The facultative intracellular pathogen
Salmonella enterica serovar Typhimurium causes gastroenteritis in humans and a lethal systemic disease in certain strains of mice (
74). These bacteria use two separate type III secretion systems (T3SS) to manipulate host cell machinery and direct their own entry into and replication in host cells. Invasion of epithelial cells is directed by the
Salmonella pathogenicity island 1 (SPI-1)-encoded T3SS, a protein delivery system that translocates bacterial proteins (effectors) across the plasma membrane and into the host cell cytosol (
22). These effectors interact with host signaling proteins, causing rearrangement of the underlying actin cytoskeleton (
21). Cell surface ruffling drives uptake of the bacteria into
Salmonella-containing vacuoles (SCVs) (
23). Effectors of the SPI-1 T3SS also direct early maturation of the SCV (
34,
64,
65,
67). Several hours after invasion, a second T3SS, encoded by SPI-2, delivers a separate set of effectors that aid in bacterial replication and survival in the host cell (
30,
32,
62,
63).
The SCV is segregated from the normal host cell endocytic pathway, allowing the bacteria to replicate intracellularly and avoid lysosomal fusion (
65). At early time points postinfection (p.i.; 0 to 60 min p.i.), the SCV migrates to the juxtanuclear region (
28). Previous studies have suggested that this centripetal movement is mediated by microtubule-based motors (
28). At later stages of infection, the SCV is maintained at the microtubule organizing center and associates with the Golgi apparatus (
61). Positioning of the SCV is thought to be important for intracellular pathogenesis, as close apposition of SCV to the Golgi apparatus is associated with maximal intracellular bacterial growth (
61).
Maintenance of SCV positioning at the juxtanuclear region is mediated by various SPI-2 T3SS effectors that manipulate microtubule motor activity at the SCV. SseF and SseG recruit dynein (
1), while SifA recruits the host protein SKIP (SifA and kinesin-interacting protein) to uncouple kinesin (
10). Another SPI-2 effector, PipB2, was shown to work antagonistically with SifA to promote kinesin accumulation on the SCV (
29). Molecular motors also regulate vacuolar membrane integrity; inhibition of dynein or kinesin leads to destabilization of the SCV (
25). In addition, kinesin activity is required for centrifugal extension of membranous
Salmonella-induced filaments (Sifs) from the SCV (
24,
28). Thus, serovar Typhimurium maintains a delicate balance of both plus- and minus-end-directed microtubule-based motors at the SCV to control its positioning and stability.
The SCV is associated with a network of actin, termed vacuole-associated actin polymerizations (VAP), that also extends along Sifs (
12,
46,
48). VAP formation is largely dependent on the SPI-2 T3SS effector SteC; however, the mechanism by which it acts remains unclear (
55). Depolymerization of VAP by cytochalasin D treatment causes serovar Typhimurium to escape from the SCV into the cytoplasm, a toxic environment in some cell types (
8,
76). Despite the importance of VAP, a role for actin-based motors in serovar Typhimurium pathogenesis has not been examined to date.
Myosins are actin-based motors constituting a large superfamily of more than 15 members (
6). All myosins contain an ATPase motor domain which binds actin and drives movement, a neck domain that binds two light chains, and a tail domain that interacts with a specific cargo (
47). The filament-forming class II myosin (conventional myosin II) is involved in muscle contraction (
41) and cytokinesis (
45), while unconventional nonmuscle myosin II participates in many diverse cellular functions, including phagocytosis, organelle transport, and signal transduction (
47). Some pathogenic organisms use myosins to drive invasion and aid their movement within host cells. Cossart and colleagues have shown that uptake of
Listeria monocytogenes requires myosin VIIA (
66), while
Shigella flexneri dissemination and murine leukemia virus infection were shown to be dependent upon myosin II (
40,
59). In addition, several different myosins have been shown to be recruited to the phagocytic cup and localize to model phagosomes in macrophages (
4,
16,
52,
72).
In this study, we asked whether nonmuscle myosin II plays a role in serovar Typhimurium pathogenesis. We demonstrate that myosin II maintains the SCV in a juxtanuclear position and maintains the integrity of this compartment during infection. Furthermore, we provide evidence that the SPI-1 effector SopB can regulate SCV positioning during early and late stages of infection through the activation of a Rho/Rho kinase (ROCK)/myosin II pathway. Thus, our findings reveal a central and previously unappreciated role for myosin II in controlling SCV dynamics within infected host cells.
MATERIALS AND METHODS
Cell culture.
HeLa and RAW 264.7 cells were obtained from the ATCC and maintained in Dulbecco's modified Eagle's medium (HyClone) supplemented with 10% fetal bovine serum (Wisent) at 37°C with 5% CO2 and without antibiotics. Cells were used between passages 5 and 30.
Strains, plasmids, and transfection.
The serovar Typhimurium strains used in this study were CS401 (wild type [WT]) (
49), CS800 (Δ
ssaT) (
31), MBO87 (Δ
sseJ) (
51), MBO107 (Δ
sseJ; complemented with a plasmid carrying
sseJ) (
51), and MBO106 (Δ
sseJ; complemented with a plasmid encoding a catalytically inactive mutant of SseJ [with mutations S151A, D247N, and H384N]) (
51), which have all been described previously. MBO207 (Δ
pipB2) was generated in the CS401 background. WT serovar Typhimurium SL1344 (
35) and isogenic mutants including M202 (Δ
sopE Δ
sopE2 mutant) (
70), Δ
sopB (
68), a Δ
sopB mutant complemented with
sopB (
68), and a Δ
sopB mutant complemented with a plasmid carrying the catalytically inactive (C462S) mutant of
sopB (
68) have all been described previously. Serovar Typhimurium SL1344/p
sifA-
2HA (
12) was used for the SifA colocalization experiments.
Bacteria were grown in Luria-Bertani (LB) broth supplemented with streptomycin, carbenicillin, kanamycin (all at 50 μg/ml), or chloramphenicol (30 μg/ml) as required.
Plasmids used were pegfp-N1 (Clontech); pmrlc2-WT and pmrlc2-AA (
19) (generously provided by G. Egea, University of Barcelona); psopBΔGFP (
42); prhoA-Q63L, pCdc42-Q61L, and prhoG-Q61L (
33) (generously provided by W. D. Heo and T. Meyer, Stanford University); and pDN-ROCK (
44), pCAT-ROCK (
2), and pMLC-DD (
36) (generously provided by A. Kupas, University Health Network, Toronto, Ontario, Canada). Plasmids were transfected into cells 16 h before infections/immunostaining using Fugene 6 (Roche) or Gene Juice (Promega) transfection reagents according to manufacturers’ instructions.
Bacterial infection of cultured cells.
Salmonella infections were performed as previously described (
69). Briefly, bacteria were grown for 16 h at 37°C with shaking and then subcultured (1:33) for 3 h in LB broth. Late-log-phase bacteria were used at a multiplicity of infection of approximately 300:1 to infect cells with a brief (10-min) exposure to bacteria. Drugs (from Sigma; see below) were added at 10 min p.i. and cells were fixed at 2 h p.i. or drugs were added at 2 h p.i. and cells were fixed at 8 h p.i. in 20 μM blebbistatin, 0.5 μg/ml cytochalasin D, 2 μg/ml nocodazole in 1% dimethyl sulfoxide (DMSO), 5 μM Y-27632 in 1% DMSO, or 2 nM calyculin A in 1% DMSO.
siRNA.
To knock down expression of myosin IIA and ROCK I and ROCK II, small inhibitory RNAs (siRNAs) (Dharmacon) were used. HeLa cells were seeded into 24-well culture plates at 2.5 × 104 cells per well. The following day, cells were transfected using Oligofectamine (Invitrogen) with either control siRNA (pool of four nontargeting siRNAs [siCONTROL, D-001206-13-05; Dharmacon]) or siRNA directed against myosin IIA (siGENOME SMARTpool, M-007668-00; Dharmacon) or both ROCK I (5′-GAG GCT CAA GAC ATG CTT A-3′) and ROCK II (5′-GGC ATC GCA GAA GGT TTA T-3′) (Dharmacon). A concentration of 50 nM of total siRNA was used in each knockdown. Medium was changed 24 h after transfection, and HeLa cells were infected with serovar Typhimurium 48 h after transfection.
SCV position measurements.
Intracellular SCV positions were determined by measuring the distances from lysosome-associated membrane protein-1-positive (LAMP-1+) SCVs to the nearest edge of the nucleus (labeled by DAPI [4′,6′-diamidino-2-phenylindole]). Images of infected cells were acquired by epifluorescence microscopy. Measurements were determined using Openlab 3.1.7 software (Improvision). The distances from ≥100 LAMP-1+ SCVs to the nucleus were measured for each time point. Average distances ± standard errors (SE) for three separate experiments were determined. Average cell surface areas were also determined by tracing cell outlines (>30 cells per time point or drug treatment for each of three separate experiments) imaged by epifluorescence or differential interference contrast microscopy and determining surface area by using Openlab software.
Immunofluorescence.
Cells were fixed for 10 min at 37°C in phosphate-buffered saline (PBS), pH 7.2, containing 2.5% paraformaldehyde. Fixed cells were washed with PBS and then permeabilized/blocked in PBS-10% normal goat serum containing 0.2% saponin. For phospho-myosin II light chain (PP-MLC) staining, cells were washed for 10 min with PBS-100 mM glycine prior to permeabilization. Samples were incubated separately with primary and secondary antibodies diluted in the PBS-10% normal goat serum containing 0.2% saponin for 1 h each and then washed three times with PBS. Coverslips were mounted onto glass slides by using antifade mounting reagent (DakoCytomation) and were analyzed by using a Leica DMIRE2 epifluorescence microscope.
For confocal microscopy, cells were immunostained as described above and then analyzed on a spinning-disk confocal microscope. Confocal sections of 0.25 μm were acquired by using a Leica DMIRE2 inverted fluorescence microscope equipped with a Hamamatsu back-thinned electron multiplying charge-coupled-device camera and spinning-disk confocal scan head. Volocity 4 (Improvision) software was used to assemble confocal z sections into flattened projections and for movie construction. Image assembly was done using Adobe PhotoShop and Adobe Illustrator software.
Antibodies.
Mouse anti-LAMP-1 H4A3 monoclonal antibody (used at a dilution of 1:50) developed by T. August was obtained from the Developmental Studies Hybridoma Lab under the auspices of the NICHD, National Institutes of Health Sciences, and maintained by the University of Iowa, Iowa City, IA. Bacteria were detected by using either anti-Salmonella group B rabbit polyclonal antibodies from Difco at 1:100 or a monoclonal anti-Salmonella antibody (BioDesign International) at 1:100. Actin was detected by using Alexa Fluor 568-conjugated phalloidin (Molecular Probes) at 1:50, and rabbit polyclonal myosin IIA heavy chain antibodies from Covance were used at 1:200. Transfected cells were detected using either a monoclonal antibody to the myc epitope tag (1:50) (Covance), a polyclonal anti-green fluorescent protein (GFP) antibody (1:200) (Clontech), or a monoclonal anti-GFP antibody (1:200) (Invitrogen). Phospho-myosin II was detected with the rabbit Ser 18/Thr19 phospho-myosin II antibody from Cell Signaling (1:50 for immunofluorescence, 1:1,000 for Western blotting). Hemagglutinin (HA) was detected using the monoclonal HA.11 (Covance) at 1:100 for 4 h. A rabbit polyclonal kinesin antibody against the kinesin heavy chain was used at 1:500 (PCP42; a gift from Ron Vale, University of California). The following secondary antibodies were used at 1:200: Alexa Fluor 568-conjugated goat anti-rabbit immunoglobulin G (IgG), Alexa Fluor 350-conjugated goat anti-rabbit IgG, Alexa Fluor 488-conjugated goat anti-rabbit IgG, Alexa Fluor 488-conjugated goat anti-rat IgG, Alexa Fluor 488-conjugated goat anti-mouse IgG, and Alexa Fluor 568-conjugated goat anti-mouse IgG (all from Molecular Probes). For some experiments, mouse anti-LAMP-1 H4A3 was conjugated to Alexa Fluor 568 by using the Zenon Alexa Fluor 568 mouse IgG labeling kit (Molecular Probes). To detect myosin II during Western blotting with siRNA-treated cell extracts, rabbit anti-nonmuscle myosin heavy chain (BT-561) (Biomedical Technologies Inc.), which detects both myosin IIA and IIB isoforms, was used at 1:1,000. To detect ROCK II, rabbit anti-ROKα/ROCK II (clone A9W4) (Upstate) was used at 1:1,000.
PP-MLC Western blotting.
HeLa cells were seeded into 24-well culture plates at 2.5 × 104 cells per well. The following day, cells were transfected with Lipofectamine 2000 (Invitrogen) with indicated constructs. Cells were washed with cold PBS (without magnesium and without calcium), and cell lysates were collected 17 h following transfection.
Statistics.
For all experiments, average values and standard deviations (SD)/SE for three experiments were determined, and ≥100 bacteria/SCVs/Sifs/cells were counted for each treatment/sample within an experiment. Depending on the nature of the experiment, statistical analyses were performed using either a two-tailed unpaired t test or a one-way analysis of variance (ANOVA) (Prism 4 software). P values of <0.05 were considered significant.
DISCUSSION
Recent studies have shown that positioning of the SCV near the nucleus is controlled, in part, by microtubule motors (
1,
10,
29,
58). Here, we demonstrated that an actin-based motor, myosin II, also contributes to spatial control of SCVs. We demonstrated that, as with microtubule motor activity, myosin II activity is modulated by a type III secreted effector protein (SopB). Our studies reveal a previously unappreciated intersection of the microtubule and actin cytoskeletons that impact the SCV.
Myosin II, along with kinesin and dynein, contributes to a remarkable balance of cytoskeletal motors acting on the SCV. Here, we demonstrated that myosin II counters the activity of PipB2, which recruits kinesin to the SCV (
29). Also countering the activity of the PipB2/kinesin complex is SifA, which recruits the kinesin uncoupler SKIP to SCVs (
10). SifA is recruited to the SCV by a SPI-1 effector, SipA, and cooperation between SifA and SipA is required for correct perinuclear positioning (
11). Brawn et al. suggest that SipA, SifA, and SKIP might compose a multiprotein SCV regulatory complex (
11). Deletion of
sifA (
15) or
sseF and
sseG (which recruit dynein) (
1,
15) or inhibition of myosin II (this study) leads to centrifugal SCV displacement. Therefore, it would appear that the PipB2/kinesin complex provides a dominant centrifugal force on SCVs that requires multiple complementary forces to counter it and maintain the SCV near the nucleus. However, it should be noted that inhibiting myosin II led to centrifugal SCV displacement even without significant kinesin accumulation on SCVs (Fig.
4), suggesting that basal levels of kinesin are sufficient to mediate peripheral displacement of SCVs if given the opportunity.
The balance of cytoskeletal forces acting on the SCV must be finely balanced or else its integrity is disrupted (
10,
25). Here, we showed that myosin II inhibition in HeLa cells leads to destabilization of the SCV and escape of bacteria into the cytosol. The deacylase activity of SseJ was found to mediate the escape of bacteria when myosin II was inhibited. Similarly, SseJ mediates SCV destabilization in cells infected with Δ
sifA bacteria (
51,
60). Therefore, our studies reveal how serovar Typhimurium utilizes both actin- and microtubule-based motors to compensate for the membrane disrupting ability of SseJ to maintain the SCV.
The SPI-1 effector SopB is conserved in most contemporary serovars of
Salmonella, suggesting that it has evolved as an essential component of their pathogenic repertoire (
50,
57). Indeed, SopB has been shown to play a role in
Salmonella pathogenesis in several animal models (
50,
79). Furthermore, this effector has been linked to a remarkable number of cellular phenotypes associated with
Salmonella infection, including invasion of nonphagocytic cells (
5,
54,
56,
73,
80), early maturation of the SCV (
34), modulation of ion channel activity (
7), and induction of nitric oxide synthase (
17). SopB is delivered into the host cell via the SPI-1 T3SS during invasion and persists in host cells for up to 12 h (
17). The data presented here significantly expand the number of known host cell processes that can be affected by SopB and provide evidence for cross talk between effectors of the SPI-1 and SPI-2 T3SS. This evidence supports studies by Brawn et al. showing a functional relationship between the SPI-1 effector SipA and the SPI-2 effector SifA (
11).
SopB is a phosphatase with a broad substrate specificity for inositol phosphates and phosphoinositides in vitro (
34,
42,
50,
80). However, the in vivo substrates of SopB remain unclear (
18), and it is not known how its phosphatase activity mediates Rho family GTPase activation. Data presented here suggest that SopB activates myosin II locally through a Rho/ROCK/MLC signaling pathway that leads to correct SCV positioning. Localized myosin II activity on an endosomal compartment has been observed in other systems. For example, Sturge et al. have reported that the Endo180 receptor can generate localized and sustained Rho/ROCK/MLC signaling during focal adhesion disassembly (
71). Our studies also indicated that activated Cdc42 and RhoG led to intermediate levels of myosin II activation, as indicated by PP-MLC immunofluorescence (Fig.
10B). This would explain why positioning of the Δ
sopB mutant SCVs could be complemented by the expression of these proteins (Fig.
8D). The mechanism(s) by which SopB can activate these small GTPases is presently unclear and would be interesting to investigate further.
Several reports (
37,
53) have shown that marked apoptosis of
Salmonella-infected epithelial cells occurs at time points (>12 h p.i.) later than those used in our studies (<10 h p.i.). Apoptosis of infected epithelial cells is delayed through the activation of Akt by SopB (
38). Since we observed SCV positioning effects as early as 2 h p.i., it is therefore unlikely that changes in host cell morphology caused by
Salmonella-induced apoptosis contributed to any differences noted in SCV localization. Knodler et al. noted that less than 10% of HeLa cells infected with a
sopB mutant exhibited evidence of apoptosis at 4 h p.i., in comparison to WT-infected cells, which displayed approximately 5% apoptosis (
38). It is possible that this slight difference may have affected mutant SCV localization; however, we observed no significant change in cell surface area between WT and
sopB-infected HeLa cells even at 10 h p.i.
Despite the obvious effect of myosin II on SCV stability and positioning, we were unable to detect myosin II on SCVs by using immunofluorescence microscopy, and very little association with Sifs was observed (Fig.
9). We also could not detect the association of overexpressed monomeric red fluorescent protein-myosin IIA with SCVs/Sifs (data not shown). It is possible that myosin II associates transiently with SCVs/Sifs and that its steady-state levels on these compartments are not detectable with the methods used. Interestingly, blebbistatin treatment allowed us to detect significant levels of association of myosin II with both SCVs and Sifs (Fig.
9). It is possible that blebbistatin treatment traps myosin II on a fraction of SCVs/Sifs, preventing its release by blocking its ATPase activity. A similar trapping of microtubule motors on their respective cytoskeletal filaments/cargo using specific drugs or nonhydrolyzable ATP analogues has been reported (
9,
14). The observation that myosin II can associate with SCVs/Sifs (albeit under artificial conditions) suggests that this motor acts locally at this compartment during infection. However, we cannot rule out the alternate possibility that myosin II is acting distally to the SCV to promote its positioning in host cells.
In summary, our data provide novel insight into how serovar Typhimurium maintains spatial control and stability of the SCV in host cells during infection. We have demonstrated a role for an actin-based motor in the regulation of SCV positioning, showing that microtubule motors are not the only host motor proteins involved in this process. We are beginning to understand the dynamic cytoskeletal forces that act upon the SCV during infection and how the bacteria control these forces to their benefit. Our study also reveals an unappreciated cross talk between effectors of the SPI-1 and SPI-2 T3SS. The finding that a SPI-1 effector (SopB) can impact SCV dynamics during late stages of infection (up to 10 h p.i. in this study) suggests that effector activities are even more dynamically coordinated than previously thought.
Acknowledgments
We are grateful to members of the Brumell laboratory and T. Yeung, D. Mason, N. Jones, and M. Terebiznik for helpful discussions and critical reading of the manuscript. M. Woodside and P. Paroutis provided support for confocal microscopy, and S. Singh provided technical assistance. We thank G. Egea (University of Barcelona) for providing MLC constructs, W. Do Heo and T. Meyer (Stanford University) for the Rho GTPase constructs, and A. Masszi and A. Kupas (Toronto General Hospital, University Health Network, Toronto, Ontario, Canada) for the CAT-ROCK and DN-ROCK constructs. We thank L. Knodler and O. Steele-Mortimer (Rocky Mountain Laboratory) for providing bacterial strains.
J.A.W. and J.S. were supported by a Natural Sciences and Engineering Research Council (NSERC) of Canada postdoctoral fellowships. M.A.B. was supported by an NSERC postgraduate scholarship. John H. Brumell holds an Investigators in Pathogenesis of Infectious Disease Award from the Burroughs Wellcome Fund. This work was supported by the Canadian Institutes of Health Research (project no. 151733).