By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Induction of efficient adaptive T cell-mediated immunity against the intracellular bacterium Listeria monocytogenes requires its successful invasion of host cell cytosol. However, it is not clear whether its cytosolic escape and growth are sufficient to induce T cell-mediated clearance and protection upon secondary infection. To investigate this issue, we have searched for mutants that do not induce long-term protective immunity yet invade the cytosol of infected cells. We found that mice immunized with L. monocytogenes lacking the SecA2 ATPase, an auxiliary protein secretion system present in several Gram-positive pathogenic bacteria, mounted a robust cytolytic IFN-γ-secreting CD8+ T cell response but were not protected against a secondary challenge with wild-type (wt) bacteria. Furthermore, CD8+ T cells from mice immunized with secA2– bacteria failed to transfer protection when injected into recipient mice demonstrating that they were unable to confer protection. Also, secA2– and wt L. monocytogenes spread to the same myeloid-derived cell types in vivo and SecA2 deficiency does not interfere with intracytosolic bacteria multiplication. Therefore, cytosol invasion is not sufficient for inducing secondary protective responses and induction of memory CD8+ T cells mediating long-term antibacterial protective immunity is dependent upon SecA2 expression inside the cytosol of host cells in vivo.
Deciphering the basis of protective immunological memory against intracellular pathogens is a field of intense investigations, ultimately aiming at improving vaccination strategies. To gain insight into the mechanisms by which intracellular pathogens induce protective responses, it is important to understand which bacterial factors are critical for the development and the maintenance of long-term T cell-mediated protective responses.
Listeria monocytogenes is a facultative intracellular bacterium that causes severe food-borne infections in immunocompromised patients (Taege, 1999; Portnoy et al., 2002). In contrast to intracellular persistent vacuolar bacteria, L. monocytogenes escapes primary phagocytic vacuoles and rapidly grows inside the cytosol of host cells to establish a productive infectious cycle. L. monocytogenes cytoplasmic escape and multiplication is dependent upon the pore-forming toxin listeriolysin O (LLO) (Gaillard et al., 1986), which acts by dissolving primary phagocytic vacuoles. In this setting, both primary and secondary protective CD8+ T cell responses have been well characterized and suggest to require intracytosolic bacterial growth (Berche et al., 1987). Along similar lines, immunization of mice with heat-killed L. monocytogenes (HKLM), which cannot escape phagocytic vacuoles, do not confer protective immunity (von Koenig et al., 1982; Lauvau et al., 2001). Thus, infection of mice with a sublethal dose (5 × 103) of live cytosolic invasive wild-type (wt) bacteria induces CD8+ T cell differentiation into memory cells (Busch et al., 1998) that mediate long-lived immunological protection against otherwise lethal doses (> 105) of bacteria (Mielke et al., 1989; Lauvau et al., 2001).
secA2 – bacteria do not induce a protective secondary response against L. monocytogenes
To search for bacterial mutants that do not induce a protective secondary response against wt L. monocytogenes, mice were inoculated with 0.1 × LD50 of 25 different mutants of this bacterium (Fig. 1 and Table S1). Immunized animals were infected 1 month later with 8 × 105 or 3 × 105 wt bacteria respectively, and monitored for both survival and bacterial titers in the spleen and the liver (Fig. 1). Control animals were injected with PBS, heat-killed (von Koenig et al., 1982) or live wt, actA– (Kocks et al., 1992) or plcA/B– (Smith et al., 1995) bacteria. As expected, mice immunized with HKLM and naïve mice exhibited 1000-fold more bacteria at 1.5 days than mice immunized with wt, actA– or plcA/B– bacteria, and rapidly succumbed to the infection (Fig. 1A and data not shown). Among the 25 mutants, six including one mutant lacking the master regulator of Listeria virulence gene expression (PrfA) (Leimeister-Wachter et al., 1990) and three mutants lacking LLO (Gaillard et al., 1986; Berche et al., 1987) failed to induce protection. These results were expected because neither PrfA– nor llo– mutants escaped to the cytosol, a prerequisite to induce long-term protective immunity. Interestingly, mutants lacking SecA2 (Lenz and Portnoy, 2002) did not control secondary infection with wt bacteria and exhibited almost comparable increase in bacterial counts over time as compared with mice injected with PBS, HKLM, or llo– bacteria (Fig. 1B). These results show that primary infection with secA2– bacteria did not induce a protective secondary response and further suggest that the lack of this dedicated secretion system interferes with the development of long-term protective immunity. As expected, complementation of the secA2– mutant by reintroducing the wt secA2 gene restored the phenotype of wt bacteria and induced immunity in immunized animals (data not shown). To further investigate this phenomenon, we have compared the immune response induced by secA2–, actA– and wt bacteria. We have chosen actA– bacteria because they exhibited the same LD50 of 107 in BALB/c mice than secA2– bacteria (Goossens and Milon, 1992; Smith et al., 1995; Lenz and Portnoy, 2002).
Primary and secondary CD8+ T cell responses are comparable in mice immunized with secA2–, actA– and wt bacteria
Previous results have shown that the long-term maintenance of the protective response against L. monocytogenes is dependent on CD8+ T cells (Mielke et al., 1989; Lauvau et al., 2001). To confirm these data, mice immunized with wt bacteria were depleted of either CD8+ T cells, CD4+ T cells or both (Fig. 1C). As expected, CD8+ T cells, but not CD4+ T cells, were required for protection. We next analysed bacteria-specific CD8+ T cells in mice immunized with secA2–, actA– or wt bacteria. Mice exhibited comparable frequencies of endogenous primary LLO91−99/H2-Kd-tetramer-specific CD8+ T cells (3–4%) among CD3+CD8+ T cells (Fig. 2A). Similarly, no difference between the different groups were observed following secondary infection (15–17%). Therefore, CD8+ T cells from mice immunized with secA2– bacteria were not impaired in their ability to proliferate in vivo upon primary or secondary infection as compared with those from mice immunized with actA– or wt bacteria.
Memory CD8+ T cells from mice immunized with secA2– bacteria do not confer protection
We next assessed whether CD8+ T cells from mice immunized with secA2–, actA– and wt bacteria were functionally different. The frequencies of CD8+ T cells that secreted IFN-γ or expressed granzyme B were similar in all groups (Fig. 3A and B). To investigate whether these cells were equally capable of conferring protective immunity, mice were primarily immunized with secA2–, actA– or wt L. monocytogenes or injected with PBS and secondary challenged with wt bacteria 1 month later. Mice were sacrificed 6 h after the challenge infection and CD8+ T cells were sorted by flow cytometry and transferred into naïve recipients. Animals were subsequently infected with 3 × 105 wt bacteria, and viable bacteria in the spleen and the liver were enumerated 48 h later (Fig. 3C and data not shown). Mice that received CD8+ T cells from mice immunized with actA– or wt bacteria exhibited 4.2- and 5.3-fold less bacteria in the spleen and in the liver respectively, as compared with those that received CD8+ T cells from mice immunized with secA2– bacteria or injected with PBS. Taken together, these results demonstrated that memory CD8+ T cells from mice immunized with secA2–L. monocytogenes were impaired in their ability to confer protection.
secA2 –, llo–, actA– and wt bacteria spread to the same cell types in vivo
To further investigate why secA2– bacteria failed to induce protective immunity, we have generated erythromycin-resistant bacteria expressing high levels of green fluorescent proteins (GFP). These bacteria were readily detected in infected spleen cells by fluorescence-activating cell sorting (FACS) (Fig. S1A) and by confocal microscopy (Fig. S1B). To determine whether GFP was stably expressed in vivo, mice were infected with GFP+ bacteria and live splenocytes sorted by flow cytometry according to their level of fluorescence. Cells were lysed, plated onto Petri dishes, and bacterial colony-forming units (cfu) enumerated (Fig. S1C). GFPbright cells (G4) accounted for 3.5% of GFP+ cells and 0.07% of all cells. GFPbright cells also contained 99% of all bacteria demonstrating that almost all live bacteria expressed GFP in vivo. In an alternative experiment, splenocytes from infected mice were prepared and plated on Petri dishes with or without erythromycin. Identical numbers of bacterial CFU were observed on both plates and all bacterial clones expressed GFP further confirming that GFP-expressing bacteria were highly stable in vivo (data not shown).
To determine whether secA2–, actA– and wt bacteria spread to different cell types in vivo, mice were injected with various numbers of bacteria and splenocytes were analysed 1 h later for the presence of GFP+ cells. GFP+ cells were readily detected in spleen only when high numbers (> 108) of bacteria were used (data not shown). GFP+ splenocytes were analysed for expression of various markers including F4/80, Gr1, CD11c, B220, CD49b, CD80, CD86, MHC-II, CD11b and Ly-6C that allow to distinguish neutrophils (CD11bhighLy-6Cmed/high), TNF/iNOS-producing dendritic cells (Tip-DCs) (CD11bmed/highLy-6Chigh) (Serbina et al., 2003), macrophages (CD11bmedLy-6Cmed) and conventional CD11chigh DCs (Serbina and Pamer, 2006) (Fig. 4 and data not shown). One hour after infection, wt and mutant GFP+ bacteria were similarly distributed in neutrophils (53–56%), Tip-DCs (8–10%), macrophages (< 5%) and CD11chigh DCs (2.0–3.0%) (Fig. 4B). This latter result was confirmed by analysing GFP+ cells by electron microscopy (EM) (Fig. S2). Therefore, live bacteria were mainly found in CD11b- and Gr1-expressing myeloid-derived cells early after infection. Because CD8α+ CD11chigh DCs are more potent initiators of CD8+ T cell-mediated immunity following L. monocytogenes infection (Belz et al., 2005), we have further analysed the distribution of secA2–, actA– and wt bacteria in CD8α+ and CD8α– DC subsets (Shortman and Liu, 2002) (Fig. 4C). GFP-expressing wt and mutant bacteria exhibited the same tropism and were present in both CD8α+ (29–31%) and CD8α– (69–71%) CD11chigh DCs. Similar experiments were performed in mice infected with 0.1 × LD50 of each bacteria for 24 h. As observed at 1 h after infection, secA2–, actA– and llo– mutants as well as wt bacteria were similarly distributed in neutrophils (40–50%), Tip-DCs (13–21%) and CD11chigh DC (1.5–2%) (Fig. 4D and Fig. S3).
Although the frequency of CD11chigh DCs among GFP+ was identical 1 and 24 h after infection, the total number of CD11chigh DCs was lower at 24 h as compared with 1 h after infection. Furthermore, while the frequency of GFP+ cells among splenocytes was similar for all bacteria 1 h after infection, this frequency varied substantially between < 0.001% for llo–, 0.004% for secA2– and 0.006% for actA– and 0.06% for wt bacteria at 24 h, correlating with the rate of multiplication of these bacteria in vivo (Fig. 4E). Therefore, the impaired ability of secA2– bacteria to promote the development of a secondary protective response did not result from an altered tropism of these bacteria in vivo.
secA2 – bacteria invade host cell cytosol in vivo
To investigate whether secA2– bacteria grow inside the cytosol of infected cells in vivo, GFP+ cells were purified from mice infected with GFP-expressing secA2– bacteria and analysed by EM (Fig. 5A). As observed in previous reports using in vitro infected cell lines, wt and actA– bacteria were found in the cytosol of infected splenocytes with no vacuolar membrane around them. In contrast, llo– bacteria were always localized within vacuoles in all cell types including neutrophils and monocytes/macrophages that accounted for the vast majority of infected cells at that time. Analysing more than 200 distinct cells from mice infected with wt bacteria showed that 12% were localized in the cytoplasm 1 h after infection (Fig. 5B), a result in agreement with those reported in vitro (Tilney and Portnoy, 1989). This proportion increased to 24% and 51% 3 and 6 h after infection respectively. In contrast to llo– bacteria, secA2–, actA– and plcA/B– bacteria escaped primary phagocytic vacuoles and were localized in the cytosol 6 h after inoculation (Fig. 5B).
SecA2 ATPase deficiency impairs the secretion of a subset of bacterial proteins that are exported to the cytosol of infected cells (Lenz and Portnoy, 2002). Therefore, we have investigated whether secretion of LLO was altered in secA2– as compared with wt bacteria (Fig. S4). Pellet, culture supernatant and SDS-soluble fraction of log-phase grown bacteria were analysed by Western blot using anti-LLO and anti-SecA1 rabbit sera. Both LLO and SecA1 were expressed at similar levels in wt and secA2– bacteria respectively.
We also found that secA2–, actA– and plcA/B– mutants exhibited similar growth curves in vivo with a rapid increase in bacterial numbers during the first 12–15 h followed by a rapid decrease thereafter (Fig. S5). Similar data were obtained when mice were given gentamicin over the course of this experiment demonstrating that the impaired ability of secA2– bacteria to induce a potent protective secondary response is unlikely to result from an impaired persistence of secA2– bacteria inside infected cells.
Taken together, our results demonstrate (i) that cytosol invasion is not sufficient for the development of protective immunity to secondary L. monocytogenes infection and (ii) that the SecA2 ATPase is required for successful induction of long-term protective immunity against L. monocytogenes.
Our study provides two main findings on which steps of the L. monocytogenes infectious cycle are critical for the development of antibacterial protective secondary responses: we show that cytosolic invasive infections, although required, are not sufficient, and that expression of the SecA2 ATPase by cytoplasmic L. monocytogenes is necessary for long-term, CD8+ T cell-mediated protection.
For most intracellular bacteria, previous reports have demonstrated that establishment of a successful cytosolic invasive infection is a prerequisite for inducing protective immunity. In the case of L. monocytogenes, escape and multiplication of the bacteria inside the cytosol was believed to be sufficient for inducing protective T cell-mediated memory (Berche et al., 1987). Our current results reveal an additional step in this process that depends on the cytosolic expression of the SecA2 auxiliary secretion system.
In Gram-negative pathogens, the SecA ATPase belongs to the general secretory pathway and provides translocating energy for a set of bacterial-derived proteins. An auxiliary SecA2 protein secretion system was identified in several major Gram-positive pathogenic bacteria, including L. monocytogenes, M. tuberculosis and Bacillus anthracis (Braunstein et al., 2001; Lenz and Portnoy, 2002). These Gram-positive bacteria contain at least two SecA genes, SecA1 and SecA2 (Lenz and Portnoy, 2002) that have been associated with the virulence and the ability of these bacteria to cause disease (Braunstein et al., 2003; Lenz et al., 2003). In both L. monocytogenes and M. tuberculosis, SecA2 promotes pathogenesis by modulating the secretion of virulence proteins such as autolytic enzymes and superoxide dismutases (Braunstein et al., 2003; Lenz et al., 2003). SecA2 is a soluble protein that does not exhibit any transmembrane domain (Sharma et al., 2003). Therefore, it is likely that its effect on the development of protective immunity involves one or several of the SecA2-secreted substrate proteins. As in the case of lipopolysaccharides, it is conceivable that SecA2-dependent proteins contain evolutionary conserved pathogen-associated molecular patterns. For instance, the p60 and N-acetylmuramidase autolytic enzymes of L. monocytogenes, which secretion is largely dependent upon SecA2, digest bacteria walls and release muramyl peptides into host cell cytoplasm. These peptides may further bind NOD intracellular pattern recognition receptors (Chamaillard et al., 2003; Girardin et al., 2003a,b) that could trigger adequate signalling pathways inside DC for protective T cell priming (Iwasaki and Medzhitov, 2004).
It is noteworthy that a dedicated secretion system found in several Gram-positive pathogenic bacteria strongly impacts the development of protective T cell memory against L. monocytogenes. Recent studies have suggested that ribosomal proteins and superoxide dismutase are the only conserved SecA2-dependent substrates translocated in both L. monocytogenes and M. tuberculosis (Braunstein et al., 2003; Archambaud et al., 2006). It remains to be determined if any of the other SecA2-secreted proteins may share common structural or sequence features and whether our finding can be generalized to other pathogenic Gram-positive bacteria.
Finally, although CD8+ T cells from mice infected with secA2– bacteria are cytolytic and secrete IFN-γ, expression of these effector functions is not sufficient to protect against a secondary bacterial infection. This result is relevant in the context of which effector functions should be measured to predict vaccine efficacy.
Mice and bacteria
Six- to 8-week-old BALB/c mice were used in all experiments (Janvier). Stocks of wt 10403s and its mutant derivatives secA2–, actA–, llo–, plcA/B–, were prepared from isolated bacterial clones selected out of the spleen of infected mice twice. Selected bacteria were frozen at −80°C after growth to log-phase in Broth Heart Infusion medium (BHI, Sigma). HKLM were prepared as previously described (Lauvau et al., 2001).
Determination of bacteria titers inside the spleen, the liver or flow-purified cells from infected mice
Organs dissociated on metal screens or flow-purified cells were recovered in water/0.1% X-100 triton (Sigma) and further plated onto BHI media plates. Successive serial dilutions were performed in this buffer to determine bacterial cfu.
Quantification of LLO synthesis
Bacteria were grown to log phase in LB broth, then fractionated as previously described (Lenz and Portnoy, 2002). Briefly, bacteria were pelleted by centrifugation, and supernatant proteins were precipitated with TCA. Whole cell lysates were prepared by incubating TCA-treated bacteria with lysozyme to digest the cell wall. Protein fractions were dissolved in SDS running buffer (62.5 mM Tris-Cl, pH 6.8, 2% SDS, 10% glycerol, 5% beta-mercaptoethanol, 0.1% bromophenol blue) and used for SDS-PAGE. Bacterial pellets, supernatant or SDS-soluble fractions were analysed by Western blotting using an antiserum against LLO. We have typically loaded 10 μl per lane which correspond to the amount of LLO produced by 109 log phase bacteria of each type.
Generation of GFP-expressing bacteria
Five micrograms of PEG-purified pNF8 (Fortineau et al., 2000) was electroporated into wt 10304s or llo–, actA– or secA2– bacteria. Transfected bacteria were further selected onto BHI plates with 10 μg ml−1 erythromycin and checked their GFP expression using a FACScalibur cytofluorometer.
Confocal tissue sections
Spleens were fixed in 0.1 M phosphate buffer/1% paraformaldehyde (PFA) at 4°C for 3 h and further dehydrated in this buffer completed with 30% sucrose overnight. Then, organs were frozen in OCT (DAKO) on dry ice and 15 μm spleen sections (Cryomicrotome Leica) were stained with alexa647-phalloidin (Molecular Probe). Sections were further analysed by confocal microscopy on a Leica laser scanning confocal microscope equipped with a DM-IRBE inverted microscope and an Argon-Krypton laser.
Preparation of organ suspensions
Organs were cut in very small pieces and incubated (37C, 30 min) into HBSS (Gibco) medium containing 4000 U ml−1 of collagenase IV (Roche) and 0.1 mg ml−1 of DNase (Sigma). Red blood cells were lysed for 2–3 min in 0.17 M NH4Cl/0.017 M Tris pH 7.4.
A total of 3–5 × 106 cells were stained using directly coupled fluorescent mAb combinations in 100 μl PBS 0.5% BSA 0.02% NaN3 (FACS buffer) and further collected on a FACScalibur cytofluorometer (Becton Dickinson, BD). The following mAbs were purchased from BD Pharmingen: anti-CD8α (53-6.7)-fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyll protein or allophycocyanin (APC), anti-CD3ε-FITC (145-2C11), anti-CD62L-APC (MEL-14), anti-CD11c-PE (HL3), anti-CD11b-PE or APC (M1/70), anti-Gr1-APC (Ly-6G), anti-B220-PE (RA3-6B2), anti-CD49b-PE (DX5), anti-I-Ad-PE (AMS-32.1), anti-Ly-6C-FITC (AL-21). Anti-Granzyme B-APC (mouse and human) and Anti-F4/80-FITC were purchased from Caltag (TebuBio). For flow-cell purifications, similar cell-staining conditions were used and cells were sorted on a FACSvantage SE cell sorter (Becton Dickinson). H2-Kd/LLO91−99 tetramers coupled to PE were obtained from the NIH tetramer core facility.
Intracellular cytokine staining
Splenocytes were incubated for 4–6 h in RPMI1640 5% FCS with 2 μg ml−1 Golgi Plug (BD Pharmingen) at 37°C, 5% CO2, except for granzyme B staining. Cells were further washed in FACS buffer, stained for cell surface markers before fixation in PBS/1% PFA for 15–20 min on ice and permeabilized for 30 min using a saponin-based buffer (1X Perm/Wash, BD Pharmingen in FACS buffer) containing the intracellular staining mAbs (anti-IFN-γ-APC XMG1-2, BD Pharmingen). After final fixation in PBS/1% PFA, cells were analysed on a FACScalibur cytofluorometer. No signal was detectable with the isotype control (rat IgG1-APC, BD Pharmingen).
EM sample preparation
Flow-sorted cells were primarily fixed with 1.6% glutaraldehyde in 0.1 M phosphate buffer pH 7.5, then at 4°C with 1% osmium tetroxyde in 0.1 M cacodylate buffer pH 7.5. Cell pellets were further washed with distilled water and incubated for 2 h with 0.5% water/uranyl acetate at room temperature in the dark. After several more washes in water, pellets were dehydrated in increasing acetone series and embedded in epoxy resin. Blocks were then conventionally thin-sectioned (80 nm thick) and contrasted for the observation on a Philips CM12 electron microscope.
We thank D. Portnoy (UCSF, USA) for the L. monocytogenes-targeted deletion mutants. Supported by INSERM (Avenir), Human Frontier Science Program (G.L. was awarded a Carrier Development Award), Ministère de l'Education Nationale, de la Recherche et de la Technologie (MENRT), Agence National pour la Recherche (ANR IRAP-2005) and Fondation pour la Recherche Médicale (FRM). E.M. received a FRM fellowship and E.N.M. a MENRT fellowship. H2-Kd/LLO91−99 tetramers were obtained through the NIH Tetramer Facility.