Necrotising enterocolitis (NEC), a severe inflammatory condition that usually affects the terminal ileum and proximal colon, is the most common serious gastrointestinal disease of preterm infants. It is the major cause of morbidity and mortality, affecting 1% to 7% of all neonates admitted into neonatal intensive care units (1,2). This disorder primarily affects preterm infants, mostly of very low birth weight, with an incidence of approximately 15%. The characteristic feature of NEC is bowel necrosis of variable length and depth, causing perforation in up to one-third of affected infants (1,2). Approximately 20% to 40% of neonates with severe NEC require surgical intervention, mostly those of lowest gestational age and birth weight (3). Postoperative complications in survivors include short bowel syndrome, stricture, persistent dysmotility, and failure to thrive (1–4).
Clinical associations of NEC include those affecting mesenteric blood flow (polycythaemia, umbilical vessel catherisation, congenital heart disease, maternal cocaine use, medications, blood transfusions), those affecting luminal contents (hyperosmolar formula feeds), and those associated with systemic illness (infection, hypoalbuminaemia, respiratory distress syndrome, and hypoxic ischemic encephalopathy) (1,2). The aetiology of NEC is unknown, although a combination of intestinal ischaemia, enteric dysmotility, disturbed luminal flora, and unabsorbed dietary substrates has been suggested from clinical and experimental studies (5,6). Many inflammatory mediators have been implicated, notably platelet-activating factor, nitric oxide, and tumour necrosis factor-α. (7). There is evidence of genetic predisposition based on cytokine polymorphisms (8) and that intestinal barrier function is compromised (9).
The question has arisen whether the disorder is always induced from the lumen (outside-in pathology) or can be initiated by pathology arising in the deeper tissues (inside-out pathology). Potential vascular causes of inside-out pathology were signalled by clinical findings of associated mesenteric blood flow abnormalities. There were no similar pointers toward causative neural abnormalities, until publication of an animal study, which showed that if mesenteric glial cells within the myenteric plexus were focally disrupted, then a spontaneous NEC-like lesion occurred causing ileal perforation and death (10,11). Enteric glial cells, the enteric nervous system equivalent of central nervous system (CNS) astrocytes, demonstrate macrophage-like activation and cytokine production (10). In these transgenic mice, whose glial cells had been engineered to express a herpes simplex thymidine kinase under a glial fibrillary acidic protein (GFAP) promoter (GFAP-HSV-Tk mice), the administration of the antiherpetic drug ganciclovir ablated almost all of the enteric neural glial cells (11). This unexpectedly induced a severe pathology, maximal in the terminal ileum, with both necrotic and inflammatory components. Transmural histopathological changes included epithelial disruption, fibrosis of the musculature, vasodilation, intraluminal haemorrhage, necrosis, and perforation of the bowel wall (11).
Similar findings were seen in a second model of enteric glial damage, in this case with glial disruption induced by infiltration of autoreactive T cells (12). Histological changes in both models were similar to those seen in human NEC, suggesting that disruption of enteric glia may similarly induce an NEC-like lesion in humans. Intriguingly, intestinal perforation in typhoid fever also shows substantial disruption of enteric glia with perineural T-cell infiltration (13). On the basis of these findings, and the emerging evidence for a potentially critical role for enteric glia in maintaining intestinal integrity (10), we have examined enteric glial and neural cells in a cohort of infants with NEC, and have additionally looked for evidence of glial chemokine and cytokine production and for local recruitment of inflammatory cell populations.
This study was based on analysis of resected specimens from preterm infants admitted to a tertiary paediatric surgical unit with a diagnosis of NEC of sufficient severity to require surgery. Ethical approval was given by the Hammersmith Hospital Local Research Ethics Committee. Four of the patients were reported previously in an analysis of epidermal growth factor receptor expression (14).
Full-thickness resected specimens were studied from 20 infants (12 girls, 8 boys) with a clinical diagnosis of acute NEC (Bell stage 3). Their mean gestational age was 29.5 weeks (range 24–38 weeks) and mean birth weight was 1300 g (480–2700 g), undergoing surgery at the mean age of 15 days (3–28 days). The distribution of disease was small intestinal (ileal) in 11 of 20 cases, ileocaecal in 8, and colonic in 1 case. Follow-up specimens were obtained at ileostomy closure in 6 cases, of whom 2 showed persistent inflammation.
Non-NEC controls studied included resected tissue from 13 preterm infants (4 girls, 9 boys), of mean gestational age 36 weeks (range 29–38 weeks), mean birth weight 2430 g (range 900–3700 g), and postnatal age at surgery 14.5 days (range 1–60 days). Indications for surgery included small bowel atresia (6), perforation (3), volvulus (1), and closure of ileostomy or colostomy (3).
Immunohistochemical and Histochemical Analyses
Serial transverse full-thickness biopsies were obtained from the resected intestinal segments previously fixed in formalin and embedded in paraffin and stained with haematoxylin and eosin to confirm the typical histopathological features of NEC. Paraffin-embedded sections were dewaxed and antigen retrieval was performed using Target Retrieval Solution (Dako Corporation, Oxford UK). Biotin/avidin immunohistochemistry (Vectastatin Elite, Vector Laboratories, Peterborough, UK) was used for peroxidase immunohistochemistry, with endogenous peroxidase blocked with 1% H2O2 in Tris-buffered saline. Primary antibodies included rabbit and mouse anti-CD 3 (dilutions 1:75, 1:50), and monoclonals recognising the cell proliferation markers Ki-67 (1:50) and human leucocyte antigen-DR (HLA-DR) (1:40), all from Dako (UK) and syndecan-1 (1:50, Serotec, Kidlington, UK). Neural markers studied included monoclonal anti-human neuron-specific enolase, (dilution 1:200), neurofilament protein (NFP) (1:75), glial fibrillary acidic protein (GFAP) (1:25), and nerve growth factor receptor (NGFR) (1:100), all from Dako. Cytokine and chemokine localisation was performed using monoclonals recognising human interleukin (IL)-12, transforming growth factor-β1, CCL20, and CCR6 (all R&D Systems, Abingdon, UK), as we have previously reported (15–17). Our previous study of CCL20 and CCR6 expression with these monoclonals showed findings concordant with gene array, reverse transcriptase-polymerase chain reaction, and flow cytometric analysis (17).
Quantitation of staining was performed blinded to patient details, although gross histological features necessarily limited true blinding. Quantitation of CD3+ cell density within lamina propria and submucosa was performed on a semiquantitative scale from 0 (absent) to 4 (dense infiltration). CD3+ cells per ganglion were quantitated in all visible ganglia and a mean was determined for each specimen. Staining intensity for CCL20 and CCR6 was determined on a semiquantitative scale from 0 (absent) to 4 (intensely stained), as previously reported (17).
Data were quantified as median with interquartile range for semiquantitative scale data, and as mean ± standard error (SE) for the cell quantitation. Differences between groups were determined using the Mann-Whitney U test. A difference of P < 0.05 was considered significant.
Routine histopathology confirmed NEC in all of the study patients. Among the control specimens, 1 case of perforation showed epithelial autolytic change, whereas all of the others showed regions of normal intestinal architecture, adjacent to atresias in 6 cases. Of 20 specimens with acute NEC, infiltration of the myenteric ganglia with mononuclear cells was identified in 16 cases, in comparison with 6 of 13 controls.
Among the controls, the 3 specimens taken at closure of the ileostomy showed evidence of mild prestomal ileitis, with transmural infiltration of CD3+ T cells, including within the muscular layers and around the myenteric ganglia. Myenteric ganglia in NEC showed significantly greater T-cell density compared with controls (mean 6.9 ± 1.0 CD3 cells/ganglion compared with 3.2 ± 0.5, P = 0.015). Staining for HLA-DR identified focal aggregation of DR+ cells around the myenteric glia in NEC, in some cases matching infiltrates of DR+ cells within the muscular layers, and in other cases standing out in isolation without significant HLA-DR+ cell infiltrate locally (Fig. 1).
The chemokine CCL20 was expressed with moderate intensity on surface and crypt epithelium in both patients with NEC and controls, with no significant differences between groups (data not shown). By contrast, CCL20 staining intensity was significantly upregulated on glial cells within the myenteric plexus in NEC (median score 2.5, interquartile range [IQR] 2–3 vs median score 1.5, IQR 1–2 in controls; P < 0.001). Representative staining is shown in Figure 1. The percentage of myenteric plexi showing CCL20 immunoreactive glia was also increased in NEC (median 65%, IQR 40%–70% vs median 5%, IQR 2%–15% in controls; P < 0.0001). In addition, the mean number of CCL20 glial cells within the myenteric plexus was also increased in NEC (mean 8.8 ± 0.54 vs 4.4 ± 1 in controls; P < 0.002).
Study of the cognate receptor for CCL20, CCR6, showed a similar increase in glial staining intensity (median 3, IQR 3–3 vs median 2.5, IQR 2–3; P < 0.05), percentage of glia showing immunoreactivity (median 70%, IQR 40%–80% vs median 30%, IQR 15%–45% in controls; P < 0.05) and numbers of CCR6+ cells within glia (16.6 ± 2 vs 9.8 ± 1.9 in controls; P < 0.05).
Staining with neural markers demonstrated no disruption of enteric innervation in any of the control specimens, whether or not there was evidence of mucosal inflammation. Thus, myenteric and submucous plexi were readily identified, and individual nerve fibrils could be seen (Fig. 2). Staining was consistent between the various neural monoclonals (neuron-specific enolase, NFP, and NGFP). Localisation of GFAP identified strong immunoreactivity within enteric glial cells in the myenteric plexus in controls. By contrast, neural disruption was identified frequently in NEC (Fig. 2). Among 20 cases with NEC, myenteric and submucous plexi were fully intact in 8 cases and patchily attenuated in 1. Substantial disruption of submucous plexi and mucosal nerves was identified in 11 of 20 cases. Thus, 2 patterns of disease were seen in NEC, with either preserved (9/20) or disrupted (11/20) mucosal and submucosal nerves. In some cases, the neural disturbance extended to the deep submucosa, even though inflammation appeared limited to the mucosa.
No difference in disease severity was noted between the 2 groups, and severe ulcerating disease with necrotic villi could be seen in cases with either fully intact or extensively disrupted enteric nerves (Fig. 2); however, the extent of neural disruption was associated with loss of GFAP expression among myenteric plexus glial cells. In the cases with disrupted enteric neural supply, glial cell numbers were reduced within myenteric plexi, and NGFR staining within the plexi showed marked disruption (Fig. 2). Despite the differences in glial cell numbers, no differences were seen between NEC cases and controls, or between neurally disrupted and intact cases with NEC in expression of IL-12, transforming growth factor-β, or syndecan-1, which were strongly expressed by all glial cells.
Analysis of the specimens obtained at ileostomy closure showed restoration of neural architecture in 4, but persistent disruption in the 2 cases with ongoing inflammatory change. Expression of CCL20 within the myenteric plexus had reduced to minimal levels in the 4 cases without inflammation, although the 2 cases with ongoing inflammatory change showed persisting CCL20 expression (70% and 30% plexi immunoreactive, staining intensity 3 and 2.5, mean numbers of CCL20+ cells within plexi 14 and 7, respectively).
This study identifies heterogeneity within the cases of NEC in respect of enteric neural disruption. In approximately half of the cases studied, there was evidence of extensive disruption of enteric nerves, associated with severe abnormalities within the myenteric plexus, characterised by loss of enteric glial cells and disruption of NGFR expression. These changes were associated with marked upregulation of the mucosal chemokine CCL20 by myenteric plexus glial cells and infiltration of cells expressing its cognate receptor CCR6. Among the 6 cases for whom repeat specimens at the time of ileostomy closure were available, these changes of glial chemokine production and enteric neural disruption persisted in only the 2 with ongoing inflammation. Persistent absence of ganglion cells has previously been reported in a proportion of children with late stricturing after NEC (18). Although several of the control specimens showed evidence of some inflammatory change, particularly those with prestomal ileitis, there was no evidence of enteric neural disruption.
Our work supports and extends previous observations of enteric neural abnormality in NEC. A study by Wedel et al (19) of 3 cases of NEC by whole-mount immunohistochemistry identified disruption of myenteric plexus architecture, with reduced numbers of glial cells and evidence of neural degeneration. Further study by the same group in 8 cases and 3 controls demonstrated absence of immunoreactive vasoactive intestinal peptide and nitric oxide synthase within the circular muscle layer and submucosal plexus (20). Our findings show that such neural abnormality occurs in a significant subgroup of infants with NEC, and is associated with activation of enteric glial cells and production of the chemokine CCL20.
CCL20 (macrophage inflammatory protein-3α, liver activation–regulated chemokine) is a chemokine that is particularly implicated in the homing of dendritic cells and T-cell subsets to mucosal surfaces via expression of its specific receptor CCR6 (21–23). Its expression plays a particular role in mucosal TH17 responses (23). CCL20 production has also been reported within glial cell populations within the CNS (24). Astrocytes, the CNS equivalents of enteric glia, were the major cellular source of CCL20 in experimental allergic encephalomyelitis, inducing recruitment of CCR6+ dendritic cells as well as memory T cells and B cells (25). Our findings suggest that the CCL20-CCR6 axis may play an analogous proinflammatory role within glial cells in NEC.
The authors and others have demonstrated nuclear factor-κB (NF-κB)–mediated induction of CCL20 production in gut and skin epithelial cells by tumour necrosis factor-α and IL-1β (17,26). These cytokines are also dominant inducers of CCL20 production by cultured astrocytes (24). NF-κB signalling in macroglia is also upregulated by other factors relevant for ill preterm infants, including oxidative stress and hypoxia (27,28). Progression to neural death is promoted in such circumstances through enhanced sensitivity to induced nitric oxide (29). More important, unconjugated bilirubin may not only induce astrocyte NF-κB activation but also trigger death of immature neurons, particularly synergising with bacterial LPS (30,31). Thus, the progression from glial activation and chemokine production to neural cell death in an infant may represent the result of synergy between several factors. The cause of enteric glial activation remains unknown, but may potentially include bacteria such as Escherichia coli (32) or viruses, as the dominant Toll-like receptor in macroglia is Toll-like receptor 3, which recognises double-stranded RNA (33–35).
We emphasise that myenteric plexus destruction and enteric neural loss only occurred in half of the patients studied, and severe disease occurred in infants without evidence of neural disruption. It is clear from a variety of animal models and clinical studies that immaturity, epithelial barrier disruption, the luminal flora, and formula feeding are important factors in determining the development of NEC. Our findings however build upon proof-of-principle murine studies to suggest that glial cells within the myenteric plexus may be a previously unsuspected additional site of disease pathogenesis, and that the characteristic disturbance of motility in NEC may reflect profound underlying neural disruption, potentially induced by systemic as well as luminal factors.
1. Gephart SM, McGrath JM, Effken JA, et al. Necrotizing enterocolitis
risk: state of the science. Adv Neonatal Care
2. Neu J, Walker WA. Necrotizing enterocolitis
. N Engl J Med
3. Dominguez KM, Moss RL. Necrotizing enterocolitis
. Clin Perinatol
4. Ward RM, Beachy JC. Neonatal complications following preterm birth. Br J Obstetr Gynaecol
5. Hsueh W, Caplan MS, Qu XW, et al. Neonatal necrotizing enterocolitis
: clinical considerations and pathogenetic concepts. Pediatr Dev Pathol
6. Lin J, Hackam DJ. Worms, flies and four-legged friends: the applicability of biological models to the understanding of intestinal inflammatory diseases. Dis Model Mech
7. Caplan MS, Simon D, Jilling T. The role of PAF, TLR, and the inflammatory response in neonatal necrotizing enterocolitis
. Semin Pediatr Surg
8. Treszl A, Tulassay T, Vasarhelyi B. Genetic basis for necrotizing enterocolitis
--risk factors and their relations to genetic polymorphisms. Front Biosci
9. Hackam DJ, Upperman JS, Grishin A, et al. Disordered enterocyte signaling and intestinal barrier dysfunction in the pathogenesis of necrotizing enterocolitis
. Semin Pediatr Surg
10. Savidge TC, Sofroniew MV, Neunlist M. Starring roles for astroglia in barrier pathologies of gut and brain. Lab Invest
11. Bush TG, Savidge TC, Freeman TC, et al. Fulminant jejuno-ileitis following ablation of enteric glia
in adult transgenic mice. Cell
12. Cornet A, Savidge TC, Cabarrocas J, et al. Enterocolitis induced by autoimmune targeting of enteric glial cells: a possible target in Crohn's disease? Proc Natl Acad Sci USA
13. Chanh NQ, Everest P, Khoa TT, et al. A clinical, microbiological and pathological study of intestinal perforation associated with typhoid fever. Clin Infect Dis
14. Fagbemi AO, Wright N, Lakhoo K, et al. Immunoreactive epidermal growth factor receptors are present in gastrointestinal epithelial cells of preterm infants with necrotising enterocolitis. Early Hum Dev
15. Pérez-Machado MA, Ashwood P, Thomson MA, et al. Reduced transforming growth factor-β1-producing T cells in the duodenal mucosa of children with food allergy. Eur J Immunol
16. Pérez-Machado MA, Ashwood P, Torrente F, et al. Spontaneous TH1 cytokine production by intraepithelial but not circulating T cells in infants with or without food allergies. Allergy
17. Puleston J, Cooper M, Murch S, et al. A distinct subset of chemokines, induced in colonic epithelium by IL-1β, dominates the mucosal chemokine response in ulcerative colitis. Aliment Pharmacol Therapeut
18. Schimpl G, Hollwarth ME, Fotter R, et al. Late intestinal strictures following successful treatment of necrotizing enterocolitis
. Acta Paediatr
19. Wedel T, Krammer HJ, Kuhnel W, et al. Alterations of the enteric nervous system in neonatal necrotizing enterocolitis
revealed by whole-mount immunohistochemistry. Pediatr Pathol Lab Med
20. Sigge W, Wedel T, Kuhnel W, et al. Morphologic alterations of the enteric nervous system and deficiency of non-adrenergic non-cholinergic inhibitory innervation in neonatal necrotizing enterocolitis
. Eur J Pediatr Surg
21. Izadpanah A, Dwinell MB, Eckmann L, et al. Regulated MIP-3α/CCL20
production by human intestinal epithelium: mechanism for modulating mucosal immunity. Am J Physiol Gastrointest Liver Physiol
22. Ito T, Carson WF 4th, Cavassani KA, et al. CCR6 as a mediator of immunity in the lung and gut. Exp Cell Res
23. Esplugues E, Huber S, Gagliani N, et al. Control of TH17 cells occurs in the small intestine. Nature
24. Ambrosini E, Remoli ME, Giacomini E, et al. Astrocytes produce dendritic cell-attracting chemokines in vitro and in multiple sclerosis lesions. J Neuropathol Exp Neurol
25. Ambrosinin E, Columba-Cabezas S, Serafini B, et al. Astrocytes are the major intracerebral source of macrophage inflammatory protein-3α/CCL20
in relapsing experimental autoimmune encephalomyelitis and in vitro. Glia
26. Fujie S, Hieshima K, Izawa D, et al. Proinflammatory cytokines induce liver and activation-regulated chemokine/macrophage inflammatory protein-3α/CCL20
in mucosal cells through NF-κB. Internat Immunol
27. Choi K, Krushel LA, Crossin KL. NF-κB activation by N-CAM and cytokines in astrocytes is regulated by multiple protein kinases and redox modulation. Glia
28. Zhang W, Petrovic JM, Callaghan D, et al. Evidence that hypoxia-inducible factor-1 (HIF-1) mediates transcriptional activation of interleukin-1β (IL-1β) in astrocyte cultures. J Neuroimmunol
29. Mander P, Borutaite V, Moncada S, et al. Nitric oxide from inflammatory-activated glia synergizes with hypoxia to induce neuronal death. J Neurosci Res
30. Falcao AS, Fernandes A, Brito MA, et al. Bilirubin-induced inflammatory response, glutamate release, and cell death in rat cortical astrocytes are enhanced in younger cells. Neurobiol Dis
31. Falcao AS, Fernandes A, Brito MA, et al. Bilirubin-induced immunostimulant effects and toxicity vary with neural cell type and maturation state. Acta Neuropathol
32. Kim JM, Oh YK, Lee JH, et al. Induction of proinflammatory mediators requires activation of the TRAF, NIK, IKK and NF-κB signal transduction pathway in astrocytes infected with Escherichia coli. Clin Exp Immunol
33. Jack CS, Arbour N, Manusow J, et al. TLR signaling tailors innate immune responses in human microglia and astrocytes. J Immunol
34. Scumpia PO, Kelly KM, Reeves WH, et al. Double-stranded RNA signals antiviral and inflammatory programs and dysfunctional glutamate transport in TLR3-expressing astrocytes. Glia
35. Boccia D, Stolfi I, Lana S, et al. Nosocomial necrotising enterocolitis outbreaks: epidemiology and control measures. Eur J Pediatr