New Research In
Physical Sciences
Social Sciences
Featured Portals
Articles by Topic
Biological Sciences
Featured Portals
Articles by Topic
- Agricultural Sciences
- Anthropology
- Applied Biological Sciences
- Biochemistry
- Biophysics and Computational Biology
- Cell Biology
- Developmental Biology
- Ecology
- Environmental Sciences
- Evolution
- Genetics
- Immunology and Inflammation
- Medical Sciences
- Microbiology
- Neuroscience
- Pharmacology
- Physiology
- Plant Biology
- Population Biology
- Psychological and Cognitive Sciences
- Sustainability Science
- Systems Biology
QseC inhibition as an antivirulence approach for colitis-associated bacteria
Edited by Lora V. Hooper, The University of Texas Southwestern, Dallas, TX, and approved November 17, 2016 (received for review August 4, 2016)

Significance
Bacteria use two-component quorum-sensing systems to communicate with each other and their hosts. Catecholamines are host stress signals that participate in this dialogue and drive bacterial growth and virulence. Evidence from a preclinical model of inflammatory bowel disease (IBD) revealed that Enterobacteriaceae and pathways linked to catecholamine-mediated bacterial virulence are enriched in active disease. Here we targeted the bacterial adrenergic sensor, quorum-sensing Escherichia coli regulator C (QseC) of the QseBC two-component system. Genetically inactivating qseC in a pathogenic, IBD-associated E. coli strain (LF82) reduced its virulence and ability to colonize a murine host. Furthermore, biochemically inhibiting QseC attenuated disease in multiple preclinical IBD models. This report demonstrates that QseC signaling influences IBD pathogenesis and identifies QseC blockade as a therapeutic strategy for colitis-associated bacteria.
Abstract
Hosts and their microbes have established a sophisticated communication system over many millennia. Within mammalian hosts, this dynamic cross-talk is essential for maintaining intestinal homeostasis. In a genetically susceptible host, dysbiosis of the gut microbiome and dysregulated immune responses are central to the development of inflammatory bowel disease (IBD). Previous surveys of stool from the T-bet−/−Rag2−/− IBD mouse model revealed microbial features that discriminate between health and disease states. Enterobacteriaceae expansion and increased gene abundances for benzoate degradation, two-component systems, and bacterial motility proteins pointed to the potential involvement of a catecholamine-mediated bacterial signaling axis in colitis pathogenesis. Enterobacteriaceae sense and respond to microbiota-generated signals and host-derived catecholamines through the two-component quorum-sensing Escherichia coli regulators B and C (QseBC) system. On signal detection, QseC activates a cascade to induce virulence gene expression. Although a single pathogen has not been identified as a causative agent in IBD, adherent-invasive Escherichia coli (AIEC) have been implicated. Flagellar expression is necessary for the IBD-associated AIEC strain LF82 to establish colonization. Thus, we hypothesized that qseC inactivation could reduce LF82’s virulence, and found that an absence of qseC leads to down-regulated flagellar expression and motility in vitro and reduced colonization in vivo. We extend these findings on the potential of QseC-based IBD therapeutics to three preclinical IBD models, wherein we observe that QseC blockade can effectively modulate colitogenic microbiotas to reduce intestinal inflammation. Collectively, our data support a role for QseC-mediated bacterial signaling in IBD pathogenesis and indicate that QseC inhibition may be a useful microbiota-targeted approach for disease management.
Escherichia coli species, the predominant Gram-negative aerobes of the mammalian gut microbiota, contribute to the stability of the microbial community and maintenance of intestinal homeostasis (1). Compared with their symbiotic counterparts, pathogenic E. coli strains have acquired virulence factors involved in attachment, invasion, and toxin production (2). Inflammatory bowel disease (IBD), comprising Crohn’s disease (CD) and ulcerative colitis (UC), is a relapsing-remitting chronic inflammatory disorder influenced by genetic and environmental factors (3). IBD-associated gut microbiota alterations include a high abundance of E. coli in mucosal biopsy and stool specimens (4, 5). E. coli strains isolated from patients with active IBD frequently display pathogenic properties, such as adhesion to and invasion of host cells (6). A role for adherent-invasive E. coli (AIEC) in the pathogenesis of IBD, particularly in CD (7, 8), has been supported by whole metagenome sequencing data from human cohort samples (9).
The AIEC strain LF82 was isolated from a mucosal lesion of a patient with CD (10). LF82’s virulence, including its ability to adhere to and invade intestinal epithelial cells (IECs) and to survive and replicate within macrophages, is well established (11, 12). A key mechanism driving this virulence, although undoubtedly not the sole virulence mechanism, is the production of flagella. Inactivating the flagellar gene fliC blocks LF82 invasion and reduces its adhesion to cultured IECs (13). In the setting of dextran sodium sulfate (DSS)-induced mucosal injury, LF82 can exacerbate colonic inflammation, an effect abrogated in the absence of fliC (14). This flagella-mediated mucosal immune response is elicited via direct signaling through its cognate pattern-recognition receptor, Toll-like receptor 5 (14, 15). Flagellin, a principal flagellar protein, continues to be implicated as a dominant antigen in CD and as a target of both innate and adaptive immune responses central to the immunopathogenesis of IBD (16, 17).
Beyond flagella and their proteins, the virulence of many pathogenic E. coli is dependent on quorum-sensing E. coli regulator C (QseC), a component of the quorum-sensing E. coli regulators B and C (QseBC) two-component system (TCS) (18, 19). QseC is a histidine sensor kinase activated on detection of microbiota-generated autoinducer-3 (AI-3) or host stress signals, specifically the catecholamines (CAs) norepinephrine (NE) and epinephrine (EPI) (20). Both NE and EPI induce QseC-mediated virulence in vitro and blocking QseBC signaling with the QseC inhibitor LED209 reduces the in vivo virulence of Gram-negative pathogens, including enterohemorrhagic E. coli and Salmonella enterica Typhimurium (21⇓⇓⇓–25). Previous surveys of stool from the T-bet−/−Rag2−/− preclinical colitis model following various treatment interventions revealed microbial features that discriminate between active disease and remission. Gut microbiomes of mice with persistent colitis exhibited Enterobacteriaceae enrichment and, based on genomic functional analysis, an enhanced capacity for bacterial pathogenesis, including pathways involved in benzoate metabolism, TCSs, and cell motility (26). These pathways are linked to the QseBC signaling axis used by Enterobacteriaceae; catechols are intermediates of microbial benzoate degradation and CAs are host-derived catechols with a side-chain amine. Beyond their role as host stress hormones, CAs are also produced by the enteric nervous system and are important for regulating intestinal motility, electrolyte transport, and immune homeostasis (27). Moreover, evidence that levels of biologically active CAs in the intestinal lumen require specific bacterial-encoded enzymes suggests that luminal CA levels are gut microbiota-dependent (28). CAs continue to attract interest as communication molecules between host cells and microbes; these stress signals may influence microbial dysbiosis and increase the susceptibility to infection by altering the growth and virulence of human pathogens, including Enterobacteriaceae species (29, 30).
Given the essential function of flagella in LF82 virulence, we hypothesized that targeted inhibition of QseC would reduce its virulence potential. Thus, we generated isogenic LF82 qseC deletion and complemented strains to assess the effects of qseC inactivation on LF82 virulence. We found that the absence of qseC leads to down-regulated virulence gene expression and defects in flagellar assembly and motility in vitro and reduced colonization in vivo.
Furthermore, given that (i) microbial pathways involved in CA-related metabolism are enriched in experimental colitis (26) and human IBD (9) and (ii) QseC can mediate CA-induced virulence in pathogenic bacteria (31), we investigated whether QseC blockade with LED209 could be an effective microbiota-targeted approach for disease management. QseC inhibition attenuated disease in three preclinical models of colonic inflammation, including T-bet−/−Rag2−/−, Il10−/−, and DSS-exposed mice. This study provides evidence that QseC may be an upstream virulence node used by colitogenic bacteria to survey their host and potentiate disease, and may be a useful target for microbiota-directed therapies for the treatment of IBD.
Results
Absence of qseC Suppresses LF82 Flagella Expression and Motility.
To determine whether QseC plays an essential role in mediating LF82 virulence, we generated isogenic qseC deletion (∆qseC) and in-genome complemented (ΔqseC::qseC) strains. To investigate whether qseC regulates LF82 virulence genes, we evaluated the expression of CA-mediated quorum-sensing genes (qseB, qseC, and qseE) and their transcriptional targets involved in flagellar activation, assembly, and motility (flhC, flhD, fliA, fliC, and motB) in the wild-type (WT) and mutant strains. QseC expression was not detected in ∆qseC, and qseB was up-regulated by >90-fold (Fig. 1A), consistent with previous findings indicating that QseC autoregulates QseBC TCS activation and that qseC deficiency leads to qseB up-regulation (32). Importantly, gene expression in the flagellar regulon (flhDC) and genes involved in flagellar assembly and motility (fliA, fliC, and motB) were all down-regulated in ∆qseC (Fig. 1A).
QseC regulates LF82 flagellar expression and motility. (A) RT-qPCR of LF82 strains grown in LB. Genes normalized to rpoA. Fold change relative to WT strain. Data are mean ± fractional SD from two independent experiments. ***P < 0.001; ****,++++P < 0.0001, two-way ANOVA with Tukey’s post hoc test. * and + indicate differences relative to WT and complemented strains, respectively. (B) Representative TEM images. (Scale bar: 500 nm.) (C) Swim diameters on soft agar plates. Data are mean ± SEM from two independent experiments. **P < 0.01, one-way ANOVA with Tukey’s test. ns, not significant.
We used transmission electron microscopy (TEM) to visualize morphological differences in LF82 strains and to corroborate our flagellar expression data. There was marked reduction or complete loss of flagella on ∆qseC compared with the parental and complemented strains (Fig. 1B), supporting the idea that QseC regulates flagellar protein production. Given that flagella are required for LF82 virulence (13, 14) and that qseC inactivation suppresses the expression of flagellar genes and proteins (Fig. 1 A and B), we assessed flagella function via plate-based motility assays. The ∆qseC mutant exhibited motility defects based on a significantly smaller swim diameter compared with the parental strain (Fig. 1C). Collectively, these data substantiate that LF82 modulates its expression of flagellar genes and proteins through QseC.
QseC Mediates LF82 NE-Induced Flagella Expression and Motility.
As a bacterial adrenergic receptor, QseC senses the host stress molecules NE and EPI and activates a signaling cascade that induces virulence gene expression (20). Thus, we evaluated the effects of qseC inactivation on CA-induced flagellar gene expression in the parental and ∆qseC mutant strains and selected NE rather than EPI because NE is more abundant in the intestinal lumen (28). NE up-regulated the expression of quorum-sensing and flagellar genes in the parental strain, but not in ∆qseC (Fig. 2A). Based on differential gene expression patterns with NE, we examined its effects on LF82 swimming motility. NE enhanced the flagella-mediated motility of WT LF82, but not of ∆qseC, in a dose-dependent manner (Fig. 2B). These observations confirm that QseC mediates NE’s effects on LF82 flagellar expression and motility, with the caveat that NE-induced QseC-signaling has broad effects on bacterial physiology (19).
QseC mediates NE-induced flagellar expression and motility in LF82. (A) RT-qPCR of LF82 strains grown in acidified or NE-supplemented LB. Genes normalized to rpoA. (B) Fold change in swim diameters on NE-supplemented soft agar plates. For A and B, fold change relative to WT grown in LB, pH 6. Data are mean ± fractional SD from two independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001, two-way ANOVA with Sidak’s multiple comparison test. ns, not significant.
QseC Influences LF82's Effects in a Low-Complexity Microbiota.
We next sought an in vivo system to assess whether QseC contributes to LF82 colonization. In specified pathogen-free (SPF) WT mice, LF82 colonization does not alter bacterial load, microbiota composition, or induce intestinal inflammation (33). This suggests that LF82 colonization is transient and does not activate innate immune responses in a healthy host. Thus, we hypothesized that gnotobiotic mice harboring the altered Schaedler flora (ASF) (34), an eight-bacterial strain consortium devoid of Proteobacteria (including E. coli), could provide a permissive niche for LF82 and furnish an in vivo system to assess whether qseC plays a role in LF82 colonization and persistence. This system has been successfully used to study other Gram-negative pathogens that are unable to stably colonize mice under SPF conditions (35). Using gnotobiotic-ASF mice also provides an opportunity to mirror in vivo features of a low-complexity microbiota commonly seen in IBD (36).
We transferred gnotobiotic-ASF mice from gnotobiotic isolators to our conventional animal facility, hereinafter referred to as ex-gnotobiotic-ASF (ex-ASF) mice. To maintain a low-complexity microbiota, ex-ASF mice were handled separately using animal husbandry practices that minimize microbial transfer. On transfer to conventional housing conditions, ex-ASF mice were orally inoculated with 108 colony-forming units (CFUs) of the WT or ∆qseC mutant LF82 strains. Experimental schema details are provided in Fig. S1.
Experimental schema for studying LF82’s effects in a low-complexity microbiota. On day 0, gnotobiotic-ASF mice were transferred from gnotobiotic isolators to conventional housing and were subsequently inoculated with LF82. Stool samples from LF82-inoculated ex-ASF breeding pairs (BP) and litters (L) were collected on day 113 for plating and on day 115 for sequencing.
To determine whether QseC promotes LF82 persistence within a low-complexity microbiota, we used 16S ribosomal RNA (rRNA) gene surveys to compare gut microbial communities of gnotobiotic-ASF and ex-ASF mice inoculated with the parental or ∆qseC mutant strains. DNA from the cecum of a gnotobiotic-ASF mouse and stool pooled from LF82-associated ex-ASF breeding pairs and corresponding litters were sequenced. Reads were binned into approximately species-level operational taxonomic units (OTUs) and analyzed using QIIME, a bioinformatics pipeline for microbial community analysis (37).
When phylum-level relative abundances were examined, only reads aligning to the microbial clades of the ASF were present in the gnotobiotic-ASF cecal sample, which is representative of the initial microbial community before LF82 inoculation (Fig. 3A). Higher levels of Proteobacteria were observed in mice exposed to the LF82 WT vs. the ∆qseC mutant strain (Fig. 3A). Of sequences that mapped to Proteobacteria, >98% belonged to the family Enterobacteriaceae, with WT LF82-associated mice having an average sixfold greater Enterobacteriaceae abundance than ∆qseC-associated mice (Fig. 3B).
QseC influences LF82's effects in a low-complexity microbiota. 16S rRNA gene surveys of cecal content from a gnotobiotic-ASF mouse and stool from LF82-inoculated ex-ASF breeding pairs (BP) and litters (L). Number of pooled samples for sequencing: WT LF82-associated mice: BP, n = 2; L1, n = 3; L2, n = 5; ∆qseC-associated mice: BP, n = 2; L1, n = 9. (A) Phylum-level relative abundances from QIIME-classified sequences. ASF strain list. (B) Enterobacteriaceae relative abundances. ND, not detected. (C) LF82 CFU in stool. Filled triangles represent L1 progeny; filled circles, L2 progeny. Data are mean ± SEM. **P < 0.01, Kruskal–Wallis test with Dunn’s post hoc test. (D) Alpha diversity using Chao1 and phylogenetic diversity (PD) indices. Boxplots are top, median, and bottom quartiles. Whiskers and outliers are plotted with Tukey’s method. (E) Beta diversity using unweighted UniFrac distances. (F) Histological colitis scores of DSS-treated WT LF82-associated (n = 29) and ∆qseC-associated (n = 10) ex-ASF mice. Data are mean ± SEM from two to three independent experiments. A colitis score >2 indicates active disease; a score ≤2, remission. **P < 0.01, Mann–Whitney U test.
To examine whether Enterobacteriaceae abundances correspond with LF82 viable counts, we plated stool samples from individual mice of each breeding pair and litter. LF82 CFUs were higher in WT LF82-associated progeny compared with ∆qseC-associated progeny (Fig. 3C). Thus, qseC loss may hinder LF82’s ability to colonize and persist in a host even when there is minimal niche competition.
To characterize differences in community structure in the presence of WT and mutant LF82 strains, we analyzed microbial community diversity. Within-sample diversity of gut microbiomes of WT LF82-associated mice deviated less from the original ASF input community compared with ∆qseC-associated mice (Fig. 3D), suggesting that the parental strain is more adept at colonizing and persisting in a low-complexity microbiota and may even facilitate a state of reduced diversity. Between-sample diversity and hierarchical clustering by similarity revealed that microbial communities segregated according to ASF input or presence of LF82 WT or ∆qseC, indicating that these strains induce distinct microbiota responses (Fig. 3E). Moreover, these data show that LF82-associated microbiota phenotypes may be transmissible to offspring, given the similar community composition, structure, and diversity in breeding pairs and their corresponding litters. These findings support LF82’s ability to sustain and thrive in a microbiota with reduced diversity.
Although ex-ASF mice harboring LF82 WT or ∆qseC do not display histological signs of intestinal inflammation, their reduced microbial community diversity and the fact that LF82 is an IBD-associated AIEC prompted us to ask whether LF82-associated ex-ASF mice would display exacerbated colitis in response to the mucosal disruptant and colitogenic agent DSS. Treating LF82 WT- and ∆qseC-associated ex-ASF mice with DSS could reveal whether the presence of functional qseC is sufficient to worsen colitis and whether QseC is a worthy microbial target in IBD pathogenesis. Indeed, after DSS treatment, WT LF82-associated mice had higher colitis scores than ∆qseC-associated mice (mean colitis score, 6.41 ± 0.22 vs. 4.3 ± 0.76) (Fig. 3F), and both strains were detectable in stool at similar levels (mean CFU/g stool, 7.62 ± 3.23e107 for LF82 WT vs. 4.54 ± 2.27e107 for LF82 ∆qseC).
As a virulence master regulator, QseC broadly affects bacterial physiology. Previous gene expression analysis in WT vs. ∆qseC uropathogenic E. coli strains revealed that qseC deletion results in altered nucleotide, amino acid, and carbon metabolism (19). Observing that LF82 is more adept at persisting in a low-complexity microbiota (Fig. 3), we queried whether qseC loss affects growth kinetics in vitro and survival in vivo. LF82 strains cultivated in aerobic or anaerobic conditions showed comparable growth and survival (Fig. S2). Thus, whereas the absence of qseC does not result in growth or survival defects, qseC has an appreciable influence on LF82’s persistence over time and, importantly, may affect the susceptibility to mucosal injury, as observed with DSS treatment.
Growth and survival of LF82 strains. (A and B) In vitro growth in aerobic (Left) or anaerobic (Right) conditions. Data are mean ± SEM from two independent experiments. *,#,‡P < 0.05; ++P < 0.01; +++P < 0.001; ****P < 0.0001, two-way ANOVA with Tukey’s test. *WT vs. ∆qseC; +∆qseC vs. ∆qseC::qseC; #WT vs. ∆qseC::kan; ‡WT pNG162 vs. ∆qseC::kan. (C) In vivo survival in mice inoculated with 1:1 cultures of LF82 pNG162:∆qseC::kan or ∆qseC/pNG162:∆qseC::kan. LF82 viable counts in stool are shown. Symbols represent individual mice. Data are mean ± SEM from two independent experiments. **P < 0.01, two-way ANOVA with Sidak’s multiple comparison test. ns, not significant. Detection limit, 102 CFU/g.
LED209 Inhibition of QseC Attenuates Colitis in SPF Mice.
Based on (i) our observation that QseC functions in AIEC persistence and exacerbated colitis in ex-ASF mice and (ii) work demonstrating that the QseC inhibitor LED209 reduces Enterobacteriaceae pathogen virulence in vivo (21, 22), we examined the effects of QseC blockade on host disease status in three experimental colitis models maintained under SPF conditions. This approach would allow us to explore the translational applicability of a QseC-targeted antivirulence strategy for IBD. We hypothesized that LED209 may reduce disease activity, given that the Enterobacteriaceae family and Proteobacteria phylum have been implicated in instigating and/or perpetuating intestinal inflammation in preclinical colitis models (38). Thus, LED209 was administered daily to SPF T-bet−/−Rag2−/−, Il10−/−, and DSS-treated WT mice. Study design, histological colitis scores, and other parameters of disease are provided in Fig. 4 and Fig. S3.
LED209 inhibition of QseC attenuates colitis. (A and B) Study design: T-bet−/−Rag2−/− (n = 31), Il10−/− (n = 35), and WT DSS-treated (n = 23) mice under SPF conditions. Mice were orally administered LED209, vehicle, or water (sham) daily. (C–E) Histological colitis scores. Symbols represent individual mice. Data are mean ± SEM from two to three independent experiments. A colitis score >2 indicates active disease; a score ≤2, remission. **P < 0.01, ****P < 0.0001, Kruskal–Wallis test with Dunn’s post hoc test.
LED209 inhibition of QseC in preclinical IBD models. (A) Percent change in baseline body weight in T-bet−/−Rag2−/− (n = 31), Il10−/− (n = 35), and WT DSS-exposed (n = 23) mice under SPF conditions. Data are mean ± SEM from two to three independent experiments. *P < 0.05, **P < 0.01, two-way ANOVA with Tukey’s test. Significance was observed between the LED209 and vehicle control groups. (B) Postintervention colon lengths. Symbols represent individual mice. Data are mean ± SEM from two to three independent experiments. *P < 0.01, Kruskal–Wallis test with Dunn’s post hoc test.
Treatment with LED209 conferred almost complete protection from colitis in T-bet−/−Rag2−/− mice and reduced disease severity in DSS-exposed mice (Fig. 4 A and C). In Il10−/− mice, a greater proportion of LED209-treated mice (9 of 13; 69.2%) showed no signs of mucosal inflammation or injury compared with sham (7 of 11; 63.6%) and vehicle control (6 of 11; 54.5%) mice (Fig. 4B). Il10−/− mice often show gender differences in disease activity, with males tending to have more severe disease than females, as was also observed with LED209 treatment (Fig. S4). Our histopathology data indicate that LED209 can attenuate disease severity in genetic and chemically induced models of colonic inflammation, presumably by inhibiting QseC virulence pathways.
Gender differences in disease severity of SPF Il10−/− mice following an LED209 intervention. Symbols represent individual mice. Data are mean ± SEM from two independent experiments. A histological colitis score >2 indicates active colitis; ≤2, remission.
Enterobacteriaceae can increase their colitogenic potential under inflammatory conditions and may participate in initiating and potentiating inflammation in IBD (1, 38⇓–40). LED209 studies measuring pathogen growth and survival in vitro and in infection models suggest that this antivirulence approach can reduce pathogenicity without affecting bacterial growth (21). Thus, we measured shifts in Enterobacteriaceae levels in stool from baseline using qPCR to assess whether we could relate abundance to treatment response in our preclinical colitis models. In Il10−/− and DSS-treated mice, LED209 blocked the expansion of Enterobacteriaceae compared with vehicle controls, which exhibited a >5.5-fold increase from baseline (Fig. S5 B and C). T-bet−/−Rag2−/− mice displayed an inverse trend, with levels remaining stable in vehicle controls and increasing by 2.4-fold in LED209-treated mice (Fig. S5A).
Shifts in Enterobacteriaceae abundance in stool of preclinical IBD models following an LED209 intervention. (A–C) RT-qPCR analysis of Enterobacteriaceae in preintervention (pre) and postintervention (post) stool of (A) SPF T-bet−/−Rag2−/− mice (vehicle, n = 10; LED209, n = 11), (B) SPF Il10−/− mice (vehicle, n = 11; LED209, n = 13), and (C) SPF WT DSS-exposed mice (vehicle, n = 4; LED209, n = 12). Enterobacteriaceae 23S rRNA was normalized to total 16S rRNA. Fold change relative to baseline is shown. Data are mean ± fractional SD from two to three independent experiments. ****P < 0.0001, two-way ANOVA with Sidak’s multiple comparison test. ns, not significant. (D) Enterobacteriaceae in stool of LED209-treated mice with active disease vs. those in remission. Log10 scale. Symbols represent individual mice. Data are mean ± SEM for two to three independent experiments. *P < 0.05, two-way ANOVA with Sidak’s multiple comparison test. ns, not significant.
Segregating Enterobacteriaceae relative abundances of LED209-treated mice based on active disease vs. remission for each model revealed comparable levels between disease states in T-bet−/−Rag2−/− and DSS-exposed mice (Fig. S5D). However, Il10−/− mice with active disease had marked variations in Enterobacteriaceae, with higher levels observed in several mice with active disease compared with mice in remission (Fig. S5D). Thus, LED209 has differential effects on Enterobacteriaceae levels that are distinct between models and between disease states of treated mice.
Given that QseC is important for Enterobacteriaceae virulence and that QseC senses host NE and EPI, we evaluated the effects of QseC inhibition on the luminal CA pool of the cecum and colon. Cecal CA levels were comparable between treatment groups and across models (Fig. S6A). Significant differences in cecal NE and EPI were observed only in DSS-treated mice, but this effect was driven by a single outlier (Fig. S6A). There were no significant differences or trends between treatment groups or across models in stool CA level changes from baseline (Fig. S6B). Despite differences in disease severity and Enterobacteriaceae abundances between treatment groups in these models, we did not observe major perturbations to the CA economy of the intestinal lumen.
CA levels in cecal content and stool of preclinical IBD models following an LED209 intervention. (A) CA abundance in cecal content. Symbols represent pooled samples. Data are mean ± SEM from two to three independent experiments. *P < 0.05, **P < 0.01, two-way ANOVA with Tukey’s test. Number of mice and associated number of pools were as follows. SPF T-bet−/−Rag2−/−: sham, n = 10 (7 pools); vehicle, n = 10 (8 pools); LED209, n = 11 (9 pools). SPF Il10−/−: sham, n = 11 (9 pools); vehicle, n = 11 (9 pools); LED209, n = 13 (13 pools). SPF WT-DSS: sham, n = 7 (5 pools); vehicle, n = 4 (3 pools); LED209, n = 12 (7 pools). (B) CA fold changes relative to baseline in stool. Symbols represent pooled samples. Data are mean ± fractional SD from two to three independent experiments. No significant differences were observed by two-way ANOVA. ND, outside the detection limit of HPLC. Number of mice and associated number of preintervention (pre) and postintervention (post) pools were as follows. SPF T-bet−/−Rag2−/−: sham, n = 6 (2 pre/2 post); vehicle, n = 6 (2 pre/2 post); LED209, n = 7 (2 pre/2 post). SPF Il10−/−: sham, n = 11 (4 pre/5 post); vehicle, n = 10 (7 pre/5 post); LED209, n = 13 (9 pre/9 post). SPF WT-DSS: sham, n = 6 (2 pre/3 post); vehicle, n = 4 (2 pre/1 post); LED209, n = 12 (4 pre/4 post).
To determine whether LED209 may exert direct anti-inflammatory effects on the host, we tested its effects on cytokine gene expression in immune cell populations in vitro. LED209 exposure did not alter the expression of Il10 or Tnfα in mesenteric lymph node (MLN)-derived CD11c+ dendritic cells (Fig. S7A) and had negligible effects on Il10, Tnfα, and Ifnγ in splenic-derived CD4+ cells (Fig. S7B). These observations suggest that LED209 may disrupt the ability of bacteria to communicate with each other and their host and in turn attenuate bacterial virulence and intestinal inflammation without directly influencing the luminal CA pool or host cytokine gene expression.
In vitro LED209 stimulation and cytokine gene expression in isolated immune cells. (A) RT-qPCR of MACS-enriched CD11c+ dendritic cells from MLNs. (B) RT-qPCR of MACS-enriched CD4+ cells from spleens. For A and B, genes are normalized to mouse β-actin, and fold change is relative to vehicle control. Data are mean ± fractional SD from three independent experiments (n = 4–8 mice/experiment). *P < 0.05, Kruskal–Wallis test with Dunn’s post hoc test.
Discussion
Enterobacteriaceae, particularly E. coli, have been implicated in IBD pathogenesis given their abundance in mucosal lesions of patients with CD and pathogenic properties in vitro and in vivo (4, 5, 7). The prototype AIEC strain LF82 requires flagella expression to adhere to and invade host cells (13). As a virulence master regulator, QseC plays a crucial role in promoting the virulence of pathogenic E. coli, including enterohemorrhagic and enteropathogenic E. coli strains (41), but little is known about its regulatory function in AIEC. To examine whether QseC regulates LF82 virulence, we genetically manipulated the qseC locus of the parental strain to generate an isogenic qseC deletion mutant and in-genome complemented strain. Our characterization revealed that the ∆qseC mutant had down-regulated expression of flagellar assembly and motility genes, loss of flagellar surface proteins, and reduced swimming motility, which were rescued with qseC complementation (Fig. 1). These findings support the idea that QseC regulates LF82 virulence, at least in part, by activating flagellar expression and motility.
Host–microbiota communication is increasingly recognized as an important aspect of both symbiosis and pathogenesis. Integral to the microbiota’s surveillance and collective decision making process are quorum-sensing TCSs. For QseBC, the presence of microbiota-generated hormone-like compounds (AI-3) or host stress signals (NE or EPI) can initiate a virulence program with detrimental consequences for the host (42). This property motivated us to examine whether QseC is necessary for NE-induced flagellar gene expression in LF82. We found that the parental strain, and not ∆qseC, had enhanced expression of NE-regulated TCS and flagellar genes and exhibited a dose-dependent increase in swimming motility (Fig. 2), indicating that QseC functions as an adrenergic receptor in LF82. We also observed that QseEF, an alternative NE-sensing TCS, was also regulated by QseC activation, given that expression of the qseE sensor kinase was reduced in the ∆qseC mutant and its levels did not change in the presence of NE (Figs. 1A and 2A). These findings are in agreement with other reports of QseC acting upstream of QseEF (43) and establish that NE-induced virulence in LF82 is regulated primarily by QseC.
Because LF82 is unable to colonize the gut of a healthy host (33), we established an in vivo system for examining the effects of qseC inactivation on LF82 colonization, persistence, and response to the mucosal disruptant and colitogenic agent DSS using ex-ASF mice. The absence of Proteobacteria in the ASF community enabled colonization of LF82. The 16S rRNA gene surveys revealed that total Enterobacteriaceae were more enriched in ex-ASF mice associated with LF82 WT vs. ∆qseC, which were consistent with culturable counts of LF82 (Fig. 3B). These results suggest that despite similar growth kinetics in vitro, LF82 may have a fitness advantage compared with the ∆qseC mutant within the gastrointestinal tract of a host over time.
Additional insights relevant to the underlying microbial dysbiosis associated with IBD were gleaned from our 16S rRNA gene sequence analysis. We observed that microbiotas of WT LF82-associated mice had less within-sample diversity and were more analogous to the ASF input community compared with microbiotas of ∆qseC-associated mice. Because LF82 is unable to stably colonize conventional mice under SPF conditions (33), these data show that LF82 can successfully colonize a host harboring a low-complexity microbiota and may even promote the maintenance of a less diverse state to ensure its survival. Whether the loss in microbial community diversity associated with dysbiosis is a cause or consequence of chronic intestinal inflammation remains elusive. Perhaps certain bacteria, like AIEC, are involved throughout the disease continuum, contributing to both the initiation and progression of disease by promoting and sustaining dysbiosis and altered host–microbiota interactions. Evidence that LF82 can preserve a specific microbiota phenotype was observed, with breeding pairs of LF82-associated ex-ASF mice exhibiting similar microbial community structures as their litters (Fig. 3), demonstrating that features of an LF82-associated microbiota may be transmissible. With the genetic foundation of most complex immune-mediated disorders, such as IBD, the increased risk of disease between family members may be augmented by the transmissibility of the microbiome.
Given that LF82 is more adapted to colonize the intestines of humans rather than mice, human CEACAM6 transgenic mice may be a useful model for understanding LF82–host interactions (44). However, intestinal inflammation in this model requires both antibiotic-mediated microbiome alterations and mucosal injury induced by DSS. Although this system affords opportunities for studying LF82 in vivo, its applicability as a robust preclinical model has major caveats. Our results support the idea that QseC is an important regulator of LF82 virulence and modulator of the microbiome and host–microbiota homeostasis, all of which strengthen the case for further investigation into perturbing QseC and other key virulence mediators as an approach to IBD therapeutics.
Quorum-sensing through TCSs is considered an adaptive and auxiliary function of bacteria—one that is critical for virulence, infection, and enhanced fitness but not essential for growth and survival (45). This concept has been strengthened by studies demonstrating that quorum-sensing inhibitors, like LED209, can reduce bacterial virulence without affecting growth (21). Thus, we investigated whether inhibition of QseC could reduce disease severity in three preclinical models of colonic inflammation under SPF conditions. We performed LED209 interventions in the (i) T-bet−/−Rag2−/− model, which has defects in both innate and adaptive immunity; (ii) Il10−/− model, in which mice develop spontaneous colitis from loss of regulatory immune function; and (iii) DSS-induced acute colonic injury model, which recapitulates histological features of human UC. Using these distinct models of experimental colitis allowed us to assess the applicability of QseC inhibition under different genetic contexts. We observed an attenuation of disease in T-bet−/−Rag2−/− and DSS-exposed mice, and a modest benefit in Il10−/− mice (Fig. 4). Intriguingly, LED209’s ability to ameliorate disease in Il10−/− and DSS-treated mice appeared to be dependent on blocking the expansion of Enterobacteriaceae, given mice with active colitis had a greater Enterobacteriaceae abundance on average compared with mice in remission (Fig. S5). Taken together, the data from LED209 interventions in genetically distinct mouse models of colonic inflammation suggest an underlying role for QseC-mediated virulence in disease pathogenesis.
Several mechanisms for how antivirulence drugs inhibit exogenous pathogens have been suggested, including preventing the colonization of pathogens during passage through the gastrointestinal tract and facilitating the elimination of pathogens by the host immune system (46). Whether these mechanisms hold true for colitogenic bacteria, which are already a part of the endogenous microbiota, requires further study.
In summary, we have demonstrated that perturbing the Enterobacteriaceae adrenergic receptor QseC with the biochemical inhibitor LED209 can attenuate experimental colitis and that genetically inactivating qseC in a pathogenic IBD-associated E. coli strain can reduce its virulence in vitro and abrogate its ability to persist in a low-complexity microbiota in vivo. These results provide insight into the use of an antivirulence approach for targeting not only pathogens, but also a much larger collection of colitogenic bacteria.
Materials and Methods
Bacterial Strains, Plasmids, and Growth Conditions.
AIEC strain LF82 was used in this study (10). Bacterial strains and plasmids are listed in Table S1. Details are provided in SI Materials and Methods.
Bacterial strains and plasmids
Gene Inactivation and Complementation.
Isogenic mutants were generated with PCR products using the method described by Datsenko and Wanner (47), with modifications for pathogenic E. coli (48). Primers for gene manipulation and verification are listed in Table S2. More details are provided in SI Materials and Methods.
Primers
RT-qPCR of Bacterial Cultures.
RNA was isolated and RT-qPCR was performed as described in SI Materials and Methods. The RT-qPCR primers are listed in Table S2.
TEM Analysis.
Bacterial cells were visualized using a JOEL 1200 EX transmission electron microscope, as detailed in SI Materials and Methods.
Plate-Based Bacterial Motility Assays.
Bacterial motility assays are described in detail in SI Materials and Methods.
Animal Husbandry.
Gnotobiotic BALB/cByJ mice harboring the ASF consortium (49) were bred and maintained in gnotobiotic isolators. SPF BALB/cByJ T-bet−/−Rag2−/−, Il10−/−, WT, and ex-ASF mice were housed in a barrier facility. All animal experiments were approved and conducted in accordance with guidelines of Harvard Medical School’s Standing Committee on Animals and the National Institutes of Health. Details of the ASF consortium and maintenance of ex-ASF mice are provided in SI Materials and Methods.
Infection of ex-ASF Mice with LF82.
On transfer from gnotobiotic isolators to conventional housing, 8-wk-old ex-ASF mice were orally inoculated with 108 CFU of LF82 WT or ∆qseC, as described in SI Materials and Methods.
16S rRNA Gene Surveys of Stool and Sequence Analysis.
DNA was isolated from stool, amplified for the V4 region of the 16S rRNA gene, and sequenced using an Illumina MiSeq instrument. Reads were processed and analyzed with QIIME (37). Details are provided in SI Materials and Methods.
In Vivo Competition.
JAX BALB/cByJ (7- to 8-wk-old) mice were inoculated with 1:1 cultures of 109 CFU of LF82 pNG162:∆qseC::kan or ∆qseC/pNG162:∆qseC::kan, as described in SI Materials and Methods.
DSS Treatment of ex-ASF Mice.
ex-ASF mice (5- to 7-wk-old) underwent a 7-d DSS intervention, as described in detail in SI Materials and Methods.
LED209 Interventions.
Mice were orally administered an equal volume of LED209 (0.4 mg/mouse), vehicle, or water (sham). Details are provided in SI Materials and Methods.
Histology.
Sections were examined and degree of colitis was scored as described previously (50) and in SI Materials and Methods.
qPCR Analysis of Enterobacteriaceae in Stool.
DNA was isolated from stool and qPCR was performed, as described in SI Materials and Methods.
Luminal CA Measurements by HPLC.
CAs were measured as described previously (28) and in SI Materials and Methods.
In Vitro LED209 Stimulation and RT-qPCR for Cytokine Gene Expression.
MLNs and spleens were harvested from 5- to 7-wk-old JAX BALB/cByJ mice to enrich for CD11c+ dendritic cells and CD4+ cells, respectively (SI Materials and Methods).
Statistical Analysis.
All statistical tests were performed in Graphpad Prism 6.0h. Averages are reported as mean ± SEM except for fold change, for which averages are mean ± fractional SD. Absence of error bars indicates minimal SEM or SD.
SI Materials and Methods
Bacterial Strains, Plasmids, and Growth Conditions.
For maintenance of ampicillin- and erythromycin-resistant AIEC strain LF82 (10), LB broth, LB agar [1.5% (wt/vol) Bacto agar], and MacConkey agar (all Difco) were used. When appropriate, the following antibiotics were added: 100 μg/mL ampicillin (Amp), 20 μg/mL chloramphenicol (Cm), 20 μg/mL erythromycin (Eryth), 100 μg/mL hygromycin B (Hyg), 50 μg/mL kanamycin (Kan), and 100 μg/mL spectinomycin (Spc). Bacterial growth was measured as optical density at 600 nm (OD600). Plasmid DNA was isolated using Miniprep or Midiprep Kits (Qiagen). Experiments were routinely performed under aerobic conditions. For in vitro growth dynamics, single colonies were grown statically in 3–5 mL LB overnight at 37 °C aerobically or in an anaerobic hood (Coy Laboratory Products). Overnight cultures were back-diluted to OD600 of 0.01 and grown statically in LB for 8 h under aerobic or anaerobic conditions. Colony-forming units (CFU) per milliliter were determined by plating serial dilutions on MacConkey agar.
Gene Inactivation.
For generating LF82 ∆qseC, the qseC chromosomal sequence in LF82 was replaced with a selectable resistance gene generated by PCR (47, 48, 51). This resistance gene was flanked by flippase recognition target (FRT) sites to facilitate flip recombinase (FLP)-mediated excision. The PCR product was generated using 80-bp primers with homology to regions adjacent to qseC and plasmid pKD4 harboring a KanR cassette as a template. PCR of pKD4 was performed with primers LF82_qseC_80bp_up-For and LF82_qseC_80bp_dwn-Rev. Each 50-μL reaction contained 1× AccuPrime Pfx reaction mix and 0.4 μL (1 U) of AccuPrime Pfx DNA polymerase (Invitrogen), 300 nM of each primer, and 50 ng of pKD4 template. PCR cycling consisted of initial denaturation at 94 °C for 2 min; 30 cycles of denaturation at 94 °C for 30 s, annealing at 50 °C for 30 s, and extension at 68 °C for 2 min; and a final extension at 68 °C for 5 min. PCR fragments were gel-purified using the QIAquick Gel Extraction Kit (Qiagen) and underwent two consecutive DpnI (New England BioLabs) restriction digestions (0.2 U/μL in 1× CutSmart Buffer) at 37 °C for 60 min, with final heat inactivation at 80 °C for 20 min. Fragments were purified using the QIAquick PCR Purification Kit (Qiagen) and desalted by drop dialysis (EMD Millipore). In parallel, LF82 was transformed with plasmid pTKred encoding λ-Red enzymes synthesized under control of an isopropyl-β-d-1-thiogalacto-pyranoside (IPTG)-inducible promoter. pTKred was maintained at 30 °C with 100 μg/mL Spc. λ-Red expression was induced with 1 mM IPTG (Invitrogen).
After transformation with the PCR product, isogenic KanR mutants were selected on LB agar containing 50 μg/mL Kan. Gene replacement by KanR cassette in mutants was confirmed by colony PCR using ChoiceTaq DNA polymerase (Denville). LF82 ∆qseC::kan was transformed with plasmid pCP20 harboring the FLP enzyme. pCP20 was maintained at 30 °C with 10 μg/mL Cm. After transformation with pCP20, bacteria were plated and grown at 37 °C. LF82 ∆qseC mutants (CmSKanS) were confirmed by plating, colony PCR, and Sanger sequencing. Primers are listed in Table S2.
Gene Complementation.
For generating LF82 ∆qseC::qseC, LF82 ∆qseC::kan/pTKred was transformed with a 5.1-kb fragment from plasmid pMGR3 generated by Gibson cloning (New England BioLabs). Plasmid pDONR221 (Invitrogen) with ApaI and EcoRV restriction sites served as the vector backbone. Plasmid pTKIP harboring a HygR cassette flanked by FRT sites served as the selectable resistance gene to screen for chromosomal integration. PCR products were generated with primers designed by Gibson Assembly software that flanked regions adjacent to the qseBC operon and incorporated a 1-kb region downstream of qseBC. DNA fragments were prepared with 18-bp homology. PCR was performed with Q5 High-Fidelity DNA polymerase (New England BioLabs) using the LF82 parental strain as a template. The Gibson assembly reaction, containing a 2.5-fold molar excess of insert fragments per 100 ng of vector backbone, was incubated in a thermocycler at 50 °C for 60 min and then directly transformed into chemically competent E. coli DH5α (New England BioLabs).
HygR colonies were selected on low-salt LB agar (0.5% yeast, 1% tryptone, 0.5% NaCl, and 1.5% agar) (all Difco) containing 100 μg/mL Hyg. Isolated plasmid DNA was screened for correct size and sequence. Plasmid pMGR3 was digested with ApaI and EcoRV (New England BioLabs) to generate the 5.1-kb linear fragment. λ-Red expression was induced in LF82 ∆qseC::kan/pTKred with 1 mM IPTG. After transformation with the 5.1-kb fragment, HygR mutants were selected and transformed with pCP20 before being plated and grown at 37 °C. LF82 ∆qseC::qseC mutants (CmSHygS) were confirmed by plating, colony PCR, and Sanger sequencing. Primers are listed in Table S2.
RT-qPCR of Bacterial Cultures.
Single colonies were grown statically in 4–5 mL of LB overnight at 37 °C. Overnight cultures were back-diluted to an OD600 of 0.02 and grown statically at 37 °C for 6 h in LB, acidified LB (pH 6), or LB supplemented with 5 μM NE (Sigma-Aldrich). Total RNA was extracted from bacterial pellets using phenol-chloroform bead-beating, followed by DNA/RNA separation and purification using the AllPrep DNA/RNA Mini Kit (Qiagen). Rigorous DNase treatments were performed both on-column (Qiagen) and in-solution (Ambion) before cDNA synthesis and RT-qPCR. RNA was quantified using a NanoPhotometer Pearl spectrophotometer (Denville). cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad). RT-qPCR was performed using the KAPA SYBR FAST Universal qPCR Kit (Kapa Biosystems). An annealing temp of 60 °C was used across primer sets (200 nM each). All reactions were performed in duplicate using a Stratagene Mx3005P qPCR system (Agilent). Genes were normalized to rpoA (52⇓–54). Fold change was analyzed by the 2−ΔΔCt method, ΔΔCt = (Ct,target gene – Ct,rpoA)experimental − (Ct,target – Ct,rpoA)reference. Error was calculated as fractional SD = (SD)/(mean value). Primers are listed in Table S2.
TEM Analysis.
Single colonies were grown statically in 4–5 mL of LB overnight at 37 °C. Cells were applied to carbon-formvar gold grids and washed with sterile PBS. Negative staining was performed with 1% uranyl formate (pH 4) for 30 s. Grids were washed twice with sterile water, blotted, and then visualized. Representative micrographs were obtained at a magnification of 12,000×.
Plate-Based Bacterial Motility Assays.
Soft (0.3%) LB agar plates were prepared. As indicated, the plates contained 50–500 μM NE (Sigma-Aldrich) or equivalent amounts of acidified water to normalize pH to 6. Single colonies were grown statically in 4–5 mL of LB overnight at 37 °C. Overnight cultures were adjusted to an OD600 of 1.0. Plates were inoculated with 3 μL of culture and then incubated at 37 °C for 8–16 h. Diameters of motility halos were measured to assess differences in motility between strains.
Animal Husbandry.
Altered Schaedler flora strains included Clostridium sp., ASF356; Lactobacillus sp., ASF360; Lactobacillus murinus, ASF361; Mucispirillum schaedleri, ASF457; Eubacterium plexicaudatum, ASF492; Firmicutes bacterium, ASF500; Clostridium sp., ASF502; and Parabacteroides sp., ASF519 (49). Gnotobiotic-ASF mice (6 wk old) were transferred from gnotobiotic isolators at Children’s Hospital Boston to the SPF barrier facility at Harvard T.H. Chan School of Public Health. Ex-gnotobiotic-ASF (ex-ASF) mice were maintained using prudent husbandry practices to minimize microbial transfer from other SPF mouse lines, including thoroughly spraying hoods with detergent/disinfectant (Quatricide) before handling ex-ASF cages and handling ex-ASF mice before any other SPF mice.
Infection of ex-ASF Mice with LF82.
Single colonies of LF82 parental strain or LF82 ∆qseC were grown statically in 4–5 mL of LB overnight at 37 °C. Cultures were harvested by centrifugation at 4,000 × g for 10 min at 4 °C. Supernatants were discarded, and bacterial pellets were resuspended in LB broth and stored at −80 °C. The CFU/mL of each strain was confirmed before orally gavaging mice with 108 CFU. Stool was collected on postinoculation day (pid) 113, homogenized, serially diluted in Dulbecco's PBS (dPBS; Cellgro), and plated on MacConkey agar (100 μg/mL Amp, 20 μg/mL Eryth) to enumerate LF82. CFU/g stool was normalized by sample wet weight.
16S rRNA Gene Surveys of Stool and Sequence Analysis.
Stool was collected and pooled on pid 115, homogenized in RNAlater (Ambion), held at 4 °C overnight, and stored at −80 °C before processing. Nucleic acids were extracted using phenol-chloroform bead-beating, followed by DNA/RNA separation and purification using the AllPrep DNA/RNA Mini Kit (Qiagen). DNA was quantified using a NanoPhotometer Pearl spectrophotometer (Denville) and stored at −20 °C before undergoing 16S rRNA gene amplification by PCR. Primers (515F/806R) targeting the V4 hypervariable region and incorporating Illumina adapters and sample barcode sequences were used, following a previously described protocol with minor modifications (55). In brief, each 25-μL reaction contained 10 μL of diluted template (1:50), 10 μL of HotMaster Mix with HotMaster Taq DNA Polymerase (5 PRIME), and 5 μL of primer mix (2 μM each). PCR cycling consisted of initial denaturation at 94 °C for 3 min; 30 cycles of denaturation at 94 °C for 45 s, annealing at 50 °C for 60 s, and extension at 72 °C for 5 min; and a final extension at 72 °C for 10 min.
Amplicons were quantified using the Quant-iT DNA HS Assay Kit (Life Technologies), pooled in equimolar concentrations, and size-selected to 375–425 bp on Pippin Prep (Sage Science). Final library size and DNA were quantified using the Quant-iT DNA HS Assay. Paired-end sequencing for 150-bp reads was performed on an Illumina MiSeq platform (v2) according to the manufacturer’s specifications, with the addition of 15% PhiX. Paired-end reads were stitched together (∼97-bp overlap) and processed with pick_closed_reference_otus.py pipeline in QIIME v1.6.0 (37). Taxonomy was assigned using the Greengenes reference set (vGG_13_5). A mean sequence depth of 65,130 reads per sample was obtained. OTUs with fewer than two reads in fewer than two samples and OTUs with fewer than five reads across all samples were excluded from downstream analysis.
Microbial diversity was evaluated within samples (alpha-diversity) and between samples (beta-diversity) with QIIME. To account for variations in sequencing depth, OTU tables were rarefied to 40,000 sequences per sample with 10 iterations for alpha-diversity and a single iteration for beta-diversity. Alpha-diversity was analyzed using Chao1 (richness) and Faith’s phylogenetic diversity (PD) estimates. Beta-diversity was analyzed using unweighted UniFrac distances to compare samples based on presence/absence of microbial community members. Samples were also hierarchically clustered based on intersample UniFrac distances using UPGMA. Sequence reads have been deposited in the National Center for Biotechnology Information’s Sequence Read Archive (BioProject ID PRJNA351873).
In Vivo Competition.
LF82 parental strain and LF82 ∆qseC were transformed with plasmid pNG162 harboring a SpcR cassette (56). SpcR mutants were selected on LB agar containing 100 μg/mL Spc and grown statically in 4–5 mL of LB overnight at 37 °C. Cultures were harvested by centrifugation at 4,000 × g for 10 min at 4 °C. Bacterial pellets were resuspended in LB broth and stored at −80 °C. The CFU/mL of each strain was confirmed before orally gavaging mice with 1:1 cultures of 109 CFU of LF82 pNG162:∆qseC::kan or ∆qseC/pNG162:∆qseC::kan. Stool was collected on pid 1–5 and 14, homogenized, serially diluted in dPBS, and plated on MacConkey agar (100 μg/mL Amp, 20 μg/mL Eryth) supplemented with 100 μg/mL Spc or 50 μg/mL Kan to differentiate the CFU of each LF82 strain. CFU/g of stool was normalized by sample wet weight.
DSS Treatment of ex-ASF Mice.
DSS (Affymetrix) [3% (wt/vol)] was added to drinking water on experiment days 0–5. Body weight and body condition were measured frequently to monitor disease activity. Stool was collected on day 7, homogenized, serially diluted in dPBS, and plated on MacConkey agar (100 μg/mL Amp, 20 μg/mL Eryth) to enumerate LF82. CFU/g stool was normalized by sample wet weight.
LED209 Interventions in Preclinical Colitis Models.
LED209 [N-phenyl-4-(3-phenylthioureido) benzenesulfonamide] (Cayman Chemical) was dissolved in vehicle containing 70% sodium bicarbonate pH 9, 23% polyethylene glycol (PEG), 5% dimethyl sulfoxide (DMSO), and 2% Tween-80 (all Sigma-Aldrich) (21). SPF BALB/cByJ T-bet−/−Rag2−/− and Il10−/− mice underwent a 28-d intervention from postnatal day (P) 42 ± 4 until P70 ± 4. SPF BALB/cByJ mice underwent a 10-d intervention from P34 ± 2 until P44 ± 2, with 3% (wt/vol) DSS (Affymetrix) added to drinking water on experimental days 2–7. Body weight and body condition were measured frequently to monitor disease activity.
Histology.
After sacrifice, tissues were resected, fixed in 4% paraformaldehyde (Sigma-Aldrich), and embedded in paraffin. Sections were H&E-stained and evaluated in a blinded fashion for epithelial hyperplasia (0–3), epithelial injury (0–3), polymorphonuclear infiltration (0–3), and mononuclear infiltration (0–3). These indices were summed to generate the histological colitis score (50).
qPCR Analysis of Enterobacteriaceae in Stool.
Preintervention and postintervention stool from preclinical colitis models was homogenized in RNAlater (Ambion), held at 4 °C overnight, and stored at −80 °C before processing. Nucleic acids were extracted using phenol-chloroform bead-beating, followed by DNA/RNA separation and purification using the AllPrep DNA/RNA Mini Kit (Qiagen). DNA was quantified using a NanoPhotometer Pearl spectrophotometer (Denville) and stored at −20 °C. In vivo DSS treatment can completely inhibit activity of polymerases and affect PCR amplification of extracted nucleic acids from exposed tissues (57).
To remove residual DSS, postintervention stool DNA from DSS-exposed mice underwent extraction using a lithium chloride (LiCl)-based method (57). In brief, DNA was incubated with 0.1 vol of 8 M LiCl (Cellgro) diluted in RNase-free water (Ambion) at −20 °C for 30 min and then centrifuged at 14,000 × g for 30 min at 4 °C. DNA pellets were resuspended in 200 μL of water. The 30-min incubation with LiCl, centrifugation, and pellet resuspension steps were repeated. DNA was precipitated at −20 °C for 30 min in 200 μL of 3 M sodium acetate, pH 5.2 (Ambion) and 400 μL of 100% ethanol. DNA was centrifuged for 30 min at 4 °C. DNA pellets were washed with 500 μL of 70% ethanol, centrifuged for 10 min at 4 °C, resuspended in water, and stored at −20 °C.
Quantification of Enterobacteriaceae involved 23S rRNA-targeted primers En_Isu3-For and En_Isu3-Rev with an annealing temperature of 60 °C (50), using the KAPA SYBR FAST Universal Kit (Kapa Biosystems). Quantification of total bacteria involved 16S rRNA-targeted primers and a probe with an annealing temperature of 60 °C (58), using the KAPA PROBE FAST ROX Low qPCR Kit (Kapa Biosystems). Each reaction contained 15 ng of extracted DNA and 200 nM of each primer or probe, and all reactions were performed in duplicate using a Stratagene Mx3005P qPCR system (Agilent). Fold change was analyzed by the 2-ΔΔCt method:
ΔΔCt = (Ct,Entero 23S rRNA – Ct,Total 16S rRNA)postintervention – (Ct,Entero 23S rRNA – Ct,Total 16S rRNA)baseline.
Error was calculated as fractional SD = (SD)/(mean value). Primers are listed in Table S2.
Luminal CA Measurements by HPLC.
Stool (preintervention and postintervention) and cecal content (on sacrifice) were collected and immediately flash-frozen in liquid nitrogen. Samples from each treatment group were weighed, combined into >100-mg pools, and stored at −80 °C. Pooling was required to reach assays limit of detection. CAs were extracted and measured by postcolumn HPLC (HLC-8030; Tosoh) using diphenylethylenediamine as a fluorogenic reagent (28). In brief, samples were homogenized by vigorous pipetting and vortexing in 1 mL of 0.01 M PBS. Supernatants were collected by centrifugation at 13,000 × g for 15 min at 4 °C and then mixed with 1 mL of 0.2 M perchloric acid (Sigma-Aldrich) for deproteinization. Solutions were centrifuged at 13,000 × g for 15 min. Deproteinized supernatants were processed for CA analysis by HPLC. Quantities were weight-corrected to ng/g of stool based on initial pooled sample wet weights.
In Vitro LED209 Stimulation and RT-qPCR for Cytokine Gene Expression.
MLNs and spleens were harvested from mice. To enrich for MLN-derived CD11c+ dendritic cells, MLNs (four to seven per mouse) were removed, placed in 10 mL of MACS buffer (0.5% BSA, 2 mM EDTA in dPBS), crushed through a 40-μm cell strainer, and centrifuged at 400 × g for 5 min at 4 °C. Cell pellets were resuspended in MACS buffer and Fc-blocked with anti-mouse CD16/32 (BioLegend, clone 93). CD11c+ cells were positively selected on LS columns using mouse CD11c microbeads (Miltenyi Biotec). CD11c+ cells were resuspended in cell buffer (10% heat-inactivated FBS, 1% penicillin/streptomycin, 1 mM sodium pyruvate, 50 μM 2-mercaptoethanol in RPMI 1640 with l-glutamine) and plated at 2.5 × 105 cells/well.
To enrich for splenic CD4+ cells, spleens were removed, placed in 10 mL of MACS buffer, crushed through a 40-μm cell strainer, and centrifuged at 400 × g for 5 min at 4 °C. Red blood cells were lysed using an ammonium chloride solution (STEMCELL Technologies). CD4+ cells were negatively selected on LD columns using a mouse CD4+ T-Cell Isolation Kit (Miltenyi Biotec). CD4+ cells were resuspended in cell buffer and plated at 1 × 106 cells/well. Cells were incubated with vehicle (DMSO; Sigma-Aldrich) or LED209 (50 pM or 50 nM; Cayman Chemical) [final DMSO concentration of 0.5% across all samples] for 6 h at 37 °C in 5% CO2. Cells were transferred to 1.5-mL tubes, washed with dPBS, and resuspended in 350–600 μL of RLT buffer containing 2-mercaptoethanol.
Total RNA was extracted using RNAeasy Micro or Mini Kits (Qiagen), with optional on-column DNase treatment. RNA was quantified using a NanoPhotometer Pearl spectrophotometer (Denville). cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad). RT-qPCR was performed using the KAPA SYBR FAST Universal qPCR kit (Kapa Biosystems). An annealing temperature of 60 °C was used across primer sets (200 nM each). All reactions were performed in duplicate. Primers were obtained and validated using MGH Primer Bank (https://pga.mgh.harvard.edu/primerbank). Genes were normalized to mouse β-actin. Fold change was analyzed by 2-ΔΔCt method, ΔΔCt = (Ct, target gene – Ct, β-actin)treated – (Ct, target gene– Ct, β-actin)control. Error was calculated as fractional SD = (SD)/(mean value). Primers are listed in Table S2.
Acknowledgments
We thank members of the W.S.G. laboratory for discussions, M. Ericsson for TEM assistance, Dr. A. Darfeuille-Michaud for providing LF82, and Drs. A. Ramer-Tait, M. Wannemuehler, and G. Phillips for providing ASF strains. This study was supported by National Institutes of Health Grants R01 CA154426 and R01 GM099531 (to W.S.G.).
Footnotes
- ↵1To whom correspondence should be addressed. Email: wgarrett{at}hsph.harvard.edu.
Author contributions: M.G.R., P.V., and W.S.G. designed research; M.G.R., S.L., M.M., L.W.-S., and C.A.G. performed research; A.Z.R., N.S., C.H., and C.F.L. contributed new reagents/analytic tools; M.G.R., P.V., K. Yasuda, Y.A., K. Yoshihara, J.N.G., and W.S.G. analyzed data; and M.G.R. and W.S.G. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: Sequences have been deposited in the National Center for Biotechnology Information’s Sequence Read Archive (BioProject ID: PRJNA351873).
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1612836114/-/DCSupplemental.
Freely available online through the PNAS open access option.
References
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵.
- Agus A,
- Massier S,
- Darfeuille-Michaud A,
- Billard E,
- Barnich N
- ↵
- ↵
- ↵.
- Boudeau J,
- Glasser AL,
- Masseret E,
- Joly B,
- Darfeuille-Michaud A
- ↵.
- Glasser AL, et al.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵.
- Clarke MB,
- Hughes DT,
- Zhu C,
- Boedeker EC,
- Sperandio V
- ↵.
- Rasko DA, et al.
- ↵
- ↵
- ↵
- ↵.
- Moreira CG, et al.
- ↵
- ↵
- ↵.
- Asano Y, et al.
- ↵
- ↵.
- Sandrini S,
- Aldriwesh M,
- Alruways M,
- Freestone P
- ↵.
- Kendall MM,
- Sperandio V
- ↵
- ↵.
- Chassaing B,
- Koren O,
- Carvalho FA,
- Ley RE,
- Gewirtz AT
- ↵
- ↵.
- Wymore Brand M, et al.
- ↵
- ↵
- ↵
- ↵.
- Winter SE,
- Lopez CA,
- Bäumler AJ
- ↵
- ↵
- ↵.
- Sperandio V,
- Torres AG,
- Jarvis B,
- Nataro JP,
- Kaper JB
- ↵.
- Reading NC, et al.
- ↵.
- Carvalho FA, et al.
- ↵
- ↵
- ↵.
- Datsenko KA,
- Wanner BL
- ↵
- ↵
- ↵.
- Veiga P, et al.
- ↵.
- Kuhlman TE,
- Cox EC
- ↵.
- Barnich N,
- Bringer M-A,
- Claret L,
- Darfeuille-Michaud A
- ↵.
- Reiss DJ,
- Mobley HLT
- ↵
- ↵
- ↵
- ↵
- ↵
Citation Manager Formats
Sign up for Article Alerts
Article Classifications
- Biological Sciences
- Microbiology