Escape from the Primary Vacuole 921 Macrophages

The fate of L. monocytogenes in a macrophage depends upon both the cell lineage and its history. Primary peritoneal macrophages are capable of killing approximately 90% of entering L. monocytogenes (Portnoy et al. 1989), bone marrow-derived macrophages kill approximately half the bacteria, whereas cell lines such as J774 and RAW 264.7 murine macrophage-derived cells are weakly cidal (Camilli et al. 1993; De Chastellier 1994). Treatment of peritoneal macrophages with y interferon (IFN-y) prior to infection resulted in complete suppression of bacterial growth (Portnoy et al. 1989), consistent with the ability of activated macrophages to kill L. monocytogenes (Cossart and Portnoy 2000). The generation of reactive oxygen and nitrogen intermediates within the vacuole contributes to the retention of L. monocytogenes within phagosomes in activated macrophages (Myers et al. 2003).

9.2.1.1. Requirements 9.2.1.1.1. Listeriolysin O

Two of the gene products of the PrfA regulon, listeriolysin O (LLO), the product of hly, and a phosphatidylinositol-specific phospholipase C (PI-PLC), the product of plcA, play key roles in escape from the primary vacuole of a macrophage. LLO is a member of a large family of cholesterol-dependent pore-forming toxins or cytolysins (CDC) which includes streptolysin O (SLO) secreted by Streptococcus pyogenes, perfringolysin O (PFO) secreted by Clostridium perfringens, anthrolysin O (ALO) secreted by Bacillus anthracis, and pneumolysin, which is produced by Streptococcus pneumoniae. These multidomain proteins insert into cholesterol-containing eukaryotic cell membranes forming large pores of 25-30 nm in diameter by oligomerization of approximately 50 monomers (Alouf 1999; Giddings et al. 2004). LLO deletion mutants are avirulent and fail to escape from the primary vacuole of a macrophage (Gaillard et al. 1987; Kuhn et al. 1988; Portnoy et al. 1988). Expression of hly in Bacillus subtilis and Salmonella conferred the ability to escape from the phagosome of J774 cells to this species; however, the expression of LLO did not convert B. subtilis into a pathogen (Bielecki et al. 1990; Gentschev et al. 1995). Although the three-dimensional structure of LLO is not known, that of PFO has been established (Rossjohn et al. 1997) (Figure 9.1.). Molecular simulation based on the structure of PFO suggests that the structure of LLO is similar, consisting of four domains. Domain four anchors the oligomers to the membrane (Ramachandran et al. 2002) whereas domain three regulates polymerization (Ramachandran et al. 2004) and pore formation (Shatursky et al. 1999).

LLO, 60 kDa, is uniquely structured to function in the acidic environment of the vacuole by virtue of its low pH optimum (pH 5.5). Leucine 461 was found to be critical for the acidic pH optimum of LLO. When leucine 461 was changed

Cereus Toxin

Figure 9.1. (Left) Structure of the perfringolysin O molecule. Taken from Rossjohn et al. (1997), with permission from Elsevier. The comparative structures of B. cereus PI-PLC (center) and L. monocytogenes PI-PLC (right) with the active site pocket containing myo-inositol in the center. The hydrophobic ridge is designated by A and the extended GPI-binding site by B. Taken from (Moser et al. 1997), with permission of the publisher. Reprinted from Journal of Molecular Biology, Vol 273:1, J. Moser et al, Crystal Structure of the Phosphatidylinositol, pages 269-282, Copyright 1997, with permission from Elsevier.

Figure 9.1. (Left) Structure of the perfringolysin O molecule. Taken from Rossjohn et al. (1997), with permission from Elsevier. The comparative structures of B. cereus PI-PLC (center) and L. monocytogenes PI-PLC (right) with the active site pocket containing myo-inositol in the center. The hydrophobic ridge is designated by A and the extended GPI-binding site by B. Taken from (Moser et al. 1997), with permission of the publisher. Reprinted from Journal of Molecular Biology, Vol 273:1, J. Moser et al, Crystal Structure of the Phosphatidylinositol, pages 269-282, Copyright 1997, with permission from Elsevier.

to threonine, LLO became about 10-fold more active at neutral pH (Glomski et al. 2003). This increase in activity results from an increase in protein stability at neutral pH (Schuerch et al. 2005). Substitution of either PFO or ALO for LLO in L. monocytogenes permits escape of bacteria from the primary phagosome of a macrophage (Jones and Portnoy 1994; Wei et al. 2005a). Both of these proteins are much more active than LLO at neutral pH, consequently their expression leads to membrane permeabilization upon growth of L. monocytogenes in the cytosol (Jones and Portnoy 1994; Wei et al. 2005a). Like PFO or ALO, which have threonine at the equivalent position, LLO L461T was able to mediate escape of bacteria from the primary vacuole, but it became cytotoxic to the host cell by permeabilizing the cell membrane (Glomski et al. 2002; Glomski et al. 2003).

LLO synthesis is upregulated in the phagosome (Bubert et al. 1999; Freitag and Jacobs 1999; Gray et al. 2006; Klarsfeld et al. 1994; Moors et al. 1999). Recently evidence has emerged indicating that the nucleotide sequence of the hly transcript is involved in translational repression of LLO in the host cytosol. This regulation is dictated by sequences within the coding region of hly mRNA (Schnupf et al. 2006). Thus, the potential cytotoxicity of LLO in the cytosol is modulated both by its lower activity at neutral pH and by repression of its synthesis.

The CDCs are characterized as oxygen-labile or thiol-activated. A characteristic motif in the carboxyl terminal of domain 4 is a conserved undecapeptide, ECTGLAWEWWR. This sequence contains the only cysteine residue present in most of the toxins in this family (Alouf 1999). Although reduction of this cysteine by sulfhydryl reagents is necessary for the activity of CDCs, replacement of Cys-484 by Ala in LLO showed that a thiol group is not essential for hemolytic activity (Michel et al. 1990).

LLO and other members of the CDC family interact with receptors on mammalian cell surfaces. Intermedilysin, a CDC toxin secreted by Streptococcus intermedius, interacts with human CD59 on erythrocytes (Giddings et al. 2004). ALO, a recently discovered member of the CDC family secreted by Bacillus anthracis (Shannon et al. 2003), interacts with the macrophage Toll-like receptor 4 (TLR4), known as a specific receptor for lipopolysaccharide of gram-negative bacteria. The interaction of ALO with TLR-4 resulted in typical signaling through p38 MAPK. In addition to ALO, activation of macrophages through TLR-4 was also observed with LLO, SLO, PFO, and pneumolysin (Park et al. 2004). These recent findings are consistent with the known ability of noncytolytic cholesterol complexes of LLO to induce cytokine expression (Nishibori et al. 1996), lipid second messengers (Sibelius et al. 1996), and an IL-1 response (Yoshikawa et al. 1993), and suggest that LLO is multifunctional by virtue of its ability to form pores and its ability to interact with surface receptors on eukaryotic cells. Binding of LLO to membrane cholesterol has been attributed to domain 4, whereas cytokine induction was attributed to domains 1-3 ( Jacobs et al. 1999; Kohda et al. 2002).

9.2.1.1.2. Phosphatidylinositol-Specific Phospholipase C

Phosphatidylinositol-specific phospholipase C of L. monocytogenes (LmPI-PLC), encoded by plcA (Camilli et al. 1991; Leimeister-Wachter et al. 1991; Mengaud et al. 1991a), is highly specific for phosphatidylinositol (PI) with relatively weak activity on glycosyl-PI-(GPI)-anchored proteins (Gandhi et al. 1993; Goldfine and Knob 1992). Mutants with deletions of plcA escape from the primary vacuole of macrophages less efficiently than the wild type. At 90 min after infection between 30 and 65%, fewer of these mutants have escaped from the phagosome compared to wild type (Bannam and Goldfine, 1999; Camilli et al. 1993; Smith et al. 1995; Wadsworth and Goldfine 1999).

The weak activity of LmPI-PLC differentiates it from the classical bacterial PI-PLCs from Bacillus species and Staphylococcus aureus, which have strong activity on GPI-anchored proteins (Low 1989; Wei et al. 2005b). Although LmPI-PLC shares only 24% identity with PI-PLC from B. cereus (BcPI-PLC), the overall three-dimensional structures are highly homologous (Moser et al. 1997). Both consist of a single (Pa)8-barrel domain. The active site pocket is highly conserved with only two differences in amino acids involved in inositol binding, but complete conservation of the residues thought to be involved in catalysis. An important structural difference is the absence in LmPI-PLC of the Vb ^-strand of BcPI-PLC, which supports the edge of a shallow groove extending from the active site pocket and is predicted to promote interactions with the glycan of GPI-anchored proteins (Figure 9.1.) (Moser et al. 1997). Removal of the Vb ^-strand of BcPI-PLC resulted in somewhat decreased activity on PI and essentially complete loss of activity on the GPI-anchored protein Thy-1 on mouse splenocytes (Wei et al. 2005b). Like other bacterial PI-PLCs, LmPI-PLC does not cleave polyphosphoinositides such as PI 4,5-bisphosphate (PI-4,5-P2), an important substrate of eukaryotic phospholipases involved in intracellular signaling (Goldfine and Knob 1992). Although both types of PI-PLC produce the second messenger diacylglycerol (DAG), the action of LmPI-PLC on PI produces inositol-1-P and not inositol 1,4,5-trisphosphate (IP3), another second messenger that releases Ca2+ from the endoplasmic reticulum. Another significant difference between LmPI-PLC and those from eukaryotic cells is the absence of a divalent cation requirement. Instead, LmPI-PLC shows a strong dependence on high ionic strength salts, e.g., 100-200 mM NaCl, KCl, NH4Cl, or (NH4)2SO4, which is needed for disaggregation of multimers (Goldfine and Knob 1992).

9.2.1.2. Permeabilization of the Phagosome

Upon internalization of L. monocytogenes, the J774 phagosome is rapidly acidified to pH 4.8-6.5. During this time the synthesis of LLO is greatly increased (Chap. 7). Soon after internalization the phagosomal membrane is permeabilized to the dye HPTS (8-hydroxypyrene-1,3,6-trisuolfonic acid), and the pH of the vacuole rapidly increases. Agents that inhibit acidification, such as bafilomycin, inhibit perforation and escape (Conte et al. 1996; Beauregard et al. 1997; Glomski et al. 2002). Permeabilization of the vacuole is absolutely dependent on the expression of LLO (Beauregard et al. 1997). These findings are consistent with a model in which LLO forms pores in the phagosome and permits a two-way exchange of small molecules. It appears, however, that host factors play a role in this process. A mutant lacking PI-PLC produced about 65% fewer permeabilized vacuoles than the wild type between 30 and 60 min after infection of J774 cells (Poussin and Goldfine 2005). This finding could be consistent with a model in which PI-PLC, by hydrolyzing a minor component of the vacuolar membrane, assists in the degradation of the membrane (Villar et al. 2000). However, inhibition of host PKC P isoforms by treatment of J774 cells with the inhibitors RO-31-8425 and Go-6983, produced 29 and 62% fewer permeabilized vacuoles, respectively, than in untreated cells during the same time frame. Inhibition of host calcium signaling by treatment with thapsigargin or SK&F 96365 produced even greater inhibition of vacuolar permeabilization by wild type L. monocytogenes (Poussin and Goldfine 2005). As will be discussed below, activation of PKC P isoforms is dependent on expression of both LLO and PI-PLC and on the elevation of host intracellular calcium. Inhibition of both calcium signaling and PKC P activation also inhibits escape from the primary vacuole of a macrophage.

Perforation of the Lm phagosome in macrophage-like cells which permits the escape of small molecules like HPTS (524 MW) (Beauregard et al. 1997; Poussin and Goldfine 2005) and Lucifer Yellow (522 MW) into the cytosol is followed 5-9 min later by larger pores that permit the escape of molecules like fluorescent dextrans (average MW 10,000). This exchange of small molecules, protons, and calcium ions is postulated to inhibit vacuole fusion with lysosomes (Shaughnessy et al. 2006).

9.2.1.3. Mechanism of Escape

It is well known that LLO is indispensable for escape from the primary phagosome of a macrophage. From this fact a predominant hypothesis has emerged which states that the perforation of the phagosomal membrane by LLO leads to escape. It is possible that the physical act of perforation is the mechanical means of disruption of the membrane. Further considerations argue against this hypothesis. The phagosomal membrane must be larger than the bacterium it contains; therefore, its surface area should be approximately 1-4 ^m2. A pore formed by LLO is approximately 25 nm in diameter or about 500 nm2. To cover the surface of the phagosome completely would require approximately 2000 pores. Yet, once the phagosome is permeabilized by LLO, there is a rapid increase in pH (Beauregard et al. 1997) which presumably results in greatly reduced LLO activity. Unless there is a concerted action of LLO resulting in the simultaneous formation of multiple pores, these calculations strongly argue against mechanical disruption of the phagosome by LLO.

LLO along with PI-PLC leads to the rapid activation of host functions in macrophages including opening of calcium channels and release of Ca2+ from stores, activation of host polyphosphoinositide-specific PLC, phospholipase D (PLD), and PKC isoforms. Inhibition of calcium elevation or activation of PLD or PKC P leads to strong, but not complete inhibition of escape from the phagosome (Wadsworth and Goldfine 1999; Goldfine et al. 2000; Wadsworth and Goldfine 2002; Poussin and Goldfine 2005). PKC P I and II are found on early endosomes which fuse to form large vesicles, within minutes of infection of J774 macrophage-like cells with wild type L. monocytogenes. LLO and PI-PLC expression are needed for mobilization of PKC PII, and LLO expression is required for mobilization of PKC P I (Wadsworth and Goldfine 2002). The activation of host polyphosphoinositide-specific PLC and phospholipase D (PLD) also requires expression of LLO, but not PI-PLC. These findings suggest a model in which LLO activates host functions that are needed for disruption of the phagosome. At this time, there is no available information on how PKC P influences phagosomal perforation and disruption. Hannun and colleagues have recently shown that classical PKC isoforms a and P II appear on a juxtanuclear subset of recycling endosomes, called the pericen-trion, after treatment of a variety of cell types with PMA, a known activator of classical PKC isoforms (Becker and Hannun 2003; Becker and Hannun 2004). Compared to the mobilization of PKC P II upon infection of J774 cells with L. monocytogenes, which takes place within the first minute of infection, juxtanuclear translocation after treatment with PMA requires 30-60 min treatment. PKC translocation may regulate the clustering of recycling endosomes in the perinuclear region (Idkowiak-Baldys et al. 2006). These findings suggest that L. monocytogenes subverts another normal process in cells which results in a specific outcome, i.e., delay of maturation of the phagosome and subsequent escape.

It appears that L. monocytogenes controls vesicular trafficking in the host by controlling the activity of Rab5, a small GTPase involved in the regulation of phagosome-endosome fusion and phagosomal maturation (Desjardins 1995; Alvarez-Dominguez et al. 1996). The ability of LLO-negative bacteria to survive requires inhibition of phagosome maturation by a mechanism involving Rab5a (Alvarez-Dominguez et al. 1997; Alvarez-Dominguez and Stahl 1999). In these two studies, the authors used a LLO-minus strain. Treatment of J774 cells with IFN-7 increases the association of lysosomal markers such as cathepsin-D, lysosome-associated membrane protein-1 (LAMP1), and Limp-II with phagosomes and this also appears to be controlled by Rab5a, which is increased on phagosomes from cells treated with IFN-7 (Prada-Delgado et al. 2001). Evidence has been presented implicating inhibition of Rab5a exchange activity by L. monocytogenes as the means for controlling phagosome-lysosome trafficking (Prada-Delgado et al. 2005). On the other hand, Henry et al. observed that L. monocytogenes escapes from Rab5a-negative, LAMP1-negative, Rab7-positive, and PI-3-P-positive vacuoles in a manner that is LLO-dependent. When hly, the gene coding for LLO, was expressed under control of an inducible promoter and its expression in the vacuole was delayed until the bacteria were in LAMP1-positive vacuoles, escape was less efficient than when hly was expressed normally (Henry et al. 2006).

The exclusion of Rab5a from L. monocytogenes vacuoles may be related to the absence of the GTP-exchange mechanism needed for bringing Rab5a to the vacuolar membrane (Prada-Delgado et al. 2005). Overexpression of Rab5Q79L, which is locked in the GTP-bound state, resulted in association of Rab5a with the L. monocytogenes vacuole, but did not affect L. monocytogenes escape from vacuoles (Henry et al. 2006). Although Rab5a is important for vesicular trafficking, its role in macrophage killing of L. monocytogenes is at present unclear.

9.2.2. Nonphagocytic Cells 9.2.2.1. Requirements 9.2.2.1.1. LLO and PI-PLC

The two bacterial factors contributing to escape from primary vacuoles of macrophages, LLO and PI-PLC, also contribute to escape from primary vacuoles of non-phagocytic cells (Marquis et al. 1995; Dancz et al. 2002), with the exception that LLO is not essential in human epithelial and fibroblast cells, as well as in Potoroo tridactylis kidney (Ptk2) cells (Portnoy et al. 1988; Marquis et al. 1995; Dancz et al. 2002; Mueller and Freitag 2005). This phenomenon was also observed in human dendritic cells (Paschen et al. 2000). In Henle 407 cells, LLO-negative bacteria show a twofold reduction in escape from primary vacuoles, whereas a double LLO-PI-PLC mutant shows a fourfold defect. In absence of LLO and PI-PLC, the broad-range phospholipase C (PC-PLC) and a metalloprotease (Mpl) of L. monocytogenes mediate escape from vacuoles (Marquis et al. 1995; Grundling et al. 2003).

L. monocytogenes secretes a PLC that has the ability to hydrolyze a large variety of phospholipids including phosphatidylcholine (PC), phosphatidylethanolamine, phosphatidylserine, phosphatidylinositol, and sphingomyelin (Sm) (Geoffroy et al. 1991; Goldfine et al. 1993). The activity of this PLC is zinc-dependent and optimum at pH 5.5-8. The enzyme was initially named PC-PLC because of its similarity to other bacterial PLCs for which PC is the preferred substrate. However, the L. monocytogenes enzyme is often referred to as BR-PLC because of its activity on a broad range of substrates. In addition, it is occasionally named PlcB in reference to the gene coding for it, plcB.

The three-dimensional structure of the L. monocytogenes PC-PLC (LmPC-PLC) is not known, but that of the Bacillus cereus (BcPC-PLC) and of the Clostridium perfringens (a-toxin) orthologs have been established. The compact catalytic domain of these PLCs is composed uniquely of a-helices, and the nine zinc-coordinating amino acid residues are conserved among them (Hough et al. 1989; Naylor et al. 1998; Zuckert et al. 1998). Exceptionally, BcPC-PLC and LmPC-PLC have an extra N-terminal propeptide-regulating enzyme activity (Johansen et al. 1988; Vazquez-Boland et al. 1992), whereas the C. perfringens a-toxin contains an additional C-terminal domain implicated in calcium-dependent membrane binding (Guillouard et al. 1997). This extra C-terminal domain is essential for a-toxin sphingomyelinase and hemolytic activities, conferring toxicity. BcPC-PLC and LmPC-PLC are not considered toxins.

The LmPC-PLC is made as a preproenzyme of 289 amino acid residues. It contains a canonical Sec-dependent signal sequence of 27 residues and a propeptide of 24 residues (Figure 9.2.) (Vazquez-Boland et al. 1992). The N-terminus of the active enzyme was sequenced and the first three residues (WSA) are identical to that of BcPC-PLC (Niebuhr et al. 1993). The secreted proenzyme has no enzymatic activity (Marquis et al. 1997). Interestingly, an LmPC-PLC mutant with a complete deletion of the propeptide is secreted as an

|~j Signal sequence jj Prodomain | Catalytic domain

Figure 9.2. Schematic representation of LmPC-PLC and Mpl domain organization. Both proteins are comprised of a signal sequence, a prodomain, and a catalytic domain. LmPC-PLC is 289 aa long, with a signal sequence of 27 aa, a prodomain of 24 aa, and a catalytic domain of 238 aa. Mpl is 510 aa long, and the size of the respective domains are predicted to be 24 aa for the signal sequence, 180 aa for the prodomain, and 306 aa for the catalytic domain.

|~j Signal sequence jj Prodomain | Catalytic domain

Figure 9.2. Schematic representation of LmPC-PLC and Mpl domain organization. Both proteins are comprised of a signal sequence, a prodomain, and a catalytic domain. LmPC-PLC is 289 aa long, with a signal sequence of 27 aa, a prodomain of 24 aa, and a catalytic domain of 238 aa. Mpl is 510 aa long, and the size of the respective domains are predicted to be 24 aa for the signal sequence, 180 aa for the prodomain, and 306 aa for the catalytic domain.

active enzyme suggesting that the propeptide does not contribute to folding of the catalytic domain (Yeung et al. 2005). Presumably, the function of the propeptide is to prevent activity by interfering with substrate binding in the active site.

The Mpl, a secreted zinc-dependent metalloprotease of L. monocytogenes, is involved in the proteolytic activation of LmPC-PLC (Poyart et al. 1993; Yeung et al. 2005). Mpl is made as a preproenzyme of 510 aa residues. It is predicted to contain a Sec-dependent signal sequence of 24 residues, a prodomain of 180 residues, and a catalytic domain of 306 residues (Figure 9.2.) (Mengaud et al. 1991b). The large prodomain is typical of metalloproteases, and is thought to function as a protease inhibitor and as an intramolecular chaperone facilitating folding of the catalytic domain (Braun and Tommassen 1998). The three-dimensional structure of thermolysin, an ortholog of Mpl and the prototype for this class of bacterial metalloproteases, indicates that the catalytic domain is comprised of two spherical subdomains, which together form a deep cleft containing the active site (Holmes and Matthews 1982). In addition to the active-site zinc ion, thermolysin contains four calcium-binding sites, which confer high thermal stability. Similarly to thermolysin, Mpl exhibits high thermostability. It is active at pH 5-9, but its activity is optimum at pH 7 (Coffey et al. 2000).

The genes coding for Mpl and PC-PLC, mpl and plcB, localize to the PrfA regulon (Portnoy et al. 1992). The mpl gene is immediately upstream of the actA and plcB genes, and promoters from the mpl and actA genes contribute to transcription of the promoterless plcB gene (Vazquez-Boland et al. 1992). However, intracellular expression of plcB is primarily under the control of the actA promoter (Shetron-Rama et al. 2002) (see Chap. 7).

Listeria ivanovii, a ruminant pathogen, expresses an additional sphin-gomyelinase encoded by smcL, which is flanked by genes encoding members of the internalin family. The sphingomyelinase appears to improve by twofold to -fourfold bacterial escape from vacuoles and intracellular growth in bovine epithelial cells. It is possible that this additional sphingomyelinase activity is important in ruminants because of the high level of sphingomyelin in their cell membranes (Gonzalez-Zorn et al. 1999).

9.2.2.2. Mechanism

The LLO-independent escape from vacuoles of human epithelial cells and Ptk2 cells requires high-level expression of PC-PLC (Grundling et al. 2003; Mueller and Freitag 2005). Expression from the actA promoter, which is responsible for the intracellular transcription of plcB, is very low in broth culture (Grundling et al. 2003; Moors et al. 1999) and in primary vacuoles of macrophages (Freitag and Jacobs 1999), but increases by «200-fold later during infection. Based on these observations, it is difficult to conceive that PC-PLC would be able to mediate LLO-independent escape from primary vacuoles. Perhaps the vacuolar makeup and/or environment of human epithelial cells are different than that of macrophages. To address this question, Cheng et al. (Cheng et al. 2005) used RNA interference to identify host knockdowns that bypass the need for LLO in vacuolar escape of Drosophila S2 cells. Knockdowns in components controlling trafficking to and from multivesicular bodies/late endosomes were identified as being permissible for escape of LLO-negative bacteria from primary vacuoles of Drosophila S2 cells. Concomitantly, it was shown that the efficacy of bacteria to escape primary vacuoles in mouse macrophages correlates with the inhibition of vacuolar maturation to LAMPl-positive compartments, and that in the absence of LLO, bacteria-containing vacuoles mature slightly faster than conventional phagosomes (Henry et al. 2006). Perhaps, factors other than LLO influence the kinetics of vesicular trafficking in human epithelial cells making time for increased expression of PC-PLC prior to reaching a stage that is no longer permissible for vacuolar escape of L. monocytogenes.

Was this article helpful?

0 0

Responses

  • Henri
    How does listeria escape from vacuole?
    8 years ago
  • RALPH
    How does listeria escape the vaculoe in epithelial cells?
    8 years ago

Post a comment