Fig 1.
S. flexneri regulates macropinosome formation at the invasion site.
(A) Confocal microscopy of the invasion site of WT fluorescent bacteria at 30 min infection in the presence of dextran Alexa Fluor-647. Hela, NRK and Caco-2 cells are compared. Actin rich invasion site ruffles are labeled with phalloidin. Scale bars are 10μm and 5μm (inset). Distribution of vesicle diameters at Hela cell invasion sites is presented (n = 2300 vesicles in 3 independent experiments). (B) WT compared to ΔipgD invasion at 30 min infection. Bacteria are labeled with DAPI (0.5 ϒ-corrected images presented). Scale bars are 10μm and 2μm (inset). (C) Comparison of the number of vesicles per bacterium in WT vs. ΔipgD strains. Distribution of all invasion sites analyzed (WT n = 324 sites, ΔipgD n = 256 sites in six independent experiments) and averages are presented. (D) Sequential labeling experiment. Cells were incubated with dextran Alexa Fluor-488 before infection (green) then washed and infected in the presence of dextran Alexa Fluor-647 (magenta), washed, fixed and imaged. 92% of vesicle volume at the invasion site around bacteria (red) is occupied by vesicles formed during infection. (n = 476 vesicles in 34 invasion sites). Scale bars are 10μm and 2μm (inset). Z projections are presented in all panels.
Fig 2.
C-FIB/SET reveals the BCV is a tight compartment structurally distinct from surrounding vesicles.
(A) Confocal microscopy of an early stage invasion site, labeled for actin (magenta) and DNA (cyan). Actin enrichment is observed around a bacterium at the site of interest (Inset, green box). This site is correlated with the corresponding 3D FIB/SET data allowing ultrastructural visualization of the entire invasion site, presented in three orthogonal views. Surface membrane ruffling (gold) and inner cell structures are observed in detail. The same bacterium is identified (panel YZ, green) in division at the center of the volume. See also S1 Movie. (B) Typical structural components of the S. flexneri BCV. Bacterial cytosol and membrane (blue); LPS layer (white); BCV membrane (yellow); and an actin layer (magenta), observed only prior to vacuolar rupture. The BCV is surrounded by vesicles (orange). Additional views place BCV in cellular context. See also S2 Movie. Overall 15 FIB/SET data sets of invasion sites were acquired.
Fig 3.
Macropinosome availability is correlated with the efficiency of vacuolar rupture and macropinosomes border the BCV at the onset of vacuolar rupture.
(A) Effector mutant library macropinosome formation screen. Number of macropinosome per bacteria at the invasion site was quantified for infection with WT and various effector mutant strains (n≥70 invasion sites per strain from 2 independent experiments). (B) Effector mutant library vacuolar rupture timing screen (n≥50 invasion events per strain, from 3 independent experiments. Significance of mutant vs. WT was determined using one way ANOVA.). Each data point represents a single rupture event recorded during live imaging. (C) Confocal time-lapse microscopy of cells transfected with the PI3P macropinosome marker 2XFYVE-GFP and vacuolar rupture marker galectin-3-mOrange, and infected with WT bacteria. Typical events are presented: onset of membrane ruffling (0 s), formation of macropinosomes (210 s), vacuolar rupture in close proximity to macropinosomes (240 s), ongoing macropinosome presence and progression of rupture (510 s). Images were taken every 30 s, z-projections are presented, scale bar is 5μm. See also S3 Movie. (D) Gallery of different rupture events where macropinosomes (labeled by 2XFYVE-GFP) border rupturing BCVs (labeled by galectin-3-mOrange). Scale bar is 2 μm. (E) Frequency of macropinosomes bordering (within 1 μm) the rupturing BCV from total rupture events observed (left). In 73% of vacuolar rupture events analyzed macropinosomes were found bordering the BCV (n = 30 rupture events in 3 independent experiments). Frequency of vesicles present at rupture onset based on high temporal resolution imaging (right). Macropinosome were found at least one frame before rupture in 92% of rupture events containing bordering macropinosomes (n = 36 rupture events in eight independent experiments).
Fig 4.
Rab11 is directly and dynamically recruited to macropinosomes and its activity is required for efficient vacuolar rupture.
(A) Dextran pulse chase live imaging experiment. Overview of invasion site. (B) Time series of S. flexneri invasion. Inset from A is presented. Minute 0 represents the time when dextran was washed out and imaging was initiated. The onset of vacuolar rupture is observed in frame “6”. (C) Inset from B showing Rab11 is dynamically and directly recruited to a dextran labeled macropinosome (highlighted in yellow). Max projection of three slices is presented. See also S5 Movie. (D) Rab11S25N-GFP GDP locked dominant negative (Rab11 DN) is not recruited to invasion site and causes a significant delay in vacuolar rupture timing. Cells were transfected with Rab11-GFP or Rab11 DN and galectin-3-mOrange followed by live imaging during S. flexneri invasion. Two representative single slices from movies showing the localization of Rab11 and Rab11 DN are presented (left). Comparison of vacuolar rupture timing (right). Wild-type infections of Rab11-GFP transfected (Rab11/wt) and Rab11 DN (Rab11 DN/wt) cells are presented. ΔipgD infection of Rab11-GFP transfected cells (Rab11/IpgD) is used as a control. Data obtained from three independent experiments with 946 overall rupture events measured. See also S6 Movie, S7 Movie. Scale bars in all panels are 10 μm.
Fig 5.
Macropinosomes come into direct contact with the bacteria containing vacuole during vacuolar rupture.
Vacuolar rupture occurring during of WT and ΔipgD infections was studied using C-FIB/SET. Confocal microscopy was used to identify bacteria (cyan) within rupturing BCVs labeled by galectin-3-mOrange (red). (A) 3D ultrastructural data of WT reveals several macropinosomes (mp, orange) in direct contact (white arrow heads) with the BCV membrane (yellow), surrounding the bacteria (blue -cytosol, white- LPS). A dissociated BCV membrane is observed (A, white arrow) as well as smaller intraluminal vesicles within the macropinosome (A, black arrowhead). See also S8 Movie. (B) The ΔipgD strain exhibits less vesicles in contact with the BCV than the ΔipgD mutant strain, yet contact morphology is maintained (white arrow heads). Rupture on the opposing end to the contact point is observed (black arrow). See also S9 Movie. Overall 15 FIB/SET data sets of invasion sites were acquired (11 WT, 4 ΔipgD). For each strain, a representative data set of a site containing a rupturing BCV is presented.
Fig 6.
New model of S. flexneri early invasion into epithelial cells.
Wild-type (A) is compared with ΔipgD (B). Altered steps are indicated in red with references to relevant work. S. flexneri (blue) induces membrane ruffling resulting in macropinosome formation (pink) in conjunction with its entry into a tight BCV. Rab GTPases (dark green) are then recruited to macropinosomes. Finally macropinosomes and the BCV come into direct contact at the onset of vacuolar rupture (dashed line) and bacteria escape to the cytosol. Pre-existing endocytic vesicles (light green) are not recruited to the invasion site. ΔipgD reduces membrane ruffling, resulting in fewer macropinosomes that can then be targeted by Rab GTPases (observed as on overall reduction of Rab recruitment to the invasion site). As a result less macropinosome- BCV contacts are formed and vacuolar ruptured is delayed.