Vacuolar escape of foodborne bacterial pathogens

The USA Center for Disease Control and Prevention (CDC) estimates that 48 million people get sick from a foodborne illness in the USA annually, leading to over a hundred thousand hospitalizations and deaths caused by the presence of intracellular pathogens in raw or undercooked foods. Known foodborne intracellular pathogens include Listeria monocytogenes, Salmonella enterica, Shigella spp. and Staphylococcus aureus (Havelaar et al., 2012; Hoffmann et al., 2012; Horn et al., 2018; Scallan et al., 2011). All these bacteria evolved mechanisms to overcome innate host defense factors, such as saliva, gastric acid and the intestinal mucous layer, and invade intestinal epithelial cells, macrophages and other host cells (Case and Samuel, 2016; Castanheira and García-del Portillo, 2017; Flieger et al., 2018; Mitchell et al., 2016). Within these infected host cells, intracellular pathogens multiply and subsequently transit to other host cells, propagating the infection. They exit the cells via activating host-cell programmed cell death, active breaching of cell membranes or an exocytosis-like mechanism (Flieger et al., 2018).

Intracellular pathogens can invade both immune phagocytes (macrophages, dendritic cells and neutrophils) and non-phagocytic cells (e.g. epithelial cells) (Case and Samuel, 2016; Castanheira and García-del Portillo, 2017; Cossart and Helenius, 2014; Mitchell et al., 2016; Pucciarelli and García-del Portillo, 2017). Although the entry mechanism differs between these cell types, in both cases, after uptake, the bacterium initially localizes in the lumen of a membrane-bound compartment, called a vacuole, which are more bactericidal in phagocytes than in non-phagocytic cells (Günther and Seyfert, 2018) (Box 1; Fig. 1). Intracellular pathogens have evolved two mechanisms to subvert these cellular bactericidal conditions in order to survive and propagate within the host cell (Sprenger et al., 2018). First, some pathogens can actively create a less-bactericidal vacuolar niche to destabilize host defense responses by arresting phagolysosomal maturation (e.g. prevent lysosome fusion) and/or redirecting host organellar trafficking, as reviewed elsewhere (Case and Samuel, 2016; Creasey and Isberg, 2014; Sprenger et al., 2018). Second, pathogens can disrupt the vacuolar membrane to escape into the nutrient-rich cytosol, which allows them to avoid the harsh conditions in endolysosomal compartments (Case and Samuel, 2016; Creasey and Isberg, 2014; Mitchell et al., 2016).

Historically, intracellular pathogens are classified according to their virulence strategies into vacuolar (or phagolysosomal) and cytosolic. Of the intracellular foodborne bacterial pathogens, Salmonella spp. (Jantsch et al., 2003), Campylobacter spp. (Watson and Galán, 2008), Brucella spp. (Celli, 2015), S. aureus (Perskvist et al., 2002; Schröder et al., 2006) and Yersinia enterocolitica (Connor et al., 2018) have been considered to be vacuolar, whereas L. monocytogenes (Bierne et al., 2018) and Shigella spp. (Schroeder and Hilbi, 2008) were considered cytosolic pathogens. For Cronobacter sakazakii, which causes systemic infections and thus must penetrate the intestinal epithelial cell barrier and can also invade epithelial cell lines in vitro (Kim and Loessner, 2008; Mohan Nair and Venkitanarayanan, 2007), the intracellular localization is not well characterized, but electron microscopy analysis has shown that it resides in membrane-bound vacuoles in human brain microvascular endothelial cells (de Mey et al., 1995). However, the boundaries between vacuolar and cytosolic pathogens are fading, and cytosolic pathogens, such as L. monocytogenes, have been found to build vacuolar niches in certain cell types (Birmingham et al., 2008; Kortebi et al., 2017). Conversely, it is increasingly clear from electron microscopy and fluorescence approaches (Creasey and Isberg, 2014; Quereda et al., 2016) that the vacuolar pathogens Campylobacter spp. (Nemelka et al., 2009), S. aureus (Giese et al., 2011) and Salmonella spp. (Birmingham and Brumell, 2006; Knodler et al., 2014; Thurston et al., 2016) can also gain access to the cytosol. Thus, the concept is emerging that all foodborne intracellular pathogens can escape to the cytosol and that this is important for their virulence. With this Review, we provide an overview of the molecular mechanisms that enable foodborne intracellular pathogenic bacteria to escape the vacuoles into the cytosol in macrophages and non-phagocytic cell types.

Infection with the Gram-positive bacteria L. monocytogenes occurs mainly following the ingestion of contaminated raw dairy and seafoods (Radoshevich and Cossart, 2018; Schuppler and Loessner, 2010). Infections with L. monocytogenes are a cause of febrile gastroenteritis and can cause a severe invasive disease called listeriosis, hallmarked by blood poisoning and encephalomeningitis (Radoshevich and Cossart, 2018). Owing to the high mortality rate of listeriosis (20–30%), L. monocytogenes is one of the leading causes of death from a foodborne pathogen (de Noordhout et al., 2014). L. monocytogenes is an archetypal cytosolic pathogen that invades both macrophages and non-phagocytic cells, particularly epithelial cells of the gastrointestinal tract (Radoshevich and Cossart, 2018; Schuppler and Loessner, 2010). Macrophages are generally invaded by host-cell-driven phagocytosis, whereas entry of epithelial cells is a bacteria-driven process (Flieger et al., 2018). In both cases, Listeria enters the first host cell in a so-called primary vacuole, which it escapes from within minutes after uptake, to propagate within the cytosol (Bierne et al., 2018; Radoshevich and Cossart, 2018; Schuppler and Loessner, 2010). Listeria express the actin assembly-inducing protein ActA, which promotes Arp2/3-mediated formation of filamentous (F-)actin. This drives the formation of so-called comet-tails that propel the bacteria through the cell and allow egress of Listeria within cell bodies that are detached from the host cell (Bierne et al., 2018). This not only enables infection of neighboring epithelial cells, but also leads to the infection of macrophages, which ingest these cell bodies through efferocytosis (phagocytosis of apoptotic cells) (Czuczman et al., 2014). The L. monocytogenes bacteria enter these cells in secondary vacuoles that have two membranes: one from the original host cell and one from the recipient cell (Radoshevich and Cossart, 2018).

In both macrophages and epithelial cells, L. monocytogenes can escape from primary and secondary vacuoles using the same well-established mechanism (Nguyen et al., 2019). Its primary component is the pore-forming toxin listeriolysin O (LLO), which together with the phosphatidylinositol-specific phospholipase C (PI-PLC or PlcA) and the broad-range phospholipase C (PC-PLC or PlcB) disrupts the vacuolar membrane, resulting in the escape of the bacteria into the cytosol (Bierne et al., 2018). Expression of LLO, PI-PLC and PC-PLC is controlled by the transcription factor positive regulatory factor A (PrfA), which also regulates transcription of ActA and other virulence genes (de las Heras et al., 2011). LLO promotes L. monocytogenes infection not only via disruption of the vacuolar membrane; in macrophage-like cells, LLO also causes a Ca²⁺ influx that prevents the fusion of late endosomes and/or lysosomes to the pathogen-containing vacuoles (Kühbacher et al., 2018; Shaughnessy et al., 2006). In HeLa cells, these compartments are instead redirected to the plasma membrane, resulting in altered exposure of glycoproteins associated with infection on the host-cell surface (Kühbacher et al., 2018).

The roles of PI-PLC and PC-PLC in vacuolar escape are still unclear. It is possible that the destabilization of the vacuolar membrane is a direct effect of PLC-mediated cleavage of vacuolar lipids, promoting membrane fusion and thereby directly supplementing the activity of LLO, which results in the escape of L. monocytogenes from the vacuole (Alberti-Segui et al., 2007; Montes et al., 2004; Sibelius et al., 1996). A second, not mutually exclusive, possibility is that the effect of PI-PLC and PC-PLC is more complex and involves host-cell signaling; it has been shown in macrophages that PI-PLC promotes vacuolar escape by activating the host-cell protein kinase C-β (PKC-β) via a poorly understood mechanism, which involves translocation of PKC-β to early endosomes (Poussin et al., 2009; Wadsworth and Goldfine, 2002). A third possibility is that the action of PC-PLC might be inhibitory; phosphocholine, produced by PC-PLC from hydrolysis of the phospholipid phosphatidylcholine, directly binds to LLO in HeLa cells and inhibits its activity (La Pietra et al., 2020). This inhibition has been proposed to prevent the premature destruction of the plasma membrane of the host cell (La Pietra et al., 2020). Moreover, the off-target disruption of cellular membranes by LLO is prevented by its acidic pH optimum, which restricts LLO activity to the acidic lumen of the vacuoles (Glomski et al., 2002; Shaughnessy et al., 2006). In macrophages, γ-interferon-inducible lysosomal thiol reductase (GILT; also known as IFI30) and cystic fibrosis transmembrane conductance regulator (CFTR) are two host proteins that are also involved in the regulation of LLO activity (Radtke et al., 2011; Singh et al., 2008). GILT is a lysosomal protein present in the lumen of L. monocytogenes-containing vacuoles in murine bone marrow-derived macrophages that activates LLO by reducing one of its disulfide bonds (Singh et al., 2008). CFTR is a Cl⁻ channel localized on L. monocytogenes-containing vacuoles in macrophage-like cells and might potentiate LLO activity by increasing the vacuolar chloride concentration (Radtke et al., 2011). Conversely, host factors can also inhibit the vacuolar escape; it has been shown that heat-shock protein 70 (Hsp70) family proteins may limit damage to vacuolar membranes in murine bone-marrow derived macrophages by making them more resistant to pressure, potentially via activation of lysosomal acid sphingomyelinase, which alters their membrane composition (Davis et al., 2012).

In epithelial cell lines, but not in macrophages, the escape of L. monocytogenes from vacuoles is enhanced by peptide pheromone from lipoprotein A (pPplA), which is produced as a result of cleavage by signal peptidase II and the protease Eep, and might indicate to a L. monocytogenes bacterium whether the pathogen is present in a vacuole (Xayarath et al., 2015). In the confined vacuolar lumen, the secreted pPplA pheromone readily increases to high levels, which facilitates its re-uptake via the cysteine transport-associated protein (CtaP), thereby signaling in an autocrine fashion. This accumulation of pPplA in L. monocytogenes eventually promotes its vacuolar escape via an unknown mechanism, possibly by upregulating expression of a factor that stabilizes LLO-generated membrane pores (Xayarath et al., 2015).

For a long time, L. monocytogenes was thought to always escape to the cytosol following cellular invasion. However, it is now clear that there is also a population of bacteria residing in vacuoles in both macrophages and epithelial cells (Birmingham et al., 2008; Kortebi et al., 2017). These vacuolar bacteria are in a quiescent non- (or slow-)growing state and can be less susceptible to antibiotics and immune clearance (Birmingham et al., 2008; Kortebi et al., 2017). Moreover, Listeria also avoids escape from the entry vacuole in intestinal goblet cells by transcytosis, which allows its transmission to a neighboring cell without passing through the cytosol (Nikitas et al., 2011). This means that L. monocytogenes can survive the harsh conditions within the vacuoles. Although not studied in this context, it is interesting that L. monocytogenes possesses an acid-tolerance response (ATR) system that allows it to survive the low pH conditions in food or in the stomach (Smith et al., 2013), which might also be responsible for its survival in acidic vacuoles. In line with this hypothesis, it has been reported that exposure of L. monocytogenes to acid stress, which upregulates the ATR system, increases invasion of epithelial cells and bacterial proliferation in macrophage-like cells (Conte et al., 2000). However, the resistance to acid stress is also beneficial for the cytosolic population of bacteria, because, as mentioned above, LLO has a low pH optimum and the vacuole is already acidified prior to bacterial escape (Glomski et al., 2002; Shaughnessy et al., 2006).

The gram-negative bacteria S. flexneri and S. dysenteriae are major causes of bacillary dysentery and outbreaks in developed countries occur mainly through contaminated food (Niyogi, 2005). S. dysenteriae type I produces potent poisons known as Shiga toxins, which cause hemolytic uremic syndrome (HUS), a severe kidney complication (Tesh, 2012). Although neutrophils and monocytes can effectively kill ingested S. flexneri (Hathaway et al., 2002; Weinrauch et al., 2002), the bacteria can escape from phagosomes into the cytosol in macrophages (Schroeder and Hilbi, 2008). However, macrophages are probably not the primary cell type targeted by Shigella spp., as their infection induces macrophage cell death (Schroeder and Hilbi, 2008). Instead, macrophage infection might serve to traverse the epithelial barrier, as bacteria released from the dying macrophages can invade nearby epithelial cells through the basolateral membrane (Ashida et al., 2015). In both macrophages and epithelial cells, ingested bacteria can disrupt the vacuolar membrane and propagate in the cytosol (Ashida et al., 2015; Carayol and Tran Van Nhieu, 2013) and this can happen shortly after invasion (within 10 min in HeLa cells) (Mellouk et al., 2014). The entry of the bacterium in non-phagocytic cells and the subsequent lysis of the vacuole are promoted by effector proteins that are injected into the cytosol of the host cell by a type III secretion system (T3SS) (Ashida et al., 2015; Kuehl et al., 2014). Similar to what is seen for L. monocytogenes, the bacteria propel within the host cells due to actin polymerization on one pole of the bacteria, using the membrane protein IcsA (also known as VirG), which also promotes Arp2/3-mediated formation of F-actin, despite having no sequence similarity to ActA of Listeria (Kocks et al., 1995); this allows infection of neighboring epithelial cells via secondary vacuoles (Ashida et al., 2015).

Lysis of the vacuolar membrane has been assigned to the T3SS effector invasion plasmid antigen (Ipa) proteins IpaB, IpaC and IpaD and the invasion plasmid gene D (IpgD) (Mellouk et al., 2014; Picking and Picking, 2016; Ramel et al., 2011). IpaB and IpaC form a complex in the vacuolar membrane that binds cholesterol, resulting in membrane degradation, and this is regulated by IpaD (Picking and Picking, 2016). IpgD is a phosphoinositide phosphatase that converts phosphatidylinositol (4,5)-biphosphate (PIP₂) into phosphatidylinositol 5-phosphate (PI5P), and, in infected HeLa cells, this results in the recruitment of Rab11a, a small GTPase of recycling endosomes, to the bacteria-containing vacuole (Mellouk et al., 2014). The subversion of the intracellular trafficking enables the disruption of the vacuolar membrane leading to bacterial dissemination (Mellouk et al., 2014); this process involves the formation of Rab11a-containing macropinosomes at the bacterial invasion site in HeLa cells (Weiner et al., 2016). The production of PI5P by IpgD at the bacterial entry sites also causes the mislocalization of the epidermal growth factor receptor (EGFR) to early endosomes, which prevents termination of host-cell survival signaling via phosphoinositide 3-kinase and protein kinase B (PI3K-Akt) and enables prolonged bacterial colonization (Ramel et al., 2011). In addition to Rab11a, a small interfering RNA (siRNA) screen, also performed in HeLa cells, revealed other host proteins involved in vacuolar rupture, including factors regulating early endosomal trafficking, such as Rab4a, Rab5 (forms Rab5a–c), early endosome antigen 1 (EEA1), synaptojanin 1 and 2, and the actin cytoskeleton, including the Rho-GTPase Cdc42 and several subunits of the Arp2/3 complex (Mellouk et al., 2014).

In various epithelial cell lines, the disruption of the vacuolar membrane allows the diffusion of the cytosolic lectin galectin-8 into the lumen of the disrupted phagosome, where it binds to glycoproteins and glycolipids and triggers danger-associated molecular pattern (DAMP) signaling (Thurston et al., 2012). In epithelial cell lines and mouse embryonic fibroblasts, remnants of the disrupted phagosome are also ubiquitylated, which triggers the recruitment of the LC3 autophagy machinery already prior to full bacterial escape into the cytosol (Campbell-Valois et al., 2015; Dupont et al., 2009; Thurston et al., 2012). LC3 family protein recruitment likely functions to repair the damaged vacuole and prevent bacterial escape, but Shigella counteracts this with the release of the bacterial virulence proteins IcsB and VirA (Campbell-Valois et al., 2015). IcsB is a cholesterol-binding protein and has been proposed to act as a protease or acyltransferase (Baxt and Goldberg, 2014; Kayath et al., 2010), whereas VirA is a GTPase-activating protein (GAP) for the small-GTPase Rab1 (Dong et al., 2012). The precise roles of IcsB and VirA in preventing LC3 recruitment are still unclear, but their activity is required for bacterial escape from the vacuole in epithelial cells (Campbell-Valois et al., 2015).

The gram-positive bacterium S. aureus is one of the most common foodborne pathogens worldwide (Scallan et al., 2011). Infections by S. aureus lead to abdominal cramps, nausea, vomiting and diarrhea, and have a high morbidity and mortality rate (Ho et al., 2008). For a long time, S. aureus was considered as an extracellular pathogen and staphylococcal foodborne disease was mainly related to heat-resistant enterotoxins produced by this pathogen (Argudín et al., 2010). However, although direct proof of intestinal cell invasion is lacking, it is now clear that S. aureus is in fact a facultative intracellular pathogen in both macrophages and non-phagocytic cells, and intracellular S. aureus bacteria have been detected in various tissues (Horn et al., 2018; Moldovan and Fraunholz, 2019).

S. aureus can survive killing by macrophages and neutrophils (Blättner et al., 2016; Flannagan et al., 2016), even though S. aureus-containing phagosomes readily mature into phagolysosomes in those immune phagocytes (Hagiwara et al., 2014; Schröder et al., 2006). It is unclear whether S. aureus-containing phagolysosomes are fully acidified (Flannagan et al., 2016; Giese et al., 2009; Lâm et al., 2010) or whether acidification is reduced (Jubrail et al., 2016). Regardless, S. aureus has developed defense mechanisms to resist the bactericidal endolysosomal compartment and to replicate inside as a vacuolar pathogen (Tranchemontagne et al., 2016). The acidic phagolysosomal environment is essential for this, as it triggers expression of the quorum-sensing accessory gene regulator (agr) locus which encodes several Agr virulence factors of S. aureus (Tranchemontagne et al., 2016). Other factors also play a role in vacuolar survival, including sigma-B, catalase, superoxide dismutase and peroxiredoxins for resistance against acidic and oxidative stress (Clauditz et al., 2006; Cosgrove et al., 2007; Karavolos et al., 2003; Liu et al., 2005; Richardson et al., 2008), as well as specific inhibitory proteins against vacuolar enzymes, such as serine proteases and myeloperoxidase (de Jong et al., 2017; Stapels et al., 2014). Modification of the bacterial membrane and the peptidoglycan layer by the lipid-modifying enzyme multiple peptide resistance factor (MrpF), and the cell-wall modifying enzymes D-alanine-D-alanyl carrier protein ligase (DltA) and O-acetyltransferase (OatA) contribute to resistance against antimicrobial peptides, such as defensins and lysozyme released from neutrophilic and cytoplasmic granules (Peschel et al., 1999, 2001). Eventually, S. aureus replicates within professional phagocytes and egresses from the cells, persisting in a continuous cycle of phagocytosis, host-cell death and bacterial release (Moldovan and Fraunholz, 2019).

Following infection of various epithelial cell lines and a keratinocyte cell line (Box 1), some strains of S. aureus escape from the vacuole and proliferate in the cytosol (Chi et al., 2014; Giese et al., 2011; Strobel et al., 2016) and these strains are more lethal than those that only reside in the vacuoles (Strobel et al., 2016). In fibroblasts and keratinocyte-like cells, replication of S. aureus was observed in the cytosol after escape from autophagosomes (Neumann et al., 2016). In addition, ubiquitylation of S. aureus was detected as taking place in the cytosol (Neumann et al., 2016). Cytosolic localization of S. aureus has also been detected in macrophage-like cells (Münzenmayer et al., 2016).

The key player in vacuolar escape in both professional and nonprofessional phagocytes are Agr virulence proteins; S. aureus mutants without a functional agr locus cannot escape from vacuoles (Münzenmayer et al., 2016; Neumann et al., 2016; Shompole et al., 2003). The exact mechanism by which Agr virulence proteins induce phagosomal escape is still unknown, but this might involve upregulation of the lytic proteins ɑ-haemolysin (Giese et al., 2009; Jarry et al., 2008; Lâm et al., 2010) and δ-toxin together with the sphingomyelinase β-toxin (Giese et al., 2011). The agr locus also controls expression of phenol-soluble modulins (PSMs) (Münzenmayer et al., 2016), which are small amphipathic peptides with cytolytic activity that are upregulated in methicillin-resistant S. aureus (MRSA) strains (Queck et al., 2008; Wang et al., 2007). The deletion of the ABC transporter Pmt, which exports PSMs from the bacteria, disrupts vacuolar escape of S. aureus in epithelial cell lines (Blättner et al., 2016). Additionally, the nonribosomal peptide synthetase (NRPS) complex AusAB is involved in phagosomal escape, but its mode of action is still unclear (Blättner et al., 2016).

According to the CDC, species of the gram-negative bacteria genus Salmonella cause 1.35 million infections annually in the USA, and especially S. enterica serovar Typhimurium leads to hundreds of deaths. S. Typhimurium invades both macrophages and non-phagocytic cells and was previously considered a vacuolar pathogen that resides exclusively in membrane-enclosed vacuoles within the host cell, called Salmonella-containing vacuoles (SCVs) (Jantsch et al., 2003). Work in macrophages and non-phagocytic cell lines showed that the pathogen establishes these replication vacuoles by rearranging the actin cytoskeleton and redirecting intracellular trafficking to prevent fusion with bactericidal lysosomes (Castanheira and García-del Portillo, 2017).

However, it recently has become clear that a fraction (10–20% in epithelial cells; 2–6% in macrophages) of S. Typhimurium cells traverses into the cytosol (Knodler et al., 2010; Malik-Kale et al., 2012). Despite these fractions being small, they are important for pathogenesis, because the cytosolic population of S. Typhimurium replicates more rapidly than the vacuolar population, possibly due to the higher availability of nutrients, and most Salmonella replication in epithelial cells occurs within the cytosol (Brumell et al., 2002; Knodler et al., 2010; Malik-Kale et al., 2012).

The invasion of Salmonella spp. involves many effector molecules that are injected in the host cell by type III secretion systems 1 and 2 (T3SS1 and T3SS2) (Bayer-Santos et al., 2016; McSorley, 2014). S. Typhimurium residing in the cytosol of various epithelial cell lines primarily expresses T3SS1, whereas the vacuolar population also expresses T3SS2 (Knodler et al., 2010). Although effector molecules injected by T3SS1 are thought to mediate the disruption of the vacuolar membrane for escape of the bacteria (Birmingham and Brumell, 2006; Miki et al., 2004), T3SS2-translocated effectors also play a role in this process, especially Salmonella-induced filament A (SifA) and Salmonella-secreted effector SseJ (Beuzón et al., 2000; Dumont et al., 2010; Lossi et al., 2008). SseJ exhibits phospholipase A and cholesterol transferase activities, which destabilize the vacuolar membrane in both epithelial cells and macrophages (Lossi et al., 2008; Ohlson et al., 2005). At least in HeLa cells, host factors are also important for vacuolar lysis of Salmonella-containing phagosomes, including Rab5 and Rab7a, and coat protein complex II (COP-II) (Santos et al., 2015), but the underlying mechanism is still unclear.

The disruption of the vacuolar membranes can trigger the host-cell defenses via macroautophagy in both macrophages and epithelial cell lines (Birmingham and Brumell, 2006; Thurston et al., 2016), which in HeLa cells and mouse embryonic fibroblasts can lead to membrane repair (Kreibich et al., 2015). Electron microscopy showed that this macroautophagy process requires the ubiquitylation of proteins of the ruptured Salmonella-containing vacuole, similar to what was described for Shigella spp. above (Kishi-Itakura et al., 2020). The survival and replication of the bacteria in the cytosol of infected mouse embryonic fibroblasts depend on the Salmonella invasion protein A (SipA), a multi-function effector protein secreted by both vacuolar and cytosolic Salmonella via T3SS1 (Chong et al., 2019; Klein et al., 2017). SipA prevents macroautophagy by an unknown mechanism (Chong et al., 2019), possibly by diverting organellar trafficking through effects on the cytoskeleton, as SipA also has a role in cell invasion by promoting F-actin formation and by subverting microtubule-dependent motors (Brawn et al., 2007; Chong et al., 2019). Cytosolic bacteria also secrete the T3SS1-effector SopB in infected epithelial and macrophage cell lines (Finn et al., 2017; Klein et al., 2017). SopB delays cell death in epithelial cells containing hyper-proliferating cytosolic Salmonella, presumably to enable longer replication times (Finn et al., 2017). A genetic screen in epithelial cell lines recently identified several other S. Typhimurium proteins required for efficient cytosolic proliferation, such as the transcription modulator YdgT and the DNA recombinase RecA, as well as the Mg²⁺ transport protein CorA (Wrande et al., 2016).

Thus, S. Typhimurium seems to have evolved two parallel strategies to survive and proliferate in host cells – not only can it survive in vacuoles but it can also escape to the cytosol using a variety of mechanisms. This feature is shared in common with most other foodborne pathogenic bacteria described in this Review (L. monocytogenes, Campylobacter spp. and S. aureus). Only Shigella spp. has been exclusively described as a cytosolic pathogen, but, since all other pathogens have now been found in both vacuoles and the cytosol, our prediction is this species might also be able to persist in vacuoles.

For all bacteria described in this Review, vacuolar escape is observed in both infected macrophages and non-professional phagocytes (Cossart and Helenius, 2014; Haas, 2007) (Box 1). To escape from vacuoles, all these foodborne intracellular pathogens use a combination of three different effector proteins to escape from the vacuole (Figs 1 and 2). The first are hemolytic proteins or peptides that form pores and/or mechanically rupture membranes, such as LLO of L. monocytogenes (Bierne et al., 2018) and PSMs of S. aureus (Münzenmayer et al., 2016). The second category of proteins modifies phospholipids and thereby presumably weakens the integrity of the vacuolar membrane, such as PC-PLC of L. monocytogenes (Alberti-Segui et al., 2007). The third category of proteins promotes vacuolar escape by modifying host-cell mechanisms, such as PI-PLC of L. monocytogenes (Poussin et al., 2009) and IpgD of Shigella spp. (Mellouk et al., 2014). These three strategies are not exclusive features of foodborne pathogenic bacteria, as many non-foodborne intracellular pathogens use similar mechanisms to escape to the cytosol (Box 2). Thus, most, if not all, intracellular pathogens seem to have evolved three different mechanisms to actively disrupt the vacuolar membrane. Nevertheless, especially in macrophages and other immune phagocytes, evidence suggests that disruption of vacuolar membranes might actually be a common outcome of phagosomal maturation processes, and intracellular pathogens need to counteract this to prevent their vacuolar escape.

Several vacuolar pathogens might not be directly responsible for the rupture of the vacuolar membrane, but rather be involved in the maintenance of its integrity (Creasey and Isberg, 2014). For Legionella pneumophila, another canonical vacuolar pathogen, this maintenance requires effector proteins injected by a type 4 secretion system (T4SS) (Creasey and Isberg, 2012), because Legionella mutants with a dysfunctional T4SS are unable to maintain the integrity of their vacuoles in both infected macrophage-like cells and epithelial cells (Molmeret et al., 2007). One candidate for this maintenance of vacuolar membrane integrity is the T4SS-secreted protective factor SdhA, which counteracts the activity of the T2SS-secreted phospholipase A2 (Pla2) in macrophages and macrophage-like cells (Creasey and Isberg, 2012). Pla2 destabilizes the vacuolar membrane to allow L. pneumophila escape upon host-cell exit (Hiller et al., 2018); however, since it recently has become clear that a fraction of L. pneumophila escapes into the cytosol in human macrophage cell lines (Truchan et al., 2017), Pla2 might also mediate vacuolar escape for the cytosolic population. Evidence suggests that S. Typhimurium also actively maintains the host-cell vacuolar membrane to prevent its escape into the cytosol; in infected HeLa cells, the Salmonella T3SS2 effector protein SifA counteracts the membrane-destabilizing activity of SseJ and thereby stabilizes the vacuolar membrane, possibly by binding to SifA, the kinesin-interacting protein (SKIP) and the microtubule motor kinesin-1 (Boucrot et al., 2005; Dumont et al., 2010). This stabilization of Salmonella-containing vacuoles might be essential to prevent bacterial escape, as the switch of Rab5 to Rab7 can result in disruption of the vacuolar membrane and escape of S. Typhimurium into the cytosol (Brumell et al., 2002).

A potential mechanism for the nonspecific rupture of the vacuolar membrane is mediated by reactive oxygen species (ROS) produced by phagosomal NADPH oxidases and other cellular sources (Fig. 3) (Davis et al., 2012). ROS are highly reactive and indiscriminately oxidize lipids and membrane proteins (Girotti, 1998), which leads to the disruption of the integrity of biological membranes (Cwiklik and Jungwirth, 2010; Wong-Ekkabut et al., 2007). Phagosomal membrane components could be particularly vulnerable to oxidation, as NOX2 (a multi-protein complex composed of CYBA, CYBB, NCF1, NCF2 and NCF4) locates to their membranes (Buvelot et al., 2019). In macrophages, the rupture of phagosomal membranes is commonly observed when these cells ingest micrometer-sized inorganic particles (Davis et al., 2012), indicating that this can occur independently of pathogen-driven processes and that host-cell responses suffice to disrupt phagosomal membranes. Similarly in apoptotic cells, vacuolar cathepsins have been shown to be released into the cytosol due to vacuolar rupture (Johansson et al., 2010; Terman et al., 2006). This rupture might well be the result of oxidation of the phagosomal membrane, because the production of ROS by NOX2 can result in lipid peroxidation and permeabilization of the phagosomal membrane in phagocytic cells (Dingjan et al., 2016, 2017). This loss of integrity of phagosomal membranes can be prevented by the lipophilic radical scavenger vitamin E and is not observed in patients suffering from chronic granulomatous disease (CGD), a genetic disease resulting in dysfunctional NOX2 (Dingjan et al., 2016, 2017), further supporting the idea that NOX2-produced ROS can disrupt the integrity of membranes. These findings show that the loss of membrane integrity by ROS-induced oxidation is a common process that might contribute to the vacuolar escape of intracellular pathogens.

Indeed, several studies provide direct evidence that ROS play a role in the escape of intracellular pathogens into the cytosol. First, in macrophages derived from mice lacking NOX2, the escape of L. monocytogenes from vacuoles is inhibited (Davis et al., 2012). Second, ROS might play a role in the rupture of the vacuolar membrane by the basidiomycetous fungus Cryptococcus neoformans (Tucker and Casadevall, 2002). C. neoformans is mainly a vacuolar intracellular pathogen, but it was also recently observed to reach the cytosol in macrophages (Coelho et al., 2014). The mechanism of vacuolar rupture is unclear, but in murine alveolar macrophages, lipid peroxidation was observed upon C. neoformans infection and this could be attributed to ROS production by the macrophage and not the pathogen (Gross et al., 2000). Moreover, in order to survive in the hostile intracellular environment of the macrophage, C.neoformans within infected macrophage-like cells upregulates expression of genes involved in responses to oxidative stress (Fan et al., 2005), including SOD2 (Giles et al., 2005). Finally, as previously mentioned, S. aureus also expresses ROS-neutralizing enzymes in order for it to persist in the vacuole.

Thus, ROS production by infected host cells, particularly immune phagocytes, might well contribute to, and in some cases even suffice for, the rupturing of the vacuolar membrane and escape of intracellular pathogens into the cytosol. Intracellular pathogens seem to have adapted to tolerate the high levels of ROS, which actually facilitate the vacuolar escape of the pathogen. Thereby, intracellular pathogens might not only have evolved a variety of mechanisms to disrupt the vacuolar membrane themselves, but also to hijack host defense responses for propagation in the cytosol.

It has become apparent that the classical distinction between vacuolar and cytosolic intracellular pathogens does not hold. Thus, the paradigm is changing, and evidence shows that many vacuolar pathogens in fact reach the cytosol; many canonical vacuolar pathogens, such as Mycobacterium spp. (Jamwal et al., 2016), Salmonella spp. (Birmingham and Brumell, 2006; Knodler et al., 2014; Thurston et al., 2016), Campylobacter spp. (Nemelka et al., 2009) and S. aureus (Giese et al., 2011; Strobel et al., 2016) also reside in the cytosol. Accessing the nutrient-rich cytosol therefore appears to be a general characteristic for intracellular pathogens; our prediction is that other foodborne vacuolar pathogens not addressed in this Review, such as Y. enterocolitica (Connor et al., 2018), Brucella spp. (Celli, 2015) and C. sakazakii (de Mey et al., 1995; Mohan Nair and Venkitanarayanan, 2007), might also gain access to the cytosol at some point during their life cycle. The vacuolar and cytosolic populations might each have unique contributions to the disease pathology, as described for S. Typhimurium (Brumell et al., 2002; Knodler et al., 2010; Malik-Kale et al., 2012). It is therefore important to delineate the impact of the different stages in term of immune responses, namely how they contribute to bacterial hyper-replication in acute infections or to bacterial dormancy for long-term colonization in chronic infections, and resistance to antibiotic treatment.

A key open question is which conditions drive pathogens to become cytosolic or remain vacuolar and how this depends on both microbial and host-cell factors. One prediction that follows from this Review is that the vacuolar escape might be more efficient in dendritic cells than macrophages; both mouse and human dendritic cells have higher activity of NOX2 than macrophages (Mantegazza et al., 2008; Savina et al., 2006), raising the possibility that vacuolar escape by NOX2-mediated oxidation of vacuolar membranes might be more efficient in dendritic cells. However, the process and contribution of peroxidation to vacuolar escape are incompletely understood, and more research is needed to determine if and to what extent this contributes to the vacuolar escape of each pathogen.

Another limitation is that most studies have been performed in vitro with model cell lines, often cancerous and derived from organs other than the gut. As enterocytes and intestinal macrophages are often the main entry point for foodborne pathogens (Case and Samuel, 2016; Castanheira and García-del Portillo, 2017; Cossart and Helenius, 2014; Flieger et al., 2018; Mitchell et al., 2016; Pucciarelli and García-del Portillo, 2017), it is important to validate findings with relevant cell lines. For instance, although LLO is important for bacterial escape from the primary vacuoles (Nguyen et al., 2019), it is not always strictly essential, since vacuolar escape has been observed in various epithelial cell lines for L. monocytogenes mutants lacking LLO production (Marquis et al., 1995; Mueller and Freitag, 2005), although the physiological relevance of this finding is unclear. Moreover, although cytosolic populations have been observed in vitro for many pathogens that were previously believed to be exclusively vacuolar, this has to be confirmed in vivo and it has to be established in which tissue this phenomenon occurs.

The escape of pathogens from vacuoles of infected host cells can have therapeutic relevance. It is clear that although pathogens have evolved highly complex and unique vacuolar escape mechanisms, the main processes are similar across species – most (but not all) intracellular pathogens rely on a combination of lytic pore-forming proteins and enzymes that modify the vacuolar membranes, such as phospholipases. In addition, ROS production by immune phagocytes might also contribute to their escape, as argued in this Review. Understanding the fundamental features of how these mechanisms enable the escape of pathogens from vacuoles, as well as the differences and similarities between different pathogenic species, aids in development of new therapeutic approaches for the treatment of both foodborne and other intracellular pathogens.

The authors thank the members from the van den Bogaart lab for helpful discussions.