Navigating contradictions in enteric chemotactic stimuli

Abstract

Previously, we showed Enterobacteriaceae use chemotaxis and the chemoreceptor Tsr for attraction to blood serum (Glenn et al., 2024). Here, we investigated the complementary role of Tsr in mediating chemorepulsion, a behaviour by which motile bacteria avoid deleterious compounds to locate permissive niches. In the gut, indole is a bacteriostatic compound produced by microbiota and thought to act as a chemorepellent for invading pathogens, thereby protecting the host against infection. The principal reservoir of intestinal indole is fecal matter, a complex biological material that contains both attractant and repellent stimuli. Whether indole in its natural context is sufficient for pathogen chemorepulsion or host protection has remained unknown. Using an intestinal explant system, we show that pure indole suppresses an infection advantage mediated through chemotaxis for the enteric pathogen Salmonella Typhimurium, but this effect is abolished in the presence of other fecal chemoeffectors, including the chemoattractant L-Serine (L-Ser), dependent on the chemoreceptor Tsr. Live imaging reveals that although S. Typhimurium is repelled by pure indole, the pathogen is actually strongly attracted to human fecal matter despite its high indole content, and this response is mediated by Tsr, which simultaneously senses both stimuli. Fecal attraction is conserved across diverse Enterobacteriaceae species that harbor Tsr orthologues, including Escherichia coli, Citrobacter koseri, Enterobacter cloacae, and clinical isolates of non-typhoidal Salmonella. In a defined system of fecal chemoeffectors, L-Ser and other fecal chemoattractants override indole chemorepulsion, but the magnitude of bacterial chemoattraction is controlled by indole levels. Together, these findings suggest chemorepulsion elicited by indole is not protective against enteric infection and actually benefits pathogens by helping them locate niches with lower competitor density. Our study highlights the limitations of applying single-effector studies in predicting bacterial behavior in natural environments, where chemotaxis is shaped by the integration of multiple, often opposing, chemical signals.

Keywords: Citrobacter koseri; E. coli; Enterobacter cloacae; Escherichia coli; Salmonella enterica Typhimurium; infectious disease; microbiology; microbiome; molecular biophysics; structural biology.

Figure 1.. Chemotaxis-mediated infection advantages in the presence of fecal effectors.

(A) Overview of the role of Tsr in coordinating responses to conflicting stimuli. (B) Experimental design of colonic explant infections. See Materials and methods for experimental details such as tissue dimensions. (C) Conceptual model of the explant infection system. The effectors from the treated tissue (gray) diffuse into the surrounding buffer solution providing a gradient. Note that the bacteria are not immersed in the effector solution and experience a local concentration during infection much lower than the effector pretreatment. Quantifications of tissue-associated bacteria reflect the ability of chemotaxis to provide an advantage (black arrow) in accessing the intestinal mucosa (reddish brown). (D) Serine (presumed to be nearly 100% l-Ser, see Materials and methods) and indole content of liquid human fecal treatments, as measured by mass spectrometry. (E–I) Competitive indices (CIs) of colony-forming units (CFUs) recovered from co-infected swine explant tissue, either from the total homogenate (open box and whiskers plots), or from tissue washed with gentamicin to kill extracellular and attached cells, which we refer to as the ‘invaded’ intracellular population (checkered box and whisker plots), as indicated. Each data point represents a single experiment of a section of tissue infected with bacteria, normalized by tissue weight, and the CI of CFUs recovered from that tissue (n = 7–10). Boxes show median values (line) and upper and lower quartiles, and whiskers show max and min values. Effect size (Cohen’s d) and statistical significance are noted for each experiment in relation to competitive advantage, that is deviation from a CI of 1 (not significant, ns; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). See also Figure 1 - figure supplement 1 for competition between wildtype (WT) and an invasion-inhibited mutant invA, and Figure 1 - figure supplement 2 for disaggregated CFU enumerations for each experimental group prior to CI calculation. Source data 1 contains all numerical CFU measurements.

Figure 1—figure supplement 1.. Competition of wildtype (WT) and invA in explant infections.

(A) Competitive indices (CIs) of colony-forming units (CFUs) recovered from co-infected swine explant tissue, either from the total homogenate (total, open box and whiskers plots), or from tissue washed with gentamicin to kill extracellular and attached cells (invaded, checkered box and whisker plots), as indicated. Each data point represents a single experiment and the CI of CFUs recovered from that tissue (n = 9–10). Boxes show median values (line) and upper and lower quartiles, and whiskers show max and min values. Effect size (Cohen’s d) and statistical significance are noted for each experiment in relation to competitive advantage, that is deviation from a CI of 1 (not significant, ns; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). (B) Disaggregated CFU enumerations for each experimental group prior to CI calculation.

Figure 1—figure supplement 2.. Colony-forming units (CFUs) recovered from swine colonic explant infections.

A-J. Shown are disaggregated CFU enumerations for each co-infection experiment with S. Typhimurium IR715 wildtype (WT) and mutant strains following tissue treatments, as indicated. CFUs extracted either using the entire tissue homogenate (total) or extracted from tissue treated with a gentamicin wash to kill exterior non-invaded cells (invaded) are noted separately (see Materials and methods). Each data point represents CFUs recovered from a single explant experiment (n = 8–10). Box and whisker plots represent the sample median (line), the edges are the upper and lower quartiles, and whiskers extend to the max and min values. The limit of detection was 1 × 10⁵ CFUs, and experiments at or below this limit are not included here. See Source data 1Data S1 for numeric values.

Figure 2.. Salmonella Typhimurium exhibits attraction toward indole-rich liquid human fecal material.

(A) Diffusion modeling showing calculated local concentrations in chemosensory injection rig assay (CIRA) experiments with liquid human fecal material based on distance from the central injection source. (B) Max projections of representative S. Typhimurium IR715 responses to a central source of injected liquid human fecal material. (C–E) Bacterial population density over time in response to fecal treatment. The initial uniform population density in these plots is indicated with the blue line (time 0), and the final mean distributions with the red line (time 280 s), with the mean distributions between these displayed as a blue-to-red spectrum at 10-s intervals. (F–G) Temporal analyses of area under the curve (AUC) or relative number of bacteria within 150 µm of the source. Effect size (Cohen’s d) comparing responses of wildtype (WT) and tsr attraction at 120 s post-treatment is indicated. Data were collected at 30°C. Data are means and error bars are standard error of the mean (SEM, n = 3–5). See also Figure 2—figure supplement 1, Video 1, Supplementary file 1.

Figure 2—figure supplement 1.. Chemosensory injection rig assay (CIRA) design, diffusion modeling, and responses to l-Ser or indole.

(A) CIRA experimental design. (B) CIRA microgradient diffusion model, simulated with a source of 1.13 mM A488 dye after 30 s of injection. Image rendered on a pink (100% source concentration)–blue–black (0%) color intensity scale, based on previous work (Glenn et al., 2024). (C) CIRA quantification overview, with blue lines corresponding to chemoattraction, null chemotactic response corresponding to gray dashed lines, and chemorepulsion corresponding to brown lines. (D–G) Microgradient models of the local effector concentration for treatments with 5 mM l-Ser or indole, respectively, at Δ300 s. Color gradient corresponds to gradually decreasing effector concentrations from yellow to blue. (H, I) Representative 10 s max projections of S. Typhimurium IR715 response to sources of l-Ser (strong chemoattraction) or indole (strong chemorepulsion) using CIRA. The glass microcapillary that injects the treatment solution is centered within the field of view. Note that cells contain a plasmid expressing mPlum (see Materials and methods), and fluorescence data are collected in the far-red channel, so the glass microcapillary is poorly visible.

Figure 3.. Non-typhoidal Salmonella exhibit fecal attraction.

(A–E) Dual-channel imaging of chemotactic responses to solubilized human feces by wildtype (WT) S. Typhimurium IR715 (pink) and isogenic mutants or clinical isolate strains (green), as indicated. Max projections are shown at time 295–300 s post-treatment. Data were collected at 37°C. Data are means and error bars are standard error of the mean (SEM, n = 3–5). See also Videos 2 and 3.

Figure 4.. Representative Enterobacteriaceae exhibit fecal attraction.

Max projections are shown from chemosensory injection rig assay (CIRA) experiments over 5 s before fecal treatment and after 5 min of treatment, as well as quantifications of bacteria within 500 µm of the treatment source at these same time points for E. coli MG1655 (A, B, GFP-reporter), E. coli NCTC 9001 (C, D, phase), and C. koseri CDC 4225-83 (E, F, phase). Data were collected at 37°C. Data are means and error bars are standard error of the mean (SEM, n = 3–5) with statistical significances denoted (not significant, ns; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). See also Videos 4–6.

Figure 4—figure supplement 1.. Enterobacter cloacae exhibits fecal attraction.

(A) Shown are max projections over 5 s from a chemosensory injection rig assay (CIRA) experiment from time −5 s before fecal treatment and after 5 min of treatment with liquid human feces. Data were collected using phase-contrast.

Figure 5.. Chemotactic responses to defined fecal effector mixtures.

Chemosensory injection rig assay (CIRA) experiments with S. Typhimurium IR715 were performed with different combinations of fecal effectors (n = 3–5). Max projections are shown from experiments over 5 s before fecal treatment and after 5 min of treatment as well as quantifications of bacteria within 500 µm of the treatment source at these same time points. Data are means and error bars are standard error of the mean (SEM, n = 3–5), with statistical significance denoted (not significant, ns; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). To achieve the greatest degree of sensitivity to differences in responses, experiments were performed using the same culture on the same day. The complete fecal effector mixture consists of indole (862 µM), l-Ser (338 µM), d-glucose (970 µM), d-galactose (78 µM), ribose (28.6 µM), and L-Asp (13 µM), modified to include or exclude certain effectors as indicated. See also Videos 7–10. Data were collected at 30°C.

Figure 6.. Tsr mediates indole chemorepulsion in S. Typhimurium.

(A) Representative max projections of responses at 295–300 s of indole treatment. (B, C) Quantification of chemorepulsion as a function of indole concentration (n = 3–5). (D–F) Comparison of wildtype (WT) and tsr mutant responses to l-Ser or indole. (E) Quantification of the relative number of cells in the field of view over time following treatment with 5 mM indole for a competition experiment with WT and tsr (representative image shown in F). Data were collected at 30°C. (G, H) Isothermal titration calorimetry (ITC) experiments with 50 μM S. Typhimurium Tsr ligand-binding domain (LBD) and indole, or with l-Ser in the presence of 500 μM indole. Data are means and error bars are standard error of the mean (SEM, n = 3–5). AUC indicates area under the curve.

Figure 7.. S. Typhimurium mediates distinct chemotactic responses based on the ratio of l-Ser to indole.

(A–D) Representative max projections of responses to treatments of l-Ser and indole at 295–300 s, as indicated. (E) Relative bacterial distribution in response to treatments of 500 µM l-Ser and varying amounts of indole, from panels A–D, with the mean value normalized to 100%. Data were collected at 30°C. Data are means and error bars are standard error of the mean (SEM, n = 3–5). (F) Diffusion modeling of local effector concentrations based on sources of 5 mM indole (dark brown), 500 µM l-Ser (blue), 500 µM indole (light brown), and 50 µM indole (yellow) are shown as dashed lines. The approximate local concentration of indole that elicits a transition in chemotactic behavior is highlighted in light blue. (G, H) Bacterial growth as a function of l-Ser or indole, at the time point where the untreated culture reaches A₆₀₀ of 0.5. (I, J) Bacterial growth ± pretreatment with 500 µM indole or l-Ser, and increasing concentrations of indole or l-Ser, as indicated at the time point where the untreated culture reaches A₆₀₀ of 0.5. Data are means and error bars are standard error of the mean (SEM, n = 8–24). See also Video 11.

Figure 8.. Model and summary for explant infection data.

Based on analyses in this study, we provide this summary of the role of chemotaxis in mediating infection advantages for different tissue pretreatments. The strength of chemotactic advantage for transiting each chemical gradient and accessing the host tissue is indicated by the width of the solid black arrows. The baseline level of chemotactic advantage seen in buffer treatments may be from effectors emitted from the host tissue (gray gradient). Other gradients containing fecal chemoattractants show a similar level of chemotactic advantage, with the highest being for fecal treatment (purple). Only soaking the tissue in pure indole results in a chemical gradient for which chemotaxis and Tsr do not provide an infection advantage. Note that bacteria are exposed only to low concentrations of residual effectors that remain after the tissue is soaked and then transferred to 300 µl of buffer for infection; they are not immersed in the more concentrated effector solution. See also Figure 1, Figure 1—figure supplement 1.