If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Legionella pneumophila is a facultative intracellular pathogen that uses the Dot/Icm Type IV secretion system (T4SS) to translocate many effectors into its host and establish a safe, replicative lifestyle. The bacteria, once phagocytosed, reside in a vacuolar structure known as the Legionella-containing vacuole (LCV) within the host cells and rapidly subvert organelle trafficking events, block inflammatory responses, hijack the host ubiquitination system, and abolish apoptotic signaling. This arsenal of translocated effectors can manipulate the host factors in a multitude of different ways. These proteins also contribute to bacterial virulence by positively or negatively regulating the activity of one another. Such effector–effector interactions, direct and indirect, provide the delicate balance required to maintain cellular homeostasis while establishing itself within the host. This review summarizes the recent progress in our knowledge of the structure–function relationship and biochemical mechanisms of select effector pairs from Legionella that work in opposition to one another, while highlighting the diversity of biochemical means adopted by this intracellular pathogen to establish a replicative niche within host cells.
Bacterial secretion systems are complex cellular machines used to translocate toxins and virulence factors into host cells. Nine major secretion systems (Types I–IX) have been described in Gram-negative and Gram-positive bacteria. Specialized systems, such as type III and type IV secretion apparatus, have been central to the evolution of many intracellular pathogenic Gram-negative bacteria (
). Pathogenic bacteria such as Legionella pneumophila exploit eukaryotic cell functions and influence multiple signaling events by translocating over 330 effectors via its Type IV secretion system, also known as the Icm/Dot transporter (
) while sharing little sequence homology with eukaryotic enzymes of the same function. Another common theme seen among bacterial effectors is redundancy, which extends beyond gene duplication. The most well-recognized form is functional redundancy, whereby two effectors may catalyze the same reaction and have similar substrate specificities, thus allowing one to compensate the absence of the other, as exemplified by the SidE family proteins from L. pneumophila that catalyze ubiquitination of host proteins associated with the endoplasmic reticulum. Redundancy can also exist between unrelated bacterial proteins. For example, Legionella effectors such as SidM, AnkX, and SidE proteins are all known to target Rab1; however, the mechanisms by which each of these effectors modulates the Rab protein are entirely different, resulting in different modifications on the GTPase.
Among the diverse ways of interaction with their host, pathogens have evolved a variety of means of manipulating host pathways by targeting their posttranslational modifications (PTMs) (
). PTMs can range from addition of relatively small chemical groups, such as acetyl, hydroxyl, phosphate, AMP, ADP-ribose, or phospholipids, to more complex forms involving conjugation of small proteins, such as ubiquitin (Ub) or ubiquitin-like proteins (Ubls), to other proteins. Modification of host targets via effector-mediated enzymatic activities allows intracellular bacteria to remodel cellular processes relatively quickly and reversibly, if needed. While most effector activities are usually directed against host proteins, it is becoming increasingly evident that pathogens have developed another layer of complexity by regulating effector–effector interactions. Such modulation occurs when the translocated bacterial proteins either indirectly counterbalance their activity in a shared host pathway or directly interact with one another to suppress or enhance the associated function.
Legionella represents one of the most elaborate cases of cross talk between host cellular processes and the effectors it translocates. This pathogen has become a popular model for understanding both the role of the effectors in infection and the affected host signaling mechanisms, examples of which have been covered extensively in several past reviews (
). In this review we focus on functionally antithetic effectors from L. pneumophila with well-established biological function. This review intends to provide structural and mechanistic insights into specific examples (Table 1) that best illustrate the existence of effectors carrying out opposing functions in this organism. These effectors seem to work in concert to help the bacteria establish a balanced lifestyle within the host while avoiding catastrophic effects on the host environment.
Table 1A list of all effector-effector pairs discussed in this review
). Ubiquitin is covalently attached to protein substrates via an isopeptide bond linking the last carboxylate of Ub (on Gly76) to the ε-amino group of lysine residues of substrates through the sequential actions of a trio of enzymes, an ATP-dependent Ub-activating E1 enzyme, a Ub-conjugating E2 enzyme, and an E3 Ub ligase (Fig. 1). This covalent attachment can be reversed when desired by deubiquitinases (DUBs) which catalyze the hydrolysis of the isopeptide bond. In addition to the vast array of other functions regulated by ubiquitination, it serves as the first line of defense against invading pathogens by mediating signaling events leading to innate immune response and xenophagy (
) and even manipulate the ubiquitination machinery to serve their purpose, sometimes using means outside the eukaryotic repertoire. In two recent examples, L. pneumophila was shown to ubiquitinate host targets by employing mechanisms independent of the classical three-enzyme system of eukaryotes: the SidE effectors use nicotinamide adenine dinucleotide (NAD+) to ubiquitinate serine residues of host targets via a phospho-ribose (PR) linker connecting the hydroxyl group with Arg42 of Ub (
), a reaction that does not even require a nucleotide cofactor (Fig. 1). These orthogonal modes of ubiquitination cannot be reversed by host DUBs, allowing the bacteria to remodel specific cellular pathways at will. However, aggressive manipulation of the Ub system and associated cellular processes can be detrimental to the pathogen. It could result in the host succumbing to the offense, ultimately limiting bacterial replication. Accordingly, L. pneumophila has evolved distinct strategies for a balanced control: in some cases, one effector switching off another to block its ubiquitinating activity, or, in other examples, one effector reversing the ubiquitin modification installed by another through a deubiquitinase-like reaction. In addition to these atypical ubiquitinating and deubiquitinating enzymes, L. pneumophila also possesses several effectors that mimic components of the host ubiquitination machinery, including the classical DUBs and E3 ligases, to co-opt the Ub system and interfere with Ub signals used in cellular defense (
MavC and its paralog MvcA are two effectors in L. pneumophila present on the same locus adjacent to each other with a 76-base pair intergenic space (Fig. 2A). These effectors were first described as cysteine-dependent deamidases based on their structural similarity to a group of bacterial deamidases of the Cif family, such as CHBP. They catalyze the conversion of Gln40 of Ub or the Ubl modifier NEDD8 to Glu40 (
). Subsequently, Gan et al. showed that MavC could catalyze monoubiquitination of Ube2N, a Ub-conjugating E2 enzyme essential for the synthesis of Lys63-linked polyubiquitin chains in the NFκB activation pathway (
) via an isopeptide crosslink between Gln40Ub and Lys92Ube2N. This atypical ubiquitination at Lys92 renders the active site of the E2 enzyme inaccessible for its catalytic function of mediating Ub transfer from the E1 enzyme to the next recipient in the ubiquitination transfer cascade, resulting in blockade of Lys63-linked polyubiquitin chain synthesis (
). But unlike the Cif deamidases, MavC seems to have evolved to target Ube2N through the acquisition of a novel insertion domain (INS domain) that is absent in the deamidases. It is this insertion (INS) domain that enables MavC to carry out atypical ubiquitination of Ube2N (
). The transglutaminase activity of MavC proceeds with, first, the formation of an acyl-enzyme thioester intermediate with Ub in which the carbonyl group of Gln40Ub is linked to the S-atom of the catalytic thiol (Cys74MavC) (Fig. 2B). Formation of this intermediate is accompanied by release of ammonia, aided by His231MavC acting as a proton donor to the leaving group. This intermediate is subsequently attacked by the ε−amino group of Lys92Ube2N resulting in an isopeptide cross-link between Ub and the E2 target. The same intermediate is also prone to a nucleophilic attack by water, especially in the absence of the amine nucleophile, resulting in the deamidation of the glutamine side chain (
), a complex often referred to as the charged E2 complex. The charging of E2 (denoted by Ube2N∼Ub) occurs when Ub is transferred from E1 in the context of the E1-E2-E3 transfer cascade. Since the dimerization interface does not overlap with MavC binding, the Ub-charged heterodimer could be the actual physiological target of the effector. In that case, the transglutamination reaction likely occurs in an intramolecular fashion between Ub and Ube2N while being covalently tethered via the active-site thioester linkage in Ube2N∼Ub (
). The intramolecular reaction improves the probability of the transamidation reaction over the futile hydrolysis of the thioester intermediate that would otherwise result in Ub deamidation, which may cause a broader cellular impact, since the deamidated Ub derivative is substantially less useful in cellular ubiquitination events of the host (
), allowing MavC to effectively engage its target amidst the host protein partners of Ube2N.
Remarkably, despite striking structural similarity with MavC, including identical catalytic residues and a similar INS domain (Fig. 2, A and C), MvcA catalyzes the removal of Ub from the ubiquitinated Ube2N (Ub-Ube2N), the product of the MavC-catalyzed modification. Cleavage of the isopeptide cross-link by MvcA leads to regeneration of native Ube2N in a reaction akin to cysteine-dependent deubiquitinase activity of eukaryotic DUBs (
). The thiol group of the catalytic cysteine of MvcA attacks the scissile isopeptide bond, forming a thioester intermediate with Ub as Ube2N leaves as the amine fragment (Fig. 2C). Hydrolysis of this intermediate releases Ub as the Q40E derivative. Thus, MvcA and MavC function similarly during the first step of their catalysis, forming the thioester intermediate with Ub accompanied by the departure of an amine group (ammonia in the case of MavC and Ube2N in the case of MvcA). The difference lies in the second step, wherein the MvcA catalysis involves water as the nucleophile in contrast to the Lys92Ube2N amine nucleophile in the MavC reaction (Fig. 2, B and C). In addition to the same catalytic triad, contact regions between MvcA and its substrates also mimic those in the MavC-substrate complexes (Fig. 2D), pointing to an evolutionary adaptation that enables two enzymes sharing a common fold and mechanistic features to catalyze opposite reactions.
The MavC/MvcA pair provides a remarkable example of temporal regulation by Legionella necessitated by a specific requirement for Ub attachment and removal. In the initial stages of infection, MavC ubiquitinates Ube2N and dampens NF-κB signaling (
). However, the regular catalytic activity of Ube2N leading to NF-κB activation is beneficial to long-term intracellular growth of L. pneumophila even though its inhibition is necessary for blocking immune response in the early phase of infection (
), giving MavC ample time to blunt the immune response of the host while the bacteria are trying to establish a replicative niche within the host cell. The MavC-MvcA pair of effectors illustrate an important example of the subtle and precise interplay of bacterial effectors with specific host posttranslational pathways while avoiding systemic effects on a broad array of cellular processes.
Another gene sharing the same locus with mavC and mvcA is lpg2149 (Fig. 2A) (
). The structures show that lpg2149 inhibits the enzymes by binding to a conserved structural element called the helical extension, thereby preventing Ub from binding (Fig. 2D). Thus, unlike MavC and MvcA, which have evolved to specifically inhibit Ube2N and restore it, respectively, at different time points during Legionella infection (
). Gan et al. have shown that in a laboratory setting, lpg2149 expresses only in the early exponential phase and not in the early stages of Legionella infection, suggesting that lpg2149 exerts its inhibitory effects only when the bacteria have started to replicate (
). Functioning as molecular switches, they cycle through two different nucleotide bound states to regulate protein–protein interactions necessary for vesicular trafficking events and other membrane-associated functions. The largely cytosolic, GDP-bound inactive Rab is activated by a guanine nucleotide exchange factor (GEF), which catalyzes the exchange of GDP for GTP to turn the protein to its active, membrane attached form. The GTP-bound Rab recruits specific protein partners to control docking and tethering steps between membrane compartments and cytoskeleton during vesicular trafficking events. The activated state of Rabs is temporally regulated by GTP-hydrolysis-activating proteins (GAPs), which promote GTP hydrolysis and return the Rab to its GDP-bound inactive form, which is subsequently extracted from the membrane by a guanine dissociation inhibitor (GDI) protein (
Rab GTPases exploit their similar overall fold and conserved residues for nucleotide binding and catalysis while using individual structural differences in key variable regions to interact with specific binding partners, such as the GEFs and GAPs (
). There are three essential recognition motifs in Rab variable regions: The P-loop (contacts the α and β-phosphates of the guanine nucleotide); Switch I (involved in Mg2+ coordination), and Switch II (consists of the DXXG motif that links binding of Mg2+ and the γ-phosphate of GTP). The inactive and active states of these GTPases are distinguished by the conformation of the switch loops (
), which along with the interswitch region, form interactions with almost all binding partners, including GEFs and GAPs. In the GTP-bound form, both Switch I and Switch II are held in place by interactions with the γ-phosphate group of GTP. Upon GTP hydrolysis, loss of these interactions and the release of the γ-phosphate group allow both switch regions to settle into their GDP-bound conformations.
One of the characteristic features of Legionella infection is the acquisition of an ER-like membrane coat on the LCV as the plasma-membrane-derived organelle matures into a phagosomal compartment supportive of bacterial replication (
). Among the several host proteins sequestered to the LCV is Rab1, a critical player during the initial stages of secretory pathways by promoting the transport and fusion of vesicles exiting the ER to the Golgi apparatus (
). The recruitment of Rab1 is essential for further maturation of the LCV to a more ER-like organelle. SidM, SidD, and LepB are a unique trio of L. pneumophila effectors known to modulate Rab1 function to recruit membranes from the ER and the pre-Golgi intermediate compartment to the LCV (
) (Fig. 3A). Neunuebel et al. showed that translocation of SidM, SidD, and LepB is temporally regulated. Levels of SidM are high immediately following infection, commensurate with the recruitment of Rab1 to the LCV. Two hours postinfection, the levels of SidM start to decline as the levels of SidD and LepB rise, both of which are required to release Rab1 from the maturing LCV back to the host cytosol (
SidM, also known as DrrA (defects in Rab1 recruitment protein A), features three distinct functional domains (Fig. 3A): an N-terminal adenylyl-transferase domain (ATase), a C-terminal lipid phosphatidylinositol-4-phosphate binding domain of SidM (P4M), and a central GEF domain (
). It initiates Rab1 activation and subsequent LCV localization by first displacing Rab1 from the Rab1-GDI complex, followed by catalyzing the GDP to GTP exchange (Fig. 3B; Step 1). As far as GEF activity is concerned, Rab1 substrate specificity for SidM stems from interactions with residues in the Switch I loop (Asp34 to Ile41). Crystal structures show substantial conformational reorganization in the Rab1 switch regions upon SidM binding (Fig. 4, A and B). Although the mode of Rab1 activation is similar to that observed in eukaryotic GEFs, the GEF domain of SidM is structurally distinct from the eukaryotic GEFs (
) (Fig. 3C). Conformational changes during eukaryotic GEF-catalyzed nucleotide exchange involve structural rearrangements within the switch regions, with a more pronounced change in Switch I (Fig. 4A). When a GEF binds to the Switch I loop of the Rab GTPase, it destabilizes the interaction of the GTPase with the phosphate and the Mg2+ ion, pulling Switch I into an open conformation (Fig. 4, A and B). This destabilization also displaces the conserved Tyr36 (or Phe in some GTPases) from its interaction with the guanine nucleobase while causing the P-loop to lose its interactions with the phosphate groups of the nucleotide, thus lowering the affinity for GDP even further (
). Interestingly, the Switch I region in the SidM-Rab1 complex also disengages from the main body of Rab1 and rotates to face the opposite direction compared with its conformation in the Rab1-GDP and Ypt1-GDI complex (
). This rearrangement causes the guanosine binding site in Rab1 to distort, displacing Tyr36 and pushing Ser25 in the P-loop into the nucleotide-binding pocket, which induces Rab1 to adopt a more open conformation comparable to the other known GEF-Rab1 complexes, facilitating GDP release (Fig. 4, A and B). SidM binding also affects the Switch II region, stabilizing it in a conformation reminiscent of a GTP-bound Rab1 (
), consequently locking Rab1 in its GTP-bound state. This AMPylation activity of SidM toward the GTP-bound Rab1 is nearly 270-fold higher than the GDP-bound form of Rab1, which implies that SidM preferentially targets active Rab1 (
) and AMPylation is preceded by the GEF function. SidM uses the classic GX11DXD motif in this reaction, where the aspartates coordinate the catalytic Mg2+ as seen in the E. coli glutamine synthetase-adenylyltransferase enzyme (GS-ATase) that catalyzes AMPylation of glutamine synthetase for regulating its activity (
) (Fig. 4C). Two additional aspartates (Asp150SidM and Asp249SidM) contribute to the binding of the Mg2+ ion, with Asp112SidM serving as a general base to promote nucleophilic attack by the phenolic OH group of Tyr77Rab1 (
) (Fig. 4C). Although the attachment makes little difference to the conformation of Rab1 and hence its GTP affinity, the modification of Tyr77 does prevent the Legionella GAP, LepB, or possibly host GAPs from binding to Rab1 prematurely (
). Thus, the GEF and ATase domains of SidM appear to function collaboratively to extend retention of active Rab1 on the LCV membrane, at the same time thwarting the access of host GAPs to the GTPase.
Overactivation and prolonged LCV retention of Rab1 may cause a drastic effect on host vesicular traffic events that rely on this critical ER-associated GTPase. SidD counteracts SidM-catalyzed AMPylation via its N-terminal enzymatic domain through deAMPylation activity (Fig. 3B; Step 3) while a putative membrane targeting segment in the C-terminal region may assist in the LCV localization (Fig. 3A) (
). The catalytic pocket of SidD features a negatively charged region with two Mg2+ binding sites such as the binuclear metal-binding sites in PPMs (Fig. 5A). Five catalytic aspartates (Asp91, Asp92, Asp110, Asp112, and Asp326) and a binuclear bridging water, that acts as the nucleophile, coordinate the two metal ions. Hydrolysis of the adenylyl-O-tyrosyl linkage (AMP-Tyr77) releases AMP and restores Tyr77Rab1 to its unmodified form (Fig. 5A).
This deAMPylation frees up Rab1 from its continuous state of activation, thus allowing LepB to trigger GTP hydrolysis through its GAP activity (Fig. 3B; Step 4), leading to inactivation of Rab1 and their subsequent removal from the LCV. LepB is mechanistically homologous to the eukaryotic Rho/Ras-GAP rather than a Rab-GAP. Traditional Rab-GAPs (the so-called TBC GAPs) and some bacterial GAPs (of the VirA/EspG family) feature a catalytic glutamine finger in addition to the canonical arginine finger (Arg finger) (
). LepB, however, features a glutamate residue (Glu449LepB) instead of the glutamine finger. Glu449LepB occupies a structural position in apo-LepB equivalent to the canonical glutamine in the traditional Rab-GAPs (
). The catalytic Arg444 finger of LepB mediates a two-pronged polar interaction with the β-and γ-phosphates of the GTP (Fig. 5B). Upon binding to Rab1, Glu449LepB undergoes a pronounced movement that triggers the side chain of Gln70Rab1 to flip toward the γ-phosphate of GTP. As a result, Gln70Rab1 adopts a similar position with respect to the γ-phosphate and a water molecule to what has previously been observed with the catalytic in-trans Gln finger (contributed by the substrate GTPase) in Ras-GAPs (Fig. 5B). The side chain of Gln70Rab1 would orient the water molecule for nucleophilic attack on the γ-phosphate center of GTP to facilitate its hydrolysis much like the catalytic in-cis Gln finger (contributed by the GAP) of TBC-like Rab GAPs (
). Perhaps, a Ras GAP-like mechanism confers certain advantages for better kinetic control of Rab1 dynamics over the host GAP.
AnkX and Lem3: parallel modulators of Rab1 function
The AnkX and Lem3 effector pair represents a sophisticated example of functional redundancy used by Legionella to subvert the function of Rab GTPases and facilitate LCV maturation (Fig. 6A). A time-resolved analysis of Legionella effectors that modulate Rab1 function showed that these effectors differ in the specific timing of increase in their levels, which agrees with the role played by them in recruiting Rab1. Allombert et al. demonstrated that AnkX levels begin to rise only after SidM has been translocated and has had a chance to release Rab1 from the Rab1-GDI complex (
), most commonly in bacterial proteins, especially those involved in targeting host GTPases. The only known human protein containing this domain is the HypE protein, known to regulate protein stress response (
). The characteristic structural elements defining the Fic family of proteins are the presence of a bundle of six α-helices and a loop region with a highly conserved motif, HXFX(D/E)(A/G)N(G/K)R, the so-called Fic motif. The Fic domain of AnkX (Fig. 6A) catalyzes phosphocholination of Rab1 using CDP-choline, a modification that is ultimately reversed by Lem3 (
). It resides within the CMP (cytidine monophosphate)-binding domain of AnkX and is unique because, unlike other Fic-domain-containing effectors, the Fic domain of AnkX transfers the phosphocholine moiety and not the nucleotide monophosphate. The C-terminal region of AnkX also contains ankyrin repeat domains (ARDs) and a PI3/4P-binding domain (Fig. 6A) (
). Akin to their role in eukaryotes, the ARDs in AnkX facilitate protein–protein interaction as observed between the ankyrin repeats 10 to 13 with the C-terminal of Rab1b in a crystal structure of the AnkX-Rab1b complex (
). This domain is divided into two subdomains separated by a unique 70-residue-long insert that masks the traditional binding site for GTPases on the Fic domain (Fig. 6, A and B). Phe107, Ile109, and Asp265 in the active site of AnkX sterically occlude the base of the CMP substrate from binding in the same orientation as seen with nucleotide substrates in other Fic enzymes. Instead, the choline group nestles in this region while the cytidine group stacks against Tyr41 of the CMP-binding domain, forcing a flipped orientation of the bound CMP moiety compared with the nucleotide-binding in AMPylating Fics. Thus, AnkX behaves as a phosphocholine transferase instead of a nucleotide transferase. Upon binding to AnkX, Switch II of Rab1b undergoes a significant structural rearrangement when Phe143AnkX sticks into a hydrophobic pocket formed between Switch II and helix-3 of Rab1b. This encroachment by Phe143AnkX displaces Tyr78Rab1b from its highly conserved position in this hydrophobic cavity (Fig. 6C) (
). The displacement of Tyr78Rab1b causes local unfolding of the Switch II region and the residues adjacent to it (Fig. 6C). As a result, the otherwise structurally restricted Ser76Rab1 of Switch II can now reach into the active site of AnkX for phosphocholination. The catalytic His229AnkX of the Fic motif acts as a general base in deprotonating the OH group of Ser76Rab1, promoting a nucleophilic attack on the β-P center of the nucleotide (Fig. 6D). The Asp233 residue of the Fic motif positions the catalytic Mg2+, while the Asn235 and Arg237 play a critical role in interactions with the phosphocholine moiety.
The Legionella effector Lem3 (Fig. 6A), a phosphodiesterase whose structure is yet to be determined, can reverse the effects of AnkX by removing the phosphocholine group from Ser76 of Rab1 (
), making the GTPase accessible to other Legionella effectors such as LepB. Although phosphocholination on Ser76Rab1 does not affect the GEF activity of SidM, it does profoundly affect the adenylation of Tyr77Rab1, indicating that these modifications can be mutually exclusive. Ser76Rab1 modification by AnkX also negatively impacts the interactions of the modified Rab1 with its GDI, which are restored upon Lem3-catalyzed dephosphocholination of Rab1 (
) pointing to the existence of yet-to-be identified effector selective for the Rab35 modification. The importance of the AnkX/Lem3 pair in the hijacking of Rab1 is not as well understood as the role of the SidM/SidD pair. Nevertheless, given the importance of phosphocholination in the modulation of the eukaryotic immune system and the opposing activities of AnkX and Lem3, they are essential for the bacteria during infection.
The eight different activities mediated by seven Legionella effectors, SidM, AnkX, SidD, LepB, Lem3, SetA, and LidA (SetA and LidA have not been discussed here), known so far to modulate Rab1 function, point to the importance of the GTPase in the intracellular lifecycle of the bacteria. Upon activation, Rab1 interacts with other proteins, such as p115 or GM130, in order to guide ER-derived vesicles to fuse with the Golgi apparatus (
). Rab1 seems to play a similar role of docking ER-derived vesicles with the LCV during Legionella infection. Recruitment of Rab1 may also contribute to bypassing the default maturation of such organelles along the endocytic pathway for lysosomal degradation. SidM is translocated to the host cell within minutes of Legionella infection coinciding with Rab1 recruitment to the LCV (
). It seems that once the GDP to GTP exchange occurs, mediated by SidM GEF domain, other effectors such as SetA, AnkX as well as the ATase domain of SidM are able to carry out specific PTMs targeting the Switch II residues Thr75, Ser76, and Tyr77, respectively (
). These mutually exclusive modifications might trap the activated Rab1 on the LCV membrane by preventing deactivation or dissociation by the eukaryotic GAPs and GDIs. On the other hand, detection of effectors such as SidD, LepB, and Lem3 in the later stages of infection suggests that these effectors temporarily control the recovery of Rab1 from the membrane by removing the modifications and deactivating the GTPase. Further studies are needed to reveal whether these effectors are specific for Rab1 alone or if they are responsible for modulating the broader landscape of GTPases, as indicated by their promiscuity toward several Rab proteins observed in vitro (
), belong to yet another unique group of bacterial effectors that target several host proteins associated with the ER, including ER-Rabs (Rab1, for example) and reticulon, through phosphoribosyl-linked (PR-Ub) ubiquitination, a type of PTM for which there are no parallels in eukaryotes so far (
). In contrast to the archetypical E1-E2-E3 three-enzyme system of eukaryotes, the SidE effectors use an all-in-one ubiquitinating machinery that utilizes NAD+, instead of ATP, to target serine residues of host proteins through PR-linked ubiquitination via Arg42 of Ub. This orthogonal mode of ubiquitination bypasses the eukaryotic machinery and produces a linker resistant to host DUBs (
SidE family members are large, functionally redundant proteins that share more than 40% sequence identity. These modular proteins function via the concerted action of four domains (Fig. 7A): a deubiquitinase domain (DUB), a phosphodiesterase domain (PDE), a mono-ADP-ribosyltransferase domain (mART), and a C-terminal coiled-coil domain (CC). Biochemical studies have shown that the DUB domain is not essential for the ubiquitinating activity of these proteins (
). SdeA likely uses its DUB domain to produce free Ub at the LCV to make it readily available for its ligase machinery. Dong et al. have shown that the CC domain interacts with parts of the Dot/Icm translocation machinery and may thus be required to inject the SidE ligases into the host cytosol (
). The mART domain, which features a characteristic RSE motif typically found in arginine-targeting mART enzymes, activates Ub by transferring the ADP-ribose group from NAD+ on to Arg42Ub, forming ADP-ribosylated Ub (Ub-ADPR) as an intermediate. Subsequently, the PDE domain catalyzes a phospho-transferase-like reaction where the PR-Ub of Ub-ADPR is transferred to the hydroxyl group of a serine residue of the host protein, accompanied by the release of AMP. Several structures of different constructs of these proteins (
) (Table 1), along with biochemical studies, have allowed elucidation of some critical aspects of the various catalytic steps involved in recognition of Ub by the mART domain and those involved in recognition of Ub-ADPR by the PDE domain. The initial discovery of the five substrates: Rtn4 (reticulon 4) Rab1a, Rab6a, Rab30, and Rab33b (
), was quickly followed by the understanding that the SidE proteins are tolerant of any serine that is a part of an unstructured/flexible region, provided it can be accommodated in the PDE active site (
) at their interface. The nucleotide cofactor binds at this pocket in a strained conformation (Fig. 7B), which facilitates the departure of nicotinamide (Nic) prior to the attack of the Arg side chain on the resulting oxocarbenium center of the Nic-bearing ribose of NAD+ (
). This reaction is similar to the one performed by bacterial mono-ADP-ribosyltransferase toxins, such as the Iota toxin from Clostridium perfringens, which utilizes the characteristic RSE active-site motif to ADP-ribosylate Arg177 of actin via the so-called SN1 strain alleviation mechanism of ADP-ribosylation (
). The conserved Arg and the Ser residues of SdeA help position and stabilize the strained conformation of the nucleotide in the active site, while the first Glu residue (Glu860 in SdeA) of the EXE dyad promotes the nucleophilic attack by the substrate Arg (Fig. 7B) and the second Glu stabilizes the oxocarbenium ion. The conformation of NAD+ observed in SidE mART crystal structures and the placement of critical residues of the RSE motif are consistent with the strain alleviation model of ADP-ribosylation in this enzyme.
The PDE domain of SdeA bears distinct sequence homology with other bacterial phosphodiesterases, such as the PDE domain of the well-known cyclic di-3′,5′-GMP phosphodiesterase PA4781 from Pseudomonas aeruginosa (
), with which it shares 23% sequence similarity. The PDE domain of SdeA has the same three catalytic residues, His277, Glu340, and His407, conserved in bacterial PDEs. Mechanistically, the reaction proceeds through covalent catalysis via the formation of a phospho-His-like intermediate (Fig. 7C) (
). This attack results in the formation of a transient His277-PR-Ub intermediate and AMP release, enabled by proton donation by His407 to the leaving group. The His277-PR-Ub intermediate then reacts with the serine OH group of the substrate leading ultimately to the transfer of phosphoribosylated-ubiquitin (PR-Ub) onto the host protein (
). From the position of His407 relative to the APDR moiety, it appears to be the most likely candidate for serving the role of the general base in activating the OH group of the target serine (Fig. 7C). Thus, despite the similarity of the PDE domain of SdeA with the bacterial phosphodiesterases, the catalytic motif in the SdeA PDE domain catalyzes a (substituted) phospho-transfer to a serine residue in the ubiquitination reaction instead of water, which would result in hydrolysis of the phosphodiester bond. Incidentally, the PDE domain of the SdeA can also catalyze phospho-transfer to water, ensuing from water attacking the phospho-His intermediate, resulting in the PR-Ub hydrolysis product (
). The biological significance of this side reaction in the context of L. pneumophila infection remains to be determined, as PR-Ub can be toxic to host cells when the SidE members are ectopically expressed in mammalian cells.
Even though the SidE proteins are essential for Legionella replication in eukaryotes, unchecked activity of these proteins may result in the accumulation of free PR-Ub that would inhibit the host ubiquitination machinery (
). Ubiquitination of substrates such as Rab1 by members of the SidE family may also affect the activity of other effectors such as SidM and AnkX that need Rab1. The PR-ubiquitination of host proteins triggered by the SidE family members is regulated by two effectors: SidJ and SdeD. sidJ resides in the same genetic locus as sdeC, sdeB, and sdeA, whereas sdeD is located much further down in a distinct genomic locus (Fig. 7A) (
). Studies have shown that SidJ can inactivate SidE proteins by directly inhibiting the mART activity, thereby shutting off the ubiquitination reaction, whereas SdeD, also known as DupB (deubiquitinase of phosphor-ribosyl linked ubiquitination), and its paralog DupA, can counterbalance the activity of the SidE members by acting on the phosphoribosylated host proteins (
SidJ is an 873 amino-acid protein expressed in the later stages of Legionella infection. Inhibition of SidE proteins by SidJ is both temporally regulated and spatially restricted as it requires the host calmodulin (CaM) for its activity (
). Compartmentalization of SidJ's activity in the host cytosol prevents premature inactivation of SidE effectors before being injected into the host cell. Biochemical studies have revealed that the association of SidJ with CaM stabilizes the active conformation of the effector (
) revealed SidJ to be a pseudokinase that utilizes ATP to catalyze the polyglutamylation of SdeA (and other SidE members). CaM binding via the IQ motif located at its C-terminal end of SidJ presumably opens the kinase-like active site of SidJ for ATP. In the first step of catalysis, SidJ uses ATP to acyl-adenylate the carboxylate group of SdeA Glu860 with the release of pyrophosphate (Fig. 8A), a reaction akin to the acyl-adenylation step of the reaction catalyzed by aminoacyl-tRNA synthetases or activation of Ub by E1. The activated carbonyl of this unstable intermediate is then attacked by the amino group of a free glutamate residue, leading to glutamylation of the SdeA catalytic residue via an isopeptide linkage and the release of AMP (Fig. 8A). This second step likely involves another essential region in SidJ, named the “migrated” nucleotide-binding site (
), which may help in the optimal positioning of the acyl-adenylated SdeA and the free glutamate to take the reaction to completion. A recent cryo-EM reconstruction of the SidJ:CaM:SdeA intermediate complex revealed that while the pseudokinase active site is responsible for the acyl-adenylation reaction, it is the migrated nucleotide binding pocket that carries out the glutamylation reaction, with Arg522SidJ playing the crucial role of positioning the donor Glu to attack the acyl-adenylate intermediate and subsequent formation of the Glu-Glu isopeptide bond on SdeA (
) along with a highly conserved catalytic core made up of His67, Glu126, and His189 (His277, Glu340, and His407 in SdeA-PDE domain) (Fig. 7A). SdeD acts as a hydrolase to catalyze the hydrolysis of PR-ubiquitinated substrates, presumably proceeding via the formation of a histidine-based phosphoramidate intermediate such as the SdeA-PDE mechanism (
) (Fig. 8B). SdeD can remove the AMP moiety from Ub-ADPR, like the hydrolysis reaction catalyzed by the PDE domain of SidE members. However, it cannot transfer the PR-Ub onto target proteins, likely due to the structural differences that alter the conformation and accessibility to the catalytic center in the two effectors. The SdeD loop comprising residues 26 to 48 is missing in SdeA, and conversely, the SdeA loop comprising residues 465 to 513 does not exist in SdeD. It is possible that because of this structural difference, in SdeD, only water can act as the ubiquitin acceptor instead of the hydroxyl group of a serine residue (Fig. 8B) (
). The SdeA-PDE domain and SdeD pair is another example, much like the MavC-MvcA pair discussed above, of Legionella adopting the same catalytic machinery to mediate opposite reactions.
The ligase activity of SidE members results in serine ubiquitination of several structurally diverse substrates, the consequences of which affect a variety of cellular processes, ranging from autophagy and vesicular trafficking to tubular ER dynamics and inhibition of protein synthesis. It appears that the seemingly indiscriminate nature of the SidE proteins contributes to the virulence of the pathogen and enables the establishment of the LCV. For example, by PR-ubiquitinating Rab6a and Rab33b, these effectors modulate the Golgi-to-ER retrograde trafficking and prevent the formation of autophagosomes ((
) and all the references therein), an essential first step in autophagy. Similarly, by ubiquitinating Rag GTPases, the SidE effectors inactivate mammalian target of rapamycin complex 1 (mTORC1) to inhibit host protein synthesis (
), effectively allowing the pathogen to consume the host amino acids as nutrients for its survival. It is believed that PR-ubiquitination by the SidE members is most likely regulated at the early stages of infection by SdeD to prevent unchecked accumulation of Ub-ADPR and depletion of cellular Ub (
). However, despite understanding the biochemistry of the reactions mediated by these effectors, it is not yet clear whether these activities are exerted simultaneously or if they are a cascading consequence of one another. Further studies are required to fully appreciate the spatiotemporal regulation of cellular processes modulated by them.
Other examples of effector–effector pairs
Identification of effector–effector interactions is often the consequence of studying the function of an individual effector. However, recently, studies have systematically explored the interactions between effector pairs by combining a gain-of-function genetic screen in yeast with cellular and biochemical approaches. Urbanus et al. (
) successfully identified novel, direct pairwise effector–effector interactions by carrying out a comprehensive analysis of all possible pairwise interactions between the 330 effectors secreted by L. pneumophila. Pairwise interactions that merit a mention even though there is not enough structural information available at present to describe their mechanism of action in detail are (1) inactivation of RavJ by LegL1 (2), polyubiquitination of SidH by LubX, and (3) inhibition of MavQ by SidP.
RavJ, a small Legionella effector, rearranges the actin cytoskeleton leading to the accumulation of more F-actin on the plasma membrane (
). It consists of two domains: An N-terminal papain-like cysteine protease domain and a C-terminal domain that interacts with various cytoskeleton-associated components of the eukaryotic septin and elongator complexes (
). For example, YopT, an effector protein from Yersinia, functions as a cysteine protease to cleave Rho GTPases and inhibits phagocytosis by disrupting the actin cytoskeleton. The N-terminal domain of RavJ has the requisite active-site elements that can potentially disrupt the actin cytoskeleton such as YopT, although the specific catalytic activity remains to be demonstrated. LegL1 is a leucine-rich repeat (LRR)-containing Legionella effector that binds RavJ (
) and inhibits its activity by blocking the putative active site. It is unclear at which point, postinfection, RavJ and LegL1 are employed during the intracellular life cycle of L. pneumophila and temporal regulation mediated by the effector–effector interaction.
Historically, it is the discovery of the SidH-LubX effector pair that first brought to light that Legionella effectors can regulate the activity of each other. LubX is a U-box containing bacterial E3 ligase (
) mediates the physical interaction between the two effectors, while in vitro ubiquitination assays and structural studies showed that the U-box1 is responsible for the polyubiquitination property of the effector via recruitment of the host Ub-conjugating E2 enzyme, Ube2d3 (
)). The presence of these motifs in SidH and its appearance in the early stages of Legionella infection indicate that perhaps SidH functions to regulate cell surface dynamics via these motifs. Further structure–function studies are required to understand the functional importance of SidH and its inhibition by LubX.
The partnering between a phosphatase (SidP) and a kinase (MavQ) is involved in phosphatidylinositol polyphosphate (PIP) modulation. The amino acid sequence of SidP bears no homology to any known phosphatidylinositol phosphatase (PI phosphatase), while the MavQ sequence is an atypical kinase. Biochemically, SidP is similar to the CX5R-based PI phosphatases that belong to the myotubularin family (
) but is able to hydrolyze PI(3)P and PI(3,5)P2. SidP inhibits MavQ by binding to its C-terminal domain, indicating that the PI phosphatase activity, which resides in the N-terminal domain of SidP, is distinct from its role of MavQ inhibition. A recent study by Hsieh et al. (
) showed that MavQ and SidP work alongside each other, the former adding a phosphate to PI(3)P moieties and the latter removing phosphates from higher-order phospholipids or even converting PI(3)P to simple PI entities and altering the lipid composition of the host membrane, in the context of infection.
Legionella effectors and cellular homeostasis
L. pneumophila allocates about 10% of its protein-coding ability toward functions that require direct interactions with host cellular processes. One of the most remarkable features of Legionella is its capacity to both temporally and spatially regulate the dynamics of its effectors during infection. Three regulatory systems (the PmrAB two-component system, the CpxRA two-component system, and the LetAS-RsmYZ-CsrA regulatory cascade) directly or indirectly regulate the expression of several effector-encoding genes (
). These regulatory systems work cohesively to allow the pathogen to enter the host cell, adapt to the new environment, and regulate host mechanisms that promote multiplication and survival within the cell. For example, Legionella triggers the nuclear localization of NF-κB in macrophages to positively upregulate antiapoptotic genes in a Dot/Icm-dependent manner to support intracellular bacterial growth (
). Another essential cellular machinery targeted by L. pneumophila is the host amino acid metabolism regulated by mTORC1. As alluded to before, the pathogen frees up host amino acids for its consumption via a concerted albeit temporally regulated action of SidE and Lgt family of effectors (
). Another well-studied example is the co-option of the host ubiquitin network by Legionella. On the one hand, the translocated effectors such as MavC and SidE proteins ligate Ub onto host proteins, inactivating or altering their function in the process (Fig. 9). On the other hand, ubiquitination is also used as a targeting signal for other effectors, as exemplified by the negative regulation of PR-ubiquitination by SidJ.
It is not surprising that there is a constant battle between pathogens and their hosts to develop means to adapt and counter-adapt during evolution. In contrast to the high conservation of the secretion system itself among different Legionella species, the effector arsenal is quite varied, suggesting that Legionella's genomic flexibility is because of its coevolution with numerous protists species that belong to the phyla Amoebozoa and Percolozoa (
). Legionella effectors can regulate a myriad of cellular functions (Fig. 9) because the bacteria may have acquired numerous genes from a range of primitive eukaryotes through horizontal gene transfer during evolution (
). These building blocks can be rearranged to generate new domain/motif combinations, contributing to the evolution of this unexpectedly large arsenal of functionally diverse and seemingly redundant effectors. Despite a high rate of evolution over long periods, the amino acids involved in protein–protein interaction have undergone positive selection pressure (
). This suggests that sometimes despite low overall sequence homology between bacterial proteins and their eukaryotic counterparts, the residues at protein interfaces are well conserved, pointing to an essential role for bacterial effectors in interfering with host pathways (Fig. 9).
Many L. pneumophila effectors have been identified using various genetic and biochemical techniques. However, it is still unclear why the pathogen requires so many effectors for its survival within its hosts. Perhaps it is to ensure sustained intracellular replication and is simply an indication of the numerous eukaryotic pathways regulated by the bacteria (Fig. 9), making the study of these effectors quite complex, albeit exciting. Our understanding of bacterial virulence and its impact on host signaling is hampered by the built-in functional redundancy exhibited by many of these effectors. The recent discovery of functional interplay between bacterial proteins to modulate each other has given rise to the concept of effector–effector synergism that adds another intriguing dimension to the various modes of survival adopted by Legionella. Despite the diverse nature of the different eukaryotic functions targeted by these effectors, one emerging theme is the existence of a yin and yang type of mechanism. It would not be unreasonable to assume that these principles are co-opted by pathogens in general and not just by bacteria. Future structure–function studies will not only help in evaluating the role of pathogenic effectors but will also aid us in understanding how essential eukaryotic cellular homeostasis is maintained.
Conflict of interest
The authors declare that they have no conflict of interest with the contents of this article.
S. I. and C. D. conceptualization; S. I. writing–original draft; S. I. and C. D. writing–review and editing.
Funding and additional information
This work was funded by the National Institutes of Health Grant R01GM126296 (to C. D.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Salmonella interactions with host cells: Type III secretion at work.