The conceptual framework for a gut–brain axis has existed for decades. The Human Microbiome Project is responsible for establishing intestinal dysbiosis as a mediator of inflammatory bowel disease, obesity, and neurodevelopmental disorders in adults. Recent advances in metagenomics implicate gut microbiota and diet as key modulators of the bidirectional signaling pathways between the gut and brain that underlie neurodevelopmental and psychiatric disorders in adults. Evidence linking intestinal dysbiosis to neurodevelopmental disease outcomes in preterm infants is emerging. Recent clinical studies show that intestinal dysbiosis precedes late-onset neonatal sepsis and necrotizing enterocolitis in intensive care nurseries. Moreover, strong epidemiologic evidence links late-onset neonatal sepsis and necrotizing enterocolitis in long-term psychomotor disabilities of very-low-birth-weight infants. The notion of the gut–brain axis thereby supports that intestinal microbiota can indirectly harm the brain of preterm infants. In this review, we highlight the anatomy and physiology of the gut–brain axis and describe transmission of stress signals caused by immune-microbial dysfunction in the gut. These messengers initiate neurologic disease in preterm infants. Understanding neural and humoral signaling through the gut–brain axis will offer insight into therapeutic and dietary approaches that may improve the outcomes of very-low-birth-weight infants.
The concept of gut–brain axis (GBA) has existed for more than three decades (1). Contributions from many scientific disciplines have now resulted in the acceptance of a multicomponent bidirectional signaling system between the gut and the brain (2). The initiation of the Human Microbiome project (3) has been associated with recent and rapid advances in GBA-related research (4,5). We propose that the GBA plays a vital role in adverse neurodevelopmental outcomes in preterm infants. This review focuses on the impact of GBA in the health and disease of preterm infants, although clinical research on the GBA is sparse in this population. Hence, we include GBA-related investigations in neonatal animals using relevant experimental models that allowed us to highlight gaps in knowledge that require additional investigation.
Gastrointestinal motor and sensory components send messages to the central nervous system and the return response to the intestine is the definition of the brain–gut axis (6). Figure 1 shows the complexity and the modifiers of GBA-related signaling pathways and the multiorgan connections that influence the balance of health vs. disease in the preterm newborn. The brain is the central component of the GBA and includes connections between the cerebral cortex, the limbic system, the hypothalamic–pituitary axis, and the brain stem. The limbic cortex regulates olfaction and integrates sensory and motor functions. The limbic system receives input from other brain regions responsible for a range of behaviors. Maternal deprivation and pain studies in neonatal mice (7,8) indicate that the limbic system plays an important role in development of the hippocampus. In preterm infants without intracranial hemorrhage or ischemia, injury to the hippocampus is likely responsible for neurobehavioral disorders seen in preterm infants during childhood (9). When studying insults to the limbic system, researchers must consider gender-specific effects of stress in human preterm infants (10).
The peripheral components of the GBA communicate with the central nervous system through the enteric, autonomic, and sympathetic nervous systems (11,12). Evidence that gut microbes modulate neural signaling indicates that they may alter brain development and function (12) via the enteric nervous system (ENS). The ENS resides within intestinal wall and abnormalities in the ENS are associated with a wide spectrum of gastrointestinal disorders in adults (13). The GBA and neonatal diseases section discusses newborn diseases associated with the ENS. The ENS communicates with the brain via the vagus nerve and dorsal root and nodose ganglia (11). The hypothalamic–pituitary axis, the autonomic nervous system, and the sympathetic nervous system are integrated peripheral components of the GBA (14). The afferent vagus nerve is a major retrograde signaling system from gut to brain (15). The efferent vagus nerve-based cholinergic anti-inflammatory pathway is known to regulate the balance of tumor necrosis factor-alpha, high mobility group box 1 (a nonhistone nuclear protein), and other cytokines secreted by macrophages in response to stress signals in the gut (16). This inflammation can result in the loss of intestinal epithelial barrier function allowing bacterial invasion (17). A concurrent increase in intestinal permeability leads to activation of immune and somatic cells through pathogen-associated molecular pattern recognition receptors that trigger inflammation in the gut (17). Signals sent through the systemic and intestinal immune system via the GBA cause alterations in brain function and disease (18).
The ENS, autonomic nervous system, and the hypothalamic–pituitary axis transduce neural and humoral signals about nutrients and microbes in the gut to the brain and the limbic system (18). The hypothalamus possesses reciprocal connections between the higher cortical centers such as the reward-related limbic system and the brain stem (19). During a stressful state, hormone and neuropeptide secretion in the gut ultimately invokes cortisol release from the adrenal gland via signals through the hypothalamus. Gut hormones, namely peptide YY, pancreatic polypeptide, glucagon-like peptide-1, oxyntomodulin, are components of the GBA associated with eating and satiety, while fasting increases ghrelin secretion (19,20). The GBA also influences intestinal immune cells with neuropeptide messages. Specific neuropeptides, namely vasoactive intestinal peptide and norepinephrine, modulate functions of dendritic cells and T cells located throughout the wall of the intestine and in secondary lymphoid tissues like Peyer’s patches (21). Depending on the balance of neuropeptides and other immunomodulatory molecules, dendritic cells orchestrate the differentiation of distinct effector lymphocytes populations by either triggering inflammatory or anti-inflammatory (tolerogenic) responses to microbial and/or dietary antigens. Finally, adrenergic stimulation of intestinal enterocytes through the sympathetic nervous system can alter gut microbial composition and the change can result in a loss of barrier function and increased gut permeability (12).
GBA Interactions with Gut Microbiota, Epithelia, and Immune Cells
This subject considers the interactions between the gut microbiota, intestinal epithelia, and the immune system related to physiology in the GBA (Figure 2). Since they are in close proximity, gut microbiota can stimulate and regulate gut epithelia, intestinal immune cells and tissues, and the ENS (32,33,34). Publications involving preterm infants are limited on this topic; however, a review describes the important role played by intestinal microbiota in the postnatal development of gastrointestinal functions (35). Recent studies suggest that postnatal events, such as antibiotic administration, modify neonatal gut microbiota and are associated with obesity in infancy and childhood (36). Strict antibiotic stewardship is mandatory in neonatal intensive care units until the links between postnatal antibiotics and future obesity have been verified (37). The basis for an association between gut microbiota and obesity is increased energy harvest (38). Studies in obese vs. lean twins show phylum-level alterations in the gut microbiota, including reduced bacterial diversity, an altered representation of bacterial genes in the microbiome and abnormal metabolic pathways suggesting that these differences account for the lean vs. the obese state (39). Obese individuals have fecal microbial biomarkers related to their diet, and these microbiota correlate with impaired metabolic health (40). Intestinal microbes can create an environment that promotes energy salvage from the diet. For example, Faecalibacterium prausnitzii is a bacterium that promotes epithelial health and is a prominent intestinal inhabitant of south Indian children with obesity (41). The effects of the gut microbes extend beyond the pathophysiology of obesity and influence the incidence of type 2 diabetes (42) and demyelinating illnesses, psychiatric disease, behavioral disorders, and irritable bowel syndrome (43).
Figure 2 depicts the recognition of commensal and pathogenic microbiota and the communication-related pathways between intestinal epithelia, dendritic cells, and other underlying immune cells. The development of tolerance shows M cells overlying Peyer’s patches and dendritic cells engaging antigens for processing and presentation. The recognition of antigens associated with pathogenic bacteria in the intestinal lumen stimulates production of secretory IgA. Binding of secretory IgA to invasive microbes in intestinal fluid and in the mucin layer blocks invasion of intestinal epithelia (17). Antimicrobial peptides secreted by enterocytes into the mucin layer also hinder microbial invasion of the gut barrier (17). Lipopolysaccharide from the cell wall of Enterobacteriaceae living in the gut binds to Toll-like receptor-4 on enterocytes. This pathogen-associated molecular pattern recognition causes neural transmissions via the vagus, while intestinal immune cells are stimulated to produce cytokines (17,44). Mice that have ileitis caused by Toxoplasma gondii infection have activated Toll-like receptor-9 in their enterocytes. In the absence of cerebral infection caused by T. gondii, the brain tissue of these mice had elevated levels of proinflammatory cytokines (45). This study persuasively reveals pathogen-specific infection and inflammation in the gut can cause immune-mediated inflammation in the brain.
Additional data suggest innate immune cells in the gut recognize microbial and other damage-related signals with help from the inflammasome. The inflammasome is a multiprotein cytoplasmic complex that activates one or more caspases that process and enhance secretion of proinflammatory cytokines in response to microbial pathogens and other stress response proteins. Characteristically, innate immune cells release interleukin-1, interleukin-18, and interleukin-33 (46); however, inhibitory signals from the inflammasome may also downregulate inflammation in the gut (47). For example, in a mouse model of murine enteritis, efferent vagal signaling via corticotropin-releasing hormone inhibits inflammasome-induced inflammation, thereby reducing intestinal enteritis (16,48). The GBA thereby can regulate inflammasome-mediated activation of innate immune cells in the gut through the action of the hypothalamic–pituitary axis, thereby abrogating inflammation in the intestine.
Other model systems indicate that gut microbiota transduce either inflammatory or protective (anti-inflammatory) signals in the GBA (49). Escherichia coli interactions with gut epithelia transmit inflammatory responses, whereas F. prausnitzii stimulate anti-inflammatory responses akin to other commensal bacteria like Lactobacillus casei and Bacteroides thetaiotaomicron. Commensal flora may promote protective responses by increasing mucin production by intestinal epithelia (50,51), a critical host defense against microbial invasion (17). Moreover, B. thetaiotaomicron heightens expression of angiogenin4 in crypt Paneth cells (52); angiogenin4 is a dual function protein that acts as an antimicrobial peptide (52) and as an angiogenic factor promoting growth and health in intestinal villi (53). Thus, commensal bacteria provide homeostasis and health in the developing gut.
Necrotizing enterocolitis (NEC) is a disease characterized by intestinal dysbiosis and intestinal inflammation affecting mostly preterm newborns (17). When NEC complicates the hospital stay, there is a higher risk for substantial long-term neurodevelopmental abnormalities (54). The stress created by NEC has adverse consequences in the brain via the GBA (55). Microbial toxicants and proinflammatory cytokines released into the systemic circulation from infected and damaged intestine have a causal relationship with the long-term psychomotor and intestinal disabilities seen after NEC (55). The degree of injury to the intestine in NEC correlates with the levels of oxidant stress and other mediators associated with brain injury (56). The role of the GBA in the pathogenesis of cerebral palsy and ASD requires more research in immature infants.
Infectious and noninfectious diseases, however, may activate intestinal inflammation in utero (57). Of course, fetal gut infection and inflammation assumes the fetus swallows bacterial pathogens into the stomach before birth. Fetal swallowing does not occur before 29–31 wk of gestation. During antenatal infections, experts have proposed the cholinergic, anti-inflammatory system sends signals to intestinal macrophages to inhibit cytokine secretion (58). This notion is conceptually correct because Ureaplasma parvum and Ureaplasma urealyticum are the most common organisms isolated from infected amniotic fluid, and are associated with preterm labor (59). Respiratory colonization with Ureaplasma doubles the risk of developing NEC in very preterm infants (60). The association of Ureaplasma with NEC still requires the molecular presence of this microbe in either surgical specimens or feces of infants who develop NEC.
After birth, however, alterations in the intestinal microbiota and the development of “dysbiosis” is probably the key event associated with the pathogenesis of NEC (61). Fecal samples have identified a bloom or sharp increase in the phylum Proteobacteria between 7 d and <72 hr before the onset of NEC cases (62). The study found unclassified members of the family Enterobacteriaceae as potential pathogens, but no particular genus and species were consistently associated with infants that developed NEC. Dysbiosis of the intestinal microbiota is also a risk factor for late-onset neonatal sepsis in very preterm infants (63,64). These insights about the gut microbiome and the pathogenesis of late-onset neonatal sepsis or NEC in very preterm infants have come forth because of metagenomics. Metagenomics is the study of microbes, their genes, and their metabolites from a defined environment, like the intestines. Metagenomic methods utilize analytical instruments that can define the molecular signatures of microbes in specific habitats. Analyses of those signatures involve software applications that characterize the microbial ecology in environmental samples. The analytical instruments, the molecular methods, the bioinformatics, and the representations of the data associated with metagenomic studies are often unfamiliar to caregivers in the neonatal intensive care units. A recent review of metagenomics describes the culture-independent methods used to identify microbes in human neonatal organs (65). The interstitial cells of Cajal are the pacemaker of the intestine and activate the muscularis mucosae in a rhythmic fashion (66). Gut motility can help remove pathogenic bacteria from the bowel lumen by defecation. Bacterial toxins in the intestinal fluid, such as a high endotoxin content, reduce ENS activity and are likely responsible for ileus before the onset of NEC (67). Ileus in very preterm infants should alert caregivers to the scenario that toxins in the intestine are inhibiting interstitial cells of Cajal activity. Most neonatologists know that ileus points to NEC on the horizon, but the preceding rationale defines the physiology. The gut microbiota are essential for the development and function of the gastrointestinal and the systemic immune system (12,34,43). A Th2:Th1 lymphocyte bias is associated with pregnancy maintenance and fetal well-being (68). The Th2 bias seen in neonates is probably a spillover of immunosuppression during human pregnancy. Studies in neonatal mice suggest the immunologic milieu renders neonates susceptible to infection. Postnatal immune events in the intestine are associated with microbial colonization that reverses neonatal Th2:Th1 bias and lowers the risk of postnatal infection (69).
Table 1 summarizes stressors before and after birth that may cause brain injury in conjunction with the GBA concept. The list of stressors is not exhaustive, since this paper describes a number of adverse neurologic consequences of common intestinal diseases throughout the review. Examples shown in Table 1 are illustrative of the unusual conditions seen with diseases involving the GBA. We selected stress states in the gut before and after birth, which were relevant to pathophysiology associated with the GBA (70,71,72,73,74,75,76,77,78,79,80,81,82,83,84).
Neonatal Nutrition and Modifications of GBA Function
While microbial colonization of the gut is important for immune system development, it also acts in concert with diet to promote healthy brain development (85,86). Human milk contains many biofactors that improve health and brain development (87). To assure its availability, neonatal intensive care unit caregivers often freeze a mother’s milk before use; however, freezing reduces the immunologic protective components in human milk. Compared with pasteurization of human milk, freezing maintains more immune properties in human milk than does heat treatment (88,89). In other words, fresh colostrum and milk contains lactoferrin, lysozyme, other antimicrobial proteins and immunomodulatory agents that prevent NEC (89,90). A human milk oligosaccharide , disialyllacto-N-tetraose, prevents NEC in a neonatal rat model (91). Several mechanisms define the beneficial effects of human milk oligosaccharides in the immature intestinal tract of human preterm infants. NEC is an intestinal disease of preterm infants, which produces intense inflammation (92) and indirect brain injury via the GBA (55,72). The preceding reasons are why we emphasize feeding colostrum right after birth and give freshly expressed mother’s milk during the neonatal intensive care unit stay (90,93,94). The use of fresh maternal milk will result in colonization of the gut with healthy commensal bacteria, particularly probiotic bacteria that utilize human milk oligosaccharides for their nutrition (90,93). For this reason, fresh human milk enables development of a healthy ENS and GBA (5). In the absence of a mother’s own milk, neonatal caregivers propose using donor human milk as the nutrient of choice instead of infant formula. When human donor milk undergoes holder pasteurization before use, the thermal processing significantly reduces the immunomodulatory and antimicrobial properties of proteins in the milk (89). Hence, pasteurization and freeze thawing make donor human milk a suboptimal antimicrobial and immunomodulatory nutrient. When human donor milk is compared with a mother’s own milk, the composition of human milk oligosaccharides was significantly different (94) suggesting that the antimicrobial benefits of donor milk and its ability to promote the assembly of a commensal microbial flora need more scrutiny if a healthy GBA is to be established.
The preceding reports led us to examine how the diet of very-low-birth-weight preterm infants played a role in the GBA. Fresh and unadulterated mother’s milk is the ideal nutrient to prevent NEC and to support optimal development and outcomes. Friel (95) found high levels of urinary F2-isoprostane, a biomarker of oxidant stress, when preterm infants received mother’s milk with >50% fortification. This investigation did not examine urinary isoprostanes when preterm infants received regular or high-density caloric formulas. However, a recent study found digested formula, but not digested fresh human milk caused death of neutrophils, endothelia, and intestinal epithelia in vitro (96). The researchers implied that these findings have relevance to the pathogenesis of NEC. The stress response seen after feeding fortified human milk and preterm cow milk-based formula is likely detrimental to brain development by way of harmful signaling through the GBA (43,97,98). Figure 3 describes the dietary, microbial, and immune events associated with brain injury in preterm infants. We emphasize that these factors may also influence intestinal barrier function and promote invasion and translocation of pathogenic microbes that cause late-onset neonatal sepsis and NEC (17).
Feeding probiotics to very preterm infants has reduced the occurrence of NEC (99). Recent studies show probiotic bacteria release inhibitors of tumor necrosis factor-alpha and its downstream target nuclear factor-kappa B (100). A recent review also says probiotic bacteria release inhibitors of tumor necrosis factor-alpha and its downstream target nuclear factor-kappa B (101). Researchers also propose that probiotics prevent brain injury by blocking the transport of damaging biomolecules via the GBA (43,86).
If one excludes intracranial hemorrhage and ischemia as principal contributors to brain injury in preterm infants, these infants still have three times the risk of developing ASD (102). A recent and compelling report showed that oral treatment of offspring with Bacteroides fragilis prevented autistic behavior in an animal model of ASD (103). The researchers proposed that B. fragilis reduced intestinal permeability and improved behavioral symptoms suggesting an altered gut microbiome plays a role in the pathogenesis of ASD. The mechanism of prevention may be an enteric anti-inflammatory milieu (interleukin-10 production) in neonatal mice after they received enteral B. fragilis (104).
Central nervous system injury in preterm infants often localizes to the white matter tracts (i.e., connectome) in the brain (105). Hence, imaging techniques that define injury to the connectome is receiving high interest (106,107). Injurious agents or neurotransmissions from the gut that reach the brain of preterm infants via the GBA are probably causes of brain injury (45,55). Advanced imaging techniques have used functional magnetic resonance imaging, magnetoencephalography, and positron emission tomography for interrogating neuronal pathways. If metagenomic techniques applied to the gut microbiota in conjunction with brain imaging, researchers will surely identify instigators of injury associated with the GBA, and the findings will correlate with human health and disease (106,107,108).
Conclusions and Future Research Considerations
Pediatric investigators must focus on the GBA of the fetus or preterm neonate, because adverse effects in their environment may be the origins of neurodevelopmental disease seen during childhood or adult life. In this review, examples of the developmental origins of disease that involve GBA have included autism and obesity. In the future, research must be devoted to neurologic disease arising from abnormalities in the GBA and obtain concrete proof of mechanisms operative in early life. The Human Microbiome Project collected and analyzed feces from preterm and term infants. The molecular findings related to gut microbes in those analyses need correlation with the development of cerebral palsy, autism, or other neurologic abnormalities that occur in very-low-birth-weight infants (109).
Statement of Financial Support
A University of Missouri Research Board grant and the Leda Sears Trust support M.P.S. The J. Lavenia Edwards endowment and Leda Sears Trust funds support H.Z. The Sears Trust and J. Lavenia Edwards endowment are funds provided by the Department of Child Health, University of Missouri– Columbia.
The authors have no financial disclosures or conflicts of interest.