The agent of Lyme disease, Borrelia burgdorferi, produces membrane lipoproteins possessing potent inflammatory properties linked to disease pathology. The recent association of toll-like receptors (TLR) 2 and 4 with LPS responses prompted the examination of TLR involvement in lipoprotein signaling. The ability of human cell lines to respond to lipoproteins was correlated with the expression of TLR2. Transfection of TLR2 into cell lines conferred responsiveness to lipoproteins, lipopeptides, and sonicated B. burgdorferi, as measured by nuclear translocation of NF-κB and cytokine production. The physiological importance of this interaction was demonstrated by the 10-fold greater sensitivity of TLR2-transfected cells to lipoproteins than LPS. Futhermore, TLR2-dependant signaling by lipoproteins was facilitated by CD14. These data indicate that TLR2 facilitates the inflammatory events associated with Lyme arthritis. In addition, the widespread expression of lipoproteins by other bacterial species suggests that this interaction may have broad implications in microbial inflammation and pathogenesis.
Lyme disease is caused by infection with the tick-borne spirochete Borrelia burgdorferi (1). A subacute inflammatory arthritis develops in 60% of individuals not treated at the time of the tick bite (2), and is associated with invasion of the joint tissue by spirochetes (3, 4). B. burgdorferi possess surface-associated proteins with the tripalmitoyl-S-glyceryl-cysteine (Pam3Cys)3 modification common to many bacterial species (5), and numerous inflammatory activities have been ascribed to the Pam3Cys modification (6, 7, 8, 9). Unique to B. burgdorferi is the presence of 130 genes, accounting for >10% of the genome, that possess a sequence encoding the signal peptidase II cleavage site necessary for synthesis of the lipid modification (10). Expression of lipoprotein genes is environmentally regulated (11), with the outer surface protein (Osp)A lipoprotein expressed primarily in the tick mid-gut and late in infection of humans (11, 12, 13). Serological evidence indicates that other B. burgdorferi lipoproteins are preferentially expressed during infection of mammals (11, 13), and it is likely that they possess similar stimulatory properties (6, 7, 8, 9).
The inflammatory activities attributed to B. burgdorferi lipoproteins include the ability to directly induce NF-κB nuclear translocation, resulting in cytokine production, adhesion molecule expression, and generation of nitric oxide and superoxide (6, 7, 8, 9). Lipoproteins or their derivatives have been shown to activate endothelial cells, neutrophils, macrophages, and B lymphocytes in vitro, and to induce localized inflammatory infiltrate into knee joints and dermal sites in vivo (6, 7, 8, 9, 14, 15, 16). B. burgdorferi does not produce LPS (17), nor does its genome encode the enzymes required for LPS synthesis (10). Thus, lipoproteins provide the major inflammatory stimulus associated with chronic infection. The existence of a specific receptor for the highly inflammatory lipoproteins has been hypothesized, and would provide a mechanism by which the localized bacteria could directly activate inflammation.
Bacterial lipoproteins and LPS share many characteristics, including a biologically active lipid modification, the cell types that are responsive, and the types of responses that are induced (18). These observations argue that a similar molecule may be involved in signaling, and is supported by the finding that the LPS coreceptor CD14 also facilitates lipoprotein signaling in several cell types (19, 20, 21). Toll-like receptor (TLR)2 and TLR4 are two candidate proteins that have recently been implicated in LPS signaling (22, 23, 24, 25, 26, 27). TLR4 has been identified by positional cloning as the gene responsible for LPS hyporesponsiveness in C3H/HeJ mice (24, 25, 26). These mice respond normally to lipoproteins (7, 14), arguing that TLR4 is not required for lipoprotein signaling. Transfection of either TLR2 or TLR4 confers responsiveness to LPS in cell lines (22, 23, 27). In this study, the possibility that members of the toll-receptor-like family are involved in signaling by bacterial lipoproteins was investigated.
Materials and Methods
Cell lines and reagents
U373 cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA). 293 cells were a subclone from Tularik (South San Francisco, CA). The TLR, endothelial cell-leukocyte adhesion molecule (ELAM-1) luciferase, and Rous sarcoma virus (RSV)-β-galactosidase DNA constructs are described by Kirschning et al. (22). Polymyxin B, LPS from Salmonella typhimurium, and human serum were obtained from Sigma (St. Louis, MO). Pam3Cys-Ser-Lys4-OH was obtained from Boehringer Mannheim (Indianapolis, IN). Sonicated B. burgdorferi and OspA purified from B. burgdorferi were isolated as described (14). Recombinant OspA, OspB, and unlipidated OspA were provided by John Dunn (Brookhaven National Laboratories, Upton, NY) (28). The recombinant and native OspA and sonicated B. burgdorferi contained <0.3 endotoxin units (EU) per 500 ng, as determined by Limulus amebocyte assay (Associates of Cape Cod, Woods Hole, MA). The 60bca cell line was obtained from ATCC, and mopc 21 Ab was from Sigma.
Quantification of TLR cDNA
mRNA and cDNA were prepared (29) from U373 cells cultured in DMEM with serum replacement Nutridoma-HU (Boehringer Mannheim). Transcript levels were quantified using the Light Cycler (Idaho Technology, Idaho Falls, ID) and normalized to hypoxanthine phosphoribosyl transferase (HPRT). This technique continuously monitors the accumulation of fluorescently labeled product to calculate starting transcript numbers and is described in detail by Morrison et al. (29). The oligonucleotide primers used to quantify HPRT were CTGGCGTCGTGATTAGTG and CCAACACTTCGTGGGGTC; TLR1, CTATACACCAAGTTGTCAGC and GTCTCCAACTCAGTAAGGTG; TLR2, TCACCTACATTAGCAACAG and GATCTGAAGCATCAATCTC; TLR3, GATCTGTCTCATAATGGCTTG and GACAGATTCCGAATGCTTGTG; TLR4, GCTTACTTTCACTTCCAAC and GCCATTGAAGATGCCATTG; and TLR5, CTAGCTCCTAATCCTGATG and CCATGTGAAGTCTTTGCTGC.
Luciferase and cytokine assays
IL-6 (U373) and IL-8 (293) production was measured by ELISA using matched-pair Abs (Endogen, Woburn, MA). 293 cells were then lysed using reporter lysis buffer (Promega, Madison, WI), and 20 μl of lysate was assayed for luciferase and β-galactosidase activity using a Dynatec MLX luminometer after incubation in luciferase assay reagent (Promega) or Galacto-Light with light emission accelerator (Tropix, Bedford, MA), respectively.
The human glioma cell line U373 responds vigorously to LPS, but not to OspA, as assayed by IL-6 production (Fig. 1⇓B). In addition, U373 cells produce detectable levels of transcript for TLR1, TLR3, TLR4, and TLR5 (Fig. 1⇓A). No transcript was detected for TLR2 in U373 cells (Fig. 1⇓A). Interestingly, the human monocytic line THP-1 was OspA-responsive and possessed very high levels of TLR2 transcript (data not shown), while the nonresponsive U373 cells produced none. These findings suggest a correlation between expression of TLR2 and the ability of cell lines to respond to the B. burgdorferi lipoprotein OspA.
To test whether the lipoprotein unresponsive phenotype could be corrected in U373 cells, expression constructs encoding four of the TLR proteins were transfected individually into this cell line, and the ability to respond to lipoproteins was measured by IL-6 production. Transfection of TLR2 conferred responsiveness to OspA in U373 cells (Fig. 1⇑B), whereas transfection of vector alone, TLR1, TLR3, or TLR4 did not. TLR4 did significantly increase IL-6 production of U373 cells when stimulated with LPS (Fig. 1⇑B), a result that agrees with previous reports suggesting that TLR4 is involved in LPS signaling (24, 25, 26, 27).
To test this effect in a second TLR2-negative cell line (22), expression constructs encoding the four TLR proteins were transiently transfected into the human embryonic kidney line 293. The TLR expression plasmids were cotransfected with a NF-κB-dependent luciferase reporter plasmid that contained the E-selectin (ELAM-1) promoter. To normalize for transfection efficiency, an RSV-β-galactocidase control plasmid was also cotransfected. Only transfection of the TLR2 expression construct conferred responsiveness to OspA in 293 cells (Fig. 2⇓, A and B) and was reflected in both luciferase and endogenous IL-8 production. As reported previously, TLR2 also conferred responsiveness to LPS (22, 23). Because TLR2 conferred responsiveness to both lipoproteins and LPS, polymyxin B was added to all lipoprotein samples to eliminate stimulation due to possible LPS contamination during experimental manipulation. Transfection of TLR4 resulted in constitutive activation, as reported by others in this cell type (22), and was not up-regulated by either LPS or OspA. None of the other constructs, pFLAG vector alone, TLR1, or TLR3, conferred responsiveness to either agonist (Fig. 2⇓, A and B).
The experiments in Figs. 1⇑, 2⇑A, and 2B used recombinant lipidated OspA, which was previously shown to possess the same stimulatory properties as OspA purified from B. burgdorferi (30). This was confirmed in Fig. 2⇑C, where OspA purified from B. burgdorferi was found to activate NF-κB in 293 cells transfected with TLR2. Sonicated B. burgdorferi, an abundant source of lipoproteins, was also a strong stimulant in TLR2-expressing cells. A second recombinant lipoprotein, OspB, which possesses the same lipid modification on a distinct polypeptide, also strongly activated TLR2-expressing cells. The fact that the synthetic lipopeptide Pam3Cys-Ser-Lys4-OH also possesses this activity further supports the requirement for the lipid modification. These findings indicate that several distinct Pam3Cys-containing molecules induce inflammatory signaling through TLR2, and suggest that the interaction is dependent on the common lipid modification. This specificity was confirmed with unlipidated OspA, a recombinant protein that possesses the same amino acid sequence as lipidated OspA, but lacks the amino-terminal cysteine required for the Pam3Cys modification (31). The unlipidated OspA did not activate either NF-κB translocation (Fig. 2⇑C) or IL-8 production (data not shown) in TLR2-expressing 293 cells. These results indicate that the lipid modification is required for cellular activation via TLR2, and that several lipoproteins and a synthetic lipopeptide possess this activity.
Further evidence that lipoproteins are a biologically important ligand for TLR2 was sought by comparing the dose required for NF-κB translocation and IL-8 production by LPS and OspA. Human neutrophils have been found to respond to 10-fold lower molar concentrations of LPS than OspA when adequate amounts of LPS binding protein and CD14 are present (9). In contrast, TLR2-transfected 293 cells responded to at least 10-fold lower concentrations of OspA than LPS (Fig. 3⇓, A and B). The greater sensitivity of TLR2-expressing cells to OspA than LPS strongly argues that TLR2 is important in mediating inflammatory signaling by lipoproteins in humans.
The finding that TLR2-negative U373 cells are responsive to LPS but not lipoproteins suggests that endogenously expressed TLR2 mediates inflammatory signaling by bacterial lipoproteins. The extremely high expression of TLR2 obtained in transiently transfected 293 cells may promote responsiveness to LPS that is not observed under physiological conditions. When the amount of TLR2 construct transfected into 293 cells was reduced, responsiveness to both LPS and OspA were conferred by transfection with as little as 0.03 μg of TLR2-expressing plasmid (Fig. 4⇓). Very low concentrations of plasmid conferred maximal responsiveness in 293 cells, and luciferase units remained high at the highest concentrations of plasmid, arguing that the difference in dose response seen between LPS and OspA did not reflect differential saturation of cofactors by either agonist. This finding further suggests that the dependence of LPS responsiveness on TLR2 expression in transfected 293 cells and its independence of TLR2 in U373 cells may reflect the influence of other essential components of the signaling pathway present in these different cell types.
OspA activation of human endothelial cells, neutrophils, monocytes, and mouse monocytes has been shown to be facilitated by the LPS coreceptor CD14 (20, 21). Although 293 cells do not express CD14 on their surface (22), human serum present in the culture medium provides soluble CD14. OspA activation of 293 cells transfected with the TLR2 expression plasmid was diminished in the presence of the CD14-blocking mAb, 60bca, when compared with an isotype control (Fig. 5⇓). The diminished activity in the presence of blocking Ab is consistent with the 10- to 20-fold inhibition previously reported in endothelial cells, monoctyes, and neutrophils (20, 21). These findings indicate that TLR2-mediated lipoprotein signaling can be enhanced by soluble CD14 in a manner consistent with that exhibited by cell types naturally responsive to bacterial lipoproteins.
Innate recognition of microbial products forms the initial defense against pathogen infection. In Drosophila, activation of Toll controls the induction of the antifungal peptide drosomycin in response to fungal infection (32), while two of the mammalian homologues, TLR2 and TLR4, have been associated with LPS responses. Our results demonstrate that TLR2 is required for bacterial lipoprotein signaling in two different TLR2-negative cell lines, U373 and 293, and that signaling requires the lipid modification. Although both LPS and lipoproteins can induce signaling through TLR2, transfected 293 cells responded to at least 10-fold lower concentrations of lipoprotein than LPS, supporting the physiological relevance of the lipoprotein interaction. Additionally, the defect in TLR4 that has been associated with the inability to C3H/HeJ mice to respond to LPS while responding normally to B. burgdorferi lipoproteins suggests that additional regulatory pathways are present in mammalian cells.
Other properties have been described for the Pam3Cys modification of bacterial lipoproteins, including that of immunological adjuvant (31). Animal studies of the OspA-based Lyme vaccine have demonstrated that its immungenicity is dependent on this lipid modification (31). The identification of TLR2 as a critical component of Pam3Cys signaling may provide a mechanism by which lipoproteins stimulate humoral immune responses without additional adjuvant.
Lyme arthritis, the most common manifestation of late Lyme disease, is associated with the presence of B. burgdorferi in the joint (3, 4). The abundantly expressed lipoproteins are intricately involved in the pathology of Lyme disease. These lipoproteins possess potent stimulatory properties for inflammatory cells that are associated with affected joints: namely neutrophils, mononuclear cells, and endothelial cells (6, 7, 8, 9, 15, 16). Thus, understanding the molecular basis of the signaling events caused by lipoproteins will lead to a greater understanding of the inflammatory events associated with Lyme arthritis in general. In addition, lipoproteins are broadly expressed by numerous pathogen species, many of which have been associated with inflammatory pathology (33, 34, 35). Thus, an understanding of lipoprotein-induced signaling events will have widespread implications in the understanding of the diverse pathologies caused by these microbial species.
We thank John Dunn for providing the recombinant OspA, OspB, and unlipidated OspA, and both Andy Hines and the members of the Weis labs for helpful discussions.
↵1 This work was supported by U.S. Public Health Service Grants AI-32223 and AI-43521 (to J.J.W.), AI-24158 (to J.H.W.), and 5P30-CA-42014 (to the University of Utah). The project described was also supported in part by an award from the American Lung Association (to J.H.W.) and an Arthritis Foundation Postdoctoral Fellowship (to R.M.W.).
↵2 Address correspondence and reprint requests to Dr. Janis J. Weis, Department of Pathology, University of Utah School of Medicine, 50 North Medical Drive, Salt Lake City, UT 84132. E-mail address:
↵3 Abbreviations used in this paper: Pam3Cys, tripalmitoyl-S-glyceryl-cysteine; TLR, toll-like receptor; RSV, Rous sarcoma virus; Osp, outer surface protein; ELAM-1, endothelial cell-leukocyte adhesion molecule (E-selectin); HPRT, hypoxanthine phosphoribosyl transferase.
- Received June 9, 1999.
- Accepted July 8, 1999.
- Copyright © 1999 by The American Association of Immunologists