The Campylobacter jejuni CiaD effector co-opts the host cell protein IQGAP1 to promote cell entry

Campylobacter jejuni is a foodborne pathogen that binds to and invades the epithelial cells lining the human intestinal tract. Maximal invasion of host cells by C. jejuni requires cell binding as well as delivery of the Cia proteins (Campylobacter invasion antigens) to the host cell cytosol via the flagellum. Here, we show that CiaD binds to the host cell protein IQGAP1 (a Ras GTPase-activating-like protein), thus displacing RacGAP1 from the IQGAP1 complex. This, in turn, leads to the unconstrained activity of the small GTPase Rac1, which is known to have roles in actin reorganization and internalization of C. jejuni. Our results represent the identification of a host cell protein targeted by a flagellar secreted effector protein and demonstrate that C. jejuni-stimulated Rac signaling is dependent on IQGAP1.

A common strategy for disease-causing microbes is to manipulate the behavior and function of host cells within infected tissues 1 . Bacterial cells hijack a broad array of host cell functions, including blocking immune signaling, promoting fluid secretion, and co-opting cytoskeletal elements to promote their spread and dissemination [2][3][4] . Importantly, bacterial factors directly target and modulate host cell signaling pathways to redirect cellular processes. Many pathogens have evolved specialized secretion systems to deliver proteins into the cytosol of host cells and rewire the cellular signaling networks in a way that benefits the pathogen 5 .
Campylobacter jejuni is among the most common bacterial causes of foodborne illness and diarrhea worldwide due, in part, to its ability to colonize food animals (particularly chickens) as a commensal organism of the intestinal tract 6 . Individuals with campylobacteriosis often develop a fever, headache, abdominal pain, nausea, and diarrhea that is frequently bloody 7 . C. jejuni infection is associated with the development of Guillain-Barré syndrome (GBS) and Miller Fisher syndrome, autoimmune disorders affecting the peripheral nervous system 8 . C. jejuni is also implicated in the development of irritable bowel disease 9 . C. jejuni invasion of intestinal epithelial cells has been observed in natural human cases, nonhuman primate infection studies, and mouse models of infection [10][11][12] . Furthermore, the presence of blood in the stool of infected individuals is indicative of intestinal cell invasion 13 . These observations support the proposal that human disease is dependent on successful intestinal colonization and subsequent invasion of the epithelial cells lining the intestinal tract.
The results from the past 30 years of work on C. jejuni pathogenesis have repeatedly reminded researchers that C. jejuni is a unique pathogen. Despite causing similar disease symptoms to other Gram-negative enteric pathogens, the genome of C. jejuni is relatively small at only~1.6 Mbp (~31% GC) compared to that of Salmonella enterica, Yersinia enterocolitica, and Shigella flexneri, which range between 4.6 and 4.8 Mbp (47-52% GC). Moreover, the process of effector protein delivery and the mechanistic basis underpinning the cellular invasion of C. jejuni differs from other intestinal pathogens. There is no evidence that C. jejuni has a secretion system dedicated to the export of virulence proteins, like the specialized type III secretion system (T3SS) injectisomes of the pathogens listed above. However, C. jejuni possesses a bifunctional flagellar export apparatus that secretes structural components of the flagella in addition to virulence proteins. Dedicated T3SS injectisomes and flagellar secretion systems share analogous proteins, and there is evidence that proteins are capable of being secreted from both the dedicated T3SS injectisome and the flagellar system [14][15][16] . To date, it is known that C. jejuni synthesizes at least eight non-flagellar proteins that are secreted from the bacterium in a manner that is dependent on the flagellar T3SS [17][18][19] . A subset of these proteins, known as the Campylobacter invasion antigens (Cia), exert their actions within the cytosol of host cells to manipulate signaling processes 20,21 . Three of these Cia proteins (CiaC, CiaD, and CiaI) have been demonstrated to be delivered to the cytosol of host cells, using a process that is dependent on direct bacterial contact and the presence of a functional flagellar export apparatus 22,23 .
C. jejuni possesses a few well-characterized virulence factors that facilitate cellular invasion. For example, the C. jejuni CadF and FlpA Fibronectin-binding proteins (FNBPs) facilitate C. jejuni attachment to cells, effector protein delivery, and, in part, initiation of the cell signaling that leads to bacterial invasion 24,25 . The C. jejuni adhesins promote the formation of a three-component bridge comprised of the FNBPs, fibronectin, and the host cell fibronectin-binding integrin receptors. Signals from the integrins are transmitted through a multiprotein complex known as the focal adhesion, a structure that links the extracellular environment to the actin cytoskeleton, which is central to cellular invasion by C. jejuni 26,27 . The signaling events initiated by the C. jejuni adhesins are further enhanced by the C. jejuni effector proteins CiaC and CiaD to promote maximal C. jejuni uptake by host cells 22 . Of these proteins, CiaD (also known as Cj0788c in C. jejuni strain NCTC 11168) is most well understood. CiaD has been shown to contribute to cellular invasion in vitro in C. jejuni strain F38011 by targeted mutagenesis 28 and in strain 81-176 by screening a transposon library 29 . Perhaps most importantly, CiaD was demonstrated to contribute to C. jejuni disease in a mouse model by investigators blinded to the identities of the infectious isolates 23 . Compared to infection with the C. jejuni wild-type strain, mice infected with the ciaD deletion mutant have reduced GI tract wall thickening and reduced inflammation of the ileocecocolic lymph node 23 . More recently, the application of a pig ligated loop model also revealed that a C. jejuni ΔciaD mutant does not induce the characteristic robust inflammatory response observed with a C. jejuni wild-type strain, as demonstrated by the production of inflammatory cytokines and neutrophil markers 30 . Laboratory studies have shown that CiaD is mechanistically involved in host cell uptake of C. jejuni by promoting actin reorganization and membrane ruffling via Erk 1/2 activation 28 . Activated Erk 1/2 phosphorylates (activates) cortactin, which promotes membrane ruffles by binding to the actin remodeling proteins N-WASP and WAVE2 31 . These ruffles are evident in scanning electron micrographs of the infected cells and promote C. jejuni uptake.
Campylobacter researchers continue to debate the mechanistic basis for C. jejuni invasion of epithelial cells, largely because no direct cellular target has been identified for an effector protein.
The goal of this study was to identify a host cell target of the C. jejuni CiaD effector protein. Previous findings demonstrated that cortactin is activated in a CiaD-dependent manner and that cortactin acts in concert with N-WASP to activate the Arp2/3 complex involved in the initiation of actin polymerization and the reorganization of actin filaments 28 . Therefore, we hypothesized that CiaD interacts with a host cell protein central to actin rearrangement. Additionally, it is known that actin reorganization mediators, including the small GTPases Rac1 and Cdc42, play important roles in the internalization of C. jejuni 32 .
In this work, we show that CiaD binds to the host cell protein IQGAP1 to displace RacGAP1 using affinity pull-down experiments coupled with LC-MS/MS and immunoblots, yeast-two hybrid (Y2H), and C. jejuni-host cell infection assays. We demonstrate that CiaD promotes C. jejuni cell invasion by modifying the IQGAP1 complex to promote signaling through the Rac1 pathway. This represents an important identification of a direct host cell interaction partner for a flagellar effector protein from a bacterial pathogen. This finding demonstrates that C. jejuni invasion of cells is a pathogen driven process that is dependent on cell binding and effector delivery (a binding and effector mechanism).

Results
Identification of eukaryotic targets of CiaD. It is known that the delivery of CiaD from C. jejuni to the cytosol of host cells is necessary for cellular invasion 23 . The purpose of this study was to determine the host cell target protein of CiaD, which is 163 amino acids long with a calculated molecular mass of 19.4 kDa. There are no apparent conserved domains in CiaD that are detected by the NCBI Conserved Domain Search 33 , the InterPro protein sequence analysis and classification system 34 , or when searching Pfam 35 . Although there are no conserved domains, previous studies have found that CiaD contains a Mitogen-activated protein kinase docking (MKD) motif that promotes cellular invasion 23,36 . Due to the presence of the MKD motif, and the phenotypes of the ΔciaD mutant, we focused our efforts on identifying CiaD-interacting proteins that were related to the MAPK pathway.
We initially used purified CiaD to pull-down proteins from lysates prepared from cultured human INT 407 epithelial cells. The pull-down experiments were performed with a glutathione-Stransferase (GST)-tagged CiaD protein, and the resulting protein complexes were subjected to LC-MS/MS for identification using a spectral counting strategy with the search engine MS-GF+ . The spectral counts were then normalized by the total number of spectra obtained for each assay. The cellular proteins from the CiaD pull-down (n = 2) were compared to the cellular proteins pulled down in the absence of recombinant CiaD (n = 1) to identify those proteins that bound in a CiaD-dependent manner to the GST affinity column. By comparing the normalized spectral counts for the proteins present in the pull-down experiments to the proteins non-specifically bound to the affinity column, we found that CiaD and 17 human proteins were enriched at least 2-fold in the pull-downs ( Fig. 1A and Supplementary Data 1). While these proteins had different levels of enrichment, it was not clear whether this was a result of the number of copies in the protein complex, affinity of the interaction, or expression level in the original sample. Therefore, we considered all 17 candidate proteins with equal likelihood.
Functional enrichment analysis of the 17 human proteins was performed using DAVID 37 (Fig. 1B). Among the enriched InterPro domains were "Spectrin/alpha-actinin," "Calponin homology domain," and "Actinin-type/actin-binding/conserved site." The latter domain binds to F-actin and was first described in α-actinin, although it is also found in many other actin-binding proteins 38 . The "Spectrin/alpha-actinin" domain forms a three-helix bundle that is commonly found in proteins involved with cytoskeletal structure 39 . The "Calponin homology domains" are found in proteins that link cellular signaling processes to the actin cytoskeleton 40 . The enrichment of actin-binding domains is consistent with the hypothesis that CiaD modifies the activity of the actin cytoskeleton, as the major phenotypes of a ΔciaD mutant are the lack of actin-driven membrane ruffles resulting in reduced C. jejuni cellular invasion 28 . Among the cytoskeleton related proteins identified in the pull-down were IQGAP1, which is a scaffolding protein that coordinates actin reorganization, Filamin A and B, which connect cell membrane constituents to the actin cytoskeleton, and Fascin, which organizes filamentous actin into parallel bundles.
A commercial yeast-two hybrid assay (Y2H) was used to screen for direct protein-protein interactions between CiaD and host cell targets (Supplementary Data 2). The CiaD protein was fused to the activation domain of LexA and used as bait to screen a human placenta cDNA library because this library represents a wide variety of transcripts 41,42 . The screen resulted in three high confidence hits (prey proteins): dynein light chain roadblock-type 1 (DYNLRB1), which is an accessory protein of the dynein 1 complex that participates in TGFβ signaling 43 ; nuclear receptorbinding protein (NRBP1), which has a role in trafficking between the endoplasmic reticulum and Golgi apparatus; and the Ras GTPase-activating-like protein IQGAP1, which is a scaffolding protein that regulates the actin cytoskeleton. The protein IQGAP1 was identified in both the commercial Y2H and LC-MS/MS screens.
Amino acid alignments between the three potential CiaD targets showed no significant sequence similarity (alignment in Supplementary Fig. 1). Therefore, we examined the region targeted by CiaD within the DYNLRB1, NRBP1, and IQGAP1 proteins. The full-length DYNLRB1 protein was identified in the screen, so binding may occur anywhere in the protein. However, DYNLRB1, at 96 aa, does not contain any specific conserved protein domains. Similarly, NRBP1, for which the CiaD target region was localized between amino acids 457 to 523, does not contain any known protein domains. For IQGAP1, the interacting fragment was localized to the Ras GTPase-activating domain (GRD domain) located between amino acids 964 and 1365. This domain is known to play a role in modulating the activity of Ras GTPases, including Rac1 and Cdc42, which are the critical mediators of actin reorganization.
Because IQGAP1 was the only protein identified in the screening experiments (LC-MS/MS and Y2H) that has a role in the Erk 1/2 signaling pathway and actin reorganization, it was considered a top candidate. The GRD domain of IQGAP1, which is the region that is predicted to interact with CiaD, regulates the activities of the Rac1 and Cdc42 GTPases involved in actin  To confirm the CiaD-IQGAP1 interaction using a third method, a targeted pull-down assay was performed coupled with immunoblot analysis. Pull-down experiments were performed by mixing purified recombinant 6His tagged CiaD with cell lysates from INT 407 cells and then pulling down the resulting protein complexes with TALON affinity resin, which binds to the tagged CiaD. Samples were tested for the presence of IQGAP1 using an immunoblot. In the samples that had the CiaD protein bound to the TALON affinity resin, IQGAP1 was pulled down; IQGAP1 was not identified in the pull-down experiments where CiaD was not added to the reaction mixture (negative control) ( Fig. 2A). In this targeted assay, we observed that CiaD bound to IQGAP1, which is in agreement with the data from the non-targeted LC-MS/MS and Y2H screens. Furthermore, we found that IQGAP1 is not pulled down with other C. jejuni proteins, including CiaC and FlpA, indicating a specific interaction with CiaD ( Supplementary Fig. 2).
To determine if CiaD was specifically binding to the domain identified in the Y2H, the GRD domain of IQGAP1 was expressed and purified from E. coli. Using an ELISA, it was found that CiaD was able to bind to the GRD domain of IQGAP1, and that this saturable interaction was stronger than other C. jejuni proteins, including CiaC and FlpA (Fig. 2C). This finding supports the conclusion that CiaD directly interacts with IQGAP1 and that this interaction occurs at the GRD domain.
Based on the findings from the pull-down experiments coupled with LC-MS/MS, Y2H, and directed pull-down experiments, CiaD binds to IQGAP1. However, the context of cellular infection is different from protein:protein interactions in vitro. The purpose of the infection experiments was to test whether the CiaD protein was binding to IQGAP1 during cellular infection. In addition, if we observed binding of the C. jejuni CiaD protein to the IQGAP1 protein, this would indicate that the CiaD protein was successfully translocated across the host cell plasma membrane to interact with a cytosolic protein. Therefore, we infected cells with a C. jejuni ΔciaD mutant harboring a plasmid with either the CiaD-FLAG or CiaDΔMKD-FLAG insert driven by a constitutive promoter. After C. jejuni infection, the cells were lysed, and immunoprecipitations were performed with an anti-FLAG affinity gel. We found that when the cells were infected with C. jejuni expressing CiaD-FLAG or CiaDΔMKD-FLAG, IQGAP1 was immunoprecipitated using the anti-FLAG affinity gel (Fig. 2B). This finding indicates that during the course of a C. jejuni-host cell infection, the CiaD protein is delivered to the cytosol of the host cells and interacts with IQGAP1. In addition, the data indicates that the interaction of CiaD with IQGAP1 is independent of its MKD domain. The MKD domain of CiaD is necessary to promote Erk 1/2 activation and is involved in cellular invasion by C. jejuni 23 . Importantly, this finding shows that the native C. jejuni protein interacts with IQGAP1 during the course of normal cellular infection.
Role of IQGAP1 for cellular invasion. It is well established that CiaD facilitates invasion into host cells, as the deletion of ciaD reduces cellular invasion in multiple strains of C. jejuni 23,29 . Because we have found that CiaD interacts with IQGAP1, we predicted that IQGAP1 is required for maximal C. jejuni cell invasion. Using lentiviral delivered shRNA, we generated clonal INT 407 cell lines that synthesized reduced levels of IQGAP1. Two independent shRNAs were used, which had different targets on the IQGAP1 mRNA. The shRNA labeled 928 targets the coding sequence of IQGAP1, while the shRNA labeled 930 targets the 3′-UTR. Both of the shRNA treatments resulted in nondetectable levels of IQGAP1 in the two clonal lines tested (Supplementary Fig. 3). We then measured the ability of C. jejuni to adhere to the INT 407 cells and the ability of C. jejuni to invade the cells using the gentamicin protection assay. In all four clonal cell lines tested, there was no reduction in the adherence of C. jejuni compared to cells expressing a non-targeting shRNA. However, there was a statistically significant reduction in invasion when compared to a cell line that was expressing a non-targeting shRNA (p < 0.05 as determined by ANOVA followed by Sidak's test). The average reduction in cell invasion was approximately 50% (or 0.5 log). There was no significant difference in the invasion between the different clones that were targeting two different portions of the mRNA. Given the similar results observed with each of the two independent INT 407 IQGAP1 knockdown cell lines (each generated with the two different IQGAP1 shRNA constructs), we concluded that the reduction in cellular invasion by C. jejuni was specific to the reduction of IQGAP1 levels.
To confirm that the knockdown of IQGAP1 was specifically responsible for the reduction in cellular invasion, the expression of IQGAP1 was restored in a knockdown cell line (named 930-C5). Because the IQGAP1 expression vector does not contain the IQGAP1 3′ UTR that is present on the native mRNA, no modification of the complementation vector was needed to avoid the effects of the shRNA. Cellular invasion was increased in the knockdown cell line expressing IQGAP1-Myc but was not changed in the cells expressing the vector that contained EGFP in place of IQGAP1, as judged by the gentamicin protection assay (Fig. 3). While expression of IQGAP1-Myc in the knockdown cells significantly increased invasion compared to the knockdown cells alone, it was not completely restored to the level of cells expressing the non-targeting shRNA. The partial restoration of invasion is consistent with the fact that transient transfection is not completely efficient and that the level of IQGAP1 expression was not the same as in wild-type cells. This finding is supported by immunoblot, where the amount of IQGAP1 protein was lower than in the wild-type cells (Fig. 3C). Finally, the invasion of C. jejuni into the cell line producing the non-targeting shRNA was not affected when IQGAP1 was overexpressed by transfection with the pCMV-IQGAP1-Myc vector (Fig. 3D). This indicates that a biologically relevant amount of IQGAP1 is necessary for C. jejuni cellular invasion and that increased levels of IQGAP1 within a cell do not enhance cell invasion. Most importantly, these findings demonstrate that IQGAP1 has a functional role in C. jejuni invasion. While the gentamicin protection assay is the most widely used method to assess the efficiency of bacterial internalization, it only provides global information about the number of bacteria internalized by a population of cells and does not address bacterial internalization into a single cell. To test if the varying levels of IQGAP1 expression were responsible for the partial complementation of cellular invasion by C. jejuni, we used confocal microscopy to assess bacterial internalization in the 930-C5 cell line that was transfected with the IQGAP1-Myc expressing plasmid. The C. jejuni infected cells were immunostained for actin, C. jejuni, and with an anti-Myc antibody to detect IQGAP1 expression ( Fig. 3D-G). Confocal examination of the cells revealed that the number of C. jejuni internalized was significantly greater in cells expressing IQGAP1 than in cells lacking IQGAP1. Consistent with the gentamicin protection assay, the number of internalized C. jejuni was reduced by approximately 50% in cells lacking IQGAP1. Importantly, the total number of cell-associated C. jejuni (adherent and internalized) was similar for the IQGAP1positive and negative cells (no significant difference, p = 0.415 by Student's t-test). Consistent with previous observations, the C. jejuni ΔciaD mutant did not invade cells regardless of the presence of IQGAP1 (Fig. 3F, G). Taken together, these findings demonstrate that IQGAP1 is necessary for C. jejuni to maximally invade cells at high efficiency.
Signaling events associated with cellular invasion. One of the most striking phenotypes of a C. jejuni ΔciaD mutant is the lack of activation (phosphorylation) of the MAPK signaling cascade, as evidenced by the lack of Erk 1/2 phosphorylation 23 . Given that IQGAP1 acts as a scaffold for Erk 1/2, we tested if IQGAP1 was facilitating the CiaD mediated activation of ERK signaling. Cells expressing non-targeting shRNA and IQGAP1 knockdown cells were serum-starved for 4 h and infected with C. jejuni. As reported previously, Erk 1/2 is activated by infection with a C. jejuni wild-type strain when compared to non-infected cells. However, the amount of Erk 1/2 activation by C. jejuni was not reduced when IQGAP1 was knocked down (Fig. 4A, B). This is consistent with the finding that the MKD domain of CiaD is not involved in IQGAP1 binding and suggests that the CiaD protein activates ERK signaling upstream or independently of IQGAP1.
As ERK signaling was not impaired in the IQGAP1 knockdown cells, we wanted to determine the functional reason for the plasmid. These constructs express the CiaD proteins from a constitutive promoter. The CiaDΔMKD-FLAG construct is missing the Mitogen-activated protein kinase docking motif (MKD) within CiaD. IPs were probed with an antibody for IQGAP1. It was found that CiaD-FLAG was delivered by C. jejuni and associates with IQGAP1. The inputs for the FLAG IPs were also probed for IQGAP1 and FLAG to determine initial protein abundance. Arrows indicate expected molecular weight for each blot. Molecular weights are indicated in kDa. C An ELISA was performed to determine the binding between the recombinant GRD domain from IQGAP1 (IQGAP1-GRD), and recombinant C. jejuni proteins CiaD, CiaC and FlpA (n = 2 independent samples). Binding between the C. jejuni proteins was assessed against increasing concentrations of IQGAP1-GRD. CiaD showed the greatest binding potential. Values are shown as normalized absorbance that has been scaled from zero (minimum assay value) to one (maximum assay value). Error bars represent the standard error of the mean. actin reorganization, so we tested the amount of Rac1 and Cdc42 activation in wild-type and IQGAP1 knockdown cells. Rac1 and Cdc42 G-LISA kits were used to determine the amount of active (GTP bound) Rac1 and Cdc42 present in the cells after infection. We found that there was nearly no activation of either Rac1 and Cdc42 in the IQGAP1 knockdown cells, while there was robust activation of both Rac1 and Cdc42 in the cells expressing the nontargeting shRNA during infection (Fig. 4C, D). This finding indicates that IQGAP1 is facilitating actin reorganization by signaling through both Rac1 and Cdc42 during the process of C. jejuni cellular invasion. This also supports the conclusion that IQGAP1 mediates the Rho GTPase signaling that participates in C. jejuni invasion.
To test if the signaling through Rac1 and Cdc42 is sufficient to facilitate invasion, we expressed constitutive-active forms of Rac1 and Cdc42 in the IQGAP1 knockdown cells (Fig. 4E). In the IQGAP1 knockdown cells, we found that expressing the constitutive-active (CA) Cdc42 construct did not increase invasion above wild-type (WT) Cdc42, while the dominantnegative (DN) reduced invasion. In contrast, we found that the does not alter the cellular adherence of C. jejuni to any of the cell types tested (n = 3 independent samples). B Knockdown of IQGAP1 reduces C. jejuni internalization while expressing IQGAP1-Myc improves invasion. In panels A and B, error bars represent the standard deviation; differences were compared using an ANOVA on log-transformed counts followed by a two-tailed Sidak's test (*p < 0.05) (n = 3 independent samples). Comparing NT to 930-C5, p = 0.0007. Comparing 930-C5 to 930-C5 + IQGAP1, p = 0.0472. C The amount of IQGAP1 in the cell lines was determined by immunoblot. Actin was used as a loading control for the cell lysates. Molecular weights are indicated in kDa. D To test if the mosaic nature of transient transfection was responsible for the incomplete restoration of invasion in the IQGAP1 knockdown cells that were transfected with IQGAP1, we used microscopy to assess the internalization of C. jejuni into cells. IQGAP1 knockdown cells were transfected with pCMV-IQGAP1-Myc, infected with C. jejuni for 10 h, then fixed and stained for C. jejuni (yellow), actin (cyan), and with an anti-Myc antibody (magenta). A z-section was constructed from the section in the white box (right). From the zsection, it is apparent that the C. jejuni associated with the cells expressing IQGAP1 are internalized. E Three biological replicates were imaged, and more than 15 cells were analyzed for each condition (IQGAP1 expressing and non-expressing). We found that the number of internalized C. jejuni was higher in the cells expressing IQGAP1. White bars represent 5 µm. Comparing internalized bacteria in IQGAP1 negative to IQGAP1-positive cells, p = 0.0061. F Cells were infected with a C. jejuni ΔciaD mutant and imaged. From the z-section, it is clear that the C. jejuni ΔciaD mutant does not invade cells, regardless of whether IQGAP1 is present. White bars represent 5 µm. G From three biological replicates, greater than 15 cells were analyzed, and the number of internal and external C. jejuni were determined. No internalized bacteria were observed in cells infected with the C. jejuni ΔciaD mutant. In panels E and G, each dot represents a biological replicate, the column represents the mean, error bars represent standard deviation, and significant differences were assessed using a two-tailed Student's t-test (*p < 0.05).
Rac1-CA increased invasion above Rac1-WT, and the Rac1-DN did not reduce invasion. Relevant to this finding is that cell viability and transfection efficiency varied by less than 7% among all of the samples (Supplementary Fig. 4). This suggests that while both Rac1 and Cdc42 activation are impaired in the IQGAP1 knockdown cells, Rac1 is primarily responsible for facilitating C. jejuni invasion through IQGAP1 signaling.
Although we have demonstrated that CiaD binds to IQGAP1, this observation, in itself, does not explain why this proteinprotein interaction is necessary for C. jejuni invasion. To explore the mechanistic basis for CiaD function (promotion of actin-based invasion by C. jejuni), we performed IQGAP1 immunoprecipitation experiments to determine the composition of the IQGAP1 complex. We chose to determine the abundance of three 'actin regulator proteins' based on the finding that Rac1-CA enhanced the invasion of the C. jejuni in IQGAP1 knockdown cells versus cells transfected with Rac1-WT construct and because the Cdc42-CA construct did not rescue cell invasion. Specifically, we probed for three of the components of the IQGAP1 complex (Rac1, Erk 1/2, and RacGAP1). Rac1 and Erk 1/2 are mediators of C. jejuni dependent actin reorganization, and RacGAP1 is a mediator of Rac1 activity that associates with IQGAP1 to constrain Rac1 activity 45 . Immunoprecipitations were performed on lysates from noninfected cells, cells infected with the C. jejuni wild-type strain, and cells infected with the ΔciaD mutant. We did not observe a difference in the abundance of Rac1 and Erk 1/2 in the IQGAP1 complex (Fig. 5A). However, we found that the association of RacGAP1 with IQGAP1 was reduced approximately 50% upon infection with a C. jejuni wild-type strain but not upon infection with a ΔciaD mutant (p < 0.05). This result was consistent with both the C. jejuni F38011 and C. jejuni 81-176 strains, demonstrating that at least two pathogenic strains that cause bloody diarrhea in infected individuals utilize the same mechanism for cell invasion (Fig. 5A, B). In summary, this finding indicates that the presence of CiaD excludes RacGAP1 from the IQGAP1 complex to potentially increase the local Rac1 activity.

Discussion
Most of the knowledge about host cell components involved in C. jejuni internalization comes from knockdown or inhibitor studies 23,27,28,32,[46][47][48][49][50] . These studies have revealed that C. jejuni targets focal adhesions, a large multiprotein complex in host cells that links the extracellular matrix to the actin cytoskeleton. The first step in C. jejuni-induced cell invasion is adherence to the cell-associated fibronectin via the CadF and FlpA FNBPs 47,51 , which initiate outside-in signaling through the focal adhesions as evidenced by paxillin phosphorylation 27 . Direct bacteria-host cell contact further permits the translocation of the flagellardependent Cia effector proteins, including CiaD, into the host cell cytosol. Cytosolic CiaD facilitates the activation of the Erk 1/2 signaling pathway and cellular invasion in a mechanism dependent on its MKD motif 23 . The presence of CiaD results in the phosphorylation and activation of cortactin, an actin remodeling protein 28 . Using both non-targeted assays (affinity pulldown experiments coupled with LC-MS/MS and Y2H), as well as targeted pull-down experiments coupled with immunoblots, we identified IQGAP1 as a binding partner of CiaD. We further found that IQGAP1 is required for the maximal internalization of C. jejuni through its interaction with Rac1 and Cdc42. Importantly, we also found that CiaD keeps Rac1 active by displacing RacGAP1 from the IQGAP1 complex and that CiaD has an IQGAP1 independent mechanism to activate the Erk 1/2 signaling pathway. These findings point to a mechanism whereby IQGAP1 is downstream in the signaling cascade that promotes C. jejuni uptake, with additional upstream or parallel pathways also acting to facilitate invasion. We show a model of C. jejuni internalization, where CiaD binds to IQGAP1, increasing Rac1 activation by displacing RacGAP1, and promoting C. jejuni internalization in Fig. 6. IQGAP1 plays a central role in integrating the signals from focal adhesions to actin through Rac1 and Cdc42. While previous studies have sought to define the host proteins necessary for C. jejuni internalization, these results now demonstrate that a C. jejuni effector protein binds to a host cell component and alters its inherent activity. In addition, we now implicate both IQGAP1 and RacGAP1 in C. jejuni internalization. Taken together, the finding that CiaD displaces RacGAP1 from IQGAP1 to promote Rac1 activity supports the proposal that C. jejuni internalization is a bacterial driven process, where the bacterium utilizes a binding and effector mechanism for cell invasion.
The terms 'type III secretion system' (T3SS) and 'injectisome' have been used to describe related sets of contact-dependent secretions systems 52 . The best-understood T3SSs are present in human pathogens that have the most tools for genetic manipulation, including E. coli, Yersinia, Pseudomonas, and Salmonella 53-56 . Among bacteria, more than 504 experimentally verified type III secreted effector proteins have been identified that facilitate a variety of host phenotypes (http://effectors.bic. nus.edu.sg). Identification of these effector proteins and their apparent phenotypes is a major milestone; however, simply identifying the effectors has not led to new advances in disease treatment. Indeed, contact-dependent effectors that are not secreted into the extracellular space make poor targets for intervention. While determining the phenotypes attributable to specific effector proteins is useful, it becomes necessary to identify the host target proteins to gain a mechanistic understanding of effector protein function 57 .  5 The CiaD effector protein binds to IQGAP1 to exclude RacGAP1. A IQGAP1 immunoprecipitations (IPs) were performed with INT 407 cells that were non-infected, infected with the C. jejuni F38011 wild-type strain, or infected with the C. jejuni ΔciaD mutant. Pull-downs were probed by immunoblot (IB) with antibodies for IQGAP1 and RacGAP1. The amount of RacGAP1 in the IQGAP1 complex was reduced during infection with the C. jejuni wild-type strain but not with the ΔciaD mutant. The amount of Erk 1/2 and Rac1 was not changed. The protein mixture used as the input for the IQGAP1 IPs were also probed for IQGAP1, RacGAP1, Erk 1/2, and Rac1 to determine initial protein abundance. Molecular weights are indicated in kDa. B Similar IQGAP1 IPs were performed with INT 407 cells that were non-infected, and infected with the C. jejuni 81-176 wild-type strain. Quantitation of the amount of RacGAP1 in three biologically independent IP experiments indicated that the amount of RacGAP1 in the IQGAP1 complex was reduced approximately 50% during infection with the C C. jejuni F38011 wild-type strain and the (*p = 0.0186) Molecular weights are indicated in kDa. D C. jejuni 81-176 wild-type strain. Bars represent the mean and error bars represent the standard deviation; significant differences from the non-infected samples were determined by ANOVA followed by a two-tailed Dunnet's multiple comparisons test (*p = 0.0401).
Because the genome of C. jejuni does not have any apparent pathogenicity island or dedicated virulence-associated secretion system, it has been thought that C. jejuni was incapable of secreting and delivering effector proteins. While it is often overlooked, the flagellar apparatus is, in fact, a T3SS that functions in much the same way as a dedicated virulence-associated T3SS to secrete proteins. Studies with the Salmonella effector proteins SptP and SopE provide support for this observation; removing the chaperone binding domains from SptP and SpoE causes the effector proteins to be secreted through the flagellar export pathway rather than the dedicated injectisome 58 . The secretion of virulence proteins by C. jejuni was first described in 1999 and was found to be dependent on the flagellar secretion apparatus in 2001 21,59 . These flagellar-dependent secreted virulence proteins were termed the Campylobacter invasion antigens (Cia), and three of these proteins (CiaC, CiaD, and CiaI) have been demonstrated to be delivered to the cytosol of host cells 22,23 . Since the initial report of the Cia proteins, several research groups have identified additional virulence proteins secreted by the flagellar export apparatus 17,18,20 . Importantly, all of these secreted proteins mediate the interaction between C. jejuni and host cells.
Invasion of intestinal epithelial cells is a phenotype of several enteric pathogens that harbor effector proteins to promote cellular uptake. This strategy is an effective means to maintain residence in the intestine and avoid the innate immune response, at least temporarily 60 . Noteworthy is that the Salmonella effector protein SseI, which only has 11.6% identity to CiaD at the deduced amino acid level, binds to IQGAP1 61-63 . Moreover, Kim et al. (2011) reported that the ability of Salmonella typhimurium to invade cells and activate both Rac1 and Cdc42 was significantly attenuated in IQGAP1-null mouse embryonic fibroblasts (MEFs) 63 . However, in contrast to our study, there was a lack of Erk 1/2 activation in IQGAP1 knockout cells upon infection with Salmonella 63 . These findings demonstrate that while both Salmonella and C. jejuni target IQGAP1 to promote bacterial uptake, they modulate the IQGAP1-mediated signaling differently. Furthermore, our studies have uncovered the mechanistic basis for the modulation of the IQGAP1-related Rac1 signaling by C. jejuni, which is by displacing RacGAP1, a suppressor of Rac1 activation.
RacGAP1 modulates the activity of Rac1 during cellular migration 45,64 , which requires integrin recycling. It is thought that during integrin activation, IQGAP1 is recruited to the sites of activated integrins, activating Rac1, which further progresses into the recruitment of RacGAP1 to prevent unconstrained Rac1 activity 45,64 . Our findings support a model whereby CiaD binds to IQGAP1 and prevents, or excludes, RacGAP1 from associating with the IQGAP1 complex. This blockage of a negative regulator of Rac1 activity (a GTPase-activating protein) then promotes localized Rac1 activity. We propose that this event leads to exaggerated actin reorganization and uptake of the associated C. jejuni. Furthermore, normal integrin recycling would likely be disrupted, leaving activated integrins on the surface of cells. The data from our study indicate that the CiaD effector protein alters the composition and function of the IQGAP1 complex by excluding RacGAP1. A surprising finding in our study is that C. jejuni activates the MAPK signaling pathway in the absence of IQGAP1. It is well established in the literature that IQGAP1 is capable of scaffolding the upstream signaling molecules MEK 1/2 as well as Erk 1/2 65 . However, because Erk 1/2 activation was independent of IQGAP1, CiaD must facilitate Erk 1/2 activation through a mechanism independent of IQGAP1. Raza and colleagues recently reported that the protein DYNLRB1, which was a high confidence hit in our Y2H screen, modulates the Erk 1/2 signaling pathway 66 . Thus, it is possible that CiaD binds to DYNLRB1 and modulates Erk 1 /2 signaling to promote C. jejuni uptake. Additional research may also uncover the possibility that one or more of the other proteins identified in the pull-down experiment can modulate the MAPK signaling pathway. Furthermore, while we did not directly identify a component of the MAPK signaling pathway in the pull-down or Y2H experiments performed with CiaD, it is possible that CiaD binds only transiently to a component of the MAPK pathway. Due to experimental limitations, it is likely that our experiments did not identify every binding partner of CiaD. Moreover, the diversity of proteins in the intestinal epithelium may not have been captured by the epithelial cell lysates and cDNA libraries used in this study. However, it is also clear that targeted assays have demonstrated that CiaD binds to IQGAP1, providing strong evidence for additional biological roles and binding partners for CiaD and other C. jejuni effector proteins.
We have demonstrated that C. jejuni possesses a flagellardependent effector that directly interacts with a host cell protein to modulate the actin cytoskeleton through the Rac1 signaling pathway. These findings are important because they demonstrate the possibility that a motile bacterium may manipulate the behavior of host cells through flagellar-secreted effectors. There is remarkable progress being made using information from highly studied bacteria that have dedicated T3SSs. However, there is nearly no information about secreted effectors in bacteria lacking dedicated secretion systems. Moreover, secreted bacterial effector proteins are still nearly impossible to identify through sequence analysis alone. In this study, we have identified a host cell target of a C. jejuni flagellar-secreted protein. The identification of IQGAP1 is particularly interesting because it is a well-known regulator of signaling events involved in cytoskeletal rearrangement. Moreover, this finding, in combination with the results from inhibitor studies and knockdown experiments that demonstrate the necessity of focal adhesion components, provides a framework for molecular dissection of C. jejuni invasion. Our findings also raise the possibility that numerous pathogenic and commensal bacteria alike may have secreted effector proteins hidden in their genomes that are modifying the behavior of the host cells.

Methods
Bacterial culture conditions. The C. jejuni F38011 67 and 81-176 68 strains were cultured on Mueller-Hinton agar plates supplemented with 5% citrated bovine blood (MHB) at 37°C in a microaerobic environment (5% O 2 , 10% CO 2 , 85% N 2 ). The C. jejuni F38011 ciaD deletion mutant and complemented isolate was generated as described previously 23 . The C. jejuni 81-176 ΔciaD strain was created using the same plasmid described previously 23 , with the tetracycline resistance cassette swapped for a chloramphenicol resistance cassette at the SacII sites. Where necessary, tetracycline (128 µg/mL) or chloramphenicol (8 µg/mL) was added to the growth media. All Escherichia coli strains were grown on LB agar plates at 37°C supplemented with carbenicillin (100 µg/mL) as necessary.
CiaD pull-down assays and sample preparation. CiaD was cloned into the EcoRI and XhoI sites of pGEX-5X-1 with the primers 5′-ATA ATG AAT TCA TGA ATT TGG AAG ATT TAG CTA AAA AAA C-3′ and 5′-ATA ATC TCG AGA AGC TTA TCT TCG ATA TTT GCA AGC-3′ (Also indicated in Supplementary Table 1). The CiaD-GST fusion was expressed in E. coli BL21(DE3) and induced with 1 mM of Isopropyl-β-D-thiogalactoside (IPTG) for 16 h at 4°C. Hostinteracting proteins were collected by loading CiaD onto a glutathione-Stransferase (GST) column and adding an INT 407 cell lysate. Proteins were eluted using reduced glutathione. Negative controls were performed by loading GST alone, expressed from pGEX-5X-1. A Pierce TM BCA Protein Assay Kit (Thermo Scientific, Waltham, MA) was used to determine the concentration of the proteins in the pull-down eluate. The eluates were denatured for 30 min at 58°C in the following solution: 500 mM of ammonium bicarbonate, 8 M urea, and 500 mM of DTT. The sulfur groups of cysteine were then derivatized using 400 mM Iodoacetamide for 30 min at 37°C. Trypsin was added to the denatured proteins at a sample protein to trypsin ratio of 50:1. The proteins were digested for 4 h at 37°C and were desalted on a reverse-phase Supelco C18 column and stored at −80°C before analysis.
MS analysis was performed using a Velos Orbitrap mass spectrometer (Thermo Scientific, San Jose, CA) outfitted with a custom electrospray ionization (ESI) interface. Electrospray emitters were custom made by chemically etching 150 µm OD × 20 µm ID fused silica 70 . The heated capillary temperature and spray voltage were 350°C and 2.2 kV, respectively. Data were acquired for 100 min after a 15 min delay from the start of the gradient. Orbitrap spectra (AGC 1 × 106) were collected from 400-2000 m/z at a resolution of 60k followed by data-dependent ion trap MS/ MS (collision energy 35%, AGC 1 × 104) of the ten most abundant ions. A dynamic exclusion time of 60 s was used.
Proteomics data analysis. The 'Thermo.Raw' files were analyzed with MS-GF+ 71 against a database comprising the Uniprot non-redundant SwissProt human proteome (20,216 sequences, downloaded in April 2015) and the sequence of Campylobacter CiaD. MS-GF+ parameters were the following: semi-tryptic peptides were authorized, methionine oxidation was set as variable modification, cysteine carbamidomethylation was set as fixed modification. A decoy database was used to generate false-positive identifications. Results were filtered to have a false discovery rate below 1%. Using a MS-GF+ score below 4.77E-10 resulted in a False Discovery Rate below 1% at the spectral, peptide and protein level (proteins were only considered identified when at least 2 unique peptides were attributed to them). A spectral counting strategy was used to estimate abundance all the unique peptides attributed to a given protein were used for quantification (e.g., unmodified, modified, tryptic and semi-tryptic peptides). Only proteins with at least 10 spectral counts in any of the three pull-down experiments were considered for quantification.
Yeast two-hybrid screening. Yeast two-hybrid screening was performed by Hybrigenics Services, S.A.S., Evry, France (http://www.hybrigenics-services.com). C. jejuni ciaD gene was PCR-amplified and cloned into the plasmid pB27 as a Cterminal fusion to LexA (LexA-CiaD). The construct was checked by sequencing the entire insert and used as a bait to screen a random-primed placenta cDNA library constructed into the plasmid pP6. pB27 and pP6 derive from the original pBTM116 72 and pGADGH 73 plasmids, respectively. A total of 104 million clones (10-fold the complexity of the library) were screened using a mating approach with YHGX13 (Y187 ade2-101::loxP-kanMX-loxP, matα) and L40ΔGal4 (mata) yeast strains as described previously 74 . 285 His+ colonies were selected on a medium lacking tryptophan, leucine, and histidine, and supplemented with 2 mM 3aminotriazole to handle bait autoactivation. The prey fragments of the positive clones were amplified by PCR and sequenced at their 5′ and 3′ junctions. The resulting sequences were used to identify the corresponding interacting proteins in the GenBank database (NCBI) using a fully automated procedure. A confidence score (Predicted Biological Score) was attributed to each interaction, as previously described 75 .
The Predicted Biological Score relies on two different levels of analysis. Firstly, a local score takes into account the redundancy and independency of prey fragments, as well as the distribution of reading frames and stop codons in overlapping fragments. Secondly, a global score takes into account the interactions found in all the screens performed at Hybrigenics using the same library. This global score represents the probability of an interaction being nonspecific. For practical use, the scores were divided into four categories, from A (highest confidence) to D (lowest confidence). A fifth category (E) specifically flags interactions involving highly connected prey domains previously found several times in screens performed on libraries derived from the same organism. Finally, several of these highly connected domains have been confirmed as false-positives of the technique and are now tagged as F. The Predicted Biological Score scores have been shown to positively correlate with the biological significance of interactions 76,77 .
Purification of CiaD, CiaC and FlpA recombinant proteins for pull-down assays. C. jejuni CiaD, CiaC and FlpA recombinant proteins were purified by using the following expression vectors; pMCGS68 for His6-CiaD and pET-24b(+) for His6-CiaC and His6-FlpA. Briefly, the E. coli BL21 (DE3) cells containing either plasmid pMCGS68-His6-CiaD, pET24b-His6-CiaC, or pET24b-His6-FlpA were grown in LB broth for overnight at 37°C supplemented by either ampicillin (for pMCGS68-His6-CiaD) or kanamycin (for pET-24b-His6-CiaC and pET-24b-His6-FlpA). The overnight cultures were inoculated in two 250 mL LB broth at a ratio of 1:50 and incubated at either 25°C (for pMCGS115-His6-CiaD) or at 37°C (for pET-24b-His6-CiaC and pET-24b-His6-FlpA) with appropriate antibiotic selection. After 4 h of incubation or when the culture was in the exponential growth phase (OD 600 = 0.35-0.5), 1 mM IPTG was added into the culture for induction and incubated for another 4 h at the appropriate condition. After the induction, cells were pelleted by centrifugation and mixed with IP buffer (25 mM Tris, 150 mM NaCl, 1% Triton X-100, 5% Glycerol, 1 mM PMSF) and lysed by sonication with appropriate settings. The cell lysates were cleared by centrifugation followed by filtration through a 0.45 µm syringe filter. The clear cell lysates were passed through the 2 mL TALON metal affinity resins (Takara Bio) in a 10 mL chromatography column (Bio-Rad). The eluted products were further cleared and concentrated by using the Amicon Ultra 10,000 MW filter (EMD Millipore). A BCA assay was done to determine the protein concentration. The purity of the protein products were determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).
Pull-downs with purified CiaD, CiaC and FlpA. Pull-down assays were performed on a 10 mL chromatography column (Bio-Rad). Briefly, 100 µL of TALON metal affinity resins were added to the column followed by washing the beads with 10 mL wash buffer (25 mM Tris, 150 mM NaCl). An equal amount of purified recombinant His6-CiaD, His6-CiaC, and His6-FlpA proteins or bovine serum albumin (BSA, negative control) were suspended in 5 mL IP wash buffer (25 mM Tris, 150 mM NaCl, 1% Triton X-100, 5% Glycerol) and added to a column. The flowthrough of the proteins were loaded into the columns two more times for optimal binding of His-tagged proteins to the TALON resin. Then, equal amount of INT 407 cell lysates at a ratio of 1: 4 of the purified protein was loaded onto the column. The flow-through of the lysates was loaded four more times for optimal binding of lysates to either CiaD, CiaC, FlpA, or BSA. Finally, the columns were washed with 10 mL IP wash buffer (25 mM Tris, 300 mM NaCl, 0.1% Triton X-100, 5% Glycerol) followed by elution with 5 mL elution buffer (150 mM imidazole, 25 mM Tris, 300 mM NaCl, 0.1% Triton X-100, 5% Glycerol). The eluted proteins were concentrated by using the Amicon Ultra-filter by centrifugation at 3000 × g for 40 min. Samples were suspended in 2X sample buffer and boiled prior to immunoblot analysis. The immunoblots were probed with mouse anti-IQGAP1 antibody (sc-376021, Santa Cruz Biotechnology) at a dilution of 1:1000. An anti-mouse IgG (A4416, Sigma) was used for the secondary antibody at a dilution of 1:4000 and the ECL Prime Detection Reagent (Amersham) was used to visualize the immunoblot.
Purification of IQGAP1-GRD. The gene sequence encoding IQGAP1-GRD was amplified from the pCMV-IQGAP1-Myc plasmid (see below) using the forward primer: 5′-GGA TCC GAA TTC GAG CTC CGA GAA GAG AGA GAA GTT GGA AGC TTA CC-3′ and the reverse primer: 5′-GTG CTC GAG CTT GTC ATC GTC ATC CTT GTA GTC GTC GAC CAG GGT GAG AGA CAC TTC CGT-3′, resulting in a FLAG-tagged gene fragment. This fragment was cloned into the pET24b vector for expression in E. coli. Protein was produced by adding 1 mM IPTG to log-phase E. coli and incubating at 37°C for 4 h. E. coli was lysed in IP Wash Buffer, and protein was collected by running the lysate over a TALON affinity Column. Proteins were eluted in elution buffer for further use.
IQPAG1-GRD interaction ELISA. Polystyrene plates were coated with 3 µg per well of either His6-CiaD, His6-CiaC, His6-FlpA, or BSA (negative control) overnight in 0.1 M carbonate buffer (pH 9.5) at 4°C. Plates were washed in PBS + 0.1% Tween buffer (above), followed by blocking with 1% BSA (w/v) in PBS. IQGAP1-GRD was added to the wells in 2-fold serial dilutions from 25 µg/mL to 0.390625 µg/mL and incubated for 2 h at room temperature. Wells were then washed in PBS + 0.1% Tween buffer before adding anti-FLAG antibody (Cat #F3165, Sigma-Aldrich) at a concentration of 1:2000, followed by a two-hour incubation at room temperature. Wells were washed, and secondary HRP-conjugated anti-mouse antibody was added (Cat # A4416, Sigma-Aldrich) at a concentration of 1:4000. After a two-hour incubation, wells were washed with PBS + 0.1% Tween and developed with TMB substrate according to the manufacturer's instructions (Cat # 34021, Thermo Fisher). Wells were read in a BioTek EL X 808 plate reader at 450 nm.
INT 407 lysate preparation. INT 407 cells were harvested with trypsin, pelleted by centrifugation, and washed with 10 mL of PBS containing EDTA-free protease inhibitor cocktail (Halt™ Protease Inhibitor Cocktail, EDTA-Free, Thermo Fisher Scientific, Waltham, MA). The cell pellet was then resuspended in 1-3 mL of immunoprecipitation (IP) buffer (25 mM Tris pH 7.5, 150 mM NaCl, 5% Glycerol, 1% Triton X-100) with protease inhibitor cocktail and lysed by repeated passage through a 21-gauge needle. Whole-cell lysates (WCLs) were clarified by centrifugation at 14,000 × g for 20 min at 4°C. Total protein content was determined by a BCA assay.
Pull-downs with purified CiaD. Purified recombinant His6-CiaD protein or an equivalent volume of PBS (negative control) was added to an INT 407 lysate at a ratio of 4 times more His6-CiaD protein than INT 407 protein. The mixed sample was incubated with rotation at 4°C for 4 h. Following incubation, 50 µL of IP buffer-equilibrated TALON metal affinity resin was added and incubated for 4 h with rotation at 4°C. After incubation, the TALON resin was collected by centrifugation at 2,000 × g for 5 min, and proteins were suspended in 2X sample buffer and boiled prior to immunoblot analysis.
Creation of stable cell lines. In a 6-well tissue culture dish, an overnight culture of INT 407 cells (4 ×105 cells per well) was infected with approximately 10 6 Lentiviral transduction particles (Mission, Sigma) in MEM supplemented with 10 µg/mL polybrene (Sigma, St. Louis, MO). Cells were infected with lentiviral particles carrying shRNA that was non-targeting (Sigma Cat# SHC016V) or targeting IQGAP1 (Sigma Mission shRNA library cat # TRCN0000298928, TRCN0000298930). Throughout the paper, these are referred to as '928' and '930'. Cells were incubated for two days with the virus, and the medium was changed for MEM supplemented with 2.5 µg/mL puromycin (Sigma). After two days of selection, cells were trypsinized and seeded at~1 cell/well in a 96-well tissue culture plate in MEM with puromycin. Only wells that were verified by microscopy to contain one INT 407 cell were used for clonal expansion. The expression of IQGAP1 in the clonal cell lines was determined by SDS-PAGE and immunoblot analysis, as described below.
IQGAP1 plasmid construction. The IQGAP1 cDNA was amplified from the plasmid HsCD00347159 from the Harvard PlasmID database using the primers 5′-GCC ATG GAG GCC CGA ATT CGG ATG TCC GCC GCA GAC GAG GTT -3′ and 5′-GAT CCC CGC GGC CGC GGT ACC TTA CTT CCC GTA GAA CTT TTT GTT GAG AA -3′. In-Fusion cloning was used to insert IQGAP1 into the pCMV-Myc vector that was linearized using EcoRI and KpnI. The pCMV-Myc-EGFP plasmid was generated by insertion of the EGFP gene into the SalI and NotI restriction sites of the pCMV-Myc vector.
Gentamicin-protection assays. The gentamicin protection assay was used to assess cellular adherence and internalization in knockdown cell lines, as described previously 78 . Briefly, 1.5 × 10 5 of each cell line was seeded in eight wells of a 24-well tissue culture tray overnight. Cells were rinsed with MEM containing 1% FBS and infected with~1 × 10 7 CFU of C. jejuni (OD 540 = 0.03). The adherence assays were performed by incubation of C. jejuni with the epithelial cells for 3 h. The cells were then rinsed three times with PBS, lysed with 0.1% Triton X-100, and samples serially diluted and spread onto MHB plates. The invasion assays were performed by incubating C. jejuni with the cells for 3 h, washing the cells three times with PBS, and adding MEM with 250 µg/mL of gentamicin for an additional 3 h to kill extracellular bacteria. Following incubation, the cells were rinsed three times with PBS, lysed with 0.1% Triton X-100, and samples were serially diluted and spread onto MHB plates. The numbers of adherent (cell-associated) and internalized bacteria were determined by counting the resulting colonies. To test if the expression of IQGAP1 rescued invasion in the knockdown cell lines, cells were seeded at 0.7 × 10 5 cells/well in a 24-well tissue culture tray and transfected with pCMV-IQGAP1-Myc or pCMV-EGFP-Myc the following day using Lipofectamine 3000 (Thermo Fisher Scientific) following the manufacture's instructions. Similarly, the effects of Rac1 activation and Cdc42 activation were tested by transfecting the cells with plasmids expressing GFP tagged wild-type Rac1, constitutive-active Rac1 (L61), dominant-negative Rac1 (N17), wild-type Cdc42, constitutive-active Cdc42 (L61), and dominant-negative Cdc42 (N17) 79 . Approximately 18 h after transfection, the adherence and internalization assays were performed as outlined above. All assays were performed in biological triplicate, and representative assays are shown. Cell viability was determined by collecting three triplicate wells of transfected cells by trypsinization, staining with trypan blue, and counting the cells in an EVE Automated Cell Counter (NanoEnTek, Inc, Waltham, MA). incubation, the cells were transfected with pCMV-IQGAP1-Myc using Lipofectamine 3000 (Thermo Fisher Scientific). Approximately 18 h after transfection, cells were washed in MEM with 1% FBS and infected with~1 × 10 5 CFU of C. jejuni (OD 540 = 0.003) for 10 h. The cells were fixed with 4% paraformaldehyde in PBS for 5 min, permeabilized using PBS with 0.1% Triton X-100 and 0.3% BSA, and then stained with antibodies. Primary antibodies against C. jejuni (1:4000, rabbit polyclonal) 80 and c-Myc (1:500 mouse monoclonal, Takara Bio) were added for 45 min. After washing three times with PBS, AlexaFlour 680 anti-rabbit and AlexaFlour 488 anti-mouse secondary antibodies (1:1000, Jackson ImmunoResearch, West Grove, PA), along with Texas Red-X conjugated phalloidin (1:1000, Thermo Fisher Scientific), were added for 30 min. Coverslips were washed three times with PBS and mounted on slides using ProLong Diamond Antifade Mountant (Thermo Fisher Scientific). After curing overnight, slides were imaged using a Leica SP5 laser scanning confocal microscope. Images were quantitated by counting C. jejuni in ImageJ.
Erk 1/2, Cdc42, and Rac1 activation experiments. Stable human INT 407 cells expressing a non-targeting shRNA or an shRNA targeting IQGAP1 were seeded to a density of 1 × 10 5 cells per well in a 24-well tissue culture tray in MEM supplemented with 10% FBS. Prior to C. jejuni infection, the cells were serum-starved in MEM without FBS for 4 h. An overnight biphasic culture of C. jejuni was then used to infect the cells at an OD 540 of 0.3. For testing of Erk 1/2 activation, samples were collected, and immunoblots were performed as described below. For testing of Cdc42 and Rac1 activation, samples were collected and processed in accordance with the instructions provided with the Cdc42 G-LISA and Rac1 G-LISA kits (Cytoskeleton Inc., Denver, CO). The cell lysates were adjusted to a concentration of 0.4 mg/mL prior to the assays. Samples were flash-frozen in liquid nitrogen and stored at −80°C. Three biological samples were collected on different days for analysis.
Immunoprecipitating the IQGAP1 complex. INT 407 cells were seeded at a density of 5.5 × 10 6 cells in a 100 mm dish and incubated overnight. The cells were then serum-starved for three hours, followed by infection with C. jejuni at an OD 540 of 0.35 in 10 mL of MEM 1% FBS. After 45 min of infection, cells were lysed on ice in IP buffer (see above) with 50 mM NaCl that was supplemented with a protease inhibitor cocktail and phosphatase inhibitor cocktail (Halt Phosphatase Inhibitor Cocktail, Pierce, Rockford IL). Lysates were clarified by centrifugation at 13,000 × g. IQGAP1 complexes were immunoprecipitated by incubating the clarified lysates with mouse anti-IQGAP1 (Clone C-9, Santa Cruz Biotechnology, Dallas, TX) overnight at 4°C with rotation, followed by overnight incubation with Protein A/G agarose (Santa Cruz Biotechnology). Beads were collected by centrifugation at 2,000 × g and washed two times in IP buffer with 50 mM NaCl. Proteins were dissociated from the beads by boiling in 2X sample buffer for five minutes before analysis by immunoblot (as described below). Antibodies were used to detect RacGAP1 (Thermo Fisher), IQGAP1 (Abcam, Cambridge, MA), Rac1 (Cell Signaling Technology), and Erk 1/2 (Santa Cruz Biotechnology).
Immunoprecipitating IQGAP1 using C. jejuni produced CiaD. For IQGAP1 coimmunoprecipitation assays with C. jejuni expressing CiaD, a 60% confluent 100-mm dish of INT 407 cells were infected with a C. jejuni ciaD deletion mutant harboring a plasmid containing either ciaD-FLAG or ciaDΔMKD-FLAG. The construction of these recombinant strains was described previously 23 . The cells were infected in MEM with 1% FBS at an OD 540 of 0.3 for 2 h before lysis in IP buffer (see above) on ice. The entire contents of the dish were harvested by scraping in 1 mL of cold IP buffer and then centrifuged at 13,000 × g to clarify the lysate. The lysate was incubated with anti-FLAG M2 affinity gel (Sigma) at 4°C overnight. Samples were washed three times in IP buffer, placed in a boiling waterbath for 5 min and subjected to discontinuous SDS-PAGE in a 10% polyacrylamide gel. The proteins were transferred to membranes and probed with IQGAP1 and FLAG antibodies.
Immunoblotting and SDS-PAGE. For immunoblotting IQGAP1, phospho-Erk 1/2, and total-Erk 1/2, cell lysates were collected from 24-well plates by adding 300 µL of single-strength sample buffer and agitation. Samples were boiled for 5 min, and proteins were separated by SDS-PAGE with 12.5% polyacrylamide described by Laemmli 81 . Gels were either stained with Coomassie brilliant blue R250 (CBB-R250) or transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were blocked with 5% non-fat dry milk and incubated with antibodies against IQGAP1 (1:1000, Abcam) or total-Erk 1/2 (1:2000 Santa Cruz Biotechnology). Membranes probed with antibodies against phospho-Erk 1/2 (1:2000 Cell Signaling Technology) with blocked with 5% BSA. The blots were then incubated with a goat anti-rabbit antibody conjugated to horseradish peroxidase (1:2500, Sigma), and reactive bands were visualized using Amersham ECL prime reagent (GE Healthcare). Images of blots were acquired on a GE Healthcare ImageQuant LAS 4000 Mini system.
Statistics and reproducibility. All statistical analysis presented was performed with GraphPad Prism 6.0 g (La Jolla, CA). Unless indicated otherwise, all experiments presented were performed at in at least three independent replicates.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
Mass spectrometric raw data were deposited in the MassIVE (https://massive.ucsd.edu/) repository, a member of the proteomeXchange consortium, under the accession number MSV000086662. The identified peptide, protein, and quantified protein lists with expression values and associated statistics are attached to this manuscript in the Supplementary Data 1. Any plasmids, mutants, or cell lines created for this study will be made available upon request. Source data are provided with this paper.