AMPylation of Rho GTPases Subverts Multiple Host Signaling Processes*

Background: VopS is a bacterial effector that AMPylates Rho GTPases. Results: AMPylation of Rho GTPases inhibits their interaction with multiple downstream effectors. Conclusion: VopS conducts a multifaceted host signaling inhibition program by targeting the critical Rho GTPases. Significance: Broadens the understanding of how switch-1 modification of Rho GTPases affects host responses. Rho GTPases are frequent targets of virulence factors as they are keystone signaling molecules. Herein, we demonstrate that AMPylation of Rho GTPases by VopS is a multifaceted virulence mechanism that counters several host immunity strategies. Activation of NFκB, Erk, and JNK kinase signaling pathways were inhibited in a VopS-dependent manner during infection with Vibrio parahaemolyticus. Phosphorylation and degradation of IKBα were inhibited in the presence of VopS as was nuclear translocation of the NFκB subunit p65. AMPylation also prevented the generation of superoxide by the phagocytic NADPH oxidase complex, potentially by inhibiting the interaction of Rac and p67. Furthermore, the interaction of GTPases with the E3 ubiquitin ligases cIAP1 and XIAP was hindered, leading to decreased degradation of Rac and RhoA during infection. Finally, we screened for novel Rac1 interactions using a nucleic acid programmable protein array and discovered that Rac1 binds to the protein C1QA, a protein known to promote immune signaling in the cytosol. Interestingly, this interaction was disrupted by AMPylation. We conclude that AMPylation of Rho Family GTPases by VopS results in diverse inhibitory consequences during infection beyond the most obvious phenotype, the collapse of the actin cytoskeleton.

Rho GTPases are proteins that are integral in diverse signaling pathways, including the maintenance of cell structure, coordination of cell motility, response to extracellular stimuli, and activation of host defenses to pathogens. Accordingly, they represent a ripe target for virulence mechanisms of pathogenic bacteria and are frequently modified with post-translational modifications, such as glucosylation, deamidation, and AMPylation (1)(2)(3). The switch-1 loop of GTPases is a frequent target of such modifications because this region of the protein mediates interactions with downstream signaling partners, such as p21-activated kinase (PAK) 3 4). The attractiveness of this family of proteins as targets for pathogens is highlighted by recent discoveries of host pattern recognition receptors (PRRs) that monitor the activation status of Rho GTPases. Nucleotidebinding oligomerization domain (NOD) family proteins were shown to be PRRs that sense-activate Rac1 and stimulate NFB signaling (5). Inversely, Rho proteins that have been inactivated through modification of the switch-1 loop can trigger the inflammasome through the novel PRR protein Pyrin via an unknown mechanism (6). The sensitivity of the host immune system to modulation of Rho GTPases may merely be a convenient way to sense infection or it may instead suggest that these modifications are exceptionally deleterious to the host. The majority of investigations about the effect of Rho GTPase posttranslational modifications, including AMPylation, have focused on the collapse of the actin cytoskeleton, but other signaling pathways in which these GTPases are involved may also be affected (7,8).
One example of Rho GTPase targeting by virulence factors is the targeting of Rho, Rac, and CDC42 by the type three secreted effector VopS from Vibrio parahaemolyticus. This marine pathogen causes gastroenteritis as a result of the consumption of contaminated undercooked seafood and septicemia in cases of wound infection or immune compromised patients (1). VopS along with a few other effectors is delivered into host cells via the first of two type three secretion systems (T3SS1) encoded in the pathogen's genome. The T3SS1 causes an orchestrated death of the host cell in less than 3 h that involves effectorinduced apparent autophagy by VopQ, plasma membrane blebbing by VPA0450, cell rounding by VopS, and finally, host cell lysis caused by all secreted effectors working together in concert (9 -11). The second type three secretion system of V. parahaemolyticus (T3SS2) is responsible for gastroenteritis in infected individuals and has been implicated in invasion of infected host cells (12). Upon delivery into the host cell, VopS localizes to the plasma membrane via its bacterial phosphoinositide binding domain where it modifies Rho GTPase pro-teins by adding an adenosine monophosphate (AMP) to a threonine located in the switch-1 loop of the GTPase (13,14). This modification has been demonstrated to cause actin cytoskeletal collapse by blocking interaction of Rac1 with PAK (1). This phenotype has obvious implications for an infected host, but its drastic nature likely masked other important cellular consequences. In support of this, VopS has previously been shown to reduce cytokine production during infection through an unknown mechanism (15).
Rho family GTPases are known to have many functions in the cell beyond control of the actin cytoskeleton. For example, Rac has been shown to perform several functions in innate immunity, such as activation of the phagocytic NADPH oxidase complex, which is important for microbial killing by lymphocytes. Association of activated Rac is essential for the recruitment of the p67phox subunit to the membrane, allowing generation of the killing superoxides at the phagocytic cup (16). Rac is also a known ubiquitination substrate for the inhibitors of apoptosis proteins (IAPs), although the full implications of this modification are unclear (17). IAPs are known to ubiquitinate TRAF6 and several other proteins as a result of microbial and other stimuli to initiate downstream signaling of NFB and MAPK pathways, leading one to speculate that Rho GTPase ubiquitination might play a role in these pathways (18). The NFB and MAPK signaling cascades are critical systems that mediate cell survival outcomes for a variety of cell responses to outside stimuli, and their importance during bacterial infection is well established (19). Manipulating these pathways are a common goal of many pathogenic bacteria, and several other diverse strategies have been elucidated including but not limited to Ser/Thr acetylation, ubiquitination, phosphothreoninelyation, deamidation, and ADP-ribosylation (20).
We sought to determine whether AMPylation of the Rho GTPase switch-1 region by VopS had effects beyond the collapse of the actin skeleton. To this end we monitored the effects of VopS during infection on NFB and MAPK signaling pathways, binding of IAP proteins to Rac1, and the ability of Rac1 to activate the phagocytic oxidase complex. Each of these signaling processes plays important roles in the ability of a host to clear infection, and we found that VopS had striking inhibitory effects on all of them. We also utilized a broad proteomic screen to identify a novel Rac1-binding protein C1qA. This interaction is hampered by VopS-mediated AMPylation and thus may have implications in immunity that are yet to be explored.
Plasmid Constructs-C-terminal GST fusions of cIAP1, XIAP, C1qA, RhoB, and all printed nucleic acid programmable protein array (NAPPA) cDNAs used for cell-free expression assays were encoded in the pANT7-cGST vector from the Arizona State University DNASU program. Rac1 V12 was cloned into pET-28a. VopS N⌬30 was cloned into pGEX-TEV for protein purification, and full-length VopS was cloned into pLAFR and pBAD33 along with its 1-kb upstream genomic sequence for V. parahaemolyticus reconstitution as detailed in Bacterial Strains and Table 1.
Bacterial Strains-Escherichia coli strain Rosetta (DE3) used for protein purification was cultured with chloramphenicol in addition to antibiotics for plasmid selection. V. parahaemolyticus strains are listed in Table 1 and were derived from POR1 (RIMD2210633 ⌬tdhAS), which was generously provided by Drs. Tetsuya Iida and Takeshi Honda. Strain genotypes and characteristics are detailed in Table 1. The CAB5 strain contains a deletion for the transcriptional regulator VtrA to prevent expression of the T3SS2 components as well as deletions of the effectors VopQ, VopR, and VPA0450 to leave VopS as the only known T3SS effector expressed and secreted. VopS was further deleted in the CAB5⌬vopS strain and complemented with a pBAD-VopS expression plasmid containing VopS wild type or VopS-H348A preceded by 1 kb of its upstream sequence, which typically allows expression and secretion levels similar to genes encoded in the genome. CAB5⌬vopS was complemented in the same manner but using the pLAFR cosmid. Routine culture, deletions, and reconstitution via plasmid mating were performed as previously described (22). Bacterial Infection Experiments-For immunoblot experiments, cells were seeded onto 6-well plates at a density of 4 ϫ 10 5 cells per well the day before infection. For immunofluorescence experiments, cells were seeded on 6-well plates containing sterile coverslips at a density of 1 ϫ 10 5 cells per well. Bacteria were induced before infection by diluting overnight culture 1:10 into plain DMEM followed by incubation at 37°C with shaking for 30 min. Bacteria were then diluted to a multiplicity of infection (m.o.i.) of 10 relative to host cells in fresh prewarmed DMEM. Culture media on cells was then replaced with this infection media and were spun at 1000 rpm for 5 min before return to the incubator for the indicated infection times. All infection experiments were performed three or more times with consistent results.
Immunoblot and Immunofluorescence-For immunoblotting, cells were lysed in buffer containing 50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM EDTA, 5% glycerol, and 1% Nonidet P-40 with protease and phosphatase inhibitors (Roche Applied Science) on ice before electrophoresis and immunoblotting. PVDF membranes were blocked with 5% milk for 30 min and incubated with primary antibodies at 4°C overnight with agitation. Secondary HRP-linked antibodies (donkey anti-rabbit, GE Healthcare; goat anti-mouse, Sigma; donkey anti-goat, Santa Cruz Biotechnology) were incubated at room temperature for 30 min at 1:10,000 dilutions in 5% milk followed by thorough washing in TBST. Chemiluminescence was generated using the Pierce ECL2 HRP substrate. For immunofluorescence experiments, cells were washed once with PBS followed by fixing in 3.2% paraformaldehyde. Fixed cells were permeabilized with 0.5% Triton X-100 in PBS before immunostaining with anti-p65 (Santa Cruz Biotechnology) at a 1:200 dilution for 1 h. Staining was detected with anti-rabbit Alexa Fluor 568 secondary antibody (Invitrogen) incubation at a 1:500 dilution for 1 h. DNA was visualized by staining with Hoechst at 1 g/ml for 10 min. Confocal images were taken using a Zeiss LSM 710 microscope and converted using ImageJ.
Recombinant Protein Purification-Tag-free VopS protein was purified as previously described (14). AMPylated and unAMPylated His-Rac1 V12 were purified by coexpression in Rosetta (DE3) E. coli with a GST-VopS fusion or empty vector. Cultures were grown to approximately an A 600 of 0.6 at 37°C, and protein expression was induced by the addition of 0.4 mM isopropyl ␤-D-thiogalactopyranoside for 4 h at room temperature. His-Rac1 proteins were isolated on Ni 2ϩ -agarose (Pierce) followed by gel filtration chromatography on Superdex 75 PG with an AKTA FPLC system. Fractions containing pure Rac1 were pooled, concentrated, and frozen in 10% glycerol at Ϫ80°C. The secondary AMPylation assay was performed by incubating Rac proteins with VopS and 0.5 Ci of [␣-32 P]ATP for 30 min at 30°C before electrophoresis, transfer, and phosphorimaging. GST-PAK3 p21-binding domain (PBD) and GST-p67 were expressed as above and purified with glutathione agarose (Pierce). GST-cIAP1, XIAP, C1qA, and RhoB fusions were expressed in the human cell-free system (Thermo Fisher) before immunoprecipitations.
Pulldown Assays and Immunoprecipitations-GST pulldown assays were performed by binding purified GST proteins to glutathione beads followed by incubation with His-Rac1 V12 at 4°C with gentle agitation for 2 h. Beads were washed four times before elution in Laemmli buffer. Beads were loaded and washed with Tris-buffered saline (300 mM NaCl) plus 0.5% Triton X-100 and 5 mM MgCl 2 . Immunoprecipitations were performed by loading anti-GST-bound magnetic protein A/G beads with GST fusions expressed in cell-free lysates. GST beads were then incubated with purified His-Rac1 V12 and washed as above.
Superoxide Generation Assay-COSphox cells were seeded at a density of 2 ϫ 10 5 cells per well in a 6-well plate followed by infection as described above. Cells were then collected, and superoxide generation upon stimulation with phorbol 12-myristate 13-acetate (PMA, Sigma) was measured using the Diogenes kit (National Diagnostics) according to the manufacturer's protocol. Luminescence was monitored using a FLUOstar OPTIMA (BMG LabTech) plate reader over 30 min. Superoxide assays were performed three times in triplicate, and a representative experiment is shown.
NAPPA Interaction Screen-The fabrication of NAPPAs comprising 10,000 highly purified DNA plasmids with fulllength human ORF sequences and the procedure of proteinprotein interaction screens were reported as previously described (23,24). Briefly, to accomplish the expression and display of proteins on NAPPA, the array was blocked with the Superblock solution (Pierce) for 1 h at 23°C and then incubated with the HeLa lysates based cell-free expression system for 1.5 h at 30°C and 0.5 h at 15°C. After a brief washing with PBST (PBS, 0.2%Tween) the resulting protein array was blocked with PPI COLD blocking buffer (1 ϫ PBS, 1% Tween 20, and 1% BSA, pH 7.4) for 1 h at 4°C. In parallel, 100 ng/l DNA of Rac1-Halo proteins was added in 170 l of a human HeLa lysate-based cell-free expression system, and expression was performed at 30°C for 2 h. To execute protein-protein interaction screening, the expressed Rac1-Halo proteins was incubated with a NAPPA for 16 h at 4°C. After reaction, the array was washed 3 times with PPI washing buffer (PBS, 5 mM MgCl 2 , 0.5% Tween 20, 1% BSA, and 0.5% DTT, pH 7.4). The binding of Rac1-Halo to its interactors on NAPPA was detected using 12.5 M Alexa 660 conjugated Halo ligand (Promega, Madison, WI). Finally, the microarray fluorescent images were scanned using Tecan's PowerScanner (Männedorf, Switzerland), and the signal intensity was quantitated using Array-Pro Analyzer (Media Cybernetics, Bethesda, MD).
Selection of Rac1 Interactors from NAPPA Screening-The selection of interaction targets was executed as previously described (24). Briefly, all NAPPA images were visually examined to eliminate the false positive signals caused by the spot shape, dust, and nonspecific bindings. Signal normalization was performed using the raw signal intensity of each spot divided by the median background-adjusted value of all features on the array. Finally, the Z-score was calculated for each protein and the selection of Rac1 interactor candidates using the following criteria: 1) a Z-score of Ն2.5, 2) a Z-score ratio of query protein to Halo negative control of Ն2, 3) targets that meet previous criteria in two independent experiments.
wNAPPA Assays-The details of wNAPPA were executed as previously described with minor modifications (25). Briefly, the anti-GST antibody-coated 96-well plate (GE Healthcare) was blocked with 5% milk at 4°C overnight. In parallel, the RhoGTPase proteins containing a C-terminal Halo tag and their interaction proteins containing a C-terminal GST tag were expressed in 30 l of a human HeLa lysate-based cellfree expression system using 40 ng/l DNA with or without 5 g of purified VopS added to the lysate. After expression, they were diluted with 100 l of PPI blocking buffer and added into the 96-well plate. The protein-protein interaction reaction was then executed at 15°C for 2 h on the shaker (Eppendorf, Hamburg, Germany) at 1000 rpm. After washing 3 times with PBST, the detection was accomplished by incubation with a 1:3000 diluted anti-Halo tag antibody (Promega), 1:000 diluted HRP-labeled goat anti-rabbit IgG antibody (Jackson ImmunoResearch, West Grove, PA), and TMB colorimetric HRP substrates (Thermo Scientific, Waltham, MA).

VopS AMPylation Inhibits NFB, Erk, and JNK Signaling-
The mitogen-activated protein kinase and NFB signaling pathways are central in pathogen recognition and response by the immune system and are known to be activated downstream of multiple receptors during bacterial infection (26). NFB is an important modulator of cell survival during microbial infection. When activated, Rel proteins in this system propagate anti-apoptotic and proinflammatory signals by translocating to the nucleus and activating the transcription of relevant factors, such as immune regulating cytokines (26). Translocation of Rel proteins, such as p65, are inhibited when such stimulus is absent by the binding of the inhibitory protein IB␣, which sequesters them in the cytosol. When the NFB pathway is stimulated, IB␣ is phosphorylated by IB kinase, targeting it for ubiquitination and degradation. A common method of measuring NFB activation under various stimuli is to monitor the phosphorylation and degradation of IB␣, which allows p65 and other Rel proteins to translocate to the nucleus.
To assess the effect of VopS on host signaling, we infected cells with V. parahaemolyticus strains rather than transfecting cells with VopS. Transfection of cells with VopS causes cell rounding within 8 h, precluding our ability to follow the kinetics of VopS-induced inhibition of a signaling pathway. For analysis of NFB signaling, we infected HEK293T cells with the V. parahaemolyticus strain Cab5, which is an attenuated clinical isolate that is deleted for its TDH toxins, the second T3SS (T3SS2), and all known T3SS effectors except for VopS, thereby allowing T3SS1 and VopS to be studied independent of the other known V. parahaemolyticus virulence factors. During infection with Cab5 we found that IB␣ phosphorylation and degradation of IB␣ was markedly reduced relative to a Cab5 strain that is deleted for VopS (Cab5⌬vopS). Unlike CAB5, infection with CAB5⌬vopS resulted in strong IB␣ phosphoryl-ation concomitant with decreased IB␣ protein levels within 1 h of infection (Fig. 1). IB␣ phosphorylation and degradation during infection with CAB5⌬vopS complemented with wild type VopS (CAB5⌬vopS/pVopS) was similar to the parental CAB5 strain.
To corroborate this observation, we performed anti-p65/ RelA immunofluorescence of HEK293T cells infected with the CAB5 strain. After 2 h of infection with this strain, only 18% of cells showed strong p65 nuclear localization ( Fig. 2A). In contrast, Ͼ95% of cells infected with CAB5⌬vopS, which contains a further deletion for VopS, showed nuclear p65 localization. Complementation of this phenotype was observed in cells infected with the CAB5⌬vopS strain reconstituted for wild type VopS, CAB5⌬vopS/pVopS, with ϳ48% p65 translocation. Therefore, we observed p65 localization during infection in the absence of VopS and inhibition of this nuclear localization in the presence of VopS (Fig. 2B).
The mitogen-activated protein kinases Erk, p38, and JNK are commonly activated during infection and along with NFB share many upstream activators that can be activated by Rho GTPases (27). Activation of these proteins is characterized by phosphorylation of their activation loops on their signature TXY motifs. With profiles similar to IB␣ activation, we found that during infection with V. parahaemolyticus CAB5 and CAB5⌬vopS/pVopS, JNK and Erk activation was inhibited (Fig.  1). p38 activation did not appear to be strongly affected by VopS, and the CAB5⌬vopS strain induced activation of all mitogen-activated protein kinases tested. This observation indicates that NFB, Erk, p38, and JNK are activated during V. parahemeolyticus infection and that VopS disrupts an upstream activator of shared by NFB, Erk, and JNK but not p38.
AMPylation of Rho GTPases Blocks Their Interaction with Inhibitor of Apoptosis Proteins-IAP proteins are well studied factors in their inhibition of apoptosis and activation of NFB, but the breadth of their involvement in these processes is not well understood. Their E3 ligase activity is important for this function, as ubiquitination of target substrate proteins is required to create the necessary docking sites for interactions between proteins in the NFB pathway, such as TRAF (TNF receptor-associated factor) proteins (28). Interestingly, IAPs were recently found to also be E3 ligases for Rac1 (17). It is not known if the ubiquitination of Rac proteins by IAPs might play a role in infection and NFB activation or if it merely targets them for proteasomal degradation. To test if AMPylation of Rac1 might disrupt this interaction, we generated AMPylated His-Rac1-V12, a dominant active mutant, by coexpression with VopS in E. coli. Rac-V12 previously coexpressed with VopS is insensitive to further AMPylation in a secondary AMPylation assay, indicating that nearly all of the protein is modified (Fig.  3A). AMPylated Rac1 also showed no interaction with PAK as demonstrated previously (Fig. 3B; Ref. 1). GST-fused cIAP and XIAP were then incubated with recombinant purified His- Rac1-V12. Unmodified Rac1 bound readily to both cIAP and XIAP, but AMPylated His-Rac1-V12 did not bind either protein (Fig. 4A), demonstrating that AMPylation may disrupt this direct interaction.
To determine if IAPs might bind other Rho GTPases, we prepared a NAPPA interaction assay in a 96-well microplate format (wNAPPA) (25,29). After incubation of the ELISA plate with the Halo-tagged Rac1, CDC42, or RhoA with or without the addition of recombinant VopS, we confirmed that IAP proteins were able to bind unAMPylated but not AMPylated Rac1, CDC42, and RhoA (Fig. 4B). CDC42 and RhoA have not previously been shown to be binding partners of cIAP.
Next we analyzed whether disruption of the interaction between IAPs and GTPases might have functional consequences during an infection. We infected HEK293T cells with CAB5 and found that although the expression of cIAP1 was enhanced during infection, we observed only a small reduction in the levels of Rac1 and Rho. By contrast, the increase in cIAP was coincident with a marked decrease in total levels of Rac1 and RhoA protein when infected with CAB5⌬vopS (Fig. 4C) and by 4-h Rac and Rho were nearly undetectable. Infection with CAB5⌬vopS/pVopS prevented this degradation. Based on our previous observations that the cIAPs can no longer interact with AMPylated Rac1 and Rho, we propose this modification is preventing targeted degradation of these RhoGTPases by cIAP1.
The Phagocytic NADPH Oxidase Complex Is Inhibited by AMPylation-Microbial killing by lymphocytes is largely a function of phagocytosis followed by generation of reactive oxygen species to kill the engulfed microbes (30). The phagocytic NADPH oxidase (NOX2) complex, which is composed of at least five proteins, generates these superoxides at the site of engulfment, which activates proteolytic enzymes that mediate killing and immune signaling (31). The complex is composed of gp91, p22, p47, p67, and Rac1 or Rac2. Rac1/2 is essential for the function of the complex as its activation, localization to the membrane, and binding to p67 activates the complex (32). Accordingly, we hypothesized that AMPylation may inhibit superoxide generation by this complex, as the interaction between p67 and Rac1 could be disrupted. To test this we utilized the COSphox cell line, which is stably transfected with all required components of the NOX2 system mentioned above (33). Superoxide generation was monitored by luminescence for 30 min using the Diogenes kit, and this cell-based assay was validated by ensuring that generation of oxygen radicals upon stimulation with the NOX2 activator PMA could be inhibited with superoxide dismutase (Fig. 5A).
COSphox cells were infected with V. parahaemolyticus strains for 2 h, collected, and stimulated with PMA for 5 min before measurement. We found that cells infected with CAB5 reduced superoxide generation to levels similar to cells that were uninfected and uninduced (Fig. 5B). By contrast, cells infected with CAB5⌬vopS strain generated robust superoxide signal upon PMA stimulation. Similarly, infection with CAB5⌬vops/pVopS-H348A strain containing a plasmid encoding the catalytically inactive VopS mutant H348A did not inhibit superoxide generation, whereas CAB5⌬vopS/pVopS strain reduced superoxide generation to levels similar to cells that were unstimulated (Fig. 5B). These results indicate that targeting of Rho GTPases by pathogens may allow them to avoid killing by lymphocytes. Interestingly, AMPylation of Rac1 appears to modestly but consistently reduce binding to the NOX2 subunit p67 in a GST pulldown assay (Fig. 5C). This may indicate that AMPylation prevents the association of Rac proteins with the NOX2 complex.
Switch-1 AMPylation Blocks Novel Rac Interactions-Although the above findings support the notion that AMPylation of Rho GTPases perturb cell signaling beyond a simple loss of actin cytoskeleton control, we further hypothesized that there may be additional unknown roles of Rho GTPases that might be FIGURE 3. Purification of AMPylated and unmodified His-Rac1 V12. A, purification of AMPylated Rac1. His-Rac1 V12 proteins were expressed in Rosetta E. coli cells with empty vector (His-Rac1 V12) or GST-VopS (His-Rac1 AMP V12) captured on Nickel affinity resin (Pierce) and applied to a Superdex 75 column. Rac1-containing fractions were pooled, and the purity of final product is shown. His-Rac1 AMP V12 is richly AMPylated as seen with T-AMPylation antibody. A secondary VopS AMPylation assay of the proteins with [␣-32 P]ATP shows that His-Rac1 AMP V12 cannot be AMPylated further, indicating Ͼ95% AMPylation from coexpression. B, AMPylated Rac1 does not bind PAK3. 25 g of His-Rac1 V12 proteins were incubated with glutathione beads loaded with GST-PAK3 Rho binding domain or GST for 2 h at 4C. Beads were washed, and co-eluting Rac1 proteins were observed by immunoblot with Rac antibody. His-Rac1 V12 but not His-Rac1 AMP V12 bound PAK3 efficiently.
disrupted by switch-1 modification. To screen for potentially novel interactions, we utilized a NAPPA system to probe for Rac1 interacting proteins. NAPPA arrays are a robust and rapid method of producing thousands of proteins in situ for use in personalized diagnostic, post-translational modification, or protein interaction screens, among other functions (23,34).
Compared with protein microarrays fabricated with purified proteins, NAPPA instead prints plasmid cDNA that can be transcribed and translated using a mammalian cell-free expression system on an amino group-coated slide. The tagged fulllength proteins expressed are then captured and displayed in situ by an anti-tag antibody that was printed together with the . AMPylation blocks Rac1 binding to IAP proteins and inhibits Rac1 degradation. A, AMPylation inhibits Rac1 binding to IAP proteins. IAP proteins expressed in human cell free lysates were incubated with recombinant purified His-Rac1-V12 or His-Rac1 AMP -V12. IAP proteins were then immunoprecipitated with GST antibody and immunoblotted for interacting Rac1 protein with His antibody. B, unAMPylated but not AMPylated CDC42, RhoA, and Rac1 are binding partners of IAPs. IAP-GST proteins presented in wNAPPA format were incubated with Halo-Rac1, CDC42, or RhoA with and without the presence of 5 g of VopS. Binding was measured by fluorescein signals after incubation with fluorescein-conjugated Halo ligand. All GTPases were able to bind cIAP, but the addition of VopS abrogated binding. NC, negative control (empty vector). C, VopS inhibits degradation of Rac1 and RhoA during infection. HEK293T cells were infected with indicated strains at an m.o.i. of 10 for the indicated times, and cells were lysed and immunoblotted. Rac and RhoA protein levels were considerably higher at late time points in infection in the presence of VopS. IP, immunoprecipitation. Asterisk, nonspecific band.
plasmid DNA (Fig. 6A). NAPPA has become a rapid method of producing thousands of proteins for use in the discovery of disease-related antibody biomarkers and the identification of novel protein-protein interactions (24,35,36).
To probe for novel Rac1 interactions, we expressed a Halo-Rac1 fusion using a human cell-free expression system. Lysates containing GTP-loaded Halo-Rac1 or Halo alone were incubated overnight with NAPPA slides presenting Ͼ10,000 unique proteins followed by washing and labeling with Halo ligand fused to Alexa 660 (Fig. 6, A and B). As expected, known interactors such as PAK6, cIAP1/BIRC2, and XIAP/BIRC4 gave positive signals (Fig. 6C). Additionally, a few potentially novel interactors were observed, including another Rho GTPase, RhoB, and the complement subunit C1qA (Fig. 5C). C1qA is the first component of the serum complement system, a secreted ligand binding factor that bridges the innate and adaptive immune systems (37).
We sought to test if AMPylation of the switch-1 loop might disrupt any of these novel interactions and used a wNAPPA assay to test for disrupted interactions as in Fig. 3B. Interestingly, the novel interactors RhoB and C1qA were able to bind Rac1 in the absence but not the presence of VopS (Fig. 7A). This observation was further confirmed in an immunoprecipitation experiment in which recombinant AMPylated Rac1 was unable to bind either GST-fused RhoB or C1qA (Fig. 7B). These results support the conjecture that Rac1 mediates additional unknown function(s) in innate immunity by binding to C1qA and Rho proteins, and these interactions are inhibited by modification of the switch-1 loop on Rho GTPases.

DISCUSSION
Bacterial pathogens and their multicellular hosts are involved in a molecular arms race that has stretched through the millennia. Host strategies to recognize and clear infections have appeared in many evolutionary conserved systems, including PRRs, adaptive immunity, and inflammatory responses. Pathogens are forced to evolve counterstrategies that can weaken or thwart these host responses, and they have found many critical nodes in signaling pathways to subvert for their own survival (26).
One of these nodes is the Rho family of GTPases. Although the susceptibility of Rho GTPases in relation to their control of the actin cytoskeletal has been well documented, it is possible that the other functions of these proteins in host defense have been overlooked (Fig. 8). Here, we sought to explore other potential effects of Rho GTPase targeting by a bacterial pathogen. Our findings suggest that targeting the Rho GTPases may provide some of the same NFB and MAPK inhibition advantages as effectors that target the pathways directly (Figs. 1 and 2).
We found that AMPylation by VopS inhibits activation of the NFB pathway, measured by the phosphorylation state of the NFB inhibitor IB␣ and the nuclear translocation of p65. Activation of the mitogen-activated protein kinases JNK and Erk were also inhibited, but little change was observed in p38 phosphorylation. Loss of NFB, Erk, and JNK activation by VopS may be explained by the previous observation that AMPylation of Rho GTPases inhibits interaction with PAK ( Fig. 3C; Ref. 1), which along with Rhotekin is known to signal through these FIGURE 5. Superoxide generation by the NOX2 complex is inhibited by VopS during infection. A, reactive oxygen species assay validation. COSphox cells stably transfected with gp91, p22, p47, and p67 were stimulated in triplicate by the addition of 2 M PMA (blue). Cells were collected and induced with PMA for 5 min in 96-well plates, and production of superoxide was measured over 30 min using a plate reader. The addition of 50 units of superoxide dismutase (SOD; red) completely inhibited the assay, demonstrating that the chemiluminescence is superoxide-dependent. Data points are the mean, and error bars are S.D. RLU, relative light units. B, VopS inhibits reactive oxygen species produced by NOX2. Before stimulation with PMA as above, COSphox cells were infected in triplicate with the indicated strains at a m.o.i. of 10 for 2 h. Complementation with wild type VopS reconstituted the inhibition, but complementation with catalytically inactive VopS H348A did not. A representative experiment of three trials is shown. C, AMPylation reduces binding of p67 to Rac1. 25 g of His-Rac1 V12 proteins were incubated with glutathione beads loaded with 50 g of GST-p67 or GST for 2 h at 4C. Beads were washed, and co-eluting Rac1 proteins were observed by immunoblot with Rac antibody. GST-p67 showed a modest but consistent reduction in binding His-Rac1 AMP -V12 in comparison to His-Rac1-V12.
pathways (27). However, the cross-talk in these immune-sensing pathways is complex and not completely understood. For example, Rac1 is known to be involved in pathogen sensing by NOD1 and is also a substrate of the NFB mediator cIAP1. It is possible that AMPylation of Rho GTPases inhibits these pathways at multiple levels.
Other factors in the NFB pathway that may be inhibited during V. parahaemolyticus infection are the IAP proteins. Ubiquitination by IAP proteins is known to be important for NFB activation, and Rac1 is a known substrate of these proteins (17). Cellular Rac and RhoA levels were observed to decrease during the late stages of V. parahaemolyticus infection with a strain lacking VopS; however, the presence of VopS appeared to inhibit their degradation. In agreement, recombinant Rac1 AMPylated with VopS was unable to bind to this E3 ligase, which could explain the persistence of Rac1 levels in the presence of VopS (Fig. 4). However, the importance of Rho GTPase ubiquitination by IAP proteins for NFB activation has not been determined. The combination of NFB inhibition and loss of IAP binding as a result of Rho GTPase AMPylation observed in this work indicates that IAP ubiquitination of Rho GTPases could be important for immune signaling and merits further study. Such ubiquitination could be a form of immune signaling, or it is also possible that the ability of the cell to turnover Rho GTPases is important. The accumulation of AMPylated and sterically blocked Rho GTPases in the cell is potentially deleterious. Interestingly, the region of Rho GTPases involved in binding IAP proteins is not known, but these results suggest that the switch-1 loop plays some role in the interaction, or AMPylation causes a conformational change that does not allow binding of the E3 ligases.
Perhaps the most direct and obviously beneficial effect of AMPylation by VopS is the inability of the phagocytic NADPH oxidase to be activated (Fig. 5). This complex is essential for fighting bacterial pathogens, as patients deficient in any of the complex subunits suffer from chronic granulomatous disease with recurrent and often deadly infections (38). Disabling this host weaponry by targeting the Rac GTPases offers bacteria a distinct survival advantage. As other bacterial effectors modify the switch-1 region of RhoGTPase, this may be an overlooked common mechanism used to disrupt the host cell activation of reactive oxygen species.
Our NAPPA interaction screen has also revealed another potential role of Rac1 in innate immunity, which is the binding of C1qA (Fig. 6). As part of the large, secreted complement complex, C1qA makes little sense as a signaling partner of Rac1, but recent research has revealed that C1qA also plays a key cytosolic role in innate immunity. C1qA was shown to enhance the retinoic acid-inducible gene-I-like receptor (RLR) pathway through the binding of several cytosolic factors, leading to reduced viral replication (39). However, little else is known about the cytosolic role of C1qA, particularly in regard to bacterial infection. It is possible that binding to Rac1 is important for an undiscovered function of C1qA in the innate immune system. Considering that at least two PRRs for altered Rho GTPases exist in the cell, it is possible that binding of C1qA to Rac1 is important for one of these processes. We observed that AMPylation disrupts this interaction, which may serve to inhibit or promote the activation of downstream signaling (Fig.  7). Dimerization of Rho GTPase proteins have been previously observed, but the full implication of these and potential Rac1 and Rho heterodimers identified in this screen are not fully understood. This NAPPA interaction system can also be used in the future to screen for Rac1 and other Rho GTPase interactions against a larger pool of proteins to identify novel binding partners and possible activities.
The sum of the findings presented here indicates that the targeting of Rho family GTPases provides complex and multifaceted advantages for microbial pathogens, highlighting the status of Rho GTPases as an Achilles' heel in the host immunity ( Fig. 8). At least six different signaling functions mediated by Rho GTPases are sabotaged: collapse of the actin cytoskeleton, inactivation of NFB, Erk, and JNK pathways, lack of degradation of Rho GTPases, and lack of superoxides produced by the NOX2 complex. The critical nature of this modification is further supported by recent findings that cytosolic PRRs are dedicated to sensing both deactivated and over-activated versions of Rho GTPases (5, 6). As other bacteria use virulence factors to inactivate Rho GTPases by modifying the switch-1 loop with a variety of post-translational modifications, the conserved FIGURE 7. AMPylation inhibits interaction of Rac1 with C1qA and RhoB. A, VopS inhibits binding of C1qA and RhoB to Rac1. The indicated prey plasmids were printed in a 96-well ELISA plate coated with anti-GST antibody before cell free expression. Plates were then incubated with Halo-Rac1 with or without the addition of 5 g of VopS, washed, and conjugated with fluorescein-Halo ligand before analysis. The addition of VopS inhibited the interaction of each of the prey with Halo-Rac1. NC, negative control (empty vector). B, AMPylated Rac1 does not bind C1qA or RhoB. C1qA and RhoB proteins expressed in human cell-free lysates were incubated with recombinant purified His-Rac1 V12 or His-Rac1 AMP V12. GSTfused proteins were then immunoprecipitated with GST antibody and immunoblotted for interacting Rac1 protein with His antibody. IP, immunoprecipitation.  (1). AMPylation inhibits GTPase binding to PAK (2), which compromises cell control of the actin cytoskeleton (3). Binding of the E3 ligase to Rac1 is also inhibited, reducing ubiquitin mediated degradation (4). Phosphorylation of JNK, Erk, and IB␣ are reduced by VopS (5 and 6) as is degradation of IB␣ and nuclear translocation of p65. Production of reactive oxygen species by the Nox2 complex is abrogated, likely promoting bacterial survival from lymphocytes (7). Although the implications of the C1qA-Rac1 interaction are unknown, AMPylation blocked this interaction and could inhibit downstream signaling (8). mechanisms disrupted by VopS will likely be applicable for many other bacterial infection models.