All animal studies were carried out in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) of the Research Institute at Nationwide Children's Hospital (protocol #AR18-00062). White Yorkshire × Landrace sows were purchased from Oak Hill Genetics (Ewing, IL). Sows were acclimatized to our facility for five days before terminal cesarean section delivery, performed on gestational day E104, with full term gestation being E114. Piglets delivered at E104 are roughly equivalent to premature infants born at 32 weeks gestation (16).

Sows were sedated with an intramuscular injection of telazol (0.4–1.0 mg/kg), ketamine (1–2.5 mg/kg), and xylazine (1–2.5 mg/kg), followed by an intramuscular injection of glycopyrrolate (0.01 mg/kg). After endotracheal intubation, anesthesia was maintained throughout the cesarean section using 1%–4% isoflurane. After induction of adequate anesthesia, laparotomy was performed through a lower midline incision to expose the uterus and deliver the piglets. The umbilical cord of each piglet was milked to ensure adequate placental blood transfusion prior to cord ligation. The time from induction of anesthesia to piglet delivery was kept at a minimum to avoid excess exposure of the premature piglets to anesthesia. After all piglets were delivered (typically within 10–15 min after skin incision), the sow was euthanized with IV or intracardiac Euthasol® (1 ml/4.5 kg) and thoracotomy, and the laparotomy closed.

Piglets were immediately dried and resuscitated after birth by clearing the remaining mucous from the airways using suction and providing positive pressure bag-mask ventilation as needed. Once piglets showed signs of independent breathing, 1 ml of iron dextran (Vedco, Saint Joseph, MO) was administered intramuscularly to prevent iron deficiency anemia, one drop of sublingual Doxapram (WEST-WARD, Eatontown, NJ) was administered to stimulate respiration, and sublingual glucose paste (Insta-Glucose, Valeant Pharmaceuticals, Bridgewater, NJ) was provided to prevent hypoglycemia. Doxapram was re-dosed as needed to piglets that were slow to breathe independently and require additional stimulation. Once resuscitated, piglets were placed in temperature- and oxygen-controlled small animal intensive care units (Suburban Surgical Co, Wheeling, IL) to maintain a temperature of 35.5°C–39.5°C with FiO2 of 40%. Once stabilized, typically within 60 min, each piglet was weighed and sexed. A 6-French transbuccal orogastric feeding tube (Cardinal Health, Dublin, OH) was introduced into the stomach through a small puncture made with the needle in the cheek. 3-0 polypropylene sutures were used to secure the feeding tube to the cheek, snout, and forehead, with additional securement of the tubes using Elastoplast tape. After tube placement was confirmed by auscultation, piglets received 10 ml of dextrose (25 g) suspended in 500 ml Pedialyte (Abbott, Columbus, OH). Piglets from 4 separate experiments were randomized regardless of sex into two different experimental groups—the colostrum-fed group (total n = 17) and the formula-fed group (total n = 18) (Supplementary Figure S1).

Piglets were fed every 3–4 h for a total of 6–8 feeds/day. The initial volume goal of colostrum or formula feeds was 240 ml/kg/day and was decreased to 120 ml/kg/day in later experiments to avoid potential over-hydration. Oral glucose paste was administered if there was concern for hypoglycemia. Feeding volumes were scaled up over the first two days. 60% of goal volume was administered as colostrum or formula on day 1, with the remaining 40% of the goal volume achieved with Pedialyte supplementation. On day 2, 75% of goal volume was administered as colostrum or formula, with the remaining 25% of the goal volume achieved with Pedialyte. On days 3 to 5, the complete goal volume was administered as colostrum or formula.

Fresh frozen bovine colostrum (BC) was obtained from OSU Waterman Farms (Columbus, OH). The colostrum provided was medium grade in quality, between 20 and 50 mg/ml of immunoglobulins, as assessed on the farm by density measurement using a Brix Refractometer. The colostrum was diluted to a 50% concentration with distilled autoclaved water upon thawing since higher concentrations led to gastric bezoar formation. Piglets in the colostrum-fed group received colostrum for the entire experiment. In contrast, piglets in the formula-fed group received fresh BC for 24 h, followed by Neocate Jr (1.0 kcal/ml; 120 kcal/kg/day) (Nutricia, Zoetermeer, Netherlands) for the remainder of the experiment to induce intestinal injury. All milk products were prepared daily and stored in a 4°C refrigerator. Aliquots were taken before each feed and warmed for 20–30 min prior to administration.

IgG is the major immunoglobulin present in BC (32, 33), and its use as a marker for BC quality is well-documented in the literature (32–36). To further assess colostrum quality, samples were thawed and centrifuged four times at 10,000 g for 20 min each at 4°C to remove cells and fat. BC IgG was quantified using a bovine IgG ELISA kit (Bethyl Laboratories Inc., Montgomery, TX) following the manufacturer's instructions with minor modifications. Samples were diluted to 1:500,000 and plated in triplicate on 96-well plates using 3,3′,5,5′tetramethylbenzidine (TMB) as substrate. Absorbance OD_{450 nm} was measured using a Spectramax M2 microplate reader equipped with SoftMax Pro 5.4 software (Molecular Devices, San Jose, CA). IgG concentrations were calculated using standard curves.

Piglets were monitored continuously throughout the experiment. Although unblinded to groups, we noted observable differences in the piglets in the two different groups. Thus, we initiated a new scoring system, the clinical sickness score (CSS), to define our observations. CSS were assigned at the time of each feed, beginning at 12 h of life. The CSS was comprised of 4 different categories: motor/tone, verbal, alertness, and body color, with scores ranging from 0 to 2 per category depending on severity, for a maximal total CSS of 8 (Table 1). This was reported as an average CSS for all piglets in each group at each time point. If a piglet was euthanized prior to the end of the experiment, that piglet was given a score of an 8 for all remaining time points of the experiment.

In alignment with our IACUC protocol, piglets were euthanized if they lost more than 25% of their birth weight, had a single temperature >40.5°C, or were noted to have extreme lethargy. Additionally, piglets were euthanized if two of the following were observed during two consecutive feeds: loss of >20% of birth weight, significant lethargy/decreased activity, temperature >40°C, bloody diarrhea, abdominal distension, emesis, or rapid, labored, or shallow breathing. Euthanasia was performed using an intramuscular injection of ketamine (0.3–0.6 ml/kg) and xylazine (0.1–0.2 ml/kg), followed by an intraperitoneal injection of Euthasol® (1–2 ml/kg).

After euthanasia, a midline laparotomy was performed, and the intestine examined in situ. A gross injury score was assigned to the small intestine (proximal, mid, and distal segments) and colon based on the most severely affected area assessed independently by two graders using an adaptation of a published scoring system (16, 37). Scores ranged from 1 (no injury) to 6 (most severe injury) using the criteria shown in Table 2. The highest overall score assigned to any segment was used to assign the animal to either no gross injury (grades 1–3) or gross injury (grades 4–6) (Figures 1A,B) (16, 37).

Tissue was collected from the small intestine (proximal, mid, and distal) and colon, fixed in 10% formalin (Fisher Scientific, Pittsburgh, PA) for 24 h, followed by 70% ethanol (Fisher Scientific, Waltham, MA). Samples were paraffin-embedded and stained with hematoxylin and eosin (H&E). Histologic sections were reviewed by light microscopy by at least two independent graders, including a board-certified pediatric pathologist, in a blinded fashion. Histologic injury scores were assigned based on a previously published grading system (28). Scores ranged from 1 (no damage) to 5 (transmural necrosis) (Table 3). The final score assigned to any animal was the highest score identified in any intestinal segment, with scores of 3 or above consistent with NEC. Scores were categorized as no histologic injury (grades 1–2), moderate histologic injury (grades 3–4), or severe histologic injury (grade 5) (Figures 1C–N). Images were taken using an Olympus U-Tv0.5XC-3 microscope (Tokyo, Japan).

Liver, spleen, and mesenteric lymph nodes were collected in a sterile fashion at the time of necropsy, snap frozen, and stored at −80°C until use. Samples were thawed on ice, weighed, and homogenized in 2 ml of PBS before plating 50 µl of the homogenized mixture on brain heart infusion (BHI) plates for overnight, aerobic incubation at 37°C. BHI is a general-purpose growth medium for culturing non-fastidious Gram-positive and Gram-negative bacteria (38). The number of colony-forming units (CFUs) for each organ was recorded after 24 h of growth and standardized per gram of tissue.

Piglets were determined to have D-NEC if they met at least 3 of the following four criteria: Gross injury score ≥4 of a maximum of 6, histologic injury score ≥3 of a maximum of 5, CSS ≥5 of a maximum of 8 in the last 12 h of life, and bacterial translocation of ≥1 CFU/g of tissue to two or more internal organs (liver, spleen, mesenteric lymph nodes).

Intestinal specimens of proximal, mid, and distal small bowel and colon were collected during necropsy and stored at −80°C until use. Total RNA was isolated using the Purelink RNA mini kit (Thermo Fisher, Waltham, MA) according to the manufacturer's instructions. Samples were thawed on ice and weighed. Tissues (0.5–0.9 g) were homogenized in lysis buffer with 1.0 mm Zirconia beads (Biospec Products, Bartlesville, OK) in a TissueLyser II (Qiagen, Germantown, MD) at 30 Hz for 1 min, repeated six times. cDNA was synthesized using the Superscript IV VILO cDNA synthesis kit with ezDNAse Enzyme to remove genomic DNA (Thermo Fisher, Waltham, MA). Quantitative real-time PCR (qRT-PCR) was performed using primers for interleukin-1α (IL-1α) and IL-10, and PowerUp SYBR Green PCR Master mix (Thermo Fisher, Waltham, MA) using a QuantStudio™ 3 System (ThermoFisher, Waltham, MA) (Table 4) (39–41). Hypoxanthine-guanine phosphoribosyltransferase (HPRT) (Integrated DNA Technologies, Coralville, IA) was used as an endogenous control The cycling program consisted of a 2-minute hold stage at 50°C, a 10-minute hold stage at 95°C, followed by 40 cycles of 15 s at 95°C followed by 1-min at 60°C. Target gene expression was analyzed using the 2^−ΔΔCt method.

Intestinal contents were aseptically collected from the colon of all piglets at sacrifice and frozen at −80°C until use. Approximately 100 mg of colonic content was subjected to DNA extraction using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) per manufacturer's instructions with slight modifications. Contents were incubated for 45 min at 37°C in lysozyme-mutanolysin buffer (pH 8.0) containing 22 mg/ml lysozyme, 0.1 U/ml mutanolysin, 20 mM TrisHCL, 1.2% Triton-x (Sigma Aldrich, St. Louis, MO), and 2 mM EDTA (Thermo Fisher Scientific, Waltham, MA), followed by homogenization for 150 s with 0.1 mm zirconia beads. Samples were incubated at 95°C for 5 min with InhibitEx Buffer, then incubated at 70°C for 10 min with Proteinase K and Buffer AL (Qiagen). Following this step, the QIAamp DNA Mini Kit isolation protocol was followed beginning with the ethanol step. DNA was quantified with the Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA) using the dsDNA Broad Range Assay Kit and stored at −20°C prior to 16S rRNA gene sequencing.

Colonic content DNA samples were submitted to the Genomic Services Core at the Institute for Genomic Medicine at Nationwide Children's Hospital (Columbus, OH) for library preparation and high-throughput sequencing. Paired-end (300 nt forward and reverse) sequences of the V4 hypervariable region of the 16S rRNA gene (515F-806R) were generated by Illumina MiSeq. Quantitative Insights Into Microbial Ecology (QIIME) 2.0 (42) and DADA2 (43) were utilized for downstream amplicon processing, quality control, diversity analyses, and taxonomic assignment using a trained classifier built from the ribosomal RNA database SILVAv138 (44, 45). For quality control, sequences were truncated from the 3′ end to 280 nt (forward reads) or 250 nt (reverse reads) to achieve a quality score of at least 20, and the first 20 nt were trimmed from the 5′ end of both forward and reverse reads as these early positions may contain errors. Sequences that did not meet these criteria were discarded. After quality control, taxa denoted as eukaryotic or “unassigned” were also removed. Subsequent diversity analyses were carried out at a sequencing depth of 18,755 sequences per sample. Evenness, richness (number of observed species), and Shannon's Diversity Index (46) were calculated to measure alpha diversity within samples on the basis of feeding regimen and occurrence of NEC. The weighted UniFrac distance matrix (47) was applied to measure beta diversity, and the EMPeror software package (48) was used to construct three-dimensional principal coordinate (PCoA) plots to visualize differences based on feeding regimen and occurrence of NEC. Relative abundance was calculated for the class, family, and genus taxonomic levels on an individual sample basis as well as abundances within feeding regimen and occurrence of NEC.

All statistical analyses were performed using GraphPad Prism, Version 9 (GraphPad Software, Inc. La Jolla, CA). Nonparametric analysis between groups was performed using the Mann–Whitney U-test. Survival data were analyzed using a log-rank (Mantel-Cox) test.

Differences in bacterial community composition were assessed by comparing alpha and beta diversity between feeding regimens (i.e., colostrum fed vs. formula fed) and between pigs that developed NEC vs. those with No NEC collapsed across the two feeding regimens. Evenness, richness, and Shannon Diversity Index alpha diversity metrics used the Kruskal-Wallis test for pairwise comparisons. Differences in beta diversity weighted UniFrac distances were analyzed by permutational multivariate analysis of variance (PERMANOVA) with 999 randomizations of the data. Relative abundance data were analyzed using the Kruskal-Wallis test for pairwise comparisons, and taxa that were present in fewer than 10% of samples (present in fewer than 3/26 samples) were omitted from abundance analyses. The statistical alpha level was set to 0.05 for all analyses.

In preliminary experiments, piglets that were exclusively fed with commercial human formulas were found to have a rapid deterioration of their health with early death within the first 24 h of life. We investigated the use of Sow Colostrum Replacer (APS LaBelle, Phoenix, AZ), but this had a minor effect on preventing early death. Unlike infants, piglets do not undergo maternal-placental transfer of immunoglobulins and instead depend almost exclusively on postnatal transfer of maternal IgG via colostrum (13, 49–52). We ultimately chose bovine colostrum because of its high levels of beneficial immunoglobulins to help protect against early animal mortality, and human breast milk or porcine colostrum were not available in the volumes needed. Piglets subsequently received BC for 24 h prior to the introduction of formula feeding to induce NEC. This led to a significant improvement in early mortality, however, the effect was variable depending upon the lot of BC used. Based on this, we investigated the IgG content of the different BC lots by ELISA. Consistent with previous reports, we found that the efficacy of the colostrum in preventing early death was dependent upon its IgG content (33). We measured the IgG content in 8 different batches of colostrum from 7 sows. The range of IgG varied widely, from 1.123 mg/ml to 94.898 mg/ml. We found that colostrum with IgG levels greater than 5 mg/ml reduced early death (data not shown). Thus, for all subsequent experiments, we used colostrum that contained a minimum of 5 mg/ml of IgG.

In addition to the use of BC, we found that rapid piglet delivery to decrease anesthesia exposure time, the use of oxygen and temperature-controlled incubators, and the administration of iron dextran to prevent iron deficiency anemia, doxapram to simulate respiration, and sublingual glucose to prevent hypoglycemia significantly improved piglet survival during the first 24 h of life.

Piglets in the formula-fed group had increased all-cause mortality compared to piglets in the colostrum-fed group, regardless of sex (Figure 2A). Furthermore, piglets in the formula-fed group lost significantly more weight compared to piglets in the colostrum-fed group (−9.9% vs. +5.3%, p = 0.0001) (Figure 2B) and had significantly higher CSS than piglets in the colostrum-fed group (Figure 3A).

Upon necropsy, we found that 61% (11 of 18) of animals in the formula-fed group had gross evidence of intestinal injury consistent with NEC, compared to 41% (7 of 17) of animals in the colostrum-fed group (Figure 3B). The highest degree of gross injury in both groups was seen in the mid-small bowel, with an average gross injury score of 3.3 in the formula-fed group and 2.5 in the colostrum-fed group. The lowest degree of gross injury in both groups was seen in the proximal small bowel, with an average score of 1.9 in the formula-fed group and 1.7 in the colostrum-fed group (Supplementary Figure S2). While these findings were not statistically significant, average gross injury in all sections of bowel was greater in the formula-fed group than the colostrum-fed group.

Histological examination of the intestine revealed that 78% (14 of 18) of animals in the formula-fed group had histologic intestinal injury consistent with NEC, with 56% having moderate NEC and 22% having severe NEC (Figure 3C). 71% (12 of 17) of animals in the colostrum-fed group had histologic injury consistent with NEC, with 59% having moderate NEC and 12% having severe NEC. In both groups, the mid-small bowel was found to have the highest degree of histologic injury, with an average score of 3.2 in the formula-fed group and 2.9 in the colostrum-fed group (Supplementary Figure S2). While these findings were not statistically significant, average histologic injury in all sections of bowel was greater in the formula-fed group than the colostrum-fed group.

Since the intestinal injury that occurs with NEC results in gut barrier failure, we next assessed intestinal barrier function by detecting bacterial translocation to the liver, spleen, and mesenteric lymph nodes. Bacterial translocation was detected in 89% of piglets in the formula-fed group compared to only 35% of piglets in the colostrum-fed group (p = 0.0016) (Figure 3D).

D-NEC criteria were met if a piglet met at least 3 of the following 4 criteria: gross injury score of ≥4, histologic injury score of ≥3, CSS of ≥5 in the last 12 h of life, or bacterial translocation to ≥2 internal organs (liver, spleen, mesenteric lymph nodes). 77% (12 of 18) of animals in the formula-fed group had D-NEC compared to 24% (4 of 17) of animals in the colostrum-fed group (p = 0.0176) (Figure 4A). This was not affected by the sex of the piglet.

Piglets in the formula-fed group tended to die during the experiment due to D-NEC more frequently than piglets in the colostrum-fed group (33.3% vs. 11.8%) (Figure 4B). qRT-PCR was performed on samples of the mid and distal small bowel and colon from 10 piglets in each group. There was significantly increased mRNA expression for IL-1α (p = 0.0115) and IL-10 (p = 0.0005) in colon specimens from the formula-fed group compared to the colostrum-fed group (Figure 4C).

To determine the influence of the feeding regimen and the impact of NEC on colonic microbial diversity within samples, alpha diversity was assessed based on evenness, richness, and Shannon Diversity Index, which accounts for both evenness and richness. Formula-fed piglets had a lower mean evenness score than colostrum-fed piglets (p = 0.054) (Figure 5A), as well as a significant decrease in richness (p = 0.002) (Figure 5B). Consistent with the observations of evenness and richness, the overall diversity within samples as denoted by the Shannon Diversity Index was significantly lower in the formula-fed relative to the colostrum-fed piglets (p = 0.003) (Figure 5C). When collapsed across feeding types, piglets that developed NEC during the experimental period had significantly reduced evenness (p = 0.012) and richness (p = 0.038) within colonic microbial communities relative to piglets that did not develop NEC (Figures 5D,E). Similarly, the Shannon Diversity Index was significantly reduced in piglets that developed NEC (p = 0.006) (Figure 5F). Analogous significant results were obtained when other alpha diversity measures were evaluated, including ACE, and Simpson Diversity Indices, but not Faith Phylogenetic Diversity (Supplementary Figure S3).

The weighted UniFrac distance matrix was constructed to characterize changes in the community and phylogenetic composition. Both feeding regimens (p = 0.019) and NEC (p = 0.002) were associated with significant changes in colonic content beta diversity. Importantly, a notable amount of clustering was associated with a positive NEC diagnosis (Figure 6). Thus, feeding regimen and NEC exerted independent influences on the bacterial composition within the colonic environment resulting in phylogenetically distinct microbial profiles. Comparable results were observed upon assessment of additional beta diversity metrics, including Bray-Curtis dissimilarity and Jaccard distances, but no differences were observed for the unweighted UniFrac distance matrix (Supplementary Figure S4).

The relative abundances of specific bacterial taxa (at the class, family, and genus levels) were compared between the formula-fed and the colostrum-fed piglets, and between piglets that had or had not developed NEC (collapsed across the two feeding regimens). Taxa relative abundance bar plots were also generated at each of the previously described phylogenetic levels on an individual sample basis for feeding regimen (Supplementary Figures S5A–C) and occurrence of NEC (Supplementary Figures S6A–C).

Class level taxa were primarily dominated by Bacilli, Gammaproteobacteria, and Clostridia (Figures 7A,B). While there were no differences in the relative abundances of bacterial classes based on feeding regimen, there was a significant increase in the relative abundance of Gammaproteobacteria in piglets with NEC relative to piglets without NEC (p = 0.038) (Figure 7C). Enterobacteriaceae, Clostridiaceae, and Enterococcaceae predominated at the family classification level (Figures 8A,B). However, significant changes in the relative abundances of bacterial families occurred based on the two experimental factors (Figures 8C,D). The relative abundance of Lactobacillaceae was surprisingly higher in formula fed piglets than in colostrum fed piglets (p = 0.017). In contrast, colostrum-fed piglets had significantly increased relative abundance of Lachnospiraceae (p < 0.001), Peptostreptococcaceae (p < 0.001), Ruminococcaceae (p = 0.030), Moraxellaceae (p = 0.040) and Pseudomonadaceae (p = 0.001), (Figure 8C). Further, Lachnospiraceae (p = 0.013) and Peptostreptococcaceae (p = 0.047) were significantly more abundant in piglets without NEC. However, in piglets that developed NEC, Clostridiaceae (p = 0.029), and, importantly, Enterobacteriaceae (p = 0.01) abundances were significantly elevated relative to healthy piglets (Figure 8D). Similar to family level taxonomy, the most prevalent genera were similar when assessing based on feeding regimen or occurrence of NEC (Figures 9A,B). However, the relative abundances of bacterial genera differed between the two factors (Figures 9C,D). While ten genera were significantly different based on feeding (Figure 9C), the most notable was Lactobacillus (p = 0.017), which was elevated in formula-fed piglets, thus recapitulating family findings. Epulopiscium (p < 0.001), Pseudomonas (p = 0.001), Pygmaiobacter (p = 0.030), and Lactococcus (p = 0.001) were increased in colostrum-fed piglets (Figure 9C). Epulopiscium was also significantly higher in piglets without NEC (p = 0.013) (Figure 9D). In piglets that were positive for NEC, there was a significant increase in the abundance of Clostridium sensu stricto 1 (p = 0.025), Clostridium sensu stricto 2 (p = 0.021) and Clostridium sensu stricto 13 (p = 0.017) (Figure 9D). Additionally, an undefined genus of Enterobacteriaceae was increased in NEC-positive piglets with a trend towards significance (p = 0.056). Collectively, taxonomic abundance data indicate that colostrum and formula feeding differentially influence pioneer colonization within the colonic environment. Additionally, this pioneer colonization may, to an extent, also affect the distribution and prevalence of taxa seen here to be associated with the onset of NEC.