|Home | About | Journals | Submit | Contact Us | Français|
Yersinia spp. undermine the immune responses of infected animals by translocating Yops directly into host cells with a type III secretion system. YopM, a leucine-rich repeat protein, is a critical virulence factor in infection. YopM localizes to both the nucleus and the cytoplasm in cultured cells, interacts with mammalian p90 ribosomal S6 kinase 1 (RSK1), and causes a decrease in NK cell populations in spleens. Little is known about the molecular interaction between YopM and RSK1 and its significance in pathogenesis. We performed a systematic deletion analysis of YopM in Yersinia pseudotuberculosis to determine which regions are required for RSK1 interactions, nuclear localization, virulence, and changes in immune cell populations during infection of mice. Full-length YopM associated with RSK1 in at least two protein complexes in infected cells, and deletion of its C-terminal tail abrogated all RSK1 interactions. The C-terminal tail was required for tissue colonization, as yopM mutants that failed to interact with RSK1 were as defective for tissue colonization as was a ΔyopM mutant; however, nuclear localization of YopM was not dependent on its RSK1 interaction. Mutants expressing YopM proteins which do not interact with RSK1 caused more pathology than did the ΔyopM mutant, suggesting that there are other RSK1-independent functions of YopM. Histopathological and flow cytometric analyses of spleens showed that infection with wild-type Y. pseudotuberculosis caused an influx of neutrophils, while mice infected with yopM mutants had increased numbers of macrophages. Decreases in NK cells after Y. pseudotuberculosis infection did not correlate with YopM expression. In conclusion, the C terminus of YopM is essential for RSK1 interactions and for virulence.
Three bacterial species in the genus Yersinia are pathogenic to humans: the enteropathogens Y. enterocolitica and Y. pseudotuberculosis and the plague-causing bacterium Y. pestis (30). All three species inhibit the host immune response by the actions of effector proteins, called Yops, which are translocated directly into host cells by a type III secretion system (59). Once in target cells, the Yops usurp cellular functions by interfering with signal transduction pathways, enabling bacterial survival and replication (59). The type III secretion system and effector Yops are encoded on a 70-kb virulence plasmid common to all three species (20, 39-42). All three pathogenic species translocate at least 5 Yops, YopE, -H, -J, -O, and -M, which contribute to pathogenesis (10, 29, 36, 53, 55, 56).
A catalytic function(s) has been demonstrated for each of the Yops with the exception of YopM (8, 19, 21, 38, 59, 61, 65). Although less is known about the effects of YopM on host cell function than about those of the other Yop proteins, several reports have given insights into its structure, interactions with mammalian proteins, and role in animal infections (15, 25, 32, 63). YopM is an acidic protein of ~46 kDa that is a member of the bacterial family of leucine-rich repeat (LRR) effector proteins. Homologs include the internalin proteins of Listeria monocytogenes (7), the IpaH proteins of Shigella flexneri (49), and the SspH and SlrP proteins of Salmonella enterica serovar Typhimurium (35, 58). YopM is composed of a central domain of 15 LRRs flanked by an N-terminal domain, which is required for its translocation into host cells, and a short 24-amino-acid C-terminal tail (15). YopM has been crystallized (15), and the solved structure revealed a horseshoe-like shape similar to those of some other LRR proteins (4) but also a paucity of secondary structure on its convex surface and an overall twist which allowed the formation of dimers and tetramers (15). The C-terminal portion of YopM was not resolved in the crystal structure and has no significant homology to other proteins. In contrast to YopM, all the other known bacterial LRR proteins contain larger carboxy-terminal domains (6, 7, 35, 44, 46, 49, 62). For example, the IpaH family members of Shigella and the Salmonella effectors SspH1, SspH2, and SlrP are E3 ubiquitin ligases (6, 44, 49, 50, 66).
YopM is unique among the Yops in that it traffics to the nucleus of eukaryotic cells (27, 51). Regions of YopM sufficient to promote nuclear localization of YopM fusion proteins have been examined, but results have been somewhat contradictory. In one study, both the N-terminal and C-terminal halves of Y. pestis YopM fused to green fluorescent protein (GFP) directed GFP to the nucleus of Saccharomyces cerevisiae cells (52), while the C-terminal tail of 32 amino acids did not. In a second study, the last 32 residues of Y. enterocolitica YopM were sufficient to target GFPs to the nucleus of both yeast and mammalian cells (5). Regardless of the location of the nuclear localization signal, however, the significance of YopM trafficking to the nucleus for virulence is not known.
In HEK293 cells, YopM forms a complex with two mammalian serine/threonine protein kinases, p90 ribosomal S6 kinase 1 (RSK1) and protein kinase C-related 2 (PRK2). RSK1 is a substrate of extracellular signal-regulated kinase (ERK) in the Ras/mitogen-activated protein kinase (MAPK) signal transduction pathway and plays an important role in cell survival and growth (18). RSK1 has two distinct kinase domains separated by a linker region, and its activation is regulated by multiple phosphorylation events (14, 24, 48). Upon activation of ERK, the C terminus of RSK1 is phosphorylated, and subsequently residues in the linker and N terminus become phosphorylated, resulting in a fully activated RSK1 (24, 48). After activation by ERK, a pool of RSK1 traffics to the nucleus, where it phosphorylates nuclear substrates, including CREB (13). It is possible that YopM migrates to the nucleus in association with RSK1 or RSK1-like proteins in yeast. PRK2 is a member of the protein kinase C family involved in cytoskeletal changes, receptor tyrosine kinase signaling, and activation of translation (12, 26, 60). PRK2 is activated by the GTPase RhoA, phospholipids, and MEK kinase 2 and interacts with the tyrosine kinase adaptor proteins Nck and Grb2 (11, 45, 54, 60, 64). In vitro kinase assays indicated that YopM activates the kinase activity of RSK1 (32); however, the significance of these interactions for Yersinia pathogenesis is not known (32).
The importance of YopM in survival and growth of Y. pestis and Y. enterocolitica in mice has been demonstrated in studies showing 1,000-fold to 100,000-fold increases in 50% lethal dose (LD50) for strains deleted of YopM after Y. pestis infection, as well as significant attenuation in colonization of the Peyer's patches (PP), spleen, and liver after Y. enterocolitica infection (25, 28, 37, 57). During infection of spleens with Y. pestis, YopM is injected into neutrophils, macrophages, and dendritic cells (31), and deletion of yopM results in increased mRNA for proinflammatory cytokines in macrophages (25). Studies of intravenous (i.v.) infection with Y. pestis have demonstrated that there is a decrease in NK1.1+ cells in the spleens of mice infected with strains expressing YopM (25); however, depletion of NK cells was recently shown to have no effect on growth or survival of wild-type or yopM Y. pestis during systemic infection, while depletion of GR1+ cells rescued the growth defect of yopM bacteria (63).
Here, we sought to determine the regions of YopM required for its interaction with RSK1 and to determine whether YopM-RSK1 interactions are necessary for YopM localization to the nucleus, YopM-dependent changes in numbers of immune cell populations, and/or virulence. We observed two RSK1-YopM-containing protein complexes in cells infected with Y. pseudotuberculosis and identified regions of YopM which are necessary for the formation of these RSK1 complexes, wild-type levels of nuclear localization, alteration of immune cell populations, and virulence in mice.
The Y. pseudotuberculosis strains used in this study are described in Table Table1.1. Escherichia coli SM10 λpir and SY327 λpir were grown in L (10 g of Bacto tryptone, 5 g of Bacto yeast extract, and 5 g of NaCl per liter) broth or L agar plates at 37°C. The two wild-type (WT) Y. pseudotuberculosis strains used in these experiments were YPIIIpIB1 and IP2666pIB1. Y. pseudotuberculosis strains were grown at 26°C in L broth, Luria-Bertani (LB) broth (10 g of Bacto tryptone, 5 g of Bacto yeast extract, and 10 g of NaCl per liter), or 2× YT broth (16 g of Bacto tryptone, 10 g of Bacto yeast extract, 5 g of NaCl per liter) or on L agar plates containing Irgasan at 0.5 μg/ml. Where applicable, ampicillin was added to the growth medium at 100 μg/ml. For in vitro induction of Yop expression and secretion on the day of infection of cultured cells, overnight cultures of Y. pseudotuberculosis were diluted in 2× YT supplemented with 20 mM sodium oxalate and 20 mM MgCl2 (low-Ca2+ medium). Cultures were incubated with aeration at 26°C for 2 h and then shifted to 37°C for an additional 2 h. For Yersinia strains containing expression plasmids, the shift to 37°C was accompanied by addition of isopropyl-β-d-thiogalactoside (IPTG) to a final concentration of 1 mM.
A yopM deletion of 388 amino acids was constructed in both the YPIII and IP2666 backgrounds to yield MM42 and MM114, respectively (Table (Table1),1), by allelic exchange using primers and methods described in references 29 and 34. Briefly, the 5′ and 3′ flanking regions of YopM were cloned into pCVD442 using primers YopM1, YopM2, YopM3, and YopM4, described in reference 29, resulting in the pYopMKO plasmid, which was then conjugated into YPIIIpIB1 and IP2666pIB1 as described in reference 29. Potential yopM deletion strains were confirmed by PCR, Southern blotting, Coomassie blue staining of secreted proteins precipitated by trichloroacetic acid, and Western blotting using antibody (Ab) to YopM (data not shown).
Deletions of LRRs in YopM were constructed as follows: codons 17 and 18 or codons 18 and 19 of each odd-numbered LRR were replaced with an XmaI site (CCCGGG), generating eight different plasmids (Table (Table2),2), each with an XmaI site in either LRR 1, 3, 5, 7, 9, 11, 13, or 15, using primers CLr1 to CLr15. Consecutive pairs of plasmids containing XmaI sites at LRRs 1 and 3, 3 and 5, 5 and 7, etc., were digested with XmaI and ligated together, resulting in yopM clones encoding proteins with 13 instead of 15 LRRs. In each case, the new LRR contained one amino acid change at position 18 or 19 to a glycine (GGG). Clones were PCR amplified with Pfu Turbo polymerase (Stratagene) using primers P8 and P70 and inserted into the EcoRI site of pFLAG-CTC (Sigma). These constructs were engineered to contain two stop codons upstream of the FLAG peptide tag of this vector, to preclude the formation of any fusion proteins with the FLAG epitope. The plasmids were named pYopM, pΔ2-3, pΔ4-5, pΔ6-7, etc., and were verified by DNA sequencing and introduced into MM42 by electroporation. The pYopMC1 to pYopMC8 plasmids were constructed by site-directed PCR mutagenesis of pYopM using primer pairs P96/P97 through P110/P111. All clones were verified by DNA sequencing.
To recombine constructs of interest into the yopM locus of pIB1, a derivative of pYopMKO, called pYopMrec, was constructed in which a single T→G base pair change 5 nucleotides upstream of the yopM start codon was engineered to create an SphI site. yopM mutations were PCR amplified with primers P69/P70 to contain the SphI site, verified by DNA sequencing, and then inserted in the SphI site engineered in pYopMrec. These plasmids were introduced into IP2666pIB1, and double-crossover clones containing the mutant genes were selected for and confirmed by PCR, Coomassie blue staining of secreted proteins precipitated by trichloroacetic acid, Western blotting with YopM antibody, and sequencing.
RAW 264.7 cells were maintained in Dulbecco modified Eagle medium (DMEM; Mediatech, Inc.) supplemented with 10% fetal bovine serum (FBS) at 37°C in a humidified chamber under 5% CO2. Before infection, cells were plated in 6-well or 24-well dishes (Corning) and grown to 90 to 95% confluence. On the day of the infection, overnight cultures of Y. pseudotuberculosis were diluted 1:40 into low-Ca2+ medium and grown with shaking for 2 h at 26°C followed by 2 h at 37°C. After being washed in phosphate-buffered saline (PBS), bacteria were spun onto the cells (195 × g) at a multiplicity of infection (MOI) of 10:1 (translocation assay) or 50:1 (native Western analysis) and placed in the tissue culture chamber for 30 min.
RAW 264.7 cells were infected for 30 min with MM42 expressing plasmid-borne wild-type YopM or one of the YopM mutants. Cells were washed in PBS and then scraped into 0.4 ml of lysis buffer (20 mM HEPES, pH 7.4, 10 mM NaCl, 0.5% saponin, 5 μg/ml aprotinin, 10 μM leupeptin, 1.4 μM pepstatin, 200 μM AEBSF [4-(2-aminoethyl)-benzenesulfonyl fluoride]). Lysis proceeded for 20 min at room temperature with rotation, and the suspensions were cleared by centrifugation at 16,100 × g at 4°C. The soluble fractions (supernatants) were separated in nondenaturing and denaturing polyacrylamide gels, transferred to polyvinylidene difluoride (PVDF) membranes (Immobilon-P; Millipore), and probed with antibodies for YopM, RSK1 (Santa Cruz Biotechnology; clone C-21), PRK2 (Cell Signaling Technology), and β-actin (Sigma; clone AC-74). In some cases, blots were stripped and reprobed with the other antibody. The YopM antibody was a rabbit polyclonal antibody generated to full-length YopM, and the IgG fraction was purified by Covance (Denver, PA). Proteins were visualized with goat anti-rabbit horseradish peroxidase (HRP)-conjugated secondary Abs followed by chemiluminescence (Pierce).
RAW 264.7 cells were infected for 30 min. Cells were washed with cold PBS and then lysed for 20 min with rocking at 4°C in 20 mM HEPES, pH 7.4, 1% Nonidet P-40 (NP-40), 150 mM NaCl, 5 μg/ml aprotinin, 10 μM leupeptin, 1.4 μM pepstatin, and 200 μM AEBSF. Lysates were cleared by centrifugation at 16,100 × g at 4°C, and proteins in the soluble fraction (supernatants) were separated by SDS-PAGE, transferred to PVDF membranes, and probed with YopM antibody.
The yopM and yopMΔC genes were amplified with primer pairs P8/17 and P8/67 (Table (Table3),3), respectively, and ligated into the EcoRI and BamHI sites of pGEX-5-1 (Amersham Biosciences). E. coli strain BL21 was transformed with these plasmids, and expression of the glutathione S-transferase (GST)-YopM and GST-YopMΔC fusion proteins was induced in 1 liter of exponentially growing cells (A600 of ~0.5) for 2 to 3 h with 0.5 mM IPTG. Cells were harvested, resuspended in 10 ml of PBS, and lysed with a French press at 10,000 lb/in2. The lysates were cleared by centrifugation at 15,000 × g and incubated with 1 ml of glutathione-Sepharose 4B resin (Amersham Biosciences) for 30 min at room temperature with rotation. The resin was washed six times with 10 ml of cold PBS and resuspended in 2 ml of factor Xa cleavage buffer (50 mM Tris-HCl, pH 8.0, 1 mM CaCl2, 100 mM NaCl). Approximately 10 units of factor Xa (Amersham Biosciences)/mg of fusion protein was added to the suspension, and cleavage was allowed to proceed for 16 h at room temperature with rotation. The following day, the cleaved YopM and YopMΔC proteins were eluted. Proteins were dialyzed three times against 300 volumes of 20 mM HEPES, pH 7.4, in the presence of protease inhibitors for a total of 40 h at 4°C. Twenty-five micrograms of each protein was analyzed by nondenaturing PAGE, followed by staining with Coomassie brilliant blue.
RAW 264.7 cells were infected for 30 min with YPIIIpIB1 or YPIIIpIB1ΔyopM expressing one of the plasmid-borne YopM mutants. Cells were washed in cold PBS, scraped into 1 ml of cold homogenization buffer (20 mM HEPES, pH 7.4, 10 mM NaCl) with protease inhibitors (5 μg/ml aprotinin, 10 μM leupeptin, 1.4 μM pepstatin, and 200 μM AEBSF), and lysed using a Dounce homogenizer (Wheaton). Phosphatase inhibitors were included unless indicated; these inhibitors were 0.2 mM Na3VO4, 0.5 mM dithiothreitol, 50 mM NaF, and 14 mM beta-glycerol phosphate. Samples were examined microscopically to ensure the presence of free, intact nuclei. The nuclei were pelleted at 1,000 × g at 4°C, and supernatants (cytoplasmic extracts) were further clarified by centrifugation at 16,100 × g at 4°C. Nuclear pellets were washed in the homogenization buffer and resuspended in 1 ml of homogenization buffer containing 0.5% saponin. After rotation at room temperature for 20 min, nuclear extracts were clarified by centrifugation at 16,100 × g at 4°C. Cytoplasmic and nuclear extract proteins were separated by SDS-PAGE and transferred to a PVDF membrane (Immobilon-P; Millipore). Blots were probed with antibodies specific for YopM and Rho GDP dissociation inhibitor (GDI; Santa Cruz Biotechnology; clone A-20). The YopM-probed membrane was then stripped and reprobed with CREB antibody (Cell Signaling Technology).
For intravenous infections, 7- to 8-week-old female BALB/c mice (National Cancer Institute) were injected in the tail vein with 100 μl of PBS containing 100 CFU of stationary-phase IP2666pIB1 grown at 26°C in LB broth. Actual doses were determined by plating inputs onto L plates containing Irgasan. Four days postinfection, mice were sacrificed by CO2 asphyxiation and spleens, livers, and lungs were aseptically removed and placed in preweighed tubes containing 1 ml of PBS-10% glycerol. Tissues were weighed, homogenized using a Tissue Tearor apparatus (BioSpec Products), and plated onto L plates containing Irgasan to determine the CFU/gram of tissue. For all experiments, at least two mice were infected with each strain being tested, and all experiments were repeated at least two times. Data were transformed logarithmically and expressed in graphs as log10 CFU/gram of tissue. P values were determined using analysis of variance (ANOVA) followed by Dunnett's multiple-comparison test of the WTrec strain with the mutant strains.
For oral infections, 7- to 8-week-old female BALB/c mice (National Cancer Institute) were fasted for 16 h and subsequently inoculated through a 20-gauge feeding needle with 200 μl PBS containing 2 × 109 CFU of YPIIIpIB1 or 7 × 108 CFU of IP2666pIB1 which had been grown to stationary phase at 26°C in LB broth. Mice were then provided with food ad libitum. Actual doses were determined by plating inputs onto L plates containing Irgasan. Four days (IP2666) or five days (YPIII) postinfection, mice were sacrificed by CO2 asphyxiation, and the spleen, liver, contents of the terminal third of the small intestine (SI), contents of the cecum, Peyer's patches (PP), and mesenteric lymph nodes (MLN) were harvested as described for intravenous (i.v.) infections. For all experiments, at least two mice were infected with each strain being tested, and all experiments were repeated at least two times. Data were transformed logarithmically and expressed in graphs as log10 CFU/gram of tissue. P values were determined using ANOVA followed by Dunnett's multiple-comparison test of the WTrec strain with the mutant strains or using a two-tailed, unpaired Student t test by comparing colonization levels of the YPIII strain and the IP2666 strain.
The Institutional Animal Care and Use Committee of Tufts University approved all animal procedures.
Mice were intravenously infected with the IP2666pIB1 WTrec, ΔyopM mutant, C7rec, or C8rec strain as described above or mock infected with PBS. At 4 days postinfection, spleens were harvested for histology. Tissues were placed in histocassettes (Fisherbrand), fixed in 10% buffered neutral formalin for 24 to 48 h, washed in 70% ethanol, and embedded in paraffin. Sections 8 to 10 μm thick were stained with hematoxylin and eosin. Slides were evaluated in a blind fashion by two independent investigators. Images were acquired with a 10× objective on a Nikon Eclipse TE2000-U microscope. Pictures were taken with a Nikon DS-M5 color camera, and scale bars in the figures represent 100 μm.
Splenocytes were prepared from mice intravenously inoculated with the IP2666pIB1 WTrec, ΔyopM mutant, C7rec, or C8rec strain or PBS as described above. Spleens were treated with 100 U/ml of collagenase D (Roche) and passed through a 70-μm filter (BD Biosciences) to obtain a single-cell suspension. Red blood cells were lysed with BD Pharmlyse (BD Biosciences), and nonlysed cells were washed in RPMI by centrifugation and resuspended in PBS containing 1% FBS and 0.01% sodium azide. Cells from each sample were treated for 5 to 20 min at 4°C with Fc-block (BD Biosciences). Subsequently, splenocytes were stained with fluorophore-conjugated antibodies (BD Biosciences) for 30 min at 4°C; specifically, macrophages and neutrophils/inflammatory monocytes were stained with phycoerythrin (PE)-CD11b and fluorescein isothiocyanate (FITC)-GR1 antibodies, and NK cells were stained with PE-CD49b (clone HMα2) and FITC-CD3 or FITC-CD69 antibodies. Fluorescently labeled cells were quantified in a FACSCalibur flow cytometer (BD Biosciences) and analyzed using FlowJo software (Tree Star, Inc.).
We first investigated whether we could detect proteins in complex with YopM during infection of the macrophage cell line RAW 264.7 using native protein electrophoresis of cell lysates and Western blot analysis. When macrophages were infected with a yopM deletion strain of YPIIIpIB1 (ΔyopM) expressing wild-type (WT) YopM from a plasmid, YopM was detected in two complexes, YMC1 and YMC2 (for YopM complex), as well as in its monomeric form (Fig. (Fig.1C,1C, lane b). No complexes were observed in macrophages infected with the ΔyopM strain, which contains only the parental plasmid lacking YopM (Fig. (Fig.1C,1C, lane k), or a ΔyscBCDEFGHIJKL (ΔyscB-yscL) strain, which expresses YopM but lacks the type III secretion system, so that it cannot translocate YopM into host cells (Fig. (Fig.1C,1C, lane l). The observation that YMC1 and YMC2 were larger than the multimeric forms observed with purified YopM (Fig. (Fig.1C,1C, lane a) suggested that an additional protein(s) was present in these complexes. These complexes were also seen in infection of macrophages with a Y. pseudotuberculosis strain lacking four effector Yops, yopHEOJ, suggesting that other Yops were excluded from the YopM complexes (data not shown). Since a FLAG-YopM fusion protein forms complexes with RSK1 and PRK2 in HEK293 cells (32), we tested whether these proteins were found in YMC1 and YMC2. We detected RSK1 but not PRK2 in both complexes (Fig. (Fig.1D1D and data not shown). The lack of detection of PRK2 could reflect inefficient binding of the PRK2 antibody to these complexes, differences between HEK293 cells and RAW 264.7 cells, differences between FLAG-YopM and native YopM, or the method of delivery of YopM (transfection versus infection).
To identify the region(s) of YopM necessary for complex formation with RSK1, RAW 264.7 cells were infected with ΔyopM strains expressing a series of YopM mutant proteins with deletions in two tandem LRRs (Δ2-3, Δ4-5, Δ6-7, etc.) or a deletion of the last 24 amino acids (ΔC). All the mutant proteins with the exception of Δ2-3 were expressed and translocated into RAW 264.7 cells at levels similar to that of WT YopM (Fig. (Fig.1B).1B). Of note, weak expression has been previously observed with deletions in LRRs 1 to 4 of Y. pestis YopM (52). The observed alterations in migration of the monomeric form (Fig. (Fig.1C)1C) likely reflect changes in the pI or the structure of the mutant proteins. None of the LRR deletions abrogated YopM's interaction with RSK1 to form YMC1. This result suggests that there are no essential determinants for binding RSK1 within the 4th to 15th LRRs; however, given the multiple repeats, it is plausible that some LRRs have redundant roles in binding (Fig. 1C and D). However, deletions of LRRs 8 and 9 through 14 and 15 resulted in loss of detection of YMC2. This indicates that detection of YMC2 requires LRRs 8 to 15 of YopM, which are dispensable for the formation of the smaller complex. The C terminus of YopM was required for detectable interaction with RSK1, as neither complex was observed in the presence of the ΔC mutant.
To further define the C-terminal residues of YopM required for interaction with RSK1, the last 24 amino acids of YopM were sequentially changed in triplicate to alanines, resulting in eight different mutants of YopM, C1 to C8 (Fig. (Fig.2A).2A). RAW 264.7 cells were infected with ΔyopM strains expressing plasmid-borne copies of the C1 to C8 mutant YopM proteins. Although each mutant was translocated efficiently into RAW 264.7 cells (Fig. (Fig.2B),2B), all of the C-terminal YopM mutants failed to form detectable levels of YMC2, except for the C3 mutant. C3 formed very low levels of YMC2 that were observed only with the RSK1 antibody (Fig. 2C and D). Furthermore, the interaction between YopM and RSK1 in YMC1 appeared weaker with each sequential set of mutations, significantly diminishing with C6 and completely disappearing with C7 and C8 (Fig. (Fig.2),2), indicating that the last six amino acids of YopM are essential for detectable interaction with RSK1.
Since multimers of YopM were observed in solution (Fig. (Fig.1C),1C), YMC2 could be comprised of multimers of YopM and RSK1. To test whether ΔC failed to form YMC2 because it cannot multimerize, we tested whether the ΔC mutant formed multimers in solution. ΔC was purified and analyzed after electrophoresis on a nondenaturing gel. Both WT and ΔC YopM formed higher-molecular-weight complexes in solution (Fig. (Fig.3).3). A series of Ferguson gels (16, 22) indicated that the apparent molecular weights of these complexes were consistent with dimers and tetramers (data not shown). These results suggest that the inability of ΔC to interact with RSK1 is not due to its being unable to multimerize with itself.
YopM fusion proteins have been localized to both the cytosol and the nucleus of yeast and mammalian cells (5, 27, 51, 52). Since RSK1 shuttles between the cytoplasm and nucleus of cells (13, 47) and alanine substitution for residues within YopM's last 6 amino acids prevented YopM's interaction with RSK1 (Fig. (Fig.2),2), we tested whether the interaction of YopM with RSK1 was required for translocation of YopM to the nucleus. We analyzed the location of YopM mutants by Western blot analysis of cytoplasmic and nuclear cell fractions after infection of RAW 264.7 cells (Fig. (Fig.4).4). WT YopM partitioned equally in the two fractions (Fig. (Fig.4).4). Likewise, the C7 and C8 mutants of YopM were not defective in nuclear localization, indicating that binding to RSK1 is not required for YopM trafficking to the nucleus. On the other hand, Δ8-9, Δ10-11, Δ12-13, and Δ14-15 were impaired in nuclear localization, while Δ4-5, Δ6-7, and ΔC had modest decreases in nuclear localization. Detection of Rho GDI in only the cytoplasmic extracts and of CREB in the nuclear extracts demonstrated that the fractionation procedure separated these two compartments successfully (Fig. (Fig.4).4). These results indicate that the formation of YMC1 or YMC2 was not required for nuclear localization, as mutants that were unable to bind to RSK1 were able to reach the nuclei. In addition, these results suggest that the regions of YopM important for nuclear localization are embedded in the LRRs, particularly those in the second half of the protein.
YopM from Y. pestis and Y. enterocolitica plays a critical role in the development of systemic infection in mice after intravenous injection (25, 28, 37, 57). To investigate the role of YopM in infection of mice with Y. pseudotuberculosis, isogenic strains of IP2666pIB1 were constructed that encode either WT or mutant alleles of yopM at the yopM locus, resulting in a series of yopM recombined (rec) strains (Table (Table1).1). The mutants were chosen to test whether the formation of YMC2 or wild-type levels of nuclear localization (Δ8-9, Δ10-11, Δ12-13, and Δ14-15) or formation of either complex with RSK1 (ΔC, C7, and C8) is important for colonization by Y. pseudotuberculosis. All of the mutant YopM proteins that were expressed from the yopM locus were translocated into RAW 264.7 macrophages at similar levels (Fig. (Fig.5A5A).
Four days post-infection of BALB/c mice with 100 CFU of the different IP2666pIB1 strains, the WT strain colonized the spleen and lungs about 25-fold better than did the isogenic ΔyopM strain (Fig. 5B and C). The ΔyopM deletion strain showed a slight but statistically significant defect in colonization of the liver (Fig. (Fig.5D),5D), about 30% of wild-type levels. These results indicate that after i.v. infection, YopM is important for colonization of the spleen and lungs by Y. pseudotuberculosis. By comparing the abilities of Δ6-7rec, Δ8-9rec, Δ10-11rec, Δ12-13rec, and Δ14-15rec strains to colonize the spleen, liver, and lungs, we found that formation of YMC2 and wild-type levels of nuclear localization were not essential for virulence. The Δ6-7rec, Δ8-9rec, Δ10-11rec, and Δ12-13rec strains all colonized the spleen, liver, and lungs as well as did the WT YopM strain, while the Δ14-15rec strain colonized the spleen and lungs as poorly as did the ΔyopM strain. While the phenotype of the Δ14-15rec strain is intriguing and suggests that this mutant is defective for some critical function of YopM for colonization, overall these observations suggest that an inability to form YMC2 and decreased translocation into the nucleus of cells do not impair YopM's role in colonization of either the spleen or the lungs. On the other hand, in mice infected with strains carrying the ΔCrec, C7rec, or C8rec mutants, these strains were all significantly attenuated in the spleen and lungs. These results suggest that the ability to form YMC1 is important for survival of Y. pseudotuberculosis in systemically infected mice.
A second route of infection, the orogastric route, was studied to determine whether YopM of Y. pseudotuberculosis was required for colonization of the gastrointestinal tract and lymph nodes (Fig. (Fig.6).6). First, colonization by the WTrec strain of IP2666pIB1 was compared to that by its parental wild-type (WT) IP2666 strain, and no differences were found. We then compared the levels of colonization of WTrec with those of ΔyopM, Δ6-7rec, Δ8-9rec, and ΔCrec strains. No statistically significant differences were found between the WTrec strain and any of the yopM mutant strains in any tissue, with the exception of Δ6-7rec, which appeared defective in colonizing the small intestine and cecum. However, this apparent defect was due to two mice (of eight mice) whose ileum and Peyer's patches were not detectably colonized in one experiment, although their ceca and MLN were robustly colonized (Fig. (Fig.6).6). These results indicate that YopM is not critical for colonization of these tissues 4 days after oral inoculation.
Mice were also orogastrically infected with the WT and the ΔyopM mutant of the YPIII strain of Y. pseudotuberculosis, as YPIIIpIB1 has been previously studied in oral infections of BALB/c mice (3, 29, 33). WT YPIIIpIB1 colonized the lymph and systemic tissues at lower levels than did WT IP2666pIB1, consistent with our observations that mice infected with IP2666pIB1 reach morbidity faster than do mice infected with YPIIIpIB1. In addition, YopM played a more significant role in lymph tissue colonization by the YPIII strains, as the yopM deletion strain of YPIIIpIB1 was significantly attenuated in colonization of the PP, MLN, and spleen (Fig. (Fig.66).
Although Y. pseudotuberculosis strains carrying mutations in the C terminus of yopM were impaired for colonization of BALB/c mice after i.v. infection, mice infected with these strains showed more outward signs of illness than did those infected with the ΔyopM strain. Thus, we investigated the histopathology of spleens of mice infected with the WTrec, ΔyopM, C7rec, and C8rec strains of IP2666pIB1 at 4 days post-i.v. infection. WTrec usually caused significant neutrophilic inflammation and necrosis in the spleen, often accompanied by severe congestion, and thrombosis (Fig. (Fig.7;7; also Table Table44 ), while the ΔyopM strain generally caused low levels of inflammation and necrosis. The damage caused by infection with the C-terminal mutants was intermediate between that caused by the ΔyopM strain and that caused by the WTrec strain despite the observation that the CFU in the spleen were comparable to those for the ΔyopM strain (Fig. (Fig.7).7). In general, these mutants caused more inflammation and necrosis than did the ΔyopM mutant, but not as much congestion and widespread damage as did the WTrec strain (Fig. 7D and E and Table Table44).
Because of the differences in inflammation and necrosis between the WTrec, ΔyopM, C7rec, and C8rec strains, we asked whether there were differences in the immune cells recruited to the spleens infected with these different strains by using flow cytometric analysis (Fig. (Fig.8).8). Four days postinfection, WTrec-infected spleens had significantly higher percentages of GR1high cells (neutrophils and inflammatory monocytes) than did mock-infected spleens. In contrast, spleens infected with C7rec, C8rec, and ΔyopM strains had significantly higher percentages of macrophages than did mock-infected spleens. Consistent with the histopathology findings (Table (Table4),4), there was more variation in the amount of GR1high cells in spleens of mice infected with C7rec and C8rec strains than in spleens of mice infected with the ΔyopM strain, making the influx, on average, more like that caused by WTrec. The spleens of WTrec-infected mice were not notably different from spleens of yopM mutant-infected mice in terms of size or weight (data not shown). Combined, these results suggest that expression of WT YopM dampens increases in macrophages, perhaps by preventing their recruitment, and that expression of WT YopM, and in some cases C7 and C8, increases recruitment of neutrophils compared to that for a strain lacking YopM.
Previous reports have demonstrated a depletion of NK cells in mice systemically infected with WT Y. pestis but not with a yopM mutant (25, 63). Therefore, we determined whether a similar effect occurred in mice i.v. infected with Y. pseudotuberculosis. Four days postinfection, the number of CD49bhigh CD3− NK cells present in WTrec-infected spleens had significantly decreased, on average, by 64% compared to mock-infected spleens (Fig. (Fig.9B),9B), but much less of a reduction was observed in mice infected with ΔyopM, C7rec, or C8rec strains. Since WTrec reaches 25-fold-higher levels of splenic colonization than do these yopM mutant strains, we wanted to differentiate between the possibilities of YopM-dependent and high-bacterial-load-dependent reduction of NK cells. To do this, mice were infected with ΔyopM, ΔyopE, and ΔyopH strains of IP2666pIB1 at various doses (between 10× and 100× the normal dose of 100 CFU) in an attempt to obtain higher colonization levels by these mutants, and numbers of splenic CD49bhigh CD3− NK cells and CFU were determined at days 2, 3, and 4 postinfection. For all three mutants, increased colonization significantly correlated with decreases in NK cells (Fig. 9C and D), indicating that YopM, YopE, and YopH were not necessary for this effect. Since CD49b expression can be downregulated on activated NK cells (1), NK cell activation was measured by monitoring the expression of the activation antigen, CD69, on the NK cells after infection with the WTrec, ΔyopM, ΔyopE, or ΔyopH strain. Greater decreases in numbers of CD49bhigh CD3− NK cells correlated with higher expression of CD69 on the NK cells (Fig. (Fig.9E).9E). This result supports the idea that the reduced detection of NK cells in mice infected with higher levels of bacteria was due to activation-dependent downregulation of CD49b in spleens of mice.
Pathogenic Yersinia bacteria produce a number of Yops with different biochemical functions and cellular targets which disable the host immune response. In general, deletion of a single Yop reduces the severity of infection but does not completely attenuate the pathogen, whereas deletion of all the Yops severely abrogates virulence, indicating that each Yop plays a unique role in infection (29, 36, 57). Each Yop has been studied in a variety of cell culture systems, and most Yops have multiple protein targets and several cellular phenotypes which could be important for some or all stages of infection (55). To understand the functions of each Yop during infection, a combination of cell biological, bacterial genetics, and animal infection models are required to determine whether a specific biochemical function, protein target, and cellular phenotype are relevant during infection. Previous studies have shown that YopM is an essential virulence factor in Y. pestis and Y. enterocolitica (25, 28, 37, 57) and that YopM localizes to the nucleus and cytoplasm in eukaryotic cells, binds to RSK1 in cultured HEK293 and J774.A1 cells, and increases the in vitro kinase activity of RSK1 (27, 32, 51). Here, we have detected two YopM-RSK1 complexes in infected macrophages (YMC1 and YMC2), defined sites of YopM that are required for its interaction with RSK1 in each of these complexes, and found that mutants unable to form detectable YMC1 complexes are attenuated for virulence in mice. In particular, the C-terminal tail of YopM is necessary for interaction with RSK1 and virulence, since alanine substitution of the last and penultimate set of three amino acids of YopM prevented both its ability to form detectable levels of YMC1 and YMC2 and its ability to promote colonization of spleens and lungs of mice. However, strains expressing YopM mutants that failed to bind detectably to RSK1 caused more pathology and inflammation than did a ΔyopM deletion mutant, suggesting that binding to RSK1 is not the only consequential action of YopM during infection. Combined, these data support the hypothesis that RSK1 is a relevant, but not the only, target of YopM during infection of mice. Alternatively, the C-terminal mutants may bind to RSK1 at low levels, which may in turn cause the intermediate pathology observed with these mutants.
Previous work demonstrated that YopM interacts with both RSK1 and PRK2 in HEK293 cells transfected with a FLAG-YopM fusion; however, PRK2 was not detected after immunoprecipitation of YopM in infected J774A.1 cells (32). While RSK1 was seen in both complexes in infected RAW 264.7 cells, PRK2 was not detected in either complex. The complete composition of these complexes is unknown, but based on their migration during native electrophoresis, the smaller complex could be composed of a monomer of YopM with RSK1 while the larger could be multimers of one of both proteins. Alternatively, both might contain one or more unidentified targets. In fact, YMC2 could consist of more than one complex, because multiple high-molecular-weight complexes may not be resolved on these gels. McDonald et al. used Nonidet P-40 (NP-40) to prepare lysates for coimmunoprecipitation of FLAG-YopM and potential targets in HEK293 cells (32); however, we observed that lysis with NP-40 disrupts the higher-molecular-weight complex in our native binding assay (data not shown). Thus, the possibility that there are novel targets in these complexes is consistent with the experiments carried out by these authors. Since no studies have yet identified putative target proteins from the nuclear compartment of cells and since our lysis method should release proteins from the nuclear compartment of cells, this remains an open area of investigation.
We found that deletion of LRRs 8 and 9, 10 and 11, 12 and 13, and 14 and 15 reduced the amount of YopM detected in the nuclei of macrophages, while deletions of the other LRRs or the 24 C-terminal amino acids had only modest effects on nuclear localization of YopM, suggesting that LRRs 8 to 15 are important for nuclear localization. Mutants with mutations in LRRs 8 to 15 also did not form detectable levels of YMC2, raising the possibility that the formation of the higher-molecular-weight complex might depend on translocation to the nucleus. Mutants C7 and C8 were detected in the nucleus even though they were unable to bind to RSK1, suggesting that YopM is not shuttled to the nucleus through its association with RSK1. Other studies have examined the requirements for nuclear localization of YopM by fusing YopM to reporter proteins and/or by detecting the presence of YopM proteins by microscopy (5, 52). Skrzypek et al. showed by indirect immunofluorescence that YopM mutants from Y. pestis lacking LRRs 4 to 7 or 7 to 10 were detected in the nuclei of infected HeLa cells and that the N-terminal half of YopM fused to enhanced GFP (EGFP; yEGFP-YopM ΔLRR8-end) localized to the nucleus of yeast (52). While these results appear to conflict with ours, immunofluorescence is not quantitative and our Y. pseudotuberculosis YopM mutants with deletions in LRRs 8 and 9 through 14 and 15 were detectable in the nuclear fraction of macrophages, albeit at much-reduced levels. In addition, there may be differences between nuclear trafficking of YopM in HeLa and yeast cells and that in macrophages and/or between that in Y. pseudotuberculosis- and that in Y. pestis-infected cells. Benabdillah et al. showed that trafficking of an EGFP-YopM fusion protein into the nucleus of transfected HEK293T cells was almost completely abolished by specific mutations within the last six amino acids of YopM to alanines (5). In addition, these authors found that the last 32 amino acids of YopM act as a nuclear localization sequence for LexA-AD and GFP fusion proteins in yeast, as well as an EGFP-LexA-AD fusion protein in transfected HEK293T cells. While our YopM C7 and C8 mutants were not identical to those of Benabdillah et al., C7 and C8 trafficked efficiently to the nuclei of macrophages, and ΔC did so only slightly less. In agreement with our data, Skrzypek et al. also found that the 32-residue tail of Y. pestis YopM was not sufficient to concentrate yEGFP in the nucleus of yeast cells and that YopM lacking the tail (yEGFP-YopM-NT) efficiently localized to the nucleus (52). Finally, if the C terminus of YopM were sufficient to direct YopM to the nucleus, one would expect that the mutants with deletions in LRRs 8 and 9, 10 and 11, 12 and 13, and 14 and 15 would traffic to the nuclei. The failure of these mutants to efficiently localize to the nucleus indicates that in the context of cell culture infection, the C terminus of YopM is not sufficient to promote normal levels of nuclear localization.
An important conclusion that can be drawn from the fitness of the Δ6-7rec, Δ8-9rec, and Δ10-11rec mutants is that YopM can undergo a decrease of two LRRs in length without being affected in its contribution to colonization of the tissues that we have studied here. Results from the work of Hines et al. support this conclusion. When LRRs 4 to 7 and LRRs 8 and 9 were deleted, YopM could be cross-linked to thrombin. However, if LRRs 4 to 9, 6 to 9, or 7 to 10 were deleted, cross-linking was reduced (Δ6-9) or abolished (Δ4-9 and Δ7-10), and a deletion of 3 LRRs resulted in a 1,000-fold increase in the LD50 of Y. pestis after i.v. infection of mice (23).
Deletion of LRRs 8 and 9, 10 and 11, 12 and 13, or 14 and 15 disrupted YMC2 and reduced nuclear localization of YopM; however, Δ8-9, Δ10-11, and Δ12-13 had no effect on colonization of Y. pseudotuberculosis. The simplest conclusion from these data is that YMC2 and wild-type levels of nuclear localization are not essential for YopM's contribution to tissue colonization and virulence. Alternatively, YMC2 formation by Δ8-9, Δ10-11, and Δ12-13 might occur at an undetectable but sufficient level for tissue colonization, or colonization 4 days postinfection may be the wrong readout for detecting a phenotype of these mutants; time-to-death or LD50 assays for virulence might reveal differences between these strains and the wild type. An observation which lends support to the possibility that Δ8-9 and Δ10-11 mutations did attenuate the virulence of Y. pseudotuberculosis in other ways is that the mice infected with the Δ6-7rec strain, which expresses a form of YopM that can take part in both complexes with RSK1, consistently showed greater signs of pathology, including a more scruffy appearance, bleeding in various organs, and cases of premature death, than did those infected with the Δ8-9rec or Δ10-11rec strain.
The histopathological and flow cytometric results for the C-terminal mutants also highlight the fact that a phenotype in one assay does not necessarily describe all aspects of a mutant. Intriguingly, infection with the C7rec and C8rec strains caused more pathology and inflammation in spleens than did infection with the yopM deletion strain, although the bacterial loads were similar. This suggests that YopM may have multiple functions in cells in the spleen and that the C7 and C8 mutants retain some of this activity. Alternatively, YopM itself may trigger an immune response regardless of its ability to interact with RSK1, or the C-terminal mutants may retain RSK1 binding activity which was below our limit of detection but which was sufficient to cause increases in pathology. In either case, the WTrec strain caused an increase in the percentage of GR1high cells in the spleens of infected animals, and this change was partially dependent on YopM. While the total numbers of splenocytes were not determined, spleens from WTrec-infected mice were not different in size or weight from spleens from yopM mutant-infected mice, and so it is likely that the changes in percentages of GR1high cells reflect changes in total numbers of GR1high cells. This is an interesting result in light of the recent finding that a yopM mutant of Y. pestis was rescued by depletion of GR1+ cells (63). With this in mind, one interpretation of our results is that WT Y. pseudotuberculosis can tolerate higher levels of neutrophil recruitment because YopM functions to inhibit neutrophil functions.
Our flow cytometry results revealed out that while infection with wild-type Y. pseudotuberculosis results in a decrease of CD49bhigh CD3− NK cells, this effect is not dependent on YopM, since mice highly colonized with a ΔyopM mutant also showed a decrease in NK cells. Decreased numbers of NK cells were also observed in mice highly infected with ΔyopE and ΔyopH mutants, adding further support to the notion that high bacterial CFU counts, similar to WT-like levels, are the cause of the depletion of NK cells. This result displays how the uncoupling of bacterial burden from the effect of removal of proteins can reveal more about the direct functions of proteins by removing the secondary effects of changes in colonization levels. However, these results do not rule out the possibility that YopM contributes to a decrease in NK cells during Y. pestis infection and/or in C57BL/6 mice, as previously reported (25, 63).
This study also reveals some of the differences between the Y. pseudotuberculosis strains YPIII and IP2666. After oral infection, wild-type IP2666 grew to slightly higher levels than did wild-type YPIII in all tissues examined, and differences between the wild-type and yopM strains were more apparent in the less virulent YPIII strain (43). An overall reduction in virulence of YPIII could render YopM more valuable for survival of YPIII. Another difference between the strains was that the IP2666 strain was consistently found in the spleen and liver of orally infected mice, whereas the YPIII strain often did not reach these systemic tissues.
In conclusion, the demonstration that YopM binding with RSK1 correlates with virulence in mice should prove useful in the continuing goal of understanding the mechanism by which YopM works. Future experiments include identifying the differences between the two YopM/RSK1 complexes, analysis of the downstream effects of complex formation, and further characterization of the cell types affected by the wild-type and yopM mutant strains during infection.
We thank Sam Miller and Marie Pierre Blanc for assistance in the design and construction of several of the YopM LRR deletion constructs. We thank members of the Mecsas lab and the Yersinia Research Group at Tufts Medical School for useful discussion and critical reading of the manuscript.
This work was supported by NIH A1056058 and AI073759 awarded to J.M. M.W.M. was supported by T32GM07310; M.L.M. was supported by T32AI007077.
Editor: A. J. Bäumler
Published ahead of print on 5 April 2010.