Abstract
Synthetic type B phosphorothioate oligodeoxyribonucleotides (ODN) activate mouse B cells via Toll-like receptor 9 (TLR9). Starting with closely related 15-mer prototype ODN, the sequence requirements for stimulatory (ST-) and inhibitory (IN-) activity were contrasted, by measuring apoptosis protection, G1 entry and interleukin-6 secretion. ST-ODN and IN-ODN differ in that (1) ST-ODN require a 5′ T, (2) the central CG is obligatory, (3) CG must be flanked 3′ specifically by TT at the position where IN-ODN have GG, and (4) IN-ODN tolerate truncation of the 3′ end better than ST-ODN. Features shared by ST-ODN and IN-ODN include (1) requiring CC adjacent to the 5′ end, and (2) avoiding CC immediately 5′ to the CG. This pattern is used to create a model of how ST-ODN binding might function to aggregate TLR9 so as to initiate the signal, and how the 5′ ends of ST-ODN and IN-ODN compete for binding. Further justification for considering TLR9 to be the ODN receptor was provided by a demonstration that in HEK293 cells transfected with TLR9, the potency of a panel of ODN for activating NF-κB roughly parallels that seen in the biological assays in mouse B cells.
Keywords: apoptosis, B cells, cell cycle entry, CpG-DNA, interleukin-6
Introduction
Our innate immune receptors have evolved to recognize a variety of molecules made by pathogens but not by us. The importance of these ‘danger signals’ for accelerating and directing the adaptive immune response is increasingly being recognized.1
Among these receptors is Toll-like receptor 9 (TLR9), which mediates the response to bacterial DNA.2,3 Short, single-stranded oligonucleotides are able to duplicate the response to bacterial DNA.2 A target motif was identified by a computer search for short sequences that are present less frequently in mammalian DNA than in bacterial DNA, where frequencies approximate the chance distribution. CG pairs were shown to be about one-quarter as frequent in mammalian DNA. If the C were unmethylated, and the adjacent bases were two purines on the 5′ side and two pyrimidines on the 3′ side, the motif frequency in mammalian DNA was decreased to about one-sixteenth that of bacterial DNA.4 An alternative formulation expressed the motif as ‘not C, unmethylated C, G, not G’.5 Because of the ease of degradation of the native oligodeoxyribonucleotides (ODN) with phosphodiester (O-) backbones by nucleases, the nuclease-resistant sulphur-substituted phosphorothioate (S-) series has been commonly used in experiments and in vaccine protocols. In mice, on a molar basis, the stimulatory S-ODN are approximately 100–200 times more potent than O-ODN with identical sequences.3
Quantitative species-related sequence preferences exist such that GACGTT appears in a variety of S-ODN that stimulate mouse cells best, and GTCGTT appears in those that stimulate human cells best.6 ODN-responsive cell types (B cells and dendritic cells in both species, and macrophages in mice) all appear to share this preference. However, species-related preferences are inapparent with O-ODN.7
Previous attempts to define the optimal stimulatory sequence have focused mainly on the central five or six bases, noting major potency loss associated with methylating the central C, reversing the CG to GC, and reversing the adjacent base substitutions to give CCG or CGG.4 Our surprising discovery that the sequence requirements for inhibitory (IN-) ODN are restricted to only six bases8 led us to re-examine the sequence requirements for stimulatory activity, which proved to be much more extensive than previously thought. These new insights generated a model of the putative TLR9-ODN binding site.
Materials and methods
Source and characterization of resting B cells
B6 D2 F1 female mice at 8–12 weeks of age from Jackson Laboratories, Bar Harbor, ME, were housed in a specific pathogen-free facility with HEPA-filtered air, autoclaved bedding and water. Spleen cell suspensions underwent red blood cell lysis, then negative selection with anti-CD43-coated magnetic beads (Midi-Macs by Miltenyi Biotec, Auburn, CA). The yield of 25 × 106 total B cells per spleen consisted of > 97% CD19+ cells, of which 1% were apoptotic, 1% in G1 and 98% in G0 by acridine orange flow cytometry.9 A human embryonic kidney cell line (HEK293) stably transfected with TLR9 and a nuclear factor-κB (NF-κB) -promoter luciferase construct10 were the kind gifts of Dr Grayson Lipford (Coley Pharmaceuticals, Wellesley, MA). HEK293 cells expressing the NF-κB-promoter-luciferase construct but lacking TLR9 were purchased from Panomics.
Oligonucleotides (ODN)
Synthetic phosphorothioate (S-) ODN were purchased from Integrated DNA Technologies, Coralville, IA. The prototype ST-ODN was 20845 and appeared for reference in all experiments: TCCTGACGTTGAAGT. Titrations were usually performed with five concentrations spread 0.5 log apart in the 10–1000 nm range.
Flow cytometry
By a modification of the method of Traganos and colleagues11 apoptosis and phases of the cell cycle were determined after 18 hr of culture by acridine orange flow cytometry, which measures the RNA and DNA content of each cell. For each run, freshly isolated B cells were used to set the G0/G1 and G0/apoptosis boundaries. Cells in G1, G2, S and M phases were included in ‘% cell cycle entry’; the denominator was total cells counted.
Interleukin-6 enzyme-linked immunosorbent assay (IL-6 ELISA)
Supernatants collected at 18 hr were incubated for 2 hr at room temperature in 10% fetal calf serum/phosphate-buffered saline on Greiner Bio-One plates which were coated with anti-IL-6 antibodies MP5-20-F3 (Bio Legend, San Diego, CA) overnight, then washed three times. After washing MP5-32C11-biotin and extravidin-peroxidase were used to detect bound IL-6. The assay sensitivity was 16 pg/ml using recombinant IL-6 (BioLegend) as standard. Examples of dose–response curves for these three assays appear in Figure 1 in ref. 9.
Figure 1.
The mean concentration of each ODN which produced 50% of maximal apoptosis protection (□) and G1 entry (○) by flow cytometry with acridine orange, and 50% of the IL-6 secretion by ELISA (Δ), generated by 100 nm of the prototype 2084, was determined in triplicate experiments. The geometric means of these three assay means is called the ‘potency index’. The potency index for 2084/the potency index of the test ODN = % activity. In addition to symbols for the four bases (A, C, G and T), × indicates a missing base, and R indicates a position containing random bases. All sequences have the phosphorothioate (S-) backbone. Potency differences of approximately two-fold are significant at P < 0·005.8
NF-κB activation
Parental HEK293 cells and HEK293 cells stably expressing TLR9 and an NF-κB-promoter-luciferase construct were incubated for 24 hr with a panel of ODN with different stimulatory potencies for B cells or with 50 ng/ml of tumour necrosis factor-α as a positive control. After three washes in Ca2+/Mg2+-free medium, the cells were resuspended in lysis buffer containing 1% Triton-X-100 and centrifuged to remove debris. Twenty microlitres of supernatant was then mixed with 100 μl luciferin using the Promega assay system and light was quantified by luminometry.
Data analysis
The geometric mean of the concentrations giving 50% of a maximal response to 2084 in each of three experiments was taken as the potency of an ODN. The maximal IL-6 response was taken as that given by 100 nm 2084, whereas the other curves gave distinct plateaus. The potency of the prototype ODN 2084 was set at 100% activity. The % activity was given by the equation: (potency of 2084/the potency of a test ODN) × 100%.
Results
Apoptosis protection, entry into G1 and IL-6 secretion yielded the same rank order of ODN activity. Apoptosis protection required the least ODN 2084 for 50% maximal activity (14 nm), IL-6 secretion required the most (119 nm) and G1 entry an intermediate amount (34 nm), a rank order that was also seen with all the other ST-ODN (Fig. 1).
Truncation studies
Base positions were numbered according to the scheme at the top of Table 1 (Fig. 1). Loss of only the T at −7 (4168) decreased activity by 98%. At the 3′ end, loss of the first base (+ 8) only reduced activity by 25% (4170), but activity was progressively lost with further 3′ truncation (4209, 4167). Interestingly, the length, not the base sequence, was critical at the 3′ end, because ODN 4215 with random bases from +5 to +8 had normal activity. When an ODN lacked the T at − 7, the loss of one or two bases at the 3′ end made no further difference (4149, 4169). When three bases at the 3′ end and at least two at the 5′ end were deleted, there was no detectable activity (4148, 4147 and 4146), meaning > 99·8% loss. Thus the central six-base motif GACGTT, which contains several structural features necessary for activity,4 is not sufficient for activity on its own (4146).
Table 1.
Comparisons between IN-ODN and ST-ODN activities
The 5′ half
Filling the − 7 position was essential to activity (4168 in Fig. 1) and the T was strongly preferred to other bases (2084 versus 4271, 4272, and 4291). But substitutions at either − 6 or − 5 produced a greater loss of activity (4277, 4278, 4293) than substitutions at − 7, and substituting both − 6 and − 5 reduced activity to around 1–3% (4150, 4276). Interestingly, 15 bases may be the optimal length, because single base extension (4273, 4274) or duplication of positions − 7 to − 4 (4275) reduced activity somewhat. In contrast, position − 4 tolerated an A for T substitution well (95% for 4263). Random bases at − 4 to − 7 led to < 2% activity, emphasizing the importance of base sequence near the 5′ end. A C for G substitution at − 3 retained 60% activity (4213). Position − 2 was more sensitive; G for A at − 2 retained 58% of activity (2147), but T at − 2 retained only 17% (4212) and C at − 2 retained 5% (4151). CC at − 2 − 3 gave only 3% activity (4161).
The central CG
The central CG was very unforgiving of changes as expected.4,5 An A or T substitution at − 1 (4156 and 4157) or an A at + 1 (4164) showed no more than 1% activity, as did the much used reversal of CG to GC (1% activity for 2087).
The 3′ half
Positions + 2 and + 3 were very sensitive to substitutions. CC at + 2, + 3 destroyed activity (0·6% for 4162), whereas AA left 5% activity (4214), and random bases left <0·2% (4216). When in 4295, TT and AA are exchanged within the prototype sequence, about 1% activity remains. Since GG at this location makes an ODN inhibitory,7 TT appears to be strictly required for stimulation. Single base changes also had an effect. C at + 2 had 31% activity (4153) but a G at the same position reduced activity to 1.9% (2086). At + 3, C reduced activity to 3.9% (4158). In contrast, G at + 5 was fairly harmless (55% for 4165). GG at + 5, + 6 had only 14% activity (4166), showing sequence preferences extended from − 7 to + 6. Interestingly, random bases at + 4 to + 8 had essentially full activity (91% of the prototype in 4215), showing that length was more important than sequence at the 3′ end (compare 4170, 4209 and 4215).
Covalent linkage requirement
Comparison was made between intact 2084 and an equimolar mixture of ‘half ODNs’ prepared by dividing 2084 between positions + 1 and + 2 or between − 3 and − 2. These mixtures had no detectable activity (< 0·2% of 2084). The same was true for the closely related inhibitory sequence 2114 when divided at the same positions (Fig. 2).
Figure 2.
Areas 1 and 3 must be linked covalently for activity in both ST-ODN and IN-ODN. (a) B-cell G1 entry at 18 hr was compared between ST-ODN 2084 and its fragments: 4267, TCCTG; and 4268, ACGTTGAAGT; 4269, TCCTGACG; and 4270, TTGAAGT. Fragments of 2084 were tested separately or together in an equimolar mixture. (b) Inhibition of B-cell G1 entry at 18 hr was compared between IN-ODN 2114 and its fragments: 4267, TCCTG; and 4288, GAGGGGAAGT; 4289, TCCTGGAG; and 4290, GGGAAGT; when all cultures contained 100 nm of ST-ODN 2084. The % inhibition of GI entry is plotted.
NF-κB activation
A panel of ODNs was chosen from Fig. 1 showing widely different potencies in B cells associated with minor sequence changes. In contrast to tumour necrosis factor-α, representative ODN failed to induce measurable NF-κB activity in parental HEK293 cells lacking TLR9 (Fig. 3a). Their relative potencies in HEK293 cells transfected with TLR9 and expressing an NF-κB-binding promoter sequence linked to luciferase (Fig. 3b) was similar to their potency in B cells (Fig. 1). The three most potent ODN ranked in the same order, whereas the five ODN with less than 5% activity in B cells (Fig. 1) also ranked the lowest in the NF-κB assay (Fig. 3b).
Figure 3.
In HEK293 cells TLR9 confers similar relative ODN potencies to those seen in B cells. The untransfected HEK293 kidney cell line responds to TNF-α but does not respond to ODN (a). After transfection with TLR9 it becomes responsive to ODN (b), measured as NF-κB activation detected with a transgenic NF-κB-binding promoter linked to the luciferase gene.10 Luciferase activity was assessed after a 24-hr exposure to a panel of ODN at 100 nm (a, b) and 330 nm (b), listed in order of their activity relative to ODN 2084 from Fig. 1: 2084: 100%; 4209: 36%; 4271: 16%; 4168: 2%; 4278: 4%; 4151: 5%; 4156: < 0·2%; 4214: 5%. Error bars represent the standard errors of the mean for three experiments.
Discussion
This reassessment of the sequence requirements for stimulatory activity in phosphorothioate (S-) ODN was prompted by our recent analysis of the sequence requirements for inhibitory activity.8,12 The experiments reported here significantly extend our knowledge of the sequence requirements for ODN stimulatory activity, revealing that (in contrast to IN-ODN) nearly every region of a prototype strong ST-ODN is essential for optimal activity in mouse B cells. Some ODN with numbers < 3000 which had been tested before were included for comparison.9 While the central six bases that received most attention in published data are important,4,5 regions closer to the ends are also. Table 1 summarizes the differences and similarities in the sequence requirements for ST-ODN and IN-ODN activity.
Focusing on differences, for example, truncation of only T-7 from the 5′ end reduced activity to 2% (4168 in Fig. 1) consistent with a recent report that a 5′-TC is important.13 As bases were deleted from the 3′ end, the loss of activity was more gradual, reaching 11% when four bases were deleted (4167 in Fig. 1). In contrast, IN-ODN activity is not affected by loss of T-7 and is minimally affected by 3′ truncation.8,12
Extending IN-ODN 2114 at the 5′ end by duplicating area 1 doubled activity8 but a similar 5′ extension of ST-ODN 2084 reduced activity (4275). Modest reductions were also seen with single base 5′ extensions (4273, 4274). Together with the sensitivity to truncation of even one 5′ base (4168) or two 3′ bases (4209) this establishes the optimal length for an ST-ODN at 14–15 bases, whereas IN-ODNs retained significant activity down to 10–12 bases.8,12
All of the substitutions for either the central C at − 1 or G at + 1 destroyed activity (4157, 4156, 4164), as did the well-known reversal of CG to GC (2087 in Fig. 1).4,5 These requirements for ST-ODN activity contrast with those of IN-ODN, which tolerate deletion of T at −7 and substitutions at −1 and + 1 with no significant loss of activity, and which show a lesser impact of serial deletions at the 3′ end.8,12
However, there are also sequence features shared by ST-ODN and IN-ODN. For example, when four bases were removed at the 3′ end, restoring the length of ST-ODN with random bases restored full activity (4215), just as it did with IN-ODN,8 revealing that the ODN sequence in this region is not recognized by the ODN-receptor. The importance of ODN length could reflect binding between the phosphorothioate backbone and the receptor, or an effect on orientation within the binding site. More importantly, our analysis of the impact of single base changes on IN-ODN activity revealed that only three pairs of the 15 bases contribute substantially to activity.8,12 These ‘active areas’ were at positions − 5, − 6 (area 1) and − 2, − 3 (area 2) and + 1–3 or + 2–4 (area 3) of the prototype sequence 2114 (Table 1). Comparing Fig. 1 to ref. 8 shows that in areas 1 and 2, the base preferences for ST-ODN and IN-ODN are essentially the same. Examples are the major decline in ST-ODN activity with AA or GG for CC at − 5, − 6 (4150, 4276), with C at − 2 (4151) or CC at − 2, − 3 (4161). These results suggest that area 1 may be where ST-ODN and IN-ODN compete for binding to the ODN receptor, whereas CC at area 2 eliminates both ST-ODN and IN-ODN activity.
Positions + 2 and + 3 (area 3) are also vital to both ST-ODN and IN-ODN, but here their sequences must differ. Substitutions of AA (4214), CC (4162), or random bases (4216) at these positions destroyed ST-ODN activity (Fig. 1), whereas the GG substitution was essential for inhibitory activity.8 Only TT at + 2, + 3 allowed full stimulatory activity (prototype 2084 in Fig. 1). Thus the choice of TT or GG at area 3 determines whether a signal is initiated. To show that not every base contributes, the A at −4 had no significant impact on activity of ST-ODN (4263 in Fig. 1) or IN-ODN.8 Also, truncation or base changes at the 3′ end have much less impact on activity than similar changes at the 5′ end. This finding agrees with the results of others who in seeking optimal design for drugs have found that linking the CpG ODN tail to tail at the 3′ end14 or conjugating them to protein antigens at the 3′ end15 produces good activity, whereas linking to the 5′ end destroys activity.
Several of our observations with phosphorothioate ODN confirm conclusions reached earlier with phosphodiester sequences, which on average are < 1/100 as potent.5 These include the reduction in activity with C at − 2 or G at + 28, the substitution of T or A for C at − 14 or the substitution of A for T at + 24. Nuclease resistance is the most obvious reason for the increased potency of S-ODN, but with certain sequences their decreased flexibility may also be influential. Furthermore, S-ODN may be taken up more rapidly by cells.7 The relative activity is not always the same between the same sequences with different backbones.16
ST-ODN activity appears to have many more stringent sequence requirements than IN-ODN activity. For example, 38 of the 60 single base sequence variants of the prototype 15-mer IN-ODN 2114 did not differ significantly in activity from the prototype8 a finding that contrasts with the wide variety of minor changes that reduced stimulatory activity in Fig. 1. It would certainly be advantageous for our own DNA not to stimulate TLR9, since innate immune activation of antigen-presenting cells can circumvent self-tolerance17 and cause B-cell proliferation and T-independent immunoglobulin secretion4 and inflammatory cytokine release.18 Since, except for the ‘CpG motif’4 and the telomeric hexamer, short sequences appear at chance frequency in all mammalian DNA, the simplest strategy for preventing autostimulation during a heart attack or crush injury would be for mammalian DNA to have a net inhibitory effect on TLR9 activation, which it does.16 The simple fact that inhibitory sequences are more abundant than stimulatory sequences can at least partially account for this. But a second contributing factor is that the most abundant hexamer sequence in mammalian DNA, the telomeric hexamer TTA GGG, as a tandem triplet is inhibitory19 about 60% as active as our prototype IN-ODN 21148.
Increasing evidence points to TLR9 as the receptor for ST-ODN. Embryonic kidney cells lack TLR9 and are unresponsive to CpG DNA (Fig. 3). However, when TLR9 is transfected together with a luciferase gene linked to an NF-κB-binding promoter, CpG-DNA can be shown to activate NF-κB in those cells.10 We have shown that ODN 2084 triggers the disappearance of Iκβα and Ikβ, and the cleavage of p105, causing the translocation of NF-κB components p50, p65 and c-Rel into the B cell nucleus20. Furthermore, blocking NF-κB blocks ODN-driven 2084-cell cycle entry and apoptosis protection20. We selected a set of ODN from Fig. 1 which showed widely varying potency in the B-cell assays, and showed in Fig. 3 not only that transfection of TLR9 conferred responsiveness [as previously shown by Bauer et al. 10], but that the order of potency of the most active ODN in the NF-κB assay with TLR9 (Fig. 2) closely paralleled that seen in the biological assays with B cells (Fig. 1).
Direct binding studies with TLR9 protein have shown higher affinity for active CG sequences than for the inactive GC sequence.21 Since the inhibition of ST-ODN by IN-ODN is fully competitive and reversible8 and since all downstream events appear to vary proportionately when IN-ODN is added8 or when the ST-ODN sequence is changed (Fig. 1), it is likely that the proportion of TLR9s bound to ST-ODN determines the biological response. With antigen–antibody interactions, minor sequence changes in the binding site can significantly affect binding avidity, partly because the binding site consists of multiple loops whose orientation relative to one another is critical.
Based on our data on the sequence requirements for ST-ODN activity (this work) and IN-ODN activity8,12 we created a unimolecular or conformational TLR9 site model, wherein several different areas of a ST-ODN must engage TLR9 properly for the binding site for the adaptor protein MyD88 to be created.8 The ODN sequence features that make positive contributions (areas 1 and 3) were postulated to interact with binding subsites on TLR9. CC at area 2 (−2, −3) however, makes a negative contribution by binding outside the site, rendering the ODN incapable of either stimulating or inhibiting. IN-ODN, which share most features with ST-ODN in the 5′ half but have GG at positions +2, +3 may compete with ST-ODN for binding via their 5′ half, but may block or distort the MyD88 binding site via area 3.
However, TLR9 is now thought to be a membrane protein once it is recruited from the endoplasmic reticulum to endosomal vesicles.22 In this orientation, it is likely that both ST-ODN and IN-ODN would bind TLR9 within the vesicle, while MyD88 would be recruited to the outside (cytoplasmic) domain of TLR9. Figure 4 shows how ST-ODN binding could dimerize TLR9 in the orientation that creates a MyD88 site, whereas IN-ODN dimerizes TLR9 in a different orientation so that MyD88 cannot bind. ODN with CC at area 2 bind decoy site δ, and thus are unable to inhibit or stimulate. In either the associative model (Fig. 3) or the conformational model,8 areas 1 and 3 must be covalently linked at the correct distance for the MyD88 site to form. There is evidence for both Type A12 and Type B8 ODN that the distance between areas 1 and 3 is important. The associative model in Fig. 3, in which the ODN ties two TLR9s together, predicts that even if areas 1, 2, and 3 are intact, breaking the covalent linkage between areas 1 and 3 would destroy activity for both ST-ODN and IN-ODN. Figure 2 demonstrates that this prediction is correct. These data would also be compatible with a model wherein sites 1 and 3 bind the same TLR9, provided that binding enabled TLR9s to assemble so as to create the MyD88 binding site (a conformational-associative hybrid model). Despite the evidence we have presented supporting various features of these models, X-ray crystallographic modelling of TLR9 with ODN will be necessary for confirmation or modification.
Figure 4.
The Associative Model postulates that ST-ODNs bind to the N-terminal intravesicular portion of TLR9 to induce the formation of the MyD88 binding site on the outside of the endosomal vesicle. The α-site binds area 1, and therefore is the site of competition between ST-ODN 2084 and molecules binding simultaneously. One or two ST-ODN could link two TLR9s from α to β inside the vesicle in the orientation that generates the MyD88 binding sites outside the vesicle. IN-ODN link site α to γ, which rotates TL9 in an orientation that destroys the MyD88 binding site and prevents the binding of ST-ODN. If Cs are present in area 2 (4161) then the ODN binds instead at the δ-site, leaving the α, β, or γ-sites available for interaction with ODN.
Acknowledgments
The authors thank Jill Kinnaird and Dede Ollendick for skilful manuscript preparation, Dr Grayson Lipford for providing the TLR9-transfected HEK293 cells, and Shawn Roach and Teresa Ruggle for graphics assistance. Supported by R01AI047374 from NIAID/GM to RFA/PL and VA Merit Grant to RFA.
Abbreviations
- DC
dendritic cell
- IN
inhibitory
- O-
phosphodiester
- ODN
oligodeoxyribonucleotides
- S-
phosphorothioate
- ST
stimulatory
- TLR
toll-like receptor
References
- 1.Takeda K, Kaisho T, Akira S. Toll-like receptors. Annu Rev Immunol. 2003;21:335–76. doi: 10.1146/annurev.immunol.21.120601.141126. [DOI] [PubMed] [Google Scholar]
- 2.Krieg AM. CpG motifs in bacterial DNA and their immune effects. Annu Rev Immunol. 2002;20:709–60. doi: 10.1146/annurev.immunol.20.100301.064842. [DOI] [PubMed] [Google Scholar]
- 3.Hemmi H, Takeuchi O, Kawal T, et al. A toll-like receptor recognizes bacterial DNA. Nature. 2000;408:740–5. doi: 10.1038/35047123. [DOI] [PubMed] [Google Scholar]
- 4.Krieg AM, Yi AK, Matson S, Waldschmidt TJ, Bishop GA, Teasdale R, Koretzky GA, Klinman D. CpG motifs in bacterial DNA trigger direct B cell activation. Nature. 1995;374:546–9. doi: 10.1038/374546a0. [DOI] [PubMed] [Google Scholar]
- 5.Yi AK, Chang M, Peckham DW, Krieg AM, Ashman RF. CpG oligodeoxyribonucleotides rescue mature spleen B cells from spontaneous apoptosis and promote cell cycle entry. J Immunol. 1998;160:5898–906. [PubMed] [Google Scholar]
- 6.Hartmann G, Weeratna RD, Ballas ZK, et al. Delineation of a CpG phosphorothioate oligodeoxynucleotide for activating primate immune responses in vitro and in vivo. J Immunol. 2000;164:1617–24. doi: 10.4049/jimmunol.164.3.1617. [DOI] [PubMed] [Google Scholar]
- 7.Roberts TL, Sweet MJ, Hume DA, Stacey KJ. Cutting Edge. Species-specific TLR9-mediated recognition of CpG and non-CpG phosphorothioate-modified oligonucleotides. J Immunol. 2005;174:605–8. doi: 10.4049/jimmunol.174.2.605. [DOI] [PubMed] [Google Scholar]
- 8.Ashman RF, Goeken JA, Drahos J, Lenert P. Sequence requirements for oligodeoxyribonucleotide inhibitory activity. Int Immunol. 2005;17:411–20. doi: 10.1093/intimm/dxh222. [DOI] [PubMed] [Google Scholar]
- 9.Stunz LL, Lenert P, Peckham D, Yi AK, Haxhinasto S, Chang M, Krieg A, Ashman RF. Inhibitory oligonucleotides specifically block effects of stimulatory CpG oligonucleotides in B cells. Eur J Immunol. 2002;32:1212–22. doi: 10.1002/1521-4141(200205)32:5<1212::AID-IMMU1212>3.0.CO;2-D. [DOI] [PubMed] [Google Scholar]
- 10.Bauer S, Kirschning CJ, Hacker H, Redecke V, Hausmann S, Akira S, Wagner H, Lipford GB. Human TLR-9 confers responsiveness to bacterial DNA via species-specific CpG motif recognition. Proc Natl Acad Sci USA. 2001;98:9237–42. doi: 10.1073/pnas.161293498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Traganos F, Darzynkiewicz Z, Sharpless T, Melamed MR. Simultaneous staining of ribonucleic acid and deoxyribonucleic acid in unfixed cells using acridine orange in a flow cytometric system. J Histochem Cytochem. 1977;25:46–56. doi: 10.1177/25.1.64567. [DOI] [PubMed] [Google Scholar]
- 12.Lenert P, Rasmussen W, Ashman RF, Ballas ZK. Structural characterization of the inhibitory DNA motif for the type A (D-CpG-induced) cytokine secretion and NK-cell lytic activity in mouse spleen cells. DNA Cell Biol. 2003;22:621–31. doi: 10.1089/104454903770238094. [DOI] [PubMed] [Google Scholar]
- 13.Vollmer J, Weeratna R, Jurk M, et al. Oligodeoxynucleotides lacking CpG dinucleotides mediate Toll-like receptor 9 dependent T helper type 2 biased immune stimulation. Immunol. 2004;113:212–23. doi: 10.1111/j.1365-2567.2004.01962.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Yu D, Kandimalla ER, Bhagat L, Tang JY, Cong Y, Tang J, Agrawal S. Immunomers-novel 3′−3′ linked CpG oligodeoxribonucleotides as potent immunomodulatory agents. Nucl Acids Res. 2002;30:4460–9. doi: 10.1093/nar/gkf582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Datta SK, Cho HJ, Takabayashi K, Horner AA, Raz E. Antigen-immunostimulatory oligonucleotide conjugates: mechanisms and applications. Immunol Rev. 2004;199:217–26. doi: 10.1111/j.0105-2896.2004.00149.x. [DOI] [PubMed] [Google Scholar]
- 16.Stacey KJ, Young GR, Clark F, Sester DP, Roberts TL, Naik S, Sweet MJ, Hume DA. The molecular basis for the lack of immunostimulatory activity of vertebrate DNA. J Immunol. 2003;170:3614–20. doi: 10.4049/jimmunol.170.7.3614. [DOI] [PubMed] [Google Scholar]
- 17.Pasare C, Medzhitov R. Toll-like receptors: balancing host resistance with immune tolerance. Curr Opin Immunol. 2003;15:677–92. doi: 10.1016/j.coi.2003.09.002. [DOI] [PubMed] [Google Scholar]
- 18.Zitvogel L. Dendritic and natural killer cells cooperate in the controls/switch of innate immunity. J Exp Med. 2002;195:F9–14. doi: 10.1084/jem.20012040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Gursel I, Gursel M, Yamada H, Ishii KJ, Takeshita F, Klinman DM. Repetitive elements in mammalian telomeres suppress bacterial DNA-induced immune activation. J Immunol. 2003;171:1393–400. doi: 10.4049/jimmunol.171.3.1393. [DOI] [PubMed] [Google Scholar]
- 20.Lenert P, Stunz L, Yi A, Krieg AM, Ashman RF. CpG stimulation of primary mouse B cells is blocked by inhibitory oligodeoxyribonucleotides at a site proximal to NF-κB activation. Antisense Nucl Acid Drug Dev. 2001;11:247–56. doi: 10.1089/108729001317022241. [DOI] [PubMed] [Google Scholar]
- 21.Rutz M, Metzger J, Gellert T, Luppa P, Lipford GB, Wagner H, Bauer S. Toll-like receptor 9 binds single-stranded CpG-DNA in a sequence- and pH-dependent manner. Eur J Immunol. 2004;34:2541–50. doi: 10.1002/eji.200425218. [DOI] [PubMed] [Google Scholar]
- 22.Latz E, Schoenemeyer A, Visintin A, et al. TLR9 signals after translocating from the ER to CpG DNA in the lysosome. Nat Immunol. 2004;5:190–8. doi: 10.1038/ni1028. [DOI] [PubMed] [Google Scholar]