The experiment was conducted on a commercial farm of Western Paraná (Toledo, PR, Brazil, 24° 42'50" S, 53° 44' 34" O), with lactating sows from the weaned piglet production unit. The Animal Use Committee of the Universidade Estadual do Oeste do Paraná approved the protocol for the experiment (number 17-2022). All methods were carried out in accordance with relevant guidelines and regulations. All methods were reported in accordance with ARRIVE guidelines (https://arriveguidelines.org/arrive-guidelines).

A total of 60 hybrid sows at 110 days of gestation (248.4 ± 2.4 kg BW) from a commercial lineage (Landrace × Large White, Agroceres PIC^®) were selected for the study. The sows were housed in farrowing crates for a period of 26 days (5 days of acclimatization and 21 days of lactation). The sows were allocated to a randomized block design to 1 of 3 dietary treatments, with 5 sows per treatment in each block (n = 4 blocks of rounds in time), resulting in 20 sows/treatment. The sows were classified according to their BW as light-, medium-, or heavy-weight and according to farrowing order into 3 groups: P1, primiparous (13 sows); P2, sows with 2 or 3 farrowings (29 sows); and P3, sows with 4 or 5 farrowings (18 sows). The average farrowing order of sows assigned to each treatment was 2.8, 2.6, and 2.7 farrowings for the groups fed the diet containing xylanase (CD40), CD40 + β-mannanase (CD85), and CD40 + β-mannanase (CD100), respectively.

Five days before farrowing, all sows were transferred to the same house in individual conventional farrowing crates (2.4 × 1.6 × 1.3 m), with a slatted metal floor and equipped with front gutter feeders and nipple drinkers, with water provided ad libitum. A heated creep area (0.98 × 0.68 × 0.66 m) was accessible to piglets to maintain thermal comfort between 28 °C and 32 °C. The farrowing crates were installed in a commercial facility with a concrete floor and side curtains. The ambient temperature (25.3 ± 0.4 °C) and relative humidity (60.0 ± 1.8%) were recorded throughout the experimental period using a datalogger (UNI-T^®, model UT330B digital USB; Beijing, China) positioned at the center of the facility.

The experimental diets were corn and soybean meal-based, supplemented with synthetic amino acids, and formulated to meet the nutrients requirements of the animals (Rostagno et al., 2017). The diets were provided in mash form, varying only in soybean oil and neutral energy replacement (kaolin, an inert compound) contents (Table 1). The experimental diets were provided according to the reproductive phase of the sows, beginning at 110 days of gestation and ending at 21 days of lactation. In the prepartum period (5 days before parturition), feeding was controlled by providing 2 kg of diet per animal once a day. The daily diet was weighed and stored in labeled plastic bags. Sows were not fed on the day of farrowing. During lactation, the sows were fed 4 times a day (07:30, 11:30, 14:00, and 17:30), and the average daily diet consumption was 7 kg.

Table 1

Item	Experimental diets1
	CD40	CD85	CD100
Ingredient (%)
Corn grain (7.88% CP)	55.43	55.43	55.43
Soybean meal (45.4% CP)	27.24	27.24	27.24
Whole soybeans (36% CP)	6.00	6.00	6.00
Sugar	3.00	3.00	3.00
Soybean oil	2.97	2.42	2.24
Fish meal (53% CP)	2.00	2.00	2.00
Dicalcium phosphate	1.61	1.61	1.61
Inert (kaolin)	-	0.51	0.69
Calcitic limestone	0.45	0.45	0.45
Common salt, NaCl	0.44	0.44	0.44
Premix, mineral and vitamin supplement2	0.30	0.30	0.30
Adsorbent, Mycofix®	0.20	0.20	0.20
L-lysine (54.6%)	0.177	0.177	0.177
DL-methionine (99.5%)	0.056	0.056	0.056
L-threonine (96.8%)	0.061	0.061	0.061
L-tryptophan (99.0%)	0.017	0.017	0.017
L-valine (95.5%)	0.049	0.049	0.049
β-mannanase	-	0.030	0.030
Calculated nutritional requirements
Metabolizable energy (kcal/kg)	3,360	3,315	3,300
Crude protein (%)	20.50	20.50	20.50
SID lysine (%)	1.080	1.080	1.080
SID methionine + cysteine (%)	0.602	0.602	0.602
SID threonine (%)	0.700	0.700	0.700
SID tryptophan (%)	0.240	0.240	0.240
SID valine (%)	0.892	0.892	0.892
Total calcium (%)	0.80	0.80	0.80
STTD phosphorus (%)	0.43	0.43	0.43
Total sodium (%)	0.200	0.200	0.200
Starch (%)	34.21	34.21	34.21
Crude fat (%)	6.89	6.34	6.16
Crude fiber (%)	2.99	2.99	2.99

SID - standardized ileal digestible; STTD - standardized total tract digestible.

The tested dietary treatments were: 1) a diet containing xylanase (Control; reduction of 40 kcal of ME/kg of diet, CD40), 2) CD40 + β-mannanase (0.3 g/kg, reduction of 45 kcal of ME/kg of diet, CD85), and 3) CD40 + β-mannanase (0.3 g/kg, reduction of 60 kcal of ME/kg of diet, CD100). The ME reduction in each dietary treatment was based on the sows’ requirement of 3,400 kcal/kg of diet (Rostagno et al., 2017). The ME of the diets was reduced by reducing the inclusion of soybean oil.

Xylanase (Sunhy Biology Co., Ltd, Wuhan, HB, China; registration no. PR-08978 03462) was obtained from Trichoderma longibrachiatum with an enzyme activity of 10,000 U/g. One unit of xylanase refers to the amount of enzyme capable of releasing 1 micromole of reducing sugar from a xylan solution (5 mg/mL) at 37 °C and pH 5.5 (Genova et al., 2023a,b). β-mannanase (Elanco Animal Health, Inc., São Paulo, SP, Brazil; registration no. SP-59122 30011, Hemicell™ HT) was obtained from Paenibacillus lentus and has an enzyme activity of 160,000 U/g. One unit of β-mannanase is defined as the amount of enzyme required to release 0.72 mcg of reducing sugars (equivalent to D-mannose) per minute from goma locust (mannans concentration of 88%) at 40 °C and pH 7.5 (Genova et al., 2023a,b).

The procedures adopted are in accordance with those described by Genova et al. (2023a,b). Blood collection (without fasting) was performed on day 18 of lactation in 8 sows per treatment, using the cranial vena cava collection technique, with a 10 mL syringe and a 1.2 × 40 mm needle. The samples were transferred to tubes (glass vacuum blood collection tube, Labingá; Maringá, PR, Brazil) without the addition of anticoagulants. Immediately after collection, the samples were stored in thermal boxes (4 °C) until the end of the blood sampling. Subsequently, blood samples were centrifuged at 1,000 × g for 10 min at room temperature, and the serum was carefully collected and stored at −80 °C being sent to the IMUNOVA laboratory (Curitiba, PR, Brazil).

Serum cytokines concentrations were analyzed in technical duplicates, as follows: interleukin-1β (IL-1β), interleukin-6 (IL-6), interleukin-8 (IL-8/CXCL8), interleukin-12/interleukin-23 (IL-12/IL-23p40), interferon-α (IFN-α), interferon-γ (IFN-γ), tumor necrosis factor-α (TFN-α), interleukin-4 (IL-4), and interleukin-10 (IL-10), using specific ELISA kits (ThermoFisher Scientific, Vienna, Austria) according to the manufacturer’s instructions.

On day 21 of lactation, 10 g of fecal samples were manually collected from 8 sows per treatment. The samples were collected by rectal stimulation, using swabs, and placed in sterile 4 mL Eppendorf-type tubes. Immediately, the sampled material was stored in thermal boxes (4 °C) until the end of the collection period. Subsequently, the samples were stored at −80 °C until the analyses.

A commercial kit (ZR Fecal DNA MiniPrep^© from Zymo Research) was used to extract DNA from fecal samples following the manufacturer’s instructions. DNA quality and quantification were evaluated using NanoDrop spectrophotometry (Thermo Scientific™). A segment of approximately 460 base pairs of the V3–V4 hypervariable region of the 16S ribosomal RNA gene was amplified using the 375F/805R primer pair for bacterial analysis, under the following PCR conditions: 95 °C for 3 min; 25 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, followed by a step at 72 °C for 5 min. The amplicons were ligated to Illumina^® Nextera dual index barcodes under the following conditions: 95 °C for 3 min; 8 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s, followed by a step at 72 °C for 5 min. The products were then purified, pooled together, and subsequently sequenced on the Illumina^® NextSeq sequencer (Degnan and Ochman, 2012) as paired-end reads of 300 base pairs.

The reads obtained by the sequencer were analyzed on the Quantitative Insights Into Microbial Ecology (QIIME 2) platform, following a workflow that uses forward and reverse sequencing (R1 and R2) in bacterial analysis, including the removal of low-quality sequences, filtration, chimera removal, and taxonomic classification. The sequences were classified into bacterial genera using amplicon sequence variants (ASVs), which involved comparing the homology of the sequences against a database. To compare the amplicons of 16S rRNA gene regions, the 2022 update (Genome Taxonomy Database, GTDB 207 version) of the GTDB bacterial taxonomy database (Degnan and Ochman, 2012) was used, with in silico reads extracted from the same amplified regions.

To generate the classification of the bacterial communities through the identification of ASVs, 75,732 reads per sample were used for the bacterial analysis, with all samples being analyzed. The use of a minimum number of reads per sample aimed at normalizing the data and avoiding statistical comparisons among samples with different sequencing depths.

Families and higher classifications that had uppercase letters as suffixes represented non-monophyletic groupings in the construction of the GTDB reference phylogeny, although there is evidence that these groups are monophyletic. Genera with the same suffix pattern were polyphyletic or subdivided according to the database’s reference phylogeny. Species names can also have the same suffix pattern if they categorize a group that may contain ambiguity in the correct nomenclature (Parks et al., 2022). Other codes present in the classifications (e.g.: NSJ-53 or sp001543345) represent the first material used as a reference for clustering into the respective operational taxonomic units (OTUs).

The serum cytokine concentration was analyzed using GraphPad Prism software version 8. Outliers were identified using the ROUT test (Q = 1%). The data were tested for normality using the Shapiro-Wilk test. Based on the results, if the data met the normality assumption, a linear mixed model was applied, with treatment as a fixed effect and block as a random effect, expressed as:

in which Y_ijk = observed value of the dependent variable in the i-th treatment, in the j-th block and in the k-th replicate; μ = overall average; T_i = fixed effect of treatment (group, i = 1, 2, and 3); b_j = random effect of block (j = 1, 2, 3 and 4); e_ijk = residual error term, assumed to be independently and normally distributed with mean zero and constant variance. If the data did not follow normality, an equivalent non-parametric model was used, based on rank analysis, whose null hypothesis is that the distributions between the groups are equal. For both tests, a difference was considered significant when P<0.05.

The statistical analyses and graphs of the fecal microbiome were performed using R software (R Core Team, 2020). The statistical comparison between groups in alpha diversity analyses was conducted using the non-parametric Kruskal-Wallis test followed by Dunn’s post hoc. Beta diversity was analyzed using permutational multivariate analysis of variance (PERMANOVA) in the QIIME2 pipeline, with 10,000 permutations, which was expressed as:

in which Y = matrix of pairwise dissimilarities (e.g. Bray-Curtis); X = design matrix encoding group membership; β = estimated coefficients (vector) for group effects; e = error term (residual matrix). Alpha diversity analyses were calculated using the “phyloseq” (McMurdie and Holmes, 2013), “vegan” (Oksanen et al., 2007), and “Microbiome” (Lahti and Shetty, 2018) libraries. To identify specific taxa that may have undergone modulation between the groups, differences in relative abundances were analyzed using the Kruskal-Wallis test with Dunn’s post hoc at a 95% confidence level for lower taxonomic classifications: family, genus, and species.

There was no effect (P>0.05) of dietary treatments on serum cytokines concentrations in lactating sows (Figure 1), but the cytokines IL-12/IL-23p40 exhibited the highest concentrations among all those evaluated (Figure 2).

Figure 1 Data were presented as means ± SEM. Differences in serum concentrations of (A) interleukin-1β (IL-1β), (B) interleukin-4 (IL-4), (C) interleukin-6 (IL-6), (D) interleukin-8 (IL-8/CXCL8), (E) interleukin-10 (IL-10), (F) interleukin-12/interleukin-23p40 (IL-12/IL23p40), (G) interferon-α (IFN-α), (H) interferon-γ (IFN-γ), and (I) tumoral necrosis factor-α (TNF-α) in sows fed to one of three dietary treatments: 1) a diet containing xylanase (Control; valorization of 40 kcal of ME/kg diet, CD40), 2) CD40 + β-mannanase (0.3 g/kg, valorization of 45 kcal ME/kg diet, CD85), and 3) CD40 + β-mannanase (0.3 g/kg, valorization of 60 kcal ME/kg diet, CD100).

Figure 2

Sows fed the CD85 diet showed (P = 0.014) higher alpha diversity richness than those fed the CD100 diet based on the Simpson index (0.98 vs. 0.97, Figure 3D). The remaining alpha diversity indices were not altered by the diets (Chao1, observed OTUs, Fisher, Shannon, and Pielou) (Figure 3). No significant effect was observed on beta diversity assessed using Bray-Curtis (P = 0.527), Jaccard (P = 0.526), UniFrac (P = 0.687), and Weighted UniFrac (P = 0.547) parameters (Figure 4).

Figure 3

Figure 4

The phyla, classes, orders, families, genera, and species with an average relative abundance above 2% in at least one of the groups tested were plotted in the graphs (Figure 5). In the fecal samples analyzed, the most abundant phylum was Firmicutes, followed by Bacteroidota, Spirochaetota, Actinobacteriota, and Cyanobacteria. Furthermore, the classes Clostridia, Bacilli, Bacteroidia, Negativicutes, Spirochaetia, Coriobacteriia, and Vampirovibrionia showed the highest abundances. The most abundant orders were Christensenellales, Oscillospirales, Peptostreptococcales, Clostridiales, Lachnospirales, Bacteroidales, TANB77, Lactobacillales, Treponematales, Coriobacteriales, Haloplasmatales_A, Gastranaerophilales, and Erysipelotrichales. The families with the highest relative abundance were Peptostreptococcaceae, Clostridiaceae, Oscillospiraceae, Lachnospiraceae, CAG-74, CAG-508, Treponemataceae, Lactobacillaceae, Muribaculaceae, Ruminococcaceae, Acutalibacteraceae, Turicibacteraceae, NSJ-53, Christensenellaceae, Gastranaerophilaceae, Erysipelotrichaceae, and Bacteroidaceae. The most abundant genera were Clostridium, Terrisporobacter, CAG-83, Romboutsia, Sodaliphilus, GCA-900199385, Limivicinus, Onthenecus, Limosilactobacillus, Turicibacter, Merdicola, Fimivivens, and Fimenecus. Species that showed relative abundance were Sodaliphilus sp004557565, GCA-900199385 sp900322155, Clostridium baratii, Limivicinus sp002320035, Onthenecus sp900199405, Turicibacter sp001543345, CAG-83 sp900549395, NSJ-53 sp014384795, NSJ-63 sp014384805, Merdicola sp001915925, Fimivivens sp900113995, Clostridium butyricum, and Fimenecus sp004556705.

Figure 5

No differences were observed between the groups for the Firmicutes:Bacteroidetes ratio (Figure 6). The Acutalibacteraceae family was more abundant in the sows fed CD100 diet than in those fed CD85 diet (P = 0.044, Figure 7A). Furthermore, sows fed CD85 diet had a higher abundance of the CAG_508 (P = 0.012, Figure 7B) and NSJ_53 (P = 0.044, Figure 7C) families than those fed CD100 diet.

Figure 6

Figure 7 Statistical tests were performed using the Kruskal-Wallis test and Dunn's post hoc test, with P<0.05.

The Fimenecus genus exhibited a lower abundance in sows fed CD85 diet compared to sows fed CD40 (P = 0.007, Figure 8A) or CD100 diets (P = 0.005, Figure 8A), while the NSJ-53 genus showed a higher abundance in sows fed CD85 diet than in those fed CD100 diet (P = 0.044, Figure 8B). Similarly, these results followed (P<0.05) for the Fimenecus sp004556705 (Figure 9A) and NSJ-53 sp014384795 (Figure 9B).

Figure 8 Statistical tests were performed using the Kruskal-Wallis test and Dunn's post hoc test, with P<0.05.

Figure 9 Statistical tests were performed using the Kruskal-Wallis test and Dunn's post hoc test, with P<0.05.

In the present study, dietary treatments, including β-mannanase supplementation in ME-reduced diets containing xylanase, did not affect serum cytokine concentrations in lactating sows. Cytokines such as IL-1β, IL-6, IL-10, TNF-α, and IL-12/IL-23p40 serve as indicators of immune status and inflammation (Meizlish et al., 2021). The absence of treatment effects on these markers indicates that, under the present experimental conditions, the enzymatic strategy adopted did not alter systemic inflammatory or regulatory immune pathways. Monitoring cytokine profiles provides information about the immune response, but may not fully capture localized or subtle immune modulations driven by the microbiota, which could occur at the intestinal level without systemic reflection.

In the present study, IL-12/IL-23p40 showed the highest concentrations among the cytokines evaluated in all treatment groups, which is consistent with its central role in innate immune activation and modulation of inflammatory responses (Cavaillon et al., 2020; Splichalova et al., 2023). However, because IL-12/IL-23p40 and the other cytokines did not differ between dietary treatments, these patterns likely reflect the physiological immune status of lactating sows rather than a diet-induced response. Thus, the combination of xylanase and β-mannanase in ME-reduced diets did not induce overt systemic inflammation or cytokine “storm”.

The relationship between dietary enzymes and immune responses is complex and may depend on the specific substrates available in the gastrointestinal tract (Kiarie et al., 2022). Our data suggest that even though β-mannanase and xylanase altered the intestinal environment, these effects were not enough to translate into measurable changes in systemic cytokines. Therefore, any potential immune modulation is likely to be local or of low magnitude and cannot be demonstrated by the current serum data.

The reduction of ME at up to 100 kcal/kg in the diets did not appear to compromise the sows’ immune status, as serum cytokine levels remained unaffected. Immune activation, while essential for pathogen defense, involves energetic costs that can impact nutrient partitioning and overall productivity (Wang et al., 2023). For example, changes in the dietary starch-to-fat ratio altered inflammatory cytokine concentrations during critical periods such as parturition (Wang et al., 2023). This suggests that within the parameters evaluated, moderate reductions in energy do not cause systemic immune activation (Rodrigues et al., 2022). The moderate reduction in ME applied here did not cause detectable systemic immune activation, suggesting that the nutritional adjustment was well tolerated by lactating sows based on the serum cytokine profile. Nonetheless, localized or subclinical effects cannot be excluded and would require tissue-level assessments in future studies.

While a systematic review indicates that β-mannanase supplementation reduced inflammatory responses caused by β-mannan and released prebiotic-like products into the intestinal tract (Kiarie et al., 2022), our findings do not support systemic immunomodulation by β-mannanase under the current conditions, and fecal microbiota data point to limited microbial ecological adjustments rather than functional changes.

In the present study, the hypothesis that β-mannanase supplementation could influence the fecal microbiome composition and diversity in lactating sows was supported by the observed increase in microbial alpha diversity, specifically based on the Simpson index. The fecal microbiota is influenced by several factors, including host genetics (Spor et al., 2011), diet (Wu et al., 2011), and the immune system (Slack et al., 2009). The microbiota plays an important role in various aspects of animal physiology, such as the inflammation process, metabolic syndromes (Chassaing et al., 2014), energy metabolism (Donohoe et al., 2011), and immune responses (Peng et al., 2021). In this study, β-mannanase supplementation in ME-reduced diets containing xylanase resulted in limited changes within the fecal microbial community.

Lactating sows fed the CD85 diet exhibited a slight but statistically significant increase in alpha diversity based on the Simpson index compared to those fed the CD100. This difference was slight in magnitude and was not accompanied by changes in other alpha diversity indices (Chao1, observed OTUs, Shannon, Pielou) or in beta diversity metrics (Bray-Curtis, Jaccard, UniFrac, weighted UniFrac), indicating that the overall profile of the microbial community remained stable across all treatments.

Consistent with previous research (Vojinovic et al., 2019), the observed microbial modifications highlight the ability of the microbiota to adapt subtly in response to dietary interventions. These modest modifications are compatible with the concept that the intestinal microbiota of lactating sows is relatively resilient and may adjust to energy-reduced, enzyme-supplemented diets with only limited changes in dominant taxa.

The balance between Firmicutes and Bacteroidetes in fecal microbiota is often studied for its potential influence on intestinal health and metabolism, given that their proportions can be associated with several physiological and pathological conditions. In this study, the Firmicutes-to-Bacteroidetes ratio was not significantly affected by the dietary treatments, suggesting that the energy reduction and supplementation strategies employed did not markedly alter these predominant microbial groups. This lack of response may suggest the resilience of these microbial profiles to dietary interventions or indicate that the modifications primarily impacted other microbial populations or functions without significantly changing the relative abundances of these major phyla.

Across all treatment groups, Firmicutes remained the dominant phylum, accounting for over 80% of the microbial community. These findings are consistent with previous research documenting the predominance of Firmicutes in the intestinal microbiota of lactating sows (Liu et al., 2019). The high relative abundance of Firmicutes is often associated with an enhanced capacity for energy extraction from the diet (Turnbaugh et al., 2006), which may support the increased energetic demands during lactation. However, the observed microbial composition remained stable across diets, indicating that the dietary modifications did not substantially alter the profile at the phylum level.

At the class level, Clostridia was the predominant bacterial class in all groups of lactating sows, aligning with prior studies that analyzed the dynamic changes in intestinal microbiota during the perinatal period (Sun et al., 2020) and lactation (Xiong et al., 2019). Additionally, the Clostridia class of bacteria has a multitude of enzymes (e.g.: butyryl-CoA:acetate-CoA transferase) that facilitate the breakdown of polysaccharides into short-chain fatty acids (SCFA). Consistently, research on the Clostridium genera showed that they can produce butyrate (Holman et al., 2017; Vital et al., 2014; Parada Venegas et al., 2019) and have associated butyrate with the prevention and treatment of intestinal disorders, by reducing inflammation in the host’s intestine, allowing intestinal epithelial cells to use it as a source of energy (Hamer et al., 2008). Nevertheless, the present data do not allow us to infer functional consequences at the host level, particularly because systemic cytokines remained unchanged among treatments.

Regarding specific microbial families, sows fed the CD100 diet exhibited a higher relative abundance of the Acutalibacteraceae family compared to the CD85 diet. However, the information available on the relevance of this family to the porcine intestinal microbiota is limited (Barducci et al., 2024). On the other hand, sows fed the CD85 diet exhibited a higher relative abundance of the CAG-508 and NSJ_53 families compared to the CD100 diet. These taxa remain poorly characterized in pigs, and their biological relevance in lactating sows is currently uncertain (Qing-Bo et al., 2023). Therefore, the observed differences in their relative abundance should be interpreted as exploratory signals that warrant further investigation rather than as evidence of specific functional changes.

The Christensenellaceae family, represented by the NSJ_53 genus, has been proposed as a potential probiotic candidate in other species with some evidence linking it to metabolic regulation and intestinal homeostasis (Waters and Ley, 2019; Alemán et al., 2018). In pigs, the specific role of Christensenellaceae remains to be elucidated. While the literature suggests potential beneficial effects (Waters and Ley, 2019; Alemán et al., 2018), the current data do not support a definitive functional role for this taxon in systemic health or immune modulation.

The Fimenecus genus has limited characterization in the context of porcine microbiota, though studies in humans report high pectin-degrading activities, including arabinases, galacturonases, and galacturonidases (Calvete-Torre et al., 2023). Its potential role in pig intestinal health and immune function remains speculative (Wiese, 2019), warranting further research. Therefore, the microbial shifts observed, such as modest increases in diversity and specific taxa, likely reflect minor ecological adjustments. These subtle microbial adaptations may play a key role in maintaining intestinal stability during dietary interventions.