Identification of Francisella tularensis Lipoproteins That Stimulate the Toll-like Receptor (TLR) 2/TLR1 Heterodimer*

Francisella tularensis (Ft)² is a Gram-negative facultative intracellular bacterium that causes tularemia. Because Ft is highly infective, easy to propagate and disseminate by aerosol, and because treatment of infection often requires intensive medical care, it is a prime candidate for use in biological warfare and bioterrorism. Ft infects various eukaryotic cell types including professional phagocytes and non-phagocytic cells and is able to escape the phagosome/lysosome and replicate within the host cell cytoplasm. Very little is known at the molecular level about how Ft causes disease, and only a few potential virulence factors have been identified. The interaction of Ft with the innate immune system (reviewed in Ref. ,1) likely plays a major role in the pathogenesis of tularemia, yet it remains an area that has received little attention.

Host cells use several types of pattern recognition receptors to detect microbial products. Toll-like receptors are expressed on the host cell surface or in endosomal compartments where they detect products derived from bacteria, viruses, yeast, and protozoans (reviewed in Ref. 2). For example, the TLR4/MD-2 complex recognizes the lipopolysaccharide derived from Gram-negative bacteria while TLR2 mediates response to a large variety of microbial products, and most prominently to bacterial lipoproteins. The signature motif that characterizes lipoproteins is the lipobox, a short sequence of amino acids at the C-terminal end of their leader peptide that contains an invariant cysteine residue (3). The sulfhydryl group of this cysteine is post-translationally modified by addition of an N-acyldiacyl-glyceryl group. The leader peptide of the lipoprotein is then cleaved and the free N-terminal group of the cysteine can be further modified by addition of a third acyl chain, resulting in a triacylated lipoprotein. This type of protein lipidation is unique to bacteria and allows these proteins to be anchored to the membrane. Lipidation is essential for lipoprotein function and ability to activate TLR2. Several other microbial products can act as TLR2 agonists including lipoteichoic acid of Gram-positive bacteria, mycobacterial lipoarabinomannan, Neisserial porins, yeast cell wall products, glycosylphosphatidylinositol-anchored proteins from the protozoa Tripanosoma cruzi, Gram-positive peptidoglycan, and atypical LPS such as that of P. gingivalis. It has been demonstrated that TLR2 activation by some of these products (e.g. PGN or P. gingivalis LPS) is due to contaminating lipoproteins (4–6). The ability of TLR2 to mediate responses to such a varied group of molecules is in part due to the fact that this receptor must form heterodimers with either TLR6 or TLR1 to signal (7). TLR2/TLR1 heterodimer recognizes triacylated lipopeptides (8, 9) while TLR2/TLR6 mediates response to diacylated lipopeptides (10).

Infection with Ft is associated with a pronounced inflammatory response characterized by production of several cytokines (1). However, the components of the bacterium that induce inflammation remain unknown. We have recently reported that the inflammatory response resulting from infection of dendritic cells or PBMC is mediated by TLR2 and caspase-1 (11). These studies revealed that caspase-1 activation (and concomitant IL-1β secretion) was dependent on infection and escape of bacteria from the phagosome. In contrast, killed Ft retained the ability to activate TLR2 inducing a variety of cytokines and chemokines. The nature of the Ft-derived products that act as TLR2 ligands is unknown. In contrast to the LPS of most Gram-negative bacteria, Ft LPS does not possess proinflammatory activities (12, 13) and is unable to act like other atypical LPS that stimulate (or antagonize) TLR4/TLR2. The inability of Ft LPS to induce inflammatory responses suggests that other Ft-derived products are responsible for triggering the innate immune response. Lipoproteins are among the most likely candidates as TLR2 agonists, though until now no Ft-derived lipoproteins that possess this ability have been identified. In this article, we report the identification of two Ft lipoproteins that can activate TLR2-mediated signaling.

EXPERIMENTAL PROCEDURES

Reagents—Triacylated Pam₃-CSK4 (Pam) and diacylated synthetic lipopeptides FSL-1 and MALP2 were purchased from Invivogen (San Diego, CA). Escherichia coli LPS (K12 LCD25) was from List Biological Laboratories (Campbell, CA) and was purified from contaminant lipoproteins normally found in commercially available LPS preparations by double phenol extraction. Proteinase K was from Fisher Biotech. Lipoprotein lipase and PMA were from Sigma.

Bacterial Strains, Growth Condition, and Plasmids—E. coli BL21 lpxM strain (14) was grown in Luria-Bertani LB broth. F. tularensis LVS was grown in modified Brain Heart Infusion (BHI) broth (BHI supplemented with 50 mg/ml hemin and 1% IsoVitalex; BBL, Cockeysville, MD) or on BHI-chocolate agar (BHI agar supplemented with 1% hemoglobin and 1% IsoVitalex).

For expression in E. coli, Ft genes were PCR-amplified from Ft Schu4 genomic DNA using Pfu polymerase and cloned into the NcoI-XhoI sites of pET-28a (Novagen) resulting in addition of a C-terminal His₆ tag. Primer sequences are available upon request.

For expression of His₆-tagged TUL4 and FTT1103 in Ft we used the suicide vector M720 (a derivative of pXB136). This vector lacks the Ft origin of replication and contains the gene for kanamycin resistance under the control of the orf5 promoter. A 2,368-bp genomic fragment terminating at the 3′-end of Tul4 was PCR-amplified from Ft Schu4 genomic DNA using the following primers: F-SacI, GGGAGCTCACCATCTTTTGAGGGTGGTAGG and the primer R-6H-BamHI, GGGGATCCTTAATGATGGTGATGATGATGAATATTTATTGAATCAGAAGCGATTA, which adds the His₆ tag at the C terminus of TUL4. The resulting PCR product was digested with BamHI and SacI and cloned into the M720 vector. For Ftt1103 expression, the plasmid pET 28a-Ftt1103 was digested with SmaI-XbaI, and the 2,310-bp fragment containing the Ftt1103 sequence was cloned into M720. The resulting vectors were electroporated into Ft and cointegrants selected on BHI chocolate agar containing 10 μg/ml kanamycin. One clone from each transformation was selected and expression of the His₆-tagged protein was confirmed by Western immunoblotting using anti-His₆ antibody (Qiagen).

For expression of recombinant proteins in E. coli BL21 lpxM, overnight cultures were diluted ten times in warm LB broth and incubated at 37 °C. After 1 h, recombinant protein expression was induced by addition of 100 mm IPTG for 3 h. Ft clones expressing His₆-tagged TUL4 or FTT1103 were grown overnight in BHI broth.

Lipoprotein-enriched Fractions and Protein Purification—The lipid-associated membrane proteins were prepared using the Triton X-114 phase separation procedure (15) as described by T. Shimizu et al. (16). Briefly, Ft or E. coli overnight cultures were pelleted by centrifugation and washed twice with cold PBS. The cell pellets were resuspended in cold PTX (PBS supplemented with 350 mm NaCl, 2% Triton X-114, protease inhibitor mixture) and incubated at 4 °C for 1 h. Samples were centrifuged at 12,000 rpm 4 °C for 30 min, and the supernatants were incubated at 37 °C for 10 min to induce detergent phase separation. After centrifugation at 14,000 rpm for 10 min at room temperature, the upper aqueous phase was discarded and replaced with a similar volume of PBS supplemented with 350 mm NaCl. The procedure of phase separation was repeated twice, and the final detergent phase was resuspended in PBS to the original volume. To obtain the crude membrane/lipoprotein fraction, the detergent was removed by precipitation with 2.5 volumes of ethanol and incubated at –20 °C overnight. After centrifugation, the pellet was washed with ethanol, air-dried, and resuspended in PBS. Protein concentration of the suspension was measured with the Bradford assay.

For purification of the recombinant lipoproteins, E. coli IPTG-induced cultures or Ft overnight cultures were centrifuged and washed in cold PBS. The pellets were resuspended in PTX, and the lipoprotein-enriched fractions were obtained as described above. After two rounds of detergent phase partitioning, 10 mm imidazole, and 400 μl of Probond Ni-Sepharose resin (Invitrogen) were added. After 1 h of incubation at 4 °C, the resin was washed with 20 ml of PTX containing 20 mm imidazole. Bound lipoproteins were eluted with 2 ml of PTX containing 300 mm imidazole, pH 8.0. The eluted fractions were incubated at 37 °C for 10 min to induce detergent phase partitioning. After centrifugation at room temperature for 10 min at 14,000 rpm, the upper phase was discarded, and the detergent fraction was diluted with 2.5 ml of PBS supplemented with 350 mm NaCl and subjected to a second affinity chromatography with 200 μl of Ni²⁺ resin. The beads were washed with 20 ml of PTX pH 5.5. The bound lipoproteins were eluted in PTX by stepwise decreasing the pH to 4.0. Each fraction was precipitated by addition of 2.5 volumes of ethanol, and the pellets resuspended in sterile PBS.

The cysteine mutant lipoproteins and the chaperones were purified from bacterial pellets by freeze-thawing four times and sonication in lysis buffer (1× PBS, 350 mm NaCl, 0.5% Nonidet P-40, 10 mm imidazole, protease inhibitor mixture) and were further purified by affinity chromatography with Probond Ni-Sepharose. The resin was washed twice with lysis buffer and elution was accomplished by stepwise decreasing the pH to 4.0. The purity of each fraction of recombinant proteins was determined by Coomassie Blue staining. Recombinant proteins were digested with proteinase K (1 μg/ml, 2 h, 42 °C), lipoprotein lipase (10 μg/ml, 18 h 37 °C,) or NaOH 0.2 n, 2 h at room temperature.

Mammalian Expression Vectors and Cell Lines—The chimeric pCMV-Flag-TLR[2-1], pCMV-Flag-TLR[1-2], and pEF-Flag-TLR[6-2] constructs have been described previously (11, 17). The chimeric constructs LRR1–4, LRR5–8, LRR9–12, LRR13–17, and LRR18-C were constructed by overlapping PCR using human TLR1 or TLR6 cDNA templates and cloning back into the CMV-Flag-TLR[1-2]. Primer sequences are available upon request. A subclone of HeLa cell line, previously isolated in our laboratory, was grown in Dulbecco's modified Eagle's medium, 10% fetal calf serum. The HeLa-TLR2, or HeLa-TLR4/MD2 stable cell lines have been described previously (18). To obtain the HeLa-TLR[2-1] stable cell line, the parental HeLa cells were co-transfected with the pCMV-Flag-TLR[2-1] and the pCMV-puro vectors. Stable transfectants were selected in presence of puromycin (2 μg/ml) and tested for responsiveness to Pam or MALP2 stimulation after transient transfection with the pCMV-Flag-TLR[1-2] or pEF-Flag-TLR[6-2] vectors.

Luciferase Assay—The HeLa cell lines were transiently transfected in 24- or 48-well plates using Effectene reagent (Qiagen) with 0.4 μg of ELAM-luciferase and 0.2 μg of pcDNA-CD14, 0.1 μg of CMV-β-Gal, and 0.7 μg of empty vector or other chimeric constructs (see figure legends). Forty-eight hours after transfection, cells were stimulated for 6 h with different agonists. Lucif-erase assay was performed using Promega (Madison, WI) reagents according to the manufacturer's recommendations. Efficiency of transfection was normalized by measuring β-Gal in cell lysates. All experiments were repeated at least three times. Results from three independent experiments were combined and expressed as lucifer-ase fold induction (compared with the unstimulated cells). Values are mean ± S.D.

Primary Cell Isolation—Human PBMC were isolated from Leuko-packs by Ficoll-Hystopaque density gradient centrifugation. To obtain mouse dendritic cells from wt or TLR2-deficient mice, the femur/tibiae were removed and freed of muscles and tendons. Bone ends were cut with scissors and bone marrow cells were flushed out of the bone cavity with D-PBS using a syringe with a no. 25-gauge needle. The cell suspension was filtered through a 70-μm cell strainer, washed once, and resuspended in RPMI 1640-10% fetal calf serum supplemented with penicillin (100 units/ml), streptomycin (100 μg/ml), and rmGM-CSF (20 ng/ml) (R&D Systems, Minneapolis, MN). The cell suspension was plated at 1 × 10⁶ cells/ml in 100-mm Petri dishes. The medium was replaced on day 4, and suspension cells collected and stimulated on day 8. This procedure routinely results in >80% CD11c+ cells. The use of animals for these experiments was approved by the University of Tennessee Animal Care and Use Committee.

RNase Protection Assay—Total RNA was isolated using TRIzol reagent (Invitrogen). RNase protection assay was performed using 4 μg of total RNA using BD-PharMingen (San Diego, CA) Riboquant kit as described previously (18). The hCK-1 and hCK-5, multi-probe template sets were used.

Western Blot Analysis—Purified lipoproteins or lipoprotein fractions were separated on SDS-PAGE, transferred to nitrocellulose membranes, and immunoblotted using serum of Ft-immunized mice (1:500).

RESULTS

Ft Lipoprotein Fraction Activates Both TLR2 Heterodimers—We have previously shown that the innate immune response to Ft infection is mediated by TLR2 (11). To identify the nature of the Ft-derived factors that activate TLR2, we obtained a Ft lipoprotein-enriched fraction using Triton X-114 detergent extraction and partition procedure (15, 16). This procedure selectively enriches for integral membrane proteins and lipoproteins. A similar fraction was obtained for comparison from E. coli BL21 lpxM, a strain that produces a modified form of lipid A that is unable to stimulate TLR4 (14). The ability of these membrane protein fractions to stimulate TLR2 or other TLR was tested using HeLa cell lines expressing different TLR. As shown in Fig. 1A, both fractions were able to activate cells expressing TLR2. In contrast, neither fraction activated cells expressing TLR4/MD-2 (Fig. 1B), TLR5 (Fig. 1C), or TLR7/TLR8 (Fig. 1D). TLR2 activation by both lipoprotein fractions was abolished by alkaline treatment. This treatment hydrolyzes the acyl groups of lipoproteins destroying their ability to stimulate TLR2. The ability of lipoproteins to activate TLR2 is known to be resistant to proteinase K treatment (4), and as expected, the TLR2 agonist activity of the E. coli fraction was not affected by this treatment. In contrast, digestion of the Ft fraction with proteinase K reduced TLR2 activation suggesting that the Ft fraction may contain two types of TLR2 agonists, one type that is resistant to protease digestion and that may represent a lipopeptide and a second type that is susceptible to protease treatment and that may represent an integral membrane protein.

TLR2 is known to recognize its ligands by forming heterodimers with either TLR1 or TLR6. The TLR1/TLR2 heterodimer mediates response to triacylated bacterial lipopeptide while cell activation by diacylated lipopeptide typically occurs through the TLR6/TLR2 heterodimer (8–10) (activation of the TLR1/TLR2 heterodimer has also been reported in response to some diacylated lipopeptides (19)). Because our HeLa or HEK293 cell lines express endogenous TLR1 and TLR6, to determine which TLR2 heterodimer is activated by the Ft lipoprotein fraction we employed an assay that relies on chimeric constructs that swap the extracellular and intracellular domains between TLR1, TLR2, and TLR6. As we previously demonstrated (11, 17) and illustrate in Fig. 2B, these chimeric receptors cannot function in isolation and restore responsiveness to TLR2 agonists only when used in complementing pairs that simultaneously provide both the cytoplasmic and the extracellular domain of TLR1 and TLR2 (or TLR6 and TLR2). As expected (Fig. 2A), the chimeric pair TLR[1-2] plus TLR[2-1] was activated by the triacylated lipopeptide Pam but not by the diacylated lipopeptide MALP-2. In contrast, the chimeric pair TLR[6-2] plus TLR[2-1] was activated exclusively by diacylated synthetic lipopeptide MALP-2. In this assay, the Ft lipoprotein fraction stimulated both heterodimers, in agreement with our previous results using the whole bacterium (11). Interestingly, digestion with proteinase K severely diminished the ability of the Ft lipoprotein fraction to stimulate the TLR[6-2] plus TLR[2-1] heterodimer while it affected the TLR[1-2] plus TLR[2-1] agonism to a much lower extent. This effect was not observed with the E. coli lipoprotein fraction whose ability to stimulate both TLR2 heterodimers was not affected by proteinase K treatment. These results are in agreement with the notion that Ft may produce triacylated lipoproteins that activate preferentially the TLR2/TLR1 heterodimer, and other proteins, possibly unlipidated ones, that stimulate the TLR2/TLR6 heterodimer.

Identification of Ft Lipoproteins That Stimulate the TLR2/TLR1 Heterodimer—A search of the DOLOP and the NCBI databases identified several Ft genes predicted to encode putative lipoproteins. We decided to produce in recombinant form a number of these proteins (see Table 1). Because several bacterial chaperones have been reported to activate TLR (20–22) we also decided to express three Ft chaperone proteins, GroES, GroEL, GrpE. In addition, an E. coli lipoprotein, Lip19, which is known to activate TLR2 (23), was produced and used as positive control. C-terminal histidine-tagged version of these proteins were constructed and expressed in E. coli BL21 lpxM by IPTG induction. Several of the selected Ft proteins were expressed at levels too low to attempt purification (lipoproteins are notoriously difficult to express in high amount) and, therefore, were excluded from further analysis (see Table 1). The remaining proteins were purified using an established procedure that relies on Triton X-114 detergent phase partitioning and nickel resin chromatography. Each purified protein was tested on HeLa-TLR2 cells for the ability to activate NF-κB-luciferase. Among the proteins tested (Table 1), only two (encoded by the genes Ftt0901 and Ftt1103) were able to stimulate TLR2 in a dose-dependent fashion (Fig. 3B). Ftt0901 encodes TUL4, a 17-kDa protein also known as LpnA, which is one of the few known Ft targets for T cell-mediated immune responses in humans (24). TUL4 was previously shown to be a lipoprotein (25). The protein encoded by Ftt1103 has not been previously characterized. TUL4 and FTT1103 failed to stimulate TLR4, TLR5, or TLR7 (data not shown). Other predicted lipoproteins such as FTT0165c, or the three Ft chaperones GroES, GroEL, GrpE, did not show TLR2 agonism regardless of the dose used (Fig. 3B). Because the same procedure was used to purify the three lipoproteins, it is unlikely that the TLR2 agonism of FTT1103 and TUL4 was due to contaminating E. coli lipoproteins, though this remains a possibility. As expected, mutation of the lipobox cysteine residue (the amino acid to which N-acyl diacyl glyceryl group is attached) abolished detergent phase partitioning of FTT1103-Cys and TUL4-Cys and their ability to stimulate TLR-2. Moreover, delipidation of FTT1103 and TUL4 using alkaline hydrolysis or lipoprotein lipase digestion abolished their stimulation of TLR2 (Fig. 3D). In contrast, digestion with proteinase K, which completely destroyed the proteins (Fig. 3C), did not reduce (but in fact, increased) their ability to serve as TLR2 agonists.

The chimeric TLR assay was used to determine the composition of the TLR2 heterodimer stimulated by TUL4 and FTT1103. As shown in Fig. 4, both proteins stimulated exclusively the TLR2/TLR1 heterodimer suggesting that these proteins may be triacylated. Together, these findings demonstrate conclusively that TUL4 and FTT1103 are lipoproteins that are capable of TLR2 stimulation.

TUL4 and FTT1103 Induce TLR2-dependent Proinflamma-tory Chemokine Expression—We next analyzed the ability of TUL4 and FTT1103 to induce a panel of chemokines in human PBMC or mouse bone marrow-derived DC. As shown in Fig. 5, both proteins, but not the cysteine mutants, were powerful inducers of several chemokines. Importantly, chemokine induction by the purified proteins or the Ft lipoprotein fraction was abrogated in BMDC from TLR2-deficient mice confirming our previous demonstration that the inflammatory response to Ft is primarily mediated by TLR2 (11). Remarkably, the chemokine profile induced by purified TUL4 and FTT1103 closely resembled that induced by the crude Ft lipoprotein fraction or the live bacterium, suggesting that these proteins play a predominant role in the innate immune response to Ft infection.

TUL4 and FTT1103 Expressed in Either Ft or E. coli Display Similar TLR2/TLR1 Agonist Activity—Although previous work has demonstrated that Ft and E. coli use similar mechanisms to process lipoproteins (25), it was important to demonstrate that TUL4 and FTT1103 expressed in Ft behave similarly to the E. coli-expressed proteins. We first attempted to express histi-dine-tagged versions of TUL4 and FTT1103 in Ft LVS using the pXB168 vector (26), a high copy number plasmid that carries a strong constitutive Ft promoter. Although we could obtain Ft LVS transformants expressing EGFP using the same vector, we were unable to recover transformants overexpressing Tul4 or Ftt1103, suggesting that overexpression of these genes is lethal. Because a system for regulated gene expression in Ft does not exist, we circumvented this problem by using a suicide vector that integrates into the chromosome of LVS by homologous recombination. This procedure resulted in the generation of merodiploid Ft strains that expressed both the wild type and His₆-tagged versions of Tul4 and Ftt1103 using their respective endogenous promoters. Expression of the tagged proteins was confirmed by immunoblot using anti-His₆ antibody (not shown). Mice immunized with killed Ft have been shown to develop anti-TUL4 antibodies (27). Using a mouse serum that contains anti-TUL4 antibodies we were able to demonstrate that the His₆-tagged TUL4 is expressed at levels comparable to the endogenous protein (Fig. 6A). The same serum did not recognize recombinant FTT1103, though this protein was reported to induce humoral responses in mice immunized with Ft LVS (28) and in tularemia patients (29). Ft-expressed TUL4 and FTT1103 were purified using Triton X-114 detergent phase partitioning and nickel resin chromatography procedure (Fig. 6B) and tested for the ability to stimulate TLR2. As shown in Fig. 6, C and D, TUL4 and FTT1103 expressed in Ft behave similarly to the respective proteins expressed in E. coli being able to stimulate exclusively the TLR2/TLR1 heterodimer.

Identification of a Domain of TLR1 That Confers Responsiveness to TUL4 and FTT1103—One of the most puzzling aspects of TLR biology concerns the way TLR recognize their agonists. In the case of TLR2, it is believed that subtle changes in sequence between TLR1 and TLR6 (which share the highest homology between TLR) must be responsible for the ligand specificity of the heterodimer. To identify the region of TLR1 that is responsible for recognition of TUL4 and FTT1103, we constructed a series of chimeric TLR that swapped groups of LRR between TLR1 and TLR6. These constructs were engineered in the chimeric TLR[1-2]. As shown in Fig. 7, substitution of LRR 1–4 of TLR1 with the corresponding LRR of TLR6 modestly affected the response to Ft lipoproteins or Pam. The response to these agonists was reduced by substitution of LRR 5–8 and completely abolished by substitution of LRR 9–12 and LRR 13–17. Substitution of LRR 18–20 and the membrane-proximal Cys-rich region did not affect the response. These results indicate that an extended region that comprises LRR 9–17 of TLR1 is important for discrimination of and responsiveness to Ft lipoproteins and triacylated lipopeptide by the TLR2/TLR1 heterodimer. The importance of this region of TLR1 for Ft lipoproteins recognition was demonstrated by the fact that grafting TLR1 LRR 9–17 into the TLR6 ectodomain imparted it with the ability to respond to the Ft lipoproteins. Grafting LRR 9–12 or LRR 13–17 in isolation did not restore responsiveness (not shown) suggesting that an extended portion of the extracellular domain of TLR1 determines ligand specificity and responsiveness.

DISCUSSION

We and others (11, 30, 31) have previously shown that the innate immune response to Ft is mediated by TLR2. These studies did not offer clues as to the nature of the Ft-derived product that acted as TLR2 agonist. In this article, we report for the first time the identification of two Ft proteins that specifically stimulated TLR2. Several observations suggest that TUL4 and FTT1103 are lipoproteins. First, bioinformatic analysis identifies the presence of a classical lipobox in both sequences. Second, TUL4 was previously shown to be a triacylated lipoprotein (25). Third, both proteins were previously shown to be located in the outer membrane of Ft (32), in agreement with our results that showed that they partition in the Triton X-114 detergent phase. Finally, the modality by which they stimulate TLR2 (resistance to proteinase K, but not to lipoprotein lipase digestion) strongly suggests that both proteins are in fact lipi-dated. By employing an innovative assay based on chimeric TLR proteins we were able to further demonstrate that TUL4 and FTT1103 activate exclusively the TLR2/TLR1 heterodimer. This result is in agreement with the demonstrated triacylated nature of TUL4 (25) and suggests that FTT1103 may also be triacylated. Interestingly, our results indicate that the Ft lipoprotein fraction contains determinants other than TUL4 and FTT1103 that also stimulate the TLR2/TLR6 heterodimer. Ongoing studies are aimed at identifying this factor(s). The fact that proteinase K treatment reduced the TLR2/TLR6 agonism suggests that this factor may not be a lipoprotein. (TLR2-stimulating activity is not abolished by proteinase K treatment.) Rather, more likely candidates may be bacterial porins, highly hydrophobic integral membrane proteins several of which have been demonstrated to activate TLR2. Interestingly, Shigella porins are reported to stimulate the TLR2/TLR6 heterodimer (33) while the Neisseria porins stimulate the TLR2/TLR1 pair (34). Recently, it was reported that TLR2 activation by Ft can occur in the absence of bacterial replication, yet it requires de novo bacterial protein synthesis (35). This observation contrasts with the results presented here and with our previous work that showed that bacterial viability and host cell infection are required for caspase-1 activation but not for TLR2 activation or cytokine induction (11). It is also interesting to note that immunization of mice with killed Ft elicits protective immunity comparable to that promoted by immunization with the live organism (36, 37) suggesting that active bacterial metabolism is not necessary to stimulate an immune response.

Infection with Ft is associated with a powerful inflammatory response that plays a prominent role in the pathogenesis of tularemia. Our results showed that purified TUL4 and FTT1103 are powerful stimulators of TLR2-dependent cytokine production, suggesting they may represent virulence factors. Future studies will determine whether ablation of these genes from the Ft genome results in decreased pathogenicity.

Through domain-exchange analysis, we determined that an extended region that comprises LRR 9–17 in the extracellular portion of TLR1 mediates response to Ft lipoproteins and triacylated lipopeptide. Our results also showed that grafting this region of TLR1 into the corresponding region of TLR6 imparted it the ability to respond to Ft lipoproteins and lipopeptide indicating that residues outside LRR 9–17 are not involved in the ability of TLR1 and TLR6 to discriminate between triacylated and diacylated lipopeptides. Previous studies that analyzed the structural requirements for lipopeptide recognition determined that the first 7 N-terminal LRR of TLR2 are not involved in ligand recognition (38) and that LRR 9–12 of TLR1 are required for responsiveness to lipopeptide (39). However, this last study concluded that this region alone is not sufficient to mediate lipopeptide responses, suggesting that additional LRR surrounding this area are important. Our results have now demarcated the boundaries of the TLR1 domain that imparts responsiveness to lipoproteins and triacylated lipopeptide. During the review of our manuscript, the crystal structure of the TLR1/TLR2 heterodimer bound to triacylated lipopeptide has been published (40). That work revealed that the two ester-bound lipid chains of the lipopep-tide are inserted into a pocket in TLR2 while the amide-bound lipid chain is inserted into a hydrophobic channel in TLR1. Both TLR1 and TLR2 hydrophobic pockets are formed by residues contained in LRR 9–12. In addition, that work showed that the dimerization interface between TLR1 and TLR2 is composed by residues found in LRR 11 through LRR14, a result that is in agreement with our finding that LRR 9–17 of TLR1 is required for responsiveness to Ft lipoproteins and triacylated lipopeptide.