The effect of Cif was first observed in 1997 by De Rycke et al. in HeLa cells transiently infected with enteropathogenic Escherichia coli (EPEC) strains isolated from weaning rabbits or human infants with diarrhea [1]. This so-called cytopathic effect (CPE) was first characterized by the progressive formation of actin stress fibers together with focal adhesions, and a dramatic cell enlargement (Figure 1). It was further shown that these large cells were irreversibly impaired for cell proliferation, with a complete lack of mitotic figures [2] (Figure 1). This effect is dependent on the locus of enterocyte effacement (LEE), the cluster of genes responsible for the attaching and effacing lesion, the hallmark of EPEC and enterohemorraghic E. coli (EHEC) infection [3,4]. The protein responsible for the CPE identified by Marches et al. was called Cif (cycle inhibiting factor). The gene coding for Cif is located outside the LEE, on a lambdoid prophage [5].

Two epidemiological studies showed that 60 to 70% of EPEC and EHEC strains are cif-positive [6,7]. However, in one of these studies, it was shown that two thirds of these cif-positive E. coli strains do not induce a typical CPE due to frameshift mutations or insertion of an IS element in the cif gene. Strikingly, cif is always found as a pseudogene in EHEC strains; this could be due to the small size of the strain collection, or could suggest an evolutionary counter-selection of functional Cif in Shiga-like producing EHEC strains [6]. Nonetheless, plasmid-mediated expression of wild type Cif in EHEC confers the CPE phenotype [5].

The E. coli Cif protein is 282 amino-acids long, showing no similarity with any protein except several homologues in the human pathogens Yersinia pseudotuberculosis (56% of identity) and beta-proteobacterium Burkholderia pseudomallei (26%). Cif homologues are also found in the entomopathogens Photorhabdus luminescens (23%) and Photorhabdus asymbiotica (26%). The Cif proteins are thus named Cif _Ec , Cif _Yp , Cif _Bp , Cif _Pl and Cif _Pa respectively. In all these bacteria the cif genes are located in prophage or prophage-like region (Escherichia and Photorhabdus species), bordered by two vestigial transposase genes (Burkholderia) and in a locus frequently rearranged nearby the yrs region homologous to a recombination target of bacteriophages (Yersinia) [8]. Thus Cif-encoding regions are found in DNA domains prone to rearrangement, suggesting that the cif genes were acquired by horizontal gene transfer. Consistent with this hypothesis, the Cif-producing EPEC strain E22 was shown to produce cif-carrying infectious phage particles supporting that the cif gene was disseminated by horizontal gene transfer among EPEC and enterohemorrhagic (EHEC) strains, as shown for other non LEE-encoded effectors [6,9,10].

Analysis of the Cif xenologues showed that they are all functional as they reproduced the typical Cif _Ec -associated CPE when transfected into HeLa cells, or lipofected following purification [8]. Importantly, Cif _Pl was shown to be expressed in vivo in an insect model, the natural host of Photorhabdus, and reproduced the CPE in the non-mammalian insect Sf9 cell line in vitro [11]. Similarly, Cif _Bp is also active in vitro when strains of E. coli (EPEC) or Burkholderia thailandensis are transformed with a plasmid coding for Cif _Bp and used to infect mammalian cell lines [8,12]. Such conservation of an effector function among very distant pathogens belonging to β-protobacteria and γ-protobacteria that infect invertebrate and mammal hosts is unique and remarkable. It also supports the role of Cifs as virulence factors and could reveal an original strategy of pathogenesis. However, Photorhabdus spp. are symbiotic of the nematode vector Heterorhabditis (that releases the bacteria into the insect) suggesting that Cif might also act as a fitness or symbiotic factor in the nematode host.

Initial experiments showed that the CPE triggered by cif-positive EPEC strains was dependent on the type III secretion system (T3SS) encoded by the LEE as mutants of the secretion and translocation apparatus (EscN and EspA, EspB, EspD, respectively) were unable to induce the CPE [1,2,5,13]. This T3SS injects an arsenal of effector proteins into the host cell that hijack cellular functions for the pathogen’s benefit [3,4,14]. Using a reporter system based on a translational fusion of the effector protein with β-lactamase that allows detection of translocation in living host cells, Charpentier and Oswald showed that an exchangeable N-terminal translocation signal comprising the first 16 amino acids of Cif _{E c} is necessary and sufficient to mediate its translocation by the EPEC T3SS [15]. Using a lipid-based cell delivery system bypassing T3SS requirement, purified Cif _Ec induced a CPE similar to that observed upon EPEC infection. This result indicated that Cif _Ec is necessary and sufficient and does not require other bacterial molecules to induce the CPE [16]. Once injected into the host cell cytoplasm by the T3SS, Cif _Ec is rapidly sorted to the peri-nuclear area and the nucleus, as observed by confocal imaging and cell fractionation [17]. As Cif _Ec lacks classical nuclear localization sequences, the mechanism of nuclear targeting of Cif remains to be elucidated.

While Cif _Bp has been shown to be injected by the T3SS of Escherichia and Burkholderia species and successfully reproduces the CPE [8,12], we and other have been unable to translocate Cif _Yp and Cif _Pl by the E. coli T3SS into the host cells. However, we have indirect evidence that they are bona fide type 3 effectors as both Yersinia and Photorabdhus strains possess T3SS. Moreover, none of the Cifs could induce the CPE without electroporation, transfection or lipofection suggesting that Cif activity depends on its injection into the host cell [8,11,16].

Despite a low level of sequence similarity between Cif homologs, the crystal structure determination of Cif _Ec , Cif _Pl and Cif _Bp showed that these proteins are structurally well-conserved (Figure 2) [18,19,20]. The overall structure of these Cif can be divided in two lobes with head and tail domains where the N-terminal part corresponds to the tail and contains the potential substrate binding site (see below) whereas the C-terminal part forms the head domain hosting the enzymatic site. The enzymatic site consists in a catalytic triad composed of residues Cys, His and Gln strictly conserved in all Cifs homologs (Figure 2).

The conserved catalytic triad is located in a negative charged cleft with a putative occluding loop near the active site that might control the substrate accessibility to the catalytic site [19]. The structural organization of this triad reveals that Cifs are divergent members of the superfamily of enzymes that comprises papain-like cysteine proteases, acetyl transferases, deamidases and transglutaminases. In particular, Cif shares homology with the avirulence effector AvrPphB from Pseudomonas syringae that belongs to the same superfamily and with the cysteine protease YopT from Yersinia spp. [19]. Mutation of one of the residues of the catalytic triad abolishes Cif _Pl , Cif _Bp , Cif _Ec and Cif _Yp activity, corroborating the implication of an enzymatic-dependent activity [8,16,19,20]. Further analysis of the tail domain showed that deletion of the N-terminal protruding α4 and α5 helix of Cif _Bp (Figure 2) and of the corresponding α4 domain of Cif _Ec suppressed the toxin activity suggesting that this domain could mediate substrate recognition [17,20].

Based on the structure and function conservation, Cif homologs constitute a family of cyclomodulins that likely target a molecule/pathway conserved in a wide range of eukaryotic cells from invertebrates to mammals [8,21].

Upon injection into host cells by the T3SS (or by transfection or lipofection of purified proteins) Cif proteins induce an irreversible cell cycle arrest with complete inhibition of mitotis entry (Figure 1: absence of mitotic figures in cells injected with Cif). This property makes Cifs as members of the cyclomodulin family defined as bacterial molecules that hijack host cell cycle functions [22,23].

The cell cycle arrest induced by Cif _Ec is associated to the inhibitory phosphorylation of the cyclin dependent kinase (CDK) 1-CyclinB complex whose activation is necessary for the cell cycle G₂/M transition [1,2,5]. Other cyclomodulins, such as CDT or Colibactin, also block the G₂/M transition through inhibition of CDK1 activity. However, in contrast to CDT and Colibactin, which have been shown to be bona fide genotoxins, Cif _Ec does not induce DNA damage nor activate the DNA damage response pathway (ATM/ATR, CHK1/2, CDC25 sequestration) that leads to the inhibitory phosphorylation of CDK1 [16]. Therefore, Cif _Ec induces the cell cycle arrest by activating a pathway independent from the canonical activation of the G₂/M checkpoint. Further studies, using synchronized cells, have shown that Cif _Ec also blocks S-phase entry. Depending on the stage of the cell cycle during infection with cif-positive EPEC, cells arrest either at the G₁/S or G₂/M transition. The analysis of host cell proteins regulating both S- and M- phases entries demonstrated a Cif-dependent accumulation of p21^Waf1/Cip1 and p27^kip2 (hereafter called p21 and p27) [24]. These proteins belong to the Cip/Kip family of CDK inhibitors (CKI). p21 is an inhibitor of the CDK1/CyclinB complex. This complex is the eukaryotic universal inducer of mitosis entry and p21 accumulation is known to inhibit G₂/M transition. Both p21 and p27 inhibit CDK2-CyclinE and A complexes whose activation are also required for G1/S and S-phase progression. p21 also binds to PCNA, and the latter is required for S-phase progression [25]. Therefore, Cif-dependent accumulation of CKI is consistent with the observed G₁/S and G₂/M cell cycle arrests (Figure 3). The inhibitory phosphorylation of CDK1 observed upon Cif _Ec injection could be explained by the disruption of the CDK1 positive feed-back loop leading to CDK1 dephosphorylation via CDC25. Further investigation revealed that CKI accumulation does not depend on transcriptional activation but relies on their stabilization in infected cells [24]. Since p21 and p27 are degraded by the ubiquitin-proteasome system, these results strongly suggested that Cif _Ec controls this proteolytic pathway. However, suppression of p21 and/or p27 accumulation did not alleviate Cif-induced inhibition of G₁/S or G₂/M transitions implying an alternative mechanism for blockage of the cell cycle [24]. Whether this arrest is associated to stress fibers induction was excluded since disruption of stress fiber formation in HeLa cells by C3 exoenzyme-mediated inhibition of RhoA, did not alleviate cell cycle inhibition [2]. Morevover, only HeLa and RK13 cells exhibit stress fibers following cif ⁺ EPEC infection suggesting that these two phenotypes are not related [16]. Nonetheless, these results indicated that accumulation of CKI participates in a multi-step process leading to the Cif-induced arrest of cell proliferation.

This cell cycle inhibition and CKI accumulation was observed with all Cif homologues [8] and in various cell lines derived from colonic epithelium (differentiated or not differentiated Caco-2 cells, DLD1, HCT116 wild-type, p21^−/− or p53^−/−) showing the independency of cell-cycle arrest from the p53 transcriptional program [16,24]. The Cif-induced cystostatic effect is also observed in non-transformed IEC-6 cells derived from rat intestinal crypt epithelium. However, the outcome of this arrest is cell type dependent as it persists for more than 3 days in HeLa cells without evident cell death, whereas it culminates in apoptosis 48 h after infection of IEC-6 cells [26]. Whether this Cif-induced cell death is a consequence of the cytostatic effect or relies on an independent pathway is still unclear, but it is tempting to propose that the functionality of p53 pathway, intact in non-transformed cells, might be decisive for the fate of Cif-arrested cells.

Three independent teams recently deciphered the molecular mode of action of Cif _Ec and Cif _Bp [12,17,27]. Despite few differences mainly related to the species origin of Cif and diverse approaches, these complementary studies showed a common mechanism of action between the two xenologs.

The role of NEDD8 on CRL activation offers to the bacteria injecting Cif into the host cells an “Achilles heel” to hijack various cellular signaling functions. Indeed, CRLs represent the largest subfamily of E3s and therefore play regulatory roles in numerous and diverse cellular functions. The diversity of substrate recognition modules associated to the different cullin subunits gives to the CRL the capacity to ubiquitylate and thus to control the stability of hundreds of proteins [35,36]. Cif capacity to induce stress fibers in certain eukaryotic cells could be related to Cif-stabilization of RhoA [12,17], which is targeted by CUL3-associated CRL [42]. Prevention of inappropriate replication depends on the degradation of the licensing factor Cdt1 during S-phase [43]. As CUL4-associated CRL targets Cdt1 for ubiquitylation-dependent degradation [44,45], inhibition of CRL by Cif provides a satisfactory explanation for occurrence of re-replication observed 3 days following infection as Cdt1 is stabilized in infected HeLa cells [2,16,17]. Moreover, Cif was shown to prevent the degradation of IκB, a central inhibitor of NF-κB pro-inflammatory response [12,17]. Down-regulation of the NF-κB pathway has been proposed to mediate the inflammatory tolerance of commensal bacteria in the mammalian intestinal epithelia [46,47]. In addition, stabilization of β-catenin by Cif might participate in dendritic cells tolerogenicity as it was recently shown that the WNT signaling pathway regulates immunosuppressive responses [48].

In conclusion, Cif might represent a fitness factor that facilitates bacterial evasion from the host immune response, via inhibition of inflammatory pathways in dendritic cells of the gut-associated lymphoid tissue. Cif-inhibition of the host cell cycle might slow down multiplication of intestinal progenitors to delay epithelial cells renewal, thus favoring gut colonization (Figure 5). We could also speculate about the role of Cif as a stabilizer of other type III co-injected effectors, to modulate their effect on host cells and boost the virulence of cif-expressing strains. Further studies of the impact of Cif in the pathogen-host interactions will undoubtedly contribute to our knowledge of bacterial pathogenic strategies. It also appears that the continuous “arm race” that characterizes host-pathogen relationship has lead the bacteria to develop sophisticated “weapons” (virulence factors) capable of controlling host’s functions for their own benefit. Therefore, much can be learned from eukaryotic functions using bacterial toxins such as Cif. While most pharmacological inhibitors or bacterial/viral proteins inhibit CRL by decreasing cullin neddylation [49,50,51,52], Cif is the first example of a bacterial effector that inhibits CRL functions by stabilizing cullin neddylation. Thus Cif represents a unique tool to study the regulation of the ubiquitin proteasome system. Lately, it was reported that inhibitors of NEDD8 represent a promising approach for cancer treatment [52] opening a potential development of innovative therapeutics using Cif-producing bacteria.