Preterm Birth by Cesarean Section: The Gut-Brain Axis, a Key Regulator of Brain Development

Complex microbial populations within the gastrointestinal tract exist in symbiosis with the host and contribute to physiological homeostasis and life-long health [1, 2]. The seeding of the microbiome and the impact of interventions such as cesarean section (C-section) delivery are the focus of intense research. When a C-section is performed, babies do not pass through the genital tract and vertical transmission of vaginal microbiota from the mother is circumvented. For these babies, microbial colonization occurs only from environmental (dietary sources) and skin bacteria. Little emphasis has been placed on the small but significant number of babies born prematurely by C-section and the potential impact of a different colonization of the intestinal microbiota (skin vs. vaginal origin) on the intestinal-brain axis, which is the subject of this review.

Preterm birth is defined as delivery occurring before 37 weeks post-menstrual age (PMA). In 2014, preterm birth rates per live births ranged from 8.7% to 13.4% and affected overall around 15 million infants [3]. Subcategories of preterm birth are based on PMA including (i) extremely preterm (before 28 weeks), (ii) very preterm (28–32 weeks), and (iii) moderate to late preterm (32–37 weeks). Medically indicated preterm birth corresponds to induced preterm delivery due to maternal or fetal distress and represents around 25% of all preterm births [4].

During the last trimester of pregnancy, major events contributing to brain development including synaptogenesis and dendritic arborization [5], migration and maturation of interneurons [6], and maturation of oligodendrocytes and myelination occur [7]. Preterm delivery disturbs this critical neurodevelopmental program, and the constellation of gray and white matter changes caused by preterm birth and related events is termed “encephalopathy of prematurity” (EoP; Fig. 1) [8]. These brain disturbances can lead to cognitive and neurobehavioral impairments involving social behavior, executive functions, and attention [9]. Increased rates of autism spectrum disorder (ASD) and other neurodevelopmental disorders (NDDs) are significantly associated with extreme preterm birth (reviewed in [10]). Later in this review, we will discuss the central role of the gut-brain axis in influencing health outcomes of preterm birth. Our broad hypothesis is that microbiome colonization after preterm C-section significantly contributes to brain maldevelopment associated with EoP.

Preterm birth is divided into spontaneous and medically indicated prematurity (Table 1). Spontaneous prematurity results from preterm labor or pre-labor rupture of membrane (PPROM) [11]. Numerous maternal factors are associated with an increased risk of spontaneous preterm birth, including infection, low BMI, and age (Table 1). PPROM complicates 2.8% of all pregnancies [12] and accounts for 32.6% of preterm births [13]. Inflammation or infection is more frequent in the case of PPROM [14]. Two different World Health Organization (WHO) surveys demonstrated an increased rate of C-sections worldwide between 2004 and 2010 [15]. In 2018, C-sections represented up to 58% of all births in some countries [16]. Despite an initial contribution to mother and child health improvement, the current C-section rate is potentially associated with long-term complications including NDDs (reviewed in [17]).

Preterm infants that are delivered by C-section because of the gestational age strongly influence statistics of prematurity. In 2011, the French EPIPAGE Cohort demonstrated that depending on the PMA, preterm birth by C-section represented up to 72% of live birth [18]. Similarly, 73.5% of 1,320 singleton preterm infants born in Austria between 2003 and 2011 were delivered by C-section, and 26.5% vaginally [19]. Whether or not C-sections for preterm born infants should be routine practice is hotly debated, and a recent analysis of WHO Global Survey and Multi-Country Survey databases investigated outcomes of preterm birth by mode of delivery [20]. An analysis of more than 30,000 newborns found that C-section was associated with a significant increase chance of maternal ICU admission, maternal near miss, and NICU admission, but significantly lower odds of a recent stillbirth and perinatal death. Although 30%–37% of preterm infants are delivered by C-section, there is no information to justify the mode of delivery in this cohort [20]. Rising numbers of C-section births are correlated to adverse effects on immune development including predisposition to infections, allergies, and inflammatory disorders as well as hypertension or diabetes [21, 22]. Epidemiological studies have shown an association between C-section and ASD or attention-deficit/hyperactivity disorder [23]. A multinational cohort consisting of 5 million births showed a significant association between ASD and C-section, even after adjustment for PMA [24].

The first colonization of the gastrointestinal tract is generally considered to occur at birth and predominantly comprises aerobic bacteria including Staphylococci and Enterococci which belong to the Firmicutes phylum. Subsequently, when most of the oxygen within the gastrointestinal tract has been consumed, anaerobic bacteria colonize the gut: Bacteroides (of the Bacteroidetes phylum), Bifidobacterium (actinobacteria division), and Clostridium (firmicute division). Studies have however discussed in utero microbes transmission from mothers to infants [25]. A recent report examining the gastrointestinal tract of unborn calves suggests the presence of a fetal microbiome with gut region-specific characteristics [26]. However, the in utero transmission of gut microbiome in humans is considered highly unlikely. For a recent and thorough discussion of the topic, see commentary by Blaser et al. [27] and research article by de Goffau et al. [28]. Events during pregnancy such as infections or inflammation, notably those associated with PPROM, may possibly influence the colonization of the gastrointestinal tract prior to birth.

Preterm and term newborn infants present different gut microbial communities, with preterm gut microbiota (GM) being characterized by low diversity and high interindividual variation compared to term infants [29‒31]. A study on the 2002–2005 Norway population showed that PMA at birth is a dominant factor contributing to microbiota composition, and that is independent of confounding factors such as delivery mode, breastfeeding duration, or antibiotic exposure [29]. The observed differences in GM were not observed at 4 months postpartum, confirming that the maturation of the GM in preterm infants is age dependent. GM was dominated by Staphylococci between 25 and 30 PMA, Enterococci between 30 and 35 PMA, and Enterobacteriaceae between 35 PMA [32]. The Bifidobacterium dominance found in term infants developed progressively at 30 weeks, but this progressive dominance was delayed in preterm infants due to persistent dominance of Enterobacteriaceae after 25 weeks [33]. Furthermore, preterm GMs exhibited lower diversity, fewer Firmicutes bacteria, and a higher abundance of Proteobacteria compared with term neonates [29]. In 2021, a worldwide study demonstrated that preterm infants during their 3 first months of life present mostly Firmicutes in comparison with term infants that present mostly Bifidobacterium and Bacteroides [34]. New insights using a scalable multi-kingdom quantification method in a 6-week longitudinal cohort of preterm infants allowed Rao et al. [35] to (i) confirm the previously described GM colonization but also (ii) to demonstrate that microbe-microbe interactions, including fungi, affect GM assembly in preterm infants.

C-section delivery is associated with GM alterations [36] and a decrease in bacterial diversity [37]. GM alterations due to C-section persist during the first year of life. When a C-section is performed, babies do not pass through the genital tract and vertical transmission of vaginal microbiota is circumvented. For these babies, microbial colonization occurs only from environmental and skin bacteria. Fecal microbial communities exhibit a lower abundance of Bifidobacterium spp, Bacteroides, and Lactobacillus and increased levels of Enterococcus and Klebsiella spp in C-section-delivered infants. Likewise, Xiao et al. [34] showed delayed Bacteroides colonization in infants delivered by C-section in the first year of life compared with infants delivered vaginally. In a cohort of children that did not receive antibiotics during the first year of life, children born by C-section, when compared to vaginal delivery, had a significantly delayed gut colonization with Bifidobacterium. These differences persisted regardless of variations in the feeding mode (breast milk or formula) [23].

Several psychiatric pathologies, such as major depressive disorder (MDD) [38, 39] or ASD [40], could be influenced by changes in GM composition. Accordingly, fecal samples from patients diagnosed with MDD exhibit an increased level of Prevotella, Bacteroides, and Proteobacteria and decreased abundance of Lactobacillus, Bifidobacterium, Faecalibacterium, and Ruminococcus. In other studies, feces from MDD patients were transplanted into microbiota-deficient rats and germ-free (GF) mice, respectively [41, 42]. These studies demonstrate that fecal microbial dysbiosis provokes a depressive-like behavior in mice and alterations in tryptophan metabolism [38, 41, 42]. A recent meta-analysis showed dysbiosis in fecal samples from ASD children and a frequent occurrence of gastrointestinal symptoms. Specifically, children with ASD had a higher abundance of Bacteroides, Parabacteroides, Clostridium, Faecalibacterium, whereas Phascolarctobacterium, Coprococcus, and Bifidobacterium were observed at lower abundance in neurotypical controls [43]. These data suggest that GM alteration could interact with brain cellular and molecular elements, leading to behavior alteration.

Experimental data suggest that GM plays an essential role in the immune system maturation, the hypothalamic-pituitary-adrenocortical (HPA) axis, the endocrine system, the formation, and maintenance of the blood-brain barrier, and brain neurogenesis and myelination [1, 44]. Manipulation of GM affects early programming of brain circuits involved in emotion, motor activity, and cognitive function by modulating gene expression and neurotransmitter levels in the rodent brain [1, 45‒47]. An imbalance of the commensal GM could directly affect brain functioning. Indeed, exposure to microbial pathogens during development leads to impaired behavior and cognition in rodents. Colonization of the gastrointestinal tract by Campylobacter jejuni leads to anxiety-like behavior in mice [48], and early postnatal exposure to infectious agents alters the structural and functional properties of hippocampal glia, contributing to memory impairment in rats [49]. It has also been proposed that the GM provides an interface between maternal stress and neurodevelopmental programming of the offspring by acting on the HPA axis [44]. Indeed, GF mice have an exacerbated HPA axis response to acute restraint stress when compared to specific pathogen-free (SPF) mice that have a commensal GM [50]. Microbial transfer experiments in GF mice (SPF mice treated with antibiotics) showed that colonization of GF BALB/c mice (with high levels of anxiety-like behavior) with microbiota from NIH Swiss mice (exhibiting “non-anxious” behavior) increased exploratory behavior, a sign of decreased anxiety. Conversely, 8 colonizations of GF NIH Swiss mice with BALB/c microbiota resulted in reduced exploratory behavior [46]. Similarly, GF mice display increased motor activity and reduced anxiety-like behavior compared with SPF mice; these behavioral changes are associated with increased synaptophysin and PSD-95 expression in the striatum of GF mice [51], and fecal microbial transfer from SPF mice normalizes synaptic protein expression in the GF mice. The offspring of GF mice that had been conventionalized with SPF microbiota showed behavior like SPF controls, suggesting that the microbiota may influence the brain during the period of development [51]. Sharon et al. [52] transplanted the GM of human donors with ASD or TD control GF mice and showed that such colonization was sufficient to induce characteristic autistic behaviors.

In a preclinical model of birth by C-section, mice also showed altered learning and behaviors compared to vaginal delivery [53]. Some of these effects have been associated with methylation alterations of glucocorticoid signaling genes potentially leading to insufficient glucocorticoid function and a hyperactive HPA axis, causing damage in the hippocampal brain region [54], to increased cell death in many brain regions including the hypothalamic paraventricular nucleus that plays an important role in stress response and brain-immune interactions [55], and to a transient underdeveloped dendritic arborization [56].

Approximately 80% of the immune system is located within and adjacent to the intestinal mucosa. Through its biological and metabolic functions, the GM plays an important role in immune system maturation and modulation [57]. Immune system abnormalities including abnormal cytokine levels are observed in ASD [58], and increased levels of proinflammatory cytokines (e.g., MCP-1, IL-1B, or IL-4) in neonatal blood have been correlated with a higher risk of developing ASD [59, 60]. The GM influences immune system maturation through specific strains (i.e., including Clostridium and Ruminococcus) which produce pro-inflammatory cytokines [61]. According to Cao et al. [62], there may be a link between certain cytokine imbalances and changes in the amounts of GM seen in youth with ASD. Peripheral inflammation induces central nervous system (CNS) inflammation or neuroinflammation. Microglia and astrocytes are key cells for brain homeostasis but also for neuroinflammation. In EoP, activated microglia and astrocytes contribute to altered developmental brain trajectory (Fig. 1), which will contribute to NDDs [10, 63‒67]. Microglia are CNS resident macrophages that are generally the first responders to the brain through several receptor classes that are triggering responses to infection, including the release of cytokines and chemokines.

GM plays a key role in modulating brain microglia development and function, and microbial metabolites modulate microglia-mediated inflammatory responses [68]. For example, in maternal immune activation models with ASD-relevant phenotypes in rodents due to neuroinflammation mediated by microglial reactivity [69], oral treatment with commensal Bacteroides fragilis corrects gut permeability, modulates microbial composition, and ameliorates ASD-related behavior in offspring [70]. One hypothesis is that microglia could play a crucial role as mediators linking the gut microbiome and NDDs relevant to caesarian delivery through molecules described hereafter. Specifically, birth by C-section may amplify the effects of microbial dysbiosis and microglial dysfunction on brain homeostasis and contribute to or exacerbate phenotypes associated with NDDs.

The microbiota is one of the key regulators of gut-brain function in maintaining homeostasis. There are multiple gut-brain pathways which include neural communication by the vagal nerve, endocrine as well as metabolic and immunological pathways (Fig. 2) [71]. The important complex and bidirectional signaling between the gut and brain via the microbiome in adulthood is becoming relatively well understood, but effects during development are emerging [72, 73]. How these signals are interpreted by microglia is a new source of questioning [74].

Here, we describe three types of gut-brain axis mediators well known in the literature: bacterial peptidoglycans (BPGs), short-chain fatty acids (SCFAs), and serotonin. BPGs derive from commensal GM and are translocated to the brain by the intestinal peptide transporter 1 [75]. BPGs also interact with two families of specific pattern recognition receptors [76]: peptidoglycan (PGN)-recognition proteins (PGLYRP1-4) and NOD-like receptors (NOD1-2). In turn, GM changes such as those occurring in GF mice or following the administration of antibiotics modulates brain pattern recognition receptor expression [75]. The C-MET gene plays a significant role in various biological processes, such as gastrointestinal and brain development, including synaptic formation. Studies have shown that individuals with an increased risk of ASD diagnosis may have an association with this gene [77]. Ablation of a PGN-sensing protein (PGLYRP2) impacts cerebral gene expression, including C-Met gene expression, and social behavior [1]. Overall, PGN produced by commensal GM modulates the expression of multiple genes involved in shaping brain development and social behavior. Recent hypotheses highlighted the role of BPGs in neuroinflammation in adults’ pathologies via the peripheral immune system (reviewed in [78]).

SCFAs [79], such as butyrate or acetate (bacterial dietary fiber metabolites), modulate chromatin and gene expression and likely play a significant role in gut health which may further impact brain development. For example, butyrate modulates histone acetylation in the hippocampal region and improves memory in a mouse model of Alzheimer’s disease [80]. Interestingly, SCFA concentration is increased in stool samples from ASD children, further supporting a link between GM and brain [42]. Similar to PGNs, SCFAs produced by commensal GM are able to modulate developing brain gene expression. On the contrary, mice lacking the SCFA receptor free fatty acid receptor 2 display immature microglia phenotype leading to impaired innate immune responses. Recolonization with conventional GM partially restores such microglial phenotype [79].

Serotonin is a key hormone of physiological homeostasis with widespread effects throughout the body, particularly in the CNS where it stabilizes mood and behavior. More than 80% of serotonin is produced in the gastrointestinal tract, and this synthesis is strongly influenced by microbes [81]. Serotonin precursors such as tryptophan are modulated by GM as demonstrated by increased tryptophan concentrations in the plasma of rats treated with Bifidobacteria infantis [82]. This suggests that the serotonin pathway forms another component of the communication between GM and the brain.

In this review, we outline the potential impact of the microbiota-gut-brain axis on premature birth and the developing brain in addition to influences on neuroinflammation processes. We emphasize the potential for microglia to play a role as crucial mediators linking the gut microbiome and NDDs relevant to preterm caesarian delivery births. Specifically, birth by C-section may cause microbial alterations that contribute to or exacerbate phenotypes associated with NDDs. Importantly, the effects of microbial dysbiosis and microglial dysfunction on physiological homeostasis may be amplified in infants born via C-section delivery. These findings are relevant to enhancing our understanding of neurodevelopmental impacts and the influence of maternal infection on premature birth.

Figures were created using BioRender.

The authors declare no conflict of interest.

Pierre Gressens, Juliette Van Steenwinckel, Cécile Morin, and Cindy Bokobza research is funded by Inserm, Université de Paris, Horizon 2020 (PREMSTEM-874721), Fondation de France, Fondation ARSEP, Fondation pour la Recherche sur le Cerveau, Fondation Princesse Grace de Monaco, and an additional grant from “Investissement d’Avenir -ANR-11-INBS-0011-NeurATRIS” and “Investissement d’Avenir -ANR-17-EURE-001-EUR G.E.N.E.” Bobbi Fleiss research is funded by Horizon 2020 (PREMSTEM-874721), RMIT University’s Biomedical Health and Engineering Enabling Capability Platform, and the Cerebral Palsy Alliance. Elisa Hill research is funded by a National Health and Medical Research Council (NHMRC) Ideas (Grant No. APP2003848).

C.M., C.B., B.F., E.L.H.-Y., J.V.S., and P.G. made substantial contributions to the conception, design, and writing of the review.

Cécile Morin, Cindy Bokobza, Juliette Van Steenwinckel, and Pierre Gressens contributed equally to this work.

This is a review paper; we do not present any data.