Gram negative bacteria are associated with the early stages of necrotizing enterocolitis

Abstract

Introduction: Necrotizing enterocolitis (NEC) affects 5-10% of infants born weighing less than 1500 g. Most models of NEC recapitulate late-stage disease with gut necrosis and elevated inflammatory mediators. Evaluation of NEC at earlier, less lethal stages of disease will allow investigation of initial disease triggers and may advance our understanding of temporal relationships between factors implicated in NEC pathogenesis. In this manuscript, we describe our investigation of early NEC and test the hypothesis that bacteria and inflammatory mediators differ between animals with early NEC and disease free animals.

Methods: On DOL7 C3HeB/FeJ pups were fed liquid formula with 1×10(4) Streptococcus thoraltensis, Serratia marcescens, and Pseudomonas aeruginosa every 3 h. To initiate NEC, pups underwent asphyxia (100% N(2) for 90 s) and hypothermia (4°C for 10 min) after feeding. Pups were euthanized at 72 h. Intestines were collected for histologic NEC scoring and DNA/RNA extraction. Bacterial populations were identified by 16S rRNA pyrosequencing and principal component analysis (PCA). RNA isolates underwent QRT-PCR for Toll-like Receptor 4 (TLR4) and inducible nitric oxide synthase (iNOS).

Results: Despite histologic, intestinal damage in mice with NEC, no gross necrosis was observed suggesting early disease. QRT-PCR yielded no difference between groups in TLR4 or iNOS mRNA levels. PCA demonstrated relative clustering of microbial communities based on presence or absence of NEC. 16S pyrosequencing demonstrated similar phyla between groups (Firmicutes and Proteobacteria predominated in all animals). However, the colonic microbiota of animals with NEC had more Citrobacter (p<0.01), Klebsiella (p<0.05), and Tatumella (p<0.05), while that of animals without NEC had more Streptococcus (p<0.01) and Enterococcus (p<0.01).

Conclusion: Citrobacter, Klebsiella, and Tatumella are associated with NEC. Differential colonic bacteria were identified despite the lack of inflammatory mediator elevation traditionally associated with NEC. This suggests a temporal relationship between bacteria and inflammatory mediators such that alterations in gut microbiota are associated with early NEC, while inflammatory mediator elevation is associated with advanced NEC.

Figure 1. Histology.

Representative histology from mice with (c,d) and without (a,b) NEC. After 72 h of exposure to the experimental feeding protocol, intestines were collected and embedded in OCT as described in Materials and Methods . Frozen sections were prepared and processed for H&E staining. Sections were scored by a blinded observer based on a well established NEC scoring scale .

Figure 2. TLR4 and iNOS QRT-PCR Analysis.

Colonic TLR4 and iNOS mRNA levels in mice with and without NEC. Following total colonic RNA isolation, cDNA was synthesized, and copy numbers of TLR4 and iNOS were determined using QRT-PCR. Data shown are mean ± SEM of TLR4 and iNOS copy numbers relative to GAPDH. Each column represents n = 4 animals with duplicate measurements.

Figure 3. Principal Coordinate Analysis.

Results of weighted principal coordinate analysis of the colonic microbiota of mice with NEC (▪N) as compared to mice without NEC (○NN). PC1 = 28.9% and PC2 = 25.1%. This analysis was performed via the mothur analytical pipeline as described for phylotype-based analysis .

Figure 4. Temporal Temperature Gradient Gel Electrophoresis (TTGE) Analysis.

Following total colonic DNA extraction and PCR amplification, PCR products were separated via TTGE to identify bacterial community structures in mice with NEC (e–h) as compared to mice without NEC (a–d).

Figure 5. Analysis of Colonic Microbiota.

Following total colonic DNA extraction, PCR was performed with 8 bar coded Bact27 and UNIV788 primers to amplify the bacterial 16S rRNA gene. PCR products were purified and underwent subsequent sequencing on the Roche GS-FLX pyrosequencing platform. Taxonomy was assigned to sequences using the Ribosomal Database Project classifier tool. Mice with NEC (a) are compared to mice without NEC (b). Statistical tests for differentially abundant 16S-based taxonomic annotations between populations were performed using Metastats methodology to compute nonparametric p-values. Differences were considered significant at p<0.05. *indicates statistically significantly more in mice with NEC. **indicates statistically significantly more in mice without NEC.