The Gram-positive bacterium Listeria monocytogenes is a ubiquitous pathogen that thrives in diverse environments such as soil, water, various food products, humans and animals. The disease caused by this bacterium, listeriosis, is acquired by ingesting contaminated food products and mainly affects immunocompromised individuals, pregnant women and newborns. Listeriosis manifests as gastroenteritis, meningitis, encephalitis, mother-to-fetus infections and septicaemia, resulting in death in 25–30% of cases. The diverse clinical manifestations of infection with L. monocytogenes reflect its ability to cross three tight barriers in the human host. Following ingestion, L. monocytogenes crosses the intestinal barrier by invading the intestinal epithelium, thereby gaining access to internal organs. During severe infections, crossing the blood–brain barrier results in infection of the meninges and the brain, and in pregnant women, crossing the fetoplacental barrier leads to infection of the fetus1.

L. monocytogenes infection has been a useful model for evaluation of the cellular interactions that are crucial for the initiation of the host T-cell response. However, this aspect of listerial biology is well documented and has recently been reviewed elsewhere2. This Review highlights the many ways in which L. monocytogenes manipulates its mammalian host, and it focuses on the lessons this bacterium has taught us in cell biology, bacterial pathophysiology, virulence-factor regulation, and bacterial adaptation to the host cytosol.

Indeed, the ability of L. monocytogenes to invade and replicate in different cell types has been extensively studied and has revealed the sophisticated relationship between the bacterium and its host. The breadth of information gathered on the elaborate mimicries that are used by L. monocytogenes to subvert host processes has made it an exceptional tool for the study of cellular processes such as actin-based motility, growth-factor-mediated signalling, endocytosis and cellular adhesion. Furthermore, available animal models, genetic tools and genomics have facilitated the compilation of information on different aspects of L. monocytogenes biology and have made this bacterium one of the most useful model organisms for the study of bacterial pathogenesis and pathophysiology.

An intracellular bacterium and a cell biologist

L. monocytogenes is a facultative intracellular bacterium. Its life cycle reflects its remarkable adaptation to intracellular survival and multiplication in macrophages and other cell types1,3,4 (Fig. 1). Similar to the situation for most pathogens, the invasion of macrophages by L. monocytogenes is a passive process, but entry into non-professional phagocytes is induced by binding of the bacterial surface proteins internalin A (InlA) and InlB to receptors on the host cell. Both of these invasins are necessary and sufficient for bacterial entry into cell types such as enterocytes, hepatocytes, fibroblasts, epithelial cells and endothelial cells, but InlA-mediated entry is restricted to the smaller number of cell types that express its receptor. Entry of L. monocytogenes into mammalian cells is a dynamic process that requires actin polymerization and membrane remodelling, and is an excellent example of how a bacterium can manipulate host-cell signalling and endocytic pathways to its advantage. L. monocytogenes can also harness the actin-polymerization machinery in the cytoplasm to facilitate intracellular and intercellular movement. New mechanisms by which L. monocytogenes manipulates the host cell are emerging through the use of microarray analyses that aim to determine the genes that are specifically activated by bacterial entry into the host cell5,6.

Figure 1: Schematic representation and electron micrographs of the Listeria monocytogenes life cycle.
figure 1

a | L. monocytogenes induces its entry into a non-professional phagocyte. b | Bacteria are internalized in a vacuole (also known as a phagosome). c,d | The membrane of the vacuole is disrupted by the secretion of two phospholipases, PlcA and PlcB, and the pore-forming toxin listeriolysin O. Bacteria are released into the cytoplasm, where they multiply and start to polymerize actin, as observed by the presence of the characteristic actin tails (see Supplementary information S3 (figure)). e | Actin polymerization allows bacteria to pass into a neighbouring cell by forming protrusions in the plasma membrane. f | On entry into the neighbouring cell, bacteria are present in a double-membraned vacuole, from which they can escape to perpetuate the cycle. F-actin, filamentous actin. Electron micrographs ac,ef are reproduced with permission from Ref. 113 © (1998) European Molecular Biology Organization, and d is reproduced with permission from Ref. 30 © (1992) Elsevier.

InlB: subverting cellular-signalling and endocytic pathways. Binding of InlB to its cellular receptor Met results in the entry of L. monocytogenes into different cell types. Met is a protein tyrosine kinase, and the endogenous ligand of this receptor is hepatocyte growth factor (HGF)7 (Fig. 2). In vivo, Met is expressed mainly by cells of epithelial origin, whereas HGF is produced mainly by fibroblasts and stromal cells. The binding of HGF to Met activates cellular survival and proliferation signals, and it induces cytoskeletal rearrangements that function in cellular motility and differentiation. Binding of InlB activates the protein-tyrosine-kinase activity of Met, as well as the phosphatidylinositol 3-kinase (PI3K) and the Ras–mitogen-activated protein kinase (MAPK) pathways, all of which are required for the uptake process7,8,9. Interestingly, although InlB and HGF both bind and activate Met, InlB does not strictly mimic HGF. Indeed, the kinetics of InlB-induced signalling are different from those of HGF-induced signalling7, and at an equal concentration, InlB seems to induce a more potent activation of the Ras–MAPK pathway than does HGF10. Differences in signalling might be explained by the finding that InlB also binds gC1qR (the receptor for the globular part of complement component C1q), which might therefore function as a co-receptor for InlB11. Furthermore, InlB and HGF do not compete for binding to Met7, indicating that they might bind distinct sites on Met. These results are consistent with the fact that InlB and HGF have no sequence homology and are structurally unrelated. Crystal structures of InlB bound to Met could provide valuable information about the molecular mechanism that underlies the activation of Met by InlB.

Figure 2: Met signalling induced by hepatocyte growth factor (HGF) and internalin B (InlB).
figure 2

a | Phosphorylation of Met leads to the recruitment and activation of many transducers, which in turn recruit cytosolic signalling proteins. Signalling mediated by HGF activates survival and proliferation signals, and it induces cytoskeletal rearrangements that are important for cellular motility and differentiation. On stimulation with HGF, the endocytosis of Met, similar to most signalling receptors, is an important regulatory mechanism that downregulates the cell-surface expression of the activated receptor. b | Met signalling mediated by the Listeria monocytogenes protein InlB induces cytoskeletal rearrangements that are important for bacterial entry into non-phagocytic cells. Clathrin components of the endocytic machinery are also recruited to the site of entry. The link between the cytoskeletal machinery (shown on the right) and the endocytic machinery (shown on the left) is still unclear. InlB, through the GW repeats at its C terminus, also binds gC1qR (the receptor for the globular part of complement component C1q) and glucosaminoglycans (GAGs), which are negatively charged polysaccharides that are present at cell surfaces. Both components might contribute to entry of L. monocytogenes by modulating the interaction of InlB with Met. ABI, Abl interactor 1; Arp, actin-related protein; CD2AP, CD2-associated protein; CIN85, Cbl-interacting protein of 85 kDa; EPS15, epidermal-growth-factor-receptor substrate 15; F-actin, filamentous actin; GAB1, GRB2-associated binding protein 1; GGA3, Golgi-localized, γ-ear-containing, ADP-ribosylation-factor-binding protein 3; GRB2, growth-factor-receptor-bound protein 2; HRS, HGF-regulated tyrosine-kinase substrate; N-WASP, neural Wiskott–Aldrich syndrome protein; PI3K, phosphatidylinositol 3-kinase; PLCγ, phospholipase C-γ; PtdIns(3,4,5)P3, phosphatidylinositol-3,4,5-trisphosphate; PtdIns(4,5)P2, phosphatidylinositol-4,5-bisphosphate; SHC, SRC-homology-2-domain-containing transforming protein C; SOS, son of sevenless; Ub, ubiquitin; WAVE, Wiskott–Aldrich syndrome protein (WASP)-family verprolin homologous protein.

The signalling pathways that are activated by InlB ultimately lead to cytoskeletal rearrangements and entry of L. monocytogenes. How activation of the PI3K and the Ras–MAPK pathways leads to cytoskeletal rearrangements has been extensively studied, and many of the proteins that are crucial for invagination and internalization have been identified. Local actin remodelling at the site of InlB attachment is mediated by the recruitment and activation of the actin-nucleation complex, Arp2/3, which promotes actin nucleation and polymerization (discussed later). The mechanism of Arp2/3 activation seems to be cell-type dependent, but it involves a combination of the small GTPases Rac and CDC42 and proteins of the Wiskott–Aldrich syndrome protein (WASP) family, which includes neural WASP (N-WASP) and WAVE12. Proteins of the Ena/VASP family (enabled homologue/vasodilator-stimulated-phosphoprotein family), which promote actin-filament elongation, are also central to the process. In addition, cofilin, which is essential for depolymerization of actin filaments, functions successively as a stimulator and a downregulator of actin rearrangements that occur during the internalization process12,13. All of these components that are recruited by the binding of InlB to Met also have a role in growth-factor-receptor activation. Therefore, although there are differences between the kinetics of signalling mediated by InlB and HGF, the machinery that is recruited to the site of activation seems to be the same for both molecules, showing the utility of L. monocytogenes as a tool to study cellular signalling by Met or other growth-factor receptors.

Recently, the study of InlB-induced internalization revealed an unexpected mechanism used by L. monocytogenes during host-cell entry. L. monocytogenes invades epithelial cells by subverting clathrin-mediated endocytosis14 (Fig. 2b; see Supplementary information S1 (figure)), a process that is used by mammalian cells to take up nutrients and to recycle membrane proteins. Owing to their size, which is usually 1–3 μm, bacteria were thought to enter cells through a mechanism related to phagocytosis (that is, an actin-dependent mechanism), which differs from 'endocytosis', a process that is thought to internalize particles no larger than 120 nm and that was considered to be actin independent until recently. Therefore, the finding that L. monocytogenes can induce its internalization by using clathrin-dependent machinery was surprising and indicated that clathrin-coated structures can engulf much larger particles than previously thought. Cell-surface expression of Met is downregulated by HGF, and similarly, soluble InlB induces the degradation of Met, through monoubiquitylation and clathrin-mediated endocytosis14. Furthermore, as shown using RNA-interference-mediated knockdown, important components of the endocytic machinery are required for internalization of L. monocytogenes. Although the underlying molecular mechanisms have not been defined, other invasive bacteria had previously been reported to use clathrin-mediated entry, implying that this mechanism is not restricted to Listeria species and could be a commonly used mechanism for bacterial entry15,16,17. Although the endocytic machinery is important for entry of L. monocytogenes, this bacterium might exploit other mechanisms, because inhibitors of endocytosis reduce bacterial entry but do not completely abolish it14. Plasma-membrane microdomains known as lipid rafts have also been shown to be important for the entry of L. monocytogenes18. Because lipid rafts are usually associated with clathrin-independent endocytic pathways, this indicates either that L. monocytogenes exploits more than one endocytic mechanism or that the separation among the classes of endocytosis is not as well demarcated as conventionally thought. Further work is necessary to decipher how lipid rafts, clathrin-mediated endocytosis and actin-mediated phagocytosis combine to enable listerial entry and cellular invasion.

InlA: exploiting intercellular junctions. Similar to InlB, InlA induces local cytoskeletal rearrangements in the host cell to stimulate uptake of L. monocytogenes by epithelial cells. InlA is a covalently linked bacterial cell-wall protein that binds the host epithelial-cell protein E-cadherin19 (Fig. 3). E-cadherin is a transmembrane protein that belongs to a large family of cell–cell adhesion molecules that are required for the correct formation of adherens junctions between epithelial cells. E-cadherin is localized at these cellular junctions, where its intracellular domain forms a complex with the cytoskeleton through the catenins (which are cadherin-binding proteins), and its extracellular domain is in contact with E-cadherin molecules on neighbouring cells20. InlA binds the extracellular domain of E-cadherin, but it is the intracellular domain of E-cadherin that is essential for the cytoskeletal rearrangements that are required for bacterial entry21. Because all of the components of the endogenous machinery of cell–cell junctions seem to be recruited on binding of InlA, L. monocytogenes is a good system for the study of cellular adhesion and identification of components that are involved in this process.

Figure 3: Adherens junction and internalin A (InlA)-induced bacterial entry.
figure 3

a | Adherens junctions hold adjacent cells together through the transmembrane protein epithelial cadherin (E-cadherin). The intracellular domain of E-cadherin recruits α-catenin and β-catenin, and α-catenin bridges the actin cytoskeleton and E-cadherin. Formins, which interact directly with α-catenin, are also essential for forming actin cables at cell–cell junctions, although the mechanism by which they achieve this is not understood. b | The receptor for the Listeria monocytogenes protein InlA is E-cadherin. Many components that are important for adherens junctions are recruited to the site of bacterial entry, where the cytoskeletal rearrangements that are required for invasion occur. ARF6, ADP-ribosylation factor 6; ARHGAP10, Rho GTPase-activating protein 10; Arp, actin-related protein; F-actin, filamentous actin; G-actin, globular actin.

Although it is known that assembly and attachment of the E-cadherin, α-catenin and β-catenin complex to the cytoskeleton is central to both intercellular adhesion and L. monocytogenes entry21, neither the mechanisms that hold cells together nor the molecular mechanisms that are required for InlA-dependent entry are as yet fully understood. Until recently, the accepted model of intercellular adhesion proposed that α-catenin anchors the E-cadherin–β-catenin complex to the actin cytoskeleton, providing a stable structure that maintains tissue integrity. However, it has become apparent that the dynamics of this process are more complex. As was recently shown, α-catenin cannot simultaneously interact with actin filaments and the E-cadherin–β-catenin complex, indicating that α-catenin is not an actin anchor but is, instead, an actin-filament organizer22,23. Further investigation is required to understand this mechanism fully; however, the dynamic interactions between the cadherin–catenin complex and the underlying actin cytoskeleton are consistent with the findings that L. monocytogenes can regulate and rearrange actin structures at intercellular junctions through adhesion to E-cadherin, and further emphasize the validity of the listerial model for analysing events at intercellular junctions.

The validity of using L. monocytogenes as a tool to study cell–cell junction formation was shown by a recent study of InlA-dependent entry that identified the protein ARHGAP10 (Rho GTPase-activating protein 10) as a novel cellular component that is involved in the recruitment of α-catenin to cell–cell junctions. This study also showed that ARHGAP10 was essential for listerial entry (see Supplementary information S2 (figure))24. Overexpression of ARHGAP10 disrupted the cytoskeleton and increased the local concentration of α-catenin, indicating that ARHGAP10 has a direct role in regulation of the dynamics of cell–cell junction formation. Furthermore, ARHGAP10 was shown to control the activity of RhoA and CDC42, two proteins that regulate cell–cell junction formation.

Components that generate the tension that is required to hold neighbouring cells together, that is, myosin VIIA and its ligand vezatin25, have been found to be important for L. monocytogenes entry, indicating that these components might generate the force that is necessary for engulfment of the bacterium through phagocytosis26. These findings also indicate that the tension that holds two cells together could be similar to the tension that is exerted during phagocytosis, as if each cell is attempting to engulf its neighbour. An analogous process known as 'frustrated phagocytosis' occurs when macrophages adhere to immune-complex-coated surfaces27.

Harnessing the actin-polymerization machinery. Following internalization into a host-cell vacuole, L. monocytogenes lyses the membrane-bound phagosome (discussed later) and escapes into the cytoplasm, where it can polymerize host actin and propel itself through the cell and into neighbouring cells28. The ability to spread from cell to cell without coming in contact with the extracellular milieu allows the bacterium to propagate through tissues and avoid contact with circulating antibodies or other extracellular bactericidal compounds. At the leading edge of moving cells, the control of actin polymerization is a complex mechanism triggered by intricate signalling pathways that take place at the plasma membrane. L. monocytogenes bypasses these signalling pathways and directly nucleates actin constitutively, making it a simple and effective model for studying the dynamics of actin polymerization.

L. monocytogenes polymerizes actin asymmetrically along its surface, producing an actin tail that propels the bacterium through the cytoplasm28,29 (see Supplementary information S3 (figure)). Polymerization of host actin is mediated by the bacterial surface protein ActA, the first protein that was identified to have functions that promote actin nucleation30,31,32. It was later found that ActA mimics its eukaryotic counterparts, proteins of the WASP family (which includes N-WASP and WAVE)33. The N-terminal region of ActA, which is essential for actin-based motility34, has homology to the C-terminal region of WASP, which binds the actin-nucleation complex Arp2/3 (Ref. 35). Accordingly, both ActA and WASP-family proteins function as nucleation-promoting factors (NPFs) for the Arp2/3 complex and are involved in forming a ternary complex that is composed of Arp2/3, an NPF and actin. The Arp2/3 complex consists of seven proteins, including two proteins that are related to actin, Arp2 and Arp3, which (as a result of their structural similarity to actin) are thought to function as a template for polymerization. Discovery of the Arp2/3 complex ensued from studies of ActA partners, thus the role of Arp2/3 as a key actin-nucleating complex was consequently revealed.

The L. monocytogenes model has also been useful for understanding the importance and the function of VASP, the other main ligand of ActA. VASP binds the central proline-rich domain of ActA and promotes efficient actin-based motility34,36,37,38, highlighting the importance of this protein in actin polymerization. Only recently, however, have the complex molecular mechanisms of VASP activity begun to emerge. VASP seems to promote listerial motility by recruiting the actin-binding protein profilin, which promotes polymerization at actin-filament barbed ends39. VASP also seems to induce faster growth of the actin network at the bacterial surface by causing the release of Arp2/3 from ActA40. In addition, VASP seems to decrease Y-branch formation, thereby increasing parallel alignment of actin filaments41. Although their mechanistic role is not fully elucidated, proteins of the Ena/VASP family are important in the formation of actin fibres, filopodial tips and the lamellipodial leading edge42.

Intracellular and intercellular movement using actin polymerization is not restricted to L. monocytogenes. A growing number of intracellular pathogens, including Rickettsia species, Shigella species, mycobacteria, Burkholderia pseudomallei and vaccinia virus, show this feature during infection (see Supplementary information S3 (figure), and for recent reviews, see Refs 43,44).

A paradigm in pathophysiology

As a pathogen that displays such interesting features as strong T-cell activation, a sophisticated relationship with its host and crossing of protective barriers, L. monocytogenes has more recently also emerged as a model to study the pathophysiology of a complex bacterial infection. Findings from in vitro work have been used to bypass stringent host species specificity and generate relevant model systems to study infection in vivo. In this respect, L. monocytogenes is a good example of how in vitro studies can help to generate animal models that more closely reflect human infection.

For many years, the animal model used to study L. monocytogenes infection was intravenous infection of mice. This model provides a dose-dependent infection with dissemination of the bacteria into organs and was crucial in the discovery of cell-mediated immunity3. However, recent molecular evidence showed that the mouse model is inadequate to study the crossing of barriers that is characteristic of listeriosis. It had long been known that oral inoculation of mice (rather than intravenous infection), which most closely reflects the human mode of infection, is not efficient because only small numbers of L. monocytogenes cross the mouse intestinal barrier. The reason for this was uncovered by molecular in vitro studies that showed that a single amino-acid difference in the mouse cellular receptor for InlA, E-cadherin, prevented it from binding InlA, thereby showing the inadequacy of the mouse model for study of the invasive role of InlA45 (Fig. 4). Consequently, a novel animal system was developed to study the crossing of the intestinal barrier: a transgenic mouse that expresses human E-cadherin on the surface of enterocytes46. This model showed that InlA has a key role in disease, because it is essential for crossing of the intestinal barrier. In this model system, InlA could interact with E-cadherin, and a wild-type strain of L. monocytogenes was able to cause disease through oral inoculation.

Figure 4: Host specificity of Listeria monocytogenes proteins internalin A (InlA) and InlB.
figure 4

a | InlA and InlB can bind and induce entry of L. monocytogenes into human cells that express the respective cell-surface receptors, epithelial cadherin (E-cadherin) or Met. However, a single amino-acid change in E-cadherin (at position 16; see b) prevents InlA from binding mouse E-cadherin, and for unknown reasons, InlB cannot recognize or activate guinea pig or rabbit Met. b | A diagrammatic representation of the crystal structure of the leucine-rich-repeat region of InlA (purple) bound to E-cadherin (green) is shown. The position of the crucial proline residue at position 16 of E-cadherin is indicated. It is this residue that determines intermolecular recognition and therefore host specificity. The crystal-structure representation is reproduced with permission from Ref. 114 © (2002) Elsevier.

So far, E-cadherin has been found only at cell–cell junctions and on the basolateral face of epithelial cells, so the mechanism by which L. monocytogenes gains access to E-cadherin was enigmatic. Two hypotheses were put forward to explain how InlA could target E-cadherin46. The first hypothesis proposed that L. monocytogenes could gain access to E-cadherin at the tips of intestinal microvilli, where apoptotic epithelial cells slough off. The second hypothesis proposed a synergy between InlA- and InlB-dependent internalization, because activation of Met by HGF has been shown to stimulate the disassembly of junctions between epithelial cells46. Recent results have shown that L. monocytogenes invades the intestinal epithelium at sites of cell extrusion at the tips of villi47 and that the contribution of InlB to crossing of the intestinal barrier is insignificant in vivo48. Whether the mechanism of intestinal invasion is used to cross other barriers is unknown. Synergy between InlA and InlB could still be important for crossing of the placental barrier49.

Because the transgenic mice described here express human E-cadherin only on enterocytes, it is not possible to study the role of InlA in deeper tissues. The generation of transgenic mice that express E-cadherin on all cells is in progress. At present, the role of InlA can also be studied in guinea pigs or rabbits, because E-cadherin is recognized by InlA in these animals. However, recent studies show a species specificity for InlB: InlB does not recognize or activate guinea-pig or rabbit Met. Therefore, the role of InlA and InlB in infection cannot be studied using these animal models48. A human model remains the best model system for studying listerial infection, and human explants have been successfully used to determine the mechanism by which L. monocytogenes crosses the maternofetal barrier49. Human placental villus explants, together with primary or immortalized trophoblastic cells, were used to show that InlA is a key bacterial protein that is required for crossing of the human maternofetal barrier, although InlA is not essential for this role in the pregnant guinea-pig model49,50. Crossing of the blood–brain barrier is still poorly understood. However, the optimization of animal models should help to decipher this crucial step.

Other model systems are being developed to identify the host factors that are required for the intracellular survival of L. monocytogenes and possibly of other intracellular pathogens. Drosophila melanogaster has attracted attention as a model because of the many genetic and immunological studies that have been carried out using this organism, and it has been successfully used to test L. monocytogenes virulence51. Moreover, genome-wide RNA-interference screens in D. melanogaster S2 cells (which are macrophage-like cells) have revealed many new host factors that are important for entry into the host cell, escape from the vacuole and intracellular growth of L. monocytogenes52,53. Another noteworthy organism that has been shown to support a listerial infection is Caenorhabditis elegans54.

Novel regulatory mechanisms

L. monocytogenes is a facultative intracellular pathogen that can live both inside and outside its host. This bacterium has therefore evolved sophisticated regulatory mechanisms to ensure that virulence factors are optimally expressed when they are required. These regulatory mechanisms might prove to be general mechanisms used by other bacteria.

PrfA: a tightly regulated protein. Most of the virulence proteins that have been identified in L. monocytogenes are under the control of one transcriptional regulator, PrfA, which itself is tightly regulated by environmental conditions. During exponential growth, prfA is mainly transcribed as a bicistronic mRNA from the plcA promoter. By contrast, during stationary phase, a monocistronic mRNA is preferentially transcribed from a promoter upstream of prfA55,56,57 (Fig. 5a). Similar to many other pathogens, L. monocytogenes can sense conditions in the mammalian host and respond by expressing virulence genes. A novel regulation of PrfA by temperature, owing to the structure of the upstream untranslated region of the prfA mRNA, was recently discovered58 (Fig. 5b). At low temperature (30°C), the prfA leader transcript controls translation of the downstream mRNA by forming a secondary structure that masks the ribosome-binding site. At mammalian host temperature (37°C), this structure partially melts to expose the ribosome-binding site, thereby allowing translation to occur. Fusion of the prfA leader transcript to the gene that encodes green fluorescent protein (GFP) also resulted in thermoregulation of GFP. This mechanism might be used by other bacteria, as has previously been suggested for the Yersinia pestis activator protein LcrF59. Such post-transcriptional regulation of prfA allows rapid expression of the encoded transcription factor and therefore efficient transcription of virulence factors as soon as the bacterium enters the host.

Figure 5: The PrfA regulator.
figure 5

a | Schematic representation of the prfA region. During exponential phase, prfA is mostly transcribed as a bicistronic product from the promoter (P) upstream of plcA. By contrast, during stationary phase, prfA is mostly transcribed as a monocistronic product from P1 and/or P2. b | Mechanism that controls the thermoregulated expression of PrfA in the promoter upstream of prfA. At low temperatures (≤30°C), a secondary structure forms in the untranslated region of prfA, and this prevents ribosome binding and therefore expression of PrfA. At high temperatures (≥37°C), melting of the prfA untranslated region allows ribosomes to bind and PrfA expression to occur. SD, Shine–Dalgarno sequence. This figure is modified with permission from Ref. 58 © (2002) Elsevier.

Other environmental conditions — such as osmolarity, iron concentrations, pH, the presence of fermentable sugars, stress (through σB), and conditions in the host-cell intracellular compartment — have been shown to regulate prfA and PrfA-controlled genes through mechanisms that are not completely understood60,61. Post-translational regulation of PrfA by a putative cofactor is suggested by its structure, which resembles that of the cyclic-AMP receptor62,63. The number of mechanisms that regulate PrfA is probably indicative of the importance of this crucial virulence factor during infection.

Listeriolysin O: a pH-sensing protein. L. monocytogenes thrives in the cytoplasm of numerous cell types. Following internalization, bacteria escape from membrane-bound phagosomes by secreting two phospholipases, PlcA and PlcB, and the pore-forming toxin listeriolysin O (LLO), thereby gaining access to the cytoplasm (Fig. 1).

Although LLO is a member of a large family of cholesterol-dependent cytolysins that are secreted by numerous Gram-positive bacteria, L. monocytogenes is the only pathogen that secretes this type of toxin inside the host cell. Therefore, secretion of LLO must be tightly regulated, because the bacterium needs to balance efficient escape from the vacuole against prevention of host-cell damage to allow its intracellular survival.

Unlike other pore-forming toxins, the activity of LLO is optimal at acidic pH (<6). So, LLO is fully active in the acidic environment of the phagosome, but it is less active at the neutral pH of the host-cell cytoplasm. An L. monocytogenes strain that was constructed to express the pore-forming toxin perfringolysin O, which is active at neutral pH, can also escape the vacuole but is toxic to the host cell and is avirulent in mice, highlighting the importance of pH regulation of LLO activity64. The molecular mechanism that underlies the optimum pH for LLO activity was only recently identified: LLO was found to be stable at acidic pH, and to unfold and aggregate at neutral pH65 (Fig. 6). At acidic pH, pore formation occurs when LLO monomers oligomerize into large complexes that penetrate the membrane by extending transmembrane β-hairpins66. At neutral pH, acidic amino-acid residues in the transmembrane domain initiate irreversible denaturation of the β-hairpins of LLO, leading to inactivation of the pore-forming function of LLO65. The elaborate structure of LLO therefore allows this protein to sense and respond to its environment.

Figure 6: Listeriolysin O (LLO) pore-forming mechanism.
figure 6

At low (acidic) pH, the soluble LLO monomer interacts with the host-cell plasma membrane, presumably by binding cholesterol. On contact with the membrane, structural rearrangements in one monomer expose residues that can form hydrogen bonds with other monomers, thereby allowing oligomerization into a pre-pore complex. Following oligomerization, two α-helical bundles from each monomer extend to form transmembrane β-hairpins (red) that punch through the membrane. Pores formed by LLO and other cholesterol-binding proteins can be 250–300 Å in diameter. At neutral pH, domain 3 (D3) of the monomer prematurely unfolds, rendering the protein unable to form pores. This figure is modified with permission from Ref. 115 © (2005) American Society for Microbiology.

In addition to its pore-forming ability, LLO has the surprising ability to induce potent signalling in the host cell. It has been shown to activate NF-kB67, MAPK68, phosphatidylinositol69,70, calcium71,72,73 and protein-kinase-C74 signalling pathways. As a pore-forming toxin with cholesterol as the only identified ligand, this signalling function is puzzling. LLO could have a signalling function at neutral pH, as has been suggested for another member of the cholesterol-dependent-cytolysin family, Streptococcus intermedius intermedilysin, which binds the host cellular receptor CD59 (Ref. 75). However, it has been difficult to separate the signalling function of LLO from its pore-forming ability. For example, MAPK activation occurs on treatment of host cells with LLO, but it also occurs after exposure to a mild detergent, indicating that signalling could be the result of membrane damage rather than a specific activity of LLO68. Clearly, the signalling activity of LLO requires further investigation.

A model of bacterial adaptation

One particularly interesting feature of L. monocytogenes is its capacity to thrive both inside and outside the host. This ability to colonize a broad range of ecosystems is reflected in the genome of the organism, which contains an unusually large number of regulatory and transport proteins (11.6% of all predicted genes)76. With one-quarter of its transport proteins dedicated to carbohydrate transport, L. monocytogenes has the largest number of phosphotransferase systems that have been described for bacteria so far. Furthermore, the genome contains 16 putative two-component regulatory systems, more than are found in other pathogenic bacteria and comparable to the number found in ubiquitous bacteria such as Bacillus subtilis . Inactivation of several of these two-component regulatory systems has revealed their importance for the resistance of L. monocytogenes to various stresses and for survival in the host77,78,79,80,81,82. In addition to these conventional bacterial two-component regulatory systems, which phosphorylate histidine and aspartic-acid residues, L. monocytogenes also uses a eukaryotic-like serine/threonine protein phosphatase system, known as Stp, to regulate the translation elongation factor EF-Tu and, consequently, virulence83.

Outside the host, L. monocytogenes is particularly well adapted to grow at low temperatures, making it difficult for the food industry to rely on refrigeration to control listerial contamination84,85,86,87. More remarkable is the ability of L. monocytogenes to survive in the host. The resistance of L. monocytogenes to acidic conditions88,89 and to bile salts90,91,92 makes this pathogen particularly adept at infecting the gastrointestinal tract. Consistent with the many bile-resistance genes in the L. monocytogenes genome, a recent study using in vivo bioluminescence imaging (a non-invasive procedure that allows bioluminescent L. monocytogenes to be traced through the body of a mouse) has shown an intriguing new reservoir for L. monocytogenes, the gall bladder, where bile is stored and concentrated93. Whether this organ is indeed colonized during human listeriosis is unknown.

Survival and replication in the cytosol of many types of host cells, including macrophages and epithelial cells, is one of the distinguishing features of infection with L. monocytogenes3,4. Other extracellular bacteria and intracellular pathogens that normally reside inside a vacuole cannot replicate in the cytosol of the host cell94, highlighting that L. monocytogenes has evolved specific mechanisms to grow in the host-cell cytosol. Although intracytosolic survival remains poorly understood, two key proteins for intracellular growth have been identified: the sugar-uptake system (Hpt) and lipoate protein ligase A1 (LplA1). Hpt (which is homologous to the mammalian translocase of glucose-6-phosphate) is required for optimal intracellular replication95. Interestingly, transcription of hpt is regulated by PrfA and is therefore activated upon entry into the cytosol of the host cell. Expression of Hpt allows the bacterium to use glucose-1-phosphate, which is an available carbon source in the host-cell cytoplasm. Other bacterial pathogens such as Escherichia coli , Salmonella enterica, Shigella flexneri and Chlamydia trachomatis have a homologue of hpt, indicating that this transporter system could have a general role in intracellular survival. The other factor that has been identified to be important for intracellular growth, LplA1, catalyses the formation of a covalent link between lipoic acid and specific protein targets96. A mutant that lacks LplA1 cannot replicate in the cytoplasm of macrophages and has a 300-fold decrease in virulence96. LplA1 seems to be important for lipoylation and full activity of the pyruvate-dehydrogenase enzyme complex in the host-cell cytosol, where lipoic acid is mainly absent. LplA1 has therefore been proposed to be important for the scavenging of lipoyl groups from host molecules.

Interestingly, several reports indicate that the expression of many L. monocytogenes genes is upregulated in the cytosol. A library of Tn917-lacZ L. monocytogenes mutants was screened for higher lacZ expression when present in the cytosol of macrophages than when grown in rich broth. This approach allowed the identification of several L. monocytogenes genes, including plcA and prfA, that are activated after infection of host cells. Recently, two transcriptomic studies have advanced the understanding of adaptation to the host-cell cytosol by showing that L. monocytogenes turns on 500 genes for survival in this cellular compartment98,99. Further analysis of the genes that were identified could elucidate the metabolism of this bacterium in the host-cell cytosol.

Genomics reveals new virulence factors

Major listerial virulence factors have been identified by classical genetics; that is, the search for a mutant with a particular phenotype, followed by gene identification and characterization1. The genetic basis of L. monocytogenes virulence can now be deciphered by exploitation of the genome sequences of L. monocytogenes and the closely related non-pathogenic species Listeria innocua 76, using three main strategies: first, the analysis of genes that are present in L. monocytogenes and absent from L. innocua; second, the study of genes that encode potentially interesting proteins; and third, the use of whole-genome DNA arrays to screen a large population of strains from different Listeria species and to identify genes that are present in, and specific to, all virulent L. monocytogenes strains100.

A first glance at the global genome structure indicates that, in contrast to many other bacterial genomes, the L. monocytogenes genome (strain EGD-e) contains only three insertion sequences and five bacteriophages, none of which seems to have a role in acquisition of virulence genes. Moreover, except for genes of the major virulence locus, virulence genes seem to be dispersed on the chromosome and are not concentrated in pathogenicity islands. One of the most striking features of the L. monocytogenes genome with respect to virulence is the exceptionally large number of genes that encode surface proteins (4.7% of all predicted genes101). These proteins are among the most likely candidates to interact with the host and therefore to be virulence factors. These proteins can be divided into three families. The largest family is composed of 68 lipoproteins, and the second largest family contains 41 LPXTG proteins (including InlA), which are anchored to the cell wall by a sortase (an enzyme that is involved in the covalent linkage of Gram-positive bacterial proteins to the bacterial surface)102,103,104. Inactivation of sortase A (SrtA) or Lsp, a signal peptidase that is involved in maturation of lipoproteins, attenuates virulence, thereby pinpointing the importance of these two classes of surface protein in L. monocytogenes infection103,105,106. The third family of surface proteins includes proteins that are non-covalently attached to the bacterial surface by their C-terminal domains. This family includes GW proteins (such as InlB and Ami), which contain modules of 80 amino acids that contain the dipeptide Gly–Trp (also known as GW modules107) in their C-terminal region. An important outcome from genome-sequence analysis was the discovery of a SecA2-dependent pathway, which allows proteins that lack a typical signal-peptide sequence to be targeted to the bacterial surface or to be secreted108.

Comparison of the L. monocytogenes and L. innocua genomes has proved an efficient approach to identify new virulence factors, including bile-salt hydrolase (Bsh)90, two LPXTG proteins (Vip and InlJ) and a GW protein (Auto)105,109,110. The GW protein Auto is a novel surface-associated cell-wall hydrolase that is required for entry into eukaryotic cells. Its contribution to listerial infection is still unclear but might also be linked to the release of immunologically active cell-wall components that could interact with components of the innate immune system109. The LPXTG protein Vip, which is positively regulated by the transcriptional activator PrfA, is required for bacterial entry into some eukaryotic cells in vitro and for the infectious process in vivo110. Vip interacts with the host-cell endoplasmic-reticulum chaperone gp96, a protein that is involved in Toll-like-receptor signalling111. The LPXTG protein InlJ, the function of which remains to be elucidated, is a leucine-rich-repeat-containing protein that is structurally related to InlA and InlB105. These three proteins are members of the internalin family, which is a large family of proteins that contain leucine-rich repeats (Fig. 7). The gene that encodes InlJ, similar to the genes that encode five other proteins of the internalin family (InlA, InlB, InlE, InlH and InlI), is conserved in the genomes of pathogenic serovars of L. monocytogenes (that is, 1/2a, 1/2b, 1/2c and 4b) and absent from all other Listeria species100,105, and this is consistent with a role for InlJ in virulence.

Figure 7: The internalin family of proteins.
figure 7

Homologous regions of internalin-family members include an N-terminal signal-peptide sequence, followed by several leucine-rich tandem repeats (LRRs). Several internalins contain an inter-repeat region, which is structurally related to an immunoglobulin-like domain. The internalin family can be divided into three subfamilies on the basis of their association with bacteria. The first subfamily includes internalins that contain an LPXTG amino-acid motif (where X denotes any amino acid), and these are covalently anchored to the cell wall. The second subfamily includes internalin B (InlB), which is loosely associated with the bacterial surface through its GW modules (which contain the amino-acid motif GW). The third subfamily contains five internalins that are predicted to be secreted, because they do not have any surface-anchoring domains. It should be noted that the number of LRRs present in the proteins Lmo2445 and Lmo2470 is unclear, because these proteins contain degenerate LRRs. The number of amino acids in each protein is indicated. This figure is modified with permission from Ref. 101 © (2002) Elsevier.

Conclusions and future perspectives

The study of L. monocytogenes infection highlights the sophisticated relationship between this bacterium and its host, and reveals their long co-evolution and reciprocal adaptation. For decades, L. monocytogenes has been a tool for immunologists. It has now also become a paradigm in bacterial pathogenesis and cellular microbiology.

The availability of the genome sequence of five L. monocytogenes serovars76,100,112 and L. innocua76, and soon of Listeria ivanovii and Listeria welshimeri , will provide further insight into the molecular basis of the pathogenesis determinants of Listeria species. Comparative genomics could also reveal genetic loci that confer specific pathogenic traits to epidemic strains: for example, loci that facilitate adaptation to different environments and that influence the onset of infection. It should be noted that it is often difficult to unravel the function of virulence factors that are identified by post-genomic methods and reverse genetics. To this end, more sophisticated in vivo studies — such as measurement of cytokine production, non-invasive in vivo imaging techniques or infection of ex vivo tissue explants — need to be used to understand the role of these bacterial factors in the interplay between the bacterium and the host. Applications of these novel techniques to the study of Listeria species will probably continue to show that this bacterium is an invaluable model in the fields of cellular microbiology, bacterial pathogenesis and cell biology.