were quantified by commercial kits (Labtest, Minas Gerais, Brazil), according to manufacturer’s guidelines, using an automatic biochemical analyser LabMax 240 (Minas Gerais, Brazil).
Serum cytokine levels
The serum cytokine levels were evaluated using the protocols of the kits (R&D Systems, Minneapolis, MN), using standard capture and detection antibodies for IL-1β, TNF-α, and IL-10. The serum samples were homogenised with phosphate buffer (10 mM, pH 7.2–7.4), and centrifuged at 4000 g at 4 °C for 10 min so that the supernatant from the centrifugation could be used to determine the cytokines at an absorbance of 450 nm in an ELISA reader.
Fatty acid
Heart, liver and adipose tissue samples were freeze-dried and sent to the Faculty of Veterinary Medicine at the University of Lisbon for FA analyses. Fatty acid methyl esters (FAME) and dimethyl acetal (DMA) from the tissue samples were prepared by reaction with HCl 1.25 M in methanol for 20 h at 50◦C. They were then analysed by gas chromatography with flame ionisation detection using a Shimadzu GC 2010-Plus (Shimadzu, Kyoto, Japan) equipped with a SP-2560 (100 m × 0.25 mm, 0.20 µm film thickness, Supelco, Bellefonte, PA, United States) capillary column. The chromatographic conditions were as follows: injector and detector temperatures were set at 250 and 280 ◦C, respectively; helium was used as the carrier gas at 1 mL/min constant flow; the initial oven temperature of 50 ◦C was held for 1 min, increased by 50 ◦C/min to 150 ◦C and held for 20 min; then increased by 1◦C/min to 190 ◦C; and finally increased by 2 ◦C/min to 220 ◦C and held for 40 min. Identification of FAME and DMA were achieved by comparison with commercial standards (FAME mix 37 components, Supelco Inc, Bellefont, PA, USA), by comparison with published chromatograms24 and by using electron impact mass spectrometry using a Shimadzu GC–MS QP2010 Plus (Shimadzu). The chromatographic column and the GC conditions used in the GC-MS were similar to the GC-FID analyses. Additional mass spectrometer conditions were as follows: ion source temperature, 200 ◦C; interface temperature, 240 ◦C; and emission voltage, 70 eV.
Assessment of Lipidic Peroxidation Levels
Lipid peroxidation was measured by the chromogenic product of 2-thiobarbituric acid (TBA) reaction with malondialdehyde (MDA), a product formed as a result of membrane lipid peroxidation25. The livers were homogenised with KCl (1:1), and samples of tissue homogenate (250 μL) were incubated at 37°C for 60 min. After that, the mixture was precipitated with 35% perchloric acid and centrifuged at 1207g for 20 min at 4 °C. Then, the supernatant was collected and 400 μL of 0.6% TBA was added and incubated at 95–100 °C for 1 h. After cooling, the samples were read in a spectrophotometer at a wavelength of 532nm (Biospectro, SP-220 model-Brazil). MDA concentration was determined by substituting the absorbance values in the MDA standard curve obtained on the basis of a standard solution (1 μL of 1,1,3,3- tetramethoxypropane in 70mL distilled water) diluted in series of 250, 500, 750, 1000, 1250, 1500, 1750, 2000, 2250, 2500, 2750, and 3000 μL of distilled water.
Antioxidant Activity
The liver homogenate was assembled as previously described. In addition, an aliquot of 1.25mg of DPPH was diluted in ethanol (100 mL), kept under refrigeration, and protected from light. Then, 3.9mL of DPPH solution was mixed with 100 μL of the supernatant liver homogenate in appropriate centrifuge tubes. These tubes were vortexed and left to stand for 30 min, then centrifuged at 1207g for 15 min at 20°C. Then, the samples were read in a spectrophotometer at a wavelength of 515nm (Biospectro, SP-220 model-Brazil)26. Results were expressed as the percentage of the oxidation inhibition: AOA= 100 − (((DPPH⋅ R) S/(DPPH⋅ R) B)× 100), where (DPPH⋅ R) S and (DPPH⋅ R) B corresponding to the concentration of DPPH⋅ remaining after 30 min, measured in the sample (S) and blank (B) prepared with distilled water.
Histological evaluation of liver
The liver was removed from the rats, washed in saline solution (0•9% NaCl) and fixed in 10% buffered formalin. The major lobe was subjected to a histological procedure according to the routine technique at the Pathology Laboratory (Department of Physiology and Pathology/CCS/UFPB) for obtaining blocks from which semi-serial 4-μm sections were cut. The slides were hydrated, stained with haematoxylin–eosin (HE), dehydrated, diaphanised in xylol and mounted with Entellan® for optical microscopic analysis (Motic BA 200, Olympus Optical Co, Philippines). Inflammatory exudate, hyperaemia, haemorrhage, necrosis, preservation of the hepatic parenchyma (cell integrity, centralised nuclei and highly evident nucleoli) and degenerative processes (e.g. fat degeneration) were evaluated in these liver histological sections27.
For morphometric analysis, twenty random images from slides of liver tissue were used. Under an Axiolab light microscope (Zeiss) with 400× resolution, twenty images were relayed to an image analysis system (Kontron Elektronik image analyser; Carl Zeiss, Germany—KS300 software). Reading of slides was performed randomly by two pathologists. A 10× objective and 40× photomicrograph of liver were used to obtain the images.
Gut microbiota composition by high-throughput 16S rRNA sequencing
Stool samples from animals were collected at three consecutive days just before euthanasia. Genomic DNA was extracted from stool using a commercial kit (PowerSoil DNA, Qiagen, Germany). DNA integrity was assessed using a 1% agarose gel and quantified by fluorometry (Qubit, LifeTechnologies, USA). Amplicon libraries were prepared for the V3-V4 regions using Nextera XT Indexkit (Illumina) amd magnetic beadsfor cleaning and purification (AMPureXP, Beckman, Indianapolis, USA). Paired-end sequencing was performed in Illumina MiSeq using a 500 cycle (2 × 250) V2 kit.
The raw demultiplexed paired-end sequences were downstream processed in QIIME 2 platform v.20.828. Reads were filtered, denoised and parsed for non-chimeric sequences using DADA229, producing Amplicon Sequence Variants (ASV) and it respective feature table. For phylogeny, sequences were aligned using SEPP30 according to Green Genes v.13.8 reference sequences. Apha diversity was evaluated by Chao1, Simpson and Shannon indices, while beta diversity was analysed by means of both weighted and unweighted Unifrac distance matrices. The visualisation plots of relative abundances, alpha and beta diversity metric measures were performed using phyloseq v.1.8.231 in R v.3.5.7. Taxonomic classifications were attributed using the Naïve Bayes method based on Green Genes database v.13.8 with 99% of similarity for the V3-V4 regions32. Differential abundance was assessed using machine learning classification of ASV based on the Random Forest method using nested stratified k-fold cross validation for automated hyperparameter optimisation and sample prediction, then taxonomic classification was performed using BLASTn 33 based on each amplicon sequence.
Statistical analysis
The data were analysed using the SAS software (SAS inst., Cary) to apply a linear model where the dietary treatment was the single fixed factor and the animals as the experimental units. Whenever significant effects were detected (p < 0.05) the least square means were compared by the Tukey post hoc multiple comparison test. The non-parametric tests Kruskal-Wallis and PERMANOVA were used to assess the statistical significance of alpha and beta diversity, respectively. Machine learning based on the fandom forest method was used to perform microbial differential abundances.