Main

NEC is a serious gastrointestinal disorder of unknown etiology affecting newborn infants, particularly those born prematurely(1,2). NEC usually involves the terminal ileum and results in necrosis with hemorrhage, mucosal edema, and limited inflammation(3). The factors predisposing to NEC include intestinal injury from ischemia and/or hypoxia, bacteria and/or bacterial toxins, feeding, and prematurity with its associated immaturity of local innate defenses(4). Although controversial, a leading hypothesis also suggests a link between infection and NEC(4). Among the products thought to be integral to innate host defenses are antibiotic peptides. Antibiotic peptides are found in circulating phagocytic cells and epithelial cells of many animal species (for reviews, see5–8). Defensins are a family of antibiotic peptides that have a broad spectrum of antibiotic activity and share a characteristic six-cysteine array(9). In humans, six defensins have been identified(10–13). Four defensins (HNP 1-4) have been identified in the azurophilic granules of neutrophils, where they participate in the nonoxidative killing of engulfed microorganisms(14,15). More recently, expression of the two other human defensins (HD5 and HD6) has been identified in Paneth cells, one of four major types of epithelial cells of the human small intestine(12,13). Paneth cells of the mouse have been-shown to produce several enteric defensins, alternatively referred to as cryptdins to highlight their crypt origin(16,17). Paneth cells are located at the base of the crypts of Lieberkühn, with highest abundance in the ileum. These cells, which contain numerous apically located secretory granules, have been implicated in intestinal host defense by various studies [see Qu et al.(18) and references therein].

A developmental profile of the expression of human enteric defensins, HD5 and HD6, demonstrates that enteric defensin mRNA are expressed at extremely low levels during fetal life compared with term newborns and adults(19). Low level expression of this component of the innate immune system may contribute to the immaturity of intestinal innate defenses, a factor thought to predispose to the development of NEC. To initiate studies on the possible involvement of enteric defensins in the pathophysiology of NEC, we characterized defensin mRNA and peptide expression in the small intestine of patients with NEC and compared the levels to those in newborn control subjects.

METHODS

Study population and histologic evaluation. Six cases of NEC were selected for this study based on the availability of small intestinal tissue obtained by operative exploration and intestinal resection, performed at The Children's Hospital of Philadelphia. The diagnosis of NEC was based on clinical findings and pathologic examination. Normal small intestine for control subjects was obtained from four cases of intestinal atresia and one case of meconium ileus. The NEC and control specimens were taken from a variety of small intestinal sites, including both ileum and jejunum. Some samples were designated only as small intestine. Subjects characteristics and their histologic diagnoses are presented in Table 1 (see"Results").

Table 1 Study population
Full size table

Immunohistochemistry. Freshly obtained small intestinal tissue was fixed in 10% neutral buffered formalin before paraffin embedding. The tissue processing and methods for immunohistochemistry were as described previously(20). Briefly, sections were cut (5 µm) and collected on silylated slides (PGC Scientifics, Gaithersburg, MD). Sections were deparaffinized, hydrated in a graded series of alcohols, and washed in PBS, pH 7.4. The sections were blocked with 5% normal goat serum, and then incubated in the rabbit polyclonal HD5 antiserum(21) (diluted 1:25 000) for 20 h at 4°C in a humidified chamber. Sections were again washed in PBS, blocked with goat serum, followed by incubation with biotinylated goat anti-rabbit (diluted 1:200; Vector Laboratories, Inc., Burlingame, CA) for 20 min at room temperature. Five percent normal goat serum was used as a diluent for antisera. The sections were then treated with avidin-biotin complex, incubated with 3,3′-diaminobenzidine substrate for 8 min at room temperature, followed by light green counterstain.

In situ hybridization. The methods for in situ hybridization and for the synthesis of the HD5 and HD6 probes were as described by Mallow et al.(19). Collagen IVa probe was generated from plasmid KK4(22) generously provided by Dr. Jeanne Myers (Department of Biochemistry and Biophysics, University of Pennsylvania School of Medicine). The plasmid yielded a riboprobe comprising the collagenous domain of collagen IVa. To control for possible day to day quantitative variation of the in situ hybridization protocol, the sections of each experiment were processed together and probed in parallel with the same preparation of probe.

Image analysis. Measurements were made on each of the control and NEC specimens, using the sections from the in situ hybridization analysis. Quantitative comparisons were made on sections processed together and probed with the same preparation of labeled probe for internal consistency. Photographic emulsion, when exposed to the radioactive emission of the riboprobe, results in the deposition of dark silver grains, which appear as black dot aggregates when visualized by light microscopy. Directly over the radioactive source, silver grains will be deposited with greatest density. In cases of abundant probe hybridization, the density of silver grains may reach the point of saturation, blunting the detectable read out. Because the radioactive decay will also deflect at various angles and cause a wider zone of silver grain deposition observed for cells with abundant probe hybridization, we therefore used areas of signal as an approximation of the signal strength. This will underestimate the true magnitude of signal from cells with abundant probe and afford a conservative approximation of signal intensity differences. Measurements were made for both HD5 and HD6 probes independently. Four microscopic fields of each tissue were selected randomly. The fields were scanned using a Hamamatsu silicon intensified tube camera attached to a Nikon optiphoto microscope. The images were captured and digitized in a 640 × 580 pixel field using ARGUS imaging software (Hamamatsu) in a PC compatible computer. The threshold values were constant within each experiment analyzed. The area of the signals over the minimum threshold was measured. The minimum area measured was 50 pixels to avoid interference from nonspecific background noise. Analysis of images was performed on a Power Mac 7100/66 using National Institutes of Health Image Software (written by Wayne Rasband at the U.S. National Institutes of Health and available from the internet by anonymous ftp from ZIPPY.NIMH.NIH.GOV or on floppy disk from NTIS, 5285 Port Royal Rd., Springfield, VA 22161, Part no. PB93-504868). The numbers of Paneth cells associated with each field were counted on a parallel section stained with phloxine/tartrazine which highlights Paneth cells(23). A value of signal area (signal strength) per Paneth cell was calculated for each specimen (Xsig).

Statistical analysis. All analyses were performed using a Statistical Package for the Social Science Version PC 6.01 (SPSS Inc., Chicago, IL). Means of signal strength per Paneth cell (Xsig) were compared between the NEC and control samples individually for probes HD5 and HD6 using analysis of variance and covariance. The number of cells was considered as a covariance. The level of statistical significance was set at p < 0.05.

RESULTS

NEC study population and controls.Table 1 shows the clinical characteristics of the subjects studied. Tissue sections from six infants with necrotizing enterocolitis were evaluated. All of these subjects were premature, with gestational ages ranging from 25 to 31 wk at birth. The surgical specimens were obtained between 5 and 45 d postnatally. Tissue sections from the control group included four cases of atresia, and one case of meconium ileus. The ages of these infants ranged from 35 to 40 wk of gestation. These surgical specimens were obtained between 1 and 2 d postnatally.

HD5 peptide expression in human small intestinal development. In addition to the described study population, tissue sections of human terminal ileum from an adult, term newborn, and 24-wk gestation fetus were stained with rabbit polyclonal antiserum to HD5. Positive granular cytoplasmic staining was observed in the Paneth cells in each specimen studied (Fig. 1), suggesting localization to secretory granules. No signal was observed if preimmune serum was substituted for the primary antibody (Fig. 1, inset). Examination of multiple microscopic areas in each specimen demonstrated that all Paneth cells stained strongly with the anti-HD5 antibody. HD5 peptide is detectable at 24 wk of gestation, but in lower abundance than at term and considerably lower than in the adult (Fig. 1).

Figure 1
figure 1

HD5 peptide expression in human small intestinal development. Terminal ileum tissue was stained with polyclonal rabbit HD5 antiserum in (A) adult, (B) term newborn, and (C) 24-wk fetus, using a counterstain of light green. Arrows denote positive staining of Paneth cells by polyclonal HD5 antibody. Control slide of adult tissue was stained with preimmune serum and counterstained with light green (A, inset). Parallel tissue sections were stained with hematoxylin and eosin (D-F). The bar equals 100 µm.

Full size image

Comparison of the expression of defensin mRNA in NEC versus controls. The cellular expression of HD5 and HD6 mRNA was determined by in situ hybridization. Despite the variety of specimen locations and the a wide distribution of ages in the cases of NEC (considering times both postconception and postnatal, Table 1) there was little internal variation in the defensin mRNA signal strength(Figs. 2, A and B, and 4). Similar consistency was observed for the control cases (Figs. 2,E and F, and 4). Uniformly, the defensin mRNA signal strength, represented by the area of aggregated black silver grains, was strikingly increased in the cases of NEC compared with controls (Figs. 2,A versus E, B versus F, and 4). Parallel sections probed with an antisense riboprobe to collagen IVa shows similar levels of hybridization signal in NEC and control, suggesting that the pathologic process of NEC does not cause a generalized nonspecific increase in mRNA in the affected tissue(Fig. 2,C and G). No signal was observed when the sections were pretreated with RNase before hybridization (Fig. 2, D and H) or with the use of the sense riboprobes (not shown).

Figure 2
figure 2

Elevated HD5 mRNA levels in NEC compared with atresia. Paraffin-embedded sections of small intestine from one representative specimen of NEC (A-D) and atresia (E-H) were hybridized with 35S-UTP-labeled HD5 antisense riboprobe(A, E), 35S-UTP-labeled HD6 antisense riboprobe (B, F), and 35S-UTP-labeled collagen IVa antisense riboprobe (C, G). Sections were pretreated with RNase A before hybridization with HD5 antisense riboprobe (D, H). C and G are dark-field images where the white dots represent a positive signal. The remaining panels are bright-field images where the dense black silver grains represent a positive signal. Small arrows denote positive antisense hybridization and therefore the presence of defensin mRNA in small intestinal Paneth cells. The large arrows denote positive antisense hybridizations and therefore the presence of collagen IVa mRNA in the lamina propria of the small intestine. The bar equals 100 µm.

Full size image
Figure 4
figure 4

Comparison of HD5 mRNA and peptide in specimens from NEC and atresia. Parallel sections of each atresia specimen(columns 1 and 2) and each NEC specimen (columns 3 and 4) were hybridized with 35S-UTP-labeled HD5 antisense riboprobe and counterstained with hematoxylin and eosin (columns 1 and 3) and stained with polyclonal HD5 antiserum and counterstained with light green (columns 2 and 4). Arrows denote positive staining of Paneth cells by polyclonal HD5 antibody. The bar equals 100 µm.

Full size image

Image analysis. To quantitatively compare representative levels of defensin mRNA in NEC and in controls, image analysis was performed. Four low power microscopic fields from each specimen were digitized, and the areas of aggregated silver grains were measured. A value of average signal strength per Paneth cell (Xsig) was calculated for each specimen and for each probe (Table 2). The Xsig for NEC was approximately 3-fold greater for NEC than were control specimens (Fig. 3A, Table 2). The mean number of Paneth cells (n) was increased approximately 2-fold in the cases of NEC compared with controls(Fig. 3B). Therefore both the numbers of Paneth cells and the amount of defensin mRNA per Paneth cell are increased in association with NEC.

Table 2 Statistical analysis
Full size table
Figure 3
figure 3

Statistical analysis of (A) signal intensity per Paneth cell (Xsig) in NEC vs control and (B) average numbers of Paneth cells(n) in NEC vs control, using probes for HD5 and HD6 mRNA. Signal intensity was measured using image analysis (see "Methods"), and statistical analysis was performed using SPSS (Version PC 6.01). The error bars represent ± SD. The level of statistical significance was set at p < 0.05.

Full size image

Comparison of intracellular HD5 peptide and mRNA expression in NEC versus control. Parallel sections of specimens from patients with NEC and controls were stained with anti-HD5 antiserum and probed with HD5 antisense riboprobe (Fig. 4). Antibody staining of Paneth cells coincided with the positive hybridization signal in the parallel sections. Although there was a significant increase in HD5 mRNA in NEC compared with control subjects, the intracellular peptide detected appeared uniformly equivalent, indicating that in NEC the levels of intracellular peptide are not elevated in conjunction with the rise in defensin mRNA. These findings contrast with those in Figure 1, where levels of detectable HD5 parallel the relative level of HD5 mRNA in Paneth cells(19).

DISCUSSION

One prevailing hypothesis on the pathophysiology of NEC invokes immaturity of innate mucosal defense of the intestine(24). Previous work in our laboratory has analyzed the developmental profile of enteric defensins, HD5 and HD6(19). Expression of HD5 and HD6 mRNA was detected at 13.5 wk of gestation(19), nearly coincident with the appearance of Paneth cells in early ontogeny(25). At 24 wk of gestation, HD5 and HD6 mRNA localize to Paneth cells of the small intestinal crypt, but at lower levels than those seen in the full-term infant, and at a level approximately 200-fold lower than in the adult(19). At the time of our previous study(19), an antibody to enteric defensin was not available. Now, using immunohistochemistry we have found evidence of intracellular HD5 peptide in Paneth cells at 24 wk of gestation(Fig. 1). The intracellular levels of peptide correlated with the increase in defensin mRNA during development, with low levels at 24 wk of gestation, and higher levels at term. This suggests that the low level of defensin expression characteristic of normal intestinal development may contribute, in part, to the overall immaturity of the intestinal defenses which predisposes the premature infant to NEC.

These data also address the requirement of an exogenous stimulus for enteric defensin expression. The expression of defensin mRNA and peptide in utero, in the absence of stimulation by any inflammatory or infectious source, suggests a baseline level of constitutive expression. Constitutive expression of the neutrophil defensins, HNP 1-4, has been observed in a window of neutrophil maturation, with mRNA levels highest in promyelocytes(26).

In cases of NEC, the levels of enteric defensin mRNA expression are significantly elevated over that observed in controls (Figs. 2 and 4). This suggests that at least during the stages at which the defensins have not yet achieved adult levels, they can be induced by some exogenous trigger, such as that provided by the process of NEC. The ongoing process of NEC may also act to promote the differentiation of the crypt stem cell to produce greater numbers of Paneth cells. In these experiments the number of Paneth cells increased approximately 2-fold in NEC compared with control subjects (Table 2). Although we suggest that this increase in number of Paneth cells reflects an intestinal response to NEC, it is also possible that increased numbers of Paneth cells might contribute to the pathophysiology of the disease.

A caveat to these latter points should be addressed. Truly age-matched controls are not readily available. The tissue from the control subjects was taken 1 or 2 d postnatally, whereas the tissue from the subjects with NEC was taken from 5 d to several weeks postnatally. Viewed by this criterion, our data are consistent with the notion that the increased numbers of Paneth cells in NEC samples resulted from differences in postnatal age. However, the subjects with NEC were born prematurely, and even at the point of surgical sampling, their age postconception was significantly less than that of the normal control subjects. Therefore, based on ontogenic considerations we would have expected to see fewer Paneth cells in the infants with NEC. Given an increased number of Paneth cells in the samples from NEC patients(Table 2), we favor the interpretation that the increase is a consequence of the pathologic process. In addition, an increase in defensin expression might have been accounted for by the postnatal ages of the patients with NEC; however, the measurements by image analysis were consistent between samples and did not reflect the wide variation in postnatal (or postconception) ages of the patients with NEC. This further supports the interpretation that the increase in defensin levels and Paneth cell numbers is better attributed to the pathologic process of NEC.

In tissues analyzed from normal subjects, intracellular peptide levels correlated with defensin mRNA levels during development (Fig. 1). In contrast, the intracellular levels of HD5 peptide appeared constant despite the increase seen in the HD5 mRNA in the NEC samples(Fig. 4). Immunohistochemistry of secretory proteins is an observation of a steady state process and does not represent levels of peptide production unless there is no degradation or secretion of the peptide. The most attractive explanation for the increased mRNA with no concomitant change in steady state peptide levels is that the peptide is being actively secreted in this setting and that the Paneth cells maintain a constant intracellular level of peptide. An alternative interpretation of our data are that defensin mRNA translation is inhibited. This explanation, although plausible, seems less likely because translational regulation has not been previously described for antibiotic peptides, and would appear counter productive to an increase in mRNA level.

In conclusion, low levels of enteric defensin expression are normally found at gestational ages corresponding to those of premature neonates, a population at risk for NEC. We have found that in normal development the level of intracellular HD5 peptide in Paneth cells increases coordinately with an increase in its corresponding mRNA. This increase most likely occurs as part of a constitutive developmental timetable, rather than in response to exogenous signals. However, the levels of enteric defensin mRNA are significantly increased in association with NEC, demonstrating that the expression of HD5 and HD6 mRNA are inducible under these circumstances. Paneth cells, which produce and secrete HD5 and HD6, are also significantly increased in number in NEC. Our data comparing HD5 mRNA and peptide levels in patients with NEC suggest that the HD5 peptide may be secreted in response to the ongoing disease process. Taken together, these results propose an association between enteric defensin expression and NEC, suggesting that enteric defensins may be involved in the pathophysiology of this disease.