Exploitation of the Ubiquitin System by Invading Bacteria

Abstract

A variety of bacterial intracellular pathogens target the host cell ubiquitin system during invasion, a process that involves transient but fundamental changes in the actin cytoskeleton and plasma membrane. These changes are induced by bacterial proteins, which can be surface-associated, secreted or injected directly into the host cell. Here the invasion strategies of two extensively studied intracellular bacteria, Salmonella enterica serovar Typhimurium and Listeria monocytogenes, are used to illustrate some of the diverse ways by which bacterial pathogens intersect the host cell ubiquitin pathway.

Introduction

Many bacterial pathogens can survive and replicate within eukaryotic host cells as well as within the extracellular environment. These facultative intracellular bacteria include important human pathogens such as; Salmonella enterica (gastroenteritis and typhoid fever), Listeria monocytogenes (listeriosis), Yersinia pestis (plague), Y. enterocolitica and Y. pseudotuberculosis (gastroenteritis), Mycobacterium tuberculosis (TB) and Legionella pneumophila (Legionnaire’s disease) amongst others. All of these bacteria can be internalized into professional phagocytes, including macrophages and dendritic cells, but some have also evolved active mechanisms to invade non-phagocytic host cells such as epithelial cells. This review focuses on the invasion mechanisms of L. monocytogenes and S. enterica serovar Typhimurium (S. Typhimurium), which invade nonphagocytic mammalian cells via strikingly different processes. L. monocytogenes uses the “zipper mechanism”, characterized by the binding of proteins on the surface of the bacteria to receptors on the surface of the host cell. In contrast, Salmonella can invade cells in the absence of specific adhesion via a process known as the “trigger mechanism”. Despite the inherent differences, the host cell ubiquitin (Ub) system is involved at multiple steps in both invasion processes. This serves as an excellent illustration of the fact that, although prokaryotes do not contain Ub, many pathogens and symbionts have developed ways to exploit this complex eukaryotic system in order to evade or utilize host cell processes.

Ubiquitination (or ubiquitylation) is a reversible, post-translational modification found in all eukaryotic cells where it plays central roles in protein degradation, membrane trafficking, signal transduction and host defense. Target proteins are modified by the covalent ligation of ubiquitin (Ub), a highly conserved 8.6 kDa protein, by a variety of linkages. Ub moieties can be added singly (mono-Ub) or in the form of chains (poly-Ub) or a combination of both. In addition, poly-Ub chains can be linear or branched, depending on whether linkage is via the same or different lysine residues in each Ub molecule. The ultimate fate of the ubiquitinated target protein is determined by the topology and length of Ub chains (1). Ub chains linked through Lys48 lead to degradation of the target via the proteasome, whereas Lys63 chains are involved in signaling and mono- or multi-ubiquitination is associated with endocytosis, sorting and other cellular pathways. Recognition of the mono-, multi-, or polyubiquitinated proteins is mediated by a diverse group of proteins containing Ub-binding domains (UBDs).

The process of ubiquitination involves the sequential activity of three classes of enzymes, designated E1, E2 and E3. In the first step the Ub activating enzyme, or E1, charges Ub in an ATP-dependent manner to form an E1-Ub thioester intermediate. The activated Ub is then transferred by a transthiolation reaction to the reactive cysteine residue of a specific E2 conjugating enzyme. Finally, an E3 Ub ligase mediates the transfer of Ub from the E2 enzyme to the target protein. Almost all E3 ligases are members of either the RING (Really Interesting New Gene), U-Box or HECT (homologous to E6-AP carboxy terminus) families, which are distinct in sequence, structure and catalytic properties [for review see (2)]. Ubiquitin-protein bonds are cleaved by a large group of proteases known as de-ubiquitinating enzymes (DUBs).

Bacterial pathogens intersect the Ub pathway at multiple points. Binding of bacterial adhesins, such as the L. moncytogenes invasins, to receptors on the host cell surface can activate signaling pathways in the host cell that involve Ub. Alternatively, many gram-negative bacterial pathogens and symbionts transfer bacterial proteins directly into host cells via specialized molecular machines, known as type III, IV or VI secretion systems (T3SS, T4SS or T6SS respectively). Proteins delivered by these systems, collectively known as effectors, often act in a coordinate fashion in order to exquisitely manipulate the host cell. Many of these bacterially encoded effectors “mimic” eukaryotic protein activities without having any significant homology at the amino acid level. Included amongst this group is the NEL (Novel E3 Ligase) family of autoregulated E3 ligases, which are translocated into host cells via T3SSs (3, 4). Although NEL proteins have no sequence or structural homology with eukaryotic E3 ligases other bacterial effectors do have structural similarity to eukaryotic ligases, such as the Salmonella HECT-like ligase SopA (5). Other bacterial effectors mimic deubiquitinases or can themselves act as substrates for host Ub-ligases (for revew see (6)).

Listeria invasion utilizes the eukaryotic Ub pathway

L. monocytogenes is a food-borne pathogen that is the causative agent of listeriosis, a serious invasive disease that primarily affects pregnant women, newborns and immunocompromised individuals (7). Two distinct pathways mediate L. monocytogenes invasion each of which is initiated by the binding of a bacterial surface protein, either internalin A (InlA) or internalin (InlB), to receptors on the surface of the host cell. In fact, either internalin alone can mediate attachment and internalization of Listeria, or internalin-coated beads, as long as the appropriate receptor is present (8, 9). Both pathways involve reorganization of the F-actin cytoskeleton, tyrosine phosphorylation and ubiquitination and are at least partially dependent on clathrin, although the signaling pathways involved are quite distinct (Figures 1 and 2) (10–13).

Figure 1. InlA mediated invasion by Listeria.

1) InlA interacts with E-cadherin. 2) Activated Src kinase phosphorylates E-cadherin in the juxtamembrane domain causing displacement if p120. 3) Recruitment of the E3 ligase Hakai leads to polyubiquitination of E-cadherin. 4) Recruitment of clathrin and bacterial internalization.

Figure 2. InlB mediated invasion by Listeria.

1) Ligand mediated dimerization of Met and autophosphorylation. 2) Recruitment of Cbl monoubiquitylation of Met. 3) Recruitment of endocytic machinery and internalization of Listeria

InlA-mediated invasion

The surface-bound protein InlA binds to E-cadherin, a single-pass transmembrane protein that functions as a Ca²⁺-dependent, homophilic cell–cell adhesion molecule that forms a physical link between the cell membranes of adjacent cells (14). The cytoplasmic tail of E-cadherin interacts directly with two catenins, p120- and β-catenin, initiating the formation of the cadherin–catenin complex, which also includes α-catenin and the actin-binding protein EPLIN (15–18). Internalization of E-cadherin is important for maintaining cellular homeostasis but it is inhibited when p120 is bound to the juxtamembrane (JMD) domain of E-cadherin since this prevents association of the E3 Ub ligase Hakai and components of the endocytic machinery (19, 20).

In epithelial cells the E-cadherin complex is endocytosed following activation of the tyrosine kinase Src, which induces tyrosine phosphorylation and ubiquitination of E-cadherin (19). Internalized E-cadherin is either trafficked to late endosomes and lysosomes for degradation (21) or recycled back to the cell surface with the exocyst (22). Ubiquitination of E-cadherin at the cell surface is dependent on Hakai, which interacts with E-cadherin in a tyrosine phosphorylation-dependent manner (19). Similarly, InlA-mediated internalization of Listeria induces sequential Src-dependent phosphorylation and Hakai-dependent ubiquitination of E-cadherin (10, 23), ultimately resulting in an accumulation of clathrin at the site of entry (10). Intriguingly InlA also binds to another cell surface pool of E-cadherin that co-localizes with caveolin in cholesterol-rich domains, however, although this binding can induce invasion it has no requirement for tyrosine phosphorylation or ubiquitination of E-cadherin (10).

InlB-mediated invasion

The host cell receptor for InlB is the receptor tyrosine kinase (RTK) Met (24) that regulates epithelial remodeling, dispersal, and invasion (25). This disulfide-linked heterodimer has an extracellular region composed of a Sema domain, a cysteine-rich domain, and four immunoglobulin-like domains (26). InlB interacts tightly with the first immunoglobulin-like domain of the Met stalk, a domain which does not bind HGF and second contact between InlB and the Met Sema domain locks the receptor in a rigid, signaling competent conformation (27). However, even though InlB is not a structural mimic of HGF, binding of either ligand to Met induces strikingly similar events including; receptor dimerization and autophosphorylation and downstream signaling events involving Grb2, Gab1, phosphatidylinositol 3-kinase and MAP kinase (MAPK) (28–32). Indeed, HGF can even substitute for InlB in inducing bacterial uptake (30). One possible distinguishing feature is the role of CD44, the hyaluronic acid transmembrane receptor, which is critical in HGF induced, but not InlB induced, Met pathways (33).

Acute stimulation of the Met receptor leads to receptor internalization into clathrin-coated vesicles and its down-regulation. This process requires the E3 Ub ligase Cbl, which targets Met for ubiquitination and degradation (34). Cbl binds to Met via a DpYR motif in the cytoplasmic domain of the RTK(35). Binding of InlB induces the Cbl-dependent monoubiquitination and endocytosis of Met and over-expression or down-regulation of Cbl, increases or decreases bacterial invasion, respectively (12). Thus Cbl has an essential role in InlB-mediated invasion and this is dependent on the endocytic machinery since RNA interference-mediated knock-down of major components of the endocytic machinery (clathrin, dynamin, eps15, Grb2, CIN85, CD2AP, cortactin and Hrs) inhibit bacterial entry (12).

Although the primary receptor for InlB is Met several other cell surface proteins, including the RTK gC1qR (receptor for complement component C1q) (36) and glycosaminoglycans (37) have been shown to act as co-receptors during bacterial entry. Binding of InlB to gC1qR stimulates tyrosine phosphorylation of the adaptor proteins Gab1, Cbl and Shc and activation of phosphatidyl inositol 3-kinase (36). This interaction is specific and invasion can be inhibited by addition of anti-gC1qR antibodies or C1q (36). In fact, gC1qR interacts with and facilitates entry of numerous microbial and parasitic pathogens, including Staphylococcus aureus as well as Plasmodium falciparum and HIV (38). How gC1q-R transduces signals into the cell is unclear since it has no transmembrane domain, and it is likely that interactions with Met, or an alternative membrane bound protein, are required. One potential candidate is β1-integrin, an interesting possibility since β1-and β3-integrins have been shown to be required for efficient InlB-mediated invasion (39, 40).

The bacterial toxin Listeriolysin O is down-modulated by ubiquitination

Many Gram-positive bacterial pathogens secrete water-soluble toxins, or cytolysins, that bind to target cell membranes and form multi-molecular transmembrane pores (for review see (41)). Apart from their pore forming ability cytolysins have other activities that can affect a variety of functions in host cells. Listeriolysin O (LLO), a cholesterol-dependent pore-forming toxin, plays an essential role in the escape of L. monocytogenes from the phagosome into the host cell cytoplasm. LLO also induces NF-κB signaling in a pore-forming-independent manner, by inducing dephosphorylation of histone H3 and deacetylation of histone H4 (42), and impairs the sumoylation pathway (43). These activities are shared by a number of bacterial toxins (42, 43) and it is apparent that, if not regulated appropriately, they have the potential to do more harm than good for the pathogen. This is certainly true for LLO, which is ubiquitinated and targeted for proteolysis during replication of cytosolic L. monocytogenes. Strains overproducing LLO or producing LLO with a stabilizing mutation have increased toxicity for host cells and have a significant virulence defect (44, 45).

Salmonella triggers the Ub pathway

S. Typhimurium is a common cause of gastroenteritis in humans The ability to invade nonphagocytic cells, such as enterocytes in the intestinal epithelium, is an essential virulence determinant and is dependent on a T3SS, T3SS1. A second, functionally distinct T3SS, T3SS2, is required for post-invasion establishment of the replicative niche, a modified phagosome known as the Salmonella-containing vacuole (SCV) [for review see (46)]. Invasion is initiated when a cohort of T3SS1 effectors is translocated across the plasma membrane and into the cytosol of the host cell. The coordinated activity of at least three of these effectors leads to localized and transient changes in the actin cytoskeleton resulting in the formation of plasma membrane ruffles and engulfment of the bacteria [for review see (47)]. Several Salmonella effectors, from both T3SS1 and T3SS2, intersect the host cell Ub-pathway, although here the focus will be on T3SS1 effectors because of their role in invasion (Figure 3).

Figure 3. Salmonella T3SS1effectors that intersect the eukaryotic ubiquitin system.

See text for details.

T3SS1 effectors are substrates for the eukaryotic Ub pathway

Salmonella uses the host cell Ub system to ensure sequential activity of the T3SS1 effector proteins, SopE and SptP. SopE is the prototypical member of a family of guanine nucleotide exchange factor (GEF) mimics used by bacterial enteropathogens to activate Rho GTPases (48, 49). SopE and SopE-like GEFs are functional mimics of host cell GEFs since they have no structural or sequence identity yet function similarly by catalyzing the exchange of GDP for GTP thus switching the GTPase to its active state (48). SopE stimulates nucleotide exchange Cdc42 and Rac1, resulting in the actin polymerization necessary for invasion (50). To ensure that this actin polymerization is switched off following internalization, Salmonella translocates a GAP protein, SptP, which catalyzes the hydrolysis of GTP thus returning the GTPase to its inactive state (51, 52). The sequential activity of SopE and SptP, which are translocated simultaneously by T3SS1, is determined by ubiquitination following translocation. Both effectors are ubiquitinated and targeted to the proteosome, but SopE is degraded more efficiently and therefore inactivated more rapidly than SptP (51). This is an example of the exquisite interaction between the pathogen and host, where activation of a host protein, in this case Cdc42, is temporally controlled to stimulate the ruffles required for invasion and then switched off so that the actin cytoskeleton and membrane can rapidly revert to normal state.

Another T3SS1 effector that is ubiquitinated following translocation into host cells is SopB (also known as SigD), an inositol phosphate phosphatase that has several functions during invasion and for some time thereafter (53–56). Initially translocated SopB associates with the plasma membrane where it contributes to invasion by indirectly activating SGEF, an exchange factor for RhoG, which is involved in actin remodeling (53). Following invasion, SopB is translocated across the SCV membrane where it reduces the levels of negatively charged lipids thus playing a role in SCV maturation (56). SopB is monoubiquitinated on at least six lysine residues following translocation into host cells via an unknown mechanism that does not require any of the known Salmonella E3 Ub ligases (57, 58). Ubiquitination down-regulates SopB activity at the plasma membrane and increases retention of the effector on the SCV, thus expanding the functional repertoire of SopB, since the same enzymatic activity can be used to modulate actin-mediated bacterial internalization and Akt activation at the plasma membrane as well as vesicular trafficking and intracellular bacterial replication at the phagosome (57, 58).

A Salmonella HECT-like E3 ligase

SopA is one of several T3SS1 effectors with no defined function, although it does contribute to invasion in polarized cells (59). Based on functional and structural data SopA appears to be a novel HECT-like E3 ligase, even though it has very little sequence similarity with any eukaryotic E3 ligases. In vitro it forms an ubiquitin-thioester intermediate (5, 60) and analysis of the crystal structure of SopA revealed a C-terminal domain that has a bilobal architecture reminiscent of the N- and C-lobe arrangement of HECT domains (5, 60). The C-lobe of SopA is almost twice as large as a canonical C-lobe, 170 compared to 100 amino acids, but it has the conformational flexibility that is an important characteristic of HECT ligases (5, 60). A putative substrate-binding site, adjacent to the E2 binding domain, is a β-helix cylinder containing a sugar-binding site, suggesting that SopA may recognize carbohydrate-modified substrate proteins, although no substrates have been identified thus far (60). An additional interesting feature of SopA is that it is targeted for degradation following ubiquitination by the ER-bound RING finger protein 5 (RNF5/RMA1) (61), a protein better known for its role as part of the ER-anchored Ub ligase complex, which processes malfolded proteins (62, 63).

The NEL ligases

NEL family members, with conserved primary sequences and domain architecture, have been identified in a variety of gram-negative bacteria including Salmonella, Yersina and Shigella (64, 65). All are characterized by the presence of an NEL domain, containing a conserved catalytic cysteine residue, that is believed to mediate E2 binding and the ubiquitination reaction (66). Many also contain a leucine-rich repeat (LRR) of variable length that is likely a substrate recognition module (66). Intriguingly, LRR motifs are present in many mammalian proteins, including TLR4 (Toll Like Receptor 4), that are used to detect pathogen-associated molecular patterns (67). This has lead to the suggestion that molecular mimicry of this group of eukaryotic proteins could enable NEL ligases to subvert the immune response (64).

Three Salmonella effectors SlrP, SspH1 and SspH2, are NEL family members. All have been implicated in pathogenesis, even though SspH1 has limited distribution among S. enterica serotypes compared to SlrP and SspH2, which are widely distributed (68, 69). SspH2 is translocated by T3SS2 (68, 70) and is therefore unlikely to be involved in invasion per se, however, SlrP and SspH1 are members of a small group of “promiscuous” effectors that can be translocated into host cells via both T3SS1 and T3SS2 indicating that they may have roles both early and late in infection (68, 71). The NEL domains from these proteins have no structural resemblance to any mammalian ligases, but they have robust E3 ligase activity when combined in vitro with the human E2 ligase UbcH5 (3, 66). The catalytic mechanism of these ligases is distinct from that of the HECT E3 ligases, or SopA, since there is no thioester intermediate (66). In addition, studies on SspH2, have shown that the LRR domain appears to sequester the catalytic cysteine residue in the NEL domain such that activation of these ligases requires a substantial conformational change. Ultimately, binding of the LRR domain to proteins at the target site may be required to release the ligase domain for site-specific function (66). A surprising feature of SspH2 directed poly-Ub chain synthesis is that it results in the accumulation of K48-linked poly-Ub chains directly tethered to the E2 UBcH5 (65). Thus SspH2 may transfer completed poly-Ub chains from UbcH5 to the target protein (65).

Two hybrid systems have been used to identify eukaryotic binding partners of the Salmonella NEL effectors (70, 72, 73). SspH1 has been shown to interact with a fatty acid- and Rho-activated serine/threonine protein kinase called protein kinase N 1 (PKN1) (70), whereas thioredoxin and the endoplasmic reticulum protein ERdj3 were identified as SlrP binding partners (72, 73). The significance to Salmonella pathogenesis is unclear, since effector mediated ubiquitination of these substrates has only been shown in vitro (3, 72, 73).

Deubiquitination

S. Typhimurium has two T3 effectors, SseL and AvrA, which have been shown to have deubiquitinase activity and appear to be involved in down-regulating immune signaling (74–77). SseL is translocated via T3SS2 and is not likely to be involved in invasion, but AvrA is translocated by T3SS1 and has been shown to have a role in invasion in vivo (75). AvrA is a close homolog of YopJ a Y. pestis and Y. pseudotuberculosis effector that negatively regulates NF-κB and MAPK signaling (76, 78). The anti inflammatory effects of YopJ and AvrA may occur via slightly different mechanisms (76) but both have been shown to deubiquitinate a number of proteins such as IκB-α and β-catenin suggesting that they regulate host inflammatory responses through NF-κB and β-catenin pathways (79, 80). It has also been suggested that inhibition of β-catenin Ub might occur when AvrA blocks an E3 ligase such as β-TrCP (79). Recently the relevance of some of these studies was brought under scrutiny since AvrA does not seem to have significant anti-inflammatory effects when translocated by Salmonella at endogenous levels (81). Indeed, the role of AvrA and YopJ as DUBs is unclear since both have also been shown to function as acetyltransferases (82–85).

A common DUB thread in invasion mechanisms

Although the invasion mechanisms used by S. Typhimurium and L. monocytogenes are very different they both require extensive actin remodeling and factors that modulate actin dynamics in the host cell can affect invasion. One such factor that was recently described is Ubiquitin C-terminal hydrolase –L1 (UCHL1), a member of the UCH class of DUBs that is associated with neurodegenerative conditions, including Parkinsons disease, and a wide range of malignancies (86). UCHL1 up-regulates β-catenin signaling (87) and possibly promotes invasion of both Salmonella and Listeria by modulating actin dynamics in the host cell (86).

Conclusions

Bacteria do not possess Ub, yet many bacterial pathogens successfully utilize or intersect the host cell Ub pathway during invasion and establishment of the intracellular niche as well as for immune modulation (for review see (88)). Here only the initial invasion process has been considered, yet it goes without saying that the ability to exploit the omnipresent ubiquitin pathway is a powerful adaption that provides bacterial pathogens with the potential to access a myriad of cellular pathways.