Mucosal metabolites fuel the growth and virulence of E. coli linked to Crohn’s disease

Elucidating how resident enteric bacteria interact with their hosts to promote health or inflammation is of central importance to diarrheal and inflammatory bowel diseases across species. Here, we integrated the microbial and chemical microenvironment of a patient’s ileal mucosa with their clinical phenotype and genotype to identify factors favoring the growth and virulence of adherent and invasive E. coli (AIEC) linked to Crohn’s disease. We determined that the ileal niche of AIEC was characterized by inflammation, dysbiosis, coculture of Enterococcus, and oxidative stress. We discovered that mucosal metabolites supported general growth of ileal E. coli, with a selective effect of ethanolamine on AIEC that was augmented by cometabolism of ileitis-associated amino acids and glutathione and by symbiosis-associated fucose. This metabolic plasticity was facilitated by the eut and pdu microcompartments, amino acid metabolism, γ-glutamyl-cycle, and pleiotropic stress responses. We linked metabolism to virulence and found that ethanolamine and glutamine enhanced AIEC motility, infectivity, and proinflammatory responses in vitro. We connected use of ethanolamine to intestinal inflammation and L-fuculose phosphate aldolase (fucA) to symbiosis in AIEC monoassociated IL10–/– mice. Collectively, we established that AIEC were pathoadapted to utilize mucosal metabolites associated with health and inflammation for growth and virulence, enabling the transition from symbiont to pathogen in a susceptible host.

relative abundance and glommed to genus level prior to Kruskal-Wallis testing and beta diversity analysis.
The threshold for false discovery was q < 0.25 for all results. Kruskal-Wallis testing was followed by Dunnet's posthoc testing when metadata included multiple categories. The Bray-Curtis distance was used for all beta diversity analysis. The complete source code for this analysis and raw sequencing data is available at https://gitlab.com/morganlab/Simpson2021. The raw sequencing data can also be found at http://www.ncbi.nlm.nih.gov/bioproject/781964.

H NMR spectroscopic analysis intact ileal biopsies.
Intact snap-frozen ileal biopsies were bathed in ice cold saline D2O solution. A portion of the tissue (~15 mg) was inserted into a zirconium oxide (ZrO2) 4 mm outer diameter rotor, using an insert to make a spherical sample volume of 25 µl. In order to reduce NMR spectral peak broadening caused by any residual dipolar couplings, chemical shift anisotropy and microscopic inhomogeneities, 600 MHz 1 H NMR spectra of intact tissues were acquired with a high resolution magic-angle-spinning probe at a spin rate of 5000 Hz. 1 H NMR spectra were acquired for each sample using a Bruker Avance II 600 NMR spectrometer (Bruker Biospin, Rheinstetten). Tissue samples were regulated at 283K using cold N2 gas during the acquisition of the spectra to minimize any timedependent biochemical degradation, as previously described (6). Standard 1 H NMR one-dimensional pulse sequence with water suppression, Carr-Purcell-Meiboom-Gill (CPMG) spin-echo sequence with water suppression, and diffusion-edited sequence were acquired for sample, as described previously (6). For each sample, 256 and 16 scans were collected into 98'000 data points using a spectral width of 18028 Hz and 18315 Hz, respectively. The assignment of the 1 H-NMR resonances to specific metabolites was achieved by matching our in-house NMR database of pure compounds and using literature data.
The tissue NMR spectra were converted into 22K data points over the range of δ 0.0-10.0 using an in-house MATLAB routine excluding the water residue signal. Chemical shift intensities were normalized to the sum of all intensities within the specified range prior to chemometric analysis for tissue samples.
One glucose peak was omitted from downstream analysis due to suspected lipid contamination. Samples with negative integral values for a metabolite were omitted from analysis of that metabolite. Kruskal-Wallis tests with Benjamini-Hochberg false discovery correction and permutation-based ANOVA were used to test for associations between metabolites and metadata of interest. Each metabolite -metadata association with uncorrected p < 0.05 was included in visualization, and its effect size (η 2 / eta squared) was calculated.
The packages FactoMineR and Factoextra were used for principal components analysis to visualize relationships between metabolites and phenotypes. The complete source code for this analysis and raw data is available at https://gitlab.com/morganlab/Simpson2021. The libraries were sequenced using Illumina Hiseq 2000 platform at Cornell Biotech Genome facility with paired-end reads (50 bp) Reads cleaned and trimmed based on a quality score of 25 and minimum length of 25 bp with fastq-mcf (https://code.google.com/p/ea-utils/wiki/FastqMcf). Paired-end reads from fragments shorter than twice the read length were merged with Flash (7). RNAmmer (8) and tRNAscan (9) were used to find ribosomal RNAs and tRNAs, respectively. Predicted proteins were queried against Swiss-Prot database (10) with BLASTx to determine putative functions. Interproscan (11) was used to assign GO terms and domains. COGs were assigned by BLAST matches to the cog database at ftp://ftp.ncbi.nih.gov/pub/mmdb/cdd/big_endian/. Cleaned reads were mapped to the new genome assemblies with Tophat2 (12). Read counts to gene models were calculated using the htseq package (13) and FPKM was calculated using cuffdiff (14). Differential expression was calculated using edgeR and cuffdiff. Heatmaps were generated using the R heatmap.2 package and custom scripts. The raw sequencing data is available at https://submit.ncbi.nlm.nih.gov/subs/bioproject/SUB10604108/overview.

Construction of CUMT8 derivative strains.
An isogenic mutant of CUMT8 lacking the complete eutH gene was constructed using the λ red recombinase system using deletion primers as described previously (15). Primers contain sequences homologous to the 5′ and 3′ ends of eutH: eutH-delF (5′-

CGATATCGATACCGACGCTCAATAGCTGGCGAGTGTTCACCATATGAATATCCTCCTTAG -3′)
were used to amplify the resistance cassette from the template plasmid pKD4 encoding kanamycin resistance. PCR products were electroporated into CUMT8 containing λ red recombinase plasmid pKD46 after which kanamycin-resistant colonies were selected. The deletion of eutH was confirmed by PCR.
fucA and pduC mutants were created similarly as described previously (15). To complement the eutH mutation, the complete eutH coding region was amplified by PCR from CUMT8, restriction digested and cloned into the kpnI and hindIII sites of plasmid pBBR1MCS. This plasmid construct was electroporated into CUMT8:ΔeutH and selected for resistance to chloramphenicol. Constructs were confirmed by PCR and sequencing using primers flanking the insertion sites.

TTGCCTTCCGGCGTAACCCATCTGT-3′)
Twelve weeks after colonization, colon tissues were harvested and analyzed by histological score, cytokine qPCR, or tissue fragment culture following previously published protocols (16)(17)(18). All procedures were performed at UNC and approved by the UNC Institutional Animal Care and Use Committee. Each cDNA sample was analyzed in duplicate for quantitative assessment of RNA amplification. Melting curve analysis confirmed the presence of single products with expected melting temperatures. Expression of housekeeping gene Actb was more consistent than that of GAPDH and 18S between tissues in the model used in this study.
Targeted transcriptional analysis of AIEC. RT-PCR primers for virulence, metabolic and stress genes are listed in Table S1. E. coli mdh was used as the reference gene. Each RT-PCR reaction contained