Cycle Inhibiting Factors (CIFs) Are a Growing Family of Functional Cyclomodulins Present in Invertebrate and Mammal Bacterial Pathogens

The cycle inhibiting factor (Cif) produced by enteropathogenic and enterohemorrhagic Escherichia coli was the first cyclomodulin to be identified that is injected into host cells via the type III secretion machinery. Cif provokes cytopathic effects characterized by G1 and G2 cell cycle arrests, accumulation of the cyclin-dependent kinase inhibitors (CKIs) p21waf1/cip1 and p27kip1 and formation of actin stress fibres. The X-ray crystal structure of Cif revealed it to be a divergent member of a superfamily of enzymes including cysteine proteases and acetyltransferases that share a conserved catalytic triad. Here we report the discovery and characterization of four Cif homologs encoded by different pathogenic or symbiotic bacteria isolated from vertebrates or invertebrates. Cif homologs from the enterobacteria Yersinia pseudotuberculosis, Photorhabdus luminescens, Photorhabdus asymbiotica and the β-proteobacterium Burkholderia pseudomallei all induce cytopathic effects identical to those observed with Cif from pathogenic E. coli. Although these Cif homologs are remarkably divergent in primary sequence, the catalytic triad is strictly conserved and was shown to be crucial for cell cycle arrest, cytoskeleton reorganization and CKIs accumulation. These results reveal that Cif proteins form a growing family of cyclomodulins in bacteria that interact with very distinct hosts including insects, nematodes and humans.


Introduction
Pathogenic bacteria have developed sophisticated arsenals of virulence factors that hijack eukaryotic host functions to their own benefit. One of the pathways targeted by several bacterial effectors is the eukaryotic cell cycle. These toxins, termed cyclomodulins, can promote cell proliferation or, conversely, inhibit cell growth and modulate differentiation by blocking cell cycle progression [1,2]. The Cycle Inhibiting Factor (Cif) is a cyclomodulin injected into eukaryotic cells by the type III secretion system (T3SS) of enteropathogenic and enterohemorrhagic Escherichia coli (EPEC and EHEC). Cif from pathogenic E. coli triggers an irreversible cytopathic effect characterized by cell cycle arrests at the G 2 /M and G 1 /S phase transitions and, at least in HeLa cells, reorganization of the actin network [3][4][5][6]. In contrast to other cyclomodulins such as the cytolethal distending toxin [7] or colibactin [8], Cif is not a genotoxin nor an activator of DNAdamage checkpoint pathways that lead to phosphorylation of cyclin-dependent kinase 1 and consequent G 2 -arrest [5]. Both G 1 and G 2 arrests induced by Cif are correlated with the accumulation of the cyclin-dependent kinase inhibitors (CKIs) p21 waf1/cip1 and p27 kip1 (hereafter referred as p21 and p27), which actively participate in the control of cell cycle progression. These accumulations result from inhibition of their proteasome-mediated degradation [6].
Cif is composed of a C-terminal active domain (residues 21-282) and an exchangeable N-terminal translocation signal encoded by the first ,20 amino acids [9]. The crystal structure of a truncated form of EPEC Cif (lacking the first 99 amino acids) was recently determined. The presence of a conserved catalytic triad comprising Cys109, His165 and Gln185, revealed that Cif is a divergent member of a superfamily of enzymes that includes cysteine proteases, acetyltranferases and transglutaminases [10]. The three amino acids that comprise the triad are essential for Cif's ability to induce cytopathic effects in eukaryotic cells as mutation of these residues leads to loss of function [10].
In EPEC and EHEC, Cif is not encoded within the locus of enterocyte effacement (LEE), which includes T3SS machinery genes and other effectors, but by a temperate lambdoid phage [11]. The cif gene has been widely disseminated by phage conversion within the natural population of E. coli, but positively selected within LEE-encoding strains [11]. Since Cif targets the cell cycle, a fundamental process conserved in all eukaryotic cells, it is reasonable to speculate that Cif homologs contribute to the pathogenicity of other bacterial species.
In the present study, four homologs of Cif have been identified and characterized in pathogenic or symbiotic bacteria: Burkholderia pseudomallei, Yersinia pseudotuberculosis, Photorhabdus luminescens and Photorhabdus asymbiotica. The four Cif homologs are functional and induce cell cycle arrest, p21 and p27 accumulation and actin cytoskeleton rearrangement in HeLa cells in an identical manner to EPEC Cif. The catalytic triad identified in the EPEC Cif crystal structure is strictly conserved in the homologs (at the sequence level) and is involved in their cytopathic activity since mutation of the critical cysteine residue leads to loss of function. Therefore, Cif proteins form a conserved family of cyclomodulins present in both symbionts and pathogens of vertebrate and invertebrate hosts.

Results
Genes encoding Cif-like proteins are present in the genomes of Yersinia, Burkholderia and Photorhabdus species The cyclomodulin Cif was initially identified and characterized in pathogenic E. coli (Cif Ec ) [3]. Sequence database searches using BLAST [12] revealed that Cif Ec shares similarity with hypothetical proteins encoded by the genome of four other bacterial species (Table 1). Cif Ec exhibited a high degree of similarity with Ypk1971 (56% identity), a protein encoded by the human pathogen Yersinia pseudotuberculosis strain YPIII [13]. Y. pseudotuberculosis infection in humans causes gastroenteritis characterized by a self-limited mesenteric lymphadenitis that mimics appendicitis. Cif Ec was also similar to a protein encoded by the open reading frame bpss1385 from B. pseudomallei strain K96243 (26% identity). B. pseudomallei is the causative agent of melioidosis, an important cause of sepsis in east Asia and northern Australia [14]. Putative Cif homologs were also detected in two Photorhabdus species: P. luminescens, a symbiotic bacterium for the soil nematode Heterorhabditis and a pathogen for a broad range of insects [15] and P. asymbiotica, an emerging human pathogen [16]. The proteins encoded by plu2515 (P. luminescens) and pha4011 (P. asymbiotica) share 23 and 26% of identity with Cif Ec respectively. Interestingly, these four bacterial species in which cif Ec -like genes were found all possess at least one T3SS. Proteins Ypk1971, Bpss1385, Plu2515 and Pha4011 are hereafter referred to as Cif Yp , Cif Bp , Cif Pl and Cif Pa respectively. Finally, it should also be noted that a truncated putative protein (GOS5485515) obtained from the translation of a DNA fragment isolated from surface water marine samples [17,18] also shows sequence similarity to Cif Ec .
The degree of conservation and the phylogenetic relationship between Cif homologs were analysed by constructing a multiple sequence alignment and a phylogenetic tree using the Neighbour-Joining method (Fig. 1). Cif Pl and Cif Pa clustered together and were separated from a second group consisting of Cif Ec and Cif Yp . Cif Bp was the most divergent protein, located to a branch between the two groups. This phylogenetic tree matches the accepted bacterial taxonomy since B. pseudomallei belongs to the bproteobacteria class whereas all others are enterobacteriacae belonging to the c-proteobacteria class.
Genes encoding Cif-like proteins are found in highly rearranged DNA regions In E. coli, the cif gene is located on an inducible lambdoid prophage spread widely amongst EPEC and EHEC strains (   [11]. In Photorhabdus strains, cif Pl and cif Pa are located downstream of a region displaying a high degree of similarity to a prophage described in Serratia entomophila (Fig. 2) [19]. This prophage is integrated 5 to 6 times in the genome of both Photorhabdus species [20] and encodes genes for several putative virulence factors, notably a putative T3SS effector protein homologous to YopT from Yersiniae. This phage has no homology with the lambdoid prophage found in E. coli isolates but displays some similarity to bacteriocins and R-type pyocins [21]. In B. pseudomallei strain K96243 (Fig. 2), cif Bp is located between two vestigial transposase genes on chromosome II near the hrp cluster, which codes for one of the three T3SS present in B. pseudomallei [22]. Comparison of sequenced genomes from different B. pseudomallei strains reveals that the organization of this locus is highly variable. B. pseudomallei strains S13 and 9 contain additional genes, encoding putative transposases, which are inserted near cif Bp (Fig. 2). In B. pseudomallei strain 1106a, this region is deleted and cif Bp is absent. These data suggest that DNA transposition events could have lead to the heterogeneous distribution of the cif Ec -like gene in B. pseudomallei strains. Among the sequenced strains of Y. pseudotuberculosis, only the strain YPIII possesses a gene with similarity to cif Ec . Comparison of the genetic environment between YPIII and other Y. pseudotuberculosis strains revealed that cif Yp is positioned within a chromosomal locus previously described as the insertion site of ypm, a gene coding for a superantigenic toxin in strain AH [23]. Both ypm and cif Yp are located downstream of a 26-bp sequence called yrs which is homologous to dif, a site-specific recombination target used by filamentous bacteriophages for host chromosome integration. Deletions in the yrs locus occur at a higher frequency compared to others regions within the chromosome [23]. Genetic instability at this locus could explain the heterogeneous distribution of both cif Yp and ypm genes in the Y. pseudotuberculosis population.
In conclusion, each cif Ec -like gene is associated either with mobile genetic elements, such as phages, or is located in region of the genome prone to rearrangements, suggesting acquisition of cif by horizontal gene transfer in all these bacterial species. This is in agreement with the observation that each cif Ec -like gene has a different GC content compared to that of their cognate host genome (Table 1).

Cif proteins share common conserved motifs, including a catalytic triad
Database searches with the sequence of Cif Ec or the Cif homologs (including the truncated protein from the marine metagenome) reveals no significant matches to well-characterized proteins or motifs. However, alignment of Cif Ec and the homolog sequences reveals several well conserved positions or regions, most of which are located to the C-terminal two thirds of the proteins (Fig. 3A). The lack of sequence conservation at the far N-terminus (top panel in Fig. 3A) is consistent with the putative function of these regions as a translocation signal for the T3SS, which may have different requirements in the different parent organisms. It is now well established that regions responsible for secretion/ translocation and chaperone binding in T3SS effectors are located to the N-terminus, but often share no sequence similarity, even for effector proteins translocated by the same T3SS [24].
Sequence alignments of the Cif protein family identified a conserved cysteine residue. Conservation of cysteine residues often implies biological significance. The recent crystal structure of Cif Ec revealed that C109 forms part of a catalytic triad. Further, the other two residues that form this structural motif (H165 and Q185) are also fully conserved in all Cif homologs ( Fig 3A). In addition to the catalytic triad, several other residues are also retained. The position of these residues, when mapped onto the structure of Cif Ec (possible for all but three of the conserved residues which are not present in the construct crystallised) reveals they cluster in three regions (Fig. 3B). The first of these clusters surrounds the active site and these residues are likely essential for retaining the catalytic triad in a suitable conformation to enable catalysis, or are directly involved in substrate binding (residues with carbon atoms coloured yellow in Fig. 3B). The second cluster (residues with carbon atoms coloured green in Fig. 3B) is somewhat distant from the active site and it seems likely that this region is important for maintaining structural integrity, with a potentially important hydrogen bond identified between the O d1 atom of Asp170 (located at the end of b-strand 2) and the OH atom of Tyr265 (located at the end of bstrand 4). The importance of the third cluster (of three residues) is less apparent (residues with carbon atoms coloured purple in Fig. 3B). This region may be involved in binding substrate molecules, or it may interact with the N-terminal region of Cif Ec not present in the crystallised protein. In conclusion, in silico analyses of Cif homologs are consistent with a conserved function for these proteins, akin to Cif Ec .

Cif Bp is injected by the EPEC T3SS and induces cell cycle arrest and stress fibre formation in HeLa cells
An EPEC strain deleted for its chromosomal cif Ec gene (E22Dcif) has previously been described [3]. To test whether the Cif-like proteins are functional homologs of Cif Ec , the E22Dcif strain was complemented with a plasmid encoding each of the cif Ec -like genes, and these bacteria were used to infect cultured HeLa cells. Since the whole amino acid sequence of the putative protein derived from the marine metagenome is not available, this truncated protein was not included in these assays. Before phenotypic characterization of cells infected with EPEC producing the Cif homologs, the translocation efficiency of the proteins by the EPEC T3SS was monitored using the TEM/CCF2 assay [9].
As expected, the Cif Ec -TEM fusion protein was properly translocated, as demonstrated by detection of intracellular blactamase activity (Fig. 4A). TEM activity was also detected in cells infected with E22Dcif producing Cif Bp -TEM, but levels of blactamase activity for Cif Pl -TEM, Cif Pa -TEM and Cif Yp -TEM were similar to the basal level detected with the negative control (TEM alone, Fig. 4A). Since TEM fusion proteins were produced to similar levels in the bacteria (Fig. 4A), absence of intracellular TEM activity likely results from inefficient recognition and/or injection of Cif Pl -TEM, Cif Pa -TEM and Cif Yp -TEM by the T3SS of EPEC. The lower translocation level of Cif Bp -TEM compared to Cif Ec -TEM probably also reflects a poor recognition of the secretion/translocation signal (STS) of Cif Bp by the T3SS from EPEC. Indeed, when this fusion protein was expressed in an escN mutant (T3SS ATPase defective mutant), b-lactamase activity was no longer detected in infected cells, confirming that translocation of Cif Bp -TEM by E22 strain is T3SS-dependent (data not shown).
Since Cif Bp can be injected by the T3SS of E22, the capacity of the protein to induce cytopathic phenotypes on HeLa cells was analysed using the infection model. In contrast to cells infected with E22Dcif carrying an empty vector, cells infected with E22Dcif producing Cif Bp developed cell distension and actin stress fibres indistinguishable from those induced by a Cif Ec -expressing strain (Fig. 4B). Cif Bp also blocked cell cycle progression, as demonstrated by the accumulation of G 2 arrested cells containing 4N DNA content (Fig. 4B). These phenotypes were not induced when Cif Bp was expressed in an escN mutant (data not shown). These data clearly demonstrate that Cif Bp is a functional homolog of Cif Ec .

Cif Pl , Cif Pa and Cif Yp are functional homologs of Cif Ec
As the EPEC T3SS was not able to translocate Cif Pl , Cif Pa and Cif Yp into infected cells, the cytopathic activity of these Cif Ec -like proteins was investigated using purified recombinant samples combined with a lipid mediated delivery system (BioPORTER) as previously described [5]. The effects of Cif Bp delivered with this system were also investigated. Cif Bp , Cif Pl and Cif Pa were all readily overexpressed and purified in a soluble form (see Materials and Methods). However, despite many efforts, it was not possible to obtain a purified soluble form of Cif Yp at levels necessary for activity assays using the BioPORTER delivery system. As previously reported [5], treatment of HeLa cells with BioPORTER mixed with purified Cif Ec leads to cell enlargement and formation of actin stress fibres (Fig. 5), identical to the phenotype observed with the infection model. However, as protein delivery with BioPORTER is not as efficient as bacterial infection [5], only ,50% of the treated cells exhibit morphological alterations (not shown). Studies of cell cycle patterns were therefore realized using G 1 /S synchronized cells to improve visualization of G2 arrest. In contrast to cells incubated with the lipid delivery agent mixed with PBS alone, cells treated with BioPORTER+Cif Ec accumulated in G2 phase (38% of Cif Ectreated cells contained 4N DNA-content against 10% for PBStreated cells). Lipofection of purified Cif Bp into HeLa cells led to actin stress fibres and cell accumulation in G 2 phase (27%) (Fig. 5), confirming the functionality of Cif Bp observed with the infection assays. Introduction of purified Cif Pl or Cif Pa into HeLa cells with BioPORTER also led to cell enlargement, cytoskeleton alteration and accumulation of cells with 4N DNA content (40 and 25% for Cif Pl and Cif Pa respectively versus 10% for PBS treated cells, see Fig. 5). Therefore, Cif Pl and Cif Pa are also functional homologs of Cif Ec . As these phenotypes are observed with purified proteins, the results demonstrate that the proteins alone are sufficient to induce the Cif-associated cytopathic effects.
As it was not possible to introduce Cif Yp into cells using either the infection or BioPORTER treatments, the function of this  GFP alone was also transfected as a negative control. Among the GFP positive population, 96% of cells expressing GFP-Cif Ec had a 2N DNA content whereas the 2N population of cells expressing GFP alone was only 82% (Fig. 6). Consistent with previous studies demonstrating that Cif could also induce G 1 /S arrest [6], this result demonstrates that the cell cycle of transfected cells expressing GFP-Cif Ec was blocked in G 1 (2N DNA content). As expected, the cell cycle arrest was not observed when the critical cysteine residue from the catalytic triad of Cif Ec was substituted (Fig. 6). Expression of GFP-Cif Yp in HeLa cells also led to accumulation of GFP-positive cells with 2N DNA content (96% against 82% for cells expressing GFP alone), demonstrating that Cif Yp induced a cell cycle arrest in G 1 phase similarly to Cif Ec (Fig. 6). This result indicates that Cif from Y. pseudotuberculosis is a functional homolog of Cif Ec .

The conserved catalytic triad is critical for the activity of Cif homologs
Most of the conserved residues in Cif proteins are clustered in discrete regions (Fig. 3). The cysteine, histidine and glutamine residues forming the catalytic triad in Cif Ec were shown to be critical for activity [10]. To determine whether an equivalent functional catalytic site exists in the Cif homologs, the conserved cysteines in Cif Bp , Cif Pl and Cif Pa (C90, C128 and C123 respectively) were substituted with a serine residue, and the corresponding proteins were purified prior to delivery into HeLa cells using the BioPORTER system. In contrast to the wild-type proteins, the cysteine variants did not induce cell enlargement and stress fibre formation (Fig. 7A). Further, analysis of DNA content revealed that accumulation of G2-arrested cells did not occur when cells were treated with the cysteine variants (Fig. 7B). Expression of the cysteine variant from Cif Yp by transfection in HeLa cells also revealed that the cell cycle was not arrested in contrast to cells producing the wild-type protein (Fig. 6). These results demonstrate that the conserved cysteine residue is critical for Cif activity. Also, as the histidine and glutamine residues that complete the triad are also conserved in the sequences of the Cif homologs, this suggests that catalytic triads also exist in Cif Bp , Cif Pl , Cif Pa and Cif Yp .

Cif homologs induce p21 and p27 accumulation in cells
It has recently been shown that the cytopathic activity of Cif Ec is correlated to the accumulation of CKIs p21 and p27, two important regulators of cell cycle progression [6]. Since all Cif homologs appear to share the same catalytic triad and induce identical cytopathic phenotypes in HeLa cells, we wonder if they could hijack the same signaling pathways, despite the fact that two of these proteins are produced by bacteria colonizing insects and nematodes. Western-blot analysis of HeLa cells treated with purified Cif homologs indicated that levels of p21 and p27 increase in the presence of wild-type Cif Bp , Cif Pl and Cif Pa (Fig. 8). An intact catalytic triad is integral to this accumulation as CKIs levels were not affected when cells were treated with the cysteine variants (Fig. 8). This accumulation of p21 and p27 suggests that the molecular mechanisms involved in Cif cytotoxicity on HeLa cells are identical for Cif Ec and the Cif homologs.

Discussion
Cif Ec proteins belong to a family of cyclomodulins that inhibit host cell proliferation by inducing G 1 /S and G 2 /M phase transition blocks [3,6]. In this study, functional homologs of Cif from pathogenic E. coli have been identified in Y. pseudotuberculosis, B. pseudomallei, P. luminescens and P. asymbiotica. These homologs possess the same capacity as Cif Ec to induce cell cycle arrest, actin stress fibre formation and p21 and p27 CKIs accumulation when introduced into HeLa cells, suggesting they target the same substrates. Each of the Cif homologs possesses a predicted catalytic triad as identified in the crystal structure of Cif Ec . This triad is involved in the cytotoxic activity of each Cif homolog as substitution of the conserved cysteine residue in any of the proteins leads to inactivation.
In pathogenic E. coli, cif is located on an inducible lambdoïd prophage that has spread widely within the natural population of E. coli [11]. Analysis of the genetic locus containing cif in other bacteria reveals that cif genes from Photorhabdus species are also located downstream of a prophage, while cif from B. pseudomallei and Y. pseudotuberculosis are inserted in highly rearranged DNA regions leading to heterogeneous distribution within bacterial populations. In addition, GC content of cif genes shows substantial deviation from the general pattern within their respective genome. In light of these data, cif genes are proposed to have been acquired by horizontal gene transfer and could be defined as xenologs according to the nomenclature proposed by Koonin et al. [25]. The phylogenetic relationship between the different xenologs is in agreement with the bacterial taxonomy since Cif from B. pseudomallei, the only b-proteobacteria, is the most divergent protein. This indicates that protein sequence variation is, to some extent, a consequence of speciation events and suggests that cif genes were probably acquired early during bacterial evolution. All Cif-producing bacteria encode at least one T3SS that could inject the effector into host cells during infection. It is interesting to speculate that tight association between horizontally acquired effectors and the T3SS machinery in bacteria is a consequence of selective pressure since advantages conferred by effector acquisition will occur only if the recipient organism produces the secretion/translocation machinery. Such an association has already been described in E. coli and Salmonella enterica serovar Typhimurium where phage-encoded T3SS effectors were associated with T3SS producing isolates [11,26].
Bacteria habouring the cif gene spend part of their life cycle in association with eukaryotic organisms. While E. coli, B. pseudomallei, Y. pseudotuberculosis and P. asymbiotica are mammalians pathogens [13,14,16,27], both Photorhabdus species are pathogenic for insects and symbiotic to nematodes [15,28,29]. Like the Cif proteins, other families of T3SS effectors are produced by bacterial pathogens that target distinct hosts. For example, a number of proteins belonging to the YopT cysteine protease family have been described in mammalian, insect and plant pathogens [30]. Although the overall sequence identity at the amino acid level is not extensive, every member of the YopT family shows several invariant residues including a cysteine, a histidine and an aspartate that form a putative catalytic triad. Representatives from the YopT-like family interfere with diverse host immune responses and display protease activity dependent on an intact catalytic triad. YopT, the archetypal member of this family, is the most potent inhibitor of phagocytosis produced by Yersinia [31] and cleaves prenylated GTPases of the Rho family in host cells [32]. Similar to YopT, LopT from P. luminescens is able to release RhoA from human and insect cell membranes [33]. AvrPphB is an avirulence protein of the YopT-like family from the plant pathogen Pseudomonas syringae that triggers a disease-resistance response in a number of host plants, including Arabidopsis [30]. Searches of the Protein Data Bank with the structure of Cif Ec reveal close structural homology to AvrPphB. Although the residues that form the catalytic triad in each protein are different (C/H/D for AvrPphB and C/H/Q for Cif), the overall folds and residues comprising the catalytic triads superimpose well [10].
YopJ-like proteins form a second family of T3SS effectors produced by different animal and plant pathogens that also possess conserved residues forming a predicted catalytic triad, which is required for protease activity [34]. YopJ, the archetypal member of this family, is an essential virulence factor produced by Yersinia which blocks MAPK and NFkB pathways resulting in inhibition of host immune responses [35,36]. In contrast to members of the Cif protein family, that induce similar phenotypes in HeLa cells, proteins belonging to the YopT or the YopJ family appear to generate different responses in eukaryotic cells. For example, AvrA, a Salmonella YopJ-like T3SS effector (56% identity with YopJ), does not induce the same host responses observed for YopJ [37]. Further studies are required to determine whether the conserved cytopathic effects induced by Cif proteins in HeLa cells, notably cell cycle arrest, also occur in their respective host cells (gut enterocytes for intestinal pathogens, insect cells for Photorhabdus species, etc).
Interestingly, the plant symbiont Rhizobium sp. strain NGR234 produces NopJ, a YopJ-like protein and NopT, a T3SS effector belonging to YopT-like family [38][39][40]. Both cysteine proteases were shown to be involved in the host-specific nodulation response of legumes [38,40]. Symbiotic bacteria deploy somewhat similar strategies for colonizing host cells as those used by mammalian pathogens. The T3SS is, for example, required for host cell invasion by a variety of symbiotic bacteria [41][42][43]. As the Cif-producing Photorhabdus species are not only insect pathogens but also nematode symbionts, it is tempting to speculate that Cif may also contribute to the symbiotic process. Further, CdtB, the active monomer of the cyclomodulin CDT, is expressed in Hamiltonella defensa, a symbiont of pea aphids [44]. It maybe that symbionts use cyclomodulins like CDT and Cif to modulate, rather than globally deregulate, host signaling pathways resulting in initiation of symbiosis. Future studies will rely on further molecular (in vitro) analysis and in vivo models to achieve a full understanding of the roles of Cif in microbial pathogenesis, commensalism and symbiosis.

Cell line, bacterial strains and plasmids
HeLa cells (ATCC CCL-2) were cultured in Dulbecco's modified Eagle medium (DMEM; Invitrogen) supplemented with 10% foetal calf serum (FCS; Eurobio) and 80 mg ml 21 gentamicin at 37uC in a 5% CO 2 atmosphere. For synchronization in G 1 /S phase, HeLa cells were treated with 2 mM thymidine (Sigma) for 18 h, washed 3 times with Hank's balanced salt solution (HBSS; Invitrogen), incubated in normal medium for 9 h and treated again with 2 mM thymidine for 16 h. Bacterial strains and plasmids used in this study are listed in Table 2. Bacteria were cultured in Luria-Bertani (LB) broth or in interaction medium (DMEM with 25 mM Hepes and 5% FCS). Antibiotics were used at the following final concentrations: chloramphenicol 20 mg ml 21 and kanamycin 25 mg ml 21 .

Construction of plasmids expressing Cif and Cif-TEM proteins
To construct plasmids suitable for expressing Cif homologs from B. pseudomallei, Y. pseudotuberculosis, P. luminescens and P. asymbiotica in EPEC, cif Bp , cif Yp , cif Pl and cif Pa genes were amplified from respective genomic DNA with primers adding a XbaI restriction site at the start codon and a BamHI (or XhoI for cif Bp ) restriction site after the stop codon. PCR products were digested and ligated into the corresponding sites of the pBRSK vector [45]. The resulting plasmids pEL1, pEL2, pEL4 and pEL5 contain, respectively, cif Bp , cif Pl , cif Pa and cif Yp genes under the control of a Plac promoter.
To create the pKTEM vector necessary to construct TEM fusions, the multiple cloning site and blaM gene (encoding the blactamase TEM-1) were amplified by PCR from pCX340 ( [9]) with primers containing XhoI and XbaI restriction sites. The PCR fragment was digested and cloned into the corresponding sites of pBBR1MCS-2 ( [46]). Plasmids encoding translational fusion between the different Cif proteins and the b-lactamase TEM-1 were obtained by cloning cif genes into the pKTEM vector. Briefly, cif Bp , cif Pl , cif Ec , cif Pa and cif Yp genes were amplified from pEL1, pEL2, pEL3, pEL4 and pEL5 respectively using primers with NdeI-EcoRI restriction sites (or XhoI-HindIII for pEL4), digested and cloned into the corresponding sites of pKTEM. The resulting plasmids pGJ719, pGJ720, pGJ626, pGJ721 and pGJ803 encode Cif Bp -TEM, Cif Pl -TEM, Cif Ec -TEM, Cif Pa -TEM and Cif Yp -TEM fusion proteins respectively. All the constructs were verified by DNA sequencing (Cogenics, France).

Purification of Cif Bp , Cif Pl , Cif Pa and Cif Ec proteins
For production of recombinant protein, the genes encoding cif Bp , cif Pl , cif Pa and cif Yp were cloned into the pET28a vector (Novagen). The resulting constructs encoded proteins with an Nterminal 6xHis tag. Plasmids were named pMB1, pCC1, pCC3 and pGJ803 respectively. The plasmid for expression of 6xHis-Cif Ec has been described elsewhere [5]. Mutations of the conserved cysteine residues were obtained by inverse PCR using pET28 based constructs as a template and oligonucleotides containing specific base changes. All the constructs were verified by DNA sequencing (Cogenics, France). After transformation into the E. coli BL21-CodonPlusH (DE3)-RIPL strain (Stratagene), bacteria were grown in LB to an OD 600 nm of ,0.6 then induced with 0.5 mM IPTG for 3 h at 37uC. Purification of native proteins was achieved by Ni-NTA chromatography as recommended by the manufacturer (Qiagen) and, if necessary, gel filtration. Samples were then dialysed against PBS, aliquoted and stored at 280uC.

Construction of plasmids expressing GFP-Cif fusion proteins and transfection assays
Plasmids encoding translational fusions between the fluorescent reporter protein GFP and Cif Ec or Cif Yp were obtained by cloning cif genes (encoding the wild-type or the cysteine variant forms) into the pTagGFP-C vector (Evrogen). The resulting plasmids were verified by DNA sequencing (Cogenics, France). Transfections were performed in 6-well plates with FuGENE (Roche) according to the manufacturer's instructions. Two days after transfection, HeLa cells were exposed to trypsin, washed with ice-cold PBS, fixed for 3 h at 4uC in PBS with 1% formaldehyde, permeabilized overnight at 4uC in PBS ethanol 70% and stained with propidium iodide for 30 min at 37uC. Cells were analysed using a FACScalibur flow cytometer (Becton Dickinson) and data from at least 20 000 cells were analysed using FloJo software v8.5 (Tree Star).

Infection, translocation and BioPORTER assays
For infection experiments, bacterial strains were cultured overnight in LB broth then diluted 1:100 in interaction medium for 3 h at 37uC in a 5% CO 2 atmosphere. HeLa cells were washed with HBSS and infected for the indicated time in interaction medium with a multiplicity of infection (MOI) of 100 bacteria per cell (except as otherwise noted). After the infection, cells were washed with HBSS then cultivated for the indicated times in DMEM medium supplemented with 10% FCS and 200 mg ml 21 gentamicin.
Translocation levels of Cif-TEM fusion proteins were determined using CCF2/AM (Invitrogen) as a substrate for intracellular TEM enzyme as described previously [9]. Briefly, HeLa cells seeded in black 96-well plates were loaded for 1 h at 37uC with 1.7 mM CCF2/ AM diluted in DMEM with 2 mM probenecid and then infected for 2 and a half h with bacteria expressing TEM fusion proteins. Fluorescence was quantified in a microplate reader (TECAN Infinite M200) with excitation set at 410 nm (9 nm bandwidth) and emission at 450 nm for blue fluorescence and 520 nm for green fluorescence (20 nm bandwidth). Translocation was expressed as the emission ratio at 450/520 nm. To determine the expression level of TEM fusion proteins in bacteria, bacterial cultures with identical OD 600 nm were pelleted, resuspended in SDS-PAGE sample buffer, boiled for 5 min and subjected to western blot analysis with anti-TEM-1 antibodies (QED Biosciences).
For BioPORTER assays, 80 ml of purified proteins (250 mg ml 21 ) (or PBS as a negative control) were added to one BioPORTER tube (Genlantis) and resuspended in 420 or 920 ml of DMEM. The samples were added to the cells grown in BD Falcon culture slides or in 6-well plates and incubated for 4 h. BioPORTER mixes were replaced by fresh complete medium and the cells were incubated for 16-72 h.

Actin stress fibre and cell cycle analyses
For cell morphology and actin cytoskeleton visualization, cells were fixed for 15 min in PBS supplemented with 4% formalde-hyde, permeabilized with 0.1% Triton X-100 and stained with rhodamine-phalloidin (Molecular Probes) and DAPI (Sigma). Images were acquired with a DMRB fluorescence microscope equipped with a DFC300FX digital camera (Leica). Cell cycle distribution analyses were performed as previously described [47]. Briefly, cells were grown on 6-well plates, synchronized in G 1 /S phase and infected or treated with BioPORTER. The cells were exposed to trypsin, washed, fixed with ethanol, stained with propidium iodide and analyzed by flow cytometry. Percentages of G 2 populations were calculated using the Dean-Jett-Fox model from the FlowJo software (Tree Star).

Western Blot analyses
For Western blot analyses, 6610 5 cells were lysed in 80 ml of SDS-PAGE sample buffer, sonicated for 2 s to shear DNA and then boiled for 5 min. Protein samples were resolved on 4-12% NuPage gradient gels (Invitrogen) and blotted on PVDF membranes. Membranes were blocked in TBST (10 mM Tris pH 7.8, 150 mM NaCl, 0.1% Tween20) 5% non-fat dry milk, then probed with primary antibody (0.5 mg ml 21 ) in TBST 5% non-fat dry milk. Primary antibodies were: anti-actin (ICN), anti- DH5a Invitrogen BL21-CodonPlus (DE3) Stratagene p21 and anti-p27 (Santa Cruz Biotechnology). Bound antibodies were visualized with horseradish peroxidase-conjugated secondary antibody. Acquisitions were performed with a Molecular Imager ChemiDoc XRS system (Bio-Rad). Protein levels were quantified with Quantity One Software (Bio-Rad) and normalized with actin level.

Bioinformatic analyses
The search for proteins sharing similarity with Cif Ec was performed using BLAST on the NCBI server and MaGe system on the Genoscope server for private access to the genome of P. asymbiotica (Sanger Institute). Genetic organization of the cif-like genes loci were determined using Artemis software from the Sanger Institute and MaGe system. Multiple alignments of Cif sequences were generated with ClustalW and edited using GeneDoc software. Based on this alignment, the unrooted phylogentic tree was obtained using Phylip's Draw software.