Effects of processed soybean meal on growth performance and gut microbiome composition in pigs in regular nursery and enterotoxigenic Escherichia coli challenged conditions

Abstract

Hydrothermal-mechanical (HTM) processing of soybean meal (SBM) has been shown to enhance intestinal health and growth in post-weaning pig compared to conventional SBM. It was hypothesized that HTM processing improves protein utilization, particularly under enterotoxigenic Escherichia coli (ETEC) challenge, resulting in better growth and a more resilient hindgut microbiome. A total of 268 weaned pigs (6.82 ± 0.85 kg body weight) were allotted to regular nursery (5 pigs/pen, 10–11 pens/treatment) or ETEC challenge (3 pigs/pen, 12 pens/treatment) and fed one of three isocaloric diets with equal standardized ileal digestible lysine: SBM, HTM SBM, or enzyme-treated (Enz Trt) SBM. Test soy products replaced SBM in a wheat-barley-SBM base diet for the first 3 wk, followed by a common diet for 3 wk. On d 14 post-weaning, ileal digesta and feces were collected for crude protein (CP) digestibility, short-chain fatty acid (SCFA), and microbiome analysis. Growth performance, digestibility, and SCFA data were analyzed using general linear models and microbiome data from Nanopore shotgun sequencing were center-log-ratio transformed for statistical analysis in R. No diet × challenge interaction was observed on ADG, ADFI or BW. Pigs in regular nursery conditions had higher ADG (P < 0.01) and ADFI (P < 0.05) than ETEC challenged pigs during d 0–7. HTM SBM and Enz Trt SBM improved ADG (P < 0.05) with similar ADFI compared to SBM across conditions. From d 7–12, pigs fed HTM SBM or Enz Trt SBM had greater ADG (P < 0.05) and ADFI (P < 0.01) than SBM-fed pigs. BW remained lower (P < 0.05) in SBM-fed pigs from d 12–21 and during the final 3 wk. Under regular nursery conditions, HTM SBM improved apparent total tract digestibility of CP (P < 0.01) compared SBM, but with no difference from Enz Trt SBM. Microbiome composition was affected by diet (P < 0.01) and ETEC challenge (P < 0.01). HTM SBM and Enz Trt SBM tended to increase α-diversity (P = 0.10) of the microbiome compared to SBM, with no difference between the two treatments. HTM SBM and Enz Trt SBM increased abundance of species positively correlated with growth and beneficial SCFA, such as caproate (P < 0.05) and valerate (P < 0.05). In conclusion, HTM SBM and Enz Trt improved ADFI, ADG resulting higher BW and promoted beneficial microbes linked to performance in nursery pigs under both regular nursery and ETEC-challenged conditions.

.

Introduction

Throughout the production life, pigs are exposed to multiple stressors including physiological (eg weaning), nutritional (eg transition from highly digestible milk to plant-based solid feed), environmental (eg heat, crowding, transportation), and social factors (eg hierarchy) that can negatively affect their health and performance (Campbell et al. 2013). These stressors often reduce feed intake and contribute to impaired intestinal and immune function, ultimately leading to reduced growth in nursery pigs (Jean-Paul Lallès 2004; Campbell et al. 2013; Khafipour et al. 2014). To mitigate the adverse effects of weaning stress, the swine industry commonly uses high crude protein diets formulated with high-quality proteins to stimulate feed intake and protein digestion. Animal protein sources such as spray-dried plasma and fish meal are examples of high-quality protein that contains readily digestible protein and functional bioactive compounds (Pierce et al. 2005; Kim et al. 2022a). However, due to the high cost, limited supply, and biosafety concerns, soybean meal (SBM) is a more widely used plant-based protein source in the nursery diets for weaned pigs. Despite its widespread use, many nutritionists limit the use of SBM in the nursery feeds due to concerns regarding antinutritional factors ie β-conglycinin, trypsin inhibitor and glycinin associated with a typical SBM (Zheng et al. 2014; Ma et al. 2019). Furthermore, the immature gastrointestinal tract of newly weaned pigs frequently struggles to digest solid plant-based feed effectively, leading to a substantial amount of undigested protein entering the hindgut (Pieper et al. 2016). Environmental stressors and the presence of undigested protein in the hindgut can alter the microbiota (Pieper et al. 2016; Zhang et al. 2020). The toxic byproducts such as ammonia, biogenic amines of proteolytic fermentation often raise pH levels and contribute to the occurrence of diarrhea in pigs (Pluske et al. 2002; Kim et al. 2008; Gilbert et al. 2018).

Soybean meal is further processed to mitigate the limitations posed by antinutritional factors and to enhance protein digestibility (Cervantes-Pahm and Stein 2010; Ma et al. 2019). Common processing techniques in the industry include enzyme treatment of the SBM to hydrolyze carbohydrate fraction and reduce antinutritional factors; fermentation using various microorganisms to decrease antinutritional factors; and solvent extraction which helps retain a relatively high crude protein content while removing carbohydrate fraction and reducing soy allergens (Peisker 2001; Deng et al. 2022; Deng et al. 2023). Another method, hydro-thermal-mechanical (HTM) processing, has been employed to reduce antinutritional factors in SBM, increase nutrient digestibility, and thereby lower fermentable protein reaching the hindgut. Previous research has shown that HTM-processed soy products have higher energy content (Yanez et al. 2019) and HTM and enzyme-facilitated processing improved CP digestibility and apparent and ­standardized ileal digestibility (AID and SID) of most amino acids in pigs 17 d post-weaning (Ton Nu et al. 2020).

The F4+ enterotoxigenic Escherichia coli (ETEC) is known to be a major contributor to post-weaning diarrhea in nursery pigs (Luppi et al. 2016). Colonization of the bacteria in the small intestine results in secretion of enterotoxins, including the heat-labile (LT) and heat-stable (STa and STb) toxins, which lead to severe diarrhea and dehydration due to the loss of water and electrolytes and reduced growth in the infected pigs (Dubreuil et al. 2016; Luppi et al. 2016; Rhouma et al. 2017). Further, ETEC F4 infection was commonly reported to take place right after weaning or even pre-weaning (Garcia et al. 2020).

Building on these findings, it was hypothesized that HTM SBM would improve digestibility of CP, resulting in less proteolytic fermentation in the hind gut. This, in turn, is expected to create a better environment support a healthier, more robust microbiome in the hindgut compared to SBM. The improvement from HTM SBM would be similar or better than Enz Trt SBM. The effect of digestibility improvement would be stronger in the early weaning transition period when the gastrointestinal tract of the pig is immature. Better protein utilization and more robust microbiome would contribute to overall better growth performance of the pigs. Given ETEC F4 frequent occurrence in this phase, it would be particularly beneficial for the pigs to have reduced proteolytic fermentation when they are under post-weaning diarrhea challenge due to ETEC infection under commercial conditions. To test the hypotheses, the study was intended to evaluate the effect of HTM SBM versus Enz Trt SBM versus SBM on growth performance of pigs, CP ­digestibility, short-chain fatty acid (SCFA) production, and microbial profile of nursery pigs housed under either regular nursery conditions or challenged with ETEC.

Materials and methods

The Animal experiment was conducted at Cargill Innovation Center, Velddriel, the Netherlands, from January to March 2022. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Cargill Animal Nutrition, protocol code VLD_NP2201.

Animals, experimental design, and diets

Upon weaning, 268 nursery pigs (Yorkshire × Landrace; Topigs 20 × Pietrain, 134 barrows and 134 gilts at average 22 d of age) with initial body weight (BW) of 6.82 ± 0.85 kg were allotted to either a regular nursery condition (5 pigs/pen, 10–11 pens/treatment) or subjected to ETEC challenge (3 pigs/pen, 12 pens/treatment) in a split-plot design with initial BW and sex as blocks. Pigs in the regular nursery and ETEC challenge groups were housed in separate rooms. Before ETEC challenge, fresh pig feces were collected to confirm the absence of the targeted ETEC strain O149: K91: K88 (F4). Once confirmed negative, the O149: K91: K88 ETEC strain for oral gavage was prepared by NutriControl BV (the Netherlands) following their established protocol. On day 5 post-weaning, two out of three pigs per pen were randomly selected and individually gavaged with 10 mL of the ETEC strain at the concentration of 1.2 × 10⁹ CFU/mL to induce the challenge. Only two pigs within a pen were inoculated to maintain the ETEC challenge at a mild to moderate level, avoiding severe morbidity or mortality while meeting the experimental objectives. Two days after inoculation, some of the pigs were visually sick, with symptoms like diarrhea, little to no feed intake and increased temperature. Pigs within a pen piled together and appeared to be lethargic.

One of the three diets formulated with i) SBM, ii) a hydro, thermal, and mechanically processed SBM (HTM SBM, Provisoy, produced by van Tuijl feed enrichment, Kesteren, The Netherlands), or iii) an enzymatically treated SBM (Enz Trt SBM; HP300, Hamlet Protein) were randomly assigned to each pen in both regular nursery and ETEC conditions. A standard wheat-barley-SBM based diet was used and HTM SBM or Enz Trt SBM was added to replace SBM in the treatment diets during the first 3 wk (d 0–21 post-weaning) of nursery program. All experimental diets met or exceeded NRC (2012) requirements and diet composition is shown in Table 1. All diets were isocaloric with similar standardized ileal digestible lysine levels. A common diet was fed during the last 3 wk (d 21–39 post-weaning) of nursery. Nutrient analyses on the soy protein sources evaluated are shown in Table 2. All pigs had ad libitum access to water and experimental diets throughout the study.

Table 1.

Composition of diets used in the experiment (as-fed basis).

		First 3  wk			Last 3  wk
Item		SBM	HTM SBM	Enz Trt SBM	Common diet
Ingredient, %
SBM		26.62	0.00	0.00	19.94
HTM SBM		0.00	25.05	0.00	0.00
Enz Trt SBM		0.00	0.00	22.40	0.00
Corn		0.00	0.00	0.00	5.00
Wheat		29.05	29.05	32.37	39.18
Soy oil		3.59	3.59	3.65	1.91
Barley		25.00	25.00	25.00	30.00
Whey powder sweet		7.04	7.04	7.04	0.00
Sugar		4.00	4.00	4.00	0.00
DL-Methionine		0.28	0.28	0.25	0.23
L-Threonine		0.23	0.23	0.22	0.26
L-Tryptophan		0.04	0.04	0.03	0.03
L-Lysine HCl		0.56	0.56	0.62	0.59
L-Valine		0.09	0.09	0.07	0.08
L-Isoleucine		0.00	0.00	0.04	0.00
Vitamin E		0.02	0.02	0.02	0.02
Choline chloride		0.07	0.07	0.07	0.07
Copper sulphate		0.05	0.05	0.05	0.04
Iron sulphate		0.02	0.02	0.02	0.02
Manganese oxide		0.01	0.01	0.01	0.01
Zinc sulphate		0.03	0.03	0.03	0.03
Calcium carbonate		0.69	0.69	0.78	0.84
Monocalcium phosphate		0.65	0.65	0.60	0.56
Salt		0.62	0.62	0.59	0.22
Titanium dioxide1		0.30	0.30	0.30	0.00
Vitamin and mineral premix2		0.05	0.05	0.05	0.05
Phytase3		0.10	0.10	0.10	0.10
Organic acid blend4		0.90	0.90	0.90	0.45
Na bicarbonate		0.00	0.00	0.00	0.35
Antioxidant mixture		0.00	0.00	0.00	0.05
Water		0.00	1.57	0.80	0.00
Calculated composition
Dry matter	%	89.47	89.47	89.78	87.83
Moisture	%	10.53	10.53	10.22	12.17
Crude protein	%	20.90	20.90	20.90	18.70
Crude fat	%	4.74	4.74	4.95	3.20
Ash	%	5.54	5.54	5.47	4.61
NDF	%	9.59	9.59	9.59	11.37
ADF	%	3.32	3.32	3.16	3.70
Calcium	%	0.55	0.55	0.55	0.55
Phosphorous	%	0.52	0.52	0.53	0.47
Net energy	kcal/kg	2550.00	2550.00	2550.00	2375.00
Standard ileal digestible lysine	%	1.40	1.40	1.40	1.25
Analyzed composition (as is basis)
Dry matter	%	88.95	89.08	89.26	86.96
Moisture	%	11.05	10.92	10.74	13.04
Crude protein	%	21.69	22.15	22.26	18.71
Crude fat	%	5.20	5.03	5.12	3.5
Ash	%	5.75	6.06	5.84	4.49
Calcium	%	0.51	0.48	0.50	0.51
Phosphorous P	%	0.40	0.39	0.38	0.38
Total lysine	%	1.51	1.51	1.49
Titanium dioxide	mg/kg	2870	2700	2780

¹Titanium Dioxide was used as a marker for digestibility measurement before the European Food Safety Authority banned its use as a food additive in March 2022.

²The vitamin and mineral premix provided the following per kilogram of diet: 16,000 IU vitamin A as vitamin A; 2000 IU vitamin D3, 10 mg vitamin B1; 5 mg vitamin B2; 5 mg pyridoxine; 25 mg niacin as Niacinamide; 0.05 mg vitamin B12; 3.1 mg iodine; 0.37 mg Se as sodium selenite.

³Phytase: Natuphos 10000 (contains 10,000 FTU/g phytase) was used. Final diet phytase level was 394 FTU/kg.

⁴Organic acid blend is composed of different organic acids, ie Citric acid.

SBM: soybean meal; HTM SBM: hydro-thermal-mechanical soybean meal; Enz Trt SBM: enzyme-treated soybean meal; NDF: neutral detergent fiber; ADF: acid detergent fiber.

Table 2.

Analyzed chemical composition of the soy protein ingredients used in the experiment (as fed basis).¹

	SBM	HTM SBM2	Enz Trt SBM3
Item (%)
Moisture	12.5	6.9	7.8
CP	49.3	52.4	56.8
Fat	1.7	1.7	3.0
Ash	6.2	6.2	7.0
Crude fiber	3.5	3.5	4.0
NDF	7.1	7.0	9.2
ADF	3.5	3.4	5.8

¹Analyzed in duplicate.

²HTM SBM was produced using the same batch of SBM in this experiment.

³Values for EnZ Trt SBM for moisture, CP and ash were lab-analyzed and the rest were adapted from the product specification sheet for HP300 from HalmetProtein.com.

CP: crude protein; NDF: neutral detergent fiber; ADF: acid detergent fiber.

SBM: soybean meal; HTM SBM: hydro-thermal-mechanical soybean meal; Enz Trt SBM: enzyme-treated soybean meal.

Growth performance and fecal score

The initial BW of individual pigs was measured at the start of the study. Subsequent pen weights and feed disappearance were recorded at d 7, 12, 21, and 39 post-weaning. The ADG, ADFI and G: F were calculated on a per-pen basis. Fecal scores were recorded by the same person on d 5, 6, 8, 12, 14, 16, 18 and 21 post-weaning. Fecal scores were based on a scale of 1 to 4 according to (Faris and Hackenhaar 2023): 1) very hard and dry stool, 2) firm to normal stool, 3) loose stool, and 4) watery stool with no shape.

Sample collection

One inoculated pig per pen was randomly selected and euthanized by intravenous injection of Euthasol^® Euthanasia ­Solution (pentobarbital sodium and phenytoin sodium) per Veterinarian followed by exsanguination to collect ileal and colon digesta and rectal feces on d 14. Digesta samples were collected in a 150 mL cup, homogenized and divided over two 20 mL cups and two 2 mL Eppendorf tubes. Rectal feces were aseptically collected from each individual pig. All samples were then kept on dry ice until storage in −80°C freezer for further analyses.

Apparent ileal, total tract protein digestibility, and short-chain fatty acids

A total of 100 g of feed and 2 g of freeze dried ileal digesta and fecal samples were used for analysis of titanium level using ICP—OES method (protocol developed by NutriControl BV, Netherlands). Feed, ileal digesta, and fecal samples were first grinded, and 2 g were sub-sampled to use for CP analysis by the combustion procedure [method 990.03, (AOAC 2019)]. Crude protein digestibility was expressed as % CP digested in the ileal or total tract after correcting using the titanium concentration in feed, ileal digesta, or feces. The apparent ileal digestibility of CP was calculated using formula previously reported by Adeola (Adeola et al. 2001): AID of CP, % = [1−[(TiO_2diet/TiO_2digesta) × (CP_digesta/CP_diet)]] × 100; the apparent total tract digestibility (ATTD of CP) of protein was calculated with the formula previously reported by Adeola (Adeola et al. 2001): ATTD of CP, % = [1−[(TiO_2diet/TiO_2feces) × (CP_feces/CP_diet)]] × 100, in which TiO_2diet, TiO_2digesta and TiO_2feces represented the measured titanium dioxide concentration of diet, ileal digesta and feces, respectively; CP_diet, CP_digeseta and CP_feces represented the measured CP concentration of the diet, ileal digesta and feces, respectively.

The SCFA including acetate, propionate, butyrate, isobutyrate, valerate, isovalerate, and caproate were analyzed in 20 mL colon digesta at the lab facility of the Gent University in Belgium, using gas-chromatography equipped with a Nukol ­column (30 m × 0.25 mm × 0.25 m, Supelco) with a flame ionization detector (Shimadzu 2010, Shimadzu Corporation, ‘s-Hertogenbosch, The Netherlands). The values were expressed as µmol/g DM. Fecal and colon digesta samples were freeze-dried and weighed before and after drying to obtain DM content.

Microbiome evaluation

DNA extraction and shotgun metagenomics sequencing.

Frozen fecal samples (n = 1 pig/pen × 12 pen/treatment × 6 treatments × 1 timepoint = 72, four samples were missing due to lack of enough feces at the time of collection in pigs housed in the regular nursery condition) were thawed overnight at 4°C and used for genomic extraction using ZymoBIOMICS 96 MagBead DNA kit (Zymo Research Corporation, Irvine, CA) that included a bead-beating step for mechanical lysis of bacterial cells. Extraction was performed in a Biomek i7 workstation (Beckman Coulter, Indianapolis, IN) according to manufacturer’s instructions. Four extraction blanks were included in each 96 well plate to confirm that cross contamination did not occur. Libraries were constructed using the SQK-RPB004 Rapid PCR Barcoding kit (ONT, Oxford, U.K.). Library preparation included DNA extraction blanks for quality control. Shotgun metagenomic sequencing was performed using R9.4.1 FLO-MIN 106 flow cells on the GridION platform (ONT, Oxford, U.K.), multiplexing 12 samples in each flow cell. Each sequencing run lasted 72 h. The MinKNOW ONT software (v 4.5.4) with Guppy basecaller (v 5.1.13) was used for sequencing using the high-accuracy basecalling setting, followed by de-multiplexing, adapter trimming, and quality control using default settings. The sequence data can be found at: https://www.ncbi.nlm.nih.gov/, bioproject # PRJNA128711.

Taxonomic assignment and diversity metrics.

Fastq files obtained from the MinKNOW ONT workflow were used for microbial taxonomic classification. First, host DNA was removed by mapping fastq files to the Sus domesticus genome (ARS-UCD1.2) using Minimap2 (Li 2018), followed by the removal of reads matching the host genome using SAMtools (Li et al. 2009). The remaining reads were assumed to be from microbial DNA and used for taxonomic assignment. Taxonomic assignation was performed using Kraken2 (Wood et al. 2019) with a custom database containing high quality genomes from the RefSeq database (Tatusova et al. 2015) and published metagenome-assembled genomes (Almeida et al. 2021; Chen et al. 2021) classified according to Genome Taxonomy Database (GTBD; (Chaumeil et al. 2019). Estimation of abundance at the species and phylum levels was performed in Bracken (Lu et al. 2017).

Statistical analyses

Performance parameters, protein digestibility, and short-chain fatty acids

Data were subject to a statistical outlier analysis. Data beyond 1.5 × the IQR (Inter-Quartile Mean) as determined by Method 8 of (Hyndman and Fan 1996), were removed from the dataset before the statistical analysis.

Data from six treatments in a three dietary treatment by two challenge conditions (regular nursery or ETEC challenge) factorial arrangement of treatments were subject to Levene’s test to confirm assumption of homoscedasticity of variance. Subsequently data were analyzed using R (version 4.0.2) and the lme4 package, to deploy a General Linear Mixed Model (GLM), appropriate for a Split-Plot experimental design. Challenge condition was applied to the main plot and dietary treatment was applied to the sub-plot. Appropriate error terms were defined in the statistical model to account for differently sized experimental units according to lme4 package (Bates et al. 2015). Initial BW (light and heavy) and sex (gilts and barrows) were used as blocking factors. Dietary treatment and challenge conditions and their interaction were defined as fixed effects, and blocking factors were considered as random effects. The emmeans package was used to calculate estimated marginal mean values for each treatment by challenge condition. Pairwise comparison was used to assess the effect among dietary treatments for ileal or total tract digestibility of the dietary CP. Interaction between dietary treatment and challenge conditions was assessed and LSmeans were provided for each dietary treatment by challenge condition interaction level. In all analyses, the differences were considered statistically significant if P < 0.05, highly statistically significant if P < 0.01, and a tendency when P < 0.10. Subscripts were used to denote pairwise comparison outcome when the interaction was detected.

Stool quality by fecal scoring

Likert based fecal scoring data were collected over multiple days and analyzed using an ordinal Generalized Linear Mixed Model (GLMM) in R using the package ordinal. As described above for the GLM, appropriate design structure and experimental units were specified using a cumulative logit linkage applied by the ordinal package clmm function except that day of scoring was also included as an additional random effect. Data were reported as inverse linked probabilities associated with better (improved) stool quality for the ordered range of Likert scores observed. Similar post hoc means comparisons were made as described for the performance parameters.

Microbiome data quality control

Prior to any microbiome analysis, rare microbial taxa, with relative abundance less than 1e-5 were removed from the dataset, decreasing the dataset from 1341 to 1140 microbial species. Filtered taxa abundances were center-log ratio (CLR) transformed using the microbiome package (Lahti 2023) in R ­(version 4.1.2), after imputation of zeros using a Bayesian ­multiplicative replacement method from the zCompositions package (Palarea-Albaladejo and Martín-Fernández 2015).

Diversity of microbial taxa

Within sample (alpha) diversity of microbial taxa was calculated in R (2015) using the Phyloseq package (McMurdie and Holmes 2013; Seemann 2014) with the rarefied taxa count table from Bracken as input. Statistical analysis for diversity metrics was performed with the lme4 package (Bates et al. 2015) from R (version 4.1.2), using the linear mixed model function lmer (de Boeck et al. 2011) with model as specified above for the GLM. Pairwise comparisons were completed using the emmeans package (Lenth et al. 2022) in R. The Benjamini-Hochberg (false-discover rate; FDR) method was used for P-value adjustment to account for multiplicity (Benjamini and Hochberg 1995; Benjamini and Yekutieli 2001).

Between sample (beta) diversity of microbial taxa was tested with permutational ANOVA (PERMANOVA) using the adonis function in the vegan package (Anderson 2001) in R. A dissimilarity matrix between samples was computed using the Aitchison distance (Aitchison et al. 2000) which is the Euclidean distance between CLR-transformed data points. This matrix was used as the input for the PERMANOVA. The statistical model for beta diversity was modified from the alpha diversity model by removing random effects, as they cannot be included in the permutational test.

Data were visualized in two dimensions using the first two principal components, computed using principal component analysis implemented through the Phyloseq package ­(McMurdie and Holmes 2013; Seemann 2014). Points and density ellipses were colored based on treatment or challenge to visually assess for distinctions in grouping.

Differential abundance of microbial taxa

A modified version of the differential abundance function from the LinDA package (Zhou et al. 2022) was used to fit a linear mixed model using the same formula as described for diversity to assess for between-treatment differences. Data were CLR-transformed prior to analysis, and the response for each model was the CLR abundance of each taxon. The fitted model for each taxon was then analyzed with the emmeans package (Lenth et al. 2022) to conduct post-hoc comparisons between treatments. Comparison results are presented as differences in mean CLR abundance between dietary treatment groups or challenge conditions, these can be interpreted as the log-fold change (difference of logs = log of the ratio). The FDR correction was used to adjust P-values to account for multiple testing, and taxa that differed significantly between treatments were identified using the adjusted P-value.

Correlation between performance parameters and short-chain fatty acids with microbial taxa

To derive biological insights from the microbial taxa identified through differential abundance analysis in relation to the most significant performance parameters and short-chain fatty acids, Pearson correlation coefficients and associated P-values were calculated using R. Correlations were performed across the entire dataset, regular nursery only, and ETEC challenge only. Correlations were assessed with final BW, ADG, ADFI, acetate (µmol/g DM), propionate (µmol/g DM), and caproate (µmol/g DM). Correlations across the entire dataset were used to understand the effect microbial taxa with effect sizes higher than 5-fold change across challenge conditions within dietary treatments. Correlations within challenge conditions (regular nursery condition or ETEC challenge) were used to evaluate the effect of microbial taxa with high effect sizes across dietary treatment within challenge conditions.

Results

Growth performance and stool quality

From d 0 to 7 post-weaning, no interaction between dietary treatment and challenge conditions was observed on ADG, ADFI or BW (Table 3). Pigs in regular nursery conditions had higher ADG (P < 0.01) and ADFI (P < 0.05) than the ETEC challenged pigs, confirming the challenge model was effective. Pigs fed HTM SBM or Enz Trt SBM showed greater ADG (P < 0.05), but similar ADFI compared to those fed SBM, regardless of ETEC challenge. An interaction for G: F (P < 0.05) was identified and indicated that HTM SBM and Enz Trt SBM improved G: F under regular conditions, but not during ETEC challenge. During this period, pigs housed in the regular nursery condition also exhibited higher ADG than ADFI and resulted in a G: F greater than 1 which was not observed in the pigs housed under ETEC challenge. From d 7 to 12, no interaction was detected on growth performance parameters. Pigs fed HTM SBM or Enz Trt SBM had higher ADG (P < 0.05) and ADFI (P < 0.01) than those received SBM regardless of ETEC challenge. Pigs under ETEC challenge had lower ADG (P < 0.05) and ADFI (P < 0.01) without affecting G: F. From d 12 to 21 post-weaning, pigs went through post ETEC challenge recovery. No interaction was observed on growth performance parameters. BW remained lower (P < 0.05) in pigs fed SBM. ETEC challenge reduced BW (P < 0.01) but increased G: F (P < 0.01). From d 21 to 39 post-weaning, no interaction was detected. ETEC challenge decreased ADG (P < 0.01) and final BW (P < 0.01) without affecting ADFI or G: F. Across d 0 to 39 post-weaning, no dietary treatment × ETEC challenge interaction was detected. ETEC challenge lowered ADFI (P < 0.01) and ADG (P < 0.05), while G: F remained unaffected. The mortality rate in regular nursery condition was 0.6% and 2.7% under ETEC challenge.

Table 3.

Effect of processed SBM and challenge condition on growth performance and mortality of pigs during 39 d post-weaning.

	Regular nursery			ETEC challenge1				P-value
Item	SBM	HTM SBM	Enz Trt SBM	SBM	HTM SBM	Enz Trt SBM	SEM	Challenge	Diet	Challenge × Diet
d 0–7
BW, kg	8.6	8.9	8.9	8.0	8.2	8.1	0.5	0.006	0.005	0.790
ADG, g/d	262	308	290	168	193	187	13.4	<.0001	0.013	0.678
ADFI, g/d	242	259	241	205	217	211	12.6	0.032	0.288	0.816
2 G: F, g/g	1.1b	1.2c	1.2c	0.9a	0.9a	0.9a	0.0	0.003	0.028	0.026
d 7–12
BW, kg	10.8	11.5	11.5	9.0	9.7	9.4	0.5	0.026	0.000	0.624
ADG, g/d	436	518	526	191	285	246	59.1	0.019	0.000	0.515
ADFI, g/d	509	559	564	309	345	317	41.4	0.007	0.021	0.303
G: F, g/g	0.9	0.9	0.9	0.6	0.8	0.7	0.1	0.16	0.012	0.268
d 12–21
BW, kg	14.8	15.8	15.7	13.9	14.5	14.6	0.8	<.0001	0.015	0.831
ADG, g/d	462	486	484	520	546	573	45.0	0.205	0.281	0.771
ADFI, g/d	533	571	556	538	557	566	39.9	0.983	0.324	0.822
G: F, g/g	0.9	0.9	0.9	1.0	1.0	1.0	0.0	0.007	0.581	0.861
d 21–39
BW, kg	28.6	29.7	29.6	27.0	27.8	27.5	1.2	<.0001	0.197	0.893
ADG, g/d	811	818	816	771	779	757	30.8	0.005	0.825	0.859
ADFI, g/d	1215	1233	1250	1184	1191	1189	37.6	0.190	0.550	0.709
G: F, g/g	0.7	0.7	0.7	0.6	0.7	0.6	0.0	0.260	0.719	0.826
d 0–39
ADG, g/d	560	590	584	487	506	501	19.1	0.011	0.166	0.911
ADFI, g/d	755	777	781	667	671	679	21.4	0.006	0.309	0.750
G: F, g/g	0.7	0.8	0.8	0.7	0.7	0.8	0.0	0.195	0.310	0.615
Mortality3	1	0	0	1	1	1

¹ETEC challenge: enterotoxigenic Escherichia coli K88.

²G: F: Superscripts with different letters differ (P < 0.05).

³Mortality: pig death count per treatment.

SBM: soybean meal; HTM SBM: hydro-thermal-mechanical soybean meal; Enz Trt SBM: enzyme-treated soybean meal.

Stool quality analyzed as the probability of better stool using fecal scoring data was not impacted by dietary intervention or ETEC challenge (data not shown). There was no interaction between ETEC challenge and dietary intervention on colon or fecal DM content. Colon DM content tended to be higher (P < 0.10) in HTM SBM and EnZ Trt SBM compared to SBM treatment. But dietary intervention did not impact fecal DM. Overall, fecal DM content was lower (P < 0.01) and colon DM tended to be lower (P < 0.10) in pig under ETEC challenge compared to those housed in the regular nursery condition. However, dietary intervention did not change fecal DM content regardless of ETEC challenge (Table 4).

Table 4.

Effect of processed SBM and challenge condition on CP digestibility and colon digesta short-chain fatty acid concentration on d14 post-weaning.

	Regular nursery			ETEC challenge
Item	SBM	HTM SBM	Enz Trt SBM	SBM	HTM SBM	Enz Trt SBM	SEM	Challenge	Diet	Challenge × Diet
Colon DM (%)	18.2	20.3	19.0	14.5	16.2	17.1	1.3	0.075	0.057	0.377
Fecal DM (%)	26.2	28.1	27.1	22.1	21.9	23.1	1.2	<.0001	0.646	0.538
1 AID CP (%)	67.1	75.8	71.3	55.5	60.4	63.1	5.6	0.109	0.131	0.617
2 ATTD CP (%)	74.9	81.3	78.0	72.7	75.3	75.7	2.4	0.272	0.003	0.223
Acetate (µmol/g DM)	457.8	398.6	433.4	493.1	420.3	379.8	42.2	0.981	0.031	0.257
Propionate (µmol/g DM)	209.4	168.9	186.4	228.1	188.7	186.7	25.4	0.692	0.018	0.751
Isobutyrate (µmol/g DM)	8.7	7.8	7.4	8.2	7.6	7.5	1.1	0.855	0.443	0.920
Butyrate (µmol/g DM)	133.8	117.8	148.8	121.4	115.5	114.9	16.3	0.434	0.321	0.306
Isovalerate (µmol/g DM)	13.1	11.0	10.1	11.5	9.9	10.6	1.7	0.688	0.271	0.700
Valerate (µmol/g DM)	27.1	34.0	35.9	21.4	21.7	21.0	3.8	0.018	0.396	0.370
Caproate (µmol/g DM)	1.9	4.6	4.2	1.3	2.3	1.2	0.6	0.010	0.004	0.065
Total short-chain fatty acid (µmol/g DM)	851.1	746.3	821.3	887.9	778.4	731.9	84.3	0.948	0.076	0.370

¹AID CP: apparent ileal digestibility of crude protein.

²ATTD CP: apparent total tract digestibility of crude protein.

SBM: soybean meal; HTM SBM: hydro-thermal-mechanical soybean meal; Enz Trt SBM: enzyme-treated soybean meal.

Apparent ileal and total tract digestibility of CP

No interaction effect between dietary intervention and challenge condition was found on AID or ATTD CP in the feeds tested in this experiment on d 14 post-weaning. Neither dietary intervention nor challenge condition affected AID CP. Dietary intervention, however, impacted ATTD CP and pigs fed HTM SBM and Enz Trt SBM diet had higher ATTD CP (P < 0.01) compared to those fed SBM regardless ETEC challenge (Table 4).

Colon DM content and short-chain fatty acid

No interaction effect between dietary intervention and challenge condition was found on colon DM content (Table 4). Dietary intervention (P < 0.10) or ETEC challenge (P < 0.10) tended to impact colon DM content, respectively. Total short-chain fatty acids or individual short-chain fatty acids were not impacted by the interaction between dietary intervention and challenge conditions. Although ETEC challenge did not have an impact, dietary intervention tended to affect total short-chain fatty acids on a DM basis (P < 0.10). Acetate (P < 0.05) and propionate (P < 0.05) content in colon digesta collected on d 14 post-weaning were both greater in SBM ­compared to HTM SBM fed pigs. While acetate content did not differ among dietary treatments, propionate content was greater in pigs fed SBM compared to HTM SBM (P < 0.05). Caproate content was greater in pigs fed HTM SBM compared to SBM (P < 0.01) but not different than those fed Enz Trt SBM regardless of ETEC challenge. ETEC challenge also lowered caproate content (P < 0.05) as well as valerate content (P < 0.05) regardless of dietary intervention.

Swine fecal microbiome

A total of 68 samples were processed for sequencing, with two samples failing the sequencing process. On average, 257,789 reads were obtained per sample (mean: 257,789; median: 248,709, SD: 81,651). The N50 read lengths averaged 3672 base pairs (mean: 3672; median: 3315; SD: 1219). Each sample yielded a total of 0.84 gigabases (mean: 0.84 Gb; median: 0.85 Gb; SD: 0.25 Gb). On average, 82.6% of the reads had taxonomic classification (mean: 82.6; median: 83.8, SD: 5.1).

Under regular nursery conditions, alpha diversity (as measured by the Shannon index) tended to be higher in pigs fed HTM SBM and Enz Trt SBM compared to SBM (P = 0.1 for both comparisons). No difference in alpha diversity was observed with ETEC challenge across treatments (Fig. 1A; Table 5). Within each treatment, SBM tended to have higher diversity under ETEC challenge compared to regular nursery conditions (P = 0.07). No differences in alpha diversity were observed within the HTM SBM and Enz Trt SBM fed pigs under ETEC challenge as compared to regular nursery conditions (Fig. 1B; Table 6).

Fig. 1.

Alpha (within sample) diversity of fecal microbiome composition in nursery pigs across dietary treatments and challenge conditions. Comparison of Shannon index of diversity at the species level (A) between treatments within challenge condition. (B) within treatments across challenge conditions. Data visualized using Box-Whisker plots, where box denotes interquartile range (distance between the 25^th and 75^th percentile) with a line at the median, whiskers indicate minimal and maximal values no further than 1.5 times of the interquartile range, and data points beyond the end of the whiskers correspond to potential outlying points. A linear mixed model was fitted with treatment, challenge, and their interaction as fixed effects, and the effect of block nested within the treatment as a random factor.⁺P ≤ 0.1.

Table 5.

Effect of processed SBM within challenge condition on microbial alpha diversity at the species level using the Shannon index.

Treatment contrast	Challenge	Estimate	SE	P-value
SBM—HTM SBM	Regular nursery	−0.498	0.190	0.104
SBM—Enz Trt SBM	Regular nursery	−0.431	0.196	0.104
HTM SBM—Enz Trt SBM	Regular nursery	0.067	0.189	0.733
SBM—HTM SBM	ETEC challenge	0.190	0.184	0.715
SBM—Enz Trt SBM	ETEC challenge	0.043	0.194	0.833
HTM SBM—Enz Trt SBM	ETEC challenge	−0.147	0.191	0.715

SBM: soybean meal; HTM SBM: hydro-thermal-mechanical soybean meal; Enz Trt SBM: enzyme-treated soybean meal.

Table 6.

Effect of challenge condition across different soy protein sources on microbial alpha diversity at the species level using the Shannon index.

Challenge contrast	Treatment	Estimate	SE	P-value
Regular nursery—ETEC challenge	SBM	−0.434	0.197	0.070
Regular nursery—ETEC challenge	HTM SBM	0.254	0.183	0.218
Regular nursery—ETEC challenge	Enz Trt SBM	0.039	0.197	0.849

SBM: soybean meal; HTM SBM: hydro-thermal-mechanical soybean meal; Enz Trt SBM: enzyme-treated soybean meal.

A significant effect of dietary intervention and challenge conditions on the overall microbiome composition (beta diversity) was observed. While PC1 and PC2 accounted for only 12.8% and 5.4% of the total variation, respectively, with no clear clustering pattern (Fig. 2), PERMANOVA based on Aitchison distances revealed significant effects of both dietary treatment and challenge conditions, with no significant interaction between them (Treatment: P < 0.01, R² = 0.04; Challenge: P < 0.01, R² = 0.06; Treatment × Challenge: P = 0.81, R² = 0.02).

Fig. 2.

Beta (between sample) diversity of fecal microbiome composition in nursery pigs across dietary treatments and challenge conditions. Principal component analysis was used to visualize dissimilarity among dietary treatments based on Aitchison distance matrices. Points and density ellipses were colored based on treatment and challenge conditions to visually assess for distinctions in grouping. (A) Samples separated by challenge and colored by treatment, (B) samples separated by treatment and colored by challenge. Aitchison distances were analyzed with permutational ANOVA (PERMANOVA) model which included treatment, challenge, and their interactions.

While dietary treatments did not significantly affect taxa at the species and genus level (P > 0.1), several taxa exhibited more than 5-fold differences in abundance across treatments and challenge conditions. We evaluated the correlations of these taxa to performance parameters (BW, ADG, and ADFI) as well as with SCFA including acetate, propionate, and caproate (Table S1) which were significantly influenced by dietary ­treatment. Under the regular nursery conditions, 14 species (Phascolarctobacterium_A sp900553055, Lactobacillus absiana, Lactobacillus crispatus, Prevotella sp004556065, SFRY01 sp004562065, Prevotella sp002251435, Dialister sp000434475, Methanosphaera sp022768985, Prevotella sp905198745.

Streptococcus hyointestinalis, UMGS172 sp900539855, Prevotella GENOME088748, Collinsella sp002391315 and Eubacterium_T pyruvativorans) with large effect size differences were identified between pigs fed SBM and those fed HTM SBM, among these, 11 species showed statistically significant correlations with the performance and SCFA measures (P < 0.05) (Fig. 3A). Pigs fed SBM had a higher abundance of species negatively correlated with ADG, caproate, and valerate, notably Prevotella sp002251435 showing a strong negative correlation (P < 0.01; R² = −0.62) with ADG. In contrast, pigs fed HTM SBM exhibited a higher abundance of species with significant (P < 0.05) positive correlations with performance and SCFA; for example, Eubacterium_T pyruvativorans, demonstrated a significant (P < 0.05) and strong positive correlation with caproate (R² = 0.63), valerate (R² = 0.48), and ADG (R² = 0.33) (Fig. 3A). Eight species (Streptococcus hyointestinalis, Mitsuokella jalaludinii, Faecalibacterium prausnitzii_D, Cryptobacteroides sp000431015, Methanosphaera sp022768985, Lactobacillus crispatus, Lactobacillus absiana and SFRY01 sp004562065) with large effect size differences were identified between the SBM versus Enz Trt SBM fed pigs, all of which had statistically significant correlations with the performance and FA measures (Fig. 3B). Pigs fed SBM showed higher abundance of species that had significant (P < 0.05) negative correlations with BW, ADG, and caproate, while pigs fed Enz Trt SBM showed a higher abundance of species with significant (P < 0.05) positive correlations with BW, ADG, ADFI, caproate, and valerate (Fig. 3B). One species (CAG-510 sp000434615) with a large effect size difference was identified between the HTM SBM versus Enz Trt SBM fed pigs (Fig. 3C). However, it was not statistically significantly correlated with the performance or FA measures (Fig. 3C).

Fig. 3.

Correlations between phenotypic outputs and species with large fold differences across dietary treatment comparisons under regular nursery conditions. Correlations between species with fold change higher than |5| across treatment comparisons and performance parameters significantly affected by dietary treatment (BW, ADG, ADFI) and short-chain fatty acids (acetate, propionate, caproate, valerate). (A) Left: Lollipop plot showing species with high effect sizes in the SBM versus HTM SBM comparison under regular nursery condition, 14 species. Positive CLR difference indicates higher abundance in SBM; negative indicates higher abundance in HTM SBM. Right: Pearson correlation of species shown in the lollipop plot. (B) Left: Lollipop plot showing species with high effect sizes in the SBM versus HTM SBM comparison under ETEC challenge, 5 species. Positive CLR difference indicates higher abundance in SBM; negative indicates higher abundance in HTM SBM. Right: Pearson correlation of species shown in the lollipop plot. For all figures, color above 0 represents a postive correlative; color below 0 represents a negative correlation.Stars indicate the level of significance: * = 0.05, ** = 0.01, *** = 0.0001. Abbreviations: Acet—acetate, prop—propionate, capro—caproate, Vale—valerate.

Under ETEC challenge conditions, five species (Prevotella sp945869265, CAG-1000 sp000436975, Prevotella sp004556065, Prevotella GENOME088748 and Dysosmobacter sp900544615) with large effect size differences were identified between pigs fed SBM and those fed HTM SBM; four of these species were significantly correlated with performance or SCFA measures (Fig. 4A). Pigs fed SBM exhibited a higher abundance of species showing significant (P < 0.05) negative correlation with BW, ADG, ADFI, and caproate, alongside significant (P < 0.05) positive correlations with acetate and propionate. Among these, CAG-1000 sp000434555 from the class Bacilli displayed strong negative correlations with BW (R² = −0.59), ADFI (R² = −0.43), and ADG (R² = −0.40). Additionally, Prevotella sp004556065 was positively correlated with acetate (R² = 0.46), and propionate (R²=0.43), but negatively correlated with caproate (R²=−0.53), ADG (R² = −0.38), ADFI (R² = −0.36), and BW (R² = −0.34) (Fig. 4A). One species (CAG-605 sp000433255) with a large effect size difference was identified between the SBM versus Enz Trt SBM fed pigs but no significant correlation with the performance or FA measures were seen (Fig. 4B). Two species (UBA3789 sp900543055 and CAG-605 sp000433255) with a large effect size difference were identified between the HTM SBM versus Enz Trt SBM fed pigs. However, neither of them had significant correlations with the performance or FA measures (Fig. 4C).

Fig. 4.

Correlations between phenotype and species with large fold differences across dietary treatment comparisons under ETEC challenge. ­Correlations between species with fold change higher than |5| across treatment comparisons and performance parameters significantly affected by dietary treatment (BW, ADG, ADFI) and short-chain fatty acids (acetate, propionate, caproate, valerate). (A) Left: Lollipop plot showing species with high effect sizes in the SBM versus HTM SBM comparison, 5 species. Positive CLR difference indicates higher abundance in SBM; negative indicates higher abundance in HTM SBM. Right: Pearson correlation of species shown in the lollipop plot. (B) Left: Lollipop plot showing species with high effect sizes in the SBM versus Enz Trt SBM comparison, 1 species. Positive CLR difference indicates higher abundance in SBM; negative indicates higher abundance in Enz Trt SBM. Right: Pearson correlation of species shown in the lollipop plot. (C) Left: Lollipop plot showing species with high effect sizes in the HTM SBM versus Enz Trt SBM comparison, two species. Positive CLR difference indicates higher abundance in HTM SBM; negative indicates higher abundance in Enz Trt SBM. Right: Pearson correlation of species shown in the lollipop plot. For all figures, color above 0 represents a positive correlation; color below 0 represents a negative correlation. Stars indicate the level of significance: * = 0.05, ** = 0.01, *** = 0.0001. Abbreviations: Acet—acetate, prop—propionate, capro—caproate, Vale—valerate.

Comparing challenge conditions within each dietary treatment group, 10 species (Prevotella sp002299635, CAG-791 sp000431495, Paratractidigestivibacter faecalis, Acidaminococcus fermentans, Prevotella hominis, CAG-1000 sp000436975, Frisingicoccus sp016298055, Onthomorpha sp004551865, UBA11774 sp900556645 and Parabacteroides sp004562445) with large effect size differences were identified in pigs fed SBM, and 22 species (Onthomorpha sp004551865, CAG-245 sp900758165, Blautia_A sp900551715, Egerieousia sp004561775, Roseburia inulinivorans, CAG-349 sp003539515, Prevotella sp002299635, V9D3004 sp002349525, Acidaminococcus fermentans, CAG-492 sp000434335, Dialister sp000434475, Vescimonas sp902388735, CAG-791 sp000431495, Eubacterium_T pyruvativorans, Collinsella sp002391315, Paratractidigestivibacter faecalis, CAG-605 sp000433475, Prevotella sp900771975, Prevotella sp900556795, CAG-269 sp900551615, Catenibacterium mitsuokai and Methanosphaera sp022768985) in pigs fed HTM SBM and 28 species (Prevotella sp002299635, Dialister sp000434475, CAG-791 sp000431495, Acidaminococcus fermentans, Paratractidigestivibacter faecalis, Clostridium sp000435835, Streptococcus hyointestinalis, CAG-245 sp000435175, Methanosphaera sp022768985, Mitsuokella jalaludinii, Megasphaera elsdenii, Ruminococcus_E bovis, Porcincola intestinalis, Prevotella sp900771975, CAG-492 sp000434335, Faecalibacterium prausnitzii_D, Faecalibacterium duncaniae, Prevotella sp900556795, Mitsuokella multacida, Choladousia sp902363135, Prevotella sp002251435, CAG-1000 sp000436975, CAG-510 sp000434615, CAG-533 sp000434495, CAG-605 sp000433255, CAG-353 sp945867875, CAG-349 sp003539515 and CAG-302 sp934101345) in the Enz Trt SBM fed pigs (Fig. 5). Among SBM-fed pigs, five species (CAG-1000 sp000436975, Frisingicoccus sp016298055, Onthomorpha sp004551865, UBA11774 sp900556645 and Parabacteroides sp004562445) were more abundant under the ETEC challenge condition, and five (Prevotella sp002299635, CAG-791 sp000431495, Paratractidigestivibacter faecalis, Acidaminococcus fermentans and Prevotella hominis) under regular nursery condition; all were significantly (P < 0.05) correlated with performance or SCFA measures. Species with higher abundance in SBM-fed pigs under ETEC challenge showed negative correlations with BW, ADG, ADFI, caproate, and valerate, while one species displayed a positive correlation with propionate. Notably, UBA11774 sp900556645 from the class Clostridia exhibited strong negative correlations with ADG (R² = −0.76), ADFI (R² = −0.73), BW (R² = −0.44), caproate (R² = −0.43), and ­valerate (R² = −0.35), along with a positive correlation with ­propionate (R² = 0.30) (Fig. 5A).

Fig. 5.

Correlations between phenotypic outputs and species with large fold differences within dietary treatments across challenge conditions. Correlations between species with fold change higher than |5| in the ETEC challenge versus regular nursery comparison within treatments and performance parameters significantly affected by dietary treatment (BW, ADG, ADFI) and short-chain fatty acids (acetate, propionate, caproate, valerate). (A) SBM ETEC challenge versus SBM regular nursery comparison, 10 species. (B) HTM SBM ETEC challenge versus HTM SBM regular nursery comparison, 22 species. (C) For all figures, Enz Trt SBM challenge versus Enz Trt SBM regular nursery comparison, 28 species. Left: Lollipop plot showing species with high effect sizes. Positive CLR difference indicates higher abundance in ETEC challenge; negative indicates higher abundance in regular nursery. Right: Pearson correlation, color above 0 represents a postive correlative; color below 0 represents a negative correlation. Stars indicate the level of significance: * = 0.05, ** = 0.01, *** = 0.0001. Abbreviations: SBM—soybean meal, HTM SBM—hydro-thermal-mechanical soybean meal, Enz Trt SBM—enzyme-treated soybean meal, ETEC—enterotoxigenic Escherichia coli, acet—acetate, prop—propionate, capro—caproate, Vale—valerate.

For the pigs fed HTM SBM, six species (Onthomorpha sp004551865, CAG-245 sp900758165, Blautia_A sp900551715, Egerieousia sp004561775, Roseburia inulinivorans and CAG-349 sp003539515) were more abundant under the ETEC challenge condition, while 16 species (Prevotella sp002299635, V9D3004 sp002349525, Acidaminococcus fermentans, CAG-492 sp000434335, Dialister sp000434475, Vescimonas sp902388735, CAG-791 sp000431495, Eubacterium_T pyruvativorans, Collinsella sp002391315, Paratractidigestivibacter faecalis, CAG-605 sp000433475, Prevotella sp900771975, Prevotella sp900556795, CAG-269 sp900551615, Catenibacterium mitsuokai and Methanosphaera sp022768985) had higher abundance under regular nursery condition. Of these, 21 species showed significant (P < 0.05) correlations with performance or SCFA measures (Fig. 5B). For the pigs fed Enz Trt SBM, eight species (Prevotella sp002251435, CAG-1000 sp000436975, CAG-510 sp000434615, CAG-533 sp000434495, CAG-605 sp000433255, CAG-353 sp945867875, CAG-349 sp003539515 and CAG-302 sp934101345) were identified to have higher abundance under the ETEC challenge condition, while 20 (Prevotella sp002299635, Dialister sp000434475, CAG-791 sp000431495, Acidaminococcus fermentans, Paratractidigestivibacter faecalis, Clostridium sp000435835, Streptococcus hyointestinalis, CAG-245 sp000435175, Methanosphaera sp022768985, Mitsuokella jalaludinii, Megasphaera elsdenii, Ruminococcus_E bovis, Porcincola intestinalis, Prevotella sp900771975, CAG-492 sp000434335, Faecalibacterium prausnitzii_D, Faecalibacterium duncaniae, Prevotella sp900556795, Mitsuokella multacida and Choladousia sp902363135) had higher abundance in the regular nursery conditions (Fig. 5C). Notably, a common finding among all dietary treatment comparisons was that Prevotella sp002299635 had the highest abundance under regular nursery conditions compared to the ETEC challenge. This species showed significant (P < 0.05) positive correlations with ADFI (R² = 0.74), ADG (R² = 0.68), BW (R² = 0.62), caproate (R² = 0.55), and valerate (R² = 0.44).

Discussion

This study investigated the effects of replacing SBM entirely with HTM-processed SBM or enzyme-treated SBM (Enz Trt SBM) on growth performance, CP digestibility, SCFA production, and the fecal microbiome of nursery pigs raised under either regular (conventional) nursery conditions or challenged with ETEC. Previous studies have demonstrated that processing techniques applied to SBM can enhance its nutritional and functional properties, leading to improvements in growth performance, stool quality, SCFA production, immune response, and intestinal morphology or barrier function in nursery pigs (Webster et al. 2003; Kim et al. 2010; Ruckman et al. 2020; Li et al. 2021; Long et al. 2021; Muniyappan et al. 2023). In the study of Muniyappan et al. (2023), fermented SBM was incorporated at 3%, 6%, 9% of the diet throughout a 42-d nursery period. They reported that these diets improved growth performance, ATTD of DM, CP and gross energy, blood profiles and altered fecal microbiota by increasing Firmicutes and decreasing Bacteroidetes phyla of the weaned pigs. In contrast, Ruckman et al. (2020) showed that increasing the inclusion of enzymatically treated SBM beyond 7% had a negative impact on nursery pig performance, although it improved stool quality. Moreover, feeding enzymatically treated SBM enhanced oxidative status and intestinal barrier integrity, while having minimal effect on intestinal inflammation or morphology (Ruckman et al. 2020). Comparing enzyme-treated SBM with extruded full-fat soybean for 14 d post-weaning, Long et al. showed that enzyme-treated SBM enhanced G: F and ADG and improved antioxidant status and gut barrier function (Long et al. 2021). Similarly, Li et al. (2021) also reported higher CP digestibility and G: F in enzyme-treated SBM than pigs fed extruded full-fat soybean diet (Li et al. 2021). Notably, while formulated at similar CP and energy level, these studies have included various levels of extruded full-fat soybean or different soy product levels compared to enzyme-treated SBM treatment, and dietary fat levels of these diets were higher for extruded full-fat soy treatment. To our knowledge, the present study is the first one replacing SBM entirely with HTM processed SBM and Enz Trt SBM and comparing the effects with SBM under both regular nursery and ETEC conditions.

In the present study, piglets fed with HTM SBM or Enz Trt SBM showed improved growth performance and feed efficiency compared to those fed with SBM, as indicated by the greater BW, ADG, ADFI, and G: F during the first 2 wk of post-weaning, regardless of ETEC challenge condition. Piglets exposed to the ETEC challenge exhibited reduced growth performance, but HTM SBM or Enz Trt SBM helped mitigate some of these negative effects of ETEC challenge on growth performance. The F4+ ETEC challenge often impairs the growth performance of the nursery pigs, as the bacteria attach to the small intestine and secret the enterotoxin LT and STa that damage the tight junctions, increasing the gut permeability which leads to inflammation, diarrhea and poor nutrient digestion and absorption (Sun and Kim 2017; Kim et al. 2022b; Wensley et al. 2023). Consistent with what others reported, the reduced feed intake in ETEC challenged condition in the current study might have directly impair the ADG (Khafipour et al. 2014; Xu et al. 2022; Deng et al. 2025). Further, pigs from HTM SBM and Enz Trt SBM treatments had higher ATTD of CP and numerically higher AID of CP in ETEC challenge, indicating better nutrient utilization which also explains the mitigation effect of these two processed SBM.

During the first week of post-weaning, pigs housed under regular nursery condition showed higher ADG relative to ADFI, resulting in a G: F higher than 1 during this period. One possible explanation is that pigs consumed large amounts of water to compensate for the loss of sow milk after weaning. G: F greater than 1 was also shown by the past study, where supplementation of whey permeate led to higher G: F in nursery pigs during weaning transition (Jang et al. 2021). These authors speculated that lactose in whey permeant or solid form diets might have helped weaning pig increase water intake. Similarly, the nursery diets containing lactose in the current study might have helped pigs recover from weaning stress by increasing the water and feed intake. Another possibility is that these pigs gained extra weight due to edema caused by fasting at weaning and erratic intake of a carbohydrate-rich nursery diet, as previously theorized by van (Kempen et al. 2023).

Even though we did not observe fecal score differences in the current study due to the intended milder ETEC challenge, colon and fecal DM content was lower in ETEC challenged pigs. DM is a key indicator for post-weaning diarrhea in pigs which involves a decrease in fecal DM content, leading to watery, loose feces due to osmotic pull, reduced nutrient absorption and gut microbiota dysbiosis, often triggered by weaning stress, diet change and pathogens like ETEC (Gao et al. 2019; Zheng et al. 2021; Jenkins et al. 2024; Deng et al. 2025). Colon DM tended to be higher in HTM SBM and Enz Trt SBM treatment, likely owing to the better digestibility of nutrients in these two treatments compared to SBM. This was similar as shown in previous work that replacing SBM with processed SBM led to better nutrient digestibility and fecal consistency (Ma et al. 2019; Deng et al. 2022; Deng et al. 2023).

HTM SBM and Enz Trt SBM improved ATTD of CP regardless of ETEC challenge and numerically increased AID of CP compared to SBM under regular nursery condition, suggesting better nutrient utilization. When estimating using the difference between ATTD and AID CP digestibility, numerically less CP was fermented in the hind gut of both HTM SBM and Enz Trt SBM compared to SBM treatment, indicating less proteolytic fermentation in the hindgut. Less proteolytic fermentation would reduce the production of ammonia, biogenic amines, and phenols which can damage the gut lining, comprise barrier function, increase inflammation, and lead to post-weaning diarrhea (Pieper et al. 2016; Gilbert et al. 2018; Zhang et al. 2020). Under ETEC infection, excessive proteolytic fermentation amplifies the negative impact, causing poorer nutrient absorption, reduced growth performance and aggravation of diarrhea symptoms (Kim et al. 2022b; Duarte et al. 2023).

Previous studies and our internal analysis indicate that HTM processing alters protein structure by reducing β-sheet content and increasing α-helix and β-turn content, resulting in a less compact protein structure (Salazar-Villanea et al. 2016; Mollaei Berenti et al. 2021). A lower proportion of β-sheets is associated with greater protein solubility and improved CP digestibility in the small intestine, as the many hydrogen bonds in β-sheets hinder protease activity (Bai et al. 2016). Thus, although HTM SBM has a nutrient profile like conventional SBM—aside from lower moisture and higher CP, the enhanced CP digestibility and growth performance is likely attributable to protein structural modifications induced by HTM processing. On the other hand, higher AID and ATTD of CP in Enz Trt SBM may result from enzymatic pre-digestion, which alters the soy protein structure and generates more small peptides (Zhu et al. 1998; Carrion Lopez et al. 2022).

Dietary interventions affected the production of SCFAs. HTM SBM and Enz Trt SBM increased caproate and valerate content under regular nursey conditions, while only HTM SBM increased caproate content during ETEC challenge. SBM elevated acetate and propionate content across conditions. SCFAs such as acetate, propionate and butyrate are key energy sources for gut cells and support gut barrier integrity and immunity (Lauridsen 2020; Dong et al. 2024; Liu et al. 2024). Both acetate and propionate were higher in SBM than in HTM SBM and Enz Trt SBM, possibly due to reduced raffinose and stachyose from HTM processing and enzymatic treatment, which limits fermentable carbohydrates (Kelkar et al. 2012; Ma et al. 2019; Long et al. 2021). Caproate, part of a broader group of SCFAs with health benefits, enhances growth, immunity, and intestinal homeostasis (Jadhav and Annapure 2023). Notably, zinc caproate supplementation improved growth, gut health, and reduced inflammation in ETEC-challenged piglets, outperforming pharmacological ZnO (Xu et al. 2025). Thus, increased production of caproate from HTM SBM and EnZ Trt SBM may partly explain their positive impact on growth and microbiome composition.

Although we did not observe significant differences in species or genera within treatment or challenge conditions (Table S1), likely due to factors such as study design, high variability between the pens and smaller sample sizes, several taxa ­exhibited substantial differences (more than 5-fold change which is microbiologically meaningful) across sub-groups. Further exploration of these taxa revealed significant correlations between species abundance and performance parameters as well as several SCFA.

From microbiome diversity perspective, pigs fed processed versus unprocessed SBM showed clear differences. Under regular nursery conditions, pigs fed HTM SBM and Enz Trt SBM tended to exhibit a higher microbial diversity compared to those fed unprocessed SBM, a response consistent with previous findings (Muniyappan et al. 2023). A possible explanation is that HTM SBM and Enz Trt SBM diets enhance better accessibility of carbohydrates and proteins that escape upper gastrointestinal tract digestion and reaching the hindgut, thereby positively influencing the microbial community and promoting diversity. In contrast, SBM not only is less degradable in the upper gut but also delivers more structurally complex substrates to the hindgut, favoring specialized taxa adapted to complex carbohydrate and protein degradation under nutrient-limited conditions.

This interpretation is supported by correlations between SCFAs and diet-responsive microbial taxa. HTM SBM and Enz Trt SBM enriched microbes positively associated with valerate (C5) and caproate (C6) production, whereas unprocessed SBM favored microbes negatively correlated with these SCFAs. Higher valerate and caproate are particularly of interest, as these longer SCFAs are typically produced when microbes have access to readily fermentable carbohydrates and amino acids, enabling fermentation beyond acetate, propionate, and butyrate production (Han et al. 2018; Fitzgerald et al. 2025; Huertas-Díaz et al. 2025). Higher production of C5 and C6 SCFAs reflects a community that is not carbon- or electron-limited, meaning substrates are not monopolized by a few specialized taxa (Han et al. 2018). These conditions support microbes with more diverse metabolic capacities, allowing a wider range of taxa to thrive.

Many studies have linked increase in microbial diversity to improved gut health as well as greater microbiome robustness (the ability to resist change when exposed to stressors) and resilience (the capacity and speed with which the community returns to its original state after disturbance) (Fouhse et al. 2016; Guevarra et al. 2019; Dong et al. 2024). Accordingly, it is plausible that the higher microbial diversity observed in pigs fed HTM SBM and Enz Trt SBM may have enhanced resistance to pathogen colonization and reduced the negative effects of ETEC challenge on performance.

No differences in alpha diversity were observed under ETEC challenge conditions across different dietary groups, potentially suggesting that the ETEC challenge exerted a stronger homogenizing effect on the gut microbiome, thereby overshadowing the dietary influences. We observed that the pigs fed SBM had higher microbial diversity under ETEC challenge compared to regular nursery conditions. The presence of pathogenic E. coli can disrupt the microbiome by increasing reactive oxygen species (ROS) in the gut (Apiwatsiri et al. 2022), and stimulating the immune system, leading to heightened inflammation (Boeckman et al. 2022; Deng et al. 2022). For pigs fed SBM, the ETEC challenge may have reduced the dominance of certain microbial species, allowing a broader range of microbes to establish. Consistent with these observations, a significant effect of dietary intervention and ETEC challenge on beta diversity was detected, as indicated by PERMANOVA of Aitchison ­distances. These findings highlight potentially distinct microbial communities associated with different diets and health conditions, confirming that both diet and challenge status play crucial roles in shaping the gut microbiome.

Consistent with Muniyappan et al. (2023), most taxa enriched by HTM SBM and Enz Trt SBM under regular nursery conditions belonged to Firmicutes. For example, Eubacterium_T pyruvativorans, a species within Firmicutes, was observed to be higher in HTM SBM fed pigs compared to SBM fed pigs and strongly correlated to the caproate levels. This is consistent with previous findings from (Wallace et al. 2004), where they showed that metabolically this species can scavenge amino acids to produce caproate. Caproate and other SCFAs play important roles in piglet nutrition by enhancing growth performance, boosting immunity, and maintaining intestinal homeostasis (Lauridsen 2020). A recent study found that the addition of zinc caproate in weaned piglets challenged with ETEC resulted in improved growth performance, better intestinal health, and reduced inflammation (Xu et al. 2025).

We identified several Prevotella species within phylum Bacteroidetes that were responded differently to processed (HTM and Enz Trt) versus unprocessed SBM under regular nursery condition. Processed SBM increased the abundance of Prevotella genome088748, whereas unprocessed SBM increased Prevotella sp004556065 and Prevotella sp002251435. All three taxa are currently known only through metagenome-assembled genomes and remain poorly characterized, making direct comparison of their metabolic capabilities challenging. That said, substantial functional diversity is known to exist across Prevotella species; for example, Prevotella stercorea lacks many hemicellulose degrading enzymes present in Prevotella copri. Therefore, consistent with the earlier discussion regarding substrate avaialbility in the hindgut, we speculate that the P. genome088748 enriched by HTM SBM and Enz Trt SBM perhaps may possess fewer carbohyrdrate-active enzymes, hence, a lower capacity to degrade complex carbohydrate compared with P. sp004556065 or P. sp002251435, which were favored by unprocessed SBM.

ETEC challenge has a larger homogenizing effect on the gut microbiome, overshadowing most of the dietary differences. Nonetheless, comparing HTM SBM and SBM fed pigs under ETEC challenge, quite a few differences were identified in terms of microbial species and their potential relationships to the performance metrics. For example, higher levels of Dysosmobacter sp900544615 in the HTM SBM fed pigs were observed which was positively correlated with caproate amongst others. The negative correlation of Dysosmobacter genera and Propionate was consistent with what we observed and reported in the literature (Azad et al. 2024). Overall, these observations indicate that HTM SBM promotes a microbial community (eg higher levels of Eubacterium_T pyruvativorans) that leads to higher concentration of caproate in the gut, which could be benefiting piglets experiencing ETEC challenges.

Conclusion

In conclusion, dietary inclusion of HTM SBM or Enz Trt SBM in the present study improved growth performance of post-weaning pigs under both regular nursery and under ETEC challenge conditions compared to SBM. The inclusion of HTM SBM or Enz Trt SBM also tended to increase colon DM and alter short-chain fatty acid profile irrespective of ETEC ­challenge. Under regular nursery conditions, HTM SBM or Enz Trt SBM increased microbiome alpha-diversity compared and promoted beneficial microbes linked to growth performance. Collectively, these findings highlight the potential of HTM SBM and Enz Trt SBM to improve growth performance and positively modulating the gut microbiome in nursery pigs. Incorporating these ingredients into nursery diets may offer a practical strategy to support pig growth performance and resilience, particularly under enteric challenge.

Supplementary data

Supplementary data are available at Translational Animal ­Science online.

Funding

The study was solely funded by Cargill Inc.

Author contributions

Qiong Hu (Conceptualization [true], Investigation [true], Writing—original draft [true], Writing—review & editing [true]), Maria I. Sardi (Formal analysis [true], Methodology [true], Writing—original draft [true], Writing—review & ­editing [true]), SyedAli Naqvi (Formal analysis [true], Methodology [true], Writing—review & editing [true]), Neil D. Paton (Formal analysis [true], Methodology [true], Writing—review & editing [true]), Leandro Hackenhaar (Conceptualization [true], Writing—review & editing [true]), Patricia Pluk (Conceptualization [true], Project administration [true], Writing—review & editing [true]), John de Laat (Data curation [true], Formal analysis [true], Methodology [true], Writing—review & editing [true]), Anirikh Chakrabarti (Writing—original draft [true], Writing—review & editing [true]), Ehsan Khafipour (Methodology [true], Writing—review & editing [true])

Conflicts of interest

There is no conflict of interest related to this publication. All authors are employees or former employees of Cargill Inc at the time of completion of this manuscript.