Bioimage analysis of Shigella infection reveals targeting of colonic crypts

Abstract

Few studies within the pathogenic field have used advanced imaging and analytical tools to quantitatively measure pathogenicity in vivo. In this work, we present a novel approach for the investigation of host-pathogen processes based on medium-throughput 3D fluorescence imaging. The guinea pig model for Shigella flexneri invasion of the colonic mucosa was used to monitor the infectious process over time with GFP-expressing S. flexneri. A precise quantitative imaging protocol was devised to follow individual S. flexneri in a large tissue volume. An extensive dataset of confocal images was obtained and processed to extract specific quantitative information regarding the progression of S. flexneri infection in an unbiased and exhaustive manner. Specific parameters included the analysis of S. flexneri positions relative to the epithelial surface, S. flexneri density within the tissue, and volume of tissue destruction. In particular, at early time points, there was a clear association of S. flexneri with crypts, key morphological features of the colonic mucosa. Numerical simulations based on random bacterial entry confirmed the bias of experimentally measured S. flexneri for early crypt targeting. The application of a correlative light and electron microscopy technique adapted for thick tissue samples further confirmed the location of S. flexneri within colonocytes at the mouth of crypts. This quantitative imaging approach is a novel means to examine host-pathogen systems in a tailored and robust manner, inclusive of the infectious agent.

Keywords: Shigella flexneri; bioimage analysis; host–pathogen interactions; intestinal crypts; tissue microbiology.

Fig. S1.

Scheme of Swiss roll preparation. The distal 10 cm of the intact guinea pig colon was removed and prepared as a Swiss roll (for details, see Materials and Methods). This scheme shows the general progression for this preparation to have the final, 30-μm-thick transversal frozen sections. Asterisk indicates the sucrose incubation steps before OCT embedding.

Fig. 1.

A schematic representation of the acquisition and analysis workflow. (A) Representative brightfield image of a colonic Swiss roll. Diameters for each Swiss roll ranged from 1 to 2 cm. (B) A representative spinning-disk confocal MIP of wild-type S. flexneri infection foci from 4 h postchallenge. This image is a 2 × 2 tile scan. Bacteria intrinsically express GFP (green); F-actin is phalloidin-568 (red), and nuclei are DAPI-stained (blue). (Scale bar: 50 μm.) (C) Representative image of both manual and automatic annotations. Manual annotations drawn in Fiji included: epithelial surface (solid black line), muscle layer (dotted black line), mucosa (open circle), crypt mouths (stars), and crypt bases (black circles). Automatically detected S. flexneri are represented as red dots. This representation corresponds to the image in B.

Fig. S2.

Example image of both manual and automatic annotations for tissue delineation and bacterial segmentation, respectively. A representative spinning-disk confocal MIP of wild-type S. flexneri infection foci from 6 h postchallenge is a 3 × 3 tile scan. Bacteria intrinsically express GFP (green); F-actin is phalloidin-568 (red), and nuclei are DAPI stained (blue). (A) Manual annotations of the tissue morphology were drawn as ROIs in Fiji. The solid yellow line indicates the bottom of the mucosa; the dotted yellow line designates the surface of the colonic epithelium; the yellow circle marks the location of the mucosa; white circles mark positions of the base of crypts, and the white X’s indicate crypt mouths at the epithelial surface. (Scale bar: 50 µm.) (B) Automatic segmentation of individual S. flexneri was carried out using the Fiji plug-in TrackMate. Close-up of A. Magenta circles indicate individually segmented bacteria. (Scale bar: 20 µm.)

Fig. S3.

Comparison between methods of tissue preparation on gross tissue morphology. Classic transversal cuts of the colon were compared with Swiss roll preparations to ensure preservation of tissue morphology. (A) The mean colonic mucosa width was calculated for tissues prepared by both methods. No significant differences were determined (Student t test). (B) The distances between crypt base positions and the mucosa bottom were measured for tissues prepared by both methods. Again, no significant differences were determined (Student t test).

Fig. S4.

Representative images show the progression of S. flexneri infection over time. Representative confocal MIPs of guinea pig colons infected with GFP-expressing wild-type S. flexneri (M90T) at 2, 4, 6, and 8 h postchallenge. Bacteria intrinsically express GFP (green); F-actin is phalloidin-568 (red), and nuclei are DAPI stained (blue). (Scale bar: 30 µm.)

Fig. 2.

Quantification of key features of S. flexneri invasion of the colonic mucosa. (A) A ratio of luminal S. flexneri per condition was calculated by dividing the total luminal bacteria by the total bacterial count, including both luminal and tissue-associated S. flexneri. (B) The density of S. flexneri was calculated per condition, normalizing tissue-associated S. flexneri counts over total tissue volume sampled. Bacterial density is presented per cubed millimeter of tissue. (C) Violin plots of the distribution of S. flexneri depth penetration over time were determined by measuring the distance of each individual, tissue-associated S. flexneri’s distance from the delineated surface of the colonic epithelium. These violin plots are not normalized for bacterial numbers between conditions. The green dashed line and shaded green area reveal the average and SD of thickness of the colonic mucosa, ∼315 ± 69 µm, respectively. ***P < 1e-5 (Student t test).

Fig. 3.

Quantification of tissue damage after S. flexneri invasion of the colonic mucosa. Representative spinning-disk confocal MIPs of uninfected (A) and wild-type S. flexneri infection foci from 4 h postchallenge (B). Bacteria intrinsically express GFP (green); F-actin is phalloidin-568 (red), and nuclei are DAPI-stained (blue). (Scale bar: 50 μm.) (Left) The region delineated by yellow lines corresponds to the area occupied by the colonic mucosa; this region was defined using manual annotations for the epithelial surface and mucosa bottom. (Center Left) The yellow line indicates the mask created to define the surface of the tissue, defining the lumen. (Center Right) The intersection of the two previous delineations estimates the regions of tissue damaged by the ensuing S. flexneri infection; (Right) however, only regions ≥100 μm² were taken into account in the final measurement. (C) The area of damaged tissue is plotted in square microns per individual FOV per condition. Error bars represent SD. *P < 1e-1; **P < 1e-2 (Student t test).

Fig. 4.

Human ex vivo infection model reveals S. flexneri’s early crypt association. A representative confocal MIP of a human colonic explant infected with GFP-expressing wild-type S. flexneri at 1 h postchallenge. Bacteria intrinsically express GFP (green); F-actin is phalloidin-568 (red), and nuclei are DAPI stained (blue). (Scale bar: 50 µm.) Arrowheads indicate S. flexneri associated with crypts, whose luminal centers are marked with asterisks as defined by actin brush border staining.

Fig. S5.

Low-resolution image showing S. flexneri association with guinea pig colonic crypts. A confocal MIP of a whole-mount preparation of guinea pig colonic tissue infected with GFP-expressing wild-type S. flexneri at 7–8 h postchallenge. Bacteria intrinsically express GFP (green); nuclei are DAPI stained (blue). (Scale bar: 50 µm.) Asterisks indicate Shigella-associated crypts.

Fig. 5.

S. flexneri target crypts for early entry. (A) Example of the method applied for measurements of bacterial distances to crypt axes. The distance (d) from each individual S. flexneri from the closest crypt axis (solid white line) was measured. The crypt axis was defined by an orthogonal line drawn from the surface delineation (dotted yellow line) through the crypt mouth (indicated by X). The image presented is a close-up of a single slice from an acquisition of a wild-type S. flexneri-infected tissue at 2 h postchallenge. Bacteria intrinsically express GFP (green); F-actin is phalloidin-568 (red), and nuclei are DAPI stained (blue). (Scale bar: 10 µm.) (B) Radial probability distributions of bacterial distances to crypt axes reveal a bias of wild-type S. flexneri for shorter distances compared with both the T3SA-deficient strain and a simulated uniform distribution. (C) Density maps of radial probability distributions of bacterial distances to crypt axes (at center). White to blue color indicates higher to lower probabilities of bacterial presence, respectively. (Scale bar: 40 µm.)

Fig. 6.

CLEM reveals S. flexneri targeting colonocytes at crypt mouths. Arrows indicate wild-type S. flexneri. Arrowheads indicate crypt mouths. (A) A transmission electron microscopy image of an early invasion event of wild-type S. flexneri in guinea pig colon at 4 h postchallenge. (Scale bar: 10 µm.) (B) Close-up of A showing intracellular location of S. flexneri. (Scale bar: 5 µm.) (C) Another example of an early invasion event of wild-type S. flexneri in guinea pig colon at 4 h postchallenge. (Scale bar: 5 µm.) (D) Close-up of C showing intracellular location of S. flexneri. (Scale bar: 5 µm.) G, goblet cell.

Fig. S6.

Additional immunofluorescence staining supports S. flexneri targeting of colonocytes at crypt mouths. Confocal microscopy images of uninfected and 4 h postchallenge with GFP-expressing wild-type S. flexneri (M90T) colonic sections. (Scale bars: Upper, 50 μm.) Samples were labeled with vimentin (mesenchymal cells in the lamina propria), wheat germ agglutinin (goblet cells), Ezrin, galectin-3, and RSK1 (epithelial cells; yellow), and counterstained with DAPI (blue) and phalloidin-647 (red). (Lower) 5× close-up MIPs of the boxed area in the middle panels. (Scale bar: 10 µm.)

Fig. S7.

icsA S. flexneri strain confirms targeting of colonocytes at crypt mouths. Arrows indicate icsA S. flexneri strain. Arrowheads indicate crypt mouths. (A) Transmission electron microscopy image of an early invasion event of the mutant in guinea pig colon at 4 h postchallenge. (Scale bar: 10 µm.) (B) Close-up of A showing the same intracellular location as for wild-type S. flexneri. G, goblet cell. (Scale bar: 5 µm.)