The Toll-Like Receptor 5 Stimulus Bacterial Flagellin Induces Maturation and Chemokine Production in Human Dendritic Cells¹

Dendritic cells (DCs)⁴ represent a class of professional APCs that are critical in the initiation of a primary immune response. The proposed model for DC function is that the DCs take up foreign Ag in the periphery, process the Ag, and migrate to the T cell area of lymph nodes, where they present the Ag in the context of MHC along with costimulatory receptors, thereby activating naive T cells (1). Furthermore, innate immunity is thought to be triggered by pattern recognition receptors on APCs. Toll-like receptors (TLRs), which recognize conserved motifs on microorganisms, are thought to be largely responsible for this innate recognition. The TLR family consists of ten conserved germline-encoded type I transmembrane receptors that are present on APCs and function in response to triacylated lipopeptides (TLR1/2 heterodimers), diacylated lipopeptides (TLR2/6 heterodimers), dsRNA (TLR3), LPS (TLR4), flagellin (TLR5), imidazoquinolines (TLR7 and TLR8), and unmethylated CpG DNA (TLR9) (reviewed in Refs. 2, 3, 4, 5, 6, 7, 8).

In vitro studies of DC maturation have been conducted using cells derived from peripheral blood monocytes cultured in the presence of GM-CSF and IL-4. Such cells have an immature phenotype, having a high rate of endocytosis and expressing low levels of MHC class II, CD83, and the costimulatory molecules CD80 and CD86. Upon maturation, DCs down-regulate mechanisms of Ag capture, including endocytic activity and expression of Fc receptors, but they increase expression of costimulatory, adhesion, and MHC class II molecules (1, 9). Moreover, monocyte-derived DCs pulsed with tumor Ags ex vivo are a promising new tool in the treatment of malignant diseases (10, 11).

Recent work has shown DC maturation after exposure to microbial lipopeptides, LPS, viruses, and parasites (9, 12, 13, 14, 15, 16, 17, 18, 19, 20). Each of these microbes or their components has been found to activate cells through TLRs. This report demonstrates that human monocyte-derived DCs express TLR5, flagellin induces human DC maturation in vitro, and that this maturation is accompanied by chemokine and cytokine production.

Human and murine IL-4 and GM-CSF were purchased from PeproTech (Rocky Hill, NJ). Human IFN-β was purchased from R&D Systems (Minneapolis, MN). LPS (Escherichia coli K12) was from Sigma-Aldrich (St. Louis, MO). LPS was purified from contaminant lipoproteins normally found in commercially available LPS preparations by double phenol extraction, exactly as described (21). Synthetic lipopeptide palmitoylcysteine[(RS)-2,3-di(palmitoyloxy)propyl]serine-lysine-OH is derived from an E. coli lipoprotein and was purchased from Bachem (Torrance, CA). Flagellin was purified from Salmonella typhimurium (fliC) as previously described (22). The flagellin preparation contained <0.2 pg of LPS per microgram of protein as tested by Limulus amebocyte assay. This purified flagellin activates NF-κB in Chinese hamster ovary cells transfected with TLR5, but not with TLR2 or TLR4 (22). All other reagents were purchased from Sigma-Aldrich unless stated otherwise.

Buffy coats were obtained from healthy volunteers and fractionated over Histopaque-1077. The PBMC layer was recovered and erythrocyte depleted by incubation in RBC lysis buffer for 5 min at room temperature. PBMCs were cultured in complete medium (RPMI 1640, 1% l-glutamine, 1% penicillin/streptomycin, and 10% low-endotoxin FCS) for 2 h in T75 flasks (Corning Glass, Corning, NY). After incubation, nonadherent cells were removed by three washes with 1× PBS (Invitrogen, Carlsbad, CA). The remaining adherent cells were then cultured in complete medium supplemented with GM-CSF and IL-4 (50 ng/ml each). On days 2 and 4, the DC cultures received an additional dose of GM-CSF and IL-4 (50 ng/ml each). On day 5, nonadherent DCs were harvested by gentle pipetting, counted, and plated in fresh medium containing GM-CSF and IL-4 (50 ng/ml each). On day 6, some DCs were matured by addition of 100 ng/ml LPS or 0.2–60 nM fliC for 1–48 h.

Primary human blood DCs were purified from human peripheral blood by immunomagentic depletion of CD3-, CD11b-, and CD16-expressing cells followed by positive selection of CD4⁺ cells (Miltenyi Biotec, Auburn, CA).

Murine splenic DCs were isolated from spleens of C57BL/6 mice as previously described (23). Briefly, spleens were dissected into small pieces (1 mm³) and incubated at 37°C in complete RPMI 1640 supplemented with 2 mg/ml collagenase D for 45 min. Cell suspension was obtained by vigorous pipetting and passage through a 70-μm nylon mesh filter and was washed with complete medium. After lysis of RBCs, CD11c⁺ DCs were isolated using CD11c microbeads according to the manufacturer’s instructions (Miltenyi Biotec). The cells isolated were 90% CD11c⁺ as measured by FACS analysis.

Murine bone marrow-derived DCs were isolated from the marrow cavities of femurs and tibiae and were incubated in culture medium with 50 ng/ml GM-CSF and 50 ng/ml IL-4 at 10⁶ cells/ml. On day 5, DCs were harvested by gentle pipetting and were enriched by 14.5% (by weight) metrizamide density gradient separation. After centrifugation (15 min, 4°C, 2000 rpm), the low-density interface containing the DCs was collected by gentle pipette aspiration. The DCs were washed twice with culture medium, enumerated (purity >90% by positive-expression CD11c by FACS), and placed in culture medium with added cytokines for further studies.

Total RNA was extracted using the RNeasy kit according to the manufacturer’s protocol (Qiagen, Valencia, CA). Briefly, after DNase I (Invitrogen) treatment, 1 μg of total RNA from each sample was used as template for the reverse transcription reaction. One hundred microliters of cDNA were synthesized using oligo(dT)₁₅, random hexamers, and multiscribe reverse transcriptase (Applied Biosystems, Foster City, CA). All samples were reverse transcribed under the same conditions (25°C for 10 min, 48°C for 30 min) and from the same reverse transcription master mix to minimize differences in reverse transcription efficiency. All oligonucleotide primers for QPCR were designed using Primer Express software 1.0 (PE Biosystems, Foster City, CA) and synthesized by Invitrogen. The 25-μl QPCR contains 2 μl of cDNA, 12.5 μl of 2× SYBR Green master mix (Stratagene, La Jolla, CA), and 250 nmol of sense and anti-sense primer. The reaction conditions were as follows: 50°C for 2 min, 95°C for 10 min, then 40 cycles of 95°C for 15 s and 60°C for 1 min. Emitted fluorescence for each reaction was measured during the annealing/extension phase, and amplification plots were analyzed using the MX4000 software version 3.0 (Stratagene). A series of standards was prepared by performing 10-fold serial dilutions of full-length cDNAs in the range of 20 million copies to two copies per QPCR. Subsequent analysis was performed on the data output from the MX4000 software using Excel XP (Microsoft, Redmond, WA). The threshold cycle, i.e., the cycle number at which the amount of amplified gene of interest reaches threshold fluorescence, was determined by using the adaptive baseline algorithm in the MX4000 analysis software. This algorithm sets the threshold cycle at 10 SDs above the background fluorescence between cycles 5 through 9. Quantity values (i.e., copies) for gene expression were generated by comparison of the fluorescence generated by each sample with standard curves of known quantities. Next, the calculated number of copies was divided by the number of copies of the housekeeping gene GAPDH. In addition, we saw no significant changes in the QPCR results when the data were normalized using another constitutively active gene, β₂-microglobulin.

Immature DCs (iDCs), LPS- or fliC-treated DCs, were washed three times, diluted in fresh complete medium, and used as allogenic stimulators. Cells were seeded in 96-well round-bottom culture plates with APC serial dilutions ranging from 10,000 to 500 DCs/well and were mixed together with freshly purified CD3⁺ T cells (100,000/well). After 5 days of incubation, cells were pulsed with 1 μCi of [³H]thymidine per well for 18 h and were harvested on filter paper. Proliferative responses were measured as [³H]thymidine incorporation by an automatic beta counter. Tests were performed in triplicates, and results were expressed as the mean cpm.

Endocytic activity of DCs was measured by the uptake of FITC-conjugated dextran (molecular mass, 40,000 kDa; Molecular Probes, Eugene, OR) as previously described (9). Briefly, DCs stimulated for 24 h in the absence or presence of LPS or flagellin were incubated in complete medium plus 10% FCS plus 1 mg/ml FITC-conjugated dextran for 45 min at 4°C to measure nonspecific binding or at 37°C to measure specific uptake. Cells were then washed extensively and analyzed by flow cytometry.

Surface expression of various markers was assessed using CellQuest analysis software on a FACSCalibur (BD Biosciences, Mountain View, CA) flow cytometer. Surface expression was determined using the following FITC- and PE-conjugated Abs: CD86-FITC (Research Diagnostics, Flanders, NJ), CCR7-PE (R&D Systems), CD80-FITC, CD83-FITC, HLA-DR-FITC, and CD11c-PE (BD PharMingen, San Diego, CA). The isotype control Abs were used accordingly in all experiments and were purchased from BD PharMingen. Human DCs were incubated in 1% human AB serum/PBS, and murine DCs were incubated with rat anti-CD16/CD32 (BD PharMingen) to block nonspecific binding. Murine CCR7 was detected using macrophage-inflammatory protein 3β (MIP-3β)/CC chemokine ligand (CCL) 19-Fc fusion protein provided by Jason Cyster (University of California, San Francisco, CA) (24). MIP-3β-Fc was detected by addition of biotin-conjugated goat anti-human Fc, followed by PE-conjugated streptavidin (Caltag Laboratories, Burlingame, CA).

Human IL-8/CXC chemokine ligand (CXCL) 8 (BioSource International, Camarillo, CA), IFN-γ-inducible protein 10 kDa (IP-10)/CX chemokine ligand (CXCL) 10, IL-10, and IL-12 protein levels in the DC culture supernatants were measured by sandwich ELISA (R&D Systems).

It has been previously shown that LPS, lipopeptides, dsRNA, and unmethylated CpG DNA induce cellular activation via TLRs and that these pathogen-associated substances can induce DC maturation (9, 13, 15, 25, 26, 27, 28, 29, 30). Flagellin, a conserved protein monomer that makes up the bacterial flagellar filament, is recognized through TLR5. Therefore, we decided to test whether flagellin could also induce maturation of DCs (22, 31, 32, 33, 34). We used a common in vitro culture system for the production of human monocyte-derived DCs in which adherent PBMCs are cultured in IL-4 and GM-CSF for 5 days. As shown in Fig. 1,A, day 0 adherent cells have surface expression of CD14, CD11c, and class II MHC, but not CD83, a surface phenotype consistent with these cells being monocytes (35). After culture of these cells in IL-4 and GM-CSF for 3 and 5 days, surface expression of CD14 decreases, whereas surface expression of CD11c increases, a change also reflected at the mRNA level (Fig. 1,B). This phenotype is consistent with the differentiation of blood monocytes into iDCs (36). In agreement with other reports, these monocyte-derived human DCs gain expression of TLR3, reduce expression of TLRs 1, 2, 4, 5, and 8, and lose expression of TLR9 (Fig. 1 C) (37). This is consistent with the previously described ability of these cells to respond to dsRNA, but not unmethylated CpG DNA (38).

As a comparison to in vitro monocyte-derived DCs, we also analyzed TLR expression on primary blood DCs. As shown in Fig. 1 D, primary DCs express a TLR expression profile similar to that of monocyte-derived DCs, with the exception of a higher level of TLR10.

Analysis of TLR expression in iDCs by QPCR revealed TLR5 expression in these cells (Fig. 1,C). To determine whether TLR5 was functional in iDCs, we treated iDCs with (fliC). After 24 h, we observed an increase in the surface expression of CD83, CD86, and CCR7 in fliC-stimulated DCs (fliC-DCs) (Fig. 2, A and B), demonstrating TLR5 function in iDCs. This increase is also evident at the mRNA level (Fig. 2,C). LPS-stimulated DCs (LPS-DCs; used as positive controls in this study) also have a mature phenotype, consistent with previous reports (9, 38, 39, 40, 41). Finally, the TLR agonists LPS and flagellin induced expression of maturation markers (CD83, CD80, CD86, CCR7), the neutrophil chemoattractant (IL-8), and proinflammatory cytokine (TNF) from primary blood DCs in a manner similar to that of monocyte-derived DCs (Fig. 2 D).

To determine whether this maturation is accompanied by other hallmarks of DC maturation, we conducted two additional experiments. We measured endocytic uptake in monocyte-derived DCs using fluorescently labeled dextran. Whereas iDCs are proficient at endocytic uptake, mature DCs reduce their fluid phase uptake in preparation for Ag presentation to T cells (9). As shown in Fig. 3, we observed a significant reduction in endocytosis in fliC-DCs, similar to that seen in LPS-DCs.

Another feature of DC maturation is an enhanced ability to stimulate T cells due to increased MHC and coreceptor surface expression (42, 43). To determine whether fliC-DCs have enhanced ability to stimulate T cells, a T cell proliferation assay was performed. Immature DCs, LPS-DCs, and fliC-DCs were cocultured with T cells from a separate donor for 6 days, and allospecific T cell proliferation was measured by radiolabeled thymidine incorporation. T cell proliferation in cells cocultured with fliC-DCs was similar to that seen in T cells cocultured with LPS-DCs, and we observed a consistent, dose-dependent increase compared with iDCs (Fig. 4,A). We also assessed the T cell stimulatory activity of fliC-DCs by measuring the production of IFN-γ during the T cell proliferation assay. As shown in Fig. 4,B, T cells cultured with fliC-DCs produced 3- to 12-fold greater levels of IFN-γ relative to iDC cocultures (Fig. 4 B).

At the junction between innate and acquired immunity, DCs play an important role in initiating an immune response. The chemotaxis of effector cells to sites of infection and secondary lymphoid organs is an important aspect of immunity. To examine the potential role of DCs in trafficking of immune effector cells, we measured expression of chemokines (as well as cytokines) in fliC-DCs. We observed increased mRNA levels for a number of chemokines, notably growth-related oncogene-α (GRO-α)/CXCL1, GRO-β/CXCL2, GRO-γ/CXCL3, IL-8/CXCL8, monocyte chemoattractant protein-1 (MCP-1)/CCL2, MIP-1α/CCL3, and MIP-1β/CCL4, which are expressed quite early after stimulation (∼1–3 h) at high levels. Interestingly, the T cell chemoattractants IP-10/CXCL10, monokine induced by IFN-γ (MIG)/CXCL9, and IFN-inducible T cell α chemoattractant (I-TAC)/CXCL11, which are expressed in LPS-DCs at later time points (∼8 h), were not expressed in fliC-DCs (Fig. 5, A and B).

As shown in Fig. 5 C, fliC stimulated the expression of the proinflammatory cytokines IL-1β and TNF, as well as the Th1 cytokines IFN-γ and IL-12 p40. In contrast with LPS-DCs, fliC-DCs failed to induce expression of IFN-α and IFN-β. The inability of fliC to induce IFN-α/β expression may explain why the IFN-inducible chemokines IP-10, MIG, and I-TAC were not expressed by fliC stimulation. Recently, Vogel and coworkers (44) reported that TLR4, but not TLR2, agonists could induce IFN-β expression and subsequently STAT1α/β phosphorylation. In mice, LPS-induced IP-10 expression has been shown to be STAT1-dependent, as demonstrated by the failure of STAT1α/β^−/− macrophages to express IP-10 mRNA in response to LPS (45). To determine whether other non-IFN-β-inducing stimuli also fail to induce IP-10, MIG, and I-TAC, we stimulated iDCs with synthetic lipopeptide (palmitoylcysteine[(RS)-2,3-di(palmitoyloxy)propyl]serine-lysine-OH, a TLR1/2 stimulus). Lipopeptide-stimulated DCs failed to express IFN-α/β as previously described and also failed to express IP-10, MIG, and I-TAC (data not shown). Thus, TLR5 (fliC) and TLR1/2 (lipopeptide) agonists appear to similarly activate only a subset of TLR4 (LPS)-inducible proinflammatory genes.

To determine whether the increase in chemokine and cytokine message level is accompanied by protein secretion, ELISA was performed on the supernatants of fliC-DCs. We observed increased IL-8 in the supernatant of fliC-DCs compared with that of iDCs, whereas IP-10 protein production was not detected in fliC-DCs (Fig. 5, E and F). This pattern of chemokine secretion induced by flagellin is similar to what has been previously described for macrophages stimulated with lipopeptide (44).

Upon encounter with microbial pathogens, DCs produce immunostimulatory cytokines, such as IFN-γ and IL-12, which are important for selective activation of T cells into the Th1 phenotype. Interestingly, LPS-stimulated DCs, but not flagellin-stimulated DCs, produced bioactive IL-12 p70 (Fig. 5,G). Meanwhile, neither LPS nor flagellin induced IL-10 protein production (Fig. 5 H). These data fit with previously published data by Thoma-Uszynski et al. (12), which demonstrate that LPS and lipopeptide induced IL-12, but not IL-10, in monocyte-derived DCs (10). These data suggest that fliC-DCs may not be as potent at generating a Th1 cytokine response as LPS-DCs or lipopeptide DCs.

LPS, but not flagellin, induced the expression of IP-10, MIG, and I-TAC in a time-dependent manner (Fig. 5, A and B). Treatment of DCs with flagellin (20 nM) in combination with rIFN-β (50, 25, and 10 U/ml) induces expression of IP-10, MIG, and I-TAC mRNA, whereas stimulation with IFN-β or flagellin alone failed to induce these chemokines (Fig. 6, A–C). This suggests that IFN-β production after LPS stimulation provides an additional signal required for IP-10, MIG, and I-TAC expression. Meanwhile, addition of IFN-β had no effect on flagellin-induced IL-8 mRNA expression (Fig. 6,D). To determine whether the increase in IP-10 message level was accompanied by protein secretion, ELISA was performed on the supernatants of flagellin and IFN-β-stimulated DCs. As shown in Fig. 6 E, treatment of DCs with flagellin in combination with rIFN-β induced IP-10 protein production.

To demonstrate TLR5 function in murine DCs, we isolated murine splenic DCs and stimulated them with fliC. Surprisingly, purified bacterial flagellin was unable to mature murine splenic DCs in vitro as determined by cell surface and mRNA expression of CD80, CD86, and CCR7 (Fig. 6, A and B). As shown in Fig. 6,C, QPCR analysis of murine DCs revealed no expression of TLR5, consistent with their inability to respond to flagellin. Interestingly, murine bone marrow-derived DCs, peritoneal macrophages, and the murine macrophage cell line RAW 264.7 fail to express TLR5 mRNA (Fig. 6,D). We were only able to demonstrate positive TLR5 expression in murine skin epithelia (Fig. 6 D). These results are in agreement with a recent report by Applequist et al. (46), which also used mouse skin cDNA as a positive control and demonstrated little to no expression of TLR5 in several murine monocyte, macrophage, and DC cell lines. Also, we detected little or no TLR5 mRNA expression in resident and thioglycollate-elicited peritoneal macrophages from C57BL and BALB/c mice. Furthermore, these macrophages, as well as purified splenic DCs, failed to produce TNF or IL-8 protein production in response to flagellin stimulation (data not shown).

Interestingly, human DCs express TLR5, but not TLR9 (Fig. 1,C), whereas mouse DCs express TLR9, but not TLR5 (Fig. 6 C). The biological consequence of this species-specific difference remains unclear and will be discussed below.

In this report, we demonstrate maturation of human monocyte-derived DCs to the TLR5 stimulus bacterial flagellin. This observation helps explain previous reports regarding the high antigenicity and adjuvant activity of bacterial flagellin (47). Our results demonstrate that TLR5 is expressed on human iDCs and that these cells are sensitive to flagellin stimulation, consistent with other reports demonstrating TLR5 expression in human monocytes and iDCs by Northern blot and RT-PCR analysis (37, 38, 48).

We demonstrated a dose-dependent induction of DC maturation using 0.2–60 nM fliC. In subsequent experiments we chose to present data using 20 nM = 1 μg/ml fliC, because it was the lowest concentration that consistently gave a robust response and was equivalent to the potent maturation inducer LPS. Flagellin (20 nM) and LPS (100 ng/ml) consistently induced DC activation as measured by T cell stimulatory activity, induction of maturation and costimulatory molecules, reduction in endocytic activity, and equivalent expression of TNF and IL-1β (Fig. 5,B). However, we have consistently found that LPS induces higher levels of CD86 on LPS-DCs, which may explain the slightly higher T cell stimulatory activity of LPS-DCs compared with fliC-DCs (Figs. 2,B and 4).

Interestingly, our observations suggest that there are differences between DCs matured with LPS and DCs matured with flagellin. fliC-DCs do not express the T cell chemokines MIG, IP-10, and I-TAC or IFN-α/β, which are all expressed by LPS-DCs (Fig. 7). This suggests that there are qualitative differences between the signals generated by TLR5 compared with that provided by TLR4, a notion consistent with recent reports demonstrating the role of IFN regulatory factor 3 (IRF3) in mediating TLR3- and TLR4-specific transcriptional responses, but not TLR2 and TLR9 responses (49, 50). The simplest explanation for the chemokine disparity between fliC-DCs and LPS-DCs is that LPS stimulation results in TLR4-mediated early IFN-β secretion through an IRF3-mediated signaling pathway, leading to the expression of MIG, IP-10, and I-TAC. TLR5-mediated signals do not result in this early IFN-β production and consequently do not mediate the expression of MIG, IP-10, and I-TAC (Figs. 5, A and C, and 8). To test this hypothesis, we stimulated DCs with flagellin in combination with exogenous rIFN-β. Interestingly, stimulation with both flagellin and IFN-β resulted in IP-10, MIG, I-TAC expression, whereas flagellin or IFN-β treatment alone failed to induce these chemokines (Fig. 6). These data suggest that IFN-β production after LPS stimulation provides an additional signal required for IP-10, MIG, and I-TAC expression. The biological consequence of this disparity between the LPS and flagellin responses is not clear; however, in an actual infection, a number of different TLRs are likely to be engaged simultaneously, perhaps making the specific signaling pathways used by any one TLR biologically less relevant (2, 4).

Flagellin and LPS induced the expression of a number of other chemokines. The neutrophil-active chemokines GRO-α, GRO-β, GRO-γ, and IL-8 were expressed in both LPS- and fliC-treated DCs, as were the monocyte/macrophage/NK-active chemokines MCP-1, MIP-1α, MIP-1β, and RANTES. These findings suggest that DCs not only play a role in the initiation of an acquired immune response, but that they may also play a role in amplifying the early immune response by participating in the trafficking of innate immune cells to sites of infection. The early expression of these chemokines (1–3 h) is consistent with the notion that iDCs, upon encounter with pathogens, express these chemokines before the expression of the lymph node homing chemokine receptor CCR7, which is maximally expressed at 24 h after stimulation (Fig. 2 B).

Another intriguing finding was that murine splenic DCs do not mature to purified bacterial flagellin. Besides providing a control for the purity of our fliC preparation, this seems at odds with the recent findings by McSorley et al. (47) demonstrating the ability of bacterial flagellin to effectively elicit a robust CD4 T cell response in vivo. Though our in vitro findings predict flagellin will not directly stimulate DC maturation in vivo, it is still possible that other cells, such as skin epithelia and gut epithelia, which express TLR5 (Fig. 7 D), are able to produce cytokines (such as TNF and IFNs), which may then trigger DC maturation (34, 51). This is consistent with our previous data demonstrating IL-6 production in the serum of flagellin-injected wild-type mice, but not of MyD88 knockout mice (20).

A recent report by Renshaw et al. (52) demonstrated murine TLR5 expression in thioglycollate-elicited peritoneal macrophages from 2- to 3-mo-old C57BL/6 mice. Moreover, this group found that flagellin treatment could induce IL-6 and TNF production from thioglycollate-elicited peritoneal macrophages isolated from young mice, but not old mice. In addition, splenic macrophages were found to express lower levels of TLR5 and were nonresponsive to flagellin treatment. In Fig. 7, we demonstrated that resident peritoneal macrophages did not express TLR5 mRNA. We also assessed TLR5 mRNA expression in resident and thioglycollate-elicited peritoneal macrophages from C57BL/6 and BALB/c mice. Unlike Renshaw et al. (52), we were only able to detect low levels of TLR5 mRNA expression in these peritoneal macrophages (Fig. 7; data not shown). Furthermore, these cells did not produce TNF or IL-8 in response to flagellin stimulation (data not shown). These differences may be due to the way the cells were handled or cultured. Moreover, Renshaw et al. (52) did not provide any information about the purity of their flagellin preparation. We have found that it is very difficult to eliminate TLR2/TLR4 agonists from flagellin preparations. Our flagellin preparations are purified using extensive dialysis and size fractionation and are rigorously tested on Chinese hamster ovary cells stably transfected with TLR2, TLR4, or TLR5. Only flagellin preparations/fractions that solely activated TLR5-expressing cells were used in our studies. Alternatively, Renshaw et al.’s flagellin preparation (52) may be modified, enabling it to stimulate through murine TLR5 or in combination with another unidentified receptor. Indeed, we have found that monomeric flagellin (induced by sonicating) stimulates DCs more potently than does aggregated flagellin. However, both of these forms failed to activate murine macrophages or murine DCs in our experiments. Interestingly, Renshaw et al. (52) reported that splenic macrophages expressed only low levels of TLR5 and did not respond to flagellin, whereas thioglycollate-elicited peritoneal macrophages expressed higher levels of TLR5 and responded to flagellin. This might indicate that TLR5 varies based on the activation state (thioglycollate elicited) and anatomical location (spleen vs peritoneum). However, we have not been able to reproduce these findings.

The biological significance of murine DC unresponsiveness to flagellin is unclear. Again, a number of different TLRs are likely stimulated during exposure to microbes and the stimulation of TLR5 by flagellin is just one of these possible interactions. Therefore, it is unlikely that the lack of recognition of bacterial flagellin by murine DCs will result in mice displaying enhanced susceptibility to flagellated bacteria compared with humans; however, it may lead to qualitative differences in the immune response to flagellated bacteria between mice and humans (53, 54).

Recently, Rehli (reviewed in Ref. 54) clarified the species-specific variations of TLR expression in mice and humans. Recent data from several groups suggest that constitutive and inducible TLR expression in different cell types and in different species are controlled by transcriptional regulation (55, 56, 57). Moreover, gene regulatory elements found in the proximal promoters of TLR genes control cell-type specificity and inducible TLR expression in mice and humans. For example, the human TLR2 gene is constitutively expressed in monocytes and is not inducible by microbial pathogens. In contrast, the mouse TLR2 gene, which has NF-κB and STAT5 transcription factor binding sites in its proximal promoter, is rapidly induced by microbial stimuli (57). The human TLR3 gene is constitutively and selectively expressed in myeloid DCs, whereas the mouse TLR3 gene is inducible and expressed in macrophages. Likewise, TLR9 and TLR5 appear to be regulated in a cell type- and species-specific manner. Human DCs express TLR5, but not TLR9 (Fig. 1,C), whereas mouse DCs express TLR9, but not TLR5 (Fig. 7 C). The biological consequence of this species-specific difference remains unclear.

Our findings and other studies suggest that the stimulation of any TLR expressed on DCs is sufficient for maturation, as measured by surface expression of coreceptors, T cell stimulatory activity, and endocytic activity (9, 38). Finally, we demonstrate a qualitative difference in gene expression of IFN-α/β and the T cell chemokines IP-10, I-TAC, and MIG between fliC-DCs and LPS-DCs.