Abstract
Toll-like receptor (TLR) 2 and TLR4 have been implicated in the responses of cells to LPS (endotoxin). CD14-transfected Chinese hamster ovary (CHO)-K1 fibroblasts (CHO/CD14) are exquisitely sensitive to endotoxin. Sequence analysis of CHO-TLR2, compared with human and mouse TLR2, revealed a single base pair deletion. This frameshift mutation resulted in an alternative stop codon, encoding a protein devoid of transmembrane and intracellular domains. CHO-TLR2 cDNA failed to enable LPS signaling upon transient transfection into human epithelial kidney 293 cells. Site-directed mutagenesis of CHO-TLR2 enabled expression of a presumed full-length hamster TLR2 that conferred LPS responsiveness in human epithelial kidney 293 cells. Genomic TLR2 DNA from primary hamster macrophages also contained the frameshift mutation found in CHO fibroblasts. Nevertheless, hamster peritoneal macrophages were found to respond normally to LPS, as evidenced by the induction of cytokines. These results imply that expression of TLR2 is sufficient but not essential for mammalian responses to endotoxin.
The responses of macrophages to bacterial products underlie the pathogenesis of many acute and chronic illnesses. One macrophage receptor that appears to play a central role in innate immune responses is CD14, a GPI-linked protein (1) that is also present as a functional soluble receptor (sCD14) in serum (2, 3). CD14 binds the lipid A portion of bacterial LPS and mediates a signaling cascade that results in the production of proinflammatory cytokines (4, 5, 6, 7). Despite the known function for CD14, this receptor has no transmembrane or cytoplasmic regions. Multiple lines of evidence suggested that CD14 is by itself incapable of transmembrane signal transduction (8, 9, 10).
Toll is a type I transmembrane receptor that was first identified in Drosophila melanogaster for its role in larval development (11). Toll activation in adult flies results in the activation of a NF-κB homologue known as dorsal and the subsequent production of antimicrobial peptides (reviewed in 12). At least five mammalian toll-like receptors (TLRs)3 have recently been described (13). The known role of Toll in a primitive immune system, as well as the homology with a mammalian proinflammatory receptor (14), made TLRs excellent candidates for a CD14-associated transmembrane signal transducer. Thus, when Yang (15) and Kirschning (16) independently observed that transfection of human TLR2 into human epithelial kidney (HEK) 293 cells rendered these otherwise LPS nonresponsive cells responsive to LPS, it appeared that the CD14-associated LPS signal transducer had finally been identified.
Almost simultaneously, an alternative TLR was proposed as a candidate for the CD14-associated signal transducer. Poltorak and colleagues mapped the Lps gene, which had been previously shown to be responsible for the LPS nonresponder phenotype in C3H/HeJ mice (17), to TLR4 (18). Sequence analysis of TLR4 from the C3H/HeJ mouse demonstrated a single point mutation at aa 712 (pro to his), which was proposed to change the function of the receptor (18) by suppressing TLR-mediated LPS-induced signals. The central role of TLR4 as the LPS signal transducer was strengthened by the observation that the LPS-resistant 10ScCr mouse was a null mutant for TLR4 (18, 19). However, to date, there is little functional data demonstrating how TLR4 might function as a receptor, nor whether it can function as a dominant negative suppressor, nor whether it signals cells by forming a complex with TLR2. Furthermore, the discovery of TLR2 has not been complemented by studies in genetically targeted mutant mice (knockout animals), an approach that often clarifies the in vivo significance of in vitro findings. Thus, the relative contribution of TLR2 to LPS responses seemed uncertain.
To define the LPS signal transduction pathway, we have focused on CD14-transfected Chinese hamster ovary (CHO)-K1 fibroblasts, which are exquisitely responsive to LPS (20) and relatively easy to manipulate genetically. Recently, we reported that we had successfully generated two mutant LPS nonresponder complementation groups derived from CHO/CD14 (21) and sought to determine whether either group contained mutations for TLR2 or TLR4. Surprisingly, wild-type CHO/CD14 cells did not contain a functional TLR2. Instead, sequencing revealed a frameshift mutation due to a dropped base in the N terminus, resulting in the introduction of a stop codon at aa 504. This mutation was not unique for CHO tumor cells, as LPS-responsive peritoneal macrophages from Chinese hamsters encoded the same mutant transcript, suggesting that hamsters do not express TLR2. We conclude that TLR2 is dispensable for LPS-initiated signal transduction.
Materials and Methods
Cell culture and Chinese hamster macrophages
Unless otherwise stated, all reagents were purchased from Sigma (St. Louis, MO). Protein-free ReLPS (Salmonella minnesota, strain R595) was a gift of Dr. N. Qureshi (Middleton Veterans Administration Hospital, Madison, WI). CHO/CD14 were engineered and grown as described (20). HEK 293 cells were maintained at Tularik, as described (22). Eight- to 10-wk-old female Chinese hamsters (Cytogen Research and Development, Boston, MA) were injected with sterile 3% thioglycollate solution. After 3 or 5 days, peritoneal exudate cells were collected by lavage with 10 ml RPMI 1640. After overnight incubation, adherent cells were stimulated with 10 ng LPS/ml in RPMI 1640 containing 10% FBS (Summit Biotechnology, Greeley, CO) for the indicated time points, and total RNA and genomic DNA were subsequently isolated (TriReagent; Molecular Research Center, Cincinnati, OH).
Cloning of CHO and mouse TLR2
A complementary DNA library was constructed from CHO/CD14 cells using the ZAP Express cDNA Gigapack III Gold cloning kit (Stratagene, La Jolla, CA). The RAW 264.7 cDNA library, generated with the same cloning kit, was a gift of Drs. N. Nagan and R.A. Zoeller (Boston University, Boston, MA).
The sequence of a mouse TLR2 expressed sequence tags clone (GenBank accession no. aa863729) was used to design PCR primers to generate a CHO-specific TLR2 PCR fragment of 169 bp from CHO cDNA. The screening of the cDNA libraries was performed according to the manufacturer’s instructions using either the 32P-labeled CHO-specific TLR2 DNA probe or the random-primed mouse EST clone. The GenBank accession numbers for CHO and mouse TLR2 are AF113614 and AF124741, respectively.
Expression plasmids
The human TLR2-Flag (pTLR2hu-F) and the pELAM-luc reporter plasmids have been previously described (16). Epitope-tag hamster TLR2 (pTLR2ham-F) was constructed by inserting PCR-generated full-length cDNA fragments lacking the N-terminal signal sequence (aa 1–18) into the mammalian expression vector pFLAG-CMV-1 (Sigma). Site-directed mutagenesis (QuikChange; Stratagene) was used to introduce an additional base A at position 1758 into the pTLR2ham-F plasmid (pTLR2ham-F/1758(+A)). The fidelity of the PCR, as well as the site-directed mutagenesis, was confirmed by sequence analysis.
NF–κB reporter assay
The NF–κB reporter assay in HEK 293 cells was performed as described (16). Each experiment was repeated at least twice.
RT-PCR
First, 1 μg of total RNA was reverse transcribed in a volume of 20 μl using Superscript II reverse transcriptase according to manufacturer’s protocol (Life Technologies, Grand Island, NY). Then, 2 μl of the resulting cDNA was used in a 25-μl PCR reaction as described (23). The PCR was conducted in an automatic thermal cycler (Hybaid, Franklin, MA) using primers for CHO-TLR2 (5′-GAGTGAGTGGTGCAAGTATGAAC and 5′-GGGCCACTCCAGGTAGGTCT), GAPDH (5′-GTCATCATCTCCGCCCCTTCTGC, 5′-GATGCCTGCTTCACCACCTTCTTG), multispecies IL-1β (5′-GCATCCAGCTTCAAATCTCACA and 5′-AACCGCTTTTCCATCTTCTTCT), hamster IL-6, (5′-TTGGGAAATTTGCCTACTGAA, 5′AGGCATGACTATTTTATCTGGA), and multispecies TNF-α, (5′-GGGGCCACCACGCTCTTCTG and 5′-GGCAGGGGCTCTTGACGGC).Control PCR with total RNA preparations that were not subjected to reverse transcription revealed no genomic DNA contamination of the isolated RNA preparations (data not shown).
Results and Discussion
The recognition of the Toll receptor system is an extraordinary representation of a phylogenetically conserved cellular host defense system. Recent studies suggest that mammalian homologues of Toll appear to mediate the responses to diverse types of bacterial cell wall products including those from Gram-positive bacteria4 (28), spirochetes and mycobacteria (E. Lien et al., manuscript in preparation). Given the current knowledge about TLRs, it is difficult to understand how two similar receptors, such as TLR2 and TLR4, could apparently serve the same function. Additional genetic and biochemical evidence was certainly necessary. The approach of engineering a targeted deletion of either TLR2 or TLR4 into a mouse line, and characterizing the effect of a targeted lesion, was a reasonable priority.
The data presented herein appear to answer the question of whether TLR2 is essential for LPS-induced signal transduction, without the need to examine a knockout animal. The sensitive nature of LPS responses in CHO/CD14 cells has been well documented (e.g., Refs. 9, 20, and 24). Therefore, it is a reasonable assumption that if TLR2 were necessary for LPS signal transduction, CHO cells would express this gene product. A clone from a CHO/CD14 cell cDNA library was isolated using a CHO-TLR2-specific PCR fragment as a probe. The clone contained a 2813-bp cDNA fragment with 85% and 75% sequence identity to mouse and human TLR2 cDNA, respectively. Surprisingly, the sequence analysis of CHO-TLR2, when compared with TLR2 from humans and mice, revealed a frameshift mutation due to a dropped base at position 1758. This new frame introduced a stop codon 31 bases downstream from the mutation. Comparison of the deducted CHO-TLR2 protein sequence showed 80% identity to the mouse protein and 68% to the human protein (Fig. 1⇓A).
Primary structure of Chinese hamster TLR2. A, Alignment of CHO, mouse (mo), and human (hu) TLR2 amino acid sequences. Identical residues are shaded. B, Sequence analysis of CHO and Chinese hamster TLR2 fragments, generated from different sources and using different polymerases. The location of the deletion is indicated by an arrow. The consensus sequence (Hamster TLR2) is compared with mouse and human TLR2. Identical amino acids are printed in bold. The protein sequence resulting from the frameshift is printed in italics. C, Graphic illustration of human and predicted hamster TLR2 primary structures.
To verify that the deletion at bp 1758 was not a peculiarity of the isolated cDNA clone, we used RT-PCR to sequence TLR2 transcripts derived from four unique CHO cell lines, including a CHO cell line obtained from another laboratory (provided by Drs. P. Tobias and R. Tapping, Scripps Research Institute, La Jolla, CA) using two different polymerases: pfu and Taq. In addition, we assessed the sequence of TLR2 transcripts from native Chinese hamster exudate macrophages. Finally, genomic DNA from Chinese hamsters was also used as a template for PCR and sequenced (Fig. 1⇑B). In every case, the sequence showed the same frameshift mutation, clearly indicating that Chinese hamsters express a TLR2 mRNA that encodes for a truncated protein without transmembrane and intracellular signaling domains (Fig. 1⇑C).
The sequence of hamster TLR2 indicated that the expressed protein was not likely to be a functional LPS receptor. We wanted to test this hypothesis and used a tissue culture system for analyzing LPS signaling components in HEK 293 cells, as described (16). Transient expression of a Flag-tagged human TLR2 expression plasmid (pTLR2hum-F) strongly activated a cotransfected NF–κB-dependent luciferase reporter gene construct when cells were exposed to 1 μg of LPS/ml (Fig. 2⇓). Similarly, transient transfection with a mouse TLR2 expression plasmid (pTLR2mo) followed by LPS exposure resulted in induced luciferase activity. In contrast, neither the hamster TLR2 (pTLR2ham) nor the Flag-tagged hamster TLR2 (pTLR2ham-F) constructs conferred responsiveness to LPS (Fig. 2⇓).
LPS-induces NF-κB-activation through mouse and human, but not hamster, TLR2. HEK 293 cells were transfected with 1 μg of the indicated TLR2 plasmids or a control plasmid (pRK) together with ELAM-1 luciferase reporter plasmid as described (16). After 24 hours, 1 μg of LPS/ml was added to monolayers. After a subsequent incubation of 6 h, induced luciferase activity was determined. White bars represent cells that were untreated, and solid bars are cells stimulated with LPS.
Sequence analysis of hamster-derived TLR2 predicted that insertion of the proper base pair into the cDNA would result in expression of a full-length TLR2 protein identical in length to mouse or human TLR2. We thus “repaired” the hamster TLR2 by adding a single base A at position 1758 through site-directed mutagenesis and tested the new construct (pTLR2ham-F/1758(+A)) in HEK 293 cells. The “repaired” construct now was able to confer responsiveness to LPS similar to human and mouse TLR2 (Fig. 3⇓).
“Repair” of hamster TLR2 restores activity as an LPS signal transducer. HEK 293 cells were cotransfected as in Fig. 2⇑ with epitope tagged hamster TLR2 (pTLR2ham-F), a “repaired” version of the cDNA (pTLR2ham-F/1758(+A)), which encodes for a full-length protein, or an empty vector together with pELAM-luc. After allowing 24 h for protein expression to occur, responses to a 6-h incubation in 1 μg of LPS/ml were determined by assessing induced luciferase activity. White bars represent cells that were untreated, and solid bars are cells stimulated with LPS.
In view of our observation that hamster TLR2 was nonfunctional and that the identical mutant mRNA was expressed in native hamster cells and CHO tumor cells, we investigated if TLR2-null peritoneal macrophages from Chinese hamster would show impaired responses when stimulated with LPS. Macrophages from mice typically respond to exposure to nanogram per milliliter concentrations of LPS by up-regulating the expression of cytokine genes (6). Therefore, hamster peritoneal macrophages were stimulated for 0, 1, or 4 h with 10 ng of LPS/ml. We then used RT-PCR to analyze the LPS-induced levels of IL-1β, IL-6, and TNF-α mRNA. Unstimulated cells expressed almost undetectable levels of IL-1β mRNA and low levels of IL-6 and TNF-α (Fig. 4⇓). Stimulation with LPS induced up-regulation of IL-1β, IL-6, and TNF-α mRNA, indicating that hamsters are in fact normally responsive to LPS compared with other mammalian species.
LPS induces cytokine gene expression in TLR2-null Chinese hamster peritoneal macrophages. Chinese hamster peritoneal macrophages were stimulated with 10 ng of LPS/ml for the indicated times. Expression of GAPDH (upper left), IL-1β (upper right), IL-6 (bottom left), and TNF-α (bottom right) was analyzed by RT-PCR.
One possible interpretation of the data presented here, in light of the findings of Poltorak et al. (18), is that different species of animals may use different Toll proteins to respond to LPS. The findings in LPS nonresponder mice strongly argue for an absolute requirement of TLR4 expression for mice to respond to LPS; we have also cloned TLR4 from the hamster, and it appears to be normal (H.H., unpublished observations), suggesting that this gene may in fact be used by hamsters as well for LPS signal transduction. As of today, no similar data on naturally occurring mutants is available to determine whether either TLR2 or TLR4 is essential for LPS responses in man, although an LPS nonresponder human patient has been reported (25). Several reports over the past decade have described species-specific effects of lipid A analogues. For example, the tetraacyl bis-phosphate precursor of lipid A, lipid IVa (also known as compound 406), is a potent LPS antagonist when tested with human macrophages and an LPS mimetic when tested with mouse or hamster macrophages (8, 26, 27). In contrast, Rhodobacter sphaeroides lipid A is an LPS antagonist in human and mouse cells, but not in hamster cells (9). This complex pharmacology of the LPS antagonists has been attributed to species-specific differences in the primary structure of the LPS receptor (9). If different species of animals use different Toll molecules as signal transducers, then the pharmacology of these analogues should be defined by the species of Toll protein that is predominantly expressed. TLRs 2 and 4 have now been cloned from human, mouse, and hamster. Additional experiments with these genes are necessary to reveal which TLR is truly dominant for LPS signal transduction.
We conclude that Chinese hamsters contain a nonfunctional version of TLR2 and that TLR2 does not appear to be necessary for LPS responses, though its expression imparts responsiveness to LPS in a CD14-dependent manner (16). How the existence of multiple LPS receptors impacts upon the biology of responses to LPS is unclear. However, with the identification of the TLRs as receptors for LPS and other bacterial products, the tools to define the signal transduction apparatus are clearly at hand, and we anticipate rapid progress in understanding LPS signaling in the near future. In this way, targeted therapy for sepsis and other LPS-mediated diseases can be designed.
Footnotes
-
↵1 D.T.G. and H.H. are supported by National Institutes of Health Grants GM54060 and AI38515. E.L. is supported by The Norwegian Cancer Society and The Research Council of Norway.
-
↵2 Address correspondence and reprint requests to Dr. Douglas Golenbock, The Maxwell Finland Laboratory for Infectious Diseases, 774 Albany Street, Boston, MA 02118, E-mail address: Douglas.Golenbock{at}bmc.org
-
3 Abbreviations used in this paper: TLR, toll-like receptor; CHO, Chinese hamster ovary fibroblast; HEK, human epithelial kidney.
-
4 R. Schwandner, R. Dziarski, H. Wesche, M. Rothe, C. J. Kirschning. Peptidoglycan- and lipoteichoic acid-induced cell activation is mediated by Toll-like receptor 2. Submitted for publication.
- Received March 25, 1999.
- Accepted April 20, 1999.
- Copyright © 1999 by The American Association of Immunologists
References
In this issue
