Legionella secreted effectors and innate immune responses

Introduction

The innate immune system functions to provide the first line of defence against pathogen attack and to signal the adaptive immune system of the presence of potential infection. This system is composed of two major components. The first is the scavenging function by phagocytic cells such as macrophages that engulf and digest incoming pathogens. Phagocytosis and subsequent lysosomal digestion is one of the first challenges met by pathogens. The second component is signalling pathways consisting of germline-encoded receptors that are highly conserved across different kingdoms of organisms and, which sense pathogen-associated molecular patterns (PAMPs). These receptors, called pattern recognition receptors (PRRs), can be divided into two families: the Toll-like receptors (TLRs), which recognize ligands on the cell surface or those associated with the lysosome/endosome, and the Nod-like receptors (NLRs), which detect bacterial and viral molecules in the cytoplasm (Akira et al., 2006).

Legionella pneumophila was first recognized as a pathogen following an outbreak of pneumonia among attendees of the 1976 American Legion convention in Philadelphia (Fields et al., 2002). This bacterium is ubiquitous in aquatic environments as a parasite of freshwater protozoa. This habitat is believed to provide the evolutionary pressure for the acquisition of the traits necessary for survival and replication in phagocytes (Swanson and Hammer, 2000). Development of this potentially fatal disease occurs when susceptible individuals inhale contaminated aerosols. Instead of being digested, L. pneumophila engulfed by alveolar macrophages undergoes robust intracellular growth, leading to tissue damage and subsequent development of the disease (Fields et al., 2002). Largely due to the progress in our understanding of the cell biological and biochemical basis for its intracellular life cycle (Isberg et al., 2009), L. pneumophila has become an attractive model in the study of bacterial pathogenesis as well as the mechanisms of pathogen detection by the immune system.

The single most important pathogenic determinant of L. pneumophila is the Dot/Icm type IV secretion, which translocates effector proteins into host cells (Ensminger and Isberg, 2009). These effectors have some highly intriguing features: first, the repertoire is extremely large and more than 270 experimentally verified substrates have been identified (Luo and Isberg, 2004; Heidtman et al., 2009; Burstein et al., 2009; Huang et al., 2010; Zhu et al., 2011); this number is close to 10% of the predicted protein-encoding genes in the genome of L. pneumophila (Cazalet et al., 2004; Chien et al., 2004). Second, the structures of these proteins are highly diverse. Whereas some proteins contain readily detectable signatures suggestive of potential biochemical activity [e.g. RalF (Nagai et al., 2002)], most proteins are unique to Legionella and do not show significant homology to other proteins in thecurrent database (Zhu et al., 2011). Third, deletion of a single one of these genes rarely leads to defects in Legionella intracellular growth in standard experimental conditions (Isberg et al., 2009). Finally, host processes targeted by these proteins are very diverse and in some cases the potential benefit of such modulation is not yet understood (Dorer et al., 2006; Shin and Roy, 2008). These effectors not only play essential roles in the biogenesis of the specialized replicative organelle but also function as messengers that trigger host immune responses. The latter activity has been exploited as a highly effective tool to dissect the mechanisms used by the host to distinguish pathogenic microorganisms from non-harmful ones.

Here, I will discuss the roles of Dot/Icm effectors directly engaging the host innate immune system. The immune responses mediated by TLRs in response to Legionella PAMP molecules such as LPS and the detection of Legionella flagellin by pathways involved the NLR family protein Naip5 and Ipaf have recently been reviewed (Neild and Roy, 2004; Vance, 2010) and will not be covered here.

Inhibition of phagosome maturation

After phagocytosis, the Legionella-containing vacuole (LCV) bypasses the default phagosome maturation pathway. At least in the first 10 h after uptake, LCVs maintain a neutral pH and are devoid of typical lysosomal proteins such as LAMP-1 (Sturgill-Koszycki and Swanson, 2000). Lysosomal killing is one of the basic components of innate immunity and is the first challenge L. pneumophila faces after phagocytosis. As a parasite of freshwater amoeba, L. pneumophila has evolved sophisticated mechanisms to escape the delivery of its phagosome into the lysosomal network. Active remodelling of the phagosomal membranes to mimic those of resident organelles is a common strategy used by intravacuolar pathogens (Blander and Medzhitov, 2006). Within minutes of its formation, membranes of the LCV are loaded with some resident proteins of the endoplasmic reticulum (ER), such as calnexin and Bip, indicating that the LCV interacts with the ER network (Swanson and Isberg, 1995; Tilney et al., 2001). Further, inhibition of the activity of proteins such as the small GTPases Sar1 and Rab1, which regulate vesicle budding and trafficking from the ER, leads to the arrest of intracellular bacterial growth (Kagan and Roy, 2002). Based on these observations, Roy and colleagues proposed that L. pneumophila redirects the trafficking of ER-originated vesicles to remodel the LCV membranes (Kagan and Roy, 2002). This model was supported by the fact that the LCV is surrounded by small smooth vesicles shortly after its formation (Tilney et al., 2001; Robinson and Roy, 2006), and was greatly substantiated by the discovery of Dot/Icm effectors that directly manipulate activities of Arf1 and Rab1, small GTPases essential for regulation of vesicle trafficking between the ER and the Golgi apparatus (Nagai et al., 2002; Machner and Isberg, 2006; Murata et al., 2006).

Oscillating between an active GTP-bound form and an inactive GDP-bound form, Rab1 regulates several important steps of vesicle transport, most notably, the anterograde transport between the ER and the Golgi apparatus (Stenmark, 2009). L. pneumophila hijacks Rab1 activity with proteins that mimic the functions of its regulatory proteins. Among these, SidM/DrrA is a protein of multiple activities. Anchoring on the LCV by binding to phosphatidylinositol-4-phosphate (Brombacher et al., 2009), SidM/DrrA extracts Rab1 from the GDP association inhibitor (GDI) and recruits and activates it on the bacterial phagosome by its guanine nucleotide exchange factor (GEF) activity (Murata et al., 2006; Machner and Isberg, 2006; 2007; Ingmundson et al., 2007) (Fig. 1). The bacterial GTPase activation protein (GAP) LepB completes the activation cycle by inducing the GTPase activity of Rab1, leading to its subsequent removal from the LCVs (Ingmundson et al., 2007) (Table 1). Activated Rab1 is believed to facilitate the recruitment of membrane vesicles to the LCV, thus contributing to the membrane-remodelling process (Machner and Isberg, 2006).

It is intriguing to note that the activity of Rab1 recruited to the LCV is under another layer of regulation, which, strikingly, is imposed by SidM/DrrA. Using a catalytic mechanism similar to glutamine synthetase adenylyl transferase, SidM/DrrA post-translationally modifies tyrosine 77 of Rab1 by AMPylation, locking the GTPase into its active form which cannot be induced by LepB (Muller et al., 2010). This paradox was recently resolved by the identification of the bacterial deAMPylase SidD, which reverses the modification, making Rab1 accessible to the GAP protein (Neunuebel et al., 2011; Tan and Luo, 2011). Despite the interesting mechanisms of regulation and the apparent importance of temporal control of Rab1 activity during L. pneumophila infection, it is not clear why it is necessary to lock this small GTPase in its active form during the several hours of its association with the LCVs. It is possible that AMPylation prevents the interference of Rab1 activity by host or bacterial factors. Remarkably, a recent study showed that Rab1 is phosphorylcholinated by AnkX at Ser₇₆ (Mukherjee et al., 2011). Although the effect of this post-translational modification on Rab1 is not clear, it is possible that AMPylation may prevent the potentially negative effect of AnkX on Rab1 at certain cellular location or phase of infection. In any case, the extensive and precise regulation of the activity of a host protein critical for vesicle transport highlights the importance of efficient membrane remodelling in meeting the challenges posed by the host innate immunity system.

It is worth noting that deletion of any of the genes involved in modulating Rab1 activity did not cause defects in L. pneumophila intracellular growth (Isberg et al., 2009), indicating that modulation of Rab1 activity does not entirely represent the factors involved in membrane remodelling of the LCV. In fact, Dot/Icm effectors such as SidJ and LegK2, which are important for efficient acquisition of ER specific proteins to the LCVs, have been identified (Liu and Luo, 2007; Hervet et al., 2011) (Table 1). However, their mechanisms of action are unknown. Similarly, many effectors capable of interfering with membrane trafficking in the model system yeast have been identified but the mechanism of action of these proteins remains to be determined (de Felipe et al., 2008; Heidtman et al., 2009). Detailed analysis of their functions will very likely reveal new layers of regulation in membrane trafficking and membrane remodelling of the LCV.

Nascent phagosomes mature by sequential fusion with endosomes and ultimately with lysosomes, which allows for terminal degradation of the engulfed particles or the killing of the phagocytosized microorganisms. A notable feature associated with matured phagosomes is low luminal pH (about 4.5), which is important for their digestive capacity because the optimal activity of many phagosomal hydrolytic enzymes requires an acidic environment (Greenberg and Grinstein, 2002). Regulation of phagosomal pH is mediated by the vacuolar ATPase (v-ATPase), a multi-component machinery responsible for proton transport across membranes in a process powered by ATP hydrolysis (Forgac, 2007). After phagocytosis, wild-type L. pneumophila maintains a neutral phagosomal pH (Sturgill-Koszycki and Swanson, 2000). Unexpectedly, given the absence of other endosomal markers, v-ATPase was found on membranes of LCVs (Urwyler et al., 2009), suggesting the necessity to inhibit the proton transporter. By screening for Legionella genes capable of causing the neutral pH-sensitive phenotype exhibited by yeast v-ATPase mutants, Xu et al. identified SidK, a Dot/Icm substrate that specifically inhibits v-ATPase activity by binding to VatA, a component of the peripheral V₁ domain involved in ATP hydrolysis (Xu et al., 2010) (Fig. 1). Macrophages loaded with SidK were impaired in the digestion of internalized non-pathogenic bacteria (Xu et al., 2010), highlighting the role of v-ATPase in lysosomal killing and the importance of SidK and other yet unidentified Legionella effectors for modulating the function of the proton transporter.

Regulatory factors such as members of the Rab small GTPase family and proteins involved in modulating lipid composition of the LCV membrane are believed to contribute to the prevention of the bacterial phagosome from entering the lysosomal pathway (Russell et al., 2006). For example, kinases involved in the metabolism of phosphatidylinositol-3-phosphate (PI3P) may be important because this lipid directly affects localization and function of proteins involved in phagosome maturation (Fratti et al., 2001). In fact, anchoring of two effectors, SidC and SidM to the LCV is mediated by binding to PI4P (Weber et al., 2006; Brombacher et al., 2009). However, effectors directly responsible for the presumed enrichment of phosphatidylinositol on the vacuole surface have not been reported. Interestingly, AnkX also targets Rab35, a small GTPase involved in endosomal trafficking (Mukherjee et al., 2011) (Fig. 1). In light of our current appreciation of the redundant mechanisms used by Legionella to regulate its membrane remodelling, it is attempting to predict the existence of more effectors involved in these processes. Future identification and functional characterization of these effectors will surely lead to better understanding of mechanisms of phagosome maturation and its contribution to host defence against intravacuolar pathogens.

Activation of the NFκB pathway

The second component of the innate immune response is mediated by the activation of specific signalling pathways at the transcriptional level, leading to the production of antimicrobial molecules such as cytokines and the activation of the adaptive immune system. Infection by Legionella triggers the classical TLRs-dependent immune responses (Archer et al., 2009). It also activates caspase 1 by the Naip5/Birc1e and Ipaf-dependent pathway involved in the detection of flagellin released into the host cytosol by the Dot/Icm transporter (Vance, 2010). Microarray analyses with bacterial flagellin mutants using macrophages derived from mice defective in the TLR pathways have revealed important roles for the Dot/Icm system in the induction of several innate immunity pathways, including the NFκB pathway, MAP kinase pathway and host stress response (Losick and Isberg, 2006; Shin et al., 2008; Fontana et al., 2011).

The master regulator NFκB plays an essential role in the mammalian innate immunity response by controlling the expression of a large number of genes (Hayden et al., 2006). Signals capable of activating NFκB are very diverse, ranging from classical PAMP molecules like LPS to cellular insults such as ER stress and disruption of the actin cytoskeleton (Akira et al., 2006; Bruno et al., 2009). In most cases, these molecules trigger a signalling cascade that causes the release of NFκB from members of the IκB inhibitor family and its subsequent nuclear translocation (Akira et al., 2006). Phosphorylation of IκB by the IκB kinases (IKKs) leads to K48-linked ubiquitination and degradation by the proteasome (Akira et al., 2006). Thus, any activity that leads to IκB phosphorylation will cause an NFκB-mediated response. The importance of the Dot/Icm system in the activation NFκB suggests the existence of substrates capable of causing IκB phosphorylation. Using NFκB-specific luciferase reporters, two Dot/Icm substrates, LegK1 and LnaB, which potently activate NFκB, have been identified (Ge et al., 2009; Losick et al., 2010) (Fig. 1). LegK1 is a Ser/Thr kinase, which in vitro directly phosphorylates some members of IκB family, including IκBα and p105, leading to the activation of the non-canonical NFκB pathway (Ge et al., 2009). However, the mechanism of action of LnaB is not known (Table 1). Further, the significance of LegK1 and LnaB in the activation of NFκB during L. pneumophila infection is unclear because mutants lacking either of these proteins exhibited very marginal defects, if any, in the activation of this pathway (Ge et al., 2009; Losick et al., 2010). Of note is that a number of Dot/Icm effectors that moderately activate the NFκB reporters were identified in these screenings (Ge et al., 2009; Losick et al., 2010; Haenssler and Isberg, 2011). It is possible that the collective effects of some of these proteins contribute significantly to the NFκB activation during bacterial infection.

Fontana et al. revealed a novel mechanism of immune regulation during L. pneumophila infection by demonstrating the requirement of five effectors involved in inhibiting host protein synthesis in the production of distinctive cytokines such as IL23a and Csf2 (Fontana et al., 2011). Compared to infections with the mutant lacking all five translation inhibitors (Table 1), a minimum of a 2 h delay in IκB re-synthesis was observed in macrophages infected with the wild-type strain (Fontana et al., 2011). Because of its labile property, delayed re-synthesis of IκB after its turnover can lead to prolonged NFκB activation, which induces the expression of a set of immune genes (Fig. 1). Thus, activity that prevents efficient production of IκB can activate NFκB. These results indicate that in addition to specific microbial molecules that activate TLRs or cytosolic PRRs, a pathogen-encoded activity can be sensed by host immune surveillance mechanisms. It is important to note that mice infected by the Δ5 mutant were not defective in cytokine production, nor did IL23a^-/- mice exhibit detectably higher bacterial burden (Fontana et al., 2011). Such discrepancies may result from the many redundant pathways that recognize L. pneumophila in vivo (Archer et al., 2010; Fontana et al., 2011)

Because the death of a host cell will lead to premature termination of infection, induction of apoptotic pathways is considered a mechanism of innate immunity. In addition to the production of cytokines, NFκB activates the expression of many anti-apoptotic genes during L. pneumophila infection (Losick and Isberg, 2006; Abu-Zant et al., 2007; Shin et al., 2008). Two lines of evidence indicate that reprogramming of the host cell death pathway by NFκB activation is important for bacterial intracellular growth. First, in macrophages lacking the anti-apoptotic gene plasminogen activator inhibitor-2 (PAI-2), L. pneumophila infection caused more extensive cell death and a reduction in bacterial replication (Losick and Isberg, 2006). Second, blocking NFκB nuclear translocation by genetic or pharmaceutical agents caused extensive cell death upon low-dose bacterial challenge (Losick and Isberg, 2006). These activities act in concert with effectors such as SidF and SdhA, which inhibit host cell death by distinctive mechanisms (Laguna et al., 2006; Banga et al., 2007).

Activation of the MAP kinase pathway

Infection by L. pneumophila also strongly induces the MAP kinase pathway, which is involved in innate immunity and regulation of cell death in parallel with the NFκB pathway (Haenssler and Isberg, 2011). Unlike NFκB, which only exists in mammalian cells and is not believed to play a significant role in shaping the pathogenicity of L. pneumophila, the MAP kinase pathway is an important signalling conduit in its natural amoeba host. The NFκB and MAP kinase pathways, however, share some common properties: first, in both cases, the stimulus is relayed by specific kinases, which form the step-wise signalling cascade. Second, like the NFκB pathway, stress response signals and stimuli engaged by TLRs or NLRs also induce the MAP kinase pathway (Huang et al., 2009; Ting et al., 2010). Regulation of MAPK signalling is mainly achieved by negative feedback loops via members of the dual specificity phosphatases (DUSPs) (Jeffrey et al., 2007). Analogous to the control of IκB expression by NFκB, transcription of DUSPs is directly regulated by MAPK signalling. Dot/Icm-dependent induction of the MAP kinase pathway occurs in both amoeba and mammalian hosts (Losick and Isberg, 2006; Farbrother et al., 2006; Shin et al., 2008). For example, an amoeba mutant lacking DupA failed to support maximal intracellular bacterial growth (Li et al., 2009). However, the lack of DupA caused constitutive activation of MAPK signalling, which led to significant changes in host gene expression, making it difficult to pinpoint the host proteins directly responsible for this phenotype (Li et al., 2009). The role of the MAPK signalling cascade in Legionella infection of mammalian hosts is unknown.

Although the involvement of the Dot/Icm transporter in the induction of the MAP kinase pathway is well established, effectors specifically targeting components of this pathway have not been identified. Under infection conditions, the putative kinases LegK1, LegK2 and LegK3 translocated by the Dot/Icm system do not appear to be important in MAPK signalling (Shin et al., 2008). However, whether any one of these proteins is capable of activating this pathway when overexpressed is not known. Future studies using suitable MAPK-dependent reporters may lead to identification of such genes. It is also possible that the induction is initiated by indirect effects resulting from the perturbation of host processes by subsets of these effectors. Identification of L. pneumophila mutants unable to induce the MAP kinase pathway will provide excellent tools to probe its regulation. For example, if the mutant missing all five translation inhibitors no longer stimulates the induction, it would suggest the involvement of one or more labile proteins akin to IκB in the regulation of the MAPK signalling.

Whereas the role of the Dot/Icm transporter or its effectors in the induction of the NFκB pathway and the MAP kinase pathway has been established, it is important to note that in both cases, maximal and sustained induction of immune responses cannot be achieved by activity conferred by Dot/Icm alone. Instead, one or more pathways engaged by TLRs or NLRs are required for maximal immune responses. In fact, it has been suggested that NFκB activation during L. pneumophila infection is a two-phase process in which the classical TLR-dependent pathway occurs immediately after bacterial contact, whereas the Dot/Icm-dependent activation persists during the entire replication cycle (Bartfeld et al., 2009). For example, maximal induction of IL23a and Csf2 by macrophages only occurs in the presence of protein synthesis agents such as ExoA and Pam3 CSK4, and ligands for TLR2 and/or Nod2 ligands (Fontana et al., 2011). Similarly, full induction of some interleukins, such as Il1a, Il1b, Il6 and Il12, requires an active Dot/Icm system and the TLR pathway mediated by MyD88 and/or Rip2 (Shin et al., 2008).

Modulation of type I interferon induction

Legionella pneumophila infection leads to transcriptional activation of type I interferon (IFN) genes and restriction of bacterial replication. This observation is reminiscent to the role of type I IFN in the innate immune response against viral infection (Akira et al., 2006). Several groups have contributed to our understanding of the mechanism of the induction of type I IFNs by L. pneumophila-infected macrophages: first, this process is independent of the TLR pathway; second, a functional Dot/Icm type IV secretion system is indispensible for the induction; third, the response does not require the pathway involved the sensing of flagellin by Naip5/Nlrc4 (Opitz et al., 2006; Stetson and Medzhitov, 2006; Monroe et al., 2009). However, the nature of the ligand and receptor involved in this signalling cascade is less clear. Based on the observation that IRF-3 or IPS-1 is essential for L. pneumophila-induced IFNβ expression, it has been suggested that bacterial DNA released by the Dot/Icm transporter is the signal sensed by the cell (Opitz et al., 2006; Stetson and Medzhitov, 2006). On the other hand, another study showed that 5′-ppp RNA products generated by the host RNA polymerase III from bacterial DNA trigger IFNβ induction through the RIG-I pathway (Chiu et al., 2009). In contrast, more recent evidence indicates that in mouse macrophages, the RIG-I and Mda5-dependent type I IFN response occurs by sensing RNA molecules presumably directly released by the Dot/Icm transporter (Monroe et al., 2009). This conclusion is consistent with the fact that IPS-1/MAVS-KO mouse cells exhibited very marginal reduction in type I IFN response to the B form of DNA delivered by transfection (Kumar et al., 2006). Of greater interest, Monroe et al. found that the Dot/Icm substrate SdhA functions to suppress the type I IFN induction triggered by L. pneumophila infection (Monroe et al., 2009) (Fig. 1). SdhA also plays a role in protecting infected cells from cell death (Laguna et al., 2006), a function that appears to be independent of the suppression of type I IFN induction (Monroe et al., 2009). The suppression by SdhA protects the intracellular bacteria from the detrimental effect of type I IFN production.

The role of type I IFN induction in defence against L. pneumophila infection is a subject of debate. Whereas earlier studies suggested that mice deficient in type I IFN signalling alone are not more susceptible to Legionella infection (Monroe et al., 2009; Ang et al., 2010), a recent report showed that both type I and II IFNs protect mice during lung infections, because higher bacterial burdens were observed in mice deficient in receptors for IFNα(IFNAR^-/-) or INFγ(IFNGR^-/-) or both (Lippmann et al., 2011).

Dot/Icm-dependent transfer of DNA has been well established (Vogel et al., 1998), but the transfer of RNA by this transporter has not yet been documented. Thus, the potential discrepancy regarding the receptors involved in the type I IFN pathway may be due to the uncertainty of the ligands and the cell types used. It is possible that a subset of the many Dot/Icm effectors or the collective effects of subsets of them serve as the signal (Monroe et al., 2009). As exemplified by the activation of NFκB by bacterial proteins involved in inhibiting host protein synthesis (Fontana et al., 2011), such proteins may not directly target the RNA sensing pathways. Instead they may interfere with one or more as yet unrecognized host processes, leading to indirect activation of the pathway. In light of the emerging mechanisms for Dot/Icm effector-mediated regulation of the activity of other effectors (Kubori et al., 2010; Luo, 2011; Neunuebel et al., 2011; Tan and Luo, 2011), SdhA may function to countercheck the activity of one of more effectors directly involved in type I IFN induction. Regardless of their chemical nature, identification of the effectors responsible for the type I IFN induction will be instrumental in addressing these important questions.

Concluding remarks

Study of the function of Dot/Icm effectors in the last decade has allowed us to better understand the mechanisms employed by L. pneumophila to manipulate host processes. To some extent, this pathogen can be considered a brilliant cell biologist with the tools necessary to exploit the intricacy of host cell biology. On the other hand, in contrast to highly adapted mammalian pathogens that have evolved sophisticated strategies to avoid acute immune responses from the host, L. pneumophila is ‘naïve’ in its interactions with the immune system of higher organisms. The many ‘accidents’ caused by Dot/Icm effectors in mammalian macrophages are invaluable tools for analysing patterns of pathogenesis or activity-dependent immune surveillance mechanisms, which function in concert with the classical receptor-mediated recognition to distinguish pathogens from non-pathogens (Vance et al., 2009). Detailed analysis of the activity of the effectors involved and the cellular processes engaged by them should reveal unrecognized yet important interactions among signalling pathways that govern a variety of distinct cellular functions.