Correspondence: Ben Adler, Australian Research Council Centre of Excellence in Structural and Functional Microbial Genomics, Monash University, Clayton, Vic. 3800, Australia. Tel.: +61 3 9902 9177; fax: +61 3 9902 9222; e-mail: firstname.lastname@example.org
Melioidosis, a febrile illness with disease states ranging from acute pneumonia or septicaemia to chronic abscesses, was first documented by Whitmore & Krishnaswami (1912). The causative agent, Burkholderia pseudomallei, was subsequently identified as a motile, gram-negative bacillus, which is principally an environmental saprophyte. Melioidosis has become an increasingly important disease in endemic areas such as northern Thailand and Australia (Currie et al., 2000). This health burden, plus the classification of B. pseudomallei as a category B biological agent (Rotz et al., 2002), has resulted in an escalation of research interest. This review focuses on the molecular and cellular basis of pathogenesis in melioidosis, with a comprehensive overview of the current knowledge on how B. pseudomallei can cause disease. The process of B. pseudomallei movement from the environmental reservoir to attachment and invasion of epithelial and macrophage cells and the subsequent intracellular survival and spread is outlined. Furthermore, the diverse assortment of virulence factors that allow B. pseudomallei to become an effective opportunistic pathogen, as well as to avoid or subvert the host immune response, is discussed. With the recent increase in genomic and molecular studies, the current understanding of the infection process of melioidosis has increased substantially, yet, much still remains to be elucidated.
From environmental saprophyte to opportunistic pathogen
Melioidosis, caused by B. pseudomallei, is endemic in tropical areas between latitudes 20°N and 20°S; it is most commonly reported in South East Asia and Northern Australia. The primary reservoirs for B. pseudomallei include rice paddies, still or stagnant waters and moist tropical soils (Brett & Woods, 2000); bacteria have also been recovered from the roots of plants (Holden et al., 2004). Burkholderia pseudomallei can persist for long periods under low-nutrient conditions; the organism has been cultured from distilled water 10 years after inoculation (Aldhous, 2005). Burkholderia pseudomallei grows best in soil with a water content of 15% (Leelarasamee, 2004) and most infections occur during the rainy season when bacteria are leached from the soil (Brett & Woods, 2000; Currie et al., 2000).
Burkholderia pseudomallei is believed to obtain nutrition from rotting organic matter and opportunistically from invasion of protozoa. Initial adhesion of B. pseudomallei to the free-living protozoan Acanthamoeba astronyxis involves polar attachment via flagella (Inglis et al., 2003). Following engulfment by pseudopodia, the viable bacteria are observed both within vacuoles and free in the cytoplasm (Inglis et al., 2000). It is predicted that mechanisms similar to those used for invasion and survival within this environmental niche are also used during infection of human macrophages.
Molecular typing methods have shown that there is a significant diversity within both environmental and clinical isolates of B. pseudomallei; however, individual isolates from either grouping can be identical (Cheng & Currie, 2005). The closely related species Burkholderia thailandensis, which was initially identified as an avirulent environmental B. pseudomallei isolate, has a similar environmental niche, but is unable to cause disease. The reason for the attenuation of B. thailandensis, in comparison with B. pseudomallei, has been associated with the presence of a functional arabinose biosynthesis operon in B. thailandensis, which is largely deleted in B. pseudomallei. Introduction of the complete B. thailandensis arabinose biosynthesis operon into B. pseudomallei resulted in the downregulation of a number of type III secretion genes and the strain displayed reduced virulence in Syrian hamsters (Moore et al., 2004). Burkholderia thailandensis, which shares many similarities with B. pseudomallei, including intracellular invasion, has been used as a model organism in which to study B. pseudomallei virulence (Yu et al., 2006; Haraga et al., 2008). Significant genetic differences have been reported between B. pseudomallei strains that differ in virulence potential (Vesaratchavest et al., 2006), but the level of virulence of B. pseudomallei strains isolated from the environment is not significantly different from that of the clinical strains. Furthermore, no clear difference in virulence was observed between strains isolated from fatal and nonfatal melioidosis cases (Ulett et al., 2001). Thus, while B. pseudomallei strains differ in their individual ability to cause disease, the outcome also clearly depends on the immune status and response of the infected host.
Initial epithelial attachment
The main routes of infection with B. pseudomallei are via percutaneous inoculation, inhalation or aspiration (Currie et al., 2000). Ingestion has also been proposed as a possible route of infection (Currie et al., 2001). Thus, initial infection occurs at the epithelial cell layer of either the abraded skin or the mucosal surface (Fig. 1). The attachment of B. pseudomallei to human pharyngeal epithelial cells appears to be mediated by a thin polysaccharide layer around the bacteria, putatively identified as a capsule (Ahmed et al., 1999). The results of attachment inhibition studies suggested that B. pseudomallei binds to the asialoganglioside aGM1–aGM2 receptor complex (Gori et al., 1999). The bacterial surface molecule responsible for aGM1–aGM2 binding is unknown; however, the closely related Pseudomonas aeruginosa attaches to this same complex via type IV pili (Comolli et al., 1999). Furthermore, a B. pseudomallei K96243 type IVA pili mutant strain displayed reduced adhesion to epithelial cell lines (Essex-Lopresti et al., 2005). In B. pseudomallei strain 08, the pilA gene was upregulated at 27 °C, a temperature at which epithelial attachment is maximized, and bacteria also formed pili-mediated microcolonies. The resultant bacterium–bacterium interactions were enhanced, but not essential for bacterium–epithelial cell interactions (Brown et al., 2002; Boddey et al., 2006). However, the temperature-dependent regulation of pilA and microcolony formation does not occur in B. pseudomallei strain K96243, indicating strain-to-strain variations in attachment mechanisms (Boddey et al., 2006). While the exact mechanisms of initial attachment and the role of type IV pili remain uncharacterized, it is clear that the level of B. pseudomallei attachment and the subsequent invasion, at least in cell culture, is actually quite low (Ahmed et al., 1999).
Burkholderia pseudomallei is a facultative intracellular pathogen and is able to actively invade and multiply in phagocytic and nonphagocytic cell lines (Pruksachartvuthi et al., 1990; Jones et al., 1996). The exact mechanism of invasion remains unknown, but inhibition of actin polymerization reduces the level of invasion (Jones et al., 1996). Rearrangement of the host actin cytoskeleton is induced by an effector of the Burkholderia secretion apparatus (Bsa) type 3 secretion system (T3SS3), BopE, which has in vitro activity as a guanine nucleotide exchange factor (Stevens et al., 2003). Furthermore, a bopE mutant demonstrated reduced invasion of epithelial cells. A T3SS3 apparatus (bipD) mutant exhibited a greater impairment of invasion than the bopE mutant, suggesting that more than one T3SS3 effector is involved in invasion (Stevens et al., 2003).
A mutagenesis screen to characterize the invasion mechanisms of B. pseudomallei identified a gene encoding a predicted two-component response regulator (irlRS) that was involved in the invasion of epithelial cells, but not macrophages (Jones et al., 1997). However, the mutant remained virulent in diabetic rats, hamsters (Jones et al., 1997) and C57BL/6 mice (Wiersinga et al., 2008a). These data suggest that either the observed in vitro tissue culture invasion defect merely represents a delay in invasion, or that B. pseudomallei infection can proceed without epithelial cell invasion.
Burkholderia pseudomallei can multiply within phagocytes, including neutrophils, monocytes and macrophages without activating a bactericidal response (Pruksachartvuthi et al., 1990; Jones et al., 1996). The survival of B. pseudomallei within macrophages has been the subject of considerable study and macrophage cell lines are commonly used as an in vitro model for phagocytic cell uptake and survival. Although a degree of lysosome fusion is detected within B. pseudomallei-infected human macrophages, the proliferation of surviving bacteria eventually overwhelms the macrophage (Nathan & Puthucheary, 2005). However, if the macrophages are activated by interferon-γ (IFN-γ), an enhanced killing of B. pseudomallei occurs (Miyagi et al., 1997). Studies with chemical inhibitors suggest that macrophage-based B. pseudomallei killing is primarily due to reactive nitrogen intermediates (RNI), while reactive oxygen intermediates (ROI) play a lesser role (Miyagi et al., 1997). However, in contrast, macrophages derived from ROI-deficient, but not RNI-deficient, knockout mice displayed an impaired ability to clear B. pseudomallei. Furthermore, ROI-deficient mice demonstrated an increased mortality to B. pseudomallei infection while the RNI-deficient mice were more resistant (Breitbach et al., 2006). Thus, ROI activity appears to be important for in vivo macrophage-based killing of B. pseudomallei.
Given the importance of the ROI response in controlling B. pseudomallei intracellular replication, the ability of B. pseudomallei to modulate this bactericidal response is an important mechanism of pathogenesis. Burkholderia pseudomallei represses inducible nitric oxide synthase (iNOS) expression (Fig. 1) by activating the expression of two negative regulators, a suppressor of cytokine signalling 3 (SOCS3) and cytokine-inducible src homology 2-containing protein (CIS) (Ekchariyawat et al., 2005). These negative regulators are activated by the presence of intracellular B. pseudomallei via an unknown intracytoplasmic receptor(s) (Ekchariyawat et al., 2007). The B. pseudomallei components of this pathway may be regulated by the RNA polymerase σ factor, RpoS, as rpoS mutants induce increased levels of iNOS, which in turn limits intracellular growth (Utaisincharoen et al., 2006). Suppression of iNOS can be overcome by costimulation of macrophages with IFN-γ (Utaisincharoen et al., 2003, 2004) but not if the IFN-γ is added postinfection (Ekchariyawat et al., 2005). These data indicate that IFN-γ-dependent induction of iNOS in activated macrophages is critical for optimal clearance of B. pseudomallei (Miyagi et al., 1997).
Burkholderia pseudomallei evasion of cellular autophagy
Invasion of both epithelial and macrophage cell lines by B. pseudomallei induces autophagy, a cellular catabolic pathway, which forms a component of the innate immune response. Under normal conditions, B. pseudomallei is capable of actively evading autophagic killing (Fig. 1), but induction of autophagy with rapamycin resulted in decreased intracellular bacterial survival (Cullinane et al., 2008). T3SS3 may play a role in the evasion of autophagy, as a mutant in the effector BopA demonstrated an increased colocalization with autophagic vesicles and decreased intracellular survival (Cullinane et al., 2008). Thus, B. pseudomallei has evolved mechanisms to inhibit host innate immune responses, including the release of ROI and stimulation of autophagy, resulting in increased intracellular survival and replication.
The level of B. pseudomallei cytotoxicity for macrophages is strain dependent; some strains cause macrophage apoptosis (Kespichayawattana et al., 2000), some cause caspase-1-dependent cell lysis (Sun et al., 2005) while others have no effect on macrophage viability (Pruksachartvuthi et al., 1990). Bacterial internalization is a prerequisite for macrophage death, as no cytotoxicity was observed in an invasion-deficient B. pseudomallei T3SS3 mutant or when invasion was chemically inhibited (Sun et al., 2005). Furthermore, cell death is at least partially dependent on RpoS, as an rpoS mutant failed to induce cytotoxicity (Lengwehasatit et al., 2008). Macrophage lysis may represent an escape mechanism for B. pseudomallei once ‘sufficient’ bacterial replication has occurred (Fig. 1).
Burkholderia pseudomallei is capable of intracellular spread via membrane protrusions that extend to neighbouring cells and through which bacteria travel by actin-mediated motility (Kespichayawattana et al., 2000; Breitbach et al., 2003). Burkholderia pseudomallei recruits host actin-associated proteins Arp3, p21 (Arp2/3 complex) and α-actinin; however, these proteins are not essential for actin polymerization (Breitbach et al., 2003). BimA, a B. pseudomallei autosecreted protein, interacts with monomeric actin in vitro and localizes to the bacterial pole at which actin polymerization occurs. A B. pseudomallei bimA mutant was unable to form actin tails (Stevens et al., 2005); therefore, BimA is essential for actin-mediated intracellular motility. BimA-dependent intracellular motility allows B. pseudomallei to move efficiently through both epithelial and macrophage cells while avoiding the host immune response.
As a result of intracellular motility, cell fusion occurs and multinuclear giant cells (MNGC) are formed (Kespichayawattana et al., 2000). This phenotype may be regulated by RpoS, as rpoS mutants fail to stimulate MNGC formation (Utaisincharoen et al., 2006). These data suggest that B. pseudomallei stimulates the formation of MNGC for intracellular spread once sufficient bacterial replication has occurred within an infected cell. MNGC were found to express osteoclast, bone-remodelling, cell markers under the regulation of the B. pseudomallei lfpA gene; however, these osteoclast-like cells cannot reabsorb bone matrix. A lfpA mutant was attenuated for virulence in both hamsters and BALB/c mice (Boddey et al., 2007), consistent with the notion that the formation of MNGC and the intracellular spread of B. pseudomallei are important for progression of infection (Fig. 1).
Localized disease, such as pneumonia and abscesses are typical in melioidosis; however, B. pseudomallei can spread to secondary sites, including organs such as liver, spleen or brain, or to the blood, resulting in septicaemia (White, 2003). The exact mechanism of the secondary spread of B. pseudomallei is unknown, although invasion of macrophages would allow for transport via the lymphatic system to the spleen and other organs (Fig. 1). Burkholderia pseudomallei-containing foci are commonly found in the spleens of chronically infected mice (Hoppe et al., 1999). Furthermore, an inflammation at the site of infection provides greater access to the circulatory system; B. pseudomallei can survive within the human serum due to the protective effects of the polysaccharide capsule and lipopolysaccharide. Both capsule (Reckseidler-Zenteno et al., 2005) and lipopolysaccharide (DeShazer et al., 1998) mediate a resistance to complement-mediated killing, while lipopolysaccharide is also responsible for a resistance to cationic peptides (Burtnick & Woods, 1999) (Fig. 1).
Patients who recover from B. pseudomallei septicaemia maintain high levels of antibody for years, suggesting either continuous exposure to the organism or sequestration of bacteria in intracellular or cryptic sites (Vasu et al., 2003). Neither the site(s) of latency nor the mechanisms by which B. pseudomallei remains undetected are clear (Gan, 2005). However, recurrent disease is common, occurring in 6–13% of cases, frequently due to relapse rather than reinfection, especially when occurring within a year of primary infection (Maharjan et al., 2005). The bacteria can remain latent for prolonged periods until activated by trauma or immunosuppression; cases of 18 (Koponen et al., 1991), 26 (Mays & Ricketts, 1975) and 62 (Ngauy et al., 2005) years of latency have been documented. These long latency periods suggest that B. pseudomallei has the ability to enter a dormant state where it can avoid immune surveillance, most probably in an intracellular location.
Virulence factors of B. pseudomallei
A number of virulence factors have been proposed to be involved in the pathogenesis of B. pseudomallei. Roles for capsule, lipopolysaccharide, the T3SS3, flagella and certain quorum-sensing (QS) molecules have been demonstrated. The effect on virulence of a number of other factors including pili, the type 6 secretion system (T6SS), secreted factors and regulatory genes has also been studied, with current data indicating that each plays a moderate to minor role in virulence (Table 1).
Table 1. Identified virulence factors of Burkholderia pseudomallei
The virulence of an individual strain is further complicated by the ability of B. pseudomallei to alter surface determinants and change its observed colony morphology. Seven distinct morphotypes have been described, although the characteristic wrinkled type I morphotype predominates (Chantratita et al., 2007). Morphotype switching can be induced by a range of in vitro stresses, including starvation, heat shock, iron limitation and subinhibitory antibiotic concentrations. Phenotypic differences, including changes in biofilm formation, secreted enzymes and motility, were observed between morphotypes. These differences affected intracellular survival in epithelial and macrophage cells. Furthermore, differential lethality and persistence were seen in BALB/c mice, suggesting that morphotype switching in vivo provides a strong survival advantage for B. pseudomallei (Chantratita et al., 2007).
The first polysaccharide molecule (designated type I O-PS) is an unbranched high-molecular-weight polymer of 1,3-linked 2-O-acetyl-6-deoxy-β-d-manno-heptopyranose (Knirel et al., 1992; Perry et al., 1995). Although initially characterized as one of the O-PS of lipopolysaccharide, this polysaccharide was independently reclassified as a capsular polysaccharide by an identification of the biosynthetic locus (Reckseidler et al., 2001) and the absence of lipid A (Isshiki et al., 2001). Studies using signature-tagged mutagenesis (STM) identified a number of genes located within the type I O-PS biosynthetic locus including wcbN, wcbC, wzm2, wcbQ and wcbB; the inactivation of any of these resulted in reduced survival in mice (Atkins et al., 2002; Cuccui et al., 2007; Lazar Adler, 2009). The wcbB gene encodes the mannosyltransferase that is required for type I capsule assembly; two directed mutants were severely attenuated in the Syrian hamster model, displaying significantly reduced growth in the blood (Reckseidler et al., 2001; Reckseidler-Zenteno et al., 2005). Moreover, the wcbB mutants were more sensitive to killing by normal human serum, which was reversed by the addition of a purified capsule. Western blot analysis showed more C3b deposition on the acapsular mutants than the wild-type B. pseudomallei (Reckseidler-Zenteno et al., 2005). These data indicate that expression of the type I O-PS capsule in vivo helps B. pseudomallei to resist phagocytosis by reducing C3b deposition on the surface of the bacteria.
The genes involved in the biosynthesis of type I O-PS have been identified, although two additional putative capsule operons have been identified on the K96243 genome: BPSS0417–BPSS0429 and BPSS1825–BPSS1832 have been designated as involved in type III O-PS and type IV O-PS biosynthesis, respectively. While B. pseudomallei mutants have been constructed with disruptions in each of these clusters, structural analyses have not been completed to determine the precise polysaccharide synthesized by each cluster. These clusters are absent in Burkholderia mallei, but partially conserved in the environmental species B. thailandensis, suggesting an environmental role. A B. pseudomallei type III O-PS mutant was attenuated for virulence in BALB/c mice whereas a type IV O-PS mutant displayed wild-type virulence, but a delay in time to death (Sarkar-Tyson et al., 2007). Thus, it is clear that at least two of the B. pseudomallei polysaccharide capsules have a role in virulence, with the type I O-PS demonstrated to be involved in serum survival and resistance to phagocytosis. The exact role of type III O-PS and type IV O-PS in disease remains unknown. Furthermore, the condition(s) under which any given polysaccharide capsule is present as well as the regulatory mechanisms controlling capsule expression still requires characterization.
A confluent biofilm, comprising bacteria encased in a carbohydrate-based fibrous matrix, has been observed in electron micrographs of B. pseudomallei-infected lung tissue from guinea pigs and a human patient (Vorachit et al., 1995). A study of 50 B. pseudomallei strains confirmed significant variability in biofilm production between strains. However, there was no correlation between biofilm formation and virulence in the BALB/c mouse melioidosis model. Furthermore, two transposon mutants deficient for biofilm production (polysaccharide biosynthesis and sugar transferase genes) were not attenuated for virulence (Taweechaisupapong et al., 2005). Biofilm production may be controlled, at least in part, by the alternative σ factor RpoE, as rpoE mutants demonstrate a 50% reduction in biofilm formation. Electron microscopy indicated that the rpoE mutants were found in chains rather than aggregated clusters as observed for the wild-type B. pseudomallei (Korbsrisate et al., 2005). Therefore, biofilm formation does not appear to be essential for virulence, although it is likely to have an important role in persistence in harsh environments, thus allowing survival for later infection.
Characterization of B. pseudomallei lipopolysaccharide demonstrated unique, acid-stable structures between the inner core and lipid A linkage, and longer amide-linked fatty acids (3-hydroxypalmitic acids) (Kawahara et al., 1992). The O-antigenic polysaccharide was identified as an unbranched polymer of the repeating disaccharide unit: -3)-β-d-glucopyranose-(1-3)-6-deoxy-α-l-talopyranose-(1- (Knirel et al., 1992; Perry et al., 1995). The talose residue can be decorated with both 2-O-methyl and 4-O-acetyl groups or 2-O-acetyl groups alone (Perry et al., 1995). The inner membrane trans-acetylase, WbiA, has been shown to be responsible for the addition of the 2-O-acetyl substituents (Brett et al., 2003).
Three distinct antigenic types of lipopolysaccharide have been reported, two smooth lipopolysaccharide serotypes A and B, and a rare rough serotype (Anuntagool et al., 2000, 2006). No immunological cross reactivity was seen between any of the types, but they shared similar endotoxic levels in the Limulus amoebocyte lysate assay and in the levels of macrophage activation (Anuntagool et al., 2000). The predominant smooth serotype A accounts for 97% of strains, while clinical isolates belonging to the other two serotypes have been associated with cases of clinical relapse (Anuntagool et al., 2006). The low-level presence of unusual serotypes, especially in relapse patients, may occur due to suboptimal antibiotic treatment. Such cases demonstrate the adaptability of B. pseudomallei, particularly when changes occur in common antibiotic targets such as lipopolysaccharide. Thus, these strains should be considered as exceptions. The minimal pyrogenic lethal toxicity and macrophage activation doses of B. pseudomallei lipopolysaccharide were less than those of the enterobacterial lipopolysaccharide, which is unsurprising as the B. pseudomallei lipopolysaccharide contains longer fatty acids (Matsuura et al., 1996). These longer fatty acids are predicted to decrease the level of interaction of B. pseudomallei lipopolysaccharide with CD14 on the macrophage cell surface (Utaisincharoen et al., 2000), resulting in a reduced inflammatory response.
A genetic locus involved in lipopolysaccharide biosynthesis was identified by transposon mutagenesis of B. pseudomallei 1026b. The lipopolysaccharide-deficient wbiI (dehydratase gene) mutant was attenuated in hamsters, guinea pigs and diabetic rats, and was susceptible to complement-mediated killing by the alternative pathway (DeShazer et al., 1998). This mutant demonstrated increased internalization by RAW 264.7 macrophage cells, but displayed decreased intracellular survival between 2- and 6-h postinfection. The absence of wild-type lipopolysaccharide resulted in IFN-β stimulation, leading to iNOS expression and subsequent bactericidal activity. This phenotype could not be reconstituted with exogenous lipopolysaccharide, nor did the mutant display increased susceptibility to NO killing in vitro (Arjcharoen et al., 2007). A role for lipopolysaccharide in serum survival was further implicated by a transposon mutagenesis screen for polymyxin-B (cationic peptide) sensitivity. Burkholderia pseudomallei mutants with transposon insertions within waaF, involved in lipopolysaccharide core biosynthesis, and lytB, a regulator of peptidoglycan and phospholipid synthesis, were identified as polymyxin-B sensitive (Burtnick & Woods, 1999). Therefore, lipopolysaccharides are likely to play a role in B. pseudomallei resistance to host cationic antimicrobial peptides and complement-mediated killing.
Flagella and pili
Electron microscopy studies have demonstrated the presence of flagella and the variable expression of pili on B. pseudomallei (Vorachit et al., 1995). A mutagenesis screen for genes involved in motility identified 19 unique genetic loci (DeShazer et al., 1997), while a bioinformatic analysis of the K96243 genome identified 13 gene clusters predicted to be involved in the synthesis of type I fimbriae, type IV pili and Tad-like pili (Holden et al., 2004).
A polar tuft of two to four flagella confers temperature-independent motility on B. pseudomallei. Flagella synthesis requires the fliC gene, which encodes a 39.1-kDa flagellum protein (DeShazer et al., 1997). Polyclonal antiserum against FliC was able to inhibit motility in all but one of the 65 B. pseudomallei strains tested (Brett et al., 1994). There are conflicting data on the importance of flagella in virulence; a fliC B. pseudomallei 1026b transposon mutant was not attenuated via the intraperitoneal route in the diabetic rat or Syrian hamster melioidosis models (DeShazer et al., 1997). However, unlike wild-type B. pseudomallei, the fliC mutant was unable to adhere to cells of the free-living amoeba A. astronyxis, a critical step for efficient invasion of this organism (Inglis et al., 2003). Furthermore, an additional study showed that a B. pseudomallei KHW fliC mutant was attenuated in BALB/c mice infected by either the intranasal or intraperitoneal routes, although this mutant showed no significant difference in invasion of lung epithelial (A549) cells (Chua et al., 2003). These data indicate that flagella have an important role in virulence, but it is possible that this role can be overcome or subverted in more acute infection models, such as diabetic rats or hamsters.
The K96243 genome encodes eight type IV pili-associated loci. However, only one type IV A pilin gene, pilA, is present on the genome. A B. pseudomallei strain K96243 pilA mutant displayed reduced adhesion to epithelial cell lines. Furthermore, reduced virulence was seen in BALB/c mice infected via the intranasal route but not the intraperitoneal route, indicating a role in initial epithelial cell attachment (Essex-Lopresti et al., 2005). In studies on two B. pseudomallei pilA mutants constructed in strains 08 and K96243, the pilA expression, microcolony formation and cell adhesion varied considerably between the strains. However, an analysis of the strain 08 mutant revealed that the expression of pilA was temperature regulated and essential for microcolony formation, but was not required for adhesion to cultured human cells (Boddey et al., 2006). The type IV pilus assembly system has been proposed to be involved in the natural competency of B. pseudomallei (Thongdee et al., 2008), but no direct role has been shown.
QS is a population density-mediated form of cell–cell communication via the production, release and detection of signalling molecules such as N-acyl-homoserine lactones (AHLs). The B. pseudomallei K96243 genome has three luxI homologues, which encode the AHL synthase proteins, and five luxR homologues, which encode transcriptional regulators. These regulators become activated upon binding their cognate AHL and subsequently mediate transcription of QS-regulated genes (Ulrich et al., 2004). Seven AHLs have been detected in the B. pseudomallei supernatant from various strains (Table 2). Supernatants from a luxI (BPSS0885) and a luxR (BPSS0887) mutant have been analysed and found to be lacking two AHLs, viz. C8-HSL and 3-oxo-C8-HSL. In addition, expression of BPSS0885 in Escherichia coli resulted in the production of C8-HSL, confirming that this AHL is synthesized by BPSS0885 (Lumjiaktase et al., 2006). The BPSS0885 luxI gene was found to be activated by the BPSS0887 luxR in the presence of C8-HSL, and to a lesser extent, 3-oxo-C8-HSL and C10-HSL. Partial activation of BPSS0885 by BPSL2347 was also observed irrespective of the presence of AHLs. Both the BPSS1180 and BPSS1570 luxI genes were constitutively expressed, but the expression of BPSS1180 was increased by all of the luxR genes while BPSS1570 expression was repressed (Kiratisin & Sanmee, 2008).
Table 2. AHL production by Burkholderia pseudomallei
Burkholderia pseudomallei mutants with inactivated luxI genes (BPSS0885, BPSS1180 and BPSS1570) or luxR genes (BPSS0887, BPSS1176, BPSS1569 and BPSL2347) have been assessed for virulence. All mutants displayed reduced colonization in the BALB/c mouse model, and mice infected with these mutants survived longer than mice infected with the wild-type strain (between 1 and 3 days, with the exception of the BPSS1570 mutant, which demonstrated 70% survival at day 39). Furthermore, all mutants showed an increased lethal dose 50% (LD50) in the Syrian hamster melioidosis model (Ulrich et al., 2004) and an independently constructed BPSS0885 mutant was attenuated for virulence in the Swiss mouse model (Valade et al., 2004).
The B. pseudomallei luxI (BPSS0885) and luxR (BPSS0887) mutants demonstrated an overproduction of one or more uncharacterized metalloprotease(s) and siderophore(s) and showed reduced expression of one or more uncharacterized phospholipase(s) (Valade et al., 2004; Song et al., 2005). The oxidative stress protein, DspA, is dependent on both BPSS0887 and C8-HSL for its expression during the late exponential phase. The BPSS0885 and BPSS0887 mutants showed increased sensitivity to hydrogen peroxide; this phenotype could be complemented by the expression of dspA in trans (Lumjiaktase et al., 2006). The full range of QS-regulated genes has yet to be determined.
Extracellular secretion of B. pseudomallei QS AHLs is dependent on the BpeAB-OprB multidrug efflux system and its negative regulator BpeR (Chan et al., 2007). Expression of the bpeAB-oprB genes is induced during the stationary phase, and can also be induced in vitro by the addition of C8-HSL or C10-HSL. Studies using B. pseudomallei bpeAB mutants, or strains overexpressing bpeR, showed that a functional BpeAB-OprB efflux pump is essential for biofilm formation and optimal production of siderophores and phospholipase C. Furthermore, mutants lacking this system displayed reduced levels of invasion and cytotoxicity for both lung epithelial and macrophage cell lines; this reduced level of invasion was partially restored by addition of the AHL, C8-HSL (Chan & Chua, 2005).
Burkholderia pseudomallei has a second QS system involving the production, release and detection of 4-hydroxy-3-methyl-2-alkylquinolone (HMAQ) signalling molecules. Three families of HMAQ molecules have been identified; these differ in the presence of saturated or unsaturated alkyl chains at the 2′ position of an N-oxide group (Vial et al., 2008). Based on the homology to P. aeruginosa HMAQ biosynthetic genes, a putative biosynthesis operon was identified on the K96243 genome and designated hmqABCDEFG (BPSS0481–BPSS0487) (Diggle et al., 2006; Vial et al., 2008). A B. pseudomallei hmqA mutant was unable to synthesize HMAQ and displayed altered colony morphology. Moreover, this mutant produced increased levels of elastase, which could be restored to wild-type levels by the addition of exogenous HMAQ (Diggle et al., 2006). A role in AHL regulation has been demonstrated for HMAQ in Burkholderia ambifaria (Vial et al., 2008), but whether the B. pseudomallei HMAQ has a similar function remains to be determined.
The T3SS is a secretion apparatus which, when triggered by a close contact with host cells, translocates effector proteins into host cells. Three T3SS operons have been identified on the B. pseudomallei K96243 genome (Attree & Attree, 2001; Holden et al., 2004). T3SS1 and T3SS2 are plant pathogen-like systems similar to the T3SS from Ralstonia solanacearum (Winstanley et al., 1999; Attree & Attree, 2001; Rainbow et al., 2002). The role of these secretion systems in B. pseudomallei is unclear, but they may be involved in symbiotic or pathogenic bacterium–plant interactions during growth in the soil (Attree & Attree, 2001). Burkholderia pseudomallei T3SS1 or T3SS2 mutants were not altered for virulence in the hamster model. However, a triple mutant, containing a disrupted gene within each T3SS system demonstrated higher attenuation than a T3SS3 mutant alone, suggesting an additive effect of T3SS1 and T3SS2 (Warawa & Woods, 2005).
The B. pseudomallei T3SS3 is similar to the T3SS of the human pathogens Salmonella and Shigella (Attree & Attree, 2001). The T3SS3 locus, designated Bsa, contains genes that encode proteins predicted to be required for the synthesis of both the secretion apparatus and the effector proteins (Stevens et al., 2002). Three T3SS3 secretion apparatus genes, bsaQ (Sun et al., 2005), bsaU (Pilatz et al., 2006) and bsaZ (Stevens et al., 2002; Warawa & Woods, 2005) have been inactivated in B. pseudomallei (Table 3). A B. pseudomallei bsaQ mutant constructed in K96243 displayed reduced invasion of A549 human epithelial cells (Muangsombut et al., 2008). However, a bsaQ mutant constructed in strain KHW was shown to invade and replicate normally in HEK293T human embryonic kidney cells, but, interestingly, failed to induce an appropriate interleukin-8 (IL-8) and nuclear factor (NF)-κB response (Hii et al., 2008). This same mutant showed a loss of cytotoxic activity against macrophage-like cell lines (Sun et al., 2005). The bsaU mutant displayed a reduced LD50 via the intranasal route in the BALB/c mouse model and infected mice showed a decreased bacterial load in the spleen, liver and lungs (Pilatz et al., 2006). The bsaU mutant was unable to escape normally from endocytic vesicles, but those mutants that were released into the cytoplasm late in infection were still capable of intracellular growth and actin-mediated motility. A B. pseudomallei bsaZ mutant was unable to escape from endocytic vacuoles, replicate or form actin tails in the first 8 h after infection of J774.2 macrophage cells (Stevens et al., 2002). Another independently constructed bsaZ mutant was attenuated for virulence in syrian hamsters and displayed delays in vacuolar escape, actin-mediated intracellular motility, MNGC formation and had reduced cytotoxicity (Warawa & Woods, 2005; Burtnick et al., 2008).
Convalescent serum from melioidosis patients has been shown to react with the purified T3SS3 translocation proteins BipB, BipC and BipD, indicating a functional expression of this apparatus in vivo (Stevens et al., 2002). However, mAbs against BipB and BipD were unable to detect these proteins from B. pseudomallei lysates grown under a range of in vitro conditions (Druar et al., 2008). Two T3SS3 translocation apparatus genes, bipB (Suparak et al., 2005) and bipD (Stevens et al., 2004), have been inactivated in B. pseudomallei (Table 3). The bipB mutant was attenuated in BALB/c mice when introduced via the intranasal route. This mutant also displayed reduced MNGC formation, cell-to-cell spreading and failed to induce significant levels of apoptosis in J774A.1 cells. Complementation with a functional copy of bipB restored each of these phenotypes to near wild-type levels (Suparak et al., 2005). The bipD mutant was also attenuated in BALB/c mice when introduced via the intraperitoneal or intranasal routes, with reduced bacterial loads observed in the spleens and livers of infected mice (Stevens et al., 2004). The bipD mutant was also unable to escape from endocytic vacuoles, replicate or form actin tails within J774.2 cells (Stevens et al., 2002).
Despite numerous studies, the precise role of the T3SS3 operon remains unclear. Four mutants within the T3SS3 apparatus were attenuated for virulence, clearly demonstrating a role for the T3SS3 in pathogenesis (Stevens et al., 2004; Suparak et al., 2005; Warawa & Woods, 2005; Pilatz et al., 2006). However, it is not clear whether this reduced virulence is specifically due to the defect observed in vacuolar escape and its downstream effects or due to some additional, as yet undefined, role in pathogenesis. A recent study demonstrated that the vacuolar escape defect in the bsaZ T3SS3 mutant is a delay rather than a complete abrogation (Burtnick et al., 2008). Thus, the T3SS3 may have additional roles in virulence, or the delayed vacuolar escape may allow sufficient time for the host immune system to control the B. pseudomallei infection.
A new secretion system, designated T6SS, has recently been identified in members of the Proteobacteria (Filloux et al., 2008). Orthologues of this system have been identified in both animal and plant pathogens. Mutation of T6SS genes reduced virulence in P. aeruginosa and reduced invasion in Salmonella enterica. Six T6SS clusters are present on the B. pseudomallei K96243 genome (Holden et al., 2004). Expression of three genes within one of these T6SS clusters was induced following macrophage invasion; however, a mutant in one of these genes displayed wild-type levels of RAW macrophage cell invasion and intracellular survival (Shalom et al., 2007). The role of the different B. pseudomallei T6SS in pathogenesis thus warrants further investigation.
Burkholderia pseudomallei secretes a number of biologically active molecules, including proteases, lipases, lecithinases, haemolysins and siderophores (Ashdown & Koehler, 1990). The cell-free B. pseudomallei supernatant has cytotoxic effects on a variety of eukaryotic cell types (Ismail et al., 1987; Haase et al., 1997; Balaji et al., 2004). Lethal toxicity of the cell-free supernatant was initially demonstrated in mice and guinea pigs (Ismail et al., 1987), but subsequent researchers have been unable to reproduce this finding (Brett & Woods, 2000). Secretion of protease, lipase and phospholipase occurs via the type II general secretory pathway (Gsp); gsp mutants lack secretion, but are not attenuated for virulence in hamsters (DeShazer et al., 1999). Furthermore, no correlation has been observed between protease production in six B. pseudomallei strains and virulence in BALB/c mice (Gauthier et al., 2000). These data suggest that the exoproducts, including protease, lipase and phospholipase play, at most, a minor role in virulence.
A B. pseudomallei serine metalloprotease has been characterized; MprA is a 47-kDa protein present in all 68 B. pseudomallei isolates screened (Lee & Liu, 2000). The activity of MprA increased in the stationary phase, and its expression was negatively regulated by QS molecules. However, an mprA mutant was not attenuated in Swiss mice, suggesting that it is not required for infection in mice (Valade et al., 2004). A smaller 36-kDa zinc metalloprotease has been purified from B. pseudomallei (Sexton et al., 1994). Antibodies against the 36-kDa zinc metalloprotease also reacted with a 50-kDa protein in B. pseudomallei, which led to the suggestion that the 36-kDa zinc metalloprotease is a processed version of MprA (Lee & Liu, 2000).
An additional 52-kDa B. pseudomallei calcium-dependent serine protease has also been identified. Injection of the purified protein into guinea pigs (Tumwasorn et al., 1994) and rabbits (Tumwasorn et al., 1994; Ling et al., 2001) resulted in localized tissue necrosis. Finally, a third 65-kDa protease was identified in B. pseudomallei and was found to exhibit collagenase activity when expressed in E. coli (Rainbow et al., 2004). The role of these proteases in the pathogenesis of melioidosis in humans is presently unclear.
A cell-surface glycoprotein acid phosphatase of B. pseudomallei demonstrated substrate activity for phosphorylated tyrosine, and is recognized by melioidosis patient sera. This putative tyrosine phosphatase may be a component of a signal transduction system (Kondo et al., 1996). High phosphatase activity was found in B. pseudomallei culture filtrates, with activity observed across three distinct pH ranges; this may indicate the presence of multiple phosphatases or a single phosphatase with multiple components (Kondo et al., 1991).
The K96243 genome encodes three phospholipase C (PLC) enzymes. PLC-1 (BPSL2403) and PLC-2 (BPSL0338) are predicted to be acidic, 77-kDa proteins that hydrolyse lipids, phosphatidylcholine (PC) and sphingomyelin. Mutants of plc-1, plc-2 and a double plc-1 plc-2 mutant displayed reduced PC-PLC activity, demonstrating that these genes encode functional enzymes. Furthermore, the presence of some remaining PC-PLC activity indicated that PLC-3 (BPSS0067), a predicted basic 81-kDa protein, is also likely to be functional. The replication of the B. pseudomallei plc-1 plc-2 double mutant was reduced following starvation, suggesting a role for PLC in nutrient acquisition. The plc-2 mutant and the plc-1 plc-2 double mutant demonstrated reduced plaque formation in HeLa cells and a decreased cytotoxicity in RAW 264.7 macrophage cells (Korbsrisate et al., 2007). Furthermore, a plc-3 mutant was attenuated in hamsters (Tuanyok et al., 2006). DNA microarray analysis of B. pseudomallei isolated from infected hamsters showed that the plc-3 gene was highly upregulated compared with its expression in in vitro-grown bacteria. Further evidence of the in vivo expression of PLC enzymes was demonstrated by the presence of antibodies in melioidosis patient sera against a B. thailandensis PLC-1 orthologue (Korbsrisate et al., 1999).
A soluble hydroxamate siderophore, designated malleobactin, is expressed by B. pseudomallei during growth under iron-deficient conditions. This siderophore has been shown to remove iron from transferrin and EDTA (Yang et al., 1991). Malleobactin preferentially releases iron from transferrin, but can also remove iron from lactoferrin, and less efficiently from erythrocytes. This siderophore-dependent activity was observed in vitro even in the absence of bacteria (Yang et al., 1993). Malleobactin consists of three components of 636, 762 and 790 Da in size. The operon BPSL1774–BPSL1787 contains five genes predicted to be involved in ferric malleobactin (MbaA) biosynthesis and transport. Expression of these genes is controlled by the extracyctoplasmic function σ factor MbaS and is upregulated under iron-limiting conditions. An mbaA mutant was unable to grow under iron-limiting conditions and this defect could be complemented by the addition of purified malleobactin or Burkholderia cenocepacia MbaA-like ornibactins (Alice et al., 2006).
Burkholderia pseudomallei produces two haemolysins. The first is expressed by most strains, but displays only weak haemolytic activity that can be observed on blood plates only in areas of heavy growth. The second, seen only in some strains, is a heat-labile haemolysin that produces clear zones of haemolysis on blood agar plates around individual colonies and in broth, and has an optimal pH of 5.5 (Ashdown & Koehler, 1990). This haemolysin was putatively identified as a 762-Da rhamnolipid (Rha-Rha-C14-C14); the purified product demonstrated cytotoxic and haemolytic activity (Haussler et al., 1998). Additionally, the K96243 genome encodes three ATP-binding cassette transport systems predicted to be involved in the export of haemolysins (Harland et al., 2007).
Genes involved in the regulation of B. pseudomallei virulence factors
σ factors are transcriptional regulators that activate gene expression in response to particular environmental conditions; the B. pseudomallei K96243 genome encodes 17 σ factors (Holden et al., 2004). RpoS is involved in the response to nutrient limitation upon entry into the stationary phase; an rpoS mutant had reduced tolerance to starvation, which was restored by complementation. Furthermore, rpoS gene expression was increased in the exponential phase, reaching a peak during the stationary phase. The rpoS mutant was also more sensitive to peroxide and prolonged acid shock (Subsin et al., 2003). While the rpoS mutant demonstrated a wild-type invasion of epithelial and macrophage cell lines (Subsin et al., 2003), it failed to repress iNOS expression and induce MNGC formation and apoptosis (Utaisincharoen et al., 2006; Lengwehasatit et al., 2008). These phenotypes suggest that in the absence of RpoS-regulated genes, B. pseudomallei is unable to survive and replicate to numbers required for intracellular and extracellular spread via MNGC formation and cell lysis.
RpoE regulates the expression of proteins responsible for maintaining the integrity of the cell envelope under environmental stresses. An rpoE mutant demonstrated increased susceptibility to oxidative, osmotic and heat stress (Korbsrisate et al., 2005; Vanaporn et al., 2008). Proteomics studies demonstrated that the rpoE mutant displayed reduced expression of a number of stress response proteins and chaperones, as well as transcriptional regulators and proteins involved in cell wall synthesis (Thongboonkerd et al., 2007). The inability of the rpoE mutant to repair cell wall damage is probably responsible for the decreased intracellular survival of this mutant in macrophages and the observed attenuation in BALB/c mice (Korbsrisate et al., 2005; Thongboonkerd et al., 2007). However, the direct regulation of unknown virulence factors by RpoE cannot be completely excluded.
Analysis of the B. pseudomallei K96243 genome using MiST, a signal transduction database (Ulrich & Zhulin, 2007), identified 66 sensor kinases and 69 response regulators. Only three of these systems have been investigated. The irlRS operon is involved in the invasion of epithelial cells, but not macrophage cells, and is not involved in virulence (Jones et al., 1997). The mrgRS locus is temperature regulated with a reduced expression at 25 °C compared with 37 or 42 °C (Mahfouz et al., 2006), suggesting an in vivo role. A sensor kinase mutant with a transposon insertion in BPSS0687 was attenuated for virulence in BALB/c mice but displayed wild-type invasion of macrophage cells (Lazar Adler, 2009). The genes regulated by these two-component systems remain uncharacterized and the role in virulence for the latter systems is unknown.
The role of the host response in the molecular pathogenesis of B. pseudomallei
While healthy individuals can contract melioidosis, most patients have an underlying predisposition, suggesting that the immunological status of the patient affects the disease initiation and progression. In particular, diabetes mellitus and renal disease are common underlying conditions in melioidosis patients; other factors which result in immune suppression, such as alcohol abuse, have also been identified as risk factors (White, 2003). Melioidosis has several disease outcomes (asymptomatic, acute, chronic or latent), which are believed to be determined by the host immune response (Gan, 2005). The murine melioidosis models of acute (BALB/c) and chronic (C57BL/6) infections mimic the acute and chronic disease in humans (Leakey et al., 1998) (Table 4). The acute melioidosis observed in BALB/c mice is characterized by a significantly stronger innate immune response (Hoppe et al., 1999; Ulett et al., 2000a, b). However, hyperproduction of proinflammatory cytokines results in an inappropriate cellular response that fails to control the infection and contributes to tissue destruction and multiple organ failure. In contrast, the chronic infection observed in C57BL/6 mice demonstrates a moderate cytokine increase that allows the mice to temporarily confine B. pseudomallei within phagocytes, allowing time for an adaptive immune response to occur (Hoppe et al., 1999; Ulett et al., 2000a, b; Barnes et al., 2001). To date, however, the relative importance of the cell-mediated and the humoral arms of both the innate and the adaptive immune responses remains unclear (Cheng & Currie, 2005; Gan, 2005).
Table 4. Differences between the BALB/c and C57BL/6 murine melioidosis models
Interaction of B. pseudomallei with the innate immune response
Burkholderia pseudomallei activates the alternative complement pathway, but the membrane attack complex is deposited on an external polysaccharide and hence is not bactericidal (Egan & Gordon, 1996). Opsonization with a complement enhances, but is not essential for, uptake by phagocytes and does not necessarily result in intracellular killing of the bacteria (Harley et al., 1998; Kespichayawattana et al., 2000). Resistance of B. pseudomallei to lysosomal defensins and cationic peptides has been demonstrated (Gan, 2005). These resistance mechanisms, attributed to the presence of the capsule and lipopolysaccharide, allow B. pseudomallei to survive within phagocytes and in human serum.
Following infection with B. pseudomallei, mouse tissue shows a rapid influx and activation of neutrophils (Barnes et al., 2001). When C57BL/6 mice are depleted of neutrophils, an acute B. pseudomallei infection is established (Easton et al., 2007), indicating the importance of neutrophils in innate immunity. However, macrophages are also essential for the control of B. pseudomallei infection. Depletion of macrophages from BALB/c or C57BL/6 mice significantly increases mortality rates (Breitbach et al., 2006; Barnes et al., 2008). Burkholderia pseudomallei infection in BALB/c mice fails to attract macrophages to the same extent as in C57BL/6 mice. It has been proposed that macrophages recruited by C57BL/6 mice may temporarily contain B. pseudomallei, resulting in the chronic melioidosis seen in these animals (Barnes et al., 2001). Macrophages from melioidosis patients demonstrate a reduced level of lysosomal fusion compared with healthy individuals, resulting in higher bacterial numbers (Puthucheary & Nathan, 2006). These data suggest that acute melioidosis results from an ineffective cellular innate immune response.
Toll-like receptors (TLRs) recognize conserved pathogen-associated molecular patterns and mediate an inflammatory immune response. Activation of TLRs occurs via various signalling adaptor proteins, including MyD88 and TRIF. MyD88 knockout mice demonstrated increased susceptibility to B. pseudomallei infection as a result of reduced neutrophil recruitment and activation (Wiersinga et al., 2008c). Melioidosis patients suffering septic shock have increased expression of TLR1, TLR2 and TLR4 and its coreceptor CD14; the expression of TLR2, TLR4 and CD14 was decreased upon recovery (Wiersinga et al., 2007a). Burkholderia pseudomallei triggers TLR2, TLR4 and TLR5 receptors in epithelial reporter cell lines and induces IL-8 production via NF-κB (Hii et al., 2008). The interaction of B. pseudomallei with TLR2, TLR4 and CD14 was confirmed by a reduced tumour necrosis factor (TNF)-α expression in leucocytes from knockout mice. Following B. pseudomallei infection, TLR4 knockout mice demonstrated wild-type mortality, whereas both TLR2 and CD14 knockout mice demonstrated reduced mortality, bacterial loads and inflammation, as assessed histologically and by the measurement of cytokine levels. Purified B. pseudomallei lipopolysaccharide was found to signal via TLR2 (Wiersinga et al., 2007a, 2008b). Thus, these data confirm that B. pseudomallei lipopolysaccharide-mediated proinflammatory cytokine release contributes to disease pathology and results in acute melioidosis (Fig. 1).
Interaction of B. pseudomallei with the humoral immune response
For people living within melioidosis endemic areas, antibodies to B. pseudomallei are common, although the percentages of seropositive individuals varies significantly between regions and subpopulations (Bryan et al., 1994; Barnes et al., 2004). This variability may be due to B. pseudomallei antigens cross-reacting with related, avirulent Burkholderiaceae species (Cheng & Currie, 2005; Gilmore et al., 2007). The role of antibodies in protection from infection is equivocal. A screen of antibodies in melioidosis patients identified antilipopolysaccharide antibodies as protective (Charuchaimontri et al., 1999). However, Ho et al. (1997) found no correlation between disease severity or survival, and antibodies against the capsule or lipopolysaccharide, despite demonstrating that these antibodies mediated phagocyte killing in vitro. Notably, recurrent infections can occur in the presence of high antibody levels (Vasu et al., 2003).
Interaction of B. pseudomallei with the cellular immune response
Once intracellular invasion by B. pseudomallei has occurred, a cell-mediated immune (CMI) response, in which T cells play an important role, is required for bacterial clearance. However, melioidosis patients demonstrate reduced T-cell counts (Ramsay et al., 2002). Following stimulation with B. pseudomallei lysate, T cells from patients with subclinical melioidosis demonstrated higher proliferation levels as well as higher IFN-γ production than those from patients with clinical melioidosis. These data suggest that a strong CMI response is essential for protection against the progression of B. pseudomallei infection (Barnes et al., 2004).
CMI responses are seen in BALB/c and C57BL/6 mice following immunization with B. pseudomallei. Lymphocyte transfer to naïve mice transferred this CMI response, but no protection was seen against a subsequent challenge (Barnes & Ketheesan, 2007). Depletion of CD4 T cells, but not CD8 T cells, from BALB/c mice immunized with B. pseudomallei, resulted in increased susceptibility to infection (Haque et al., 2006). CD4 T cells are essential for B-cell isotype switching and activation of CD8 cells, as well as for the activation of phagocytes; thus, these data suggest that a comprehensive cellular response is required to control B. pseudomallei infection. Furthermore, optimal bactericidal activity against B. pseudomallei was observed only when both lymphocytes and macrophages were present (Ulett et al., 1998). A sublethal, chronic infection, suggesting containment of B. pseudomallei infection, was observed in Taylor outbred mice, whose cellular response involved an initial neutrophil infiltration followed by a macrophage and lymphocyte influx (Santanirand et al., 1999).
The role of cytokine responses in pathogenesis
Cytokines play an important role in regulating the immune response to B. pseudomallei infection. However, in acute disease, these regulatory mechanisms fail, resulting in excessive inflammation (Gan, 2005). Patients with acute melioidosis produce elevated levels of proinflammatory cytokines (IL-6, IL-12, IL-15, IL-18, TNF-α and IFN-γ); serum levels of a number of these have been shown to be significantly higher in cases of fatal melioidosis (Lauw et al., 1999; Simpson et al., 2000; Wiersinga et al., 2007a). Indeed, a high serum level of either IL-6 or IL-18 is considered a mortality predictor (Simpson et al., 2000; Wiersinga et al., 2007a). Therefore, the immune response of the host contributes significantly to the pathogenesis of melioidosis.
The use of the murine model of melioidosis has provided a detailed picture of the cytokine responses to B. pseudomallei. Increased levels of proinflammatory cytokines are observed in the acute melioidosis BALB/c mouse model, whereas moderately increased levels of proinflammatory cytokines are observed in the C57BL/6 chronic melioidosis mouse model (Ulett et al., 2000a; Koo & Gan, 2006; Tan et al., 2008). Furthermore, analysis of the cytokine expression kinetics showed that cytokines peak earlier in BALB/c mice (24–48 h) than C57BL/6 mice (48–72 h) (Ulett et al., 2000a). Cytokine levels tend to correlate with bacterial numbers rather than with the virulence of the B. pseudomallei strain. Thus, the high cytokine levels observed in BALB/c mice are a direct consequence of the higher bacterial loads observed in these animals (Ulett et al., 2002).
IFN-γ has been shown to be essential for an innate immune response against B. pseudomallei in the melioidosis mouse model. Burkholderia pseudomallei stimulates IFN-γ production, which then activates T cells and natural killer (NK) cells, perpetuating the CMI response (Lauw et al., 2000). Administration of anti-IFN-γ antibodies resulted in acute melioidosis in C57BL/6 mice at normally sublethal doses due to increased bacterial loads (Breitbach et al., 2006; Easton et al., 2007). IFN-γ knockout mice were extremely susceptible to B. pseudomallei infection, while lymphocyte-deficient mice had an intermediate resistance, highlighting the importance of NK cell-derived IFN-γ (Easton et al., 2007). NK cells are detectable at the site of infection within 5 h of B. pseudomallei infection and produce 60–80% of the secreted IFN-γ (Lertmemongkolchai et al., 2001; Easton et al., 2007). IFN-γ hyperproduction in BALB/c mice may relate to differences in receptors found on T cells and NK cells in BALB/c mice compared with C57BL/6 mice (Koo & Gan, 2006).
Administration of antibodies against TNF-α also resulted in an increased susceptibility to B. pseudomallei infection, although not to the same level as anti-IFN-γ antibodies (Santanirand et al., 1999). Knockout mice for either TNF-α or its receptors also demonstrated an increased susceptibility to B. pseudomallei infection; these mice demonstrated increased neutrophil-based inflammatory influx and associated necrosis (Barnes et al., 2008). TNF-α, predominately produced by macrophages (Easton et al., 2007), was found to be expressed as a consequence of B. pseudomallei interaction with the macrophage cell surface (Ekchariyawat et al., 2007).
The future of melioidosis research
In 2004, the first annotated genome of a B. pseudomallei strain, a Thai clinical isolate K96243, was published (Holden et al., 2004). Currently, the National Center for Biotechnology Information lists a further 20 genome sequences, of which three are fully annotated: strains 1106a, 1710a (Thai clinical isolates) and 668 (an Australian clinical isolate) (http://www.ncbi.nlm.nih.gov/). This information has resulted in the publication of a number of bioinformatics analyses (Harland et al., 2007; Lim et al., 2007), and genomic and proteomic studies have followed (Ou et al., 2005; Rodrigues et al., 2006; Harding et al., 2007; Thongboonkerd et al., 2007; Wongtrakoongate et al., 2007). These large-scale studies generate a substantial amount of data for future studies and provide a global perspective on the pathogenesis of melioidosis. Given the multifaceted nature of virulence of B. pseudomallei, genomic- and proteomic-scale studies will provide a broader understanding of the complex pathways that allow infection, invasion and persistence.
With a wealth of genomic information and a diverse array of tools for molecular manipulation and analysis of B. pseudomallei, a picture of the molecular and cellular basis of pathogenesis is beginning to emerge. However, almost a century after the initial discovery of melioidosis by Whitmore, questions still remain at every point of the B. pseudomallei infection process. Future studies should address important questions such as the mechanisms by which B. pseudomallei can attach, invade and survive within epithelial and phagocytic cells. Furthermore, an understanding of how B. pseudomallei spreads to secondary sites, and how the bacterial interaction with the immune system results in different disease outcomes, is essential for the development of a much-needed vaccine.
The original work in the authors' laboratories was supported by the Australian Research Council and the National Health and Medical Research Council, Canberra, Australia.