Inactivation by blocking effector binding
Pathogens commonly inactivate Rho GTPases by blocking their binding to effectors through cross-linking bulky molecules to amino-acid residues located in the switch I region. Threonine-37 of RhoA (threonine-35 of Rac and Cdc42) is a hot spot for modifications by bacterial virulence factors, given its major role in contacting many Rho protein effectors (Figures 2 and 3a). This threonine residue is modified by a group of LCTs (large clostridial toxins), which trigger mono-glucosylation or N-acetyl-glucosaminylation of small GTPases (Just et al., 1995) (Table 1). Intoxication of cells by LCTs leads to disruption of the actin cytoskeleton, resulting in cell retraction, rounding and detachment. Interestingly, among the small GTPases tested for their sensitivity to these toxins, Rac1 (and potentially Rac2) was the only one modified by all toxin members (Table 1). This is probably related to the central role that these GTPases play in controlling inflammatory responses and the production of reactive oxygen species for pathogen destruction (Boquet and Lemichez, 2003; Bokoch, 2005).
Table 1. Examples of post-translational modifications induced by bacterial virulence factors targeting Rho GTPasesResidues are given in single-letter code.
|Endoproteolysis||YopT||Yersinia enterocolitica||RhoA, Rac, Cdc42||C-terminus (see Figure 2)||Shao et al. (2003); Fueller and Schmidt (2008)|
|ADP-ribosylation||C3 exoenzyme (C3bot)||Clostridium botulinum||RhoA>RhoB>RhoC (>>Rac1)||N41 of RhoA (and equivalent)||Chardin et al. (1989); Wilde et al. (2003)|
| ||C3-like transferase (C3lim)||Clostridium limosum||RhoA>RhoB, RhoC|| ||Just et al. (1992b)|
| ||C3-like transferase (C3cer)||Bacillus cereus||RhoA, RhoB>RhoC|| ||Just et al. (1992a); Wilde et al. (2003)|
| ||EDIN-A, −B, −C (C3stau −1, −2, −3)||Staphylococcus aureus||RhoA>RhoB>RhoC (>>Rnd3 for EDIN-A)|| ||Yamaguchi et al. (2001, 2002); Wilde et al. (2003)|
|Glucosylation||Toxin A and B||Clostridium difficile (C. difficile VPI 10463)||Rac, RhoA, Cdc42, not RhoD||T37 of RhoA (and equivalent)||Just et al. (1995); Jank et al. (2006)|
| ||Toxin B-1470||Clostridium difficile (C. difficile strain 1470)||Rac, Cdc42, R-Ras, Ral, Rap, not RhoA||T35 of Rac1 (and equivalent)||Chaves-Olarte et al. (1999)|
| ||Lethal toxin||Clostridium sordellii||Rac, Cdc42, RhoG, Ras, Ral, Rap, not RhoA||T35 of Rac1 (and equivalent)||Popoff et al. (1996); Jank et al. (2006)|
| ||Haemorrhagic toxin||Clostridium difficile||RhoA, Rac, Cdc42||T37 of RhoA (and equivalent)||Genth et al. (1996)|
|N-acetyl-glucosaminylation||Alpha-toxin||Clostridium novyi||RhoA, Rac, Cdc42||T37 of RhoA (and equivalent)||Selzer et al. (1996)|
|AMPylation||VopS||Vibrio parahaemolyticus||RhoA, Rac, Cdc42||T35 of Rac (and equivalent)||Yarbrough et al. (2009)|
| ||IbpA||Histophilus somni||RhoA, Rac, Cdc42||Y34 of RhoA (and equivalent)||Worby et al. (2009)|
|Transglutamination/deamidation||DNT||Bordetella bronchiseptica||RhoA, Rac, Cdc42||Q63 of RhoA||Horiguchi et al. (1997)|
| ||CNF1||Uropathogenic Escherichia coli||RhoA, Rac, Cdc42, not RhoD||Q63 of RhoA (and equivalent)||Flatau et al. (1997); Schmidt et al. (1997); Jank et al. (2006)|
| ||CNF2||Enteropathogenic Escherichia coli||RhoA, Rac, Cdc42|| ||Stoll et al. (2009)|
| ||CNF3||Necrotoxigenic Escherichia coli||RhoA, Rac, Cdc42|| ||Orden et al. (2007)|
| ||CNFY||Yersinia pseudotuberculosis||RhoA>Rac, Cdc42|| ||Hoffmann et al. (2004)|
Exciting new findings have unravelled a new type of modification affecting residues of the switch I region (Figures 2 and 3a). The injected effector VopS of V. parahaemolyticus catalyses the covalent attachment of a molecule of AMP from ATP to the threonine-37 of RhoA (threonine-35 of Rac and Cdc42) (Yarbrough et al., 2009) (Table 1 and Figure 2). This covalent modification by AMPylation triggers an inactivation of Rho proteins and the ensuing destruction of the actin cytoskeleton (Yarbrough et al., 2009). The covalent attachment of AMP by a phosphodiester bond is predicted to be reversible (Kinch et al., 2009). This activity depends on a conserved histidine residue of the Fic (filamentation induced by cAMP) domain of VopS. The bacterial virulence factor IbpA of the respiratory pathogen H. somni also disrupts the actin cytoskeleton (Worby et al., 2009). It bears two Fic domains at its C-terminus. One of the two Fic domains (Fic2) of IbpA was recently shown to catalyse the AMPylation of RhoA, Rac1 and Cdc42 (Worby et al., 2009). Here, however, AMP is cross-linked to another residue of the switch I region: the conserved tyrosine residue at position 34 of RhoA (tyrosine-32 of Rac and Cdc42) (Table 1 and Figure 2). Much remains to be learnt about this newly discovered enzymatic reaction, which may be shared by eukaryotes. For example, the primary sequences of the Fic domains of IbpA are homologous with the Fic domain of the human HypE (Huntingtin yeast-interacting protein E) (Faber et al., 1998; Kinch et al., 2009). HypE, but not the mutant HypE (H295A), can adenylylate recombinant RhoA, Rac and Cdc42 and catalyses the modification of proteins of similar size in cell extracts (Worby et al., 2009) (Table 2). Nevertheless, expression of HypE in cells does not trigger a collapse of the actin cytoskeleton. This might be due to a tight regulation of its activity in cells. A Fic-like-domain-containing protein from Drosophila melanogaster shows auto-adenylylation activity in vitro (Kinch et al., 2009). Two other domains found in proteins, named Doc (death on curing), encoded by phage P1 of Escherichia coli, and the injected effector AvrB (avirulence protein B) of Pseudomonas syringae, display sequence and structural homologies with Fic domains respectively. Kinch et al. (2009) have proposed that these domains be unified with Fic domains to create a single superfamily of functional domains named Fido (for Fic, Doc and AvrB). It remains to be shown whether Doc and AvrB also confer AMPylation, and whether this activity is directed toward Rho GTPases.
Table 2. Examples of post-translational modifications of Rho GTPasesResidues are given in single-letter code. CK1, casein kinase 1.
|AMPylation||HypE||RhoA, Rac, Cdc42||Y34 (RhoA)||Worby et al. (2009)|
|Transglutamination||Transglutaminase||RhoA||Q63||Schmidt et al. (1998); Singh et al. (2001)|
| ||Transglutaminase (serotonin)||RhoA, Rac1, not Ras|| ||Walther et al. (2003); Dai et al. (2008)|
|Phosphorylation||PKA||RhoA||S188||Lang et al. (1996)|
| ||PKG||RhoA||S188||Sauzeau et al. (2000); Rolli-Derkinderen et al. (2005)|
| ||SLK||RhoA||S188||Savoia et al. (2005)|
| ||Src kinase||Cdc42||Y64||Tu et al. (2003)|
| ||Akt||Rac1||S71||Kwon et al. (2000)|
| ||ROCK||RhoE||S11||Riento et al. (2005)|
| ||CK1 Ser/Thr kinase||RhoB||S185||Tillement et al. (2008)|
|Ubiquitination||Smurf1||RhoA||K6, 7||Wang et al. (2003); Ozdamar et al. (2005); Boyer et al. (2006a)|
| ||Cul3/BACURD||RhoA|| ||Chen et al. (2009)|
| || ||Rac1||K147||Doye et al. (2002); Lynch et al. (2006); Visvikis et al. (2008)|
| || ||Cdc42|| ||Doye et al. (2002)|
Inactivation by triggering the release of Rho proteins from the membrane
One way of inactivating Rho proteins is to remove these GTPases from the membrane. This can be achieved by cleaving the C-terminal part of Rho proteins containing the CAAX-box (Figures 2 and 3b). Indeed, Yersinia enterocolitica synthesizes the injected effector YopT (Yersinia outer protein T) (Shao et al., 2003). YopT is a cysteine protease with an endopeptidase activity that cleaves the C-terminal part of RhoA, Rac1 and Cdc42, thus severing their lipid anchor (Shao et al., 2003; Fueller and Schmidt, 2008) (Figure 2). This in turn triggers a release of the active form of Rac1 from membranes, resulting in its translocation to the nucleus (Wong and Isberg, 2005).
Another way to release Rho proteins from membranes is to corrupt their interaction with GDIs (Figure 3b). This strategy is shared by both cellular and bacterial factors and is best studied for the GTPase RhoA (Boquet and Lemichez, 2003; Aktories and Barbieri, 2005; Loirand et al., 2006). Various cellular kinases can phosphorylate RhoA, notably both the cAMP- and cGMP-dependent kinases, PKA and PKG respectively (Lang et al., 1996; Sauzeau et al., 2000, 2003; Loirand et al., 2006) (Table 2 and Figure 2). Phosphorylation of RhoA at serine-188 does not block its GTP-loading or its GTPase activity (Ellerbroek et al., 2003). Instead, both in vitro and in vivo studies show that this modification triggers a tight association of RhoA with GDIs and the extraction of RhoA from membranes (Lang et al., 1996; Sauzeau et al., 2000; Ellerbroek et al., 2003) (Figure 3b). Another study shows that, in vitro, the phosphorylation of RhoA blocks its association with ROCK (p21-Rho kinase), whereas it does not impair its association with rhotekin or mDIA1 (mammalian homologue of Drosophila Diaphanous protein 1) (Nusser et al., 2006). Phosphorylation of RhoA thus inactivates or eventually modulates the activity of RhoA. In contrast with regulation by GAPs, this may maintain a cytosolic pool of RhoA inaccessible to activation by GEFs, or, as suggested previously, a reservoir of GTP-loaded RhoA that can be mobilized in the absence of activation by GEFs (Loirand et al., 2006).
Several studies have also suggested a possible phosphorylation of other Rho proteins, such as Rac1 and Cdc42, at different positions (Table 2). As for RhoA, the phosphorylation of Cdc42 on tyrosine-64 by Src tyrosine kinase increases its association with GDIs (Tu et al., 2003). It has been suggested that the phosphorylation of serine-71 of Rac1 by the serine/threonine kinase Akt impairs the GTP-binding activity of Rac1 (Kwon et al., 2000). Phosphorylation might also contribute to regulation of atypical Rho GTPases, such as Rnd3 (RhoE), RhoH and RhoB (Riento et al., 2005; Gu et al., 2006; Tillement et al., 2008). Indeed, it has been shown by Riento et al. (2005) that the phosphorylation of serine-11 of Rnd3 by ROCKs in response to cell stimulation by PDGF (platelet-derived growth factor) stabilizes this GTPase and thus potentiates Rnd3 activity.
Phosphorylation of RhoA has been implicated in the regulation of vascular homoeostasis (Pacaud et al., 2005; Loirand et al., 2006). Studies involving the cardiovascular system implicate the phosphorylation of RhoA in smooth muscle cells as a key regulatory mechanism in blood vessels to prevent vascular dysfunction (Loirand et al., 2006). Indeed, the contraction of actomyosin cables driven by RhoA and its effector ROCK triggers vasoconstriction. Abnormal vasoconstrictions lead to vascular diseases, such as pulmonary hypertension (Pacaud et al., 2005). Consistently, expression of a phospho-resistant mutant of RhoA in smooth muscle of transgenic mice remodels the vascular system by causing perivascular fibrosis and coronary artery wall thickening (Sawada et al., 2009). Two important feedback regulations have been described. Vasodilatation can be induced by the NO/cGMP-dependent kinase PKG (Pacaud et al., 2005) and can be triggered by SLK (Ste20-related kinase) upon activation by the AT2R (angiotensin II type 2 receptor) pathway (Guilluy et al., 2008). Both the NO/PKG and AT2R/SLK pathways induced the phosphorylation of serine-188 of RhoA as a mechanism aimed at counteracting the effect of actin cable contraction mediated by RhoA in order to produce smooth muscle cell relaxation (Rolli-Derkinderen et al., 2005; Savoia et al., 2005).
To date, no bacterial virulence factors have been found to inactivate Rho proteins by phosphorylation. However, a parallel can be drawn with a group of virulence factors that inactivate RhoA by mono-ADP-ribosylation (Chardin et al., 1989; Wilde et al., 2003) (Table 1 and Figure 3b). In vitro experiments show that, although these ADP-ribosyltranferases can modify various Rho members, the C3 toxin of Clostridium botulinum and EDIN (epidermal cell differentiation inhibitor) of Staphylococcus aureus have a preference for RhoA (Wilde et al., 2003). ADP-ribosylation of asparagine-41 of RhoA in vitro does not block its binding to the PLD (phospholipase D) effector protein (Sekine et al., 1989; Genth et al., 2003). In contrast, several studies have established that in vivo ADP-ribosylation of RhoA induces its tight association with inhibitory factors (GDIs), leading to RhoA inactivation (Figure 3b) (Fujihara et al., 1997; Genth et al., 2003). Inhibition of RhoA by ADP-ribosyltransferases triggers various effects that depend on host cell types, including the paralysis of immune cell functions and the disruption of tight junctions in gut epithelial cell monolayers (Nusrat et al., 1995; Boquet and Lemichez, 2003). Rho GTPases, especially RhoA, regulate the formation of intercellular junctions and consequently endothelium permeability (Wojciak-Stothard and Ridley, 2003). A new cellular outcome of RhoA inhibition by ADP-ribosylation is observed when endothelial cells are intoxicated with either the ADP-ribosyltransferase C3 of C. botulinum or EDIN of S. aureus (Boyer et al., 2006b). Intoxication of various primary endothelial cell types with either of these toxins results in a dramatic loss of cellular actin cables (Boyer et al., 2006b). Intriguingly, this loss of actin cables in primary endothelial cells does not lead to cell retraction, as observed upon RhoA, Rac1 and Cdc42 inhibition by LCTs glucosylating toxins of Clostridium difficile. This instead leads to the formation of transcellular tunnels named macro-apertures (Boyer et al., 2006b). These tunnels form transiently, due to membrane-wave-driven closure. Formation and closure of macro-apertures by EDIN can be recapitulated by knocking down RhoA using RNAi (RNA interference). Interestingly, macro-apertures show some structural homologies with transmigratory cups that form in endothelial cells during the transcellular diapedesis of leucocytes (Carman and Springer, 2008). However, the mechanism by which leucocytes drive the formation of transcellular tunnels remains largely unknown. Endothelium intoxication by EDIN leads to a loss of barrier function (Boyer et al., 2006b; Rolando et al., 2009). These tunnels confer to S. aureus the ability to bind to the extracellular matrix beneath the endothelium (Boyer et al., 2006b; Rolando et al., 2009). This phenomenon may favour the bacterial colonization of the endothelium and potentially promote the dissemination of these pathogens in host tissues.