Bacillary dysentery is an acute inflammatory bowel disease caused by an infectious agent, the enteroinvasive genus Shigella, hence the name shigellosis. Shigellae have the capacity to invade the colonic and rectal epithelium in humans, thereby causing the acute mucosal inflammation that characterizes the disease. Shigellosis is endemic throughout the world, but 99% of the 150 million annual cases and almost all of the million deaths occur in the developing world, particularly in areas where personal and general hygiene are insufficient. Shigellosis is a disease of impoverished people which in about 70% of the cases affects children between the ages of 1 and 5 years . There are four species of Shigella: S. dysenteriae, among which serotype 1 (shiga bacillus) accounts for brisk and deadly epidemics in the poorest populations, S. flexneri and S. sonnei which account for the endemic form of the disease, the latter being prevalent in the industrialized world, and S. boydii which is only observed in the Indian subcontinent. In view of increasing antibiotic resistance and persistence of the poor hygiene conditions that allow contamination, vaccination appears the most cost-effective approach in spite of tremendous difficulties in developing a new vaccine in this particular situation .
2Pathogenesis of bacillary dysentery
Shigella is a highly contagious microorganism since as few as 10–100 bacteria can cause the disease in adult volunteers. After oral contamination, bacteria pass through the stomach and the small intestine before reaching the colon where they invade the mucosa, initiating the acute destructive recto-colitis that causes the dysenteric symptoms: fever, intestinal cramps and emission of mucopurulent and bloody stools. The basis for organ specificity of shigellosis to the rectal and colonic mucosae is not understood. Shigella may express a colon-specific adhesive system, or the colonic and rectal mucosae may be more susceptible to developing acute inflammation in the presence of invasive shigellae. The disease remains essentially limited to the intestinal mucosa and septicemic dissemination is a rare event, except in malnourished children. The molecular and cellular effectors of innate immunity that eradicate the bacteria during the phase of primary infection and prevent systemic dissemination at the price of intestinal tissue destruction have yet to be fully identified and their mode of action characterized. Shigellosis can be seen as a loss of balance in the host mechanisms that regulate inflammation in the presence of an invading microorganism, emphasizing that in the elucidation of an infectious process, equal attention should be paid to the mechanisms of infection and the mechanisms of host response. A series of discoveries have recently allowed progress in our understanding of the molecular mechanisms by which Shigella disrupts, invades and destroys the intestinal barrier . Understanding the molecular and cellular bases of these mechanisms is essential to develop innovative antipathogenicity molecules and vaccine candidates against this disease.
3Determinants of pathogenesis expressed by Shigella
All virulent isolates of Shigella carry a 220-kb plasmid which encodes the ‘invasive phenotype’ of this species. In cell assay systems, expression of this invasive phenotype depends on the cell population that is used as a target (Fig. 1).
In the presence of epithelial cells, invasive Shigella induce massive reorganization of the subcortical cytoskeletal network causing the formation of a macropinocytic vacuole, entry into the cells and intracellular growth after quick lysis of the endocytic vacuole. Escape into the cytoplasm is followed by an actin-dependent motility process which leads to bacterial passage into adjacent cells . In addition, intracellular microorganisms program epithelial cells to become pro-inflammatory cells and to play a major role in elicitation of mucosal inflammation . In the presence of lipopolysaccharide (LPS)-activated macrophages, invasive Shigella survive and grow, they quickly induce apoptosis and cause massive secretion of interleukin (IL)-1β and IL-18, factors that are likely to play a central role in initiating inflammation. In the presence of polymorphonuclear leukocytes (PMN) invasive Shigella cause increased adherence and release of granular content, thereby adding to the severity of inflammation , although Shigella is unable to survive in those activated PMNs .
Although cellular approaches tend to provide an oversimplified (i.e. reductionist) view of the infectious process, it is likely that, in vivo, expression of the invasive phenotype of Shigella upon its various cellular targets, results in rupture, invasion and destruction of the mucosal barrier.
In S. flexneri as well as in the other Shigella species, a large virulence plasmid contains most of the key genes required to express the invasive phenotype [9,10]. After completion of the sequence and annotation of pWR100, the 214-kb plasmid of the S. flexneri 5a strain M90T , it appears that about one third of the sequences corresponds to various complete and incomplete insertion sequences that are likely to reflect the natural history of this plasmid's construction. The coding sequences are scattered on the entire plasmid with one block of 30 kb showing a particularly dense pattern of genes, the ipa/mxi-spa locus that can be considered the central S. flexneri pathogenicity island (PAI). This PAI, which is shown in Fig. 2, is necessary and sufficient to cause entry of Shigella into epithelial cells and macrophage apoptotic death.
Based on available data, one can consider that this PAI primarily encodes a type III secreton, in other words a flagella-like structure able to deliver Shigella proteins straight from the bacterial cytoplasm into the cytoplasmic membrane of epithelial cells, or into their cytoplasm. At least five proteins encoded by the mxi operon assemble to form a structure that can be divided into three domains: a cytoplasmic bulb, a multidisc, transmembrane domain that spans the inner and outer membrane of the bacterium, forming a basal body-like structure, and a needle of 60 nm which is expected to deliver the effector proteins through a channel of 2–3 nm [12,13]. This secretory apparatus, which is similar to the type III secreton of Salmonella, is activated upon contact with the cellular surface. There are about 20 candidate target proteins secreted through this secreton, upon contact of the bacterium with the epithelial cell surface. Five of them, IpaA–D and IpgD, are encoded by the 30-kb PAI. Others are encoded by genes scattered on the virulence plasmid. These genes share a low GC content of about 35%. These proteins have no signal peptide but are likely to share common features on their N-termini which target them to the secreton. Some of these target proteins are associated in the bacterial cytoplasm to a dedicated chaperone which prevents their proteolytic degradation. There are two categories of proteins secreted through the type III secreton:
1IpaB, IpaC and IpaD are essential to the initial events of secretion. IpaB and IpaD form a complex which controls the flux of proteins through the type III secreton. It is likely to be the primary target of the signal inducing secretion upon contact of bacteria with eukaryotic cells. IpaB and IpaC then form a complex which inserts into the eukaryotic cell membrane to form a pore [12–15]. In the presence of an epithelial cell, this pore expresses a dual function: it induces the early events of actin polymerization via the C-terminal domain of IpaC  and it is also likely to allow the injection of several proteins into the cell cytoplasm. In the presence of a macrophage, IpaB causes apoptosis of the target cell. These properties are summarized in Fig. 3. IpaA and IpgD, like IpaB, C and D, are constitutively expressed at 37°C, regardless of the activity of the type III secreton. Deletion of their respective genes does not eliminate the entry capacity of the mutants, but induces significant attenuation, indicating that these genes have an effect on the maturation of the entry focus.
2The second category of proteins encompasses two subgroups. These proteins, of still unidentified function, correspond to putative additional targets for the type III secreton, such as members of the IpaH family, SopB and VirA. Their genes are transcriptionally induced upon activation of the secreton. This illustrates how far we are from having identified the function of many secreted proteins. In any event, the type III secreton and its target proteins can now be regarded as the major weapon that Shigella uses to enter into epithelial cells and also to alter the function of others such as phagocytes.
Beside the type III secreton and its cognate target proteins, other important proteins are encoded by the virulence plasmid. Among these, SepA is a secreted serine protease whose function has not yet been established . The most important, however, is IcsA (VirG), which achieves the actin-based motility of Shigella and permits its passage from one cell to another . IcsA (VirG) is a 120-kDa surface protein which localizes at one pole of the bacterial body and is able to cause actin-dependent motility. Actin-based motility, in the context of a polarized epithelium establishing intercellular junctions, allows cell-to-cell spread in a process that involves the engagement of components of the cellular junction by the pathogen in order to form a protrusion which is internalized by the neighboring cell. Lysis of the protruding membranes allows escape of the bacterium into the cytoplasm of the newly infected cell and so on, establishing a remarkably efficient system allowing extended cellular colonization. IcsA mutants are severely impaired in pathogenicity, including in monkeys and even in human volunteers as recently tested.
3.2Chromosomal genes associated with virulence
Shigella is an interesting paradigm of co-evolution and mutual adaptation between plasmid and chromosomal virulence genes. Beside the core of virulence plasmid genes that dictate the direct interaction of bacteria with the various cell populations that make up the epithelial barrier, chromosomal genes also participate in the pathogenic process. This was strongly suggested by classical recombination experiments showing that introduction of the S. flexneri virulence plasmid into Escherichia coli K12 conferred full invasiveness on the transconjugants in in vitro systems whereas successive conjugative transfer of various portions of the chromosome in plasmid-bearing strains eventually conferred full virulence in in vivo models of infection . These chromosomal genes can be classified into two categories: (i) genes that regulate the expression of the virulence genes on the plasmid. This category is exemplified by virR, a gene encoding a histone-like molecule close to H–NS which controls the temperature-dependent expression of Ipa and Mxi-Spa proteins ; (ii) genes that are important for bacterial survival in the intestinal tract and in infected tissues such as those encoding the LPS and siderophores, several of them being located on increasingly identified PAIs. In addition, in S. dysenteriae 1, shiga toxin is encoded by a chromosomal locus. One should also mention the recent demonstration of ‘black holes’ in the Shigella chromosome, corresponding to deletions of loci which would otherwise impair expression of the full invasive phenotype of Shigella. Availability of the genome sequence of representative isolates of S. flexneri, S. dysenteriae 1 and S. sonnei should soon illuminate our understanding of the plastic events that have led to construction of Shigella as a pathogen, based on a comparison with the E. coli genome.
4Invasion of epithelial cells by Shigella, a paradigm for cellular microbiology
4.1Cytoskeletal structures are the targets for Shigella entry-associated proteins
During the initial step of the entry process, Shigella induces actin polymerization at the site of bacterial contact with the host cell membrane [21,22]. Actin polymerization forms filopodial structures that eventually give rise to lamellipodial structures. This results in a macropinocytic pocket which engulfs the microorganism. Formation of this pocket is the result of a series of maturation steps that are programmed by the Ipa effector proteins and IpgD. Rho GTPases (Cdc42, Rac and Rho itself) are essential regulators of cellular processes involving cytoskeletal rearrangements . These proteins are active in their GTP-binding form, Cdc42 programming formation of filopodial extensions from the cellular surface, Rac programming formation of lamellipodial structures and Rho, which does not directly induce actin polymerization, leading to stress fiber formation and establishment of cell adherence plaques. The three Rho GTPases are involved in the formation of the cellular structures that support formation of the Shigella entry focus . Our current model for the sequence of events that occurs is as follows. Once inserted into the eukaryotic cell membrane, as a component of the IpaB/IpaC complex that forms a pore, the C-terminal domain of IpaC which is exposed on the cytoplasmic side triggers actin nucleation/polymerization  by inducing activation of Cdc42 and Rac. The former acts through a protein of the WASP family and the actin nucleator Arp2/3. This sequential activation is likely to account for the appearance of filopodes followed by lamellipodes. The mechanisms involved in the activation of Cdc42 (Rac) by IpaC are currently unknown. Unlike SopE, a Salmonella protein injected in the host cell cytosol in the course of bacterial entry , IpaC does not show an in vitro exchange factor (GEF) activity on those GTPases. At the same time, downstream activation of Rho appears to be a major factor promoting further extension of actin structures , thus leading to giant extension from the cell surface. This is due to two major events:
•The recruitment and engagement of the proto-oncogene c-src  which expresses two major functions . (i) It phosphorylates cortactin, an actin-binding protein that is recruited largely at the tip of the projections and seems to be involved in promoting further actin polymerization. (ii) It also down-regulates Rho, partly by inducing tyrosine phosphorylation of p190RhoGAP, thus engaging a negative loop that promotes further Cdc42/Rac-induced actin polymerization .
•The recruitment of ezrin, a membrane–cytoskeleton linker, which, like cortactin, induces further extension of filopodial structures .
It appears, however, that despite inducing massive cytoskeletal extensions, Shigella does not promote a fully functional entry focus. Projections extending from the cell surface even tend to push the bacterial body away. A further step of maturation is therefore required that involves injection of IpaA into the host cell cytosol. Mutants that do not express IpaA enter epithelial cells with much lower efficiency than their wild-type counterpart, whereas they induce entry foci that are not only as numerous as those induced by the wild-type strain, but also bigger and much longer lasting . Entry foci induced by ipaA mutants show massive and persistent actin polymerization, but fail to form an ‘actin cup’ or ‘pseudo-adherence plaque’ around the endocytic vacuole. This structure clearly results from the bundling of very short actin filaments, indicating that the presence of IpaA induces both actin depolymerization and its bundling. Most of this maturation process results from the binding of IpaA to the N-terminal head of vinculin [31,32], a cytoskeleton-associated protein which orchestrates the formation of cell adherence plaques. Binding of IpaA to the head domain, with a Kd in the 5 nM range, activates the dormant form of vinculin and induces its opening, thus revealing a functional actin-binding site that participates in the bundling process in its C-terminal portion, and a binding site for α-actinin in its head. In addition, this interaction also results in actin depolymerization. This is another example of a bacterial protein taking control of cell signaling pathways via its binding with strong activity to a protein target. The mechanisms that control the timing of these maturation events remain unknown.
The complexity of the signaling processes that induce and modulate the macropinocytic vacuole internalizing Shigella are summarized in Fig. 4.
4.2Actin is the target for Shigella motility-associated IcsA/VirG protein
After lysis of the phagocytic vacuole, Shigella nucleates and assembles an F-actin comet at one of its poles. This process provides a motor which allows the bacterium to move intracellularly and spread from cell to cell . The ability of Shigella to move intracellularly via an actin-dependent mechanism is a property shared with Listeria monocytogenes, some Rickettsia and the vaccinia virus . Shigella's actin-based motility is mediated by a single outer membrane protein, IcsA/VirG [18,34,35]. This protein is exported and anchored by its C-terminal domain (IcsAβ) into the bacterial outer membrane. The N-terminal domain, which is characterized by the presence of glycine-rich repeats, represents the active site of the protein required to induce actin nucleation. IcsA shows a polar distribution on the bacterial cell surface that is essential for proper assembly of the comet tail and straight movement inside the cell cytoplasm .
As IcsA is unable to directly induce actin nucleation and polymerization, it was soon considered that this protein was able to recruit one or several ligands allowing this process to occur. Although vinculin was claimed to play a role , recent evidence indicates that the principal ligand of IcsA/VirG is N-WASP , a member of the WASP family of Cdc42-dependent mediators of actin nucleation via the Arp2/3 complex. Formation of a ternary complex between IcsA, N-WASP and Arp2/3 is necessary and sufficient to cause actin nucleation in the presence of actin monomers . The interaction between IcsA and N-WASP involves the glycine-rich repeats and seems to exclude Cdc42. These interactions leading to actin nucleation and polymerization are summarized in Fig. 5. Moreover, recent work indicates that in addition to this ternary complex, proteins regulating the treadmilling process of actin filaments such as ADF/cofilin, the capping protein and profilin are necessary to achieve bacterial motility .
4.3Adherent junction as a motor for Shigella cell-to-cell spread
A major component of the Shigella invasive phenotype is the capacity to spread from one cell to another. Cell-to-cell spread is a direct consequence of the IcsA-mediated, actin-based motility process. In the context of a polarized epithelium, as can be reconstituted in vitro by growing human colonic cell lines such as Caco2 and T84, bacteria engage the components of the adherence junction, requiring the cadherins, and form a protrusion which is actively endocytosed by the neighboring cell in a process requiring activation of the myosin light chain kinase, indicating a role for myosin II, therefore the possibility that a true phagocytic process of the protrusion is involved . Escape from the two membranes requires expression of IpaB and IpaC [42–44]. Once free in the cytoplasm of the adjacent cells, bacteria can proceed to the next cell and so on. This cycle of Shigella infection in the context of a polarized epithelium is summarized in Fig. 6.
5Crossing, disruption and invasion of the epithelial barrier by Shigella or the devious use of inflammation by a pathogen
Shigellosis is a paradigm of a pathogen manipulating innate immunity, essentially inflammation, to disrupt and invade the intestinal mucosa. However, in an attempt to destroy the invasive pathogen, inflammation goes uncontrolled and causes major tissue destruction that represents the substrate of the symptoms and lesions observed in the course of bacillary dysentery. However, before eliciting this response, Shigella must first cross the epithelial barrier, a process that is achieved with great difficulty through the apical pole of epithelial cells. Current evidence indicates that the initial entry route is the follicle-associated epithelium (FAE) that overlies the mucosa-associated lymph nodes.
5.1M cells as a route through the intestinal epithelium for Shigella and other enteroinvasive Gram-negative pathogens
M cells constitute a variable percentage of the epithelial cells of the FAE, in other words, the epithelial layer that covers the lymphoid nodules associated with the intestinal mucosa, either in aggregates in the terminal ileum (Peyer's patches), or as solitary nodules in the colon and rectum . These lymphoid structures, in toto, represent the inductive site of intestinal immunity. M cells lack mucus, glycocalyx and brush border microvilli on their apical surface. Their luminal side is therefore accessible to antigens, particles and microorganisms from the intestinal lumen. The high transcytotic activity of M cells therefore facilitates translocation of antigens and microbes through the epithelial barrier, straight into the inductive sites of mucosal immunity . M cells form a pocket filled with both lymphocytes and macrophages, thus providing direct contact between the intestinal microbial flora and the mucosal immune system. Although the entire surface of FAE is about 1/10 000 of that of the regular villous epithelial surface, it plays a strategic role in the defense of the mucosal barrier and at the same time is used by pathogens to cross the epithelial barrier since it spares this microbe the need to dissolve the mucus, resist intestinal peristaltism, invade epithelial cells through their apical brush border, or penetrate between cells by opening their tight junctions. Conversely, using this route, the microbe exposes itself to an environment characterized by the presence of numerous resident macrophages, particularly in the dome area of the follicular zone (i.e. the space between FAE and the lymph node). In this situation, an invasive pathogen must develop strategies allowing its survival to the killing mechanisms of phagocytic cells. Several lines of evidence indicate that the major Gram-negative enteroinvasive pathogens, Shigella, Salmonella, and Yersinia, primarily use the FAE as their entry route . These pathogens have consequently evolved divergent strategies to survive their encounter with phagocytes. Shigella has opted for a strategy dominated by the apoptotic killing of macrophages, escape in subepithelial tissues and basolateral invasion of the epithelial lining. Salmonella, although able to also induce macrophage apoptosis, has evolved a strategy of remodeling its intracellular vacuole and of expressing genes allowing its intracellular survival. It is possible that a compromise established between the macrophages and their bacterial hosts allows further systemic dissemination following the movement of these phagocytes. Yersinia is also able to cause macrophage death. However, it seems that the dominant feature of its interaction is the expression of a strong antiphagocytic potential which causes Yersinia to remain essentially extracellular in infected lymph nodes .
5.2The Shigella paradigm: M cell translocation–macrophage apoptosis–inflammation 
The demonstration that M cells of the FAE are the major site of initial epithelial translocation for Shigella was based on a conjunction of experiments carried out in the rabbit ligated intestinal loop model of infection, as well as in a macaque monkey model of bacillary dysentery. Shortly after their translocation, bacteria are found in macrophages and probably in dendritic cells. Transmission electron microscopy shows that many of these infected cells present major alterations, indicating their rapid death.
In vitro observations also showed that in the presence of virulent shigellae, macrophages quickly die . The killing process was identified as apoptosis. It requires phagocytosis of invasive shigellae and is caused by IpaB . IpaB alone causes apoptosis of macrophages . Therefore, macrophage apoptosis, in addition to entry into epithelial cells, represents another major facet of the Shigella invasive phenotype. In vivo, massive induction of apoptosis is characteristic of rabbit Peyer's patches infected by an invasive Shigella isolate . Many of these cells are macrophages and probably also dendritic cells which are also efficiently killed by invasive shigellae in vitro (Edgeworth and Sansonetti, in preparation), but B and T cells are also concerned, suggesting that direct infection of cells is not necessarily an absolute prerequisite for apoptosis in the course of Shigella infection. This point needs further clarification.
Macrophage apoptosis caused by invasive Shigella is not only likely to permit bacterial survival, following crossing of the epithelial lining, but is also central to early triggering of inflammation. This dual function reflects the molecular mechanisms of IpaB triggering of the macrophage death program induced by invasive Shigella. IpaB binds to caspase 1, formerly called ICE for IL-1 cleavage enzyme . This event triggers activation of this cysteine protease into its active components which self-oligomerize, thereby expressing a capacity to cleave the pro-IL-1β and pro-IL-18 cytokines , but also to engage the apoptotic program of the cell, caspase 1 being able to cause cell apoptosis, thus reflecting its homology to the cell death molecule Ced-3 of Caenorhabditis elegans. The fact that caspase 1 has been selected as a target by IpaB accounts for the dual function that concurrently associates cell death and release of mature IL-1β and IL-18, two potent pro-inflammatory cytokines. The use of YVAD-CHO peptides which inhibit caspase 1 protects against Shigella killing of macrophages and macrophages obtained from caspase 1 knockout mice are resistant to apoptotic killing by Shigella, confirming that caspase 1 is central to the process . Although other cell populations are likely to express caspase 1 or very closely related enzymes, it is not yet understood why Shigella cytotoxicity remains restricted to the monocyte/macrophage and dendritic cell populations. The effect of the increased secretion of IL-1β may be enhanced by the associated release of IL-1α, due to the massive cell death triggered by Shigella. In addition, the effect of the higher titers of IL-1 in the infected tissues may be even further enhanced by the lack of IL-1 receptor antagonist (IL-1ra) observed during the early period of development of infection. IL-1ra is a homolog of IL-1 which binds to the same receptor, but has no agonistic function, thus acting as a negative regulator of inflammation in the presence of IL-1. Intravenous infusion of IL-1ra in rabbits undergoing infection of intestinal loops by Shigella has a strong anti-inflammatory effect, thereby reducing the degree of tissue lesions. It also largely reduces the degree of bacterial invasion, thus confirming that Shigella uses inflammation-mediated disruption of the epithelial barrier to enhance its invasive capacity . Recently published data obtained in a model of murine intranasal infection causing acute tracheo-bronchitis followed by condensing alveolitis have made it possible to address the actual role of caspase 1 in the initiation of the innate response to Shigella. A combination of infections carried out in a series of genetically invalidated mice (i.e. caspase-1−/−, IL-1β−/−, IL-18−/−) indicated that IL-1β caused tissue destruction that facilitates bacterial dissemination, whereas IL-18 was involved in the control of bacterial growth, probably due to its effect on interferon γ (IFNγ) production which appears to be essential in the defense against Shigella. It is therefore likely that macrophage apoptosis occurring early after bacteria have crossed the FAE caused both IL-1β-mediated inflammation participating in the inflammatory rupture of the epithelial barrier and facilitation of bacterial dissemination and IL-18-programmed control of Shigella growth.
6Bacterial invasion causes epithelial cells to produce pro-inflammatory molecules: inflammation goes uncontrolled
The fact is now well accepted that in response to bacterial invasion, colonic epithelial cells express a large array of pro-inflammatory molecules, particularly chemokines such as IL-8 . Invasion of epithelial cells by Shigella activates the translocation of nuclear factor (NF)-κB which accounts for the transcription of IL-8. LPS released by intracellular shigellae accounts for the strong and sustained activation of NF-κB . This is likely to further aggravate inflammation by attracting more PMNs to the scene. In the rabbit ligated intestinal loop model of infection, neutralization of IL-8 causes a dramatic effect on the evolution of the infectious process with a major decrease in the number of PMNs immigrating into the lamina propria and invading the epithelium. This confirms that a major function of IL-8 produced by epithelial cells is to attract PMNs to areas where bacterial invasion of the epithelium is occurring. Interestingly, neutralization of the function of IL-8 causes a spectacular decrease in the severity of epithelial lesions, but conversely allows passage of the bacteria in the lamina propria . In consequence, the surface control of Shigella infection achieved by PMNs attracted along an IL-8 gradient occurs at the cost of severe epithelial destruction.
6.1What is the benefit for Shigella of causing inflammation? A case of ‘fatal attraction’
Triggering such a strong inflammation at the early stage of infection sounds like an error for a bacterium which primarily aims at avoiding its killing by the innate immune reaction. Even rapid killing of resident macrophages may not compensate for the bactericidal effect resulting from the strong local influx of PMNs that respond to inflammatory signals. However, infusion of IL-1ra during infection of rabbit intestinal tissue has clearly shown that not only was inflammation and tissue destruction controlled, but also the number of bacteria invading those tissues decreased dramatically . In addition, injection of a neutralizing monoclonal antibody directed against CD18 which blocks PMN migration to tissues also controls both the inflammatory destruction of tissues and the number of invading bacteria . It was then possible to model a process of ‘fatal attraction’ in vitro in which apical shigellae are able to stimulate transmigration of basal PMNs, thus disrupting the impermeability of the epithelial lining and allowing quick basolateral invasion of epithelial cells.
It is therefore likely that the spread of bacterial invasion, at a distance from the FAE, is maintained by a constant influx of PMNs which subvert the epithelial integrity, thus allowing further bacterial entry into the epithelium. Once inside epithelial cells, bacteria grow and spread from cell to cell, thus achieving efficient intracellular colonization in a protected environment. In summary, the lymphoid follicles of the colonic and rectal mucosa act as ‘Trojan horses’ allowing both crossing of the intestinal epithelial barrier and early firing of the inflammatory response following cytokine signals engaged during the process of macrophage death. Extension of infection from these sites to larger surfaces of the epithelium may proceed via a combination of inflammation and invasion for bacteria which are inefficient at invading the epithelial barrier apically. This ‘snowball’ effect, in which inflammation and invasion combine their effects, probably explains the severity of the inflammatory destruction which is characteristic of the disease. The different steps of translocation, rupture, invasion and destruction of the intestinal barrier by Shigella are summarized in Fig. 7.
7What finally controls Shigella growth?
Three recent contributions shed some light on the mechanism of the healing process, although it is still difficult to connect these observations. IFNγ seems to be essential to the control of early stages of infection as recently demonstrated in a murine model of lung infection . In this model, knockout animals for the IFNγ gene appear to be five orders of magnitude more susceptible to Shigella infection than the wild-type animals. In addition, bronchiolitis and alveolitis of increasing severity develop, compared to the wild-type animals, with uncontrolled bacterial growth. NK cells may be the major source of IFNγ in this situation. This experimental observation is worth analyzing in the context of the clinical situation since down-regulation of IFNγ transcription has been reported during the early stage of Shigella infection in patients . This down-regulation, which remains to be explained in molecular and cellular terms, may facilitate early multiplication of Shigella in tissues. PMNs are likely to play a major role in the final control of Shigella infection. As previously shown, early influx of PMNs breaks epithelial permeability, thereby facilitating invasion by opening the way to extensive epithelial colonization. Unexpectedly, shigellae do not escape the phagocytic vacuole in these cells. Trapped in a vacuole, in spite of the expression of their invasive phenotype, bacteria may undergo killing both by oxygen radicals and by antibacterial proteins such as the bactericidal permeability increasing (BPI) protein . Finally, recent evidence indicates that neutralization of CD14, a pattern recognition receptor which participates in the signalling pathway activating phagocytic cells in the presence of LPS, facilitates overgrowth of Shigella in invaded mucosal tissues. This is a strong indication that during the infectious process, LPS sensing by phagocytes is used as a signal that upgrades the antibacterial response .
It is clear that the period to come will be devoted to the understanding of how the innate immune system eventually clears Shigella infection and how this response affects the adaptive immune response, both qualitatively and quantitatively. Meanwhile, progress in the understanding of how the adaptive immune system completes bacterial clearance at the stage of primary infection and prevents infection in immunized people should help in developing improved vaccine candidates against this disease.
The author would like to express his sincere thanks to Armelle Phalipon, Claude Parsot, Guy Tran Van Nhieu and all the members, past and present, of the Unité de Pathogénie Microbienne Moléculaire as well as to our collaborators, both at the Institut Pasteur and outside, for their outstanding contribution to the understanding of the pathogenesis of shigellosis.