ROS-inhibitory activity of YopE is required for full virulence of Yersinia in mice

Summary

YopE, a type III secreted effector of Yersinia, is a GTPase Activating Protein for Rac1 and RhoA whose catalytic activity is critical for virulence. We found that YopE also inhibited reactive oxygen species (ROS) production and inactivated Rac2. How YopE distinguishes among its targets and which specific targets are critical for Yersinia survival in different tissues are unknown. A screen identifying YopE mutants in Yersinia pseudotuberculosis that interact with different Rho GTPases showed that YopE residues at positions 102, 106, 109, and 156 discern among switch I and II regions of Rac1, Rac2, and RhoA. Two mutants, which expressed YopE alleles with different antiphagocytic, ROS-inhibitory, and cell-rounding activities, YptbL109A and YptbESptP, were studied in animal infections. Inhibition of both phagocytosis and ROS production were required for splenic colonization, whereas fewer YopE activities were required for Peyer's patch colonization. This study shows that Y. pseudotuberculosis encounters multiple host defenses in different tissues and uses distinct YopE activities to disable them.

Introduction

Many gram-negative pathogens use a type III secretion system to translocate effector proteins into host cells which then modulate host cell-signaling pathways and enable the pathogen to establish an infection. Effectors with GTPase Activating Protein (GAP) activity, such as YopE of Yersinia spp., SptP of Salmonella enterica, and ExoS of Pseudomonas aeruginosa, inactivate Rho family GTPases (Black and Bliska, 2000; Fu and Galan, 1999; Goehring et al., 1999; Von Pawel-Rammingen et al., 2000). Rho GTPases affect many processes that are critical for host defenses, including actin cytoskeleton rearrangements, MAP kinase pathways, and generation of ROS (Bishop and Hall, 2000; Etienne-Manneville and Hall, 2002; Sorokina and Chernoff, 2005). Rho GTPases cycle between an active GTP-bound form and an inactive GDP-bound form. The active GTP-bound form differs structurally from the inactive GDP-bound form at two regions, switch I and switch II (Paduch et al., 2001). These switch regions interface with GAPs and are highly conserved among small GTPases (Bishop and Hall, 2000; Etienne-Manneville and Hall, 2002; Wennerberg and Der, 2004). The slow intrinsic GTP-hydrolysis activity of Rho GTPases is accelerated by GAPs (Bishop and Hall, 2000; Etienne-Manneville and Hall, 2002; Sorokina and Chernoff, 2005).

Bacterial GAPs studied to date are structurally homologous to each other, but not to eukaryotic GAPs (Evdokimov et al., 2002; Rittinger et al., 1997; Stebbins and Galan, 2000; Wurtele et al., 2001). Each bacterial GAP can inactivate several Rho GTPases. However, structural analyses of bacterial GAPs and Rho GTPases have not provided clues about how specific bacterial GAPs target different Rho GTPases. Purified YopE inactivates RhoA, Rac1, and Cdc42 whereas in cultured cells, YopE inactivates RhoA and Rac1, but not Cdc42 (Aili et al., 2006; Andor et al., 2001; Black and Bliska, 2000; Von Pawel-Rammingen et al., 2000). In contrast, SptP inactivates Rac1 and Cdc42 but not RhoA in vitro (Fu and Galan, 1999), while ExoS inactivates RhoA, Rac1, and Cdc42 both in vitro and in cultured cells (Goehring et al., 1999; Krall et al., 2002). The well-studied Rho family members, RhoA, Rac1, and Cdc42 are most frequently tested as potential targets of bacterial GAPs; however, other Rho GTPases may be important targets during infection since many control critical functions of host defenses (Condliffe et al., 2006; Filippi et al., 2004).

YopE contributes to Yersinia spp. colonization of the small intestine, Peyer's patches (PP), mesenteric lymph node (MLN), and spleen of mice (Black and Bliska, 2000; Logsdon and Mecsas, 2003; Straley and Cibull, 1989; Trulzsch et al., 2004). In mice infected intravenously with Y. pestis, a YopE-β-lactamase fusion protein is detected in neutrophils, macrophages, and dendritic cells of spleens (Marketon et al., 2005), indicating that YopE likely exerts its function directly on these cell types. Bactericidal functions of phagocytes, such as phagocytosis, ROS production, and neutrophil extracellular traps (NETs) formation, are controlled by several Rho GTPases, including RhoA, Rac1, and Rac2 (Bokoch, 2005). In particular, RhoA is important for complement-mediated phagocytosis in macrophages, while Rac1 is critical for Fc receptor- and β1 integrin-mediated phagocytosis (Alrutz et al., 2001; Caron and Hall, 1998). Rac2, which is expressed in myloid cells, also plays a role in phagocytosis driven by Fc receptor in macrophages (Hoppe and Swanson, 2004). In addition, Rac2 is a key component of the NADPH oxidase enzyme and is essential for ROS production (Filippi et al., 2004; Gu et al., 2003). ROS production is induced by agonists binding to their receptors which in turn activate signal transduction pathways to trigger NADPH oxidase assembly. These downstream signal transduction pathways can include a variety of small Rho GTPases, and thus, ROS production is not solely dependent on Rac2 (Condliffe et al., 2006; Kim et al., 2004). However, some agonists, such as PMA stimulate ROS but bypass all known Rho GTPases, except for Rac2 (Condliffe et al., 2006; Filippi et al., 2004; Glogauer et al., 2003; Gu et al., 2003; Kim et al., 2004).

The GAP activity of YopE is essential for virulence since a catalytically inactive mutant, YopER144A is attenuated during infection of mice (Aili et al., 2006; Black and Bliska, 2000). However, it is not known which GAP targets are crucial for virulence and whether different activities are important in different tissues. Initially, the antiphagocytic activity of YopE through inactivation of Rac1 was thought to be YopE's most significant contribution to virulence (Alrutz et al., 2001; Andor et al., 2001; Black and Bliska, 2000). However, the recent findings that YopE modulates translocation of other Yops by inactivation of RhoA (or potentially RhoB, or RhoC), and that avirulent yopE mutants hypertranslocate other Yops into cells infected in culture, raised the intriguing possibility that a critical function of YopE during infection is to regulate translocation rather than, or in addition to, modulating host defenses (Aili et al., 2007; Mejia et al., 2008). Here we use the enteric pathogen Yersinia pseudotuberculosis (Yptb) to dissect the role(s) of YopE during infection. YopE point mutants and a YopE-SptP chimera were investigated to identify YopE proteins that interact with and distinguish among different Rho GTPases. The YopE-SptP chimera contains the secretion and translocation domains of YopE and the GAP catalytic subunit of SptP from Salmonella. Analysis of two YopE mutants showed that distinct activities of YopE are required for colonization of different tissues during mouse infections, indicating that YopE inactivates different host defenses in these tissues.

Results

YopE inhibits ROS production and Rac2

Since Yersinia spp. inhibit ROS production in macrophages and neutrophils (Bliska and Black, 1995; Green et al., 1995; Ruckdeschel et al., 1996) and small GTPases regulate this process, we investigated whether YopE inhibited ROS production. We expressed YopE in a Yersinia pseudotuberculosis (Yptb) strain lacking 5 effector Yops, YptbΔyopEHMOJ (called Δ5) because other Yops have also been implicated in inhibiting ROS production after exposure to various stimuli in different cells (Bliska and Black, 1995; Green et al., 1995; Ruckdeschel et al., 1996). The neutrophil-like HL-60 cells were infected with wild-type Yptb, Δ5 carrying a pBAD33 (Δ5+pBAD), Δ5 carrying a plasmid expressing YopE in pBAD (Δ5+pYopE), or left uninfected for 1 hr to permit Yop delivery. ROS production was then triggered by the addition of phorbol 12-mytistate 13-acetate (PMA), which acts to stimulate protein kinase C (PKC) (Ron and Kazanietz, 1999). ROS production after PMA stimulation requires Rac2 but not RhoA, RhoG or Rac1 (Condliffe et al., 2006; Filippi et al., 2004; Glogauer et al., 2003; Gu et al., 2003; Kim et al., 2004). Uninfected cells produced ROS at higher levels than cells infected with Δ5 (Fig. 1A&B) indicating that pre-infection with Yptb reduced the ability of HL-60 cells to generate ROS in response to PMA. Expression of YopE in Δ5 led to a reduced level of ROS production comparable to the inhibition caused by Yptb (Fig. 1A), suggesting that YopE was sufficient to block PMA-dependent ROS production.

Fig. 1. Inhibition of ROS, and Rac2 by YopE.

(A-B) HL-60 cells infected with wild-type Yptb (WT) + pBAD (■), Δ5 + pBAD (▲), Δ5 + pYopE (▼) at an MOI of 20 or uninfected HL-60 cells (◆) were stimulated for ROS production with 1μg/ml of PMA. (A) One representative experiment in which the y-axis shows the relative light units (RLU) produced during 3 minutes. (B) Plot of average relative ROS level compared to Δ5+pBAD from three experiments. The data was analyzed using ANOVA with Tukey's multiple comparison. * indicates statistical significance with P<0.05. (C-D) HL-60 cells were infected with ΔyopE + pBAD or ΔyopE + pYopE at an MOI of 20. The level of Rac2-GTP was determined by effector pull-down assay. (C) The ratio of the GTP bound form of Rac2 to the total level in uninfected cells was set to 1.0. The numbers below the western blots reflect the ratio of GTP-bound/total GTPase in infected cells compared to that of uninfected cells. (D) The ratio of Rac2-GTP to total Rac2 in uninfected cells was set to 1 and then the ratio of GTP-Rac2 to total Rac2 in cells from the indicated condition (x-axis) relative to the ratio from uninfected cells was calculated. The plot shows the average ratio ± SEM from three independent experiments. The data was analyzed using ANOVA with Tukey's multiple comparison as the post test. * indicates statistical significance with P<0.05.

The superoxide-producing NADPH oxidase complex contains Rac2, which was tested for functional modification by YopE. Rac2 is a Rho family GTPase that is an essential component of the NADPH complex (Bokoch et al., 1991; Glogauer et al., 2003; Gu et al., 2003; Quinn et al., 1992). We determined whether YopE inactivates Rac2 by using effector pull-down assays. HL-60 cells were infected with ΔyopE carrying pBAD or ΔyopE carrying pYopE for various times. The ratio of Rac2-GTP to total Rac2 decreased within 15 minutes in the presence of YopE (Fig. 1C), indicating that YopE prevented ROS production by inactivating Rac2. After 120 minute exposure to YopE, the level of Rac2-GTP decreased to less than less than 15% of the Rac2-GTP level found in cells infected with ΔyopE (Fig 1D). Together, these data indicate that YopE targets Rac2 and prevents ROS formation.

Identification of YopE mutants that distinguish among Rho GTPases

To identify YopE residues that are likely to directly contact Rho GTPases, we threaded the YopE structure onto the co-crystal structure of SptP with Rac1 (Evdokimov et al., 2002; Stebbins and Galan, 2000) and predicted that residues F102, I106, L109, E112, Q151, Q155, F156, and Q180 might directly interact with Rac1 (Fig 2). Each of these residues was individually mutated to an alanine on pYopE. All of the resulting YopE mutants were secreted normally from ΔyopE and Δ5 (data not shown). To identify if these changes permitted or prevented YopE interactions with different Rho GTPases, the mutants were screened for YopE-dependent cell-culture phenotypes that rely on different Rho GTPases. Specifically, inhibition of Yptb internalization by epithelial cells was used to measure Rac1-GAP activity (Alrutz et al., 2001; Black and Bliska, 2000). Epithelial cell rounding was used to measure RhoA-GAP activity (Black and Bliska, 2000). Finally, inhibition of ROS production after PMA stimulation in HL-60 cells was used to measure Rac2-GAP activity (Fig 1). Strains expressing YopE mutants were compared to isogenic control strains containing either pBAD or expressing wild-type YopE. The GAP activities were set to 0% for strains lacking YopE protein and to 100% for strains expressing wild-type YopE protein (Table 1). Alanine substitutions at residues E112, Q151, or Q155 did not result in significant changes in any YopE-dependent phenotypes (Table 1), suggesting that either these residues do not interact with Rho GTPases or that alanines were accommodated at these locations. In contrast, strains expressing YopE mutants with alanine substitutions at F102, I106, L109, F156, or Q180 displayed changes in at least one GTPase-dependent cell-culture phenotype (Table 1).

Fig. 2. Structure model of YopE interacting with Rac1.

The YopE sequence was threaded onto the SptP-Rac1 crystal structure (Evdokimov et al., 2002; Stebbins and Galan, 2000) with YopE (grey), Rac1 (yellow), switch I (pink) and switch II (cyan) depicted. YopE point mutants listed in Tables 1 were tested in three cell-culture assays. When mutated residue results in no change in phenotypes, the residue is colored dark blue. When mutated residue results in a change in all phenotypes, the residue is colored red. When mutated residue affected some but not all YopE phenotypes, the residue is colored purple.

Table 1.

Phenotypes in cell culture of YopE mutants on pBAD33^a

	% Phagocytosis inhibitionb (Rac1-GAP) in ΔyopE	% cell roundingc (RhoA GAP) in ΔyopE	% ROS inhibitiond (Rac2 GAP) in Δ5
Plasmid	Mean ± SEM	Mean ±STD	Mean ± SEM
pYopE	100 ± 0	100 ±	100 ± 0
pBAD	0 ±2	0 ±0	0 ±0
pF102A	102 ± 1	75±20	98 ± 4
pI106A	91 ± 3	33±28	55 ± 4
pL109A	92 ± 1	92±14	73 ± 5
pE112A	99 ± 0	100	93 ± 4
pQ151Ae	99 ± 2	100	85 ± 4
pQ155A	96 ± 2	87.5	91 ± 3
pF156A	100 ± 1	62.5±14	69 ± 9
pQ180A	65 ± 6	50±0	44
pF102D	100 ± 2	41±38	65 ± 4
pF102T	84 ± 2	75±0	80 ± 7
pF102C	84 ± 3	67±14	76 ± 6
pF102R	86 ± 5	75±25	104 ± 7
pF102L	98 ± 1	92±14	94 ± 3
pF102N	100 ± 1	33±23	89 ± 6
pF102Y	100 ± 0	62.5	101 ± 5
pL109T	103 ± 2	69±12	90 ± 7
pL109C	101 ± 1	75	89 ± 6
pF156D	79 ± 3	31±24	53 ± 23
pF156S	90 ± 2	56±12	96 ± 8
pF102NF156A	12 ± 11	17±14	−123 ± 71f
pF102DF156A	−30±20f	8±14	31 ± 14
pESptP	98 ± 2	0±0	67 ± 10
pF102S	62 ± 12	56±12	64 ± 10
pS140N	95 ± 1	87±14	84 ± 4
pG145S	100 ± 2	90±14	96 ± 7
pG152N	3 ± 19	0±0	0
pP177R	99 ± 0	100±0	107
pA190R	98 ± 2	94±12	45

^aYptbΔyopE or Δ5 strains carrying plasmids expressing the indicated Yop mutant were tested for phagocytosis inhibition, cell rounding and ROS inhibition.

^bThe degree of phagocytosis inhibition for YptbΔyopE+pYopE was set to 100% and for YptbΔyopE+pBAD was set to 0%. The phagocytosis inhibition assays were performed in triplicate and repeated 3 times.

^cThe degree of cell rounding by YptbΔyopE+pYopE was determined as described in experimental procedures. All assays were performed at moi of 10-20 and repeated three to five times unless only the mean (of two experiments) is given.

^dThe degree of ROS inhibition assay was measured using Δ5 as the background strain. YptbΔyopE+pYopE was set to 100% and YptbΔyopE+pBAD was set to 0% ROS inhibition. The experiments were performed 3 times in triplicate, unless only the mean (of two experiments) is given.

^eThe Q151A mutant also had a valine substitution at L153.

^fA negative number indicated that on average the indicated strain were internalized at higher levels than the strain containing pBAD or had higher levels of ROS than the strain containing pBAD.

Interestingly, F102A, I106A, L109A, and F156A expressing-mutants behaved like wild-type YopE in some assays and were defective in others, suggesting that these residues are in positions which distinguish among Rho GTPases. For instance, I106A and F156A inhibited phagocytosis to nearly wild-type levels (91% and 100%, respectively), but caused intermediate levels of cell rounding (50% and 75%) and ROS production (55% and 69%) (Table 1). Based on our threaded model of YopE with Rac1, these four residues are predicted to form a binding pocket that interacts with switch I and switch II regions of Rac1 (Fig 2). Random mutagenesis at residues F102, L109 and L156 was performed to identify additional YopE mutants that have retained wild-type GAP activity for some, but not all YopE targets, and the resulting mutants were tested in cell-culture assays (Table 1). Many mutants retained wild-type GAP activity for one or two Rho GTPases but had reduced GAP activity for the other(s), confirming the idea that these residues distinguish among different Rho GTPases. In an attempt to enhance specific defects, double mutants (F102N or F102D with F156A) were made by combining two mutations of similar phenotypes, but both double mutants lost almost all detectable GAP activities (Table 1).

SptP is a GAP for Rac1 and Cdc42 but not RhoA (Fu and Galan, 1999), while YopE is a GAP for RhoA and Rac1 in cell culture assays. To generate a GAP lacking RhoA activity but retaining Rac1 activity in Yptb, we created a chimeric YopE_1-100–SptP_166-293 (ESptP) protein which contained the secretion and chaperone-binding domains of YopE (Sory et al., 1995) and the SptP catalytic domain (Fu and Galan, 1998; Stebbins and Galan, 2000). When expressed in ΔyopE or Δ5 from pBAD, ESptP had antiphagocytic activity (98%) comparable to wild-type YopE, was intermediate for ROS inhibition (67%), and caused no cell rounding (0%) (Table 1). This result indicated that binding to RhoA is likely mediated by its catalytic domain (amino acids 100-219). We next attempted to reduce the RhoA-GAP activity of YopE by substituting neutral or hydrophobic residues in the YopE catalytic domain with the corresponding charged or polar residue from SptP, reasoning that such changes might reduce the affinity of YopE for RhoA. The changes generated were F102S, S140N, G145S, G152N, Q155K, P177R, and A190R. Most mutants showed near wild-type activities in antiphagocytosis and cell-rounding assays, indicating that these single residue changes were not sufficient to alter the RhoA-GAP activity of YopE (Table 1). However, F102S had intermediate levels of activities in all three assays, S140N had reduced cell-rounding activity but maintained nearly wild-type levels of antiphagocytosis, and G152N was unable to cause any YopE-dependent phenotypes (Table 1). These results indicate that specific changes to positions F102 and/or S140 may reduce the RhoA GAP activity of YopE, but that no single change fully obliterated the interaction of YopE with just RhoA.

Eleven yopE mutants or the wild-type yopE gene were recombined at the native yopE locus in YptbΔyopE to generate Yptb strains expressing mutant or wild-type YopE from the yopE locus (for instance, YptbL109A or YptbyopE_rec). All YopE proteins were secreted normally from the recombined strains (data not shown). Their phenotypes were tested in the three cell-culture assays: % phagocytosis inhibition, % cell rounding and % ROS inhibition. In contrast to experiments performed in Table 1, ROS inhibition was performed after infection with Yptb strains expressing all other Yops and the indicated Yop mutant (Table 2). Since PMA stimulates PKC, which in turn triggers phosphorylation of gp47^phox, only Yops which directly alter the NADPH oxidase complex should effect ROS production after stimulation with PMA. We found that ΔyopE was unable to inhibit ROS production after stimulation with PMA indicating that the other Yops, which inhibit ROS production, most likely interfere with signal-transduction cascades leading to the assembly of the NADPH oxidase complex while YopE directly alters Rac2, a component of the NADPH oxidase complex.

Table 2.

Phenotypes of YopE mutants expressed from the yopE locus.

	% phagocytosis inhibitiona (Rac1)	% cell roundingb (RhoA)	% ROS inhibitionc (Rac2)
Strain	Mean ± SEM	Ave±SEM	Mean ± SEM
YptbyopErecd	100 ± 0	100	100 ± 1
Yptb Δ yopEd	0 ± 2	0	0 ± 3
YptbF102L	98 ± 0	100±0	99 ± 3
YptbF102N	91 ± 3	70±10	68 ± 4
YptbF102R	95 ± 1	67±20	75 ± 4
YptbF102C	93 ± 2	75±0	86 ± 6
YptbF102T	88 ± 3	75±25	79 ± 2
YptbL109Ad	97 ± 0	70±11	72 ± 5
YptbF156A	96 ± 1	70±11	72 ± 3
YptbF156S	90 ± 2	62.5±14	66 ± 5
YptbF102D	80 ± 2	50±15	52 ± 5
YptbI106A	81 ± 3	15±22	38 ± 9
YptbESptPd	83 ± 3	0±0	34 ± 6

^aThe degree of phagocytosis inhibition for YptbyopE_rec was set to 100% and the YptbΔyopE mutant was set to 0%. For phagocytosis inhibition and ROS inhibition, each strain was tested in triplicate in at least three independent experiments and the SEM was calculated.

^bThe degree of cell rounding caused by infection with YptbyopE_rec was given a score of 4 (100%), while the cell rounding observed after infection with YptbΔyopE was given a score of 0 (0%) as described in experimental procedures. A score of 0-4 was given to cells infected with YptbyopE mutants. Each strain was tested at least three times.

^cThe level of ROS inhibition in HL-60 cells infected with YptbyopE_rec upon PMA stimulation was set to 100% while the level of ROS produced by cells infected with YptbΔyopE was set to 0% ROS. The level of ROS produced by cells infected with yopE mutants was compared to levels produced by YptbΔyopE. Each strain was tested in three independent experiments and the SEM was calculated.

^dThese strains were used in animal infection experiments described in Fig 4 and Fig 5.

Three classes of YopE mutants were apparent in these three phenotypic assays (Table 2). YptbF102L was comparable to YptbyopE_rec in all assays. YptbF102N, YptbF102R, YptbF102C, YptbF102T, YptbL109A, YptbF156A and YptbF156S mutants had near wild-type Yptb levels of phagocytosis inhibition but caused less rounding (75%) and inhibited ROS production to approximately 65-75% of YptbyopE_rec. YptbF102D, YptbI106A, and YptbESptP had more severe defects. All three inhibited phagocytosis to only 80% that of YptbyopE_rec, caused significantly less cell rounding (0%-50%), and had only mild inhibitory effects on ROS production (34-52%), suggesting that they retained only modest RhoA- and Rac2-GAP activities. One mutant with defects representative of each class, YptbL109A and YptbESptP (indicated in bold in Table 2), was chosen for further investigation.

RhoGAP activities of purified L109A and ESptP

We determined whether the in vitro GAP activities of purified L109A and ESptP correlated with the phenotypes of YptbL109A and YptbESptP in cultured cells. Purified His₆-RhoA, His₆-Rac1, or GST-Rac2 was incubated with purified GST-YopE, GST-YopEL109A, GST-ESptP, GST-YopER144A (a catalytically inactive YopE), or TBS buffer at a 10:1 molar ratio of Rho GTPase to GAP. The amount of free γ-phosphate released as a result of GTP hydrolysis caused by each mutant was measured by a colorimetric assay. In general, the activities of purified L109A and ESptP mirrored the cell-culture phenotypes of YptbL109A and YptbESptP with two exceptions (Fig. 3). L109A had GAP activities on RhoA, Rac1, and Rac2 that were slightly lower but statistically indistinguishable from that of wild-type YopE protein and were significantly higher than the R144A protein. This result indicates that L109A retains most of its GAP function for RhoA, Rac1, and Rac2 in vitro (Fig 3A-C). Purified ESptP lacked all RhoA-GAP activity (Fig. 3A) and had intermediate levels of Rac1 and Rac2 GAP activity. ESptP-GAP activity was significantly different from wild-type YopE for Rac1- and Rac2-GAP activities and from R144A for Rac2-GAP activity, but ESptP was not statistically significantly different from R144A for Rac1-GAP activity (although its level of GAP activity appeared higher). The Rac2-GAP and RhoA-GAP activity results were consistent with phenotypes observed by YptbESptP regarding its partial inhibition of ROS production and its inability to induce cell rounding (Table 2). However, YptbESptP partially inhibited of phagocytosis whereas the Rac1-GAP activity of the purified protein was statistically indistinguishable from R144A. It is possible that YptbESptP retained other GAP activities that enable YptbESptP to resist phagocytosis. There was a second modest discrepancy between the cell-culture phenotypes of YptbL109A and YptbESptP and the in vitro activities of L109A and ESptP. Purified L109A and ESptP showed similar Rac2-GAP activity in vitro (Fig 3C). In contrast, YptbL109A inhibited ROS production by 72% while YptbESptP inhibited ROS production by only 34% (Table 2). This difference may reflect an inability of ESptP to traffic to Rac2-rich areas, an inability of ESptP to compete with other factors for binding to Rac2 in cultured cells, and/or that ESptP may target and be sequestered by another small Rho-GTPase in HL-60 cells.

Fig. 3. GAP activities of purified L109A and ESptP and YptbL109A and YptbESptP.

Purified GST-YopE mutants were incubated with purified His₆-RhoA (A), His₆-Rac1 (B), or GST-Rac2 (C) at a ratio of 10 Rho: 1 GAP in the presence of GTP for 20 minutes at 37°C. CytoPhos reagent was added, and the level of free phosphate released as a result of GTP hydrolysis was measured by OD₆₅₀. Experiments were repeated at least three times and shown as mean ± SEM. * indicates that the value was statistically different from the indicated protein using ANOVA with Tukey's multiple comparison as the post test with P<0.05. (D) HEp-2 cells were infected at MOI of 20 for 1 hour, lysed, run on SDS-PAGE and probed for YopM, YopE, and actin by Western blot. Translocation from each strain was normalized to the actin loading control and compared to the wild-type Yptb level, which was set to 1.0. The experiment was repeated twice and a representative blot is shown.

Translocation levels of Yops from YptbL109A and YptbESptP correlates with RhoA inactivation

To investigate whether translocation of other Yops was altered after infection with YptbL109A and YptbESptP, the amounts of translocated YopM and YopE in infected HEp-2 cells were determined by Western blot. Both ΔyopE and YptbESptP hyper-translocated YopM, consistent with their lack of RhoA-GAP activity (Fig. 3D). (While the chimeric ESptP can be detected with YopE antibody, the levels of YopE and ESptP cannot be compared directly because our YopE polyclonal antibody was raised to full length YopE, much of which is missing in the ESptP chimera.) YptbL109A translocated two times more L109A and four times more YopM into cells than wild-type Yptb (Fig. 3D), consistent with the observations that L109A has more RhoA-GAP activity than ESptP, but slightly less than wild-type YopE (Table 2 and Fig 3A).

In summary, these combined data (Table 2 and Fig 3) indicate that while more mutant YopE and other Yops are translocated into host cells by YptbL109A and YptbESptP, the cell culture phenotypes of these strains correlated well with the GAP activities of purified YopE mutants. Specifically, L109A had wild-type YopE levels of Rac1-GAP and lower but detectable Rac2-GAP and RhoA-GAP activities correlating well with the wild-type levels of antiphagocytosis and detectable but lower levels of cell rounding and ROS-inhibitory functions of YptbL109A. ESptP had lower Rac1-GAP and Rac2-GAP activities than YopE and no detectable RhoA-GAP activity correlating well with the abilities of YptbESptP to partially prevent phagocytosis and ROS production and its inability to cause cell-rounding.

YptbL109A colonizes the Peyer's patches, but not the spleen, while YptbESptP is defective in colonizing all tissues

YptbL109A and YptbESptP were tested in mouse-infection studies to determine if activity on specific GAP targets were required for colonization of different tissues. Since YptbΔyopE is more defective in colonizing tissues in competitions with Yptb compared to single strain infections (Logsdon and Mecsas, 2003, 2006), YptbyopE_rec, YptbL109A, YptbESptP, and YptbΔyopE were tested in competition infections with a kanamycin-resistant, but otherwise wild-type Yptb strain, denoted YptbKan (Mecsas et al., 2001). Five days after oral inoculation, mice were sacrificed and the competitive indices (CIs) of each mutant compared to YptbKan were determined in the PP, MLN, and spleen. YptbL109A colonized the PP as well as YptbyopE_rec (Fig. 4). The YptbL109A strain colonized the PP as well as WT and significantly better than YptbESptP and ΔyopE (Fig. 4A). These results suggest that full ROS inhibition (Rac2-GAP) and RhoA-GAP activities of YopE are not required for normal colonization of PP. In contrast to its ability to colonize PP, YptbL109A did not colonize the spleen as well as Yptb (Fig. 4C). This result indicated that wild-type levels of antiphagocytic activity alone are not sufficient for splenic colonization, and that either full ROS inhibition and/or cell-rounding activities are needed. YptbESptP failed to efficiently colonize the PP and MLN (Fig. 4A-B). The finding that YptbESptP could not colonize lymph tissues indicated that either 83% antiphagocytosis activity was not sufficient for colonization and/or that more than 34% ROS inhibitory and/or some cell-rounding activities were required for lymph-node colonization. Combined, these results demonstrate that different thresholds of YopE activities are required for colonization of different tissues. In the lymph tissues, either full Rac1-GAP activity is needed and/or some targeting of RhoA and Rac2 by YopE is required for Yptb colonization. In the spleen, either full Rac2- and/or RhoA-GAP activity is required for colonization or some Rac1-GAP activity is necessary for Yptb colonization.

Fig. 4. Competition infection of YptbKan and Yptb, YptbL109A, YptbESptP, or ΔyopE in BALB/c mice.

(A-D) BALB/c mice were infected orogastrically with a total of 2×10⁹ CFU/mouse of a 1:1 mixture of YptbKan and either Yptb, YptbL109A, YptbESptP, or YptbΔyopE. Day 5 post infection (A) PP, (B) MLN and (C) spleens, were harvested, homogenized, and plated on Yersinia selective plates. The CI from each mouse is represented as a closed circle. Open circle indicates that one strain of bacteria was not recovered and the value was calculated as if one colony was recovered from that strain. The geometric mean is shown as a horizontal bar in the graph and is listed below as the mean. Statistical analysis was determined by ANOVA followed by Tukey's as the post test with P value < 0.05. * under the strain name indicates that the strain is statistically different from the strain in the left column.

Antiphagocytic and ROS inhibitory activities of YopE are required for splenic colonization

To investigate whether ROS inhibition is sufficient for Yptb colonization of spleens, Cybb^−/− mice, which lack the component, gp91^phox, of the NADPH oxidase enzyme responsible for O₂⁻ production, were used in infection (Pollock et al., 1995). These mice lack the ability to generate ROS, but still have Rac2 and the neutrophils still migrate to infected tissues (Pollock et al., 1995). We observed high scatter of data in the spleen after oral infection (Fig. 4C) due to a bottleneck that limits the number of Yptb reaching the spleen after oral inoculation (Mecsas et al., 2001). Therefore, an intravenous infection model was used to provide Yptb direct access through the bloodstream to the spleen. If ROS inhibition is the only YopE function required for spleen colonization, then ΔyopE should colonize the spleen as well as YptbKan in Cybb^−/− mice. However, YptbΔyopE was still attenuated in the spleen of Cybb^−/− mice (Fig. 5), indicating that other YopE functions in addition to ROS inhibition, are important for surviving in the spleen.

Fig 5. Competition infection with YptbKan and YptbL109A, YptbESptP, or YptbΔyopE in C57Bl/6 and Cybb−/− mice.

C57BL6/J mice and Cybb^−/− mice were infected intravenously with a 1:1 mixture of YptbKan and each of the mutants at a total dose of 100 CFU/ mouse. Day 5 post infection, spleens were harvested, homogenized, and plated. Each closed circle represents the CI from one mouse. The fold difference between geometric mean of a strain's CI in Cybb^−/− mice compared to its CI in wild-type mice is indicated below each strain. * indicates the P value <0.05 as determined by Mann-Whitney test.

To determine whether YptbL109A and YptbESptP both of which have retained anti-phagocytic activity, could successfully colonize C57Bl/6 or congenic Cybb^−/− mice, mice were infected with a mixture of YptbKan and YptbL109A or YptbESptP. YptbL109A and YptbESptP were attenuated in colonizing the spleen of wild-type mice with CIs of 0.13 and 0.04, respectively (Fig. 5). In contrast, YptbL109 and YptbESptP colonized the spleen almost as well as YptbKan in Cybb^−/− mice with CIs of 0.7 and 0.5, respectively (Fig. 5). Since YptbESptP, which lacks RhoA-GAP activity, could colonize the spleen of Cybb^−/− mice, RhoA-GAP activity was not required for splenic colonization. These results show that both antiphagocytic and ROS inhibitory activities are required for splenic colonization.

Discussion

Many Yersinia Yops have multiple molecular targets, which could reflect the extensive animal host range of these pathogens. In addition, since Yersinia spp. colonize many tissues, the broad range of cellular targets may allow Yersinia to disable a variety of host defenses present in different tissues. Dissecting which activities are crucial for virulence should reveal insights into the host defenses encountered by Yersinia in different organs. This study expands on the number of known targets and functions of YopE to include Rac2 and inhibition of ROS production, identifies an interface in YopE that distinguishes among its Rho GTPase targets at the switch I and switch II regions, and shows that different levels of YopE activities and/or distinct YopE activities are required for infection in different tissues. Furthermore, a YopE chimera, ESptP, which had reduced Rac2-GAP activity, was virulent in mice lacking NADPH oxidase. Combined, these data strongly support the idea that Rac2 is a critical target of YopE during murine infection.

The inactivation of Rac2 by YopE adds another important facet to the ways in which YopE disables critical innate host defenses. YopE prevents phagocytosis by mammalian cells, contributing to the extracellular nature of Yersinia (Alrutz et al., 2001; Grosdent et al., 2002; Guinet et al., 2008; Rosqvist et al., 1990; Simonet et al., 1990). However, neutrophils, which migrate to tissues in response to Yersinia infection (Fisher et al., 2007; Guinet et al., 2008; Lathem et al., 2005; Logsdon and Mecsas, 2006), can kill extracellular bacteria by releasing granule contents, ROS, and NETs (Brinkmann et al., 2004; Brinkmann and Zychlinsky, 2007; Lacy and Eitzen, 2008; Nathan, 2006). By inactivating Rac2 as well as Rac1, YopE can prevent phagocytosis, degranulation of primary granules, and ROS production (Glogauer et al., 2003; Koh et al., 2005; Lacy and Eitzen, 2008). ROS production is required for NETs formation and protease activity (Reeves et al., 2002; Watts, 2006), so inactivation of Rac2 by YopE should have drastic inhibitory effects on antimicrobial functions of neutrophils and permit Yersinia survival in their presence. Since induction of NETs formation results in tissue injury and is used when preliminary mechanisms like phagocytosis does not work (Clark et al., 2007), it may behoove Yersinia and other extracellular pathogens to disable NETs production in order to prevent tissue damage and inflammation.

How does YopE distinguish among different Rho GTPases? Mutating residues F102, I106, L109, and F156 of YopE, which are in close proximity to each other and contribute to the interface of YopE with switch I and switch II, resulted in changes to some but not all YopE-dependent phenotypes. Thus, these residues distinguish among different Rho GTPases. Interestingly, the switch regions of the Rho family GTPases are highly homologous. Furthermore, the Rac1, Rac2, and RhoA residues V36, F37, D63, Y64, and L67 (by Rac1 numbering) which are predicted to make contact with F102, I106, L109, and F156 are identical. However, based on structural analysis of Rac1 and RhoA bound to GAPs (Rittinger et al., 1998; Stebbins and Galan, 2000), the conformation of these conserved switch residues are slightly different (Fig. 6). These conformational differences likely contribute to the ability of YopE mutants to interact with Rac1 versus RhoA. For example, F37 of Rac1 faces toward F102 and F156 of YopE (Fig. 5A), while F39 of RhoA faces away from F102 and F156 (Fig. 6B). When YopE residues F102 or F156 were mutated to non-hydrophobic residues, the hydrophobic interaction of YopE with Rac1 may have been better maintained than in RhoA because of the conformation of F37. Likewise, L109 is in closer proximity to D63 and Y64 in Rac1 then in RhoA (Fig. 6D-F) which is consistent with our findings that changes at L109 yielded mutants that interact with Rac1 better than RhoA.

Fig. 6. YopE residues that confer differential phenotypes interact with conserved residues in Rac1 and RhoA which are oriented differently.

Models of YopE residues interacting with Rac1 (A, D), RhoA (B, E), and merge of Rac1 and RhoA (C, F) are shown. YopE residues that confer differential phenotypes are colored purple. Rac1 residues are colored yellow. RhoA residues are colored white. Structures of SptP-Rac1 complex (Stebbins and Galan, 2000) and RhoA-RhoGAP (Rittinger et al., 1997) were used for Rac1 and RhoA models.

In addition to the F102-F156 interface identified here, there are other residues which enable bacterial GAPs to distinguish among Rho-GTPases. Residue 183 of YopE targets specific Rho-GTPases based on the observation that a YopET183A mutant retains Rac1-GAP activity but has low RhoA-GAP activity (Aili et al., 2006). Interestingly, the corresponding residue in SptP is also a threonine, but SptP does not target RhoA. Therefore, a threonine at this position is not solely responsible for recognition of RhoA by YopE and specificity is context dependent. The lack of RhoA-GAP activity from the purified ESptP chimera suggested that there are residues in the SptP catalytic domain that prevent SptP from interacting with RhoA. We were unable to convert YopE's GAP activities to SptP's by substituting any one of a number of single YopE residues with the corresponding SptP residue, which suggests that several residues may be critical.

Curiously, no YopE mutants were identified that had lower Rac1-GAP activity while maintaining wild-type levels of RhoA or Rac2 GAP activity. There are several plausible explanations for this observation. For example, YopE may have the strongest affinity for Rac1. In fact, results from several studies support the idea that Rac1 rather than RhoA is the preferred target of YopE in cultured cells, although none of the previous studies investigated Rac2 (Aili et al., 2006; Andor et al., 2001). Alternatively, after Yptb binds β1 integrin receptors, Rac1 is recruited to sites of receptor clustering and is favorably positioned to bind to most of the YopE delivered to cells (Hoppe and Swanson, 2004; Wong and Isberg, 2005). Thus Rac1 may appear to be the preferred target because YopE has immediate access to Rac1 which may sequester YopE from reaching other targets elsewhere in the cell.

What are the important activities and protein targets of YopE during infection? We favor the idea that different functions and/or different levels of YopE activities are required in different tissues during murine infections. The YopE functions required for splenic colonization were distinct from those in lymph tissues as demonstrated by the growth of YptbL109A in the PP and MLN but not the spleen. The observation that YptbL109A was attenuated after intravenous infection in wild-type mice indicates that its inability to reach the spleen after oral infection was not because of a bottleneck exiting the GI tract, but rather because YptbL109A cannot survive in the blood and/or spleen. Furthermore, this result indicated that a wild-type YopE level of antiphagocytic activity and partial ROS inhibitory and cell rounding functions were not sufficient for splenic colonization. This stricter requirement for colonizing spleens suggests a stronger host defense in reaching or replicating within spleens. YptbL109A and YptbESptP were able to survive in the spleens in Cybb^−/− mice whereas ΔyopE was not, strongly supporting the idea that both inhibition of phagocytosis as well as inhibition of ROS are critical for spleen colonization. Interestingly, YopH also inhibits ROS production stimulated by IgG opsonized Yersinia or serum opsonized zymosan (Bliska and Black, 1995; Ruckdeschel et al., 1996). During infection, YopE and YopH might act in neutrophils to rapidly and efficiently down-regulate pathways leading to ROS production in order to promote Yersinia survival.

While our finding strongly suggest that antiphagocytosis is critical for spleen colonization, a YopE mutant, F178A, which had very little Rac1-GAP activity, was almost as virulent as wild-type Yptb in LD₅₀ studies after intraperitoneal (i.p.) infection suggesting that Rac1-GAP activity is not critical for lethality (Aili et al., 2006). However, it is possible that this mutant may have retained Rac2-GAP activity which may have provided significant virulence after i.p. inoculation. Alternatively, the functions of YopE required for lethality after i.p. inoculation could be different than the functions of YopE required for lymph node, and splenic colonization after oral infection or intravenous infection. Finally, YopE recently has been shown to have RhoG-GAP activity (Mohammadi and Isberg, 2009; Roppenser et al., 2009). RhoG serves an important role in internalization of Yersinia by epithelial cells and in ROS production after stimulation with fMLP (Condliffe et al., 2006; Mohammadi and Isberg, 2009). F178F may have retained RhoG-GAP activity, which could also explain why F178A retains its virulence as prevention of ROS by YopE may be essential.

The question remains as to the role of RhoA-GAP activity of YopE during Yptb infection. RhoA-GAP activity is important for Yptb to limit translocation of Yops and cause epithelial cell rounding in cultured cells. In the lymph tissues, it is unclear whether the full Rac1 activity of YptbL109A, its partial Rac2 and/or RhoA activities, or a combination of the three activities were sufficient for YptbL109A colonization. In contrast, YptbESptP which lacked all RhoA-GAP activity and had reduced Rac1 and Rac2-GAP activities was unable to colonize the GI tract and lymph tissues. So RhoA may play a role in lymph node colonization or full Rac1-GAP and/or higher Rac2-GAP activities may be required for lymph node colonization. Our observation that YptbL109A translocates more Yops into host cells yet is able to colonize the PP as well as wild-type Yptb, suggests that limiting Yop translocation is not an essential function of YopE for GI tract and lymph node colonization. Aili et al., identified two YopE mutants, W181A and T183A, which lacked RhoA-GAP activity and were avirulent after i.p, infection, which suggests that limiting translocation is critical for lethality (Aili et al., 2006). In the light of our findings, it seems plausible that these mutants might also lack Rac2-GAP activity which could also explain their avirulence. Alternatively, their method of measuring virulence (lethality) differed from ours (colonization), and RhoA-GAP activity might be critical for lethality. Based on our findings with YptbESptP infection of Cybb^−/− mice, it is tempting to conclude that RhoA-GAP activity is not required for splenic colonization in wild-type C57Bl/6 mice. However, an important caveat to this interpretation is that in the absence of NADPH oxidase, RhoA-GAP activity may no longer be needed. For example, during infection, Yptb may stimulate ROS production in phagocytes through the RhoA-dependent complement receptor pathway (Kim et al., 2004). In mice lacking gp91^phox, no ROS is produced regardless of stimulation of complement or other receptors. However, in wild-type mice, both the Rac2- and RhoA-GAP activities of YopE may disrupt neutrophil ROS production.

In conclusion, we present evidence that different tissues may use different mechanisms to combat Yersinia and that YopE inhibits distinct mechanisms in order to cause full virulence. By identifying additional targets of YopE and dissecting various roles of YopE in virulence using bacterial and mouse mutants, our results demonstrate that Yersinia uses different YopE functions and/or different levels of YopE activities to colonize different tissues. Both the functions of YopE to prevent phagocytosis and to inhibit ROS production are critical to defuse neutrophilic attack and permit systemic infection. It seems plausible that Yersinia can successful thwart host immune defenses with relatively few effectors because effectors, such as YopE, can target multiple host processes.

Experimental Procedures

Bacteria Strains and cell lines

Strains and primers are described Table S1 and S2, respectively, and in supplemental material.

Gentamicin Protection Assay (Phagocytosis Inhibition assay)

The gentamicin protection assay was adapted from (Mecsas et al., 1998) with the following modifications. When strains containing pBAD33 plasmids were used, 50 mM arabinose was added to the cultures before shifting to 37°C, and the HEp-2 cell medium was replaced with RPMI without glucose (GIBCO) supplemented with 50 mM arabinose and 5% FBS prior to infection. The percent intracellular YptbΔyopE+pBAD was normalized to 0% antiphagocytic activity, while the percent intracellular YptbΔyopE + pYopE normalized to 100% antiphagocytic activity.

Cell-rounding Assay

HEp-2 cells were infected as in the gentamicin protection assay at an MOI of 10-20. Cells were examined at 30 minute intervals. When cells infected with YptbΔyopE+pYopE (Table 1) or infected with YptbyopErec (Table 2) were well-rounded, a score of 4 was given. Cells infected with YptbΔyopE were given a score of 0. Cells infected with other strains were compared these strains and a score of 0-4 was given depending on the degree of rounding. The scores were averaged, converted to percent and reported in the Tables. For Table 1, scores are averaged from 3-5 experiments except where indicated; for Table 2 scores are averaged from 3-6 experiments.

ROS inhibition assay

This assay was adapted from (Dahlgren and Karlsson, 1999). HL-60 cells were differentiated into neutrophil-like cells by treatment with 1 μM retinoic acid for 1 week (Breitman et al., 1980; Lawson and Berliner, 1999). Cells were washed with Hank's balanced salt solution (HBSS, Cellgro) and resuspended to 1 × 10⁶ cells/ml in HBSS with 100 μM luminol (Sigma) and 10 μg/ml HRP. 1×10⁶ cells were infected at MOI of 20 for 1 hour at 37 °C. 1 μg/ ml phorbol 12-myristate 13-acetate (PMA) was added to infected cells to induce ROS production. Infected cells were immediately read for luminol-amplified chemiluminescence in a SpectraMax M5 plate reader (Molecular Device) every 30 seconds for 3 minutes. The level of ROS production was measured as the area under curve of the chemiluminescence over time (Relative Luminescence Unit x 180 seconds). ROS inhibitory activity of YptbΔyopE+pYopE or YptbyopE_rec was set to 100%, while YptbΔyopEHMOJ (Δ5) or YptbΔyopE ROS inhibitory activity was set to 0%. Statistical differences were determined by one-way analysis of variance (ANOVA) followed by Tukey's Multiple comparison test as the post test with P<0.05 using software from GraphPad Prism 4.

Translocation Assay

The translocation assay was done as described (Davis and Mecsas, 2007) with the following modifications. Overnight bacterial cultures were diluted in 2×YT containing 5 mM CaCl₂ and grown at 26°C for 3 hrs. HEp-2 cells were infected at MOI of 20 for 1 hr at 37°C and lysed for 10 min at 4°C with 0.5% Triton-X-100 in PBS with protease inhibitors (5 μg/ml aprotinin, 5 μg/ml leupeptin, 100 μg/ml AEBSF, and 1 μg/ml pepstatin). Equal amount of proteins from each lysate determined by the Bradford assay (BioRad) were run on SDS-PAGE. Proteins were transferred onto PVDF membrane (Millipore) and probed for YopE, YopM, and actin.

Small GTPase Pull-down assays

6 × 10⁶ HL-60 cells were grown in the presence of serum and infected at MOI of 20 with no spinning. At the time points indicated, the infected cells were washed twice with cold PBS and lysed with 0.5 % Triton X-100 containing protease inhibitors as above. Hereafter, lysates were manipulated on ice or at 4°C. Lysates were clarified by spinning in a microcentrifuge for 15 minutes at maximum speed. Supernatants containing equal amount of proteins (200-300 μg) were incubated with 15 μg of beads-bound to either GST-RBD of PAK (_Cytoskeleton, Denver, CO) to isolate Rac2-GTP. Samples were run on SDS-PAGE, transferred to PVDF, and probed with Rac2 (Upstate) antibodies by Western blot.

YopE purification and in vitro GAP activity

Strains were grown overnight with aeration at 37°C. Cultures were diluted 1:50 in fresh media and grown for 2-3 hours at 37°C. Cultures were grown in the presence of 0.1 mM IPTG overnight at 26°C with the exception of BL21 expressing GST-ESptP which was grown at 20°C. Proteins were purified according to (Black and Bliska, 2000) with GST Mini-spin columns (GE Healthcare). The in vitro GAP assay was performed according to manufacturer's protocol (BK105, Cytoskeleton). Purified His₆-RhoA, His₆-Rac1, or GST-Rac2 was incubated for 20 minutes at 37°C with purified GST-YopE, GST-YopEL109A, GST-ESptP, or GST-YopER144A with an excess of GTP. GTP hydrolysis resulted in release of free phosphate which was measured by the addition of a colorimetric reagent and read at OD₆₅₀. The OD₆₅₀ values were in a linear range as determined by a standard curve. Enzymes were assayed at least three times. Statistical significance was calculated using ANOVA with Tukey's multiple comparison test as the post test with P<0.05.

Infections of mice

Seven to eight-week old female BALB/c mice (NCI) or C57BL/6J mice and Cybb^tm1Din (Cybb_−/−) mice (Jackson Laboratory) were used for infections. Oral infections were done as described (Logsdon and Mecsas, 2003). Briefly, 1:1 mixture of a total of 2×10⁹ YptbKan and YptbyopE_rec or each of the mutants tested was used to infect groups of 2-3 mice. Each strain was tested in 3-4 experiments. At day 5 post infection, mice were euthanized with CO₂. Up to two hundred colonies were patched onto L plates containing kanamycin to determine the CI for each tissue. All the data were combined and the competitive index (CI) was calculated as (ratio of mutant/YptbKan) _output/(ratio of mutant/YptbKan) _input. For statistical analysis, CFU values were transformed logarithmically. Geometric means were determined and statistical significance (p<0.05) between groups was determined by ANOVA followed by Tukey's multiple comparison test as the post test.

For intravenous infections, each mouse received 200 μl of 100 CFU via tail vein injection. Groups of 2-3 mice were used and each strain was tested in 3-4 experiments. Differences were determined by Mann-Whitney P<0.05. Experiments were done with approval of The Institutional Animal Care and Use Committee of Tufts University.