Newborns do not cultivate a fully mature immune system until a few years after birth (42). To compensate, maternal IgG and IgA antibodies are donated from the placenta and maternal breast milk (if provided) to protect against pathogens and the development of NEC (15). Maternal IgG transfer to the fetus across the syncytiotrophoblast depends on the IgG-FcRn (neonatal Fc receptor) interaction. The expression of IgG-FcRn begins during the first trimester (12 weeks) of pregnancy and continues to rise until between 17 and 41 weeks gestation. The majority of placental IgG transfer occurs after 28 weeks of gestation. IgG levels reach 50% maternal concentration between 28 and 33 weeks gestation and will rise above maternal levels by 20%–30% at term (43, 44). It is possible that low IgG levels in preterm infants may predispose these infants to develop NEC.

In addition to the transfer of maternal IgG, transfer of maternal IgA through breast milk, also protects infants from NEC. Originating from IgA+ plasma cells in the gut and educated by gut microbiota, IgA in the intestine can bind to pathogens and aid in their clearance. The ability of bacteria to bind to IgA was negatively correlated to NEC development, and the reduced stool bacterial diversity known to precede NEC was associated with a higher amount of unbound Enterobacteriaceae (15). Taken together, this data suggests that the absence of sIgA creates higher susceptibility to infections as well as delayed gut microbiota maturation which leads to gastrointestinal inflammatory diseases such as NEC.

As the primary source of nutrition, breast milk ensures proper growth and development for newborns. Human milk is composed of micro and macronutrients, bioactive components, growth factors, antibodies, and HMOs (45). HMOs, in particular, play an important role in shaping microbiome composition and modulating neonatal immunity. HMOs act as natural prebiotics, functioning as soluble decoy receptors or antiadhesives to block the adhesion of pathogens to epithelium. They also enhance commensal growth and limit pathogen growth (46). HMOs are non-digestible sugars, composed of five basic monosaccharide units: glucose, fucose, d-galactose, N-acetylglucosamine, and sialic acid (47, 48). These monosaccharide units are joined by glycosidic linkage to generate a variety of HMOs with different functions. HMOs are indigestible in the human upper digestive tract and remain intact while in the colon. Colonic microbes secrete enzymes to utilize these HMOs as nutrition (49, 50). Many of the commensals that degrade HMOs for fuel are members of the Bifidobacterium genus, mostly beneficial bacteria for infant health. Specific examples are B. longum and B. breve that are usually prominent in the digestive tract of breastfed infants.

In addition, Bacteroides species possess an excellent capacity to metabolize dietary polysaccharides to host-derived mucus-associated glycans. A study by Sodhi and colleagues has shown that HMOs 2′-fucosyllactose (2′-FL) and 6′- sialyllactose (6′- SL) can reduce NEC severity through TLR4 inhibition (51). 2′-FL also suppresses lipopolysaccharide (LPS) induced inflammation during Escherichia coli (E. coli) invasion of intestinal epithelial cells (52). Similarly, Masi et al. found significantly lower disialyllacto-N-tetraose (DSLNT) in the maternal milk given to infants prior to NEC development (53). Furthermore, authors reported that low DSLNT in milk was also associated with a significantly lower relative abundance of Bifidobacterium sp. and higher Enterobacter cloacae in the stool of infants prior to NEC (53). Fractions of HMOs were also shown to decrease mucus penetrability and bacterial attachment by enhancing the expression of Mucin 2 (MUC2) in a mouse model of NEC (54).

Other milk factors such as casein, a highly glycosylated breast milk protein, promotes intestinal defenses by increasing goblet cell numbers, enhancing Muc2 expression, and Paneth cell activity (55, 56). Additional factors found in breast milk include lactoferrin and lysozymes that possess antipathogenic properties. Enteral supplementation of lactoferrin has been shown to decrease the likelihood of late-onset bacterial and fungal sepsis in preterm infants, but meta-analysis has shown there was no significant decrease in NEC in infants who were received lactoferrin (16). Breastmilk platelet activating factor-acetyl hydrolase (PAF-AH) potentially protects preterm newborns from NEC (57). Similarly, interleukin-10 (IL-10) found in breast milk has been found protective against developing NEC in premature infants (58). In addition to IL-10, maternal transforming growth factor beta (TGF-β) provides protection by helping to increase IgA locally in the gut (59). Growth factors found in breast milk, such as insulin-like growth factor (IGF) and epidermal growth factor (EGF), support intestinal health and may protect against the development of NEC (60–65).

Mucus is one of the first lines of intestinal host defense. Mucus is produced by goblet cells, which are found in the crypts of Lieberkühn. The colonic mucus layer is divided into two layers, an outer, penetrable layer, and an inner, impenetrable layer. This contrasts with the mucus in the small intestine (SI) which is single layered and penetrable by bacteria. A protective layer of mucus keeps bacteria in the SI away from the intestinal epithelium by antimicrobial proteins (AMPs) secreted by Paneth cells (66). Studies have found defective and a significantly lower number of goblet and Paneth cells in the SI of infants with NEC compared to NEC (67). Using HT29-MTX-E12, a mucus secreting cell line, Hall and colleagues reported that breast milk significantly lowered the adherence and internalization of NEC-associated pathogenic E. coli into the mucus compared to infant formula, suggesting that breast milk enhances mucus integrity (68). Clostridium difficile (C. difficile), a known gut pathogen, also influences mucus production and composition (69).

Antimicrobial peptides (AMPs), such as defensins, including human β-defensin-3 (hBD3), cathelicidins, C-type lectin receptors (CLRs), regenerating islet-derived protein 3, and intestinal enzymes such as phospholipase A2-IIA (PLA2) and lysozyme are expressed in the gut epithelium and provide protection for the intestinal mucosa from pathogenic bacteria either by killing pathogens or by inhibiting their growth (70, 71). In addition, AMPs are involved in the immune response and shaping the microbiome (72). Using an experimental rat NEC model, Underwood and colleagues found increased intestinal mRNA expression of the AMPs lysozyme, secretory PLA2, and pancreatic-associated proteins 1 and 3 in rats with NEC compared to either dam-fed or formula-fed rats supplemented with the probiotic bacteria Bifidobacterium bifidum (B. bifidum), suggesting that AMP induction is a mucosal response to gut inflammation in NEC (73). Another study evaluated the defensin hBD3, a small cationic antimicrobial peptide that can exert multiple protective properties on the gut. Using an animal NEC model, Sheng et al., showed that hBD3 administration decreased the incidence of NEC, reduced NEC severity (decreased pro-inflammatory cytokines, intact intestinal barrier), and increased the survival rate of the animals (74). Collectively, these studies suggest a protective role for mucus and associated AMPs in neonatal mucosal defense and intestinal barrier function in NEC.

During infection, complement proteins assist in the phagocytosis of invading pathogens by opsonization, generating inflammatory responses, and altering the activity of B and T lymphocytes (75, 76). Three different pathways—lectin, alternative, and classical—activate the complement cascade. Previous studies have reported defective complement protein activity in preterm infants (77, 78). More specifically, one study reported low complement component 3 (C3) and complement component 9 (C9), intermediates of complement pathways, in preterm infants (79, 80). C5a, a cleavage product of complement component 5 (C5), is a potent chemoattractant, anaphylatoxin, and intermediary in both the conventional and non-canonical complement pathways. C5a was reported to be substantially expressed in NEC cases and could be partially responsible for inflammation in NEC. Due to its multifaceted nature, C5a is being studied for its utility as a clinical marker for the diagnosis of neonates with NEC in conjunction with radiographic evidence of disease (81). In addition, MBL-associated serine protease-2 (MASP-2), an enzyme associated with C2 and C4 cleavage and activity, is detected in higher concentrations in the cord blood of premature children who are susceptible to NEC and is linked to a threefold increased risk of developing NEC (82, 83).

Drosophila Toll was discovered as a receptor for dorso-ventral patterning during development and was later identified as a participant in immunity against fungal infections (84). Consequently, several other homologues of Toll, named Toll-like receptors (TLRs) were discovered in mammals. TLRs sense pathogen-associated molecular pattern molecules (PAMPs) and danger-associated molecular patterns (DAMPs) through their N-terminal extracellular leucine-rich repeats (LRRs) and elicit innate immunological responses, including the production and release of inflammatory cytokines (85). So far, 10 different types of TLRs have been identified in humans and 12 in mice. TLR1, TLR5, TLR6, and TLR10 are membrane receptors that may detect extracellular ligands while TLR3, TLR7, TLR8, and TLR9 work on subcellular structures. For example, TLR9 is found on endosomes and recognizes nucleic acids derived from pathogens and self-damaged cells (85, 86). TLR2 and TLR4 are expressed on the cell membrane as well as on subcellular structures.

TLR4 is a receptor for LPS, a component of Gram-negative bacteria's outer membrane that is critical for the NEC pathogenesis (87). TLR9 binds to and is activated by unmethylated cytosine-guanine oligodeoxynucleotides (CpG ODNs) in bacterial genomes, and acts as antagonist of TLR4. Activation of TLR4 in newborn mouse epithelial cells by LPS results in undesired activation of the NF-κB pathway that leads to damage of the intestinal mucosa through production of pro-inflammatory cytokines, which is one of the hallmarks of NEC (87). Several studies have shed light on the association of TLR4 with NEC (41). Recently, Liu and coworkers have shown both TLR4 and necro apoptotic protein upregulation in both NEC patients with NEC and animal NEC models (88). Egan et al., highlighted the role of TLR4 in recruiting the inflammatory CD4⁺ Th17 cells into the intestinal mucosa via activation of cognate chemokine ligand 25 (CCL25) in NEC (89). In an another study, Colliou et al., found a commensal Propionibacterium bacterial strain named UF1 that can reduce intestinal inflammation through the reduction of Th17 cell expansion in the gut of a mouse NEC model (90). TLR4 activation significantly inhibits the β-catenin signaling that is important for enterocyte proliferation in the ileum of newborn mice, which further leads to apoptosis and can lead to NEC (91). Studies have shown that activation of TLR9 can decrease experimental NEC severity, and that TLR9 activation can inhibit TLR4 signaling via IL-1R-associated kinase M (92, 93). In addition to TLR9, NOD2 reduces NEC severity via suppressing TLR4 and genetic variants in NOD2 are associated with NEC development (94, 95).

Originating from myeloid cell lineage monocytes, macrophages (Mϕ) act as a frontline guard of innate immunity against invading pathogens. Monocytes and Mϕ have several weapons in their arsenal to tackle incoming threats. By recognizing molecular patterns via toll-like receptors (TLRs), nucleotide-binding oligomerization domain-containing proteins (NOD2), and C-type lectin receptors (CLRs,) these cells either actively engage in phagocytosis or secrete various cytokines and chemokines to alert and recruit other immune cells (96). Classical monocytes (CD14⁺CD16⁻), intermediate monocytes (CD14⁺CD16⁺), and non-classical monocytes (CD14^dimCD16⁺) are the three subsets of human monocytes. In mice, monocytes are grouped based on expression levels of lymphocyte antigen 6 complex (Ly6C⁺ and Ly6C⁻) on their cell surface (97).

Several studies have suggested that tissue infiltration and enrichment of monocyte-derived Mϕ occur during inflammation in NEC (98–100). Intestine monocyte-derived Mϕ are nonproliferative, short lived and terminally differentiated, rendering their continuous replacement necessary for homeostasis. A study by Managlia et al., revealed the significance of nuclear factor kappa B (NF-κB)-driven monocyte activation, recruitment, and differentiation in neonatal intestines in NEC (99). They concluded that NF-κB-mediated activation and differentiation of Ly6c⁺ monocytes into Mϕ and their recruitment into the intestine are critical for NEC development and disease progression. Olaloye and colleagues have identified a novel subtype of inflammatory CD16⁺CD163⁺ monocytes/Mϕ associated with infants with NEC (100). In infants with NEC, peripheral monocyte counts drop due to their recruitment to the damaged intestine (101). Following recruitment, monocytes undergo differentiation to form pro-inflammatory M1-type Mϕ (102). Monocyte-derived M1 Mϕ in humans and in animal models have been linked to the severity of NEC (102, 103). Interferon regulatory factor 5 (IRF5), a factor crucial for M1 Mϕ polarization is highly expressed in infants with NEC compared to controls. Specifically, IRF5 deficiency significantly reduced M1 polarization, inflammation, and intestinal injury in experimental NEC (103). Inflammation and intestinal cell damage caused by M1 Mϕ is linked with their high level of pro-inflammatory cytokines such as IL-1, IL-6, IL-12, chemokines (Ccl4, Ccl5), and tumor necrosis factor (TNF) production. By inhibiting M1 and promoting M2 polarization of Mϕ, heparin-binding epidermal growth factor (HB-EGF) has also been found to protect against experimental NEC (102).

As one of the most abundant immune cells (nearly 70% of total leukocytes) in human blood, neutrophils are among the first responders in the fight against potential pathogens or tissue damage/injury. Neutrophils eliminate pathogens either by recruiting a wide variety of immune cells through the secretion of cytokines, chemokines, and leukotrienes or by causing direct damage to tissue or pathogens by releasing lytic proteases and neutrophil extra cellular trap (NETs) (104). In addition to their well-documented protective role, neutrophils are also able to cause significant tissue damage through the release of reactive oxygen species (ROS) in intestinal inflammation (105).

Early neutropenia has been associated with higher odds of developing NEC (106). Interestingly, neutrophils in preterm newborns have altered immunological functions, including impaired phagocytosis. Another study by Zindl and colleagues revealed the protective role of IL-22-producing neutrophils in experimental colitis by increasing AMP production and promoting mucosal repair (107). In the context of NEC, a recent study from Mihi et al., demonstrated a protective role of IL-22 treatment in attenuating intestinal injury and enhancing epithelial proliferation in experimental NEC (108). This study also found that there was a lack of IL-22 production in preterm infants or developing mice, suggesting that immunomodulatory treatments may help protect premature infants from the intestinal inflammation seen in NEC.

As specialized antigen presenting cells (APCs), dendritic cells (DCs) serve as critical link between innate and adaptive immunity. In intestine, DCs are present in Peyer's patches, mesenteric lymph nodes (MLNs), and the colonic lamina propria to provide protection against invading pathogens. To date, several studies have highlighted the protective role of DCs in regulating the gut inflammation; however, studies investigating the role of DCs in NEC is limited. In one study, which utilizes Cronobacter sakazakii (C. sakazakii) to induce NEC in mice, Emami and colleagues have reported higher DC recruitment in mouse gut. They found that DC recruitment to the gut accelerated the destruction of the intestinal epithelium and promoted NEC onset with increased TGF-β production (109). C. sakazakii also induced pyroptosis in the intestinal epithelium and promoted NEC by induction of IL-1β and Gasdermin D (GSDMD) through TLR4/MyD88 mediated activation of the nucleotide-binding oligomerization domain (NLRP3) inflammasome (110). Another study by Nolan and colleagues investigated the role of aryl hydrocarbon receptor (AhR) signaling in DCs during experimental NEC, as this signaling pathway helps regulate intestinal immunity and homeostasis. They found that a lack of AhR signaling in DCs increased NEC-mediated intestinal inflammation, and that this effect was associated with an increase in a specific subset of macrophages in the small intestinal lamina propria (111).

Adult human intestine is made of a single layer of epithelium, covering an area of 32 m² (112). The intestinal epithelium is important for digesting food and absorbing nutrients, but it is also the largest entry port for pathogens. To provide protection against these pathogens, “as a guard of port”, complex and tightly controlled innate and acquired immunity are required. Among the many different types of immune cells involved in this protection are intraepithelial lymphocytes (IELs). IELs are positioned between intestinal epithelial cells and constantly patrol the epithelial barrier (113). IEL subsets, composed of antigen-experienced T cells, are in direct contact with enterocytes and antigens in the gut lumen. These cells are classified based on the expression of T cell receptor-γδ (TCRγδ)⁺ and TCRαβ⁺ (114). Approximately 60% of small intestinal IELs are TCR⁺ cells. γδ IEL play a crucial role in mucosal defense by regulating the production of IgA, clearing and repairing damaged epithelium, increasing production of TGF-β cytokines and by decreasing IFN-γ and TNF-α in response to stress and infection (115). The protective role of IELs is also evident in TCRγδ-deficient mice, as these mice have defective gut epithelial morphology and impaired IgA production (116). When compared to non-NEC controls, surgical NEC patients with NEC had a lower number of γδ IELs in the ileum (116). Researchers have shown that subsets of IELs are dependent on AhR activation for their survival (117). However, a recent study did not find any involvement of IELs in AhR activation-mediated protection against NEC, indicating that the protective role of IELs against NEC is not AhR-mediated (118).

In addition to IELs, infants with NEC also have altered functions of some subsets of CD4⁺ T cells, Th17, and regulatory T (Treg) cells (89, 119–121). Th17 cells are strongly implicated in intestinal inflammation and are linked with the pathogenesis of NEC. In infants with NEC, Pang and colleagues found a lower percentage of Foxp3-expressing Tregs with several functional defects, including the inability to block IL-17 expression (121). In NEC tissue, Th17 cells appear to cause intestinal damage that is reduced by IL-17 receptor inhibition by STAT3 activation (122). Additionally, retinoic acid-induced polarization of CD4⁺ T cells towards Treg from Th17 resulted in reduced NEC severity (123). Furthermore, Zhao et al. reported an increased percentage of RORγt⁺ cells (inflammatory Th17 and type 3 innate lymphoid cell populations) in the intestinal lamina of mice and humans with NEC compared to those without NEC (84). Studies have also demonstrated a significant decrease in lamina propria associated Treg cells in surgical NEC specimens (85, 86, 89). In addition, a Treg/Th17 imbalance leads to the excessive proinflammatory response preceding tissue injury and necrosis associated with NEC development (122).

With the high prevalence of NEC, the need for effective in vivo models has become more important in recent years. Due to the aggressive nature of the disease and the scarcity of available human specimens, performing experiments with human samples is difficult and multi-center studies are typically needed (138). As a result, animal models are commonly used to study NEC by inducing inflammation that mimics the intestinal damage seen in human infants.

While the conditions of in vivo experimental NEC models are generally based on similar underlying principles, several different animals have been used to study NEC (Figure 2). The rat's intestinal development is similar to a human premature infant, making it an excellent model for investigating preventative measures and therapeutics for NEC (139). Early studies using a rat model concluded that the gut microbiota and the absence of breast milk are significant factors in NEC pathogenesis (140, 141). Further, several laboratories have used hypoxia, LPS, and hypothermia at different time points in a day for several days to help induce NEC in laboratory settings (142). Due to their affordability, preterm survivability, and resistance to typical stressors used to develop the disease, rat models are a desirable option when investigating NEC but rats are not ideal for research at the genomic level. Their slower development and challenges with culturing embryonic stem cells in rats makes it difficult to generate transgenic lines compared to mice (143, 144). These shortcomings necessitated the creation of other types of animal NEC models.

Although their small size makes them technically challenging to work with, mice are the preferred model for genomic studies as it is far easier to create transgenic colonies. Another appealing feature of the mouse model is its experimental flexibility, with some models successfully inducing NEC by beginning the gavage feed at postnatal day 4 while others begin at postnatal day 7 (145, 146). However, mice delivered more than one day prior to the determined due date have a 100% mortality rate (147). Because of this low viability, it is extremely difficult to use a preterm mouse model for studies that require animals to be delivered via cesarean section.

Pigs share many features of anatomy and physiology with humans, rendering them one of the more popular choices when exploring NEC pathogenesis. Additionally, the piglet's larger size affords the ability to study preterm neonates (148, 149). Piglets are a good model for testing preclinical drugs, effects of various diet formulations, and pathological manifestation on NEC (150). While it is true that hypoxia and hypothermic stress induces histological changes that resembled NEC in piglet models, the inflammation induced by this model is not always contained within the lower gastrointestinal tract, with some instances reported of inflammation spreading to the stomach and jejunum (139, 150–152).

Rabbit NEC models are infrequently used but have been used to study the effects of NEC that extend past the gut. Non-human primate models, although rare and expensive, have also been used as an experimental NEC models due to the homology to humans in both anatomy and at the genomic level (139).

In vivo animal models allow for limited NEC modeling as the cellular genetics, drug metabolism, immunology, gut microbiomes, and HMOs can differ significantly from humans. In vitro intestinal models used to study NEC are briefly summarized in this section and have been covered extensively elsewhere (153–156).

Different in vitro models such as the human epithelial cell line Caco2, colon adenocarcinoma derived cell line HT-29, IEC-6 and IEC-18 derived from rat SI, and most importantly, fetus derived FHs 74-Int and H4 cells are frequently used in in vitro NEC studies (153). These cell lines are optimized and phenotypically mimic different regions of the gut including ileum, duodenum, jejunum, and colon, each requiring specific culturing conditions.

Recent scientific advancements in culturing human intestinal organoids (enteroids) also called “mini guts”, allow investigators to recapitulate the intestinal cell morphology that is crucial for studying the molecular mechanisms of NEC. Enteroids derived from LGR5⁺ progenitor cells of the SI and colon, allow for the study of barrier function, gut inflammation, cell proliferation, drug responses, and intestinal microbial interactions characteristic of NEC (157). Further advancements of in vitro models led to the development of a “gut-on-a-chip”, a method which cultures intestinal cells to mimic the microenvironment of the intestine (158, 159). The gut-on-a-chip model provides a suitable environment to culture different human cell types including epithelial, endothelial, and immune cells with gut microbes together in a controlled environment, to explore gut physiology and inflammatory changes seen in NEC, and can also be used as a pharmacological platform to test potential drug treatments (160).

Though, these in vitro models excellently resemble human intestine, several key criteria are considered in cell culture model design. Table 1 compares common different models and devices, specifically summarizing whether the models are static or microfluidic, in vitro or ex vivo, cell differentiation, cell polarity (apical out or basal out), nutrient absorption, drug metabolism, crypt villus formation, mechanical stimulation or peristalsis, oxygen gradient modulation, measure trans epithelial electrical resistance (TEER), co-culture with endothelial, vascular, and immune cells, and co-culture with gut microbes.

Static models are standard tissue culture models which include “NEC-on-a-dish” 2D, 3D organoid and transwell culture models (175). Additionally, synthetic scaffolds, and ex vivo tissue (Ussing chambers) are used to measure live tissue (167, 168). Static models use growth factors to differentiate intestinal epithelial cells (IECs) and organoids, derived from LGR5⁺ progenitor cells, into diverse functional intestinal cells (163). Static models are generally less time consuming, less expensive, and more accessible, but are relatively limited to the degree of differentiation, co-culture, and microbiome interactions. Typically, in static models, microbiome interactions are limited to between 1 and 24 h based on the model due to rapid microbial overgrowth in static conditions.

Gut-on-a-chip microfluidic devices use soft lithography to layer polydimethylsiloxane (PDMS) or micromilling to produce luminal and vascular channels separated by a porous membrane (reviewed in (176). Short term ex vivo microfluidic devices can evaluate live tissue conditions under constant flow (169, 170). The luminal flow in a microfluidic model enhances differentiation and 3D villus and crypt-villus like topography where adjacent air channels are regulated to mimic peristalsis through mechanical stimulation, thus providing a major advantage over static models. The NEC microbiome and HMO interactions, drug metabolism, and tissue integrity assays can be measured within the microfluidic chip system (177, 178). A major advantage of the microfluidic flow is that it reduces the static overgrowth of microbes, in turn reducing the limitations on the microbial co-culturing time to more than 7 days, depending on the specifics of the model. Gut-on-a-chip models can additionally be cultured under oxygen gradient modulation. Intestinal disease pathology is increased by lower oxygen gradients which induce Hif1-α signaling (179). Oxygen gradients under aerobic, hypoxic, and anaerobic culturing conditions have also been applied to resemble microbial intestinal environments under inflammatory conditions (176).