ABSTRACT
Shigella flexneri disseminates within the colonic mucosa by displaying actin-based motility in the cytosol of epithelial cells. Motile bacteria form membrane protrusions that project into adjacent cells and resolve into double-membrane vacuoles (DMVs) from which the bacteria escape, thereby achieving cell-to-cell spread. During dissemination, S. flexneri is targeted by LC3-dependent autophagy, a host cell defense mechanism against intracellular pathogens. The S. flexneri type III secretion system effector protein IcsB was initially proposed to counteract the recruitment of the LC3-dependent autophagy machinery to cytosolic bacteria. However, a recent study proposed that LC3 was recruited to bacteria in DMVs formed during cell-to-cell spread. To resolve the controversy and clarify the role of autophagy in S. flexneri infection, we tracked dissemination using live confocal microscopy and determined the spatial and temporal recruitment of LC3 to bacteria. This approach demonstrated that (i) LC3 was exclusively recruited to wild-type or icsB bacteria located in DMVs and (ii) the icsB mutant was defective in cell-to-cell spread due to failure to escape LC3-positive as well as LC3-negative DMVs. Failure of S. flexneri to escape DMVs correlated with late LC3 recruitment, suggesting that LC3 recruitment is the consequence and not the cause of DMV escape failure. Inhibition of autophagy had no positive impact on the spreading of wild-type or icsB mutant bacteria. Our results unambiguously demonstrate that IcsB is required for DMV escape during cell-to-cell spread, regardless of LC3 recruitment, and do not support the previously proposed notion that autophagy counters S. flexneri dissemination.
INTRODUCTION
Autophagy is a cellular process that is responsible for the degradation and recycling of cellular debris and damaged organelles. Ubiquitination and recognition by adaptor proteins, such as p62, TANK binding kinase 1 (TBK1), and NDP52, can selectively target cargos for autophagy and recruit the autophagy machinery (1–4). Several members of the autophagy-related (ATG) family of proteins are involved in formation of an isolation membrane around cytosolic cargo forming an autophagosome (5). One of these proteins, ATG8/LC3, is conjugated to phosphatidylethanolamine and localizes to the isolation membrane (6), where it controls its expansion around cytosolic cargo (7). This autophagosome matures along the endo-lysosomal pathway and eventually fuses with a lysosome, resulting in degradation of the cargo.
Canonical and noncanonical autophagy pathways are increasingly recognized as host cell defense mechanisms against intracellular infection. LC3-associated isolation membranes are recruited to and entrap cytosolic bacteria in a specialized form of autophagy called xenophagy (8). Alternatively, LC3 can be recruited to a pathogen-containing vacuole in noncanonical pathways that do not require all of the classical autophagy proteins (9–11). Many intracellular bacteria, including Legionella pneumophila, Salmonella enterica, and Listeria monocytogenes, are targeted by autophagy. Pathogens such as Legionella pneumophila and Salmonella enterica are recognized by the autophagy machinery while residing in vacuoles (12, 13). Cytosolic pathogens such as Listeria monocytogenes, which quickly escape from primary vacuoles formed upon invasion, are detected by autophagy either in vacuoles prior to escape or in the cytosol after escape (11).
Shigella flexneri is also targeted by autophagy during intracellular infection. S. flexneri uses its type III secretion system (T3SS) to induce its own uptake into colonic epithelial cells (14) and escape from the primary vacuole to gain access to the host cell cytosol. In the cytosol, Shigella expresses the autotransporter protein IcsA, which promotes polar actin polymerization and cytosolic motility (15, 16). When motile bacteria encounter the plasma membrane, they form membrane protrusions that project into adjacent cells (17). These protrusions resolve into an intermediate compartment called vacuole-like protrusions (VLPs), which leads to the formation of double-membrane vacuoles (DMVs) in adjacent cells (18, 19). The bacteria subsequently escape from DMVs to gain access to the cytosol of adjacent cells, thereby achieving cell-to-cell spread (19).
The LC3-dependent autophagy machinery was first shown to target S. flexneri in the cytosol as a result of the surface expression of IcsA (20). IcsA was proposed to bind ATG5 and thus trigger autophagic recognition of cytosolic bacteria. The S. flexneri protein IcsB, which was first identified as a type III secretion system effector protein required for intracellular spread (21, 22), was proposed to prevent recognition of cytosolic bacteria by binding to IcsA and masking it from ATG5 (20). More recently, it was proposed that LC3 is recruited not to cytosolic bacteria but to bacteria located in vacuoles (23). In that study, the authors capitalized on a transcriptional reporter of the activity of the T3SS and the secreted translocator IpaB for identifying bacteria in membrane-bound compartments during intracellular infection. By combining these markers with LC3 labeling, the authors concluded that LC3 was recruited to bacteria located in the double-membrane vacuoles formed during S. flexneri spread from cell to cell. This approach also confirmed that the icsB mutant was more frequently associated with LC3, suggesting that IcsB is required for escaping LC3-positive vacuoles. Although the study clearly demonstrated that LC3 was recruited to actively secreting S. flexneri, the approach did not allow for the unambiguous identification of features of cell-to-cell spread, including protrusions, VLPs, and vacuoles.
In this study, we have used time-lapse confocal microscopy in cells expressing plasma membrane-targeted yellow fluorescent protein (YFP) to track the dissemination of individual cyan fluorescent protein (CFP)-expressing bacteria and the simultaneous recruitment of mCherry-LC3. We provide an unambiguous demonstration that LC3 is recruited to S. flexneri when located in the double-membrane vacuoles that derive from membrane protrusions formed during dissemination. Importantly, our approach uncovered that IcsB contributes to DMV escape, regardless of LC3 recruitment. Additionally, we revealed a positive correlation between the time until LC3 recruitment and the time until vacuole escape. Finally, we demonstrated that inhibiting autophagy did not rescue the spreading defect in cells infected with the icsB mutant. Moreover, inhibition of autophagy impaired cell-to-cell spread of wild-type (WT) bacteria. Collectively, these results do not support the notion that the LC3-dependent autophagy machinery acts as host cell defense mechanism against S. flexneri during cell-to-cell spread.
RESULTS
IcsB promotes S. flexneri spread from cell to cell.To study the role of IcsB in S. flexneri dissemination, we generated a mutant lacking icsB. To determine if the icsB mutant displayed an invasion and/or primary vacuole escape defect, we performed a gentamicin protection assay comparing the isogenic wild-type 2457T strain and the icsB mutant. The icsB mutant showed no significant difference compared to the wild type in CFU 1 h postinvasion (see Fig. S1A in the supplemental material) or in numbers of infection foci 8 h postinvasion (Fig. S1B), indicating that the icsB mutant was as invasive as the wild-type strain. We have previously established the intestinal HT-29 cell line as a model system for studying the dissemination of S. flexneri in epithelial cells (24, 25), which leads to the formation of large infection foci 16 h postinfection (Fig. 1, WT). We compared the area of the foci formed in cells infected with wild-type bacteria (Fig. 1, WT) and the icsB mutant (Fig. 1, icsB). While the wild-type strain formed large foci, the icsB mutant formed significantly smaller foci. Using computer-assisted image analysis, we measured the area of individual infection foci. The average focus size formed in cells infected with the icsB mutant was significantly smaller than with the wild type, revealing a 65% decrease in spreading (Fig. 1, WT versus icsB). The complemented strain (icsB/picsB) formed infection foci with average size similar to wild type (Fig. 1, icsB/picsB, +Arabinose), indicating that the spreading defect of the icsB mutant strain was rescued by expression of IcsB. These results indicate that the expression of the T3SS effector protein IcsB is not required for invasion and primary vacuole escape but promotes efficient spread from cell to cell in the intestinal epithelium.
The icsB mutant displays a spreading defect. Shown are representative images of HT-29 cells infected with CFP-expressing WT, icsB, or icsB/picsB S. flexneri. Scale bar, 200 μm. The graph shows quantification of area of infection foci for WT, icsB, and icsB/picsB bacteria. The expression of icsB is under the control of an arabinose-inducible promoter. Uninduced (−) and 1% arabinose (+)-induced conditions are shown. Arabinose was added 1 h postinfection. Fifty foci per strain/condition were measured in each experiment. Error bars represent SEMs from three independent experiments. ****, P < 0.0001, one-way analysis of variance (ANOVA) with multiple comparisons.
IcsB promotes cell-to-cell spread through escape from double-membrane vacuoles.To determine the stage of dissemination at which the icsB mutant was defective, we used time-lapse confocal microscopy to track the progression and timing of cell-to-cell spread (18, 26). We infected HT-29 cells expressing plasma membrane-targeted YFP with wild-type or icsB mutant strains expressing isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible CFP to track motile, cytosolic bacteria forming membrane protrusions that projected into neighboring cells, which was followed by the collapse of the protrusion neck into vacuole-like protrusions (VLPs) (18). VLPs then resolved into double-membrane vacuoles (DMVs) from which the bacteria subsequently escaped (Fig. 2A, DMV Escape Success). In some instances, the bacteria failed to escape from DMVs and remained trapped until the end of the movie (Fig. 2B, DMV Escape Failure). In Fig. 3, we show the representative tracking of 40 wild-type (Fig. 3A, Wild Type) and 40 icsB (Fig. 3A, icsB) bacteria from a single experiment. Dark blue at the beginning of each track indicates the bacteria in the primary infected cell cytosol. Light blue indicates the bacteria in protrusions, purple indicates the bacteria in a VLP, yellow refers to bacteria in a DMV, and green indicates that the YFP membrane was no longer visible and that the bacteria regained actin-based motility, showing that they were free in the cytosol of the adjacent cell. Tracking results showed that 76% of wild-type bacteria that formed protrusions were successful in spreading to the adjacent cell (Fig. 3B, Wild Type, green, DMV escape). Twenty-four percent of the bacteria that formed protrusions failed to spread to the adjacent cell. Thirteen percent failed in protrusions that ultimately collapsed, bringing the pathogen back to the primary infected cell (Fig. 3B, Wild Type, dark blue, protrusion failure), and 11% failed to escape from DMVs (Fig. 3B, Wild Type, yellow, DMV escape failure).
Tracking individual bacteria using time-lapse microscopy of HT-29 cells expressing plasma membrane-targeted YFP and infected with CFP-expressing S. flexneri. (A and B) Representative images showing the progression of a single bacterium over time. For each panel, the top image shows a merged image of bacteria (red, pseudo-color) and plasma membrane (green, pseudo-color) and the bottom images show plasma membrane only. The schematic at the top of each panel shows stage and corresponding color code of spread progression as used in Fig. 3. (A) Successful progression of a bacterium from the primary cell cytoplasm into a membrane protrusion (2 min, Protrusion) that resolves into a vacuole-like protrusion (14 min, VLP) and then a double-membrane vacuole (46 min, DMV) from which the bacterium escapes and gains access to the cytoplasm of the adjacent cell (70 min, Free bacteria). (B) Unsuccessful progression of a bacterium from the primary cell cytoplasm into a membrane protrusion (2 min, Protrusion) that resolves into a vacuole-like protrusion (6 min, VLP) and then a double-membrane vacuole (12 min, DMV) and remains trapped in the DMV (284 min, DMV). Green, plasma membrane; red, Shigella. Scale bar, 2 μm.
The icsB mutant displays a DMV escape defect. (A) Representative tracking analysis of WT (top) and icsB (bottom) bacteria in HT-29 cells. Each line represents tracking of one bacterium for 180 min. Forty bacteria were tracked in 8 independent infection foci per strain. (B) Graphs showing the relative proportions of fates of tracked bacteria in panel A for the wild type and icsB mutant.
Tracking analysis revealed that in contrast with wild-type bacteria, 37% of the icsB bacteria that formed protrusions successfully spread to the adjacent cell (Fig. 3B, icsB, green, DMV escape). Similar to the case with the wild type, a small percentage of bacteria failed in protrusions that collapsed, bringing the pathogen back to the primary infected cell (Fig. 3B, icsB, dark blue, protrusion failure). In contrast with wild-type bacteria, of which only a minority of spreading bacteria failed to escape vacuoles (11%), the majority of the icsB bacteria (53%) failed to escape DMVs and remained trapped in a membrane-bound compartment (Fig. 3B, icsB, yellow, DMV escape failure). A total of 147 wild-type and 147 icsB bacteria were tracked for at least 3 h in three independent experiments. We found similar values in our three independent experiments (Fig. S2). Taken together, these results indicate that the T3SS effector protein IcsB promotes cell-to-cell spread through escape from double-membrane vacuoles.
The autophagy marker LC3 is exclusively recruited to double-membrane vacuoles.IcsB has been previously suggested to counteract the recruitment of the LC3 marker to cytosolic bacteria (20). However, a recent report suggested instead that IcsB was required for escaping LC3-positive DMVs (23). In order to clarify the spatial and temporal recruitment of LC3 to S. flexneri during cell-to-cell spread, we used HT-29 cell lines that express mCherry-LC3 in combination with plasma membrane-targeted YFP. The distinction between LC3-negative (LC3−) bacteria (Fig. 4A, LC3− DMV) and LC3-positive (LC3+) bacteria (Fig. 4A, LC3+ DMV) was made based on the recruitment of mCherry-LC3 around CFP-expressing bacteria (Fig. 4A, LC3+ DMV, mCherry-LC3). We first used the approach to quantify the ratio of LC3+ bacteria per total number of bacteria in infected cells. As in previous studies (20, 23), we observed an increase in LC3+ bacteria in cells infected with icsB bacteria (Fig. 4B). We next used time-lapse confocal microscopy to track the progression of individual bacteria and simultaneously the recruitment of mCherry-LC3. Figure 5 shows the representative tracking of 40 wild-type (Fig. 5A, Wild Type) and 40 icsB (Fig. 5A, icsB) bacteria from a single experiment. Dark blue at the beginning of each track represents the location of the bacteria in the primary infected cell cytosol. Light blue indicates the bacteria in protrusions, purple indicates the bacteria in a VLP, and yellow shading refers to bacteria in a DMV. Red indicates bacteria in a DMV to which LC3 has been recruited. Green indicates that the YFP membrane was no longer visible and that the bacteria regained actin-based motility, showing that they were free in the cytosol of the adjacent cell. A total of 147 wild-type and 147 icsB bacteria were tracked for at least 3 h in three independent experiments. We found similar values in our three independent experiments (Fig. S3). In all cases of LC3 recruitment to CFP-positive bacteria (n = 57), back-tracking of the bacteria revealed that LC3 was recruited to plasma membrane YFP-positive DMVs that derived from plasma membrane YFP-positive protrusions formed during cell-to-cell spread. LC3 was never recruited to plasma membrane YFP-positive protrusions or VLPs or to plasma membrane YFP-negative (cytosolic) bacteria. These experiments provide an unambiguous demonstration that the LC3 autophagy marker is recruited to S. flexneri only when bacteria are located in plasma membrane vacuoles that derive from the membrane protrusions formed during cell-to-cell spread, i.e., DMVs.
Tracking LC3 recruitment using HT-29 cells expressing LC3-mCherry and plasma membrane-targeted YFP and infected with CFP-expressing S. flexneri. (A) Representative images of cells expressing LC3-mCherry and plasma membrane-targeted YFP infected with CFP-expressing S. flexneri. Images on the left show an example where LC3-mCherry is not recruited to DMV (LC3− DMV, yellow bar), and right images show an example where LC3-mCherry is recruited to DMV (LC3+ DMV, red bar). Scale bar, 2 μm. (B) Graph showing percentages of wild-type and icsB bacteria associated with LC3 4 h postinfection. Error bars represent SEMs from three independent experiments. Total bacteria from 8 infection foci per strain were counted in each experiment. ****, P < 0.0001, unpaired t test.
LC3 is recruited equally to double-membrane vacuoles containing wild-type and icsB mutant bacteria. (A) Representative tracking analysis of dissemination and LC3 recruitment for wild-type and icsB bacteria. Each line represents tracking of one bacterium for 180 min. (B) Graphs showing the relative proportion of fates of tracked bacteria in panel A for wild-type and icsB bacteria. (C) Graph depicting the percentage of DMVs that recruited LC3 for wild-type and icsB bacteria. (D) Graph depicting the percentage of DMVs that recruited LC3 for wild-type and icsB bacteria grouped according to DMV escape or DMV escape failure. (E) Graph depicting the percentage of DMV escape failure among wild-type and icsB bacteria grouped according to LC3+ or LC3− DMVs. For panels C to E, at least 40 bacteria were tracked per experiment. Error bars indicate SEMs from three independent experiments. n.s., not significant. *, P < 0.05; ****, P < 0.0001 (unpaired t test or one-way ANOVA with multiple comparisons).
LC3 is recruited equally to the double-membrane vacuoles containing wild-type bacteria and the icsB mutant.The increased numbers of LC3+ bacteria in cells infected with the icsB mutant (12, 13) (Fig. 4B) may reflect a potential role for IcsB in counteracting the recruitment of the LC3-dependent autophagy machinery to DMVs. Alternatively, wild-type bacteria and the icsB mutant could recruit LC3 equally in the course of cell-to-cell spread, but vacuole escape failure of the icsB mutant may lead to an accumulation of LC3+ vacuoles over time in cells infected with the icsB mutant and therefore increased numbers of LC3+ bacteria in infected cells. To determine whether DMVs harboring the icsB mutant recruited LC3 more frequently than with the wild type, we tracked individual DMVs that derived from protrusion during cell-to-cell spread and determined the percentage of DMVs that became LC3 positive. We found no significant difference in the percentage of DMVs that recruited LC3 in cells infected with the wild type or the icsB mutant (Fig. 5C, Wild Type versus icsB). We further examined whether comparing bacteria in groups according to their success or failure in escaping DMVs may reveal differences in LC3 recruitment. However, we found no difference in LC3 recruitment between the wild-type strain and the icsB mutant strain among bacteria that escaped DMVs (Fig. 5D, Wild Type versus icsB, Escape, green and red bars) or failed to escape DMVs (Fig. 5D, Wild Type versus icsB, Failure, yellow and red bars). Thus, LC3 is not recruited more frequently to the icsB mutant relative to the number of bacteria in DMVs, and the increased numbers of LC3+ bacteria in cells infected with the icsB mutant are a consequence of the increased numbers of DMVs due to vacuole escape failure.
The icsB mutant fails to escape from LC3+ as well as LC3− double-membrane vacuoles.Since IcsB has been previously implicated in escape from LC3+ vacuoles (23), we investigated whether there was a difference in DMV escape failure between the wild-type strain and the icsB mutant strain depending on their location in LC3− or LC3+ DMVs. As previously suggested (23), the icsB mutant strain was significantly impaired in escape from LC3+ DMVs compared to the wild-type strain (Fig. 5E, LC3+, Wild Type versus icsB). However, the icsB mutant was equivalently defective in escaping from LC3− DMVs compared to the wild-type strain (Fig. 5E, LC3−, Wild Type versus icsB). These results show that the icsB mutant is significantly impaired in escape from DMVs, regardless of LC3 recruitment.
IcsB promotes prompt escape from double-membrane vacuoles.We also analyzed a potential role for IcsB in the dissemination of bacteria that successfully escaped DMVs. We observed a modest increase in the time spent in protrusions by the icsB mutant compared to wild-type bacteria (Fig. S4A, gray bar, icsB) and no significant increase in the time spent in VLPs (Fig. S4B, gray bar, icsB). However, the time spent in DMVs before escape was significantly increased in cells infected with the icsB mutant compared to wild-type bacteria (65 min versus 40 min [Fig. 6A, Wild Type versus icsB]), showing that IcsB is required for prompt escape from DMVs. We next investigated the impact of LC3 recruitment on the time until DMV escape. The results revealed a significant delay in DMV escape of wild-type bacteria from LC3+ DMVs (Fig. 6B, LC3+, wild type, black) compared to LC3− DMVs (Fig. 6B, LC3−, wild type, black). The same trend was observed for the icsB mutant, although the difference was not statistically significant. Importantly, there was a significant increase in time until escape between the wild type and the icsB bacteria that resided in LC3− DMVs (Fig. 6B, LC3−, Wild Type versus icsB). Collectively, the results indicate that the icsB mutant displays delayed vacuole escape (Fig. 6) or fails to escape (Fig. 5), even in the absence of LC3 recruitment to DMVs. These results are therefore not in agreement with the notion that the role of IcsB is to counteract LC3 recruitment to DMVs.
IcsB promotes prompt escape from DMVs. (A) Graph showing the time spent by bacteria in DMVs before escape for the wild type and icsB mutant. (B) Graph showing the time spent by bacteria in DMVs before escape for the wild type and icsB mutant grouped according to LC3+ or LC3− status. Data from three independent experiments are shown. Error bars indicate SEMs. **, P < 0.005; ***, P < 0.001; ****, P < 0.0001 (unpaired t test or one-way ANOVA with multiple comparisons).
LC3 recruitment correlates with but is not the cause of vacuole escape failure.The recruitment of LC3 to S. flexneri has been proposed to reflect the role of the LC3-dependent machinery as a host cell defense mechanism. Accordingly, we found that LC3 recruitment correlated with escape failure as reflected by the observed increase in LC3+ DMVs among wild-type bacteria that failed to escape (Fig. 5D) and the observed increase in vacuole escape failure among wild-type bacteria that resided in LC3+ DMVs (Fig. 5E). In addition, we found that LC3 recruitment correlated with increased time spent in vacuole until escape of wild-type bacteria (Fig. 6B). To further investigate the exact role of LC3 recruitment, we determined the time until LC3 recruitment to DMV in groups of bacteria according to their success or failure to escape DMVs. Surprisingly, among the bacteria that recruited LC3 to DMVs, both wild-type and icsB bacteria that successfully escaped DMVs recruited LC3 more rapidly than bacteria that failed to escape DMVs (Fig. 7A, Escape versus Fail to Escape). Moreover, among bacteria that successfully escaped vacuoles, we found a positive correlation between the time of LC3 recruitment and the time until escape (Fig. 7B). Thus, early LC3 recruitment correlates with prompt escape, and late LC3 recruitment correlates with delayed escape or escape failure. These results strongly suggest that when LC3 is recruited to DMVs from which bacteria fail to escape, that recruitment is the consequence and not the cause of vacuole escape failure.
Positive correlation between the time until escape and the time until LC3 recruitment. (A) Graph showing average time in DMV before LC3 recruitment for wild-type and icsB bacteria grouped according to vacuole escape success or failure. Error bars indicate SEMs from three independent experiments. *, P < 0.05; **, P < 0.005 (one-way ANOVA with multiple comparisons). (B) Graph showing time until LC3 recruitment versus time until DMV escape for wild-type bacteria. Data from three independent experiments are shown. Pearson (r) = 0.597; correlation significance, P < 0.05.
LC3-dependent autophagy does not inhibit S. flexneri spread from cell to cell.To further investigate the role of the LC3-dependent autophagy machinery in S. flexneri spread from cell to cell, we inhibited steps of autophagy with bafilomycin A1 (Baf-A1), chloroquine, NH4Cl, or 3-methyladenine (3-MA) treatment. Baf-A1 interferes with autophagy by preventing the fusion of autophagosomes with lysosomes. We found no difference in the size of the foci formed by the icsB mutant in cells treated with dimethyl sulfoxide (DMSO) or Baf-A1 (Fig. 8A, icsB, DMSO versus Baf-A1), indicating that Baf-A1 does not rescue the spreading defect displayed by the icsB mutant. We noticed, however, that the foci formed by the wild-type strain in the presence of Baf-A1 were significantly smaller than those with DMSO treatment (35% decrease [Fig. 8A, WT, DMSO versus Baf-A1]). Additionally, treatment with Baf-A1 affected the size of the infection foci formed by the complemented icsB mutant (50% decrease [Fig. 8A, icsB picsB, DMSO versus Baf-A1]). We obtained similar results when cells were treated with the acidification inhibitors chloroquine (Fig. 8B) and NH4Cl (Fig. 8C) or when they were treated with the specific PI3KC3 inhibitor 3-MA (Fig. 8D). To specifically assess the role of LC3 in dissemination, we measured the effect of LC3 depletion by small interfering RNA (siRNA) on S. flexneri dissemination. We found no significant differences in the sizes of foci formed by the wild type, icsB mutant, or the complemented icsB mutant in mock-treated cells and LC3-depleted cells (Fig. S5). Taken together, our results indicate that siRNA or chemical inhibition of important steps of autophagy has no positive impact on the spreading efficiency of wild-type or icsB mutant bacteria, suggesting that the LC3-dependent autophagy machinery does not interfere with S. flexneri dissemination.
Autophagy does not inhibit S. flexneri spread from cell to cell. Quantification of areas of infection foci for wild-type, icsB, and icsB/picsB in HT-29 cells treated with 200 nM bafilomycin A1 (A), 10 μM chloroquine (B), 10 mM NH4Cl (C), or 6.7 mM 3-methyladenine (D). picsB is under the control of an arabinose-inducible promoter. Uninduced and 1% arabinose-induced conditions are shown. Arabinose and inhibitors were added 1 h postinfection. Computer-assisted image analysis was used to measure focus size. Fifty foci per experiment were measured. Error bars represent SEMs from three independent experiments. A.U., arbitrary units. **, P < 0.005; ***, P < 0.001; ****, P < 0.0001 (one-way ANOVA with multiple comparisons).
In summary and as depicted in Fig. 9, this study demonstrates that (i) IcsB is required for DMV escape, (ii) the autophagy marker LC3 is exclusively recruited to DMVs, (iii) IcsB does not counter LC3 recruitment to DMVs, (iv) the recruitment of LC3 to DMVs is not the cause but the consequence of vacuole escape failure, and (v) the LC3-dependent autophagy machinery does not restrain S. flexneri spread from cell to cell.
Model depicting the timing and frequency of LC3 recruitment and double-membrane vacuole escape for wild-type and icsB bacteria during S. flexneri spread. The timing and frequency of events are, respectively, depicted by the length and the width of the corresponding boxes. (Top) The large majority of wild-type bacteria successfully escape double-membrane vacuoles promptly (65%, within 40 min, green box). A subset of the population (12%, red box) recruits LC3 early (within 25 min). A small proportion of wild-type bacteria fail to escape double-membrane vacuoles (7%, yellow box). A large proportion of those recruit LC3 (5%, red box), and this recruitment is delayed (100+ min). (Bottom) In contrast to wild-type bacteria, only a small fraction of icsB bacteria successfully escape double-membrane vacuoles (28%, green box). These bacteria are delayed in escape (65 min) and a subset recruits LC3 (6%, red box) early (within 25 min). The majority of icsB bacteria fail to escape double-membrane vacuoles (40%, yellow box). A large proportion recruit LC3 (17%, red box), and this recruitment is delayed (85 min). Percentages represent proportions of total bacteria that progress to DMVs. The bacteria that fail in protrusions (11% for the wild type and 9% for icsB bacteria) are not depicted in the model.
DISCUSSION
The homeostatic cellular process of autophagy plays an important role in host defense against infection with numerous intracellular pathogens. In particular, recruitment of the LC3-dependent autophagy machinery to intracellular bacteria has been proposed to act as host cell defense against the intracellular human pathogen Shigella flexneri (27). However, a controversy recently emerged in the field regarding the subcellular compartment where this defense mechanism takes place. In a study conducted by Ogawa and colleagues, the autophagy marker LC3 was proposed to be recruited to cytosolic bacteria (20). However, in a recent study conducted by Campbell-Valois and colleagues, LC3 was shown to be recruited to bacteria displaying an actively secreting T3SS, which are located in membrane-bound compartments (23). In this study, we used live microscopy to track S. flexneri during the dissemination process and provided an unambiguous demonstration that LC3 is uniquely recruited to bacteria located in the double-membrane vacuoles that derive from the membrane protrusions formed during cell-to-cell spread.
Previous reports uncovered that bacteria lacking the type III secretion system effector protein IcsB were more frequently associated with LC3. IcsB was therefore proposed to counter the entrapment of S. flexneri in LC3-positive compartments targeted for degradation (20, 23, 28). IcsB was first proposed to prevent the autophagic recognition of cytosolic S. flexneri by blocking recognition of the autotransporter IcsA by ATG5 (20). More recently, IcsB was proposed to enable escape from LC3-positive membrane compartments through an unknown mechanism (23). Here we demonstrate that IcsB is required for efficient DMV escape during cell-to-cell spread. However, our results uncover that the role of IcsB in vacuole escape is not related to counteracting LC3 recruitment, as our tracking experiments visualizing the plasma membrane revealed that icsB bacteria were trapped in LC3-positive as well as LC3-negative DMVs. To our knowledge, this is the first demonstration that IcsB facilitates DMV escape through a mechanism that is unrelated to counteracting the LC3-dependent autophagy machinery. IcsB is also the first type III effector protein to be identified to specifically contribute to escape from the double-membrane vacuoles formed during cell-to-cell spread. Bioinformatics have predicted that IcsB could function as a protease or an acyltransferase (29). IcsB could thus degrade or modify vacuolar component(s), thereby challenging the integrity of the DMV membrane and promoting bacterial escape. Future studies are needed to uncover the mechanism supporting DMV escape and the exact function of IcsB in this process.
We note that a small proportion of icsB bacteria were delayed in vacuole escape but still able to successfully escape DMVs, indicating the existence of an IcsB-independent mechanism contributing to DMV escape. This pathway may involve additional type III effector proteins. Similar to bacteria lacking icsB, bacteria lacking virA were more frequently associated with LC3-positive vacuoles (23), suggesting that in addition to IcsB, VirA could contribute to vacuole escape. While VirA was shown to have GTPase-activating protein (GAP) activity in vitro (30), its exact role in vacuole escape is unknown. Interestingly, the type III effector protein IpgD was shown to be required for prompt escape from primary vacuoles upon invasion, and it may also be a candidate effector protein involved in DMV escape during cell-to-cell spread (31).
Surprisingly, our data uncovered that early recruitment of LC3 to the Shigella-containing vacuoles formed during cell-to-cell spread correlated with prompt escape of bacteria from DMVs. LC3-dependent autophagy has been shown to contribute to endosomal membrane integrity in the context of other intracellular pathogens (32–34). Interestingly, the LC3-dependent autophagy machinery was shown to be recruited to the Salmonella-containing vacuoles (SCVs) early during infection, in a manner dependent on the integrity of the type III secretion system (32). It was proposed that the membrane damages inflicted by the Salmonella type III secretion system lead to LC3 recruitment to SCVs, reflecting a role for the autophagy machinery in repairing damaged membrane. It is therefore tempting to speculate that the early recruitment of LC3 to S. flexneri-containing vacuoles is a consequence of the activity of the T3SS, which is required for DMV escape (26). Consistently, LC3 has been shown to be recruited to actively secreting S. flexneri during infection (23). DMV escape may thus be viewed as a consequence of failure of the autophagy machinery to repair the T3SS-damaged membrane. Whether the bacteria are actively engaged in that process remains to be determined.
While our work strongly suggests that the role of IcsB is not to counteract LC3 recruitment to DMVs, it also indicates that the recruitment of the LC3-dependent autophagy machinery does not restrain dissemination of wild-type bacteria. Importantly, our tracking experiments not only revealed that early LC3 recruitment predicted vacuole escape success but also demonstrated that vacuole escape failure correlated with late LC3 recruitment. These results clearly show that LC3 recruitment is not the cause but the consequence of vacuole escape failure. We therefore speculate that in the case of failure, prolonged residency of bacteria in vacuoles, perhaps due to deficient type III secretion, ultimately leads to DMV fusion with autophagosomes/lysosomes. It is noteworthy that these instances of LC3 recruitment represent a very small proportion of spreading bacteria (Fig. 9, Escape Failure). Neither drug-mediated inhibition of steps involved in autophagy nor genetic interference of LC3 enhanced the spreading efficiency of wild-type or icsB bacteria, supporting the notion that LC3-dependent autophagy does not significantly interfere with S. flexneri dissemination. In fact, chemical inhibition of autophagic effects had a negative impact on cell-to-cell spread of wild-type bacteria, potentially indicating a previously unappreciated role for autophagosome formation in particular, or vacuolar acidification in general, in S. flexneri dissemination, through an unknown mechanism.
MATERIALS AND METHODS
Cell lines and bacterial strains.HT-29 cells (ATCC HTB-38) were cultured at 37°C with 5% CO2 in McCoy's 5A medium (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Invitrogen). The wild-type Shigella flexneri strain used in this study is serotype 2a 2457T (35). The icsB strain of S. flexneri was generated by allelic exchange resulting in replacement of the icsB open reading frame (ORF) in the Shigella large virulence plasmid by the coding region of a kanamycin resistance cassette. The icsB strain was complemented by expressing wild-type icsB from the arabinose-inducible pBAD promoter in vector pBAD18 (ATCC 87393).
DNA constructs and cell transfection.HT-29 cell lines stably expressing yellow fluorescent protein (YFP) membrane markers were generated using the pMX-Mb-YFP vector (18). MAPLC3B was cloned into the XhoI and NotI sites of the pMX_mCherry vector. The corresponding lentiviruses were generated in 293T cells cotransfected with the packaging constructs pCMVΔ8.2Δvpr (HIV helix packaging system) and pMD2.G (a vesicular stomatitis virus glycoprotein) as previously described (36).
Bacterial infection.S. flexneri was grown overnight in LB broth at 30°C with agitation. The bacteria were diluted 1:100 and grown to exponential phase for approximately 3 h at 37°C with agitation. Cells were infected with S. flexneri expressing CFP under the control of an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter. Infection was initiated by centrifuging the plate at 1,000 rpm for 5 min, and internalization of the bacteria was allowed to proceed for 1 h at 37°C before gentamicin (50 μg/ml) and IPTG (10 mM final concentration) were added to kill the extracellular bacteria and induce CFP expression, respectively. For time-lapse microscopy, imaging began 2 h postinfection. For S. flexneri focus size analysis, infected cells were incubated at 37°C for 16 h.
siRNA and Western blotting.Cells were transfected by reverse transfection with Dharmafect1 and a pool of small interfering RNAs (siRNAs) (D1, D2, D3, and D4; 12.5 nM each; 50 nM total final concentration) targeting LC3B or siRNA buffer alone (mock) and incubated for 96 h in a 384-well format. Knockdown efficiency was determined by Western blotting for LC3B (Nanotools 0231-100/LC3-5F10 used at 1:100). Western blot quantification was performed using Fiji.
Size of infection foci.The size of infection foci formed in plasma membrane YFP-expressing HT-29 cells and infected with the listed CFP-expressing S. flexneri strains was determined in a 96-well plate format (catalog no. 3904; Corning). After fixation with 4% paraformaldehyde, the plates were imaged using the ImageXpress Micro imaging system (Molecular Devices). Margins of individual foci were determined manually, and image analysis for focus size (area) was performed with the ImageXpress imaging software (Molecular Devices) as previously described (24). Image analyses were conducted on at least 50 infection foci in each independent experiment.
Live imaging.Bacterial dissemination was monitored using time-lapse confocal microscopy. Plasma membrane YFP-expressing HT-29 cells were grown in McCoy's medium in eight-well chambers (Lab-Tek II [catalog no. 155409; Thermo Fisher Scientific]) at 37°C in 5% CO2. Cells were infected with the desired CFP-expressing S. flexneri strains and imaged with a Leica DMI 8 spinning-disc confocal microscope driven by the iQ software (Andor). Z-stacks were captured 2 h postinfection every 2 min for 6 h. The corresponding movies were generated with Imaris software (Bitplane). Protrusions were defined as plasma membrane extensions that formed as a result of bacteria reaching the cell cortex and projecting into adjacent cells. Vacuole-like protrusions (VLPs) were defined as an intermediate compartment between protrusions and vacuoles, characterized by a continuous membrane lining around the bacteria and a membranous tether. Double-membrane vacuoles (DMVs) were defined as membrane-bound compartments that derived from VLPs after resolution of the membranous tether. As opposed to VLPs, DMVs were therefore no longer connected to the primary infected cell. Free (cytosolic) bacteria were defined as bacteria that were previously observed in vacuoles but were no longer surrounded by a continuous lining of the plasma membrane.
ACKNOWLEDGMENTS
We thank members of the Agaisse and Derré laboratories for discussions and comments on the manuscript. We thank Natalie Sebeck for generating the rescue construct picsB.
This work was supported by the National Institutes of Health grants R01AI073904 (H.A.) and T32AI007046 (E.W.).
FOOTNOTES
- Received 16 February 2018.
- Returned for modification 7 April 2018.
- Accepted 22 May 2018.
- Accepted manuscript posted online 29 May 2018.
Supplemental material for this article may be found at https://doi.org/10.1128/IAI.00134-18.
- Copyright © 2018 American Society for Microbiology.
