Oligodeoxynucleotides EnhanceLipopolysaccharide-StimulatedSynthesis of Tumor Necrosis Factor:Dependence on PhosphorothioateModification and Reversal by Heparin

Gunther Hartmann, Anne Krug, Kerstin Waller-Fontaine,and Stefan EndresMedizinische Klinik, Klinikum Innenstadt of the Ludwig Maximilians-University, Munich, Germany

ABSTRACT

Background: Specific inhibition of target proteins byantisense oligodeoxynucleotides is an extensively stud-ied experimental approach. This technique is currentlybeing tested in clinical trials applying phosphorothioate-modified oligonucleotides as therapeutic agents. Thesepolyanionic molecules, however, may also exert non-antisense-mediated effects.Materials and Methods: We examined the influenceof oligonucleotides on lipopolysaccharide (LPS)-stimu-lated tumor necrosis factor a (TNFa) synthesis in freshlyisolated human peripheral blood mononuclear cells. Oli-gonucleotides (18 mer) with different degrees of phos-phorothioate modification were studied.Results: The addition of phosphorothioate oligonucleo-tides (5 ,tM) caused amplification of TNF synthesis of upto 410% compared with the control with LPS alone.Without LPS stimulation, phosphorothioate oligonucle-otides did not induce TNF production. We demonstratethat the enhancement of LPS-stimulated TNF productionby phosphorothioate oligonucleotides does not rely on

the intracellular presence of oligonucleotides and isnot mediated by LPS contamination. Partially phos-phorothioate-modified oligonucleotides and un-modified oligonucleotides did not increase TNF syn-thesis. High concentrations of the polyanion heparinreversed the oligonucleotide-induced enhancementof TNF synthesis.Conclusions: The data suggest that amplification of TNFsynthesis may be caused by binding of the polyanionicphosphorothioate oligonucleotide to cationic sites on thecell surface. Such binding sites have been proposed forpolyanionic glycoaminoglycans of the extracellular ma-trix, which have also been described to augment LPS-stimulated TNF synthesis. The present results are rele-vant to all in vitro studies attempting to influence proteinsynthesis in monocytes by using phosphorothioate oli-gonucleotides. The significance of our findings for in vivoapplications of phosphorothioates in situations wherethere is a stimulus for TNF synthesis, such as in sepsis,should be elucidated.

INTRODUCTIONAntisense oligonucleotides have been termed in-formational drugs because of their unique poten-tial to translate the molecular understanding ofdisease into new therapeutic agents. The princi-ple of antisense technique is the sequence spe-cific hybridization of complementary single-stranded oligonucleotides to target RNA thereby

Address correspondence and reprint requests to: Stefan En-dres, Medizinische Klinik der Universitat Munchen, Ziems-senstrale 1, 80336 Munchen, Germany.

Copyright 1996, Molecular Medicine, 1076-1551/96/$10.50/0Molecular Medicine, Volume 2, Number 4, July 1996 429-438

inhibiting translation of the target protein (forreview see Refs. 1-5). The efficacy of antisenseoligonucleotides against neoplasia (6-9), viralinfections (10), and post-traumatic neointimalhyperplasia ( 11) has been established in preclin-ical models. A total of seven oligonucleotideshave entered clinical trials aiming at treatment ofacute and chronic myelogenous leukemia(12-14), human immunodeficiency virus (HIV)infection (15), cytomegalovirus retinitis in ac-quired immunodeficiency syndrome (AIDS) pa-tients (clinical trial phase III [ 16,17]), and genital

429

430 Molecular Medicine, Volume 2, Number 4, July 1996

warts (18). No severe side effects have occurredin human studies to date, including those withsystemic administration (12,15).

Despite numerous reports documenting theefficacy of antisense oligonucleotides in vitro andin vivo in inhibiting gene expression, comple-mentary RNA binding is not their sole mecha-nism of action (2,19,20). Different mechanismsare responsible for non-antisense-mediated ef-fects. They can be divided into sequence-specific(so-called aptamer effect) and non-sequence-specific binding of oligonucleotides to proteins.The latter is primarily based on a charge interac-tion. In addition, hybridization to unintendedRNA targets occur. These mechanisms explainseveral examples of nonantisense effects: (i)stimulation of B lymphocytes by oligonucleo-tides containing the CpG dinucleotide (21); (ii)anti-adhesive effects (19,22); (iii) sequence-spe-cific, but not antisense-mediated, inhibition ofcell proliferation (22,23); and (iv) inhibition ofviral infection at the level of adsorption, pene-tration, or uncoating (4).

Non-antisense-mediated activities may con-tribute to side effects of oligonucleotides in clin-ical studies. Nevertheless, these effects could alsobe beneficial under some circumstances. In fact,oligonucleotides reduce melanoma growth in aSCID-hu mouse model by their interaction withbasic fibroblast growth factor (bFGF) (20). In thepresent study, we demonstrate a new non-anti-sense-mediated effect of phosphorothioate oligo-nucleotides, the enhancement of lipopolysaccha-ride (LPS)-stimulated tumor necrosis factor(TNF) synthesis.

MATERIALS AND METHODSCell CultureHuman peripheral blood mononuclear cells(PBMC) were isolated from blood of healthy fast-ing volunteers by Ficoll-Hypaque gradient cen-trifugation (24) as described previously in detail(25). As a modification of the protocol, centrifu-gation was performed in tubes containing a hor-izontal porous filter disc over the Ficoll layer(Leucosep tubes; Greiner, Frickenhausen, Ger-many), in order to facilitate layering of blood.Cells were suspended in RPMI 1640 culture me-dium (Biochrom, Berlin, Germany) supple-mented with 2 mM L-glutamine, 100 U/ml pen-icillin, 100 jig/ml streptomycin, 10 mM HEPES(all from Sigma, Munich, Germany), 1% human

albumin (Curasan, Kleinostheim, Germany),and 1% heat-inactivated human serum. All com-pounds were purchased endotoxin-tested. Cells,100 ,ul, were seeded at a final concentration of2.5 X 106/ml (370C, 5% CO2 and fully humidi-fied air) in flat bottom 96-well microtiter plates(200 ul/well). Oligonucleotides and/or heparin(Liquemin N, 10 000 IE/ml heparin-sodium,preservant-free, from porcine mucosa, Hoff-mann-La Roche AG, Grenzach-Whylen) wereadded in 100 ,ul of supplemented medium forfinal concentrations as indicated. TNF synthesiswas stimulated by LPS (Escherichia coli 055:B5;Sigma) diluted to a final concentration of10 ng/ml. After 20 hr, addition of trypan blue toselected wells showed 95-97% dye exclusion,indicating preserved cell viability. Incubationwas stopped by freezing the samples at -700C.Cells were further disrupted by completing threefreeze-thaw cycles. TNF concentrations were de-termined in combined cell lysate and superna-tant by specific radioimmunoassay. Results arereported as means of experimental duplicates,unless otherwise indicated.

Oligonucleotides

Oligonucleotides were synthesized by Eurogen-tec (Seraing, Belgium). Lyophilisized oligonucle-otides were diluted in sterile endotoxin-free wa-ter and stored at -200C. The sequences used inthis study are shown in Table 1. Oligonucleotideswere chosen to exhibit minimal self-complemen-tarity (computer program: OLIGO version 4; Na-tional Biosciences Inc., Plymouth, MN). All se-quences were compared for possible homologyto any known nucleotide sequence by a com-puter search in the GENBANK DNA database(EMBO, Heidelberg, Germany) using the com-puter program GCG (Version 7; Genetics Com-puter Group, Madison, WI). Oligodeoxynucle-otide (ODN) 7 (arbitrary numbering) is an18-mer oligonucleotide complementary to nu-cleotides 796-813 of TNF mRNA spanning theinitiation region (see Refs. 26 and 27 for num-bering). ODN 7 contains no CpG-dinucleotideand no more than two neighboring C- or G-nucleotides, motifs which have been shown toconfer a high rate of nonspecific oligonucleotideeffects. In addition, we used the sequence termedODN 19, which contains 15 mismatches in com-parison to ODN 7 (but identical qantitative nu-cleotide composition). Melting temperaturesbased on nearest neighbor thermodynamic pa-rameters (28) are similar. Sequences are shown

G. Hartmann et al.: Phosphorothioate Oligonucleotides and TNF Synthesis 431

TABLE 1. Characteristics of the oligonucleotides used.

Number ofNumber of Phosphorothioate Melting

Name Sequence Mismatches Linkages Temperature

ODN 7cm 5'-CAT GCT TTC AGT GCT CAT-3' 0 17 560CODN 7pm 5'-CAT GCT TTC AGT GCT CAT-3' 0 4 560C

ODN 7um 5'-CAT GCT TTC AGT GCT CAT-3' 0 0 560C

ODN 19cm 5'-TAC TGCAGG ATT CTC TTC-3' 15 17 510C

Nucleotides on either side of a phosphorothioate linkage are indicated in bold letters. Mismatched nucleotides are in italics. Themelting temperatures are estimated by the nearest neighbor method. The G/C content for all four oligonucleotides is 56%. A,adenosine; G, guanosine; C, cytosine; T, thymidine.

in Table 1. ODN 7 was used completely phospho-rothioate-modified (ODN 7cm), partially phos-phorothioate-modified (ODN 7pm) and unmod-ified (ODN 7um). ODN 19 was used completelymodified (ODN 19cm). In phosphorothioate oli-gonucleotides one oxygen atom of the phosphategroup is substituted by a sulfur atom.

Radioimmunoassay for TNF

TNF was determined by specific radioimmunoas-say as previously described (29). In order to ra-tionalize sample processing, a 96-microtube platesystem with single polypropylene tubes(Sarstedt, Niimbrecht, Germany) was used. Thesample (50 ,lA) was added to 50 [lI of dilutedpolyclonal anti-TNF rabbit antiserum and 50 gl1% rabbit IgG and was incubated overnight.Bolton Hunter-labeled 1251-TNF (50 ,ul) (NEN/DuPont, Munich, Germany) was added on thesecond day. After another overnight incubation,250 [lI of second antibody (sheep anti-rabbit IgG)in 6% polyethylene glycol was added. TNF con-centrations were calculated from a standardcurve of human recombinant TNF (supplied bythe National Institute for Biological Standardsand Control, Potters Bar, United Kingdom),ranging from 0.02 to 10 ng/ml. The presence ofoligonucleotide or heparin did not influence themeasurement of TNF.

Endotoxin AssayTo exclude contaminations with endotoxin, alloligonucleotides were tested in the limulusamoebocyte lysate assay (Chromogenix, Charl-ston, SC, U.S.A.) and were found endotoxin neg-ative (endotoxin content less than 6.0 pg/ml).

Statistical AnalysisResults are given as means ± SEM. The pairedtwo-tailed Student's t test was performed forcomparisons of means of TNF values. Differenceswere considered statistically significant for p <0.050. Statistical analyses were performed by us-ing Stat-View 512 software (Abacus Concepts,Calabasas, CA, U.S.A.).

RESULTSConcentration-Dependent Enhancementof LPS-Stimulated TNF Synthesis byPhosphorothioate OligonucleotidesHuman mononuclear cells stimulated with10 ng/ml of LPS for 20 hr synthesized 6.5 ng/mlTNF. Addition of increasing concentrations ofphosphorothioate oligonucleotides led to en-hanced TNF synthesis. Maximal TNF synthesis (upto 410% of control) was found for 2.5 and 5 ,uMODN 7 (Fig. 1). Higher concentrations showed re-duced efficacy in augmenting LPS-stimulated TNFsynthesis. As low as 0.078 ,uM was effective. Sub-sequent experiments were performed using theconcentration of 5 AM oligonucleotide.

Enhancement of TNF Synthesis IsDependent on the Degree ofPhosphorothioate ModificationThe phosphorothioate modification alters themolecular structure of oligonucleotides. Despiteits advantages over unmodified oligonucleotideswith respect to stability, this modification in par-ticular has been described to cause nonantisenseeffects. We studied the influence of the degree of

432 Molecular Medicine, Volume 2, Number 4, July 1996

TNF 1[ng/ml]

0 0.078 0.156 0.313 0.625 1.25 2.5 5 10

ODN 7cm [1iM]

FIG. 1. Enhancement of LPS-stimulated TNFsynthesis by phosphorothioate oligonucleotidesHuman mononuclear cells (2.5 Mio/ml) were stimu-lated with 10 ng/ml LPS in the presence of the com-

pletely phosphorothioate-modified oligonucleotideODN 7cm (0.078-10 ,uM). After 20 hr of incubation,TNF was measured using a specific RIA. Increasingconcentrations of the oligonucleotide enhance TNFsynthesis up to 410% of the control (a). WithoutLPS stimulation (C), TNF synthesis was below thedetection limit of the RIA. Data are shown as means

of duplicates of a representative experiment.

phosphorothioate modification on enhancementof LPS-stimulated TNF synthesis.

The oligonucleotide ODN 7 was compared inthree different forms: completely phosphoro-thioate-modified (ODN 7cm), partially modified(ODN 7pm) and unmodified (ODN 7um), eachwith identical nucleotide sequence. Neither theoligonucleotide with two phosphorothioate nu-

cleotides at the 5' and the 3' end nor the un-

modified oligonucleotide led to an increase inTNF synthesis in comparison to the control withLPS (Fig. 2). However, at the same concentrationthe completely modified oligonucleotide (5 ,uM)showed a reproducible enhancement of TNF syn-

thesis to a mean of 290%. Without LPS stimula-tion, little TNF was induced by completely mod-ified oligonucleotides (mean 0.78 ng/ml TNF),and still less by partially modified and unmodi-fied oligonucleotides (at the detection limit of theRIA; i.e., 0.20 ng/ml TNF).

Enhancement of TNF Synthesis Is NotDependent on the Sequence of theOligonucleotideBoth sequence-dependent and sequence-inde-pendent nonantisense effects have been de-scribed for phosphorothioate oligonucleotides. Weexamined the role of the nucleotide sequence re-

garding the capacity of oligonucleotides in aug-

menting LPS-stimulated TNF synthesis.

ODN 7 ODN 7 ODN 7 ODN 19unmodified partially completely completely

modified modified modified

FIG. 2. Enhancement of TNF synthesis is de-pendent on the degree of phosphorothioatelinkages and independent of the sequence

Human mononuclear cells (2.5 Mio/ml) were stimu-lated with 10 ng/ml LPS for 20 hr in the presence ofthe oligonucleotides indicated (all at 5 ,uM). No dif-ference of TNF synthesis was found between thecontrol with LPS alone (U) and in the presence ofthe partially modified oligonucleotide ODN 7pm(two phosphorothioate nucleotides at the 5' and the3' end) and the unmodified oligonucleotide ODN7um. LPS-stimulated TNF synthesis was enhancedonly for the completely modified oligonucleotidesODN 7cm and ODN 19cm. Without LPS stimulation(C) only small induction occured in the presence ofthe completely modified oligonucleotides ODN 7cmand ODN 19cm, whereas TNF synthesis was at thedetection limit in the presence of the partially andunmodified ODN 7. Results are given as means ofthree independent experiments. *p < 0.05; error

bars = SEM.

The sequence complementary to the startsite of TNF-mRNA (ODN 7) was compared with a

mismatched control (ODN 19, same nucleotidecontent as ODN 7). In the presence of ODN 7cm,LPS-stimulated TNF synthesis was enhancedfrom 3.3 ng/ml (LPS stimulation alone) to9.8 ng/ml. Despite the TNF sequence specifity ofODN 7, no significant difference in TNF amplifi-cation was found between ODN 7cm (mean300%) and ODN 19cm (mean 280%; Fig. 2, rightbar). No sequence-specific suppression of TNFsynthesis could be achieved by applying fiveother TNF-specific sequences in this experimen-tal system (data not shown).

Time Dependence of Enhanced TNFSynthesis by PhosphorothioateOligonucleotidesSince partially and unmodified oligonucleotidesshow no difference in TNF production from the

TNF 30-

[ng/ml] 25-

20-

15-

10-

5.

( p m

I

4 I/- 11

G. Hartmann et al.: Phosphorothioate Oligonucleotides and TNF Synthesis 433

TNF[lftq10-

14

12

10-

6-

4-

2-

ODN 7* cowI8modbdODN 19

a .-modlb

ODN 7

modleid* 00N7

WmliotmWdOO..

0 1 2 4 6 8 20Tim (h)

FIG. 3. Enhancement of TNF synthesis byphosphorothioate oligonucleotides is timedependentHuman mononuclear cells (2.5 Mio/ml) were stimu-lated with 10 ng/ml LPS in the presence of oligonu-cleotides added (5 ,uM). TNF synthesis was measuredat different time points. In the presence of com-pletely phosphorothioate-modified oligonucleotides(ODN 7cm and ODN 19cm) TNF synthesis was en-hanced at all time points 4 hr or more after LPSstimulation. No plateau of TNF production wasreached up to 20 hr. In contrast, the partially modi-fied oligonucleotide ODN 7pm and the unmodifiedoligonucleotide ODN 7um did not enhance TNF syn-thesis compared with stimulation with LPS alone.

control with LPS alone at 20 hr we investigatedwhether there was any difference at earlier timepoints. TNF synthesis was examined in the pres-ence of oligonucleotides at consecutive timepoints (1, 2, 4, 6, 8, and 20 hr after LPS stimu-lation). In the presence of completely phospho-rothioate-modified oligonucleotides (ODN 7cmand ODN 19cm), TNF synthesis was enhanced atall time points 4 hr or longer after LPS stimula-tion (Fig. 3). A plateau in TNF production was

not reached for up to 20 hr. In contrast, thepartially modified oligonucleotide ODN 7pm andthe unmodified oligonucleotide ODN 7um didnot enhance TNF synthesis compared with stim-ulation with LPS alone at all time points tested.

High Concentrations of Heparin ReverseEnhancement of TNF Synthesis byPhosphorothioate OligonucleotidesA possible mechanism for the phosphorothioateoligonucleotide-mediated enhancement of TNFsynthesis may be through their property ofpolyanionic macromolecules. We therefore in-vestigated the influence of other polyanions on

phosphrothioate oligonucleotide-enhanced TNF

TNF It[ng/ml]

1'

1:.

1

0 12.5 25 50 100 200

Heparin [IE/ml]

FIG. 4. Heparin reverses enhancement of TNFsynthesisHuman mononuclear cells (2.5 Mio/ml) were stimu-lated with 10 ng/ml LPS for 20 hr. The influence ofincreasing concentrations of heparin on the en-hancement of TNF production by phosphorothioateoligonucleotides (5 ,uM ODN 7cm) was tested. At aconcentration above 25 IE/ml heparin reverses en-hanced TNF synthesis to the level of the controlwith LPS alone. Reversal by heparin suggests dis-placement of oligonucleotides from the binding sitesthat mediate enhanced TNF synthesis. Influence ofheparin on TNF measurement by the RIA wasexcluded.

synthesis. Heparin is a ubiquitous, naturally oc-curing polyanionic polysaccharide, which is clin-ically used for its antithrombotic effect.

Mononuclear cells were stimulated in thepresence of a completely modified oligonucleo-tide (ODN 7cm, 5 ,uM) and increasing concen-trations of heparin (Fig. 4). The addition of hep-arin (12.5-200 IE/ml) without oligonucleotidesled to a small increase of TNF production whichwas not concentration dependent. The TNF syn-thesis enhanced by phoshorothioate oligonucle-otides was markedly reduced, from 14.7 ng/m to10.1 ng/ml TNF, in the presence of 12.5 IE/mlheparin. Higher concentrations of heparin(50 IE/ml and more) completely reversed theenhancement of TNF synthesis by phosphoro-thioate oligonucleotides to the level achieved byLPS alone.

DISCUSSIONAntisense oligonucleotides were initially thoughtto have the unique potential of sequence-specificinhibition of single target proteins. Today, the

n4

TrkIC xCe

--L-v 1-1 .1 I I I .I It T

434 Molecular Medicine, Volume 2, Number 4, July 1996

concept of absolute specifity has been replacedby the acceptance of nonantisense effects in ad-dition to the antisense mechanism. Non-anti-sense-mediated immunological effects includethe induction of interferon-y and of IgG and IgMsynthesis by oligonucleotides. These effects aredependent on defined motifs within the se-quence (21). Numerous sequence-independentactions of oligonucleotides have also been reportedin the literature (30-32). It has been proposed thatnonspecific effects frequently predominate whenphosphorothioate oligonucleotides are employedat concentrations higher than 5 ,uM (19).

In the present study, we describe the influ-ence of phosphorothioate oligonucleotides onLPS-induced TNF synthesis in human mononu-clear cells. We found an augmentation of TNFsynthesis of up to 410% in the presence of 5 ,uMphosphorothioate oligonucleotide. Enhance-ment of LPS-stimulated TNF synthesis was con-centration dependent, with a maximum between2.5 and 5 ,tM. It was detectable for lower con-centrations (down to 0.078 ,uM) than those pro-posed to be effective in the plasma of patients ( 12).In contrast to completely phosphorothioate-modi-fied oligonucleotides, oligonucleotides with onlytwo modified nucleotide bonds at each end or un-modified oligonucleotides showed no enhance-ment of TNF synthesis. The effect was independentof the sequence of the oligonucleotide.

The augmentation of TNF synthesis is notlikely to be mediated by a direct action at the LPSreceptor. We excluded relevant endotoxin con-tamination of the oligonucleotide preparationsby using the limulus amoebocyte lysate assay.Furthermore, reversal of enhanced TNF synthe-sis by heparin argues against an effect of LPS.

The mechanism of the observed effect re-mains unknown. It likely relates to differences inthe structures of modified versus unmodified oli-gonucleotides. In phosphorothioate oligonucleo-tides, one oxygen atom of the phosphate group issubstituted by a sulfur atom. Phosphorothioateoligonucleotides represent polyanions withstructural similarities to the naturally occuringpolyanionic glycosaminoglycans of the extracel-lular matrix, such as heparan sulfate and heparin(Fig. 5). There are other examples of polyanionicmacromolecules that enhance TNF synthesis inmonocytic cells. Heparan sulfate increases LPS-stimulated TNF release from interferon->-treated THP- 1 cells, a human monocytic cell line,by up to 500%. Chondroitin sulfate, keratan sul-fate, and dermatan sulfate enhance LPS-stimu-lated TNF synthesis in THP-1 cells up to 200%

(33). Chang et al. argued that glycosaminogly-cans assist in the priming step of THP-1 cells,which is associated with enhanced TNF synthesisand secretion.

Hershkoviz et al. described other immobi-lized components of the extracellular matrixwhich induce TNF secretion in human mononu-clear cells (34). The authors found the interac-tion between two glycoproteins of the extracel-lular matrix, laminin and fibronectin, and ,B1-integrins of the cell surface to be responsible forTNF production without requiring an additionalspecific stimulus. It has also been demonstratedthat collagen stimulates the production and se-cretion of interleukin 1 in human mononuclearcells (35). Together, these results argue that thecomponents of the extracellular matrix play animportant role in potentiating cytokine synthesisby monocytes.

With regards to the binding of phosphoro-thioates to the surface of monocytes, it is of notethat circulating human monocytes bind heparinin a rapid, saturable, and reversible manner(36,37). The affinity of heparin for the cell sur-face is independent of that for antithrombin III.In addition to sharing molecular features withheparin, phosphorothioate oligonucleotideshave been described to bind directly to heparin-binding proteins, such as basic fibroblast growthfactor (bFGF), acidic fibroblast growth factor(aFGF), FGF-4, platelet-derived growth factor(PDGF), and vascular endothelial growth factor(VEGF) (38). Oligonucleotides and sulfated gly-cosaminoglycans share binding sites on severalother proteins. These include CD4, HIV- 1 reversetranscriptase, the HIV-1 envelope glycoproteinand the protein kinase C 131-subunit (PKC11)(1 9). Since similar binding properties for heparinand oligonucleotides have been proposed, oligo-nucleotides may indeed bind to the surface ofmonocytes at the same sites as heparin.

Recently, another nonantisense mechanismmediated by the extracellular action of phospho-rothioate oligonucleotides has been described(22). The non-sequence-specific inhibition of ad-hesion of glioblastoma cells could clearly be sep-arated from the sequence-specific antisense ef-fect. An excess of fibronectin, a glycoprotein ofthe extracellular matrix, restored the capabilityof the cells to adhere to the bottom of the cultureplate. It was proposed that phosphorothioate oli-gonucleotides compete with molecules of the ex-tracellular matrix such as fibronectin for bindingsites at the cell surface. Both in that study and inthe present experiments, the effect was abolished

G. Hartmann et al.: Phosphorothioate Oligonucleotides and TNF Synthesis

3ase

H

H

0

05-

CH2

H

-01**

0

OSo-0

a-L-iduronic acid-2-sulfate a-2-sulfamidic acid-D-glucose-6-sulfate

Heparin-disaccharide

FIG. 5. Structural similarities between the polyanions phosphorothioate-deoxyribonucleic acid andheparinIn phosphorothioate-modified oligodeoxynucleic acid, one free oxygen atom of the phosphate is replaced by a sul-fur atom, providing increased stability against nucleases. Base, the linkage to one of the bases adenosine,guanosine, cytosine, and thymidin. For heparin one of several possible sulfated disaccharides of the polysaccharidechain is shown. Both heparin and phosphorothioate-deoxyribonucleic acid represent polyanions. Groups of themolecules that are negatively charged under cell culture conditions are indicated with bold letters.

using oligonucleotides with a low degree ofphosphorothioate modification or unmodifiedoligonucleotides. This suggests that the bindingsites of phosphorothioate oligonucleotides thatmediate the inhibition of adhesion and the in-crease in TNF synthesis are identical.

The priming monocytes by adherence resultsin cell activation (39). While adherence is suffi-cient to induce high steady-state levels of TNFmRNA in monocytes (40), actual translation andsecretion of TNF requires the exposure to a sec-

ond signal (39). As summarized above, glyco-saminoglycans of the extracellular matrix are

unable to induce TNF production by themselves,but promote TNF production when induced by a

specific stimulus such as LPS. A biological role forthis phenomenon may be that TNF synthesis ismainly restricted and strengthened in the extra-cellular compartment at the site of local inflam-mation. The underlying mechanism is likely toinvolve receptors that recognize components of

the extracellular matrix. In the experiments de-scribed here, an excess of heparin reversed theenhancement of TNF synthesis by phosphoro-thioate oligonucleotides and restored levels sim-ilar to those achieved by LPS alone. Since phos-phorothioate oligonucleotides and heparin are

both polyanions and share molecular features,we propose that heparin displaces the oligonu-cleotides that mediate enhanced TNF synthesisfrom their binding sites on the cellular surface.Interestingly, heparin has been demonstrated toreduce the inflammatory response in vivo (41).One may speculate that the observed TNF-en-hancing effect of phosphorothioate oligonucleo-tides, sharing molecular features with glycosami-noglycans of the extracellular matrix, representsa molecular mimicry of extravasal location.

Our findings are relevant to all in vitro stud-ies attempting to influence protein synthesis inmonocytes by phosphorothioate oligonucleo-tides. While we have examined LPS-induced

Phosphorothioate-deoxyribonucleic acid

435

I

436 Molecular Medicine, Volume 2, Number 4, July 1996

TNF synthesis, the enhancing effect may alsooccur using other stimuli such as phagocytizedbacteria, and other monocyte-released proteinssuch as interleukin 1, interleukin 6, or tissuefactor. If the enhancing effect is not accountedfor, it may mask the intended inhibition of syn-thesis of a target protein. There are several ex-perimental approaches to circumvent or mini-mize this non-antisense-mediated effect. First, aswe have shown here, the effect can be mini-mized by reducing of the number of phosphoro-thioate-modified bonds within the oligonucleo-tide. This, in turn, has to be balanced withreduced stability of the oligonucleotide. Anotherapproach, which we have shown in other studies(Hartmann et al., "Specific suppression of tumornecrosis factor a synthesis by antisense oligode-oxynucleotides," manuscript submitted), consistsof removing oligonucleotide from the extracellu-lar compartment through washing steps beforestimulation with LPS. Finally, liposomal encap-sulation of oligonucleotides facilitates penetra-tion into the cells and may avoid nonantisenseaction of oligonucleotides in the extracellularcompartment.

Whether the present findings are relevant toin vivo situations is not known. As phosphoro-thioate oligonucleotides alone did not induceTNF synthesis, one would not expect such aninduction after systemic administration underordinary conditions. In the case of exogenous orendogenous endotoxinemia (e.g., gram-negativesepsis), however, the possibility of enhanced TNFsynthesis in the presence of oligonucleotideswould have to be considered. In this context, it isof note that in monkeys a bolus injection of adose higher than that proposed to be effective inhumans of phosphorothioate oligonucleotidescaused a lethal decrease of blood pressure (42).

Finally, this study adds to the existing evidencethat sulfated polyanions enhance LPS-induced ac-tivation of monocytes. By analogy to glycosamino-glycans (33), fibronectin (34), and collagen (35),we have identified phosphorothioate oligonucleo-tides as a new class of macromolecules that partic-ipate in the priming of monocytes.

ACKNOWLEDGMENTSThe authors thank Dr. Andreas Eigler and Dr.Jochen Moller for helpful discussion and Chris-tiane Haslberger for excellent technical assis-tance. This work was made possible by Grant93.0422 from the Wilhelm Sander-Stiftung.

REFERENCES1. Askari FK, McDonnell WM. (1996) Anti-

sense-oligonucleotide therapy. N. Engl.J. Med. 334: 316-318.

2. Sharma HW, Narayanan R. (1995) The ther-apeutic potential of antisense oligonucleo-tides. Bioessays 17: 1055-1063.

3. Stein CA, Narayanan R. (1994) Antisenseoligodeoxynucleotides. Curr. Opin. Oncol. 6:587-594.

4. Wagner RW. (1994) Gene inhibition usingantisense oligodeoxynucleotides. Nature 372:333-335.

5. Rossi JJ. (1995) Therapeutic antisense andribozymes. Br. Med. Bull. 51: 217-225.

6. Kitajima I, Shinohara T, Bilakovics J, BrownDA, Xu X, Nerenberg M. (1992) Ablation oftransplanted HTLV-I tax-transformed tu-mors in mice by antisense inhibition of NF-KB. Science 258: 1792-1795.

7. Higgins KA, Perez JR, Coleman TA, et al.(1993) Antisense inhibition of the p65 sub-unit of NF-KB blocks tumorigenicity andcauses tumor regression. Proc. Natl. Acad. Sci.U.S.A. 90: 9901-9905.

8. Schwab G, Chavany C, Duroux I, et al.(1994) Antisense oligonucleotides adsorbedto polyalkylcyanoacrylate nanoparticles spe-cifically inhibit mutated Ha-ras-mediatedcell proliferation and tumorigenicity in nudemice. Proc. Natl. Acad. Sci. U.S.A. 91: 10460-10464.

9. Skorski T, Nieborowska SM, Nicolaides NC,et al. (1994) Suppression of Philadelphialleukemia cell growth in mice by bcr-abl an-tisense oligodeoxynucleotide. Proc. Natl.Acad. Sci. U.S.A. 91: 4504-4508.

10. Offensperger WB, Offensperger S, Walter E,et al. (1993) In vivo inhibition of duck hep-atitis B virus replication and gene expressionby phosphorothioate-modified antisense oli-godeoxynucleotides. EMBO J. 12: 1257-1262.

11. Morishita R, Gibbons GH, Ellison KE, et al.(1994) Intimal hyperplasia after vascular in-jury is inhibited by antisense cdk 2 kinaseoligonucleotides. J. Clin. Invest. 93: 1458-1464.

12. Bayever E, Iversen PL, Bishop MR, et al.(1993) Systemic administration of a phos-phorothioate oligonucleotide with a se-quence complementary to p53 for acute my-elogenous leukemia and myelodysplasticsyndrome: Initial results of a phase I trial.Antisense Res. Dev. 3: 383-390.

G. Hartmann et al.: Phosphorothioate Oligonucleotides and TNF Synthesis 437

13. Nichols GL. (1995) Antisense oligode-oxynucleotides as therapeutic agents forchronic myelogenous leukemia. AntisenseRes. Dev. 5: 67-69.

14. Gewirtz AM. (1994) Treatment of chronicmyelogenous leukemia (CML) with c-mybantisense oligodeoxynucleotides. Bone Mar-row Transplant. 14: S57-S61.

15. Zhang R, Yan J, Shahinian H, et al. (1995)Pharmacokinetics of an anti-human immu-nodeficiency virus antisense oligode-oxynucleotide phosphorothioate (GEM 91)in HIV-infected subjects. Clin. Pharmacol.Ther. 58: 44-53.

16. Cohen J. (1995) AIDS therapy. New hopeagainst blindness. Science 268: 368-369.

17. Hawkins JW. (1995) Oligonucleotide thera-peutics: Coming 'round the clubhouse turn.Antisense Res. Dev. 5: 1.

18. Gura T. (1995) Antisense has growing pains.Science 270: 575-577.

19. Stein CA. (1995) Does antisense exist? Na-ture Medicine 1: 1119-1121.

20. Jansen B, Wadl H, Inoue SA, et al. (1995)Phosphorothioate oligonucleotides reducemelanoma growth in a SCID-hu mousemodel by a nonantisense mechanism. Anti-sense Res. Dev. 5: 271-277.

21. Krieg AM, Yi AK, Matson S, et al. (1995)CpG motifs in bacterial DNA trigger directB-cell activation. Nature 374: 546-549.

22. Chavany C, Connell Y, Neckers L. (1995)Contribution of sequence and phosphoro-thioate content to inhibition of cell growthand adhesion caused by c-myc antisense oli-gomers. Mol. Pharmacol. 48: 738-746.

23. Obrien SG, Kirkland MA, Melo JV, et al.(1994) Antisense BCR-ABL oligomers causenon-specific inhibition of chronic myeloidleukemia cell lines. Leukemia 8: 2156-2162.

24. B0yum A. (1968) Isolation of mononuclearcells and granulocytes from human blood(paper IV). Scand. J. Clin. Lab. Invest. 21: 77-89, 107-109.

25. Endres S, Fiille HJ, Sinha B, et al. (1991)Cyclic nucleotides differentially regulate thesynthesis of tumour necrosis factor-a andinterleukin- 1O by human mononuclearcells. Immunology 72: 56-60.

26. Pennica D, Nedwin GE, Hayflick JS, et al.(1984) Human tumour necrosis factor: Pre-cursor structure, expression and homologyto lymphotoxin. Nature 312: 724-729.

27. Nedwin GE, Naylor SL, Sakaguchi AY, et al.(1985) Human lymphotoxin and tumor ne-

crosis factor genes: Structure, homology andchromosomal localization. Nucleic Acids Res.13: 6361-6373.

28. Breslauer KJ, Frank R, Blocker H, MarkeyLA. (1986) Predicting DNA duplex stabilityfrom the base sequence. Proc. Natl. Acad. Sci.U.S.A. 83: 3746-3750.

29. van der Meer JW, Endres S, Lonnemann G,et al. (1988) Concentrations of immunore-active human tumor necrosis factor a pro-duced by human mononuclear cells in vitro.J. Leukocyte Biol. 43: 216-223.

30. Perez JR, Li Y, Stein CA, Majumder S, vanOorschot A, Narayanan R. (1994) Sequence-independent induction of Spl transcriptionfactor activity by phosphorothioate oligode-oxynucleotides. Proc. Natl. Acad. Sci. U.S.A.91: 5957-5961.

31. Weidner DA, Valdez BC, Henning D, Green-berg S, Busch H. (1995) Phosphorothioateoligonucleotides bind in a non sequence-specific manner to the nucleolar protein c2 3/nucleolin. FEBS Lett. 366: 146-150.

32. Bergan RC, Kyle E, Connell Y, Neckers L.(1995) Inhibition of protein-tyrosine kinaseactivity in intact cells by the aptameric actionof oligodeoxynucleotides. Antisense Res. Dev.5: 33-38.

33. Chang NS, Intrieri C, Mattison J, Armand G.(1994) Synthetic polysulfated hyaluronicacid is a potent inhibitor for tumor necrosisfactor production. J. Leukoc. Biol. 55: 778-784.

34. Hershkoviz R, Gilat D, Miron S, et al. (1993)Extracellular matrix induces tumour necro-sis factor-a secretion by an interaction be-tween resting rat CD4+ T cells and macro-phages. Immunology 78: 50-57.

35. Pacifici R, Basilico C, Roman J, Zutter MM,Santoro SA, McCracken R. (1992) Collagen-induced release of interleukin 1 from humanblood mononuclear cells. Potentiation by fi-bronectin binding to the alpha 5 beta 1 in-tegrin. J. Clin. Invest. 89: 61-67.

36. Dawes J. (1993) Interactions of heparins inthe vascular environment. Haemostasis 1:212-219.

37. Leung L, Saigo K, Grant D. (1989) Heparinbinds to human monocytes and modulatestheir procoagulant activities and secretoryphenotypes. Effects of histidine-rich glyco-protein. Blood 73: 177-184.

38. Stein CA, Cheng YC. (1993) Antisense oli-gonucleotides as therapeutic agents-Is thebullet really magical? Science 261: 1004-1012.

39. Schindler R, Clark BD, Dinarello CA. (1990)

438 Molecular Medicine, Volume 2, Number 4, July 1996

Dissociation between interleukin- 1 betamRNA and protein synthesis in human pe-ripheral blood mononuclear cells. J. Biol.Chem. 265: 10232-10237.

40. Haskill S, Johnson C, Eierman D, Becker S,Warren K. (1988) Adherence induces selec-tive mRNA expression of monocyte media-tors and proto-oncogenes. J. Immunol. 140:1690-1694.

41. Ekre HP, Naparstek Y, Lider 0, et al. (1992)

Anti-inflammatory effects of heparin and itsderivatives: Inhibition of complement and oflymphocyte migration. Adv. Exp. Med. Biol.313: 329-340.

42. Galbraith WM, Hobson WC, Giclas PC,Schechter PJ, Agrawal S. (1994) Comple-ment activation and hemodynamic changesfollowing intravenous administration ofphosphorothioate oligonucleotides in themonkey. Antisense Res. Dev. 4: 201-206.

Contributed by A. Cerami on April 26, 1996.