Yersinia pseudotuberculosis is able to replicate inside macrophages. However, the intracellular trafficking of the pathogen after its entry into the macrophage remains poorly understood. Using in vitro infected bone marrow-derived macrophages, we show that Y. pseudotuberculosis activates the autophagy pathway. Host cell autophagosomes subverted by bacteria do not become acidified and sustain bacteria replication. Moreover, we report that autophagy inhibition correlated with bacterial trafficking inside an acidic compartment. This study indicates that Y. pseudotuberculosis hijacks the autophagy pathway for its replication and also opens up new opportunities for deciphering the molecular basis of the host cell signalling response to intracellular Yersinia infection.
After ingestion by the host, the Gram-negative enteropathogenic bacterium Yersinia pseudotuberculosis accesses the lymph nodes by moving across the epithelial barrier into Peyer's patches (Wren, 2003; Viboud and Bliska, 2005). In mesenteric lymph nodes, Y. pseudotuberculosis performs predominantly extracellular replication at late stages of infection and causes mesenteric lymphadenitis in its host (Simonet et al., 1990; Cornelis, 2002; Balada-Llasat and Mecsas, 2006). Nevertheless, experimental infection has evidenced the presence of Yersinia within intestinal macrophages in the early stages of enteric disease (cited in Pujol and Bliska, 2005) and the bacterium's ability to survive and replicate inside macrophages cultured in vitro (Isberg and Barnes, 2001; Wong and Isberg, 2005). It is thought that survival and replication inside macrophages represent an important step in early stages of infection for evading the immune response and enabling bacterial dissemination (Pujol and Bliska, 2005). Although bacteria were shown to be able to inhibit phagosome acidification in macrophages (Tsukano et al., 1999), the intracellular traffic and the bacterial intracellular niche have yet to be identified and characterized.
In the present study, we have investigated the intracellular traffic of Y. pseudotuberculosis in bone marrow-derived macrophages (BMDMs) and mouse embryonic fibroblasts (MEFs). We determined that bacteria can replicate inside autophagosomes, which are blocked in their maturation and thus are unable to fuse with lysosomes.Upon autophagy inhibition, we observed bacteria reaching an acidic compartment leading to their degradation.
Yersinia-containing vacuoles display markers of the autophagy pathway
To characterize the bacterial replication niche inside macrophages, we analysed the ultrastructure of the Yersinia-containing vacuoles (YCVs) at 4 h post infection (p.i.). As depicted in Fig. 1A and B, YCVs were delineated mainly by a double membrane or to a lesser extent by multiple membranes – a hallmark of autophagosomes. Thus, we strongly suspected that Y. pseudotuberculosis had interacted with the autophagy pathway. Autophagosome biogenesis is mediated by a specific protein conjugation system composed of several proteins encoded by the autophagy-related genes atg, of which Atg8 (also called LC3) is currently one of the few specific markers for the autophagosomal membrane (Nakagawa et al., 2004). To confirm the hypothesis of autophagy involvement in the infection process, we examined LC3 recruitment using an LC3-GFP (green fluorescent protein) construct or GFP alone and analysed the recruitment of these proteins to the YCVs. There was no association of the LC3-GFP protein with the YCVs in BMDMs at 1 h p.i. (Fig. 1C and D). Four hours p.i., we did not observe any recruitment of the GFP protein to the YCVs whereas we observed a strong association between the LC3-GFP protein and the YCVs (Fig. 1C). Additionally, 3 h p.i., we found that LAMP-1 and LC3 co-distributed in the YCVs (Fig. S1). Virulence in Y. pseudotuberculosis requires the pYV thermo-inducible plasmid, which enables the bacterium to circumvent the host's defences (notably the innate immune system) by expressing a type III secretion system (T3SS) and its associated effectors (Cornelis, 2002). However, this genetic element did not contribute to LC3-positive vacuole targeting, since the absence of pYV in infecting bacteria did not significantly influence the formation of LC3-GFP-containing YCVs (Figs 1C and D and 2). We did not perform experiments in which the T3SS was stimulated (i.e. bacterial growth at 37°C and with low calcium levels), since the bacteria stay outside the cells under these conditions due to the antiphagocytic activity of the Yop effectors. When the autophagy pathway in macrophages was activated by rapamycin treatment (Klionsky et al., 2008), we observed an increase in the size of the YCVs; the latter displayed intense LC3-GFP staining and contained many bacteria (Fig. 1E). This result led us to the hypothesis whereby YCVs are continuously supplied by autophagosomal membranes. Strikingly, we were able to track LC3-positive membranes migrating to fuse with YCVs (Movies S1 and S2). Both the origin of the donor compartment and the fusion mechanisms involved await further characterization.
To check whether or not autophagosomes indeed fused with the growing YCVs, we blocked the autophagy pathway in BMDMs [as already reported for epithelial cells (Fujita et al., 2008)] by overexpressing the enzymatically inactive Atg4B C74A mutant of Atg4B (a processing protease for pro-LC3) (Fujita et al., 2009). This mutant impairs the formation of Atg7-LC3 intermediate and thus interferes with LC3 lipidation, resulting in the absence of LC3 recruitment to the nascent autophagosomes. As shown in Fig. 2A, the recruitment of LC3-GFP protein to YCVs was impaired in the Atg4B C74A-expressing cells, regardless of whether Y. pseudotuberculosis harboured the pYV plasmid or not. Next, we used confocal microscopy to quantify the recruitment of LC3-GFP to the YCVs in infected BMDMs treated (or not) with rapamycin and expressing (or not) the Atg4B C74A mutant. In cells with unimpaired autophagy activity, almost all the bacteria were found inside LC3-positive vacuoles (Fig. 2B). In contrast, when the cells were expressing the Atg4B C74A mutant, the recruitment of LC3-GFP to the YCVs was abrogated. As a control, we did not observe any recruitment of LC3-GFP to the vacuole in BMDMs infected by a non-pathogenic Escherichia coli strain (Figs 1B and 2B), indicating that the recruitment of the autophagicmachinery was an active process and at least selective for Yersinia. Moreover, as shown in Fig. 2B and Fig. S2, this recruitment was specific for viable bacteria, demonstrating that metabolically active, pathogenic microorganisms are needed to trigger this phenomenon (which is modulated by one or more chromosomally encoded factors).
Autophagy is induced by Yersinia infection of macrophages
To determine whether the autophagy pathway was induced upon infection, we monitored the post-translational processing of the LC3-GFP protein. Indeed, upon autophagy activation, the cytosolic form of LC3 (LC3 I) targets to the newly generated autophagosomes via phosphatidylethanolamine lipidation by a ubiquitin-like conjugation system; this results in the formation of the membrane-associated form of LC3 (LC3 II) whose electrophoretic mobility was greater than that of the LC3 I form. Indeed, upon rapamycin treatment, we observed an increase of LC3 I conversion (Fig. 3A). We checked that overexpression of the Atg4B C74A mutant protein in our experimental model decreased LC3 II/I conversion following autophagy stimulation with rapamycin (Fig. 3A). Then, we infected BMDMs transiently expressing either the LC3-GFP protein alone or both LC3-GFP and Atg4B C74A. As shown in Fig. 3B, the lipidated LC3 form (LC3 II) was weakly enhanced in 3 h infected macrophages. Other authors have previously established that the LC3 II form trapped inside autophagosomes is degraded by lysosomal hydrolases after formation of an autophagolysosome (Tanida et al., 2005; Klionsky et al., 2008). Hence, the lysosomal turnover of LC3 II and not a transient increase in LC3 II is indicative of autophagy induction (Tanida et al., 2005; Klionsky et al., 2008). To assess whether Y. pseudotuberculosis rerouted the cellular pool of autophagosomes or indeed activated their formation, we studied the effect of blocking autophagosome degradation with an inhibitor of the lysosomal enzyme V-ATPase (bafilomycin A1), which notably impedes the fusion between autophagosomes and lysosomes (Yamamoto et al., 1998; Klionsky et al., 2008). When macrophages infected by Y. pseudotuberculosis were pre-incubated with bafilomycin A1, the amount of LC3 II was substantially greater, when compared either with non-infected but treated cells or with infected and untreated cells (Fig. 3B). This result indicated that macrophage infection was indeed inducing autophagy. We also observed increased accumulation of LC3 II in infected cells expressing the dominant-negative Atg4B C74A and treated with bafilomycin A1 (Fig. 3B). This result could be explained by accumulation of the unclosed autophagosomes seen during Atg4B C74A expression (Fujita et al., 2008; Fujita et al., 2009).
Once lipidated, LC3 undergoes a transition from the cytosolic form to a membrane-bound form. We also scored LC3-GFP dots in BMDMs by fluorescence microscopy (Klionsky et al., 2008). Following autophagy induction by rapamycin treatment, we observed an increase in LC3-GFP dots in uninfected cells (Fig. 3C); this indicated a good correlation between induction of autophagy and increased LC3-GFP dot formation. As expected, this increase was not noticed with Atg4B C74A expression (Fig. 3C). When cells were infected with Y. pseudotuberculosis, the number of LC3-GFP dots was significantly higher than in uninfected cells (Fig. 3C and D), regardless of the presence of the pYV in bacteria (Fig. 3C and D). The potential regulation of autophagy via T3SS overexpression was not explored herein and will be investigated in future work. Additionally, infected cells in which autophagy was abolished (i.e. during Atg4B C74A mutant expression) displayed a similar number of LC3-GFP dots to that in uninfected cells, in agreement with a previous biochemical analysis (Fig. 3B–D). Moreover, killed bacteria were unable to induce LC3-GFP dot formation in control cells (Fig. 3C and D). Lastly, BMDMs infected with a non-pathogenic strain of E. coli had much the same number of LC3-GFP dots as uninfected cells did – confirming that induction of autophagy by Y. pseudotuberculosis was an active, specific, bacterial process (Fig. 3C and D). Hence, both measuring the autophagic flux and quantification of the LC3-GFP dots led to the conclusion that Yersinia infection induced the autophagy.
Yersinia replicates within autophagosomes in macrophages
In order to ascertain the functional role of autophagy in the invasion of macrophages by Yersinia, we examined bacterial replication inside macrophages in general and inside autophagosomes in particular. First, we checked that bacterial entry into macrophages was unchanged when the cells expressed the dominant-negative Atg4B C74A or were treated with rapamycin (Fig. 4A). To test the survival of Y. pseudotuberculosis inside macrophages, we monitored bacterial replication using the classical colony-forming unit (cfu) assay and observed a decrease of bacterial yield in Atg4B C74A-expressing cells compared with control cells (Fig. 4B). However, for a more rigorous examination, we wanted to take into account the yield of transfection efficiency (i.e. about 40% in BMDMs) and the cell death induction upon the infection (data not shown) and thus used the accumulating DNA probe ethynyldeoxyuridine (EdU)-Alexa488 (Salic and Mitchison, 2008). This method allowed us to work at the single cell level, therefore increasing the specificity of the bacterial replication assay. Like bromodeoxyuridine, EdU is a thymidine analogue that is incorporated into replicating DNA. We verified the good correlation between EdU-Alexa488 accumulation and bacterial growth (data not shown). As shown in Fig. 4C, the level of replicating bacteria in BMDMs was significantly higher when autophagy was stimulated by rapamycin treatment, compared with Mock-transfected cells (Fig. 4C). In contrast, expression of the Atg4B C74A-mStrawberry construct led to a dramatic decrease in the amount of replicating bacteria (Fig. 4C and D). A similar finding was observed in BMDMs transfected with another Atg mutant (Atg5 K130R) that had previously been reported as having an impaired autophagy pathway (Mizushima et al., 2001; Fig. 4C). Rapamycin treatment was not able to restore bacterial replication in cells expressing either of the two mutants proteins (Fig. 4C). This set of results demonstrated that the autophagy pathway played an important role in bacterial replication inside macrophages.
To establish how autophagy could be involved in bacterial multiplication within the host cells, we looked at whether autophagosomes constituted the replication niche for the pathogen. To this end, macrophages were infected with Y. pseudotuberculosis transformed with p67GFP3.1 [a plasmid that encodes GFP under the control of an IPTG-inducible promoter (Pujol and Bliska, 2003)] and the co-distribution of bacteria and LC3-mRFP (monomeric red fluorescent protein) proteins was analysed using confocal microscopy (Fig. 4E and F). One hour prior to cell fixation, IPTG was added to the medium in order to induce GFP expression in metabolically active bacteria only, since dead bacteria could not produce GFP following IPTG supplementation of the cell culture medium. As shown in Fig. 4E and F, metabolically active bacteria (representing more than 80% of the bacterial population) were inside LC3-positive compartment 4 h p.i., indicating that bacterial protein synthesis did indeed occur within these organelles. Hence, autophagy facilitates Y. pseudotuberculosis replication in BMDMs, as demonstrated by both cfu counting and EdU accumulation. Moreover, the replication occurs in LC3-positive intracellular compartment demonstrated by localization of GFP under IPTG-inducible promoter-expressing bacteria in LC3-containing membrane organelles.
Pujol et al. recently reported that Y. pestis could replicate within autophagosomes; however, autophagy per se was not required for intracellular survival, since macrophages from Atg5−/− mice could sustain bacteria replication (Pujol et al., 2009). We thus decided to look at whether Y. pseudotuberculosis could replicate in autophagy-deficient Atg5−/− MEF cells (Kuma et al., 2004; Kaushik et al., 2008). In contrast to Atg5−/− MEF cells, we noted that metabolically active bacteria co-distributed with LC3 positive compartments in wild-type MEF cells (Fig. 5A and B) in which they were able to replicate (Fig. 5A–C). We still observed both metabolically active bacteria and bacterial multiplication in Atg5−/− MEF cells but the replication is lower compared with wild-type MEFs (Fig. 5C).
Autophagy inhibition leads to Yersinia trafficking inside an acidic compartment
To analyse whether autophagosomes containing replicating bacteria could fuse with lysosomes, we investigated the acidification of YCVs by using the LysoTracker probe [an acidotropic fluorescent dye that accumulates in acidic organelles (Via et al., 1998)]. In macrophages expressing the Atg4B C74A protein, bacteria were mostly included in a LysoTracker-positive compartment (Fig. 6A and B); this contrasted with control macrophages for which the bacteria-containing compartment was mostly LysoTracker-negative (Fig. 6A and B). A similar finding was seen when cells were transiently expressing the Atg5 K130R mutant (Fig. 6A). In opposite, killed bacteria reached a LysoTracker-positive compartment that was LC3-negative (Fig. S2). Rapamycin treatment did not influence the percentages of LysoTracker-positive YCVs (Fig. 6A), since the drug induces the autophagosomes but does not control the autophagosome–lysosome fusion step. This result is in agreement with Pujol et al. who have recently demonstrated that Y. pestis is not localized in an acidic compartment (Pujol et al., 2009). However, our data indicate also that autophagy inhibition leads to the trafficking of bacteria inside an acidic compartment (Fig. 6A and B). To confirm this intriguing result and to avoid classical side-effects of proteins overexpression (proteins mistargeting, homeostasis troubles, etc.), we have performed experiments using the autophagy-deficient Atg5−/− MEF cells. In a first assay, we have performed experiments in live infected cells and followed the YCVs acidification using the LysoSensor Blue dye. We recorded the distribution of IPTG-dependent GFP-expressing bacteria. The LysoSenso Blue dye is an acidotropic probe that appears to accumulate in acidic organelles as the result of protonation. Because of its low pKa value (5.1), LysoSensor Blue is almost non-fluorescent except when inside acidic compartments. A clear picture of LysoSensor-positive YCVs in Atg5−/− MEF cells 4 h p.i. is shown in Fig. 6C, in contrast to wild-type MEF cells. In order to quantify the acidic YCVs, we have used the LysoTracker probe that reaches lysosomes and fix the cells 4 h p.i. As shown in Fig. 6D and E, we have also observed the localization of YCVs in a LysoTracker-positive compartment in Atg5−/− MEF cells, confirming the data obtained on BMDMs.
Maturation of Yersinia-containing autophagosomes is impaired in macrophages
Next, to investigate specifically the maturation of Yersinia-containing autophagosomes in macrophages, we employed pmRFP-LC3-GFP (a recombinant plasmid encoding the LC3 protein fused to both the GFP and the mRFP proteins) to study the acidification of bacteria-containing autophagosomes. A tandem mRFP-GFP fluorescence microscopy analysis with this construct enables discrimination between neutral and acidified compartments: both proteins are excited (leading to yellow staining) in a neutral environment, whereas protonation of GFP at an acidic pH (due to the differences in pK values, i.e. 6.0 for GFP and 4.5 for mRFP) attenuates the fluorescence (leading to a redder staining) (Kimura et al., 2007). This is illustrated in Group A streptococci-infected HeLa cells in which GFP staining is attenuated when bacteria have reached the acidified lysosomes (Fig. S3). We observed both mRFP and GFP fluorescence in the YCVs (Fig. 6F), thus confirming the maturation impairment for Yersinia-containing autophagosomes.
As the autophagy flux is not impaired in BMDM-infected macrophages (Fig. 3B; lane 3 versus lane 4), we next analysed the maturation of bacteria-free autophagosomes inside the cells. In contrast to YCVs, the maturation of bacteria-free autophagosomes was unchanged, since some LC3 structures were only mRFP-positive (Fig. 6G) – indicating the autophagosome maturation towards an acidic compartment (i.e. the autophagolysosomes).
We have shown that in primary macrophages, Y. pseudotuberculosis replicates within autophagosomes whose ability to fuse with acidic compartments has been blocked. Interestingly, we have observed that bacteria-free autophagosomes can fuse with lysosomes, indicating that the autophagic machinery is not impaired. This result suggests that Y. pseudotuberculosis has evolved a mechanism to avoid fusion of the replicative autophagic vacuole with lysosomes (Fig. S4), the molecular basis of which remains to be deciphered. It is possible that, like Mycobacterium tuberculosis, Y. pseudotuberculosis prevents association of the vATPase with the YCV. Alternatively, the vATPase may associate with the YCV but is inactivated, as has been suggested in a previous study (Tsukano et al., 1999). Microscopy techniques should allow us to distinguish between these possibilities in the future. Moreover, when autophagy is inhibited, either by the Atg4B mutant overexpression or by the use of Atg5 knockout cells, bacteria are localized in an acidic compartment and are degraded. This result suggests that the autophagy pathway provides a survival niche for Y. pseudotuberculosis. The question related to the benefit between an autophagosome compared with a phagosome for bacterial replication is exciting and needs further investigations. Considering the work of Sanjuan et al. that demonstrates the role of autophagy in phagosome maturation (Sanjuan et al., 2007), our observation is quite surprising. However, this latter was performed using latex beads associated with Toll-like-receptor ligands whereas we used pathogenic bacteria, which reinforces the possibilities of survival strategies developed by pathogens to avoid phagosome/autophagosome maturation as discussed above.
The biogenesis of neutral pH Yersinia-containing autophagosomes may interfere with Yersinia-induced immunity: blocking autophagosome maturation might allow the bacteria to establish a replication niche which is then continuously supplied with nutrients via fusion with incoming autophagosomal membranes as a consequence of autophagy induction. Hence, Yersinia's hijacking of the autophagy pathway differs from the mechanisms employed by many other bacterial pathogens that interfere with this route. Indeed, for group A Streptococcus, Salmonella enterica and Shigella flexneri, it has been suggested that trapping into autophagosomes acts as clearance for an unsuccessful infection process (Nakagawa et al., 2004; Birmingham et al., 2006; Ogawa and Sasakawa, 2006). In contrast, M. tuberculosis is engulfed in a vacuolar compartment, which only undergoes autophagosomal fusion for degradation in activated macrophages (Gutierrez et al., 2004). Legionella pneumophila, after entry into macrophages, uses a raft-dependent pathway and reaches an autophagosome–lysosomal compartment for degradation (Amer and Swanson, 2005). Fusion of autophagosomes with bacteria-containing vacuoles is thus detrimental for these bacteria and confirms the role of autophagy as a cytoprotective pathway (Gutierrez et al., 2007; Dupont et al., 2009; Moreau et al., 2009). However, invasive bacteria like Francisella tularensis, which can grow within the cytoplasm, can be re-internalized within autophagosomes that may play a role in egress from the host cell (Checroun et al., 2006). It has been suggested that Chlamydia trachomatis uses autophagosomes as nutrient suppliers or vehicles (Al-Younes et al., 2004). Coxiella burnetii also uses autophagosomes as nutrient suppliers, in order to multiply in a large, acidic, endosome- and autophagosome-derived vacuolar compartments (Colombo et al., 2006).
The autophagy induction mechanism upon Y. pseudotuberculosis infection needs to be deciphered. Recent data obtained with Y. enterocolitica have demonstrated that autophagy stimulation is mediated by the Yersinia adhesins invasin and YadA and depends on the engagement of β1 integrin receptors (Deuretzbacher et al., 2009). This could be similar for Y. pseudotuberculosis but not for Y. pestis as this last specie is invasin- or YadA-deficient (Pujol and Bliska, 2005). Preliminaries data obtained by our group indicate a possible role of invasin in Y. pseudotuberculosis-mediated autophagy induction (data not shown).
Ruckdeschel and colleagues have recently described the degradation of Y. enterocolitica in J774 macrophage autophagolysosomes (Deuretzbacher et al., 2009). In contrast and in agreement with the report from Bliska and colleagues studying Yersinia pestis (Pujol et al., 2009), we found that Y. pseudotuberculosis can replicate in autophagosomes in BMDMs although we noticed that autophagy was important for the infectious process. This discrepancy could be due to the differential cellular models, as we observed that Y. pseudotuberculosis was localized in macrophage cell lines (RAW 264.7 and J774) in typical autophagosomes and in an additional intracellular compartment that remains to be characterized (our unpublished data). Alternatively, it might be related to distant pathogenesis evolution of Y. enterocolitica and Y. pseudotuberculosis/Y. pestis, which separated 80 million years ago whereas Y. pestis emerged from Y. pseudotuberculosis less than 20 000 years ago (Achtman et al., 1999). However, it is important to note that Y. pseudotuberculosis replicated mainly in autophagosomes in BMDMs from wild-type mice (i.e. cells lacking potential compensatory mechanisms following inhibition of gene expression). The observed difference in autophagosomal bacterial localization when comparing the report from Bliska and colleagues with the present study (30% versus 80% respectively) might be due to the Yersinia species examined (Y. pestis and Y. pseudotuberculosis respectively) and/or the gene expression system employed (retroviral transduction and plasmid transfection respectively). Moreover, although autophagy is not required for bacterial survival, it is important for Yersinia multiplication as demonstrated by both the increased rate of replication upon rapamycin treatment in BMDMs and the kinetic of bacterial replication in MEF cells.
The localization of Y. pseudotuberculosis inside autophagosomes could induce important consequences for the outcome of cellular infection. Indeed autophagy has been involved in a number of different pathways, in particular in the cell death regulation (Mizushima et al., 2008; Levine and Kroemer, 2008). The recruitment of the autophagy machinery to the YCVs and the hijacking of autophagosomes for bacterial replication may prevent the use of autophagy for its normal functions, such as damaged mitochondria clearance and survival. For example, a loss of the autophagic function has been involved in the regulation of an inflammation response (Saitoh et al., 2008; Dupont et al., 2009). We hypothesize that modulation of autophagy by Y. pseudotuberculosis infection may compromise macrophages viability. Further investigations of the interactions between the bacteria and autophagy will be required to elucidate the physiological role of autophagy in Y. pseudotuberculosis infection.
Bacterial strains and plasmids
Yersinia pseudotuberculosis strain IP32777 (bearing the virulence-associated plasmid pYV; pYV+) and its isogenic pYV-cured derivative IP32777 (pYV-) were used in this study. Strain IP32777 expressing the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible GFP was constructed as described elsewhere (Pujol and Bliska, 2003).
Bacteria were grown at 28°C in Luria–Bertani (LB) broth or on LB agar plates. For inducing GFP expression in bacteria, 300 µM IPTG was added to the medium for 1 h. GFP expression was undetectable in the absence of IPTG induction (data not shown). Bacteria were killed by treatment with 4% paraformaldehyde for 20 min and were used as a negative control to score LC3-GFP acquisition by the YCVs. E. coli strain K-12 and a clinical strain of Group A Streptococcus were also used in this study and were grown at 37°C in LB broth or brain heart infusion medium for streptococci.
Mouse antiserum against Y. pseudotuberculosis was prepared from animals sacrificed at 4 weeks post oral challenge with a sublethal dose of strain IP32777. Mouse anti-GFP antibody was purchased from Sigma Chemicals. Rabbit anti-LC3 (PM036) was purchased from MBI International. Rat monoclonal anti-mouse LAMP-1 (1D4B, developed by J.T. August) was obtained from the Developmental Studies Hybridoma Bank (Baltimore, MD, USA) under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biological Sciences of the University of Iowa (Iowa City, IA, USA). Caspase-1 antibody was a kindly gift from P. Vandenabeele (Ghent University, Belgium). The secondary antibodies used were Alexa Fluor555-conjugated goat anti-rabbit IgG, Alexa Fluor555-conjugated goat anti-mouse IgG and Alexa Fluor555-conjugated goat anti-rat IgG (Molecular Probes). Alexa Fluor555-conjugated cholera toxin B subunit, LysoTracker Red DND-99 (577/590), LysoTracker Blue DND-22 (373/422), LysoSensor Blue DND-167 (373/425), Syto61, Syto81 and Lipofectamine2000 were purchased from Molecular Probes. 4′-6-Diamidino-2-phenylindole (DAPI) and Bafilomycin A1 were obtained from Sigma Chemicals. Caspase-1 inhibitor acetyl-Tyr–Val–Ala–Asp chloromethylketone (YVAD) was purchased from Calbiochem and Caspase-1 FLICA (FAM-YVAD-FMK) was purchased from Immunochemistry Technologies, LLC.
Cell culture and transfection
To generate BMDMs, cells were collected from dissected femurs of 10-week-old BALB/c female mice. Macrophages were derived in 100 mm non-tissue-culture-treated dishes containing DMEM-Glutamax with 30% l-cell conditioned medium (as a source of CSF-1), 10% fetal calf serum (FCS) and 100 U ml−1 penicillin G/streptomycin for 5 days. HeLa cells were cultured in Eagle's minimal essential medium (EMEM) supplemented with 10% heat-inactivated FCS, 2 mM l-glutamine, 1% non-essential amino acids and 100 U ml−1 penicillin G/streptomycin. Atg5−/− MEFs were obtained from N. Mizushima (Tokyo Medical and Dental University, Tokyo, Japan) and cultured as previously described (Mizushima et al., 2001).
BMDMs were transiently transfected with plasmids using Amaxa Biosystems technology (Mouse Macrophage Nucleofector Kit), according to the manufacturer's instructions. The plasmid pcDNA3.1 (Invitrogen) was used as a control (‘mock’).
Autophagy was induced in macrophages by rapamycin treatment (Sigma Chemicals; 10 µg ml−1 for 2 h at 37°C). For inhibition of the lysosomal protease V-ATPase, bafilomycin A1 (100 nM) was added to the cell monolayer 2 h before infection. Caspase-1 inhibition was performed by incubating the macrophages with the inhibitor YVAD (100 µM for 3 h).
Macrophage and MEF infection assay
Macrophage or MEF infection was performed at a multiplicity of infection (moi) of 5 or 10 bacteria per cell respectively. Briefly, exponentially grown (OD600 = 0.8) bacteria suspended in PBS were centrifuged onto macrophages for 5 min at 200 g, in order to synchronize the infection. For video microscopy, bacteria were pre-labelled with DAPI (25 µg ml−1) for 2 h at 28°C. Infected macrophages were incubated for 30 min at 37°C in 5% CO2. After two washes with PBS, cells were further incubated with fresh cell culture medium containing gentamicin at 8 µg ml−1, in order to kill extracellular microorganisms.
Macrophage invasion assay
In order to measure the cellular uptake of bacteria, cells were infected for 30 min and then fixed with 4% paraformaldehyde for 15 min. Total bacteria were labelled with the DNA markers DAPI (25 µg ml−1) or Syto81 (25 µM) for 20 min and extracellular bacteria were labelled with a specific antibody against Yersinia for 30 min, in the absence of a membrane permeabilization step. Immunofluorescence microscopy was used to discriminate between intracellular and extracellular bacteria; live and dead bacteria did not differ dramatically in the extent of labelling (data not shown).
Bacterial intracellular replication assay
In order to determine the number of intracellular bacteria, macrophages were washed with PBS and lysed with 0.1% Triton X-100 (Sigma Chemicals). The lysates were sonicated to disperse the bacteria and then plated onto LB agar plates for viable Yersinia counting (cfu).
In order to measure the intracellular replication of bacteria, Click-iT™ EdU Alexa Fluor® Imaging Kits (Molecular Probes) were used to quantify the accumulation of ethynyldeoxyuridine (EdU) in replicative DNA, according to the manufacturer's instructions. Briefly, EdU was added to the medium 1 h before fixation. After Triton X-100 treatment, EdU was detected using an immunostaining technique based on Alexa Fluor antibodies. IPTG-inducible GFP protein was used to quantify Yersinia protein synthesis in macrophages and MEFs cells, as described by Pujol and Bliska (2003).
Proteins from cell lysates and supernatants were separated by SDS-PAGE. Proteins of interest were detected using a specific antibody and a horseradish peroxidase-conjugated secondary antibody. Immune complexes were revealed using SuperSignal West Pico Chemiluminescent Substrate (Pierce). The density of LC3-GFP bands was measured using the ImageJ software®.
Fluorescence and transmission electron microscopy
Fluorescence emission was observed with a Leica SP2 and a Zeiss LSM710 confocal microscope or a Carl Zeiss AxioImager® microscope fitted with an AxioCam® MRm camera and operated by AxioVision® software. Images were acquired and assembled using Adobe Photoshop® software.
For transmission electron microscopy analysis, cells were fixed for 2 h in 2.5% glutaraldehyde in PIPES buffer (100 mM PIPES, 80 mM NaCl, 10 mM KCl, 4 mM CaCl2, 2 mM MgCl2), post-fixed for 1 h in 1% osmium in 100 mM PIPES buffer, dehydrated and, lastly, embedded in Epon. Contrasted ultrathin sections (70 nm) were examined under a Hitachi 7500 electron microscope.
Live cell imaging
The recruitment of cell markers was tracked by time-lapse, high-speed video microscopy. Epifluorescence and total internal reflection fluorescence (TIRF) were recorded using an inverted Carl Zeiss Axio Observer Z1 microscope fitted with a Zeiss AxioCam® MRm camera and operated by AxioVision® software. Images were acquired with a Plan-Apo 100×/1.46 oil immersion objective. For video microscopy, the fluorophore excitation system was composed of a Colibri system with LEDs at 365, 470 and 530 nm. For TIRF, the fluorophore excitation system featured a laser module with a multi-line (458/488/514 nm 100 mW) argon laser (Lasos Lasertechnik GmbH) and a laser TIRF slider with a monomode fibre 561 nm 10 mW diode-pumped solid-state laser system (Melles Griot), further equipped with a lambda 10-3 optical filter changer with SmartShutterTM (Sutter Instrument Company). Images were further analysed using Adobe Photoshop® software and converted into QuickTime® movies using Adobe Flash® software.
Unless otherwise stated, all data are reported as the mean ± SEM of three independent assays. A two-tailed, unpaired Student's t-test was used for inter-group comparisons. A P-value below 0.05 was considered to be statistically significant.
We thank Joëlle Warein for expert technical assistance and Julie Bertout for FACS analysis. We are grateful to J.B. Bliska, N. Mizushima, P. Vandenabeele and F.G. van der Goot for sharing reagents. We thank Guy Tran van Nhieu for critical reading of the manuscript. We thank the Microscopy-Imaging-Cytometry Facility of the Pasteur Lille campus (MICPaL) for access to instrumentation and technical assistance. F.L. holds grants from the French Ministry for Research and Technology [Programme Chaire d'Excellence (042466)] and the ‘Fondation pour la Recherche Medicale’ (DEQ20051205758). S.L.-G. holds a Postdoctoral Fellowship from the Région Nord-Pas-de-Calais.