Perturbations to the well-being of tissues in plants and invertebrates generate fragments of endogenous molecules that are recognized by innate immune receptors. Vertebrates have homologous receptors on specialized cells such as dendritic cells, but whether these receptors respond to fragments of endogenous molecules is not known. We tested the idea that Toll-like receptors on dendritic cells might recognize polysaccharide fragments of heparan sulfate proteoglycan. Dendritic cells were found to mature in response to heparan sulfate as measured by costimulatory protein expression, morphology, and T lymphocyte stimulation, but this maturation was absent when Toll-like receptor 4 was mutated or inhibited. These findings suggest that Toll-like receptors in vertebrates may monitor tissue well-being by recognizing fragments of endogenous macromolecules.
Multicellular organisms have sentinel receptors that distinguish conditions of well-being from conditions of disease (1, 2). In vertebrate species these receptors compose a family of proteins that are homologous to Drosophila Toll and some plant resistance, or R proteins, and, accordingly, are called Toll-like receptors (TLR)3 (2, 3, 4). Vertebrate TLR recognize products of microorganisms such as LPS (3, 4, 5), intact proteins such as fibrinogen and heat-shock protein 60 (6, 7), and newly synthesized protein such as the extra domain A of fibronectin (8). In plants and Drosophila, these sentinel receptors recognize fragments of ubiquitous endogenous molecules generated in disease and tissue injury (2, 9); however, ubiquitous endogenous molecules, fragments of which are capable of stimulating vertebrate TLR, have not been identified. The disease resistance proteins in plants, the Toll receptors in Drosophila, and the TLR in vertebrates, when stimulated, trigger innate immunity (10). The response of vertebrates to TLR activation not only activates innate immunity but also activates lymphocytes which mediate “adaptive” immune responses (4).
While many types of cells may participate in vertebrate innate immune responses, adaptive immune responses are triggered predominantly by dendritic cells (DC) (11). Upon activation of TLR, DC briefly increase phagocytosis, migrate to lymph nodes (12), process and present phagocytosed Ags, and undergo a phenotypic maturation resulting in high expression of adhesive and costimulatory proteins, all of which help to activate naive T lymphocytes (13). Maturation of DC, and the subsequent activation of the adaptive immune system, must also occur in the absence of exogenous stimuli (14, 15), because T lymphocyte responses are vital to the defense against noncytopathic viruses, tumors, and transplants (16). Endogenous molecules generated at the very inception of tissue disease that are capable of activating TLR on DC, and thus leading to their maturation in the absence of exogenous stimuli, have not been identified.
Because injured or infected tissues in plants and invertebrates generate fragments of endogenous molecules that activate immunity through sentinel receptors, we questioned whether vertebrates use a similar molecular pathway to monitor tissue disease. In support of this, we recently found that fragments of heparan sulfate, an acidic polysaccharide ordinarily found in cell membranes and extracellular matrices (17), activates DC (18). Heparan sulfate is rapidly shed from cell surfaces and basement membranes as a result of tissue injury (19, 20), in the course of general inflammation (21, 22), and is also shed as a result of tumor cell migration and metastasis (23). In this work we report that soluble heparan sulfate uses TLR4 in the activation of DC, thus suggesting that vertebrate TLR monitor perturbations to the well-being of tissues. Therefore, plants, invertebrates, and highly evolved vertebrates maintain receptor systems that are capable of recognizing general signals of tissue disease manifest by fragmentation of endogenous molecules.
Materials and Methods
Reagents and Abs
Ultrapure heparan sulfate (Super Special Grade) and chondroitin sulfate B were obtained from Seikagaku (Falmouth, MA), and LPS derived from Escherichia coli. was obtained from Sigma-Aldrich (St. Louis, MO). CpG DNA, non-CpG DNA, and inhibitory CpG DNA were synthesized and phosphorothioate modified. The following FITC-conjugated mAbs were obtained from BD PharMingen (San Diego, CA): HM40-3 (anti-CD40), 16-10A1 (anti-CD80: B7-1), GL1 (anti-CD86: B7-2), G235-2356 (hamster IgG isotype standard), R35-95 (rat IgG isotype standard). The following biotin-conjugated Abs were obtained from BD PharMingen: AF6-88.5 (anti-H-2Kb) and AF6-120.1 (anti-I-Ab). PE-conjugated streptavidin and unconjugated 2.4G2 (anti-CD16/CD32) were also purchased from BD PharMingen. CD4 MicroBeads were purchased from Miltenyi Biotec (Auburn, CA). LPS removal resin (END-X B15) was from Associates of Cape Cod (Falmouth, MA). Recombinant mouse GM-CSF was purchased from R&D Systems (Minneapolis, MN). Rhodobacter sphaeroides diphosphoryl lipid A (Rs-DPLA) was a gift from N. Quereshi (University of Wisconsin, Madison, WI). Synthetic single-stranded oligonucleotides were dissolved in TE (10 mM Tris, 1 mM EDTA), purified by gel filtration chromatography, and quantitated spectrophotometrically.
Cell isolation and culture
DC were generated from murine bone marrow culture as previously described (18). Briefly, bone marrow was flushed from the long bones of C57BL10ScNCr (National Cancer Institute, Bethesda, MD), C57BL6J, BALBcJ, C57BL10SnJ, C3H/HeJ, or C3H/HeSnJ mice (The Jackson Laboratory, Bar Harbor, ME) and depleted of red cells with ammonium chloride. At day 3 of culture, floating cells were gently removed and fresh medium containing 3.3 ng/ml GM-CSF was added. At day 6 or 7 of culture, nonadherent cells and loosely adherent proliferating cell aggregates were harvested for analysis or stimulation. RAW 264.7 (ATCC no. TIB71; American Type Culture Collection, Manassas, VA) were maintained in culture as recommended by the supplier. Splenocytes from 8-wk-old BALB/cJ mice (H-2d) were purified by density centrifugation (Ficoll Paque Plus (Amersham Pharmacia Biotech, Uppsala, Sweden)), washed with PBS, and resuspended in RPMI 1640 supplemented with 10% heat-inactivated FCS, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Gaithersburg, MD), and incubated at 37°C, 5% CO2.
Stimulation of DC and RAW 264.7
Cells were stimulated with 10 μg/ml chondroitin sulfate, 150 ng/ml CpG DNA, 10 ng/ml LPS, and 10 μg/ml heparan sulfate or PBS in culture medium for 24 h, unless otherwise indicated. Rs-DPLA, or CpG inhibitory DNA, were added to cell culture as indicated 1 h before stimulation. In some experiments, stimulants were pretreated by 24-h incubation on a nutator with endotoxin removal resin, containing Limulus anti-LPS factor (LALF) covalently linked to silica beads, or were boiled for 10 min at 100°C.
Flow cytometric analysis
Flow cytometric analysis was performed as described by Kodaira et al. (25). Cells were incubated with anti-CD16/CD32 Ab, then stained with the indicated Abs and analyzed by FACScan using CellQuest software (BD Biosciences, San Jose, CA).
Phase microscopy images of cells in active culture were captured at ×100 on an inverted microscope.
Immature C3H/HeJ, C3H/HeSnJ (H-2k), or BALB/cJ (H-2d) DC were stimulated, then washed and resuspended at various concentrations in 100 μl/well of medium in 96-well round-bottom plates. Splenocytes from BALB/cJ mice (H-2d) were added at various concentrations to DC at 100 μl/well and incubated at 37°C, 5% CO2. After 3, 4, 5, and 6 days of coculture, 2 mCi/ml [3H]thymidine (ICN Pharmaceuticals, Irvine, CA) was added for 16 h. [3H]Thymidine incorporation into DNA was analyzed using a liquid scintillation counter (Wallac, Turku, Finland). Student’s t test was used for statistical analysis. In separate experiments, immature female C57BL/10SnJ and C57BL10/ScNCr DC were stimulated in the presence of multiple soluble Ags from bovine serum for 24 h, then washed three times and irradiated with 15 Gy. CD4+ cells were isolated from combined spleens and axillary lymph nodes of female C3H/HeSnJ by staining with CD4 MicroBeads and positively selected on a LS+ column using a VarioMACS instrument (Miltenyi Biotec). DC were cocultured at various concentrations with CD4+ Th cells at various concentrations, and thymidine incorporation was measured after 3, 4, 5, and 6 days as previously described.
NF-κB translocation analysis
Nuclear translocation of NF-κB was analyzed using an EMSA. DC or RAW 264.7 cells were stimulated with the indicated reagents and for the indicated times, then washed three times with ice-cold PBS. Cells were lysed by scraping and nuclear protein was extracted from washed nuclei as previously described (26).
EMSA was performed according to a modification of a previously described procedure (27). Nuclear extract (10 μg) was preincubated on ice in poly(dI-dC)-containing binding buffer (see below). Where indicated, cold competitive oligonucleotides were included during the preincubation period. 32P-labeled oligonucleotide probe containing two tandem NF-κB binding sites (28) was added (2.5 × 104 cpm or ∼2.5 fmol/reaction), and the reactions were incubated for 15 min at 25°C. The final binding reactions (20 μl) contained 12.5 mM HEPES, 87.5 mM NaCl, 1 mM DTT, 0.15 mM EDTA, 0.02% Nonidet P-40, 12.5% glycerol, and 100 μg/ml poly(dI-dC). The samples were electrophoresed through 4% polyacrylamide gels (25:1 acrylamide:bisacrylamide) in 45 mM Tris-borate buffer containing 1 mM EDTA (pH 8). Gels were dried under vacuum and radiolabeled species were detected by capture on a Kodak storage phosphor screen (Kodak, Rochester, NY) and revealed by a Bio-Rad molecular imager FX (Bio-Rad, Irvine, CA). Visualization and quantitative analysis were done using Bio-Rad Quantity One software.
Heparan sulfate and TLR agonists induce DC maturation
To explore the potential involvement of TLR in the maturation of DC, we used a model system in which DC are driven to mature by established agonists of TLR (29, 30). Fig. 1 shows that immature DC, grown from mouse bone marrow culture with GM-CSF for 6–7 days, are induced to mature by soluble heparan sulfate, by LPS, the active agent of bacterial endotoxin (31), and by single-stranded, nonmethylated DNA with a bacterial CpG motif (CpG DNA). Maturation of DC is manifest by a progressive increase in expression of cell surface molecules that aid in T lymphocyte stimulation—CD80 (B7-1), CD86 (B7-2), CD40, and MHC class II (Fig. 1 and data not shown)—consistent with findings established in previous reports (18, 29, 30). Limiting concentrations of heparan sulfate, LPS and CpG DNA induce maturation at the same rate and to the same extent (Fig. 1). Heparan sulfate, LPS, and CpG DNA (Fig. 1 D) induced the formation of cell aggregates, which are commonly seen as DC mature in culture (32). These results show that heparan sulfate, LPS, and CpG DNA induce similar maturation of DC.
Stimulation of DC is specific for fragments of heparan sulfate. DC maturation induced by heparan sulfate was not a “nonspecific” effect of a charged polyanion, as chondroitin sulfate, a saccharide with charge density similar to that of heparan sulfate (33), and heparin, a polysaccharide structurally related to heparan sulfate that contains a markedly different sulfation pattern (34), did not induce maturation of DC (Fig. 1 and data no shown). Although heparan sulfate used in these experiments was chromatography purified and deemed ultrapure by amino acid autoanalyzer and cellulose acetate strip electrophoresis, the absence of LPS was confirmed by the following experiments. First, adsorption of heparan sulfate with a LPS-binding protein, LALF, linked to silica beads that were then removed by centrifugation did not reduce, and in fact enhanced, the ability of heparan sulfate to induce DC maturation (Fig. 2). However, the same treatment of LPS eliminated its ability to induce DC maturation (18, 35 and Fig. 2). Heparan sulfate has little or no reactivity with LALF (36). Second, digestion of heparan sulfate by deaminative cleavage eliminated the stimulatory capacity of heparan sulfate but had no effect on DC maturation induced by LPS (18). Third, the inhibition and signaling kinetics of heparan sulfate differed markedly from that of LPS (see below). The absence of contaminating proteins was confirmed by the following experiment. Boiling of heparan sulfate, which inactivates the ability of proteins to stimulate TLR4 (6, 7, 8, 37), had no effect on DC maturation induced by heparan sulfate (Ref. 18 and data not shown).
Effect of TLR inhibitors on DC maturation induced by heparan sulfate
Because DC appear to respond to heparan sulfate in the same way as they do to LPS and CpG DNA, we asked whether heparan sulfate might use the same cellular receptors as these agonists. To address this question, we tested whether inhibitors of the receptors used by LPS and CpG DNA inhibit activation of DC by heparan sulfate. Rs-DPLA is a competitive inhibitor of LPS and other TLR4 agonists (5, 38, 39). As Fig. 3,A shows, DC incubated with Rs-DPLA did not mature in response to heparan sulfate. In contrast, DC incubated with a competitive inhibitor of CpG DNA4 did mature in the presence of heparan sulfate (Fig. 3 B). Taken together, these results suggest that the maturation of DC induced by heparan sulfate may require TLR4, but not TLR9, the receptor that recognizes CpG DNA (40).
The results given above do not exclude the possibility that Rs-DPLA inhibits the action of some other molecule in the pathway of heparan sulfate signaling. For example, LPS forms a soluble complex with LPS-binding protein (LBP), a serum protein that aids in transferring LPS to cellular surfaces (41). LPS also complexes with CD14, a protein found both on cell surfaces and in serum (41, 42). Both LBP and CD14 aid in low-dose LPS signaling (41, 42). Although Rs-DPLA inhibits interaction of LPS with TLR4 directly, interactions of LPS with LBP and CD14 are also competitively inhibited by Rs-DPLA (39, 43). Interactions between heparan sulfate, LBP, and CD14 were not explored in this study.
Because Rs-DPLA inhibits the action of multiple extracellular proteins that may specifically aid in LPS signaling through TLR4, one might expect Rs-DPLA to inhibit other TLR4 agonists at different doses. We found that Rs-DPLA is a more potent inhibitor of DC maturation induced by heparan sulfate, as compared with LPS (requiring a 6- to 7-fold lower dose of Rs-DPLA), despite the fact that we used limiting concentrations of heparan sulfate and LPS (Fig. 3,A). Rs-DPLA inhibition was not due to interaction with MyD88 or other intracellular signaling molecules that are shared by TLR (40), because Rs-DPLA did not inhibit CpG DNA-induced maturation (Fig. 3 A). These data are consistent with a model in which DC activation by heparan sulfate and DC activation by LPS both depend on TLR4 but differ with regard to the types of interactions with TLR4 and coreceptors.
Effect of TLR4 mutation on DC maturation induced by heparan sulfate
To determine whether DC maturation induced by heparan sulfate requires functionally active TLR4, we tested DC cultured from C3H/HeJ mice. C3H/HeJ mice have a single amino acid mutation (Pro712 to His712) in the conserved cytoplasmic Toll-IL-1R domain of TLR4 and do not respond to the presence of LPS (31). The mutation in the TLR4 gene of C3H/HeJ mice abrogates the interaction of TLR4 with MyD88, a cytoplasmic adaptor molecule shared by TLR and required for full responses (31, 44). Fig. 4,A shows that DC from C3H/HeJ mice do not mature in response to either heparan sulfate or LPS, whereas DC from TLR4 wild-type mice with the same genetic background mature normally. In contrast, C3H/HeJ DC do mature in response to CpG DNA (Fig. 4 A), showing that these mutant DC are capable of responding to stimulation via TLR9.
Effect of TLR4 deletion on DC maturation induced by heparan sulfate
Although the results given above suggest that functional TLR4 is required for DC to respond to heparan sulfate, the results do not exclude the possibility that C3H/HeJ mice may have another defect, such as a deficient production of IFN-γ (45), that may impair responses to heparan sulfate. To test whether a defect in TLR4, and not some other defect, abrogates maturation of DC induced by heparan sulfate, we tested DC from C57BL/10ScNCr mice. C57BL/10ScNCr and C57BL/10ScCr mice have a deletion in chromosome 4 that encompasses the TLR4 gene and, like C3H/HeJ mice, do not respond to the presence of LPS (31). C57BL/10ScCr have been used extensively to study the function of TLR4; however, C57BL/10ScCr mice have an additional mutation in their IL-12Rβ2 gene that is not found in C57BL/10ScNCr mice (46). Fig. 4,B shows that DC from C57BL/10ScNCr mice do not mature in response to either heparan sulfate or LPS, whereas DC from C57BL/10SnJ mice, a TLR4 wild-type strain with the same genetic background, mature normally. It should be noted that DC from C57BL/10SnJ mice appear to express higher basal levels of CD80, CD86, and CD40 than DC from C57BL/10ScNCr mice, but this does not effect the clear difference between the strains in response to stimulation with heparan sulfate or LPS. DC from C57BL/10SnJ mice also increase expression of MHC class II (I-Ab) and MHC class I (H-2Kb) in response to heparan sulfate, whereas DC from C57BL/10ScNCr mice do not (data not shown). Like DC from C3H/HeJ mice, DC from C57BL/10ScNCr mice do mature in response to CpG DNA (Fig. 4 B), indicating that DC from both strains of TLR4 mutant mice are capable of responding to stimulation via TLR9.
Effect of TLR4 mutation on activation of T lymphocytes by DC stimulated with heparan sulfate
We tested whether TLR4 is essential for the increase in T lymphocyte stimulation by DC induced to mature by heparan sulfate (18). To address this we used a modified MLR in which stimulated DC are mixed in culture with splenocytes from a different strain of mice. The cellular proliferation that ensues is measured as an indicator of T lymphocyte stimulation and acts as an in vitro model of activation of acquired immune responses that may lead to allogeneic transplant rejection. Fig. 5 A shows that DC cultured from C3H/HeJ mice do not increase their stimulatory capacity in response to either heparan sulfate or LPS, whereas DC from TLR4 wild-type mice increase their stimulatory capacity significantly in response to both heparan sulfate and LPS.
We next tested whether TLR4 is essential for adjuvant action of heparan sulfate on DC-mediated activation of CD4+ Th cells to soluble Ags. Toward this end, immature DC were stimulated in the presence of multiple soluble Ags from bovine serum (47), washed, irradiated, and combined with CD4+ T cells from syngeneic mice. The cellular proliferation that ensues acts as an in vitro model of activation of primary Th cell responses to infection, vaccination, and transplants carrying minor Ags. Fig. 5 B shows that DC expressing wild-type TLR4 stimulated with heparan sulfate and exposed to Ag caused T cell proliferation, whereas DC lacking TLR4 did not. These results show that TLR4 on DC is required for heparan sulfate to cause functional DC maturation.
Heparan sulfate induced nuclear translocation of NF-κB
Maturation of DC by TLR activation appears to require nuclear translocation of NF-κB, which is an intracellular mediator of TLR signals (3, 48). We tested whether stimulation of DC with heparan sulfate induces rapid increases in nuclear NF-κB, which is capable of binding to a DNA promoter sequence, as measured by gel EMSA. Fig. 6,A shows strong increases in NF-κB in DC stimulated by heparan sulfate, LPS, and CpG DNA. Because unstimulated DC contain basal levels of NF-κB, whereas unstimulated macrophages contain negligible levels of NF-κB (Fig. 6,B), we used macrophages, in which nuclear translocation of NF-κB occurs in response to LPS and heparan sulfate (49), to address this question in further detail. We used the murine macrophage cell line RAW 264.7, which responds to LPS in a TLR4-dependent manner (50). Fig. 6,C shows that heparan sulfate and LPS induce NF-κB translocation in a dose-dependent manner. Fig. 6 D shows that the kinetics of NF-κB translocation induced by heparan sulfate is significantly slower than those induced by LPS. These data suggest that, although heparan sulfate and LPS give rise to similar functional outcomes, the intracellular signals they generate differ, perhaps because of differences in extracellular molecular interactions with TLR4 and coreceptors.
Protection from tumors and rapidly evolving viruses, both of which may not synthesize molecules capable of stimulating sentinel immune receptors, requires a general mechanism for monitoring tissue well-being (14, 15, 16). Multicellular organisms mount innate (and, in the case of vertebrates, acquired) immune responses when challenged by foreign organisms and tissue injury. In plants and invertebrates these responses are triggered when ubiquitous endogenous macromolecules are degraded and fragments of these macromolecules interact with innate immune receptors (9, 51, 52, 53). Whether a similar system exists in higher animals is unknown. We have previously shown that heparan sulfate proteoglycan, a component of vertebrate cell membranes and extracellular matrices, is cleaved to form soluble heparan sulfate fragments in the course of inflammation and tissue damage (22, 54). In this work we report that fragments of heparan sulfate proteoglycan stimulate murine TLR4, leading to maturation of DC. This series of events links a pathway capable of activating primary immune responses in the absence of exogenous molecules that may function as a monitor of tissue well-being (Fig. 7).
Our findings may help to explain the observation that administration of heparin to experimental animals prevents generation of delayed-type hypersensitivity reactions and allograft rejection (55). Heparin, a polysaccharide structurally related to heparan sulfate that contains a markedly different sulfation pattern (34), does not stimulate DC maturation and inhibits enzymatic cleavage of heparan sulfate by heparanase, an enzyme activated by tissue damage (22). Thus, administration of heparin may prevent the release of heparan sulfate fragments that results from tissue damage and therefore block the stimulation of DC and immune responses.
Nearly every type of tissue injury, infection, and inflammation leads to shedding of heparan sulfate from mammalian cells (19, 20, 21, 22, 56). Activation of complement (21, 54, 57), neutrophils (58, 59, 60), or platelets (22), as well as the acidic environment of tissue damage (22), leads to the generation of soluble heparan sulfate proteoglycan and its fragmentation. Soluble heparan sulfate is not found in healthy tissues in significant quantities, whereas the concentration of soluble saccharide in the tissue fluid of wounds (61), the synovium of arthritic joints (62), and the urine of infected individuals (63) is within the range we observe to stimulate DC. Similar concentrations of homogalacturonan fragments are needed to stimulate host defense in plants (52, 53).
Activation of TLR by endogenous molecules such as heparan sulfate may also play a role in causing disease. Administration of high-dose CpG DNA or LPS leads to a syndrome that mimics septic shock, resulting in multiorgan failure and death (31, 40). However, humans who have survived severe injuries, as may result from burns or surgery, can also succumb to a “septic” syndrome, dying from multiorgan failure without any evidence of bacterial or fungal infection (64).
While we report studies using an in vitro model system, we believe it will be important to determine that activation of TLR4 by heparan sulfate also occurs in vivo. Our preliminary studies in several model systems are consistent with this likelihood.
We thank Nilofer Qureshi for providing Rs-DPLA and the Mayo Flow Cytometry Core Facility for their expert assistance.
This work was supported by National Institutes of Heath Grants HL46810 and HL52297.
Abbreviations used in this paper: TLR, Toll-like receptor; LBP, LPS-binding protein; LALF, Limulus anti-LPS factor; Rs-DPLA, Rhodobacter sphaeroides diphosphoryl lipid A; DC, dendritic cell.