Gene transfer in vitro from intracellular bacteria to mammalian phagocytic and non-phagocytic cells and in vivo in mice has been reported. The bacteria used as DNA delivery vectors were engineered to lyze upon entry in the cell due to impaired cell wall synthesis for Shigella flexneri and invasive Escherichia coli, or production of a phage lysin for Listeria mono- cytogenes. In vivo gene transfer was obtained with attenuated Salmonella typhimurium and resulted in stimulation of mucosal immunity. We report that wild-type intracellular human pathogens, such as L. monocytogenes EGD or LO28 and S. flexneri M90T, mediate efficient in vitro transfer of functional genes into epithelial and macrophage cell lines. A low- efficiency transfer was obtained from strain EGD to mouse peritoneal macrophages. DNA transfer with S. typhimurium was observed only from atten-uated aroA strain SL7207 into COS-1 cell line. As demonstrated by the study of listeriolysin-defective L. monocytogenes or of S. typhimurium SL7207 aroA engineered to secrete listeriolysin, escape of bacteria or of plasmid DNA from the intracytoplasmic vacuole is required for transfer of genetic information to occur.
Several bacteria, such as Shigella, Listeria, Salmonella and invasive Escherichia coli have been shown to act as gene delivery vectors in both phagocytic and non-phagocytic mammalian cells (Grillot-Courvalin et al., 1999). All the bacterial species that have been used to transfer genes to professional and non-professional phagocytes are facultative intracellular pathogens that have been designed to lyze after cell invasion. The plasmid DNA released by the attenuated intracellular bacteria is transferred from the cytoplasm to the nucleus resulting in cellular expression of the transfected genes.
These intracellular bacteria display very different invasion and survival strategies within the host cells (Finlay and Falkow, 1997). Following cell internalization, Listeria monocytogenes rapidly escapes from the phagolysosomes and multiplies in the cytosol; listeriolysin O (LLO), coded for by the hly gene, is a pore-forming cytolysin which plays a major role in lysis of the phagosomal membrane, allowing bacterial access to the cytosol (Cossart and Lecuit, 1998). Similarly, following uptake by host cells, Shigella rapidly escapes from the phagocytic vacuole after lysis of the membrane, multiplies and spreads in the cytoplasm of the cell and to adjacent cells. This invasive phenotype is conferred to S. flexneri by a 220-kb virulence plasmid which is responsible for entry, intracellular mobility and cell-to-cell spread (Sansonetti et al., 1982). Salmonella invades phagocytic and-non-phagocytic cells in a way similar to that used by Shigella. However, and as opposed to Shigella and Listeria, Salmonella remains in the primary phagosomal vacuole and divides within this compartment (Garcia-del Portillo and Finlay, 1995).
Despite differences in their intracellular pathways, attenuated mutants belonging to these three bacterial genera have been shown to transfer functional DNA into mammalian cells. We and others have shown that mammalian cells can express genes delivered intracellularly by strains of S. flexneri impaired in cell wall synthesis due to diaminopimelate (dap) auxotrophy (Courvalin et al., 1995; Sizemore et al., 1995); direct gene transfer was monitored by screening for production of nuclear β-galactosidase. Transfer of the virulence plasmid of S. flexneri to E. coli confers the ability to enter epithelial cells on this otherwise extracellular bacterial species. With such an invasive E. coli rendered dap auxotroph, strain BM2710, we have shown that bacteria that undergo lysis upon entry into mammalian cells can deliver plasmid DNA to their hosts (Courvalin et al., 1995).
In order to design a genetically better defined system we have cloned the 3.2 kb inv locus encoding the invasin of Yersinia pseudotuberculosis, alone or with the 1.5 kb hly gene coding for listeriolysin O from L. monocytogenes, in the stable dap auxotroph E. coli BM2710 (Grillot- Courvalin et al., 1998). Very efficient functional gene transfer was observed in vitro in several cell lines. An attenuated L. monocytogenes, impaired in intra- and intercellular movements and able to undergo self-destruction in the cell cytosol by production of a phage lysin could deliver functional plasmids in mouse macrophage cell lines (Dietrich et al., 1998). Following lysis by antibiotic treatment, L. monocytogenes has recently been shown to mediate plasmid delivery in a variety of cell lines but not in macrophage cell types (Hense et al., 2001). Gene transfer was observed in vitro and in vivo after oral administration of an attenuated strain of S. typhimurium that resulted in genetic immunization (Darji et al., 1997; Paglia et al., 1998).
We show here that wild-type intracellular pathogens, such as Listeria or Shigella, can also deliver functional genes into their host cells. The transfer efficiency of strains of Listeria, Shigella and Salmonella was compared using optimal infection conditions in different epithelial or macrophage cell types. Studies with Listeria mutants defective in escape from the primary vacuole or in intercellular mobility, as well as Salmonella derivatives that can escape from the vacuole of entry, have highlighted some of the requirements for gene transfer from bacteria to mammalian cells.
Wild-type intracellular pathogens can deliver functional genes to mammalian cells
Fully virulent wild-type L. monocytogenes EGD was the most efficient in gene delivery (Fig. 1 and Table 1). Following co-incubation for 2 h, up to 10% of HeLa and COS-1 cells producing the green fluorescent protein (EGFP) were detected 24 h post infection (Fig. 1 and Table 1). Similar results were obtained with another wild type strain of L. monocytogenes, L028 (Table 1). The J774 murine macrophage cell line was transfected, albeit at lower efficiency, by L. monocytogenes EGD. Low efficiency transfection was observed with normal peritoneal mouse macrophages only 48 h post invasion and a very low MOI (5:1) was optimal for this cell type; higher MOI resulting in rapid cell death. HepG2 (Fig. 1 and Table 1) and Caco-2 cell invasion with L. monocytogenes EGD resulted in EGFP expression in about 1.4 and 1.2% of the cells respectively. With Listeria, in vitro gene transfer was documented only at 24 h after cell invasion; beyond this period, intracellular multiplication of bacteria resulted in important cell death. In these experiments, no antibiotics aimed at the lysis of intracellular Listeria were added during cell culture. Of note, when tetracycline was added to the cells at the end of the bacterial invasion period (as described by Hense et al., 2001) the percentage of transfected HeLa cells dropped from 9.2 ± 0.6 to 4.9 ± 0.4. Enumeration of intracellular bacteria (data not shown) confirmed that this bacteriostatic antibiotic significantly slowed bacterial growth and thus, also likely, bacterial lysis which is required for gene transfer to occur.
Table 1. Transfer of genetic material from bacteria to mammalian cells.
. Flow cytometric analysis was performed 24 h after cell invasion except for peritoneal macrophages where it was carried out after 48 h. Mean values and standard deviations are calculated from the data obtained in a minimum of three independent experiments.
. Listeria monocytogenes were grown until late logarithmic (LL) or middle logarithmic (ML) phase. Strain EGD and derivatives in LL were added at a MOI of 15 for HeLa, 30 for COS-1, 75 for J774 (in ML for Æhly2) and in ML at a MOI of 50 for HepG2 cells; LO28 and LO28 hly were added in LL at a MOI of 50 for HeLa, and 100 for COS-1 cells.
. ND, not done.
. Shigella flexneri were grown in ML phase and added at a MOI of 250 for HeLa, HepG2 and J774, and 500 for COS-1 cells.
. Salmonella typhimurium C52 and 14028s were grown in LL phase and added at a MOI of 50 for HeLa, HepG2 and J774, and 100 for
COS-1 cells. Salmonella typhimurium SL7207 was grown in ML phase and added at a MOI of 250 for HeLa, HepG2 and J774, and 500 for
Under conditions optimal for in vitro cell invasion by virulent S. flexneri M90T, i.e. high MOI (250:1–500:1) and plate centrifugation to allow contact between donor and recipient that yielded from 0.3 to 1 bacterium per cell 30 min post invasion depending on the cell type, EGFP positive cells were detected in all the cell lines (HeLa, HepG2, COS-1 and J774) studied 24 h post infection (Fig. 1 and Table 1). The efficiency of transfer was lower than that observed with L. monocytogenes strains. No gene transfer was detected in mouse peritoneal macrophages under conditions optimal for cell invasion.
Repeated attempts to transfer the gene for EGFP, under invasion conditions adjusted to obtain initial numbers of intracellular wild-type S. typhimurium strains C52 and 14028s varying from 3 to 20 per cell, were unsuccessful in all the cell lines studied and in normal peritoneal macrophages 24 and 48 h post invasion. Prolonged cell cultures were possible because of the very low cellular toxicity following cell invasion. Moderately efficient gene transfer was observed only with attenuated aroA S. typhimurium SL7207 in COS-1 cells (Table 1). No gene transfer (≤0.01%) from SL7207 to mouse peritoneal macrophages was observed after addition of tetracycline, as opposed to a previous report (Darji et al., 1997).
Under optimal conditions of lipofection (i.e. 1 μg of purified pAT18ΩEGFP DNA in 7.5 μl of lipofectamin for 1–2 × 105 plated cells) different transfection efficiencies were observed with the various cell lines: HeLa, 63%; COS-1, 38%; J774, 15%; and HepG2, 11.5% (Fig. 1). In comparison, the 3 × 106Listeria added per well to HeLa cells correspond to an initial input of 3 ng of plasmid DNA. This indicates that the efficiency transfer per initial molecule of plasmid achieved by bacteria is 50 times higher than that obtained by lipofection.
Efficient bacterial cell invasion is necessary but not sufficient for efficient gene transfer
Efficiency of bacterial cell invasion was determined by enumeration of viable intracellular bacteria after gentle cell lysis and was carried out after an additional 30 min incubation with gentamicin to kill the remaining extracellular bacteria or after 24 h of culture. The results represent the average number of bacteria per cell calculated from the number of bacteria recovered per well. When the initial number of L. monocytogenes was determined an average of 0.6 of a viable bacterium was recovered per HeLa cell. COS-1 and J774 cells were infected more efficiently with an average of four bacteria per cell 30 min post invasion. Depending on the cell line tested, a 10 (J774) to 100 (COS-1) or 200 (HeLa and HepG-2) fold increase in the number of viable intracellular bacteria per well was observed 24 h post infection.
The variable transfection efficiency observed with the different cell lines could result in part from heterogeneity in initial cell infection. The percentage of infected cells was determined by flow cytometric analysis of fluorescent cells after invasion with bacteria expressing GFP under the control of a prokaryotic promoter. When the initial number of cells infected with L. monocytogenes EGD harbouring pNF8 (pAT18ΩdLt-gfp-mut1) was determined 30 min post invasion, only 15% of HeLa cells contained GFP-positive bacteria (Fig. 2, panel A). After 24 h of culture, intracellular bacterial proliferation and spread to adjacent cells resulted in 80% of brightly fluorescent HeLa cells. COS-1 and J774 cells were infected more efficiently with, respectively, 59 and 54% fluorescent cells 30 min post invasion. Twenty-four hours after infection, more than 80% of the cells of these two cell lines were fluorescent, and cytofluorometric analysis revealed substantial heterogeneity in the fluorescence pattern that reflected variable numbers of intracellular bacteria per cell.
Although higher numbers of COS-1 and J774 than HeLa cells were initially infected with L. monocytogenes, similar numbers of cells expressing acquired EGFP were observed 24 h post infection in COS-1 and HeLa cells. On the contrary, despite a high number of initially infected cells, only moderate transfer was observed with J774 cells (Table 1).
After cell invasion with S. typhimurium SL7207/pAT505 (pUC18Ωgfp-mut1), 38% fluorescent HeLa cells, 72% COS-1 and 82% J774 cells were recovered 30 min post infection (Fig. 2, panel B). The cytofluorometric profiles of infected cells 24 h post infection showed a percentage of positive cells lower than that after 30 min but a more intense fluorescence signal, reflecting bacterial proliferation in the infected cells. Despite efficient invasion of all the cell lines tested, gene transfer was observed only with S. typhimurium SL7207 in the COS-1 cell line (Table 1).
Despite a lower number of initial bacteria per cell (0.4 versus 2 bacteria per cell 30 min after infection), S. flexneri M90T harbouring the eukaryotic expression vector pAT18ΩEGFP was more efficient in gene transfer in all the cell lines tested than the same strain harbouring pEGFP-C1 (Table 1). The intracellular survival of bacteria in HeLa cells after 24 h of cell culture was markedly different for S. flexneri/pAT18ΩEGFP (less than 5% of initial intracellular viable bacteria) than for S. flexneri/pEGFP-C1 (50% of viable bacteria). The reason for this difference is unclear but is consistent with that in growth patterns of bacteria harbouring either plasmid (data not shown).
Taken together, these results are consistent with the notion that the ability of intracellular bacteria to transfer genetic material to the host cell is not related to the efficiency of bacterial internalization.
Escape of bacteria or plasmid DNA from the primary vacuole is required for gene transfer to occur
We have previously shown that listeriolysin production markedly enhances the efficiency of gene transfer by invasive dap– E. coli BM2710 (Grillot-Courvalin et al., 1998). Mutants of L. monocytogenes have been constructed that produce either no listeriolysin following a chromosomal deletion (strain EGDΔhly2, Guzman et al. 1995) or a listeriolysin with very low haemolytic activity resulting from amino acid substitution Trp-492-Ala in the protein (strain L028 hly, Michel et al., 1990). Despite efficient initial bacterial internalization (data not shown), no gene transfer was observed when strain EGDΔhly2 was used as a gene delivery vector (Table 1). The low-level transfer observed with L. monocytogenes L028 hly (Table 1) is probably due to the low level of listeriolysin activity in this mutant.
L. monocytogenes mutant EGDΔplcB2 is impaired in intercellular mobility and gene transfer efficiency with this strain was much lower than that obtained with wild- type strain EGD in all cell lines tested, except HepG2 (Table 1).
With dap+ parental E. coli BM2711 expressing the listeriolysin in a non-secreted form (encoded by plasmid pGB2Ωinv-hly) or in a form secreted by the E. coli haemolysin export system [coded for by pAT724 (pACYC184ΩhlyR, C, hly-hlyAs, B, D)] a moderate enhancement of gene transfer was observed for COS-1 cells only (Table 2).
Table 2. Consequence of listeriolysin production on DNA transfer from bacteria to mammalian cells.
Plasmid pAT724 was introduced into S. typhimurium strain SL7207 harbouring pEGFP-C1. This resulted in the detection of red blood cell haemolytic activity associated with bacteria in amounts similar to those published by Genstchev et al., 1995) for Salmonella dublin aroA/ pILH1(pBR322Ωhly, R, C, hly hlyAs, B, D) (data not shown). S. typhimurium SL7207 harbouring pAT724 led to a more efficient gene transfer in COS-1 cells than SL270/pAT723 (pACYC184Ωhly-hlyAs, B, D) containing a deleted gene for listeriolysin (0.25 versus 0.03%, Table 2). However, no gene transfer was observed with this strain in J774 and HeLa cell lines.
Altogether, these results indicate that listeriolysin plays a major role in the passage of plasmid DNA from the initial vacuole of internalization to the cytosol of the cell.
We have expanded our initial finding that attenuated invasive intracellular E. coli are able to deliver genes into mammalian cells (Grillot-Courvalin et al., 1998) by demonstrating that certain wild-type intracellular patho-gens such as Shigella and Listeria can also transfer functional genetic information to infected host cells.
Among the facultative intracellular pathogens studied, L. monocytogenes was the most efficient at gene delivery (Table 1). This bacterial species, once into phagocytic or non-phagocytic cells, rapidly escapes from the host phagosome after lysis of the membrane by listeriolysin O, multiplies in the cytosol and spreads from cell to cell (Cossart and Lecuit, 1998). The efficiency of gene transfer achieved by two wild-type strains of L. monocytogenes was high in epithelial cell types and moderate in murine macrophage cell lines (Table 1). The results with the macrophage cell lines are similar to those obtained with a L. monocytogenes impaired in intra- and intercellular movements and designed to undergo self-destruction in the cytosol by production of a phage lysin (Dietrich et al., 1998). Suicide or antibiotic-mediated killing of the intracytoplasmic bacteria was not required in our experiments to achieve highly efficient gene transfer (Table 1).
We observed that only a fraction of the HeLa, COS-1 or J774 cells was initially infected by L. monocytogenes but that subsequent passage from cell to cell resulted in a rapid increase in the number of infected cells, as shown 24 h post infection (Fig. 2). Consistent with this notion, lower gene transfer was obtained with L. monocytogenes EGDΔplcB2 impaired in intercellular mobility. These observations could explain why a higher efficiency of gene transfer was obtained with fully virulent Listeria than with the attenuated suicidal strain (Table 1) unable to spread from cell to cell (Dietrich et al., 1998) or when antibiotic treatment was added to the culture (Hense et al., 2001 and our data).
S. flexneri is another facultative invasive bacterial species which, soon after entry into epithelial cells, lyzes the endocytic vacuole leading to bacterial multiplication in the cytoplasm and subsequent escape from the cell or cell death. The efficiency of DNA transfer by virulent S. flexneri M90T (Table 1) was similar to that reported initially using an attenuated S. flexneri deficient in cell wall synthesis (Courvalin et al., 1995; Sizemore et al., 1995). With this species, the low initial number of cells infected in vitro is followed by a decrease in the number of viable intracellular bacteria recovered after cell culture. Cell death, in particular by apoptosis in macrophage cell type, occurs rapidly after invasion (Zychlinsky et al., 1992). Of note, rapid intracellular bacterial death was optimum for efficient DNA transfer (Table 1). It is therefore all the most remarkable that, in vivo, an attenuated Shigella was able, after oral or intranasal administration, to deliver plasmids encoding β-galactosidase (Sizemore et al., 1995) or measle antigens (Fennelly et al., 1999) and to stimulate an efficient mucosal immune response against them.
In contrast to wild-type Listeria and Shigella which were more, or at least as, efficient than their attenuated forms in gene transfer to mammalian cells, the wild-type Salmonella tested, S. typhimurium C52 and 14028s, were inefficient with every cell line studied or with normal peritoneal macrophages 24 and 48 h post invasion (Table 1). Moderately efficient gene transfer was only observed in vitro with attenuated S. typhimurium aroA strain SL7207 in COS-1 cells. Salmonella, in both epithelial cells such as HeLa or in macrophages, remains in the primary vacuole and replicates within this compartment (Méresse et al., 1999). However, whereas a-non-invasive strain of S. typhimurium was shown to reside in a similar intracellular compartment in murine macrophages, the heat-killed isogenic bacteria were processed through the normal degradation pathway (Rathman et al., 1997). The intracellular pathway of attenuated S. typhimurium SL7207 aroA, which in our experiments was able to achieve some gene transfer, has not been extensively studied.
The importance of the escape of bacteria, or of plasmid DNA, from the vacuole of internalization for transfer of genetic information to occur was clearly illustrated by the study of listeriolysin O defective Listeria derivatives. With these mutant strains, no or very reduced gene transfer was obtained (Hense et al., 2001 and Table 1). Total lack of transfer by deletion mutant EGDΔhly2 (Guzman et al., 1995) was observed with most cell lines tested whereas mutant L028hly, which produces a listeriolysin with reduced lysin activity (Michel et al., 1990), induced moderate transfer in the two cell lines tested. Invasive dap–E. coli BM2710/pGB2Ωinv-hly, designed as a gene delivery vector, expresses listeriolysin O in a non-secreted form; listeriolysin is released in the vacuole only after lysis of the bacteria, leading to pore formation in the phagosomal membrane and a markedly enhanced gene transfer (Grillot-Courvalin et al., 1998). A similar observation was made with the dap+ parental E. coli BM2711 (Table 1). Transfer efficiency to COS-1 cells was increased when listeriolysin was produced via the secretion system of E. coli haemolysin or even in a-non-secreted form (Table 2). Secretion of listeriolysin by Salmonella results in a change in its intracellular localization, with escape from the phagosome into the cytoplasm of J774 macrophages (Genstchev et al., 1995). Secretion by S. typhimurium SL7207 of listeriolysin resulted in a significant increase in gene transfer to COS-1 cells but not to the other cell lines tested (Table 2). Listeriolysin has proven to be an efficient way to provide access to cell cytosol for protein antigens, attenuated bacterial vaccine strains and DNA vaccines, particularly in professional antigen-presenting cells (reviewed in Dietrich et al., 2001).
Gene transfer efficiency was markedly dependent on the type of recipient cell (Table 1). In general, higher transfer was achieved in epithelial cells than in macrophages. This difference did not correlate with the initial efficacy of bacterial cell invasion which was high in both cell types. Of note, transfection efficiency by lipofection was also higher in epithelial than in macrophage cell types. In our experiments, moderate transfer was observed with Listeria and Shigella in the mouse macrophage cell line J774, as opposed to what was reported with Listeria by Hense et al. (2001). However, no gene transfer to macrophages was observed in vitro with wild-type S. typhimurium or attenuated strain SL7207. Gene transfer with Salmonella has been mainly observed in vivo, following oral administration of the bacteria, resulting in stimulation of an immune response (Darji et al., 2000). Macrophages are thought to be important in the dissemination of Salmonella to central lymphoid organs but dendritic cells may also play a role (Hopkins et al., 2000). No gene transfer was obtained in a mouse dendritic cell line with an invasive E. coli vector (Grillot-Courvalin et al., 1998) or with L. monocytogenes in conditions of optimal cell internalization (data not shown). Attempts to transfer DNA from the three bacterial species used in this study to human peripheral blood monocyte-derived dendritic Langerhans cells, as described by Geissman et al. (1998) were also unsuccessful (data not shown). A more sensitive detection system for gene transfer in this-non-dividing cell type may be necessary.
The results presented in this study indicate that access to the cell cytosol of the bacteria, or of the plasmid, is an important requirement for successful gene transfer, in particular since DNA has to be released into the cytosol before nuclear entry can occur. Listeriolysin has proven to be a major tool to achieve this prerequisite.
In a previous report, we suggested that DNA transfer from bacteria to mammalian cells takes place in nature and that comparison of genomic sequences may confirm acquisition of bacterial genes by mammalian cells (Grillot-Courvalin et al., 1998). It is therefore all the more interesting that comparative analysis of the human genome sequence led to the proposal that certain human genes are probably of bacterial origin and that the most likely donor candidates are human intracellular bacterial pathogens (Lander et al., 2001). Our observations are consistent with, and could account for, the presence of genes from pathogenic bacteria in the human genome, a finding which, however, remains controversial (Salzberg et al., 2001).
Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in Table 3. Escherichia coli TOP10 (Grant et al., 1990) was used as a recipient for plasmid constructions. The 2569 bp AseI (blunt ended)-StuI fragment of pEGFP-C1 (Clontech) was ligated in the Gram-positive/Gram-negative shuttle plasmid pAT18 (Trieu- Cuot et al., 1991) digested by SmaI to generate pAT18ΩEGFP (pAT18ΩPCMV-egfp) or in pBR322 digested by EcoRV-NruI to give pBR322ΩEGFP (pBR322ΩPCMV-egfp).
Table 3. Bacterial strains and plasmids used.
Source or reference
Ap, ampicillin; Cm, chloramphenicol; Em, erythromycin; Km, kanamycin; Sp, spectinomycin; Sm, streptomycin; Tc, tetracycline; R, resistant; S, susceptible.
The insert in pILH1 (hlyR, C, hly-hlyAs, B, D) directs the production and secretion of listeriolysin via the haemolysin transport machinery of E. coli (Genstchev et al. 1995). The insert was cloned in pACYC184 (Chang and Cohen, 1978) in two steps. Firstly, the 6185 bp ClaI-SalI fragment was cloned in the vector digested with the same enzymes leading to pAT723 (pACYC184ΩΔhly-hlyAs, B, D). Secondly, the 3224 bp ClaI fragment containing the 5′ region of hly-hlyAs was cloned in ClaI-digested pAT723 to give rise to pAT724 (pACYC184ΩhlyR, C, hly-hlyAs, B, D).
Plasmid DNA was introduced into bacterial strains by electrotransformation using a Bio-Rad apparatus. Bacteria were grown in brain heart infusion (BHI, Difco Laboratories, Detroit, MI) broth or agar. Shigella was isolated on trypticase soy (TCS) agar plates containing 0.01% Congo red to test for the presence of virulence plasmid pWR100. Salmonella was grown in 2X yeast tryptone (2YT) medium.
Bacterial cell invasion
Cell lines and culture conditions. The cell lines studied were HeLa (human cervix epitheloid carcinoma), HepG2 (human hepatocellular carcinoma), COS-1 (monkey African green kidney, SV40 transformed) and J774 (mouse monocyte-macrophage). Cells were maintained in complete culture Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 2 mM L-glutamine and 10% fetal calf serum (FCS, Myoclone +, Gibco-BRL, Grand Island, NY).
1 × 105 COS-1 or 2 × 105 HeLa, HepG2 or J774 cells were incubated overnight in six-well plates (Costar, Cambridge, MA). Cells were incubated after bacterial invasion with the same medium containing 20% FCS and gentamicin.
Infection of cell lines. The bacterial strains were grown at 37°C, except E. coli BM2711 and derivatives which were grown at 30°C, harvested by centrifugation in middle (ML, OD600 between 0.8 and 1.6) or late (LL, OD600 between 3 and 4) logarithmic phase of growth and suspended in DMEM. Bacteria at 1.5 × 106– 2.5 × 107 ml–1 were added to mammalian cells in DMEM at a MOI of 10–500 and the plates were incubated for 1 or 2 h at 37°C. For Shigella M90T, the plates were centrifuged for 10 min at 2000 rpm before incubation. The cells were then washed three times with DMEM and re-incubated in complete medium containing 20–50 μg ml–1 of gentamicin.
E. coli BM2711 harbouring plasmids pGB2Ωinv + pEGFP-C1 or pGB2Ωinv-hly + pEGFP-C1 was grown in BHI broth con-taining 25 μg ml–1 of spectinomycin and of kanamycin. If the strain also contained pAT724, 25 μg ml–1 of chloramphenicol were added.
The Listeria strains harbouring pAT18ΩEGFP or pNF8 were grown in BHI broth with erythromycin at 8 μg ml–1.
S. flexneri M90T harbouring pAT18ΩEGFP or pEGFP-C1 was grown in BHI broth supplemented with erythromycin 200 μg ml–1 or kanamycin 50 μg ml–1, respectively.
Most likely because of its high copy number, plasmid pEGFP-C1 (pUCΩPCMV-egfp) appeared unstable in S. typhimurium except in strain SL7207. Thus, S. typhimurium C52 and 14028s were transformed with pBR322ΩEGFP DNA and grown in BHI containing 100 μg ml–1 of ampicillin. Auxotrophic aroA S. typhimurium SL7207 harbouring pEGFP-C1 was grown in 2YT medium with kanamycin, 50 μg ml–1. When the strains also contained pAT723 or pAT724, chloramphenicol at 25 μg ml–1, was added. Strain SL7207 harbouring pAT505 was grown in the presence of ampicillin, 100 μg ml–1.
Infection of peritoneal mouse macrophages. Approximately 2 × 106 peritoneal macrophages from BALB/c mice in DMEM containing 5% FCS and 2 mM L-Glu were allowed to adhere for 2 h at 37°C in a six-well plate. After removal of the-non-adherent cells by washing with DMEM, bacteria were added (at an MOI of 5 for L. monocytogenes EGD and derivatives, 10–50 for S. flexneri M90T and 5–25 for S. typhimurium strains), the plates were centrifuged for 10 min at 2000 rpm in experiments with M90T, incubated for 1 h, washed with DMEM and incubated in complete medium containing 10 μg ml–1 of gentamicin.
Cells were incubated in culture medium without FCS with 1 μg of purified plasmid pAT18ΩEGFP mixed with 7.5 μl of lipofectamin (Gibco BRL) for 5 h at 37°C. The transfection medium was then replaced by complete culture medium. EGFP expression was assessed after 24 h.
Numeration of viable internalised bacteria
After cell invasion, the cells were incubated for 30 min at 37°C in complete culture medium containing 20–50 μg ml–1 gentamicin to kill extracellular bacteria and washed three times with DMEM. The bacteria were released from the cells with 0.2% Triton and viable bacterial counts were determined on BHI agar plates containing the appropriate antibiotics. Results are expressed as the number of viable bacteria per well divided by the number of cells per well.
Flow cytometric analysis
GFP-producing bacteria. L. monocytogenes EGD, with or without pNF8, was grown overnight at 37°C in BHI, and S. typhimurium SL7207, with or without pAT505, was grown for 3–4 h at 37°C in 2YT. Bacteria from 1 ml cultures were centrifuged, washed twice in PBS and fixed with 0.5% paraformaldehyde for 10 min at 4°C. The fluorescence of 5 × 104 bacteria at ca. 107 ml–1 was analyzed by flow cytometry using a FACScan flow cytometer with CellQuest software (Becton-Dickinson, Mountain View, CA)
Cells. One, 24 or 48 h after invasion, the cells were trypsinized, washed once with 2% FCS in phosphate-buffered saline (PBS), resuspended in the same medium at 106 cells ml–1 and 3 × 104 cells were analysed.
Determination of listeriolysin activity
Strains were grown to ML phase in 2YT medium and listeriolysin activity was assayed with 100 μl of whole-cell suspension and 900 μl of 2.5% suspension of sheep erythrocytes diluted in PBS (pH 6) for 45 min at 37°C. Intact erythrocytes were removed by centrifugation. The haemolytic activity, relative to that obtained with 1% Triton, was determined from the release of haemoglobin as measured at 543 nm.
We thank A. I. Chakraborty, P. Cossart and B. A. Stocker for their gift of strains, N. Fortineau for gift of plasmid pNF8 and F. Huetz for the gift of mouse macrophages. This work was supported in part by the French Cystic Fibrosis Foundation (AFLM) and by a Bristol-Myers Squibb Unrestricted Biomedical Research Grant in Infectious Diseases.