Gut mucosal injury in neonates is marked by macrophage infiltration in contrast to pleomorphic infiltrates in adult: evidence from an animal model

Abstract

Necrotizing enterocolitis (NEC) is an inflammatory bowel necrosis of premature infants. In tissue samples of NEC, we identified numerous macrophages and a few neutrophils but not many lymphocytes. We hypothesized that these pathoanatomic characteristics of NEC represent a common tissue injury response of the gastrointestinal tract to a variety of insults at a specific stage of gut development. To evaluate developmental changes in mucosal inflammatory response, we used trinitrobenzene sulfonic acid (TNBS)-induced inflammation as a nonspecific insult and compared mucosal injury in newborn vs. adult mice. Enterocolitis was induced in 10-day-old pups and adult mice (n = 25 animals per group) by administering TNBS by gavage and enema. Leukocyte populations were enumerated in human NEC and in murine TNBS-enterocolitis using quantitative immunofluorescence. Chemokine expression was measured using quantitative polymerase chain reaction, immunoblots, and immunohistochemistry. Macrophage recruitment was investigated ex vivo using intestinal tissue-conditioned media and bone marrow-derived macrophages in a microchemotaxis assay. Similar to human NEC, TNBS enterocolitis in pups was marked by a macrophage-rich leukocyte infiltrate in affected tissue. In contrast, TNBS-enterocolitis in adult mice was associated with pleomorphic leukocyte infiltrates. Macrophage precursors were recruited to murine neonatal gastrointestinal tract by the chemokine CXCL5, a known chemoattractant for myeloid cells. We also demonstrated increased expression of CXCL5 in surgically resected tissue samples of human NEC, indicating that a similar pathway was active in NEC. We concluded that gut mucosal injury in the murine neonate is marked by a macrophage-rich leukocyte infiltrate, which contrasts with the pleomorphic leukocyte infiltrates in adult mice. In murine neonatal enterocolitis, macrophages were recruited to the inflamed gut mucosa by the chemokine CXCL5, indicating that CXCL5 and its cognate receptor CXCR2 merit further investigation as potential therapeutic targets in NEC.

necrotizing enterocolitis (NEC), an inflammatory bowel necrosis of preterm infants, is a leading cause of death among neonates born before 32 wk of gestation or with a birth weight <1,500 g (24, 32). Although the etiology of NEC remains unclear, epidemiological studies show an association with diverse risk factors such as maternal chorioamnionitis, perinatal asphyxia, indomethacin therapy, viral infections, and blood transfusions (24). Current pathophysiological models suggest that NEC occurs when altered/disrupted gut mucosal barrier in the preterm intestine allows luminal bacteria to translocate across the epithelial barrier into the lamina propria, triggering a severe mucosal inflammatory response and tissue damage (25).

The present study was designed to investigate the cellular inflammatory response in NEC. In surgically resected tissue samples of human NEC, we observed that these infiltrates were comprised predominantly of macrophages along with a few neutrophils but not many lymphocytes. Existing clinical studies indicate that 1) the incidence of NEC peaks in premature infants at a specific postmenstrual age (gestational age at birth + postnatal age) of 32 wk of gestation (21, 46, 59), and 2) NEC is associated with a host of very diverse risk factors that may not share a plausible, unifying mechanism of injury (24, 32). Based on these observations, we hypothesized that the pathoanatomy of NEC is related not to a single etiological pathway but may instead represent a generic tissue injury response of the gastrointestinal tract at a specific stage of development to a variety of insults. To investigate the effect of age on the cellular inflammatory response during gut mucosal injury, we used the haptenic agent 2,4,6-trinitrobenzene sulfonic acid (TNBS) as a nonspecific mucosal insult in 10-day-old murine pups and adult mice. (44). Using this TNBS-injury model, archived human tissue samples of NEC, and other ex vivo models, we show that macrophage-rich leukocyte infiltrates seen in NEC are a characteristic of inflammatory mucosal injury in the developing intestine. In our murine model, the recruitment of macrophage precursors was mediated via CXCL5, which is an important chemoattractant for myeloid cells in the gastrointestinal tract (18, 56). CXCL5 expression was also significantly increased in human tissue samples, indicating that a similar pathway may be at work in NEC.

MATERIALS AND METHODS

Human NEC and controls.

Human intestinal tissues were collected after approval by the local Institutional Review Board at University of Illinois. Deidentified paraffin-embedded tissue sections of NEC were compared with healthy tissue margins resected for indications other than NEC (intestinal obstruction or spontaneous intestinal perforation; n = 12 in each group). Deidentified, frozen tissue samples were available for RNA isolation from 11 patients with NEC and 8 controls.

Mice.

To induce enterocolitis, we administered TNBS (44) in 10-day-old and adult C57/BL6 mice (n = 25 animals per group; animals reared under conventional conditions) by gavage and enema. Animals were anesthetized in an isoflurane chamber, a 3.5-gauge French silicone catheter was inserted into the stomach, the gastric contents were removed, and TNBS (50 mg/kg total body wt dissolved in 30% weight/volume ethanol) was administered by gavage. The catheter was then inserted per rectum to a length of 1–2 cm, and another 50 mg/kg total body wt dose of TNBS was injected slowly as enema. In preliminary experiments, we defined optimum doses and concentrations of TNBS concentrations to limit mortality in pups to ≤30%. Animals were killed using CO₂ inhalation at serial time points up to 48 h after TNBS administration. Control animals received vehicle alone using the same technique described above. In some experiments, we administered TNBS in 10-day-old and adult mice (n = 5 each) reared under germ-free conditions (3). We also evaluated TNBS enterocolitis in a small number of 5-day-old pups (n = 6). Intestinal injury was graded as described previously (6): grade 0: no injury; grade 1: mild separation of lamina propria; grade 2: moderate separation; grade 3: severe separation and/or edema in submucosa; grade 4: transmural injury. Severity of colitis was graded (6) as 0: no inflammation; grade 1: low level leukocyte infiltration seen in <10% high-power fields (HPF), and no structural changes; grade 2: moderate leukocyte infiltration in 10–25% HPF, crypt elongation, mucosal thickening, and no ulcerations; grade 3: high level leukocyte infiltration seen in 25–50% HPF, crypt elongation, infiltration beyond the mucosal layer, thickening of the bowel wall, and superficial ulcerations; and grade 4: marked transmural leukocyte infiltration in >50% HPF, elongated and distorted crypts, bowel-wall thickening, and extensive ulcerations.

Immunohistochemistry.

Human tissues were stained (49) for HAM56 and epithelial-derived neutrophil chemoattractant (ENA)/CXC ligand 5 (CXCL5) using our previously described protocol (25). Briefly, tissue sections were deparaffinized, and antigen retrieval was achieved using the EZ-AR Common solution (Biogenex, San Remon, CA) per the manufacturer's protocol. The slides were then treated with Proteinase K (20 μg/ml) (Promega, Madison, WI) for 10 min at room temperature. The sections were rinsed in PBS (5 min), blocked using SuperBlock T20 blocking buffer (Thermo Scientific, Rockford, IL) for 30 min at room temperature, rinsed again in PBS, and then incubated (overnight, 4°C) in appropriate primary antibody: mouse anti-human pan-macrophage marker (HAM56)IgM (Abcam, Cambridge, MA), monoclonal mouse anti-human CXCL5 (R&D, Minneapolis, MN), polyclonal rat anti-mouse F4/80 (eBiosciences, San Diego, CA), mouse monoclonal anti-human/mouse myeloperoxidase IgG2b (R&D), monoclonal rat anti-mouse CD3 IgG_2a (Santa Cruz Biotechnology, Santa Cruz, CA), goat polyclonal anti-mouse LPS-induced CXC chemokine/CXCL5 (R&D), goat polyclonal macrophage inflammatory protein-2/CXCL2 (R&D), and rabbit polyclonal anti-mouse CXCR2 (Santa Cruz Biotechnology). Secondary staining was performed at room temperature for 30 min with Alexa 488 or Alexa 568-conjugated chicken anti-rat, goat anti-mouse IgM, or rabbit anti-goat antibody (Invitrogen, San Diego, CA). Controls included slides with no primary antibody, appropriate isotype control, and with competing recombinant CXCL5. Cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; Calbiochem, San Diego, CA), diluted 1:1,000 in PBS, applied for 3 min. Imaging was performed using a Zeiss Axiovert fluorescence microscope. Macrophages were enumerated in five randomly chosen HPFs (×40) in the ulcerated area and on the edges of the lesion and normalized against the total number of nuclei in the field.

PCR-denaturing gradient gel electrophoresis.

Qualitative assessment of gut bacterial flora was performed using PCR amplification of the V6-V8 region of bacterial 16S ribosomal DNA followed by denaturing gradient gel electrophoresis (DGGE) (54). Briefly, total genomic DNA from the murine small intestine was extracted using the GenElute Genomic DNA Miniprep Kit (Sigma-Aldrich). Bacterial 16s ribosomal DNA was amplified by PCR using the primers Bac-1 GCCGCCGGGCCGCGGCCCGCCCGCCCGCGGGGG-, CACGGGGGACTACGTGCCAGCAGCC, and Bac2 -GGACTACCAGGGTATCTAATCC in a reaction mixture comprised of 22 mM Tris·HCl, pH 8.4, 55 mM KCl, 1.65 mM MgCl₂, 220 mM each dNTPs, 250 nM primers, 200 ng total small intestinal DNA and 22 U/ml Taq DNA polymerase in a 50-μl reaction. Amplification reactions were performed in 0.2-ml tubes using a Applied Biosystems thermal cycler. The cycling parameters were 1 cycle of 94°C × 8 min, 30 cycles of 94°C × 30 s, 55°C × 30 s, 72°C × 30 s, a 5-min hold at 72°C (Applied BioSystems thermocycler), and a holding temperature of 4°C following the final cycle. PCR products were resolved by DGGE using the Bio-Rad DCode apparatus. The denaturing gel consisted of 8% (vol/vol) polyacrylamide (ratio of acrylamide-bisacrylamide, 37.5:1), and 0.5 × Tris-acetate-EDTA buffer (pH 8.0). One hundred percent denaturing acrylamide was defined as 7 M urea and 40% formamide or 8 M urea, 20% deionized formamide, 2% glycerol, 1× TAE, 0.1% ammonium persulfate, and 0.05% tetramethylethylenediamine. One hour after polymerization, the gel was assembled onto the apparatus. Twenty microliters of the PCR product was loaded, and the gel was run with a denaturing gradient of 30% to 70% urea/formamide at 200 V for 5 min at and then overnight at 85 V in 0.5 × Tris-acetate-EDTA buffer at a constant temperature of 60°C. The gels were stained with AgNO₃ as previously described. To allow a comparison between PCR-DGGE gels, internal standards were used. After completion of electrophoresis, gels were stained with ethidium bromide (1 μg/ml) for 5 min.

Real-time PCR.

Real-time PCR primers were designed using the Beacon Design software (Bio-Rad, Hercules, CA). Cytokine mRNA expression was measured using our previously described SYBR green protocol; data were analyzed using the 2^−ΔΔCT method (1, 45).

Western blots.

CXCL5 expression in intestinal tissue and exfoliated intestinal epithelial cells was measured using our previously described protocol (45).

ELISA.

CXCL5 concentrations were measured in mouse sera, tissue-conditioned media (T-CMs), and epithelial-conditioned media (E-CMs) using a commercially available ELISA kit (R&D; catalog no. DY443). Optical densities and standard concentrations were log-transformed, and a linear equation was obtained (accepted if r² ≥ 0.95). CXCL5 concentrations in test samples were calculated by regression. The linear range of measurement of the assay was 31.2–1,000 pg/ml.

Primary intestinal and bone marrow-derived macrophages.

Murine intestinal macrophages were isolated by our previously described density centrifugation and adherence protocol (25). Briefly, intestinal tissue was washed with Hanks' balanced-salt solution (HBSS) containing 1 mM DTT (Sigma) to remove any mucus. Tissues were next treated with HBSS containing 1 mM EDTA (Sigma) twice for 20 min each at 37°C, washed thrice, and then incubated in HBSS containing 1 mM collagenase type IV (Sigma) for 2 h at 37°C. Isolated cells were suspended in 40% Percoll (Pharmacia Biotech, Baie d'Urfe, QC, Canada), layered on to 75% Percoll, and centrifuged at 2,000 revolution/min for 20 min. Cells recovered from the interphase were selected using CD11b microbeads (Miltenyi Biotec, Cambridge, MA) and then allowed to adhere on polystyrene plates for 1 h. More than 90% adherent cells were confirmed as CD11b⁺ F4/80⁺ CD11c^int macrophages by fluorescence-activated cell sorting (antibodies form BD Pharmingen, San Diego, CA) and immunocytochemistry (antibodies from eBiosciences).

Bone marrow-derived macrophages (BMDMs) were prepared from 10-day-old mice as described previously (58). Bone marrow cells devoid of red blood cells were cultured in BMDM media containing DMEM with 0.5% MEM essential amino acids, 0.5% MEM nonessential amino acids, 1% 1 mM HEPES, 0.1% β-mercaptoethanol, 10% heat-inactivated fetal calf serum, 1% penicillin/streptomycin (all media products from Invitrogen, Carlsbad, CA), and 40 ng/ml monocyte colony-stimulating factor (PeproTech, Rocky Hill, NJ) overnight. Nonadherent cells were collected, and 1 × 10⁷ cells were plated in BMDM media per 150-mm plate. After 5 days of incubation, BMDMs were collected from the plates by scraping. Cells were cultured in granulomonocyte colony stimulating factor-free RPMI complete media for 24 h before chemotaxis assays were performed.

E-CM and T-CM.

We prepared conditioned media from fresh murine intestinal/colonic tissue using our previously reported protocol. T-CMs were prepared by incubating intact intestinal tissue overnight in serum-free DMEM (1 ml/g tissue). To prepare E-CMs, intestinal epithelial cells were isolated by treating the tissue with 0.2 M ethylenediaminetetraacetic acid and 10 mM 2-mercaptoethanol in HBSS for 30 min, washed, and then incubated in serum-free DMEM (1 × 10⁶ cells/ml) for 18 h. Conditioned media were sterile-filtered (0.2-mm syringe filter, Corningware, Corning, NY) and frozen at −80°C.

Macrophage chemotaxis.

Macrophage chemotaxis was measured using our previously described microchemotaxis method (48). Briefly, BMDMs were stained with the fluorescence dye calcein-AM (2 μM; Molecular Probes, Eugene, OR) and suspended in HBSS with 0.1% BSA at a concentration of 1.1 × 10⁶ cells/ml. T-CMs from untreated and TNBS-treated intestines were placed in lower wells of microchemotaxis chambers (ChemoTx System; NeuroProbe, Gaithersburg, MD). In some wells, we added the T-CMs after preincubation for 30 min with excess neutralizing anti-CXCL5 antibody (10 μg/ml; R&D). Additional experiments were performed after pretreatment of BMDMs with anti-CXCR2 antibody or with SB225002 (10 nM; Sigma), a small molecule inhibitor of CXCR2. Recombinant murine CXCL5 (10 and 100 pM) standards were included for positive control. A polycarbonate filter membrane (5 μm pores; Neuro Probe, Gaithersburg, MD) was placed on the microchemotaxis chamber, and then 80,000 macrophages (72 μl, ≥98% viability, suspended in HBSS plus 0.1% BSA) were placed in the upper wells marked on the filter. Control wells contained 0–80,000 calcein-stained cells per well to generate a standard curve. The microchemotaxis chamber was incubated for 60 min (37°C, 5% CO₂, humidified air). After incubation, nonmigrated cells on the top of the filter were removed with a smooth-edged wiper, and the plate was centrifuged at 200 g for 5 min to collect migrated cells attached to the underside of the filter. The fluorescence in each well was determined at 485/530 nm, and the number of cells that migrated through the filter was computed by comparing the fluorescence signal in test wells with that of the standard curve generated from known numbers of fluorescence-labeled cells.

Statistical methods.

Parametric and nonparametric tests were applied using the Sigma Stat 3.1.1 software (Systat, Point Richmond, CA). The number of samples and statistical analyses are indicated in each figure legend. Each sample was tested in duplicate. A P value of <0.05 was considered significant.

RESULTS

Necrotizing enterocolitis is characterized by a macrophage-rich inflammatory infiltrate.

To characterize the leukocyte infiltrate in NEC, we first examined hematoxylin and eosin-stained sections from non-NEC premature intestinal tissue (n = 8; postmenstrual age 27.2 ± 3 wk; representative section shown in Fig. 1, A and B) and compared those with tissue specimens of advanced NEC (n = 15; postmenstrual age 29.1 ± 2.6 wk; representative section shown in Fig. 1, C and D). NEC was associated with a prominent inflammatory infiltrate, comprised mainly of HAM56⁺ macrophages (12.8 ± 1.1 cells/HPF in control tissue vs. 128.6 ± 9.4 cells/HPF in NEC; P < 0.001). We also detected a modest increase in the number of polymorphonuclear leukocytes (PMNs) in NEC (7.7 ± 1.7 cells/HPF in control tissue vs. 37.9 ± 5.8 cells/HPF in NEC; P < 0.001). Interestingly, there was no difference in the number of lymphocytes in NEC and control specimens.

Fig. 1.

Necrotizing enterocolitis (NEC) is characterized by a macrophage-rich inflammatory infiltrate. A: normal premature intestine (jejunum) from a 2-wk-old neonate born at 28-wk gestation showing normal cellularity and crypt-villus histoarchitecture (hematoxylin & eosin, magnification ×100); inset: high-magnification (×1,000) photomicrograph shows the normal absence of inflammatory cells in the lamina propria. B: fluorescence photomicrograph of a serial section shows HAM56⁺ macrophages (red) in the normal premature intestine. Nuclear staining was obtained with 4′,6-diamidino-2-phenylindole (DAPI) (blue); inset: high-magnification photomicrograph highlights the cytoplasmic HAM56 staining in macrophages. C: NEC in a 5-wk-old neonate born at 24-wk gestation, showing epithelial necrosis and a prominent inflammatory infiltrate; inset: many infiltrating cells showed a macrophage-like appearance (large, rounded cells with eccentrically placed vesicular nuclei; black arrows), although a few polymorphonuclear leukocytes (PMNs) were also noted (white arrow). D: fluorescence photomicrograph of a serial section highlights the predominance of HAM56⁺ macrophages (red) in NEC; inset: high-magnification photomicrograph shows HAM56-stained macrophages with better resolution. E: bar diagrams (means ± SE) show the number of HAM56⁺ macrophages, PMNs, and lymphocytes per high-power field (HPF) in premature intestine and in NEC. N = 15 cases of NEC and 8 controls. Groups were compared by Mann-Whitney U-test; ***P < 0.001; N.S. indicates that differences were not significant.

TNBS administration by gavage and enema in 10-day-old pups and adult mice induces inflammatory mucosal injury in both small intestine and colon.

To investigate the hypothesis that macrophage infiltration in NEC is a generic pathoanatomic feature of gut mucosal injury in the developing intestine, we administered TNBS by gavage and enema in 10-day-old mouse pups and adult mice and observed these animals for up to 48 h. TNBS caused inflammatory mucosal injury in both small and large intestine in pups (Fig. 2A) and adult mice (Fig. 2B). Pups sustained more severe injury in both the small intestine (median 3, range 0–4 in pups vs. 2, 0–3 in adult mice; P = 0.003) as well as colon (median 2, range 0–4 in pups vs. 1, 0–3 in adult mice; P = 0.003). Consistent with previous reports, the same dose of TNBS produced only minimal mucosal damage in germ-free pups, indicating that gut microbial flora was essential for the development of TNBS-induced enterocolitis (Fig. 2, C and D). To ensure that the inflammatory changes we observed in 10-day-old pups were not an artifact of this specific postnatal time point, we also performed the experiment in a small number of 5-day-old mouse pups (n = 6). These animals also developed gut mucosal inflammation that was similar to the 10-day-old pups with prominent, macrophage-rich leukocyte infiltrates in the lamina propria (data not depicted).

Fig. 2.

Trinitrobenzene sulfonic acid (TNBS) administration by gavage and enema in 10-day-old pups and adult mice induces inflammatory mucosal injury in both small intestine and colon. A: photomicrographs (hematoxylin & eosin, ×100) of distal jejunum (left) and colon (right) from 10-day-old pups. Tissue sections from sham-treated pups with no injury (top), TNBS-induced mild injury (middle), and TNBS-induced severe injury (bottom). B: photomicrographs of distal jejunum (left) and colon (right) from adult mice after sham (top) and TNBS treatment (bottom). Column scatter plots on the right show the frequency distribution of the injury grade in pups (top) and adult mice (bottom). N = 25 animals per group. Groups were compared by Mann-Whitney U-test, ***P < 0.001. C: representative photomicrographs of distal jejunum (left) and colon (right) from germ-free pups treated with TNBS treatment show minimal injury to intestinal villi and intact colonic mucosa. D: 16S rRNA PCR-DGGE gels run on colonic tissue from 10-day-old pups (1), adults (2), and germ-free pups (3). Gels from 10-day-old pups show numerous bands in 10-day-old pups, confirming the presence of bacterial flora. Adult mice showed greater diversity of bacterial flora than pups. No bands were detected in germ-free animals. Data represent 3 separate experiments.

TNBS-induced mucosal injury was characterized by a macrophage-rich infiltrate in pups but a pleomorphic inflammatory response in adult mice.

We next used quantitative immunofluorescence to enumerate the number of macrophages, polymorphonuclear leukocytes, and T-lymphocytes in TNBS-induced enterocolitis in pups vs. adult mice. As shown in representative photomicrographs and summarized in the bar diagrams in Fig. 3A, TNBS-induced mucosal injury was associated with increased number of macrophages in small intestine (16.3 ± 2.3 cells/HPF in control vs. 124 ± 11.1 cells/HPF in TNBS-induced injury; P < 0.001) and large intestine (26.8 ± 3.6 cells/HPF in control tissue vs. 272.6 ± 30.7 cells/HPF in TNBS-induced injury; P < 0.001). Similar to human NEC, there was a moderate increase in PMNs. There was no change in the number of lymphocytes. A similar predominance of macrophages was observed at 12 and 24 h after TNBS administration (not depicted). In contrast, TNBS-induced mucosal injury in adult mice was associated with a pleomorphic inflammatory infiltrate as shown in Fig. 3B.

Fig. 3.

TNBS-induced mucosal injury was characterized by a macrophage-rich infiltrate in pups but a pleomorphic inflammatory response in adult mice. A: TNBS-induced mucosal inflammatory response in 10-day-old mouse pups. Fluorescence photomicrographs (magnification ×250) of distal jejunum and colon from control and TNBS-treated pups show immunostaining for F4/80, myeloperoxidase (MPO), and CD3 (all green), which was used to enumerate macrophages, PMNs, and lymphocytes, respectively; nuclear staining was obtained with DAPI (blue). TNBS-induced mucosal injury was associated with an increase in macrophages in both small and large intestine. There was a small increase in PMNs, but the number of lymphocytes did not change significantly. Asterisk on the immunofluorescence panel showing myeloperoxidase staining indicates nonspecific staining on the edge of the tissue. Bar diagrams (means ± SE) show the number of F4/80, myeloperoxidase, and CD3-positive cells per HPF. Cells were counted in fields with at least 10 villi or 10 colonic crypts. N = 5 animals per group. Groups compared by Mann-Whitney U-test, *P < 0.05; **P < 0.01, and ***P < 0.001. B: TNBS-induced mucosal inflammatory response in adult mice. Fluorescence photomicrographs of distal jejunum and colon from control and TNBS-treated adult mice show immunostaining for F4/80, myeloperoxidase, and CD3. TNBS-induced mucosal injury in mature animals was associated with a smaller increase in macrophages in small and large intestine than in pups, whereas the increase in PMNs and lymphocytes was more robust. Bar diagrams (means ± SE) show the number of F4/80, myeloperoxidase, and CD3-positive cells per HPF; n = 5 per group; data were analyzed as described in A.

TNBS-induced gut mucosal injury in pups is associated with increased mRNA expression of the chemokines CXCL5 and CXCL2.

To determine the mechanism(s) by which macrophage precursors are recruited to areas of gut mucosal injury in mouse pups, we used a PCR-based array to compare tissue expression of key chemokines/cytokines in the intestine (not depicted) and colon (Fig. 4) from pups and adult animals. We detected increased interleukin (IL)-12 and interferon-γ expression in adult mice with TNBS-enterocolitis, which was consistent with previous reports (44), but not in pups. TNBS enterocolitis increased the expression of transforming growth factor (TGF)-β₂ in adult mice but inhibited its expression in pups; this dichotomous change was consistent with our previously reported findings in NEC (25). TNBS enterocolitis increased IL-1α exclusively in pups, whereas IL-1β was upregulated in both pups and adult mice. Compared with adult mice, TNBS-enterocolitis in pups was associated with increased expression of CXCL5 (84.1 ± 8.5-fold vs. 2.3 ± 2.2-fold; P < 0.05) and CXCL2 (30.6 ± 6.5-fold vs. 2.3 ± 1.3-fold; P < 0.05). There was a trend toward increased CCL4 expression that did not reach statistical significance.

Fig. 4.

TNBS-induced gut mucosal injury in pups is associated with increased mRNA expression of the chemokines CXCL5 and CXCL2. Bar diagrams (means ± SE) show fold change in mRNA expression of major cytokines/chemokines in TNBS-induced colonic injury above control in 10-day-old pups and adult mice. N = 5 animals per group. Crossing-threshold (ΔΔCT) values for genes with a ≥2-fold increase were compared by Mann-Whitney U-test; *P < 0.05. Differences that did not reach statistical significance were left unmarked.

TNBS-induced gut mucosal injury in pups is associated with increased CXCL5 expression in affected tissues and in sera.

We first used immunohistochemistry to localize CXCL5 in intestinal tissue from control and TNBS-treated mice. CXCL5 was detected in the epithelial and muscularis layers in TNBS-treated but not in control animals (Fig. 5A). In contrast, CXCL2 was detected exclusively in the infiltrating F4/80⁺ macrophages (Fig. 5A, inset). We interpreted these differences in the cellular origin of the two chemokines to identify CXCL5 to be a primary initiator of macrophage recruitment and CXCL2 as a secondary “amplifier” of this process. Therefore, we focused on CXCL5 in subsequent experiments. Using Western blots, we confirmed increased expression of CXCL5 protein in TNBS enterocolitis (Fig. 5B). We also detected increased expression of CXCL5 in T-CMs, E-CMs, and sera from mouse pups with TNBS enterocolitis (Fig. 5C).

Fig. 5.

TNBS-induced gut mucosal injury in pups is associated with increased CXCL5 expression in affected tissues and in sera. A: fluorescence photomicrographs (magnification ×1,000) of colonic tissue from control and TNBS-treated mouse pups. CXCL5 (green) was immunolocalized to the epithelium and muscularis layers. Nuclear staining (blue) was obtained with DAPI. Inset: photomicrographs (×400) show CXCL2 expression exclusively in F4/80⁺ macrophages and not in primary intestinal cells. B: western immunoblots show increased CXCL5 expression in jejunoileal tissue and colon from TNBS-treated pups. Bar-diagrams (means ± SE) show densitometric units for CXCL5 bands normalized against β-actin in each sample. N = 5 per group. Groups compared by Mann-Whitney U-test; **P < 0.01, and ***P < 0.001. C: bar diagrams (means ± SE) show that tissue-conditioned media (T-CMs) and epithelial-conditioned media (E-CMs) prepared from TNBS-treated pups and serum samples contained higher concentrations of CXCL5 than from controls. N = 5 per group; statistical analysis as in Fig. 4A.

CXCL5 recruits macrophages to the gastrointestinal tract during inflammatory mucosal injury in mouse pups.

To determine whether CXCL5 recruits macrophage precursors during TNBS enterocolitis, we first confirmed the expression of CXCR2 (cognate receptor for CXCL5) on murine intestinal macrophages using immunohistochemistry and flow cytometry (detected on 97.1 ± 2.2% cells; Fig. 6A). Because differentiated intestinal macrophages do not migrate consistently along chemokine gradients (48), we hypothesized that macrophage infiltration at sites of mucosal inflammation resulted from the recruitment of macrophage precursors and used BMDMs as a model of gut macrophage precursors. CXCR2 expression on BMDMs was confirmed by flow cytometry (48.3 ± 8.1% cells) and immunocytochemistry (variable but uniformly positive staining on all cells; not depicted).

Fig. 6.

CXCL5 recruits macrophages to the gastrointestinal tract during inflammatory mucosal injury in mouse pups. A: intestinal macrophages express CXCR2, the cognate receptor for CXCL5. Fluorescence photomicrographs (×1,000) show CXCR2 staining on intestinal macrophages in the lamina propria. Photomicrographs on the right show a cross-sectional view through a villus, where CXCR2 colocalizes with the macrophage marker F4/80. Representative histograms on the extreme right show CXCR2 expression on CD11b⁺ CD11c^int intestinal macrophages (top), and on bone marrow-derived macrophages (BMDMs) (bottom). B: T-CMs prepared from TNBS-treated colons showed greater macrophage chemoattractant activity than controls. As depicted in the schematic, macrophage chemotactic activity of T-CMs was measured using BMDMs in a fluorescence-based assay. Bar diagram (means ± SE) summarizes the number of cells migrating through a polycarbonate filter toward standards or test samples. Standards (shaded bars) included a range of concentrations of recombinant murine CXCL5 (200–2,000 pg/ml). Macrophage chemotactic activity of TNBS-T-CMs was blocked by neutralizing anti-CXCL5 antibody or when BMDMs were pretreated with anti-CXCR2 antibody or with SB225002, a small molecule inhibitor of CXCR2. Data represent 3 separate experiments. Groups were compared by Kruskal-Wallis H-test/Dunn's multiple-comparison post test; **P < 0.01, ***P < 0.001.

We next used BMDMs in a microchemotaxis assay to measure macrophage chemotactic activity of TNBS T-CMs. As shown in Fig. 6B, T-CMs prepared from TNBS mice had greater macrophage chemotactic activity than T-CMs from control mice. Macrophage chemotactic activity of TNBS T-CMs was blocked by the addition of neutralizing anti-CXCL5 antibody to the T-CMs, indicating its important role in the recruitment of macrophage precursors to the TNBS intestine. In support of these data, macrophage chemotaxis toward TNBS T-CMs was also blocked by prior treatment of macrophages with either anti-CXCR2 antibody or with SB225005, a small molecule inhibitor of CXCR2.

Human NEC is associated with increased tissue expression of ENA-78/CXCL5.

Murine CXCL5 is similar to its human homologue ENA-78/CXCL5 [identities = 60/117 (51%), positives = 76/117 (65%), gaps = 4/117 (3%); e-value = 4e-29] (11). To validate our findings in NEC, we measured CXCL5 expression in a small set of tissue samples of NEC. CXCL5 mRNA expression in NEC was increased significantly over control tissues (Fig. 7). Consistent with these data, we detected CXCL5 protein in the epithelium in tissue samples of NEC but not in normal preterm intestinal tissue (Fig. 7, inset).

Fig. 7.

Human NEC is associated with increased tissue expression of epithelial-derived neutrophil chemoattractant-78 (ENA-78)/CXCL5. Scatter bar diagram shows fold change (above control) in ENA-78/CXCL5 mRNA expression in tissue samples of human NEC vs. controls, measured by real-time PCR. Data represent 8 controls and 11 samples of NEC. Crossing-threshold (ΔΔCT) values were compared by Mann-Whitney U-test. Representative fluorescence photomicrographs (×250) in inset show immunofluorescence staining for CXCL5 in normal premature intestine (jejunum; 3-wk-old neonate born at 27-wk gestation) and in NEC (jejunum; 4-wk-old neonate born at 26-wk gestation). CXCL5 staining was detected in epithelium in NEC but not in control tissue.

DISCUSSION

We present a detailed investigation into the mechanism(s) by which macrophage precursors are recruited to areas of tissue injury during NEC and identify CXCL5 and its cognate receptor, CXCR2, as potential therapeutic targets in NEC. We also propose a novel pathophysiological model of NEC, where its pathoanatomic characteristics represent an integrated tissue injury response related to a specific stage of gut development rather than to specific etiological factor(s). Such a model would adequately explain the occurrence of NEC almost exclusively in preterm infants, even though gut barrier dysfunction and bacterial translocation are frequently encountered in critically ill patients of all ages (33). In this schema, NEC, which is a clinical and histopathological diagnosis, could be conceptualized not as a single nosological entity but as a group of conditions with shared clinico-pathological manifestations; an appropriate simile could be drawn from our current understanding of hepatic cirrhosis, where diverse etiological factors such as genetic mutations, viruses, and environmental toxins can produce similar clinical features and histoarchitectural changes in the liver.

We detected a marked increase in the number of macrophages in our tissue samples of NEC. Pender et al. (38) have previously described the characteristics of the inflammatory response in a small cohort (n = 9) of patients with NEC. Although these investigators did not investigate the mechanism of leukocyte infiltration in their study, they detected a similar increase in the number of CD68⁺ macrophages in affected tissue. They also detected increased expression of tumor necrosis factor in NEC, thereby providing indirect evidence for the activated, inflammatory nature of these macrophages. Macrophages in the adult intestine display a profound inflammatory “anergy” to bacterial products, a unique adaptation to maintain the absence of inflammation in the normal gut mucosa despite close physical proximity to luminal bacteria (49). In contrast, macrophages in the preterm intestine are yet to undergo this inflammatory downregulation and respond to bacterial products with exaggerated cytokine/chemokine responses (25). Understanding the mechanism(s) by which macrophages accumulate at sites of tissue injury during NEC is an important step in the development of novel anti-inflammatory therapies that can prevent/ameliorate tissue damage in NEC.

We used TNBS-induced enterocolitis in murine pups and adult mice as a nonspecific mucosal insult. The major advantage of using a chemical agent to induce mucosal inflammation was that the dose of the inciting agent (TNBS) could be normalized against the body weight of the animal, which allowed comparison of mucosal injury across age groups. An added theoretical advantage was that TNBS-induced enterocolitis requires the presence of luminal bacteria (44), which is also the case in NEC (24, 29). We chose TNBS enterocolitis after careful evaluation of existing animal models of neonatal gut mucosal injury, which were not suitable for age-based comparisons. In the rodent models of NEC, newborn rats (25) or 10-day-old mouse pups (25, 50) are provided formula feedings and exposed to hypoxia and hypothermia twice a day for up to 4 days, which induces intestinal injury in up to 70–80% rats or 30–40% mouse pups. This model was not suitable for the present study because it does not induce injury in adult animals. In other studies, parenteral (intra-arterial/intraperitoneal) administration of platelet-activating factor and Escherichia coli LPS was used to induce gut mucosal injury. In this model, the animals are killed after 2–3 h of treatment (16). Although this model is a useful screening tool for the investigation of inflammatory signaling, its short duration limits its value in the study of leukocyte trafficking (20). Intestinal ischemia-reperfusion injury, induced by superior mesenteric artery occlusion, is another short-duration model (2–3 h) with similar limitations (23). We opted against the use of live infectious agents such as Enterobacter sakazakii (17) because the dose of the inciting agent (density of Enterobacter population in the gut lumen) is difficult to control and normalize by body weight/size of the animal. We also evaluated models of NEC-like injury in experimental animals such as rabbits (37), quails (5), and piglets (42) but did not find a suitable way to induce comparable mucosal injury in adult animals. TNBS-mediated mucosal injury was particularly suitable for age-based comparisons because TNBS colitis in adult mice is associated with a striking increase in the number of T-lymphocytes with relatively few macrophages (7, 15, 35, 36, 44), which contrasted with the leukocyte infiltrates we observed in NEC. Therefore, if the null hypothesis was to be disproven and we were to detect abundant macrophages but few lymphocytes in neonatal TNBS enterocolitis, we anticipated these differences to be significant. We chose the postnatal age of 10 days for the pups to ensure postnatal bacterial colonization and for ease of comparison with previous studies that have used animals at this specific postnatal age (25, 50). It was convenient that the rodent neonatal intestine, which has been compared with the preterm human intestine, takes nearly 3 wk to undergo structural and functional maturation to levels seen in the full-term human neonate (6, 13, 30, 34, 55).

We detected important differences in the inflammatory response in TNBS-enterocolitis in pups vs. adult mice. In adult mice, rectal instillation of TNBS causes colitis that resembles Crohn's disease with its T-helper 1 (Th1) lymphocyte response and induction of proinflammatory cytokines such as IFN-γ, TNF-α, and IL-12 (44). To induce inflammatory changes in both the small and large intestine, we modified the colitis protocol (44) and administered a weight-normalized dose of TNBS in 10-day-old pups by both gavage and enema. We (25, 27), as others (31), have previously documented the “proinflammatory bias” in the developing intestine and had therefore anticipated the pups to sustain more severe mucosal injury than adults when challenged with a similar inflammatory stimulus. However, the mechanisms underlying the observed differences in the leukocyte infiltrates were unclear. Similar to our observations in NEC, pups with TNBS-enterocolitis showed leukocyte infiltrates comprised predominantly of macrophages and a few neutrophils. These findings in pups with TNBS enterocolitis contrasted with TNBS-treated in adult animals, which developed pleomorphic leukocyte infiltrates. Although age-related differences in cytokine/chemokine expression can account for these findings, maturational differences in the development of gut leukocyte populations are also likely to play an important role in the relative composition of leukocyte infiltrates during inflammatory injury. Macrophages are the first leukocytes to appear in the developing intestine, and the intestinal macrophage pool approaches mature levels in both its size and function by midgestation (2, 22, 26). In contrast, gut lymphocyte populations appear during fetal development a few weeks after macrophages but remain limited in size and repertoire until postnatal bacterial colonization and weaning (19, 22, 40). There were significant regional differences in the cellular inflammatory response between the small and large intestine, emphasizing the need to compare similar regions of the gastrointestinal tract when evaluating clinical specimens of NEC.

We detected important qualitative differences in the expression of chemokines and inflammatory cytokines during gut mucosal inflammation in pups and adult mice, even though the inciting agent was administered in amounts that were normalized against total body weight. Unlike in adult mice, which upregulated a wide repertoire of cytokines/chemokines, TNBS enterocolitis in the pup induced a very limited set of cytokines/chemokines. We detected increased expression of CXCL5, CXCL2, and IL-1α in pups to levels several fold higher than in adult animals. TNBS injury in pups did not induce cytokines associated with TNBS colitis in adult animals such as IFN-γ and IL-12 (44), which is consistent with existing data from human studies showing that neonates, particularly those born at preterm gestation, are relatively deficient in these two cytokines (39). The increase in IL-1α expression we detected in pups is likely to have originated in the large number of macrophages recruited to the inflamed tissues (41). Although increased IL-1α expression has not been reported previously in NEC, circulating IL-1α levels are increased in neonatal sepsis (9). These findings are of interest in view of recent data indicating that the gut microenvironment in the neonate may influence local IL-1α expression (47). Finally, pups upregulated IL-1β to levels comparable to those in adults, which is consistent with information that the inflammasome pathway is largely intact in the neonate (4, 43).

We show increased expression of CXCL5 in pups with TNBS-enterocolitis than in adult mice and also in tissue samples of human NEC. These findings are consistent with existing information that human neonates have higher circulating CXCL5 concentrations than adults (52) and detection of elevated plasma CXCL5 concentrations in patients with NEC (12). The mechanism of these developmental differences remains unclear although promoter methylation and micro-RNA-mediated mRNA instability are plausible possibilities (53). Besides functioning as a chemokine for CXCR2⁺ cells such as neutrophils and monocytes/macrophages (14), CXCL5 can also increase the bioavailability of other CXC chemokines (such as CXCL2) by binding nonsignaling chemokine sinks such as the Duffy antigen receptor for chemokines and heparan sulfate proteoglycans (28). Therefore, increased CXCL5 expression during neonatal enterocolitis provides a unifying mechanism for the recruitment of both macrophages and neutrophils. The relative predominance of macrophages in these lesions (compared to neutrophils) is consistent with differences in the functional maturation of neonatal macrophages vs. neutrophils; whereas macrophages attain functional maturity by midgestation (51, 57), neonatal neutrophils remain limited in their chemotactic capacity until up to 42 wk postmenstrual age (8, 10).

In conclusion, we show that the cellular inflammatory response during mucosal injury is affected by the stage of gastrointestinal development. We also show that macrophages were recruited to the inflamed gut mucosa in murine neonatal enterocolitis by the chemokine CXCL5, indicating that CXCL5 and its cognate receptor CXCR2 merit further investigation as potential therapeutic targets in NEC. The present study has important limitations, in its use of an animal model, the small sample size, and the lack of information on the microbiota. We acknowledge that inflammatory changes induced by chemical agents such as TNBS, which although acceptable in proof-of-concept studies, may not accurately simulate pathological changes in a natural disease process and require further corroboration in other models. Finally, unlike large animals, limited availability of clinical and demographic data in rodent pups precludes the use of robust statistical tools such as multivariate regression analysis to dissect the relative contribution of the clinical course of the animals, their feeding experience, and the microbial flora.

GRANTS

This work was supported by the National Institutes of Health awards HD059142 (A. Maheshwari), HD061607 (J.-H. Weitkamp), and a research grant from the CACA Jones Family Foundation (A. Maheshwari).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: K.M., N.K., A.R.K., and A.M. conception and design of research; K.M., N.K., R.J., S.A.G., A.B., A.R.K., K.N., J.I.R., R.B., X.F., J.-H.W., and A.M. performed experiments; K.M., S.A.G., K.N., R.B., J.-H.W., and A.M. analyzed data; K.M., K.N., and A.M. interpreted results of experiments; K.M., R.J., and A.M. prepared figures; K.M. and A.M. drafted manuscript; K.M., X.F., J.-H.W., and A.M. edited and revised manuscript; K.M., N.K., R.J., S.A.G., A.B., A.R.K., K.N., J.I.R., R.B., X.F., J.-H.W., and A.M. approved final version of manuscript.