Induction of Necrosis in Human Neutrophils by Shigella flexneri ...

INFECTION AND IMMUNITY, Mar. 2000, p. 1289–1296 0019-9567/00/$04.00⫹0 Copyright © 2000, American Society for Microbiology. All Rights Reserved.

Vol. 68, No. 3

Induction of Necrosis in Human Neutrophils by Shigella flexneri Requires Type III Secretion, IpaB and IpaC Invasins, and Actin Polymerization ´ RONIQUE LE CABEC,1 MARIE-ANGE DUPONT,2 MATHIAS FRANC ¸ OIS,1 VE PHILIPPE J. SANSONETTI,3 AND ISABELLE MARIDONNEAU-PARINI1* Institut de Pharmacologie et de Biologie Structurale, CNRS UPR 9062,1 and Laboratoire de Biologie Molećulaire Eucaryote, CNRS UPR 9006,2 Toulouse, and Unite´ de Pathogeńie Microbienne Molećulaire, Institut Pasteur, Paris,3 France Received 24 September 1999/Returned for modification 29 October 1999/Accepted 6 December 1999

Infection by Shigella flexneri is characterized by infiltration of neutrophils in the intestinal mucosa and by a strong inflammatory reaction. Although neutrophils are constitutively programmed to die by apoptosis, we show that isolated human neutrophils undergo necrosis 2 h after infection with virulent S. flexneri strain M90T but not with the virulence plasmid-cured strain BS176. This was demonstrated by the release of azurophil granule proteins concomitant with the release of lactate dehydrogenase (LDH), disruption of the plasma membrane, and absence of DNA fragmentation. Mutants with the mxiD1 gene, coding for an essential component of the secretion type III machinery, or the genes coding for IpaB or IpaC invasins deleted were not cytotoxic. Neutrophil necrosis occurred independently of the bacterial ability to leave phagosomes, and it involved actin polymerization, as the addition of cytochalasin D after phagocytosis of Shigella inhibited the release of LDH. In conclusion, Shigella kills neutrophils by necrosis, a process characterized by the release of tissue-injurious granular proteins. This probably contributes to disruption of the epithelial barrier, leading to the dysentery observed in shigellosis and allowing Shigella to enter its host cells. Bacillary dysentery caused by Shigella is generally acquired by orofecal contact or ingestion of contaminated food and water. These facultative intracellular pathogens enter the colonic epithelium, where they multiply, disseminate, and cause inflammation, which leads to necrosis and destruction of the epithelia. This tissue damage accounts for the clinical manifestations of shigellosis, which is a severe form of bloody diarrhea (2, 22). It has been shown that the uptake of shigellae occurs primarily within cells present in the follicle-associated epithelium, also called M cells (12, 22, 37). In parallel, a strong mucosal infiltration of neutrophils, considered responsible for acute inflammation of the colon and mucosal destruction, is observed (2). As these bacteria are incapable of invading epithelial cells through the apical membrane, disruption of the epithelial barrier should facilitate the entry of shigellae into epithelial cells at the basolateral pole. It is therefore of interest to examine the relationships between Shigella and neutrophils which might generate a strong inflammatory reaction accounting for the clinical manifestations of dysentery. Neutrophils constitute the first line of host defense against microorganisms (6, 30, 32). Their bactericidal functions consist of recognition and phagocytosis of the invading microbes; activation of the O2⫺-producing enzyme NADPH oxidase, which leads to the formation of other reactive oxygen species; and release of their granular contents into phagosomes and the extracellular medium (6, 30, 32). Neutrophils possess (i) primary granules, also called azurophil granules, which are specialized lysosomes containing bactericidal proteins and the hypochlorous acid-generating enzyme myeloperoxidase in addition to classical lysosomal enzymes, and (ii) secondary

granules, which include the specific- and gelatinase granule subpopulations (4). Specific granules constitute a reservoir of plasma membrane proteins and also contain lactoferrin and elastase, which exert some bactericidal functions, while the gelatinase granules are mainly devoted to extravasation of neutrophils. Exocytosis of secondary granules can be triggered either by soluble activating agents or during phagocytosis of particles, while mobilization of azurophil granules is only triggered during phagocytosis (4, 13, 31, 33, 34). Azurophil granules are promptly mobilized and fuse, about 30 s after the addition of phagocytic particles, with nascent, unclosed phagosomes (31, 34), leading to the release of their matrix proteins into the extracellular medium during phagolysosome biogenesis (20). Release of azurophil granules also occurs when neutrophils undergo necrosis as the plasma membrane becomes leaky, but in contrast to stimulus-induced degranulation, it occurs with a delay and is accompanied by the release of cytosolic proteins, such as lactate dehydrogenase (LDH) (11, 28). Normally, neutrophils are programmed to die by apoptosis (11, 28), which, in contrast to necrosis, maintains the plasma membrane integrity, thereby avoiding inflammatory reactions due to the release of tissue-injurious granule contents (11, 28). The strong inflammatory reaction is a crucial step in shigellosis. It is probably mediated by neutrophils, but the molecular mechanisms have not been defined. Activation of azurophil granule exocytosis by Shigella flexneri has been reported, but it was obtained with a high multiplicity of infection (MOI) (1,000 bacteria per neutrophil) and was detectable only 2 h after infection (24). As mentioned above, the release of azurophil granule markers occurs either very rapidly when it results from phagolysosome formation or after a delay when it results from necrosis. This led us to suspect that the release of azurophil granules was not a bactericidal response but was rather part of a necrotic process. Therefore, we decided to further investigate the effect of Shigella on granule mobilization in human neu-

* Corresponding author. Mailing address: CNRS-IPBS, 205 Route de Narbonne, 31077 Toulouse, France. Phone: 33-561 14 54 58. Fax: 33-561 17 59 94. E-mail: [email protected]. 1289

1290

FRANC ¸ OIS ET AL.

INFECT. IMMUN.

trophils to establish whether their release reflects a bactericidal response or a necrotic process. MATERIALS AND METHODS Human neutrophils. Neutrophils were isolated from the blood of healthy donors, separated by the Dextran-Ficoll method as previously described (14), resuspended in minimal essential medium–10 mM HEPES (pH 7.4), and maintained for 20 min at 37°C prior to stimulation. Bacterial strains and growth conditions. All Shigella strains used in this study are derivatives of the wild-type strain M90T. BS176 is cured of the virulence plasmid, and SF620 (ipaB2), SF621 (ipaC2), and SF401 (mxiD1) have been previously described (1, 18). Bacteria were grown overnight at 37°C in tryptic soy broth, subcultured, washed in phosphate-buffered saline (PBS), and adjusted to the appropriate concentration in minimal essential medium-HEPES just before the infection of neutrophils. Opsonization of bacteria and zymosan. Zymosan (Sigma Chemical Co., St. Louis, Mo.) was incubated in pooled human sera for 20 min at 37°C, washed twice with PBS (pH 7.4), and resuspended in PBS supplemented by 1 mM CaCl2 and 0.5 mM MgCl2 (20). Prior to the opsonisation of bacteria, the pooled serum was heat inactivated at 56°C for 30 min, and the bacteria were opsonized for 20 min at 37°C, washed, and resuspended in PBS. Phagocytosis measurement. Neutrophils were allowed to adhere on glass coverslips as previously described (20). Bacteria were added and centrifuged on the neutrophils for 10 min at 150 ⫻ g at 37°C. Phagocytosis was carried out for 1 h at 37°C. Then the neutrophils were washed to remove unincorporated bacteria and fixed with 3.7% paraformaldehyde in PBS containing 15 mM sucrose, pH 7.4, for 30 min at room temperature. After neutralization with 50 mM NH4Cl, the slides were washed with PBS (pH 7.4) and incubated with monoclonal anti-lipopolysaccharide (LPS) antibodies against S. flexneri (IgC20; 1:500) and then revealed by rabbit anti-mouse tetramethyl rhodamine isothiocyanate antibodies to stain the remaining extracellular bacteria. The cells were then permeabilized with 0.5% Triton X-100 in PBS for 2 min at room temperature and washed in PBS, and intracellular and extracellular bacteria were revealed by sequential addition of rabbit anti-LPS (1:100) and goat anti-rabbit antibodies coupled to fluorescein isothiocyanate (FITC) (29). Under these conditions, the few extracellular bacteria were FITC and tetramethyl rhodamine isothiocyanate stained while intracellular bacteria were only FITC positive. Ingested bacteria were often seen in aggregates, and therefore it was impossible to count them precisely, but the number of bacteria ingested per neutrophil was approximately the same for the different strains. For each condition, at least 100 cells were counted using a Leitz DM fluorescent microscope, and the percentage of cells that had internalized ⱖ1 bacterium was calculated. In some experiments, phagocytosis was synchronized. Neutrophils were coincubated with bacteria at 4°C for 30 min and washed at 4°C to eliminate the bacteria which did not bind to the neutrophils, and phagocytosis was carried out by transferring the cells at 37°C. Electron microscopy. Neutrophils were fixed in glutaraldehyde (2.5% [vol/ vol]) in 0.1 M phosphate buffer, pH 7.2 (buffer A), for 1 h. After three washes in buffer A, the cells were postfixed with 1% osmium tetroxide in buffer A and then dehydrated in graded ethanol. The 100% ethanol solution was then replaced by propylene oxide and embedded in Epon 812. Sections were stained with uranyl acetate and lead citrate and examined with a Jeo1 1200EX electron microscope. Protein exocytosis. Control or stimulated neutrophils (5 ⫻ 106/ml) were pelleted, and the supernatants were centrifuged (10,000 ⫻ g for 10 min) to eliminate bacteria. The cell pellets were lysed overnight in 1% Triton X-100, and the cell supernatants were stored at ⫺20°C. Lactoferrin, a marker of specific granules, was measured by enzyme-linked immunosorbent assay (10), and the enzyme activity of ␤-glucuronidase, a marker of azurophil granules, was measured as previously described (38). LDH measurement. For quantification of cell cytolysis, release of the cytosolic enzyme LDH was measured using the colorimetric assay kit from Boehringer (Meylan, France) according to the manufacturer’s instructions. DNA extraction and gel electrophoresis. DNA extraction was performed essentially as previously described (5). Briefly, cells (106) were pelleted, washed in PBS, and lysed at room temperature (10 mM Tris-HCl [pH 8], 100 mM EDTA, 0.5% sodium dodecyl sulfate), and 0.1 mg of RNAse/ml (final concentration) was added for 15 min. Then, proteinase K (0.2-mg/ml final concentration) was added for 2 h. DNA was extracted twice with phenol-chloroform-isoamyl alcohol (25: 24:1). The DNA was precipitated with 0.1 volume of 3 M sodium acetate (pH 5.5) and 2 volumes of ethanol and recovered by centrifugation, electrophoresed on a 1% agarose gel, and stained with ethidium bromide.

RESULTS Infection of neutrophils with virulent and nonvirulent strains of S. flexneri rapidly triggers exocytosis of specific but not azurophil granules. Neutrophils were incubated with the invasive serotype 5 strain M90T and the noninvasive strain BS176 (cured of the 220-kb virulence plasmid) at different MOIs, and cells having internalized bacteria were counted. As

FIG. 1. Nonopsonic phagocytosis of S. flexneri induced exocytosis of specific but not azurophil granules. Phagocytosis of the virulent strain M90T and the virulence plasmid-cured strain BS176 was performed at different MOIs for 1 h, and the percentage of cells that had internalized at least one bacterium was determined. The release of lactoferrin (a marker of specific granules) by neutrophils exposed for 1 h to the indicated strains of Shigella (100 bacteria/cell) or by noninfected neutrophils (control) was measured by enzyme-linked immunosorbent assay. The release of ␤-glucuronidase (a marker of azurophil granules) was measured 1 or 2.5 h after infection with 100 bacteria/cell or no bacteria (control). OZ (50 particles/cell), which triggers exocytosis of specific and azurophil granules, was used as a positive control. Results are expressed as the means ⫾ standard deviations of three to seven experiments.

shown in Fig. 1, both strains were ingested with the same efficiency. This differs from entry into epithelial cells, which is only observed with the wild-type bacteria while the virulence plasmid-cured strain BS176 remains in the extracellular compartment (25). Therefore, neutrophil-mediated internalization of Shigella is independent of the virulence plasmid and is probably mediated by phagocytic receptors. Next, we examined whether phagocytosis of S. flexneri triggers neutrophil degranulation. Lactoferrin, a marker of specific granules, was secreted in the extracellular medium in response to both strains 1 h after infection at an MOI of 100 (Fig. 1). There was no difference between BS176 and the virulent strain M90T. In contrast to lactoferrin, there was no release of ␤-glucuronidase (Fig. 1), a marker of azurophil granules, even when a higher MOI was used (1,000 bacteria/ cell [data not shown]). In these experiments, phagocytosis of zymosan (wall particles from Saccharomyces cerevisiae) opsonized in human serum (OZ) was used as a positive control for exocytosis of ␤-glucuronidase (Fig. 1).

VOL. 68, 2000

S. FLEXNERI INDUCES NECROSIS IN HUMAN NEUTROPHILS

FIG. 2. Human neutrophils infected with virulent S. flexneri release LDH. The release of LDH was measured at the indicated time points after infection of neutrophils with nonopsonized M90T or BS176 (100 bacteria/cell). The results are expressed as the means ⫾ standard deviations of three experiments. (Inset) Release of LDH induced by M90T at the indicated MOIs.

Only the virulent strain induced the release of azurophil granule content after a long incubation period. Next, the effect of Shigella at later time points was studied. When experiments were carried out for 2.5 h, exocytosis of ␤-glucuronidase was strongly triggered by M90T but not by BS176 (Fig. 1). The effect of M90T, which did not trigger the release of azurophil granules 1 h after infection but after 2.5 h, contrasted with the results obtained with OZ. As expected for the time course of exocytotic events associated with phagocytosis of OZ in neutrophils (31, 34), the release of ␤-glucuronidase did not increase between 1 and 2.5 h (Fig. 1). The delay for release of ␤-glucuronidase induced by the wild-type strain suggests that it was not the result of phagolysosome biogenesis but more likely was the result of a cytotoxic process, as stated in the introduction. To support this hypothesis, the release of lactoferrin after 2.5 h was more important when cells were infected with M90T (34% at 2.5 h versus 12.5% at 1 h) than with BS176 (22 versus 11.5%) (means of two separate experiments). Phagocytosis of virulent S. flexneri induces necrosis of neutrophils. The possibility that Shigella could exert cytotoxic effects on neutrophils was examined by measuring the release of the cytoplasmic enzyme LDH. During infection of neutrophils by M90T or BS176, the release of LDH was only triggered by the virulent strain, starting 2 h after the beginning of the experiment (Fig. 2). When different MOIs were used (from 3 to 100 bacteria per neutrophil), we observed that the release of LDH reached a plateau at an MOI of about 50 (Fig. 2, inset). To confirm that the neutrophils were dying by necrosis, electron microscopy was performed on neutrophils 2.5 h after infection with M90T or BS176 (Fig. 3). Control neutrophils are shown in Fig. 3A and B for comparison with infected cells. The plasma membranes of neutrophils that had ingested virulent Shigella were broken (Fig. 3C [inset], G, and H), while they remained intact in neutrophils infected with BS176 (E and F). Most of the time, infected cells presented a nuclear ultrastructure indistinguishable from that of uninfected cells, and the cytoplasm did not display vacuolation as observed in apoptotic

1291

neutrophils (see Fig. 1 in reference 28). Occasionally, an infected cell was found in an advanced stage of disintegration with a disrupted plasma membrane (Fig. 3D). Unlike BS176, which remained inside phagosomal vacuoles (Fig. 3E), M90T bacteria were free in the cytoplasm (C and D). Finally, fragmentation of DNA was assessed by agarose gel electrophoresis; no difference between control cells and cells infected with Shigella was observed, while characteristic DNA ladders were obtained in aging neutrophils maintained in suspension for 24 h (28) (Fig. 4). Together, these results indicate that M90T induced necrosis of neutrophils but not apoptosis. Neutrophil necrosis induced by M90T is dependent on phagocytosis but is not related to the ability of bacteria to leave the phagosomal vacuole. We examined whether phagocytosis of M90T is an important step in the cytolytic pathway. Experiments were performed with cytochalasin D to inhibit actin polymerization and, consequently, inhibit phagocytosis (23). Neutrophils were preincubated for 10 min with 10 ␮g of cytochalasin D/ml and then infected with Shigella. In the presence of cytochalasin D, phagocytosis of M90T was dramatically reduced and the release of LDH remained at the level of control cells (6.2 versus 7.6%, measured 2.5 h after infection; n ⫽ 2 [see results in Fig. 7, 10 min before]). Therefore, phagocytosis of Shigella and cell necrosis were both inhibited by cytochalasin D, suggesting that bacterial internalization could be a critical step to induce the necrotic response. As clearly shown in Fig. 3C and D, virulent bacteria were not inside phagosomes but free in the cytoplasm. To determine whether the cytotoxic effect of M90T is only observed when the bacteria are free in the cytoplasm, experiments were performed with serum-opsonized bacteria because it has been previously reported that, under these conditions, S. flexneri remains inside phagosomes in rabbit neutrophils (15). Experiments were first performed to determine if the M90T strain behaves similarly in human neutrophils. Since complement factors efficiently kill Shigella, the pool of human sera was heated to inactivate complement before bacterial opsonization. When electron microscopy was performed 2.5 h after infection, opsonized M90T organisms were enclosed in phagosomes (Fig. 3G and H), but the cells had the same necrotic phenotype (disrupted membrane) as those infected with nonopsonized M90T. In addition, the releases of LDH induced by opsonized and nonopsonized bacteria were similar (Fig. 5). For both strains, opsonization approximately doubled the rate of phagocytosis (Fig. 5). Therefore, the cytotoxic effect was maintained independently of the ability of neutrophils to maintain the virulent strain inside phagosomes, suggesting that M90T-induced neutrophil necrosis did not result from the ability of bacteria to escape into the cytoplasm. As expected with opsonic receptors which are coupled to the formation of phagolysosomes (34), opsonization of the virulent or nonvirulent strain triggered the release of ␤-glucuronidase 1 h after infection (Fig. 5). Thus, bacterial opsonization (i) triggered the release of lysosomal enzymes, (ii) impaired the escape of the virulent strain into the cytoplasm, and (iii) did not alter Shigella-mediated cytotoxic effects. Molecular mechanisms involved in Shigella-dependent neutrophil necrosis. Experiments were then performed to analyze the molecular mechanisms involved in necrosis. We tested whether protein synthesis is required for the induction of necrosis. Neutrophils were treated with 20 ␮g of cycloheximide/ml or 10 ␮g of actinomycin D/ml for 10 min to inhibit protein or mRNA synthesis (16), respectively, before infection with M90T. However, the release of LDH occurred as well. Then we studied whether neutrophils and bacteria could release molecules into the extracellular milieu which could exert

FIG. 3. Transmission electron micrographs of Shigella-infected human neutrophils. Control neutrophils (A and B) and neutrophils infected for 2.5 h with nonopsonized M90T (C and inset and D), BS176 (E), or opsonized BS176 (F) or M90T (G and H). (C, inset) Disrupted plasma membrane. In panels G and H, the arrows point to ruptured plasma membranes. S, Shigella. Magnification: A, ⫻8,500; B, ⫻7,500; C, ⫻9,500; D, ⫻6,500; E, ⫻5,000; F and H, ⫻7,000; G, ⫻6,500.

1292

VOL. 68, 2000


1293

index of cell toxicity (Fig. 6). Therefore, S. flexneri requires a functional secretory apparatus, IpaB, and IpaC, to induce neutrophil cytotoxicity. Since IpaB and IpaC are normally complexed (36), we studied whether infection of neutrophils with ipaB and ipaC mutants could restore cytotoxicity. However, even when these mutants were internalized together, no cytotoxic effect was obtained (data not shown). In addition, an S. flexneri mutant devoid of toxic LPS (msbB deletion; lack of lipid A myristoylation [P. J. Sansonetti, unpublished data]) induced the same pattern of responses in neutrophils as did the wild-type strain, M90T (data not shown), indicating that LPS is not implicated in the necrotic process. Shigella flexneri-dependent cytotoxicity requires actin polymerization. It has been shown that virulent Shigella induces rearrangements of F-actin during infection of epithelial cells (36), and this is clearly dependent on the IpaB and IpaC proteins (36), which are both secreted by the Mxi-Spa apparatus. Since ipaB, ipaC, and mxiD1 mutants did not induce neu-

FIG. 4. Lack of DNA fragments in neutrophils infected with S. flexneri for 2.5 h. DNAs from control cells (lane 2) and cells infected with nonopsonized BS176 (100 bacteria/cell) (lane 3) or M90T (100 bacteria/cell) (lane 4) are shown. Lane 5, DNA from neutrophils 24 h after isolation from blood. Only aging, apoptotic neutrophils generated DNA ladders (lane 5). Lane 1, 1-kb DNA ladder molecular size markers.

cytotoxic effects on naive cells. The incubation medium of neutrophils infected with virulent S. flexneri for 15 min to 2 h was transferred (after centrifugation at 200 ⫻ g to spin down neutrophils followed by a second centrifugation at 20,000 ⫻ g to spin down bacteria) to control neutrophils. However, this did not trigger the release of LDH (data not shown). Therefore, neither protein synthesis nor release of cytotoxic molecules in the extracellular medium is involved in the necrotic process. To further characterize the mechanisms used by Shigella to induce neutrophil toxicity, several mutants were studied. A mutant (SF401) with the mxiD1 gene coding for an essential component of the secretion type III machinery deleted was studied, since it is involved in secretion and probably translocation of several bacterial products of genes located in the virulence plasmid (36). Mutants (SF620 and SF621) which do not express the IpaB or IpaC invasins were tested as potential mediators of neutrophil necrosis because these products of genes located in the virulence plasmid are necessary for entryand contact-dependent hemolysis (18). First, phagocytosis of these mutants by human neutrophils was examined. All of them were internalized, and the percentage of cells having ingested bacteria was similar to that observed with the virulent strain, M90T, or the nonvirulent strain, BS176 (data not shown). Second, the release of granule markers by neutrophils exposed to these mutants was measured. As described in Fig. 1 with M90T and BS176, the release of ␤-glucuronidase was not triggered 1 h after infection (data not shown), while exocytosis of lactoferrin was elicited by all the mutants tested (Fig. 6). Third, the release of LDH was studied. ipaB, ipaC, and mxiD1 mutants did not exert toxic effects on neutrophils (Fig. 6). Similar results were obtained by measuring the release of ␤-glucuronidase 2.5 h after infection as an

FIG. 5. Opsonization of S. flexneri in heat-inactivated human serum enhanced phagocytosis and triggered exocytosis of azurophil granules, but only the virulent strain is cytotoxic. Neutrophils were exposed to no bacteria (none), to nonopsonized or opsonized bacteria (100:1), or to OZ (50:1). The percentage of cells containing ⱖ1 bacterium and the release of ␤-glucuronidase were determined 1 h after infection. The release of LDH was measured 2.5 h after infection. The results are expressed as the means ⫾ standard deviations of three to five experiments.

1294

FRANC ¸ OIS ET AL.

FIG. 6. Type III secretion pathway, IpaB, and IpaC are necessary for S. flexneri-induced neutrophil cytotoxicity. Several mutants were used under nonopsonic conditions at an MOI of 100:1 to infect neutrophils. M90T (wild type), SF401 (mxiD1), SF620 (ipaB2), SF621 (ipaC2), BS176 (virulence plasmid cured). (A) The release of lactoferrin was measured 1 h after infection. (B) The cytotoxic effect of Shigella was measured 2.5 h after infection, using LDH and ␤-glucuronidase as markers of cytolysis. The results are expressed as the means ⫾ standard deviations of three to five experiments. nd, not determined.

trophil necrosis, we considered the possibility that actin polymerization triggered by Shigella could play a role in the cytotoxic process. To distinguish between actin rearrangements needed for phagocytosis and actin rearrangements triggered by bacteria, cytochalasin D was added after the bacteria were ingested. To perform these experiments, we needed to fulfill two criteria: (i) maintain the phagocytic capacity of neutrophils and (ii) add cytochalasin D very rapidly after the bacteria had been internalized to block actin polymerization induced by Shigella. To obtain a good phagocytic rate in a short time, neutrophils were incubated in the presence of bacteria at 4°C for 45 min (100 bacteria/cell) to allow adhesion of bacteria on the neutrophils without phagocytosis. Nonadherent bacteria were then removed by washing the neutrophils at 4°C, and the phagocytic process was initiated by the addition of fresh medium at 37°C. When cytochalasin D was added 10 min before the bacteria, phagocytosis and LDH release were inhib-

INFECT. IMMUN.

FIG. 7. Cytotoxic effect of virulent S. flexneri is abolished by cytochalasin D added to neutrophils after bacterial p hagocytosis. Neutrophils were untreated or were treated with 10 ␮g of cytochalasin D/ml for 10 min before the addition of bacteria (10 min before). The neutrophils were then exposed to no bacteria (control) or to nonopsonized (100:1) or opsonized (50:1) bacteria of the M90T strain for 45 min at 4°C, and nonadherent bacteria were removed. Phagocytosis was initiated by adding fresh medium at 37°C (the medium for cells “10 min before” was supplemented by 10 ␮g of cytochalasin D/ml), and 15 min later, cytochalasin D was added to the cells (15 min after). Then, experiments were carried out for 1 h to determine the percentage of phagocytic cells or for 2.5 h for LDH measurement. none, no addition of cytochalasin D. The results are expressed as the means ⫾ standard deviations of three experiments.

ited (Fig. 7) as described above. More interestingly, when cytochalasin D was added 15 min after initiation of the phagocytic process, phagocytosis of bacteria occurred at a rate similar to that obtained in the absence of cytochalasin D (Fig. 7), but the release of LDH was back to the level of the control cells (Fig. 7). When Shigella was opsonized, the same pattern of results was obtained, but amplified. Therefore, by inhibiting actin polymerization after the phagocytic process, it was possible to maintain the neutrophil membrane integrity, indicating that polymerization of actin plays a critical role in necrosis induced by Shigella. DISCUSSION Although neutrophils are constitutively programmed to die by apoptosis (28), we report that human neutrophils exposed to virulent S. flexneri undergo necrosis. This is based on the following observations: (i) the cytoplasmic enzyme LDH, a marker of cytolysis, is released 2 h after infection of neutrophils by the virulent strain M90T but not by the virulence plasmid cured-strain BS176; (ii) release of the azurophil granule enzyme, ␤-glucuronidase, was triggered by M90T but not BS176 after a delay corresponding to the release of LDH; (iii)

VOL. 68, 2000


when cells were examined by electron microscopy, we noticed that the plasma membrane was disrupted, the nuclear morphology did not change, and the cytoplasm did not vacuolize; and (iv) no DNA fragmentation was detected in electrophoretic gels. In contrast, apoptotic cells are characterized by alteration of the nuclear morphology, DNA fragmentation, and vacuolization of the cytoplasm and by keeping plasma membrane integrity (11, 28). Because apoptotic neutrophils retain an intact plasma membrane, they do not release the contents of their granules (11, 28) and surrounding tissues are protected from proinflammatory agents. Imbalance between apoptosis and necrosis pathways is consequently important in the pathogenesis of inflammatory diseases. S. flexneri, which appears to be able to induce necrosis of neutrophils, therefore has the potential to tip the balance toward inflammation. This correlates with the clinical observations reporting destruction of the intestinal mucosa and hemorrhagic diarrhea during shigellosis (2). By inducing mucosal inflammation, responsible for major tissue destruction, Shigella’s advantage is probably to gain access to the basolateral membrane of epithelial cells. This is crucial for the survival and growth of bacteria, since they can invade these cells only through the basolateral pole and, once internalized, they start to multiply and spread from cell to cell, thus achieving extensive intraepithelial colonization (27). Similarly, S. flexneri induces a rapid cytolytic event in human monocyte-derived macrophages (8). In contrast, it induces apoptosis in rabbit or mouse macrophages (39, 40), suggesting that Shigella can act differently in different species, as expected for a bacterium pathogenic for humans. Otherwise, S. flexneri could induce distinct types of cell death, necrosis in neutrophils and apoptosis in macrophages, independently of the species origin. Indeed, when the human monoblastic cell line U937 is differentiated in macrophages with gamma interferon, it undergoes apoptosis when exposed to S. flexneri, while it undergoes oncosis-necrosis when differentiated with retinoic acid (21), an agent inducing a neutrophilic morphology (19). Necrosis of neutrophils induced by Shigella depends on the secretion type III apparatus and IpaB or IpaC, two invasins secreted through the type III machinery. M90T has been shown to induce lysis of erythrocytes (18), and IpaC interacts with lipid membranes (7). IpaB and IpaC associate as a complex and can form a pore in the membranes of eukaryotic cells (17; C. deGeyter and P. J. Sansonetti, personal communication) which could destabilize the plasma and the granule membranes. In addition, the complex formed by IpaB and -C is sufficient to initiate the cellular rearrangements necessary to achieve bacterial entry into epithelial cells (36), and recent data have demonstrated that IpaC is a direct effector of Shigella-induced actin rearrangements (35). The ability of bacteria to rearrange microfilaments is critical for entry into epithelial cells (36) and for triggering necrosis in human neutrophils (see above). Indeed, cytochalasin D, which impairs actin polymerization and blocks entry into epithelial cells, also protects neutrophils against the cytotoxic effect of Shigella. Other experiments were designed to allow internalization of bacteria by neutrophils, and actin polymerization was inhibited by the addition of cytochalasin D immediately after entry had occurred. Under these conditions, the release of LDH was not triggered by Shigella, indicating that actin reorganization, distal to the site of phagocytosis, is implicated in signals leading to necrosis. It is possible that, in addition to the cytoskeleton rearrangements, membrane destabilization by the pore-forming invasins could participate in the necrotic process. We also report that Shigella-induced necrosis does not require synthesis of proteins by neutrophils and that potentially

1295

cytotoxic products are not released into the incubation medium containing both neutrophils and bacteria (16). This suggests that a close interaction between virulent bacteria and neutrophils is probably necessary to allow injection through the type III machinery of proteins such as IpaB and IpaC. The fusion of lysosomes (azurophil granules) with phagosomes constitutes a potent bactericidal response, as proteases and bactericidal proteins are quickly released at contact with ingested microorganisms (32). We report that fusion of lysosomes with phagosomes is not triggered during the ingestion of Shigella. The observations that Shigella escapes from macropinosomes in epithelial cells (26) and from phagosomes in mouse macrophages (3) and human neutrophils (see above) suggests that the bacteria take advantage of not being exposed to lysosomal enzymes before their exit. This is supported by the observation that, when Shigella is internalized through opsonic receptors, the bacteria stay inside phagosomes (see above) (15), phagolysosomes are formed (see above), and the bacteria are killed (9, 15). Although it is an interesting finding that phagocytosis of Shigella is not coupled to lysosome fusion, it has nothing to do with the ability of these bacteria to kill neutrophils. Indeed, opsonized bacteria which were retained inside phagolysosomes induced neutrophil necrosis even better than the nonopsonized bacteria (Fig. 7). In conclusion, when human neutrophils ingest Shigella under nonopsonic conditions, they do not proceed to the biogenesis of phagolysosomes and the bacteria escape the phagosomes, while bacterial opsonization induced phagolysosome formation and the bacteria remained trapped in phagosomes. In both cases, neutrophils undergo necrosis, and this is under the control of a functional type III secretory apparatus and of IpaB and IpaC invasins. Actin polymerization is a critical step in the necrotic pathway, as treatment of neutrophils by cytochalasin D inhibited the Shigella-induced cytotoxicity. Although neutrophils are constitutively programmed to die by apoptosis, Shigella has developed a strategy to kill neutrophils by necrosis, a process characterized by the release of tissue-injurious granular proteins. As a consequence, disruption of the epithelial barrier accounts for both the dysentery observed in shigellosis and the access that Shigella gains at the basolateral poles of epithelial cells in order to invade them. ACKNOWLEDGMENTS This work was supported in part by The Ministe`re de l’Education Nationale de la Recherche et de la Technologie, Programme Microbiologie. We gratefully acknowledge Christine Bordier for expert technical assistance. REFERENCES 1. Allaoui, A., P. J. Sansonetti, and C. Parsot. 1993. MxiD, an outer membrane protein necessary for the secretion of the Shigella flexneri 1pa invasins. Mol. Microbiol. 7:59–68. 2. Anand, B. S., V. Malhotra, S. K. Bhattacharya, P. Datta, D. Datta, D. Sen, M. K. Bhattacharya, P. P. Mukherjee, and S. C. Pal. 1986. Rectal histology in acute bacillary dysentery. Gastroenterology 90:654–660. 3. Barzu, S., Z. Benjelloun-Touimi, A. Phalipon, P. J. Sansonetti, and C. Parsot. 1997. Functional analysis of the Shigella flexneri Ipac invasin by insertional mutagenesis. Infect. Immun. 65:1599–1605. 4. Borregaard, N., K. Lollike, L. Kjeldsen, H. Sengelov, L. Bastholm, M. H. Nielsen, and D. F. Bainton. 1993. Human neutrophil granules and secretory vesicles. Eur. J. Haematol. 51:187–198. 5. Canitrot, Y., P. Frit, and B. Salles. 1997. Deficient apoptotic process in cisplatin-resistant L1210 cells cannot account for the cellular response to various drug treatments. Biochem. Biophys. Res. Commun. 234:573–577. 6. Cohen, M. S. 1994. Molecular events in the activation of human neutrophils for microbicidal killing. Clin. Infect. Dis. 18(Suppl. 2):S170–S179. 7. deGeyter, C., B. Vogt, Z. Benjelloun-Touimi, P. J. Sansonetti, J. M. Ruysschaert, C. Parsot, and V. Cabiaux. 1997. Purification of IpaC, a protein involved in entry of Shigella flexneri into epithelial cells, and characterization

1296

FRANC ¸ OIS ET AL.

of its interaction with lipid membranes. FEBS Lett. 400:149–154. 8. Fernandez-Prada, C., D. L. Hoover, B. D. Tall, and M. M. Venkatesan. 1997. Human monocyte-derived macrophages infected with virulent Shigella flexneri in vitro undergo a rapid cytolytic event similar to oncosis but not apoptosis. Infect. Immun. 65:1486–1496. 9. Gbarah, A., D. Mirelman, P. J. Sansonetti, R. Verdon, W. Bernhard, and N. Sharon. 1993. Shigella flexneri transformants expressing type 1 (mannosespecific) fimbriae bind to, activate, and are killed by phagocytic cells. Infect. Immun. 61:1687–1693. 10. Gre´goire, C., H. Welch, C. Astarie-Dequeker, and I. Maridonneau-Parini. 1998. Expression of azurophil and specific granule proteins during differentiation of NB4 cells in neutrophils. J. Cell. Physiol. 175:203–210. 11. Haslett, C. 1997. Granulocyte apoptosis and inflammatory disease. Br. Med. Bull. 53:669–683. 12. Jensen, V. B., J. T. Harty, and B. D. Jones. 1998. Interactions of the invasive pathogens Salmonella typhimurium, Listeria monocytogenes, and Shigella flexneri with M cells and murine Peyer’s patches. Infect. Immun. 66:3758–3766. 13. Joiner, K. A., T. Albert, and D. Rotrosen. 1989. The opsonizing ligand on Salmonella typhimurium influences incorporation of specific, but not azurophil, granule constituents into neutrophil phagosomes. J. Cell Biol. 109: 2771–2782. 14. LeCabec, V., and I. Maridonneau-Parini. 1995. Complete and reversible inhibition of NADPH oxidase in human neutrophils by phenylarsine oxide at a step distal to membrane translocation of the enzyme subunits. J. Biol. Chem. 270:2067–2073. 15. Mandic-Mulec, I., J. Weiss, and A. Zychlinsky. 1997. Shigella flexneri is trapped in polymorphonuclear leukocyte vacuoles and efficiently killed. Infect. Immun. 65:110–115. 16. Maridonneau-Parini, I., S. E. Malawista, H. Stubbe, F. Russo-Marie, and B. Polla. 1993. Heat shock in human neutrophils: superoxide generation is inhibited by a mechanism distinct from heat-denaturation of NADPH oxidase and is protected by heat shock proteins in thermotolerant cells. J. Cell. Physiol. 156:204–211. 17. Meńard, R., P. Sansonetti, C. Parsot, and T. Vasselon. 1994. Extracellular association and cytoplasmic partitioning of the IpaB and IpaC invasins of S. flexneri. Cell 79:515–525. 18. Meńard, R., P. J. Sansonetti, and C. Parsot. 1993. Nonpolar mutagenesis of the ipa genes defines IpaB, IpaC, and IpaD as effectors of Shigella flexneri entry into epithelial cells. J. Bacteriol. 175:5899–5906. 19. Nakajima, H., M. Kizaki, H. Ueno, A. Muto, N. Takayama, H. Matsushita, A. Sonoda, and Y. Ikeda. 1996. All-trans and 9-cis retinoic acid enhance 1,25dihydroxyvitamin D3-induced monocytic differentiation of U937 cells. Leuk. Res. 20:665–676. 20. N’Diaye, E. N., X. Darzacq, C. Astarie-Dequeker, M. Daffe´, J. Calafat, and I. Maridonneau-Parini. 1998. Fusion of azurophil granules with phagosomes and activation of the tyrosine kinase Hck are specifically inhibited during phagocytosis of mycobacteria by human neutrophils. J. Immunol. 161:4983– 4991. 21. Nonaka, T., A. Kuwae, C. Sasakawa, and S. Imajoh-Ohmi. 1999. Shigella flexneri YSH6000 induces two types of cell death, apoptosis and oncosis, in the differentiated human monoblastic cell line U937. FEMS Microbiol. Lett. 174:89–95.

Editor: E. I. Tuomanen

INFECT. IMMUN. 22. Perdomo, J. J., J. M. Cavaillon, M. Huerre, H. Ohayon, P. Gounon, and P. J. Sansonetti. 1994. Acute inflammation causes epithelial invasion and mucosal destruction in experimental shigellosis. J. Exp. Med. 180:1307–1319. 23. Pryzwansky, K. B., S. Kidao, and E. P. Merricks. 1998. Compartmentalization of PDE-4 and cAMP-dependent protein kinase in neutrophils and macrophages during phagocytosis. Cell. Biochem. Biophys. 28:251–275. 24. Renesto, P., J. Mounier, and P. J. Sansonetti. 1996. Induction of adherence and degranulation of polymorphonuclear leukocytes: a new expression of the invasive phenotype of Shigella flexneri. Infect. Immun. 64:719–723. 25. Sansonetti, P. J. 1993. Molecular mechanisms of cell and tissue invasion by Shigella flexneri. Infect. Agents Dis. 2:201–206. 26. Sansonetti, P. J., A. Ryter, P. Clerc, A. T. Maurelli, and J. Mounier. 1986. Multiplication of Shigella flexneri within HeLa cells: lysis of the phagocytic vacuole and plasmid-mediated contact hemolysis. Infect. Immun. 51:461– 469. 27. Sansonetti, P. J., G. TranVanNhieu, and C. Egile. 1999. Rupture of the intestinal barrier and mucosal invasion by Shigella flexneri. Clin. Infect. Dis. 28:466–475. 28. Savill, J., and C. Haslett. 1995. Granulocyte clearance by apoptosis in the resolution of inflammation. Semin. Cell Biol. 6:385–393. 29. Skoudy, A., G. T. V. Nhieu, N. Mantis, M. Arpin, J. Mounier, P. Gounon, and P. J. Sansonetti. 1999. A functional role of ezrin during Shigella flexneri entry into epithelial cells. J. Cell Sci. 112:2059–2068. 30. Smith, J. A. 1994. Neutrophils, host defense and inflammation: a doubleedged sword. J. Leukoc. Biol. 56:672–686. 31. Suzaki, E., H. Kobayashi, Y. Kodama, T. Masujima, and S. Terakawa. 1997. Video-rate dynamics of exocytotic events associated with phagocytosis in neutrophils. Cell Motil. Cytoskeleton 38:215–228. 32. Tapper, H. 1996. Out of the phagocyte or into its phagosome: signalling to secretion. Eur. J. Haematol. 57:191–201. 33. Tapper, H. 1996. The secretion of preformed granules by macrophages and neutrophils. J. Leukoc. Biol. 59:613. 34. Tapper, H., and S. Grinstein. 1997. Fc receptor-triggered insertion of secretory granules into the plasma membrane of human neutrophils. Selective retrieval during phagocytosis. J. Immunol. 159:409–418. 35. TranVanNhieu, G., E. Caron, A. Hall, and P. J. Sansonetti. 1999. IpaC induces actin polymerization and filipodia formation during Shigella entry into epithelial cells. EMBO J. 18:3249–3262. 36. TranVanNhieu, G., and P. J. Sansonetti. 1999. Mechanism of Shigella entry into epithelial cells. Curr. Opin. Microbiol. 2:51–55. 37. Wassef, J., D. F. Keren, and J. L. Mailloux. 1989. Role of M cells in initial bacterial uptake and ulcer formation in the rabbit intestinal loop model in shigellosis. Infect. Immun. 57:858–863. 38. Welch, H., and I. Maridonneau-Parini. 1997. Hck is activated by opsonized zymosan and A23187 in distinct subcellular fractions of human granulocytes. J. Biol. Chem. 272:102–109. 39. Zychlinsky, A., M. C. Prevost, and P. J. Sansonetti. 1992. Shigella flexneri induces apoptosis in infected macrophages. Nature 358:167–169. 40. Zychlinsky, A., K. Thirumalai, J. Arondel, J. R. Cantey, A. O. Aliprantis, and P. J. Sansonetti. 1996. In vivo apoptosis in Shigella flexneri infections. Infect. Immun. 64:5357–5365.