Correspondence: Douglas A. Drevets, VAMC 111/C, 921 NE 13th Street, Oklahoma City, OK 73104, USA. Tel.: +405 270 0501, ext. 3284; fax: +405 297 5934; e-mail: email@example.com
Listeria monocytogenes is a facultative intracellular bacterium that has predilection for causing central nervous systemic infections in humans and domesticated animals. This pathogen can be found worldwide in the food supply and most L. monocytogenes infections are acquired through ingestion of contaminated food. The main clinical syndromes caused by L. monocytogenes include febrile gastroenteritis, perinatal infection, and systemic infections marked by central nervous system infections with or without bacteremia. Experimental infection of mice has been used for over 50 years as a model system to study the pathogenesis of this organism including the mechanisms by which it invades the brain. Data from this model indicate that a specific subset of monocytes, distinguished in part by high expression of the Ly-6C antigen, become parasitized in the bone marrow and have a key role in transporting intracellular bacteria across the blood-brain barriers and into the central nervous system. This Minireview will summarize recent epidemiologic and clinical information regarding L. monocytogenes as a human pathogen and will discuss current in vitro and in vivo data relevant to the role of parasitized monocytes and the pathogenetic mechanisms that underlie its formidable ability to invade the central nervous system.
Listeria monocytogenes is a Gram-positive, facultative intracellular bacterium that causes invasive diseases in humans and animals, especially central nervous system (CNS) infections (Vazquez-Boland et al., 2001). Although it is not the most common cause of CNS infection, L. monocytogenes is surprisingly good at doing so. In fact, an epidemiologic study of bacterial meningitis in the United States (US) in 1995 found that this bacterium is nearly 10-fold more efficient than other neuroinvasive Gram-positive bacteria including Streptococcus pneumoniae and group B streptococci at invading the CNS once an invasive infection has been established (Schuchat et al., 1997). Listeria monocytogenes has been extensively studied as a model pathogen largely from the perspective of innate and adaptive immune responses to intracellular infection and bacterial entry and survival within mammalian cells (Cossart, 2007). There is also a growing body of data concerning CNS infections caused by L. monocytogenes including studies on mechanisms of how the bacteria enter the CNS as well as bacterial and host responses when the bacteria enter the brain. This minireview begins with a brief overview of epidemiological data and the clinical manifestations of listeriosis then concentrates more specifically on the mechanisms by which L. monocytogenes enters the CNS, especially the pathogenetic role of infected monocytes.
Listeria monocytogenes is an uncommon human pathogen. It is mostly identified through cases of invasive disease, either associated with outbreaks or as sporadic infections, and through food safety programs. Nonperinatal cases of invasive listeriosis have an estimated worldwide incidence ranging from 0.1–1.1 cases per 105 population. CNS infections are present, on average, in 47% of infected patients with an average case-fatality rate of 36% (Siegman-Igra et al., 2002). Listeria monocytogenes usually ranks as the third or fourth most common cause of bacterial meningitis in North America and Western Europe (Durand et al., 1993; Schuchat et al., 1997; Sigurdardottir et al., 1997; Hussein & Shafran, 2000; Kyaw et al., 2002). In calendar year 2005, the most recent year for which complete statistics are available, 896 cases of invasive listeriosis were reported to the US Centers for Disease Control and Prevention for an annual incidence of 0.28 cases per 105 (CDC, 2007). In contrast, infection rates are much higher in populations considered to be at higher risk. For example, the incidence during pregnancy has been estimated at c. 12 cases per 105 (Mylonakis et al., 2002), and among patients with AIDS was estimated at 115 cases per 105 (Jurado et al., 1993). Additionally, the first year estimated risk of listeriosis after starting infliximab, an inhibitor of tumor necrosis factor-α, is c. 4.3–15.5 cases per 105 (Slifman et al., 2003; Wallis et al., 2004). Other at risk patients include those at extremes of age, those taking immunosuppressive medications following organ transplantation and those with immunocompromising illnesses including cancer, autoimmune diseases, alcoholism and diabetes mellitus (Ciesielski et al., 1988; McLauchlin, 1990a; Goulet & Marchetti, 1996; Mylonakis et al., 1998; Aouaj et al., 2002). The US Department of Agriculture and the Food and Drug Administration Center for Food Safety enforce a zero-tolerance policy for contamination of ready to eat foods with Listeria. These efforts have led to a 24% reduction in the incidence of invasive listeriosis and a 37% reduction in pregnancy associated infections since 1996 (Voetsch et al., 2007).
Another interesting epidemiologic feature of L. monocytogenes is that the organism can be detected in the stools of a small percent of asymptomatic, healthy volunteers. Grif et al. (2003) detected the organism in 3.5% of stool specimens collected from three individuals over a 1-year period using PCR and recovered it from culture in 1.15%. Prolonged shedding was not detected and shedding did not correlate with overt illness. Further studies will be necessary to determine the impact of asymptomatic shedding on the acquisition of infection in others.
Listeria monocytogenes is a food borne pathogen that is frequently identified in poultry and dairy products although it can contaminate a wide variety of food stuffs (Farber & Peterkin, 1991; Wilson, 1995). Recent data from the two federal agencies responsible for overseeing food safety in the US, the Department of Agriculture (USDA) and the Food and Drug Administration (FDA) illustrate its widespread presence in food (http://www.fda.gov/oc/po/firmrecalls/archive.html). The USDA issued 333 food recalls in the 5 year period from 2002 to 2006, 108 (32.4%) of which were due to L. monocytogenes. The quantity of meat and poultry involved ranged from 5 lbs (c. 2 kg) to 27 400 000 lbs (12 454 545 kg) per recall with a median of 1100 lbs (500 kg) per event. By comparison, the FDA issued 126 food product recalls or alerts in 2006 alone, 19 (15%) of which were due to L. monocytogenes. In these instances bacteria were isolated from a diverse array of products including dairy (cheeses, raw milk), agricultural (strawberries, cut fresh fruit, sliced mushrooms), and various ready to eat foodstuffs such as coleslaw, crab dip, smoked salmon and turkey, egg salad, potato salad, and macaroni salad. The most commonly isolated bacteria belong to the serotypes 4b, 1/2a and 1/2b (Ooi & Lorber, 2005; Swaminathan & Gerner-Smidt, 2007).
Human listeriosis usually presents as one of three clinical syndromes namely febrile gastroenteritis, maternal-fetal/neonatal listeriosis, or bacteremia with or without cerebral infections such as meningitis, meningoencephalitis, rhombencephalitis or brain abscess (reviewed in Gray & Killinger, 1966; Vazquez-Boland et al., 2001). Less common focal infections derived from hematogenous spread include endocarditis, peritonitis, septic arthritis or endopthalmitis (Doganay, 2003). Focal infections including cholecystitis, prosthetic joint infection and infections of arterial grafts have also been described (Alleberger et al., 1989, 1992; Cone et al., 2001). Lastly, cutaneous listeriosis may complicate those with eczematous skin and occupational exposure to infected animals (McLauchlin & Low, 1994).
The role of L. monocytogenes as a potential food-borne diarrheal illness was suggested by primate studies in which large bacterial inocula orally administered produced diarrhea (Farber & Peterkin, 1991). Multiple outbreaks of listerial gastroenteritis have been described and typically occur in healthy persons (Ooi & Lorber, 2005). Following an incubation period of 6–49 h (median 20–25 h), most patients present with diarrhea, fever, abdominal pain, chills, headache and myalgias (Dalton et al., 1997; Frye et al., 2002). This is a self-limited disease with median durations of fever and diarrhea of 27 and 42 h, respectively, and most patients recover without antimicrobial treatment (Dalton et al., 1997).
Infection in pregnancy
Most women present with a bacteremic illness consisting of fever, chills, headache and leukocytosis with a 6–7 day illness before diagnosis (McLauchlin, 1990b; Mylonakis et al., 2002). The organism may be recovered from cultures of the cervix, amniotic fluid and placenta. Complications may include spontaneous abortion or stillbirth in c. 20% especially if infection occurs early in the pregnancy; preterm delivery and neonatal infection are also possible (Mylonakis et al., 2002). Up to two-thirds of surviving neonates born to mothers with listeriosis develop overt neonatal infection due to transplacental transmission from maternal bacteremia, or from exposure during transit through a colonized vaginal canal. Neonatal listeriosis is classified as either early (occurring in the first 5–7 days following delivery) or late infection (Larsson et al., 1979; McLauchlin, 1990b; Mylonakis et al., 2002). Early disease is often overt at time of delivery and associated with maternal infection. It presents as pneumonia, bacteremia, or meningitis. Meconium staining, respiratory distress, fever, lethargy, jaundice and rash are typically observed. In some neonates, the infection manifests as granulomatosis infantiseptica in which there are widespread micro-abscesses and granulomata especially in the liver, spleen and lungs. In contrast, late onset infection usually occurs in full term neonates delivered from uncomplicated pregnancies. It usually presents as meningitis with the presumed source of the organism from the mother's vaginal tract acquired at the time of delivery.
Invasive L. monocytogenes infection typically presents as bacteremia with or without an evident focus of infection, or as CNS infection including meningitis, meningoencephalitis, brainstem encephalitis (rhombencephalitis) and brain abscess. Most cases of listerial meningitis/menigoencephalitis are seen in patients >50 years of age and fever, altered sensorium and headache are the predominant symptoms (Brouwer et al., 2006). Meningeal signs on presentation are demonstrable in 26–54%, less commonly in the immunosuppressed (Skogberg et al., 1992; Mylonakis et al., 1998). The typical triad of bacterial meningitis of fever, neck stiffness and mental status change is observed in 43% (Brouwer et al., 2006). Focal neurological signs may be observed 16–37% and seizures in 4–17% (Mylonakis et al., 1998; Bartt, 2000; Brouwer et al., 2006). The cerebrospinal fluid (CSF) demonstrates pleocytosis in 75% usually with a neutrophil predominance although mononuclear cells may also be seen (Mylonakis et al., 1998; Bartt, 2000). The CSF protein is invariably elevated although hypoglycorrhachia is variably observed. Gram's stain of the CSF is positive for the organism in less than one third of cases, but cultures of the blood or CSF are positive in 75% and 80%, respectively (Bartt, 2000).
Two less commonly observed CNS infections are brainstem encephalitis (rhombencephalitis) and brain abscess. Rhombencephalitis accounts for c. 10% of listerial CNS infections and is a biphasic illness characterized by a prodrome of headache, nausea, vomiting and fever, followed by progressive brainstem and cerebellar dysfunction (Armstrong & Fung, 1993; Uldry et al., 1993; Antal et al., 2005a, b). The prodrome usually lasts 4–5 days then ends with the abrupt onset of neurological deficits; most commonly asymmetric cranial nerve palsies reflecting pontomedullary involvement. Usually, cranial nerves 5, 6, 7, 9 and 10 are involved with a 7th nerve palsy most commonly observed (Armstrong & Fung, 1993). Many patients will also demonstrate cerebellar findings such as ataxia and dysmetria or long tract signs (hemiparesis and hypesthesia). At maximal illness, 80% will have cranial nerve palsies and long tract signs (Armstrong & Fung, 1993). Nuchal rigidity is observed in <50% (Armstrong & Fung, 1993). A mononuclear pleocytosis is observed in 58% of CSF specimens, while variable protein and glucose concentrations have been reported (Bartt, 2000). CSF Gram's stain rarely demonstrates the organism. Blood and spinal fluid cultures are positive in 60% and 40% respectively (Armstrong & Fung, 1993; Antal et al., 2005a, b). CT scans may show hypodense lesions in the brainstem, but the superiority of MR imaging of the posterior fossa make it the radiological image of choice (Soo et al., 1993; Hatipoglu et al., 2007). Most survivors have significant neurological sequelae.
Several investigators have described the clinical features of CNS abscesses caused by L. monocytogenes (Dee & Lorber, 1986; Mylonakis et al., 1998; Cone et al., 2003). Most patients are immunosuppressed or have comorbid diseases, especially those in which two or more brain abscesses are found (Mylonakis et al., 1998). The typical patient presents with fever, headache and focal neurological deficits. Meningismus is rarely observed. Blood cultures are usually positive while CSF cultures, when obtained, are positive in less than half (Bartt, 2000). Lesions are best localized either by CT or MR imaging.
Treatment and outcome of invasive L. monocytogenes infections
Isolates are generally susceptible to the clinically important antibiotics, however are usually resistant to the cephalosporins (Heger et al., 1997; Hof et al., 1997). Ideally, an effective antibiotic must enter the host cells, bind to the penicillin-binding protein 3 (PBP3) of Listeria and remain active in the unfavorable intracellular environment (Hof et al., 1997; Temple & Nahata, 2000). Based on animal studies, case reports and in vitro susceptibility testing, ampicillin and penicillin, often in combination with gentamicin, are the drugs of choice for these infections (Hof et al., 1997; Marco et al., 2000; Safdar & Armstrong, 2003a). In penicillin allergic patients, trimethoprim/sulfamethoxazole is an acceptable second line agent (Michelet et al., 1999). Animal studies and in vitro data suggest a potential role for the newer fluoroquinolones, but clinical data are lacking (Michelet et al., 1999; Temple & Nahata, 2000). Care must be exercised with some antimicrobial combinations because they elicit antagonistic results in vitro.
Despite the growing problems with antimicrobial resistance in gram-positive bacteria, antibiotic resistance is uncommon among human isolates of Listeria spp. (Poulsen et al., 1988; Marco et al., 2000; Safdar & Armstrong, 2003a; Hansen et al., 2005). However, isolates are often intermediate or fully resistant to cephalothin, cefotaxime, cefepime, monobactams, d-ofloxacin and nalidixic acid (Hof et al., 1997). Thus, cephalosporins are not recommended for the treatment of listerial infections. Tetracycline resistance has been described (Marco et al., 2000). An additional treatment consideration is the role of antibiotic tolerance. Several laboratories have identified putative genetic loci that may regulate tolerance to bacteriocins, ampicillin and penicillin (Stack et al., 2005; Begley et al., 2006).
The case fatality rate of listeriosis is high reflecting both the underlying health status of patients and the virulence of the organism. Skogberg et al. (1992) reported nearly 30% mortality in those with underlying disease or receiving immunosuppressant medications and in neonates while no deaths were observed in healthy patients. The mortality among those with bacteremia complicating malignancy approaches 40% (Safdar & Armstrong, 2003b). Other correlates of mortality include low Glasgow coma scale at admission, thrombocytopenia, elevated hepatic transaminases, inappropriate antibiotic treatment or use of vancomycin and extremes of age (Skogberg et al., 1992; Mylonakis et al., 1998; Safdar & Armstrong, 2003b; Brouwer et al., 2006).
Pathophysiology of CNS infections
The mouse model of experimental L. monocytogenes infection for the study of brain invasion
Listeria monocytogenes is unique among neuroinvasive bacteria in that in vitro and in vivo data suggest it has the potential to invade the CNS by at least three different mechanisms (Drevets et al., 2004a). These include (1) transport across the blood-brain or blood-choroid barriers within parasitized leukocytes, (2) direct invasion of endothelial cells by extracellular blood-borne bacteria, or (3) retrograde (centripetal) migration into the brain within the axons of cranial nerves. Before considering neuroinvasive mechanisms, we will briefly review the microbiology of L. monocytogenes as an intracellular pathogen. Readers are directed to recent reviews for in-depth discussions of this topics (Hamon et al., 2006; Bierne & Cossart, 2007; Ireton, 2007).
Listeria monocytogenes can invade and replicate within a wide variety of mammalian cells. It can invade normally nonphagocytic cells such as endothelial cells, enterocytes, and fibroblasts, utilizing certain surface invasion proteins, of which internalin A (InlA) and internalin B (InlB) are the best studied (Ireton, 2007). The cellular ligand for InlA is E-cadherin which is present at adherens junctions between epithelial cells. Thus, its role in mediating cell entry has thus far been restricted to invasion across the gastrointestinal tract (Jacquet et al., 2004; Wollert et al., 2007). In contrast, InlB mediates entry into a wider variety of cells through binding to the more widely distributed cellular receptor Met, a protein kinase which is the ligand of endogenously produced hepatocyte growth factor. InlB also binds the globular portion of the receptor for the first component of complement (C1q). Binding either InlA or InlB protein to cells triggers intracellular signaling, although with some differences between them, that causes rearrangement of the cytoskeleton and internalization of bacteria (Hamon et al., 2006).
Additionally, L. monocytogenes is avidly phagocytosed by professional phagocytes, e.g. macrophages and neutrophils, by complement-dependent and complement-independent mechanisms (Drevets et al., 1992). After entering a host cell, L. monocytogenes rapidly escape from phagosomes through the combined actions of a pore-forming hemolysin, listeriolysin O, and two phospholipases, PlcA and PlcB, and then replicates in the cytosol (O'Riordan & Portnoy, 2002; Schnupf & Portnoy, 2007). In addition, cytosolic bacteria nucleate F-actin through the action of the ActA protein and the growing F-actin filaments form a molecular motor that propels bacteria through the cytosol and then into adjacent cells thus enabling the bacteria to spread cell-to-cell (Tilney & Portnoy, 1989; Kocks et al., 1992).
The mouse model of L. monocytogenes has been used for 50 years to understand neuroinvasive mechanisms by this pathogen (Asahi et al., 1957; Cordy & Osebold, 1959). This model reasonably recapitulates the histological features of disease in humans and naturally infected animals. However, as with any model, it has caveats and limitations. One is that immunocompetent mice are resistant to lethal infection and CNS invasion following gastrointestinal challenge with bacteria. At face value this might limit its role as a research tool given that the gastrointestinal route is the typical means by which humans are infected. However a closer analysis suggests that this resistance towards developing severe systemic disease following a gut challenge also mimics the human situation. For example, recent epidemiological investigations show that ingestion of high bacterial inocula typically does not produce severe invasive disease in immunocompetent humans, but rather a self-limited febrile gastroenteritis. These outbreaks occurred in largely young and healthy populations and were linked to ingestion of contaminated chocolate milk (c. 109 CFU mL−1), corn and tuna salad (>106 CFU g−1), and delicatessen style turkey meat (c. 107 CFU g−1) (Dalton et al., 1997; Aureli et al., 2000; Frye et al., 2002). Although estimates of the amount of bacteria ingested were as high as 1011 CFU per person in those who drank chocolate milk, these outbreaks produced no deaths or CNS infections and only one case of bacteremia. Active case finding based on positive blood cultures identified three additional patients that were not associated with the initial epidemic but had consumed contaminated milk (Dalton et al., 1997). This demonstrates that the absence of invasive disease in the original outbreak was not due to reduced bacterial virulence.
Mechanisms used by L. monocytogenes to invade the gastrointestinal tract have been recently reviewed (Gahan & Hill, 2005). A key observation is that a single amino acid difference, Pro→Glu switch at position 16 of E-cadherin, between human and mouse, respectively, prevents InlA-mediated invasion of gut epithelial cells in mice (Lecuit et al., 1999). The relevance of this was established by studies demonstrating that transgenic mice that express human E-cadherin in their gut epithelium were more likely to succumb to fatal infection than were normal mice following gastrointestinal challenge with 5 × 1010 CFU bacteria (Lecuit et al., 2001). However, neuroinvasion can be achieved in a mouse model without genetic alteration. For example, Czuprynski et al. (2002, 2003) showed that the tendency for mice to develop CNS infection after gastric inoculation depends in part upon the strain of mouse as well as the strain of bacteria used. Bacteria within serotypes 4b, which are usually responsible for epidemic listeriosis, are more invasive than are serotype 1/2 strains, and A/J mice more frequently developed lethal systemic infection, including CNS infection, after gastric inoculation than did similarly infected C57BL/6 mice. In an adaptation of the mouse model, Altimira et al. (1999) showed that 25% of mice developed brain lesions after being exposed to 5 × 109 CFU bacteria via drinking water daily for 7–10 days. Lastly, we showed that mice pharmacologically immunosuppressed with cyclosporin A and hydrocortisone, which mimics organ transplantation, readily developed CNS after inoculation via gastric lavage infection despite systemic treatment with gentamicin to kill extracellular bacteria (Drevets et al., 2001). Collectively these data show that the mouse model is a reasonable mimic of the human situation with regard to the frequency of CNS invasion following gastrointestinal infection. Nevertheless, most investigators favor parental inoculation of bacteria for the study of the mechanisms of CNS entry.
Systemic L. monocytogenes infection and the key role of infected monocytes
In most clinical situations L. monocytogenes enters the CNS from the bloodstream in the context of an established systemic infection. This is readily recapitulated in experimental L. monocytogenes infection in mice. Bacteria injected intravenously. are removed from the bloodstream within a matter of hours by the liver and spleen, but the brain and bone marrow remain uninfected (Fig. 1) (Rosen et al., 1989; Berche, 1995; Gregory et al., 1996). Systemic infection is marked by bacterial replication predominantly in the liver and spleen, and later also in the bone marrow (de Bruijn et al., 1998; Join-Lambert et al., 2005). Clearance mechanisms can be overwhelmed with large amounts of bacteria, e.g. >10 lethal dose 50% (LD50), so that the bloodstream is never sterilized. However, in models in which lethal infection is produced using fewer bacteria, a secondary bacteremia develops and it is during this phase that bacteria enter the CNS (Berche, 1995; Join-Lambert et al., 2005). Interestingly, our data show that the secondary bacteremia is composed of a combination of cell-free and intracellular bacteria (Drevets, 1999). Furthermore, intracellular L. monocytogenes are clearly in a parasitic relationship with their host leukocytes in the blood as demonstrated by the ability of these bacteria to escape phagosomes and polymerize F-actin, spread to endothelial cells in vitro, and to cause disseminated infection when transferred into other animals (Drevets, 1999, 2001).
Initial phenotypic characterization showed that more than 90% of L. monocytogenes-infected phagocytes in the peripheral blood were mononuclear and expressed CD11b (Drevets et al., 2004b). Thus, they fit the broad definition of monocytes (Ziegler-Heitbrock, 2000; Hume, 2006). In the mouse, the predominate subpopulations of monocytes are distinguished in part by variable expression of the Ly-6C antigen. They are referred to as Ly-6Chigh and the more mature Ly-6Cneg/low monocytes, also known as Gr-1high and Gr-1low, respectively, with a transition population described as Ly-6Cint (Geissmann et al., 2003; Qu et al., 2004; Sunderkotter et al., 2004). Different monocyte subpopulations express different chemokine receptors with Ly-6Chigh cells bearing the CCR2highCX3CR1low phenotype whereas Ly-6Cneg cells are CCR2neg/lowCX3CR1high (Geissmann et al., 2003; Tsou et al., 2007). Similarities in the expressions of CD62L, CCR2, and CX3CR1 between mouse and human monocytes suggest that mouse Ly-6Chi monocytes correspond to human CD64+CD14+CD16neg monocytes, whereas mouse Ly-6Cneg monocytes resemble human CD64negCD14+CD16+ cells (Geissmann et al., 2003). Current data suggest that different subpopulations perform distinct roles in vivo. For example, at steady state Ly-6Chigh monocytes show little trafficking to organs other than the spleen (Geissmann et al., 2003). However, expression of CCR2 on Ly-6Chigh monocytes is thought to be crucial for mediating release of these cells from the bone marrow in response to chemokine ligands CCL2 or CCL7, and for directing them into acutely inflamed organs and spaces (Geissmann et al., 2003; Henderson et al., 2003; Serbina & Pamer, 2006; Tsou et al., 2007). By comparison, Ly-6Cneg monocytes migrate into a variety of organs at steady state, possibly to perform homeostatic functions (Geissmann et al., 2003), whereas Ly-6Cint monocytes have a predilection for becoming lymphatic-homing dendritic cells (Qu et al., 2004).
Acute L. monocytogenes infection causes a dramatic shift in the homeostatic levels of the monocyte subsets in favor of greater representation of the Ly-6Chigh subset (Fig. 2) (Drevets et al., 2004b). This is in accord with their role as the main monocyte subset that exits the bone marrow in response to peripheral demand (Sunderkotter et al., 2004). Analyses of infection according to Ly-6C expression shows that both Ly-6Chigh and Ly-6Clow monocytes are infected in vivo (Drevets et al., 2004b; Join-Lambert et al., 2005). However due to their greater representation Ly-6Chigh monocytes harbor the majority of cell-associated L. monocytogenes in the bloodstream.
The importance of bone marrow infection during systemic disease
Given the importance of parasitized monocytes, it is necessary to understand from where these cells originate as well as to identify where they go when they leave the circulation. As noted above, phenotypic similarities functional studies indicate that Ly-6Chigh monocytes are recent immigrants from the bone marrow. Similarly, recent studies show that the phenotype of infected monocytes in the peripheral blood is similar in many respects to that of monocytes in the bone marrow suggesting that L. monocytogenes-parasitized monocytes originate in the bone marrow (Drevets et al., 2004a, b; Join-Lambert et al., 2005). Literature review shows that L. monocytogenes can be isolated from the bone marrow of mice during lethal or sublethal infection although bacteremia typically occurs only in lethal infection (de Bruijn et al., 1998; Join-Lambert et al., 2005). Interestingly, data from de Bruijn et al. (1998) showed that the bone marrow was sterile 1 h following intraperitoneal injection of a sublethal amount of L. monocytogenes, but was infected 24 h later. This was confirmed by Join-Lambert et al. (2005) and is illustrated by our data in Fig. 1. These results show that the bone marrow does not simply filter bacteria from the bloodstream as do the liver and spleen and raises the question of how the bone marrow becomes infected. One possibility is that intracellular bacteria are transported within senescent neutrophils that home to the marrow after exiting heavily infected tissues (Tacke et al., 2006). Once in the bone marrow, the number of organisms increase rapidly for 72 h in sublethal infection then declines in concert with that observed in other organs, whereas in lethal infection there is unchecked replication (de Bruijn et al., 1998; Join-Lambert et al., 2005).
de Bruijn et al. (1998), analyzing the cellular composition of the marrow by flow cytometry following sublethal L. monocytogenes infection, demonstrated that ‘mature’ monocytes and neutrophils exited the marrow within the first 48 h followed by an expansion of Ly-6ChighCD31pos myeloid cells that peaked on the 7th day postinfection. The loss of neutrophils from the bone marrow partly explains its vulnerability to L. monocytogenes infection (Kratz & Kurlander, 1988; Li et al., 2004). Additionally, it shows that cells which take up L. monocytogenes lack bactericidal activity and suggests that the bone marrow is a vulnerable niche that can be exploited by this pathogen once invasive infection is established. As reflected in the bloodstream, infection also skews the myeloid component of the bone marrow in favor of Ly-6ChighCD31pos myeloid precursors (Fig. 3) (de Bruijn et al., 1998; Join-Lambert et al., 2005). Specific changes induced by infection in the monocyte compartment, identified as Ly-6ChighCD11bpos cells, include upregulation of both CD31 and CD11b which is contrary to the normal maturational pattern of increased surface expression of CD11b with decreased CD31 expression (Leenen et al., 1994).
Sophisticated studies by Join-Lambert et al. (2005) used green fluorescent protein (GFP)-expressing L. monocytogenes to show bacterial uptake by Ly-6ChighCD11bposLy-6Glow cells, i.e. monocytic cells, in the bone marrow. Electron microscopy identified bacteria within myeloid cells of both monocytic and granulocytic lineages in various stages of development. Our data show bacterial infection within Ly-6ChighCD11bpos bone marrow cells that are negative for other lineage markers including CD3, CD19, Ly-6G, NK1.1, TER 119, thus establishing their identity as cells of the monocytic lineage (Fig. 4). However, other cells such as granulocytic precursors (Ly-6CposCD11bposCD31varLy-6Gpos) also contain bacteria. These studies support the hypothesis that the organism parasitizes bone marrow myeloid cells. Further evidence for this parasitism comes from experiments in which gentamicin-treated mice are infected with L. monocytogenes NF-L512, a strain that contains an actA-gfpuv-plcB transcriptional fusion in single copy such that GFP is expressed preferentially when intracellular bacteria are engaged in their parasitic life cycle (Freitag & Jacobs, 1999; Drevets et al., 2001). The presence of GFP-expressing bacteria within cells is a reliable marker for intracellular parasitism (Fig. 4d). Collectively, this body of information demonstrates that bone marrow cells identified phenotypically and morphologically as Ly-6ChiCD31posCD11bpos monocyte precursors are a reservoir for L. monocytogenes infection. Interestingly, these cells are capable of establishing CNS infection when isolated from infected animals and inoculated into uninfected ones (Join-Lambert et al., 2005). Thus, Ly-6ChighCD11bpos monocytes are able to transport bacteria from the bone marrow to the bloodstream and from there into the brain.
Brain invasion from the bloodstream via trafficking of monocytes
The hypothesis that infected phagocytes play a key role in establishing CNS infection by L. monocytogenes has been supported in part by their appearance in the brains of experimentally infected mice (Prats et al., 1992). However, this histological study did not distinguish monocytes transporting bacteria into the brain from those recruited into it after infection was already established. To demonstrate that infected phagocytes are not mere bystanders, but rather are transporters of intracellular bacteria to the brain, cell-free bacteria were eliminated from the bloodstream during in vivo infection (Drevets et al., 2001). This was accomplished by treating mice with gentamicin, a bactericidal antibiotic that penetrates cells poorly, delivered by surgically implanted osmotic pumps. In these experiments, gentamicin-treated mice had fewer bacteria in the bloodstream than did untreated mice indicating that essentially all bacteria were cell-associated. Despite clearance of cell free bacteria, the gentamicin-treated mice developed brain infection to the same extent as did untreated mice.
Further evidence supporting the role of parasitized phagocytes in establishing CNS infection in mice comes from observations that bacterial infection of the brain by L. monocytogenes is more efficient when bacteria are injected intravenously within infected leukocytes than when injected as cell free bacteria (Drevets, 1999; Join-Lambert et al., 2005). Studies for our laboratories show that Ly-6Chigh monocytes accumulate in the brains of systemically infected mice, although neutrophils do not, and that they harbor the majority of bacteria associated with CD11bpos cells in the brain (Fig. 5) (Drevets et al., 2004a, b). Additionally, infected monocytes have been observed adhering to the walls of brain capillaries and undergoing transmigration into the subarachnoid space (Join-Lambert et al., 2005). Once infected phagocytes arrive at or in the CNS, in vitro data suggest that bacteria carried by them can enter parenchymal cells including endothelial cells and neurons by cell-to-cell spread or can be phagocytosed by microglia (Drevets et al., 1995; Dramsi et al., 1998).
What is not yet known are the mechanisms by which Ly-6Chigh monocytes are attracted to CNS during infection. Monocyte entry into the CNS, like other forms of leukocyte migration, is a sequential process of rolling along capillary endothelial cells, followed by adhesion, and then diapedesis (Imhof & Aurrand-Lions, 2004). A chemokine gradient enables directional migration of leukocytes expressing the cognate chemokine receptors. A key role for CCL2 and CCR2 in recruiting monocytes into the CNS has been shown in experiments using transgenic mice that overproduce CCL2 in the brain (Fuentes et al., 1995), and by studies that showed reduced numbers of monocytes/macrophages in the brains of CCL2−/− and CCR2−/− mice undergoing experimental autoimmune encephalitis (Fife et al., 2000; Huang et al., 2001), or experimental stroke injury (Hughes et al., 2002; Dimitrijevic et al., 2007). Additionally, previous results from our laboratories showed that CCL2 mRNA is strongly upregulated in brains of L. monocytogenes-infected mice and coincides with the monocyte influx suggesting that CCL2 and CCR2 are important in this setting as well (Drevets et al., 2004a, b). Nonetheless, ligands other than CCL2, such as CCL7 (MCP-3) (Thirion et al., 1994; Menten et al., 2001; Tsou et al., 2007) and CCL12 (MCP-5) (Jia et al., 1996; Sarafi et al., 1997), can bind and signal through CCR2 and could attract Ly-6Chigh monocytes. Furthermore, Ly-6Chigh monocytes have shown to use CX3CR1 and CCR5, as well as CCR2, to enter atherosclerotic plaque although the role of these ligands and receptors in CNS entry is not clear (Tacke et al., 2007).
Brain invasion from the bloodstream through infection of endothelial cells
The extent to which cell free L. monocytogenes in the bloodstream contribute to brain invasion also has been investigated. In vitro studies from our laboratories and from others showed that L. monocytogenes can invade and replicate within a wide variety of endothelial cells such as umbilical vein and brain microvascular endothelial cells (reviewed in Vazquez-Boland et al., 2001; Drevets et al., 2004a). However, experimental results showed that the invasion proteins, in particular InlB, did (Greiffenberg et al., 1998, 2000) or did not (Wilson & Drevets, 1998) mediate invasion of brain microvascular endothelial cells in vitro. Discrepant results were later attributed to the finding that normal human serum contains antibodies that strongly inhibit L. monocytogenes invasion of human brain microvascular endothelial cells in vitro (Hertzig et al., 2003). Thus, invasion of these microvascular cells in serum-free conditions in vitro is dependent upon InlB, but its role in vivo would be limited. The conclusion that antibody inhibits CNS invasion by extracellular bacteria is supported by experiments showing that IgM-deficient (μMT) mice had 13-fold more L. monocytogenes in the brain, but fewer organisms in the spleen, than did control (C57BL/6) animals 6 h after injection with a high inoculum (2 × 107 CFU) of bacteria (Ochsenbein et al., 1999).
Epidemiologic data in humans also suggest InlA does not contribute to CNS invasion from the bloodstream. Jacquet et al. (2004) analyzed protein expression of full-length internalin in 300 clinical strains of L. monocytogenes and a representative set of 150 strains collected from food products during the same time, but which were not associated with clinical outbreaks. They found that 96% of clinical strains expressed full-length internalin compared with 65% of isolates from food products (odds ratio, 12.73; 95% confidence interval, 6.27–26.34; P<1 × 10−7). Further analysis showed that 100% of strains from pregnancy-related cases, 98% from patients with CNS infections, and 93% of strains from patients with bacteremia expressed full-length internalin. The authors identified a statistically significant role for InlA in invasion across the gastrointestinal tract and for entering the placenta from the bloodstream. However, no significant correlation was established for the role of InlA for crossing into the CNS from the bloodstream. These data were recently recapitulated in the mouse by Wollert et al. (2007) who constructed a L. monocytogenes strain expressing a mutated InlA protein that binds murine E-cadherin with similar affinity as it does to human E-cadherin. In vivo experiments showed that the LD50 of bacteria expressing mutated InlA was c. 1000-fold lower than the parent strain following intragastric infection, but there was no significant difference between them in mortality or in bacterial loads in liver, spleen, or mesenteric lymph nodes after intravenous infection. The authors concluded that after the gastrointestinal tract is colonized and the intestinal epithelium is infected, dissemination of L. monocytogenes is largely independent of InlA.
Brain invasion via the neural route
In addition to neuroinvasion from the bloodstream, L. monocytogenes also invades the CNS through a neural route. The neural route refers to a mechanism in which bacteria enter axons in the periphery then centripetally migrate within them into the brainstem. This process was identified by finding L. monocytogenes in peripheral nerve axons of the trigeminal nerve in pathological specimens derived from sheep and goats with naturally acquired Listerial encephalitis (Charlton & Garcia, 1977; Otter & Blakemore, 1989). Several investigators have experimentally recapitulated this by inoculation of bacteria into oral and ocular tissues of goats and mice, or by direct injection into peripheral neurons (Asahi et al., 1957; Antal et al., 2001; Jin et al., 2001). In vitro experiments by several different investigators evaluated the ability of L. monocytogenes to enter cultured brain parenchymal cells, such as neurons and glial cells including microglia, astrocytes and oligodendrocytes (Peters & Hewicker-Trautwein, 1994, 1996; Dramsi et al., 1998; Dons et al., 1999). These studies revealed that non-neuronal cells take up bacteria more avidly than do neurons, with microglia being the most easily infected. In contrast, binding and invasion of neurons occurs to some extent and is probably internalin-independent, but the frequency by which it happens is quite variable depending upon the origin of the neurons, how they are cultured, and how bacterial infection is measured (Dramsi et al., 1998; Dons et al., 1999; Jin et al., 2004). A more reliable means for L. monocytogenes to infect neurons is through cell-to cell spread from an infected macrophage as demonstrated by Dramsi et al. (1998). Importantly, a key role for macrophages in facilitating neural invasion in vivo was established by Jin et al. (2002) who showed that L. monocytogenes inoculated into oral tissues of macrophage-deficient op−/− (M-CSFR−/−) mice were less able to enter neurons than after injection into normal mice.
Collectively these data provide a clearer understanding of the neural route of invasion in animals and a mechanism by which these cells can be infected, but the applicability to human infection remains unclear. Recently, however, careful analysis of autopsy material from nine humans who died of brainstem encephalitis revealed inflammatory infiltrates within nuclei, tracts, and intra-parenchymal parts of cranial nerves V, VII, IX, X, and XII, which innervate different areas of the oropharynx (Antal et al., 2005a). These patients were part of a retrospective study of invasive listeriosis in Norway during the years 1977–2000 which suggested that brainstem infection was present in 11% of cases in nonpregnant adults (Antal et al., 2005b). This number is considerably higher than previously reported and needs to be considered in light of the difficulties of making a retrospective diagnosis of brainstem encephalitis in the absence of modern imaging techniques, e.g. magnetic resonance imaging with gadolinium enhancement, or at autopsy. Nonetheless, it suggests that humans as well as ruminants are attacked by the neural route of CNS invasion.
There have been many exciting new developments in the study of L. monocytogenes as an intracellular pathogen and as a neuropathogen. Certainly many questions remain to be answered and many yet to be asked. Regarding clinical disease and at-risk populations, there may be polymorphisms in key immune response genes that contribute to the susceptibility of seemingly healthy individuals to severe L. monocytogenes infection, as have been recently found for other pathogens (Texereau et al., 2005). Because monocytes have dual roles in the pathogenesis of L. monocytogenes infections and in host defenses against this organism, in-depth studies of monocyte biology are likely to be even more important. For example, it is not fully understood how bone marrow monocytes first become infected or why these cells are susceptible to intracellular infection rather than being able to kill the bacteria. The mechanisms by which infected monocytes are recruited to and enter the CNS have yet to be elucidated. Additionally, given the mismatch between human and murine E-cadherin, are there other differences between mice and humans that influence the means by which L. monocytogenes invades the brain in the mouse model?
One of the strengths of experimental L. monocytogenes infection is that it is a useful model for understanding the pathogenesis of diseases caused by intracellular bacteria. Thus, one of the key questions to be answered is to what extent does L. monocytogenes invasion of the CNS via infected monocytes share common features with other intracellular pathogens that use a similar mechanism? For example, recent data show that Toxoplasma gondii enters the brain within parasitized CD11bpos monocytes (Courret et al., 2006). This interesting paper showed that CD11cnegCD11bpos monocytes and to a lesser extent CD11cpos CD11b+/− dendritic cells, transported intracellular tachyzoites in the bloodstream and then into the brain. Similarly, lentiviruses, for example HIV-1, likely enter the CNS through infected leukocytes and have been a model for other pathogens (Gonzalez-Scarano & Martin-Garcia, 2005). In this respect, it will be important to understand the importance of pathogen-specific features relative to commonalities in the host responses to them.