Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mol Microbiol. Author manuscript; available in PMC Nov 26, 2009.
Published in final edited form as:
PMCID: PMC2783611
Shigella flexneri T3SS effectors OspB and OspF target the nucleus to down-regulate the host inflammatory response via interactions with retinoblastoma protein
Daniel V. Zurawski,1 Karen L. Mumy,2 Christina S. Faherty,1 Beth A. McCormick,2 and Anthony T. Maurelli1*
1 Department of Microbiology and Immunology, Uniformed Services University of the Health Sciences, Bethesda, Maryland
2 Mucosal Immunology Laboratory, Massachusetts General Hospital, Charlestown; and Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, Massachusetts
* Corresponding author, Uniformed Services University of the Health Sciences, Bethesda, Maryland, Department of Microbiology and Immunology, 4301 Jones Bridge Rd., Bethesda, MD 20814-4799, Telephone: 301-295-3415, Fax: 301-295-1545, amaurelli/at/
Present address: Division of Bacterial and Rickettsial Disease, Walter Reed Army Institute of Research, Rm#3S29, 503 Robert Grant Ave., Silver Spring, MD 20910-7500, Daniel.Zurawski/at/
OspF, OspG, and IpaH9.8 are type III secretion system (T3SS) effectors of Shigella flexneri that down-regulate the host innate immune response. OspF modifies mitogen-activated protein kinase (MAPK) pathways and polymorphonuclear leukocyte (PMN) transepithelial migration associated with Shigella invasion. OspF also localizes in the nucleus to mediate chromatin remodeling resulting in reduced transcription of inflammatory cytokines. We now report that OspB can be added to the set of S. flexneri T3SS effectors required to modulate the innate immune response. T84 cells infected with a ΔospB mutant resulted in reduced PMN transepithelial migration and MAPK signaling. Tagged versions of OspB localized with endosomes and the nucleus. Further, T84 cells infected with the ΔospB mutant showed increased levels of secreted IL-8 compared to wild-type infected cells. Both GST-OspB and GST-OspF co-precipitated retinoblastoma protein (Rb) from host cell lysates. Because ΔospB and ΔospF mutants share similar phenotypes, and OspB and OspF share a host binding partner, we propose that OspB and OspF facilitate the remodeling of chromatin via interactions with Rb resulting in diminished inflammatory cytokine production. The requirement of multiple T3SS effectors to modulate the innate immune response correlates to the complexity of the human immune system.
Shigella species are Gram-negative, facultative intracellular enteric bacteria that specifically infect and cause disease in the large intestine of human hosts. Shigella spp. are the causative agent of bacillary dysentery (shigellosis) with an estimated 165 million cases per year leading to approximately 1.1 million deaths worldwide (Kotloff et al., 1999). The majority of infections occur in countries with inadequate health care and unsafe food and water supplies; however, Shigella outbreaks are a significant problem for industrialized nations as well. For example, the United States has an annual incidence rate of 5.6 infections per 100,000 people, with the majority of infections found in child daycare facilities (Gupta et al., 2004). The persistence and prevalence of shigellosis worldwide underscores the importance or studying of Shigella pathogenesis, especially given the increase in antibiotic resistance, the potential to be used as a bio-weapon, and the lack of an efficacious vaccine (Moran, 2002; Pickering, 2004; Levine et al., 2007).
Shigella flexneri enter the human host by ingestion of contaminated food or water. After passage through the stomach, and small intestine, S. flexneri invade epithelial cells of the large intestine only at the basolateral membrane. Therefore, access to the submucosa is a requirement for full virulence. S. flexneri reach the submucosa via three separate mechanisms: transcytosis through M cells, S. flexneri-induced tight junction disruption, or S. flexneri-induced polymorphonuclear (PMN) leukocyte transepithelial migration. The latter two mechanisms create a breach in the colonic epithelium providing entry to the submucosa (Perdomo et al., 1994; McCormick et al., 1998a; Sakaguchi et al., 2002).
Shigella flexneri harbors a large virulence plasmid which encodes a Type III secretion system (T3SS) and the majority of T3SS-secreted effectors responsible for S. flexneri pathogenesis. These effectors include the Ipa proteins (IpaA-D) that facilitate bacterial uptake via a macropinocytosis-like mechanism resulting with the invading bacterium residing in a vacuole. S. flexneri escape from the vacuole to avoid autophagy and reach the cytoplasm where they replicate. Once in the cytoplasm, S. flexneri use IcsA, an outer membrane protein, to recruit actin and initiate actin polymerization, which promotes the spread of the bacteria into adjacent cells (reviewed in Schroeder and Hibli, 2008).
A second set of T3SS effector proteins, the Osp (OspB-OspG) and IpaH proteins, play a complementary role to invasion which is to modulate host cell signaling pathways and host cell transcription. A common function shared by most of these proteins is the post-invasion manipulation of the host inflammatory response. For example, the IpaH and OspG proteins modify signaling pathways by targeting inflammatory signaling molecules such as NF-κB for ubiquitination and degradation (Rohde et al., 2007; Kim et al., 2005). IpaH9.8 also localizes in the nucleus where it reduces IL-8 transcript levels by interfering with U2AF35, a transcription factor required to splice IL-8 mRNA (Okuda et al., 2005). Another effector, OspF, also localizes in the host nucleus (Zurawski et al., 2006) and plays a role in remodeling host chromatin, which leads to a reduction of IL-8 transcripts (Arbibe et al., 2007). Specifically, OspF is required for the dephosphorylation/deacylation of histone H3; however, it is unable to dephosphorylate histone H3 in vitro (Arbibe et al., 2007) suggesting that other proteins (bacterial or host) are required to mediate this process. In a separate, but possibly linked activity, OspF acts as a lyase that irreversibly dephosphorylates mitogen-activated protein kinases (MAPK) in the nucleus of the host cell, where this enzymatic activity also reduces the host inflammatory response (Arbibe et al., 2007; Li et al. 2007, Kramer et al., 2007; Zhu et al., 2007).
Conversely, OspF is required for the full induction of PMN transepithelial migration mediated by S. flexneri infection (Zurawski et al. 2006). S. flexneri infection of polarized, colonic epithelial cells activates extracellular signal-regulated kinase (ERK1/2) signaling and this activation is required for efficient PMN migration (Kohler et al. 2002). Recent results have linked mitogen-activated extracellular kinase/ERK kinase (MEK/ERK) pathway activation to the activation of cPLA2, which leads to the production of hepoxilin A3, a potent PMN chemoattractant secreted apically by epithelial cells (McCormick 2007; Mumy et al., 2008). Since an ΔospF mutation results in decreased ERK1/2 phosphorylation and PMN transepithelial migration (Zurawski et al., 2006), OspF must have another function in the cytoplasm to drive the phosphorylation of ERK1/2, separate and opposite from the lyase activity in the nucleus. In support of this model, MEK, the kinase directly upstream of ERK1/2 in the MEK/ERK pathway, has reduced phosphorylation in cells infected with the ΔospF mutant (Kramer et al. 2007). Activation of MEK/ERK pathway is a property shared with OspC1 (Zurawski et al. 2006) and the recently characterized OspZ (Zurawski et al. 2008). Therefore, the activation of the MEK/ERK pathway and the induction of PMN transepithelial migration requires multiple T3SS effectors and cannot be solely attributed to lipopolysaccharide (LPS), which has also been shown to contribute to ERK1/2 activation (Kohler et al., 2002).
Both ospF and ospC1 are regulated by VirB and MxiE, transcriptional activators that are induced when Shigella are grown at 37°C and inside host cells, respectively (Marvis et al., 2002; Kane et al., 2002; Le Gall et al. 2005). OspB, another T3SS effector (Buchrieser et al., 2000; Santapaola et al., 2006), is also co-regulated by VirB and MxiE (Le Gall et al., 2005); however, the function of OspB with respect to S. flexneri pathogenesis has not been determined. In this study, we generated an ospB deletion mutation which allowed us to define roles for OspB during S. flexneri infection. We discovered that ospB is required for efficient PMN migration similar to ospF, ospC1 and ospZ. Tagged OspB, secreted from S. flexneri or ectopically expressed, localizes in the host nucleus. Lastly, we determined that OspB, along with OspF, down-regulates the production of IL-8 at later time points of infection of epithelial cells and shares a common association with retinoblastoma protein (Rb). Since Rb is known to recruit factors which modify histones (Macaluso et al. 2006), we propose that Rb is one of the missing components linking OspF to Shigella-induced chromatin remodeling.
OspB has homologues in other Gram negative bacteria
OspB is encoded on the virulence plasmid of S. flexneri in an operon that includes phoN2, which encodes apryase, an enzyme required for the proper localization of IcsA (Santopaola et al., 2002; Santopaola et al., 2006). The ospB-phoN2 operon is >100 kilobases from the ipa-mxi-spa pathogenicity island that encodes the T3SS and its secreted effectors (Buchrieser et al., 2000). This separation from the major cluster of virulence genes and the fact that the operon is flanked by a set of IS elements suggests that these genes were acquired via horizontal transfer (Santopaola et al., 2002). Therefore, we searched the database of microbial sequences utilizing the BLAST search engine for other Gram-negative bacteria species that contain OspB homologues since Gram-negative pathogens often share T3SS effectors via horizontal gene transfer (Brown et al. 2006). As expected, all Shigella spp. have ospB. However, the annotated start codon for S. dysenteriae serotype 1 is an alternative start ~150 bp distal to the start codon annotated in the other Shigella spp. (Fig. 1). Further analysis revealed that the open reading frame (orf) extends farther upstream in the S. dysenteriae nucleotide sequence. The corresponding encoded peptide matches the N-terminus of OspB found in the other Shigella spp., and also the N-terminal sequenced from the secreted protein from S. flexneri 5a (Buchrieser et al., 2000). Since translation from the upstream start codon of OspB in the S. dysenteriae sequence would yield an OspB that matches what is found in the other Shigella spp., we conclude that the current annotated start codon of the S. dysenteriae ospB sequence in GenBank is incorrect. Taking this into consideration, the homology of OspB among all Shigella spp. is 99% conserved (Fig. 1).
Fig. 1
Fig. 1
OspB and its homologues. Sources (GenBank accession numbers) are in parentheses. Sequence from S. flexneri 2a 2457T (our sequence), S. dysenteriae (YP406222), S. sonnei (YP313493), S. boydii (YP406371), S. Paratyphi A (NC_006511), V. cholerae 1587 (ZP_01947905), (more ...)
The BLAST search also revealed OspB homologues in other Gram-negative bacteria that possess T3SS. OspB homologues are present in both Salmonella enterica serovar Typhi and Paratyphi A. The S. Typhi protein is truncated and shares homology only with 47 amino acids (44 - 91) of OspB from Shigella (data not shown). However, S. Paratyphi A appears to have a complete copy of OspB sharing 32% identity and 52% similarity to the S. flexneri protein (Fig. 1). OspB homologues were also found in Vibrio parahaemolyticus (31% identity 50% similarity) and Vibrio cholerae O12 strain 1587 (33% identity, 51% similarity), a strain isolated from a cholera outbreak in Peru (Dalsgaard et al. 1995). Recent studies have shown that both V. parahaemolyticus and non-O1, environmental strains of V. cholerae such as V. cholerae 1587 have functional T3SS (Makino et al., 2003; Kodama et al., 2007; Tam et al., 2007); therefore, it is possible these Vibrio species also secrete their OspB homologue.
OspB localizes to endosomes and the nucleus of host cells
Although OspB is a T3SS effector protein secreted by S. flexneri (Buchrieser et al., 2000; Santapaola et al., 2006), its function is unknown. Previously, we used two-hemagglutinin (2HA) tagged versions of OspF and OspC1, to demonstrate protein localization inside the host cell (Zurawski et al., 2006). Therefore, we made an OspB-2HA fusion plasmid (pDZ8) and transformed it into an ospB deletion mutant, ΔospB (BS818) that was generated by allelic exchange. This plasmid was also transformed into a Δspa47 mutant (BS652). Spa47 is the ATPase required for a functional T3SS of S. flexneri (Jouihri et al., 2003), and this genetic background is useful for testing the T3SS-dependent secretion of putative effectors (Zurawski et al., 2006; Zurawski et al., 2008). After one hour of exposure to Congo red to induce the S. flexneri T3SS, OspB-2HA was found in both the whole cell and supernatant fractions (Fig. 2A). In contrast, no signal was detected in either the untransformed 2457T (wild type) or in the supernatant of the Δspa47 mutant (Fig. 2A). These results confirm that OspB is secreted by the S. flexneri T3SS (Santapaola et al., 2006) and that the C-terminal 2HA-tagged construct did not impair its secretion.
Fig. 2
Fig. 2
OspB-2HA is secreted by S. flexneri and localizes to the nucleus. A. 2457T, ΔospB/pOspB-2HA, and Δspa47/pOspB-2HA whole cell fractions and supernatant were prepared after one hour of induction with Congo red. Samples were run on a 13% (more ...)
We infected a semi-confluent monolayer of HeLa cells with S. flexneri containing the OspB-2HA plasmid. After four hours of invasion, we fixed the infected cells and detected OspB-2HA using indirect immunofluorescence with anti-HA antibody and secondary antibody coupled to Alexa Fluor 488. The anti-HA signal was localized primarily inside the host nucleus (Fig. 2B) with a lesser amount also observed in the cytoplasm. Cells infected with wild-type 2457T (no OspB-2HA plasmid) did not display anti-HA signal above background levels and served as a negative control (Fig. 2B).
To confirm the localization of OspB in the nucleus, we ectopically expressed full-length OspB along with N-terminal and C-terminal truncations fused to green fluorescent protein (GFP) at the N-terminus of the OspB polypeptide (Fig. 3A). A semi-confluent monolayer of HeLa cells was transiently transfected with the GFP-OspB producing plasmids and cells were fixed after 16 hours. Two distinct localization patterns in the host cell were observed: punctate staining in the cytoplasm of transfected cells (Fig. 3B), and localization in the nucleus (Fig. 3B). The nuclear localization was confirmed with DAPI staining (Fig. 3C).
Fig. 3
Fig. 3
Fig. 3
Ectopic expression of GFP-OspB in HeLa cells. A. Diagram showing the three GFP-OspB fusion contructs. B. Semi-confluent monolayers of HeLa cells were transiently transfected with a mammalian expression vectors encoding GFP-FL-OspB, GFP GFP-N-OspB, GFP-C-OspB, (more ...)
We suspected that the punctate pattern of GFP-OspB staining in the cytoplasm represented localization to endosomes based on the size of the compartment and because S. flexneri come in contact with the endosomal pathway during infection (Ogawa and Sasakawa, 2006). Transfection experiments were repeated with the GFP-OspB expressing plasmids and, after fixation, the transfected cells were counter-stained with an antibody against Rab5, a marker for early endosomes. As expected, some of the GFP-OspB signal co-localized with the anti-Rab5 staining (Fig. 3D, merged yellow signal), however there was also a smaller proportion of GFP-OspB signal in the cytoplasm in undefined compartments. We examined and quantitated the localization for each construct in three separate transfection experiments by assessing the signal localization in 300 cells per experiment (Fig. 3E).
From the GFP-tagging experiments we conclude the N-terminus of OspB is responsible for targeting the protein to the nucleus and early endosomes. The C-terminal fusion to GFP was found predominantly in the cytoplasm and not localized to a specific cellular compartment (Fig. 3B). In contrast, the N-terminal fusion resembled the full-length GFP-OspB construct (Fig. 3B). It should also be noted that unlike the GFP alone control which had equal amounts of signal in the nucleus and cytoplasm, the GFP-FL-OspB and the GFP-N-OspB appeared to concentrate in the nucleus over time (Fig. 3B, 3C). We searched for a nuclear localization signal (NLS) in the N-terminus of OspB using predictNLS (Cokol et al., 2000) and WoLF PSORT (Nakai and Horton, 1999), but no canonical NLS was found. Therefore, we hypothesize OspB has either a non-canonical NLS or interacts with a host protein that targets the nucleus after its expression in the cytoplasm.
Virulence phenotypes of the ΔospB mutant
Localization of OspB to the nucleus is a property previously reported for other Osp proteins (Zurawski et al., 2006). Therefore, we next determined if a ΔospB mutant also behaved similarly to the other S. flexneri osp deletion mutants in assays that assess Shigella virulence. When the ΔospB mutant (BS818) was compared to 2457T in invasion and plaque assays, no discernible difference was seen between the mutant and wild-type parent (data not shown), as was shown previously with a ΔospB mutant in S. flexneri 5a (Santapaola et al., 2006). The ΔospB mutant, along with ΔospF and ΔospC1 mutants, was also tested in an apoptosis protection assay, which was shown to be dependent on one or more mxiE-regulated genes (Clark and Maurelli, 2007). None of these osp genes were required for host cell survival in staurosporine-treated cells.
PMN transepithelial migration, which is required for S. flexneri virulence (Perdomo et al., 1994), can be measured using an in vitro tissue culture assay (McCormick et al., 1998a) and can detect subtle differences in mutants of T3SS effectors when compared to wild-type (Zurawski et al., 2006; Zurawski et al., 2008). Therefore, the ΔospB mutant was tested in the PMN migration assay. Mutant and wild-type strains of S. flexneri were used to basolaterally infect polarized T84 monolayers. The infection with the ΔospB mutant resulted in a statistically significant (P < 0.001) ~60% reduction of PMN transepithelial migration (Fig. 4) when compared to wild-type infection (normalized to 100% ± 7.3%). A result that was similar to what was observed with the ospF and ospC1 mutants (Fig. 4). The PMN migration defect could be complemented by transforming the ΔospB mutant with pOspB-2HA (pDZ8) (Fig. 4). Since, the complementation only reached ~80% of wild-type S. flexneri infection, it is possible that the 2HA-tag may partially interfere with OspB folding or function although secretion appears unimpaired (Fig. 2A). It is also possible that the amount of OspB generated by the pOspB-2HA plasmid was not sufficient to complement the ΔospB mutant or that the parental ΔospB mutant background may influence the secretion or function of other effectors required for PMN migration.
Fig. 4
Fig. 4
OspB is required for efficient Shigella-induced PMN transepithelial migration. T84 polarized monolayers were infected basolaterally with 2457T (wild type), BS103 (plasmid-cured, non-invasive 2457T), ΔospB, ΔospB + pOspB2HA, ΔospC1 (more ...)
Since the infections using the single deletion mutants of ΔospB, ΔospF, and ΔospC1 all led to a reduction of PMN migration, we speculated that OspB, OspC1, and OspF may work together to modulate signaling in the host, either at different steps in the same signaling pathway or entirely different signaling pathways each of which feed into the cPLA2 pathway which is required for neutrophil migration (Mumy et al., 2008). To address these hypothesis two different double deletion mutants (ΔospBF, ΔospC1F) were generated and compared to infection with wild-type 2457T and BS103, a strain that lacks the virulence plasmid and induces only minor PMN migration above buffer alone controls (Maurelli et al., 1984; McCormick et al., 1998a). Interestingly, there were diverse results. First, an infection with the ΔospBF double mutant drastically reduced PMN migration to nearly that of BS103-induced levels (Fig. 4). Surprisingly, the other double mutant ΔospC1F did not display a further reduction in PMN transepithelial migration compared to infection with the single mutants; however, all of the mutants induced significantly less PMN transepithelial migration than 2457T (Fig. 4).
Infection with the ΔospB mutant displays reduced ERK1/2 signaling
S. flexneri osp deletion mutants deficient in PMN transepithelial migration are also deficient in the activation of the MEK/ERK pathway (Zurawski et al. 2006; Zurawski et al. 2008). These results correlate with the requirement of ERK1/2 signaling for the generation of hepoxilin A3 leading to the PMN migration associated with S. flexneri infected cells (Kohler et al., 2002, Mumy et al., 2008). ERK1/2 phosphorylation, a marker of MEK/ERK pathway activation, was evaluated in polarized T84 cells infected with the ΔospB mutant. T84 cells were basolaterally infected with 2457T or ΔospB and ERK1/2 phosphorylation was evaluated by Western blot after 45 and 90 minutes using a phospho-specific antibody that recognizes the activated form of ERK1/2. A blot for total ERK1/2 protein was done in parallel as a loading control. There was a reduction in ERK1/2 phosphorylation after 90 minutes of infection in T84 cells infected with ΔospB compared to 2457T (Fig. 5). Densitometric analysis indicated that T84 cells infected with ΔospB had 15% less (45 min) and 30% less (90 min) ERK1/2 activation compared to wild-type infected cells. Therefore, we conclude that OspB plays a role in activating a host inflammatory pathway(s), and specifically, the MEK/ERK pathway required for PMN transepithelial migration.
Fig. 5
Fig. 5
ERK1/2 phosphorylation is reduced in cells infected with a S. flexneri ΔospB mutant. T84 polarized monolayers were infected basolaterally with wild-type 2457T or ΔospB (BS818). Cells were lysed at each time point, boiled, and run on a (more ...)
Evaluation of osp mutants in the Serény test
Having established that ospB is important for S. flexneri-induced PMN migration, the ΔospB, ΔospBC1, ΔospBF, ΔospC1F mutants were compared to the wild-type parent 2457T using the Serény test, one of the few animal infection models suitable for S. flexneri (Serény, 1957). Three guinea pigs were infected with each mutant or 2457T at a dose of 2.0 × 108 CFU in one eye of each animal. Symptoms were monitored and scored based on double-blinded observations over the course of four days. There was no difference between ΔospB and 2457T infections (Table 1). However, a marked delay was observed in the onset of swelling and inflammation with the double mutants, and particularly, in the guinea pigs infected with the ΔospBF double mutant (Table 1). Interestingly, by day 2 infections with ΔospBF displayed wild-type symptoms (Table 1) suggesting rapid inflammation occurred after the initial delay that was observed. The inflammation continued to rise at a faster rate than that of wild-type infection the following two days, resulting in a higher amount of swelling and conjunctiva destruction by day 4 compared to infection with 2457T (Table 1). In contrast, inflammation due to infection with the ΔospC1F double mutant never reached that of wild-type (Table 1). A similar delay and decrease in inflammation was observed previously with the ΔospC1 single mutant (Zurawski et al., 2006). The infection with ΔospBF double mutant was the first time we observed significant differences in the Serény reaction with strains harboring an ospB or ospF deletion mutation, since the single ΔospF mutant (Zurawski et al., 2006) and the single ΔospB mutant (this study) had the same results as a wild-type infection. These results suggest a role for the OspB and OspF proteins, together, on the first day of infection with respect to S. flexneri virulence, and this result correlates to the severe drop in PMN migration that is induced by the ΔospBF double mutant.
Table 1
Table 1
Phenotypes of the osp mutants in the Serény test
OspB and OspF reduce the amount of IL-8 secreted by epithelial cells
The ΔospB and ΔospF mutants had similar phenotypes in virulence assays, both OspB and OspF localized in the nucleus, and the double mutant showed the most significant delay in host inflammation followed by a higher amount of swelling and conjunctiva destruction at later time points during the Serény test. Based on these observations, we hypothesized that OspB was initially required for a pro-inflammatory response at early time points of infection, but then, OspB must also play an anti-inflammatory role at later time points. Such a property of OspB function would explain the increase of the Serény response observed at day 3 and beyond with the ΔospBF double mutant. It would also fit a model similar to OspF, which has these same phenotypes (Zurawski et al., 2006; Arbibe et al., 2007, Kramer et al., 2007).
Polarized T84 cells were infected with 2457T, ΔospB, ΔospF, and ΔospBF, and the amount of IL-8 secreted into the basolateral compartment was measured by ELISA. 2457T infection generated a 40-fold increase of IL-8 production over uninfected cells (HBSS, buffer alone) after 180 min of infection (a time point that matches infection time during the PMN migration assay). Cells infected with ΔospF displayed greater amounts of IL-8 secretion in comparison to wild-type infection, as was shown in a previous study (Arbibe et al., 2007) (Fig. 6). Interestingly, the ΔospB mutant resulted in a similar amount of increased IL-8 secretion indicating that OspB also plays a role in down-regulating the inflammatory response (Fig. 6). Infection with the ΔospBF double mutant resulted in even more IL-8 release than infection with the ΔospF single mutant (P = 0.01) and wild-type (P < 0.01) suggesting that the proteins may work independently to down-regulate IL-8 production (Fig. 6). Not all osp mutants showed a similar phenotype. Levels of IL-8 secreted from polarized T84 cells infected with ΔospC1 and ΔospZ mutants were equivalent to those from cells infected with wild-type bacteria (data not shown). Even though these effector proteins localize in the nucleus, OspC1 and OspZ do not appear to modulate IL-8 levels.
Fig. 6
Fig. 6
OspB reduces the amount of IL-8 secreted by infected epithelial cells. T84 polarized monolayers were infected basolaterally with 2457T, ΔospB, ΔospF, and ΔospBF. HBSS represents uninfected monolayers (buffer only). After four hours (more ...)
Again, these experiments suggest a duel role for the OspB and OspF T3SS effector proteins in Shigella pathogenesis. First, these proteins appear to activate the MEK/ERK pathway in the cytoplasm to induce PMN migration, but there is a second function where infections with the single ΔospB mutant, the single ΔospF mutant and the ΔospBF double mutant resulted in significant increases in IL-8 levels compared to wild-type infections (Fig. 6). We wanted to further explore the potentially-linked relationship in the host cell.
GST-OspB and GST-OspF associate with Retinoblastoma protein
Because OspB and OspF appeared to have common functions, we hypothesized that OspB and OspF may share a host binding partner. We took a directed approach to identify host binding partners based on the localization and the amino acid sequences of OspB and OspF. First, we used a protein prediction server, Eukaryotic Linear Motif (ELM), which identifies short linear domains in amino acid sequences that are known to mediate protein-protein interactions or that are subject to post-translational modifications (Puntervoll et al., 2003). OspF had ~10 predicted sites for protein interactions. Interestingly, one of the predicted sites was a MAPK binding site in the N-terminal portion of OspF (Fig. S1).
RFP-OspF co-localizes with microtubules (Zurawski et al., 2006). Since ERK1/2 interacts with microtubules in a complex with MEK and Raf, the kinases upstream of ERK1/2 in the MEK/ERK pathway (MacCormick et al., 2005), we hypothesized OspF may interact with ERK1/2 as well. We discovered that OspF interacted directly with ERK2 and dephosphorylates the active form of ERK2 (data not shown), a result which is now consistent with other reports (Arbibe et al., 2007; Kramer et al., 2007; Li et al., 2007). Recently, it was also shown that the site predicted by the ELM server is required for the OspF and MAPK interaction (Zhu et al., 2007). Taken together, these results validate the use of the ELM server to predict protein-protein interactions.
When analyzing OspF lyase activity to remove phosphate groups, we discovered that when MEK and OspF were present at a 1:1 molar ratio in the same in vitro reaction, we still observed ERK2 phosphorylation (Fig. S2A). However, when the concentration of OspF was increased to ≥ 2:1 molar ratio or if no MEK was added to the reaction, we observed dephosphorylation of ERK2 (Fig. S2A). To confirm that MEK kinase activity can outcompete OspF activity, the MEK inhibitor PD98059 was used to inhibit MEK kinase activity. In this background, OspF actively removed phosphate groups on ERK2 as opposed to the untreated control (Fig. S2B). These results suggest that the concentration of OspF has to be higher than the concentration of endogenous host MEK in the host cytoplasm or be in a different compartment separate from MEK (i.e. the nucleus) to efficiently remove the phosphate groups from activated MAPK. However, it is also possible that only half or less than half of the purified OspF is active in vitro.
Since OspF interacts directly with ERK2 (Li et al., 2007; Zhu et al., 2007) and co-localizes with microtubules in vivo (Zurawski et al., 2006), we speculated that OspF may also interact directly with microtubules like the members of the MEK/ERK complex. We found that GST-OspF precipitated β-tubulin from HeLa lysates (Fig. 7A) which suggested an interaction between OspF and microtubules. To determine if the interaction was the result of direct binding, we combined purified, pre-formed microtubules with GST and GST-OspF in vitro and discovered that GST-OspF could precipitate the purified microtubules while the GST alone control did not precipitate these microtubules to the same degree (Fig. 7B). These results suggest that in the cytoplasm of the host cell OspF is most likely in a protein complex with Raf, MEK, ERK1/2, and microtubules (MacCormick et al., 2005) since it directly interacts with two of the four complex components.
Fig. 7
Fig. 7
GST-OspF interacts directly with microtubules. A. GST-OspF and GST were mixed in equal amounts with HeLa lysate and precipitated using glutathione beads. Beads were washed and resuspended in SDS-PAGE sample buffer, boiled, and run on 12% SDS-PAGE gel. (more ...)
The microtubule interaction, while direct, is not likely significant with respect to reducing the host inflammatory response. Instead, the microtubule interaction, along with ERK1/2 interaction, may be required to drive OspF to the nucleus where it performs its anti-inflammatory function. Another Shigella T3SS effector, IpaH9.8, has similar properties. IpaH9.8 is an E3 ubiquitin ligase that targets proteins for destruction in the cytoplasm (Rhode et al., 2007), but it also has a second function in the nucleus to reduce IL-8 transcript (Okuda et al., 2005), which requires microtubules for its nuclear localization (Toyotome et al., 2001). Therefore, we searched for other host binding partners to OspF that were relevant to its chromatin remodeling function and the possible link to OspB.
Another interaction site predicted in the OspF sequence by the ELM server was the Rb binding site (LXCXE). OspF contains the sequence IMCLE located from amino acids #178-182 in its C-terminus (Fig. S1). A potential interaction with Rb was of particular interest because proteins that associate with Rb can promote histone modification (Macaluso et al., 2006), an OspF-dependent phenotype observed during S. flexneri infection (Arbibe et al., 2007). To determine if OspF interacts with Rb, we combined bacterial lysates containing GST-OspF with HeLa cell lysates, and added glutathione beads to precipitate GST-OspF from the lysates along with any associated proteins. Bacterial lysates with GST-OspB, and GST alone were used as controls. As predicted by the ELM server, GST-OspF precipitated Rb from HeLa cell lysates (Fig. 8A) suggesting that the proteins interact. GST alone did not precipitate Rb from HeLa lysate. However, although no Rb binding site was found in the OspB amino acid sequence (ELM server), GST-OspB also precipitated Rb (Fig. 8A). To determine if this was a direct interaction, equal amounts of GST-OspF and GST-OspB were mixed in vitro with equal amounts of a purified MBP-Rb C-terminal fusion (Rb amino acids #701-928) and the precipitation with glutathione beads was repeated. The C-terminus of Rb is known to interact with the LXCXE binding site (Macaluso et al., 2006). Under these conditions, GST-OspB did not precipitate pure Rb suggesting that the interaction was not due to direct binding or that OspB may interact with the N-terminus of Rb. However, GST-OspF did precipitate the pure Rb indicating that the interaction between OspF and Rb is a result of direct binding, and most likely, the LXCXE site was required for this interaction (Fig. 8B).
Fig. 8
Fig. 8
OspB and OspF associate with Rb. A. Equal amounts of bacterial lysate containing GST, GST-OspF, or GST-OspB were mixed with HeLa lysate. Subsequently, GST and the GST-Osp fusion proteins were precipitated with glutathione beads. Samples were washed, resuspended (more ...)
To further verify the OspF/Rb interaction, we made GST fusions to N-terminal (N), middle (M), and C-terminal (C) domains of OspF (Fig. S1). The N-OspF domain (aa #1 – 83) contains the predicted MAPK binding site, while the C-OspF domain (aa #152 – 239) has the predicted Rb binding site. As a control to show our constructs were functional, we mixed the GST fusions with pure ERK2. As predicted and demonstrated previously (Zhu et al.; 2007), the N-terminal OspF domain bound to pure ERK2 (Fig. S3). The experiment was repeated using the pure MBP-Rb C-terminal fusion. The C-terminal OspF domain bound to pure Rb as predicted (Fig. 8B). These results suggest that OspF associates directly with Rb via the binding site predicted by the ELM server in the C-terminus.
To verify that the LXCXE was responsible for the Rb interaction, we generated a site-directed mutation to this site in OspF. We changed three nucleotides in the OspF sequence with site-directed mutagenesis that in turn changed the amino acids of the Rb pocket binding site IMCLE to RMGLA. After verifying protein expression of the new clone (GST-OspFMutRb) (data not shown), we mixed GST-OspF and GST-OspFMutRb in equal amounts with purified Rb and subsequently precipitated the GST fusions with glutathione beads. As predicted, GST-OspFMutRb did not precipitate Rb from solution (Fig. 8B). Therefore, mutation of the Rb pocket binding site I/LXCXE in OspF abolished the ability of OspF to associate directly with Rb. The site-directed mutant binding result provides strong evidence that OspF is using a canonical site I/LXCXE required for an Rb association.
Lastly, we wanted to see if mutation of this site had in vivo consequences for the PMN migration assay and IL-8 secretion. The ospFMutRb was cloned into pBAD24 and transformed into the ΔospF mutant (BS771). The resulting strain (BS873) was used to infect polarized monolayers of T84 cells, and PMN migration and secretion of IL-8 were assessed and compared to the ΔospF mutant expressing and secreting wild-type ospF encoded in pBAD24 (BS802) (Zurawski et al., 2006). PMN migration was unchanged when measured after cells were infected with BS873 or BS802 (data not shown). However, the amount of secreted IL-8 measured by ELISA in basolateral compartments of Transwells infected with BS873 (ΔospF/ospFMutRb) was 370.06 ± 37.79 pg mL-1 as compared to 231.81 ± 16.34 pg mL-1 for BS802 (ΔospF/wild-type ospF). Therefore, the site-directed mutation of the OspF Rb binding site did not affect PMN migration but it did prevent the binding to Rb and reduced the capacity of OspF to repress IL-8 secretion in vivo.
It is clear from the recent work on the Osp and IpaH T3SS effector proteins of Shigella that these proteins play a pivotal role in manipulating host cell signaling and the innate immune response (Okuda et al., 2005; Kim et al., 2005; Zurawski et al., 2006; Arbibe et al., 2007, Rohde et al., 2007). OspB can now be added to a growing number of T3SS effectors from Shigella and other Gram-negative bacteria species that target the host cell nucleus and modulate the host inflammatory response. However, it is important to recognize that while these T3SS effectors are involved in dampening host inflammation during infection, most Gram-negative pathogens still induce a significant inflammatory response which contributes to damage and/or disease in the host. In the case of Shigella, the recruitment of neutrophils is a hallmark of its virulence, and the host inflammatory response contributes to increased invasion and dysentery. However, by reducing inflammation with a combination of the Osp and IpaH proteins, Shigella may prevent premature clearance from the host and prolong its proliferation.
OspB is highly conserved between Shigella species, and Salmonella and Vibrio species also have OspB homologues. It is interesting that the human host-specific S. Typhi and S. Paratyphi (as opposed to S. Typhimurium) share a T3SS effector with another human host-specific pathogen such as Shigella because it suggests these pathogens evolved a unique effector specific for their human host. In light of recent evidence that V. parahaemalyticus, and non-O1 strains of Vibrio cholerae have functional T3SS (Kodama et al., 2007; Tam et al., 2007), it is not surprising that an ospB homologue is shared between other T3SS containing Gram-negative bacteria that also reside in the human intestine (Brown et al., 2006). Future studies of S. Typhi, S. Paratyphi, V. parahaemalyticus, and other non-O1 strains of Vibrio cholerae may reveal that the OspB homologues contribute to the pathogenesis of these Gram-negative organisms.
Like OspF and OspC1 (Zurawski et al., 2006), OspB also localized in the nucleus of host cells during infection with S. flexneri or when transfected cells expressed the GFP-OspB fusions. No obvious NLS was detected, but it is clear that a sequence in the N-terminal half of the molecule is required for the nuclear localization because the C-terminal GFP fusion did not localize in the nucleus (Fig. 3). Nuclear targeting is a common trait shared by the IpaH and Osp proteins of Shigella (Toyotome et al., 2001; Zurawski et al., 2006; Zurawski et al., 2008). One can envision that Shigella requires a wide array of effectors in order to efficiently target the numerous genes in the nucleus involved in the innate immune response to completely down-regulate their transcription. Microarray data confirms that the transcription of >100 host genes in the nucleus are altered by Shigella infection, and MxiE-regulated genes such as ospB and ospF are required for this process (Pédron et al., 2003; Arbibe et al., 2007; Sperandio et al., 2008; Faherty and Maurelli unpublished results).
The GFP-OspB signal also localized in the cytoplasm with early endosomes, and this result is similar to the other Osp proteins which have separate cytoplasmic localizations apart from nuclear localization (Zurawski et al., 2006; Zurawski et al., 2008). We are still trying to determine the significance of the co-localization with Rab5, but there is a correlation between host signaling pathways and endosomes. Endosomes have been shown to serve as a scaffold concentrating the proteins of signaling pathways and enhancing their activity (von Zastrow and Sorkin, 2007), and this includes members of the MEK/ERK pathway (Anderson 2006).
The activation of the MEK/ERK pathway is required for cPLA2 activation and the production of hepoxilinA3 which ultimately results in PMN migration (Kohler et al., 2002; Mumy et al., 2008). Therefore, osp mutants that display a deficiency in MEK/ERK activation are linked to their deficiency to induce the PMN migration phenotype, and this relationship may also explain the delay in the inflammatory response observed in the Serény test with the double mutants. What is clear however is that OspF binding to Rb is not a requisite for induction of PMN migration since OspFMutRb complemented the ΔospF mutant similar to wild type OspF. The importance of Rb binding lies elsewhere as discussed below. The exact mechanism of how OspF and OspB activate the MEK/ERK pathway remains to be elucidated and several candidate proteins upstream of the pathway with cytoplasmic localization are under investigation (Zurawski and Maurelli, unpublished results).
In a separate and opposite phenotype, both OspB and OspF reduce the amount of IL-8 secreted during Shigella infection (Fig. 6). While this result appears counter-intuitive, it is important to stress that IL-8 is secreted only basolaterally by epithelial cells and is responsible for attracting neutrophils from the bloodstream and not across the epithelial barrier (McCormick 2007). Hepoxilin A3 is a neutrophil attractant secreted only apically, and in the case of Gram-negative bacterial infections, it is required to attract PMN across the epithelial barrier (McCormick 2007; Mumy et al., 2008). Therefore, these processes are not linked and underscore the dual role for T3SS effectors OspB and OspF in Shigella pathogenesis, one in the cytoplasm and one in the nucleus. Further support of this model for OspF can be found in the increased levels of IL-8 released from cells infected with Shigella expressing the Rb binding site mutant form of OspF (OspFMutRb) as compared to a strain expressing the wild type OspF. This result suggests that loss of the Rb binding site restricts the ability of OspF to reduce IL-8 message in vivo and is consistent with our model that Rb is required to recruit factors that repress IL-8 transcript in the nucleus.
The localization of OspF could be a crucial step with regard to its lyase activity. Two pieces of evidence substantiate this hypothesis. First, immunofluorescence with anti-phospho ERK1/2 antibody showed that dephosphorylation occurs only in the nucleus of Shigella-infected cells (Arbibe et al., 2007). Second, when MEK and OspF were mixed at equimolar concentrations in vitro, the presence of MEK prevented ERK2 from being dephosphorylated by OspF (Fig. S2). It was not until higher concentrations of OspF (> the MEK concentration) were added that OspF dephosphorylated ERK2 in the presence of MEK (Fig. S2). These results imply that even though OspF lyase activity is irreversible (Li et al., 2007), the OspF concentration in vivo must be higher than MEK to out compete the endogenous MEK kinase activity. In support of this model, the activation of the Salmonella OspF homologue, SpvC, is only detected in infected cells when SpvC is overexpressed (Mazurkiewicz et al., 2008). Therefore, the localization of OspF to a different subcellular compartment from the cytoplasm would allow OspF-mediated dephosphorylation in the absence of a competing MEK kinase activity. One subcellular compartment where MEK may not be as concentrated as OspF is the nucleus because MEK is actively exported via a nuclear export signal (Jaaro et al., 1997) and OspF is actively transported to the nucleus (Zurawski et al., 2006; Arbibe et al., 2007). We hypothesize that the lyase activity of OspF would be more effective in the nucleus where the concentration of MEK is substantially lower than in the cytoplasm.
The localization of OspF in the nucleus would also promote the chromatin remodeling activity which reduces IL-8 transcript (Arbibe et al. 2007). It is possible that OspB uses a similar mechanism to reduce IL-8 given the localization of OspB in the nucleus and given an interaction with Rb. The reduction of IL-8 is also consistent with the results of the Serény test where hyper-inflammation was observed two days after infection with the ΔospBF mutant.
Our results suggest that OspF is the first example of a secreted bacterial protein that interacts with Rb via the pocket binding domain (LXCXE), but there are many examples of other pathogens that interact with this binding site and exploit Rb function. Viruses encode proteins which bind to the Rb pocket to promote its degradation and stimulate the cell-cycle (Du and Pogoriler, 2006; Felsani et al., 2006; White and Khalini, 2006). Degradation of Rb does not occur during Shigella infection as we have found the total amount of Rb is unchanged over six hours of infection (Zurawski and Maurelli, unpublished results). However, Rb has another function which is to recruit histone deacetylases (HDACs), histone methyltransferases, and other factors which are required for chromatin remodeling and transcription repression (Du and Pogoriler, 2006; Macaluso et al., 2006). Since Shigella infection instigates histone modification, and this phenotype requires OspF (Arbibe et al., 2007), we speculate that Shigella secretes OspF to interact with Rb to recruit the other host proteins required for Shigella-induced histone modification. The interaction of OspF and Rb appears to be mediated by the Rb binding site (IMCLE) based on our site-directed mutagenesis and truncation studies (Fig. 8). Based on the structure of the SpvC homologue (Zhu et al. 2007), the IMCLE sequence would be found in alpha-helix 4 (α4), exposed on the surface of OspF, and adjacent to the MAPK binding site. It should be noted that Rb also interacts with all of the MAPK, including ERK1/2, but only in the active state (Chauhan et al., 1999; Wang et al., 1999; Guo et al., 2005). Therefore, it is possible that all of these proteins are in a complex, and after OspF cleaves the MAPK phosphate group with lyase activity, the MAPK disassociates from Rb leaving OspF to promote a repressive activity rather than an activating one (Wang et al., 1999; Guo et al., 2005) (Fig. 9). It is also important to note that OspF appears to only influence p38 MAPK and ERK1/2 phosphorylation in vivo, but not JNK (Arbibe et al., 2007; Kramer et al., 2007). This point is also significant with respect to Rb since JNK signaling through Rb leads to a different outcome than p38 signaling through Rb (Wang et al., 1999).
Fig. 9
Fig. 9
Model of S. flexneri T3SS effectors OspB and OspF in host epithelial cells. OspF and OspB are secreted via the Shigella T3SS and enter the nucleus by an unknown mechanisms. In the case of OspB, the N-terminus is required for nuclear import. OspF associates (more ...)
Further investigation is needed to determine the nature of the interaction between OspB and Rb. OspB does not have the LXCLE pocket binding site so the lack of an interaction in vitro with the C-terminus of Rb was not surprising. Nonetheless, the results of the OspB/Rb binding experiments (Fig. 8) suggest that OspB either interacts with the N-terminus of Rb or that another host protein acts as an intermediary between OspB and Rb (Fig. 9).
Future studies will focus on determining how Rb is modified to induce transcriptional repression via the OspF/Rb and OspB/Rb interactions. Because Rb plays a role in activating IL-8 transcription and inducing PMN transepithelial migration (Zhang et al., 2000), modification of Rb by Shigella T3SS effectors could just as easily result in the reverse outcome, i.e. repression of IL-8 transcription. It is clear that OspB and OspF are required for Shigella pathogenesis, and the interaction with Rb suggests that Rb is yet another host factor exploited by Shigella to repress host transcription. Future research will determine the distinct activities of OspF and OspB in different compartments of the host cell (cytoplasm vs. nucleus), and the other proteins that interact with these Osp proteins. Understanding how OspB and OspF manipulate host cell signaling and transcription in separate host compartments could serve as a paradigm for how Gram-negative bacteria modulate the host innate immune system.
Bacterial strains and growth conditions
All E. coli and S. flexneri strains used in this study are listed in Table 2. E. coli and S. flexneri were cultured in Luria broth (LB) or in tryptic soy broth (TSB), both at 37°C with aeration. Antibiotics were used in growth media and plates when required for selection at the following concentrations unless stated otherwise: kanamycin, 50 μg mL-1; chloramphenicol, 25 μg mL-1; spectinomycin, 50 μg mL-1; streptomycin, 25 μg mL-1; and ampicillin, 200 μg mL-1.
Table 2
Table 2
Bacterial strains and plasmids used in this study.
Tissue culture
HeLa cells and L2 mouse fibroblast cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). The human epithelial colon cancer-derived cell line T84 (passages 46 to 66) was maintained in DMEM/F-12 supplemented with 15 mM HEPES (pH 7.5) and 10% FBS. To obtain polarized monolayers, T84 cells were grown on 0.33 cm2 (PMN migration and IL-8 secretion) or 4.7 cm2 (harvesting, immunoblotting) collagen-coated permeable polycarbonate filters (Costar) as previously described (McCormick et al., 1998a; Kohler et al., 2002), and were utilized after they reached a confluent and differentiated state (~14 days). All tissue culture media were purchased from Invitrogen, and all cell lines were grown in the presence of 5% CO2 at 37°C.
Plasmid and mutant construction
All the plasmids, and the primers used to construct these plasmids, are described in Table 2 and Table 3. ospB and its native promoter along with the ospF truncations were amplified by colony PCR using Taq polymerase (Qiagen), cloned into pGEM-T (Promega), and sequenced. Subsequently, ospB and its promoter were subcloned into pDZ2 (BamHI and BglII). Full-length ospB and the ospF truncations were cloned into pGEX6P-1 using the BamHI and XhoI restriction sites, except for the C-terminal OspF construct where a SalI site was used at the 3′ end because of the XhoI site present in the C-terminal ospF sequence.
Table 3
Table 3
Oligonucleotide primers used in this study.
Full-length ospB and the ospB truncations were amplified by PCR using Vent polymerase (New England Biolabs) with the appropriate attB1 and attB2 sequences engineered into the primers required for BP recombination into pDONR221 plasmid of the Gateway® system (Invitrogen). A BP reaction was performed using the PCR products and transformed into the E. coli strain DB3.1 (Table 1). Clones were verified by restriction digest using BsrG1. Subsequently, an LR reaction was performed to recombine the ospB constructs into pDEST53 resulting in N-terminal GFP fusions. Clones were verified by PCR and restriction digests.
The deletion mutants of S. flexneri were generated by allelic exchange using a modification of the lambda red system of Datsenko and Wanner previously described (Datsenko and Wanner, 2000; Murphy and Campellone 2003; Zurawski et al., 2006). Primers were used to amplify the chloramphenicol resistance cassette (cat) from pKD3 with sequences at the 5′ and 3′ ends identical to sequences 20 bp internal to and 30 bp upstream of ospB (Table 3). After transformation of BS766, bacteria were plated on TSB plates containing Congo red and chloramphenicol at a concentration of 5 μg mL-1 for selection of recombinants. Double mutants were constructed by allelic exchange using ospF and ospC1 primers (Zurawski et al., 2006) after single mutants were transformed with pCP20 and incubated at 42°C to remove the cat cassette (Datsenko and Wanner, 2000). Cmr colonies on these plates were purified and screened by PCR using three different primer sets to identify deletion mutants. We verified that all mutations created in S. flexneri did not have an effect on growth in LB or TSB broth.
Site-directed mutagenesis on pGST-OspF was performed using the QuikChange site-directed mutagenesis kit according to manufacturer's instructions (Stratagene). Positive colonies were sequenced to verify the mutations that changed ospF sequence from 5′- ataatgtgtctcgag -3′ (amino acids IMCLE) to 5′ - agaatgggtctcgcg – 3′ (amino acids RMGLA) resulting in the pGST-OspFRbMut mutant strain. Subsequently, pGST-OspFMutRb was transformed into BL21 pLysS for protein binding studies.
Secretion assay
Secretion of T3SS effector proteins from S. flexneri was analyzed as previously described (Zurawski et al., 2006). Bacterial cultures were subcultured in TSB and grown at 37°C. Once late log phase was reached, bacterial samples were normalized using OD600. Congo red (7 μg mL-1) was added, and after 1 h, whole-cell lysates and supernatant fractions were prepared. Supernatant fractions were prepared by trichloroacetic acid (TCA) precipitation of 25 ml of culture passed through a 0.2 micron filter to ensure the removal of bacteria. Following an acetone wash, samples were resuspended in SDS polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer and run on 12% SDS-PAGE gels.
Immunofluorescence and transfection analysis
Immunofluorescence was carried out as previously detailed using mouse anti-HA monoclonal antibody HA.11 (Covance) and goat anti-mouse antibody conjugated to AlexaFluor 488 (Invitrogen) (Zurawski et al., 2006). Transfections were carried out as previously described (Zurawski et al., 2006) except Lipofectimine 2000 was used as the transfection reagent according to the manufacturer's instructions. Cells were counter-stained with 4′,6′-diamidino-2-phenylindole (DAPI) (0.5 μg/ml) for 20 min, rabbit anti-Rab5 antibody (Cell Signaling, Inc.) diluted 1:50, and goat anti-rabbit secondary coupled to AlexFluor 594 (Invitrogen) and diluted 1:500. Shigella-infected cell images were acquired with an Olympus BX51 microscope using an Olympus DP-70 digital camera and merged using DP controller/manager software version Images of transfected cell were acquired on an epifluorescence Olympus 1×81 microscope using a SensiCam charge-coupled device camera (Cooke) and merged with IPLabs software version 3.1.
Shigella virulence assays
S. flexneri invasion assays were carried out as previously described (Hale and Formal, 1981), where colony forming units recovered from infected cells were counted and compared to total input bacteria to calculate the invasion efficiency. Plaque assays and the apoptosis protection assays were performed as previously described (Oaks et al., 1985; Clark and Maurelli, 2007). The Serény test was used to assess invasion and the in vivo inflammatory response in guinea pigs (Serény, 1957). Three guinea pigs were used to evaluate each strain used, and symptoms were monitored for four days. Reaction ratings presented are based on one representative animal for each group of three infected animals and defined in Table 1. The extra rating (4) was added to account for additional swelling/inflammation seen with some of the osp mutants as was done previously for ipaH mutants (Fernandez-Prada et al., 2000).
PMN transepithelial migration assay
The PMN transepithelial migration assay was performed as previously described (McCormick et al., 1998a). Briefly, inverted T84 polarized monolayers seeded on 0.33 cm2 filters were basolaterally infected with 25 μL S. flexneri wild-type or mutant suspension in HBSS (with Ca2+ and Mg2+) at 37°C for 90 min at a multiplicity of infection (MOI) of ~100. After 90 min, extracellular bacteria were removed by thorough washing.
Human PMNs were purified from whole blood (anticoagulated with 13.2 g of citrate and 11.2 g of dextrose in 500 mL of water, pH 6.5) collected by venipuncture from healthy human volunteers of both sexes, in accordance with the Massachusetts General Hospital IRB approval (protocol number P-007782/7). The buffy coat was obtained by centrifugation at 400 × g at RT. Plasma and mononuclear cells were removed by aspiration, and the majority of erythrocytes were removed by a 2% gelatin sedimentation technique (Parkos et al., 1991). Residual erythrocytes were removed by lysis in cold NH4Cl lysis buffer. After isolation, PMNs were suspended at a concentration of 5 × 107 mL-1 in modified HBSS (without Ca2+ and Mg2+) and kept at 4°C until used. PMNs prepared in this manner were 95% pure with 98% viability (McCormick et al., 1998b). 20 μL of PMN suspension was added to 100 μl HBSS on the basolateral surface and Transwells were incubated at 37°C for 3 h to allow PMNs to migrate to the apical compartment. Positive control transmigration assays were performed by the addition of the potent PMN chemoattractant, 1 μM N-formylmethionylleucyl phenylalanine (fMLP; Sigma, St. Louis, MO) to the apical chamber. Transmigration to the apical compartment was quantified by assaying for the PMN azurophilic granule marker myeloperoxidase (Parkos et al., 1991). Unpaired Student t tests were used to analyze the statistical significance of the mutant S. flexneri strains vs. wild-type 2457T.
IL-8 secretion
Inverted T84 monolayers seeded on 0.33 cm2 filters were infected basolaterally with 25 μl of S. flexneri wild-type or mutant suspension in HBSS for 90 min at 37°C at an MOI of 100. As a positive control, monolayers were incubated with 1 μg/ml phorbol 12-myristate 13-acetate for 90 min. Following infection, monolayers were washed and transferred to fresh 24-well trays with 1 ml HBSS in the well and 100 μl HBSS in the upper chamber (basolateral surface). Monolayers were incubated at 37°C for 4 h to allow the basolateral secretion of IL-8. After incubation, 100 μl were collected from the basolateral chamber and IL-8 was measured per manufacturer's instructions using the Human IL-8 ELISA Kit (Pierce Endogen, Rockford, IL). Data were analyzed using Tukey's analysis of variance post hoc test (ANOVA) on the SPSS program, version 12.0.1 for Windows.
Protein binding assays
To generate HeLa lysates, HeLa cells were washed with ice-cold phosphate-buffered saline (1 × PBS) and lysed for 30 min in a lysis buffer [1 × PBS 1% Triton X-100 plus a protease inhibitor tablet (Roche) and phosphatase inhibitor mixture including activated 4 mM Na3VO4 and 40 mM NaF] at 4°C. The lysates were centrifuged at 15,000 × g for 30 min at 4°C. The supernatant was stored at -80°C or used immediately. E. coli BL21 pLysS bacteria with pGEX 6P-1 (GST alone control) or pGST-Osp plasmids grown overnight were subcultured 1:100 and expression was induced using 1 mM IPTG when the culture reached an OD600 between 0.6 and 0.7. After two hours of expression, bacteria were washed with 1× PBS, pelleted, and frozen at -80°C. Following two cycles of freeze-thaw (-80°C/4°C), bacterial pellets were resuspended in lysis buffer. Bacterial lysates were centrifuged at 15,000 × g for 30 min at 4°C. The supernatant was stored at -80°C or used immediately.
Bacterial and HeLa cell lysates were mixed at equal amounts at 4°C with constant end-over-end mixing with a rotisserie in 15 mL conical tubes. The next day lysates were mixed with 100 μL of prewashed glutathione beads (50% slurry). Samples were centrifuged at 2000 × g and glutathione beads were washed in cold lysis buffer and transferred to microfuge tubes. Following two more washes, the absorbed complexes were removed from the beads by heating for 5 min at 100°C in 1× SDS sample buffer and separated on 7.5% or 10% polyacrylamide gels. Proteins were transferred to nitrocellulose membranes and detected with appropriate antibodies by immunoblot.
For pure protein binding experiments, GST fusions were purified as above from bacterial lysate, but after washes, GST fusions were eluted with 50 mM glutathione buffer according to manufacturer's instructions (Amersham). After elution, proteins were dialyzed in 1× PBS and mixed with purified microtubules resuspended in 1× PBS (Cytoskeleton Inc.) or purified Rb C-fusion, ERK2, or MEK (Cell Signaling, Inc.) at equimolar concentrations measured by ND-1000 Spectrophotometer (Nanodrop) and bicinchoninic acid (BCA)-containing protein assay (Pierce) OspF dephosphorylation of ERK2 was analyzed by mixing pure OspF, cleaved from GST using PreScission protease (Amersham) and dialyzed into kinase buffer [25 mM Tris pH 7.5, 5 mM β-glycerolphosphate, 10 mM MgCl2, 0.1 mM Na3VO4, 2 mM DTT, 200 μM ATP], with pure ERK2 and MEK (Cell Signaling, Inc.). MEK was inhibited with PD98059 according to manufacturer's instructions (Cell Signaling, Inc.). Phosphorylation was measured using monoclonal Phospho-p44/42 MAP Kinase Thr202/Tyr204 (Anti-ERK1/2-P) antibody (Cell Signaling, Inc.)
SDS-PAGE and immunoblot analysis of proteins
For protein analysis, samples were resolved on 7.5%, 10%, or 12% Tris-glycine SDS-PAGE gels (Laemmli 1970). For immunoblotting, proteins were transferred to pure nitrocellulose membranes (Biorad), OspB-2HA was detected using mouse anti-HA monoclonal antibody HA.11 (Covance). ERK1/2 signaling was measured as previously described (Kohler et al., 2002). Briefly, T84 cells were seeded on collagen-coated filters on 4.7-cm2 Transwells, infected, and cells were lysed in a lysis buffer [1% Triton X-100, 100 mM NaCl, 10 mM HEPES, 2 mM EDTA, 4 mM Na3VO4, 40 mM NaF, 200 mM phenylmethylsulfonyl fluoride, protease inhibitor cocktail (Roche)]. Samples were centrifuged, and the supernatant representing the cytosol was collected and stored at -80°C until they were boiled and run on SDS-PAGE. Samples were immunoblotted using mouse monoclonal antibody against Anti-ERK1/2-P (Santa Cruz or Cell Signaling, Inc.). Total protein amounts were evaluated using rabbit polyclonal antibody against ERK1/2 (Santa Cruz or Cell Signaling, Inc.). Rabbit polyclonal antibody against ERK1/2 (Cell Signaling, Inc.) was also used to detect ERK1/2 in binding experiments. Rb was detected using anti-Rb mouse monoclonal antibody (Cell Signaling, Inc.), and tubulin was detected with anti-β-tubulin polyclonal antibody (Sigma). Bands were visualized using sheep anti-mouse and sheep anti-rabbit secondary antibodies conjugated to horseradish peroxidase (Amersham). Primary and secondary antibodies were used at a 1:1000 dilution. Blots were developed using Visualizer (Upstate) or NCT (Pierce), and images were captured with a charge-coupled device camera from the LAS-3000 CH imaging system (Fuji) or captured on film.
Supplementary Material
Supp Mat
We would like to acknowledge Nancy Adams and Reinaldo Fernandez for technical assistance, Dr. Joseph Giam for use of the inverted fluorescence microscope, and Dr. Anne-Laure Prunier for critical reading of this manuscript. We would also like to thank our colleagues at the Walter Reed Institute of Army Research for reagents and the time to allow D.V.Z. finish this manuscript, especially the members of the Oaks laboratory and Dr. Ryan Ranallo for supplying the Gateway® vectors.
This work was supported by National Institutes of Allergy and Infectious Diseases Grant AI24656 (to A.T.M.). B.A.M. is supported by National Institutes of Health Grants DK56754 and DK33506. K.L.M. is supported by a T32 Training Grant sponsored by Harvard Medical School and the Department of Surgery at Massachusetts General Hospital. C.S.F. is supported by the Val G. Hemming Fellowship sponsored by the Henry M. Jackson Foundation and Uniformed Services University of the Health Sciences.
  • Anderson DH. Role of lipids in the MAPK signaling pathway. Prog Lipid Res. 2006;45:102–119. [PubMed]
  • Arbibe L, Kim DW, Batsche E, Pedron T, Mateescu B, Muchardt C, et al. An injected bacterial effector targets chromatin access for transcription factor NF-κB to alter transcription of host genes involved in immune responses. Nat Immunol. 2007;8:47–56. [PubMed]
  • Brown NF, Wickham ME, Coombes BK, Finlay BB. Crossing the line: selection and evolution of virulence traits. PLoS Pathog. 2006;2:e42. [PMC free article] [PubMed]
  • Buchrieser C, Glaser P, Rusniok C, Nedjari H, D'Hauteville H, Kunst F, et al. The virulence plasmid pWR100 and the repertoire of proteins secreted by the type III secretion apparatus of Shigella flexneri. Mol Microbiol. 2000;38:760–771. [PubMed]
  • Cokol M, Nair R, Rost B. Finding Nuclear localization signals. EMBO Rep. 2000;1:411–415. [PubMed]
  • Chauhan D, Hideshima T, Treon S, Teoh G, Raje N, Yoshihimito S, et al. Functional interaction between retinoblastoma protein and stress-activated protein kinase in multiple myeloma cells. Cancer Res. 1999;59:1192–1195. [PubMed]
  • Clark CS, Maurelli AT. Shigella flexneri inhibits staurosporine-induced apoptosis in epithelial cells. Infect Immun. 2007;75:2531–2539. [PMC free article] [PubMed]
  • Dalsgaard A, Albert MJ, Taylor DN, Shimada T, Meza R, Serichantalergs O, Echeverria P. Characterization of Vibrio cholerae non-O1 serogroups obtained from an outbreak of diarrhea in Lima, Peru. J Clin Microbiol. 1995;33:2715–2722. [PMC free article] [PubMed]
  • Datsenko KA, Wanner BL. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A. 2000;97:6640–6645. [PubMed]
  • Du W, Pogoriler J. Retinoblastoma family genes. Oncogene. 2006;25:5190–5200. [PMC free article] [PubMed]
  • Felsani A, Mileo AM, Paggi MG. Retinoblastoma family proteins as key targets of the small DNA virus oncoproteins. Oncogene. 2006;25:5277–5285. [PubMed]
  • Fernandez-Prada CM, Hoover DL, Tall BD, Hartman AB, Kopelowitz J, Venkatesan MM. Shigella flexneri IpaH7.8 facilitates escape of virulent bacteria from the endocytic vacuoles of mouse and human macrophages. Infect Immun. 2000;68:3608–3619. [PMC free article] [PubMed]
  • Formal SB, Dammin GJ, Labrec EH, Schneider H. Experimental Shigella infections: characteristics of a fatal infection produced in guinea pigs. J Bacteriol. 1958;75:604–610. [PMC free article] [PubMed]
  • Gupta A, Polyak CS, Bishop RD, Sobel J, Mintz ED. Laboratory-confirmed shigellosis in the United States, 1989-2002: epidemiologic trends and patterns. Clin Infect Dis. 2004;38:1372–1377. [PubMed]
  • Guo J, Sheng G, Warner BW. Epidermal growth factor-induced rapid retinoblastoma phosphorylation at Ser780 and Ser795 is mediated by ERK1/2 in small intestine epithelial cells. J Biol Chem. 2005;280:35992–35998. [PubMed]
  • Guzman LM, Belin D, Carson MJ, Beckwith J. Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol. 1995;177:4121–4130. [PMC free article] [PubMed]
  • Hale TL, Formal SB. Protein synthesis in HeLa or Henle 407 cells infected with Shigella dysenteriae 1, Shigella flexneri 2a, or Salmonella typhimurium W118. Infect Immun. 1981;32:137–144. [PMC free article] [PubMed]
  • Hanahan D. Studies on transformation of Escherichia coli with plasmids. J Mol Biol. 1983;166:557–580. [PubMed]
  • Jaaro H, Rubinfeld H, Hanoch T, Seger R. Nuclear translocation of mitogen-activated protein kinase kinase (MEK1) in response to mitogenic stimulation. Proc Natl Acad Sci USA. 1997;94:3742–3747. [PubMed]
  • Jouihri N, Sory MP, Page AL, Gounon P, Parsot C, Allaoui A. MxiK and MxiN interact with the Spa47 ATPase and are required for transit of the needle components MxiH and MxiI, but not of Ipa proteins, through the type III secretion apparatus of Shigella flexneri. Mol Microbiol. 2003;49:755–767. [PubMed]
  • Kane CD, Schuch R, Day WA, Jr, Maurelli AT. MxiE regulates intracellular expression of factors secreted by the Shigella flexneri 2a type III secretion system. J Bacteriol. 2002;184:4409–4419. [PMC free article] [PubMed]
  • Kim DW, Lenzen G, Page AL, Legrain P, Sansonetti PJ, Parsot C. The Shigella flexneri effector OspG interferes with innate immune responses by targeting ubiquitin-conjugating enzymes. Proc Natl Acad Sci U S A. 2005;102:14046–14051. [PubMed]
  • Kodama T, Rokuda M, Park KS, Cantarelli VV, Matsuda S, Iida T, Honda T. Identification and characterization of VopT, a novel ADP-ribosyltransferase effector protein secreted via the Vibrio parahaemolyticus type III secretion system 2. Cell Microbiol. 2007;9:2598–2609. [PubMed]
  • Kohler H, Rodrigues SP, McCormick BA. Shigella flexneri interactions with the basolateral membrane domain of polarized model intestinal epithelium: role of lipopolysaccharide in cell invasion and in activation of the mitogen-activated protein kinase ERK. Infect Immun. 2002;70:1150–1158. [PMC free article] [PubMed]
  • Kotloff KL, Winickoff JP, Ivanoff B, Clemens JD, Swerdlow DL, Sansonetti PJ, et al. Global burden of Shigella infections: implications for vaccine development and implementation of control strategies. Bull World Health Organ. 1999;77:651–666. [PubMed]
  • Kramer RW, Slagowski NL, Eze NA, Giddings KS, Morrison MF, Siggers KA, et al. Yeast functional genomic screens lead to identification of a role for a bacterial effector in innate immunity regulation. PLoS Pathog. 2007;3:e21. [PMC free article] [PubMed]
  • Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature. 1970;227:680–685. [PubMed]
  • Le Gall T, Mavris M, Martino MC, Bernardini ML, Denamur E, Parsot C. Analysis of virulence plasmid gene expression defines three classes of effectors in the type III secretion system of Shigella flexneri. Microbiology. 2005;151:951–962. [PubMed]
  • Levine MM, Kotloff KL, Barry EM, Pasetti MF, Sztein MB. Clinical trials of Shigella vaccines: two steps forward and one step back on a long, hard road. Nat Rev Microbiol. 2007;5:540–553. [PubMed]
  • Li H, Xu H, Zhou Y, Zhang J, Long C, Li S, et al. The phosphothreonine lyase activity of a bacterial type III effector family. Science. 2007;315:1000–1003. [PubMed]
  • MacCormick M, Moderscheim T, van der Salm LW, Moore A, Pryor SC, McCaffrey G, Grimes ML. Distinct signaling particles containing ERK/MEK and B-Raf in PC12 cells. Biochem J. 2005;387:155–164. [PubMed]
  • Macaluso M, Montanari M, Giordano A. Rb family proteins as modulators of gene expression and new aspects regarding the interaction with chromatin remodeling enzymes. Oncogene. 2006;25:5263–5267. [PubMed]
  • Makino K, Oshima K, Kurokawa K, Yokoyama K, Uda T, Tagomori K, et al. Genome sequence of Vibrio parahaemolyticus: a pathogenic mechanism distinct from that of V. cholerae. Lancet. 2003;361:743–749. [PubMed]
  • Maurelli AT, Blackmon B, Curtiss R., III Loss of pigmentation in Shigella flexneri 2a is correlated with loss of virulence and virulence-associated plasmid. Infect Immun. 1984;43:397–401. [PMC free article] [PubMed]
  • Mavris M, Page AL, Tournebize R, Demers B, Sansonetti P, Parsot C. Regulation of transcription by the activity of the Shigella flexneri type III secretion apparatus. Mol Microbiol. 2002;43:1543–1553. [PubMed]
  • Mazurkiewicz P, Thomas J, Thompson JA, Liu M, Arbibe L, Sansonetti P, Holden DW. SpvC is a Salmonella effector with phosphothreonine lyase activity on host mitogen-activated protein kinases. Mol Microbiol. 2008;67:1371–1383. [PMC free article] [PubMed]
  • McCormick BA, Siber AM, Maurelli AT. Requirement of the Shigella flexneri virulence plasmid in the ability to induce trafficking of neutrophils across polarized monolayers of the intestinal epithelium. Infect Immun. 1998a;66:4237–4243. [PMC free article] [PubMed]
  • McCormick BA, Parkos CA, Colgan SP, Carnes DK, Madara JL. Apical secretion of a pathogen-elicited epithelial chemoattractant activity in response to surface colonization of intestinal epithelial by Salmonella typhimurium. J Immunol. 1998b;160:455–466. [PubMed]
  • McCormick BA. Bacterial-induced hepoxilin A3 secretion as a pro-inflammatory mediator. FEBS J. 2007;274:3513–3518. [PubMed]
  • Moran GJ. Threats in bioterrorism. II. CDC category B and C agents. Emerg Med Clin N Am. 2002;20:311–330. [PubMed]
  • Murphy KC, Campellone KG. Lambda Red-mediated recombinogenic engineering of enterohemorrhagic and enteropathogenic E. coli. BMC Mol Biol. 2003;4:11. [PMC free article] [PubMed]
  • Mumy KL, Bien JD, Pazos MA, Gronert K, Hurley BP, McCormick BA. Distinct isoforms of phospholipase A2 mediate the ability of Salmonella enterica serotype typhimurium and Shigella flexneri to induce the transepithelial migration of neutrophils. Infect Immun. 2008;76:3614–3627. [PMC free article] [PubMed]
  • Nakai K, Horton P. PSORT: a program for detecting the sorting signals of proteins and predicting their subcellular localization. Trends Biochem Sci. 1999;24:34–35. [PubMed]
  • Oaks EV, Hale TL, Formal SB. Plaque formation by virulent Shigella flexneri. Infect Immun. 1985;48:124–129. [PMC free article] [PubMed]
  • Ogawa M, Sasakawa C. Shigella and autophagy. Autophagy. 2006;2:171–174. [PubMed]
  • Okuda J, Toyotome T, Kataoka N, Ohno M, Abe H, Shimura Y, et al. Shigella effector IpaH9.8 binds to a splicing factor U2AF35 to modulate host immune responses. Biochem Biophys Res Commun. 2005;333:531–539. [PubMed]
  • Parkos CA, Delp C, Arnaout MA, Madara JL. Neutrophil migration across a cultured intestinal epithelium. Dependence on a CD11b/CD18-mediated event and enhanced efficiency in physiological direction. J Clin Invest. 1991;88:1605–1612. [PMC free article] [PubMed]
  • Pédron T, Thibault C, Sansonetti PJ. The invasive phenotype of Shigella flexneri directs a distinct gene expression pattern in the human intestinal epithelial cell line Caco-2. J Biol Chem. 2003;278:33878–33886. [PubMed]
  • Perdomo JJ, Gounon P, Sansonetti PJ. Polymorphonuclear leukocyte transmigration promotes invasion of colonic epithelial monolayers by Shigella flexneri. J Clin Invest. 1994;93:633–643. [PMC free article] [PubMed]
  • Pickering LK. Antimicrobial resistance among enteric pathogens. Semin Pediatr Infect Dis. 2004;15:71–77. [PubMed]
  • Puntervoll P, Linding R, Gemund C, Chabanis-Davidson S, Mattingsdal M, Cameron S, et al. ELM server: A new resource for investigating short functional sites in modular eukaryotic proteins. Nucleic Acids Res. 2003;31:3625–3630. [PMC free article] [PubMed]
  • Rohde JR, Breitkreutz A, Chenal A, Sansonetti PJ, Parsot C. Type III secretion effectors of the IpaH family are E3 ubiquitin ligases. Cell Host Microbe. 2007;1:77–83. [PubMed]
  • Sakaguchi T, Kohler H, Gu X, McCormick BA, Reinecker HC. Shigella flexneri regulates tight junction-associated proteins in human intestinal epithelial cells. Cell Microbiol. 2002;4:367–381. [PubMed]
  • Santapaola D, Casalino M, Petrucca A, Presutti C, Zagaglia C, Berlutti F, et al. Enteroinvasive Escherichia coli virulence-plasmid-carried apyrase (apy) and ospB genes are organized as a bicistronic operon and are subject to differential expression. Microbiology. 2002;148:2519–2529. [PubMed]
  • Santapaola D, Del Chierico F, Petrucca A, Uzzau S, Casalino M, Colonna B, et al. Apyrase, the product of the virulence plasmid-encoded phoN2 (apy) gene of Shigella flexneri, is necessary for proper unipolar IcsA localization and for efficient intercellular spread. J Bacteriol. 2006;188:1620–1627. [PMC free article] [PubMed]
  • Schroeder GN, Hilbi H. Molecular pathogenesis of Shigella spp.: controlling host cell signaling, invasion, and death by type III secretion. Clin Microbiol Rev. 2008;21:134–156. [PMC free article] [PubMed]
  • Serény B. Experimental keratoconjunctivitis shigellosa. Acta Microbiol Acad Sci Hung. 1957;4:367–376. [PubMed]
  • Sperandio B, Regnault B, Guo J, Zhang Z, Stanley SL, Jr, Sansonetti PJ, Pédron T. Virulent Shigella flexneri subverts the host innate immune response through manipulation of antimicrobial peptide gene expression. J Exp Med. 2008;205:1121–1132. [PMC free article] [PubMed]
  • Tam VC, Serruto D, Dziejman M, Brieher W, Mekalanos JJ. A type III secretion system in Vibrio cholerae translocates a formin/spire hybrid-like actin nucleator to promote intestinal colonization. Cell Host Microbe. 2007;1:95–107. [PubMed]
  • Toyotome T, et al. Shigella protein IpaH9.8 is secreted from bacteria within mammalian cells and transported to the nucleus. J Biol Chem. 2001;276:32071–32079. [PubMed]
  • Wang S, Nath N, Minden A, Chellappan S. Regulation of Rb and E2F by signal transduction cascades: divergent effects of JNK1 and p38 kinases. EMBO J. 1999;18:1559–1570. [PubMed]
  • White MK, Khalili K. Interaction of retinoblastoma protein family members with large T-antigen of primate polyomaviruses. Oncogene. 2006;25:5286–5293. [PubMed]
  • von Zastrow M, Sorkin A. Signaling on the endocytic pathway. Curr Opin Cell Biol. 2007;19:436–445. [PMC free article] [PubMed]
  • Zhang H, Wei S, Sun J, Coppola D, Zhong B, Wu GD, et al. Retinoblastoma protein activation of interleukin 8 expression inhibits tumor cell survival in nude mice. Cell Growth Differ. 2000;11:635–639. [PubMed]
  • Zhu Y, Li H, Long C, Hu L, Xu H, Liu L, et al. Structural insights into the enzymatic mechanism of the pathogenic MAPK phosphothreonine lyase. Mol Cell. 2007;28:899–913. [PubMed]
  • Zurawski DV, Mitsuhata C, Mumy KL, McCormick BA, Maurelli AT. OspF and OspC1 are Shigella flexneri type III secretion system effectors that are required for postinvasion aspects of virulence. Infect Immun. 2006;74:5964–5976. [PMC free article] [PubMed]
  • Zurawski DV, Mumy KL, Badea L, Prentice JA, Hartland EL, McCormick BA, Maurelli AT. NleE/OspZ family of effector proteins is required for PMN transepithelial migration, a characteristic shared by enteropathogenic Escherichia coli and Shigella flexneri infections. Infect Immun. 2008;76:369–379. [PMC free article] [PubMed]