Abstract
Members of the Toll-like receptor (TLR) family are components of the mammalian anti-microbial response, signaling with a domain closely related to that of IL-1 receptors. In this report the expression and function of TLR1, a TLR of unknown function, are examined. TLR1 is expressed by monocytes, as demonstrated using a novel mAb. Monocytes also express TLR2. TLR1 transfection of HeLa cells, which express neither TLR1 nor TLR2, was not sufficient to confer responsiveness to several microbial extracts. However, cotransfection of TLR1 and TLR2 resulted in enhanced signaling by HeLa cells to soluble factors released from Neisseria meningitidis relative to the response with either TLR alone. This phenomenon was also seen with high concentrations of some preparations of LPS. The N. meningitidis factors recognized by TLR1/TLR2 were not released by N. meningitidis mutant in the LpxA gene. Although LpxA is required for LPS biosynthesis, because cooperation between TLR1 and TLR2 was not seen with all LPS preparations, the microbial component(s) TLR1/2 recognizes is likely to be a complex of LPS and other molecules or a compound metabolically and chemically related to LPS. The functional IL-1R consists of a heterodimer; this report suggests a similar mechanism for TLR1 and TLR2, for certain agonists. These data further suggest that mammalian responsiveness to some bacterial products may be mediated by combinations of TLRs, suggesting a mechanism for diversifying the repertoire of Toll-mediated responses.
An important component of mammalian anti-microbial defenses is the innate immune system, a group of proteins that recognizes conserved features of pathogens and triggers host responses. These microbe recognition proteins have been termed pattern recognition receptors (1). The innate immune system is found in plants and animals, and in both kingdoms Toll-like receptor (TLR)3 signaling molecules play crucial roles in the induction of anti-microbial responses (2). Animal TLRs are transmembrane molecules comprising multiple extracellular leucine-rich repeats, a single transmembrane domain, and an intracellular signaling domain. The signaling domain has been termed TIR (Toll and IL-1 receptor related) (3), because closely related domains are found in the IL-1R family.
At least six mammalian TLRs have been characterized (4, 5), and it seems likely that others remain to be identified. It is clear from genetic, biochemical, and Ab blockade experiments that the physiological roles of the two best-characterized TLRs, TLR2 and TLR4, differ: TLR4 is required for in vivo LPS responsiveness, whereas TLR2 is required for responsiveness to microbial products other than LPS. TLR2 agonists include lipoproteins (6, 7, 8, 9). The current data are compatible with a model in which TLRs are receptors for conserved microbial determinants, while other pattern recognition receptors, such as CD14, lack signaling domains but present microbial determinants to TLRs (10, 11, 12).
The functional form of transmembrane receptors commonly comprises several different proteins, which form heteromeric complexes. Individual components may express poorly on their own (13, 14). Binding and signaling functions may be subserved by different proteins of the complex. An example of this is the mammalian T and B cell Ag receptors, in which covalently linked binding chains engage ligand (15), while other receptor chains are required for signaling, but not for ligand binding (16). In other cases, although binding and signaling domains may exist on the same protein, ligand specificity is determined by combinations of multiple receptor chains, as seen in the cytokine (17, 18) and TGF-β receptor families (19).
Heterodimerization is important in the IL-1R family, which, like the TLR family, contains receptors with TIR signaling domains. IL-1 and IL-18 signaling is mediated not by a single IL-1R family member, but by a heterodimer consisting of two family members. Only one chain is a high affinity cytokine binding protein (20, 21), although the accessory member of the heterodimer is essential for cytokine signaling (22, 23). Accessory proteins are structurally very similar to other family members, and their accessory role cannot be distinguished from that of other IL-1R family members by primary sequence analysis.
In this study we observed that responsiveness to some products of Gram-negative bacteria by transfected HeLa cells is influenced by coexpression of both TLR2 and TLR1, a TLR of previously unknown function. Using an mAb generated against TLR1, we showed that TLR1, like TLR2 (10), is expressed in human mononuclear cells and propose a model involving functional heterodimerization between TLRs to explain the cooperative effect observed between TLR1 and TLR2.
Materials and Methods
cDNA preparation
RNA was extracted from HeLa cells and from Ficoll-Hypaque (Lymphoprep, Nycomed)-purified human PBMC using RNAzol B (Biogenesis, Bournemouth, U.K.). cDNA was made using AMV retrotranscriptase (Promega, Madison, WI) with oligo(dT) priming, according to the manufacturer’s recommendations. Controls without retrotranscription were also generated.
Vectors
pCMV-TLR1.
A clone of the human TLR1 open reading frame was obtained from human PBMC cDNA by PCR using Pfu polymerase (Stratagene, La Jolla, CA) and a TA cloning kit (TA-TOPO, Invitrogen, San Diego, CA). Primers are shown in Table I⇓. Three independent TLR1 clones, obtained by PCR from PBMC cDNA of a single individual, were sequenced. The sequence found differs from that published (GenBank D13637), having a G at position 302. This codon (AGA) specifies R (not T). HinfI restriction of TLR1 fragments amplified from PBMC cDNA from healthy volunteers confirmed that the sequence at position 302 of our clone was also present in all six individuals examined (data not shown). pCMV-TLR1 was generated by inserting the cloned full-length TLR1 open reading frame into pCDNA3 (Invitrogen).
Primers and amplification conditionsa
pCMV-TLR1Δcyt.
pCMV-TLR1Δcyt encoding only the extracellular and transmembrane portions of TLR1 was generated by joining the Asp718I-HindIII fragment of pCMV-TLR1 with a HindIII-XbaI-digested mutant TLR1 fragment. PCR mutagenesis with primers 5′-TCAGATGTTTACCTCCCAGGATCAAG and the reverse mutagenic primer 5′-TCTAGATTATTACCTGCGCCGGGTCTGGGTCCACTG-3′ was used to insert stop codons after the stop transfer signal RRRAR.
pFLAG-TLR2.
The TLR2 open reading frame was amplified as described for TLR1. Primers used are shown in Table I⇑. The product has the sequence previously described (24). pFLAG-TLR2 is the amplified TLR2 NotI-XbaI fragment subcloned into pFLAG-CMV1 (Sigma, St. Louis, MO), which replaces TLR2 aa 1–17 (24) with a preprotrypsin leader and FLAG epitope.
Other vectors.
pFLAG-TLR4 vector (pFLAG-hToll) (25) was a gift from Dr. M. Muzio (Instituto Medic Negri, Milan, Italy). pFLAG-MD2 was constructed as described for TLR2, using primers shown in shown in Table I⇑. The amplified sequence found was identical with that described previously (26). pFLAG-IL-1RAcP, which produces flag-tagged IL-1R accessory protein (27), was a gift from Dr. F. Volpe, Glaxo-Wellcome (Stevenage, U.K.). pCDM8-CD14 (28) was a gift from Prof. B. Seed (Massachusetts General Hospital, Boston, MA). pPL-pLUC consists of the −174 to +45 region of the human IL-8 gene (GenBank M28130, bases 1310–1645) subcloned into the firefly luciferase vector pGL3-Basic (Promega), creating pIL-8-pLuc. pTK-rLUC was obtained from Promega. Plasmid identities were confirmed by restriction mapping; the sequences of all constructs and mutants constructed here were confirmed by sequencing. All plasmids were prepared using Qiagen endotoxin-free Maxiprep kits (Valencia, CA).
TLR1Fc production
An XhoI site was introduced by PCR into the TLR1 sequence at the predicted extracellular-transmembrane junction. The TLR1Fc expression vector pCMV-TLR1Fc was constructed by ligation of the TLR1 extracellular domain into a pCDNA3 vector containing an engineered XhoI site 5′ to the hinge of the human IgG1 heavy chain. The predicted amino acid sequence of the junction is SELSCNSSGDKKV. A TLR1Fc-producing NSO/1 murine myeloma line was produced as previously described (29). Briefly, cells were electroporated with pCMV-TLR1Fc, selected with 1 mg/ml G418 (Life Technologies, Gaithersburg, MD), and screened for Fc production. A secreting clone was established, subcloned twice, and grown in Spinner cultures in RPMI plus 1% low IgG FBS (both from Life Technologies). TLR1Fc was purified from batches of 20–30 liters of conditioned medium by protein A affinity chromatography. Elution with 25 mM MOPS and 4 M MgCl2 (pH 6.0) was followed by gel filtration into PBS (pH 7.2) and freezing at −80°C. Purity, as assessed by SDS-PAGE, was >95%. Concentrations of TLR1Fc were determined with a human Fcγ-specific sandwich ELISA. This used specific polyclonal goat anti-human Fcγ antisera (Jackson ImmunoResearch Laboratories, West Grove, PA) and was standardized with human IgG1 myeloma protein (Calbiochem, Torrance, CA).
mAb production
Using standard techniques, mice were immunized repeatedly with TLR1Fc. Following splenocyte fusion with the partner NSO/1, hybridoma supernatants were screened for anti-TLR1Fc reactivity by ELISA and for reactivity with TLR1-transfected HeLa cells by flow cytometry as described below. Here we describe one reactive clone, GD2.F4, of isotype IgG1, as determined by an isotyping kit (Serotec, Oxford, U.K.).
HeLa cell culture
HeLa cells (American Type Culture Collection, Manassas, VA) were maintained in DMEM (Life Technologies) plus 10% FBS. All FBS used was heat treated for 30 min at 56°C. The medium and serum were endotoxin free, as tested by the manufacturer. All cell lines used tested negative for mycoplasma.
Analysis of TLR mRNA expression in HeLa cells and PBMC
Expression of TLR1, TLR2, TLR4, MD-2, and the housekeeping gene encoding the ribosomal L41- homologue protein (GenBank Z12962) was analyzed in PBL and HeLa cells. The retrotranscript of 40 ng of total RNA was used as template for PCR with Taq polymerase (Promega). Conditions and primers used are shown in shown in Table I⇑. The TLR4 primers span the apparent unexcised intron in GenBank sequence U88880, but not U93091.
Flow cytometry of HeLa cells
HeLa cells were grown in 100-mm dishes to 50% confluence and transfected with pCMV-TLR1 (10 μg) using the calcium phosphate method (30). For comparison of expression levels on cotransfection of different TLRs, 5 μg each of TLR expression vector or pCDNA3 were used. Control transfections with a pCMV-enhanced green fluorescent protein vector showed >70% transfection efficiency under these conditions. Thirty-six hours after transfection cells were washed off the plate using PBS and 5 mM EDTA (pH 7.2) and resuspended in staining buffer (PBS, 5% goat serum, and 0.1% sodium azide) on ice. mAb GD2.F4 or IgG1 anti-keyhole limpet hemocyanin isotype control (PharMingen, San Diego, CA) was added to 5 μg/ml. A second layer consisted of staining buffer containing 3 μg/ml biotinylated goat anti-mouse IgG (Jackson ImmunoResearch). The third layer was staining buffer containing streptavidin-PE (Sigma; 10%, v/v). Cells were washed twice between each step. Fluorescence was measured with a FACSCalibur cytometer (Becton Dickinson, Mountain View, CA) with logarithmic FL2 amplifier.
Flow cytometry of mononuclear cells
Mononuclear cells from venous blood of healthy donors were purified by centrifugal elutriation. The monocyte population was identified by size gating, and purity was confirmed by anti-HLA-DR and anti-CD14 immunostaining. One hundred percent of the monocytes expressed these markers (data not shown). Cells (5 × 105) were suspended in staining buffer (PBS, 0.1% BSA, and 0.02% sodium azide) with 100 μg/ml human IgG and 15 μg/ml GD2.F4 or isotype control. Secondary Ab was a one-eighth dilution of goat anti mouse IgG F(ab′)2-FITC conjugate (Roche, Indianapolis, IN). Cells were washed twice with staining buffer between each step and were analyzed by flow cytometry.
Bacterial growth and LPS preparations
Meningitidis strain H44/76 (pLAK33) was derived from wild-type clinical isolate H44/76 by targeted disruption of the lpxA gene as previously described (31). N. meningitidis H44/76 LPS, isolated by the hot phenol extraction method (32), was a gift from Dr. P. van der Ley (National Institute of Public Health and Environment, Bilthoven, The Netherlands). Highly purified Escherichia coli K235 LPS, prepared using a modified extraction procedure (33), was a gift of Dr. S. Vogel (Uniformed Services University of the Health Sciences, Bethesda, MD). Other LPS preparations used were purified by phenol/water extraction and purchased from Sigma (Gillingham, Dorset, U.K.). Stock solutions were stored dissolved in PBS at 5 mg/ml. Soluble factors released by live N. meningitidis were collected as follows. Soluble factors released by N. meningitidis were collected by growing H44/76 N. meningitidis in Mueller-Hinton broth (Oxoid; Basingstoke, Kent, U.K.) to mid-log phase, pelleting, washing twice in DMEM, and resuspending at ∼2 × 108 CFU/ml in a 5-ml sterile endotoxin-free Sterilin (Stone, Staffs, U.K.) vessel. After incubation at 200 rpm at 37°C in 5% CO2 for 1 h, conditioned culture medium was filtered using Acrodisc 0.2-μm pore size filters (Gelman, Oberlin, OH).
Transfection and reporter assays in HeLa cells
HeLa cells (ECACC, 85060701; 1.5 × 104/well) were seeded into 96-well tissue culture plates 24 h before transfection. Transfections were performed using SuperFect (Qiagen) according to the manufacturer’s advice; each well received 25 ng of pCDM8-CD14, 400 ng of pPL-IL-8, 100 ng of pTK-rLuc for normalization of transfection efficiency, and 50 ng of the TLR-expressing vector under test. Sufficient pcDNA3 (Invitrogen; empty vector) was added to keep the total DNA dose constant. After transfection, cells were washed, and 100 μl of fresh medium was added. Triplicate wells were transfected for each treatment. Twenty-four hours later, agonists were added. These were prepared and added as 10× stock solutions in PBS. Thirty-six hours following transfection reporter levels were measured using the dual luciferase system (Promega) as recommended by the manufacturer. Normalized IL-8 promoter activity is the ratio of firefly to Renilla luciferase activity.
IL-8 ELISA
HeLa cells were transfected as described above, except that 500 ng of pCDNA3 was used in place of the reporter vectors, and 200 μl of medium was added following transfection. Twenty-four hours later 100 μl was aspirated, and agonists were added as 20× solutions in 5 μl of PBS. Twelve hours subsequently the media were aspirated. After removal of any cellular material by centrifugation, media were stored at −20°C until assayed. IL-8 immunoreactivity was determined using specific polyclonal sheep anti-IL-8 antisera (generated and donated by Dr. S. Poole, National Institute for Biological Standards and Control, Potter’s Bar, U.K.).
Results
TLR1 is a cell surface protein expressed by monocytes
To define which subpopulations in human peripheral blood leukocytes express TLR1, monoclonal antibodies were generated against recombinant soluble TLR1 extracellular domain (TLR1-Fc). TLR1Fc was produced in a murine myeloma line as a fusion to the human IgG1 Fc region. The protein migrates as two major bands on nonreducing SDS-PAGE gels (Fig. 1⇓A) and one smaller band on reducing gels (Fig. 1⇓B). This may be due to different patterns of disulfide bonding. Sequencing of the band migrating faster on nonreducing gels showed the amino terminus of the recombinant protein to be SEFLVDRSKN. This corresponds to aa 25 of the predicted propeptide and is three amino acids from the predicted signalase cleavage site (SigCleave, Genetics Computer Group, Madison, WI). TLR1Fc was used as an immunogen to generate mAb against TLR1; Ab GD2.F4 does not recognize human IgG1, but recognizes both TLR1Fc forms (data not shown). It specifically affinity purifies both forms of TLR1Fc from TLR1Fc containing tissue culture medium (data not shown). It recognizes HeLa cells transfected with TLR1 expression vectors (Fig. 2⇓A), but does not bind to mock transfected HeLa, a cell line that lacks TLR1 mRNA (see below). Higher surface GD2 epitope expression is seen in cells transfected with the deletion construct pCMV-TLR1Δcyt than with the full-length construct pCMV-TLR1 (Fig. 2⇓A). No binding of isotype control Ab to transfected HeLa cells was observed (data not shown). The specificity of GD2.F4 was tested by staining cells transfected with other TLRs. It does not recognize TLR2, TLR4, or TLR4 and MD2 (Fig. 2⇓B) (Table II⇓).
TLR1Fc analysis by SDS-PAGE. Purified TLR1Fc fusion protein was analyzed on 7.5% SDS-PAGE gels following adjustment to 1× Laemmli buffer without (A) and with (B) reduction with 2-ME. BSA and IgG are included as controls.
Recognition of TLR1 by mAb GD2.F4. A, HeLa cells transfected with expression vectors pcDNA3, pCMV-TLR1, and pCMV-TLR1Δcyt. Single-cell suspensions were stained with mAb GD2.F4 or isotype control at 5 μg/ml, followed by secondary reagents. No binding of isotype control to any transfected population was observed. B, HeLa cells transfected with expression vectors pFLAG-CMV, pFLAG-TLR2, pFLAG-TLR4, and pFLAG-MD2. Single-cell suspensions were stained with mAb GD2.F4 or isotype control at 5 μg/ml followed by secondary reagents.
TLR1 mRNA is expressed in PBMC (5); hence, TLR1 expression on leukocyte populations was studied using mAb GD2.F4. Fig. 3⇓ shows that all cells in a monocyte population, purified from the blood of healthy volunteers by elutriation, express TLR1. A more detailed analysis will be reported elsewhere (A. Visintin et al., manuscript in preparation).
Expression of TLR1 on monocytes. A mononuclear fraction from a healthy donor was purified by centrifugal elutiation and stained with 15 μg/ml GD2.F4 or isotype control in the presence of 100 μg/ml normal human IgG. Binding was detected by anti-mouse FITC-conjugated secondary Ab, and the monocyte population was identified by forward and side scatter profiles.
TLR1 modifies TLR2-dependent responses
The effect of TLR1 on responses to bacteria was investigated. The cell line HeLa was used, because this cell line does not express detectable TLR1 or TLR2 messages (Fig. 4⇓). Although TLR4 message is detectable, message for MD-2, a protein required for TLR4 function (26), is not (Fig. 4⇓). We used luciferase controlled by an IL-8 promoter fragment as a reporter. This reporter responds to LPS in RAW264.7 cells (data not shown). The cell culture medium used included 10% FBS, a source of LPS-binding protein (34). We consistently observed, as previously reported (11, 35), that cells transfected with TLR1 alone did not respond to E. coli or N. meningitidis bacteria (data not shown). However, Fig. 5⇓A shows that in cells transfected with TLR1 and TLR2, induction of the IL-8 promoter was elicited by soluble factors released by a growing wild-type H44/76 organism. Efficient induction depended on transfection of both TLR1 and TLR2 (Fig. 5⇓B).
TLR message expression in HeLa cells and PBMC. cDNA from PBMC or HeLa cells was used as templates in PCR amplifying various TLRs and related molecules. L41 is a control housekeeping gene encoding a ribosomal protein. Retrotranscriptase (RT) negative samples are included as controls for genomic contamination.
TLR1/2 responds to lpxA-dependent components of N. meningitidis. HeLa cells cotransfected with TLR1 and TLR2 respond to soluble components released from N. meningitidis strain H44/76, but not from the isogenic mutant pLAK33, which lacks LPS due to a targeted deletion of the lpxA gene (A). Responses to these soluble products and to an LPS preparation derived from H44/76, are dependent on both TLR1 and TLR2 (B). A, Means of duplicate wells from one of two similar experiments. B, Mean ± SD of three transfected HeLa cell pools.
Dependency of TLR2 stimulation on N. meningitidis LpxA gene
Cells transfected with TLR2 have been reported to respond to LPS preparations (11, 24). LPS is released by growing N. meningitidis (36); thus, the structure recognized by TLR2 and TLR1 in the Neisseria supernatants might be LPS. Therefore, the stimulatory effect of soluble factors released from mid-log phase wild-type N. meningitidis strain H44/76 and from its LPS-free isogenic mutant pLAK33 (31), prepared as described in Materials and Methods, were compared. An LPS preparation, prepared by phenol-water extraction from strain H44/76, was also studied. Fig. 5⇑B shows that induction of the IL8 gene depended on transfection of both TLR1 and TLR2 and on the presence of N. meningitidis LPS. Compatible with results from observations with other LPS preparations used in this study (below), none of the Neisseria preparations tested caused IL-8 promoter induction in cells transfected with TLR1 alone (data not shown).
TLR1/2 agonist activity resides in only some LPS preparations
The influence of TLR1/2 cotransfection on responses of HeLa cells to other preparations of LPS was investigated, because copurification of other bacterial products with LPS can contribute significantly to bioactivity of LPS preparations (37). N. meningitidis contains LPS of the rough type (38); signaling to another rough LPS preparation, from S. minnesota Re595, was also dependent on both TLR1 and TLR2. The increases in IL-8 promoter activity induced by S. minnesota Re595 LPS preparation in four independent experiments are summarized in Table III⇓; a similar pattern of responses was observed when IL-8 production was measured directly by ELISA (Table III⇓), confirming that the reporter and the genomic IL-8 gene behave similarly. By contrast, the responses to two smooth LPS preparations were not affected by TLR1 (Fig. 6⇓). Likewise, responses to a highly repurified E. coli K235 LPS preparation (a gift from Dr. S. Vogel) did not depend on TLR1. Note that weak stimulation via TLR2 was observed at high doses in this system (Fig. 7⇓). By contrast, when HeLa cells were transfected with an expression vector producing MD-2, a component of the TLR4 receptor complex that is not expressed in HeLa cells (Fig. 4⇑), the repurified LPS produced a strong response (Fig. 7⇓). The effect of TLR1 cotransfection was not due to alteration in cell surface expression of TLR2 (Table II⇓), nor was it bypassed by increasing the TLR2 plasmid dose (data not shown). The TLR1/2 cooperative effect did not depend on whether the Re595 stock solution was dissolved in PBS or in PBS/0.2% triethylamine as previously described (39) (data not shown).
Signaling by Salmonella minnesota Re595 LPS preparation is influenced by both TLR1 and TLR2. HeLa cells were transfected with empty vector, TLR1, TLR2, or TLR1 plus TLR2, and reporter plasmids. Stimulation with various doses S. minnesota Re595 LPS preparation (A), S. minnesota wild-type LPS preparation (B), or LPS prepared from E. coli O55:B5 (C) was followed by reporter measurement. Shown are the mean ± SD of normalized IL-8R activity 12 h after stimulation obtained from three triplicate wells from a representative experiment of two to four similar assays.
TLR stimulation by highly purified LPS is TLR1 independent. HeLa cells were transfected with vector, TLR1, TLR2, TLR1 and 2, or MD2, an obligate component of the TLR4 signaling complex not expressed by HeLa cells. A, Cell stimulation by preparations containing 10 μg/ml S. minnesota Re595 LPS or repurified protein free E. coli K235 LPS. B, Concentration relationship of cell stimulation by repurified E. coli K235 LPS of transfected cells. Cell stimulation is the mean ± SD of normalized IL-8R activity obtained from three triplicate wells from a representative experiment of two similar assays.
Effect of 10 μg/ml S. minnesota Re595 LPS on transfected HeLa cellsa
Cotransfection of TLR1 and TLR2 does not increase TLR2 expression levels
Full-length TLR1 is required for TLR2-dependent responses to Re595 LPS preparations
To investigate the mechanism of the observed TLR1-TLR2 cooperation, TLR2 was cotransfected with vectors directing expression of TLR1, TLR1Δcyt, or the accessory chain of the IL-1R (IL-1RAcP) (27), which contains a TIR signaling domain related to those of TLRs (3). We reasoned that if the increase in signaling was a nonspecific effect due to increased expression of TIR domains, then IL-1RAcP might substitute for TLR1. Flow cytometry showed IL-1RAcP to be expressed at the cell surface in HeLa (data not shown). Despite cell surface expression of both TLR1Δcyt (Fig. 2⇑) and IL-1RAcP (data not shown), neither could substitute for full-length TLR1 in the cooperativity effect (Fig. 8⇓). Thus, the intracellular domain of TLR1 is required for cooperativity with TLR2.
TLR1/2 cooperation in Re595 LPS signaling requires full-length TLR1. HeLa cells were transfected with combinations of cell surface molecules with TIR signaling domains and were stimulated with S. minnesota Re595 LPS (10 μg/ml; A). Shown are the mean ± SD of normalized IL-8R activity obtained from three triplicate wells from a representative experiment of two similar assays.
Discussion
In this report it is shown that the pattern of microbial responsiveness in cells cotransfected with TLR1 and TLR2 differs from that in cells transfected with either molecule alone. This is the first demonstration of TLR1-dependent signaling and also suggests that combinations of TLRs can recognize agonists not effective with individual TLRs. Molecular studies of IL-1R family members, the signaling domain of which are closely related to those of TLRs, suggests that agonist-driven heterodimerization is a the mechanism for signal activation in TIR domain-containing proteins (40). In the case of the receptors for IL-1 and IL-18, both chains of a heterodimer are known to be required for cytokine signaling (22, 23). A similar mechanism in the TLR family seems plausible given the results obtained in this study and the close sequence homology (41) between TIR domains of IL-1R family members and TLRs. A recent study of RP105, a transmembrane molecule with extracellular domain similarity to TLRs (42), demonstrated functional cooperation between RP105 and TLR4 on cotransfection of these molecules into Ba/F3 cells. Thus, comparison with closely related receptor systems suggests that TLR1-TLR2 heterodimerization is the most likely explanation for the phenomenon described here, although we do not show whether this occurs directly or indirectly. The results presented in this work together with the recent demonstration by Ogata and colleagues that both TLR4 and RP105 contribute to B cell LPS recognition in vivo (42) support a combinatorial model of TLR function. We would speculate that deployment of a limited number of TLRs in a combinatorial fashion might be a means of diversification of the repertoire of the innate microbe detection system of mammals. Further, we have recently found by BLAST searching that there are about 15–20 TLRs in the draft and finished sequence of the human genome on current release (∼66% complete), suggesting that there may be 25–30 TLRs in the complete genome. Even if combinatorial function is limited to dimerization, this would generate 500–1,000 different receptors if any pair is functional. Further, similar database searching shows that there are least six MDs (MD-1, MD-2, and four others) in the human genome. If, as current data suggest, these are sensitizers/modifiers for TLR function, then a further diversification into the 1–10,000 range is potentially possible.
This report leaves several issues unresolved. The first is whether TLR2 is the physiological partner of TLR1. However, Muzio and colleagues have recently shown that TLR1 mRNA, or a closely related sequence, is expressed on all leukocyte subsets analyzed, as judged by Northern blotting (43). We demonstrate TLR1 monocyte surface expression in this work; TLR2 expression by monocytes has previously been reported (10). Therefore, TLR1 and TLR2 are coexpressed in normal human monocytes. However, we cannot exclude the possibility that, as occurs with components of several multichain cytokine receptor families (18, 44), TLR1 interacts functionally with other molecules as well. Study of the phenotype of mice with targeted deletions of TLR1 will help resolve this issue.
Abundant data are now available to show that in vivo responses to LPS depend on TLR4, not TLR2. This has been shown by genetic means (7, 45), by Ab blockade of TLR4 (9), and by demonstration that species-specific variation in LPS structures recognition can be reproduced by transfection of cell lines with TLR4 clones from that species (12). By contrast, LPS responses of TLR2 knockout animals appear normal (6). Rather, TLR2 responds to synthetic lipoproteins in a stereospecific fashion (46), and lipoprotein effects on cell lines can be blocked by mAbs to TLR2 (47). TLR2 responses to several other bacterial preparations of less defined composition have also been described (35, 48, 49, 50). Therefore, a second unresolved issue concerns the physiological nature of the TLR1/2 agonist. Although we show that such an entity exists in some LPS preparations and is secreted by N. meningitidis in an LpxA-dependent fashion, the precise nature of the agonist is not known. Previous work has shown that LPS preparations from rough Salmonella mutants include proteins, and perhaps other cell wall constituents, that remain tightly associated with LPS during conventional phenol extraction (37). These factors act via TLR4-independent receptor systems (33) and contribute to the bioactivity of the crude LPS preparation, acting synergistically with the LPS, which is absolutely dependent on TLR4 for its effect
The presence of the TLR1/2 agonist in only some LPS preparations suggests that it is unlikely to be LPS per se, a conclusion supported by the weak activity of a repurified LPS preparation in the TLR2-transfected or TLR1- and TLR2-transfected cells and compatible with the demonstration of normal LPS responsiveness in TLR2 knockout animals (6, 51). Genetic analysis of genes required for secretion of TLR1/2 agonist by N. meningitidis showed requirement for LpxA, a gene that encodes part of the LPS biosynthesis pathway of Gram-negative organisms (52). This suggests that the TLR1/2 agonist is metabolically related to LPS or requires LPS for efficient expression or secretion, as would be the case if the TLR1/2 agonist were a lipoprotein attached to LPS in a complex. There is evidence that such complexes exist and that LPS and the other associated component(s) act synergistically to activate macrophages (33, 37). Our data show that HeLa cells express TLR4, but not MD-2, and therefore do not respond to LPS alone (Figs. 4⇑ and 7⇑), but become highly LPS responsive when transfected with MD2 and CD14, but not with CD14 alone (Fig. 7⇑). Therefore, it is possible that when HeLa cells are transfected with TLR1 and TLR2 (+CD14), either TLR1/TLR2 form a receptor that responds to such a complex agonist or, alternatively, TLR1/TLR2 mediate responses to a specific agonist and a separate LPS signal (subthreshold by itself, without MD-2) is mediated by TLR4, both being required for detectable effects in our assays. Detailed biochemical studies and a greater understanding of the extent of functional cooperation between different TLR family members (42) will be required to resolve these issues. Finally, given the findings alluded to earlier that up to 30 TLRs and six MDs may be encoded in the human genome, some caution must be exercised in interpreting the data on TLR function published to date, because any cell line or population may express multiple TLRs and MDs not yet accounted for that contribute to the observed pattern of responses. Further, interpretation of studies on knockout and mutant mice become more complex if functional responses to pathogens and their products are mediated by oligomeric structures containing two or more TLRs and/or MDs.
Acknowledgments
We thank all our colleagues in the Division of Molecular and Genetic Medicine for advice, R. C. Read and A. Pridmore for Neisseria culture, L. Steeghs and P. van der Ley for the N. meningitidis mutant, Dr. A. Moir for protein sequencing, Dr. S. Poole for ELISA reagents, and Dr. M. Muzio, Dr. F. Volpe, and Professor B. Seed for expression vectors. Dr. S. Vogel generously provided highly purified LPS and disclosed research results before publication.
Footnotes
-
↵1 This work was supported by a Wellcome Trust Training Fellowship for Medical Graduates (to D.W.), a Vacation Studentship (to S.B.) and a Medical Research Council project grant (to E.K.-T.). A.V. is a Fogarty Fellow.
-
↵2 Address correspondence and reprint requests to Prof. Steven Dower, Functional Genomics Group, Division of Molecular and Genetic Medicine, University of Sheffield, Royal Hallamshire Hospital, M Floor, Sheffield, U.K. S10 2JF. E-mail address: s.dower{at}sheffield.ac.uk
-
3 Abbreviations used in this paper: TLR, Toll-like receptor; TIR, Toll and IL-1 receptor related.
- Received January 5, 2000.
- Accepted September 21, 2000.
- Copyright © 2000 by The American Association of Immunologists
References
In this issue
