Isolauri E, Walker WA (eds): Allergic Diseases and the Environment. Nestlé Nutrition Workshop Series Pediatric Program, Vol. 53, pp. 153–177, Nestec Ltd.; Vevey/S. Karger AG, Basel, © 2004.
The Role of Bacteria in the Development of Intestinal Protective Function
N. Nanda Nanthakumar and W. Allan Walker
Harvard Medical School and Developmental Gastroenterology Laboratory of the Combined Program in Pediatric Gastrointestinal and Nutrition, Massachusetts General Hospital, Charlestown, Mass., USA
Introduction
The primary function of the gastrointestinal tract is to complete the digestion and absorption of nutrients so as to provide a source of energy and substrate for growth and maintenance of the complete organism. Therefore, diseases that affect intestinal function have a major impact on body systems [1, 2]. This challenge is further compounded by the fact that the gut is directly in contact with a microbial and nutritional rich external environment. Under normal circumstances, a large number of bacterial species reside in the intestinal lumen in a symbiotic relationship with the host [3]. In addition, the gut is continuously exposed to foreign antigens, derived from luminal microbes, diet and ingested toxic substances [4]. In contrast to other organ systems, with the exception of skin, the gut is continually exposed to this external environment with an epithelial surface juxtaposed between the lumen and the interstitium and circulation. Unlike the skin, the intestinal epithelium is made up of a single polarized monolayer [2]. The apical surface of the epithelium is exposed to luminal contents including commensal flora [2, 3] but these substances are restricted from the basolateral surface by tight and adherent junction proteins [5]. These two junctional complexes are specialized structures unique to polarized cells and provide not a rigid structure, but an active flexible surface that allows migration of activated polymorphonuclear cells [6] during infection and access by dendritic cells to sample foreign antigen in the lumen [7].
153 Microbes in Gut Development
Structure and Function of the Intestine
The intestinal epithelium which separates luminal contents from the underlying mucosa consists of absorptive enterocytes (93–95% of cells), mucus-secreting goblet cells (3–5% of cells) and gastrointestinal hormone- producing enteroendocrine cells (1–2% of cells) [2]. Unlike the colon, the surface area of the small intestine is increased by invagination into ‘tongue- like’ structures called villi. Mucus produced by the goblet cells is secreted as a layer of highly glycosylated proteins onto the intestinal surface and functions as a lubricant and protective layer on the epithelial surface. Undifferentiated proliferative cells including stem cells exit in a pit-like structure called the crypts of Lieberkühn both in the small intestine and colon [8, 9]. However, only in the small intestine unique cells called paneth cells are located at the bottom of each crypt [10]. Paneth cells produce a number of unique antibacterial proteins which act to protect nearby stem cells from microbial damage. The crypt epithelium is also polarized but only acquires microvilli, also known as the brush border, on its apical surface as cells migrate from the crypt to villus [2, 8, 9]. As these cells emerge from the crypts and undergo epithelial differentiation they begin to express specialized apical proteins such as digestive enzymes and transporters [1, 2]. These glycoproteins, anchored on the apical surface of the epithelium, are responsible for the digestive and absorptive functions of this tissue. These highly glycosylated proteins and glycolipids that enrich the apical surface of the epithelium can also function as receptors for commensal microflora that begin to colonize the gut lumen shortly after birth [11, 12]. In addition to the multi-lineage epithelial cells described, the epithelial monolayer infrequently displays a dome-like surface called the follicle- associated epithelium (FAE) [2, 13]. Unlike the villus epithelium, the epithelial surface of those domes display, with varying frequency, a unique type of epithelial cell called the microfold cell (M cell). In humans and rodents 10% of FAE are made up of M cells [2, 13]. These cells have no lysosomes and are capable of invaginating upon attachment of microorganisms and large proteins. The M cells are a specialized lineage of epithelium dedicated to antigen sampling [13]. Unlike the adjacent enterocytes, M cells have fewer and shorter microvilli on their apical surface and the basolateral surface display numerous invaginations in which mucosal lymphocytes reside [14]. M cells are never seen on differentiated villus epithelium. The FAE and M cells appear above aggregates of lymphocytes in Peyer’s patches [14]. Since a single stem cell resides in each crypt, how these two lineages are derived during epithelial differentiation is not known, because the lack of a suitable in vitro model system has precluded the elucidation of the mechanism of this form of epithelial differentiation. However, recently exciting in vitro studies have shown that a differentiated enterocyte cell line can trans-differentiate into M cells under the influence of luminal pathogens and basolateral exposure of B cells suggesting
154 Microbes in Gut Development that luminal microbial attachments and paracrine action by B cell may be responsible for M-cell differentiation [15].
Development of the Small Intestine
Morphological development, cytodifferentiation and enterocyte-specific dif- ferentiation are established by the end of the first trimester in humans [16] and at birth in rodents (rats and mice) [2, 8]. Gestation is 21 days in rodents, whereas it is 40 weeks in humans. The functional maturation of the gut is divided into 2 periods. Details of the development of the gastrointestinal tract is beyond the scope of this review and provided in several recent reviews [1, 2, 9, 16]. By the end of the first trimester the epithelium begins to form a monolayer and a crypt–villus architecture appears. The epithelium starts to differentiate and tissue-specific markers appear [2]. Proliferating epithelium is confined to the crypts where multiple stem cells reside [9, 16]. This phase of development occurs during the second and third trimester in humans and during the first 2 weeks of postnatal development in rodents [2, 9]. The early phase of func- tional maturation of the small intestine can be defined using differentiation- specific enterocyte markers [1, 2]. For example, disaccharidases are first detected with initial cytodifferentiation of the enterocytes, but the levels of disaccharidases vary depending on species. In humans, lactase remains low in utero but sucrase is high during this period, e.g. equivalent to levels found in infants [16]. In contrast, in rodents sucrase is undetectable with high- lactase activity until weaning [2]. The second and final phase of development begins at birth for humans and at the time of weaning (3rd postnatal week) in rodents [2]. During this period, lactase activity rapidly declines to the levels seen in adult rodents but in humans this enzyme increases and reaches a maximal level in the newborn [16]. In rodents the expression of sucrase increases to adult levels by the end of wean- ing [2]. At the same time, most of the enzymes and transporters responsible for digesting solid food are rapidly established at mature levels. Terminal matura- tion of the small intestine temporally coincides with weaning in rats and mice. These changes, coinciding with ‘hard-wired’ development of enzymatic expres- sion, reflect the adaptive process necessary for survival on solid food [1, 2].
Regulation of Intestinal Development
The functional development of the gut is regulated by a number of factors. To unravel the complex mechanism(s) of development, extensive studies have been done in the rodent model [1, 2, 16]. However, little objective data are available for human gut development because of the inaccessibility of human tissues and inadequate intestinal models. The regulators of intestinal
155 Microbes in Gut Development development can be either extrinsic (luminal) factors such amniotic fluid, colostrum/milk and microbial flora or intrinsic factors such as circulating growth factors, e.g. glucocorticoids, intrinsic timing mechanisms (a biological clock), and/or epithelial–mesenchyme interactions. The role of these divergent regulators are briefly discussed below.
Colostrum and Mature Milk Colostrum and mother’s milk are complex biological fluids that contain many substances which provide nutrition but also protect and stimulate cell turnover including proteins such as casein, micelles, membranes, membrane- bound globules, and viable cells [4]. A complete description of the macro- and micronutrients in milk has been published recently [17]. However, in this review we will focus only on trophic factors present in colostrum/breast milk that play a critical role in intestinal development. These factors are present in physiologic quantities and their role(s) in intestinal development is not fully understood, again in part because of lack of availability of a model that recapitulates the newborn human gut.
Commensal Flora At birth, commensal bacteria begins to colonize the gastrointestinal tract [18]. The composition of the flora changes at the time of weaning [19]. This is in part due to the changing luminal environment contributed to by diet and the epithelium itself. However, a symbiotic relationship likely exists between microbes and the developing gut and will be discussed in detail later. Studies with germ-free rats and mice show no difference in the final stage of development [20], but in the absence of microbes the proliferation rate and epithelial migration are significantly reduced [9]. When germ-free animals are conventionalized to a normal environment, proliferation and migration return to mature levels within 2 weeks. Therefore, maturation of the small intestine at weaning in rodents, an event similar to changes in the human at birth, is a complex process likely involving both secondary hormonal stimuli and alterations in the composition of the luminal microbial flora.
Circulating Growth Factors Thus far, several factors have been implicated as potential growth factors regulating the development of the small intestine [1, 2]. Among them, only glucocorticoids appear to have a primary influence on the developing intestine. The other factors that have been studied are thyroxine, insulin, gastrin, epidermal growth factor, transforming growth factors (TGF- and TGF- ). In rodents both endogenous as well as the exogenous glucocorticoids have a potent effect on the rate of intestinal maturation [2]. These effects can be recapitulated in both in vitro and in vivo model systems. They appear to regulate all facets of intestinal maturation, enabling precocious development of digestive absorptive function and activation, proliferation as well as
156 Microbes in Gut Development epithelial migration from the crypt to the villus. These effects, however, are restricted to a narrow period of 2 weeks in the rodent and after the 3rd postnatal week the intestine loses its responsiveness to glucocorticoids [2]. Indirect evidence in the human suggests that a similar period of glucocorticoid sensitivity might exist during the third trimester (details will be discussed below). Non-circulating autocrine or paracrine factors may also play a role in the ontogeny of the intestine. As in vitro models and human intestinal cell lines have been established [21], the additive role of other growth factors, e.g. epidermal growth factor, TGF- , etc., in concert with corticosteroids can be determined and the effect of these growth factors at other phases of development understood. This information may provide not only a better understanding of gut development but also can be used in gut repair after resection or vascular insufficiency and necrosis.
The Intrinsic Timing Mechanism (a Biological Clock) A number of investigators have sought to determine the intrinsic factors controlling terminal maturation of the rodent small intestine, but these changes are not primarily dependent on hormones. Using intestinal isografts, studies have established an intrinsic timing mechanism that initiates the development of the gut at a precise time [22]. A similar hypothesis for other intrinsic timing mechanisms has been proposed in other organ systems. The existence of a ‘hot-wired local trigger’ mechanism in the small intestine has been proposed [23] but the cellular basis and the molecular nature of this biological clock is unknown and remains to be elucidated.
Epithelial–Mesenchyme Interactions The importance of epithelial–mesenchymal interactions during intestinal development has been extensively analyzed [24]. The data suggest an instruc- tive role for the epithelium and a permissive role for mesenchyme in epithelial cytodifferentiation, i.e. mesenchyme is necessary but not sufficient for devel- opment. For example, glucocorticoids most likely mediate their effect on the epithelium via epithelial–mesenchymal interactions [25]. However, the mechanism of mesenchyme action in hormonal regulation is not completely understood. It has been postulated that corticosteroids may affect direct cell interaction via specific extracellular matrix components [25]. However, new techniques such as immortalized, non-malignant human cell lines are needed to help answer this question. Unfortunately, the role of epithelial–mesenchymal interaction in the development of the mucosal immune system remains an unknown process.
Development of the Mucosal Immune System
The epithelial barrier of the gut separates luminal antigens and microbes from underlying lymphoid elements and acts as a first responder in the
157 Microbes in Gut Development mucosal immune system [26]. Generally, food antigen and commensal microbes do not elicit an immune response but instead evoke oral tolerance [27]. However, when an immune reaction occurs in response to these stimuli in lamina propria lymphocytes, the activation can lead to food intolerance and intestinal inflammation. In contrast, enteric pathogens can stimulate a self- limited mucosal and systemic immune inflammatory response after breaching the epithelial barrier by using various adaptive techniques as an attempt to eliminate these pathogens [6, 28]. Depending on the pathogen, intestinal damage may vary from mild inflammation to full-fledged tissue damage. This ability to distinguish nutrients and luminal commensal bacteria from enteric pathogens and their toxins is an important feature of the normal mucosal immune response and generally results in intestinal immune homeostasis [6, 26]. The development of this balance is initiated during fetal maturation but requires initial bacterial colonization [18]. The luminal environment changes from a sterile surface bathed in amniotic fluid to bacterially contaminated food sources, initially milk and then solid foods, during the adaptation to the extrauterine environment [17]. Briefly the elements of the mucosal immune system are described below, but details can be obtained from recent reviews [4, 14, 26, 27]. As stated, a polarized epithelial monolayer exists as a tight barrier that prevents movement of various molecules from the lumen to the interstitium [5]. Direct sampling of luminal food and bacterial antigens is carried out by M cells which act as a conduit to underlying antigen-presenting cells, e.g. macrophages and dendritic cells that lie beneath the epithelium directly in contact with M cells [13, 26]. The lamina propria also contains mast cells, and various lymphocytes, particularly IgA-producing B cells. Moreover, the epithelium rather than being simply a physical barrier to antigens is also an active participant in the mucosal immune response by releasing cytokines and by transcytosing dimeric IgA produced by B cells via the polymeric immunoglobulin receptor. IgA acts on luminal bacteria to prevent their adherence to the enterocyte surface. In addition, specialized T cells, known as intraepithelial lymphocytes and juxtaposed between the epithelium at the basolateral surface, recognize bacterial antigens and/or immune-activated epithelial surface molecules so that they could respond rapidly upon stimulation [29]. The epithelium, together with these components of the immune system, modulates luminal antigen and bacterial activation of the peripheral immune system [26, 27]. This physical and functional barrier constitutes the mucosal immune system of the gut that continuously samples the luminal environment without disrupting the digestive and absorptive function of the gut as well as modulating a systemic immune response that can be muted or lead to exces- sive inflammation and tissue damage. These cells constitute the innate and adoptive immune system of the gut that is defined as a rapidly responding antigen nonspecific response and a delayed, antigen-specific memory cell
158 Microbes in Gut Development response, respectively [26, 27, 30]. These two arms of the mucosal immune system initiate a self-limited inflammatory response by recruiting activated neutrophils and monocytes into the lamina propria from blood vessels but yet does not sustain a chronic inflammatory response. The intestinal epithelium is closely associated with several cell types involved in innate immunity that include macrophages, dendritic, mast and paneth cells [4, 7, 10, 26, 27]. The developmental response of these cells has not been well studied. Macrophages are the major resident phagocytic lymphoid cell in the gut. They exist diffusely in the lamina propria, but mainly in Peyer’s patches. In human fetal intestine, macrophages can be detected from about 12 weeks of gestation onwards, although specific subpopulations differ with gestational age and location in the gut. Of particular interest to this review is that lamina propria macrophages lack CD14, a cell surface co-receptor for lipopolysaccharide binding, and therefore these cells are hypo-responsive to endotoxin [31]. This regional difference in the macrophage may be important in preventing inappropriate activation of inflammatory signals in response to gram-negative bacteria continuously present in the gut lumen and potentially available to penetrate the epithelial barrier. Antigen- processing dendritic cells function at the interface between innate and adaptive immunity. Immature dendritic cells rapidly sample luminal food and microbial antigens and then become activated, migrate to regional lymph nodes where they present the processed antigen to T cells with co-stimulatory signals under the influence of regulatory cytokines [7, 26]. How they discrim- inate pathogenic and commensal bacteria is as yet not known. Recently dendritic cells from the intestinal lamina propria have been shown to protrude as cytoplasmic extensions across tight junction barriers into the lumen presumably to sample luminal antigens [32]. This is done without disrupting the polarized epithelium. The development of these mucosal dendritic cells is unknown. Mast cells which are present in significant numbers in the intestinal mucosa are considered to be effector cells in IgE-mediated allergic responses, but also play a role in defense against intestinal parasites and enteric bacterial pathogens via pathogen-induced tumor necrosis factor- release [33]. However, the development as well as the mode of distribution of these mast cells in the intestinal mucosa is not completely understood. The role of the epithelium in the development of the mucosal immune system is better characterized than other lymphoid cells in the mucosal immune system [26]. Tight and adherent junctions are formed during the second trimester [2, 5]. However, the functional property of transepithelial resistance and the passage of macromolecular markers during the second and third trimester of development is not known. Recently, the major proteins in the junctional complex, e.g. occludin and claudins, have been identified [34] but their developmental regulation functions have not been elucidated. Possibly because of incompletely developed tight junctions in newborn infants, immature enterocytes have the potential to translocate macromolecules such
159 Microbes in Gut Development as maternal antibodies and trophic factors from colostrum and mother’s milk [35] to function passively and act in initial mucosal defense. This process is partially facilitated by lower gastric acidity and pancreatic enzymes in the lumen preventing destruction of these bioactive molecules before reaching the small intestine for translocation into the mucosa [35, 36]. Gradually, within weeks, this capacity is lost and the gut becomes less permeable. This process has been termed ‘closure’, and the period of this uptake is well characterized in rodents, ending just before weaning [2, 36]. Exogenous glucocorticoids are capable of accelerating closure precociously along with terminal maturation in the developing rodent intestine. Though it is believed to be developmen- tally regulated, the mechanisms by which bioactive macromolecules are taken up, e.g. phagocytosis versus pericellular transport, by the small intestine is not known. The ability to produce chemokines that can recruit activated polymorphonuclear cells to the site of inflammation may be developmentally regulated. Using intestinal explants of 20-week-old fetal intestine and biopsies of infants, we were able to demonstrate an excessive IL-8 response to endotoxins and endogenous pro-inflammatory stimuli [37]. The highest amount of IL-8 was produced in the epithelium in the developing gut and these results were confirmed using human fetal intestinal epithelial cell lines. This difference may in part be responsible for initiating the excessive inflammatory response in the premature infant gut seen in diseases such as necrotizing enterocolitis, which will be discussed below. Sampling of luminal antigens are carried out by M cells so that the adaptive immune system can be tolerized to food and unknown bacterial antigens [26, 27]. M cells begin to appear by 17 weeks of gestation in the human, after lymphoid aggregates begin to appear [2]. This temporal sequence of events is of some significance since mice lacking B cells have either completely or partially impaired development of Peyer’s patches, FAE and M cells [12, 38]. Though other lymphoid cells beside B cells have recently been shown to induce M-cell differentiation, antigenic macromolecules and a few pathogenic bacteria, viruses and protozoans preferentially penetrate the M-cell epithelial barrier by endocytosis or phagocytosis [4, 26]. M cells facilitate the sampling of these pathogens and luminal antigens by the mucosal and systemic immune system via the subepithelial lymphocytes [26, 27]. Unfortunately, the mechanism(s) by which either the macromolecules or pathogens penetrate the cells is not know and remains an active area of investigation.
Bacterial Colonization and the Role of Glycosylation during Gut Development
The lumen of the gut contains a complex ecosystem of microbial flora composed of an excess of 500 species with 100-fold more anaerobes than
160 Microbes in Gut Development aerobes [3, 11, 18]. However, about 30–40 species comprise 99% of the total bacteria. These microbes facilitate colonic function, such as absorption, secretion of electrolytes and water, as well as storage and excretion of waste materials. The gut microflora plays an important role in tissue homeostasis because of fermentation which results in the production of compounds that have a positive and a negative influence on gut function. For example, the fermentation of residual undigested complex carbohydrate content (such as fiber) to short-chain fatty acids can provide an energy source for the colon while pathogenic species may produce toxic compounds. The human gut is a complex microbial ecosystem in a multiplicity of different microhabitats and metabolic niches [11, 17, 18]. This observation has largely arisen by analyzing the bacteria content in feces. Gram-negative anaerobes of the genus bacteroides are the most prevalent microbe in the gut (30% of the total fecal flora). Other predominant groups are gram-positive rods (bifidobacteria, eubacteria, clostridia, lactobacilli) and gram-positive cocci (ruminococci, peptococci, peptostreptococci). Chief among these are the bifidobacteria, which may constitute as much as 25% of total fecal contents. A number of other groups exist in lower proportions, including enterococci, coliforms, methanogens and dissimilatory sulfate- reducing bacteria (fig. 1). More recently molecular techniques have been used to identify previously unknown bacteria. However, the unique microenvironment in which these microbes grow could not be recapitulated in vitro, thus preventing comprehensive identification of these bacteria and elucidating their role in microbial homeostasis in the lumen of the gut. Functionally, these bacteria are divided into species that exert either harmful or beneficial effects on humans (fig. 1). The pathogenic consequences of an imbalance in bacterial content include diarrhea, inflammation, necrosis, ulceration and intestinal perforation. Growth inhibition of harmful bacteria may promote better health by stimulating appropriate immune functions, decreasing gas production, improving digestion and absorption of essential nutrients as well as the synthesis of vitamins B and K. At birth the gut is initially exposed to maternal bacteria that are present in the vagina and maternal colon. However, a delayed colonization occurs if the infant is delivered by cesarean section. The newborn intestine is first colonized with enterobacteria and their number attains 109/g of feces. On day 6, bifidobacteria become the predominant microbe in the breast-fed infants, exceeding enterobacteria by a ratio of 1,000:1, whereas enterobacteria remain predominant in formula-fed infants, exceeding bifidobacteria by 10:1 (fig. 2). By the end of a month, bifidobacteria are the predominant organism in both fed groups. However, these organisms in formula-fed infants are approximately one tenth that of breast-fed infants, presumably the breast milk creates an environment favoring the development of a simple flora such as bifidobacteria and few other anaerobic and small numbers of facultatively anaerobic bacteria. In contrast, formula-fed infants have complex microbiota,
161 Microbes in Gut Development
Harmful/pathogenic effects Health-promoting functions 2 Ps. aeruginosa Diarrhea/ constipation Vibrionaceae infections Pathogenic systemic (incl. Stapylococci effects production of toxins) Clostridia 6 Veillonellae
Inhibition of growth Production of Enterobacteria of exogenous and/or potential harmful bacteria carcinogens E. coli Stimulation of immune Production of 8 Lactobacilli functions through non- toxic H2S Sulfate reducers pathogenic means, anti-tumor properties, cholesterol reduction Intestinal Anaerobic G ve cocci putrefaction Methanogens Lower gas distention Aid in digestion and/or Eubacteria absorption of food ingredients/minerals Bifidobacteria Synthesis of vitamins
Bacteroides
11 Number/g feces
log10 scale
Fig. 1. Generalized scheme of the human gut microbial composition. The different bacterial groups are divided on the basis of whether they exert properties that are potentially damaging or health-promoting for the host. The central vertical line gives approximate number in feces. H2S Hydrogen sulfide. Adapted from Fuller and Gibson [40]. with bifidobacteria, bacteroides, clostridia and streptococci all prevalent (fig. 2). The introduction of solid foods to breast-fed infants caused a major re-disturbance of the microbial ecology of the colon as numbers of enterobacteria and enterococci colonized by Bacteroides spp., clostridia and anaerobic streptococci occurs. This was not observed when formula-fed infants began to take solids. Instead, counts of facultative anaerobes remain high while colonization by anaerobes other than bifidobacteria continue. By the end of a year the composition of the anaerobic bacteria of healthy infants in both groups resemble that of adults with a corresponding decrease in the number of facultative anaerobes [17, 39]. The incidence of pure cultures of aerobic bacteria is higher in the stool of the premature neonate and critically
162 Microbes in Gut Development
Pre-wean Early wean Late wean Week 1 Week 4 or or or 12 months pre-solids after solids 9 months
Bifidobacteria 7 * * * *
1 Bacteroides spp. 7 * * * 1 * Anaerobic gram-positive cocci 7 * 1 * Clostridia * 7 * * Number of babies colonized by indicated bacteria 1 Enterobacteria 7 * * * a /g 8 *
10 1