Subversion of Innate Immune Responses by Brucella through the Targeted Degradation of the TLR Signaling Adapter, MAL

Detection of microbes by TLRs is a critical step in activation of the innate immune response and is essential for robust priming of the adaptive immune response (1, 2). TLRs localize to the plasma membrane or to endocytic membranes and recognize certain molecular elements of pathogens called pathogen associated molecular patterns (PAMPs). On ligation with PAMPs, TLRs undergo conformational changes that allow them to engage intracellular adapter molecules, thus initiating a cascade of signaling events that culminates in activation of key transcription factors, such as NF-κB and AP-1. These transcription factors in turn trigger the production of cytokines, chemokines, and antimicrobial peptides that eventually help contain and clear the infection (3). So far, 13 TLRs have been identified in mammals and they all share a leucine-rich extracellular region (LRR) and an intracellular signaling domain known as the Toll/IL-1 receptor (TIR) domain. Although the LRR domain is responsible for binding to PAMPs, it is the TIR domain that binds to TIR-domain containing adapter molecules to initiate signaling (2, 4) Five TIR-domain containing adapter molecules have been described: MyD88 (5), MyD88-adapter–like (MAL)(6) [also known as TIRAP (7)], TIR-domain containing adapter protein-inducing IFNβ, TRIF (also known as TICAM1) (8, 9), TRIF-related adaptor molecule (TRAM; also known as TICAM2) (10–12), and sterile armadillo-motif–containing protein (SARM) (13, 14). Although MyD88 and TRIF can bind to TLRs independent of other adapters, MAL and TRAM can bind only in conjunction with MyD88 and TRIF, respectively. Therefore, MAL and TRAM act as bridging adapters that connect MyD88 and TRIF to their respective TLRs (12, 15). SARM functions only by interaction through TRIF and acts a negative regulator (16). These adapters have unique signaling properties, and their differential use by TLRs results in specialized signaling outcomes that permit a tailored, robust response against the pathogen (2, 17)

The genus Brucella is comprised of Gram-negative bacteria that survive and replicate predominantly in macrophages in a broad range of mammalian hosts from cows to humans, causing brucellosis, a zoonotic disease in the latter host (18). The hallmarks of animal brucellosis, in both domesticated and wild animals, are abortion, infertility, and reproductive failures. Infection in humans causes chronic debilitating fever (also known as Malta fever), chills, malaise, arthritis, dementia, and in serious cases, endocarditis and neurologic disorders (19, 20). Human brucellosis primarily occurs either through contact with infected animals or through consumption of dairy products from infected animals, and is caused primarily by B. melitensis, B. abortus, and B. suis strains (21). Brucellosis is the leading zoonosis on a worldwide scale and constitutes a major public health threat in regions of the world where Brucella infections are uncontrolled in food animals (20). Brucella sp. have a low infectious dose and are easily aerosolized, and human infections with these bacteria are debilitating and difficult to treat (22, 23). Consequently, these bacteria are presently listed as class B agents on the National Institute of Allergies and Infectious Diseases list of etiologic agents of concern with respect to their potential use in bioterrorism (24, 25).

There is a considerable amount of evidence that indicates that the capacity of Brucella to avoid or interfere with components of the host innate, and acquired immune responses plays a critical role in their virulence. The lipid A moiety of the LPS of these bacteria, for instance, elicits a reduced and delayed inflammatory response in infected hosts compared with the endotoxin of some other Gram-negative bacteria and this property has been proposed to allow the brucellae to use a mechanism for a “stealthy” mode of entry into host macrophages (26). The Brucella LPS O-side chain also forms complexes with MHC class II and interferes with the capacity of Brucella-infected macrophages to serve as APCs (27). Once the brucellae have been engulfed by host macrophages, they use a type IV secretion system to redirect the membrane-bound compartment within which they reside from the endolysosomal pathway into a pathway where this compartment maintains extensive interaction with the endoplasmic reticulum (28–30). Two effectors that appear to be secreted into the host cell cytoplasm by the Brucella T4SS have recently been described but the biological function of these effectors is presently unknown (31, 32). The brucellae also produce a periplasmic cyclic β-1,2-glucan that assists these bacteria in modifying their phagosome in such a fashion that it avoids fusion with lysosomes in host cells (33).

The interactions of Brucella strains with TLR4 and TLR2 on host cells has been shown to influence the induction of innate immune responses during infection (34). Several recent reports have described the activity of a Brucella protein designated TcpB (also known as Btp1) (35, 36) that shares significant amino acid with mammalian TIR domains and interferes with TLR2 and TLR4 signaling pathways when expressed in mammalian cells. TcpB has also been shown to resemble the TIR domain adapter protein MAL, by being able to bind phosphoinositides (37). Brucella tcpB mutants do not exhibit attenuation in cultured murine macrophages or immunocompetent mouse models of chronic infection. These mutants do, however, display delayed virulence in the IFN regulatory factor (IRF-1^−/−) mouse model of infection, and unlike wild-type (wt) strains they do not inhibit the maturation of murine dendritic cells suggesting that TcpB plays a role in the early interactions of Brucella strains with their host cells. In this study, we present evidence that the Brucella TcpB targets the TLR adapter protein MAL for degradation, thereby suppressing TLR4 signaling and providing a molecular mechanism for subversion of innate immune responses by these bacteria.

Rabbit polyclonal anti-TcpB antisera (raised by injecting rabbits with GST-TcpB protein) were affinity purified against GST-TcpB. Anti-HA mouse ascites fluid, anti-HA rabbit polyclonal Ab, anti-M2 mAb, and anti–β-tubulin mAb was from Sigma-Aldrich (St. Louis, MO). Rabbit monoclonal anti-MAL Ab for detecting endogenous MAL was from Epitomics (Burlingame, CA). Rabbit polyclonal anti-VSV Ab was from Zymed (San Francisco, CA). Anti-MyD88 Ab for detecting the endogenous protein was from Santa Cruz Biotechnology (Santa Cruz, CA). Protein-G Sepharose (PGS) and GST Sepharose beads were from Pharmacia (Peapack, NJ). HRP-labeled secondary Abs were purchased from The Jackson Laboratory (Bar Harbor, ME). Dual luciferase assay kit and TNT in vitro transcription and translation kits were from Promega (Madison, WI). Restriction enzymes and other DNA-modifying enzymes were from New England Biolabs (Ipswich, MA).

TcpB was amplified by PCR from genomic DNA of B. melitensis 16M and cloned into pcDNA3.1 V5B as described later in this paper. TcpBm was created by deleting 9 bp from box 1 of wt TcpB using the QuikChange II Site-Directed Mutagenesis Kit from Stratagene (La Jolla, CA) and wt TcpB as template.

CD4/TLR4 construct has been described previously. HA-tagged MAL and MAL-DN clones were kindly provided by L. O’Neill (Trinity College, Dublin, Ireland).

Multiple alignment of TIR domain was domain was performed by Align X software (Invitrogen, San Diego, CA). Tertiary structure of TIR domain of TcpB, MAL, and TLR4 were predicted by 3D-JIGSAW (version 2.0) server with TLR 2-TIR domain structure (PDB ID: 1FYX) as a model. The predicted structures were further refined for energy minimization by using deepview/Swiss-pdbviewer version 3.7.

DNA was transfected into HEK 293 cells using polyethylenimine (PEI), lipofectamine, or Fugene as indicated in the figure legends. For PEI-mediated transfection, 24 h after seeding HEK 293 cells, DNA was mixed with PEI (2 μg PEI per μg DNA used) and diluted with Optimem. After incubation for 20 min at room temperature, the DNA-PEI complexes were added to the cells Where lipofectamine or Fugene were used, transfections were performed according to manufacturer’s protocol. Cells were harvested 24–48 h later in lysis buffer containing 20 mM Tris, pH 8.0, 150 mM NaCl, and 1% Tirton-X100 (TNT) plus protease inhibitors. Equal amounts of each sample was fractioned by 12% SDS-PAGE and transferred to Immobilon. After blocking in Tween 20 and Tris-buffered saline (TTBS) with 5% nonfat milk for 30 min at room temperature, the blots were incubated with primary Ab (in TTBS + 5% nonfat milk) overnight at 4°C (unless noted otherwise), washed three times with TTBS, and incubated with appropriate HRP-conjugated secondary Ab (1:10,000 dilution) for 30 min at room temperature. After three washes with TTBS, bands were detected by ECL according to manufacturer’s recommendation.

HEK 293 cells were seeded (0.8 × 10 ⁶ cells/well) into 6-well plates 24 h prior to transfection. DNA was transfected using either lipofectamine or PEI as described previously. Cells were harvested after ∼30 h in 250 μl TNT buffer + protease inhibitors. After 15 min incubation on ice, samples were spun at 14,000 rpm for 15 min at 4°C to remove cellular debris. Supernatants (300 μg) were then precleared with PGS beads with rotation for 30 min at 4°C. Precleared lysates (50 μl) were saved for immunoblotting later. The remainder of the precleared lysates were then incubated overnight with appropriate Ab and PGS at 4°C. Immune complexes were collected by spinning at 5000 × g for 2 min at 4°C. Washed beads three times with TNT, added 30 μl 6× SDS-loading dye, boiled for 1 min, and fractionated on SDS-PAGE. Samples were immunoblotted with the indicated Abs as described previously. For analysis with the same Ab as used for immunoprecipitation, loaded only 5 μl of each sample. Used the reminder for analysis with other Abs. Preimmune lysates were analyzed in identical manner.

Indicated preimmune lysates were treated with 50 U calf intestinal phosphatase (CIP) in presence or absence of 4 mM sodium orthovanadate (Na₂V₂O₅) as inhibitor of CIP and incubated for 3 h at 37°C. Samples were analyzed thereafter as described previously.

HEK 293 cells (0.8 × 10⁵/well) were seeded onto 24-well plates 24 h before transfection. Increasing amounts of TcpB (50, 100, and 200 ng/well) or TcpB mutant (TcpBm), one stimulant plasmid DNA (CD4/TLR4 [250 ng/well], MAL[25 ng/well], MyD88 [5 ng/well], TRIF [5 ng/well], Irak-1, TRAF 6), NF-κB–luciferase reporter construct (100 ng/well) and Renilla luciferase reporter (50 ng/well) constructs were cotransfected using PEI or Fugene according to the manufacturer’s recommendation. Total amount of DNA transfected was maintained constant (800 ng) by addition of various amounts of the empty vector. Cells were washed once with PBS 24 h later and then harvested in 100 μl of passive lysis buffer from the dual luciferase assay kit. After incubation on ice for 15 min, 5 μl of the lysate was assayed using the dual luciferase assay kit according to the suggested protocol. The level of Renilla luciferase activity was used to normalize NF-κB–luciferase activity to serve as control for transfection. Results are expressed as mean fold stimulation over unstimulated controls. Each assay was repeated at least three times, each time in triplicate (CD4/TLR4) or duplicate (all others).

HEK293 cells (0.8 × 10⁶ cells/well in 6-well plates) were transfected with plasmids encoding HA MAL (300 ng/well) and VSV-Ubiqitine (1.0 μg/well) in the presence (1.2 μg/well) or absence of TcpB as described previously. Half of the samples were treated 24 h later with MG132 (20 mM) for 4 h. Cells were washed and harvested in PBS and spun at 2500 rpm for 3 min. Pelleted cells were resuspended in 300 μl SDS lysis buffer containing 1% SDS, 20 mM Tris-HCl, pH 8.0, and boiled for 10 min. They were then sonicated for 5-s pulses and 1-s pauses for a total of 30 s, followed by centrifugation at 16,000 × g for 15 min. Supernatant was transferred to a new tube. Five hundred microliters 2 × IP buffer (2% Triton X-100, 150 mM NaCl, 20 mM Tris-HCl, 8.0, 1 mM EDTA, and 0.5% NP-40) was added to 300 mg total protein (volume made up to 500 μl with ddH₂O). Preclearing and immunoprecipitation of the lysates thereafter were carried out as described previously.

TcpB open-reading frame was amplified from B. melitensis 16M genomic DNA using the following primers: Fwd: 5′ CGGGGTACCATGTCTAAAGAGAAACAAAGCC-3′ and Rev: 5′ GAAAGAACTGCATTCCCTTATCAAGAATTCCG-3′.The ∼750 nt, amplified product was gel purified and ligated to pCDNA 3.1-V5-His B vector and transformed into Escherichia coli DH5α cells. Putative colonies containing the correct plasmids were identified by analysis of the minipreps of the resulting bacterial colonies.

The tcpB open-reading frame was PCR amplified from B. abortus 2308 chromosome, using the following primers: Fwd: 5′-CTTGAGAGGCACCGCTCAAT-3′ and Rev: 5′-CAGATTTGCGCAGATTGCAT-3′.

This resulted in a product of 1961-bp that was ligated into pGEM-T Easy (Promega) according to manufacturer protocols. The ligation mixture was transformed into E. coli DH5a with blue/white selection, putative colonies containing the desired plasmid were grown and stocked in 25% glycerol/Brucella broth. After confirmation of the transformation product, the plasmid pTBtcpB was cut with the blunt cutters EcoRV and NruI, which removed an internal fragment of 577 bp from the tcpB coding region. The kanamycin resistance cassette (aph3a) from pKS-Kn was inserted into the resulting blunt site and this plasmid was given the designation pTBtcpBD. This plasmid was dialyzed and electroporated into B. abortus 2308 using previously described methods (38). The resulting transformants were cultured on Schaedler agar supplemented with 5% defibrinated bovine blood (SBA) and 45 μg/ml kanamycin and subsequently patched onto Schaedler agar supplemented with 5% defibrinated bovine blood (SBA)SBA with containing kanamycin and 25 μg/ml ampicillin. Colonies that were resistant to kanamycin, but sensitive to ampicillin were selected for further characterization, grown overnight in Brucella broth and stored at −80°C in Brucella broth supplemented with 25% glycerol. Chromosomal DNA was isolated from all of these putative mutants and PCR analysis of the mutated tcpB region was performed using primers specific for the kanamycin resistance gene (aph3a), the ampicillin resistance gene (bla) from the pGEM plasmid, the entire tcpB gene, and a 400-bp region representing the sequences deleted from the tcpB coding regions. This analysis yielded the expected products with the aph3a- and complete tcpB-specific primers, but no products with the bla- or 400-bp deleted region-specific primers. One mutant was given the designation KB2 and selected for further analysis. The crystal violet exclusion assay (39) was performed on KB2 to ensure that this strain had retained is smooth lipopolysaccharide phenotype. Brucella strains that spontaneously lose their LPS O-chain (this changes them from the “smooth” to the “rough” LPS phenotype) are highly attenuated in animal models.

Infection of macrophages with Brucella was performed in a biosafety level 3 laboratory according to the protocol described by Gee et al. (40). Briefly, J774 cells grown in T75 flasks were cultivated in DMEM supplemented with 5% FCS at 37°C in 5% CO₂. After 24 h of growth, the confluent monolayers were infected with 2308 or the tcpB mutant. Infectious strains were opsonized for 30 min with a subagglutinating dilution (1:1,500) of hyperimmune C57BL6J mouse serum at a multiplicity of infection (MOI) of 50 Brucellae per macrophage (MOI of 50:1) and incubated at 37°C with 5% CO₂ for 1h to allow for phagocytosis of the Brucellae. Cell culture medium was removed and replaced with fresh medium supplemented with 5% FCS and 50 μmg/ml gentamicin to kill any remaining extracellular bacteria and incubated at 37°C with 5% CO₂ for 2 h. The macrophages were then washed with PBS supplemented with 0.5% FCS and maintained thereafter in DMEM with 5% FCS and 12.5 μg/ml gentamicin. Cell culture medium was replaced with fresh medium every 24 h. At 48 h postinfection, macrophages were washed with PBS supplemented with 0.5% FCS and lysed with protein sample buffer supplemented with 0.1% deoxycholic acid. After 5 min incubation at room temperature, macrophage lysates were centrifuged for 30 min at 10,000 rpm, supernatant collected, filtered, and bacterial and cellular debris discarded. J774 lysates were then boiled for 15 min. J774 lysate protein concentrations were then determined by the Bradford method. Where necessary, the lysates were further concentrated by acetone precipitation as follows. Four volumes of ice cold acetone added to one volume of lysate, and stored overnight at −20^oC. Proteins were recovered by centrifugation at 14,000 rpm for 15 s and samples were resuspended in protein lysis buffer (0.3% [w/v] SDS, 200 mM DTT, 22 mM Tris-base, 28 mM Tris-HCl, pH 8.0), Samples were then analyzed by Western blot analysis with anti-MAL Ab (1:1000 dilution) as described in 1Materials and Methods.

Brucella abortus cultures were inoculated at 1 × 10³ CFU/ml in Brucella broth and incubated at 37°C under 5% CO₂ with shaking. Bacteria were pelleted at 48, 72, and 96 h, resuspended in 1 ml 1× protein lysis buffer (0.3% [w/v] SDS, 200 mM DTT, 22 mM Tris-base, 28 mM Tris-HCl, pH 8.0), boiled for 30 min, and lysed by bead beating six times for 40 s. Samples were then centrifuged for 20 min at 14,000 rpm and supernatant collected. Total cell protein concentrations were determined by the Bradford method and equal amounts of protein were separated onto a SDS-PAGE gel. After transfer to nitrocellulose membrane, TcpB protein was detected using anti-TcpB Ab (1:1000) and goat anti-rabbit Ig G conjugated to HRP (1:10,000). Bands were visualized using SuperSignal West Pico Chemiluminescent Substrate (Pierce Protein Research Products, Rockford, IL).

Spleen colonization data were obtained as described previously (40). Briefly, 6–8-wk-old female BL6 mice were infected via the i.p. route with approximately 5 × 10 ⁴ brucellae. At 1, 2, 3, and 4 wk postinfection, five mice per experimental group were sacrificed by isoflurane overdose. Immediately after euthanasia, spleens were harvested aseptically and homogenized in sterile PBS. Spleen homogenates were then serially diluted 10-fold in sterile PBS and plated onto SBA to determine the number of brucellae present at each time point. Spleen homogenates were also plated in parallel on SBA supplemented with the appropriate antibiotics to confirm the identity of the isolates and monitor the stability of the plasmid in MEK2Cm. Mean averages of counts from each test group were determined, and the data were expressed as log₁₀ brucellae/spleen.

Using the sequence of the TIR domain of hTLR4 to search for related coding sequences in the B. melitensis 16M genome sequence, we identified a coding sequence homologous to mammalian TIR-domain sequences and originally gave it the designation TIRB (TIR-domain-like protein of Brucella). This protein, encoded by BMEI1674 in the B. melitensis 16M genome and BAB1_0279 in the B. abortus 2308 sequence, has subsequently been given the designations TcpB and BtpA in the literature (35, 36, 41). For clarity, we will use the designation TcpB. As shown in Fig. 1A,, box 1, alignment of the TIR domain of TcpB with other TIR domains from various TLRs and adapters such as MyD88 and MAL indicates that TcpB is most similar to the box 1 region. However, unlike the TIR domains of these other proteins, TcpB lacks a well-defined box 3 region, missing even the most highly conserved amino acid residues of the canonical TIR sequence (Fig. 1A). The box 2 region is also poorly conserved, and lacks some characteristic residues such as the conserved proline that is absolutely critical for signaling via TLR4 and MAL (Fig. 1A). Despite the lack of a box 3 region and divergence in the box 2 region, computer modeling of the TIR region of TcpB predicted a conformation similar to TIR domains from MyD88, MAL, and other TLRs (42, 43) (Fig. 1B).

To further investigate the potential functional role of TcpB, we amplified the TcpB coding region from B. melitensis 16M genomic DNA and cloned it into the pcDNA3 expression vector. In vitro transcription/translation of this plasmid yielded a protein of MW ∼26 kDa (Fig. 1C), as predicted by the open-reading frame in the DNA sequence. The open-reading fame was then fused in frame with GST in a pGEX vector, and recombinant GST-TcpB was generated in E.coli BL21 cells. The purified GST-TcpB was used to generate rabbit polyclonal antisera, which recognized two bands of ∼25 kDa when TcpB was expressed in transfected 293 cells (Fig. 1D, lane 1). Both of these bands are specific for TcpB as they are not detected by the preimmune sera (Fig. 1D, lane 4), nor are they detected in lysates transfected with an empty pcDNA3 plasmid (Fig. 1D, lane 2). Moreover, preincubation of the Ab with the GST-TcpB protein abolishes the detection of TcpB expressed in 293 cells (Fig. 1D, lane 5). These two TcpB bands could result from differential posttranslational modifications, or might represent a product of premature termination of translation, but are unlikely to be due to differential phosphorylation (Supplemental Fig. 4).

It has been reported previously that B. abortus is recognized by TLR2 and TLR4 (44, 45). To test whether TcpB might have any effect on activation of NF-κB via the TLR4 signaling pathway, 293T cells were transfected with a constitutively active CD4/TLR4 chimera in the presence or absence of TcpB, and activation of NF-κB was determined by measuring activity of a NF-κB driven luciferase reporter construct. Coexpression of TcpB inhibited the activation of NF-κB by CD4/TLR4 in a dose dependent manner, as shown in Fig. 2A. Expression of a mutated form of TcpB, TcpBm, that contains a replacement of 3 aa (phenylalanine-isoleucine-serine) in box 1 region of TcpB-TIR (Fig. 1D, lane 3, 1E) did not have a similar inhibitory effect (Fig. 2A), suggesting that the box 1 region of TcpB is necessary for its function. Interestingly, however, TcpB could not inhibit the activation of NF-κB by overexpression of MAL, MyD88, TRIF, IRAK-1, or TRAF6 (Fig. 2C–F, respectively). These data indicate that TcpB likely inhibits TLR4 signaling by interfering with an event preceding the binding of TIR-containing adapters MAL, TRAM, MyD88, or TRIF.

It has recently been demonstrated that MAL contains an N-terminal phosphoinositide- binding domain that allows it to recruit MyD88 from the cytosol and target it to domains close to TLR2 and TLR4 on the plasma membrane. Remarkably, TcpB was also shown to resemble MAL by being able to bind phosphoinositides (37). The TIR domains of TLR2/4, MyD88 and MAL then create a platform that enables MyD88 to bind IRAK4 via its death domain to initiate signaling (46). We therefore wanted to test whether TcpB might inhibit signaling by interfering with the interactions between the TIR domain of TLR4 with MAL or MyD88. We transfected cells with TcpB, HA-MAL, or Flag-MyD88, immunoprecipitated TcpB, and analyzed the presence of coimmunoprecipitated MAL or MyD88 by immunoblotting with Abs against the HA or Flag epitopes. The results show that TcpB interacts with MAL (Fig. 3A), but not with MyD88 (Fig. 3B). In addition, neither TLR4 nor MyD88 immunoprecipitates TcpB (Fig. 3C) suggesting that TcpB functions through its interaction with MAL. Moreover, TcpB expression had no effect on interactions of MyD88 and MAL (Fig. 3D), which suggests that TcpB does not block TLR4 signaling by disrupting this interaction. Finally, TcpB was unable to interact directly with IRAK-1, IRAK-2, or TRAF-6, other key players in the TLR4 signaling pathway (data not shown).

Previous studies have shown that two forms of MAL are detected on expression in cells, where the slower migrating band corresponds to a phosphorylated form (47). As expected, in the absence of TcpB, expression of HA MAL generated two bands (Fig. 4B, lane 1), but in presence of increasing amounts of TcpB, the level of the upper band was dramatically reduced (Fig. 4A, lanes 2 and 3, top panel; 4B, lanes 1–4, top two panels). At the highest concentration of TcpB tested, the lower band is also reduced (Fig. 4B, lanes 1 and 4). The extent of reduction was more pronounced after 45 h than after 24 h (Fig. 4B). Treatment with CIP abolishes the slower migrating band confirming that the upper band corresponds to a phosphorylated version of MAL (Fig. 4E and Supplemental Fig. 4). The effect of TcpB on MAL is dependent on the intact box 1 region of TcpB, as a mutation in this region abrogated the effect of TcpB on MAL (Fig. 4D). This effect of TcpB was specific for MAL because TcpB had no detectable effect on coexpressed MyD88 (Fig. 4C). Remarkably, TcpB has no effect on a MAL DN mutant in which proline 125 has been changed to histidine (P125H), thus rendering this mutant defective in both phosphorylation and signaling (Fig.4F) (6, 7, 48). We also wanted to test whether TcpB had a similar effect on endogenous MAL. Using a rabbit mAb that recognizes endogenous MAL in 293 cells, we observed that the slower migrating form was abolished in presence of TcpB (Fig. 5A). Therefore, these results suggest that TcpB specifically abolishes the signal-competent, phosphorylated form of MAL, thereby affecting signaling through TLR4.

The results presented previously indicate that TcpB predominantly affects the phosphorylated version of MAL, although it is unclear if the observed reduction in the slower migrating form of MAL reflects reduced phosphorylation, enhanced dephosphorylation, or enhanced degradation. It has been shown previously that suppressor of cytokine stimulation-1 (SOCS-1) can negatively regulate signaling by TLR2 and TLR4 through degradation of MAL by an ubiquitination-dependent process (49). To test whether TcpB might also lead to enhanced, ubiquitin-dependent degradation of MAL, we expressed HA-MAL, TcpB, and VSV-ubiquitin in the presence or absence of the proteasome inhibitor, MG132. Although MAL expressed by itself is ubiquitinatinated to a low level, the extent of ubiquitination is dramatically increased in the presence of TcpB (Fig. 6, lane 5). This suggests that the phosphorylated form of MAL is most likely degraded by TcpB through a process that involves polyubiquitination. However, unlike SOCS-1, which is an E3 ubiquitin ligase (50, 51), TcpB has no obvious similarity to known ubiquitin ligases. Hence, it is unclear whether TcpB itself is responsible for enhanced ubiquitination of MAL, or whether it acts by recruiting some other ubiquitin ligase. If TcpB itself induces degradation of MAL via polybiquitination, then the functional domain responsible remains to be identified.

To determine whether the native TcpB has an effect on MAL levels in mammalian cells infected with Brucella strains, we evaluated the levels of this protein in the murine macrophage-like cell line J774 infected with virulent B. abortus strain 2308 and an isogenic tcpB deletion mutant designated KB2 (Supplemental Fig. 1). This cell line have been used extensively as a model for the interactions of bacterial pathogens, including Brucella strains with host macrophages (30, 52). Immunoblotting of Brucella lysates with the TcpB-specific Ab revealed that TcpB could only be detected in wt B. abortus 2308 but not the tcpB mutant (Supplemental Fig. 2). To test the effect of TcpB encoded by the bacteria on endogenous MAL, we infected J774 macrophages with wt and tcpB mutant of B.abortus 2308. As shown in Fig. 5B, the MAL-specific Ab could detect only one band in the J774 cell lysates. However, when J774 cells were infected with wt B. abortus 2308, MAL was significantly diminished suggesting that the visible band in these cells was most likely the phosphorylated form (Fig. 5B, lane 1 versus lane 3). In contrast, infection with the B. abortus tcpB mutant showed significantly less reduction of MAL levels (Fig. 5B, lane 2 versus lane 3). These results further support the model that TcpB causes the degradation of MAL. The level of MyD88 expression in the J774 cells remained unchanged in absence or presence of B. abortus 2308 or tcpB infection (Fig. 5C). These results further bolster our contention that TcpB does not exert its effect through interaction with MyD88.

The recombinant Brucella TcpB exhibits a clear impact on the activity of MAL when this protein is expressed in mammalian cells. The differential effects of infection of J774 cells with virulent B. abortus 2308 and the isogenic tcpB mutant also support the proposition that TcpB is secreted into the cytoplasm of infected host phagocytes where it has an effect on cellular levels of MAL. On the basis of these results, it was initially somewhat surprising to find that B. abortus 2308 and the isogenic tcpB mutant exhibit equivalent patterns of intracellular survival and replication in cultured murine peritoneal macrophages and spleen colonization profiles in experimentally infected BALB/c and C57BL6 mice (Supplemental Fig. 3). These findings, however, are consistent with other published reports that other Brucella mutants do not display attenuation in cultured murine macrophages, HeLa cells or in mice with intact immune systems.

The Gram-negative bacteria Brucella encodes a protein named TcpB, that has significant homology to TIR domains, particularly in the box 1 region. Remarkably TcpB acts as a negative regulator of TLR4-mediated signaling by inducing targeted degradation of MAL, one of the components of the TLR signaling cascade that shares phosphoinositide binding property of TcpB (37). Other components of this pathway, such as MyD88, TLR4, IRAK 1 and 4, and TRAF6 are not affected by TcpB, as shown by luciferase reporter assays as well as immunoprecipitation studies with overexpressed proteins. Immunoblotting analysis of endogenous MAL also showed targeted degradation of MAL but not of other adapters. Given this effect of TcpB on MAL it might seem surprising that when coexpressed in HEK 293 cells, TcpB does not inhibit the activation of NF-κB. We believe that in transfected cells there is sufficient level of other signaling competent TIR-domain adapters, thereby making it difficult to detect significant inhibition of NF-κB activation by MAL in presence of TcpB.

Unlike infection by Salmonella, infection by Brucella is not usually fatal, but causes chronic debilitating disease (20). This bacteria, just like many other intracellular microbes, tries to avoid elicitation of a strong immunologic response that might destroy its replicative niche. However, the overall survival of the host could be compromised if the bacteria completely cripple the host immune system. Therefore, many microbes have evolved finely tuned mechanisms that interfere with host immunity to create a replication permissive environment, but not act to completely destroy host immune function. By using TcpB to specifically inhibit MAL signaling, Brucella sp. can effectively dampen a critical microbial detection mechanism, while keeping the larger innate immune response intact. Degradation of the phosphorylated MAL molecules would afford an excellent means of exerting regulated control on the TLR4-MAL-MyD88 signaling pathway. As MAL is a unique adapter in TLR2 and TLR4 signaling, downregulation of MAL is a potent means of inhibiting the expression of proinflammatory cytokines and chemokines downstream of these receptors, while not inhibiting signaling via other TLRs. In fact, SOCS-1–mediated degradation of MAL after stimulation by LPS serves a similar purpose by helping to reset the cell to its basal state. TcpB in Brucella also acts as a negative regulator of TLR4-mediated responses by degrading the signal competent form of MAL. By targeting MAL, it ensures that only the TLR2 and TLR4 pathways (that respond to Gram-negative bacteria) are affected but not the other pathways that respond to different pathogens via different PAMPs.

Recently two other groups also have reported on TcpB (35, 39). A more recent study reported that TcpB was most similar to MAL, in particular in its ability to bind phosphoinositides (37). In agreement with our work, these studies also reported that TcpB inhibits TLR4 and TLR2-mediated signaling to NF-κB although the mechanism by which TcpB functioned was not determined. Our work is the first demonstration of a mechanism by which TcpB inhibits signaling by TLR4 and activation of NF-κB, namely, through targeted degradation of MAL. Moreover, in contrast to findings by the other groups, we see no effect of TcpB on activation of NF-κB activation by MyD88. Our immunoprecipitation data also shows that overexpressed MyD88 and TcpB do not interact, nor does TcpB cause degradation of MyD88.

It has been demonstrated that on LPS stimulation, MAL gets phosphorylated by Bruton’s kinase at Tyr-86, Tyr-106, and Tyr-187 and such phosphorylation is critical for signaling and activation of NF-κB. Substitution of tyrosine with phenylalanine or alanine at positions 86 and 187 not only abolishes phosphorylation of MAL but also renders MAL signaling-deficient (47, 48, 53). LPS stimulation of cells also leads to SOCS-1 mediated degradation of MAL and this degradation is absolutely dependent on phosphorylation of MAL (49). Thus, phosphorylation of MAL, followed by its degradation affords the cells an efficient means of downregulating robust MAL-MyD88–mediated TLR signaling. The recent finding that LPS-tolerized cells do not exhibit MAL phosphorylation and are signaling-deficient on secondary stimulation with LPS also supports such a model (47, 48, 53). Our data showing that TcpB degrades phospho-MAL and inhibits TLR4 signaling is also consistent with this notion. Despite its effect on signaling by MAL, Piao et al. (48) have shown that the phosphorylation-deficient MAL mutants Y86A, Y106A, and Y187A, have no effect on post receptor stimulation of NF-κB by MyD88, IRAK-2, or TRAF-6. This result also strongly supports our finding that TcpB has no effect on postreceptor activation of NF-κB by MyD88 and other adapters.

MAL is required for recruitment of MyD88 from the cytosol and delivering it to the plasma membrane via its phosphoinositide-containing domain. There, a tertiary complex between TLR4, MAL, and MyD88 is presumably formed through interaction of their TIR domains. Such complexes can provide the scaffold for formation of further signaling complexes containing IRAK4, IRAK1, etc. (46, 54). The phosphoinositide binding property of TcpB (37) probably allows it to colocalize to the same region in the membrane and facilitate its ability to interact with MAL and cause its degradation. Conformational changes in TLR (induced by ligation with PAMP or other means) probably result in release of these complexes from TLR4, as well as activation of the signaling cascade. Although nothing is known about the fate of the released MAL molecule at this stage, it is possible that they are recycled back to the plasma membrane to be repositioned and be ready to initiate another cycle of signaling. It has been shown that in 293 cells unphosphorylated MAL shows increased association with TLR4 and MyD88 (48, 54), suggesting that perhaps phosphorylation of MAL is necessary for the release of MAL from the TLR complex. By degrading MAL, TcpB reduces the amount of available phosphorylated MAL, thereby, slowing down the signaling process, without completely shutting down the system. In mouse macrophage cell lines, where we do not detect phosphorylated forms of MAL, TcpB could be exerting similar effect simply by reducing the amount of available MAL molecules by degrading them. Such slowdown would allow the bacteria to survive inside the cell without the possibility of causing cell death because of unregulated overactive signaling event (as seen in case of Salmonella poisoning) or because of fatal or severe sepsis, as seen in case of LPS tolerization.

We thank Drs. Ian Strickland and Matthew Hayden for their help with experiments as well as stimulating discussions.

Disclosures The authors have no financial conflicts of interest.