YopE of Yersinia pseudotuberculosis inactivates three members of the small RhoGTPase family (RhoA, Rac1 and Cdc42) in vitro and mutation of a critical arginine abolishes both in vitro GTPase-activating protein (GAP) activity and cytotoxicity towards HeLa cells, and renders the pathogen avirulent in a mouse model. To understand the functional role of YopE, in vivo studies of the GAP activity in infected eukaryotic cells were conducted. Wild-type YopE inactivated Rac1 as early as 5 min after infection whereas RhoA was downregulated about 30 min after infection. No effect of YopE was found on the activation state of Cdc42 in Yersinia-infected cells. Single-amino-acid substitution mutants of YopE revealed two different phenotypes: (i) mutants with significantly lowered in vivo GAP activity towards RhoA and Rac1 displaying full virulence in mice, and (ii) avirulent mutants with wild-type in vivo GAP activity towards RhoA and Rac1. Our results show that Cdc42 is not an in vivo target for YopE and that YopE interacts preferentially with Rac1, and to a lesser extent with RhoA, during in vivo conditions. Surprisingly, we present results suggesting that these interactions are not a prerequisite to establish infection in mice. Finally, we show that avirulent yopE mutants translocate YopE in about sixfold higher amount compared with wild type. This raises the question whether YopE’s primary function is to sense the level of translocation rather than being directly involved in downregulation of the host defence.
The Gram-negative pathogen Yersinia pseudotuberculosis causes gastrointestinal infections in animals and humans, characterized by diarrhoea, abdominal pain and fever. Although human infections usually are self-limiting, the bacteria can cause a lethal infection in rodents (Cover and Aber, 1989). Thus, mice constitute a suitable animal model for Y. pseudotuberculosis infection. The major virulence determinants are found on a 70 kb virulence plasmid encoding a functional type III secretion system (T3SS). Five different effector proteins, called Yops, are translocated into the target cells by the T3SS enabling the pathogen to resist the innate immune system of the host, including antiphagocytosis and downregulation of the proinflammatory cytokine response (Rosqvist et al., 1988; Schesser et al., 1998; Black and Bliska, 2000; Schotte et al., 2004). One key effector in these processes is the 25 kDa YopE protein (Rosqvist et al., 1990; Black and Bliska, 2000).
Several studies have established that mutation of a critical arginine residue in the GAP domain of the bacterial GAPs completely abolishes their in vitro GAP activity towards RhoA, Rac1 and Cdc42 (Fu and Galan, 1999; Goehring et al., 1999; Von Pawel-Rammingen et al., 2000). We and others have shown that the YopE GAP mutant (R144A) is avirulent in mice (Black and Bliska, 2000; Aili et al., 2002) and as such is unable to induce a cytotoxic effect on HeLa cells (Black and Bliska, 2000; Von Pawel-Rammingen et al., 2000; Aili et al., 2002). This has led to the proposal that the YopE GAP activity towards Rac1, RhoA and Cdc42 is essential for induction of the cytotoxic phenotype on HeLa cells and virulence. This is supported by the finding that pre-treatment of HeLa cells with cytotoxic necrotizing factor 1 (CNF1), resulting in constitutive activated Rho proteins, is able to protect infected cells from the YopE cytotoxic effect (Von Pawel-Rammingen et al., 2000). While these studies collectively support the view that the GAP activity towards RhoA, Rac1 and Cdc42 is important for the cytotoxic effect of YopE on eukaryotic cells, there is no formal evidence for a direct interaction between YopE and any of these three RhoGTPases during infection.
It has also been reported that constitutively active RhoA can rescue cells from YopE-induced cytotoxicity (Black and Bliska, 2000). However, transfection of active RhoA may alter the balance of other substrates involved in the actin homeostasis and thus generate a misleading result. Andor et al. (2001) reported that YopE selectively targets Rac1-mediated signalling pathways, leaving RhoA- and Cdc42-controlled pathways untouched. The discrepancies between the results of these two studies indicate that neither method is optimal for resolving the question of YopE’s in vivo target and that further studies are required.
A previous study identified five regions in the C-terminal of YopE to be necessary for in vitro GAP activity towards Rac1, RhoA and Cdc42. Surprisingly, mutations in these regions resulting in YopE variants devoid of in vitro GAP activity towards RhoA, Rac1 and Cdc42 were still cytotoxic to HeLa cells (Aili et al., 2003). While the GAP activity of YopE is essential, it is not clear whether RhoA, Rac1 and/or Cdc42 are the functional targets during in vivo conditions.
To further dissect the substrate specificity of YopE we have investigated the region between amino acids 178 and 183, which is suggested to be involved in substrate recognition (Stebbins and Galan, 2000; Aili et al., 2003). Single-amino-acid substitutions were created in cis in this region and were used to elucidate the role of YopE with respect to cytotoxicity, virulence and GAP activity, both in vitro and in vivo. We identified such amino acid mutations essential for virulence in mice yet these mutants have no effect on the ability of YopE to downregulate RhoA and Rac1 in vivo during infection of HeLa cells. Importantly, we show that there is a lack of correlation between in vivo and in vitro GAP activity and virulence. Our results suggest that YopE targets an as yet unidentified GTPase. We have also analysed the contribution of YopE in regulating translocation and expression of Yops in the presence of eukaryotic cells. Although YopE has no effect on the in vitro regulation of the yop virulon, avirulent yopE mutants show higher translocation of YopE than wild type, indicating a regulatory role for YopE in effector translocation into HeLa cells.
Construction of single-amino-acid substitution mutations in YopE
YopE shares a high degree of similarity with the N-terminal half of two other bacterial toxins, ExoS of P. aeruginosa and SptP of S. typhimurium, both possessing in vitro GAP activity towards members of the RhoGTPase family (Fu and Galan, 1999; Goehring et al., 1999). Sequence alignment revealed that the region between amino acids 178 and 186 of YopE have at least six out of nine amino acids identical to the corresponding regions of ExoS and SptP, called here the ‘homology domain’ (Fig. 1), suggesting that this region is of importance for the function of these three bacterial GAPs. In line with this, a previous mutational analysis showed that amino acids 166–183 were important for the activity of YopE (Aili et al., 2003). To gain a deeper insight into the YopE function, we focused on this region. Six different substitution mutations were constructed in the homology domain (Fig. 1) and introduced in cis into the wild-type virulence plasmid (pIB102) generating yopE mutants YPIII(pIB566) [F178A], YPIII(pIB567) [S179A], YPIII(pIB568) [Q180A], YPIII(pIB569) [W181A], YPIII(pIB570) [G182A] and YPIII(pIB571) [T183A] (Table 1). Identical mutations were made in the YopE expressing pAF19 (Rosqvist et al., 1995), where yopE and yerA are divergently transcribed from their native promoter resulting in overexpression of the proteins, and introduced in trans in the yopE null mutant, YPIII(pIB522) (Table 1).
Table 1. Phenotypes of wild-type and mutant strains of Yersinia pseudotuberculosis.
Cytotoxicity in cis was assayed after infection of HeLa cells with the different strains. Cytotoxicity is characterized by rounding up of the HeLa cell as analysed by phase contrast microscopy 3 h after infection. +++, complete rounding up, –, no changed morphology.
The LD50 value was determined using C57BL/6 mice infected with wild type and different yopE mutants. The calculated LD50 value should not be seen as a standard LD50 experiment (Reed and Muench, 1938) as mice were sacrificed when they showed signs of severe sickness. The LD50 value is the mean value of three independent trials. We define a strain virulent if the LD50 value is within five times of the wild-type strain.
pAF19 (Rosqvist et al., 1995) carries the wild-type yerA and yopE genes under their native promoters. When expressed in a yopE null mutant this plasmid expresses, secretes and translocates high levels of YopE. Point mutations were constructed in the pAF19 background.
Cytotoxicity in trans was assayed with the yopE null mutant [YPIII(pIB522)] complemented with the different plasmids in the table.
n.d., not done.
2.8 × 103
1.0 × 107
1.0 × 107
1.3 × 104
1.1 × 103
4.3 × 105
1.0 × 107
9.2 × 104
8.0 × 106
Virulence of yopE mutants
We have earlier shown that arginine R144 of YopE is essential for Yersinia to cause systemic infection in mice (Aili et al., 2002). This mutation abolishes the YopE GAP activity towards RhoA, Rac1 and Cdc42 in vitro (Black and Bliska, 2000; Von Pawel-Rammingen et al., 2000). To further evaluate the function of YopE during infection, we challenged mice with the single-amino-acid substitution mutants YPIII(pIB566) [F178A], YPIII(pIB567) [S179A], YPIII(pIB568) [Q180A], YPIII(pIB569) [W181A], YPIII(pIB570) [G182A] and YPIII(pIB571) [T183A] by intraperitoneal injection.
As observed with the GAP mutant (R144A), mutants expressing W181A and T183A were avirulent showing a LD50 value of about 1 × 107 (Table 1). Mutants expressing Q180A and G182A showed about 2 log (LD50– 4 × 105 and LD50– 9 × 104) attenuation compared with the wild-type strain (3 × 103). Mutant strains expressing F178A and S179A were both virulent showing an LD50 value of 1 × 104 and 1 × 103 respectively. Interestingly, the latter strain showed consistently a lower LD50 value than the wild-type strain.
Effect of single-amino-acid substitution mutants on HeLa cells
Wild-type Yersinia induces a YopE-mediated cytotoxic effect on HeLa cells. The cells round up as early as 45 min after addition of wild-type bacteria due to a collapse of the actin cytoskeleton and characteristic retraction tails are left (Rosqvist et al., 1991). Plasmids expressing single-amino-acid substitutions R144A, F178A, S179A, Q180A, W181A, G182A and T183A of YopE were expressed in trans in the yopE null mutant and allowed to infect cultured HeLa cells. The morphology of the cells was observed by phase contrast microscopy. As expected the GAP-deficient arginine finger mutant (R144A) showed no cytotoxic activity (Aili et al., 2002), whereas the other six mutants induced a cytotoxic response identical to wild type 45 min after initiation of infection (Table 1). All YopE variants are under the control of their native promoter on a high-copy plasmid, which results in overexpression of the protein. In order to analyse the effect of the mutations during native conditions, each mutation was therefore introduced in cis into the yopE gene on the virulence plasmid by homologous recombination (see Experimental procedures). When the identical mutations were analysed in cis, surprisingly, the W181A mutant did not induce a cytotoxic response on infected HeLa cells (Table 1). The other five mutants F178A, S179A, Q180A, G182A or T183A all induced a cytotoxic response on the HeLa cells with similar kinetics as the wild-type strain (Table 1), showing full cytotoxicity after 45 min of infection.
All YopE variants localizes to the perinuclear region
Previous studies have shown that translocated YopE localizes to the perinuclear region of HeLa cells (Rosqvist et al., 1994). We wanted to test whether the single-amino-acid substitution mutants affected the intracellular localization of YopE. Infected HeLa cells were analysed 3 h after infection with wild-type and mutant Yersinia. Cells were fixed, permeabilized and a rabbit anti-YopE antibody was added to the cells followed by a secondary donkey anti-rabbit Alexa488-conjugated antibody. Visualization of the HeLa cell membrane was carried out by using Wheat Germ agglutinin conjugated to Rhodamine. The samples were then analysed by laser scanning confocal microscopy. To facilitate the analysis the HeLa cells were infected with the mutated YopE variants provided in trans. Wild-type YopE as well as all the single-amino-acid substitution mutants analysed showed the same perinuclear localization of YopE 3 h after infection (Fig. 2). Thus, introduction of mutations in the ‘homology domain’ of YopE does not alter the intracellular localization of YopE.
YopE has a regulatory effect on translocation in vivo
The difference in cytotoxicity between overproduction of the protein in trans and native regulation in cis made it important to measure the amount of YopE translocated by the different cis mutants. The digitonin protease K protection assay (Nordfelth and Wolf-Watz, 2001; Aili et al., 2003) was employed and we found to our surprise that the three avirulent mutants (R144A, W181A and T183A) repeatedly translocated about five to six times more YopE when compared with the wild type (Fig. 3). Quantification was carried out using a Fluor-STM MultiImager (Bio-Rad, Richmond, CA). In contrast, the virulent and attenuated point mutants all translocated YopE at levels between wild type and the avirulent mutants. Thus, no mutant translocated YopE at lower levels than wild type (Fig. 3).
Yop expression were also analysed both in vitro and in vivo. The yopE mutants analysed in this study show both similar Ca2+ regulation, Yop expression and secretion in vitro as the wild-type strain YPIII(pIB102) (Fig. 4A and B). Analysis of the in vivo expression of YopE after infection of HeLa cells was assessed (Fig. 4C), and avirulent mutants (R144A, W181A and T183A) showed a twofold increase in total YopE expression (intrabacterial and intracellular) when compared with wild type. Remaining four mutants showed intermediate levels (Fig. 4C). These results indicates that YopE has the potential to regulate at least its own expression after infection of cultured cells. The results obtained with the W181A mutant with respect to cytotoxicity show also that in some cases different results can be obtained when the particular protein is expressed in cis or in trans. This is an important point to consider in these kinds of studies and indicate that the experiments should be carried out in cis.
Non-cytotoxic YopE variants induce LDH release
Recently it was shown by Viboud and Bliska (2001) that YopE of Yersinia prevented YopB-induced LDH release from HeLa cells. We employed this assay to analyse our mutants and LDH release was measured from HeLa cells infected with different strains (Fig. 5). The multiple Yop mutant (pIB29MEKA) and the isogenic yopB strain confirm previous study that YopB is essential for Yersinia-induced LDH release from HeLa cells (Fig. 4; Viboud and Bliska, 2001). The pIB29MEKBA strain was used as a negative control strain and release from cells infected with this strain was used to normalize the values. Wild-type Yersinia blocked LDH release while the GAP mutant did not, confirming the results of Viboud and Bliska (2001). All substitution mutants were tested for LDH release, and only one, the W181A mutant, showed an increase in LDH release compared with wild type (Fig. 5). Thus, the two non-cytotoxic mutants (R144A and W181A) stimulated LDH release in contrast to the cytotoxic mutants which were able to block LDH release from infected HeLa cells. This observation was in line with the findings of Bliska who argued that LDH release was connected to rearrangement of the actin cytotoskeleton in the HeLa cells (Viboud and Bliska, 2001). This suggests that the cytotoxicity induced by all strains in this study is not due to unspecific membrane damage, but show similar characteristics as wild type-induced cytotoxicity. It also singles out the W181A mutant as it has lost the ability to prevent LDH release, presumably by loss of activity towards the YopE target that prevents LDH release from infected cells.
In vitro GAP activity
We have earlier analysed in vitro GAP activity of different deletion and substitution mutations located between amino acids 162 and 192 of YopE (Aili et al., 2003). We found four substitution mutants (YopE166-68A, YopE169-71A, YopE175-77A and YopE178-80A) with a lowered level of GAP activity (10-fold decrease) towards RhoA, Rac1 and Cdc42 when compared with wild-type YopE, but interestingly the corresponding YopE mutant strains still induced a full cytotoxic response on HeLa cells (Aili et al., 2003), indicating that the cytotoxic response can be uncoupled from the in vitro GAP activity of YopE on RhoA, Rac1 and Cdc42. To extend this analysis, single-amino-acid substitution mutants was used to measure the YopE in vitro GAP activity towards these substrates. To facilitate purification of the mutant proteins, 6xHis-tagged variants were constructed and expressed in the MYM strain (lacking the Yop effectors) as earlier described for the wild-type YopE (Von Pawel-Rammingen et al., 2000). These YopE(90-219) variants were thus purified and tested for in vitro GAP activity towards RhoA, Rac1 and Cdc42 (Fig. 6).
The in vitro GAP activity of YopE was measured as hydrolysis of 32P-GTP after 20 min at 16°C (Rac1 and Cdc42) or 20°C (RhoA). To reduce the amount of 32P-GTP bound to RhoA to 50%, 11 nM wild-type YopE was required (Aili et al., 2002). The GAP-defective R144A mutant protein has been tested before for in vitro activity towards RhoA, Rac1 and Cdc42 (Aili et al., 2002) and no activity was seen upon addition of 300 nM YopER144A, the highest concentration of mutant protein we were able to purify. Complete hydrolysis of 32P-GTP to GDP was obtained after addition of 80 nM wild-type YopE. Addition of the W181A mutant YopE protein did not increase the GTPase activity of RhoA, even at concentrations as high as 5000 nM showing the similar kinetics in its inability to stimulate hydrolysis as the YopE arginine finger mutant (Fig. 6A). The F178A and the S179A variants showed a low GAP activity and high concentrations (1200 nM and 2500 nM) were required to reduce the level of bound 32P-GTP to 50%. Even high concentrations (5000 nM) were insufficient to completely hydrolyse the bound 32P-GTP (Fig. 6A). The Q180A variant was also defective in stimulating hydrolysis and only minor GAP activity could be seen at concentrations of 5000 nM where 60% of the bound 32P-GTP was hydrolysed (Fig. 6A). The in vitro GAP activity of the purified YopE variants was also tested on Rac1 and Cdc42. A pattern similar to that of RhoA was found for all of the YopE variants tested as well as the wild-type YopE protein (Fig. 6B and C).
Thus, amino acids 178–181 are of importance for the in vitro GAP activity of YopE towards RhoA, Rac1 and Cdc42. Mutation of a single amino acid at position 178, 179 or 180, respectively, did not change the in vitro substrate specificity between these Rho GTPases. However, these mutations dramatically reduced the GAP activity of YopE towards all three GTPases tested. Amino acid W181 was found to be crucial for the GAP activity of YopE, as the W181A mutant protein was as inactive as the arginine finger mutant protein. YopE variants corresponding to these two latter mutant proteins lacked the ability to induce a cytotoxic response, whereas the other three mutants induced cytotoxicity at the same level as the wild-type strain. This latter observation supported our earlier findings showing that yopE mutants lacking in vitro GAP activity towards RhoA, Rac1 and Cdc42 but still being cytotoxic for HeLa cells can be isolated.
Surprisingly, we found that purified YopE protein from virulent mutants (F178A and S179A) possesses a very low in vitro GAP activity towards RhoA, Rac1 and Cdc42 (Fig. 6). We had anticipated that if YopE targets one of these proteins during infection, the substantial reduction in GAP activity of these mutants would influence their ability to cause disease in mice. However, GAP activity was only measured during in vitro conditions, and we could not at this point exclude the possibility that there was a difference between in vitro and in vivo activity of YopE.
YopE is a functional GAP towards RhoA and Rac1, but not towards Cdc42 in vivo
Investigation of the effect of YopE on RhoA, Rac1 and Cdc42 after infection with Y. pseudotuberculosis was carried out using the pull-down technique employing GST fusions of the binding domain of downstream targets of the active form of each GTPase to measure the activation state of endogenous GTPases in HeLa cells (Ren et al., 1999; Sander et al., 1999).
Cells were infected with wild-type YopE and the GAP arginine finger (R144A) mutant strain and lysed after 30 min. Cleared supernatants were incubated with GST-PAK beads and bound GTPase was separated on SDS-polyacrylamide electrophoresis (SDS-PAGE) and analysed by immunoblotting. The results showed that wild-type YopE efficiently reduced the activation state of Rac1 as measured by GST-PAK-Rac1-BD pull-downs in cells (Fig. 7A), whereas the GAP mutant (R144A) had no effect on the activity of Rac1 (Fig. 8B).
By analysing the kinetics of YopE-induced downregulation of Rac1 during infection, we observed that already 5 min after infection the level of active Rac1, as compared with uninfected cells, was greatly reduced (about 90%) (Fig. 7A). At 15 min after infection, only very low levels of active Rac1 could be detected (Fig. 7A). Therefore, we can conclude that YopE efficiently downregulates Rac1 at an early stage after onset of infection in eukaryotic cells.
Cells were infected with wild-type Y. pseudotuberculosis and the activation state of RhoA was measured by GST-Rhotekin-Rho-BD pull-downs at different time points after infection. As expected, the GAP mutant (R144A) strain showed no effect on the activation stage of RhoA after infection (Fig. 7B). In contrast, at 30 min after infection with the wild type, there was almost no detectable active RhoA in the lysates (Fig. 7B), suggesting that YopE acts as a GAP for RhoA in vivo during the intermediate early phase of infection.
Cells infected with wild-type Y. pseudotuberculosis show no reduction in Cdc42 activation, as measured by GST-WASP-Cdc42-BD precipitation (data not shown). This result that YopE does not function as a GAP for Cdc42 in vivo is in disagreement with our in vitro observation (see Fig. 6).
YopE-mediated inactivation of RhoA and Rac1 is not a prerequisite for virulence
Based on the results obtained above, the in vivo YopE GAP activity towards RhoA and Rac1 of the single-amino-acid substitution mutants was further analysed.
As above, we used the Rhotekin-binding assay to assess the activation state of RhoA after infection of eukaryotic cells (Ren et al., 1999). Cells were infected with strains carrying wild-type YopE, the arginine R144A finger mutant or the single-amino-acid substitutions F178A, S179A, Q180A, W181A, G182A or T183A respectively. Thirty minutes after onset of infection cells were lysed and active RhoA was precipitated using GST-Rhotekin-coupled agarose-beads and analysed by Western blot with anti-RhoA antibodies. In cells infected with the wild-type strain only low amounts of active RhoA could be detected (Fig. 8A). Cells infected with strains expressing the arginine mutant R144A or either of the single substitution mutants F178A, W181A or T183A showed a high level of active RhoA after infection. Mutation of S179 and G182 did not change the ability of YopE to downregulate RhoA, whereas the Q180A mutation slightly lowered the GAP activity of YopE towards RhoA (Fig. 8A). Collectively these data show that amino acids in position 144, 178, 181 and 183 are essential for the YopE GAP activity towards RhoA. Interestingly, the F178A mutant strain was as virulent as the wild type although its activity to downregulate RhoA in cells was severely impaired suggesting that inactivation of RhoA is not a prerequisite for virulence. Even after 60 min of infection with the F178A YopE mutant active RhoA could still be detected at similar levels as after 30 min infection (data not shown). In contrast, the G182A mutant showed similar GAP activity as the wild-type strain, but was attenuated for mice compared with the wild type and the F178A mutant (Table 1). These findings suggest that RhoA is a non-essential target of YopE during in vivo conditions.
The single-amino-acid mutants’ ability to affect Rac1 activity during infection of HeLa cells was also analysed as described above. No effect of the F178A mutant on Rac1 activity could be detected 30 min after infection, whereas the G182A mutant was as efficient as wild-type YopE strain in inactivating Rac1 (Fig. 8B). As observed with RhoA, the F178A mutant was unable to downregulate Rac1 even after 60 min infection (data not shown). These data favours the hypothesis that Rac1 is a non-essential target of YopE, as the F178A mutant was fully virulent and the G182A was attenuated. In addition, the T183A mutant did not affect the GAP activity of YopE towards Rac1, while it was essential for the ability to downregulate RhoA (Fig. 8), identifying the importance of amino acid T183 in recognition of RhoA.
All mutants were also analysed for in vivo GAP activity towards Cdc42, with ExoS of Pseudomonas as a positive control for Cdc42 inactivation. None of the mutants showed any change of the activation state of Cdc42 (data not shown), supporting the results described above that Cdc42 is not targeted by YopE during infection in vivo.
YopE of Y. pseudotuberculosis possesses in vitro GAP activity towards three members of the small Rho family of GTPases, Rac1, RhoA and Cdc42 (Black and Bliska, 2000; Von Pawel-Rammingen et al., 2000; Aili et al., 2002). A critical mutation of the ‘arginine finger’ at position 144 to alanine abolishes the in vitro GAP activity and renders the corresponding mutant avirulent (Aili et al., 2002). Thus, the GAP activity of YopE per se is essential for virulence. It has been suggested that YopE specifically targets Rac1 in the eukaryotic target cell (Andor et al., 2001; Schotte et al., 2004) which, also as a result of the YopE activity, becomes cytotoxically effected, due to depolymerization of the actin cytoskeleton (Rosqvist et al., 1991). However, we recently reported that mutants of YopE lacking in vitro GAP activity towards RhoA, Rac1 and Cdc42 were still cytotoxic for HeLa cells (Aili et al., 2003), raising doubts that these small GTPases are the sole targets for YopE during infection. In this study, amino acid changes in YopE were localized between amino acids 166 and 180 (Aili et al., 2003). This region overlapped with a region of YopE (amino acids 178–186), the ‘homology domain’, showing high similarity between the three identified bacterial GAPs YopE, ExoS and SptP respectively (see Fig. 1). Interestingly, the corresponding region in SptP was recently shown by Stebbins and Galan (2000) to be in direct contact with the proposed substrate Rac1, suggesting that this domain is involved in substrate recognition.
To further dissect the functional role of YopE, we therefore constructed six single-amino-acid alanine substitution mutants in amino acids 178–183 of YopE. Structural modelling of these mutations revealed that four of the targeted amino acids had the potential to affect the substrate binding of YopE while the remaining two amino acids (F178 and W181) faced the interior of the YopE GAP domain and would likely not directly affect the interaction with the substrate but would rather have structural implications for the YopE tertiary structure (Koradi et al., 1996; Guex and Peitsch, 1997; Guex et al., 1999).
We found that the W181A mutant was devoid of in vitro GAP activity towards RhoA, Rac1 and Cdc42, similar to the arginine finger mutant (R144A). Mutants F178A, S179A and Q180A showed 100-fold reduced GAP activity compared with the wild-type YopE protein with no particular alteration in substrate specificity towards the substrates. For unknown reasons we were unable to purify two of the mutants (G182A and T183A). Nevertheless, our results indicated that we had targeted an essential domain of YopE involved in the interaction with small RhoGTPases.
Wild-type YopE inactivated Rac1 and RhoA in infected cells, whereas no effect on the activity of Cdc42 was found, excluding Cdc42 as a target of YopE. YopE is not unique in showing specificity difference in vitro and in vivo, as eukaryotic GAPs have been shown to target more members of the small RhoGTPase family in vitro, than in vivo (Shang et al., 2003). Although BPGAP1 targets both RhoA and Cdc42 in vitro, it is only able to downregulate the activity of RhoA in vivo (Shang et al., 2003). This clearly demonstrates that relying solely on in vitro data can lead to artificial conclusions. It also shows that results obtained from in vitro assays can differ from studies conducted during in vivo conditions. These results highlight the importance of performing analysis in biological relevant models and it is obvious that conclusions based on in vitro results may be misleading.
Our result that YopE inactivated Rac1 and RhoA in infected cells corroborates the findings by others that YopE of Yersinia enterocolitica specifically targets Rac1 during infection of eukaryotic cells (Andor et al., 2001). In addition, transfection of a constitutive active form of Rac1 has been shown to interfere with the antiphagocytic effect of YopE (Black and Bliska, 2000). Thus, the results from the three studies collectively suggest that among the three GTPases, YopE preferentially targets Rac1 after infection of cultured cells. Importantly, this effect seems to be independent of the cell type used.
Studies of the in vivo GAP activities of the single-amino-acid substitution mutants revealed a range of interesting phenotypes. Surprisingly, the attenuated yopE mutant G182A showed wild-type activity towards RhoA and Rac1 in vivo. This result points towards a redundant function of YopE against RhoA and Rac1 in vivo. In support of this conclusion was the finding that the virulent F178A mutant showed reduced in vivo GAP activity towards RhoA and Rac1 up to at least 1 h after infection. Obviously, HeLa cells are not the natural target cell for Yersinia and effects seen in this model system may be misleading. Nevertheless, Y. enterocolitica translocate YopE protein into a wide range of cultured eukaryotic cells, including haematopoietic cells, the identified target cells of Yersinia in mice (Boyd et al., 2000; Marketon et al., 2005). In addition, Yersinia-induced YopE-dependent cytotoxicity is a general effect on eukaryotic cells, implying that YopE is active towards all cells (Boyd et al., 2000). Hence, it is reasonable to assume that the differences in GAP activity of the YopE variants are conserved in all eukaryotic cells including the target cells. Therefore, a relatively low activity in HeLa cells would also imply a low activity in the target cell and vice versa. In addition, the membrane localization domain (MLD), identified in the N-terminus of YopE (Krall et al., 2004), is intact in our mutants and all YopE variants, as well as the wild type, were shown to localize to the same perinuclear region of infected HeLa cells, showing that the mutations did not affect the localization of the respective mutant protein.
It could be argued that the physical interaction between the YopE variants and the activated GTPase could prevent binding of the GST-PAK-BD or GST-Rhotekin-BD, leading to a seemingly loss of active Rac1 or RhoA in the pull-down experiments. This is unlikely, as this effect should be more pronounced when the R144A mutant was employed. This was, however, not the case, as the GAP mutant was clearly unable to downregulate RhoA and Rac1 both in vitro and in vivo (compare Figs 6 and 8). It could be anticipated that the interaction between the R144A GAP mutant and the GTPases should be tighter as the R144 is mainly involved in stabilizing the GTP during hydrolysis, not in actual substrate binding while the other mutations are localized in a region involved in substrate recognition (Stebbins and Galan, 2000). In addition, single-amino-acid substitution in the ‘homology domain’ should rather weaken the interaction with the GTPase and interfere less with the binding capacity of the GTPase to downstream targets, which would result in the observed inability to downregulate RhoA and Rac1. Therefore, we argue that our in vivo measurements reflect the state of activity of the different GTPases analysed.
Mutants with reduced in vitro GAP activity still induce a cytotoxic response on cultured HeLa cells (Aili et al., 2003). In line with this, the mutants studied here showing 100-fold reduced in vitro GAP activity are still cytotoxic towards HeLa cells, confirming that the cytotoxic activity of YopE can be uncoupled from the in vitro GAP activity towards RhoA, Rac1 and Cdc42. Interestingly, two of these mutants (F178A and S179A) maintained full virulence in mice, while one mutant (Q180A) was attenuated and another avirulent (W181A). Thus, in vitro results can not be directly translated into in vivo conclusions. Moreover, all mutants that induced cytotoxicity also blocked LDH release. This latter activity has been linked to YopB-induced pore formation in the target cell membrane and the status of cellular stress fibres (Viboud and Bliska, 2001). In contrast, only the avirulent mutants R144A and W181A failed both to induce HeLa cell cytotoxicity and to block LDH release. Clearly, these two mutants are unable to block essential cellular functions of the eukaryotic cell that, when disrupted, would promote HeLa cell cytotoxicity and blockage of LDH release. As both these cellular effects have been linked to the regulation of the actin cytoskeleton (Rosqvist et al., 1991; Viboud and Bliska, 2001), our results suggest that YopE targeting of Rac1, rather than RhoA, is responsible for these effects on infected HeLa cells. Additional support for this conclusion is seen with the avirulent T183A mutant, which is cytotoxic, blocks LDH release and can inactivate Rac1, but not RhoA, after infection of HeLa cells. However, we hypothesize that neither Rac1 nor RhoA is the critical target of YopE during infection. This idea is supported by the T183A mutant that was fully active towards Rac1, but still avirulent in mice.
We also observed that all avirulent mutants (R144A, W181A and T183A) translocated elevated levels (about five times) of YopE into the target cell. This indicated that the bacteria can monitor the level of translocation and respond accordingly, as we earlier have suggested (Holmstrom et al., 1997). This was also supported by the fact that total expression of YopE was higher in the avirulent mutants (R144A, W181A and T183A) compared with wild type. In addition, the attenuated mutants showed intermediate levels with a small tendency for the attenuated strains to express higher levels of YopE. We therefore suggest that the GAP activity of YopE is involved in mediating a regulatory signal from the inside of the eukaryotic cells back to the extracellular bacteria, instructing the bacteria that translocation should be terminated.
In conclusion, no direct link between in vitro and in vivo results with respect to the YopE GAP activity could be established. This argues against the use of in vitro results when trying to dissect the functional role of YopE in vivo. Furthermore, the importance of the ‘homology domain’ in substrate recognition by YopE has been demonstrated and points towards a possible conserved function of these amino acids in the different bacterial GAP domains. Minor changes in this domain can have profound effects on substrate recognition, and can explain why the bacterial GAP domains, although likely of a common evolutionary origin, could possess different substrate-binding properties. Our results argue for the hypothesis that neither RhoA nor Rac1 is essential targets for the GAP activity of YopE in causing disease in mice. We also challenge the view that the function of YopE is solely to disrupt the actin cytoskeleton and prevent phagocytosis during infection. Rather we favour the idea that YopE is a sensor involved in regulating the level of translocated effectors during infection. This hypothesis is supported by the avirulent, fully cytotoxic T183A mutant showing increased translocation during infection, yet retaining in vivo activity towards Rac1. In this context, both ExoS and SptP are interesting proteins, as they are YopE-X hybrid proteins where ‘X’ have either ADP-ribosyltransferase activity (ExoS) or tyrosine phosphatase activity (SptP). Perhaps the YopE-like domains of these hybrid proteins are not essential for virulence by interfering with host cell signalling, but rather act as a sensor for translocation. We are currently exploring this hypothesis.
PCR-amplified DNA fragments used for constructing the single-amino-acid mutants of yopE were generated by overlap PCR (Horton and Pease, 1991) using the pAF19 plasmid (Rosqvist et al., 1995) as a template for the amino acid substitutions within the yopE gene with mutation-specific oligoncleotides. Oligonucleotide sequences can be obtained by request. Derivatives of pAF19 were made for each single-amino-acid mutation with yopE under the control of its native promoter (Table 1).
Isolation of plasmid DNA from Escherichia coli strains was performed using the Quantum Prep Plasmid Miniprep kit (Bio-Rad) as described by the manufacturer. Standard DNA manipulation techniques were essentially used as described (Sambrook et al., 1989). Recovery of DNA fragments from agarose was achieved by spin-column purification (Amicon) as detailed by the manufacturer. The PCR-generated DNA fragments were gel purified, digested with restriction enzymes PstI and SphI and cloned back into the pAF19 plasmid. The clones were screened by PCR and plasmid DNA from one positive clone was isolated, and the sequence was confirmed using the DYEnamic ET terminator cycle sequencing kit (Amersham Biosciences). Plasmids carrying the different YopE variants were introduced into the Yersinia strains by electroporation (Conchas and Carniel, 1990) using a Gene Pulsar apparatus (Bio-Rad).
Construction of single-amino-acid substitution mutants in cis was facilitated by lifting the mutated region from the corresponding pAF19-derived construct by PCR and subcloning it into the pCR4-TOPO TA cloning vector (Invitrogen). After confirmation of the mutation by sequencing, the fragment was cloned into the XbaI- and SphI-digested suicide mutagenesis vector pDM4 (Milton et al., 1996). E. coli S17-1λpir was used as the donor strain in conjugal mating experiments with Y. pseudotuberculosis YPIIIpIB102 (Bolin and Wolf-Watz, 1984) as the recipient strain. For selection of appropriate allelic exchange events, we used established methods (Milton et al., 1996). All mutants were confirmed by sequencing the targeted area of yopE.
Bacteria grown overnight in LB broth at 26°C were harvested and resuspended in PBS. Groups of C57BL/6, Scanbur BK, female mice with three animals in each group were administrated intraperitoneally with increasing doses [from 104 to 107 colony-forming units (cfu) per ml] of the different strains. The LD50 calculation (Reed and Muench, 1938) should not be seen as a standard LD50 experiment as mice were sacrificed when they showed signs of severe sickness. The animal infection study was approved by the Swedish National board for Laboratory Animals.
Cultivation of HeLa cells
HeLa cell cultures were routinely maintained as previously described (Rosqvist et al., 1990). For the in vivo GTPase activation assay, cells were seeded on 100 mm tissue culture plates and incubated at 37°C with 5% CO2 until they reached 70% confluency. The cells were washed twice with PBSA and antibiotic-free medium was added 16 h before bacterial infection.
The cytotoxicity assay
For the cytotoxic assay, cells were plated on coverslips (12 mm diameter) placed in 24-well tissue culture plates. HeLa cells were seeded on coverslips and incubated overnight at 37°C with 5% CO2. Bacteria grown overnight were diluted in fresh cell culture media and incubated for 30 min at 26°C, followed by 1 h at 37°C before infection of HeLa cells were carried out. The HeLa cells were observed over a 3 h period. Coverslips were fixed in 2% paraformaldehyde at 45 min and 2 h after infection, respectively, and the cytotoxicity was assessed by phase contrast microscopy. Infection of HeLa cells and analysis of the cytotoxic effect (altered morphology of cultured cells) were performed as described previously (Rosqvist et al., 1990). Cytotoxicity is characterized by a destruction of the actin cytoskeleton of the target cell, resulting in rounding up of the cell (Rosqvist et al., 1991).
Localization of YopE in infected HeLa cells
The HeLa cells were seeded on coverslips as described above. Bacteria grown overnight at 26C were diluted to OD600 0.1 in fresh LB media and incubated for 1 h at 26°C, followed by 1 h at 37°C before infection of HeLa cells was carried out with a multiplicity of infection (moi) of 10. Three hours after infection the infected cells were fixed in 2% paraformaldehyde, permeabilized in 0.5% Triton X-100, and processed for indirect immunofluorescence labelling using affinity purified rabbit anti-YopE antibodies followed by a donkey anti-rabbit antibody conjugated to Alexa488 (Molecular Probes, Eugene, OR). The HeLa cell membrane was labelled with Wheat Germ agglutinin conjugated to Rhodamine (Molecular Probes, Eugene, OR). The specimen were analysed using a Leica SP2 laser scanning confocal microscope and the green and the red channels were scanned sequentially and stacks of images were obtained with a image size of 512 × 512 pixels and a step size of 0.40 µm.
Quantification of translocated YopE
Protease protection and digitonin extraction of translocated Yersinia effector proteins were carried out as previously described (Nordfelth and Wolf-Watz, 2001; Aili et al., 2003). In short, bacteria grown overnight at 26°C in LB were subcultured in MEM supplemented with 10% heat-inactivated fetal calf serum and pre-grown for 30 min at 26°C and 1 h at 37°C. HeLa cell monolayers grown to 80% confluence in 10 cm diameter tissue culture dishes were infected in duplicates with either wild-type or mutant Yersinia at a moi of 10. After 3 h infection, all monolayers were treated with proteinase K (Roche, Indianapolis, USA) in order to remove extracellular proteins. One set of monolayers were lysed with 1% digitonin (Fluka, Buchs, Switzerland) and the cytoplasmic fraction was separated from membranes and bacteria by centrifugation at 4°C for 10 min at 13 000 r.p.m. The supernatant was analysed by SDS-PAGE and ECL-Western blotting (Amersham Biosciences) using antiserum raised against YopE (AgriSera, Sweden). The other set of monolayers were not treated with digitonin, but incubated with PBSA before sample collection and centrifuged at room temperature for 5 min at 5000 r.p.m. to clear the supernatant of bacteria, cell debris and intact cells. The supernatant was analysed as described above. All experiments were reproduced at least three times. Quantification of the YopE signal was performed with a Fluor-STM MultiImager (Bio-Rad) and the Quantity One software.
Analysis of Yop secretion and expression
Bacterial strains were grown overnight at 26°C in BHI media with or without Ca2+. An 0.05 volume of overnight culture was diluted into 2 ml of fresh BHI media and incubated for 30 min at 26°C and 2 h at 37°C. The cultures were centrifuged at 13 200 r.p.m. for 2 min. After centrifugation, 2 µl of sample from the supernatant was collected into loading buffer and analysed using 12% SDS-PAGE followed by immunoblotting with total Yop antisera (AgriSera, Sweden). For analysis of Yop expression and secretion, bacterial strains were grown in BHI media depleted of Ca2+ and treated as described above. Supernatant (secreted bacterial proteins) and total sample (secreted and bacterial-associated proteins) were analysed as described above.
Bacterial protein expression in the presence of eukaryotic cells
A total of 5 × 105 HeLa cells were seeded in four of the wells in a six-well tissue culture tray (plus cells), while the remaining wells were overlaid with media alone (minus cells). The next day, wells were washed twice with PBSA and overlaid with 500 µl of antibiotic-free culture media. A volume of 500 µl of pre-induced bacteria (as above) was added to triplicate wells, two with and one without cells. In addition, to one of the ‘plus cell’ wells, chloramphenicol was added at 25 µg ml−1 to inhibit bacterial protein synthesis. After 90 min incubation, 250 µl of 4x loading buffer was added to each well and the total content recovered. Equal amount of protein was analysed by 12% SDS-PAGE and immunoblotted with total Yop antisera (AgriSera, Sweden).
LDH release assay
A total of 2 × 104 HeLa cells were seeded into wells of flat-bottomed 96-well plates and grown overnight as described above. The cells were washed twice in PBSA and once in RPMI media, before 150 µl of RPMI was added to each well. Bacteria grown overnight at 26°C were diluted to 0.05 OD600 in 4 ml of fresh Ca2+-depleted BHI media and then incubated for 30 min at 26°C and 1 h at 37°C. Cells were infected with 20 µl of bacteria and the infected cells were centrifuged for 5 min at 1400 r.p.m. and incubated at 37°C with 5% CO2. After 3 h of incubation, the plates were centrifuged for 5 min at 1400 r.p.m. and the supernatants were assayed for LDH release by using the CytoTox 96 assay kit (Promega) accordingly to the manufacturer’s instructions. After 30 min incubation with the substrate, the reaction was stopped and absorbance at 492 nm was determined in an ELISA reader. The experiment was repeated four times and the amount of LDH released from cells infected with the respective strain was normalized to the amount released from cells infected with the pIB29MEKBA strain. The percentage of LDH release was calculated using the following formula: percentage of LDH released = (LDH release sample – LDH release pIB29MEKBA) × 100/(LDH release pIB29MEKBA).
In vitro GTPase assay
Wild-type and mutant YopE GAP domains (amino acids 90–219) were expressed and purified as described previously (Von Pawel-Rammingen et al., 2000). Recombinant wild-type Rho GTPases were produced and purified as GST fusion proteins (Self and Hall, 1995a) and in vitro GAP activity was measured as described previously (Self and Hall, 1995b). In short, Rho GTPases were loaded with [γ-32P]-GTP for 5 min at 37°C in loading buffer (50 mM Hepes, pH 7.3, 5 mM EDTA, 5 mg ml−1 BSA). Cold GTP to a final concentration of 0.1 mM was added together with hydrolysis buffer (50 mM Hepes, pH 7.3, 10 mM MgCl2, 1 mM DTT, 100 mM KCl, 0.1 mg ml−1 BSA). Purified YopE protein was added to 25 µl of loading reaction with a final concentration of Rho GTPases of 11 nM and incubated at 16°C for Cdc42 and Rac1, and 20°C for RhoA. The hydrolysis reaction was stopped after 20 min by adding 1 ml of ice-cold stop solution (50 mM Hepes, pH 7.3, 20 mM MgCl2, 1 mM DTT, 10 µg ml−1 BSA, 0.1 mM cold GTP) and the GTPase activity was analysed by a filter-binding assay (Self and Hall, 1995b).
In vivo Rac1, RhoA and Cdc42 activation assays
GTPase fusion proteins were purified and employed as described previously: GST-PAK-Rac-BD and GST-WASP-Cdc42-BD (Sander et al., 1999) and GST-Rhotekin-Rho-BD (Ren et al., 1999) (where GST is glutathione S-transferase, PAK is p21-activated kinase, WASP is Wiskott–Aldrich syndrome protein and BD is binding domain). Cells were infected as indicated in the figure legends (Figs 7 and 8). Cells were infected with bacteria at an moi of 10 as indicated in the figure legends (Figs 7 and 8). After infection cells were washed twice in ice-cold PBSA and lysed on ice in lysis buffer (Rac1/Cdc42: 1% Triton X-100, 100 mM NaCl, 50 mM Tris/HCl, pH 7.5, 15 mM MgCl2 and 1 mM EDTA, RhoA: 1% Triton X-100, 500 mM NaCl, 50 mM Tris/HCl, pH 7.2, 0.5% sodium deoxycholate, 0.1% SDS and 10 mM MgCl2). Lysates were cleared by centrifugation at 15 000 g for 10 min at 4°C. Equal amounts of cell lysates were incubated with GST fusion protein beads at 4°C for 30 min (Rac1 and Cdc42) or 45 min (RhoA), washed and bound GTPase was separated on SDS/PAGE, followed by immunoblotting. Whole-cell lysates (5 µg) were also analysed for the presence of GTPase, as indicated in the figure legends (Figs 7 and 8). Each series of infection and subsequent GST fusion pull-down assay was carried out at least three times for each endogenous GTPase. Depending on cell conditions and different batches of GST-GTPase-BD, the GTPase-bound protein accounts for 1–5% of total GTPase.
Polyclonal YopE antibodies were produced by Agrisera (Umeå, Sweden). Monoclonal antibodies towards RhoA, Cdc42 and Rac1 were purchased from Santa Cruz Biotechnology. Glutathione-Sepharose 4B was obtained from Amersham Biosciences.
We thank Dr Ruth Palmer and Dr Matthew Francis for critical review of the article. This work was supported by grants from the Swedish Research Council, the Swedish Foundation of Strategic Research, the Swedish Cancer Society and the Kempe Foundation.