The Intracellular Fate of Salmonella Depends on the Recruitment of Kinesin

See allHide authors and affiliations

Science  20 May 2005:
Vol. 308, Issue 5725, pp. 1174-1178
DOI: 10.1126/science.1110225


Salmonella enterica causes a variety of diseases, including gastroenteritis and typhoid fever. The success of this pathogen depends on its capacity to proliferate within host cells in a membrane-bound compartment. We found that the Salmonella-containing vacuole recruited the plus-end–directed motor kinesin. Bacterial effector proteins translocated into the host cell by a type III secretion system antagonistically regulated this event. Among these effectors, SifA targeted SKIP, a host protein that down-regulated the recruitment of kinesin on the bacterial vacuole and, in turn, controlled vacuolar membrane dynamics.

Intracellular replication of Salmonella enterica serovar Typhimurium (S. typhimurium) takes place in a membrane-bound compartment, the Salmonella-containing vacuole (SCV) and requires the type III secretion system (TTSS) encoded by the Salmonella pathogenicity island 2 (SPI-2) (1). The SPI-2 TTSS is activated intracellularly and mediates the translocation of bacterial proteins into host cells. SifA is encoded outside of SPI-2 but is tightly controlled by the SPI-2 regulatory system SsrAB, and SifA is delivered into host cells via the SPI-2 TTSS (2, 3). SifA is required for the formation of Salmonella-induced filaments (Sifs) (4, 5) and is necessary to maintain the integrity of the SCV (6). A sifA mutant is strongly attenuated in virulence in mice, indicating that the control of vacuolar membrane dynamics is a key aspect of the Salmonella virulence process (7).

The molecular mechanisms underlying the formation of Sifs and the maintenance of SCV integrity are still unresolved, and host proteins with which SifA interacts are unknown. We performed a yeast two-hybrid screen of a human mammary gland cDNA library using SifAΔ6 (8) as bait. We isolated three independent cDNA clones, all encoding the C-terminal region of a protein identical to KIAA 0842 (9) (Fig. 1A). Kiaa 0842 encodes a 113-kD ubiquitously expressed polypeptide of unknown function that we have named SKIP (for SifA and kinesin–interacting protein). SKIP contains an N-terminal RPIP8, UNC-14, and NESCA (RUN) motif (10) and a C-terminal pleckstrin homology (PH) motif (11) (Fig. 1A).

Fig. 1.

SifA interacts with SKIP. (A) Schematic representation of native SKIP and derived polypeptides. Arrows indicate the regions corresponding to the cDNAs cloned in a two-hybrid screen. The first and last amino acid residue numbers of truncated polypeptides are shown in parentheses. (B) SifA recruits myc-SKIP on SCVs and Sifs. HeLa cells expressing moderate levels of myc-SKIP (5 to 10 times greater than the endogenous level) were infected for 14 hours with either a S. typhimurium strain secreting SifA-2HA (sifApsifA-2HA) or the sifA mutant strain, and were immunostained for HA (green), myc (red), and lipopolysaccharide (LPS) (blue). Scale bar, 5 μm. (C) SKIP immunoprecipitates SifA but not SseJ. HeLa cells were infected with Salmonella strains secreting SifA-2HA or SseJ-2HA for 14 hours. Cell lysates were incubated with a rabbit antibody to SKIP or the corresponding preimmune serum (P.I.), precipitated using Protein-A beads (co-IP) and immunoblotted with an antibody to HA. The background at the top of the gels corresponds to rabbit immunoglobulin G heavy chains. (D) Pull-down experiments with various myc-tagged SPI-2 TTSS effector proteins with GST.SKIP-derived polypeptides. Extracts of HeLa cells expressing SPI-2 effector proteins were incubated with either GST or GST.SKIP-derived polypeptides immobilized on beads. Bound proteins were immunoblotted with an antibody to myc. GST.SKIP(762-885) encoding the PH motif is sufficient to pull down SifA (left panel). GST.SKIP(518-1019) specifically pulls down SifA (right panel). The positions of molecular mass markers are indicated.

We analyzed whether SifA interacted with SKIP in cultured cells. In HeLa cells, SKIP was essentially a cytosolic protein, and it was partially recruited to membranes upon ectopic expression of SifA (fig. S1). In S. typhimurium 12023–infected cells, secreted double hemagglutinin (HA) epitope–tagged SifA (SifA-2HA) was present on SCVs and Sifs (3), and overexpressed myc-SKIP was recruited on these Salmonella-induced membranous structures (Fig. 1B). However, myc-SKIP remained evenly distributed in HeLa cells infected with a sifA mutant and could not be detected on bacterial vacuoles (Fig. 1B). The SifA-SKIP interaction was investigated by coimmunoprecipitation analysis of infected HeLa cell lysates. S. typhimurium strains secreting SifA-2HA or SseJ-2HA, another SPI-2 effector protein, were used. SifA was specifically detected in SKIP immunoprecipitates (Fig. 1C). Thus, SKIP was recruited to SCVs and Sifs in a SifA-dependent manner. We further analyzed the interaction between SifA and SKIP in vitro and showed that purified glutathione S-transferase (GST).SKIP, but not GST, pulled down SifA (Fig. 1D). Various fragments of SKIP were tested to localize the SifA-binding site. The PH motif (amino acid residues 762 to 885) was sufficient for the interaction with SifA (Fig. 1D). None of the other SPI-2 effector proteins tested interacted with SKIP (Fig. 1D). Thus SKIP, through its PH motif, interacts specifically with SifA.

The physiological relevance of the SKIP-SifA interaction was studied by examining the functional consequences of RNA interference (RNAi) depletion of SKIP on SifA-dependent phenotypes. Transfection of HeLa cells with a small interfering (si) RNA directed against SKIP (SKIP-RNAi) reduced the expression of endogenous SKIP or overexpressed myc.SKIP to about 20% of control levels (Fig. 2A). RNAi-treated HeLa cells were infected with wild-type S. typhimurium or a sifA mutant for 14 hours, and SifA-dependent phenotypes were scored. The ability of wild-type S. typhimurium to induce the formation of Sifs was strongly reduced in SKIP-RNAi cells as compared to cells transfected with a control oligoribonucleotide (ctrl-RNAi). In addition, depletion of SKIP decreased the stability of SCVs, resulting in the release of wild-type bacteria into the cytosol, pheno-copying the effect of the sifA mutation (Fig. 2B). We then tested whether SKIP was required for intracellular replication of Salmonella in MelJuSo human melanoma cells. We found that wild-type Salmonella displayed a replication defect in SKIP-depleted cells, which was comparable to that of the sifA mutant in control cells (fig. S2). Thus SKIP-depleted cells are unresponsive to SifA, and SKIP acts as an essential mediator of SifA functions.

Fig. 2.

SifA functions are mediated by its interaction with SKIP. (A) Transfection of HeLa cells with siRNA directed against SKIP leads to a strong decrease in endogenous or overexpressed SKIP levels. HeLa cells were either not transfected (N.T.) or transfected with siRNAs directed against SKIP (SKIP-RNAi) or a corresponding scrambled RNA (ctrl-RNAi) for 48 hours and then transfected or not with a plasmid encoding myc.SKIP for 24 hours. 40 μg of total proteins were loaded per lane and were immunoblotted with appropriate antibodies. (B) Depletion of SKIP and absence of sifA result in the same phenotype in Salmonella-infected cells. RNAi-treated HeLa cells were infected with GFP-expressing wild-type (wt) or sifA mutant Salmonella for 14 hours, fixed, and immunostained for the SCV marker lamp1. Sifs, bacteria in SCVs, and the position of SCVs were scored (23). Representative confocal microscopy images of SKIP-depleted or control cells infected with wild-type Salmonella are shown. Scale bar, 10 μm.

Although most wild-type SCVs had a juxtanuclear localization in control cells, we found SCVs scattered throughout SKIP-depleted cells (Fig. 2B). Scattering of SCVs was also observed in control cells infected with the sifA mutant (Figs. 2B and 3A), indicating that SifA via SKIP may also control the intracellular positioning of SCVs. These and other findings (1214) suggest the importance of the intracellular positioning and of microtubule motors in SCV membrane dynamics. We investigated the role of SifA and SKIP in the regulation of SCV-associated molecular motor activities. Conventional kinesin and dynein are the major plus-end–directed and minus-end–directed microtubule motors, respectively. We analyzed their distribution during the course of a Salmonella infection. Dynein was equally distributed on wild-type and sifA SCVs (fig. S3). Conversely, most sifA SCVs were strongly decorated by an antibody to kinesin beyond 6 hours of infection, whereas wild-type SCVs displayed a limited association with kinesin (Fig. 3A). We examined the kinetics of kinesin acquisition. As early as 4 hours after invasion, kinesin started to be associated with sifA SCVs and continued to accumulate, so that 86 ± 6% of SCVs labeled positively for kinesin at 14 hours after invasion (Fig. 3B). Selective accumulation of kinesin on sifA SCVs was also observed in MelJuSo cells, as well as in mouse bone marrow–derived macrophages (Fig. 3C). The wild-type phenotype was rescued by introducing a plasmid encoding the sifA allele into the mutant strain (Fig. 3B). Thus, recruitment of kinesin to SCVs is specifically associated with the absence of SifA. Vacuoles enclosing a ssaV mutant, which is defective for the secretion of all SPI-2 effector proteins, were seldom positive for kinesin (Fig. 3B), thereby indicating that the absence of SifA per se is not sufficient to induce the recruitment of kinesin. Rather, this phenomenon requires the translocation of other effector proteins via the SPI-2 TTSS. Thus, the formation of the kinesin coat on SCVs requires a functional SPI-2 secretion system and is negatively regulated by SifA.

Fig. 3.

Vacuoles containing the sifA mutant accumulate kinesin during the course of an infection. (A) HeLa cells were infected with GFP-expressing wild-type or sifA Salmonella (green) for 10 hours, fixed, and immunostained for lamp1 (blue) and kinesin (red) and observed by confocal microscopy. Wild-type SCVs are enriched in lamp1 but seldom positive for kinesin. Most sifA bacteria are present in the cytosol (arrowhead) as shown by the absence of surrounding lamp1. sifA mutants still present in vacuoles are associated with kinesin (arrows). Scale bar, 10 μm. (B) The percentage of kinesin-positive SCVs at different time points of an infection of HeLa cells was scored (23) for wild-type Salmonella (squares) and ssaV (diamonds), sifA (circles), and sifApsifA (triangles) mutants. (C) sifA SCVs accumulate kinesin in various cell types. MelJuSo cells and mouse bone marrow–derived macrophages (BMM) were infected with sifA mutant bacteria (green) for 10 hours, fixed, and immunolabeled for kinesin (red). Scale bar, 10 μm.

We next explored whether the SifA-mediated recruitment of SKIP promotes the negative regulation of kinesin recruitment on SCVs. The infection of SKIP-depleted HeLa cells with wild-type S. typhimurium for 14 hours resulted in 61 ± 7% SCVs coated with kinesin, compared to 18 ± 4% in control cells (Fig. 4A). Thus, the depletion of SKIP, like the absence of SifA, leads to an accumulation of kinesin on SCVs. We next investigated the impact of high kinesin motor activity on SifA/SKIP-dependent phenotypes. Overexpression of the tetratricopeptide repeat (TPR) cargo-binding domain of the mouse kinesin light chain 2 exerts a dominant negative effect on kinesin motor functions (15). In sifA infected cells, overexpression of a green fluorescent protein (GFP)–tagged TPR domain (GFP.TPR) was sufficient to relocalize SCVs to the juxtanuclear area and to preserve the integrity of vacuoles (13) (fig. S4). Thus, in infected cells, SKIP is an inhibitor of kinesin recruitment, and its activity is required for the positioning and maintenance of the SCV.

Fig. 4.

SKIP is a down-regulator of kinesin motor activity. (A) Depletion of SKIP results in the accumulation of kinesin on wild-type SCVs. Control or SKIP-depleted HeLa cells were infected with GFP-expressing wild-type Salmonella (green) for 14 hours and immunostained for lamp1 (not shown) and kinesin (red). Most SCVs were kinesin-positive (arrows) in SKIP-depleted cells. Scale bar, 10 μm. (B) Scattering of the Golgi apparatus in SKIP-depleted cells is due to a high level of kinesin activity. Control or SKIP-depleted HeLa cells were transfected with plasmids encoding GFP or HA.TPR and immunostained for the Golgi marker giantin (red) and possibly for HA.TPR (green). Nontransfected (NT) and GFP- or HA.TPR-expressing cells were scored (23) for the presence of a stacked Golgi apparatus. Representative confocal microscopy images of SKIP-depleted or control cells are shown. Scale bar, 20 μm. (C) SKIP pulls down kinesin. An extract from HeLa cells was incubated with GST.SKIP-derived polypeptides immobilized on beads. Bound proteins were immunoblotted with an anti-kinesin heavy chain (HC). SKIP and SKIP(1-310), but not SKIP(518-1019), pulled down kinesin.

Conventional kinesin is involved in the transport of vesicles, organelles, chromosomes (16), and intermediate filaments (17) and in maintaining the structure and position of the Golgi apparatus (18). To test whether SKIP also functions as a kinesin regulator in the context of uninfected cells, we analyzed the fate of the Golgi apparatus in RNAi-treated HeLa cells. Depletion of SKIP induced a scattering of the giantin labeling, whereas control cells presented a regular stacked Golgi (Fig. 4B). An ongoing tug of war between dynein and kinesin occurs at the Golgi level. Golgi scattering has been observed upon disruption of dynein function, presumably because the plus-end–directed motor is still active (19). In contrast, the disruption of kinesin function causes the collapse of the Golgi apparatus (18, 20). Thus, scattering of the giantin labeling in absence of SKIP is consistent with elevated kinesin motor activity associated with the Golgi apparatus. To confirm this hypothesis, we inhibited kinesin motor activity by expressing HA.TPR in SKIP-depleted cells. Expression of HA.TPR but not GFP restored the stacked Golgi apparatus (Fig. 4B). Thus, SKIP acts as a down-regulator of kinesin.

Little is known about how kinesin interacts with its cargos and how this process is regulated. Several proteins mediating the binding of kinesin onto cargos have been identified (21). Among them are kinectin, the amyloid precursor protein, and JNK-interacting proteins (JIP-1, JIP-2, and JIP-3). The latter are scaf-folding proteins that may assemble linkers and regulatory proteins. We investigated the possibility that SKIP may form a complex with kinesin in such a regulatory complex of proteins. We tested the capacity of recombinant SKIP fragments to pull down kinesin from HeLa cellular extracts. SKIP and its N-terminal domain, but not its C-terminal domain, pulled down kinesin (Fig. 4C). Thus, the RUN-containing N-terminal domain of SKIP interacts with kinesin.

We observed an inverse correlation between the recruitment of SKIP and the recruitment of kinesin, thereby ruling out the possibility of a sole direct interaction between both molecules. Because the recruitment of SKIP on membranes displaces kinesin in vivo and because SKIP pulls down kinesin in vitro, we propose a model in which the RUN-containing domain of SKIP binds kinesin in a regulatory complex of proteins (fig. S5). Thus, kinesin displacement would result from the interaction of SKIP with accessory proteins. The RUN motif–containing region of UNC-14 interacts with UNC-16, the Caenorhabditis elegans JIP-3 (22). UNC-14 also binds kinesin and regulates synaptic vesicle transport (22). Thus, RUN motifs may have a role in the regulation of the dynamics of cargo/kinesin interactions.

A dynamic process of kinesin recruitment in Salmonella-infected cells is mediated by the secretion of unidentified SPI-2 TTSS effectors and is down-regulated by the SifA-mediated recruitment of SKIP on membranes. S. typhimurium is thereby able to fine-tune the SCV-associated kinesin motor activity by regulating the secretion of its own effector proteins. The consequences of an increased kinesin motor activity associated with SCVs may be an excessive formation of outgoing tubules and vesicles, eventually leading to SCV disruption. The identification of SKIP as a SifA interactor and the demonstration that it acts as a negative regulator of kinesin motor activity provide a key to understanding the molecular mechanism of SifA action.

Supporting Online Material

Materials and Methods

Figs. S1 to S5


References and Notes

View Abstract

Navigate This Article