Salmonella enterica serotypes Dublin and Typhimurium were utilized in a preliminary study to determine the capacity of multiple adsorbents in vitro. An overnight culture from each strain was created by inoculation of 20 ml of sterile tryptic soy broth and incubation at 37°C. The overnight culture was quantified by spread plate serial dilutions on tryptic soy agar plates. Dilutions with 30–150 colony forming units (CFU) were counted to determine the concentration of each isolate (CFU/ml). Approximate working concentrations of each serotype were performed in sterile phosphate-buffered saline (PBS) (2 × 10⁶ CFU/ml). PowerGuard (PG; MB Nutritional Sciences, Lubbock, TX, USA), activated carbon (Sigma Aldrich, St. Louis, MO, USA), bentonite #1, and bentonite #2 were all diluted to 5 mg/ml concentrations in sterile PBS. Equal volumes of working bacterial solution and binder solutions were mixed into 50 ml conical tubes, along with negative control samples mixing an equal volume of each bacterial solution and sterile PBS. All cultures were performed in triplicate, and each tube was rocked for 2 h at 37°C. Then, each culture, including negative controls, was filtered through a series of sterile Whatman filter paper. First, a 25-micron paper (Whatman Grade 54, Whatman plc, Little Chalfont, Buckinghamshire, United Kingdom) followed by an 8-micron paper (Whatman Grade 2) to remove most of the adsorbent particles but to still allow un-adsorbed Salmonella spp. to flow through the filter paper into the filtrate. The final filtrate was clear; however, this methodology may underestimate the adsorptive capacities due to some adsorbent particle sizes being smaller than 8 microns, leading to some smaller adsorbent particles remaining in the filtrate. Each filtrate sample was serially diluted in sterile PBS and spread plated on tryptic soy agar. Plates were incubated overnight at 37°C and plates with CFU counts of 30–150 were counted. Data were reported as the log₁₀ reduction from the control to the adsorbent solution for each isolate.

For imaging purposes utilizing scanning electron microscopy, samples were prepared by the addition of 1.75 × 10⁹ CFU/ml of Salmonella typhimurium in 10 ml of deionized water with 25 mg of PowerGuard. The samples were incubated at 37°C for 2 h before being fixed. The samples were then chemically fixed with 2% glutaraldehyde for 1 h, then washed with deionized water three times while centrifuging between each wash. The solutions were then quickly frozen in a dry ice and ethanol bath for 10 min followed by freezing overnight. The dried samples were mounted onto sample mounts with double-sided carbon tape and coated with a thin layer of Iridium to provide conductivity. Scanning electron microscopy (SEM) images were taken by Zeiss crossbeam 540 at 5 kiloelectron volts (kV) using secondary electrospray ionization detector (Zeiss Meditec Inc., Dublin, CA, USA). Samples were observed at the Texas Tech University Microscopy Core Laboratory by SEM with the assistance of Dr. Zhao.

All procedures in this study were ap-proved by the USDA-ARS, Livestock Issues Research Unit's Institutional Animal Care and Use Committee (IACUC protocol #LIRU-2120S). The study was conducted in May 2021. Forty crossbred newly weaned barrows (L280 × Camborough, PIC, Inc., Hendersonville, TN, USA) were weighed (1.98 ± 0.09 kg) and transported 9 km from the Texas Tech Swine Unit to the swine barn at the Livestock Issues Research Unit. Pigs were transported mid-day when the temperature was 27°C. Pigs were randomly assigned to one of five treatments (n = 8): (1) Uninfected Control (CON), no Salmonella typhimurium (ST) administered and no treatment in the diet; (2) Infected Control (ST), infected with ST on day 7 but no treatment in the diet; (3) PG0.05, infected with ST on day 7 with 0.05% PG of the diet; (4) PG0.25, infected with ST on day 7 and 0.25% PG of the diet; and (5) PG0.5, infected with ST on day 7 and 0.5% PG of the diet. PowerGuard is a thermally processed ceramic particle that is micronized to a median particle size of 40 μm, with 90% of the particles being <100 μm. The barn was maintained at 32.2°C for days 0 and 1. The environmental temperature was then decreased by 0.56°C each day. The final barn temperature of 23.8°C was reached on day 16 and maintained for the duration of the study. Pigs were housed in individual stainless-steel pens (1.2 × 0.6 m) equipped with stainless steel automatic feeders and nipple waterers. All pigs had ad libitum access to feed and in order to allow ad libitum access to the feed, the automatic feeders had the gate open 7.5 cm, which likely allowed for pigs to waste some feed. Therefore, feed intake data and feed efficiency data are reported as feed disappearance. All pigs were enrolled on the same day, 21.5 ± 1.33 days of age, and immediately upon arrival to the facility were individually anesthetized briefly to insert temperature measuring loggers (Star-Offi DST micro-T; Metermall USA, Marysville, Ohio) into their abdominal cavity. This methodology is previously described (Burdick Sanchez et al., 2017). Peripheral blood (10 ml) was drawn in EDTA vacutainer tubes while each pig was anesthetized via jugular venipuncture. The entire process was completed in <10 min. Pigs had ad libitum access to feed and water. Diets were formulated to meet or exceed the NRC (2012) nutrient recommendations (NRC, 2012). They received the experimental diets over two dietary phases from day 0 to 6 (P1) and day 7 to 21 (P2), respectively. Samples of each diet were collected at each feeding and composited by period before being analyzed for dry matter, ash, crude protein, and ether extract by a commercial lab using wet chemistry methods (ServiTech, Amarillo, TX, USA). Formulated ingredients and diet nutrient compositions for both P1 and P2 are shown in Table 1.

Pigs in treatments ST, PG0.05, PG0.25, and PG0.5 were all infected on day 7 with 1.75 × 10⁷ CFU of Salmonella typhimurium that was resistant to nalidixic acid and in the mid-logarithmic phase of growth ATCC 14028, which was originally isolated from cardiac tissue in chickens. The infection and identification methods were adapted from Broadway et al. (2015). To summarize, an individual colony from a streak plate was placed into trypticase soy broth with nalidixic acid (50 mg/L) and incubated at 37°C agitating at 200 rpm overnight. Then, 100 μl of the overnight culture was added to 20 ml of fresh trypticase soy broth with nalidixic acid (50 mg/L) incubated at 37°C agitating at 200 rpm until reaching an optical density of 0.8 at 450 nm. This was determined previously to be the mid-logarithmic phase of growth. This broth was diluted by the addition of 1 ml of the bacterial broth with 9 ml of sterile saline for a final concentration of 1.75 × 10⁷ CFU per the 10 ml. Sodium bicarbonate was also added to buffer the bacteria at 1 g/10 ml. This buffered bacterial solution was then administered by oral gavage utilizing a 10 ml syringe with an attached 1 mm internal diameter silicone tube, 6 mm in length, administered behind the tongue slowly as the pig swallowed within 10 min of creating the working bacteria solution. Uninfected control pigs (CON) were administered 10 ml of sterile saline with 1 g/10 ml of sodium bicarbonate prior to any handling of infectious doses or pigs in any other treatment. To estimate the infection dose, the working bacteria solution was serially diluted on trypticase soy agar with nalidixic acid at 50 mg/L and incubated overnight at 37°C. Plates were counted with colony counts between 30 and 150, and the infection dose was determined to be 1.75 × 10⁷ CFU.

Individual pig body weights and peripheral blood samples were collected on days 0, 7, 10, 14, and 21. Peripheral blood (10 ml) was collected by jugular venipuncture via a 20-gauge needle 1.5 inches in length in EDTA vacutainer tubes while the pig was restrained in a v-trough. Complete blood count analysis was performed on the samples using an IDEXX analyzer with the swine-specific algorithm (ProCyte Hematology Analyzer, IDEXX, Westbrook, ME, USA). Fecal swabs for bacteria prevalence were collected following Broadway et al. (2015). Two fecal swabs were collected from fresh feces from each crate beginning on day 7 prior to Salmonella typhimurium administration. As described by Broadway et al. (2015), one swab was placed directly into Rappaport-Vassiliadis (RV) broth (Thermo Fischer Scientific, Waltham, MA, USA) and the other swab was placed directly into Difco Tetrathionate (TT) broth (Thermo Fischer Scientific, Waltham, MA, USA). Fresh fecal swabs were collected every morning from day 7 through day 21 by two trained individuals only swabbing fresh feces. If no fresh feces were present, a fecal swab was taken directly from the rectum for that sample day. The fecal swabs in the RV broth were incubated overnight at 42°C, and the TT broth swabs were incubated overnight at 37°C. All samples were then streak plated on Xylose Lysine Deoxycholate Agar (XLD) (Thermo Fischer Scientific, Waltham, MA, USA). The plates were then incubated overnight at 37°C to determine if there was Salmonella colony growth via examination for the presence or absence of black colonies. Enrichment samples were recorded as either positive or negative. At harvest, a section of the ileum, as well as associated mesenteric lymph nodes, were collected, rinsed, and immediately dipped into boiling water for no more than 5 s to remove potential external contamination from the harvest process for later bacterial quantification and enrichment. These samples were then stomached in a 1:10 dilution of sterile 1X PBS and sampled three subsequent times. Two swabs were collected from each tissue sample and grown overnight in both TT and RV enrichment broth and subsequently streak plated on differential XLD agar as stated previously in the fecal swabs. These samples provided enrichment data on only the presence of ST in the tissue samples and not quantification. The third sample from these stomached tissues was serially plated for Salmonella typhimurium quantification purposes on XLD agar as well.

On day 21, pigs were humanely euthanized. Tissue samples were collected from both the ileum and liver, washed with sterile saline to remove blood and digesta, then placed directly into 10% buffered formalin for histological analyses after hematoxylin and eosin staining. Slides were created at the Anatomical Pathology Lab at the Texas Tech University Health Sciences Center (Texas Tech University Health Science Center, Lubbock, TX, 79415). Slides were analyzed with an anatomical histopathologist at the Texas Tech School of Veterinary Medicine and scanned via an Olympus slide scanner for imaging and count purposes (Olympus Life Sciences, Center Valley, PA, 18034). The method of scoring histology slides is as follows. Inflammation scoring was on a 0 to 3, scale with 0 = normal tissue; 1 = mild inflammation; 2 = moderate inflammation; and 3 = marked inflammation. Villi blunting: 0 = long, slender villi; 1 = 75% of normal height; 2 = 50% of normal height; and 3 = 25% of normal height. Villi epithelium score: 0 = single layer of the columnar epithelium; 1 = degeneration, vacuolation, or separation of focal areas of the superficial epithelium; 2 = more marked changes and some focal loss of epithelium; and 3 = widespread ulceration of the epithelium surface. Lacteal dilation score: 0 = central lacteal represents approximately 25% of the width of the villi; 1 = represents approximately 50% of the villi width; 2 = 75% of the villi width; and 3 = dilated to 100% of the villi width. Goblet cell count in the crypt and intraepithelial lymphocyte counts are the number of cells counted per 50 epithelial cells. Lamina propria counts were the number of each leukocyte (lymphocytes, neutrophils, and eosinophils) observed between crypts in an x40 field. Connective tissue (CT) prominence scores ranged from 0 to 3, with a score of 0 = no prominent CT, 1 = slightly prominent CT, not diffuse; 2 = more prominent CT, throughout the sample, and 3 = all CT present is severe and prominent. Lymphocyte foci scores ranged from 0 to 3, with 0 = no lymphocyte aggregates present; 1 = few lymphocyte aggregates present; 2 = multiple lymphocyte aggregates present; and 3 = severe and diffuse lymphocyte aggregates present throughout the sample. Lymphocyte foci area was measured in μm and conducted on up to 5 foci/samples and averaged per sample. The neutrophil scores ranged from 0 to 3, with 0 = no neutrophils present in lymphocyte aggregates, 1 = <5 neutrophils within lymphocyte aggregates; 2 = between 5 and 10 neutrophils within the lymphocyte aggregates; and 3 = more than 10 neutrophils within the lymphocyte aggregates.

Preliminary in vitro adsorption data were analyzed using the GLM procedure of SAS with adsorbent, Salmonella enterica serotype, and their interaction as the fixed effects. Continuous, repeatedly measured data were analyzed using the Mixed procedure in SAS (SAS 9.4, Cary, NC, USA). The model included fixed effects of treatment, time, and treatment × time. Initial body weight was included in the model as a covariate and retained in the model if it was significant. The subject of the repeated statement was pig nested within treatment. All covariance and variance structures were analyzed, and the most appropriate model was selected based on the smallest Bayesian information criterion. All continuous, non-repeated data were analyzed using a restricted maximum-likelihood ANOVA with treatment included as the fixed effect using the MIXED procedure of SAS. Count data were analyzed using the GLIMMIX procedure of SAS with treatment as the fixed effect and fit with a Poisson distribution using the log link function. Differences of P ≤ 0.05 were considered significant and a tendency was reported when 0.05 < P ≤ 0.1. Treatment × time interactions found to be significant were further evaluated using a Duncan adjustment to control for familywise error. All pairwise comparisons at each significant time point were reported.

Measures of the capacity for different adsorbents to reduce bacteria concentrations reported as log base 10 reductions, in the final filtrate are reported in Figure 1 for Salmonella enterica serotypes Dublin and Typhimurium. There was no adsorbent treatment × Salmonella enterica serotype interaction (P = 0.696); however, there was an adsorbent treatment difference for log₁₀ reduction of both Salmonella enterica serotypes (P < 0.0001; Figure 1). Both PowerGuard and activated carbon had a greater log₁₀ reduction of both Salmonella enterica serotypes when compared to two bentonite clays. The reported electron micrograph shows Salmonella typhimurium, highlighted in yellow, adsorbed to the surface of PowerGuard (Figure 2).

Measures of performance are reported in Table 2. There was no difference among treatments for initial body weight or final body weight (P ≥ 0.201). There was a treatment × time interaction for feed disappearance (P = 0.028; Figure 3). There was no treatment or treatment × time interaction for ADG (P = 0.312). There was a tendency for a treatment × time interaction for feed disappearance to gain (P = 0.059; Figure 4); where on days 11-14 PG0.25 had decreased feed disappearance to gain when compared to both PG0.5 and CON treatments.

Hematological counts are reported in Table 3. There was a treatment difference for red blood cell count (P = 0.03), where the CON treatment had increased red blood cell counts when compared to the ST and PG0.5 treatments, and PG0.05 and PG0.25 treatments were intermediate. There was a tendency for a treatment × time interaction for hematocrit (P = 0.0837; Figure 5). There was a treatment difference for hemoglobin (P = 0.0003), where the ST and PG0.5 treatments had decreased hemoglobin concentrations when compared to all other treatments. There was a tendency for a treatment × time interaction for reticulocyte counts (P = 0.1; Figure 6). There was no treatment difference for platelet count (P = 0.163). There were treatment differences in mean corpuscular volume (MCV) and mean corpuscular hemoglobin (MCH) (P ≤ 0.019), where the ST and PG0.5 had decreased MCV and MCH when compared to all other treatments. There was a treatment × time interaction for total leukocyte count (P = 0.0074; Figure 7), where leukocyte counts were increased post-infection for the ST pigs compared to all other treatments. There was a treatment × time interaction for neutrophil count (P = 0.0026; Figure 8), where neutrophil counts were increased post-infection for the ST pigs compared to all other treatments. There was a treatment × time interaction for monocyte count (P = 0.0048; Figure 9), where monocyte counts were increased post-infection for the ST pigs compared to all other treatments. There was no treatment difference for lymphocyte count (P = 0.512).

Measures of health are reported in Table 4. There was a tendency for treatment difference in the fecal score (P = 0.087), where the ST and PG0.5 treatments tended to have increased fecal scores when compared to the CON and PG0.25 treatments. There were treatment differences for both fecal shedding in RV and TT broth (P < 0.0001), where the CON treatment shed less ST in feces than every other treatment. There was no difference among treatments for the presence of ST in either lymph node tissue or ileum tissue samples (P ≥ 0.311). There was a treatment × time interaction for intraperitoneal temperature (P < 0.0001; Figure 10), where the addition of PG to the diet attenuated the post-infection febrile response.

Ileal histology data are reported in Table 5. There was a treatment difference for villi length (P = 0.0482) where the PG0.05 treatment had the shortest villi compared to all other treatments. There was no treatment difference in crypt depth (P = 0.213). There was a treatment difference in villi length:crypt depth (P = 0.0115) where the PG0.05 treatment had a decreased ratio compared to all other treatments. There was a treatment difference for villus blunting score (P = 0.0169) where the CON and PG0.25 treatments had decreased blunting scores, ST and PG0.5 had intermediate scores, and the PG0.05 treatment had an increased villus blunting score. There was no treatment difference in villus epithelium score (P = 0.7053). There was a treatment difference for lacteal dilation score (P = 0.0383), where the ST treatment had an increased score when compared to all other treatments. There was no treatment difference for goblet cell count (P = 0.7856). There was a treatment difference for intraepithelial lymphocyte count (P = 0.0017), where the ST treatment had increased cell counts when compared to all other treatments. There were no treatment differences for lamina propria lymphocyte, neutrophil, or eosinophil cell counts (P ≥ 0.1425).

Liver histology data are presented in Table 6. There was no treatment difference in the hazy vacuolization score (P = 0.159). There was a treatment difference in connective tissue prominence score (P = 0.009), where the CON treatment had a decreased score and PG0.5 had an intermediate score compared to all other treatments. There was a treatment difference for lymphocyte foci score (P = 0.02), where the CON treatment had a decreased score, PG0.25 had an increased score, and all other treatments were intermediate. There was no treatment difference for the lymphocyte foci area (P = 0.199). There was a treatment tendency for neutrophil score (P = 0.052), where the CON treatment had a decreased score and PG0.25 had an increased score, with all other treatments intermediary.