The in vivo trial was approved by the Ethics Committee for Experiments on Animals at the University of Bologna, Italy and by the Italian Ministry of Health (Authorization n. 287/2021 PR, issued in compliance with art. 31 of the D.lgs. 26/2014) and complied with the Animal Research Reporting of In Vivo Experiments guidelines (Percie du Sert et al., 2020).

A total of 64 piglets (Large white × Landrace) were selected from multiparous sows (5 ± 1.5 parity order) reared on an Italian commercial farm. Healthy piglets from each litter were weighed at birth and were categorized as LBBW when the BBW was <1 kg (0.92 ± 0.09 kg; n=32) or NBBW when the BBW was >1 kg (1.37 ± 0.09 kg, n=32). The piglets were then monitored and weighed 3 days after birth (LBBW: 1.23 ± 0.00014 kg; NBBW: 1.74 ± 0.00021 g) and again at weaning (25 days of age; LBBW: 6.284,52 ± 0.75 kg and NBBW: 7.777,50 ± 0.77 g). No creep feed was provided during the suckling phase.

At weaning (d0), the piglets were transported to the experimental unit at the University of Bologna and arranged in a 2 x 2 factorial design that included BBW (LBBW or NBBW) and diet (ZnSO₄ or pZnO) (HiZox, Animine, Annecy, France) as factors. Therefore, the piglets were divided into four experimental groups (8 replicates of 2 piglets/group) balanced for litter of origin (both BBW classes were represented in each litter) and BW (within each BBW class): 1) LBBW piglets receiving a diet with ZnSO_4; 2) LBBW piglets receiving a diet with pZnO; 3) NBBW piglets receiving a diet with ZnSO_4; and 4) NBBW piglets receiving a diet with pZnO.

To prepare the two diets, the basal diet was initially produced in a single batch without Zn addition. After production, Zn concentrations were determined in the basal diet. Then, the basal diet was divided into two sub-batches. Each sub-batch was integrated with one of the two Zn sources to reach approximately 150 mg/kg of total Zn using a small mixer.

Humidity, crude protein, crude fat, and crude fiber in the diet were analyzed following the methods approved by the European (2009). The zinc content of the diets was analyzed using inductively coupled plasma mass spectrometry ICP/MS (Wilschefski and Baxter, 2019). The final Zn levels of the two diets, obtained after supplementation with HiZnO and ZnSO₄, respectively, fall within the EU authorized range, considering the analytical error of the analysis. Calcium and phosphorus concentration analyses were carried out using inductively coupled plasma optical emission spectroscopy (ICP-OES, AOAC Official Method 984.27; AOAC Int., 2007). Metabolizable and Net energy (ME and NE, respectively) were estimated using the InraPorc software (Version 1.8.0.0) (van Milgen et al., 2008) based on the Noblet et al. (1994).

The estimated and analyzed composition values of the basal diet are reported in Table 1 .

During the study, the piglets were fed ad libitum with continuous access to water. The piglets were kept in pens with a slatted floor containing enrichment material consisting of metal chain and cotton rope. The room temperature was kept at 30°C at the beginning of the trial and then gradually decreased to 26°C by the end of the trial (day 21). In addition, infrared lamps were used during the first 10 post-weaning days. Temperature and humidity were recorded daily.

Piglets were weighed individually on d0 and then every week until d21 post-weaning (end of the trial). Feed intake (FI), general health status, and fecal consistency (5-point scale: 1 hard feces – 5 watery feces) were recorded daily. The piglets were considered to have diarrhea when the fecal score was > 3 (Correa et al., 2022). The piglets (one piglet/replicate) were slaughtered at two time points: d9 (total of 32 piglets; acute post-weaning phase) and d21 (total of 32 piglets; post-weaning recovery phase). At slaughter, the piglets were sedated and euthanized as previously described by Trevisi et al. (2023).

At d9 and d21, the piglets were fasted for 12 hours in order to reach the basal physiological level of Zn in the blood circulation and to avoid confounding factors regarding the determination of Zn in the target tissues. Blood samples were collected at d9 (from all the piglets) and d21 (from the remaining piglets) in 10 mL tubes with a clot activator (Vacutest Kima Padova, Italy). After 2 hours of clotting, the blood samples were centrifuged at 2,000 x g at room temperature for 10 minutes to obtain the serum. The serum was then stored at –80°C until the analyses for reactive oxygen metabolite (ROM), haptoglobin (Hp), and Zn concentrations.

At slaughter, the intestinal content from the distal jejunum, cecum, and colon was immediately sampled and processed for pH determination (Vio, Giorgio Bormac S.r.l., Carpi, MO, Italy). For microbiota profile characterization and short-chain fatty acid (SCFA) analyses, a sample of the colon content was collected in a sterile tube, immediately frozen in liquid nitrogen, and then stored at -80°C. Two samples of the distal jejunal mucosa were collected from each piglet. One was gently scraped, promptly frozen in liquid nitrogen, and then stored at -80°C for gene expression analyses, and the second one was fixed in formalin for immunohistochemical and morphological analyses. Finally, a liver sample (from the same lobe) was collected, frozen in liquid nitrogen, and then stored at -80°C for Zn concentration analyses.

ROM concentration was determined colorimetrically from the serum samples using the d-ROMs test kit (Diacron International Sr1, Grosseto, Italy). Regarding the automatic analysis of the ROMs (Brambilla et al., 2002), the analyses were carried out as described by Correa et al. (2022). The concentrations of Hp were determined in the serum samples using the Tridelta Phase Hp Kit (Tridelta Development Ltd., Maynooth, Co. Kildare, Ireland) as previously described by Trevisi et al. (2017). Zn concentrations in the serum and liver were analyzed using inductively coupled plasma optical emission spectrometry (ICP-OES, Method 985.01 A, B, and C; 135 AOAC Int., 2007) after wet ash sample preparation (Method 975.03 B(b); AOAC Int., 2007).

Total RNA was isolated from the jejunal mucosal samples of the piglets and transformed into complementary DNA (cDNA) using the method described by Luise et al. (2023). Duplex real-time PCR was carried out for the following specific genes: IAP (intestinal alkaline phosphatase) as an intestinal development marker; CLAUD4 (Claudin-4) for its role in the regulation of the tight junction pathway; NFKB2 (Nuclear Factor Kappa B Subunit 2), to assess the activation of inflammatory pathways; GPX2 (Glutathione Peroxidase 2), as a marker of oxidation status; SLC39A4 (Solute Carrier Family 39 Member 4), and SLC30A7 (Solute Carrier Family 30 Member 7) as Zn transporters. Each reaction consisted of 2 µL of cDNA and an 8 µL mix containing probe assays ( Supplementary Table S1 ), along with 2X TaqMan Mastermix. The assays were carried out in triplicate as previously described by Trevisi et al. (2023). Target gene expression was presented as a fold change, calculated using the formula 2-ΔΔCt.

Samples from the jejunum were preserved in 10% buffered formalin for 48 hours and subsequently embedded in paraffin for the morphological assessments. Sections of the paraffin were then stained with hematoxylin and eosin. Measurements of the height and width of 30 villi and the width and depth of 30 crypts were taken from each sample. Only villi and crypts aligned perpendicularly to the muscularis mucosae were selected for the morphometric study. Using a conventional microscope linked to a digital camera and computer with cytometry software (Byk Gulden, Milan, Italy), the sections were examined at a lower magnification. Villus height was defined as the distance from the base of the crypt to its top, while crypt depth was measured from its base to the level of the crypt opening. The mucosal-to-serosal amplification ratio (M), indicating the absorptive mucosal area in the jejunum, was computed as described by Correa et al. (2022).

The samples were observed using a Nikon Eclipse Ni microscope, and the images were captured with a Nikon DS-Fi2 digital camera using NIS Elements software BR 4.20.01 (Nikon Instruments Europe BV, Amsterdam, The Netherlands). A 20X objective lens was utilized for morphometric assessments.

Regarding the immunohistochemical procedure, the avidin-biotin peroxidase complex (ABC) technique was utilized. The paraffin sections initially underwent deparaffinization and rehydration. Antigenic sites were exposed by heating the slides in sodium citrate buffer (pH 6.0) using a microwave. To inhibit endogenous peroxidase, the sections were treated with a 1% methanol peroxidase solution for 30 minutes at room temperature. Subsequently, a 30-minute incubation in 0.01 M phosphate-buffered saline containing 10% goat or horse serum was carried out to block non-specific antibody binding. The slides were then left overnight at 4°C with the following antibodies: polyclonal goat anti-porcine immunoglobulin A (IgA) (1:2000; Novus Biologicals, NB724), polyclonal rabbit anti-CD3 serum (1:1500; Sigma Aldrich, C7930) and monoclonal mouse anti-Claudin-4 (1:300; ThermoFisher -Invitrogen, 32-9400, clone 3E2C1). After rinsing, the slides were incubated at RT for 1 hour with biotin-conjugated secondary antibodies [(goat anti-rabbit IgG, horse anti-goat IgG, and goat anti-mouse IgG, each diluted 1:200) (Vector)], followed by treatment with the ABC complex (Vector Elite Kit, Vector Laboratories). Visualization of the immune reactions was achieved using a 3,3’-diaminobenzidine chromogen solution (Vector DAB Kit, Vector Laboratories), and Toluidine blue was used to counterstain the sections. Figure 1 shows a representative image of porcine jejunal mucosa stained with anti-IgA, CD3, and Claudin-4 antibodies.

Quantitative assessment of the IgA cells followed the protocol described by Bianco et al. (2014). The IgA-positive cell counts were carried out in two functional regions of the intestinal lamina propria: 1) along the villus axis and 2) interspersed within the crypts. The IgA-positive cell density was quantified and expressed as the number of immunoreactive IgA cells per 4,000 μm². The lamina propria region was manually outlined to exclude significant blood vessels and epithelial cells.

The distribution and count of jejunal intra-epithelial lymphocytes, T lymphocytes along the villi, and the lamina propria between the crypts were assessed in three distinct areas: the first within the enterocytes above the epithelial basal lamina, the second within the lamina propria along the villus axis, and the third within the lamina propria interspersed with the crypts, as previously defined by Vega-López et al. (2001).

For Claudin-4 immunoreactivity assessment, 10 villi from each sample were examined. The sections were graded based on the presence, distribution, and intensity of Claudin-4, using the following scores: 1: light/limited staining, 2: moderate staining, and 3: intense staining, as detailed by Correa et al. (2022). The analyses involved 10 villi and 10 crypts per piglet.

The DNA extraction was carried out on 62 samples in total. The bacterial DNA present in the colon content was isolated using the FastDNA SPIN Kit for Soil (MP Biomedicals, Santa Ana, CA, USA), following the instructions provided. To assess DNA concentration and purity, a spectrophotometric analysis was carried out using the NanoDrop instrument (Thermo Fisher Scientific Inc., Waltham, Massachusetts, United States).

The V3-V4 regions of the 16S rRNA gene (~460 bp) were targeted for amplification. The amplification process utilized universal primers Pro341F and Pro805R as detailed by Takahashi et al. (2014). Platinum™ Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific Inc., Waltham, Massachusetts, United States) was used. The Illumina MiSeq platform was utilized for sequencing with a 300x2bp configuration. Library preparation adhered to the standard MiSeq Reagent Kit V3 protocol, and sequencing took place on the MiSeq instrument (Illumina Inc., San Diego, CA, USA).

For bioinformatic analyses, the DADA2 pipeline (Callahan et al., 2016) was used, and the Silva database (Quast et al., 2013) (version 138.1) was used for the taxonomic assignments. After primer removal, both forward and reverse reads were trimmed at positions 290 and 250, respectively, based on the average quality scores. Default settings for the other parameters within the DADA2 analysis were retained.

Examination of SCFAs (including acetate, propionate, isobutyrate, butyrate, valerate, and isovalerate) and lactic acid in the colon content samples was conducted using HPLC, as previously reported by Trevisi et al. (2023). Quantitative analysis was carried out based on an external calibration curve using standard solutions (Sandri et al., 2017).

R version 3.6 was used to carry out the statistical analyses, with the “car” 3.1-1 (Fox and Weisberg, 2019), “lm4” 1.1-31 (Bates et al., 2015), and “lsmeans” 2.30–0 packages (Lenth, 2021).

The four groups were arranged based on a 2 x 2 factorial design, taking into consideration the BBW class and diet. Prior to initiating the statistical assessments, the data distributions were examined. Parameters including BW, average daily gain (ADG), fecal score, intestinal content pH, ROMs, Hp, gene expression, morphological metrics, SCFAs, and lactic acid underwent analysis using a linear mixed model. This model included BBW class, diet, and their interaction as fixed factors, with the litter as a random factor. For the FI and the gain to feed (G:F) ratio, the pen was considered the experimental unit. These parameters were analyzed using a linear model with diet and BBW class as fixed factors. The difference between the groups was tested using Tukey’s post hoc test.

For the microbiota analysis, the vegan 2.6 (Dixon, 2003), phyloseq 1.38.0 (McMurdie and Holmes, 2011), and microbiomeMarker 1.0.2 (Cao et al., 2022) R packages were used for statistical analyses on alpha diversity, beta diversity, and taxonomic composition. For alpha diversity, the Chao1, Shannon, and Simpson indices were calculated, and differences among the groups were analyzed using a linear model. Diet, BBW class, their interaction, and sequencing depth were included in the model. The beta diversity was calculated utilizing the Bray–Curtis distance matrix, visualized via a non-metric multidimensional scaling (NMDS) plot. The effects of BBW class and diet were tested using a non-parametric PERMANOVA model (Adonis test) with 999 permutations. Differential abundance analysis concerning various taxa was carried out using Linear discriminant analysis Effect Size (LefSe) (Segata et al., 2011), integrated within the microbiomeMarker package (v 1.0.2). Data were aggregated at the genus level, with a cutoff value set at 3, and a P.adj < 0.05 was considered significant for the LefSe analysis (Correa et al., 2022).

The results were considered significant at P < 0.05, while P ≥ 0.05 and P < 0.10 were considered tendencies (Correa et al., 2022).

Two piglets out of 64 were excluded from the study due to reduced FI during the first week post-weaning and the consequent loss in BW (one in the ZnSO₄ group and one in the pZnO group from the LBBW class). No effect of diet and BBW class was observed on the fecal score. Overall, the piglets were healthy since the average fecal score was below 3 (cutoff for diarrhea).

Table 2 reports the effects of the diet and the BBW class on growth performance. The interaction between diet and BBW class was never significant. The BW was not affected by diet at d7, d9, and d21, while at d14, the piglets in the pZnO group tended to be heavier than the piglets in the ZnSO₄ group (P = 0.09). The BBW class significantly influenced the BW of the piglets; the NBBW piglets were heavier than the LBBW piglets at all the time points (P < 0.001).

The ADG was not affected by diet in the periods from d0 to d7, from d9 to d21, and from d0 to d21, while it tended to be higher in the pZnO group compared with the ZnSO₄ group when considering the periods from d0 to d9 (P = 0.07) and from d9 to d14 (P = 0.08). No differences were observed in the ADG in the periods from d0 to d7, from d0 to d9, and from d0 to d21 for the BBW, while it affected the ADG in the periods from d9 to d14 (P < 0.01) and from d9 to d21 (P = 0.01); the NBBW piglets had a higher ADG compared with the LBBW piglets (P < 0.01).

The FI was never affected by diet or by the interaction between diet and BBW in any of the periods analyzed, while the BBW class influenced piglet FI from d9 to d14, from d9 to d21, and from d0 to d21, resulting in a higher FI in the NBBW group (P < 0.01).

The G:F ratio was affected by diet considering the periods from d0 to d7 (P = 0.05), from d0 to d9 (P = 0.04), and from d0 to d21 (P = 0.04), resulting in a higher G:F ratio for the pZnO group as compared with the ZnSO₄ group. The BBW class never affected the G:F ratio except for a tendency from d0 to d9 (P = 0.09) when the LBBW piglets had a higher G:F ratio as compared to the NBBW piglets.

The effects of diet, BBW class, and their interaction on Zn concentrations in serum and liver, and on Hp and ROM concentrations are reported in Table 3 . The effects of interaction between diet and BBW class, diet, and BBW class were never significant.

Table 4 reports the effects of diet, BBW class, and their interaction on the gene expression in the jejunal mucosa at d9 and d21. The expression of NFKB-2 was significantly affected by the interaction between diet and BBW class (P = 0.03) at d9. The expression of NFKB-2 was also affected by diet (P = 0.02) and BBW class (P = 0.005); the pairwise contrasts showed that the LBBW group tended to have a higher NFKB-2 than the NBBW group but only in the ZnSO₄ group (P = 0.09). The other genes were not affected by diet, BBW class, or their interaction at d9. At d21, the interaction between diet and BBW class did not affect the expression of any gene. Diet tended to affect the SLC39A4 expression, resulting in higher expression in the ZnSO₄ group than in the pZnO group (P = 0.09). The other genes were not affected by diet at d21.

Table 5 reports the effects of diet, BBW class, and their interaction on the distal jejunum morphology and immunohistochemical parameters at d9 and d21. At d9, the interaction between diet and BBW class tended to affect the crypt depth (P = 0.08) and the number of T lymphocytes in the villi (P = 0.050). The T lymphocytes in the villi were lower in the LBBW piglets fed pZnO than in all the other groups. At d9, the diet almost entirely did not affect the morphological and immunological parameters, with the only exception of the number of T lymphocytes in the crypts (P = 0.0005), which was higher in the ZnSo4 group than in the pZNO group. The BBW class tended to affect the villus width (P = 0.08) at d9, resulting in wider villi in the NBBW than in the LBBW piglets. The BBW class tended to affect the IgA cell count in the villi at d9 (P = 0.059) and significantly affected the IgA in the crypts (P = 0.003); the NBBW piglets had higher values in both the villi and the crypts as compared to the LBBW piglets. The BBW class also affected the number of T lymphocytes in the crypts (P = 0.016), which was higher in the LBBW piglets.

At d21, the interaction between diet and BBW class tended to influence the villus height (P = 0.05) and the height-to-crypt-depth ratio (VH: CD) (P = 0.068), which were also affected by the BBW class (P = 0.005 and P = 0.008, respectively); the LBBW piglets had lower villus heights and a lower VH: CD compared to the NBBW piglets, and the LBBW piglets in the ZnSO₄ group tended to have lower villus height and lower VH: CD than the NBBW piglets in the ZnSO4 group (P = 0.06). The absorptive mucosal surface was affected by the interaction between diet and BBW class (P = 0.02) and by the BBW class (P = 0.008); it was higher in NBBW piglets than the LBBW piglets, and when examining the differences among the groups, it tended to be higher in the NBBW piglets than in the LNNB piglets fed ZnSO₄ (P = 0.08), while no differences were observed between the NBBW and LBBW piglets fed pZnO. The interaction between diet and BBW class tended to influence the IgA in the villi (P = 0.06), which were also affected by BBW class (P = 0.04); the NBBW piglets had higher IgA than the LBBW piglets. The interaction between diet and BBW class affected the number of T lymphocytes in the epithelium (P = 0.049), which was also affected by the BBW class (P <0.001); the LBBW piglets had a higher number of epithelial T lymphocytes. However, the LBBW piglets fed pZnO had a comparable number of epithelial T lymphocytes as the NBBW piglets fed pZnO. Finally, diet (P = 0.04) and BBW class (P = 0.002) influenced the number of T lymphocytes in the crypt, which was higher in the ZnSo₄ groups than the pZnO groups and in the LBBW piglets compared to the NBBW piglets. BBW tended to influence the Claudin-4 score, which was higher in the NBBW group (P = 0.09).

The effects of diet, BBW class, and their interaction on the pH values of the jejunum, cecum, and colon contents are reported in Table 6 . The interaction between diet and BBW class was never significant. Diet and BBW class did not influence the pH values at d9. At d21, the pH of the jejunum contents was lower in the piglets in the pZnO group compared with those in the ZnSO₄ group (P = 0.02). The BBW class did not influence the pH of the gut contents at d21.

Table 7 reports the effects of diet, BBW class, and their interaction on the SCFA and lactic acid concentrations in the colon content at d9 and d21. At d9, the interaction between diet and BBW class was never significant. The isovalerate tended to be influenced by diet, resulting in a higher value in the ZnSO₄ group (P = 0.06). The BBW class significantly affected the acetate (P = 0.03) and isobutyrate (P = 0.01) concentrations, and tended to influence valerate concentration (P = 0.09); acetate and valerate concentrations were higher in the LBBW piglets, while isobutyrate concentration was higher in the NBBW piglets.

At d21, the interaction between diet and BBW class tended to affect the lactic acid and acetate concentrations (P = 0.09). Lactic acid also tended to be affected by BBW class (P = 0.06) and it was affected by diet (P = 0.04). The LBBW piglets fed ZnSO₄ had higher lactic acid values than the NBBW piglets fed ZnSO_4. Acetate concentrations were also affected by the BBW class and were higher in the LBBW piglets (P = 0.03). The interaction between diet and BBW class also affected the level of valerate (P = 0.005), which was also affected by BBW class (P = 0.004) and tended to be affected by the diet (P = 0.06). Valerate concentration was higher in the LBBW piglets compared to the NBBW piglets only in the ZnSO₄ group (P = 0.02). Diet also affected the isovalerate concentration (P = 0.02), with higher values in the piglets that received pZnO. BBW class also affected the butyrate concentration (P = 0.02) and the total SCFA value (P = 0.01), and tended to affect the isobutyrate concentration (P = 0.06), which were all higher in the LBBW group than in the NBBW group.

A total of 2,728,354 raw reads were obtained, which were attributed to a total of 6803 amplicon sequence variants (ASVs). The ASVs were associated with 21 phyla, 89 families, and 225 genera. The predominant phyla comprised Firmicutes at 57 ± 0.11%, Bacteroidota at 31 ± 0.10%, Spirochaetota at 6 ± 0.05%, and Proteobacteria at 2 ± 0.02%. Among the most prevalent families were Prevotellaceae at 20 ± 0.09%, Lachnospiraceae at 17 ± 0.06%, Spirochaetaceae at 6 ± 0.05%, and Oscillospiraceae at 6 ± 0.03%. Notably, the most abundant genera included Prevotella at 12 ± 0.08%, Lactobacillus at 8 ± 0.12%, and Treponema at 5 ± 0.05%.

The alpha and beta diversity results are shown in Figures 2A, B and 3A, B , respectively. Alpha diversity was not affected by diet, BBW class, or their interaction. Regarding the beta diversity, the Adonis test showed that the BBW class tended to affect the microbial composition at d9 (R² = 0.04, P = 0.07) and significantly affected the bacterial composition at d21 (R² = 0.05, P = 0.03). Figures 4A and B show the results of the LefSe analyses. At d9, the LBBW piglets fed ZnSO₄ were characterized by a greater abundance of the Lachnospiraceae XPB1014 group (LDA_score=3.51, P.adj=0.03), Monoglobus (LDA_score=3.35, P.adj=0.01), and Shuttleworthia (LDA_score=3.20, P.adj=0.03); the NBBW piglets fed ZnSO₄ were characterized by a greater abundance of Phascolarctobacterium (LDA_score=3.85, P.adj=0.04); and the LBBW piglets fed pZnO were characterized by a greater abundance of Intestinimonas (LDA_score=3.13, P.adj=0.03). At d21, the piglets in the NBBW group fed ZnSO₄ were characterized by a greater abundance of Succinivibrio (LDA_score=3.95, P.adj=0.02), the LBBW piglets fed ZnSO₄ were characterized by a greater abundance of Streptococcus (LDA_score=3.97, P.adj=0.03) and Peptococcus (LDA_score=3.15, P.adj=0.01), and the piglets in the NBBW group fed pZnO were characterized by a greater abundance of Ruminococcus gauvreauii (LDA_score=4.09, P.adj=0.01).