Identification and Characterization of a Novel Bacterial Virulence Factor That Shares Homology with Mammalian Toll/Interleukin-1 Receptor Fa
http://www.100md.com
感染与免疫杂志 2006年第1期
Burnham Institute for Medical Research, 10901 North Torrey Pines Road, La Jolla, California 92037
ABSTRACT
Many important bacterial virulence factors act as mimics of mammalian proteins to subvert normal host cell processes. To identify bacterial protein mimics of components of the innate immune signaling pathway, we searched the bacterial genome database for proteins with homology to the Toll/interleukin-1 receptor (TIR) domain of the mammalian Toll-like receptors (TLRs) and their adaptor proteins. A previously uncharacterized gene, which we have named tlpA (for TIR-like protein A), was identified in the Salmonella enterica serovar Enteritidis genome that is predicted to encode a protein resembling mammalian TIR domains, We show that overexpression of TlpA in mammalian cells suppresses the ability of mammalian TIR-containing proteins TLR4, IL-1 receptor, and MyD88 to induce the transactivation and DNA-binding activities of NF-B, a downstream target of the TIR signaling pathway. In addition, TlpA mimics the previously characterized Salmonella virulence factor SipB in its ability to induce activation of caspase-1 in a mammalian cell transfection model. Disruption of the chromosomal tlpA gene rendered a virulent serovar Enteritidis strain defective in intracellular survival and IL-1 secretion in a cell culture infection model using human THP1 macrophages. Bacteria with disrupted tlpA also displayed reduced lethality in mice, further confirming an important role for this factor in pathogenesis. Taken together, our findings demonstrate that the bacterial TIR-like protein TlpA is a novel prokaryotic modulator of NF-B activity and IL-1 secretion that contributes to serovar Enteritidis virulence.
INTRODUCTION
The innate immune response is the first line of host defense against microbial and fungal invasion. Components of the innate immune pathway are evolutionarily conserved in insects, plants, and vertebrates. One of the primary mediators of the innate immune response is the Toll-like/interleukin-1 receptor (IL-1R) superfamily of cell surface receptors (reviewed in references 1, 2, 3, 4, and 7). In mammals, members of this superfamily are divided into three subgroups. Subgroup 1 is comprised of the interleukin receptors, which bind cytokines. Subgroup 2 is comprised of Toll-like receptors, which display extracellular leucine-rich repeats that bind microbial products. The third subgroup is comprised of the cytosolic adaptor proteins used to convey extracellular signals from both subgroup 1 and 2 receptors to downstream effectors. The structural motif common to all members of the Toll-like/IL-1R superfamily is a cytosolic domain called the Toll/IL-1R (TIR) domain. This domain is required for interaction of TIR-containing Toll-like/IL-1R surface receptors to downstream TIR-containing signaling partners through homotypic interactions among TIR domains. The three-dimensional structure of the TIR domain of IL-1R consists of five parallel -sheets surrounded by two layers of parallel -helices (32).
TLR/IL-1R family proteins activate signal transduction pathways in response to proinflammatory cytokines or microbial derived substances such as lipopolysaccharide, peptidoglycan, and other cell wall components. Binding of appropriate ligands to these receptors engages a signaling cascade that converges upon the NF-B family of transcription factors (reviewed in reference 27). This cascade involves assembly of a multiprotein complex that includes (in mammals) TIR-containing adapters such as MyD88 or TIRAP, IRAK family kinases, and TRAF6, resulting in activation of the IB kinase complex (IKK) that phosphorylates the NF-B inhibitor IB and targets it for ubiquitination and proteasome-dependent destruction. Activation of NF-B in turn leads to production of proteins involved in inflammation, host defense, apoptosis suppression, and tissue repair (reviewed in reference 15).
Both gram-negative and gram-positive bacterial pathogens produce a variety of virulence factors that interact with host molecules and modulate specific cellular processes important for immune defense including phagocytosis, signal transduction, cytokine production, and cytoskeletal rearrangements (reviewed in reference 12, 16, 25, 28, and 29). Virulence factors can be presented either on the surface of the bacterium, injected into the host cell upon bacterial attachment to the cell surface, or secreted through specialized secretion systems. Bacterial virulence factors that interfere with NF-B induction have been identified. For example, the YopJ protein of Yersinia blocks signal transduction events required for NF-B induction by cytokine receptors and Toll-like receptors (TLRs). YopJ is a putative cysteine protease that cleaves the ubiquitin-like protein SUMO from target molecules, although the specific host cell substrates of this bacterial SUMO lyase remain unclear (23, 24). Similarly, Salmonella enterica serovar Typhimurium encodes a homologous putative cysteine protease AvrA that also thwarts TLR and cytokine-induced activation of NF-B, apparently acting at a point distal to YopJ in the signal transduction pathway responsible of NF-B activation (5). By suppressing NF-B activation, virulence factors such as YopJ and AvrA not only reduce the host inflammatory response but also promote host cell apoptosis due to the importance of NF-B in inducing expression of antiapoptotic genes required for protection against cytokine-induced apoptosis (15).
We postulated that identification of bacterial proteins showing sequence homology with components of the innate immune pathway would lead to the identification of novel bacterial virulence factors. By searching for proteins with structural homology to TIR-like domains, we have identified such a protein from the serovar Enteritidis genome, which we designated TlpA (for TIR-like protein A). We show that the TlpA protein suppresses NF-B induction by stimuli that involve TIR domain proteins, and we provide evidence that TlpA is required for full virulence. In addition to suppressing TIR-dependent NF-B activation, however, TlpA possesses other functions. Analogous to the previously characterized SipB protein of serovar Typhimurium, TlpA expression promotes activation of the protease caspase-1, resulting in caspase-dependent secretion of IL-1 and host cell apoptosis. Thus, the TlpA protein of serovar Enteritidis represents a novel bacterial virulence factor possessing at least two potent mechanisms for modification of host defense.
MATERIALS AND METHODS
Bioinformatics. Amino acid sequences of human TIR domains were used to search public protein sequence databases, including the database of unfinished microbial genomes at National Center for Biotechnology Information (http://ncbi.nig.gov) and the SEED database and genome annotation system (http://theseed.uchicago.edu/FIG/index.cgi). A cascade of PSI-BLAST searches, as implemented in the SaturatedBLAST program (18), were used to identify bacterial proteins with potential TIR domains. Up to five iterations of PSI-BLAST were run, and all proteins with an e value below 0.005 were taken for further consideration. All thus identified proteins were subject to further analysis and verification by reverse PSI-BLAST searches, distant homology analysis, and three-dimensional model building using the FFAS server (13, 26).
Cell culture. HEK293N and HEK293T cells were maintained in Dulbecco modified Eagle medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen), 100 U of penicillin/ml, and 100 μg of streptomycin/ml at 37°C in 5% CO2. THP-1 cells were maintained in RPMI supplemented with 10% FBS, penicillin, and streptomycin at 37°C in 5% CO2. All bacterial strains were grown in standard LB broth unless otherwise indicated.
Bacterial strains. Wild-type serovar Enteritidis LK5 was used for infection experiments (gift of S. Maloy, San Diego State University, San Diego, Calif.). The LK5 tlpA::kan mutant strain was constructed by the method of Datsenko and Wanner (6). Briefly, PCR primers tlp/P1 (5'-ATGTCAGAGATCTCTCGTCATAAAAATGATATTTCAAGATGTAACTCTGTGTAGGCTGGAG CTGCTT-3') and tlp/P4 (5'-TCATTTCAATAATGATTCAAGCTGTTCAGCAATTTCATCAATGGATTCCGGGGATCCGTCGACC-3') were synthesized with homology to the FLP recognition target sites flanking the kanamycin resistance (Kanr) gene of template plasmid pKD13. These primers also contained 44- to 48-nucleotide extensions (underlined) that are homologous to regions flanking the tlpA gene. The PCR product generated using these primers with pKD13 as a template was gel purified and electroporated into the LK5 strain carrying the Red recombinase helper plasmid pKD46. The resulting Kanr transformants were used for P22 transduction of the tlpA::kan allele into a fresh LK5 background and subsequent selection for Kanr colonies. Loss of the wild-type tlpA allele and introduction of the Kanr gene was verified by PCR analysis with the primers tlpA204 (5'-CACACGAGAATTGAAATCGC-3') and tlpA711 (5'-CGGCTCGAGTCATTTCAATAATG-3') in the tlpA gene and the primers kt (5'-CGGCCACAGTCGATGAATCC-3') and k2 (5'-CGGTGCCCTGAATGAACTGC-3') in the Kanr gene.
Plasmid construction. To construct a TlpA mammalian expression vector, genomic DNA from the serovar Enteritidis strain LK5 was prepared by using the QIAGEN genomic DNA isolation kit according to the manufacturer's instructions and used for PCR amplification of the tlpA gene with the primers tlpAFL/F (5'-CCGGGATCCATGTCAGAGATCTCTCG-3') and tlpAFL/R (5'-CGGCTCGAGTCATTTCAATAATG-3'). The resulting fragment was digested with BamHI and XhoI and ligated into either pcDNA3myc vector (Invitrogen) digested with BamHI and XhoI to give pMycTlpA, into pEGFP-C1 vector (Invitrogen) digested with BglII and SalI to give pGFPTlpA, or into pcDNA3HA digested with BamHI and XhoI to give pT7HATlpA. To construct a tlpA bacterial expression vector pHATlpA, pT7HATlpA was digested with HindIII and XhoI, and the resulting tlpA-containing fragment ligated to pSC101 was digested with HindIII and SalI. To construct a SipB mammalian expression plasmid, the sipB gene was PCR amplified from serovar Typhimurium LT2 genomic DNA (ATCC) by using primers the SipB/F (5'-CCGGGATCCATGGTAAATGACGC-3') and SipB/R (5'-CCGCTCGAGTTATGCACGACTCTG--3'). The resulting fragment was digested with BamHI and XhoI and cloned into pcDNA3HA digested with the same to give plasmid pHASipB.
Luciferase gene reporter assays. Luciferase reporter assays were performed as previously described (10). Briefly, HEK293N cells were seeded in 12-well plates and transfected by using Superfect transfection reagent (QIAGEN) according to the manufacturer's recommended protocol. Cells were transfected with 50 ng of pNF-B-luc and 10 ng of pTK-RL reporter vectors (Stratagene) and various amounts of the relevant expression plasmids as described in the figure legends, maintaining the total amount of DNA constant by using pcDNA3Myc empty vector. At 36 h after transfection, cells were treated with tumor necrosis factor alpha (TNF-) or IL-1 (at 20 and 50 ng/ml, respectively) for 6 h where indicated. Activities from firefly and Renilla luciferases were assayed by using the dual-luciferase reporter assay system (Promega).
EMSA. HEK293N cells were seeded into 10-cm2 plates and transfected with relevant plasmids by using Superfect transfection reagent (QIAGEN) according to the manufacturer's recommendations. Nuclear extracts were prepared from these cells, and electrophoretic mobility shift assays (EMSAs) were carried out as described previously (10). Briefly, double-stranded oligonucleotides containing a consensus NF-B binding site (Promega) or STAT-1 binding site (Santa Cruz Biotechnology) were end labeled with [-32P]ATP (Perkin-Elmer Life Sciences) using T4 polynucleotide kinase (Promega). After purification with MicroSpin G-25 columns (Amersham Biosciences), the labeled probe (15 fmol) was incubated with 2 μg of nuclear extracts for 25 min at room temperature. The nuclear extracts were also incubated with specific antibodies recognizing the NF-B subunits p65 and p50 (Santa Cruz Biotechnology) before the binding reaction or with a 100-fold molar excess of unlabeled DNA probe as competitor. All complexes were separated by electrophoresis in nondenaturing 5% polyacrylamide gels at 4°C. After drying, gels were exposed to X-ray film at –70°C.
Bacterial infection of THP-1 macrophages. THP-1 monocytes were grown in RPMI containing 10% FBS and seeded into 24-well dishes and allowed to differentiate to macrophages in the presence of 75 ng of phorbol 12-myristate 13-acetate (Calbiochem)/ml for 24 h. Prior to infection, serovar Enteritidis strains were grown overnight in high-salt LB broth (LB broth plus 300 mM NaCl) and then subcultured 1:20 and allowed to grow for an additional 2 to 3 h. Bacteria were washed twice in phosphate-buffered saline and then resuspended in THP-1 growth media. THP-1 cells were washed twice in phosphate-buffered saline and then overlaid with bacterial suspension such that the multiplicity of infection was 10 bacteria per mammalian cell. After 30 min of invasion, the monolayer was washed three times in RPMI, and extracellular bacteria were killed by incubating the cells in growth medium containing 100 μg of gentamicin (Sigma)/ml. After 90 min, media was replaced with THP-1 growth medium containing 10 μg of gentamicin/ml. At times 2, 4, 8, and 12 h after bacterial addition, mammalian cells were lysed in 1% Triton X-100. Serial 10-fold dilutions of lysates were plated on LB plates or LB plates containing 30 μg of kanamycin/ml, and viable colonies were counted to determine intracellular bacterial numbers. For IL-1 measurements, 0.1 ml of culture supernatant was removed prior to cell lysis and used for enzyme-linked immunosorbent assay (ELISA).
Bacterial infection of mice. Female BALB/c mice between 6 and 8 weeks old were purchased from Charles River Laboratories and housed at The Burnham Institute according to local Animal Care Advisory Committee guidelines. The estimated lethal dose of serovar Enteritidis LK5 in BALB/c mice has previously been determined (8). Overnight bacterial cultures grown in LB medium were diluted in sterile 0.15 M NaCl to yield 106 CFU per ml. Mice were inoculated orally with 0.1 ml of the diluted bacteria by using a feeding needle. All inocula were quantified by plating serial 10-fold dilutions of the cultures onto LB and counting the viable colonies.
RESULTS
Identification of a S. enterica serovar Enteriditis protein containing a TIR-like domain. A cascade of PSI-BLAST searches (18) were used to search for TIR domain homologs in a set of sequences available from a variety of microbial DNA sequence databases, as described in Materials and Methods. Potential homologs identified in this search were further validated by using more sensitive profile-profile searches and structure modeling. This computational approach can detect homology between proteins that have little primary sequence similarity by using additional information from multiple alignments, structure, and secondary-structure features. From these searches, more than 200 bacterial TIR homologs were identified, expanding on the previously noted protein family (17). Proteins from this family display a very diverse domain structure and an unusual phylogenetic distribution. Proteins from several human and animal pathogens form a well-defined subfamily, with their TIR-like domains displaying many features of the eukaryotic TIRs (unpublished observations). Among the latter group was a hypothetical protein from the human pathogen S. enterica serovar Enteritidis. We have named the gene encoding this protein TlpA for TIR-like protein A.
TlpA is a 236-amino-acid protein, annotated in sequence databases as a hypothetical protein of unknown function. Sequence comparison of the hypothetical TIR domain of TlpA (amino acids 101 to 236) with the human TIR-containing proteins MyD88, TIRAP, and TLR4 showed significant sequence similarity (sequence identity of 28% and PSI-BLAST e.value of E-4 in second iteration) with conservation of the box 1 and box 2 sequences that are characteristic of all known members of the TLR/IL-1R superfamily (Fig. 1a). The domain architecture of TlpA indicated that it is comprised of a C-terminal TIR domain and an N-terminal 100-residue coiled-coil domain (Fig. 1b). Additional comparison between Salmonella genomes reveals that the tlpA gene is located in a DNA island that is unique to the S. enterica serovars Enteritidis and Dublin (S. Maloy, unpublished data). Sequence analysis of the regions flanking tlpA reveals the presence of genes, indicating phage origin of tlpA (integrases and phage signature proteins), as well as several proteins of unknown function and pilin-related domains (9). The latter are extracellular proteins involved in cell adhesion, suggesting that proteins in this genomic region are secreted. However, because the genome sequence of serovar Enteritidis is discontinuous in this region, we cannot determine whether tlpA resides within a cluster of genes involved in type III or type IV secretion.
Close homologs of serovar Enteritidis with the same domain organization and strong sequence similarity (>40% sequence identity) were found in an invasive, community-acquired methicillin-susceptible Staphylococcus aureus strain (MSSA476) and in Brucella melitensis, the agent of brucellosis, a deadly disease of humans and animals. In phylogenetic analysis, this group of bacterial TIR domains clusters together despite the distant relation between the organisms (gram positive, gram negative), supporting the hypothesis of their spread by lateral transfer.
Suppression of TIR-dependent NF-B activation by TlpA. Since TlpA protein was identified on the basis of its homology to the TIR domain of signaling proteins of the innate immune pathway, we explored the effects of TlpA on TIR-dependent activation of NF-B in mammalian cells. Accordingly, HEK293 cells were cotransfected with increasing amounts of plasmid encoding TlpA and a fixed amount of a luciferase gene reporter plasmid containing five NF-B-binding sites for measuring transcriptional activity. Despite confirmation of TlpA protein production in HEK293 cells by immunoblotting, expression of TlpA itself had no effect on basal levels of NF-B activity (data not shown). We therefore assessed the ability of TlpA to suppress NF-B activation induced by a variety of proteins that contain TIR domains, including a constitutively active TLR4 receptor (CD4:TLR4) and the MyD88 adaptor protein. Cells were also stimulated by incubation with the proinflammatory cytokine, IL-1, whose receptor contains TIR domains, and TNF-, which signals via TIR-independent mechanisms. Using the same luciferase-based reporter gene assay we found that TlpA suppressed NF-B activation induced by components of the TLR signaling pathway TLR4, MyD88, and IL-1 in a dose-dependent manner (Fig. 2a). However, TlpA did not suppress activation induced by TNF-, suggesting specificity for the TLR/IL-1R pathway. In contrast to its effects on NF-B, TlpA expression did not interfere with the activity of other transcription factors (as assessed by reporter gene assays), including AP-1 and STAT-1 (not shown). Furthermore, in experiments where TlpA was cotransfected with plasmids encoding CD4:TLR4 or MyD88, we show the effects on NF-B activation were not due to alterations in protein expression (Fig. 2b).
To determine whether the observed reduction of NF-B activity correlated with reduced NF-B DNA-binding activity in TlpA-expressing cells, we performed EMSAs with 32P-labeled, double-stranded DNA oligonucleotides containing a consensus NF-B-binding site. Nuclear extracts were prepared from unstimulated cells and cells that had been stimulated by culturing with TIR-dependent and TIR-independent cytokines, IL-1 and TNF-, respectively. Incubation of the nuclear extracts with 32P-labeled NF-B probe, followed by gel electrophoresis, then allowed visualization of the relative levels of the NF-B DNA-binding activity. Extracts from cells exposed to IL-1 or TNF- contained abundant NF-B DNA-binding activity when derived from cells transfected with control plasmid (Fig. 3a). In contrast, NF-B-binding activity was markedly reduced in extracts derived from cells expressing TlpA when IL-1 was applied as a stimulus but not when TNF- was used, a finding consistent with luciferase reporter gene results. Competition experiments with unlabeled wild-type and mutant NF-B oligonucleotide probes confirmed the specificity of the EMSA results (Fig. 3a). Moreover, TlpA had no effect on the DNA-binding activity of an unrelated transcription factor STAT-1, further confirming specificity.
Immunoblot analysis of the nuclear levels of NF-B subunits p65 and p50 were compared to control and cytokine-treated cells (Fig. 3b). Nuclear levels of p65 and p50 in cells treated with TNF- were comparable in TlpA-expressing and control-transfected cells, confirming that TNF--mediated NF-B activation is unaffected by TlpA expression. In contrast, levels of p65 and p50 were reduced in the nuclei of IL-1-treated cells expressing TlpA compared to control-transfected cells. Levels of another nuclear transcription factor TFII-I were not affected by TlpA overexpression, confirming the specificity of the results. These data suggest that expression of TlpA suppresses nuclear translocation of the NF-B complex.
TlpA contributes to virulence of serovar enteritidis in both a cell culture and a mouse infection model. To determine whether TlpA plays a role during infection, the tlpA gene of serovar Enteritidis LK5 was disrupted by replacement with a Kanr cassette using the "Red Swap" method of Datsenko and Wanner (6). Loss of the allele was confirmed by PCR with locus-specific primers.
The resultant strain (tlpA::kan) was used to infect human monocyte-derived macrophage cell line THP-1. Bacteria were cocultured with macrophages in the absence of antibiotic and allowed to invade for 30 min prior to antibiotic treatment to kill extracellular bacteria. At 2, 4, 8, and 12 h after bacterial addition, macrophages were lysed, and intracellular CFU were enumerated by dilution plating of the lysates. Infection results indicate that although the tlpA::kan strain has no defect in invasion, it is unable to survive and/or replicate in macrophages compared to wild-type, isogenic serovar Enteritidis strain LK5 (see Fig. 5a). This effect is specific for TlpA, as ectopic expression of TlpA on a low-copy plasmid restored intracellular survival to the tlpA::kan strain (Fig. 4a).
To further test the role of TlpA in disease, mice were orally infected with a lethal dose of wild-type serovar Enteritidis LK5 and the tlpA::kan mutant strain. Survival of infected mice was then compared. Disruption of the tlpA gene lead to prolonged survival of infected mice compared to animals infected with the isogenic wild-type serovar Enteritidis strain LK5. Typical results from an infection experiment are shown in Fig. 5b. The average survival times derived from three such experiments were 4.3 ± 0.3 days for mice infected with wild-type bacteria versus 7.8 ± 0.5 days for mice infected with the tlpA mutant. Analysis of animals 5 days postinfection with either LK5 or tlpA::kan showed that there was a significant decrease in the number of bacteria recovered per spleen from mice infected with the mutant strain (Fig. 4c).
TlpA induces caspase activation and IL-1 secretion. Several studies have shown the importance of caspase-1 activation for in vivo dissemination of Salmonella (14, 20-22). Caspase-1, also known as IL-1 converting enzyme (ICE), is a cysteine-specific protease that cleaves and activates proinflammatory cytokines pro-IL-1 and pro-IL-18 (31). The Salmonella SipB protein and the closely related Shigella IpaB protein have been shown to bind and activate caspase-1 in host cells (11, 30), providing examples of virulence factors that regulate this protease involved in cytokine elaboration. Activation of TLRs also causes activation of caspase-1, although the mechanism is unknown (31). We therefore explored the effects of TlpA expression on caspase-1 activity and IL-1 secretion by using a cell transfection method previously used for studies of SipB (11).
Accordingly, HEK293 cells expressing recombinant murine caspase-1 and murine pro-IL-1 were transfected with a plasmid encoding myc-tagged TlpA. Quantitative ELISA was used to measure the relative amount of mature IL-1 secreted into the cell culture media 24 h after transfection. Expression of TlpA together with procaspase-1 increased IL-1 secretion 8-fold over vector control, representing a level comparable to that generated by serovar Typhimurium SipB (Fig. 5a). In contrast, transfecting cells individually with plasmids encoding either TlpA or procaspase-1 did not result in significant IL-1 secretion (see supplemental data at http://www.burnham.org/papers/IAI1013-05/SupplementalFigures.ppt), demonstrating that TlpA only induces IL-1 secretion when procaspase-1 is present. The ability of TlpA to induce IL-1 secretion was not a promiscuous feature of TIR domains because expression of the TIR domain of MyD88 did not induce IL-1 secretion. TlpA-induced secretion of mature IL-1 was shown to be caspase dependent, since both the caspase-1/8-specific protein inhibitor CrmA and a chemical pan-caspase inhibitor zVAD-fmk suppressed TlpA-induced IL-1 secretion (Fig. 5a). Immunoblot analysis confirmed that caspase inhibitors did not interfere with expression of TlpA (not shown), thus excluding a trivial explanation for the results. Taken together, these data indicate that TlpA induces caspase-dependent IL-1 secretion.
To extend these gene transfer results to a more physiological context, we compared IL-1 secretion induced by infection of THP-1 macrophages with wild-type versus mutant serovar Enteritidis strains. Accordingly, the relative amounts of IL-1 in culture supernatants were measured by quantitative ELISA at various times after bacterial addition. Infection with the wild-type serovar Enteritidis strain resulted in steadily increasing levels of secreted IL-1 starting approximately 4 h after bacterial addition and continuing through the 12 h period assessed (Fig. 5b). After 12 h, infected THP-1 macrophages began to die, making later measurements of IL-1 production problematic (not shown). Disruption of the tlpA gene rendered the mutant strain (tlpA::kan) unable to sustain the high levels of IL-1 secretion after infection of macrophages. The diminution of IL-1 levels at 8 h after bacterial addition coincides with a decrease in the numbers of intracellular bacteria (as seen above), suggesting the two events may be connected. Introduction of the tlpA gene into the mutant strain on a low-copy plasmid restored IL-1 secretion to the levels observed with infections using the wild-type strain (Fig. 5b). Taken together, these cell infection data confirm a role of TlpA in regulating secretion of the proinflammatory cytokine IL-1, a finding consistent with the gene transfection data above showing that TlpA can induce caspase-1-dependent IL-1 secretion.
DISCUSSION
To identify candidate novel, virulence genes, we examined bacterial genomes for hypothetical proteins with homology to mammalian TIR domains. We postulated that bacterial proteins mimicking these eukaryotic signaling molecules would block inflammatory responses and thus blunt host defense. The serovar Enteritidis TlpA protein was thus identified. Genetic and biochemical characterization of TlpA suggests that it is important for bacterial virulence in vivo and that it modulates host defense mechanisms involved in regulation of NF-B and caspase activation. The observation that TlpA can suppress NF-B while inducing caspase-1 activation suggests that this protein may act in different ways during the course of infection to either promote or suppress cytokine induction, perhaps as a mechanism to disseminate after successful invasion.
TlpA is a novel suppressor of NF-B activation. Given the homology of TlpA to mammalian TIR-containing proteins, we considered that it might suppress NF-B by binding to members of the TLR/IL-1R superfamily and preventing downstream signaling. Consistent with this possibility, TlpA suppressed NF-B activation induced by TIR-dependent stimuli such as TLR4, MyD88, and IL-1, but not by the TIR-independent cytokine TNF-. However, we have been unable to detect direct interaction between TlpA and a variety of mammalian TIR-family proteins.
We found that TlpA is required for induction of IL-1 secretion in the context of infection of cultured macrophages with serovar Enteritidis. Bacteria lacking TlpA induced far less IL-1 production compared to the wild-type strain, even though such gram-negative organisms produce lipopolysaccharide and should trigger caspase-1 activation pathways via TLRs. Moreover, ectopic expression of TlpA was sufficient to induce caspase-dependent elaboration of IL-1 from mammalian cells, indicating a direct effect. Taken together, these experiments indicated that TlpA is both necessary and sufficient for inducing IL-1 production. How TlpA induces caspase activation leading to IL-1 secretion is unclear. Some bacterial virulence proteins, such as SipB from serovar Typhimurium and IpaB from Shigella species, appear to directly bind pro-caspase-1 and induce its activation, but TlpA shares neither sequence similarity nor predicted structural similarity to these proteins. Cellular TIR domain-containing proteins are known to transduce signals that trigger caspase-1 activation; thus, the predicted structural similarity of TlpA to animal TIRs might allow it to modulate TIR-dependent signaling events linked to caspase-1 activation.
It is not entirely clear why bacteria would find it advantageous to induce IL-1 production. In the case of serovar Typhimurium, it has been suggested that IL-1 may recruit additional host inflammatory cells that can serve as reservoirs for bacterial replication or alter vascular blood flow in ways that promote deep tissue invasion (14, 21, 22). Alternatively, IL-1 production may represent a by-product of caspase activation, where bacterial virulence factors use caspase-1 or other members of the caspase family to kill host defense cells.
Regardless of the mechanisms by which TlpA activates caspases and suppresses NF-B, our comparisons of genetically engineered bacteria indicate that TlpA represents a virulence gene. Without TlpA, mice survive serovar Enteritidis infection longer. Curiously, TlpA deficiency reduces accumulation of serovar Enteritidis inside infected macrophages, implying that TlpA is required for intracellular growth or survival of bacteria. We speculate therefore that TlpA is required for thwarting activation of endogenous host cell pathways that suppress accumulation of intracellular bacteria. Given that the genomes of many other pathogenic and nonpathogenic microbial species contain open reading frames potentially encoding TIR-like proteins (unpublished observations), it will be interesting to compare the functions of these gene products to ascertain whether they exhibit conserved functions.
It is interesting that proteins containing predicted TIR-like domains are widespread in bacteria (unpublished observations). These proteins show an unusual distribution in bacterial species, with very close homologs found in unrelated bacteria and, conversely, TlpA proteins from strains of the same bacteria being distantly related. In addition, all genes predicted to encode proteins with TIR-like domains are found in genomic regions with phage origins. These observations suggest that the family of bacterial TIR-like genes evolved by lateral transfer. It is unknown whether the proteins of this family fulfill intrinsic bacterial functions, or whether they represent a widespread, proto-pathogenic adaptation of bacteria. At least for the TIR-like protein of S. enterica serovar Enteritidis, it appears that members of this family represent a new class of virulence factors adapted to alter intracellular signaling pathways of eukaryotic host cells and thereby afford bacteria with survival and dissemination advantages in vivo.
ACKNOWLEDGMENTS
We thank Loredana Fiorentino, Marcin Feder, Juan Zapata, Jason Damiano, Rhonda Croxton, April Stanley, and Stanley Maloy for helpful discussion. We also thank Christina Kress and Adriana Charbono for expert assistance with mouse infections.
This study was supported by a grant from the National Institutes of Health (AI-055789) and a Kirchstein NRSA/NIH fellowship (5T32AG00252).
Present address: Harvard Medical School/NEPRC, 1 Pine Hill Dr., Southborough, MA 01772.
REFERENCES
1. Akira, S., and K. Takeda. 2004. Toll-like receptor signalling. Nat. Rev. Immunol. 4:499-511.
2. Athman, R., and D. Philpott. 2004. Innate immunity via Toll-like receptors and Nod proteins. Curr. Opin. Microbiol. 7:25-32.
3. Beutler, B. 2004. Inferences, questions and possibilities in Toll-like receptor signalling. Nature 430:257-263.
4. Beutler, B., K. Hoebe, X. Du, and R. J. Ulevitch. 2003. How we detect microbes and respond to them: the Toll-like receptors and their transducers. J. Leukoc. Biol. 74:479-485.
5. Collier-Hyams, L. S., H. Zeng, J. Sun, A. D. Tomlinson, Z. Q. Bao, H. Chen, J. L. Madara, K. Orth, and A. S. Neish. 2002. Cutting edge: Salmonella AvrA effector inhibits the key proinflammatory, antiapoptotic NF-B pathway. J. Immunol. 169:2846-2850.
6. Datsenko, K. A., and B. L. Wanner. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97:6640-6645.
7. Dunne, A., and L. A. O'Neill. 2003. The interleukin-1 receptor/Toll-like receptor superfamily: signal transduction during inflammation and host defense. Sci. STKE 2003:re3.
8. Edwards, R. A., D. M. Schifferli, and S. R. Maloy. 2000. A role for Salmonella fimbriae in intraperitoneal infections. Proc. Natl. Acad. Sci. USA 97:1258-1262.
9. Fellowship for Interpretation of Genomes. 2005. The SEED: an annotation/analysis tool. Fellowship for Interpretation of Genomes, University of Chicago, Chicago, Ill. [Online.]. http://theseed.uchicago.edu/FIG/index.cgi.
10. Fiorentino, L., C. Stehlik, V. Oliveira, M. E. Ariza, A. Godzik, and J. C. Reed. 2002. A novel PAAD-containing protein that modulates NF-B induction by cytokines tumor necrosis factor- and interleukin-1. J. Biol. Chem. 277:35333-35340.
11. Hersh, D., D. M. Monack, M. R. Smith, N. Ghori, S. Falkow, and A. Zychlinsky. 1999. The Salmonella invasin SipB induces macrophage apoptosis by binding to caspase-1. Proc. Natl. Acad. Sci. USA 96:2396-2401.
12. Hornef, M. W., M. J. Wick, M. Rhen, and S. Normark. 2002. Bacterial strategies for overcoming host innate and adaptive immune responses. Nat. Immunol. 3:1033-1040.
13. Jaroszewski, L., L. Rychlewski, Z. Li, W. Li, and A. Godzik. 2005. FFAS03: a server for profile-profile sequence alignments. Nucleic Acids Res. 33:W284-W288.
14. Jarvelainen, H. A., A. Galmiche, and A. Zychlinsky. 2003. Caspase-1 activation by Salmonella. Trends Cell Biol. 13:204-209.
15. Karin, M., and A. Lin. 2002. NF-B at the crossroads of life and death. Nat. Immunol. 3:221-227.
16. Knodler, L. A., J. Celli, and B. B. Finlay. 2001. Pathogenic trickery: deception of host cell processes. Nat. Rev. Mol. Cell. Biol. 2:578-588.
17. Koonin, E. V., and L. Aravind. 2002. Origin and evolution of eukaryotic apoptosis: the bacterial connection. Cell Death Differ. 9:394-404.
18. Li, W., F. Pio, K. Pawlowski, and A. Godzik. 2000. Saturated BLAST: an automated multiple intermediate sequence search used to detect distant homology. Bioinformatics 16:1105-1110.
19. Lupas, A., M. Van Dyke, and J. Stock. 1991. Predicting coiled coils from protein sequences. Science 252:1162-1164.
20. Monack, D. M., C. S. Detweiler, and S. Falkow. 2001. Salmonella pathogenicity island 2-dependent macrophage death is mediated in part by the host cysteine protease caspase-1. Cell Microbiol. 3:825-837.
21. Monack, D. M., D. Hersh, N. Ghori, D. Bouley, A. Zychlinsky, and S. Falkow. 2000. Salmonella exploits caspase-1 to colonize Peyer's patches in a murine typhoid model. J. Exp. Med. 192:249-258.
22. Monack, D. M., W. W. Navarre, and S. Falkow. 2001. Salmonella-induced macrophage death: the role of caspase-1 in death and inflammation. Microbes Infect. 3:1201-1212.
23. Orth, K. 2002. Function of the Yersinia effector YopJ. Curr. Opin. Microbiol. 5:38-43.
24. Orth, K., Z. Xu, M. B. Mudgett, Z. Q. Bao, L. E. Palmer, J. B. Bliska, W. F. Mangel, B. Staskawicz, and J. E. Dixon. 2000. Disruption of signaling by Yersinia effector YopJ, a ubiquitin-like protein protease. Science 290:1594-1597.
25. Rosenberger, C. M., and B. B. Finlay. 2003. Phagocyte sabotage: disruption of macrophage signalling by bacterial pathogens. Nat. Rev. Mol. Cell. Biol. 4:385-396.
26. Rychlewski, L., L. Jaroszewski, W. Li, and A. Godzik. 2000. Comparison of sequence profiles. Strategies for structural predictions using sequence information. Protein Sci. 9:232-241.
27. Silverman, N., and T. Maniatis. 2001. NF-B signaling pathways in mammalian and insect innate immunity. Genes Dev. 15:2321-2342.
28. Stebbins, C. E., and J. E. Galan. 2001. Structural mimicry in bacterial virulence. Nature 412:701-705.
29. Tato, C. M., and C. A. Hunter. 2002. Host-pathogen interactions: subversion and utilization of the NF-B pathway during infection. Infect. Immun. 70:3311-3317.
30. Thirumalai, K., K. S. Kim, and A. Zychlinsky. 1997. IpaB, a Shigella flexneri invasin, colocalizes with interleukin-1-converting enzyme in the cytoplasm of macrophages. Infect. Immun. 65:787-793.
31. Tschopp, J., F. Martinon, and K. Burns. 2003. NALPs: a novel protein family involved in inflammation. Nat. Rev. Mol. Cell. Biol. 4:95-104.
32. Xu, Y., X. Tao, B. Shen, T. Horng, R. Medzhitov, J. L. Manley, and L. Tong. 2000. Structural basis for signal transduction by the Toll/interleukin-1 receptor domains. Nature 408:111-115.(Ruchi M. Newman, Prabhaka)
ABSTRACT
Many important bacterial virulence factors act as mimics of mammalian proteins to subvert normal host cell processes. To identify bacterial protein mimics of components of the innate immune signaling pathway, we searched the bacterial genome database for proteins with homology to the Toll/interleukin-1 receptor (TIR) domain of the mammalian Toll-like receptors (TLRs) and their adaptor proteins. A previously uncharacterized gene, which we have named tlpA (for TIR-like protein A), was identified in the Salmonella enterica serovar Enteritidis genome that is predicted to encode a protein resembling mammalian TIR domains, We show that overexpression of TlpA in mammalian cells suppresses the ability of mammalian TIR-containing proteins TLR4, IL-1 receptor, and MyD88 to induce the transactivation and DNA-binding activities of NF-B, a downstream target of the TIR signaling pathway. In addition, TlpA mimics the previously characterized Salmonella virulence factor SipB in its ability to induce activation of caspase-1 in a mammalian cell transfection model. Disruption of the chromosomal tlpA gene rendered a virulent serovar Enteritidis strain defective in intracellular survival and IL-1 secretion in a cell culture infection model using human THP1 macrophages. Bacteria with disrupted tlpA also displayed reduced lethality in mice, further confirming an important role for this factor in pathogenesis. Taken together, our findings demonstrate that the bacterial TIR-like protein TlpA is a novel prokaryotic modulator of NF-B activity and IL-1 secretion that contributes to serovar Enteritidis virulence.
INTRODUCTION
The innate immune response is the first line of host defense against microbial and fungal invasion. Components of the innate immune pathway are evolutionarily conserved in insects, plants, and vertebrates. One of the primary mediators of the innate immune response is the Toll-like/interleukin-1 receptor (IL-1R) superfamily of cell surface receptors (reviewed in references 1, 2, 3, 4, and 7). In mammals, members of this superfamily are divided into three subgroups. Subgroup 1 is comprised of the interleukin receptors, which bind cytokines. Subgroup 2 is comprised of Toll-like receptors, which display extracellular leucine-rich repeats that bind microbial products. The third subgroup is comprised of the cytosolic adaptor proteins used to convey extracellular signals from both subgroup 1 and 2 receptors to downstream effectors. The structural motif common to all members of the Toll-like/IL-1R superfamily is a cytosolic domain called the Toll/IL-1R (TIR) domain. This domain is required for interaction of TIR-containing Toll-like/IL-1R surface receptors to downstream TIR-containing signaling partners through homotypic interactions among TIR domains. The three-dimensional structure of the TIR domain of IL-1R consists of five parallel -sheets surrounded by two layers of parallel -helices (32).
TLR/IL-1R family proteins activate signal transduction pathways in response to proinflammatory cytokines or microbial derived substances such as lipopolysaccharide, peptidoglycan, and other cell wall components. Binding of appropriate ligands to these receptors engages a signaling cascade that converges upon the NF-B family of transcription factors (reviewed in reference 27). This cascade involves assembly of a multiprotein complex that includes (in mammals) TIR-containing adapters such as MyD88 or TIRAP, IRAK family kinases, and TRAF6, resulting in activation of the IB kinase complex (IKK) that phosphorylates the NF-B inhibitor IB and targets it for ubiquitination and proteasome-dependent destruction. Activation of NF-B in turn leads to production of proteins involved in inflammation, host defense, apoptosis suppression, and tissue repair (reviewed in reference 15).
Both gram-negative and gram-positive bacterial pathogens produce a variety of virulence factors that interact with host molecules and modulate specific cellular processes important for immune defense including phagocytosis, signal transduction, cytokine production, and cytoskeletal rearrangements (reviewed in reference 12, 16, 25, 28, and 29). Virulence factors can be presented either on the surface of the bacterium, injected into the host cell upon bacterial attachment to the cell surface, or secreted through specialized secretion systems. Bacterial virulence factors that interfere with NF-B induction have been identified. For example, the YopJ protein of Yersinia blocks signal transduction events required for NF-B induction by cytokine receptors and Toll-like receptors (TLRs). YopJ is a putative cysteine protease that cleaves the ubiquitin-like protein SUMO from target molecules, although the specific host cell substrates of this bacterial SUMO lyase remain unclear (23, 24). Similarly, Salmonella enterica serovar Typhimurium encodes a homologous putative cysteine protease AvrA that also thwarts TLR and cytokine-induced activation of NF-B, apparently acting at a point distal to YopJ in the signal transduction pathway responsible of NF-B activation (5). By suppressing NF-B activation, virulence factors such as YopJ and AvrA not only reduce the host inflammatory response but also promote host cell apoptosis due to the importance of NF-B in inducing expression of antiapoptotic genes required for protection against cytokine-induced apoptosis (15).
We postulated that identification of bacterial proteins showing sequence homology with components of the innate immune pathway would lead to the identification of novel bacterial virulence factors. By searching for proteins with structural homology to TIR-like domains, we have identified such a protein from the serovar Enteritidis genome, which we designated TlpA (for TIR-like protein A). We show that the TlpA protein suppresses NF-B induction by stimuli that involve TIR domain proteins, and we provide evidence that TlpA is required for full virulence. In addition to suppressing TIR-dependent NF-B activation, however, TlpA possesses other functions. Analogous to the previously characterized SipB protein of serovar Typhimurium, TlpA expression promotes activation of the protease caspase-1, resulting in caspase-dependent secretion of IL-1 and host cell apoptosis. Thus, the TlpA protein of serovar Enteritidis represents a novel bacterial virulence factor possessing at least two potent mechanisms for modification of host defense.
MATERIALS AND METHODS
Bioinformatics. Amino acid sequences of human TIR domains were used to search public protein sequence databases, including the database of unfinished microbial genomes at National Center for Biotechnology Information (http://ncbi.nig.gov) and the SEED database and genome annotation system (http://theseed.uchicago.edu/FIG/index.cgi). A cascade of PSI-BLAST searches, as implemented in the SaturatedBLAST program (18), were used to identify bacterial proteins with potential TIR domains. Up to five iterations of PSI-BLAST were run, and all proteins with an e value below 0.005 were taken for further consideration. All thus identified proteins were subject to further analysis and verification by reverse PSI-BLAST searches, distant homology analysis, and three-dimensional model building using the FFAS server (13, 26).
Cell culture. HEK293N and HEK293T cells were maintained in Dulbecco modified Eagle medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen), 100 U of penicillin/ml, and 100 μg of streptomycin/ml at 37°C in 5% CO2. THP-1 cells were maintained in RPMI supplemented with 10% FBS, penicillin, and streptomycin at 37°C in 5% CO2. All bacterial strains were grown in standard LB broth unless otherwise indicated.
Bacterial strains. Wild-type serovar Enteritidis LK5 was used for infection experiments (gift of S. Maloy, San Diego State University, San Diego, Calif.). The LK5 tlpA::kan mutant strain was constructed by the method of Datsenko and Wanner (6). Briefly, PCR primers tlp/P1 (5'-ATGTCAGAGATCTCTCGTCATAAAAATGATATTTCAAGATGTAACTCTGTGTAGGCTGGAG CTGCTT-3') and tlp/P4 (5'-TCATTTCAATAATGATTCAAGCTGTTCAGCAATTTCATCAATGGATTCCGGGGATCCGTCGACC-3') were synthesized with homology to the FLP recognition target sites flanking the kanamycin resistance (Kanr) gene of template plasmid pKD13. These primers also contained 44- to 48-nucleotide extensions (underlined) that are homologous to regions flanking the tlpA gene. The PCR product generated using these primers with pKD13 as a template was gel purified and electroporated into the LK5 strain carrying the Red recombinase helper plasmid pKD46. The resulting Kanr transformants were used for P22 transduction of the tlpA::kan allele into a fresh LK5 background and subsequent selection for Kanr colonies. Loss of the wild-type tlpA allele and introduction of the Kanr gene was verified by PCR analysis with the primers tlpA204 (5'-CACACGAGAATTGAAATCGC-3') and tlpA711 (5'-CGGCTCGAGTCATTTCAATAATG-3') in the tlpA gene and the primers kt (5'-CGGCCACAGTCGATGAATCC-3') and k2 (5'-CGGTGCCCTGAATGAACTGC-3') in the Kanr gene.
Plasmid construction. To construct a TlpA mammalian expression vector, genomic DNA from the serovar Enteritidis strain LK5 was prepared by using the QIAGEN genomic DNA isolation kit according to the manufacturer's instructions and used for PCR amplification of the tlpA gene with the primers tlpAFL/F (5'-CCGGGATCCATGTCAGAGATCTCTCG-3') and tlpAFL/R (5'-CGGCTCGAGTCATTTCAATAATG-3'). The resulting fragment was digested with BamHI and XhoI and ligated into either pcDNA3myc vector (Invitrogen) digested with BamHI and XhoI to give pMycTlpA, into pEGFP-C1 vector (Invitrogen) digested with BglII and SalI to give pGFPTlpA, or into pcDNA3HA digested with BamHI and XhoI to give pT7HATlpA. To construct a tlpA bacterial expression vector pHATlpA, pT7HATlpA was digested with HindIII and XhoI, and the resulting tlpA-containing fragment ligated to pSC101 was digested with HindIII and SalI. To construct a SipB mammalian expression plasmid, the sipB gene was PCR amplified from serovar Typhimurium LT2 genomic DNA (ATCC) by using primers the SipB/F (5'-CCGGGATCCATGGTAAATGACGC-3') and SipB/R (5'-CCGCTCGAGTTATGCACGACTCTG--3'). The resulting fragment was digested with BamHI and XhoI and cloned into pcDNA3HA digested with the same to give plasmid pHASipB.
Luciferase gene reporter assays. Luciferase reporter assays were performed as previously described (10). Briefly, HEK293N cells were seeded in 12-well plates and transfected by using Superfect transfection reagent (QIAGEN) according to the manufacturer's recommended protocol. Cells were transfected with 50 ng of pNF-B-luc and 10 ng of pTK-RL reporter vectors (Stratagene) and various amounts of the relevant expression plasmids as described in the figure legends, maintaining the total amount of DNA constant by using pcDNA3Myc empty vector. At 36 h after transfection, cells were treated with tumor necrosis factor alpha (TNF-) or IL-1 (at 20 and 50 ng/ml, respectively) for 6 h where indicated. Activities from firefly and Renilla luciferases were assayed by using the dual-luciferase reporter assay system (Promega).
EMSA. HEK293N cells were seeded into 10-cm2 plates and transfected with relevant plasmids by using Superfect transfection reagent (QIAGEN) according to the manufacturer's recommendations. Nuclear extracts were prepared from these cells, and electrophoretic mobility shift assays (EMSAs) were carried out as described previously (10). Briefly, double-stranded oligonucleotides containing a consensus NF-B binding site (Promega) or STAT-1 binding site (Santa Cruz Biotechnology) were end labeled with [-32P]ATP (Perkin-Elmer Life Sciences) using T4 polynucleotide kinase (Promega). After purification with MicroSpin G-25 columns (Amersham Biosciences), the labeled probe (15 fmol) was incubated with 2 μg of nuclear extracts for 25 min at room temperature. The nuclear extracts were also incubated with specific antibodies recognizing the NF-B subunits p65 and p50 (Santa Cruz Biotechnology) before the binding reaction or with a 100-fold molar excess of unlabeled DNA probe as competitor. All complexes were separated by electrophoresis in nondenaturing 5% polyacrylamide gels at 4°C. After drying, gels were exposed to X-ray film at –70°C.
Bacterial infection of THP-1 macrophages. THP-1 monocytes were grown in RPMI containing 10% FBS and seeded into 24-well dishes and allowed to differentiate to macrophages in the presence of 75 ng of phorbol 12-myristate 13-acetate (Calbiochem)/ml for 24 h. Prior to infection, serovar Enteritidis strains were grown overnight in high-salt LB broth (LB broth plus 300 mM NaCl) and then subcultured 1:20 and allowed to grow for an additional 2 to 3 h. Bacteria were washed twice in phosphate-buffered saline and then resuspended in THP-1 growth media. THP-1 cells were washed twice in phosphate-buffered saline and then overlaid with bacterial suspension such that the multiplicity of infection was 10 bacteria per mammalian cell. After 30 min of invasion, the monolayer was washed three times in RPMI, and extracellular bacteria were killed by incubating the cells in growth medium containing 100 μg of gentamicin (Sigma)/ml. After 90 min, media was replaced with THP-1 growth medium containing 10 μg of gentamicin/ml. At times 2, 4, 8, and 12 h after bacterial addition, mammalian cells were lysed in 1% Triton X-100. Serial 10-fold dilutions of lysates were plated on LB plates or LB plates containing 30 μg of kanamycin/ml, and viable colonies were counted to determine intracellular bacterial numbers. For IL-1 measurements, 0.1 ml of culture supernatant was removed prior to cell lysis and used for enzyme-linked immunosorbent assay (ELISA).
Bacterial infection of mice. Female BALB/c mice between 6 and 8 weeks old were purchased from Charles River Laboratories and housed at The Burnham Institute according to local Animal Care Advisory Committee guidelines. The estimated lethal dose of serovar Enteritidis LK5 in BALB/c mice has previously been determined (8). Overnight bacterial cultures grown in LB medium were diluted in sterile 0.15 M NaCl to yield 106 CFU per ml. Mice were inoculated orally with 0.1 ml of the diluted bacteria by using a feeding needle. All inocula were quantified by plating serial 10-fold dilutions of the cultures onto LB and counting the viable colonies.
RESULTS
Identification of a S. enterica serovar Enteriditis protein containing a TIR-like domain. A cascade of PSI-BLAST searches (18) were used to search for TIR domain homologs in a set of sequences available from a variety of microbial DNA sequence databases, as described in Materials and Methods. Potential homologs identified in this search were further validated by using more sensitive profile-profile searches and structure modeling. This computational approach can detect homology between proteins that have little primary sequence similarity by using additional information from multiple alignments, structure, and secondary-structure features. From these searches, more than 200 bacterial TIR homologs were identified, expanding on the previously noted protein family (17). Proteins from this family display a very diverse domain structure and an unusual phylogenetic distribution. Proteins from several human and animal pathogens form a well-defined subfamily, with their TIR-like domains displaying many features of the eukaryotic TIRs (unpublished observations). Among the latter group was a hypothetical protein from the human pathogen S. enterica serovar Enteritidis. We have named the gene encoding this protein TlpA for TIR-like protein A.
TlpA is a 236-amino-acid protein, annotated in sequence databases as a hypothetical protein of unknown function. Sequence comparison of the hypothetical TIR domain of TlpA (amino acids 101 to 236) with the human TIR-containing proteins MyD88, TIRAP, and TLR4 showed significant sequence similarity (sequence identity of 28% and PSI-BLAST e.value of E-4 in second iteration) with conservation of the box 1 and box 2 sequences that are characteristic of all known members of the TLR/IL-1R superfamily (Fig. 1a). The domain architecture of TlpA indicated that it is comprised of a C-terminal TIR domain and an N-terminal 100-residue coiled-coil domain (Fig. 1b). Additional comparison between Salmonella genomes reveals that the tlpA gene is located in a DNA island that is unique to the S. enterica serovars Enteritidis and Dublin (S. Maloy, unpublished data). Sequence analysis of the regions flanking tlpA reveals the presence of genes, indicating phage origin of tlpA (integrases and phage signature proteins), as well as several proteins of unknown function and pilin-related domains (9). The latter are extracellular proteins involved in cell adhesion, suggesting that proteins in this genomic region are secreted. However, because the genome sequence of serovar Enteritidis is discontinuous in this region, we cannot determine whether tlpA resides within a cluster of genes involved in type III or type IV secretion.
Close homologs of serovar Enteritidis with the same domain organization and strong sequence similarity (>40% sequence identity) were found in an invasive, community-acquired methicillin-susceptible Staphylococcus aureus strain (MSSA476) and in Brucella melitensis, the agent of brucellosis, a deadly disease of humans and animals. In phylogenetic analysis, this group of bacterial TIR domains clusters together despite the distant relation between the organisms (gram positive, gram negative), supporting the hypothesis of their spread by lateral transfer.
Suppression of TIR-dependent NF-B activation by TlpA. Since TlpA protein was identified on the basis of its homology to the TIR domain of signaling proteins of the innate immune pathway, we explored the effects of TlpA on TIR-dependent activation of NF-B in mammalian cells. Accordingly, HEK293 cells were cotransfected with increasing amounts of plasmid encoding TlpA and a fixed amount of a luciferase gene reporter plasmid containing five NF-B-binding sites for measuring transcriptional activity. Despite confirmation of TlpA protein production in HEK293 cells by immunoblotting, expression of TlpA itself had no effect on basal levels of NF-B activity (data not shown). We therefore assessed the ability of TlpA to suppress NF-B activation induced by a variety of proteins that contain TIR domains, including a constitutively active TLR4 receptor (CD4:TLR4) and the MyD88 adaptor protein. Cells were also stimulated by incubation with the proinflammatory cytokine, IL-1, whose receptor contains TIR domains, and TNF-, which signals via TIR-independent mechanisms. Using the same luciferase-based reporter gene assay we found that TlpA suppressed NF-B activation induced by components of the TLR signaling pathway TLR4, MyD88, and IL-1 in a dose-dependent manner (Fig. 2a). However, TlpA did not suppress activation induced by TNF-, suggesting specificity for the TLR/IL-1R pathway. In contrast to its effects on NF-B, TlpA expression did not interfere with the activity of other transcription factors (as assessed by reporter gene assays), including AP-1 and STAT-1 (not shown). Furthermore, in experiments where TlpA was cotransfected with plasmids encoding CD4:TLR4 or MyD88, we show the effects on NF-B activation were not due to alterations in protein expression (Fig. 2b).
To determine whether the observed reduction of NF-B activity correlated with reduced NF-B DNA-binding activity in TlpA-expressing cells, we performed EMSAs with 32P-labeled, double-stranded DNA oligonucleotides containing a consensus NF-B-binding site. Nuclear extracts were prepared from unstimulated cells and cells that had been stimulated by culturing with TIR-dependent and TIR-independent cytokines, IL-1 and TNF-, respectively. Incubation of the nuclear extracts with 32P-labeled NF-B probe, followed by gel electrophoresis, then allowed visualization of the relative levels of the NF-B DNA-binding activity. Extracts from cells exposed to IL-1 or TNF- contained abundant NF-B DNA-binding activity when derived from cells transfected with control plasmid (Fig. 3a). In contrast, NF-B-binding activity was markedly reduced in extracts derived from cells expressing TlpA when IL-1 was applied as a stimulus but not when TNF- was used, a finding consistent with luciferase reporter gene results. Competition experiments with unlabeled wild-type and mutant NF-B oligonucleotide probes confirmed the specificity of the EMSA results (Fig. 3a). Moreover, TlpA had no effect on the DNA-binding activity of an unrelated transcription factor STAT-1, further confirming specificity.
Immunoblot analysis of the nuclear levels of NF-B subunits p65 and p50 were compared to control and cytokine-treated cells (Fig. 3b). Nuclear levels of p65 and p50 in cells treated with TNF- were comparable in TlpA-expressing and control-transfected cells, confirming that TNF--mediated NF-B activation is unaffected by TlpA expression. In contrast, levels of p65 and p50 were reduced in the nuclei of IL-1-treated cells expressing TlpA compared to control-transfected cells. Levels of another nuclear transcription factor TFII-I were not affected by TlpA overexpression, confirming the specificity of the results. These data suggest that expression of TlpA suppresses nuclear translocation of the NF-B complex.
TlpA contributes to virulence of serovar enteritidis in both a cell culture and a mouse infection model. To determine whether TlpA plays a role during infection, the tlpA gene of serovar Enteritidis LK5 was disrupted by replacement with a Kanr cassette using the "Red Swap" method of Datsenko and Wanner (6). Loss of the allele was confirmed by PCR with locus-specific primers.
The resultant strain (tlpA::kan) was used to infect human monocyte-derived macrophage cell line THP-1. Bacteria were cocultured with macrophages in the absence of antibiotic and allowed to invade for 30 min prior to antibiotic treatment to kill extracellular bacteria. At 2, 4, 8, and 12 h after bacterial addition, macrophages were lysed, and intracellular CFU were enumerated by dilution plating of the lysates. Infection results indicate that although the tlpA::kan strain has no defect in invasion, it is unable to survive and/or replicate in macrophages compared to wild-type, isogenic serovar Enteritidis strain LK5 (see Fig. 5a). This effect is specific for TlpA, as ectopic expression of TlpA on a low-copy plasmid restored intracellular survival to the tlpA::kan strain (Fig. 4a).
To further test the role of TlpA in disease, mice were orally infected with a lethal dose of wild-type serovar Enteritidis LK5 and the tlpA::kan mutant strain. Survival of infected mice was then compared. Disruption of the tlpA gene lead to prolonged survival of infected mice compared to animals infected with the isogenic wild-type serovar Enteritidis strain LK5. Typical results from an infection experiment are shown in Fig. 5b. The average survival times derived from three such experiments were 4.3 ± 0.3 days for mice infected with wild-type bacteria versus 7.8 ± 0.5 days for mice infected with the tlpA mutant. Analysis of animals 5 days postinfection with either LK5 or tlpA::kan showed that there was a significant decrease in the number of bacteria recovered per spleen from mice infected with the mutant strain (Fig. 4c).
TlpA induces caspase activation and IL-1 secretion. Several studies have shown the importance of caspase-1 activation for in vivo dissemination of Salmonella (14, 20-22). Caspase-1, also known as IL-1 converting enzyme (ICE), is a cysteine-specific protease that cleaves and activates proinflammatory cytokines pro-IL-1 and pro-IL-18 (31). The Salmonella SipB protein and the closely related Shigella IpaB protein have been shown to bind and activate caspase-1 in host cells (11, 30), providing examples of virulence factors that regulate this protease involved in cytokine elaboration. Activation of TLRs also causes activation of caspase-1, although the mechanism is unknown (31). We therefore explored the effects of TlpA expression on caspase-1 activity and IL-1 secretion by using a cell transfection method previously used for studies of SipB (11).
Accordingly, HEK293 cells expressing recombinant murine caspase-1 and murine pro-IL-1 were transfected with a plasmid encoding myc-tagged TlpA. Quantitative ELISA was used to measure the relative amount of mature IL-1 secreted into the cell culture media 24 h after transfection. Expression of TlpA together with procaspase-1 increased IL-1 secretion 8-fold over vector control, representing a level comparable to that generated by serovar Typhimurium SipB (Fig. 5a). In contrast, transfecting cells individually with plasmids encoding either TlpA or procaspase-1 did not result in significant IL-1 secretion (see supplemental data at http://www.burnham.org/papers/IAI1013-05/SupplementalFigures.ppt), demonstrating that TlpA only induces IL-1 secretion when procaspase-1 is present. The ability of TlpA to induce IL-1 secretion was not a promiscuous feature of TIR domains because expression of the TIR domain of MyD88 did not induce IL-1 secretion. TlpA-induced secretion of mature IL-1 was shown to be caspase dependent, since both the caspase-1/8-specific protein inhibitor CrmA and a chemical pan-caspase inhibitor zVAD-fmk suppressed TlpA-induced IL-1 secretion (Fig. 5a). Immunoblot analysis confirmed that caspase inhibitors did not interfere with expression of TlpA (not shown), thus excluding a trivial explanation for the results. Taken together, these data indicate that TlpA induces caspase-dependent IL-1 secretion.
To extend these gene transfer results to a more physiological context, we compared IL-1 secretion induced by infection of THP-1 macrophages with wild-type versus mutant serovar Enteritidis strains. Accordingly, the relative amounts of IL-1 in culture supernatants were measured by quantitative ELISA at various times after bacterial addition. Infection with the wild-type serovar Enteritidis strain resulted in steadily increasing levels of secreted IL-1 starting approximately 4 h after bacterial addition and continuing through the 12 h period assessed (Fig. 5b). After 12 h, infected THP-1 macrophages began to die, making later measurements of IL-1 production problematic (not shown). Disruption of the tlpA gene rendered the mutant strain (tlpA::kan) unable to sustain the high levels of IL-1 secretion after infection of macrophages. The diminution of IL-1 levels at 8 h after bacterial addition coincides with a decrease in the numbers of intracellular bacteria (as seen above), suggesting the two events may be connected. Introduction of the tlpA gene into the mutant strain on a low-copy plasmid restored IL-1 secretion to the levels observed with infections using the wild-type strain (Fig. 5b). Taken together, these cell infection data confirm a role of TlpA in regulating secretion of the proinflammatory cytokine IL-1, a finding consistent with the gene transfection data above showing that TlpA can induce caspase-1-dependent IL-1 secretion.
DISCUSSION
To identify candidate novel, virulence genes, we examined bacterial genomes for hypothetical proteins with homology to mammalian TIR domains. We postulated that bacterial proteins mimicking these eukaryotic signaling molecules would block inflammatory responses and thus blunt host defense. The serovar Enteritidis TlpA protein was thus identified. Genetic and biochemical characterization of TlpA suggests that it is important for bacterial virulence in vivo and that it modulates host defense mechanisms involved in regulation of NF-B and caspase activation. The observation that TlpA can suppress NF-B while inducing caspase-1 activation suggests that this protein may act in different ways during the course of infection to either promote or suppress cytokine induction, perhaps as a mechanism to disseminate after successful invasion.
TlpA is a novel suppressor of NF-B activation. Given the homology of TlpA to mammalian TIR-containing proteins, we considered that it might suppress NF-B by binding to members of the TLR/IL-1R superfamily and preventing downstream signaling. Consistent with this possibility, TlpA suppressed NF-B activation induced by TIR-dependent stimuli such as TLR4, MyD88, and IL-1, but not by the TIR-independent cytokine TNF-. However, we have been unable to detect direct interaction between TlpA and a variety of mammalian TIR-family proteins.
We found that TlpA is required for induction of IL-1 secretion in the context of infection of cultured macrophages with serovar Enteritidis. Bacteria lacking TlpA induced far less IL-1 production compared to the wild-type strain, even though such gram-negative organisms produce lipopolysaccharide and should trigger caspase-1 activation pathways via TLRs. Moreover, ectopic expression of TlpA was sufficient to induce caspase-dependent elaboration of IL-1 from mammalian cells, indicating a direct effect. Taken together, these experiments indicated that TlpA is both necessary and sufficient for inducing IL-1 production. How TlpA induces caspase activation leading to IL-1 secretion is unclear. Some bacterial virulence proteins, such as SipB from serovar Typhimurium and IpaB from Shigella species, appear to directly bind pro-caspase-1 and induce its activation, but TlpA shares neither sequence similarity nor predicted structural similarity to these proteins. Cellular TIR domain-containing proteins are known to transduce signals that trigger caspase-1 activation; thus, the predicted structural similarity of TlpA to animal TIRs might allow it to modulate TIR-dependent signaling events linked to caspase-1 activation.
It is not entirely clear why bacteria would find it advantageous to induce IL-1 production. In the case of serovar Typhimurium, it has been suggested that IL-1 may recruit additional host inflammatory cells that can serve as reservoirs for bacterial replication or alter vascular blood flow in ways that promote deep tissue invasion (14, 21, 22). Alternatively, IL-1 production may represent a by-product of caspase activation, where bacterial virulence factors use caspase-1 or other members of the caspase family to kill host defense cells.
Regardless of the mechanisms by which TlpA activates caspases and suppresses NF-B, our comparisons of genetically engineered bacteria indicate that TlpA represents a virulence gene. Without TlpA, mice survive serovar Enteritidis infection longer. Curiously, TlpA deficiency reduces accumulation of serovar Enteritidis inside infected macrophages, implying that TlpA is required for intracellular growth or survival of bacteria. We speculate therefore that TlpA is required for thwarting activation of endogenous host cell pathways that suppress accumulation of intracellular bacteria. Given that the genomes of many other pathogenic and nonpathogenic microbial species contain open reading frames potentially encoding TIR-like proteins (unpublished observations), it will be interesting to compare the functions of these gene products to ascertain whether they exhibit conserved functions.
It is interesting that proteins containing predicted TIR-like domains are widespread in bacteria (unpublished observations). These proteins show an unusual distribution in bacterial species, with very close homologs found in unrelated bacteria and, conversely, TlpA proteins from strains of the same bacteria being distantly related. In addition, all genes predicted to encode proteins with TIR-like domains are found in genomic regions with phage origins. These observations suggest that the family of bacterial TIR-like genes evolved by lateral transfer. It is unknown whether the proteins of this family fulfill intrinsic bacterial functions, or whether they represent a widespread, proto-pathogenic adaptation of bacteria. At least for the TIR-like protein of S. enterica serovar Enteritidis, it appears that members of this family represent a new class of virulence factors adapted to alter intracellular signaling pathways of eukaryotic host cells and thereby afford bacteria with survival and dissemination advantages in vivo.
ACKNOWLEDGMENTS
We thank Loredana Fiorentino, Marcin Feder, Juan Zapata, Jason Damiano, Rhonda Croxton, April Stanley, and Stanley Maloy for helpful discussion. We also thank Christina Kress and Adriana Charbono for expert assistance with mouse infections.
This study was supported by a grant from the National Institutes of Health (AI-055789) and a Kirchstein NRSA/NIH fellowship (5T32AG00252).
Present address: Harvard Medical School/NEPRC, 1 Pine Hill Dr., Southborough, MA 01772.
REFERENCES
1. Akira, S., and K. Takeda. 2004. Toll-like receptor signalling. Nat. Rev. Immunol. 4:499-511.
2. Athman, R., and D. Philpott. 2004. Innate immunity via Toll-like receptors and Nod proteins. Curr. Opin. Microbiol. 7:25-32.
3. Beutler, B. 2004. Inferences, questions and possibilities in Toll-like receptor signalling. Nature 430:257-263.
4. Beutler, B., K. Hoebe, X. Du, and R. J. Ulevitch. 2003. How we detect microbes and respond to them: the Toll-like receptors and their transducers. J. Leukoc. Biol. 74:479-485.
5. Collier-Hyams, L. S., H. Zeng, J. Sun, A. D. Tomlinson, Z. Q. Bao, H. Chen, J. L. Madara, K. Orth, and A. S. Neish. 2002. Cutting edge: Salmonella AvrA effector inhibits the key proinflammatory, antiapoptotic NF-B pathway. J. Immunol. 169:2846-2850.
6. Datsenko, K. A., and B. L. Wanner. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97:6640-6645.
7. Dunne, A., and L. A. O'Neill. 2003. The interleukin-1 receptor/Toll-like receptor superfamily: signal transduction during inflammation and host defense. Sci. STKE 2003:re3.
8. Edwards, R. A., D. M. Schifferli, and S. R. Maloy. 2000. A role for Salmonella fimbriae in intraperitoneal infections. Proc. Natl. Acad. Sci. USA 97:1258-1262.
9. Fellowship for Interpretation of Genomes. 2005. The SEED: an annotation/analysis tool. Fellowship for Interpretation of Genomes, University of Chicago, Chicago, Ill. [Online.]. http://theseed.uchicago.edu/FIG/index.cgi.
10. Fiorentino, L., C. Stehlik, V. Oliveira, M. E. Ariza, A. Godzik, and J. C. Reed. 2002. A novel PAAD-containing protein that modulates NF-B induction by cytokines tumor necrosis factor- and interleukin-1. J. Biol. Chem. 277:35333-35340.
11. Hersh, D., D. M. Monack, M. R. Smith, N. Ghori, S. Falkow, and A. Zychlinsky. 1999. The Salmonella invasin SipB induces macrophage apoptosis by binding to caspase-1. Proc. Natl. Acad. Sci. USA 96:2396-2401.
12. Hornef, M. W., M. J. Wick, M. Rhen, and S. Normark. 2002. Bacterial strategies for overcoming host innate and adaptive immune responses. Nat. Immunol. 3:1033-1040.
13. Jaroszewski, L., L. Rychlewski, Z. Li, W. Li, and A. Godzik. 2005. FFAS03: a server for profile-profile sequence alignments. Nucleic Acids Res. 33:W284-W288.
14. Jarvelainen, H. A., A. Galmiche, and A. Zychlinsky. 2003. Caspase-1 activation by Salmonella. Trends Cell Biol. 13:204-209.
15. Karin, M., and A. Lin. 2002. NF-B at the crossroads of life and death. Nat. Immunol. 3:221-227.
16. Knodler, L. A., J. Celli, and B. B. Finlay. 2001. Pathogenic trickery: deception of host cell processes. Nat. Rev. Mol. Cell. Biol. 2:578-588.
17. Koonin, E. V., and L. Aravind. 2002. Origin and evolution of eukaryotic apoptosis: the bacterial connection. Cell Death Differ. 9:394-404.
18. Li, W., F. Pio, K. Pawlowski, and A. Godzik. 2000. Saturated BLAST: an automated multiple intermediate sequence search used to detect distant homology. Bioinformatics 16:1105-1110.
19. Lupas, A., M. Van Dyke, and J. Stock. 1991. Predicting coiled coils from protein sequences. Science 252:1162-1164.
20. Monack, D. M., C. S. Detweiler, and S. Falkow. 2001. Salmonella pathogenicity island 2-dependent macrophage death is mediated in part by the host cysteine protease caspase-1. Cell Microbiol. 3:825-837.
21. Monack, D. M., D. Hersh, N. Ghori, D. Bouley, A. Zychlinsky, and S. Falkow. 2000. Salmonella exploits caspase-1 to colonize Peyer's patches in a murine typhoid model. J. Exp. Med. 192:249-258.
22. Monack, D. M., W. W. Navarre, and S. Falkow. 2001. Salmonella-induced macrophage death: the role of caspase-1 in death and inflammation. Microbes Infect. 3:1201-1212.
23. Orth, K. 2002. Function of the Yersinia effector YopJ. Curr. Opin. Microbiol. 5:38-43.
24. Orth, K., Z. Xu, M. B. Mudgett, Z. Q. Bao, L. E. Palmer, J. B. Bliska, W. F. Mangel, B. Staskawicz, and J. E. Dixon. 2000. Disruption of signaling by Yersinia effector YopJ, a ubiquitin-like protein protease. Science 290:1594-1597.
25. Rosenberger, C. M., and B. B. Finlay. 2003. Phagocyte sabotage: disruption of macrophage signalling by bacterial pathogens. Nat. Rev. Mol. Cell. Biol. 4:385-396.
26. Rychlewski, L., L. Jaroszewski, W. Li, and A. Godzik. 2000. Comparison of sequence profiles. Strategies for structural predictions using sequence information. Protein Sci. 9:232-241.
27. Silverman, N., and T. Maniatis. 2001. NF-B signaling pathways in mammalian and insect innate immunity. Genes Dev. 15:2321-2342.
28. Stebbins, C. E., and J. E. Galan. 2001. Structural mimicry in bacterial virulence. Nature 412:701-705.
29. Tato, C. M., and C. A. Hunter. 2002. Host-pathogen interactions: subversion and utilization of the NF-B pathway during infection. Infect. Immun. 70:3311-3317.
30. Thirumalai, K., K. S. Kim, and A. Zychlinsky. 1997. IpaB, a Shigella flexneri invasin, colocalizes with interleukin-1-converting enzyme in the cytoplasm of macrophages. Infect. Immun. 65:787-793.
31. Tschopp, J., F. Martinon, and K. Burns. 2003. NALPs: a novel protein family involved in inflammation. Nat. Rev. Mol. Cell. Biol. 4:95-104.
32. Xu, Y., X. Tao, B. Shen, T. Horng, R. Medzhitov, J. L. Manley, and L. Tong. 2000. Structural basis for signal transduction by the Toll/interleukin-1 receptor domains. Nature 408:111-115.(Ruchi M. Newman, Prabhaka)