Nucleic acid-binding polymers asanti-inflammatory agentsJaewoo Leea,b,1, Jang Wook Sohnc, Ying Zhangd, Kam W. Leongb,d, David Pisetskye, and Bruce A. Sullengera,b,1

aDuke Translational Research Institute, Duke University Medical Center, Durham, NC 27710; bDepartments of Surgery and dBiomedical Engineering, DukeUniversity, Durham, NC 27710; eMedical Research Service, Durham VA Hospital, Durham, NC 27705; and cDepartment of Internal Medicine, College ofMedicine, Korea University, Seoul 136-701, Korea

Edited by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved July 14, 2011 (received for review April 11, 2011)

Dead and dying cells release nucleic acids. These extracellular RNAsand DNAs can be taken up by inflammatory cells and activatemulti-ple nucleic acid-sensing toll-like receptors (TLR3, 7, 8, and 9). Theinappropriate activation of these TLRs can engender a variety ofinflammatory and autoimmune diseases. The redundancy of theTLR family encouraged us to seek materials that can neutralizethe proinflammatory effects of any nucleic acid regardless of itssequence, structure or chemistry. Herein we demonstrate thatcertain nucleic acid-binding polymers can inhibit activation of allnucleic acid-sensing TLRs irrespective of whether they recognizessRNA, dsRNA or hypomethylated DNA. Furthermore, systemicadministration of such polymers can prevent fatal liver injuryengendered by proinflammatory nucleic acids in an acute toxicshock model in mice. Therefore these polymers represent a novelclass of anti-inflammatory agent that can act as molecular scaven-gers to neutralize the proinflammatory effects of various nucleicacids.

damage-associated molecular pattern ∣ inflammation ∣pattern-recognition receptor

Pattern-recognition receptors (PRRs) allow immune cells torecognize and protect tissues from various harmful stimuli,

such as pathogens and damaged cells (1, 2). They recognize adiverse set of molecular signatures termed pathogen-associatedmolecular patterns (PAMPs) and damage-associated molecularpatterns (DAMPs) (2), and act as innate and adaptive immunesensors by recognizing such PAMPs and DAMPs (3). Toll-likereceptors (TLRs) are the best characterized PRRs (4). At least10 human and 12 mouse TLRs have been identified. Each TLR isable to recognize a particular molecular pattern. For instance,TLR2, 4, 5, 6, and 11 bind to bacterial membrane-associatedmolecules such as lipoprotein, lipopolysaccharide (LPS), andpeptidoglycan whereas TLR3, 7, 8 and 9 recognize bacterial,viral, or endogenous nucleic acids including ssRNA, dsRNA,and unmethylated CpG-containing DNA (5). Moreover, TLRscan be classified based on their cellular localization: TLR1, 2,4, 5, and 6 are expressed on the cell surface whereas TLR3, 7, 8,and 9 are localized mostly in the endosomal compartment (5).Activation of TLRs upon ligand binding results in signalingevents that lead to the expression of immune response genes in-cluding inflammatory cytokines, immune stimulatory cytokines,chemokines, and costimulatory molecules, which augment thekilling of pathogens and initiates the process of developing adap-tive immunity (6, 7). Inappropriate activation of TLRs, on theother hand, contributes to development of a variety of humandiseases including systemic lupus erythematosus (SLE), bacterialsepsis, inflammatory bowel disease, psoriasis, multiple sclerosis,rheumatoid arthritis, and atherosclerosis (8–10). Upon infectionor injury of tissue, the infected or damaged tissue releases variousintracellular factors that can recruit and activate innate immunecells. Nucleic acids that originate from host cells or intracellularmicroorganisms can be recognized by nucleic acid-sensing TLRsand can result in the induction of pathological inflammatoryresponses (Fig. S1) (2). In patients with bacterial sepsis, toxic

shock, SLE or rheumatoid arthritis, hypomethylated CpG DNAs,or RNAs have been detected in extracellular compartments andthese extracellular nucleic acids have been correlated to thepathogenesis of these diseases (11–13). Although inhibition ofone or two nucleic acid-sensing TLRs using receptor antagonistshas been demonstrated to attenuate disease progression in poly-microbial sepsis and autoimmune disease models to some extent(12, 14–16), the redundancy of the TLR family of proteinssuggests that concurrent inhibition of all nucleic acid-sensingTLRs would be the most effective means to control nucleicacid-induced inflammation in humans. Because all the nucleicacid-sensing TLRs bind to RNA or DNA, albeit each TLR ligandcontains a distinguishable molecular pattern, we hypothesizedthat agents that bind to DNAs and RNAs regardless of their se-quence, structure or chemistry might be able to inhibit nucleicacid-mediated activation of all RNA- and DNA-sensing TLRs.Herein we demonstrate that certain cationic polymers can act asmolecular scavengers and block the immune stimulatory effectsof extracellular ssRNA, dsRNA, and unmethylated DNA.

ResultsNucleic Acid-Binding Polymers Inhibit Nucleic Acid-Mediated Activa-tion of TLRs. We initially evaluated six agents known to bindnucleic acids for their ability to attenuate nucleic acid-mediatedactivation of TLRs on macrophages: polyphosphoramidate poly-mer (PPA-DPA), polyamidoamine dendrimer, 1,4-diaminobutanecore-PAMAM-G3 (PAMAM-G3), poly-L-lysine, β-cyclodextrin-containing polycation (CDP), hexadimethrine bromide (HDMBr),and protamine sulfate (Table S1). All of the compounds exceptprotamine sulfate inhibit TLR3 activation by synthetic dsRNA,polyinosinic-polycytidylic acid (poly I∶C) as measured by TNFαand IL-6 production and CD80 expression (Fig. 1A and Fig. S2).Moreover three of the cationic polymers, CDP, HDMBr, and PA-MAM-G3 inhibit the ability of synthetic CpG DNAs (CpG 1668)to activate TLR9 (Fig. 1A). Similarly these three polymers inhibitthe ability of ssRNA-lipid complexes (ssRNA40) to activateTLR7 (Fig. 1A). Such cationic polymers are specific for nucleicacid-mediated activation of the TLRs. They do not inhibit activa-tion of TLRs that recognize nonnucleic acid-based TLR agonistssuch as bacterial LPS, which activate TLR4, and Pam3CSK4, asynthetic bacterial lipoprotein, which activates the TLR2/1 com-plex (Fig. 1A and Fig. S2). Furthermore CDP, HDMBr, and PA-MAM-G3 neutralize immune stimulatory activity of all types ofnucleic acid-based TLR agonists in a variety of primary cells in-cluding B cells, fibroblasts and dendritic cells (DCs) (Fig. S3).

Author contributions: J.L., K.W.L., and B.A.S. designed research; J.L., J.W.S., and Y.Z.performed research; K.W.L. contributed new reagents/analytic tools; J.L., K.W.L., D.P.,and B.A.S. analyzed data; and J.L. and B.A.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence may be addressed. E-mail: bruce.sullenger@duke.edu orjaewoo.lee@duke.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1105777108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1105777108 PNAS ∣ August 23, 2011 ∣ vol. 108 ∣ no. 34 ∣ 14055–14060

CHEM

ISTR

YAPP

LIED

BIOLO

GICAL

SCIENCE

S

Next we asked whether the cationic polymers could neutralizethe immune stimulatory activity of endogenous nucleic acids. Asexpected, the three polymers, CDP, HDMBr, and PAMAM-G3inhibited bacterial DNA-induced cellular activation (Fig. 1B).Unlike bacterial DNAs, mammalian genomic DNAs are highlymethylated and immunologically inert. In various pathologicalconditions, autoantibody, nuclear protein or endogenous cationicprotein form a complex with mammalian DNAs, and theseDNA-containing complexes are able to activate innate and adap-tive immune cells through TLR9 signaling (reviewed in ref. 17).Consistent with previous findings, mammalian DNA (CT DNA)only weakly activated mouse macrophages to produce TNFα, butCT DNA/DOTAP (N-[1-(2,3-Dioleoyloxy)propyl]-N,N,N-tri-methylammoniummethyl-sulfate) complexes induced much high-er levels of TNFα production from these cells (Fig. 1B). Thisinflammatory cytokine production was inhibited upon treatmentwith cationic polymers, CDP, HDMBr, or PAMAM-G3. More-over, the inhibitory cationic polymers did not inhibit nonnucleicacid-based endogenous DAMPs. Heparan sulfate is a stronglynegatively charged polysaccharide found in cell surface or extra-cellular matrix. When harmful stimuli insults tissues the heparansulfates are enzymatically cleaved from cellular or extracellularmatrix macromolecules and recognized by TLR4 of innateimmune cells (18). The CDP, HDMBr, and PAMAM-G3 didnot inhibit heparan sulfate-mediated TNFα production by cells(Fig. 1B). This data suggest that the nucleic acid-binding poly-mers specifically neutralize immune stimulatory activity of ncleic

acids, and this target specificity is not just governed by the elec-trostatic interaction between oppositely charged molecules.

It has been shown that binding of CpG oligodeoxynucleotides(ODNs) to TLR9 in the early endosomal compartment leadspredominantly to a type I IFN response whereas activation inthe late endosomal compartment induces inflammatory cytokinesand immune stimulatory receptors but not type I IFN (19, 20).Multiple types of CpG ODNs have been identified that prefer-entially activate the early, late or both endosomal responses. TypeA CpG ODNs are characterized by poly-G tails at both ends anda palindromic sequence in the middle, and they activate plasma-cytoid DCs (pDCs) to produce type I IFNs through MyD88-IRF7signaling pathway, but only weakly activate other immune cells toproduce inflammatory cytokines (21, 22). Type B CpG ODNscontain linearized sequences and activate B cells, conventionalDCs and other inflammatory cells to produce inflammatory cyto-kines (e.g., TNFα and IL-6) through MyD88-NF-κB signalingpathway, but not pDCs to produce type I IFN (23, 24). Interest-ingly, lipoplex can regulate endosomal location of CpG ODNsand consequently alter the downstream signaling of TLR9 (19).To determine if the inhibitory nucleic acid-binding polymers,CDP, HDMBr, and PAMAM-G3, form complexes with CpGODNs that are retained in the early endosome, we evaluatedwhether these polymers would skew TLR9 signaling toward typeI IFN expression and away from TNFα and IL-6 production. CDPinhibited TNFα production by macrophage cells after stimulationwith either type A CpG ODN (CpG 1585) or type B CpG ODN(CpG 1668) but did not inhibit IFNα production from pDCs after

Fig. 1. Nucleic acid-binding polymers neutralize the ability of nucleic acids to activate inflammatory cells. Nucleic acid-binding polymers, CDP, HDMBr andPAMAM-G3, (20 μg∕mL) inhibited the activation of the murine macrophage cell line, RAW264.7, by (A) synthetic nucleic acid-based TLR agonists, ssRNA40(TLR7), poly I∶C (TLR3), CpG1668 (TLR9) or (B) genomic DNAs from bacteria (E. coli) and calf thymus (CT DNA) (TLR9) but did not inhibit cell activation by (A) thesynthetic nonnucleic acid TLR agonists, Pam3CSK4 (TLR2) and LPS (TLR4) and (B) endogenous DAMP, heparan sulfate (TLR4). To transform immunological inertmammalian CT DNA into immune stimulatory DNA CT, a DNA/DOTAP complex was prepared by incubation of 5 μg of CT-DNAs with 15 μg of DOTAP in 30 μLHepes-buffered saline buffer (20 mM Hepes, 150 mM NaCl, pH 7.4). (C, D) To study spatiotemporal regulation of TLR9 signaling (C) bone marrow-derivedplasmacytoid DCs and (D) RAW cells were incubated with CpG 1585 (1 μM) in the presence or absence of cationic polymers, CDP, HDMBr or PAMAM-G3(20 μg∕mL). (E) To study activation of cytosolic dsRNA sensors, PKR and MDA5, 1 μg of poly I∶C was complexed with 1 μL of DharmaFECT (Thermo Scientific)in 50 μL of Opti-MEM Imedia (Invitrogen). RAW264.7 cells andMEFs were stimulatedwith the poly I∶C/DharmaFECTcomplex, poly I∶C alone, DharmaFECTaloneor PBS in the presence or absence of HDMBr or PAMAM-G3. TNFα, IL-6 and IFNα production were measured by ELISA to monitor the activation of the cells.Data represents the mean of three individual experiments. Error bar is SD; ∗ P < 0.05. ‡ Amount of cytokine ¼ ½cytokine�poly I∶C or poly I∶C∕DharmaFECT−½cytokine�PBS or DharmaFECT

14056 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105777108 Lee et al.

stimulation with either type A or type B CpG ODN. By contrast,HDMBr and PAMAM-G3 inhibit both TNFα and IFNα produc-tion by pDCs and macrophage cell lines after stimulation with alltypes of CpG ODNs (Fig. 1 C and D).

In addition to endosomal nucleic acid-sensing TLRs microbialdsRNAs can be recognized by cytosolic dsRNA-sensing PRRs,including protein kinase R (PKR) and RIG-I-like receptor familymembers such as melanoma differentiation-associated gene 5(MDA-5) and retinoic acid inducible gene-I (RIG-I). Althoughboth endosomal TLRs and cytosolic dsRNA sensors are ableto recognize exogenous dsRNAs, their activation leads to distinc-tive downstream signaling pathways resulting in expression ofdifferent types of genes. For example, cytosolic dsRNA-sensingPRRs induce type I IFN production through IFN regulatory fac-tor 3 (IRF-3)-dependent pathways whereas endosomal dsRNA-sensing TLR3 induce inflammatory cytokine production throughIRF-3-independent but NF-kB-dependent pathways (25). It iswell established that treatment of cells with poly I∶C dsRNAcomplexed with a transfection reagent activates cytosolic PKRand MDA5 signaling pathways that leads to type I IFN produc-tion; whereas cells treated with poly I∶C alone predominantlyinduce endosomal TLR3 signaling pathways (26, 27). As shownin Fig. 1E, HDMBr and PAMAM-G3 can block poly I∶C/Dhar-maFECT-mediated activation of type I IFN production fromtreated cells. Thus HDMBr and PAMAM-G3 are able to effec-tively and specifically inhibit the activation of multiple nucleicacid-sensing PRRs on a variety of inflammatory cell types. There-fore, these two polymers were chosen for further study.

Nucleic Acid-Binding Polymers Neutralize Extracellular InflammatoryNucleic Acids, Rather than Directly Inhibiting Nucleic Acid-SensingTLRs. The two polymers, HDMBr and PAMAM-G3, were able toneutralize the immune stimulatory activity of nucleic acid-basedTLR agonists whose receptors are located in endosomal com-partments. This inhibitory effect can be similarly achieved bytreatment of cells with chloroquine that is an inhibitor of endo-somal acidification (28–30). Nucleic acid-binding polymers con-taining protonable amines (−NH2) have been shown to induce aproton sponge effect inside endosomes, which promotes osmoticswelling and disruption of the endosomal membrane (31). Todetermine if HDMBR and PAMAM-G3 were working by thismechanism, we studied the inhibitory effect of polymers onTLR7 activated by either a nucleic acid-based agonist (ssRNA40)or nonnucleic acid-based agonist (imidazoquinoline compound,R848). Although the polymers inhibited the activation of macro-phages treated with ssRNA40, they did not inhibit stimulation ofcells treated with R848. If macrophages were coincubated withssRNA40, R848, and either nucleic acid-binding polymer, these

cells generated reduced levels of inflammatory cytokines produc-tion comparable to the level from cells treated with R848 alone.By contrast, treatment of macrophages with chloroquine inhib-ited inflammatory cytokine production from cells treated withssRNA40 and R848 (Fig. 2A). Furthermore, pretreatment of cellswith chloroquine prior to stimulation prevented the activationof the cells with CpG ODNs. By contrast, preincubation ofmacrophages with the polymer-based inhibitors and subsequentremoval of the polymers did not prevent inflammatory cytokineproduction from the cells subsequently stimulated with CpGODNs even though pretreatment and continuous treatment ofmacrophages with the cationic polymers prevents CpG ODN-induced TNFα production (Fig. 2B). Thus HDMBr and PA-MAM-G3 do not directly inhibit the nucleic acid-sensing TLRsbut rather neutralize the nucleic acid-based receptor ligands.

Nucleic Acid-Binding Polymers can Alter the Uptake and IntracellularDistribution of Immune Stimulatory Nucleic Acids. Next we wantedto explore the mechanism(s) that nucleic acid-binding polymersutilize to neutralize the immune stimulatory activity of nucleicacids. We speculated that the HDMBr and PAMAM-G3 mayact by inhibiting the cellular uptake of immune stimulatorynucleic acids. To evaluate this possibility we performed cellularuptake studies using fluorescently labeled CpG ODNs. PAMAM-G3 and HDMBr slightly reduced cellular uptake of CpG ODNswhereas noninhibitory cationic molecules, PPA-DPA, poly L-lysine and protamine, significantly increased intracellular uptakesof CpG ODNs (Fig. 3A). Nevertheless the mean fluorescenceintensity of cells coincubated at 37 °C with combination of CpGODNs and HDMBr or PAMAM-G3 was significantly greaterthan that of cells at 4 °C (CpG 1668 alone vs. CpG 1668 +HDMBr: P ¼ 0.0493; CpG 1668 + PAMAM-G3: P ¼ 0.0165),which suggests that cells internalized significant amounts ofCpG ODNs even in the presence of HDMBr and PAMAM-G3. Thus, the polymers do not appear to exclusively act bylimiting the ability of nucleic acids to associate with cells. Next,because different nucleic acid-binding polymers are known todeliver genes into cells through diverse internalization and traf-ficking pathways, which is influenced by molecular composition,surface charge and size of complexes (32), we investigatedwhether the subcellular localization of CpG ODNs was alteredby the polymers. As shown in Fig. 3B, when no polymer is added,a CpG ODN is internalized into macrophages and is found in thecytosol as well as the nucleus. In the cytosol, the DNA appears toform a punctate pattern that at least in part colocalizes withTLR9. Addition of HDMBr or PAMAM-G3 to the CpG ODNsappears to not only reduce internalization of the DNA intocells but the polymers also alter CpG distribution largely toward

Fig. 2. Nucleic acid-binding polymers selectively inhi-bit nucleic acid-based TLR ligands. (A) RAW264.7 cellswere coincubated with nucleic acid-binding polymers(HDMBr or PAMAM-G3), nucleic acid TLR7 agonist(ssRNA40) or nonnucleic acid TLR7 agonist (R848). En-dosomal acidification inhibitor, chloroquine, wasused as an inhibitor of TLR7. Cationic polymers inhib-ited TLR7 activation by the nucleic acid-based agonistbut not by the nonnucleic acid agonist. (B) Cells werepreincubated for 2 h with nucleic acid-binding poly-mer, chloroquine or PBS. CpG 1668 (1 μM), then, wereadded into cell culture after 3× washing with com-plete medium or nonwashing. Cells were incubatedfor additional 6 h and TNFα production was analyzedby ELISA. Data represents the mean of two individualexperiments. Error bar is SD.

Lee et al. PNAS ∣ August 23, 2011 ∣ vol. 108 ∣ no. 34 ∣ 14057

CHEM

ISTR

YAPP

LIED

BIOLO

GICAL

SCIENCE

S

the nucleus where it does not colocalize with TLR9. By contrast,addition of noninhibitory polymer, PPA-DPA, to CpG ODNenhances this punctate appearance in the cytosol and appears tofurther colocalize the DNA with TLR9 (Fig. 3B). Thus HDMBrand PAMAM-G3 appear to alter the uptake and distribution ofCpG ODNs in cells that likely accounts at least in part for theirability to limit nucleic acid-mediated activation of TLR9.

Cationic Polymers can Protect Mice from Nucleic Acid-Induced ToxicShock and Fatal Inflammation. Finally, we evaluated the abilityof the nucleic acid-binding polymers to control pathogenic acti-vation of TLRs by freely circulating nucleic acids in an in vivotoxic shock model. When mice that have been sensitized withD-galactosamine (D-GalN) are challenged with TLR agoniststhese mice will succumb to TNFα-dependent lethal liver damageand die shortly (33, 34). Consistent with previous reports, we

observed that greater than 90% of mice die following administra-tion of D-GalN and CpG DNA or poly I∶C RNA within 48 h.Strikingly, coadministration of the nucleic acid-binding polymers,HDMBr or PAMAM-G3, following D-GalN and CpG DNA orpoly I∶C injection resulted in significant protection of animals,in a dose-dependent manner, and reduced mortality by 100%in several cases (Fig. 4). By contrast such polymers are unableto protect mice from nonnucleic acid TLR activators such asLPS (Fig. S4). Thus certain nucleic acid-binding polymers arealso able to bind freely circulating nucleic acids in vivo and there-by inhibit the potent nucleic acid-induced toxic shock responsethat such free nucleic acids engender in mice.

DiscussionOur studies demonstrate that certain nucleic acid-binding poly-mers can act as molecular scavengers and inhibit the ability ofcirculating immune stimulatory nucleic acids, regardless of theirsequence, chemistry, or structure, to activate a variety of nucleicacid-sensing TLRs and cytoplasmic PRRs in vitro and in vivo. In-creasing evidence indicates that inappropriate activation ofnucleic acid-sensing TLRs and cytoplasmic PRRs can result in avariety of human inflammatory and autoimmune diseases includ-ing SLE, sepsis, inflammatory bowel disease, psoriasis, multiplesclerosis, and rheumatoid arthritis (8–10). Thus, unique strategiesto control such pathogenic activation of these receptors areneeded. Unfortunately, the interconnectedness and redundancyof nucleic acid-induced TLR and PRR signaling pathways willlikely limit the therapeutic efficacy of single or dual TLR/PRRinhibitors (Fig. S1). TLR signaling, irrespective of which TLRis activated, culminates in activation of MAP kinases, NF-κB,and IFN regulatory factors and engenders the production ofinflammatory cytokines and/or type I IFN (5). Our proof-of-concept studies describe a unique strategy that employs nucleicacid-binding polymers to simultaneously inhibit the ability offreely circulating nucleic acids to activate all nucleic acid-sensingTLRs and cytoplasmic PRR.

We have previously shown that cationic polymers can serve asantidotes for nucleic acid aptamers and rapidly reverse the effectsof such agents in vivo (35). Results in this study complement ourprevious findings and demonstrate that these polymers can alsointeract with endogenous nucleic acids to inhibit activation ofnucleic acid-sensing TLRs and cytplasmic PRRs. One of themost exciting applications of nanomedicine has been to engineernanocomplexes that can deliver exogenous drugs and genes totarget tissues in a spatially and temporally controlled manner.This study shows that nucleic acid-binding polymers used for suchnanotherapeutic applications can instead be applied in a uniquemanner to form complexes with various extracellular nucleicacids in situ. This unique concept may have several therapeuticapplications. Extracellular nucleic acids released by dead anddying cells play a role not only for the aberrant inflammatoryresponses by overactivation of endosomal TLRs, but also forother pathological responses; e.g., microvascular permeability,blood coagulation, and thrombosis. Extracellular RNAs has beenshown to directly bind to vascular endothelial-cell growth factor(VEGF) and stabilize the interaction of VEGF and cognate re-ceptor, VEGF-R2, which lead to significantly increase the perme-ability of microvascular endothelial cells that is one of commonpathologic conditions associated with brain tumor, head injuryand ischemic stoke (36). Moreover, extracellular RNAs werefound to mediate blood coagulation by activation of Factor XIIand XI, proteases of the contact factor pathway of blood coagu-lation (37), and treatment of mice with RNase significantly delaysocclusive thrombus formation in a murine model of arterialthrombosis (37). Though not directly evaluated herein, we spec-ulate that this same strategy may also be useful for inhibiting theability of extracellular nucleic acids from stimulating endothelialcells and blood coagulation pathways as well.

Fig. 3. Modification of cellular internalization of CpG DNAs by nucleic acid-binding polymers. (A) Cellular uptake of nucleic acids was studied by quan-tification of Alexa488-CpG ODNs in cells. RAW264.7 cells were incubated at37 °C for 1 h with Alexa488-conjugated CpG 1668 (1 μM) in the presence orabsence of HDMBr, PAMAM-G3, PPA-DPA, poly-L-lysine or protamine sulfate(10 μg∕mL). CpG DNA internalization was analyzed by flow cytometry. Filledgray histogram: unlabeled CpG 1668. Open red histogram: Alexa488-CpG1668 + cationic polymer. Open black histogram: Allexa488-CpG 1668 alone.To assess nonspecific uptakes or surface binding of CpG ODNs, the cells wereincubated at 4 °C for 1 h with Alexa488-CpG 1668 and nucleic acid-bindingpolymers. (B) To study intracellular trafficking of CpG ODNs, RAW cells wereincubated at 37 °C for 1 h with Alexa488-CpG 1668 and HDMBr, PAMAM-G3or PPA-DPA. After extensive washing, the cells were stained with an anti-TLR9Antibody (red) and DAPI (blue). Intracellular localization of CpG DNAs wasanalyzed by a confocal microscopy. Colocalization of CpG 1668 and TLR9in endosomal compartments formed fluorescent puncta in yellow. Cells co-incubated with CpG 1668 and noninhibitory polymer, PPA-DPA increased sizeand intensity of puncta comparing CpG 1668 alone whereas coincubationwith CpG 1668 and HDMBr or PAMAM-G3 did not form puncta (X63).

14058 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105777108 Lee et al.

HDMBr is broadly used to facilitate viral transduction, and itcan bind to and neutralize negatively charged viral membrane,resulting in decreasing of virus-cell repulsion and increasing ofvirus absorption and transduction (38). HDMBr is also knownto neutralize highly negatively charged anticoagulation agent,heparin (39). The PAMAM dendrimer is a macromolecularplatform for carrying drug, gene, siRNA and imaging agents invivo (40). Depending on their surface chemistry PAMAM dendri-mers can also act as active therapeutic agents. For example, PA-MAM-G5 with alkylation at surface primary amines neutralizesLPS. Treatment with the PAMAM-G5 derivative could protectD-GalN-sensitized mice from toxic shock upon LPS challenge(41). Furthermore, Shaunak et al. has shown that anionic poly-mer, PAMAM-G3.5, covalently conjugated with glucosamineinhibited a variety of cellular activity including LPS-mediatedTLR4 activation, fibroblast growth factor-2 (FGF-2)-mediatedendothelial-cell proliferation and allogeneic mixed leukocytereaction (42). In the current study, we found that cationic poly-mers, PAMAM-G3 and -G5 and HDMBr inhibited nucleicacid-sensing TLRs by neutralization of immune stimulatory nu-cleic acids. Although we do not clearly understand the inhibitorymechanism of polymers the HDMBr and PAMAM seems to alterintracellular distribution of immune stimulatory nucleic acids andsequester the nucleic acids from their corresponding TLRs. In thethermodynamic and stoichiometric studies, we have observedcorrelation between the polymer-mediated neutralization ofCpG ODNs and the strength of binding between the polymerand CpG ODN (Table S2 and Fig. S5). Additional studies willbe required to develop a thorough understanding of the mechan-ism(s) employed by various nucleic acid-binding polymers toinhibit activation of nucleic acid-sensing TLR.

Though the development of any unique therapeutic strategyrequires extensive testing and development, the current study

suggests that certain cationic polymers that are being utilizedin a variety of nanomedicine applications may be useful for atotally unique application, to act as molecular scavengers andthereby limit nucleic acid-induced inflammation in patients witha broad spectrum of inflammatory and autoimmune diseases. It isworth noting that the nucleic acid-binding materials we evaluatedherein were not originally designed with the intent of using themto control nucleic acid-induced inflammation. Therefore weanticipate that a significant translational research opportunitynow exists to develop optimized nucleic acid-binding polymersfor the treatment of patients with nucleic acid-induced inflamma-tory and autoimmune diseases.

Materials and MethodsIn Vitro TLR Activation and Inhibition.Mouse embryonic fibroblasts (MEFs), 5 ×105 of RAW264.7 cells, immature DCs, or B cells were incubated for 18 h incomplete medium in a 48-well culture plate. In a 96-well flat-bottom cultureplate, 8 × 104 of pDCs were incubated for 48 h. To activate TLR9, the completemedium was supplemented with phosphorothioate type B CpG oligodeoxy-nucleotide (ODN) 1668 (CpG 1668) (5′-TCCATGACGTTCCTGATGCT-3′) (IDT),type A CpG ODN 1585 (CpG 1585) (5′-GGGGTCAACGTTGAGGGGGG-3′)(1 μM) (InvivoGen) or DNAs isolated from calf thymus or Escherichia coli(5 μg∕mL) (Sigma). Phosphorothioate GpC DNA 1720 (GpC 1720) (5′-TCCAT-GAGCTTCCTGATGCT-3′) (IDT) and CpG 1585 control (5′-GGGGTCAAGCTT-GAGGGGGG-3′) (InvivoGen) were used as control DNAs. For activation ofTLR2, 3 and 4, Pam3CSK4 (1 μg∕mL) (InvivoGen), poly I∶C (10 μg∕mL) (Amer-sham/GE Healthcare), LPS serotype 026:B6 (100 ng∕mL) (Sigma) and heparansulfate (kindly provided by Todd Brennan, Duke University) were added.Finally, phosphorothioate ssRNA40/LyoVec (5 μg∕mL) (5′-GCCCGUCUGUUGU-GUGACUC-3′), a control ssRNA41/LyoVec (5 μg∕mL) (5′-GCCCGACAGAAGA-GAGACAC-3′) or a synthetic Imidazoquinoline compound, R848 (5 μg∕mL)(all from InvivoGen) were used to activate mouse TLR7. To inhibit TLR activa-tion PAMAM-G1, -G3 and -G5, poly-L-lysine, HDMBr (all from Sigma), CDP(kindly provided by Mark Davis, California Institute of Technology), prota-mine sulfate (APP), PPA-DPA [MW30 kDa, synthesized as previously described(43)] at various concentrations were added and cells were treated with TLR

Fig. 4. Nucleic acid-binding polymers limit nucleic acid-induced acute liver injury. (A) Prevention of CpG DNA- and (B) poly I∶C RNA-mediated liver injury. Micewere injected i.p. with D-galactosamine (D-GalN) + CpG 1668 or D-GalN + poly I∶C. Subsequently, PBS (black circle), HDMBr (red triangle) or PAMAM-G3 (darkgreen square) at the indicated amounts was administered i.p. Mice injectedwith D-GalN, CpG 1668, poly I∶C, HDMBr or PAMAM-G3 alone or D-GalN + GpC 1720were used as controls (gray asterisk). Mice were monitored daily for survival.

Lee et al. PNAS ∣ August 23, 2011 ∣ vol. 108 ∣ no. 34 ∣ 14059

CHEM

ISTR

YAPP

LIED

BIOLO

GICAL

SCIENCE

S

agonists. After incubation, cells were analyzed for CD80 expression by flowcytometry after staining with either phycoerythrin (PE)-labeled anti-mouseCD80 or PE-hamster IgG isotype control (all from eBioscience). Culture super-natants were collected and analyzed. The production of TNFα and IL-6were analyzed with BD OptEIA™ ELISA sets (BD Biosciences); IFNα productionwas determined with a Mouse IFNα ELISA kit (PBL Biomedical Laboratories)following the manufacturer’s instructions.

Cellular Uptake of CpG DNAs. In a 24-well culture plate, 2 × 105 of RAW264.7cells were cultured overnight. After washing three times with fresh culturemedium 300 μL of cold complete medium were added onto each welland cells were incubated for 15 min at 4 °C. After cold incubation, CpG1668 conjugated at 3′ end with Alexa-488 (1 μM) (IDT) in the presence orabsence of cationic polymers (20 μg∕mL) was supplemented into culturemedia. Fluorescently unlabeled CpG 1668 was used to assess autofluores-cence. Cells were incubated at either 37 °C or 4 °C. After a 1-h incubation,cells were washed five times with cold PBS and harvested by treatment with0.25% typsin-EDTA. After washing once with cold PBS, fluorescence intensityof cells was measured by flow cytometry.

Intracellular Localization of CpG DNA. On a 1 mm glass cover slip in a 24-wellculture plate, 2 × 105 RAW264.7 cells were cultured overnight. The cells were

incubated with CpG 1668-Alexa 488 (1 μM) at 37 °C. After a 1-h incubation,cells were washed five times with cold PBS and fixed with a 4% paraformal-dehyde solution followed by permeablization with 0.25% Saponin and 1%BSA in PBS. The cells were blocked by incubation for 30 min with Image-iT FXsignal enhancer (Invitrogen) and stained withmouse anti-human TLR9 (1∕100dilution) (Imgenex) and Cy5-conjugated goat anti-mouse IgG secondaryantibody (1∕1000 dilution) (Abcam). Localization of CpG DNAs and TLR9was observed under a Zeiss LSM 510 inverted confocal microscope andimages was analyzed using a Zeiss LSM image software.

Mouse TNFα-Mediated Acute Liver Injury. TNFα-mediated acute liver injury in aD- galactosamine-sensitized mice was performed as previously described (33,34). Briefly, C57BL/6 mice were injected i.p. with PBS (100 μL), CpG 1668(51 μg), GpC DNA (51 μg), poly I∶C (200 μg), or LPS (2 μg) with or withoutD(+)Galactosamine, Hydrochloride (D-GalN) (EMD Biosciences) (20 mg).HDMBr or PAMAM-G3 (200 to 400 μg) or PBS were injected i.p. 5–10min aftertoxin challenge. Viability of mice was monitored daily for a week.

ACKNOWLEDGMENTS. We thank Dr. Mark Davis for kindly providing CDP andDr. Il-Hwan Kim for technical assistance with immunohistochemistry. Thiswork was supported by a grant from the National Heart, Lung, and BloodInstitute NHLBI (B.A.S.).

1. Kawai T, Akira S (2007) Antiviral signaling through pattern recognition receptors.J Biochem 141:137–145.

2. Lotze MT, et al. (2007) The grateful dead: Damage-associated molecular patternmolecules and reduction/oxidation regulate immunity. Immunol Rev 220:60–81.

3. Palm NW, Medzhitov R (2009) Pattern recognition receptors and control of adaptiveimmunity. Immunol Rev 227:221–233.

4. Leulier F, Lemaitre B (2008) Toll-like receptors–taking an evolutionary approach. NatRev Genet 9:165–178.

5. Kawai T, Akira S (2007) TLR signaling. Semin Immunol 19:24–32.6. Takeda K, Akira S (2005) Toll-like receptors in innate immunity. Int Immunol 17:1–14.7. Akira S, Uematsu S, Takeuchi O (2006) Pathogen recognition and innate immunity. Cell

124:783–801.8. Marshak-Rothstein A, Rifkin IR (2007) Immunologically active autoantigens: The role

of toll-like receptors in the development of chronic inflammatory disease. Annu RevImmunol 25:419–441.

9. Chen K, et al. (2007) Toll-like receptors in inflammation, infection, and cancer. IntImmunopharmacol 7:1271–1285.

10. O’Neill LA (2008) Primer: Toll-like receptor signaling pathways–what do rheumatolo-gists need to know? Nat Clin Pract Rheumatol 4:319–327.

11. Barrat FJ, et al. (2005) Nucleic acids of mammalian origin can act as endogenousligands for toll-like receptors and may promote systemic lupus erythematosus.J Exp Med 202:1131–1139.

12. Cavassani KA, et al. (2008) TLR3 is an endogenous sensor of tissue necrosis duringacute inflammatory events. J Exp Med 205:2609–2621.

13. Hajizadeh S, DeGroot J, TeKoppele JM, Tarkowski A, Collins LV (2003) Extracellularmitochondrial DNA and oxidatively damaged DNA in synovial fluid of patients withrheumatoid arthritis. Arthritis Res Ther 5:R234–240.

14. Barrat FJ, Meeker T, Chan JH, Guiducci C, Coffman RL (2007) Treatment of lupus-pronemice with a dual inhibitor of TLR7 and TLR9 leads to reduction of autoantibodyproduction and amelioration of disease symptoms. Eur J Immunol 37:3582–3586.

15. Plitas G, Burt BM, Nguyen HM, Bamboat ZM, DeMatteo RP (2008) Toll-like receptor 9inhibition reduces mortality in polymicrobial sepsis. J Exp Med 205:1277–1283.

16. Lenert P, et al. (2009) DNA-like class R inhibitory oligonucleotides (INH-ODNs) prefer-entially block autoantigen-induced B-cell and dendritic cell activation in vitro andautoantibody production in lupus-proneMRL-Faslpr/lpr mice in vivo.Arthritis Res Ther11:R79.

17. Pisetsky DS (2008) The role of innate immunity in the induction of autoimmunity.Autoimmun Rev 8:69–72.

18. Tang AH, Brunn GJ, Cascalho M, Platt JL (2007) Pivotal advance: Endogenous pathwayto SIRS, sepsis, and related conditions. J Leukoc Biol 82:282–285.

19. Honda K, et al. (2005) Spatiotemporal regulation of MyD88-IRF-7 signalling for robusttype-I interferon induction. Nature 434:1035–1040.

20. Lande R, et al. (2007) Plasmacytoid dendritic cells sense self-DNA coupled withantimicrobial peptide. Nature 449:564–569.

21. Krug A, et al. (2001) Identification of CpG oligonucleotide sequences with high induc-tion of IFN-alpha/beta in plasmacytoid dendritic cells. Eur J Immunol 31:2154–2163.

22. Honda K, et al. (2005) IRF-7 is the master regulator of type-I interferon-dependentimmune responses. Nature 434:772–777.

23. Krieg AM, et al. (1995) CpG motifs in bacterial DNA trigger direct B-cell activation.Nature 374:546–549.

24. Ballas ZK (2007) Modulation of NK cell activity by CpG oligodeoxynucleotides. Immu-nol Res 39:15–21.

25. Kalali BN, et al. (2008) Double-stranded RNA induces an antiviral defense status inepidermal keratinocytes through TLR3-, PKR-, and MDA5/RIG-I-mediated differentialsignaling. J Immunol 181:2694–2704.

26. Gitlin L, et al. (2006) Essential role of mda-5 in type I IFN responses to polyriboinosinic:polyribocytidylic acid and encephalomyocarditis picornavirus. Proc Natl Acad Sci USA103:8459–8464.

27. Gilfoy FD, Mason PW (2007) West Nile virus-induced interferon production ismediated by the double-stranded RNA-dependent protein kinase PKR. J Virol81:11148–11158.

28. Ahmad-Nejad P, et al. (2002) Bacterial CpG-DNA and lipopolysaccharides activateToll-like receptors at distinct cellular compartments. Eur J Immunol 32:1958–1968.

29. Matsumoto M, et al. (2003) Subcellular localization of Toll-like receptor 3 in humandendritic cells. J Immunol 171:3154–3162.

30. Melchjorsen J, et al. (2005) Activation of innate defense against a paramyxovirus ismediated by RIG-I and TLR7 and TLR8 in a cell-type-specific manner. J Virol79:12944–12951.

31. Boussif O, et al. (1995) A versatile vector for gene and oligonucleotide transfer intocells in culture and in vivo: Polyethylenimine. Proc Natl Acad Sci USA 92:7297–7301.

32. Morille M, Passirani C, Vonarbourg A, Clavreul A, Benoit JP (2008) Progress in devel-oping cationic vectors for non-viral systemic gene therapy against cancer. Biomaterials29:3477–3496.

33. Alexopoulou L, Holt AC, Medzhitov R, Flavell RA (2001) Recognition of double-stranded RNA and activation of NF-kappaB by Toll-like receptor 3. Nature413:732–738.

34. Duramad O, et al. (2005) Inhibitors of TLR-9 act on multiple cell subsets in mouse andman in vitro and prevent death in vivo from systemic inflammation. J Immunol174:5193–5200.

35. Oney S, et al. (2009) Development of universal antidotes to control aptamer activity.Nat Med 15:1224–1228.

36. Fischer S, et al. (2007) Extracellular RNA mediates endothelial-cell permeability viavascular endothelial growth factor. Blood 110:2457–2465.

37. Kannemeier C, et al. (2007) Extracellular RNA constitutes a natural procoagulantcofactor in blood coagulation. Proc Natl Acad Sci USA 104:6388–6393.

38. Davis HE, Rosinski M, Morgan JR, Yarmush ML (2004) Charged polymers modulateretrovirus transduction via membrane charge neutralization and virus aggregation.Biophys J 86:1234–1242.

39. Kimura ET, Young PR, Stein RJ, Richards RK (1959) Some pharmacologic characteristicsof hexadimethrine bromide (polybrene); a new antiheparin agent. Toxicology1:185–202.

40. Biricova V, Laznickova A (2009) Dendrimers: Analytical characterization and applica-tions. Bioorg Chem 37:185–192.

41. Cromer JR, Wood SJ, Miller KA, Nguyen T, David SA (2005) Functionalized dendrimersas endotoxin sponges. Bioorg Med Chem Lett 15:1295–1298.

42. Shaunak S, et al. (2004) Polyvalent dendrimer glucosamine conjugates prevent scartissue formation. Nat Biotechnol 22:977–984.

43. Wang J, et al. (2004) Polyphosphoramidate gene carriers: effect of charge group ongene transfer efficiency. Gene Ther 11:1001–1010.

14060 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1105777108 Lee et al.