TLRs discriminate foreign from self via their specificity for pathogen-derived invariant ligands, an example being TLR9 recognizing bacterial unmethylated CpG motifs. In this study we report that endosomal translocation of CpG DNA via the natural endocytotic pathway is inefficient and highly saturable, whereas endosomal translocation of DNA complexed to the cationic lipid N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP) is not. Interestingly, DOTAP-mediated enhanced endosomal translocation of otherwise nonstimulatory vertebrate DNA or of certain noncanonical CpG motifs triggers robust dendritic cell activation in terms of both up-regulation of CD40/CD69 and cytokine production, such as type I IFN and IL-6. We report that the stimulatory activity of phosphorothioated noncanonical CpG oligodeoxynucleotides is TLR9 dependent, whereas phosphodiester DNA, such as vertebrate DNA, in addition trigger TLR9-independent pathways. We propose that the inefficiency of the natural route for DNA internalization hinders low affinity TLR9 ligands in endosomes to reach threshold concentrations required for TLR9 activation. Endosomal compartmentalization of TLR9 may thus reflect an evolutionary strategy to avoid TLR9 activation by self-DNA.

Bacterial and viral DNA rich in CpG motifs or small synthetic oligodeoxynucleotides (ODN) 4 containing CpG motifs activate innate immune cells such as dendritic cells (DCs) via TLR9 (reviewed in Refs. 1 and 2). Cellular activation by immunostimulatory CpG-ODN appears to be sequence specific, because inversion within CpG motifs of CG to GC or methylation of cytosine (in this study termed noncanonical CpG motifs) ablates its immunostimulatory activities (3, 4, 5, 6). Surprisingly, vertebrate DNA fails to activate innate immune cells (2, 6). To date, this lack of immunostimulatory potential has been explained by CpG suppression (7), CpG methylation (8), and/or inhibitory motifs (9, 10), even though vertebrate DNA still contains some unmethylated CpG motifs (6).

Unlike TLR1, -2, -4, -5, and -6, which recognize invariant foreign bacterial and viral constituents at the cell membrane, TLR3, -7, -8, and -9 are expressed endosomally in innate immune cells and recognize distinct patterns of nucleic acids at late endosomal compartments (11, 12, 13, 14, 15). Although in the former cases self-nonself discrimination can be easily explained by the invariant molecular nature of the respective foreign ligands, structural differences between pathogen and host nucleic acids appear less prominent. This raises the question of whether it is the endosomal compartmentalization and thus the limited accessibility of TLR3 and -7 to -9 that help to control foreign vs self nucleic acid discrimination. Specifically, pathogen-derived DNA may access the TLR9-expressing endosomal compartment of infected cells in the course of infections, whereas host-derived DNA may not, because it becomes rapidly degraded by extracellular DNase. Recent evidence implies that upon internalization of chromatin-IgG immune complexes (ICs) via BCR or FcγRIII of DCs, vertebrate DNA displays robust immunostimulatory activities toward B cells or DCs (16, 17). In the case of DCs, both TLR9-dependent and -independent signaling pathways appear to operate (17). Furthermore, the sera of patients with systemic lupus erythematosus containing ICs consisting of autologous DNA and anti-DNA Abs effectively activate plasmacytoid DCs (pDCs) to produce type I IFNs (18, 19). Like other researchers (20, 21, 22, 23), we have noted that high concentrations of noncanonical CpG DNA sequences may moderately activate TLR9-expressing immune cells even though they are sequence-divergent from canonical unmethylated CpG motifs to date considered as natural TLR9 ligand. Because canonical and noncanonical CpG motifs as well as vertebrate DNA first need to translocate to the endosome to trigger TLR9 (14, 15, 24), we analyzed whether the limited accessibility to TLR9 prevents vertebrate DNA to unveil immunostimulatory effects.

To clarify aspects of these matters, it seemed desirable to establish a system in which the potentially rate-limiting step of natural DNA internalization is bypassed. To do this, we complexed either vertebrate DNA or canonical or noncanonical CpG motifs to the cationic lipids N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP) known to enforce endosomal translocation (25, 26). To analyze the DC stimulatory capacity of DNA and to exclude interference by potentially contaminating non-TLR9 ligands such as lipopeptides or endotoxin, we generated DCs lacking either TLR2 and -4 or TLR9 in addition. We now describe that upon DOTAP-mediated enforced endosomal translocation, both vertebrate DNA as well as phosphodiester (PD) canonical and noncanonical CpG motifs can activate DCs via TLR9-dependent and -independent pathways to up-regulate costimulatory cell surface molecules. In contrast, production of DC subset-specific cytokines primarily requires signaling via TLR9. Interestingly, the immunostimulatory activities of phosphorothioated (PTO) canonical and noncanonical ssCpG motifs are found to be entirely TLR9 dependent.

C57BL6 wild-type (wt) mice were purchased from Harlan. TLR2−/−/TLR4−/−/TLR9+/+ mice were crossed with TLR9-deficient mice to yield triple TLR2−/−/TLR4−/−/TLR9−/− mice. Mice used as donors of bone marrow cells were on a mixed genetic background (129SV × C57BL/6).

The ODN, listed in Table I, were purchased from TIB Molbiol or MWG Biotech. Plasmid DNA (pcDNA4 Myc His; Invitrogen Life Technologies) was purified with an Endofree Plasmid kit (Qiagen). Plasmid DNA and calf thymus (CT) DNA (Sigma-Aldrich) were also subjected to phenol/chloroform/isoamylalcohol (25/24/1) extraction, followed by a Triton X-114-based purification step to remove LPS, as described previously (26).

Table I.

Sequences of ODN used in this studya

ODN NameSequences
1668 PTO tccatgacgttcctgatgct 
1720 PTO tccatgagcttcctgatgct 
NAOS-1 PTO gctcatgagcttcctgatgctg 
AP-1 PTO gcttgatgactcagccggaa 
2216 ggGGGACGATCGTCgggggg 
1668 PD TCCATGACGTTCCTGATGCT 
1720 PD TCCATGAGCTTCCTGATGCT 
m-1668 PD TCCATGA[meth]CGTTCCTGATGCT 
ODN NameSequences
1668 PTO tccatgacgttcctgatgct 
1720 PTO tccatgagcttcctgatgct 
NAOS-1 PTO gctcatgagcttcctgatgctg 
AP-1 PTO gcttgatgactcagccggaa 
2216 ggGGGACGATCGTCgggggg 
1668 PD TCCATGACGTTCCTGATGCT 
1720 PD TCCATGAGCTTCCTGATGCT 
m-1668 PD TCCATGA[meth]CGTTCCTGATGCT 
a

Lowercase letters indicate PTO, and capital letters show PD backbone.

In vitro Flt3 ligand (FL)-dependent DC (FL-DC) were generated as described previously (27). In short, bone marrow cells were cultured in the presence of murine rFL for 8 days. Resulting cells were >90% CD11c positive, and 30–40% of cells displayed a plasmacytoid phenotype (CD11cposCD45RAhighB220highCD11blow). DCs (3 × 105 to 5 × 105 cells/200 μl) were incubated with the respective stimuli in 96-well plates.

Human PBMC were generated by Ficoll gradient centrifugation and were seeded at 3 × 105 to 5 × 105 cells/200 μl cells for cytokine induction. Human pDC were enriched or depleted by incubation with PE-labeled Abs to CD123, followed by anti-PE beads (Miltenyi Biotec) as described previously (28) and were seeded at 5 × 104 cells/200 μl.

Human or murine cells were incubated for 18–24 h with 12.5 μg/ml DNA as indicated in the figures. Complexes of DNA with DOTAP (Roche) were prepared according to the manufacturer’s instructions. In brief, 5 μg of DNA in 50 μl of HBS (20 mM HEPES and 150 mM NaCl, pH 7.4) was combined with 10 μg of DOTAP in 50 μl of HBS. After 15 min of incubation, 100 μl of complete RPMI 1640 medium (10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 50 μM 2-ME) was added to the mixture of DNA and DOTAP. Cells in 100 μl of complete medium were incubated with 100 μl of DNA-DOTAP complexes for 18–24 h in 96-well culture plates., then collected for FACS analyses. The cytokines in the supernatants were quantified by ELISA as described previously (28).

Before stimulation, FL-DC cultures were routinely checked for the relative percentages of pDC and conventional DC (cDC) and the respective DC activation status by staining with mAb to CD11c, CD45RA, B220, CD11b, and CD62L. After stimulation, FL-DC were labeled with a combination of Abs to B220 (RA3-6B2-allophycocyanin), CD40 (FGK45.5-PE), and CD69 (H1.2F3-FITC; all from BD Biosciences) and analyzed on a FACSCalibur flow cytometer (BD Biosciences). For uptake analyses, FL-DC were exposed to 5′-Bodipy-tetramethylrhodamine (TMR)-conjugated CpG-ODN 1668-PTO (IBA Naps) in the presence or the absence of DOTAP for various time periods. Afterward, cells were extensively washed, chased for 3 h, stained with ethidium monoazide (Molecular Probes) to exclude dead cells, and subsequently analyzed.

DNA (30 μg) was diluted in 75 μl of HBS. DOTAP (60 μg) was diluted in 75 μl of HBS. Subsequently, the DNA solution was mixed with the DOTAP solution and incubated for 15 min. One hundred and fifty microliters of DNA-DOTAP complex containing solution was injected i.v. After 2 h, mice were killed, blood was collected, and cytokines in the plasma were determined by ELISA.

Ten thousand RAW264.7 murine macrophage cells per well were cultured overnight in eight-well chamber slides (Falcon; BD Biosciences). For stimulation, cells were incubated with 2 μM 3′-fluorescein-labeled CpG-ODN 1668-PTO (TIB Molbiol) for 60 min at 37°C and thereafter washed three times with ice-cold PBS. Cells were fixed (PBS and 4% formaldehyde) for 15 min and permeabilized with staining buffer (PBS, 0.4% saponin, and 2% normal goat serum (Invitrogen Life Technologies)) for 15 min. Incubation with primary Ab (anti-lysosome-associated membrane protein (LAMP1); BD Pharmingen; 1/500) and secondary Ab (anti-FITC-Alexa 488 (Molecular Probes; 1/400) and anti-rat IgG-Alexa 546 (Molecular Probes; 1/400)) was performed at room temperature for 1 h and 30 min, respectively. Microscope slides were mounted and analyzed at room temperature using a Zeiss LSM500 confocal microscope (software version 2.2) equipped with an argon/krypton laser (458/488 nm) and two helium/neon lasers (543 and 633 nm). The lens used was a Plan-Neofluar (Zeiss) 40 1.3 oil lens, and the pinhole was set to scan layers <1 μm at a resolution of 1024 × 1024 pixels. Pictures were then imported into Photoshop CS (Adobe Systems).

Recently, we and others used the cationic lipid DOTAP to translocate ssRNA to TLR7-expressing endosomes of DCs (28, 29). In this study we asked whether DOTAP can also be used to translocate DNA to endosomes expressing TLR9 (reviewed in Refs.2 and 24). In addition, we addressed the question of how the efficacy of DOTAP-mediated endosomal translocation compares with that of the natural translocation route known to be mediated by TLR9-independent DNA receptor-driven endocytosis (24, 30, 31, 32). Using FACS and confocal microscopy to quantitate natural vs DOTAP-mediated internalization of Bodipy-TMR-tagged CpG-ODN, we established that CpG DNA uptake via the natural internalization pathway is highly saturable at higher CpG DNA concentrations, i.e., 1–2 μM, whereas DOTAP-mediated internalization is not (Fig. 1). These differences were observed with both canonical and noncanonical CpG-ODN (data not given) and allowed DCs to internalize 10- to 50-fold more DNA complexed to DOTAP (Fig. 1). Confocal microscopy revealed that DOTAP translocated its cargo DNA into intracellular vesicular structures did not deposit CpG DNA at the cell surface (data not shown). In fact, colocalization analysis demonstrated that CpG DNA complexed to DOTAP translocated primarily to LAMP1-positive late endosomal compartments (Fig. 2), a site at which TLR9 is expressed and becomes activated (14, 15, 30). We concluded that by complexing DNA to DOTAP, up to 10- to 50-fold more DNA can be translocated to the late endosomal compartment of DCs compared with the natural route of CpG DNA uptake.

FIGURE 1.

Kinetics of CpG-DNA internalization by DCs. FL-DC (2 × 106/ml) were exposed for different time periods (10 min up to 3 h) with Bodipy-TMR-conjugated CpG-ODN 1668-PTO (0.2 and 2 μM). A, Cells were washed at the individual time points indicated and incubated for additional 3 h. B, FL-DC were exposed to Bodipy-TMR-conjugated CpG-ODN 1668-PTO complexed to DOTAP, washed, and incubated for an additional 3 h. The fluorescence of internalized ODN was determined by FACS. Similar results were obtained with TLR9-deficient FL-DC (data not given). One representative experiment of at least three is shown.

FIGURE 1.

Kinetics of CpG-DNA internalization by DCs. FL-DC (2 × 106/ml) were exposed for different time periods (10 min up to 3 h) with Bodipy-TMR-conjugated CpG-ODN 1668-PTO (0.2 and 2 μM). A, Cells were washed at the individual time points indicated and incubated for additional 3 h. B, FL-DC were exposed to Bodipy-TMR-conjugated CpG-ODN 1668-PTO complexed to DOTAP, washed, and incubated for an additional 3 h. The fluorescence of internalized ODN was determined by FACS. Similar results were obtained with TLR9-deficient FL-DC (data not given). One representative experiment of at least three is shown.

Close modal
FIGURE 2.

Liposomes (DOTAP) translocate CpG-DNA to LAMP1-positive endosomal compartments. RAW264.7 cells were pulsed for 60 min with 2 μM FITC-conjugated CpG-ODN 1668-PTO with or without DOTAP. Thereafter, cells were fixed, permeabilized, stained for LAMP1 (Alexa 546; red) and FITC (anti-FITC Alexa 488; green), and subjected to confocal laser scanning microscopy (CLSM). Representative pictures with pinhole settings for <1-μm layers are shown.

FIGURE 2.

Liposomes (DOTAP) translocate CpG-DNA to LAMP1-positive endosomal compartments. RAW264.7 cells were pulsed for 60 min with 2 μM FITC-conjugated CpG-ODN 1668-PTO with or without DOTAP. Thereafter, cells were fixed, permeabilized, stained for LAMP1 (Alexa 546; red) and FITC (anti-FITC Alexa 488; green), and subjected to confocal laser scanning microscopy (CLSM). Representative pictures with pinhole settings for <1-μm layers are shown.

Close modal

Canonical ssCpG motifs efficiently activating murine TLR9 contain the hexameric sequence GACGTT present in ODN 1668, whereas the prototypic ODN sequence for activation of human or primate TLR9 displays a GTCGTT sequence (reviewed in Refs. 1 and 2). Upon methylation of CpG-ODN 1668 or inversion of its central dinucleotide from CG to GC (herein termed noncanonical CpG motifs), the immunostimulatory activity of these ODNs is ablated (2). These findings are based on experiments using the natural DNA uptake pathway by exposing innate immune cells in vitro or in vivo to the respective DNAs (1, 2, 24). Under such experimental conditions, vertebrate DNA fails to stimulate innate immune cells (2, 6). Unexpectedly, upon DOTAP-mediated endosomal translocation of graded concentrations of the noncanonical CpG-ODN 1720 or CT DNA, these otherwise nonstimulatory DNAs triggered production of IL-6 in wt FL-DCs. Furthermore, both pDNA and CT DNA complexed to DOTAP caused low, but significant, IL-6 production in DCs from TLR9-deficient cells (Fig. 3,B). In contrast, DOTAP enhanced the stimulatory potential of the canonical CpG motif 1668 only at low concentrations (Fig. 3,A), whereas the noncanonical CpG motif 1720 complexed to DOTAP unfolded stimulatory activity at higher concentrations (Fig. 3,A), a finding also observed with the methylated CpG motif 1668 (data not given) or in part with bacterial pDNA (Fig. 3,B, inset). Of note, under conditions in which vertebrate DNA exhibited stimulatory activity (Fig. 3,B), the control ODNs, AP-1 or nonactivating ODN phosphorothioated (NAOS)-1, remained inactive (Fig. 3,C). The lack of stimulatory activity of the control ODNs, AP-1 and NAOS-1, indicates sequence specificity. Both control ODNs displayed a guanine at the 5′ end, and we noted that addition of a 5′ G (guanine) and 3′ G ablated the stimulatory potential of ODN 1720 (data not given). In quantitative terms, vertebrate DNA complexed to DOTAP displayed ∼10–15% the activity of the canonical CpG motif 1668 or of complexed pDNA, whereas complexed noncanonical CpG motif 1720 (or methylated ODN 1668; data not shown) unfolded ∼50% of this activity (Fig. 3).

FIGURE 3.

Endosomal translocation of canonical and noncanonical CpG-DNA, pDNA, and vertebrate DNA triggers IL-6 production. A, Respective DNA was mixed at graded concentrations with 10 μg of DOTAP in 50 μl of HBS. After incubation for 15 min at 37°C 100 μl of complete RPMI 1640 medium was added. The mixture was added to wt or TLR9-deficient FL-DC cultures (2.5 × 106/ml). B and C, ODN (0.8 nmol) or 5 μg of bacterial DNA or vertebrate DNA in 50 μl of HBS was combined with 10 μg of DOTAP (in 50 μl of HBS). After 15-min incubation, the DNA complexed to DOTAP was successively diluted, as indicated, and complete RPMI 1640 medium (100 μl) was added and thereafter cultured together with wt or TLR9-deficient FL-DCs. The supernatants were analyzed for IL-6 production by ELISA. One representative experiment of at least three is shown. Error bars represent the range of duplicate samples.

FIGURE 3.

Endosomal translocation of canonical and noncanonical CpG-DNA, pDNA, and vertebrate DNA triggers IL-6 production. A, Respective DNA was mixed at graded concentrations with 10 μg of DOTAP in 50 μl of HBS. After incubation for 15 min at 37°C 100 μl of complete RPMI 1640 medium was added. The mixture was added to wt or TLR9-deficient FL-DC cultures (2.5 × 106/ml). B and C, ODN (0.8 nmol) or 5 μg of bacterial DNA or vertebrate DNA in 50 μl of HBS was combined with 10 μg of DOTAP (in 50 μl of HBS). After 15-min incubation, the DNA complexed to DOTAP was successively diluted, as indicated, and complete RPMI 1640 medium (100 μl) was added and thereafter cultured together with wt or TLR9-deficient FL-DCs. The supernatants were analyzed for IL-6 production by ELISA. One representative experiment of at least three is shown. Error bars represent the range of duplicate samples.

Close modal

To exclude the effects of potential contamination with, for example, endotoxin or lipopeptides, we reanalyzed, at a 2-μM concentration, the immunostimulatory potential of methylated CpG-1668-PTO, 1668 PD, methylated 1668-PD, noncanonical CpG-ODN 1720-PTO, or 1720-PD, pDNA, and CT DNA using FL-DCs generated from TLR2 and -4 double-deficient or from TLR2, -4, and -9 triple-deficient bone marrow cells. Under these conditions, mixed DC populations of pDC and cDC were generated (33). As a marker for DC stimulation, we analyzed up-regulation of CD69 and CD40 (Fig. 4,A) as well as IFN-α (prototypic cytokine produced by pDCs) or IL-6 production (a cytokine preferentially produced by cDCs; Fig. 4,B). In the absence of DOTAP, only canonical CpG-ODN 1668-PTO and bacterial pDNA activated the TLR2- and TLR4-deficient DC populations to up-regulate CD40 and CD69. This was TLR9 dependent. Upon DOTAP-mediated endosomal translocation, however, the noncanonical CpG-ODN 1720-PTO, methylated CpG-1668-PD, as well as vertebrate CT DNA also caused strong up-regulation of CD40 and CD69, whereas the control ODN AP-1 did not. Fig. 4,A presents additional information. The immunostimulatory activity of PTO canonical and noncanonical CpG DNA complexed to DOTAP was entirely TLR9 dependent, yet PD DNA (ODN 1668, ODN 1720, bacterial pDNA, and vertebrate DNA) displayed TLR9-independent stimulatory activity, because they caused up-regulation of CD40 and CD69 in TLR2, TLR4-, and TLR9-deficient DCs (Fig. 4,A). Fig. 4,B illustrates IFN-α and IL-6 production triggered by the various stimuli in these mixed DC cultures. As expected, only the canonical CpG-ODN 1668-PTO and bacterial pDNA (PD) caused, in the absence of DOTAP, TLR2- and TLR4-deficient DC to produce substantial amounts of IL-6. However, when complexed to DOTAP, vertebrate DNA as well as noncanonical CpG motifs, such as methylated ODN 1668-PD or noncanonical CpG-ODN 1720-PTO, were also effective (Fig. 4,B). Again, the control ODN AP-1 was negative. IL-6 production was largely TLR9 dependent, because TLR2, -4, and -9 triple-deficient DC produced only minor amounts (1–5%) of IL-6. We were surprised to note that CpG-ODN 1668-PTO failed to trigger IFN-α production in TLR2- and TLR4-deficient DCs, whereas noncanonical CpG-ODN 1720-PTO was effective (Fig. 4,B). We therefore titrated the concentrations of these ODN. As shown in Fig. 4,C, the biological activity of ODN-1668 PTO was suppressed at high (2 μM) concentrations, whereas that of ODN 1720-PTO was not. Yet, both ODN triggered IFN-α at intermediate ODN concentrations. Thus, both vertebrate DNA as well as the canonical CpG motif 1668 and the noncanonical CpG motifs (methylated ODN 1668 and ODN 1720) complexed to DOTAP efficiently triggered IFN-α production in TLR2- and TLR4-deficient DCs. In contrast, IFN-α production in TLR2-, TLR4-, and TLR9-deficient DCs could not be detected. Taken together, these data allowed three major conclusions. First, the sequence restrictions described for canonical CpG motifs (reviewed in Refs.1 and 2) become less stringent upon DOTAP-mediated enhanced endosomal delivery. Second, upon endosomal translocation, vertebrate DNA (known to contain noncanonical CpG motifs) displays DC stimulatory activities. Third, PTO canonical and noncanonical CpG DNA stimulate DCs in a TLR9-dependent fashion, whereas PD DNAs, in addition, trigger a TLR9-independent pathway that drives up-regulation of CD40 and CD69, but is poor in inducing cytokine production (Fig. 4, A and B).

FIGURE 4.

DOTAP-complexed vertebrate DNA or noncanonical CpG-DNA activate DCs. TLR2- and TLR4-deficient or TLR2-, TLR4-, and TLR9-deficient murine FL-DC (2.5 × 106ml) were stimulated as indicated (2 μM). After 18 h, cells were stained with Abs directed to CD69 or CD40. ▪, Stimulated cells; □, cells in medium only (A). The culture supernatants were analyzed for IFN-α or IL-6 content by ELISA (B). C, Respective DNA was mixed at graded concentrations with 10 μg of DOTAP in 50 μl of HBS. After incubation for 15 min at 37°C 100 μl of complete RPMI 1640 medium was added. The mixture was added to wt FL-DC cultures (2.5 × 106/ml). One representative experiment of at least three is shown. Error bars represent the SD of duplicate values; the asterisk indicates that cytokines were not detectable by ELISA.

FIGURE 4.

DOTAP-complexed vertebrate DNA or noncanonical CpG-DNA activate DCs. TLR2- and TLR4-deficient or TLR2-, TLR4-, and TLR9-deficient murine FL-DC (2.5 × 106ml) were stimulated as indicated (2 μM). After 18 h, cells were stained with Abs directed to CD69 or CD40. ▪, Stimulated cells; □, cells in medium only (A). The culture supernatants were analyzed for IFN-α or IL-6 content by ELISA (B). C, Respective DNA was mixed at graded concentrations with 10 μg of DOTAP in 50 μl of HBS. After incubation for 15 min at 37°C 100 μl of complete RPMI 1640 medium was added. The mixture was added to wt FL-DC cultures (2.5 × 106/ml). One representative experiment of at least three is shown. Error bars represent the SD of duplicate values; the asterisk indicates that cytokines were not detectable by ELISA.

Close modal

In humans, pDCs and B cells express TLR9 (2, 24). We tested whether upon DOTAP-mediated endosomal translocation of noncanonical PD or PTO CpG motifs triggers type I IFN release from human PBMCs. As expected, without DOTAP, only type A CpG-ODN 2216 (34, 35) triggered IFN-α production (Fig. 5,A). Unexpectedly, upon DOTAP-mediated endosomal translocation, murine CpG-ODN 1668-PD, noncanonical CpG-ODN 1720-PD, and methylated CpG 1668-PD as well as bacterial pDNA and vertebrate DNA, in part, also caused robust release of IFN-α from human PMBC, whereas PTO-ODNs were poor inducers, and control ODN AP-1 was inactive. Dose-response experiments confirmed that noncanonical PD CpG motifs complexed to DOTAP triggered robust IFN-α production comparable to the type A ODN 2216 (data not shown). Within human PMBC, only cells positively selected for CD123, a marker typifying pDCs (36), produced IFN-α (Fig. 5 B). We concluded that DOTAP-mediated endosomal translocation of murine canonical or noncanonical CpG motifs as well as of vertebrate DNA overcomes the species and sequence restrictions described for TLR9 activation (reviewed in Refs.2 and 24), thus allowing murine PD DNA motifs to trigger efficient type I IFN induction in human pDCs.

FIGURE 5.

Noncanonical CpG-DNA complexed to DOTAP induces IFN-α production of human pDC. A, Human PBMC (1.5 × 106 cells/ml) were stimulated with ODN complexed or not complexed to DOTAP. B, CD123-negative or -positive selected cells (2.5 × 105/ml; □ and ▪, respectively). Cells were stimulated with ODN, complexed or not complexed to DOTAP. The supernatants were analyzed for their contents of IFN-α by ELISA. Error bars represent the SD of duplicate values. The results of one experiment (shown) are representative of results for at least three replicate experiments with different donors.

FIGURE 5.

Noncanonical CpG-DNA complexed to DOTAP induces IFN-α production of human pDC. A, Human PBMC (1.5 × 106 cells/ml) were stimulated with ODN complexed or not complexed to DOTAP. B, CD123-negative or -positive selected cells (2.5 × 105/ml; □ and ▪, respectively). Cells were stimulated with ODN, complexed or not complexed to DOTAP. The supernatants were analyzed for their contents of IFN-α by ELISA. Error bars represent the SD of duplicate values. The results of one experiment (shown) are representative of results for at least three replicate experiments with different donors.

Close modal

In mice, systemic delivery of cationic lipid-pDNA complexes induces acute, TLR9-dependent inflammatory responses, which in the context of therapeutic gene delivery represent dose-limiting toxicities (37). Following up our results in vitro, we explored whether systemic delivery of vertebrate DNA or of the noncanonical CpG motif 1720 complexed to DOTAP triggers IL-6 production in vivo. As shown in Fig. 6, upon i.v. challenge of wt mice, significant amounts of serum borne IL-6, a bona fide marker of inflammation, were detectable.

FIGURE 6.

Non-CpG-DNA and vertebrate DNA complexed to DOTAP trigger systemic IL-6 release. Serum cytokine levels in mice after systemic delivery of DNA or DNA complexed to DOTAP. DNA (30 μg) or DNA (30 μg) complexed to DOTAP (60 μg) in 150 μl was injected i.v. into wt mice. Blood was collected 2 h after injection. n = 4 mice (1720 or 1720 and DOTAP), 5 mice (CT-DNA or CT-DNA and DOTAP), or 9 mice (AP-1 and DOTAP or DOTAP only) per group. The bar indicates the mean of the data. Differences in cytokine levels in the samples treated with DNA and DNA plus DOTAP were statistically analyzed by Mann-Whitney rank- sum test. ∗, p = 0.008; #, p = 0.029.

FIGURE 6.

Non-CpG-DNA and vertebrate DNA complexed to DOTAP trigger systemic IL-6 release. Serum cytokine levels in mice after systemic delivery of DNA or DNA complexed to DOTAP. DNA (30 μg) or DNA (30 μg) complexed to DOTAP (60 μg) in 150 μl was injected i.v. into wt mice. Blood was collected 2 h after injection. n = 4 mice (1720 or 1720 and DOTAP), 5 mice (CT-DNA or CT-DNA and DOTAP), or 9 mice (AP-1 and DOTAP or DOTAP only) per group. The bar indicates the mean of the data. Differences in cytokine levels in the samples treated with DNA and DNA plus DOTAP were statistically analyzed by Mann-Whitney rank- sum test. ∗, p = 0.008; #, p = 0.029.

Close modal

Our experiments report the surprising finding that upon endosomal translocation, both vertebrate DNA and certain noncanonical CpG motifs cause activation of murine and human DCs via TLR9-dependent and -independent pathways. These findings apparently contrast with the current paradigm of species- and sequence-specific TLR9 DNA ligand interactions (reviewed in Refs. 1, 2 , and 24). Nonetheless, our results are consistent with aspects of the recently described immunobiology of chromatin-IgG ICs, as observed upon BCR- or FcγRIII-mediated cellular internalization (16, 17).

The paradigm of species- and sequence-specific CpG DNA-mediated immune cell activation via TLR9/MyD88 rests on a large body of experimental data in which innate immune cells have been challenged with CpG motifs (reviewed in Refs. 2 and 24). Under such conditions, vertebrate DNA and noncanonical CpG DNA lack stimulatory activities. However, because TLR9 recognizes its ligands at a late endosomal compartments (reviewed in Ref.24), we hypothesized that poor endosomal translocation of putative TLR9 ligands via the natural uptake pathway might be rate limiting. Accordingly low affinity TLR9 ligands might fail to reach the threshold concentrations required for TLR9 activation. Under natural conditions, endosomal translocation of DNA is brought about by an as yet ill-defined sequence-nonspecific DNA receptor that TLR9-independently drives receptor-mediated endocytosis (31, 32 ; reviewed in Ref.24). In this study we show that this natural endocytosis pathway is highly saturable, because at 1–2 μM ODN, saturation is reached within 10–20 min (Fig. 1). At present, we explain this saturation by the loss of cell surface DNA receptors due to ligand-driven endocytosis; DOTAP-mediated endosomal DNA translocation is also known to use endocytotic pathways (25), yet, as shown in this study, this endocytotic pathway is not saturable. As a consequence, up to 10- to 50-fold higher concentrations of potential TLR9 ligands become translocated. Under such conditions, endosomally translocated, noncanonical PTO CpG motifs trigger DC activation in a TLR9- (Figs. 3 and 4) and MyD88 (data not shown)-dependent fashion, whereas control ODNs, such as AP-1 and NAOS-1, do not. At present, we explain the inactivity of AP-1 or NAOS-1 with the guanine at the 5′ position because we observed that modification of, for example, ODN 1720 by addition of a G nucleotide at the 5′ and/or 3′ end abolishes its stimulatory potential (data not shown). Interestingly, PD DNA, such as vertebrate DNA and bacterial pDNA, displayed, in addition, TLR9-independent immunostimulation. Together, these results support the hypothesis that upon enhanced endosomal translocation, low affinity ligands in endosomes can reach the threshold concentrations required to drive TLR9 activation. Accordingly, our results confine the accepted paradigm of species- and sequence-specific TLR9 DNA ligand interactions to the natural uptake route, because upon DOTAP-mediated enhanced endosomal translocation, noncanonical CpG motifs or vertebrate DNA also trigger TLR9 activation. Vertebrate DNA is known to be rich in noncanonical and methylated CpG motifs (6). We therefore assume that vertebrate DNA donates sufficient low affinity ligands upon fragmentation by DNase within late endosomes, a process likely to be facilitated by liposome-driven DNA-helix denaturation in ssDNA (38). In support of this idea, plasmon resonance (Biacore)-based analysis of sequence-dependent TLR9-CpG-ODN interactions revealed low affinity interactions of noncanonical CpG-ODN 1720 PD with recombinant TLR9 (39) (data not shown), whereas PTO ODN allowed no analysis of CpG sequence specificity, because it bound nonspecifically even to TLR2 (39). Perhaps the latter observations might explain why PTO ODN sequence nonspecifically precipitates TLR9 from cellular lysates (15).

Several lines of evidence imply immune stimulatory activities of vertebrate DNA contingent upon translocation to TLR9-expressing endosomes. For example, a role of self-DNA as an indicator of tissue damage has been suggested, although self-DNA failed to trigger cytokine production (40). Furthermore, mouse embryos deficient in DNase II contain macrophages carrying undigested DNA intracellularly, and this causes lethal type I IFN production (41). Perhaps the strongest evidence has been provided by Marshak-Rothstein et al. (16), who described a pathway involved in the means by which vertebrate DNA within chromatin-IgG ICs enter B cell endosomes via BCRs or enter endosomes of DCs via FcγRIII, the latter triggering both TLR9-dependent and -independent cell activation (17). However, FcγRs often contain ITAM (42), which may cloud interpretation of cell activation. Our data, therefore, extend the elegant findings of Marshak-Rothstein et al. (16), because we show that mere endosomal translocation is the key for vertebrate DNA to induce immunostimulation via TLR9-dependent and -independent pathways.

Our report is the first to describe factors that determine initiation of TLR9-dependent vs TLR9/MyD88-independent DC activation pathways. For example, inactivated DNA viruses, such as HSV-1, activate DC solely via TLR9, whereas infectious HSV-1 recruits an TLR9-independent pathway in addition (33). In this study we report that a chemical modification of the DNA backbone represents a determining factor for TLR9 dependency, because canonical and noncanonical PTO CpG motifs activate DCs via TLR9, whereas PD DNA (as contained in bacterial pDNA or vertebrate DNA) in addition triggers a TLR9/MyD88-independent DC activation pathway. Future studies are needed to analyze the molecular relatedness of the TLR9-independent pathway with the signal pathway driven by TLR9.

We observed that canonical and noncanonical murine PD CpG motifs displayed powerful interferonic activities toward human pDCs. The pioneering work of Alm and Rönnbloom (18, 19) pointed out that inversion of CpG→GpC or methylation of a CpG motif (identified in serum of systemic lupus erythematosus patients (43)) did not ablate the ability to trigger type I IFN production in human pDCs. The latter reports speculated that interferonic DNA sequences may be more frequent in eukaryotic DNA than anticipated. Indeed, upon enhanced endosomal translocation into human pDCs, vertebrate DNA, pDNA, and canonical and noncanonical murine CpG-ODN PD triggered robust type I IFN production (Fig. 5). The magnitude of type I IFN production was equal to the amount triggered by type A CpG-ODN 2216, which is generally selected for efficient IFN-α induction in pDCs (34, 35). We found that type A CpG-ODN is notoriously poor in TLR9-dependent NF-κB activation (unpublished observations) as are murine CpG motifs toward human TLR9 (4). Therefore, these ligands may be catalogued as low affinity ligands for human TLR9. It is thus possible that activation of pDCs for IFN-α production preferentially requires TLR9 activation via low affinity ligands, whereas high affinity ligands such as type B CpG-ODN (34, 35) trigger too rapidly the maturation of pDCs into professional APCs able to secrete Th1-polarizing cytokines such as IL-12 (36, 44).

Because TLR1, -2 and -4 to -6 recognize invariant foreign bacterial and viral products at the cell membrane, one may ask why TLR3 and TLR7 to -9 recognize distinct patterns of nucleic acids at late endosomes (reviewed in Ref. 24). Although in the former case, self-nonself discrimination can easily be envisaged, structural differences between pathogen- and host-derived nucleic acids appear less prominent, because mere endosomal translocation of vertebrate DNA or of noncanonical CpG motifs unveils strong stimulatory activity toward DCs. We therefore propose that it is the restricted accessibility of endosomally expressed TLR9 that hinders activation of TLR9-expressing immune cells by host-derived self-DNA, because the latter is likely to be rapidly degraded via DNases.

We acknowledge Cornelia Wagner, Hendrikje Drexler, and Monika Hammel for excellent technical assistance, and Meredith O’Keeffe (The Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia) for help with DC characterization.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by Deutsche Forschungsgemeinschaft (FORIMMUN SFB391 and SFB456) and by the Coley Pharmaceutical Group.

4

Abbreviations used in this paper: ODN, oligodeoxynucleotide; CT, calf thymus; DC, dendritic cell; cDC, conventional DC; CT DNA, calf thymus DNA; FL, Flt3 ligand; FL-DC, FL-dependent DC; IC, immune complex; PD, phosphodiester; pDC, plasmacytoid DC; DOTAP, N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate; pDNA, plasmid DNA; PTO, phosphorothioated; TMR, tetramethylrhodamine; LAMP, lysome-associated membrane protein; NAOS, nonactivating ODN phosphorothioated; wt, wild type.

1
Wagner, H..
1999
. Bacterial CpG DNA activates immune cells to signal infectious danger.
Adv. Immunol.
73
:
329
-368.
2
Krieg, A. M..
2002
. CpG motifs in bacterial DNA and their immune effects.
Annu. Rev. Immunol.
20
:
709
-760.
3
Hemmi, H., O. Takeuchi, T. Kawai, T. Kaisho, S. Sato, H. Sanjo, M. Matsumoto, K. Hoshino, H. Wagner, K. Takeda, et al
2000
. A Toll-like receptor recognizes bacterial DNA.
Nature
408
:
740
-745.
4
Bauer, S., C. J. Kirschning, H. Hacker, V. Redecke, S. Hausmann, S. Akira, H. Wagner, G. B. Lipford.
2001
. Human TLR9 confers responsiveness to bacterial DNA via species-specific CpG motif recognition.
Proc. Natl. Acad. Sci. USA
98
:
9237
-9242.
5
Krieg, A. M., A. K. Yi, S. Matson, T. J. Waldschmidt, G. A. Bishop, R. Teasdale, G. A. Koretzky, D. M. Klinman.
1995
. CpG motifs in bacterial DNA trigger direct B-cell activation.
Nature
374
:
546
-549.
6
Stacey, K. J., G. R. Young, F. Clark, D. P. Sester, T. L. Roberts, S. Naik, M. J. Sweet, D. A. Hume.
2003
. The molecular basis for the lack of immunostimulatory activity of vertebrate DNA.
J. Immunol.
170
:
3614
-3620.
7
Bird, A. P..
1993
. Functions for DNA methylation in vertebrates.
Cold Spring Harb. Symp. Quant. Biol.
58
:
281
-285.
8
Ramsahoye, B. H., C. S. Davies, K. I. Mills.
1996
. DNA methylation: biology and significance.
Blood Rev.
10
:
249
-261.
9
Krieg, A. M., T. Wu, R. Weeratna, S. M. Efler, L. Love-Homan, L. Yang, A. K. Yi, D. Short, H. L. Davis.
1998
. Sequence motifs in adenoviral DNA block immune activation by stimulatory CpG motifs.
Proc. Natl. Acad. Sci. USA
95
:
12631
-12636.
10
Shirota, H., M. Gursel, D. M. Klinman.
2004
. Suppressive oligodeoxynucleotides inhibit Th1 differentiation by blocking IFN-γ- and IL-12-mediated signaling.
J. Immunol.
173
:
5002
-5007.
11
Heil, F., P. Ahmad-Nejad, H. Hemmi, H. Hochrein, F. Ampenberger, T. Gellert, H. Dietrich, G. Lipford, K. Takeda, S. Akira, et al
2003
. The Toll-like receptor 7 (TLR7)-specific stimulus loxoribine uncovers a strong relationship within the TLR7, 8 and 9 subfamily.
Eur. J. Immunol.
33
:
2987
-2997.
12
Lee, J., T. H. Chuang, V. Redecke, L. She, P. M. Pitha, D. A. Carson, E. Raz, H. B. Cottam.
2003
. Molecular basis for the immunostimulatory activity of guanine nucleoside analogs: activation of Toll-like receptor 7.
Proc. Natl. Acad. Sci. USA
100
:
6646
-6651.
13
Matsumoto, M., K. Funami, M. Tanabe, H. Oshiumi, M. Shingai, Y. Seto, A. Yamamoto, T. Seya.
2003
. Subcellular localization of Toll-like receptor 3 in human dendritic cells.
J. Immunol.
171
:
3154
-3162.
14
Ahmad-Nejad, P., H. Hacker, M. Rutz, S. Bauer, R. M. Vabulas, H. Wagner.
2002
. Bacterial CpG-DNA and lipopolysaccharides activate Toll-like receptors at distinct cellular compartments.
Eur. J. Immunol.
32
:
1958
-1968.
15
Latz, E., A. Schoenemeyer, A. Visintin, K. A. Fitzgerald, B. G. Monks, C. F. Knetter, E. Lien, N. J. Nilsen, T. Espevik, D. T. Golenbock.
2004
. TLR9 signals after translocating from the ER to CpG DNA in the lysosome.
Nat. Immunol.
5
:
190
-198.
16
Leadbetter, E. A., I. R. Rifkin, A. M. Hohlbaum, B. C. Beaudette, M. J. Shlomchik, A. Marshak-Rothstein.
2002
. Chromatin-IgG complexes activate B cells by dual engagement of IgM and Toll-like receptors.
Nature
416
:
603
-607.
17
Boule, M. W., C. Broughton, F. Mackay, S. Akira, A. Marshak-Rothstein, I. R. Rifkin.
2004
. Toll-like receptor 9-dependent and -independent dendritic cell activation by chromatin-immunoglobulin G complexes.
J. Exp. Med.
199
:
1631
-1640.
18
Vallin, H., A. Perers, G. V. Alm, L. Ronnblom.
1999
. Anti-double-stranded DNA antibodies and immunostimulatory plasmid DNA in combination mimic the endogenous IFN-α inducer in systemic lupus erythematosus.
J. Immunol.
163
:
6306
-6313.
19
Magnusson, M., S. Magnusson, H. Vallin, L. Ronnblom, G. V. Alm.
2001
. Importance of CpG dinucleotides in activation of natural IFN-α-producing cells by a lupus-related oligodeoxynucleotide.
Scand. J. Immunol.
54
:
543
-550.
20
Wang, Y., A. M. Krieg.
2004
. Induction of autoantibody production but not autoimmune disease in HEL transgenic mice vaccinated with HEL in combination with CpG or control oligodeoxynucleotides.
Vaccine
22
:
2641
-2650.
21
Cornelie, S., O. Poulain-Godefroy, C. Lund, C. Vendeville, E. Ban, M. Capron, G. Riveau.
2004
. Methylated CpG-containing plasmid activates the immune system.
Scand. J. Immunol.
59
:
143
-151.
22
Elias, F., J. Flo, R. A. Lopez, J. Zorzopulos, A. Montaner, J. M. Rodriguez.
2003
. Strong cytosine-guanosine-independent immunostimulation in humans and other primates by synthetic oligodeoxynucleotides with PyNTTTTGT motifs.
J. Immunol.
171
:
3697
-3704.
23
Vollmer, J., R. D. Weeratna, M. Jurk, U. Samulowitz, M. J. McCluskie, P. Payette, H. L. Davis, C. Schetter, A. M. Krieg.
2004
. Oligodeoxynucleotides lacking CpG dinucleotides mediate Toll-like receptor 9 dependent T helper type 2 biased immune stimulation.
Immunology
113
:
212
-223.
24
Wagner, H..
2004
. The immunobiology of the TLR9 subfamily.
Trends Immunol.
25
:
381
-386.
25
Almofti, M. R., H. Harashima, Y. Shinohara, A. Almofti, Y. Baba, H. Kiwada.
2003
. Cationic liposome-mediated gene delivery: biophysical study and mechanism of internalization.
Arch. Biochem. Biophys.
410
:
246
-253.
26
Yasuda, K., Y. Ogawa, M. Kishimoto, T. Takagi, M. Hashida, Y. Takakura.
2002
. Plasmid DNA activates murine macrophages to induce inflammatory cytokines in a CpG motif-independent manner by complex formation with cationic liposomes.
Biochem. Biophys. Res. Commun.
293
:
344
-348.
27
Spies, B., H. Hochrein, M. Vabulas, K. Huster, D. H. Busch, F. Schmitz, A. Heit, H. Wagner.
2003
. Vaccination with plasmid DNA activates dendritic cells via Toll-like receptor 9 (TLR9) but functions in TLR9-deficient mice.
J. Immunol.
171
:
5908
-5912.
28
Heil, F., H. Hemmi, H. Hochrein, F. Ampenberger, C. Kirschning, S. Akira, G. Lipford, H. Wagner, S. Bauer.
2004
. Species-specific recognition of single-stranded RNA via Toll-like receptor 7 and 8.
Science
303
:
1526
-1529.
29
Diebold, S. S., T. Kaisho, H. Hemmi, S. Akira, S. Reis e Sousa.
2004
. Innate antiviral responses by means of TLR7-mediated recognition of single-stranded RNA.
Science
303
:
1529
-1531.
30
Hacker, H., H. Mischak, T. Miethke, S. Liptay, R. Schmid, T. Sparwasser, K. Heeg, G. B. Lipford, H. Wagner.
1998
. CpG-DNA-specific activation of antigen-presenting cells requires stress kinase activity and is preceded by non-specific endocytosis and endosomal maturation.
EMBO J.
17
:
6230
-6240.
31
Shirota, H., K. Sano, N. Hirasawa, T. Terui, K. Ohuchi, T. Hattori, G. Tamura.
2002
. B cells capturing antigen conjugated with CpG oligodeoxynucleotides induce Th1 cells by elaborating IL-12.
J. Immunol.
169
:
787
-794.
32
Heit, A., T. Maurer, H. Hochrein, S. Bauer, K. M. Huster, D. H. Busch, H. Wagner.
2003
. Cutting edge: Toll-like receptor 9 expression is not required for CpG DNA-aided cross-presentation of DNA-conjugated antigens but essential for cross-priming of CD8 T cells.
J. Immunol.
170
:
2802
-2805.
33
Hochrein, H., B. Schlatter, M. O’Keeffe, C. Wagner, F. Schmitz, M. Schiemann, S. Bauer, M. Suter, H. Wagner.
2004
. Herpes simplex virus type-1 induces IFN-α production via Toll-like receptor 9-dependent and -independent pathways.
Proc. Natl. Acad. Sci. USA
101
:
11416
-11421.
34
Krug, A., S. Rothenfusser, V. Hornung, B. Jahrsdorfer, S. Blackwell, Z. K. Ballas, S. Endres, A. M. Krieg, G. Hartmann.
2001
. Identification of CpG oligonucleotide sequences with high induction of IFN-α/β in plasmacytoid dendritic cells.
Eur. J. Immunol.
31
:
2154
-2163.
35
Verthelyi, D., K. J. Ishii, M. Gursel, F. Takeshita, D. M. Klinman.
2001
. Human peripheral blood cells differentially recognize and respond to two distinct CPG motifs.
J. Immunol.
166
:
2372
-2377.
36
Bauer, M., V. Redecke, J. W. Ellwart, B. Scherer, J. P. Kremer, H. Wagner, G. B. Lipford.
2001
. Bacterial CpG-DNA triggers activation and maturation of human CD11c, CD123+ dendritic cells.
J. Immunol.
166
:
5000
-5007.
37
Zhao, H., H. Hemmi, S. Akira, S. H. Cheng, R. K. Scheule, N. S. Yew.
2004
. Contribution of Toll-like receptor 9 signaling to the acute inflammatory response to nonviral vectors.
Mol. Ther.
9
:
241
-248.
38
Prasad, T. K., V. Gopal, N. M. Rao.
2003
. Cationic lipids and cationic ligands induce DNA helix denaturation: detection of single stranded regions by KMnO4 probing.
FEBS Lett.
552
:
199
-206.
39
Rutz, M., J. Metzger, T. Gellert, P. Luppa, G. B. Lipford, H. Wagner, S. Bauer.
2004
. Toll-like receptor 9 binds single-stranded CpG-DNA in a sequence- and pH-dependent manner.
Eur. J. Immunol.
34
:
2541
-2550.
40
Ishii, K. J., K. Suzuki, C. Coban, F. Takeshita, Y. Itoh, H. Matoba, L. D. Kohn, D. M. Klinman.
2001
. Genomic DNA released by dying cells induces the maturation of APCs.
J. Immunol.
167
:
2602
-2607.
41
Yoshida, H., Y. Okabe, K. Kawane, H. Fukuyama, S. Nagata.
2005
. Lethal anemia caused by interferon-β produced in mouse embryos carrying undigested DNA.
Nat. Immunol.
6
:
49
-56.
42
Takai, T..
2002
. Roles of Fc receptors in autoimmunity.
Nat. Rev. Immunol.
2
:
580
-592.
43
Sato, Y., M. Miyata, Y. Sato, T. Nishimaki, H. Kochi, R. Kasukawa.
1999
. CpG motif-containing DNA fragments from sera of patients with systemic lupus erythematosus proliferate mononuclear cells in vitro.
J. Rheumatol.
26
:
294
-301.
44
Boonstra, A., C. Asselin-Paturel, M. Gilliet, C. Crain, G. Trinchieri, Y. J. Liu, A. O’Garra.
2003
. Flexibility of mouse classical and plasmacytoid-derived dendritic cells in directing T helper type 1 and 2 cell development: dependency on antigen dose and differential Toll-like receptor ligation.
J. Exp. Med.
197
:
101
-109.