Minimal Sequence Requirements for Oligodeoxyribonucleotides Activating Human TLR9

This information is current as Jelka Pohar, Alenka Kuznik Krajnik, Roman Jerala and of October 2, 2021. Mojca Bencina J Immunol 2015; 194:3901-3908; Prepublished online 16 March 2015; doi: 10.4049/jimmunol.1402755 http://www.jimmunol.org/content/194/8/3901 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2015/03/14/jimmunol.140275 Material 5.DCSupplemental http://www.jimmunol.org/ References This article cites 44 articles, 14 of which you can access for free at: http://www.jimmunol.org/content/194/8/3901.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on October 2, 2021

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2015 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Minimal Sequence Requirements for Oligodeoxyribonucleotides Activating Human TLR9

Jelka Pohar,*,1 Alenka Kuznik Krajnik,*,1 Roman Jerala,*,† and Mojca Bencina*,†

Synthetic oligodeoxyribonucleotides (ODNs) containing CpG (unmethylated deoxycytidylyl-deoxyguanosine dinucleotide) motifs activate endosomal TLR9. The nucleotide sequence, length, and dimerization properties of ODNs modulate their activation of TLR9. We performed a systematic investigation of the sequence motifs of B-class and C-class phosphodiester ODNs to identify the sequence properties that govern TLR9 activation. ODNs shorter than 21 nt and with the adenosine adjacent to the cytidine- guanosine (CG) dinucleotide motif led to a significant loss of the propensity to activate TLR9. The distance between the stimulatory CpG motifs within the ODN fine-tunes the activation of B cells. The minimal ODNs that activate human TLR9 comprise 2 CG dinucleotides separated by 6–10 nt, where the first CpG motif is preceded by the 59-thymidine and the elongated poly-thymidine tail at the 39 end of the ODN. The minimal sequence provides insight into the molecular mechanism of TLR9 ligand recognition. On the basis of sequence requirements, we conclude that two binding sites with different affinities for CG Downloaded from are formed in the human TLR9 dimer, with a very stringent binding site interacting with the 59 CpG motif. The Journal of Immunology, 2015, 194: 3901–3908.

ingle-stranded DNAwith CpG (unmethylated deoxycytidylyl- ulators of B cell proliferation as well as the secretion of IL-6, IL- deoxyguanosine) dinucleotide motifs, characteristic of 10, and IL-12 through NF-kB signaling (1, 11). A-class ODNs S bacteria and virus genomes, activates the immune response (alias D-type), which are poor stimulators of B cells, contain http://www.jimmunol.org/ against invading pathogens (1). The frequency of the cytidine- a palindromic sequence at the center of the ODN and poly-G guanosine (CG) dinucleotide is strongly underrepresented in the motif, which can form quadruplexes at each end. These ODNs human genome, and ∼80% of the cytidine in the CG dinucleotide activate NK cells, and in pDCs, A-class ODNs induce secretion of is methylated, which establishes the recognition of DNA-comprising type I IFNs (IFN-a/b), TNF-a, IL-12, and IFN-g–IP10 (12–14). CpG motifs as the structural pattern for the innate immune response. C-class ODNs are chimeras of A- and B-class ODNs with one or Synthetic ssDNA fragments, oligodeoxyribonucleotides (ODNs) two CGT motifs at the 59 end and a CG-rich palindromic sequence containing stimulatory CpG motifs, also initiate immune signaling at the 39 end of the ODN. This class of CpG ODNs possesses the in a TLR9-MyD88–dependent fashion, leading to the activation immune-stimulatory properties of the A- and B-class ODNs and of transcription factors NF-kB and IRF7 and, consequently, the induces proliferation of B cells and the production of low amounts by guest on October 2, 2021 secretion of and chemokines by B cells, plasmacytoid of IFN-a from pDCs (15, 16). A-, C-, and P-class (17, 18) CpG dendritic cells (pDCs), NK cells, and macrophages that express ODNs form intra- and intermolecular duplexes. These multi-ODN TLR9 (2, 3). Membrane receptors—mannose receptor 1 (4), Ig- structures are suggested to take a different route of endocytosis like receptor KIR3DL2 (5), multilectin receptor DEC-205 (6), and and, therefore, induce the synthesis of type I IFNs (17, 19). In scavenger receptor CXCL16 (7)—enhance internalization and addition to unmethylated CpG motifs within the ODN, other define the route of endocytosis of ODNs. factors determine the immune response of TLR9-expressing cells. Several types of CpG ODNs, each with a distinct sequence, The position, number, and accessibility of CpG motifs, the secondary structure, and specificity for TLR9 activation (8), are nucleotides adjacent to the CG dinucleotide, and the secondary potent activators of Th1-type cytokines and so are used as adju- structures of the ODNs influence agonist potency for TLR9 acti- vants (9, 10). The B-class CpG ODNs (alias K-type) with the CG vation (15, 20–23). Moreover, CpG ODNs show sequence-species motif and a phosphorothioate (PTO) backbone are strong stim- specificity for the activation of TLR9. GACGTT has been pro- posed as the optimal TLR9 recognition hexamer for mice (1, 24), *Department of Biotechnology, National Institute of Chemistry, SI-1000 Ljubljana, whereas GTCGTT is optimal for humans (12) and other vertebrate Slovenia; and †EN-FIST Centre of Excellence, SI-1000 Ljubljana, Slovenia species (25). 1J.P. and A.K.K. share first authorship. The naturally occurring agonist of TLR9 is ssDNA with a Received for publication October 30, 2014. Accepted for publication February 8, phosphodiester (PD) backbone. To prevent degradation of ODNs 2015. by deoxyribonucleases, synthetic ODNs may consist of a partial or This work was supported by programs and projects from the Slovenian Research complete PTO backbone. Some important differences are noted in Agency and the EN-FIST Centre of Excellence and partially financed by European Union Structural Funds. terms of TLR9 activation between ODNs based on a PD backbone and those based on a PTO backbone. A base-free PD-deoxyribose Address correspondence and reprint requests to Dr. Roman Jerala and Dr. Mojca Bencina, National Institute of Chemistry, Hajdrihova 19, SI-1000 Ljubljana, Slovenia. homopolymer is a very weak TLR9 agonist (26), whereas the base- E-mail addresses: [email protected] (R.J.) and [email protected] (M.B.) free PTO homopolymer does not activate TLR9 but instead The online version of this article contains supplemental material. competitively inhibits TLR9 activation in the presence of B-class Abbreviations used in this article: CG, cytidine-guanosine; ECD, ectodomain; h, ODN, which is not the case for the PD homopolymer (26). human; HA, hemagglutinin; ODN, oligodeoxyribonucleotide; PD, phosphodiester; In this study, we aimed to define the sequence and functional pDC, plasmacytoid ; PTO, phosphorothioate. features of ODNs that govern TLR9 activation and the minimal Copyright Ó 2015 by The American Association of Immunologists, Inc. 0022-1767/15/$25.00 sequence requirements for strong ODN stimulation of the immune www.jimmunol.org/cgi/doi/10.4049/jimmunol.1402755 3902 MINIMAL ODN AS AGONIST OF HUMAN TLR9 response in human B cells and human TLR9-expressing HEK293 HEK293T cells seeded onto eight-well tissue culture chambers (Ibidi) at 2.2 3 105 cells per well were transfected with the plasmids expressing cells. We systematically analyzed which sequence motifs in B-class HA ODN2006PD and C-class ODN2395PD are necessary to effectively hTLR9 (140 ng DNA per well) and an EEA1-mTomato (80 ng). Lo- calization was visualized 48 h post transfection. The hTLR9HA was stained activate TLR9. The impact of the nucleotide context and the using rabbit anti-HA (H6908, 1:50; Sigma-Aldrich) and secondary Abs distance between the CpG motifs, as well as the effect of ODN Alexa Fluor 647 F(ab9)2 fragment of goat anti-rabbit IgG (H + L) (A21246, duplex formation on TLR9 activity, were determined. Finally, 10 mg/ml; Invitrogen). Successive images excited at 543 nm (fluorescence we established the minimal ODNs defined by the sequence emission, 560–620 nm) and 633 nm (fluorescence emission, 650–700 nm) . were captured using the Leica TCS SP5 inverted laser-scanning micro- TCG[T]6–10CG[T]9–19 and a total length of 21 nt. These mini- scope on a Leica DMI 6000 CS module equipped with an HCX Plane- mal ODNs induce production of TNF-a in human PBMCs. Apochromat l blue 63 3 oil-immersion objective with NA 1.4 (Leica Microsystems). Images were processed with LAS AF software (Leica Microsystems) and ImageJ software (National Institute of Mental Health, Materials and Methods Bethesda, MD). Cell cultures and reagents AlphaScreen TLR9 binding assay Human PBMCs and B cells were cultivated in RPMI 1640 medium (Life Technologies, Invitrogen) supplemented with 10% (v/v) heat-inactivated The binding assays were performed in a final reaction volume of 25 ml assay FBS (Life Technologies, Invitrogen) at 37˚C in 5% CO2. B lymphocyte buffer, which contained 50 mM MOPS-NaOH (pH 6.5), 100 mM NaCl, Ramos-Blue cells (InvivoGen), which stably expressed an NF-kB/AP-1– 0.01% (w/v) Tween 20, and 1 g/l BSA in a half 96-well plate (Perkin- inducible secreted embryonic alkaline phosphate (SEAP) reporter gene, Elmer). The ODNs labeled with biotin-TEG at 39 (Integrated DNA Tech- were cultured in IMDM (Life Technologies, Invitrogen) supplemented nologies) were used. In the assay, 5 ml cell lysate (prepared as described with 10% (v/v) heat-inactivated FBS (Life Technologies, Invitrogen) at below) was incubated with an ODN (#75 mM final concentration) for 1 h

37˚C in 5% CO2. Human embryonic kidney cells HEK293, transfected at 23˚C. Anti-HA acceptor beads (20 mg/ml final concentration) were Downloaded from with human (h) TLR9 and used as control to confirm the results obtained added and incubated for 1 h at 30˚C. Then, streptavidin-coated donor beads with the Ramos-Blue cells, were cultivated in DMEM (Life Technologies, (20 mg/ml final concentration) were added and incubated for 30 min at Invitrogen) supplemented with 10% (v/v) heat-inactivated FBS (Life 30˚C before the signals were measured with EnVision (PerkinElmer). HA ectodomain (ECD) Technologies, Invitrogen) at 37˚C in 5% CO2. The binding of ODN to wt-hTLR9 or wt-hTLR9 -Fc after PD Cells were treated with different CpG-ODNs: ODN2006 (B-type), subtraction of intrinsic activity (no ODN) are shown as AlphaScreen units PD ODN2395 (human/mouse C-type), their variants, and other ODNs all (all n =3,6 SD; two independent experiments). with the PD backbone, synthesized by Sigma-Aldrich and Integrated DNA HEK293T cells were transiently transfected with plasmid expressing

PTO PTO PTO http://www.jimmunol.org/ Technologies. The ODN2006 , ODN2395 , and minH ODNs with hTLR9HA (735 mg DNA per well) using Lipofectamine Transfection Re- the PTO backbone were from Integrated DNA Technologies. ODN dimers agent (Invitrogen). After 48 h, cells were lysed in 0.2 ml lysis buffer [50 were formed by mixing pairs of ODNs, heating to 95˚C, and then slowly mM Tris-HCl (pH 7.4), 1 mM EDTA, 150 mM NaCl, 0.5% Triton X-100, cooling down before use. 20% glycerol, 1 mM Na3VO4, 25 mM NaF, and 1 mM PMSF] containing a cOmplete Mini Protease Inhibitor (Roche) for 10 min on ice. Then, cell Activity assay lysates were centrifuged (10.000 rpm, 10 min, 4˚C). 3 5 Human B lymphocyte Ramos-Blue cells were seeded at a density of 2 10 Internalization of ODNs cells per well onto Costar 96-well plates (Corning). The cells were im- mediately stimulated with ODNs (5 mM final concentration, if not stated The Ramos-Blue cells at 1 3 106 cells per well were treated with ODNs otherwise). After 20 h, the supernatants were collected and heated to 56˚C conjugated at 39 with fluorescent dye (1 mM final concentration): minM- for 1 h, and then NF-kB/AP-1 activation linked to secreted alkaline 75ATTO647 and ODN2006ATTO488 or polyTATTO488. As negative control, by guest on October 2, 2021 phosphatase was determined using a QUANTI-Blue reagent according to untreated cells were used. At 5 h after treatment, the cells were washed, the manufacturer’s instructions (InvivoGen). Normalized alkaline phos- and internalization of ODNs was analyzed using flow cytometry and cells phatase was calculated relative to the ODN2006 output. Error bars rep- were imaged with confocal microscopy. Successive images excited at resent the SD obtained from three experimental replicates. The data are 488 nm (fluorescence emission, 500–570 nm) and 633 nm (fluorescence representative of at least two independent experiments. emission, 650–700 nm) were captured. HEK293 cells were harvested from actively growing culture and plated onto Costar Clear White 96-well plates (Corning) at 3 3 104 cells per Fluorescence spectroscopy well. After 24 h at 70% confluency, they were transfected with a mixture of jetPEI reagent and plasmid DNA according to the manufacturer’s The dsDNA in solution was analyzed using the Quant-iT dsDNA High- instructions (PolyPlus Transfection). For the TLR9-expressing cells, we Sensitivity Assay Kit (Molecular Probes, Invitrogen). A PicoGreen dye, used 50 ng pUNO-hTLR9-HA plasmid (InvivoGen), 150 ng ELAM1-luc when bound to dsDNA, fluoresced at 523 nm when excited with 502-nm reporter plasmid expressing firefly luciferase under control of the NF-kB light. The formation of dsDNA was determined with PicoGreen dye at 37˚C promoter, and 5 ng phRL-TK (Promega) transfection control plasmid (or at 22˚C) with the Synergy Mx (BioTek) spectrofluorimeter according to expressing Renilla luciferase. At 24 h after transfection, the culture me- the manufacturer’s guidelines. dium was replaced with fresh DMEM with 10% FBS, and the cells were treated with ODNs (5 mM final concentration). After 18–20 h, the cells detection were lysed in passive lysis buffer (Promega), and the expression of the Human PBMCs were extracted from whole blood using density gradient luciferase reporter genes was analyzed using Dual-Glo Luciferase Assay medium Lymphoprep and SepMate tubes (STEMCELL Technologies). System reagents (Promega) and the Orion luminometer plate reader PBMCs were cultured in 96-well round-bottom plates (0.2 ml per well; 0.5– (Berthold Detection Systems). Relative luciferase activity was calculated 1 3 106 cells per well) and stimulated with ODNs. Culture supernatants by normalizing each sample’s firefly luciferase activity with the constitu- were collected after 16 h. The secreted human TNF-a and IFN-g were tive Renilla luciferase activity determined within the same sample. Nor- determined with ELISA (human TNF-a and human IFN-g Ready-SET-Go malized relative luciferase activity was calculated relative to the ODN2006 ELISA; eBioscience). output. Error bars represent the SD obtained from three experimental Untouched B cells were isolated from PBMCs by negative selection with replicates. The data are representative of at least three independent CD43 beads (Miltenyi Biotec). B cells were cultured in 96-well round- experiments. bottom plates (0.5–1 3 105 cells in 0.1 ml per well) and stimulated with ODNs. Cells were treated with human FcR blocking reagent and stained Immunoblotting and confocal microscopy with anti-hCD19–FITC, anti-hCD20–VioBlue and anti-hCD69–PE according to the manufacturer’s instructions (Miltenyi Biotec). HEK293T cells were transiently transfected with plasmids expressing HA hTLR9 (1470 ng DNA per well) and mCerulean (10 ng DNA per well) Software and statistics (for transfection control) using Lipofectamine 2000 Transfection Reagent (Invitrogen). The expression of hTLR9 and mCerulean was determined Program IDT OligoAnalyzer 3.1 (www.idtdna.com/analyzer/applications/ with primary rabbit anti–hemagglutinin (HA) Abs diluted 1:1000 (H6908; oligoanalyzer/) using the program settings, assuming a temperature of Sigma-Aldrich); rabbit anti-GFP diluted 1:1000 (A11122; Invitrogen); and 37˚C, 150 mM Na+, and 0.5 nM Mg2+, was used to calculate the Gibbs free secondary Abs goat anti-rabbit IgG-HRP diluted 1:4000 (ab6721; Abcam). energy of the ODN homodimer. Graphs were prepared with Origin 8.1 The Journal of Immunology 3903

(www.originlab.com/), and GraphPad Prism 5 (www.graphpad.com/) was When the guanosines within the GG, GC, and CG dinucleotides, used for statistics. The Student unpaired two-tailed t test was used for except the 59-CG, were substituted with C, the efficacy in acti- statistical comparison of the data. vating NF-kB signaling decreased significantly (Supplemental Fig. 2A; CG1-l32-cont, CG6-l32-cont, and CG8-l40-cont). The re- Results sults confirmed the crucial role of the 59-CG and the requirement PD CpG motif at the 59 end of the ODN is essential for TLR9 of at least another CpG within the ODN to activate B cells and activation hTLR9-expressing HEK293 cells. The number and position of the CpG motifs of ODN2006PD and The adenosine adjacent to the CG dinucleotide reduces the ODN2395PD were varied to determine the minimal sequence efficacy of ODNs necessary for activating the immune response of B cells and For human cells, the species-specific motif that activates TLR9 was PD hTLR9-transfected HEK293 cells. B-class ODN2006 and, to proposed to be GTCGTT (12), whereas GACGTT was proposed PD a lesser degree ODN2395 (C-class), activated NF-kB/AP-1 of for mice (1), (24). Therefore, we analyzed the consequences of the Ramos-Blue cells (Fig. 1). The well-established B-type agonist replacing thymidine with the adenosine adjacent to the CG in the ODN2006 consists of three CpG-comprising motifs: a 59 located CpG motifs on TLR9 activation. These substitutions decreased TCGTCGT motif with two CG dinucleotides, followed by the TLR9 activation. In particular, the adenosine adjacent to the first middle GTCGT sequence, and a GTCGT sequence at the 39 end. CG dinucleotide decreased NF-kB/AP-1 activity by .70% We substituted one or more CG sequences with TT, and TLR9 (Fig. 2, ODN-23, -11, -25). Several adenosines are present within activation (NF-kB/AP-1) of the Ramos-Blue cells was measured the sequence of long ODNs; however, the adenosines that are

PD Downloaded from after the cells were treated with ODN2006 and its variants remote from the CpG motifs in CG6-l32-woA and CG8-l40-woA (Fig. 1). A minimum of the 59-CGTCG motifs and another CG had almost no influence on the stimulation (Supplemental Fig. 2B). dinucleotide were required for efficient TLR9 activation (Fig. 1A, Nonetheless, substitution of five adenosines within the CG1-l32- ODN-57, -13 compared with -58). The impaired activation of woA improved TLR9 activation. This finding suggests that the TLR9 was detected with the 16-nt-long ODN (Fig. 1A, ODN-14). A adjacent to 59-CG had a strong negative effect on activation of A marked decrease in TLR9 activation was observed for the ODN the hTLR9, whereas the adenosines placed elsewhere had only variants featuring substitution of the 59-CG dinucleotide with TT, a minor impact. http://www.jimmunol.org/ GG, or GC, even if the other CG dinucleotides remained intact (Fig. 1B, ODN-3, -19, -20). Similar results for TLR9 activation by Separation of the CpG motifs modulates TLR9 activation the ODN variants (Supplemental Fig. 1A, 1B) were also obtained Initial analysis of the CpG motifs required for TLR9 activation using hTLR9-expressing HEK293 cells (Supplemental Fig. 1C suggested that the distance between the CpG motifs is not a par- and 1D demonstrates localization and expression). ticularly critical parameter (Fig. 1A, ODN-13, -57). To elaborate We also examined the impact of the CpG motif (59-TCGTCGT) the requirements for the distance between the 59, middle, and 39 at the 59 end of the C-type ODN2395PD on hTLR9 activation. CpG motif in more detail, we designed ODN2006PD variants with Substitution of the first CG with TT or GC led to a strong decrease two to nine thymidines between the CpG motifs. As a conse- in TLR9 activation (Fig. 1C, ODN-C2, -C7 compared with quence of the thymidine insertions, the lengths of the ODNs by guest on October 2, 2021 ODN2395PD), whereas the second CG dinucleotide was much less varied from 20 to 34 nt (Fig. 3A). The ODN-5 with only two important (ODN-C3, -C8 compared with ODN2395PD). However, thymidines between the CpG motifs (20 nt long) preserved 60% no TLR9 activation was observed when both 59-CGs were TLR9 activity. Insertion of three to eight thymidines between CpG substituted with TT (ODN-C9). motifs (ODN-7, -8, -6) preserved or even slightly increased the We examined the effect of the increased number of CG motifs ability of the agonist to activate TLR9 (Fig. 3A). Similarly, an by designing longer ODNs: 32-nt-long ODNs CG1-l32-F and CG6- insertion of seven or nine thymidines between the 59 CpG motif PD l32-F, and 40-nt-long CG8-l40-F with 1, 6, and 8 CG dinucleotides, and the CG-rich palindrome of ODN2395 improved the TLR9 PD respectively. The activity of CG1-l32-F with only a single CpG was activation compared with that of ODN2395 (Fig. 3B). PD still 50% of the activity of ODN2006 . The CG6-l32-F and CG8- Initial screening (Fig. 3C) suggested that only two CpG motifs l40-F with six and eight CGs activated TLR9 of the Ramos-Blue within the ODN are sufficient to activate B cells. Therefore, we cells better than the reference ODN2006PD (Supplemental Fig. 2A). analyzed the minimal number of CG dinucleotides and their po-

FIGURE 1. The position and number of CpG motifs define the potency of ODNs to activate TLR9. Ramos-Blue cells were stimulated with (A) ODN2006PD variants with individual CpG motifs substituted with thymidines; (B) ODN2006PD variants with individual CpG motifs substituted with GG, GC, or TT; and (C) ODN2395PD variants with two 59 CpG motifs substituted with TT or GC. The predicted dimer is underlined. **p , 0.05. ns, p . 0.5. 3904 MINIMAL ODN AS AGONIST OF HUMAN TLR9

from C-type to B-type ODN, which is a more potent activator of B cells (Fig. 4A). To demonstrate that the decrease in activity is due to the for- mation of the duplex and the decreased content of single-stranded DNA rather than to a particular sequence, we used a mixture of long ODNs with their reverse complement sequences. Other TLR9 stimulatory ODNs reported in the literature that contain a duplex all formed homodimers. ODNs that form a heterodimeric duplex over the 39 segment of the ODN with the remaining single-stranded 59- TCGT motif showed a negative impact of ODN dimerization on FIGURE 2. The thymidine adjacent to the 59 CpG motif is important for TLR9 activation. Although each of the components (CG1-l32-F/R, PD TLR9 activation. Ramos-Blue cells were stimulated with ODN2006 CG6-l32-F/R, CG8-l40-F/R) stimulated TLR9, the mixture of each variants with adenosine next to the CpG motif. ***p , 0.005. strongly impaired TLR9 activation (Supplemental Fig. 2E) (15, 28), with ,30% TLR9 activity compared with that of ODN2006PD or the activity of individual monomers (Supple- sition within the ODN that promote TLR9 activation comparable mental Fig. 2F). This result reflects the transition from B- to to that of ODN2006PD. Substitution of the second CG within the C-type ODN, demonstrating the preference for single-stranded 59 CpG motif (59-TCGTCGT) of ODN2006 with TT, accompa- conformation over the entire length of the ODN. nied by the replacement of the 39 CpG (GTCG) with TTTT, was The minimal sequence requirements of ODNs for TLR9 Downloaded from well tolerated, with only a minor decrease in TLR9 activation activation (Fig. 3D, ODN-53, -54). To summarize, only two CG dinucleo- tides within the ODN were required for receptor activation. We On the basis of these results, we selected pairs of CG dinucleotides also determined that the 59-CG dinucleotide and the CpG motif and systematically modified their position and distance within separated by four to eight thymidines were well tolerated. the ODN. An ODN comprising only two TCGT motifs had $80% of the activity of ODN2006PD (Fig. 5A, ODN-46). Shortening of k Dimer formation decreases NF- B/AP-1 activation the poly-T tail at the 39 end strongly decreased activity (Fig. 5B, http://www.jimmunol.org/ The ssDNA is a ligand for TLR9 (1); however, intra- and inter- ODN-74, ODN-73). Any other nucleotide except T adjacent to the molecular secondary-structure–forming ODNs stimulate type I 59-CG sequence considerably impaired the activation of TLR9 IFN secretion (12–16, 19, 27). ODN2395 forms a homodimer (Fig. 5E, ODN-86 U,-87 deoxyribose backbone, -83, -64, -82). In mediated by the 39-CG–rich palindromic sequence. For the addition, shifting 59-CG farther toward the 39 end, but retaining ODN2395 dimer, the predicted DG was 236.75 kcal/mol. For- the length of the ODN, led to a stepwise decrease in receptor mation of the dsDNA was monitored using PicoGreen, a dsDNA- activity (Fig. 5C). Therefore, the 59-CG dinucleotide must be specific dye. Variants of ODN2395PD were prepared with mod- positioned next to the 59 thymidine. As seen in the ODN2006PD ifications in the palindromic segment of ODN2395, rendering variants (Fig. 1A, ODN-20; Fig. 1C, ODN-C9,-C2,-C7; Supplemental the duplex less stable (Fig. 4A, ODN-C4, -C6, -C10, -C5; Fig. 2A, long ODNs), as well as for minH, at least two CG by guest on October 2, 2021 Supplemental Fig. 2C, TLR9-expressing HEK293 cells). As ex- dinucleotides were required for TLR9 activation (Fig. 5D). The pected, gradual replacement of the palindromic sequence with length of minH and the distance between the CG dinucleo- thymidines decreased the intensity of fluorescence, which indi- tides fine-tune the ODN’s activation potency (Fig. 5F, 5G; cates the disrupted stability of the dimer (Fig. 4B, 4D). We ob- Supplemental Fig. 3A, 3B). The CG dinucleotide motifs should be served a negative correlation between the stability of the dimer separated by 6–10 nt for the best activity (Fig. 5F). This finding formation and hTLR9 activation at 37˚C and 22˚C (Pearson’s was also confirmed for minH ODNs that are 30 nt long (Fig. 5G). correlation coefficient 20.88 and 20.65, respectively) (Fig. 4C, We replaced the poly-T spacer between the CG motifs and the 4D; Supplemental Fig. 2D). This finding reflects the transition 39 poly-T tail with poly-C, G, or A sequences. These nucleotide

FIGURE 3. The number and position of CpG motifs within ODN fine-tune TLR9 activation. Ramos-Blue cells were stimulated with (A) ODN2006PD and its variants with two to nine thymidines between the CpG motifs; (B) ODN2395PD and its variants with four to nine thymidines between the CpG motif and the CG palindrome; (C) ODN2006PD variants with 59-CG and an additional CG dinucleotide; and (D) ODN2006PD variants with four to eight thymidines between 59-CG and an additional CG dinucleotide. The predicted dimer is underlined. *p , 0.5. ns, p . 0.5. The Journal of Immunology 3905

FIGURE 4. Dimerization of ODNs impairs activation of TLR9 in B lymphocytes. (A) Ramos-Blue cells were stimulated with ODN2395PD and its variants with a modified palindrome motif. (B) Scheme of possible homodimers of ODN2395PD and its variants. Correlation between TLR9 activation and (C) calculated DG and (D) dimer formation in solution measured with the PicoGreen fluorescence dye of ODN2395PD and its variants (measured at 37˚C). The predicted dimer is underlined. Downloaded from substitutions considerably diminished the potency of minH to triggered the expression of the cell surface marker CD69 on activate TLR9 (Fig. 5H). Cytidines replacing the poly-T spacer or B cells (Fig. 6D). tail had a less detrimental effect than the larger purines A and G. In summary, the minimal sequence requirement for ODNs to Consequently, the poly-T tail at the 39 end should be between stimulate human TLR9 is two CpG motifs, separated preferably 9 and 19 nt long. As determined by AlphaScreen assay, the with the oligo-T spacer and the oligo-T tail at the 39 end (Fig. 6E). 24-nt-long ODNs with different numbers of CG dinucleotides For efficient activation, the CG motifs should be separated by http://www.jimmunol.org/ (ODN2006, minH-75) bound TLR9 with similar affinity regard- between 4 and 10 nt, and the ODN should be longer than 21 nt. less of the sequence (Supplemental Fig. 4A). The minH-30 with The sequence requirement for the 59-TCGT motif is very stringent only the 59 TCG motif and the minH-64 with the ACG motif at the because the substitution of the T next to CG with any other nu- 59 end bound TLR9 with lower affinity. The 15-nt-long ODN cleotide strongly reduced the ability to activate TLR9. The (minH-70) and poly-T did not bind to TLR9. The internalization requirements for the second CG motif and nucleotides adjacent to of ODNs (ODN2006 and minH-75 or poly-T) to B cells was not the CG dinucleotide are less rigorous, and substitutions of the sequence dependent, and 5 h after ODN treatment, all Ramos- nucleotides next to CG only mildly affect the efficiency of acti- Blue cells contained both ODNs (Supplemental Fig. 4B, 4C). To vating human TLR9. These facts suggest two probable binding by guest on October 2, 2021 summarize, the minimal sequence of minH that activates TLR9 is sites within the TLR9 dimer with different binding affinities for TCG[T]6–10CG[T]10–19, with the length of the ODN ranging from the CG dinucleotide: one that binds the 59-TCG motif and one that 21 to 30 nt. should bind less stringently.

Response of human PBMCs and B cells to the minimal ODN Discussion Because internalization and digestion might play a role in Synthetic CpG ODNs are potent stimulators of immune cells TLR9 activation, we tested in parallel the activation of TLR9 by expressing TLR9; however, different classes of ODNs that differ in minH ODN with PD and PTO backbones (Fig. 6A). The responses terms of their sequence elicit diverse cytokine patterns (1, 11, 15, of the B cells triggered by minH ODN and ODN2006 were 16). In this study, we identified the minimal sequence motifs of B- comparable. It was found that 5–10 times less ODN with the PTO class ODNs that effectively activate B cells, PBMCs, and HEK293 backbone was needed for the full activation of TLR9 than was cells transfected with human TLR9. We identified minH by the needed for ODNs with the PD backbone (Fig. 6A), suggesting that sequence pattern TCG[T]6–10CG[T]9–19 that contains two CG the recognition site is optimized for PDs. sequences separated with a 6- to 10-nt-long thymidine spacer. The Substitution of the PTO backbone with the PD backbone for only first CpG motif starts at the 59 end of ODN, and at the 39 end, the 59-CG dinucleotide in the P-class ODN triggers stronger TLR9 a poly-T tail is required. The minH ODN should be longer than signaling (17). Therefore, we examined whether the type of 21 nt for the efficient activation of TLR9, which might be due backbone determines the efficiency of TLR9 activation. The minH to either increased protection against exonucleases or sequence- ODNs with mixed PD-PTO backbone content were tested on independent binding to the receptor. The minHPD ODN activates B cells. The minH with the PD-linked 59-TCGT tetranucleotides, NF-kB/AP-1 signaling in B cells as efficiently as ODN2006PD. PTO PD with the rest of the nucleotides linked by PTO (minH-75 -CG1 ), Although we cannot exclude the possibility that the length re- was the most efficient activator of B cells (Fig. 6B). quirement and the addition of the 39 oligo-dT tail may be due to Finally, we analyzed minH ODN efficiency to promote a cyto- stability against the exonuclease, the similar activation of the kine response in human PBMCs and B cells (CD43-negative) PTO-based activation by minH and ODN2006 suggests that this is (Supplemental Fig. 4D, CD19- and CD20-positive cells). Min- not the case. HPTO was as efficient as ODN2006PTO in stimulating TNF-a, The 59-CG seems to be most important, but not sufficient, for IFN-g, and IL-6 release from PBMCs. The minHPD also triggered TLR9 activation, as previously determined in the design of the the secretion of TNF-a at a 4-fold higher concentration compared generally used ODN2006PTO and ODN2395PTO (12, 23, 29– with the PTO ODNs (Fig. 6C). IFN-a was determined in the 31). The backbone modification to 29-O-methyl ribose within presence of the ODN2006 and minH with PD backbone 16 h after the 59 CpG motif, but not within the 39-CG–rich sequence, of the PBMCs were stimulated. ODN2006PTO and minH-75PTO the C-class ODN led to the complete loss of IFN-a secretion 3906 MINIMAL ODN AS AGONIST OF HUMAN TLR9 Downloaded from http://www.jimmunol.org/ by guest on October 2, 2021

FIGURE 5. The minimal ODN sequence requirements for activation of TLR9. Ramos-Blue cells were stimulated with ODN2006PD and (A)its variants lacking the last CpG motif; (B) minH ODNs with different T-tail lengths at the 39 end; (C) minH ODNs with different T-tail lengths at the 59 end; (D) minH ODNs with only one CG sequence and different T-tail lengths at the 59 end; (E) minH ODNs with different nucleotides adjacent to the CG motifs; (F) minH ODNs, 24 nt long, with separation of CG sequences from 5 to 12 nt; (G) minH ODNs, 30 nt long, with separation of the CG sequences from 4 to 24 nt; and (H) minH ODNs with a different 39 tail sequence and a different spacer sequence between the CG motifs. *p , 0.5. ns, p . 0.5; d, deoxyribose.

(30), confirming the importance of 59-CG and deoxyribose in Nucleotides adjacent to the CG dinucleotide are also important the activation of TLR9. Moreover, CpG ODNs containing a 39 for agonist potency and are an integral part of the CpG motif (12), hairpin structure, but not the 59, are immunostimulatory (20), probably reflecting the binding specificity of the site on the TLR9 which is consistent with the finding that the accessible 59 CpG ECD. Substitution of the thymidine adjacent to the 59-CG with motif is necessary for activating TLR9. The binding of ODNs any other nucleotide strongly reduced the ability to activate human to TLR9 showed weak sequence dependence, and shorter minH TLR9. In contrast, the T to A substitution next to any other CG ODN and ODN with 59-ACG motif poorly bound TLR9. Latz had only a moderate negative effect on activation. Moreover, the et al. (32), Ashema et al. (33), and Kindrachuk et al. (34) poly-T tail at the 39 end of the minH ODN, which defines the identified that ODNs bind mTLR9 in a sequence-independent length requirement of the ODN, is necessary for efficient activa- manner. However, Rutz et al. (35) demonstrated the sequence- tion. The distance between two essential TCGT motifs fine-tunes dependent binding of ODN to mTLR9. The number of CG TLR9 activation. We showed that TCGT sequences four to eight dinucleotides within the ODN is another parameter that defines thymidines apart in minH and ODN2006PD led to the highest the agonist’s efficacy in activating TLR9 of B cells as well TLR9 activity, as shortening or extending the thymidine spacer as PBMCs. More than two CpG motifs improve TLR9 activa- outside this range decreases the ODN’s efficacy to activate TLR9. tion, which is understood not only for ODN2006PD and The sequence difference in the A-, B-, C-, and P-classes pre- ODN2006PTO (23, 29) but also for long ODNs. The CD32 and dominantly defines the secondary structure of ODNs, which, CD40 with six and eight CG sequences activate B cells better according to the literature, is the main reason for differences in the than ODN2006PD (per mole, but not per mass), although three activation of B cells and pDCs as well as the secretion of different or more CG dinucleotides within a 22- to 25-nt-long ODN does cytokines (36–39). The synthesis of type I IFNs is primarily linked not enhance activity any further (2). to the A-class ODNs (12–14), most likely owing to the different The Journal of Immunology 3907 Downloaded from http://www.jimmunol.org/

FIGURE 6. The minimal ODN activates PBMCs to a similar extent as ODN2006. (A) The minH activates Ramos-Blue cells similarly to ODN2006 based on the same type of polynucleotide backbone. (B) The CpG motif with a PD backbone compared with a PTO backbone is a more potent activator of Ramos- by guest on October 2, 2021 Blue cells. (A and B) The Ramos-Blue cells were stimulated with ODN2006 and minH-75 with a PD or PTO backbone or a combination of PTO and PD backbones. Uppercase letters indicate PD and lowercase letters indicate PTO backbone. (C) Human PBMCs stimulated with ODN2006, minH-75 (PTO or PD) secrete TNF-a, IFN-g, and IL-6. Results of three independent experiments are shown. (D) Human B cells (CD19/CD20-positive) express the cell surface marker CD69 after stimulation with ODN2006PTO and minH-75PTO.(E) The features of the ODN sequence motifs required for activation of TLR9. *p , 0.5, ***p , 0.005. ns, p . 0.5. cellular internalization pathway. The latter was also associated correlation between dimer formation and NF-kB/AP-1 activation with the route of endocytosis and the location of ODNs in dif- was also confirmed with long ODNs and C274 ODN (28). Weaker ferently matured endosomes (14, 40). Nanoparticles loaded with activation by C-type–like ODNs compared with B-type–like B-class CpG ODNs that mimic A- and P-class ODN aggregates ODNs can be explained by binding to only one CpG binding site, are as effective as A-class ODNs at inducing IFN-a in pDCs and leaving the other CpG binding site unoccupied, which leads to less PBMCs (36, 41). This finding implies that the propensity of ODNs effective activation of TLR9. to internalize into early endosomes, rather than the specific DNA There are differences in the TLR9 activation between ODNs binding motif, drives MyD88/IRF7-dependent signaling and IFN-a based on the PD and PTO backbones. Introduction of the nuclease- synthesis (40). The B-class ODNs, like minH-75, internalized resistant PTO backbone improved the ODN’s uptake and hindered sequence independently; however, the route of internalization is degradation by . A disadvantage of PTO modifications probably not linked to IRF7-dependent activation, which would is the induction of an immune response independent of the CpG trigger type I IFNs. The C-class ODN, in addition to the 59 CpG motif (42–44), which obscures the motif of the physiologically motif identical to ODN2006, contains a 39-CG–rich palindrome relevant agonists. The PTO-based minH ODN chimera comprising that participates in forming homodimers (16). The substitutions PD-based 59-TCGT maintained the ODN’s species-specific spec- within the CG-rich palindromic sequence that preserves the pal- ificity while retaining protection of the ODN against nucleases. indrome were well tolerated. We demonstrated that it is possible to Therefore, such chimeric ODNs are as efficient stimulators of design an apparently smooth transition from C- to B-type ODNs immune cells as ODNs with the PTO backbone. by modulating the stability of the palindromic segment. Although The minimal sequence motif of ODN required for TLR9 acti- many ODNs have been investigated, few systematic studies have vation contributes to elucidation of the recognition specificity of been conducted to unravel the key sequence determinants that the ligand–TLR9 interactions. Requirements for the minimal ODN govern activation of TLR9 by ODNs. The substitutions that de- motif suggest that at least two nonequivalent sites at the hTLR9 stroy the palindrome, however, decrease IFN-a secretion and in- ECD dimer must be present to interact with CG dinucleotides. crease the secretion of IL-6 cells in B cells (16). Negative One CG recognition site of TLR9 must recognize the 59-CG motif 3908 MINIMAL ODN AS AGONIST OF HUMAN TLR9 and specifically distinguish thymidine as the first nucleotide. The 18. Struthers, M., A. J. Bett, T. Wisniewski, S. A. Dubey, M. Precopio, W. Jiang, Z. Sun, H. Wang, I. Nowak, M. R. Putta, et al. 2010. Synthesis and immunological other CpG binding site, where the recognition of CG bases is activities of novel agonists of toll-like receptor 9. Cell. Immunol. 263: 105–113. probably less stringent, is located apart from the first CG recog- 19. Yu, D., M. R. Putta, L. Bhagat, M. Dai, D. Wang, A. F. Trombino, T. Sullivan, nition site. The best activation of TLR9 is achieved with ODNs E. R. Kandimalla, and S. Agrawal. 2008. Impact of secondary structure of toll- like receptor 9 agonists on interferon alpha induction. Antimicrob. Agents with CG dinucleotide motifs separated with a spacer of at least six Chemother. 52: 4320–4325. nucleotides, which corresponds to a distance of ∼40–50 A˚ and is 20. Kandimalla, E. R., L. Bhagat, Y.-P. Cong, R. K. Pandey, D. Yu, Q. Zhao, and probably the distance between the binding sites on the TLR9 ECD S. Agrawal. 2003. Secondary structures in CpG oligonucleotides affect immu- nostimulatory activity. Biochem. Biophys. Res. Commun. 306: 948–953. dimer. In addition, the poly-T tail at the 39 end of ODN that is 21. Agrawal, S., and E. R. Kandimalla. 2007. Synthetic agonists of Toll-like $10 nt long probably serves as an initial nonspecific binding of receptors 7, 8 and 9. Biochem. Soc. Trans. 35: 1461–1467. ssDNA to TLR9 through a negative phosphate backbone and a 22. Cong, Y.-P., S. S. Song, L. Bhagat, R. K. Pandey, D. Yu, E. R. Kandimalla, and S. Agrawal. 2003. Self-stabilized CpG optimally activate human B cells and positive interface on TLR9. Those structural requirements will plasmacytoid dendritic cells. Biochem. Biophys. Res. Commun. 310: 1133–1139. also support the design of improved activation because we dem- 23. Klinman, D. M., and D. Currie. 2003. Hierarchical recognition of CpG motifs expressed by immunostimulatory oligodeoxynucleotides. Clin. Exp. Immunol. onstrated that some ODNs, particularly those that are longer, 133: 227–232. cause stronger activation than ODN2006. 24. Yi, A. K., M. Chang, D. W. Peckham, A. M. Krieg, and R. F. Ashman. 1998. CpG oligodeoxyribonucleotides rescue mature spleen B cells from spontaneous apoptosis and promote cell cycle entry. J. Immunol. 160: 5898–5906. Acknowledgments 25. Rankin, R., R. Pontarollo, X. Ioannou, A. M. Krieg, R. Hecker, L. A. Babiuk, We thank Katja Leben and Ana Kunsek for technical help with the cytokine and S. van Drunen Littel-van den Hurk. 2001. CpG motif identification for measurements. veterinary and laboratory species demonstrates that sequence recognition is highly conserved. Antisense Nucleic Acid Drug Dev. 11: 333–340. 26. Haas, T., J. Metzger, F. Schmitz, A. Heit, T. Muller,€ E. Latz, and H. Wagner. Downloaded from Disclosures 2008. The DNA sugar backbone 29 deoxyribose determines toll-like receptor 9 activation. Immunity 28: 315–323. The authors have no financial conflicts of interest. 27. Gay, N. J., and M. Gangloff. 2007. Structure and function of Toll receptors and their ligands. Annu. Rev. Biochem. 76: 141–165. 28. Marshall, J. D., K. Fearon, C. Abbate, S. Subramanian, P. Yee, J. Gregorio, References R. L. Coffman, and G. Van Nest. 2003. Identification of a novel CpG DNA class and motif that optimally stimulate B cell and plasmacytoid dendritic cell func- 1. Krieg, A. M., A. K. Yi, S. Matson, T. J. Waldschmidt, G. A. Bishop, R. Teasdale, tions. J. Leukoc. Biol. 73: 781–792.

G. A. Koretzky, and D. M. Klinman. 1995. CpG motifs in bacterial DNA trigger http://www.jimmunol.org/ 29. Meng, W., T. Yamazaki, Y. Nishida, and N. Hanagata. 2011. Nuclease-resistant direct B-cell activation. Nature 374: 546–549. immunostimulatory phosphodiester CpG oligodeoxynucleotides as human Toll- 2. Krieg, A. M. 2002. CpG motifs in bacterial DNA and their immune effects. like receptor 9 agonists. BMC Biotechnol. 11: 88. Annu. Rev. Immunol. 20: 709–760. 30. Jurk, M., B. Schulte, A. Kritzler, B. Noll, E. Uhlmann, T. Wader, C. Schetter, 3. Hornung, V., S. Rothenfusser, S. Britsch, A. Krug, B. Jahrsdo¨rfer, T. Giese, A. M. Krieg, and J. Vollmer. 2004. C-Class CpG ODN: sequence requirements S. Endres, and G. Hartmann. 2002. Quantitative expression of toll-like receptor Immu- 1-10 mRNA in cellular subsets of human peripheral blood mononuclear cells and and characterization of immunostimulatory activities on mRNA level. sensitivity to CpG oligodeoxynucleotides. J. Immunol. 168: 4531–4537. nobiology 209: 141–154. 4. Moseman, A. P., E. A. Moseman, S. Schworer, I. Smirnova, T. Volkova, U. von 31. Hartmann, G., R. D. Weeratna, Z. K. Ballas, P. Payette, S. Blackwell, I. Suparto, Andrian, and A. Poltorak. 2013. Mannose receptor 1 mediates cellular uptake W. L. Rasmussen, M. Waldschmidt, D. Sajuthi, R. H. Purcell, et al. 2000. De- and endosomal delivery of CpG-motif containing oligodeoxynucleotides. J. lineation of a CpG phosphorothioate oligodeoxynucleotide for activating primate Immunol. 191: 5615–5624. immune responses in vitro and in vivo. J. Immunol. 164: 1617–1624. 32. Latz, E., A. Verma, A. Visintin, M. Gong, C. M. Sirois, D. C. G. Klein,

5. Sivori, S., S. Carlomagno, M. Falco, E. Romeo, L. Moretta, and A. Moretta. by guest on October 2, 2021 2011. Natural killer cells expressing the KIR2DS1-activating receptor efficiently B. G. Monks, C. J. McKnight, M. S. Lamphier, W. P. Duprex, et al. 2007. kill T-cell blasts and dendritic cells: implications in haploidentical HSCT. Blood Ligand-induced conformational changes allosterically activate Toll-like receptor 117: 4284–4292. 9. Nat. Immunol. 8: 772–779. 6. Lahoud, M. H., F. Ahmet, J.-G. Zhang, S. Meuter, A. N. Policheni, S. Kitsoulis, 33. Ashman, R. F., J. A. Goeken, E. Latz, and P. Lenert. 2011. Optimal oligonu- C.-N. Lee, M. O’Keeffe, L. C. Sullivan, A. G. Brooks, et al. 2012. DEC-205 is a cell cleotide sequences for TLR9 inhibitory activity in human cells: lack of corre- surface receptor for CpG oligonucleotides. Proc.Natl.Acad.Sci.USA109: 16270–16275. lation with TLR9 binding. Int. Immunol. 23: 203–214. 7. Gursel, M., D. M. Klinman, and I. Gursel. 2013. Plasmacytoid dendritic cell 34. Kindrachuk, J., J. E. Potter, R. Brownlie, A. D. Ficzycz, P. J. Griebel, response to CpG ODN correlates with CXCL16 expression and is inhibited by N. Mookherjee, G. K. Mutwiri, L. A. Babiuk, and S. Napper. 2007. Nucleic acids ox-LDL. Mediators Inflamm. 2013: 312590. exert a sequence-independent cooperative effect on sequence-dependent acti- 8. Krieg, A. M. 2006. Therapeutic potential of Toll-like receptor 9 activation. Nat. vation of Toll-like receptor 9. J. Biol. Chem. 282: 13944–13953. Rev. Drug Discov. 5: 471–484. 35. Rutz, M., J. Metzger, T. Gellert, P. Luppa, G. B. Lipford, H. Wagner, and 9. Hartmann, G., G. J. Weiner, and A. M. Krieg. 1999. CpG DNA: a potent signal S. Bauer. 2004. Toll-like receptor 9 binds single-stranded CpG-DNA in a se- for growth, activation, and maturation of human dendritic cells. Proc. Natl. Acad. quence- and pH-dependent manner. Eur. J. Immunol. 34: 2541–2550. Sci. USA 96: 9305–9310. 36. Kerkmann, M., L. T. Costa, C. Richter, S. Rothenfusser, J. Battiany, V. Hornung, 10. Hanagata, N. 2012. Structure-dependent immunostimulatory effect of CpG oligo- J. Johnson, S. Englert, T. Ketterer, W. Heckl, et al. 2005. Spontaneous formation of deoxynucleotides and their delivery system. Int. J. Nanomedicine 7: 2181–2195. nucleic acid-based nanoparticles is responsible for high interferon-alpha induction 11. Klinman, D. M., A. K. Yi, S. L. Beaucage, J. Conover, and A. M. Krieg. 1996. CpG by CpG-A in plasmacytoid dendritic cells. J. Biol. Chem. 280: 8086–8093. motifs present in bacteria DNA rapidly induce lymphocytes to secrete interleukin 6, 37. Keniry, M. A. 2000-2001. Quadruplex structures in nucleic acids. Biopolymers interleukin 12, and interferon gamma. Proc. Natl. Acad. Sci. USA 93: 2879–2883. 56: 123–146. 12. Hartmann, G., and A. M. Krieg. 2000. Mechanism and function of a newly 38. Dalpke, A. H., S. Zimmermann, I. Albrecht, and K. Heeg. 2002. Phosphodiester identified CpG DNA motif in human primary B cells. J. Immunol. 164: 944–953. CpG oligonucleotides as adjuvants: polyguanosine runs enhance cellular uptake 13. Vollmer, J., M. Jurk, U. Samulowitz, G. Lipford, A. Forsbach, M. Wullner,€ and improve immunostimulative activity of phosphodiester CpG oligonucleo- S. Tluk, H. Hartmann, A. Kritzler, C. Muller,€ et al. 2004. CpG oligodeox- tides in vitro and in vivo. Immunology 106: 102–112. ynucleotides stimulate IFN-gamma-inducible protein-10 production in human 39. Gursel,€ M., D. Verthelyi, I. Gursel,€ K. J. Ishii, and D. M. Klinman. 2002. Dif- B cells. J. Endotoxin Res. 10: 431–438. ferential and competitive activation of human immune cells by distinct classes of 14. Guiducci, C., G. Ott, J. H. Chan, E. Damon, C. Calacsan, T. Matray, K.-D. Lee, CpG oligodeoxynucleotide. J. Leukoc. Biol. 71: 813–820. R. L. Coffman, and F. J. Barrat. 2006. Properties regulating the nature of the 40. Honda, K., H. Yanai, H. Negishi, M. Asagiri, M. Sato, T. Mizutani, N. Shimada, plasmacytoid dendritic cell response to Toll-like receptor 9 activation. J. Exp. Y. Ohba, A. Takaoka, N. Yoshida, and T. Taniguchi. 2005. IRF-7 is the master Med. 203: 1999–2008. regulator of type-I interferon-dependent immune responses. Nature 434: 772– 15. Hartmann, G., J. Battiany, H. Poeck, M. Wagner, M. Kerkmann, N. Lubenow, 777. S. Rothenfusser, and S. Endres. 2003. Rational design of new CpG oligonu- 41. Chinnathambi, S., S. Chen, S. Ganesan, and N. Hanagata. 2012. Binding mode of cleotides that combine B cell activation with high IFN-alpha induction in CpG oligodeoxynucleotides to nanoparticles regulates bifurcated cytokine in- plasmacytoid dendritic cells. Eur. J. Immunol. 33: 1633–1641. duction via Toll-like receptor 9. Sci Rep 2: 534. 16. Vollmer, J., R. Weeratna, P. Payette, M. Jurk, C. Schetter, M. Laucht, T. Wader, 42. Suwarti, S., T. Yamazaki, C. Svetlana, and N. Hanagata. 2013. Recognition of S. Tluk, M. Liu, H. L. Davis, and A. M. Krieg. 2004. Characterization of three CpG oligodeoxynucleotides by human Toll-like receptor 9 and subsequent cy- CpG oligodeoxynucleotide classes with distinct immunostimulatory activities. tokine induction. Biochem. Biophys. Res. Commun. 430: 1234–1239. Eur. J. Immunol. 34: 251–262. 43. Baek, K. H., S. J. Ha, and Y. C. Sung. 2001. A novel function of phospho- 17. Samulowitz, U., M. Weber, R. Weeratna, E. Uhlmann, B. Noll, A. M. Krieg, and rothioate oligodeoxynucleotides as chemoattractants for primary macrophages. J. Vollmer. 2010. A novel class of immune-stimulatory CpG oligodeoxy- J. Immunol. 167: 2847–2854. nucleotides unifies high potency in type I interferon induction with preferred 44. Wagner, H. 2011. New vistas on TLR9 activation. Eur. J. Immunol. 41: 2814– structural properties. Oligonucleotides 20: 93–101. 2816.