Salmonella enterica serovar Typhimurium binds to HeLa cells via Fim-mediated reversible adhesion and irreversible type three secretion system 1-mediated docking

Abstract

The food-borne pathogen Salmonella enterica serovar Typhimurium invades mammalian epithelial cells. This multistep process comprises bacterial binding to the host cell, activation of the Salmonella type three secretion system 1 (T1), injection of effector proteins, triggering of host cell actin rearrangements, and S. Typhimurium entry. While the latter steps are well understood, much less is known about the initial binding step. Earlier work had implicated adhesins (but not T1) or T1 (but not other adhesins). We have studied here the Salmonella virulence factors mediating S. Typhimurium binding to HeLa cells. Using an automated microscopy assay and isogenic S. Typhimurium mutants, we analyzed the role of T1 and of several known adhesins (Fim, Pef, Lpf, Agf, and Shd) in host cell binding. In wild-type S. Typhimurium, host cell binding was mostly attributable to T1. However, in the absence of T1, Fim (but not Pef, Lpf, Agf, and Shd) also mediated HeLa cell binding. Furthermore, in the absence of T1 and type I fimbriae (Fim), we still observed residual binding, pointing toward at least one additional, unidentified binding mechanism. Dissociation experiments established that T1-mediated binding was irreversible ("docking"), while Fim-mediated binding was reversible ("reversible adhesion"). Finally, we show that noninvasive bacteria docking via T1 or adhering via Fim can efficiently invade HeLa cells, if actin rearrangements are triggered in trans by a wild-type S. Typhimurium helper strain. Our data show that binding to HeLa cells is mediated by at least two different mechanisms and that both can lead to invasion if actin rearrangements are triggered.

FIG. 1.

Binding of S. Typhimurium to cells via T1-dependent and -independent mechanisms. (A) Model of S. Typhimurium invasion into epithelial cells. In order to invade epithelial cells, S. Typhimurium first binds to the surface by an incompletely understood mechanism. It is not known whether reversible adhesion and/or stable association mechanisms might be involved. In the present study, we define “docking” as a stable association (i.e., able to be sustained considerably longer than the invasion process which happens within several minutes). In addition, a “docked” bacterium should be committed to invasion and be able to invade the bound cell without another round of dissociation and association. Bound S. Typhimurium then uses a molecular syringe (type three secretion system 1 [T1]) to inject effector proteins (shown in red) into the cytosol of the cell. Some of the effectors activate actin polymerization, which leads to visible ruffles on the surface of the cell. Ruffles are the site of Salmonella entry; the bacteria end up in a Salmonella-containing vacuole (SCV). (B) Binding of S. Typhimurium to cells. HeLa cells were incubated with the indicated S. Typhimurium strains at an MOI of 128 for 12 min before fixation and staining. S. Typhimurium strains carry plasmid pM965 for constitutive gfp expression and are shown in green. Extracellular bacteria are stained in blue due to staining by an anti-S. Typhimurium antibody; intracellular bacteria are protected by the membrane from antibody staining. The actin cytoskeleton is shown in red. Scale bar, 50 μm.

FIG. 2.

Automated analysis of S. Typhimurium binding. (A) Image analysis strategy to quantify S. Typhimurium binding. HeLa cells were incubated with the Δ4 strain (MOI = 62.5) for 10 min at 37°C. (Upper left panel) Bacteria were stained by an anti-S. Typhimurium antibody (green), nuclei were stained with DAPI (gray), and the actin cytoskeleton was stained with TRITC-phalloidin (red). The following panels demonstrate image analysis by the open source program CellProfiler and custom algorithms. (Middle left panel) Recognition of nuclei. (Lower left panel) Expansion of the nuclear area to estimate the dimensions of cells. (Upper right panel) Overlay of the cell borders over the actin stain for illustration (the actin stain was not used for the image analysis). (Middle right panel) Detection of spots in the Salmonella channel using a threshold that is calculated by comparing infected and noninfected wells. (Lower right panel) Spots and cells are superimposed. Spots are allocated to the cell with the greatest overlap. Cells containing at least one spot are scored as having bound bacteria (red outlines), and cells without S. Typhimurium are scored as noninfected (blue outlines). Only a small part of one image is shown. In a typical well, 10,000 cells are identified within the four acquired images. The automated analysis for this well yielded a ratio of infected cells of 49.6%. Scale bar, 100 μm. (B) S. Typhimurium binding to cells via T1-dependent and -independent mechanisms. HeLa cells were infected with the different S. Typhimurium strains at the indicated MOIs and incubated for 12 min. The medians and standard deviations of five independent experiments are shown. An asterisk (*) indicates a P value below 0.05 (Mann-Whitney U test) comparing the tested data point to the Δ4 strain at the same MOI; an asterisk in parentheses indicates a P value below 0.05 comparing the T1⁻ Fi⁻ strain to the T1⁻ strain. Points not marked by an asterisk of the same curve were also tested but did not yield significant P values. (C) Efficient S. Typhimurium binding requires an intact T1 system. HeLa cells were infected as in panel B using either the Δ4 strain or an isogenic strain carrying a combined knockout of sipB, sipC, and sipD (the SipBCD⁻ strain). The T1⁻ strain and a mutant lacking the three translocases as well as T1 (the T1⁻ SipBCD⁻ strain) are shown for comparison. An asterisk next to the curve indicates a P value below 0.05 (Wilcoxon signed-rank test) comparing the tested curve (median values) to the Δ4 strain.

FIG. 3.

Fundamental characteristics of the automated S. Typhimurium binding assay. (A and B) Increasing numbers of HeLa cells were seeded and infected with the indicated S. Typhimurium strain for 12 min. S. Typhimurium binding was determined as described in the text and plotted as a function of the number of nuclei detected within the respective well. The data represent the medians of three independent experiments and the standard deviations. (C) Comparison of S. Typhimurium binding to HeLa cells and “empty” areas of the same well. The spot density of an empty area of a well was calculated and expressed as a fraction of the spot density of the cell area of the same well. The data represent the medians and standard deviations of 87 to 90 wells at various MOIs from three independent experiments.

FIG. 4.

Role of type I fimbriae for S. Typhimurium binding to HeLa cells. HeLa cells were infected with the indicated S. Typhimurium strains for 12 min, and binding was analyzed. (A) Deletion mutants of type I fimbriae, shdA, long polar fimbriae, plasmid-encoded fimbriae, and thin aggregative fimbriae were analyzed in the background of the T1⁻ mutant. (B) Deletion mutants of the fim operon were generated in the background of the T1⁻ strain (fimA, fimD, and fimH), the Δ4 strain (fimD, fimA, and fimH), and the wild-type strain (fimD), respectively, and tested for HeLa cell binding. An asterisk next to the curve indicates a P value below 0.05 (Wilcoxon signed-rank test) comparing the indicated curve (median values) to the T1⁻ strain.

FIG. 5.

Inhibition of Fim-mediated binding by α-methyl-mannose. HeLa cells and bacteria were preincubated with the inhibitor α-methyl-mannose or glucose before infection of cells at an MOI of 1,000 for 10 min. The medians and standard deviations of six data points from four independent experiments are shown. An asterisk indicates binding in the presence of α-methyl-mannose that is significantly different compared to binding in the presence of glucose at the same concentration (Mann-Whitney U test). Pairs of data points not marked by an asterisk are not significantly different.

FIG. 6.

Time course of the S. Typhimurium binding to host cells. HeLa cells were infected with the indicated S. Typhimurium strains at an MOI of 62.5 for the indicated times followed by analysis of S. Typhimurium binding.

FIG. 7.

Bound S. Typhimurium can proceed with cellular invasion if ruffling is triggered in trans. (Left panel) Scheme of the helper assay. S. Typhimurium carrying plasmid pM975 for intracellular GFP production was allowed to bind, followed by washing and addition of the helper strain. The helper strain induces ruffling and internalization of both the previously bound S. Typhimurium and the helper bacteria. S. Typhimurium was allowed to express gfp during another 4 h in medium containing gentamicin for killing extracellular S. Typhimurium. Binding of the Δ4 strain and the T1⁻ strain (black symbols) was evaluated by anti-LPS staining and automated microscopy. (Middle panel) Microscopy images showing induction of gfp expression by the Δ4 strain (pM975) with the wild-type strain as the helper and no gfp expression with the Δ4 strain as a helper strain. Scale bar, 100 μm. (Right panel) Quantification of binding at an MOI of 125 for various time points (black dashed lines) and invasion (red lines) of the indicated combinations of noninvasive S. Typhimurium and helpers. The helper was tested at various concentrations, and the maximum invasion for each concentration was plotted. For curves marked by two asterisks, a P value of less than 0.001 was obtained when invasion in the presence of the wt strain as a helper was compared to the Δ4 strain or the T1⁻ strain. For these analyses, all values of the three biological replicas were used in a paired test (Wilcoxon signed-rank test).

FIG. 8.

Dissociation kinetics of T1- and Fim-mediated binding. The indicated S. Typhimurium strains were incubated with HeLa cells for 12 min. Plates were subsequently washed and incubated for 10 min at 37°C in medium containing α-methyl-mannose. Several rounds of medium exchange and incubation were performed. The appropriate wells were fixed. Finally, binding was measured as described in the text.