Intracellular replication of the well-armed pathogen Burkholderia pseudomallei.

The Burkholderia genus contains a group of soil-dwelling Gram-negative organisms that are prevalent in warm and humid climates. Two species in particular are able to cause disease in animals, B. mallei primarily infects Equus spp. and B. pseudomallei (BPS), that is able to cause potentially life-threatening disease in humans. BPS is naturally resistant to many antibiotics and there is no vaccine available. Although not a specialised human pathogen, BPS possesses a large genome and many virulence traits that allow it to adapt and survive very successfully in the human host. Key to this survival is the ability of BPS to replicate intracellularly. In this review we highlight recent advances in our understanding of the intracellular survival of BPS, including how it overcomes host immune defenses and other challenges to establish its niche and then spread the infection. Knowledge of these mechanisms increases our capacity for therapeutic interventions against a well-armed foe.


Introduction
Burkholderia pseudomallei (BPS) is a motile, Gram-negative saprophyte that dwells in the soil and surface water of hot and humid countries, notably in South East Asia and Northern Australia. It is capable of infecting a wide variety of mammals and causes the emerging disease, melioidosis, in humans. Pathogenicity manifests as either a chronic pulmonary disease when particles of contaminated soil are inhaled, alternatively as life-threatening septicaemia if the bacteria are introduced into the blood stream through dermal puncture.
Humans are an accidental host and the lack of human-tohuman transmission precludes host-pathogen evolution. Yet, BPS is very successful at survival in vivo, in part due to genomic islands acquired through horizontal gene transfer from other bacteria [1], and in part from adaptation to environmental stress factors that confer dual protection against host immune mechanisms. BPS possesses a large genome for a bacterium of around 7.25 million bases spread over two chromosomes. Accordingly, it possesses numerous virulence factors and demonstrates remarkable plasticity in adapting to microenvironmental conditions. Several tools are available that help to identify virulence factors in the context of intracellular survival. A common approach is to compare the genome of BPS with a closely related avirulent species, B. thailandensis. A conceptually similar experiment is to compare virulent and attenuated clinical isolates derived from the same strain. Molecular based approaches such as PCR-based subtractive hybridisation have identified factors such as the membrane transporter, BPSL2033 that reduces the ability of BPS to replicate in macrophages [2]. More recently, the creation of initially small transposon-mediated signature-tagged mutant libraries in BPS [3] and now large scale libraries coupled with next generation sequencing [4] have opened new avenues to reveal the genes required by BPS to survive intracellularly.
In this review, we highlight recent reports relating to the mechanisms employed by BPS to survive and thrive inside host cells; summarised in Figure 1 and Table 1. Not only is this characteristic an essential requirement for pathogenicity, it also presents targets for the development of novel antimicrobial therapeutics against a hardy organism that naturally tolerates many existing antibiotics.
First contact with the host cell BPS is not an obligate intracellular pathogen, but if cellular defense mechanisms are overcome it offers several advantages over an extracellular lifestyle, such as the evasion of humoral immunity, reduced exposure to antibiotics, and the potential establishment of latency. Intracellular living is also essential for localised spread from the site of infection, and eventual systemic dissemination.
The initial infection of a host cell depends upon invasion and adhesion factors that have been well reviewed [5]. In brief, BPS has the capability to invade cells either passively in the case of uptake by phagocytes such as resident tissue macrophages and recruited neutrophils; or actively in the case of epithelial cell infection. An advantageous trait of BPS in this regard are its flagellae, which in wet soil or aquatic environments propel the bacilli against a concentration gradient toward nutrients that may otherwise be scarce; and in the mammalian host, enable penetration of protective mucous linings and surfactant to reach epithelial cells. Other factors such as the expression of Type IV pilli are well-recognized adhesion factors, and BPS also possesses two further proteins, BoaA and BoaB that promote attachment to epithelial cells [6]. Host cell factors also play a role, such as protease-activated receptor-1, expressed by various cell types, that promotes cell ingress and is associated with BPS growth and dissemination [7].
The benefits of intracellular growth seem at odds with the characteristic of BPS to induce cell death of the infected cell. BPS is cytotoxic to dendritic cells (DC), monocytes and macrophages within hours of infection, with an absolute requirement for internalisation, caspase-1 activation, and the Type Three Secretion System (T3SS) in what has been described as a pyroptotic cell death mechanism [8]. The BPS flagellar hook-associated protein also induces cytotoxicity [9]. This rapid cell death is strongly linked to the multiplicity of infection and cell type. For example, epithelial cells seem somewhat resistant to BPS-mediated killing. Therefore, a low level infection that includes infection of epithelial cells may contribute to chronic persistence in the host, in contrast to highly inflammatory acute infection. Concurrently, BPS inhibits mechanisms of cell death that may control the infection though bacterial killing; infected neutrophils, for example, normally undergo apoptosis and are subsequently engulfed and degraded by macrophages, destroying their     pseudomallei is held within a phagosome, where the T3SS is required for BPS escape. Escape occurs before maturation and killing. Resistance to oxidative stress response from the host, and production of the serine protease ecotin are also crucial steps for phagosome survival and subsequent escape. (III) BPS replication and dissemination to surrounding cells, and fusion of neighbouring cells into MNGC's. Once outside the phagosome, B. pseudomallei is able to replicate host actin is utilized for the polymerisation of elegantly structured Ena/VASP-like tails. These projections alongside the T6SS-1 result in successful dissemination to surrounding cells, resulting in the formation of multi-nucleated giant host cells (MNGC). Black text describes intracellular process of BPS, navy text lists genes important for intracellular survival are listed, references for which can be found in Table 1. Some of the host-combative responses are shown in red text, and the dashed-arrow indicates the connectivity between T3SS and T6SS. Actin tails are not structurally accurate. contents within a few hours, but this is delayed by BPS infection [10,11].

The host cell response to BPS infection
The engagement of host pattern recognition receptors during BPS infection have been well characterized, and it has been found that the toll-like receptor repertoire offer little protective benefit or worse, contribute to disease severity and mortality [12][13][14][15][16][17]. This may in part be due to the ability of BPS to resist various TLR-dependent innate immune killing mechanisms. One such mechanism, particularly in the murine model of infection, is the generation of reactive nitrogen species (RNS). In contrast to other Gram-negative bacteria, the LPS in the cell wall of BPS is a weak inducer of the enzyme, iNOS, partly due to its inability to stimulate IFN-b [18]. Accordingly, impairment of iNOS activity in a murine model of infection does not significantly alter killing capability [19]. A related mechanism is the respiratory burst, triggered by the engagement of pathogen associated molecular patterns (PAMP) by TLR. Reactive oxygen species (ROS) are able to denature enzymes as well as damage fatty acids and DNA, and direct exposure is lethal for BPS. While the effect of ROS-mediated killing of BPS in ex vivo isolated macrophages or macrophage-like cell lines is modest, in vivo studies have shown a significant role of ROS in host survival [19].
Unlike RNS, BPS is an inducer of ROS, perhaps unavoidably so. A key protein in BPS invasion of host cells is BopE, which acts as a guanine nucleotide exchange factor for RhoGTPases, promoting actin cytoskeleton rearrangement (reviewed by Allwood et al. [5]). However, RhoGTPases also stimulate NAPH oxidase [20], promoting the respiratory burst. Therefore, the ability of BPS to neutralise ROS using proteins such as superoxide dismutase is a key aspect of their virulence in vivo [21]. Whole genome analysis has further revealed that during oxidative stress, BPS upregulates genes involved in cell wall repair and biosynthesis, as well as the peroxide scavenging enzymes, KatG and AhpC. Also discovered was an oxidative stress regulon under the control of the sigma factor, RpoE. RpoE is conserved among Gramnegative bacteria, and deletion in BPS resulted in reduced intracellular survival in macrophages and virulence in the murine model. RpoE seems to be important for maintaining expression of spermidine acetyltransferase, speG, under oxidative stress. Spermidine is useful for the bacteria as it acts as a free radical scavenger but excessive levels are toxic, and so SpeG is required to prevent its accumulation to damaging levels [22,23].
Restricting the exposure of BPS to toxic levels of ROS in monocytes and macrophages is potentially realised through an additional mechanism. During the initial hours post-BPS infection, there is a reduction in CFU followed by growth recovery [24][25][26]; this temporally correlates with activation of NADPH-oxidase that causes rapid, but only transient ROS production. Although direct modulation of NADPH-oxidase by BPS has not been experimentally demonstrated, its activity, as well as a large component of TLR-mediated signalling, depends on the transcription factor, NfkB, which BPS has been shown to actively inhibit. By undermining the transcriptional response to TLR-activation, BPS is able to disarm a key aspect of host immunity. NfkB activity is by phosphorylation, and subsequent inactivation of, glycogen synthase kinase 3b (GSK3b), which usually supports NfkB activity. This is dysregulated in a rat model of diabetes, resulting in excessive TNF-a and IL12 cytokine production by peripheral blood mononuclear cells that may contribute to the higher risk of septic shock in diabetic patients with melioidosis [31]. Remarkably, BPS can alter the immune function of not only the infected host cell, but also uninfected bystander cells. BPS-infected neutrophils are able to locally inhibit CD4 + T-cell proliferation and IFN-g production via increased surface expression of programmed death ligand-1 (PDL1) [32 ]. This trait is exacerbated diabetes mellitus patients, and the resulting loss of effective CD4 + T-cell function may contribute to disease severity and the ability to prevent disease progression.
Phagosome escape and the type three secretion system Following uptake by a professional phagocyte, BPS is contained in a membrane-bound vesicle that matures through endosomal contact to eventual fusion with the lysosome, an organelle that by virtue of acidic pH and degradative enzymes destroys the contents of the phagosome, allowing denatured proteins to be loaded onto MHCII molecules for surface expression and engagement of CD4 T-cell immunity.
As the phagosome matures, BPS encounters host proteases that are able to kill the compartmentalised invader.
To counter this, BPS expresses a serine protease inhibitor called ecotin that enables growth in macrophages and contributes to virulence in the murine model [33]. Ultimately, however, BPS is able to escape the phagosome before phagolysosomal fusion occurs, and for this, the T3SS is essential.
The role of the T3SS in BPS intracellular fitness, and therefore virulence cannot be overstated. To provide some context, the capsule of BPS is a well-studied virulence factor that confers protection against antibiotics and host immune effectors such as complement and antibody deposition. In a murine model of acute respiratory BPS infection, capsule mutants were shown to be attenuated 6.

Metabolic challenges in the intracellular environment
Any intracellular organism finds itself in a conflict for resources with the host to maintain its metabolic activity. This is especially true for essential micronutrients such as iron, which is incorporated into molecules involved in multiple biological processes. Pathogens such as Mycobacterium tuberculosis achieve the sequestration of iron through the expression of siderophores that have a higher affinity for the molecule than equivalent host molecules such as transferrin and lactoferrin. Ferric siderophores are then recognized by specific receptors on the bacterial surface. The BPS malleobactin siderophore family is crucial for its virulence and intracellular survival. They appear to be remarkably diverse in their peptide structure, suggesting characteristic adaptability of BPS to different microenvironmental conditions [45]. Genes that encode malleobactins as well as other BPS proteins relating to iron binding and uptake have been shown to be upregulated in experimentally induced models of iron deprivation, including oxidative stress [23,46,47]. Besides the dedicated iron-binding molecules, BPS expresses proteins with dual function such as BPSL2825 (PabB), a para aminobenzoate synthetase component important for metabolism and growth in vitro that takes on a critical role utilizing hemin as an iron source in vivo [47].
Metabolic flexibility is key to the ability of BPS to adapt as it moves from in terra to in vivo and it remains one of its defining features. While several Burkholderia spp. are similarly well equipped to thrive on multiple carbon sources, it has been postulated based on genomic interrogation that virulent species may regulate these pathways differently from avirulent species, with consequential effects on the expression of virulence factors [48]. An important requirement of pathogens, whether they are obligate or facultatively intracellular is their ability to metabolise fatty acids [49][50][51]. In bacteria, fatty acids are required for the formation of cell wall phospholipids and may be acquired either from existing host-derived fatty acids, through de novo synthesis or a combination of both. Since there is no evidence that BPS is able to acquire host-derived fatty acids, de novo fatty acid synthesis an attractive target for chemotherapeutic intervention. FabI is an enoyl-acyl carrier protein reductase, a key enzyme of the type II fatty acid synthesis system, and has been shown to be required for BPS virulence in the murine model of infection. It has since been successfully targeted by specific chemical inhibitors [52,53], representing a promising approach to drug design.
In addition to these processes, anabolic pathways, while perhaps not traditionally considered virulence factors are nonetheless essential for intracellular fitness. Anabolism requires ATP energy, for which the synthesis of purine, encoded by the pur gene cluster is required, purine also being a vital component of nucleic acids [54]. BPS differs from its host in that it is able to synthesise several amino acids and vitamins that humans cannot and instead acquire from dietary intake. This also makes them an attractive target for antimicrobial therapy. For example, histidine synthesis, encoded by the his gene cluster, is required for intracellular growth of BPS, and this area is also being explored for potential future vaccine design [54].

Cell to cell spread and the Type Six Secretion System
BPS is able to spread beyond the site of initial infection, disseminating through the host and potentially infecting any organ or tissue. One strategy is to use the infected host cell itself as a vehicle. BPS has been shown to remain viable within, and induce maturation and migration of, immature DCs in vitro, a phenomenon seen in other infections as a route for transport to secondary lymphoid organs in vivo; although seemingly only a minority of BPS-infected DCs translocate to the lymph in vivo [55].
More localised spread occurs via a distinctive feature of BPS infection, namely its ability to create MNGC. Arising from the fusion of cell membranes between an infected cell and a neighbouring, uninfected cell, it is a phenomenon that enables continued intracellular replication, and cell to cell spread without the requirement of an extracellular step and the consequences that entails regarding host immunity. BPS is able to form MNGC in both phagocytic and non-phagocytic cells within hours of infection [56], and the eventual death of the MNGC results in plaque formation [57 ] that is damaging to the host, and may serve as a niche for further BPS replication.
It is well established that cell-cell spread and induction of MNGC requires bacterial motility. In addition to the two flagellin encoding genes, fla2 on chromosome 2 and fla1 on chromosome 1, BPS achieves intracellular motility through the action of the autotransporter, BimA. Localising to one pole of the bacillus, it binds host actin, polymerisation of which pushes the bacteria through the cytoplasm, leaving in its wake characteristic 'comet tails'. BimA localisation is not itself directed to the poles, rather it is transported equally across the bacterial inner membrance [58]. BimA localisation depends on the polarity of its partner protein, BimC, which functions through an iron-dependent mechanism that remains to be fully described.
Interestingly, BimA is also expressed by B. thailandensis which is avirulent in humans. Recent published work has revealed the finer distinction responsible for virulence: whereas B. thailandensis BimA mimics the host Arp3/3 actin polymerase complex, BimA from BPS and B. mallei, another pathogenic member of the genus that causes the disease, glanders, in horses, instead mimic the far more efficient host Ena/VASP complex to initiate and elongate actin filaments [59,60 ]. The result is that while B. thailandensis BimA is able to generate branched actin networks, BPS BimA generates multiple bundles of filaments; moreover, BPS BimA is able to generate more actin tails and larger MNGC compared with B. mallei BimA, highlighting the key importance of actin-based motility for cell-cell spread in the host.
In addition to bacterial motility, a key requirement for cell fusion and MNGC formation is the Type Six Secretion System (T6SS), disruption of which attenuates virulence in vivo. BPS possesses a predicted six T6SS gene clusters, which are thought to have evolved as a weapon for inter-bacterial competition [61] within the rhizosphere, the native microbiome of BPS.
While the T3SS is not directly required for MNGC formation, it is a prerequisite that the bacteria are able to escape the phagocytic vacuole [39,57 ]. Furthermore, there is cross-talk between the T6SS and T3SS; for example, T6SS1 expression is regulated by BprC, which is located in the T3SS gene cluster. The T6SS is upregulated by several orders of magnitude upon infection of host cells, and this is largely regulated by the two-component histidine kinase sensor, VirAG [62], which is itself regulated by a T3SS regulator, BsaN [63 ]. Mutagenesis studies have revealed that abrogation of both virAG and bprC function attenuates virulence in the murine model of BPS infection [63 ].
VirAG activates the promoter of a key T6SS effector, hemolysin-coregulated protein (Hcp), located in gene cluster five out of the six BPS T6SS gene clusters [63 ]. Solving the crystal structure of BPS Hcp has revealed unique variations in its tertiary structure that are not present in orthologues from avirulent species, hinting at its key role in virulence.
Hcp polymerases the hexameric ring structures that comprise the T6SS, forming an overall structure that is akin to the bacteriophage tailspike. Hcp is also secreted by the T6SS, dependant on the protein, BPSS1504 [64], and functions as a chaperone for other secreted effectors, localising to the infected cell plasma membrane. Exogenous recombinant Hcp also localises to the host cell surface, preferentially targeting antigen presenting cells. Accordingly, it has been found to be highly immunogenic and reacts strongly with sera from melioidosis patients [65]. The consequence of this is not clear, but it has been reported that binding of Hcp induces the anti-inflammatory cytokines, IL-10 and TGF-b, potentially interfering with the mounting of a robust immune response to other epitopes besides Hcp.
Activation of T6SS is a dynamic, multi-step process that begins with assembly of the inner tube, comprised of Hcp, its outer sheath and baseplate at the bacterial cell wall-host cell membrane interface. The tip of the T6SS spike is comprised of a trimer of the protein, Valineglycine repeat protein G (VgrG) [66]. VgrG tranverses the inner membrane, periplasmic space and peptidoglycan layer of the bacterial cell wall, while the elongation of the tail spike extends into the bacterial cytoplasm.
Upon contact with the host cell membrane, the T6SS outer sheath contracts and provides the kinetic energy for the inner tube to propel the VgrG tip, penetrating the host cell and potentially into the neighbouring cell membrane [67 ]. At this stage, T6SS effector proteins are also delivered. The recruitment of ATPase then disassembles the outer sheath proteins, which are recycled [68]. The contraction of the T6SS may serve to pull the two membranes close enough whereupon, according to the hemifusion paradigm, membrane fusion occurs when proximity is <1 nM [69], resulting in a fused, multinucleated cell. Although VgrG has been shown to be functionally conserved across Burkholderia spp., some species such as B. oklahomensis lack the ability of cell-cell spread, suggesting other requisites than VgrG alone [66].

Conclusions and future perspectives
Much has been learned in recent years about how BPS is able to infect human cells, evade immune killing mechanisms and replicate intracellularly. This review is not an exhaustive list of BPS virulence factors. Several are the subject of ongoing research to fully explain their function, such as boaB, bpaC and bpaE, which all contribute to virulence in the murine infection model [70]. Still more targets remain to be identified and understood, and newly emerging tools will help to expedite these efforts. For example, transposon-directed insertion-site sequencing (TraDIS) also known as TnSeq, has facilitated the discovery of the essential gene set of BPS [4] and B. thailandensis [71] and has very recently been applied to elucidate virulence genes in the acute model of in vivo infection [72]. A particular remaining challenge is to understand the establishment of chronic infection with BPS, which may demand novel treatment therapies in the clinical setting [73]. We have gained important insights into the pathogenicity of BPS, for example the significance of the T3SS and T6SS, and some of the basic cell biology of infection. Further work is needed to understand the subtleties of host-pathogen interactions that result in, or fail to result in, protective host immunity. Together these efforts will contribute to our ability to counteract the emerging threat of this well-armed pathogen.