Dietary glycerides of valerate ameliorate diarrhea and impact intestinal physiology and serum biomarkers in weaned piglets infected with enterotoxigenic Escherichia coli F18

Abstract

In the commercial swine farm setting, the postweaning period is a critical window during which piglets are highly susceptible to infection and enterotoxigenic E. coli (ETEC)-associated diarrhea. Short-chain fatty acids and their glycerides are compounds that may influence intestinal health; however, valerate is one that has not been well-characterized for its role as a dietary supplement. Therefore, the major objective of this experiment was to investigate two forms of valerate glycerides on diarrhea, intestinal physiology, and systemic immunity of weaned pigs experimentally infected with ETEC F18. Dietary treatments included a control diet and three additional diets supplemented with 0.075% monovalerin, 0.1% monovalerin, or 0.1% trivalerin, respectively. Piglets were weaned (21 d to 24 d of age), individually housed, and experimental diets were fed through the 28-d trial period. After a 7-d period, all piglets were inoculated on three consecutive days with 10¹⁰ CFU ETEC F18/3 mL. Growth performance was monitored throughout the trial, and daily diarrhea scores were recorded. Rectal swabs were collected for bacterial culture to confirm the presence or absence of β-hemolytic coliforms throughout the trial. Serum samples were collected and analyzed for inflammatory biomarkers on days 0, 3, 6, and 21 postinoculation (PI) and untargeted metabolomics on day 6 PI. Intestinal mucosa and tissue sections were harvested from pigs sacrificed on day 7 PI for gene expression and histology analysis. All data, except for frequency of diarrhea and metabolomics, were analyzed by ANOVA using the PROC MIXED of SAS. Dietary trivalerin reduced (P < 0.05) the frequency of severe diarrhea over the entire trial period and the frequency of β-hemolytic coliforms on day 7 PI compared with the control. The intestinal villus height on day 7 PI in jejunum tissue was increased (P < 0.05) in pigs fed trivalerin. The mRNA expression of TNF-α was decreased (P < 0.05) in the trivalerin group, while that of ZO1 was increased (P < 0.05) compared with control. Throughout the trial, serum TNF-α was reduced in pigs fed trivalerin compared with control. Serum metabolites, adenosine, inosine, and shikimic acid were reduced (P < 0.05) on day 6 PI in all treatment groups compared with control. In conclusion, the present results indicate supplementing dietary valerate glycerides exhibited beneficial impacts on diarrhea, inflammation, and intestinal gene expression of piglets during the postweaning period.

Valerate, a short-chain fatty acid, has received limited study regarding its impacts on intestinal health and disease resistance of weaned piglets when incorporated into diets. This study administered two different glyceride forms of valerate to evaluate their impacts on frequency of diarrhea, inflammatory status, systemic biomarkers, and intestinal physiology. Overall, the use of valerate glycerides exhibited a positive influence on disease status in pigs under enteric infection conditions.

Introduction

Postweaning diarrhea presents a substantial challenge in commercial swine production. Piglets are typically weaned between 21 and 24 d of age in the United States (Faccin et al., 2020). The process of weaning involves an amalgam of external stimuli, including maternal separation, transportation, and introduction to new social groups, which can induce physiological stress and provide opportunities for conflict and competition for access to feed. At this stage, the gastrointestinal tract is highly vulnerable to invasion by a variety of pathogens. The function of the intestinal tract in pigs is still developing upon weaning, as the epithelial barrier and population of immune cells in the lamina propria continue to develop during the first 12 wk of life (Smith et al., 2010). Even without infection, the stress of weaning alone can lead to diarrheal disease (Pohl et al., 2017). Therefore, multiple factors predispose piglets to disease after weaning, often manifesting as diarrhea associated with enterotoxigenic E. coli (ETEC) infection. Contraction of ETEC in piglets not only impacts intestinal health, thus affecting feed efficiency and growth performance, but can also lead to morbidity, secondary infection, and sudden death (Luppi, 2017; Fairbrother and Nadeau, 2019). The use of subtherapeutic-dose antibiotics in feed for growth-promoting purposes is restricted in efforts to mitigate antimicrobial resistance (AMR; Medicine, 2019). Prior to increased regulation of antibiotic use, between 2000 and 2012, E. coli-associated diarrhea occurred in 32–45% of commercial swine operations in the United States (NAHMS, 2012). Postweaning diarrhea is most commonly associated with ETEC F18, in particular (Fairbrother and Nadeau, 2019). With the reduced use of antimicrobials, the development of alternative prophylactic measures is vital to promote animal welfare and support the livestock industry and our food production systems (Kil and Stein, 2010). Zinc oxide included at pharmacological levels in diets is one approach to reduce the instance of diarrhea in piglets; however, this practice raises concerns regarding the development of resistance and heavy metal accumulation in the environment (García et al., 2020).

Nondrug feed additives for weaned piglets, such as pre and probiotics and plant-derived compounds, may protect animals from disease due to immunomodulation, increased nutrient absorption, or effects on the commensal microbial profile of the intestine (Guerra et al., 2007; Kim et al., 2022; Jinno et al., 2023). The benefits conferred to organisms by the consumption of pre and probiotics are largely attributed to the activities of microbial fermentation end-products in the gastrointestinal environment, such as lactic acid (Tejero-Sariñena et al., 2012) or short-chain fatty acids (SCFAs; Wang et al., 2004). The majority of endogenous microbial fermentation occurs in the large intestine, where acetate (2:0), propionate (3:0), and butyrate (4:0) are the primary SCFA end products. Acidification of the luminal environment by SCFAs prevents colonization by acid-intolerant pathogenic bacterial species (Stecher and Hardt, 2011). These compounds have also demonstrated direct antimicrobial activity against pathogenic bacteria in vitro (Huang et al., 2011; Kovanda et al., 2019). Formic acid (1:0) and butyric acid are fed directly to pigs, typically in salt forms, and may enhance growth performance or reduce diarrhea (Canibe et al., 2005; Feng et al., 2018).

Valerate (5:0) is a SCFA that is not well understood for its role as a bacterial metabolite or dietary feed additive. In our previous work, we observed a direct bacteriostatic effect of valerate against ETEC F18 (Kovanda et al., 2019), suggesting that the compound may impact bacterial proliferation during enteric infection. Our pilot study demonstrated that dietary valerate glycerides reduced serum TNF-α and the frequency of diarrhea in weaned piglets coinfected with ETEC F4 and ETEC F18 (Kovanda et al., 2023). The colonization of ETEC F18 begins by binding to F18 receptors expressed by enterocytes of the pig small intestine via fimbriae (Cheng et al., 2006). Therefore, the oral administration of dietary SCFAs aims to reach the site of infection without degradation in the gastric environment, which may be achieved by using glyceride forms of SCFAs. Dietary monoglycerides of valerate have been tested in our previous research (Kovanda et al., 2023); however, triglycerides of valerate have not been evaluated as feed additives for swine. The aim of the current study was to expand our understanding of the impacts of dietary valerate monoglycerides and triglycerides on diarrhea, fecal shedding, intestinal physiology, systemic inflammatory status, and serum metabolomic profile of weaned pigs infected with F18 ETEC. We hypothesized that glycerides of valerate would reduce the frequency of severe diarrhea and fecal shedding while reducing inflammatory status, modifying serum metabolites, and altering intestinal physiology in terms of gene expression and morphology.

Materials and Methods

Animals, housing, experimental design, and diet

This experiment was reviewed and approved by the Institutional Animal Care and Use Committee (IACUC #20809) at the University of California, Davis. A total of 60 pigs (30 barrows and 30 gilts, 21d to 24 d of age) provided by the Swine Teaching and Research Center at the University of California, Davis, were used in this experiment. The procedures for this experiment were adapted from the methods of Liu et al. (2013). Pigs selected were not given E. coli vaccines or antibiotics. Piglets were tested for susceptibility to ETEC F18 infection by checking the F18 receptor status based on methods by Kreuzer et al. (2013). Additionally, fecal cultures were performed prior to allotment in the experiment to confirm the absence of β-hemolytic coliforms indicative of enteric infection. Only piglets susceptible to ETEC F18 were used. Piglets were randomly assigned to one of four dietary treatments in a randomized complete block design. Pigs were blocked by bodyweight (BW) within sex and litter, and the pig was used as the experimental unit. There were 15 replicate pigs per treatment. Pigs were individually housed in disinfected crates and rooms under a controlled environment for 28 d with free access to feed and water offered. Dietary treatments were fed starting on day −7 postinoculation (PI), during a 7-d adaptation period, and throughout the entire trial period. After the 7-d adaptation, oral E. coli inoculation was administered on three consecutive days (0, 1, 2 PI). The ETEC F18 was originally isolated from a field disease outbreak in piglets by the University of Montreal in 2017 (isolate number: ECL22131). The ETEC F18 expresses heat-labile toxin and heat-stable toxins a and b. The inoculums were prepared at 10¹⁰ CFU per 3 mL dose in phosphate buffered saline (PBS) according to our previous published research (Liu et al., 2013; Kim et al., 2019).

The four dietary treatments were: 1) positive control: control diet with E. coli challenge; 2) low dose monovalerin diet: control diet, including 0.075% monovalerin with E. coli challenge; 3) high dose monovalerin diet: control diet, including 0.1% monovalerin with E. coli challenge; 4) trivalerin diet: control diet, including 0.1% trivalerin-based product with E. coli challenge. Experimental diets were fed throughout the trial, and all diets were provided in mashing form. All diets were formulated to meet pig nutritional requirements (Table 1, NRC, 2012). The organic acid products were generously provided by Perstorp Animal Nutrition.

Table 1.

Ingredient compositions of experimental diets¹

Ingredient, %	Control, phase I	Control, phase II
Corn	44.41	57.27
Dried whey	15.00	10.00
Soybean meal	18.00	22.00
Fish meal	10.00	7.00
Lactose	6.00	-
Soy protein concentrate	3.00	-
Soybean oil	2.00	2.00
Limestone	0.56	0.70
L-Lysine·HCl	0.21	0.23
DL-Methionine	0.08	0.05
L-Threonine	0.04	0.05
Salt	0.40	0.40
Vit-mineral, Sow 62	0.30	0.30
Total	100.00	100.00
Calculated energy and nutrient
Metabolizable energy, kcal/kg	3,463	3,429
Net energy, kcal/kg	2,601	2,575
Crude protein, %	22.27	20.80
Arg,3 %	1.23	1.15
His,3 %	0.49	0.47
Ile,3 %	0.83	0.76
Leu,3 %	1.62	1.55
Lys,3 %	1.35	1.23
Met,3 %	0.45	0.39
Thr,3 %	0.79	0.73
Trp,3 %	0.23	0.21
Val,3 %	0.91	0.84
Met + Cys,3 %	0.74	0.68
Phe + Tye,3 %	1.45	1.38
Ca, %	0.80	0.70
Total P, %	0.68	0.59
Digestible P, %	0.47	0.37

¹In each phase, three additional diets were be formulated by adding valerate products to the control.

²Provided by the United Animal Health (Sheridan, IN, USA). Provided the following quantities of vitamins and micro minerals per kilogram of complete diet: Vitamin A as retinyl acetate, 11,136 IU; vitamin D3 as cholecalciferol, 2,208 IU; vitamin E as DL-alpha tocopheryl acetate, 66 IU; vitamin K as menadione dimethylprimidinol bisulfite, 1.42 mg; thiamin as thiamin mononitrate, 0.24 mg; riboflavin, 6.59 mg; pyridoxine as pyridoxine hydrochloride, 0.24 mg; vitamin B12, 0.03 mg; D-pantothenic acid as D-calcium pantothenate, 23.5 mg; niacin, 44.1 mg; folic acid, 1.59 mg; biotin, 0.44 mg; Cu, 20 mg as copper sulfate and copper chloride; Fe, 126 mg as ferrous sulfate; I, 1.26 mg as ethylenediamine dihydriodide; Mn, 60.2 mg as manganese sulfate; Se, 0.3 mg as sodium selenite and selenium yeast; and Zn, 125.1 mg as zinc sulfate.

³Amino acids are indicated as standardized ileal digestible AA.

Clinical observations and sample collection

The daily diarrhea score of each pig was assessed visually by two independent evaluators to provide an average daily score ranging from 1 to 5 (1 = normal feces, 2 = moist feces, 3 = mild diarrhea, 4 = severe diarrhea, and 5 = watery diarrhea). The alertness score of each pig was assessed visually with a score from 1 to 3 (1 = normal, 2 = slightly depressed or listless, and 3 = severely depressed or recumbent). Pigs were weighed on day −7 (weaning day), 0 (first inoculation day), 7, 14, and 21 PI and feed intake was recorded throughout the experiment.

Rectal swabs from each pig were collected on days −7, 0, 3, 7, and 14 PI to perform fecal culture. Blood samples were collected from the jugular vein of all pigs on day 0 before E. coli inoculation and days 3, 6, and 20 PI to further process serum samples. On day 7 PI, 24 animals (6 piglets per treatment) were euthanized and intestinal tissues were collected. The remaining piglets were euthanized on day 21 PI. Prior to euthanasia, pigs were anesthetized by intramuscular injection of 1 mL telazol, ketamine, and xylazine (2:1:1) per 23 kg BW. Then, intracardiac injection with 78 mg sodium pentobarbital (Sleepaway, Vortech Pharmaceuticals, Ltd., Dearborn, MI) per kg BW was used for euthanasia. From each timepoint, 5 cm intestinal sections of jejunum and ileum were harvested, washed with PBS, and fixed for 48 h in 10% neutral buffered formalin for histological analysis. Intestinal mucosa samples were collected from jejunum and ileum by scraping the luminal layer with glass microscope slides and immediately flash freezing in liquid N and stored at −80 °C for future analysis.

Fecal culture

Rectal swabs from days −7, 0, 3, 7, and 14 PI were cultured on E. coli selective media. MacConkey agar was used for confirmation of E. coli, while blood agar containing 5% sheep blood was used to detect absence or presence of β-hemolytic coliforms indicative of ETEC fecal shedding. The percentage of coliforms exhibiting β-hemolysis from total coliforms was calculated for all fecal swabs (Song et al., 2012; Liu et al., 2013). Primers for F18 + E. coli were used for PCR analysis to confirm ETEC F18 presence in pooled samples of β-hemolytic coliforms after inoculation (Zhang et al., 2007).

Serum sample analysis

Serum samples from all blood collection timepoints were analyzed for the concentrations of proinflammatory markers using commercial enzyme-linked immunosorbent assay (ELISA) kits. TNF-α and CRP were purchased from R&D Systems, Inc., Minneapolis, MN, and haptoglobin ELISA kits were purchased from Aviva Systems Biology Corporation, San Diego, CA. All procedures were performed according to manufacturer’s instructions. Samples were tested in duplicate, and concentrations were calculated from a standard curve.

Serum D-lactate concentrations from days 3 and 6 PI were analyzed by a commercial colorimetric assay kit (Sigma-Aldrich, St. Louis, MO, USA; Hu et al., 2012; Rong et al., 2015). Serum samples were diluted 2 × using milliQ water. Reactions were performed in duplicate, where 20 μL of diluted sample was mixed with an 80 μL reaction mixture prepared according to manufacturer’s instructions. Absorbance at 565 nm was measured immediately after mixing and again at 565 nm for final absorbance after a 20 min incubation period. The difference between final and initial absorbance was calculated, and the sample D-lactate concentrations were interpolated from a standard curve.

Serum samples (6 replicates per treatment) from day 6 PI were submitted for untargeted metabolomics analysis at the NIH West Coast Metabolomics Center at UC Davis. Individual metabolites were detected by gas chromatography (Agilent 6,890 gas chromatograph controlled using Leco ChromaTOF software version 2.32, Agilent, Santa Clara, CA, USA) and time-of-flight mass spectrometry (GC/TOFMS; Leco Pegasus IV time-of-flight mass spectrometer controlled using Leco ChromaTOF software version 2.32, Leco, Joseph, MI, USA). Briefly, samples were vaporized under pressure, separating compounds by mass, and individual compounds were detected based on the ratio of mass to charge. Metabolites were quantified by peak heights, and individual metabolites were identified.

Intestinal morphology

Fixed intestinal tissue sections collected from day 7 PI were dehydrated by graded ethanol exposure, then trimmed and embedded in paraffin. Samples were sectioned at 4 μm and mounted on microscope slides, then stained by hematoxylin and eosin. Images were taken using an Olympus BX51 microscope at 10 × magnification, then measurements were made using image processing and analysis software, Image J, NIH. Ten villi with associated crypts per sample were used to measure villus height and crypt depth. Villi height to crypt depth ratio was calculated accordingly.

Gene expression

All intestinal mucosa samples were analyzed in duplicate. The relative gene expression in jejunum and ileum mucosa from day 7 PI was analyzed by quantitative real-time PCR (qRT-PCR). Tissue samples were homogenized using TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA), and total RNA was extracted and used for reverse transcription. The RNA was quantified, and quality was confirmed by measuring an absorbance ratio of approximately 2 for 260 nm and 280 nm using a Thermo Scientific NanoDrop 2,000 Spectrophotometer (Thermo Scientific, Inc., Waltham, MA). cDNA was generated from 1 μg of total RNA per sample using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems; Thermo Fisher Scientific, Inc., Waltham, MA). The 18s rRNA gene was utilized as a housekeeping gene to analyze relative mRNA expression of target genes. Primer sequences, accession numbers, and cycling conditions are described in Supplementary Table S1. Standard curves were generated using serially diluted concentrations of a pooled cDNA sample to confirm the efficiency of the reaction conditions (90–110%). Relative quantification of mRNA encoding for the gene compared to the control group was calculated by the 2−^ΔΔCT method.

Statistical analysis

Normality of data was verified and outliers were identified using the UNIVARIATE procedure (SAS Inst. Inc., Cary, NC, USA). Outliers were identified and removed as values that deviated from the treatment mean by more than three times the interquartile range. Data were analyzed by ANOVA using the PROC MIXED of SAS (SAS Inst. Inc., Cary, NC, USA) in a randomized complete block design with the pig as the experimental unit. The statistical model included diet as a fixed effect and blocks as random effects. Treatment means were separated by using the LSMEANS statement and the PDIFF option of PROC MIXED. The Chi-square test was used for analyzing the frequency of diarrhea. Statistical significance and tendency were considered at P < 0.05 and 0.05 ≤ P < 0.10, respectively.

Metabolomics data were analyzed using several modules on the web-based platform, MetaboAnalyst 5.0 (Pang et al., 2021). Data were filtered to include peaks with detection rates of less than 30% missing abundances, then normalized using logarithmic transformation and autoscaling. For individual metabolites, fold change (FC) in treatment groups compared with control was calculated and considered significantly different when FC > 2, and log2(FC) > 1. Metabolites with significantly different FC and variable importance in projection scores > 1 were used in pathway analysis to associate patterns in metabolites with biological pathways. Metabolic pathways associated with significantly impacted individual metabolites were identified using Metabolite Set Enrichment Analysis (MSEA). Finally, the entire serum metabolomic profile from day 6 PI for each treatment compared with control was assessed using principal component analysis plots.

Results

Growth performance, diarrhea, and fecal culture

There were no differences in BW in any treatment group compared with control pigs (Supplementary Table S2). The daily diarrhea scores from days −7 to 20 PI are shown in Figure 1. On day 2 PI, pigs fed 0.1% trivalerin exhibited reduced (P < 0.05) diarrhea score compared with control pigs. Diarrhea scores peaked in each treatment group between days 2 and 4 PI, where an overall downward trend was observed for the remainder of the trial period. Over the entire trial period, the frequency of a diarrhea score ≥ 4 was reduced (P < 0.05) in the 0.1% trivalerin group compared with control (Figure 2).

Figure 1.

Mean diarrhea scores of piglets infected with enterotoxigenic Escherichia coli F18 throughout the trial period (days −7 to 20 post-inoculation). Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin. *P < 0.05 indicates that the diarrhea scores were different among treatments. Each least squares mean represents 9–15 observations.

Figure 2.

Frequency of severe diarrhea (score ≥ 4) in piglets infected with enterotoxigenic Escherichia coli F18 over the entire trial period (days −7 to 20 post-inoculation). Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin. Frequency of severe diarrhea was calculated as the percentage of pigs with diarrhea score ≥ 4 out of total pigs. *P < 0.05 indicates that the frequency of severe diarrhea was different among treatments. Each least squares mean represents 9–15 observations.

On day 3 PI, the percentage of β-hemolytic coliforms in fecal cultures was 92.9% in control, 81.8% in low monovalerin, 100% in high monovalerin, and 90.1% in trivalerin, respectively. On day 7 PI, there were fewer (P < 0.05) β-hemolytic coliforms in fecal cultures from pigs fed 0.1% trivalerin compared with control (Figure 3). No difference (P > 0.05) was observed in fecal cultures across dietary treatments on days 14 and 21 (data not shown).

Figure 3.

The percentage of β-hemolytic coliforms in fecal cultures from rectal swabs in piglets infected with enterotoxigenic Escherichia coli F18 on day 7 post-inoculation. Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin. *P < 0.05 indicates that the percentage of β-hemolytic coliforms in fecal culture was different among treatments. Each least squares mean represents 9–15 observations.

Serum biomarkers

Pigs fed 0.1% trivalerin showed significantly reduced (P < 0.05) serum TNF-α on days 3, 6, and 21 PI compared with control (Table 2). Additionally, reduced serum TNF-α was also observed on day 3 PI in pigs fed 0.1% monovalerin compared with control. No differences (P > 0.05) in serum haptoglobin, CRP, or D-lactate were detected across treatment groups at tested timepoints.

Table 2.

Serum biomarkers in weaned pigs under enterotoxigenic Escherichia coli F18 challenge conditions fed diets supplemented with valerate glycerides

Item1	Control	Low monovalerin	High monovalerin	Trivalerin	SEM	P-value
TNF-α, pg/mL
Day 0	989	809	695	658	104	0.116
Day 3 PI	905a	749a,b	554b	530b	108	0.063
Day 6 PI	1,033a	893a,b	773a,b	582b	113	0.045
Day 21 PI	606a	465a,b	375a,b	293b	131	0.079
CRP, ng/mL
Day 0	5.94	5.50	7.35	6.08	0.887	0.501
Day 3 PI	8.76	6.93	6.50	8.33	0.939	0.287
Day 6 PI	7.52	5.85	5.46	7.04	0.989	0.432
Day 21 PI	8.76	8.24	6.83	7.96	1.357	0.207
Hp, mg/mL
Day 0	2.83	2.70	2.50	2.27	0.405	0.843
Day 3 PI	4.18	4.87	4.25	3.70	0.535	0.401
Day 6 PI	2.07	2.85	2.45	2.85	0.352	0.370
Day 21 PI	0.100	1.67	1.01	0.97	0.348	0.316
D-Lactate, mM
Day 3 PI	1.25	1.72	1.63	1.51	0.181	0.307
Day 6 PI	1.43	1.24	1.55	1.52	0.162	0.542

^a,bMeans without a common superscript are different (P ≤ 0.05).

¹TNF-α = tumor necrosis factor-alpha; CRP = C-reactive protein; Hp = haptoglobin. Each least squares mean represents 12–15 observations for day 0 before inoculation and days 3 and 6 post-inoculation (PI), and 7–9 observations for day 21 PI. Control = basal diet formulation; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin.

Metabolomics

Individual metabolites (features) that were significantly reduced according to FC analysis when comparing serum from each treatment with control are shown in Table 3. Several metabolites were commonly affected in all treatment groups compared with control: adenosine, putrescine, and shikimic acid. Meanwhile, inosine, nicotinic acid, and palmitoleic acid were impacted in both monovalerin groups compared with control. Both high monovalerin and trivalerin groups showed impacted levels of 2-hydroxybutanoic acid, alpha-aminoadipic acid, and salicylic acid. When performing MSEA, the set of metabolites associated with methionine metabolism was significantly affected in each treatment group compared with control, while purine metabolism was impacted in both monovalerin groups compared with control but not in the trivalerin group (Supplementary Figure S1). No differential metabolomic serum profiles by treatment were identified as indicated by a lack of cluster separation using principal component analysis (Supplementary Figure S2).

Table 3.

FC calculated compared with control for individual metabolites identified in serum from weaned pigs under enterotoxigenic Escherichia coli F18 challenge conditions fed diets supplemented with valerate glycerides on day 6 post-inoculation

1 Feature	2 FC	log2(FC)	3 VIP
Control vs. Low monovalerin
Putrescine	4.0838	2.0299	1.6194
Adenosine	3.4669	1.7936	1.662
Shikimic acid	3.3322	1.7365	1.662
Inosine	3.1207	1.6419	1.5861
Nicotinic acid	2.3778	1.2496	1.6878
Phenaceturic acid	2.1859	1.1282	1.4195
Glycerol	2.1688	1.1169	2.1907
Palmitoleic acid	2.0889	1.0627	1.011
Control vs. High monovalerin
Putrescine	4.1285	2.0456	1.5021
Montanic acid	4.0352	2.0456	2.1291
Adenosine	3.8015	1.9266	1.4937
Shikimic acid	3.701	1.8879	1.5839
2-hydroxybutanoic acid	3.3298	1.7355	0.98324
Inosine	2.9298	1.5508	1.4173
Alpha-aminoadipic acid	2.9144	1.5432	0.92767
Salicylic acid	2.7564	1.4628	1.5137
Oleic acid	2.7094	1.438	1.7513
Palmitoleic acid	2.5875	1.3716	1.0819
Nicotinic acid	2.5022	1.3232	1.5973
Sorbitol	2.4129	1.2708	2.3221
Control vs Trivalerin
Adenosine	3.6569	1.8706	1.4853
Putrescine	3.3509	1.7446	1.4182
2-hydroxybutanoic acid	3.3184	1.7305	0.99172
Shikimic acid	3.1163	1.6398	1.5002
Alpha-aminoadipic acid	2.4283	1.2799	0.8431
2-aminobutyric acid	2.1749	1.1209	1.0878
Salicylic acid	2.0852	1.0602	1.2559
Hippuric acid	2.0852	1.0602	1.8911

¹Feature = individual metabolite.

²FC = fold change.

³VIP = variable importance in projection. Significant differences were considered when FC > 2 and log2(FC) > 1. Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin.

Intestinal morphology and gene expression

In the jejunum, pigs fed 0.1% monovalerin exhibited increased (P < 0.05) villus height and crypt depth compared with control pigs; thus, no difference (P > 0.05) in villus height:crypt depth was observed when comparing these two treatments (Supplementary Table S3). No differences (P > 0.05) in ileal villus height and crypt depth were observed across treatments on day 7 PI.

The mRNA expression of TNF-α in jejunum mucosa was not statistically impacted by dietary treatments, while supplementing 0.1% monovalerin reduced (P < 0.05) TNF-α expression in ileal mucosa compared with control on day 7 PI (Figure 4). No differences (P > 0.05) in the expression of IL10 nor IL12 were detected in jejunal and ileal mucosa across treatment groups. In jejunum mucosa, the expression of ZO1 was upregulated (P < 0.05), but the gene expression of OCLN was downregulated (P < 0.05) in pigs fed 0.1% trivalerin compared with control (Figure 5). No differences (P > 0.05) were observed in the mRNA expression of ZO1 and OCLN in ileal mucosa and CLDN1 gene expression in jejunal and ileal mucosa. The mRNA expression of genes related to G-protein coupled receptors (GPR41 and GPR43) and chloride secretion (CFTR) were not different (P > 0.05) in jejunum and ileum mucosa across treatment groups (Figure 6). The mRNA expression of CCK was upregulated (P < 0.05) in the low (0.075%) monovalerin group in ileum mucosa compared with control.

Figure 4.

The relative mRNA expression of genes related to cytokine production from jejunum and ileum mucosa on day 7 post-inoculation in piglets infected with enterotoxigenic Escherichia coli F18. Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin. TNF-α = tumor necrosis factor-alpha; IL12 = interleukin 12; IL10 = interleukin 10. *P < 0.05 indicates that the relative fold gene expression was different among treatments. Each least squares mean represents six observations.

Figure 5.

The relative mRNA expression of genes related to tight junction proteins from jejunum and ileum mucosa on day 7 post-inoculation in piglets infected with enterotoxigenic Escherichia coli F18. Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin. ZO1 = zona-occludens 1; CLDN1 = claudin-1; OCLN = occludin. *P < 0.05 indicates that the relative fold gene expression was different among treatments. Each least squares mean represents six observations.

Figure 6.

The relative mRNA expression of regulatory genes from jejunum and ileum mucosa on day 7 post-inoculation in piglets infected with enterotoxigenic Escherichia coli F18. Control = basal diet; Low monovalerin = 0.075% monovalerin; High monovalerin = 0.1% monovalerin; Trivalerin = 0.1% trivalerin. GPR41 = G-protein coupled receptor 41; GPR43 = G-protein coupled receptor 43; CCK = cholecystokinin; CFTR = cystic fibrosis transmembrane regulator. *P < 0.05 indicates that the relative fold gene expression was different among treatments. Each least squares mean represents six observations.

Discussion

The present experiment evaluated dietary supplementation of valerate glycerides on diarrhea, intestinal physiology, and serum metabolomic profile in piglets challenged with an enterotoxigenic strain of F18 E. coli. Overall, trivalerin had more impacts on the measured parameters than monovalerin. This finding is likely due to the potential higher net level of dietary valerate delivered in the trivalerin treatment compared with monovalerin treatments. Triglycerides undergo hydrolysis by lipase, yielding diglycerides and free fatty acids, with diglycerides subsequently cleaved into one glycerol and one monoglyceride. Between 10% and 30% of dietary triglycdrides undergo lipolysis in the stomach, while brush border and pancreatic enzyme secretion complete digestion (Lauridsen, 2020). Meanwhile, monoglycerides are either directly absorbed by enterocytes or cleaved into glycerol and a free fatty acid and then absorbed in the small intestine (Carreiro and Buhman, 2019). Therefore, the end-product composition of the trivalerin supplement after enzymatic digestion provides monoglycerides and free forms of valerate to the small intestine. Increased enzymatic activity is necessary for the digestion of dietary triglycerides. When considering the application of dietary triglycerides, it should therefore be noted that weaning stress can cause an immediate reduction in pancreatic lipase activity (Jensen et al., 1997), though the enzyme activity slowly recovers as the postweaning period progresses. Further research that aims to determine the dynamics of triglycerides and monoglycerides in different segments of the small intestine is suggested.

Diarrhea and fecal shedding

Diarrhea is commonly observed after weaning, independent of bacterial infections (Moeser et al., 2007; Medland et al., 2016). In the current study, F18 ETEC infection was confirmed by the continuing increase in diarrhea score over the days following inoculation and the presence of β-hemolytic coliforms in feces (PI). Supplementation of 0.1% trivalerin reduced the frequency of diarrhea over the entire trial period and the fecal shedding of β-hemolytic coliforms on day 7 PI. The high dose (0.1%) of monovalerin showed a tendency toward these outcomes. The reduced fecal shedding of ETEC on day 7 PI suggests the potential increased clearance of infection and may be ascribed to several modes of action. Firstly, the direct antimicrobial activity exerted by organic acids must be considered. Valeric acid is considered a weak organic acid that resists dissociation (to valerate and a proton) in environments with a pH lower than its dissociation constant (pKa, 4.482; Sue et al., 2004). The pH of the small intestine is maintained between ~4.5 and 7 (Braude et al., 1976); therefore, we can assume that a proportion of valeric acid dissociates in this environment. Yet, the proportion of protonated fatty acids may be sufficient to implement antimicrobial effects on susceptible bacteria. This antimicrobial process is theorized to involve the bypass of fatty acids across the bacterial cell membrane and subsequent intracellular dissociation, where the need to efflux excess protons can disrupt energy balance, and anion accumulation can alter cytoplasmic pH, affecting enzyme activity and nucleic acid synthesis (Mani-López et al., 2012). However, it is unclear whether the concentration of valerate in its fatty acid form provided in the experimental diets from the current study is sufficient to exert direct antimicrobial effects on ETEC, which warrants further investigation. As such, measuring the concentration of valeric acid in the intestinal lumen would be helpful.

Intestinal physiology and immunity

The reduced overall frequency of diarrhea and fecal shedding of β-hemolytic coliforms on day 7 PI in the trivalerin group may also be attributed to the modulation of host physiology. The presence of SCFAs elicits regulatory properties both in vitro and in vivo in a variety of organ systems, especially the intestinal environment (Grilli et al., 2015; Yan and Ajuwon, 2017; Feng et al., 2018). In the current study, intestinal morphology and several genes related to intestinal physiology and barrier function were analyzed.

The intestinal villi structures increase surface area of the epithelium, promoting sufficient nutrient absorption (Miller et al., 1986). An increased ratio of functional absorptive enterocytes (located at the villus) to undifferentiated cells of the intestinal crypts is used as an indicator for intestinal function and is usually decreased alongside increased severity of diarrhea (López-Colom et al., 2019), indicating a positive relationship between increased crypt depth and increased severity of diarrhea. The current study found that villus height was increased in the 0.1% monovalerin group on day 7 PI. However, the crypt depth in this group was also increased, and therefore, villus height:crypt depth was not affected. New enterocytes are generated at the intestinal crypt and as more cells originate and maturation and differentiation occur, cells are rearranged to migrate further up the villus structure over time (Ensari and Marsh, 2018). Apoptosis of the cells of the villus tip causes damage to the integrity of epithelium and reduces the height of villi (Chen et al., 2018). Cell death and turnover rate are mediated in part by TNF-α in a pleiotropic manner. Normal production of TNF-α contributes to homeostasis, regulating cell cycle progression and optimal cell turnover rate; however, high levels can induce apoptosis in the epithelium (Parker et al., 2019; Ruder et al., 2019), illustrating the complex relationship between inflammation and intestinal function. When visualizing the pattern in diarrhea severity over time, it is evident that the recovery from the most severe diarrhea across the trial period was in progress by day 7 PI. Supplementing the high dose of monovalerin reduced the gene expression of TNF-α in ileal mucosa compared with the control. As TNF-α is a cardinal mediator of the inflammatory response, the increased TNF-α transcription in control pigs compared with the trivalerin group indicates an overall reduced inflammatory status in the ileum of treated individuals. There were no differences in TNF-α mRNA expression when comparing monovalerin groups with control. Another proinflammatory cytokine, IL-12, is involved in driving an adaptive immune response geared toward bacterial clearance by promoting differentiation of T-cells to the T_h17 subtype (Kiros et al., 2011). Antiinflammatory cytokines, such as IL-10, function to promote differentiation of T-cells to T_reg subsets, which are important to balance the proportion of proinflammatory T_h17 cells during the immune response (Yu et al., 2021). Previous research reported that SCFAs, such as butyrate, may regulate T-cell immunity by influencing IL-10 transcription (Arpaia et al., 2013). However, no difference was observed in IL10 or IL12 mRNA expression across dietary treatments in the current study.

Colonization by ETEC F18 accompanied by the secretion of toxins can alter gene expression and protein distribution of tight junction proteins and stimulate local inflammation (Ngendahayo Mukiza and Dubreuil, 2013; He et al., 2020; Kim et al., 2021). In pigs supplemented with trivalerin, the mRNA expression of ZO1 (zona-occludins 1, ZO-1) was upregulated in jejunal mucosa of pigs, while expression of OCLN (occludin) was downregulated. Both ZO-1 and occludlin are vital components of the tight junction complex of proteins. Occludin functions to form a barrier inbetween neighboring enterocytes, while ZO-1 anchors the complex to the β-actin scaffold intracellularly (Zhang and Guo, 2009). Claudin-1 is a transmembrane protein that also greatly contributes to barrier integrity (Wang et al., 2012). Transcription of CLDN1 (encoding for claudin-1) was not impacted by dietary treatment in the current study. It should be noted that the expression of tight junctions at the gene level may not completely describe the functionality of tight junctions. The protein accumulation and the distribution/localization of tight junctions between adjacent cells should be considered for further evaluation (Slifer and Blikslager, 2020). With interest in generating data to assess intestinal integrity, the concentration of D-lactate was measured. Increased circulating levels of the metabolite have been positively correlated with reduced intestinal barrier integrity (Jin et al., 2007; Zhao et al., 2011), although no changes in serum D-lactate were observed as an effect of dietary treatment.

Additional physiological factors can influence diarrheal disease. For example, activity of cystic fibrosis transmembrane conductance regulator (CFTR) and the downstream effects of G-protein coupled receptor (GPR) agonism can impact the flux of ions across the intestinal epithelium (Kunzelmann and Mall, 2002) or intestinal motility (Dass et al., 2007), respectively. As such, the expression of these transmembrane factors at the gene level may be altered alongside incidence of diarrhea. The transmembrane channel, CFTR, is a cyclic adenosine monophosphate (cAMP)-dependent conductor of chloride ions to the intestinal lumen, contributing to maintenance of electrolyte balance under normal conditions (Zhu et al., 2017). Alternatively, upon stress- (Boudry et al., 2004) or toxin-induced (Sheikh et al., 2020) increases in cAMP, chloride ions can accumulate in the intestinal lumen, leading to passive transepithelial transport of water from the epithelium. Despite this, we did not observe any changes in the gene expression of CFTR in piglets fed trivalerin, although we observed reduced overall frequency of diarrhea. However, cAMP-dependent, CFTR-mediated chloride secretion is just one of several regulatory mechanisms involved in electrolyte balance. Another potential mechanism by which SCFAs affect intestinal health is by binding GPRs. Valerate has displayed agonistic activity for GPR41 (Brown et al., 2003) and GPR43 (Poul et al., 2003). Wu et al. (2012) showed that GPR41 binding downregulated intracellular cAMP, suggesting that activation of GPR41 by SCFAs in the intestine may regulate cAMP-dependent CFTR chloride secretion. However, consistent with the expression of CFTR, supplementing valerate glycerides did not impact the mRNA expression of GPR41 or GPR43.

The interface between SCFAs present in the lumen and intestinal physiology is also linked by enteroendocrine cell activity. Enteroendocrine cells are dispersed among the intestinal epithelium and secrete a variety of peptide hormones, including secretin, glucagon-like peptides, and cholecystokinin (CCK; Egerod et al., 2012). Enteroendocrine cells express GPRs, where the downstream effects of SCFA binding stimulate the secretion of hormones. Therefore, it was anticipated that the mRNA expression of CCK, encoding for the production of the peptide, may be upregulated with the supplementation of dietary valerate glycerides. Although no changes in GPRs at the transcriptional level were observed, the mRNA expression of CCK was upregulated in ileal mucosa in pigs supplemented with 0.075% monovalerin compared with control. The paracrine function of CCK targets the gallbladder inducing contraction and the pancreatic tissue stimulating secretion of enzymes (Chandra et al., 2010). A subpopulation of enteroendocrine cells (I cells) is classically defined by their production of CCK and has been described as localized mainly to the duodenum (Egerod et al., 2012). However, evidence also supports the expression of CCK-producing enteroendocrine cells in regions distal to the duodenum (Agersnap and Rehfeld, 2015).

Systemic immunity

Innate immunity, including the acute phase response, is vital to respond to and clear bacterial infection. As a result of signaling by proinflammatory cytokines, acute phase proteins are produced by the liver, including CRP, which functions to upregulate opsonization of pathogens, drive cytokine production by immunoglobulins, and increase phagocytic activity in macrophages (Du Clos, 2000). Haptoglobin is another indicator of disease, usually increased in serum alongside bacterial infections (Raju et al., 2019). Although the inflammatory response functions to clear infection, uncontrolled production of proinflammatory cytokines can inflict tissue damage, exacerbating diarrhea or even leading to shock (Fink, 1991). To assess the degree of systemic inflammation, circulating levels of TNF-α, CRP, and Haptoglobin were evaluated. Pigs fed trivalerin exhibited reduced TNF-α compared with control at each timepoint, including prior to inoculation on day 0. Weaning piglets, regardless of bacterial infection, are associated with increased proinflammatory cytokine production (Lallès et al., 2004; Hu et al., 2013). These results indicate that sufficient inclusion of valerate glycerides may downregulate inflammation associated with weaning and the postweaning period in nondiseased or infected piglets. In order to confirm the potential benefits of this feed additive, further research using increased sample sizes, expanded trial periods, and exploring inclusion rate and composition of valerate derived feed additives is needed.

Serum metabolites

Although the entire metabolomic profiles overlapped across treatment groups on day 6 PI, individual metabolites were significantly impacted by dietary treatments. Supplementing valerate glycerides reduced serum adenosine levels regardless of dose and type. Upon pathogenesis, intracellular and extracellular dephosphorylation of adenosine triphosphate (ATP), adenosine diphosphate (ADP), and adenosine monophosphate (AMP) result in the formation of adenosine (Yegutkin, 2008). Adenosine is deeply connected to inflammation, exerting effects downstream of binding adenosine receptors expressed on immune cells and is understood to reduce effector cell functions and cytokine production, thereby resolving inflammation (Antonioli et al., 2008, 2014). Adenosine-dependent inhibition of lymphocyte activation has also been observed (Johnston et al., 2005). In addition, adenosine activates GPRs, which can either upregulate or downregulate intracellular cAMP, as demonstrated by Calker et al. (1979) in murine brain cell cultures. In humans, immunohistochemistry analyses revealed the presence of both adenosine receptors and nucleoside transporters localized to the mucosa of the small intestine (Antonioli et al., 2014). Adenosine is also understood to influence gut motility by interaction with the enteric nervous system (Christofi et al., 2001; Antonioli et al., 2006). It is evident that adenosine is tightly linked to important factors in diarrheal disease; however, the implications of the affected serum adenosine levels in pigs fed valerate glycerides are not entirely clear. Comparatively lower serum adenosine on day 6 PI among treated pigs may be related to reduced dephosphorylation of ATP systemically, indicating reduced pathogenesis according to the model proposed by Antonioli et al. (2007). This theory is corroborated by reduced frequency of severe diarrhea over the trial period and reduced ETEC shedding on day 7 PI.

Inosine is a product of the deamination of adenosine, which may occur spontaneously, due to environmental factors, or by activities of nitric oxide produced by leukocytes during the inflammatory response (Nguyen et al., 1992). Inosine can pair with cytosine, providing opportunity for mutation, which is regularly amended by DNA repair enzymes (Alseth et al., 2014). Interestingly, inosine levels were significantly reduced in pigs fed with monovalerin (0.075% or 0.1%) compared with control, while the trivalerin group did not exhibit a difference in serum inosine. Together, changes in adenosine and inosine for monovalerin groups resulted in significant impacts on purine metabolism, according to MSEA. Purines include adenine and guanine, comprising major building blocks of DNA, in addition to other metabolites important for cell survival such as adenosine and inosine (Pedley and Benkovic, 2017). While the trivalerin group exhibited changes in adenosine levels, increased deamination of adenosine to inosine compared with control is not supported, as inosine levels were not different compared with control and the purine metabolism overall was not impacted.

Serum putrescine levels were also reduced by valerate supplements. Putrescine, along with its precursors spermidine and spermine, are classified as polyamines that are required for intestinal epithelial cell differentiation and are especially important during the development of the intestine (Johnson, 1988). Endocrine activity is tightly linked to polyamine synthesis (Johnson, 1988). For example, cortisol challenge in porcine enterocytes resulted in increased concentrations of putrescine, spermidine, and spermine (Wu et al., 2000). Although the glucocorticoid, cortisol, is associated with physiological stress, its production within a range conducive to homeostasis is important for cell growth (Igarashi and Kashiwagi, 2000). Although the reduced levels of circulating putrescine in treatment groups compared with control may support the notion that glucocorticoid activities were impacted by treatment, our current findings lack sufficient supportive data. Together, adenosine and putrescine changes contributed to a significant impact on methionine metabolism according to MSEO when comparing each treatment group with control. Methionine is an indespensable amino acid for pigs, which is among the most limiting amino acids and must be balanced for in swine feed (NRC, 2012). Putrescine is a precursor for the methionine salvage pathway, a highly conserved means of converting 5’-methylthioadenosine (MTA) to methionine (Albers, 2009). MTA has demonstrated the ability to inhibit TNF-α production in LPS-challenged mice (Hevia et al., 2004). Putrescine is a precursor to MTA and was downregulated in treated animals in the current study (Albers, 2009). Perhaps conversion of putrescine to MTA was favored in animals fed valerate glycerides, which would explain the reduced putrescine levels and theoretically provide increased MTA levels and therefore contribute to the reduced serum TNF-α observed. However, MTA was not a detected compound based on untargeted metabolomics analysis.

In the current study, shikimic acid was downregulated in animals treated with dietary valerate glycerides alongside reduced frequency of diarrhea, fecal shedding, and systemic inflammatory markers. Certain strains of E. coli produce shikimic acid and are cultivated for this purpose (Martínez et al., 2015; Hou et al., 2020). Therefore, the reduced serum shikimic acid in treated animals may be directly linked to bacterial load, which is supported by reduced frequency of β-hemolytic coliforms detected on day 7 PI.

Conclusions

In summary, dietary supplementation of valerate glycerides impacted the disease status of weaned pigs experimentally infected with ETEC F18, including reducing frequency of severe diarrhea and fecal shedding of β-hemolytic coliforms. Limited impacts were observed on intestinal physiology and barrier function at the transcriptional level, although changes in CFTR, TNF-α, and ZO1 expression were observed in the intestinal mucosa on day 7 PI. As such, it would be interesting to have evaluated the intestinal morphology of intestinal tissue collected at a timepoint when diarrhea peaked (between days 1 and 5 PI). It is possible that recovery from diarrhea as well as recovered villus height were underway by day 7 PI, rendering differences between treatments less demonstrable. Systemic changes in pigs supplemented with valerate glycerides were observed. In particular, pigs fed 0.1% trivalerin exhibited reduced serum TNF-α at each blood collection timepoint and reduced adenosine on day 6 PI. Meanwhile, pigs supplemented with monovalerin reduced serum adenosine and inosine on day 6 PI, regardless of dose, and 0.1% inclusion maintained reduced serum TNF-α on day 3 PI. In all treatment groups, shikimic acid, a potential byproduct marker of E. coli, was also reduced on day 6 PI. The current study expands on the investigation of dietary valerate glycerides, confirming the efficacy of trivalerin as a suitable feed additive for weaned pigs. Further research is needed to understand the modes of action contributing to the nonnutrient function of dietary valerate glycerides. The direct antimicrobial effects of valerate may drive ETEC clearance, while several pathways regulating host physiology (inflammation, ion secretion, and intestinal barrier function) may be manipulated by valerate interactions. Future research should aim to elucidate these mechanisms independently. In terms of feed additive capacity, the inclusion rate of valerate glycerides is also worth studying more thoroughly.

Glossary

Abbreviations

ANOVA

analysis of variance

AMR

antimicrobial resistance

BW

bodyweight

cAMP

cyclic adenosine monophosphate

CCK/CCK

cholecystokinin

cDNA

complementary deoxyribonucleic acid

CFU

colony forming unit

cm

centimeter

CFTR

cystic fibrosis transmembrane regulator

CLDN1

claudin-1

CRP

C-reactive protein

d

day

ELISA

Enzyme-linked immunosorbent assay

ETEC

enterotoxigenic Escherichia coli

FC

fold change

g

gram

GC/TOFMS

gas chromatography/time-of-flight mass spectrometry

GPR41/GPR41

G-protein coupled receptor 41

GPR43/GPR43

G-protein coupled receptor 43

IACUC

Institutional Animal Care and Use Committee

IL-10/IL10

interleukin 10

IL-12/IL12

interleukin 12

kg

kilogram

LPS

lipopolysaccharide

mg

milligram

MSEA

metabolite set encrichment analysis

MTA

5’-methylthioadenosine

N

nitrogen

NAHMS

National Animal Health Monitoring System

nm

nanometer

NRC

National Research Council

OCLN

occludin

PBS

phosphate buffered saline

PCA

principal component analysis

pH

negative log base 10 of hydrogen concentration

PI

Post-inoculation

qRT-PCR

quantitative real-time polymerase chain reaction

pKa

negative log base 10 of dissociation constant

RNA

ribonucleic acid

SCFA

short chain fatty acid

TNF-α

Tumor necrosis factor-alpha

UC Davis

University of California, Davis

VIP

variable importance in projection

ZO-1/ZO1

zona-occludens-1

μg

micgrogram

μl

microliter

μm

micrometer

Author Contributions

Lauren Kovanda (Data curation, Formal analysis, Investigation, Project administration), Sofia Rengman (Formal analysis, Writing—review & editing), Snehal Tawde (Formal analysis, Writing—review & editing), Jeroen Pos (Formal analysis, Writing—review & editing), Shuhan Sun (Data curation, Writing—review & editing), Kwangwook Kim (Data curation, Formal analysis, Methodology, Supervision, Writing—review & editing), Yanhong Liu (Conceptualization, Formal analysis, Funding acquisition, Methodology, Writing—review & editing), Sangwoo Park (Data curation, Investigation, Methodology, Writing—review & editing), Jungjae Park (Data curation, Methodology, Writing—review & editing), and Xunde Li (Data curation, Methodology, Writing—review & editing)

Conflict of interest statement

S.R., S.T., and J.P. are employees of Perstorp Animal Nutrition (Waspik BV, 5165 NH Waspik, The Netherlands) who generally donated the organic acid products. No other authors have conflicts of interest to declare.