Escherichia coli Producing CNF1 Toxin Hijacks Tollip to Trigger Rac1-Dependent Cell Invasion

Traffic 2011 © 2011 John Wiley & Sons A/S

doi:10.1111/j.1600-0854.2011.01174.x

Escherichia coli Producing CNF1 Toxin Hijacks Tollipto Trigger Rac1-Dependent Cell Invasion

Orane Visvikis1,2,3, Laurent Boyer1,2, Stephanie

Torrino1,2, Anne Doye1,2, Marc Lemonnier1,2,

Patrick Lores3, Monica Rolando1,2, Gilles

Flatau1,2, Amel Mettouchi2,4, Daniel Bouvard5,

Esteban Veiga6, Gerard Gacon3, Pascale

Cossart6 and Emmanuel Lemichez1,2,∗

1INSERM, U895, Centre Mediterraneen de MedecineMoleculaire, C3M, Toxines Microbiennes dans la relationhote pathogenes, Nice, F-06204 Cedex 3, France2Universite de Nice-Sophia-Antipolis, UFR Medecine,Nice, F-06204 Cedex 3, France3Institut Cochin, Departement de Genetique etDeveloppement, INSERM U1016, CNRS UMR8104,Universite Paris Descartes, 24 rue du Faubourg SaintJacques, 75014 Paris, France4INSERM, U634, Faculte de Medecine de Nice, 27Avenue de Valombrose, Nice, F-06107 Cedex 2, France5INSERM, U823; Institut Albert Bonniot, UniversiteJoseph Fourier; CNRS ERL 3148, UJF site Sante, BP170,F38042 Grenoble Cedex 09, France6Institut Pasteur, Unite des Interactions BacteriesCellules, INSERM, U604, INRA, USC2020, Paris,F-75015, France*Corresponding author: Emmanuel Lemichez,[email protected]

Rho GTPases, which are master regulators of both

the actin cytoskeleton and membrane trafficking, are

often hijacked by pathogens to enable their invasion

of host cells. Here we report that the cytotoxic

necrotizing factor-1 (CNF1) toxin of uropathogenic

Escherichia coli (UPEC) promotes Rac1-dependent entry

of bacteria into host cells. Our screen for proteins

involved in Rac1-dependent UPEC entry identifies the

Toll-interacting protein (Tollip) as a new interacting

protein of Rac1 and its ubiquitinated forms. We show

that knockdown of Tollip reduces CNF1-induced Rac1-

dependent UPEC entry. Tollip depletion also reduces

the Rac1-dependent entry of Listeria monocytogenes

expressing InlB invasion protein. Moreover, knockdown

of Tollip, Tom1 and clathrin, decreases CNF1 and Rac1-

dependent internalization of UPEC. Finally, we show that

Tollip, Tom1 and clathrin associate with Rac1 and localize

at the site of bacterial entry. Collectively, these findings

reveal a new link between Rac1 and Tollip, Tom1 and

clathrin membrane trafficking components hijacked by

pathogenic bacteria to allow their efficient invasion of

host cells.

Key words: bacterial invasion, bacterial toxin, clathrin,

CNF1, Rac1, Rho GTPases, Tollip, ubiquitination, UPEC

Received 28 July 2010, revised and accepted for

publication 2 February 2011, uncorrected manuscript

published online 3 February 2011

Several studies point toward an intimate link between thedynamics of the actin cytoskeleton and membrane remod-eling during vesicular trafficking (1–3). Owing to theirability to regulate actin cytoskeleton dynamics and phos-pholipid metabolism, Rho GTPases play essential rolesin endocytosis and phagocytosis (4,5). Rho GTPases arekey signaling proteins that behave as molecular switches.They oscillate between a GDP-bound form, localized inthe cytosol in association with RhoGDI and a GTP-boundform, localized in membranes, able to bind to and activateeffector proteins (4,6). The binding of Rho proteins to theireffectors triggers activation of effector proteins either byunmasking functional domains and/or by targeting themto specific membrane locations. Several pathogenic bac-teria have the ability to trigger their own uptake intonon-phagocytic cells by activating Rho GTPases such asRac1 (7–9). The cytotoxic necrotizing factor-1 (CNF1) is aparadigmatic bacterial virulence factor that directly mod-ifies the regulation of Rho proteins to induce bacterialinvasion. CNF1 toxin is produced by various uropathogenicEscherichia coli (UPEC) clinical isolates responsible forurinary tract infections (UTIs) (for reviews, see 10,11).Persistence and recurrence of UTIs have been linked tothe capacity of UPEC to invade host tissues within the uri-nary tract (11). Initial work by Falzano et al. (12) revealedthe capacity of CNF1 to trigger bacterial invasion of hostcells. The major role of CNF1 in host-cell invasion by UPECwas confirmed also suggesting a central role of Rac1 inthis process (13).

CNF1 is a polypeptide composed of three functionaldomains (14). Its first two amino-terminal domains allowthe sequential endocytosis and translocation of thecarboxy-terminal enzymatic domain into the host cellcytosol to target Rho GTPases (15–18). Upon reaching thecytosol, CNF1 catalyzes the deamidation of a glutamineresidue within the switch-II domain of Rho proteins, trig-gering their permanent activation by blocking Rho GTPaseactivity (19–21). Activation of Rho proteins, notably Rac1,by CNF1 sensitizes these GTPases to ubiquitin-mediatedproteasomal degradation (13). Rho protein activation byphysiologic stimuli, such as the hepatocyte growth fac-tor (HGF), also sensitizes these GTPases to ubiquitinationand proteasomal degradation (13,22–24). Ubiquitinationinvolves the covalent attachment of ubiquitin, an 8-kDapolypeptide, to lysine residues on the target. Conjuga-tion of ubiquitin to cellular targets is achieved througha cascade of transfer reactions between ubiquitin carrierproteins. Additional molecules of ubiquitin can be sub-sequently attached to one of the seven lysines of thepreviously conjugated ubiquitin molecule, leading to theformation of various types of polyubiquitin chains. As anexample, CNF1 triggers a classical K48 polyubiquitinationof Rac1 leading to its proteasomal degradation (13,25).

www.traffic.dk 1

Visvikis et al.

Modification of cellular proteins by ubiquitination con-trols their stability, as well as regulates their associationwith ubiquitin-binding proteins. Indeed, the sorting ofubiquitinated products depends on their recognition byproteins containing ubiquitin-binding domains such as theCUE1p-homologous [coupling of ubiquitin to endoplasmicdegradation (CUE)] domain (26). Ubiquitination and vesic-ular trafficking are intimately connected (27). For instance,ubiquitin plays a key role in the trafficking of Ras (28), andrecent studies have revealed that caveolin-1 controls theefficiency of Rac1 polyubiquitination (29).

Here, we set out to investigate the role of ubiquitin-bindingproteins in the internalization of UPEC triggered by CNF1and Rac1. We used a two-hybrid screen for proteins thatinteract with an ubiquitin-Rac1 (UbRac1L61) chimera. Thisallowed us to identify the CUE-domain containing protein[Toll-interacting protein (Tollip)] as a new partner of Rac1and its ubiquitinated form. Our functional studies demon-strated the involvement of Tollip, together with the Tollip-binding proteins, Tom1 and clathrin, in CNF1-inducedRac1-dependent entry of bacterial pathogens in host cells.

Results

CNF1-induced invasion of cells by UPEC is Rac1

dependent

We first examined the contribution of CNF1 toxin to host-cell invasion by UPEC. We deleted the cnf1 gene in thetype 1-piliated UPEC strain J96 (30) (Figure S1A–C). Cellswere infected with either J96 or J96�cnf1 bacteria, andthe efficiency of cell invasion was determined by a gen-tamicin protection assay that specifically kills extracellularbacteria (31). We characterized the involvement of cnf1in cell invasion by UPEC (Figure 1A), as previously per-formed using a J96cnf1::Tn5 strain (13). Consistent withthese earlier findings, cell invasion by J96�cnf1 could berescued by intoxication of host cells with CNF1, but notwith CNF1C866S catalytically inactive mutant (Figure 1B).CNF1 triggers an activation and ubiquitin-mediated pro-teasomal degradation of Rho proteins, maximal for Rac1,a small GTPase involved in cell invasion by numerouspathogenic bacteria (7,13). This prompted us to exam-ine the involvement of Rac1 in UPEC entry. We firstmeasured that entry of J96�cnf1 could be triggered incells expressing the activated-mutant Rac1E61, mimick-ing CNF1-induced deamidation, or expressing the broadlyused activated-mutant Rac1L61 (Figure 1C). To validatethe specificity of bacterial entry triggered by expression ofactivated mutants of Rac1 we performed additional exper-iments in beta1 integrin knock out cells, considering thatalpha3beta1 integrin is a cell receptor-associated compo-nent of invasion by type 1-piliated UPEC (32). Whereasthe knockout of beta1 integrin had no effect on bacte-rial adhesion (data not shown), it abrogated bacterial entryinduced by Rac1L61, Rac1E61 or CNF1 (Figure 1D). Theseexperiments demonstrate that Rac1 and CNF1 stimulate

a specific bacterial entry into cells rather than a broad-range endocytosis. In CNF1-treated cells, the level ofRac1 is dramatically reduced due to the fact that acti-vated Rac1 undergoes a one-round cycle of synthesis,activation by the toxin, followed by ubiquitination anddegradation by the proteasome. It was important to deter-mine whether, despite low levels of Rac1 in CNF1-treatedcells, Rac1 remains to be able to trigger bacterial entry.To this end we blocked Rac1 synthesis using an RNAiknockdown approach. Depletion of endogenous Rac1 dra-matically reduced the entry of bacteria in CNF1-treatedcells (Figure 1E). Thus, despite reduction of Rac1 levels inintoxicated cells, this GTPase remains an essential factorfor efficient cell invasion by pathogenic bacteria.

Complementary to these findings, we observed that acti-vated Rac1 colocalized with bacteria entering the cells(Figure 1F, arrow), as well as with bacteria already inter-nalized (Figure 1F, arrowhead). Rac1 did not localize withJ96�cnf1 cell-bound bacteria in the absence of CNF1treatment (Figure S1D) while colocalization is observed inCNF1-intoxicated cells (Figure 1G). This shows that acti-vation of Rac1 coincides with its recruitment to bacteriaentry sites. In addition, we observed a clustering of beta1integrin with bacteria invading cells (Figure S1E), whilealpha-5 integrin is not recruited, thereby demonstratingthe specificity of Rac1 and beta1 integrin colocalization ofinvading bacteria (Figure S1F and data not shown).

Together, our data established the requirement of Rac1 inCNF1-induced UPEC entry into host cells.

Involvement of Tollip in bacterial entry

Following its activation by CNF1, Rac1 is subjected to ubiq-uitination (13). We were interested to isolate novel hostfactors that might interfere with Rac1- and CNF1-triggeredcell invasion by pathogenic bacteria. By conducting ayeast two-hybrid screen of proteins interacting with achimera made of ubiquitin fused to the amino-terminus ofRac1L61 (UbRac1L61), we identified the carboxy-terminalpart (amino acids 138-274) of the Tollip which was firstdiscovered as a partner of the IL (interleukin)-1 receptor-1(IL-1R1) (33). Tollip is an adaptor protein, which containsa carboxy-terminal ubiquitin-binding domain CUE, pre-ceded by a C2 domain and an amino-terminal Tom-bindingdomain (TBD), which binds to proteins of the Tom1 (tar-get of the oncogene v-myb 1) family (34,35) (Figure 3A).Tollip can recruit Tom1-bound clathrin to membranes andregulate IL-1R1 sorting in late endosomes (36,37). UPECexpressing the adhesin FimH colonize the urinary tract andenter host cells in a manner dependent on Rho GTPaseand clathrin (38,39). Thus, Tollip seemed to be an attractivecandidate as a protein hijacked by pathogenic bacteria totrigger Rac1-dependent host-cell invasion. To evaluate theinvolvement of Tollip in bacterial entry, we chose to sup-press its expression by constructing two shRNAs targetingTollip. The efficiency of both Tollip shRNAs was confirmedusing quantitative real-time polymerase chain reaction(qRT-PCR) (Figure S2A). We found that knockdown of

2 Traffic 2011

Tollip Facilitates Rac1-Induced Bacterial Invasion

Figure 1: CNF1- and Rac1-dependent internalization of UPEC in host cells. A) Percentages of J96 and J96�cnf1 entry into 804Gepithelial bladder cells. B–E) Graphs show efficiencies of J96�cnf1 internalization in 804G cells (B, C and E) or mouse-derived beta1integrin floxed (β1fl/fl), KO (β1−/−) and rescued (β1resc) cells (D), expressed as arbitrary units (A.U.). A–E) Graphs correspond to onerepresentative experiment performed in triplicate ± SD, *p < 0.05. B) Cells were treated 24 h with 10−9 M of CNF1 or CNF1-C866Scatalytically inactive mutant, and then infected with J96�cnf1. C) Cells were co-transfected with pEGFP and mock (Ctrl), Rac1L61,Rac1E61 or Rac1N17 expression plasmids 24 h prior to infection with J96�cnf1. D) Cells were intoxicated for 24 h with CNF1 at 10−9 M

(CNF1) or transfected with either Rac1L61 or Rac1E61 expression plasmids. Cells were next infected with J96�cnf1. Expression ofRac1 mutants and activity of CNF1 were verified (data not shown). E) Cells were transfected with 100 μM of non-targeting (siCtrl) or Rac1specific siRNAs (siRac1). The next day, cells were treated for 24 h with 10−9 M of CNF1 prior to infection with J96�cnf1. Inset: lysatesfrom cells transfected with non-targeting or Rac1-specific siRNAs were analyzed by western blotting with anti-Rac1 antibody. F andG) Immunofluorescence analysis of HUVECs infected 20 min with J96�cnf1. Infection performed with YFP-Rac1L61-expressing cells(F) or YFP-Rac1-expressing cells intoxicated by CNF1 (G). Bacteria were labeled with anti-J96 antibodies prior to cell permeabilization(external UPEC, blue) and after cell permeabilization (total UPEC, red). Bacteria entering the cell (arrow) and internalized (arrowheads).Bar = 5 μm.

Traffic 2011 3

Visvikis et al.

Tollip inhibited bacterial entry triggered by either CNF1 oractivated Rac1 (Figure 2A–C). We verified that expressionof Tollip shRNAs had no effect on Rac1 activation, didnot trigger cell toxicity and did not impair bacterial adhe-sion to host cells (data not shown). Similar results wereobtained using siRNA targeting Tollip (Figure 2D). Thismechanism of UPEC entry presents similarities with Lis-teria internalization. Indeed, studies on Listeria expressinginternalin-B show that InlB activates the host cell receptorc-Met to trigger a Rac1-dependent internalization of bacte-ria into host cells (40,41). Moreover, Listeria subverts theclathrin-dependent endocytic machinery for InlB-mediatedhost-cell invasion (42). Using the BUG947 strain of Listeriamonocytogenes, whose uptake into host cell is mediatedby InlB, we also observed that Tollip positively inter-feres with L. monocytogenes entry (Figure 2E). We nextexamined the localization of Tollip during bacterial entry.Immunofluorescence analysis allowed us to visualize thecolocalization of green fluorescent protein (GFP)-Tollip andUPEC entering cells (Figure 2F, arrow). Tollip also localizedwith endosomes containing bacteria at early time-pointsof infection (Figure 2F, arrowhead). Similar results wereobtained with Rac1L61- or Rac1E61-induced UPEC inter-nalization (Figure 2G and data not shown). In addition,we detected a clear localization of Tollip at the site ofL. monocytogenes entry (Figure 2H). Together, this func-tional analysis shows that Tollip localizes at the site ofbacterial entry and participates in Rac1-dependent entryof UPEC and Listeria into host cells.

Requirement for the Tollip TBD and CUE domains for

UPEC entry

In order to define more precisely the involvement of Tollipin the Rac1-dependent entry of UPEC, we analyzed theeffect of expression of wild-type Tollip and various Tollipdeletion mutants (Figure 3A). We observed that expres-sion of Tollip deleted of either its TBD or CUE domainblocked the internalization of UPEC triggered by Rac1(Figure 3B). These data suggested that overexpression ofTBD or CUE domains of Tollip likely titrated essential fac-tors involved in bacterial entry. Thus, overexpression ofwild-type Tollip should also titrate these factors and thusreduce bacterial entry. Consistent with a role of Tollip, asan adaptor molecule, we observed that expression of wild-type Tollip blocked the internalization of UPEC triggeredby Rac1 (Figure 3B). Such a scaffolding function of Tollipalso predicted that the deletion of both TBD and CUEdomains should release the titration effect. Consistentwith this idea, we measured a clear decrease of inhibitionof bacterial entry in cells expressing a mutant of Tollipdeleted both TBD and CUE domains (Figure 3B). In theseexperiments, we verified equal expression of activatedRac1, as well as Tollip constructs (data not shown). Wealso verified that expression of wild type and mutatedforms of Tollip had similar effects when bacterial uptakewas driven by CNF1 (data not shown). Finally, we verifiedthat Tollip expression had no impact on the proteasomaldegradation of Rac1 in CNF1-intoxicated cells (Figure 3C).

These data clearly indicated that the CUE and TBDdomains of Tollip were both involved in bacterial entry.

Interaction of Tollip and Rac1

Here we have isolated Tollip as an interacting proteinof UbRac1L61 and determined its involvement in Rac1-induced bacterial invasion of host cells. We next charac-terized the conditions of interaction of Tollip with Rac1.By conducting co-immunoprecipitation experiments wefirst determined that Tollip associates with Rac1L61 andhigh molecular weight forms of Rac1L61 (Figure 4A). Todirectly demonstrate that these high molecular weightforms of Rac1L61 correspond to polyubiquitinated Rac1,we performed a glutathione-S-transferase (GST) pulldownwith GST-Tollip followed by His-tagged ubiquitin precip-itation of associated proteins using denaturing condi-tions to avoid co-purification of non-ubiquitinated proteins(Figure 4B). This directly confirmed the capacity of Tol-lip to associate with polyubiquitinated forms of Rac1L61.CNF1 induces a polyubiquitination of Rac1. Consistentwith the above data, we observed an increase of theassociation of Tollip with high molecular weight formsof Rac1 in CNF1-intoxicated cells (Figure 4C). Intoxica-tion of cells with CNF1 also increased the association ofRac1 to Tollip. We thus further investigated the speci-ficity of association of Tollip with Rac1. We mutated thethreonine-35 residue of Rac1 into alanine (Rac1A35), akey amino-acid residue of the effector-binding domain ofRac1 and a hot spot of modification by large glucosylatingtoxins of Clostridium (7,8). We found that mutation T35Aabolished Rac1L61 association with Tollip and Rac1L61-driven bacterial entry into cells (Figure 4A and data notshown).

Together our data establish that Tollip is an effector ofRac1 involved in cell invasion by pathogenic bacteria.

Involvement of Tom1 and clathrin in UPEC entry

The involvement of components of the endocytic machin-ery, notably clathrin, in type 1-pili-expressing E. coliprompted us to determine whether the Tollip-associatedfactor Tom1 and the Tom1-associated protein clathrincould also play a role in CNF1-dependent Rac1-inducedbacterial internalization (39). Using Tom1 siRNA, whoseefficiency was verified by qRT-PCR (Figure S2B), we ana-lyzed the involvement of Tom1 in cell invasion by UPEC.Knockdown of Tom1 reduced bacterial entry triggeredby CNF1 or by activated Rac1 (Figure 5A,B), withoutaffecting cell viability and bacterial adhesion to hostcells (data not shown). We next examined the role ofclathrin in bacterial entry. Cells were transfected withshRNA targeting the clathrin heavy chain (shCHC) andthen intoxicated with CNF1. Note that CNF1 does notenter cells by clathrin-mediated endocytosis (15). Thefunctionality of shRNA was confirmed by immunoblot-ting (Figure S2C). Results show that clathrin heavy chainknockdown reduced bacterial entry triggered by CNF1(Figure 5C). Cells were next transfected with shCHC and

4 Traffic 2011

Tollip Facilitates Rac1-Induced Bacterial Invasion

Figure 2: Functional implication of Tollip in Rac1-dependent bacterial entry. A–E) Efficiencies of bacterial entry under variousconditions. Results are expressed as arbitrary units (A.U.). Mean values are of n = 3 experiments ± SD, *p < 0.05. A–C) J96�cnf1entry in control and Tollip shRNA (shTollip) expressing cells treated 24 h with CNF1 10−9 M (A) or co-transfected with Rac1L61 (B) orRac1E61 (C). D) J96�cnf1 entry in control and Tollip siRNA (siTollip) expressing cells treated 24 h with CNF1 10−9 M. E) InlB-expressingListeria (BUG947) internalization in mock (Ctrl) or Tollip shRNA (shTollip) expressing HeLa cells. F and G) Immunofluorescence analysisof HUVECs transfected with GFP-Tollip and intoxicated by CNF1 10−9 M (F) or co-transfected with HA-Rac1L61 (G) 24 h prior to 20 mininfection with J96�cnf1. Bacteria were labeled with anti-J96 antibodies prior to cell permeabilization (external UPEC, blue) and aftercell permeabilization (total UPEC, red). Arrows show bacteria entering the cell. Arrowheads show bacteria internalized. Bar = 5 μm. H)Immunofluorescence analysis of HeLa cells expressing GFP-Tollip and infected 1 h with InlB-expressing Listeria. Bacteria were labeledwith anti-L. monocytogenes prior to cell permeabilization (external Listeria, blue) and after cell permeabilization (total Listeria, red). Arrowshows bacteria entering the cell. Arrowhead shows internalized bacteria. Bar = 5 μm.

Traffic 2011 5

Visvikis et al.

Figure 3: TBD- and CUE-dependent involvement of Tollip in

bacterial entry. A) Schematic representation of wild-type Tol-lip (Tollip WT), Tollip delta Tom-binding domain (Tollip�TBD),Tollip delta coupling of ubiquitin to endoplasmic degrada-tion domain (Tollip�CUE) or Tollip delta-TBD and delta-CUE(Tollip�TBD�CUE). B) Efficiencies of J96�cnf1 internalization inHUVECs co-transfected with expression plasmids encoding HA-Rac1L61 and with either HA-Tollip WT or Tollip mutants. Meanvalues are of n = 3 experiments ± SD, *p < 0.05 versus controlcells. C) Immunoblots (left) and quantification (right) showing theabsence of effect of Tollip expression on CNF1-induced Rac1degradation. HEK293 cells were transfected with empty vectoror HA-Tollip and treated for different periods of time with CNF1(10−9 M), as indicated. Expression of endogenous Rac1 and HA-Tollip was monitored on total protein extracts by immunoblottinganti-Rac1 and anti-HA. Levels of Rac1 were quantified and nor-malized to actin (Graph shows mean values of n = 3 experiments± SD).

activated Rac1. Figure 5D shows the inhibition of bacterialentry in clathrin heavy chain knock down cells express-ing activated Rac1. Thus, both CNF1 and activated-Rac1induced a Tom1- and clathrin-dependent entry of UPEC.We then analyzed by immunofluorescence the localizationof Tom1 and clathrin at early time-points of cell infec-tion. Cells expressing GFP fused to Tom1 (GFP-Tom1) orto clathrin light chain A (GFP-Clathrin LCA) were eitherintoxicated with CNF1 or co-transfected with activated

Figure 4: Association of Tollip with activated Rac1 and

ubiquitinated Rac1. A) Immunoblots showing the interactionof HA-Rac1 and mutants (IB : HA) with Flag-Tollip (IB : Flag).Immunoprecipitation (IP : HA) from HEK293 cells co-transfectedwith expression plasmid of HA-Rac1L61 or HA-Rac1A35 andFlag-Tollip. Input shows expression of constructs in lysates.B) Immunoblot anti-HA showing ubiquitinated forms of Rac1associated with Tollip. CHO cells were co-transfected withexpression plasmids of HA-Rac1L61, GST-Tollip and His-Ub,as indicated and processed for GST-P followed by His-taggedubiquitin precipitation (GST-P + HisP). Lower panels correspondto anti-GST and anti-HA immunoblots on total lysate. C)Immunoprecipitation (IP : HA) from HEK293 cells co-transfectedwith expression plasmid of HA-Rac1 wild type and Flag-Tollip,next intoxicated 24 h with CNF1 10−9 M. Input shows expressionof constructs in lysates.

6 Traffic 2011

Tollip Facilitates Rac1-Induced Bacterial Invasion

Rac1 and exposed to bacteria for 20 min. This allowed usto visualize a colocalization between Tom1 and bacteriaentering cells (Figure 5E,F), as well as a punctate signalof clathrin around bacteria entering cells (Figure 5G,H).Collectively, our results show that the CNF1- and Rac1-dependent entry of bacteria requires Tom1 and clathrincomponents of the endocytic machinery.

Interaction and localization of Rac1, Tollip and Tom1

at sites of bacterial entry

The aforementioned results showed that Rac1, Tollip andTom1 are host components hijacked by UPEC to triggertheir efficient internalization into cells. Together, thesefindings suggested a possible association of Rac1 andTollip with Tom1 and clathrin. In accordance with thishypothesis, we observed that Rac1 colocalizes with Tollipat the site of bacterial entry (Figure 6A). We also observeda colocalization of Rac1 together with Tom1 at the siteof bacterial entry (Figure 6B). Finally, we detected a colo-calization of Tollip with Tom1 together with bacteria atearly time-points of infection (Figure 6C). Importantly,immunoprecipitation experiments revealed an interac-tion of Rac1 together with Tollip, Tom1 and clathrin(Figure 6D).

Our data show that the active form of Rac1 interacts withTollip, Tom1 and clathrin and reveal the involvement ofthese components in bacterial entry into cells.

Discussion

We found that CNF1-producing UPEC hijack Rac1 andTollip, as well as the Tollip, Tom1 and clathrin, to triggerbacterial internalization into host cells. Indeed, by func-tional analysis we uncovered the role of Tollip, Tom1and clathrin in CNF1-induced Rac1-dependent entry ofUPEC into host cells. Complementary to these findingswe showed that Tollip, Tom1 and clathrin associate to theactive form of Rac1 and localize at the site of bacterialentry.

To date, several studies have established that Tollip,through its binding to Tom1, recruits clathrin to endo-somes (35,36) and participates to the sorting of IL-1R1in endosomes (37). Here, we provide evidence that Tol-lip is localized at the plasma membrane during infection.Indeed, we could visualize by immunofluorescence thatTollip colocalized with UPEC entering cells. In addition,we showed that Tollip is functional at the plasma mem-brane, as its knockdown significantly reduced UPEC entryinto host cell. Moreover, we show that Tollip binds theactive form of Rac1. Our results thus suggest that UPEChave hijacked Tollip, and likely Tom1 to trigger theirRac1-dependent entry into host cells. Interestingly, wehave observed similar results with the Rac1- and clathrin-dependent endocytosis of Listeria. Tollip colocalized withListeria entering cells and Tollip knockdown reduced InlB-dependent Listeria entry. This suggests that the role of

Tollip in Rac1-dependent bacterial entry does not only relyon permanent activation of Rac1, for instance by CNF1-induced deamidation, but also depends on Rac1 activationby InlB-induced activation of c-Met. This suggests a pos-sible role for Tollip in cell receptor endocytosis. It will thusbe interesting to investigate whether Tollip is involvedin the clathrin-dependent endocytosis of c-Met when itis activated by its physiologic ligand HGF (43). Clathrin-adaptor molecule, AP2, as well as the alternative clathrinadaptors Numb, ARH and Dab2 are involved in cell entry bytype 1-pili-expressing E. coli (39). These alternative clathrinadaptors recognize NPXY motifs found within the cytosolictail of beta1 integrin. We show that beta1 localizes at thesite of bacterial entry and is essential for Rac1- and CNF1-triggered cell invasion. In addition, we have observed thatthe CNF1-induced Rac1-dependent bacterial entry alsorequires AP2 (data not shown). The requirement of severalclathrin-adaptor molecules for bacterial entry may indicatethe need of high amounts of clathrin molecules to allowthe uptake of large particles such as bacteria. Rac1 maythus contribute to clathrin recruitment by binding to Tollipand Tom1. A non-exclusive hypothesis is that Rac1 modu-lates actin polymerization at the level of membrane-boundTollip. Indeed, the CNF1-induced and Rac1-dependentinvasion of host cells by bacteria is reduced by treat-ment with actin depolymerizing drugs (data not shown).In support of this hypothesis, a recent study shows thatactin polymerization induced by Rac1 facilitates the for-mation of transport carriers by the clathrin heavy chainbinding protein, CYFIP, at the trans Golgi network (1).Finally, Tollip might also organize clathrin lattice structuresat the site of bacterial entry (44). Several groups haveestablished that once Rac1 is activated, for instance byCNF1, it is ubiquitinated and conveyed to the proteaso-mal degradation machinery (13,22,23). Consequently thelevel of active Rac1 is considerably reduced in CNF1-intoxicated cells (13). This raised the question of whetherRac1 was still required for cell invasion by CNF1-producingUPEC (13). Here we provide new evidence showing therequirement of Rac1 for CNF1-induced bacterial invasion.In addition, we here isolate Tollip as a new partner of Rac1and ubiquitinated Rac1, positively involved in bacterialinvasion. This establishes that the two-hybrid screeningmethod using the Ub-RacL161 chimera as a bait is avaluable strategy to identify new partners of Rac1 andpotentially of other small Rho GTPases. Finally, Tollip bindsboth Rac1 and ubiquitinated forms of Rac1. Our data thusshow that polyubiquitination of Rac1 does not alter itscapacity to associate with Tollip. This suggests a possiblefunction of the ubiquitination of Rac1. Whether and howubiquitination of Rac1 interferes with Rac1 and/or Tollipfunction awaits the discovery of the E3-ubiquitin ligaseresponsible for Rac1 ubiquitin-mediated degradation.

In conclusion, we here identified Tollip as a new effectorof Rac1 hijacked by pathogenic bacteria for efficient cellinvasion.

Traffic 2011 7

Visvikis et al.

Figure 5: Tom1 and clathrin involvement in Rac1-induced UPEC entry. A–D) Efficiencies of J96�cnf1 internalization in HUVECs,expressed as arbitrary units (A.U.). Mean values of n = 3 experiments ± SD, *p < 0.05 versus control cells. A and B) Cells were eithertransfected with 100 μM non-targeting siRNA (siCtrl) or 100 μM siRNA targeting Tom1 (siTom1), 72 h prior to infection. Cells were eitherintoxicated with CNF1 10−9 M 24 h prior to infection (A) or co-transfected with Rac1L61-expressing plasmid (B). C and D) Cells weretransfected with control plasmid (Ctrl) or plasmid expressing clathrin heavy chain shRNA (shCHC) and either intoxicated with CNF1 10−9 M

(C) or co-transfected with Rac1L61-expressing plasmid (D). E–H) Immunofluorescence analysis of HUVECs transfected with GFP-Tom1(E and F) or GFP-Clathrin LCA (G and H) and intoxicated with CNF1 10−9 M (E and G) or co-transfected HA-Rac1L61-expressing plasmid(F and H), 24 h prior to bacterial infection with J96�cnf1. Bacteria were labeled with anti-J96 antibodies prior to cell permeabilization(external UPEC, blue) and after cell permeabilization (total UPEC, red). Bacteria entering the cell (arrow) and internalized (arrowheads).Bar = 5 μm.

8 Traffic 2011

Tollip Facilitates Rac1-Induced Bacterial Invasion

Figure 6: Colocalization and binding of Rac1 with Tollip and Tom1 during bacterial entry. A–C) Immunofluorescence analysisof HUVECs infected with J96�cnf1. Bacteria were labeled with anti-J96 antibodies after cell permeabilization (total UPEC, blue) andHA-tagged expression constructs with anti-HA antibody (red). Bar = 5 μm. A) Cells expressing YFP-Rac1L61 and HA-Tollip. B) Cellsexpressing GFP-Tom1 and HA-Rac1L61. C) Cells expressing GFP-Tom1 and HA-Tollip and intoxicated with CNF1 10−9 M, 24 h prior toinfection. D) Immunoblots showing the interaction of Flag-Rac1L61 (IB : Flag) with HA-Tollip, HA-Tom1 (IB : HA) and clathrin heavychain (IB : CHC). Flag immunoprecipitation (IP : Flag) of HEK293 cells transfected with Flag-Rac1L61, HA-Tollip and HA-Tom1 expressionplasmids. Input shows expression of constructs and endogenous CHC in lysates.

Materials and Methods

Bacterial strainsBacterial strains are shown in Table 1.

Cells, culture and transfectionCells used were human embryonic kidney cells (HEK293), human umbilicalvein endothelial cells (HUVECs) and human cervical carcinoma cells(HeLa); Chinese hamster ovary cells (CHO); rat epithelial bladder cells(804G) and human cells were grown as described (13,46). The beta1

floxed cells (β1fl/fl) correspond to primary mouse osteoblast-enrichedcells isolated from beta1fl/fl newborn calvarias, immortalized with SV40large T antigen, described in (47). Beta1 integrin KO cells (β1−/−) cellswere generated from β1fl/fl infected with an adenoviral supernatantencoding the Cre recombinase (kindly provided by Dr R. Meuwissen,Institut Albert Bonniot). Cells rescued in beta1 integrin (β1resc) weregenerated from β1−/− cells infected with the pCLMFG-beta1 vector,as described (48). For transfection, we used ExGen 500 (Euromedex)for 804G cells, lipofectamine 2000 for HEK293 cells (Invitrogen) orelectroporation for HUVECs, as described previously (46). siRNAs weretransfected using PolyMag reagent (OZ Biosciences) following themanufacturer’s procedure.

Traffic 2011 9

Visvikis et al.

Table 1: Bacterial strains

Strains DescriptionRelevant

phenotype References

J96 Uropathogenic wild-typeE. coli strain (O4:K6)

cnf1+ (30)

J96�cnf1 J96 rpsL (SmR) �cnf1 cnf1− This workBUG947 Listeria EGD�inlA InlA− (45)

Table 2: List of primer sequences

Primer Sequence

P1 5′-ATAAGGTCATCCACTGCACGP2 5′-GATTCCCGTGAAAGAGCACP3 5′-GGGGAATTCGATGGCGACCACCGTCAGCACP4 5′-GGGGGATCCTTATGGCTCCTCCCCCATCTGCAP5 5′-GGATCCATGGCGACCACCGTCAGCACP6 5′-GAATTCTTATGGCTCCTCCCCCATCTGCAP7 5′-GCGAATTCATGGACTTTCTCCTGGGGAACCCP8 5′-CGGGATCCTCATAAGGCAAACAGCATGTCATCATCP9 5′-GGATCCATGCAGATCTTCGTGAAGACP10 5′-GGATCCACCACCTCTTAGTCTTAAGP11 5′-CCGCAGAAGCGAGAATGGTTP12 5′-ACCGATGCGTTCAGCGAGTTP13 5′-GAGGACCTGAAAGCCATCCAP14 5′-CACCTCCTGGTCCATGTTGGP15 5′-GACACGCCCATAGCACCAAP16 5′-CACGTTCCCACTCACCATCTC

Plasmid constructs and two-hybrid screeningPrimer sequences are listed in Table 2. For Tollip knockdown, byshRNA expression, we used two plasmids pSuper-Tollip1 and pSuper-Tollip2 generated following the manufacturer’s procedure using sequencetargeting, P1 and P2, respectively (Oligoengine). pSuper-clathrin expressiontargeting human shCHC was constructed using the sequence as previouslydescribed (42). Plasmids pXJ-HA-Rac1 (wild type), pXJ-HA-Rac1L61, pXJ-HA-Rac1L61A35, pXJ-HA-Rac1E61, pXJ-HA-Rac1N17 and pRGB4-6His-Ubare described in (13). Flag-Rac1L61 expression plasmid was obtained bysubcloning Rac1L61 cDNA into pCMV-Tag2B plasmid (Stratagene) usingBamHI/EcoRI. YFP-Rac1 and L61 mutants were obtained by subcloningRac1 cDNAs into pEYFP-C2 (Clontech), using BamHI/EcoRI. Expressionplasmids of HA-tagged Tom1, Tollip and Tollip mutants are describedin (34). Tollip cDNA was generated by PCR using primers P3 and P4and subcloned into pCMV2-Flag and pEGFP-C1 (Clontech) plasmids usingEcoRI/BamHI to obtain Flag-Tollip and GFP-Tollip expression vectors,respectively. Mammalian expression plasmid encoding GST-tagged Tollipwas generated by PCR using primers P5 and P6 and subcloned usingBamHI/EcoRI in pLef plasmid (49). Tom1 cDNA was obtained by PCR usingP7 and P8, and subcloned into pEGFP-C2 using BamHI/EcoRI (Clontech).GFP-Clathrin LCA is described in (42). All constructions were verified bysequencing entire coding regions. For the yeast two-hybrid screen, thecDNA encoding ubiquitin was PCR amplified from the pRBG4-6HisUbusing the primers P9 and P10 and cloned BamHI into the pXJ-HA-Rac1L61in order to generate pXJ-HA-Ub-Rac1L61. The cDNA sequence of HA-Ub-Rac1L61 was subcloned by hybrigenics into the pB27 plasmid (Hybrigenics;http://www.hybrigenics.com/) in order to generate a LexA-UbRac1L61yeast expression plasmid. The yeast two-hybrid screen was performedby hybrigenics with UbRac1L61 as bait to screen a random-primed cDNAlibrary from human placenta. We have isolated 27 different UbRac1L61interacting proteins. The cDNA sequence of Tollip (411-1304) encoded thecarboxy-terminal part amino acids 138-274.

Reagents and antibodiesTollip siRNA was purchased from Ambion. siRNA pool against Rac1and Tom1 as well as control siRNA pool were purchased from Santa

Cruz. Antibodies used in this study were anti-Flag M2 (Sigma-Aldrich);anti-HA [clone 16B12] (Covance); anti-HA [clone 3F10] (Roche); anti-GST [clone 26H1] (Cell Signalling Technology); anti-Rac1 [clone 102] (BDTransduction Laboratories); anti-ubiquitin [clone P4D1] (Santa Cruz); anti-β-actin [clone AC-74] (Sigma-Aldrich); anti-clathrin heavy chain [clone 23] (BDTransduction Laboratories); anti-beta1 integrin [clone JB1A] (Chemicon);anti-alpha-5 integrin [clone SAM-1] (Chemicon); Cy™5- or Texas Red-conjugated donkey anti-rabbit and Cy™5-conjugated donkey anti-mouseantibodies (Jackson Immunoresearch Laboratories); Texas Red-conjugatedhorse anti-mouse (Vector Laboratories); horseradish peroxidase (HRP)-conjugated goat anti-mouse (DAKO); HRP-conjugated swine anti-rabbit(DAKO); HRP-conjugated rabbit anti-rat (DAKO); anti-J96 (Eurogentec) andanti-CNF1 (Agro Bio SA) were generated in rabbit using J96 purifiedmembranes and purified CNF1-C866S, respectively. Serum anti-alpha-hemolysin of E. coli was kindly provided by Dr V. Koronakis. Rabbit anti-L.monocytogenes (R11) has been previously described in (42).

Generation of a �cnf1 derivative of the UPEC strain

J96The two-step procedure for the replacement of specific DNA sequences inthe E. coli chromosome was performed as described previously (50).A PCR fragment containing the cnf1 was generated using genomicDNA from UPEC strain J96 and oligonucleotides L (see Table 2, P11)and R (see Table 2, P12). The fragment was digested with BspHI andinserted into the BspHI site of plasmid pLN135. The resulting plasmidwas digested with BstEII and BsaI to remove 2962 bps out of 3044of cnf1 (Figure S1A). The protruding ends were rendered blunt usingKlenow and ligated, to yield plasmid pMLM151. Briefly, we used pLN135,a replication temperature-sensitive plasmid bearing a chloramphenicolresistance gene (cat) and a counterselectable rpsLwt marker that confersdominant sensitivity to streptomycin (Sm) to Sm-resistant bacteria. ThepMLM151 plasmid was used to transform a spontaneous Sm-resistant(SmR) mutant of UPEC strain J96. Following transformation, plasmidintegration into the chromosome was selected by plating bacteria onchloramphenicol-containing plates at 42◦C. Excision of cnf1 was selectedby plating bacteria on medium containing Sm to yield the �cnf1 strain.Assessment of the deletion of cnf1 was carried out by means of PCRamplification of total genomic DNA from J96 and J96�cnf1 (Figure S1B)and by immunoblotting anti-CNF1 (Figure S1C).

Bacterial internalization assayMeasurements of bacterial internalization were performed by gentamicinprotection assay as described previously (13). In order to study CNF1function and to get reproducible bacterial internalization results, cellswere intoxicated with CNF1 and infected with J96�cnf1. Briefly, epithelialbladder cells (4 × 105 cells), HUVECs (5 × 104 cells) or HeLa cells (4 × 105

cells), were seeded in 12-well plates 24 h before infection. Exponentiallygrowing UPEC, grown in static conditions to DO600 = 0.8, were added ontocells (10 bacteria per cell). Cell infection by UPEC was conducted during20 min at 37◦C. Infected cells were washed three times with PBS andeither lysed for cell-bound bacteria measurements or incubated another30 min with 50 μg/mL of gentamicin before lysis for internalized bacteriameasurements. Cells were lysed in PBS 0.1% Triton-X-100 and bacteriaplated on LB plates supplemented with 200 μg/mL Sm (Sigma-Aldrich) forcounting. For Listeria infection assay, exponentially growing Listeria wereadded to cells at 100-fold excess. Cell infection was allowed during 1 h at37◦C (cell-associated bacteria), followed by 2 h of treatment with 5 μg/mLof gentamicin (internalized bacteria). Cells were lysed and bacteria wereplated on brain heart infusion plates for counting. In order to determinethe efficiency of J96 and J96�cnf1 entry (Figure 1A), overnight culturesof bacteria were diluted 1/100 and cultured 3 h in LB supplemented withSm. Bacteria were then resuspended in PBS at OD = 1.5 and incubatedfor 40 min at 37◦C. Next, epithelial bladder cells were infected for 1 hwith 1/103 cell/bacteria. Cells were washed and incubated 30 min withgentamicin. Internalization (Int.) of UPEC and Listeria represents a ratio ofinternalized bacteria to cell-bound bacteria, in the absence of differencesmeasured between the various conditions for cell-bound bacteria.

10 Traffic 2011

Tollip Facilitates Rac1-Induced Bacterial Invasion

GST pulldown of ubiquitinated-Rac1CHO cells (107 cells) were electroporated with pRBG4-6HisUb and pXJ-HA-Rac1L61 together with either pLef-GST or GST-Tollip. Cells were lysedand processed for GST pulldown, as described for GST-Pak pulldownin (46). Proteins bound to glutathione-agarose beads were resuspended in1 mL of ubiquitin lysis buffer (8 M urea, 20 mM Tris–HCl pH 7.5, 200 mM

NaCl, 10 mM imidazole, 0.1% Triton-X-100) and processed for His-taggedubiquitin precipitation and immunoblotting, as described (46).

Immunoprecipitation experimentsHEK293 cells (107 cells) were transfected using lipofectamine 2000with pXJ-HA-Rac1, pXJ-HA-Rac1L61, pXJ-HA-Rac1L61A35, pCMV2-Flag-Rac1L61, pCMV2-Flag-Tollip, pcDNA3-HA-Tollip and pcDNA3-HA-Tom1as indicated in the figure legend. Cells were lysed in 1 mL ofimmunoprecipitation buffer (IPB) (20 mM Tris pH 7.5, 150 mM NaCl, 10 mM

MgCl2, 20 mM NaF, 1% Triton-X-100, 1 mM phenylmethylsulfonyl fluoride,2 mM Na3VO4, 2 mM DTT and 20 mM β-glycerophosphate). Clearedlysates were incubated with 4 μg of mouse Flag-M2 antibody for 2 hat 4◦C. Immunoprecipitates were immobilized on Dynabeads® Protein G(Invitrogen) for 2 h at 4◦C. Beads were washed three times with 1 mLof IPB and resuspended in 50 μL of Laemmli buffer. Forty microlitersof immunoprecipitate were analyzed by immunoblotting with rat anti-HAantibody to reveal HA-tagged co-immunoprecipitates. Ten microliters ofimmunoprecipitate were analyzed using anti-Flag antibody to show controlimmunoprecipitated proteins.

qRT-PCR analysisStandard conditions of qRT-PCR described in (25) were used to measuredepletion of endogenous Tollip mRNA using shRNA at 24 h using primers,P13 and P14, and depletion of endogenous Tom1 mRNA using siRNA mixat 72 h using primers, P15 and P16.

Microscopy techniquesFor immunofluorescence studies, cells were processed using standardtechniques. In order to discriminate external from internal bacteria, priorto cell permeabilization, bacteria were stained with anti-J96 and Cyan5-conjugated anti-rabbit antibody. After cell permeabilization, total bacteriawere stained with anti-J96 and Texas Red-conjugated anti-rabbit. Pictureswere taken on an LSM510 Meta confocal microscope (Zeiss) using a 63×lens (total magnification 630×).

Statistical analysisStatistical data analysis was performed with unpaired two-tailed t-testswith Prism (GraphPad Software Inc.).

Acknowledgments

We are grateful to Dr Kazu Nakayama for sharing Tom1 and Tollipexpression plasmids and Vassilis Koronakis for anti-alpha-hemolysin anti-sera. We thank Dr Jacques Bertoglio, Dr Jacques Camonis, Dr MireilleCormont and Dr Rosine Haguenauer-Tsapis for fruitful discussions. Wethank the conseil regional PACA and the Conseil general des Alpes-Maritimes for their financial support and the MiCa microscopy facilityplatform at the C3M Research Center. Our laboratory is supported by aninstitutional funding from the INSERM, a grant and fellowship to O.V. fromthe Agence Nationale de la Recherche (ANR A05135AS, R07120AA andR07113AS), a grant and fellowship to O.V. from the Association pour laRecherche sur le Cancer (ARC 4906) and a fellowship to L. B. from theLigue Nationale contre le Cancer.

Supporting Information

Additional Supporting Information may be found in the online version ofthis article:

Figure S1: Description of the J96 and the J96�cnf1 bacterial strains.

A) Map of the cnf1 region in the parental strain J96 and J96�cnf1strain. BstEII and BsaI, the restriction enzymes used for the deletion,are shown. The sequences flanking the cnf1 gene, including the intergenicsequence (IGS) as well as the cnf1 promoter (Pcnf1) and ribosome-bindingsite (RBS) remain intact in the �cnf1 construct. The oligonucleotidesused for the PCRs are shown by arrows (L and R). B) Agarose gelelectrophoresis of PCR products obtained using J96 and J96�cnf1 totalgenomic DNA. Size in base pair of PCR fragments is indicated by arrows.C) Immunodetection of CNF1 and alpha-hemolysin A (HlyA) of strainsJ96 and J96�cnf1. Bacteria were lysed in Laemmli buffer and proteinresolved on a 7% SDS–PAGE. D–F) Immunofluorescence analysis ofHUVECs infected 20 min with J96�cnf1. Bacteria were labeled with anti-J96 antibodies prior to cell permeabilization (external UPEC, blue) and aftercell permeabilization (total UPEC, red). Bacteria entering the cell (arrow)and internalized (arrowheads). Bar = 5 μm. Infection performed on YFP-Rac1 expressing (D) or CNF1-treated (E and F) cells. E) Beta1 integrin waslabeled with anti-beta1 antibodies after cell permeabilization (green). F)Alpha-5 integrin was immunolabeled with anti-alpha-5 antibodies after cellpermeabilization (green).

Figure S2: Validation of RNAi experiments. A) Graphs show mRNAlevels of Tollip in HUVECs transfected with control or plasmids expressingshRNAs against Tollip (shTollip). Total mRNA were extracted 24 h aftertransfection and processed for RT-qPCR analysis. B) Graphs show Tom1mRNA levels in HUVECs transfected with 100 μM non-targeting siRNA(siCtrl) or 100 μM siRNA targeting specifically Tom1 (siTom1). Total mRNAwere extracted 72 h after transfection and processed for RT-qPCR analysis.C) Immunoblot anti-CHC showing CHC expression in HUVECs transfectedwith control vector or plasmid expressing shRNA against CHC. Lowerpanel: immunoblot anti-β-actin shows equal loading in both conditions.

Please note: Wiley-Blackwell are not responsible for the content orfunctionality of any supporting materials supplied by the authors.Any queries (other than missing material) should be directed to thecorresponding author for the article.

References

1. Anitei M, Stange C, Parshina I, Baust T, Schenck A, Raposo G, Kirch-hausen T, Hoflack B. Protein complexes containing CYFIP/Sra/PIR121coordinate Arf1 and Rac1 signalling during clathrin-AP-1-coated carrierbiogenesis at the TGN. Nat Cell Biol 2010;12:330–340.

2. Merrifield CJ, Perrais D, Zenisek D. Coupling between clathrin-coated-pit invagination, cortactin recruitment, and membrane scissionobserved in live cells. Cell 2005;121:593–606.

3. Merrifield CJ, Feldman ME, Wan L, Almers W. Imaging actin anddynamin recruitment during invagination of single clathrin-coated pits.Nat Cell Biol 2002;4:691–698.

4. Ridley AJ. Rho GTPases and actin dynamics in membrane protrusionsand vesicle trafficking. Trends Cell Biol 2006;16:522–529.

5. Kaksonen M, Toret CP, Drubin DG. Harnessing actin dynamics forclathrin-mediated endocytosis. Nat Rev Mol Cell Biol 2006;7:404–414.

6. Jaffe AB, Hall A. RHO GTPases: biochemistry and biology. Annu RevCell Dev Biol 2005;21:247–269.

7. Boquet P, Lemichez E. Bacterial virulence factors targeting RhoGTPases: parasitism or symbiosis? Trends Cell Biol 2003;13:238–246.

8. Aktories K, Barbieri JT. Bacterial cytotoxins: targeting eukaryoticswitches. Nat Rev Microbiol 2005;3:397–410.

9. Patel JC, Galan JE. Differential activation and function of RhoGTPases during Salmonella-host cell interactions. J Cell Biol2006;175:453–463.

10. Lemonnier M, Landraud L, Lemichez E. Rho GTPase-activating bacte-rial toxins: from bacterial virulence regulation to eukaryotic cell biology.FEMS Microbiol Rev 2007;31:515–534.

11. Dhakal BK, Kulesus RR, Mulvey MA. Mechanisms and consequencesof bladder cell invasion by uropathogenic Escherichia coli. Eur J ClinInvest 2008;38(Suppl. 2):2–11.

12. Falzano L, Fiorentini C, Donelli G, Michel E, Kocks C, Cossart P,Cabanie L, Oswald E, Boquet P. Induction of phagocytic behaviourin human epithelial cells by Escherichia coli cytotoxic necrotizingfactor type 1. Mol Microbiol 1993;9:1247–1254.

Traffic 2011 11

Visvikis et al.

13. Doye A, Mettouchi A, Bossis G, Clement R, Buisson-Touati C,Flatau G, Gagnoux L, Piechaczyk M, Boquet P, Lemichez E. CNF1exploits the ubiquitin-proteasome machinery to restrict Rho GTPaseactivation for bacterial host cell invasion. Cell 2002;111:553–564.

14. Lemichez E, Flatau G, Bruzzone M, Boquet P, Gauthier M. Molecularlocalization of the Escherichia coli cytotoxic necrotizing factor CNF1cell-binding and catalytic domains. Mol Microbiol 1997;24:1061–1070.

15. Contamin S, Galmiche A, Doye A, Flatau G, Benmerah A, Boquet P.The p21 Rho-activating toxin cytotoxic necrotizing factor 1 isendocytosed by a clathrin-independent mechanism and enters thecytosol by an acidic-dependent membrane translocation step. MolBiol Cell 2000;11:1775–1787.

16. Pei S, Doye A, Boquet P. Mutation of specific acidic residues of theCNF1 T domain into lysine alters cell membrane translocation of thetoxin. Mol Microbiol 2001;41:1237–1247.

17. Kim KJ, Chung JW, Kim KS. 67-kDa laminin receptor promotesinternalization of cytotoxic necrotizing factor 1-expressing Escherichiacoli K1 into human brain microvascular endothelial cells. J Biol Chem2005;280:1360–1368.

18. Knust Z, Blumenthal B, Aktories K, Schmidt G. Cleavage ofEscherichia coli cytotoxic necrotizing factor 1 is required for fullbiologic activity. Infect Immun 2009;77:1835–1841.

19. Flatau G, Lemichez E, Gauthier M, Chardin P, Paris S, Fiorentini C,Boquet P. Toxin-induced activation of the G protein p21 Rho bydeamidation of glutamine. Nature 1997;387:729–733.

20. Schmidt G, Sehr P, Wilm M, Selzer J, Mann M, Aktories K. Gln 63 ofRho is deamidated by Escherichia coli cytotoxic necrotizing factor-1.Nature 1997;387:725–729.

21. Lerm M, Selzer J, Hoffmeyer A, Rapp UR, Aktories K, Schmidt G.Deamidation of Cdc42 and Rac by Escherichia coli cytotoxicnecrotizing factor 1: activation of c-Jun N-terminal kinase in HeLacells. Infect Immun 1999;67:496–503.

22. Lerm M, Pop M, Fritz G, Aktories K, Schmidt G. Proteasomal degra-dation of cytotoxic necrotizing factor 1-activated rac. Infect Immun2002;70:4053–4058.

23. Lynch EA, Stall J, Schmidt G, Chavrier P, D’Souza-Schorey C.Proteasome-mediated degradation of Rac1-GTP during epithelial cellscattering. Mol Biol Cell 2006;17:2236–2242.

24. Visvikis O, Maddugoda MP, Lemichez E. Direct modifications of Rhoproteins: deconstructing GTPase regulation. Biol Cell 2010;102:377–389.

25. Boyer L, Turchi L, Desnues B, Doye A, Ponzio G, Mege JL,Yamashita M, Zhang YE, Bertoglio J, Flatau G, Boquet P, Lemichez E.CNF1-induced ubiquitylation and proteasome destruction of activatedRhoA is impaired in Smurf1−/− cells. Mol Biol Cell 2006;17:2489–2497.

26. Harper JW, Schulman BA. Structural complexity in ubiquitin recogni-tion. Cell 2006;124:1133–1136.

27. Mukhopadhyay D, Riezman H. Proteasome-independent functions ofubiquitin in endocytosis and signaling. Science 2007;315:201–205.

28. Jura N, Scotto-Lavino E, Sobczyk A, Bar-Sagi D. Differential modifica-tion of Ras proteins by ubiquitination. Mol Cell 2006;21:679–687.

29. Nethe M, Anthony EC, Fernandez-Borja M, Dee R, Geerts D, Hens-bergen PJ, Deelder AM, Schmidt G, Hordijk PL. Focal-adhesion tar-geting links caveolin-1 to a Rac1-degradation pathway. J Cell Sci2010;123:1948–1958.

30. Blum G, Falbo V, Caprioli A, Hacker J. Gene clusters encoding thecytotoxic necrotizing factor type 1, Prs-fimbriae and alpha-hemolysinform the pathogenicity island II of the uropathogenic Escherichia colistrain J96. FEMS Microbiol Lett 1995;126:189–195.

31. Falkow S, Small P, Isberg R, Hayes SF, Corwin D. A molecularstrategy for the study of bacterial invasion. Rev Infect Dis 1987;9(Suppl. 5):S450–S455.

32. Eto DS, Jones TA, Sundsbak JL, Mulvey MA. Integrin-mediated hostcell invasion by type 1-piliated uropathogenic Escherichia coli. PLoSPathog 2007;3:e100.

33. Burns K, Clatworthy J, Martin L, Martinon F, Plumpton C, Maschera B,Lewis A, Ray K, Tschopp J, Volpe F. Tollip, a new component ofthe IL-1RI pathway, links IRAK to the IL-1 receptor. Nat Cell Biol2000;2:346–351.

34. Katoh Y, Shiba Y, Mitsuhashi H, Yanagida Y, Takatsu H, Nakayama K.Tollip and Tom1 form a complex and recruit ubiquitin-conjugatedproteins onto early endosomes. J Biol Chem 2004;279:24435–24443.

35. Yamakami M, Yoshimori T, Yokosawa H. Tom1, a VHS domain-containing protein, interacts with Tollip, ubiquitin, and clathrin. J BiolChem 2003;278:52865–52872.

36. Katoh Y, Imakagura H, Futatsumori M, Nakayama K. Recruitment ofclathrin onto endosomes by the Tom1-Tollip complex. BiochemBiophys Res Commun 2006;341:143–149.

37. Brissoni B, Agostini L, Kropf M, Martinon F, Swoboda V, Lippens S,Everett H, Aebi N, Janssens S, Meylan E, Felberbaum-Corti M,Hirling H, Gruenberg J, Tschopp J, Burns K. Intracellular traffick-ing of interleukin-1 receptor I requires Tollip. Curr Biol 2006;16:2265–2270.

38. Martinez JJ, Hultgren SJ. Requirement of Rho-family GTPases inthe invasion of type 1-piliated uropathogenic Escherichia coli. CellMicrobiol 2002;4:19–28.

39. Eto DS, Gordon HB, Dhakal BK, Jones TA, Mulvey MA. Clathrin, AP-2, and the NPXY-binding subset of alternate endocytic adaptorsfacilitate FimH-mediated bacterial invasion of host cells. Cell Microbiol2008;10:2553–2567.

40. Shen Y, Naujokas M, Park M, Ireton K. InIB-dependent internalizationof Listeria is mediated by the Met receptor tyrosine kinase. Cell2000;103:501–510.

41. Seveau S, Tham TN, Payrastre B, Hoppe AD, Swanson JA, Cossart P.A FRET analysis to unravel the role of cholesterol in Rac1 and PI 3-kinase activation in the InlB/Met signalling pathway. Cell Microbiol2007;9:790–803.

42. Veiga E, Cossart P. Listeria hijacks the clathrin-dependent endo-cytic machinery to invade mammalian cells. Nat Cell Biol 2005;7:894–900.

43. Petrelli A, Gilestro GF, Lanzardo S, Comoglio PM, Migone N, Gior-dano S. The endophilin-CIN85-Cbl complex mediates ligand-dependent downregulation of c-Met. Nature 2002;416:187–190.

44. Raiborg C, Wesche J, Malerod L, Stenmark H. Flat clathrin coats onendosomes mediate degradative protein sorting by scaffolding Hrs indynamic microdomains. J Cell Sci 2006;119:2414–2424.

45. Dramsi S, Biswas I, Maguin E, Braun L, Mastroeni P, Cossart P. Entryof Listeria monocytogenes into hepatocytes requires expression ofinIB, a surface protein of the internalin multigene family. Mol Microbiol1995;16:251–261.

46. Doye A, Boyer L, Mettouchi A, Lemichez E. Ubiquitin-mediatedproteasomal degradation of Rho proteins by the CNF1 toxin. MethodsEnzymol 2006;406:447–456.

47. Bouvard D, Aszodi A, Kostka G, Block MR, Albiges-Rizo C, Fassler R.Defective osteoblast function in ICAP-1-deficient mice. Development2007;134:2615–2625.

48. Millon-Fremillon A, Bouvard D, Grichine A, Manet-Dupe S, Block MR,Albiges-Rizo C. Cell adaptive response to extracellular matrix densityis controlled by ICAP-1-dependent beta1-integrin affinity. J Cell Biol2008;180:427–441.

49. Rudert F, Visser E, Gradl G, Grandison P, Shemshedini L, Wang Y,Grierson A, Watson J. pLEF, a novel vector for expression ofglutathione S-transferase fusion proteins in mammalian cells. Gene1996;169:281–282.

50. Lemonnier M, Ziegelin G, Reick T, Munoz Gomez A, Diaz-Orejas R,Lanka E. Bacteriophage P1 Ban protein is a hexameric DNA helicasethat interacts with and substitutes for Escherichia coli DnaB. NucleicAcids Res 2003;31:3918–3928.

12 Traffic 2011