Oligonucleotide clamps arrest DNA synthesis on a single-stranded ...

Proc. Natl. Acad. Sci. USAVol. 90, pp. 10013-10017, November 1993Biochemistry

Oligonucleotide clamps arrest DNA synthesis on a single-strandedDNA target

(triplex/psoralen/replication)

CARINE GIOVANNANGELI*, NGUYEN T. THUONGt, AND CLAUDE HtLtNE**Laboratoire de Biophysique, Institut National de la Sante et de la Recherche Mddicale Unite 201, Centre National de la Recherche Scientifique Unit6Associde 481, Museum National d'Histoire Naturelle, 43, Rue Cuvier, 75231 Paris Cedex 05, France; and tCentre de Biophysique Moldculaire,45071 Orleans Cedex 02, France

Communicated by I. Tinoco, June 1, 1993

ABSTRACT Triple helices can be formed on single-stranded oligopurine target sequences by composite oligonu-cleotides consisting of two oligonucleotides covalently linked byeither a hexaethylene glycol linker or an oligonucleotide se-quence. The first oligomer forms Watson-Crick base pairs withthe target, while the second oligomer engages in Hoogsteen basepairing, thereby acting as a molecular clamp. The triple-helicalcomplex formed by such an oligonucleotide clamp, or "oligo-nucleotide4oop-oligonucleotide" (OLO), is more stable thaneither the corresponding trimolecular triple helix or the doublehelix formed upon binding of the oligopyrimidine complementto the same oligopurine target. Attaching a psoralen derivativeto the 5' end of the OLO allowed us to photoinduce a covalentlinkage to the target sequence. The psoralen moiety becamecovalently linked to all three portions of the triplex, therebymaking the oligonucleotide clamp irreversible. These crosslink-ing reactions introduced strong stop signals during DNA rep-lication, as shown on a plasmid containing a portion of the HIVproviral sequence ofhuman immunodeficiency virus. A 16-meroligopurine sequence corresponding to the "polypurine tract"of human immunodeficiency virus was chosen as a target for apsoralen-OLO conjugate. Three different stop signals for DNApolymerase were observed, corresponding to different sites ofpolymerase arrest on its template. Even in the absence ofphotoinduced crosslinking, the psoralen-OLO coijugate wasable to arrest DNA replication. The formation of triple-helicalstructures on single-stranded targets may provide an alterna-tive to the antisense strategy for the control of gene expression.

Oligonucleotides have been used to inhibit the biologicalactivity of RNA molecules in the so-called "antisense"strategy (for review, see ref. 1). In this strategy, an oligonu-cleotide binds to a complementary RNA sequence and in-hibits protein synthesis or viral RNA replication. Oligonu-cleotides can also bind to the major groove of double-stranded DNA, thereby forming triple helices. They are thusable to inhibit replication or transcription of specific genes, inwhat is termed the "antigene" strategy (for review, see ref.2). In both the antisense and antigene strategies, covalentattachment of an activatable reagent at one or both end(s) ofthe oligonucleotide allows irreversible reactions to occur atthe target site (1, 2).We previously showed that a dimeric oligonucleotide ("oli-

gonucleotide-loop-oligonucleotide," or OLO) displayedstrong affinity for single-stranded DNA by forming bothWatson-Crick and Hoogsteen hydrogen bonds with a single-stranded oligopurine target (3). A 24-mer pyrimidine oligo-nucleotide was also shown to bind an 11-mer purine oligo-nucleotide by forming a triplex structure (4). Circular oligo-nucleotides can also form triple-helical structures when

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

bound to a polypurine sequence (5, 6). Here we show thatdifferent linkers can be used to tether the Watson-Crick andHoogsteen base-pair-forming portions of the OLO withoutloss of stability of the resultant triple helix. A psoralenderivative attached to the 5' end of the OLO may becrosslinked to its target sequence in such a way that the threestrands of the resultant triplex become covalently linked toone another. These complexes act as strong stop signals,blocking chain elongation during replication. This inhibitionis much more efficient than that obtained with an antisenseoligonucleotide carrying a reactive psoralen group.

MATERIALS AND METHODSOligonucleotide Synthesis. The unmodified oligonucleo-

tides were obtained from the Pasteur Institute and purified byreverse-phase HPLC. The 16L18-mer OLO (see sequence inFig. 1) was synthesized on a Pharmacia automated synthe-sizer using phosphoramidite chemistry (3, 7). The psoralen-substituted oligopyrimidines Pso-16-mer(p) and Pso-16L18-mer were synthesized from 5-(F-iodohexyloxy)psoralen andthe corresponding unsubstituted oligomer carrying a 5'-thiophosphate group (8).

Plasmid Construction. The plasmid pLTR (a gift from thelate H. Hirel, Rhone-Poulenc-Rorer) was constructed byinsertion of human immunodeficiency virus (HIV) BRUCGprovirus restriction fragments (BamHI-HindIII and HindIlI-Cla I) into pBR328 by standard procedures. pLTR contains1440 bp of HIV proviral DNA carrying a 16-bpoligopurine-oligopyrimidine sequence. The HIV genome con-tains two repeats of the 16-nt oligopurine sequence 5'-AAAAGAAAAGGGGGGA-3'. One oligopurine stretch ispresent on the 5' side of the U3 sequence, within the nefgene(the so-called polypurine tract, positions 8662-8677 in HIVBRUCG or 3526-3541 in pLTR), and the second, which isabsent from pLTR, is located in the 3' region of the pol gene(positions 4367-4382 in HIV BRUCG) (9).

Irradiation Studies. Two targets, a 29R-mer single-strandedoligonucleotide and a double-stranded fragment (abbreviatedD) consisting of the 29R-mer plus a complementary 18-mer(see sequences in Figs. 1 and 3), were used as substrates forpsoralen-induced photo-crosslinking. The 29R-mer was pu-rified by gel electrophoresis and 5'-end labeled with[y-32P]ATP and polynucleotide kinase (Ozyme). The dena-tured linearized plasmid (pLTRs.s.) was also chosen as atarget to form psoralen photoadducts. A Xenon lamp (150 W)in a Cunow housing system provided the light source. Thelight was filtered through a Pyrex filter in water to removeradiation below 310 nm. Electrophoresis was carried out ineither 10% or 8% polyacrylamide gels containing 7 M urea.Quantification of gel autoradiograms was carried out bydensitometry.

Abbreviations: HIV, human immunodeficiency virus; OLO, oligo-nucleotide-loop-oligonucleotide.

10013

10014 Biochemistry: Giovannangeli et al.

Hoogsteen Portion1....... Loop

Watson-Crick Portion

r'lll 111111111lLIII IIIIII 1111II1I

Intercalator

29R-mer CCACTTTTT AAAAGAAAAGGGGGGA CTGG5' 3'

Oligonucleotide Hoogsteen Portion Linker Watson-Crick Portion

5' 3'18-mer T C6T4 CT4 A216-mer T4C T4 C6 T

16 L 18-mer T4 C T4 C6 T -O-(CH2-CH2-0)6- T C6 T4 C T4 A,16 Ts 18-mer T4CT4C6T TTTTT TC6T4 CT4 A-16 T4 18-mer T4CT4C6T TTTT TC6T4 CT4A,16 T3 19-merI T4CT4C6T TTT GTC6T4 CT4A2

FIG. 1. (Upper) Representation of the triple helix formed on a single-stranded nucleic acid containing an oligopurine stretch with anoligonucleotide clamp formed by a Watson-Crick portion and a Hoogsteen portion linked together (OLO). The nature of the linker is indicatedin Lower. The scheme at left represents an OLO whose Watson-Crick and Hoogsteen parts have the same length. The scheme at right showsan intercalator-OLO conjugate whose Watson-Crick part is two bases longer than the Hoogsteen part to allow for intercalation at thetriplex-duplex junction. The intercalator is covalently attached to the 5' end of the Hoogsteen part. (Lower) Sequence of the 29R-mersingle-stranded DNA fragment used as a substrate for binding of the oligonucleotides shown below. The 16-nt oligopurine target sequence isindicated by larger letters. Various composite oligonucleotides consisting of two portions were synthesized. The first portion is either 18 or 19nt long and can form Watson-Crick hydrogen bonds (Watson-Crick portion), while the second is 16 nt long and can form Hoogsteen hydrogenbonds (Hoogsteen portion) with the oligopurine target sequence. Different linkers were used: a hexa(ethylene glycol) linker or a stretch of nthymines (n = 3, 4, or 5). Two separate oligonucleotides corresponding to each of the two portions (18-mer for the Watson-Crick portion and16-mer for the Hoogsteen portion) were used as controls.

Replication Experiments. pLTR was digested with Bsu36IorEcoRV and denatured with 0.2 M NaOH for 15 min at 30°Cto form pLTRs.s. The denatured plasmid (10 nM) wasincubated at 30°C in a 40 mM Tris-HCl, pH 7.5/50 mMNaCl/20 mM MgCl2 in the presence of a 5'-32P-labeledprimer. Two primers were used to replicate each of the twostrands. Primer locations are indicated in Fig. 4. This incu-bation was carried out in either the absence or the presenceof oligonucleotides able to form a complex with the 16-nttarget sequence. Some of the samples were irradiated, whileothers were not. Dithiothreitol (4 mM) and T7 DNA poly-merase (Sequenase version 2.0, 0.13 unit/,ul; United StatesBiochemical) were then added and synthesis was initiated byaddition of 37.5 ,tM dNTPs. The reactions were stopped after50 min by addition of EDTA (50 mM). Sequence analysismade use ofthe same primers, and elongation was carried outin the presence of dNTPs and one ddNTP to allow for chaintermination.

Spectroscopic Methods. Absorption spectra were recordedon a Uvikon 820 spectrophotometer. Melting curves wereobtained by increasing the temperature of 500-,l samples ata rate of 0.15°C/min.

RESULTS AND DISCUSSIONOligonucleotide clamps (Fig. 1) were synthesized with aWatson-Crick oligonucleotide 2 (or 3) nt longer than theHoogsteen part in order to create a triplex-duplex junctionwhen both parts were hydrogen-bonded to the target 16-ntoligopurine sequence (corresponding to the polypurine tractofHIV proviral DNA). An intercalator can insert its aromaticring at this junction (8, 10, 11).

Triple-Helix Formation on Single-Stranded DNA with anOLO. Complex formation between the target 29R-mer andthe oligonucleotide clamps shown in Fig. 1 was followed bymeasuring the temperature dependence of A258 (Fig. 2). Twotransitions were observed when the 18-mer duplex obtainedby mixing the 29R-mer target with the complementary 18-merwas combined with the triple-helix-forming 16-mer oligonu-

cleotide. The transition in the lower temperature range(around 30°C) reflects dissociation of the psoralen-

1.1

1.0

oA-

0)

co

0.9

0.8

0.7-

A &I

10 20 30 40 50 60

temperature(OC)

FIG. 2. Transition curves obtained by measuring A258 as afunction of temperature in a buffer containing 10 mM sodiumcacodylate (pH 6.0) and 100 mM NaCl. A slight excess of 29R-merstrand was used (1.8 AM) to ensure complete binding of the 18-mer(1.5 ,uM) and of the OLO (1.3 ,uM) (see Fig. 1 for abbreviations).Triplex formation ofthe 16-mer (1.2 ,IM) with a duplex solution at 1.5,uM concentration was also followed at 258 nm. The triple-helix-forming oligonucleotide (16-mer) and the 16L18-mer were substi-tuted by psoralen at the 5' end, as described in Fig. 3. x, 29R-merplus 18-mer (mixture abbreviated as D below) (Tm = 480C); o, D plusPso-16-mer (Tm = 31°C and 48°C); A, 29R-mer plus Pso-16L18-mer("Tm" = 460C); *, 29R-mer + 16T319-mer (Tm = 380C and 50°C).When two transitions are observed the first Tm value corresponds to50% dissociation of the Hoogsteen portion from the duplex; thesecond Tm value corresponds to melting ofthe duplex. When a singletransition is observed an apparent "T." is given which correspondsto 50% of the optical transition. The melting curve of 29R-mer with16L18-mer ("Tm" = 420C), as well as that of D plus unsubstituted16-mer (Tm = 20°C and 48°C) can be found in figure 2 of ref. 3.Oligonucleotides 16T418-mer and 16T518-mer gave melting curvessuperimposable with that of 16L18-mer (data not shown).

U000000EWE 0AAAAEU AA,

A0000 AA

00

00 00 00 A,&AA

0S..00AA

u .bi . -.

Proc. Natl. Acad. Sci. USA 90 (1993)

c

Proc. Natl. Acad. Sci. USA 90 (1993) 10015

substituted 16-mer (Pso-16-mer; see Fig. 3 for nomenclature)from its double-stranded target, while the other, occurring inthe upper temperature range (around 50°C), was observed inthe absence of the 16-mer and was attributed to thermaldissociation of the 18 Watson-Crick base pairs of the duplex,as described (10, 12).When the single-stranded 29R-mer target was mixed with

either of the OLOs, the triplex-to-duplex transition wasshifted toward higher temperatures, as compared with thedissociation of the Hoogsteen portion observed when the16-mer was not covalently linked to the 18-mer. A singletransition comprised this profile, whose hyperchromicityapproximately equaled the sum of those for dissociation ofthe 16-mer from the duplex and dissociation of the 18Watson-Crick base pairs. Therefore dissociation of bothWatson-Crick and Hoogsteen hydrogen bonds formed be-tween the purine stretch of the 29R-mer target and the OLOsoccurred in the same temperature range. The same meltingprofile was obtained independent of the linker (hexaethyleneglycol, T4, or T5) used to connect the 16-mer to the 18-mer.A 1:1 OLO/target complex was formed as determined fromtitration experiments.A dimeric oligonucleotide, with a 19-nt Watson-Crick

portion and a 16-nt Hoogsteen portion, 16T319-mer, wastested; with the additional base at the 5' end of the Watson-Crick portion (as compared with the 18-mer oligonucleotide)the distance to the 3' end of the Hoogsteen portion becameminimal and a linker with three thymines was sufficient. This16T319-mer (Fig. 1) was tested for complex formation withthe 29-nt single-stranded target (29R-mer). The 16T319-merformed 19 Watson-Crick base pairs as opposed to the 18formed by the 16T418- and 16T518-mers. The additional CGbase pair on the 3' side of the oligopurine target sequencestabilized the Watson-Crick duplex and facilitated separateanalysis of the dissociations of the two portions of the16T319-mer. Dissociation of the 16-nt Hoogsteen portiontook place around 40°C, compared with 20°C for that corre-sponding to dissociation of the unsubstituted 16-mer from theduplex (result not shown).Chemical Modification of the Target by Oligonucleotide-

Psoralen Conjugates. Psoralen attached to antisense oligonu-cleotides was shown to enhance their biological activity,following UV irradiation (13, 14). A psoralen derivativeattached to the 5' end of antigene oligonucleotides, wasshown to induce efficient covalent crosslinking between thetwo strands of DNA (8, 11), provided that there was a5'-TpA-3'/3'-ApT-5' step at the triplex-duplex junction.When Pso-16L18-mer was incubated and irradiated in the

presence of the labeled 29R-mer for increasing periods oftime, two slowly migrating species were observed uponelectrophoresis (Fig. 3). One slowly migrating species formedat short irradiation times and was converted to the otherduring longer irradiation periods.

Psoralen can form two types of monoadducts with thy-mine, involving either the 3-4 (pyrone) double bond or the4'-5' (furan) double bond (15, 16) (see Fig. 3 for nomencla-ture). Only the 4'-5' (furan) monoadduct can absorb light atwavelengths longer that 310 nm and form the resultingbisadduct. Fig. 3 indicates that the slowest of the two slowlymigrating bands corresponds to the monoadduct (m) formedon the 29R-mer, involving the 4',5' double bond of psoralen.The more compact structure of the bisadduct (b) is probablyresponsible for its slightly faster migration under denaturingconditions (Fig. 3). In this product, the psoralen is linked atthree different positions to three oligonucleotide sequences.It is chemically linked to the 16-mer Hoogsteen portion of theOLO and photochemically adducted to the thymines on boththe 29R-mer and the 18-mer Watson-Crick portion of theOLO. The oligonucleotide clamp is thus made irreversible.

k- 3349 20

a

50-

e (CH2)6 POligo0

0~5'1 1 .,1 3'CCACTTTTT AAA} GAAAGGGGGGA CTGG

-TTTTTTCCCCCCTDooAA TTTTCTTTTCCCCCCT)LOOP3' _ _

2 16W.C. m

m 209

16H18w.c,

29R-mer

A

B

CD

E

16w.dc

FIG. 3. Photoinduced reactions of oligonucleotide-psoralen con-jugate with the 29R-mer. (Upper) Representation of Pso-16L18-merbound to its target 29R-mer with psoralen (rectangle) covalentlyattached (chemically) to the 5'-end of Pso-16L18-mer. 5-Methylcy-tosine was introduced in place of cytosine in the Hoogsteen portionto enhance triple helix stability (10). Two thymines at the triplex-duplex boundary can undergo photoaddition reactions with psoralen(Pso). One is located on the 29R-mer, and the other on the Watson-Crick portion of Pso-16L18-mer. The open rectangle refers to thepsoralen ring attached via its C-S atom (represented at the center ofthe rectangle) to the 16-mers or 16L18-mer. As shown in the box thereactive 4'-5' and 3-4 double bonds of the psoralen ring are positionedat the top and the bottom of the rectangle, respectively. (Lower) Timecourse of photoinduced reaction of Pso-16L18-mer. The 29R-mer (10nM) wasS'-labeled and incubated at 30°C in presence of 1.5,&Moligonucleotide-psoralen conjugate in 40 mM TrisHCl, pH 7.5/50mM NaCl/20 mM MgCl2. Irradiation was performed at 30°C with the150-W xenon lamp for 3, 10, 30, 90, or 210 sec (lanes A-E,respectively) and samples were electrophoresed in a denaturing 10%polyacrylamide gel (the top of the gel is to the right). The nature ofthe photoproducts (m, monoadduct; b, bisadduct) is representedbelow the gel. The scheme of the photoadducted products is notmeant to represent the actual structure but only to represent theoligonucleotides to which the psoralen ring is attached in the differentphotoproducts. Numbers refer to the length of oligonucleotides oneach side of the crosslinked thymines. On the 29R-mer the 4'-5'double bond of psoralen reacts with the 5-6 double bond of thethymine at position 9 from the 5' end; on the 18-mer W.C. portion the3-4 double bond reacts with the thymine at position 16 from the 5' end(see ref. 8). H. and W.C. refer to Hoogsteen and Watson-Crickinteractions, respectively, with the 16-nt-long oligopurine sequenceof the 29R-mer. The 29R-mer is represented by a thin line.

The photochemical reactions of different oligonucleotide-psoralen conjugates were studied by using the 29R-mer DNAtarget. Pso-16L18-mer achieved the highest percentage ofphotoadducts with the labeled 29R-mer. At a concentration of1.5,uM the percentage of crosslinks reached a plateau above60%, compared with either 35% for the intermolecular triplexformed by the 29R-mer, 18-mer, and Pso-16-mer(p) (synthe-sized with 5-methylcytosine) or 15% for the 29R-mer boundto the complementary (antisense) oligomer, 16-mer-Pso(ap),under the same irradiation conditions. The antisense 16-mer-Pso(ap) was synthesized in an antiparallel (ap) orientationwith respect to the 16-nt oligopurine sequence. Psoralen wasattached to the 3' end of 16-mer-Pso(ap) so that it was broughtin close proximity to the same bases of the 29R-mer as thosereached by psoralen linked to either the 5' end of the

Biochemistry: Giovannangeli et al.

10016 Biochemistry: Giovannangeli et al.

Hoogsteen portion of the OLO or to the triplex-forming16-mer, Pso-16-mer(p), which binds in a parallel (p) orienta-tion with respect to the 29R-mer.

Inhibition of Single-Stranded DNA Replication by OLO-Psoralen Conjugates: In Vitro Experiments with T7 DNA Poly-merase. Formation ofpsoralen photoproducts (15-17), as wellas a stable triplex on a duplex target, has been shown to inhibittranscription (18, 19). On a single-stranded nucleic acid, tri-plex formation might be expected to be more efficient inarresting biological processes than formation of a double helixwith an antisense oligonucleotide. To test this hypothesis,denatured, linearized pLTR plasmid (pLTRs.s.) was used as atarget for binding by OLO-psoralen conjugates. The resultingcomplexes were then irradiated, thus crosslinking OLO-psoralen conjugates to the target strand. Primers were used toinitiate replication on either strand after linearization anddenaturation of the plasmid as described in Fig. 4.With the antisense oligomer 16-mer-Pso(ap), bands corre-

sponding to chain termination (Fig. 4, lane 2) appeared at thebases preceding the thymines which had photoreacted on thetemplate strand. This result indicates that psoralen monoad-ducts formed at this site prevented read-through by DNApolymerase. The most important photoadducts were ob-tained at the first two thymines on the 5' side of the oligopu-rine stretch in the target. The hexamethylene linker betweenpsoralen and 16-mer(ap), as well as the dynamics of thesingle-stranded template, can explain the two sites of pho-toadduct formation and so the different sites of termination ofDNA synthesis observed in Fig. 4 (lane 2). Under theconditions used in these studies, the polymerase efficientlydisplaced the complementary antisense oligonucleotide whenit was not crosslinked to the template strand, and full-lengthreplication products were observed.A "standard" (trimolecular) triplex was formed on single-

stranded linearized plasmid (pLTRs.s. Bsu36I), with the

Watson-Crick complementary oligomer (18-mer) and Pso-16-mer(p) forming Hoogsteen hydrogen bonds (Fig. 4, lane3). The 18-mer formed a duplex with the strand of thedenatured plasmid containing the 16-nt oligopurine targetsequence (Fig. 4). Upon addition of Pso-16-mer(p), a triplehelix was formed. After irradiation, psoralen was crosslinkedto the duplex-triplex junction, as described (8, 11). Elonga-tion of the 20R-mer primer stopped at two sites (stops 1 and2 in Fig. 4, lane 3). Stop 1 was identical to that observed withthe antisense 16-mer-Pso(ap) bound to its complementary16-nt oligopurine target sequence, except that a single bandwas observed, instead oftwo bands as seen with the antisenseoligonucleotide. This result reflects the higher selectivity ofthe crosslinking reaction for the 5'-TpA-3' step located at theduplex-triplex junction, compared with the single-strand-duplex junction when the antisense oligomer is used. Stop 2is 7-8 nt upstream from stop 1. The active site of the DNApolymerase, at which chain elongation occurs, is 7-8 nt fromthe edge of the enzyme moving along the template (20, 21).So the simplest explanation is that DNA polymerase isphysically arrested at the crosslinked psoralen during itsprogression along the template. The positions of DNA poly-merase corresponding to stops 1 and 2 are schematicallydepicted in Fig. 4 (Center).The 29R-mer can hybridize to the strand of the denatured

plasmid which contains the 16-nt oligopyrimidine sequence(pLTRs.s. EcoRV) (Fig. 4). The binding of Pso-16-mer(p) tothis duplex was followed by photoinduced crosslinking ofpsoralen at the triplex-duplex junction. When the 20Y-merprimer was used, two efficient stops were observed duringreplication (Fig. 4, lane 6). These two stops have the sameorigin as those observed for the plasmid strand containing theoligopurine sequence under the conditions described above.DNA polymerase can be physically arrested when its activesite reaches the psoralen adduct (stop 1) and when its

[LTRIs.s.(Bsu36 1)

Comllplele 1 2cloigauion

X|-~ 142-pte'er IN 41

(86)26 A4

A4

bpurinej A4G6

(Ro'-7) L.

3 A 4 5a-

FE ~ mm..._ -ks roi'} 1

-._

ti'oI+dXI '

6 +±A

S T1O 1

16

--'~, 5 *

16 C

STOI' 2

.'~~~~~~~~~~ 5'*

16

2 _1 4 - , 5'*

PLTRs.s.(EcoR V)

w- 110f)-r',, (1- FcoR V(8713)

;, (77)

dineI' 1 _,,A662]I(11 2_ .. r*.

20Y-mer(idla 3)le

I+l-1.1 6r.:61X (R 1 ()X!~~~~~~C

t I'ol+(I XT

FIG. 4. Inhibition of DNA synthesis by oligonucleotide-psoralen conjugates. Alkaline-denatured linear pLTR was incubated at 30°C in 40mM Tris HCl, pH 7.5/50 mM NaCl/20 mM MgCl2 in the absence (lane 1) or presence of 16-mer-Pso(ap) (lane 2), 18-mer plus Pso-16-mer(p)(lane 3), Pso-16L18-mer with (lane 4) or without (lane 5) irradiation, and 29R-mer plus Pso-16-mer(p) (lane 6). 5-Methylcytosine was introducedin place of cytosine in the Hoogsteen portion. After irradiation (30°C, 150-W xenon lamp) for 5 min (lanes 1-4 and 6; in lane 5 the sample wasnot irradiated), elongation was performed in standard conditions (see Materials and Methods). In lanes 1-5 the strand containing the 16-ntoligopurine sequence was used as a template (Left). In lane 6 (Right) the strand containing the 16-nt oligopyrimidine sequence was used as atemplate. Sequences were also obtained by primer extension, and lanes A and G + A correspond to an elongation experiment performed in thepresence of ddTTP and of ddCTP plus ddTTP, respectively. Samples were analyzed in a denaturing 8% polyacrylamide gel. The mechanismsfor site-specific arrest of DNA synthesis are schematically represented (Center). In the upper two schemes, the Watson-Crick and Hoogsteenparts are not connected by a loop. They correspond to the results shown in lane 3. The same stops are observed when the Watson-Crick andHoogsteen parts are linked to each other (lane 4). Pol, polymerase.

IBuO6(8(0),))

Proc. Natl. Acad Sci. USA 90 (1993)

_--STOP'3

Proc. Natl. Acad. Sci. USA 90 (1993) 10017

forefront contacts the crosslink, but its active site is 7-8 ntupstream (stop 2).The intensity of stop 1 is higher than that of stop 2 on both

strands of the denatured plasmid (Fig. 4, lanes 3 and 6). Thisresult indicates that the front of DNA polymerase can effi-ciently move across the crosslinked psoralen until its activesite is facing the psoralen adduct, which prevents incorpo-ration of any deoxynucleotide into the growing chain.A triplex was then formed with the oligonucleotide clamp

Pso-16L18-mer on the top strand of the plasmid containingthe oligopurine target sequence. After irradiation, the twopreviously described stops were present, but a third stopappeared outside the triplex site, 12-13 nt upstream from the3' side of the purine stretch in the template (Fig. 4, lane 4).This new termination site (stop 3) is consistent with a physicalarrest ofDNA polymerase by the loop of the triple-strandedcomplex as depicted in Fig. 4 Center. When the polymeraseis arrested at this position, chain elongation should stop 7-8nt upstream from the site ofarrest (as observed for stop 2 withrespect to stop 1). The four additional bases accounting forthe position of the arrest site may be explained by the stericeffect of the linker (loop) used to tether the two oligonucle-otides in Pso-16L18-mer.The concentration of oligonucleotide-psoralen conjugate

which, after irradiation, blocked 50% of the elongation byDNA polymerase was determined from densitometric anal-ysis of the autoradiograms: S ,uM was necessary with theantisense oligonucleotide 16-mer-Pso (ap), whereas 0.1 AMwas sufficient with Pso-16L18-mer. This difference reflectsthe enhanced thermodynamic stability of the 16L18-mertriple-stranded complex and the efficiency of psoralen pho-toaddition in producing mono- and bisadducts, which de-pends in turn on the stability and stereochemistry of thecomplexes.Moreover, in contrast to the unmodified OLO, the OLO-

psoralen conjugate, even without irradiation, was able toform a sufficiently strong, noncovalent complex to arrest theDNA polymerase (Fig. 4, lane 5). Intercalation of psoralen atthe duplex-triplex junction was efficient enough to block theprogression of the enzyme even in the absence of photoprod-ucts with thymine bases. Only stop 2 was observed, inagreement with the physical arrest of the front of the enzymeat the strong duplex-triplex intercalation site. Stop 1 was notobserved, since there was no chemically modified base toinhibit deoxynucleotide incorporation at the active site of theenzyme.

Conclusion. This study shows that it is possible to designoligonucleotides capable of forming both Watson-Crick andHoogsteen hydrogen bonds with a single-stranded nucleicacid containing an oligopurine sequence. These oligonucle-otide clamps have higher binding affinities than do standardantisense Watson-Crick oligomers. These properties mayprove useful in the design of more efficient ligands of single-stranded nucleic acids.

Circular oligonucleotides have been described (5, 6) whichcan bind to an oligopurine sequence more strongly thanOLOs, probably for entropic reasons. However, one of theadvantages of OLOs, besides the ease of synthesis, is thatreactive groups may be attached to either end. The Watson-Crick part of the OLO can be extended by one, two, or morebase pairs in order to create a site for an intercalating agentattached to the 5' end of its Hoogsteen part. Triple-helixformation with an OLO-intercalator conjugate on a single-stranded target was sufficient to arrestDNA replication in theabsence of any irreversible reaction.Attachment of a psoralen moiety to the 5' end of an OLO

improves binding and brings the photoreactive agent into anappropriate position for photoadduct formation with bothstrands ofthe Watson-Crick part. Formation ofthe bisadductwas much more efficient than formation of the monoadduct

with the antisense oligomer. The efficacy of replication arrestwas strongly enhanced when an OLO-psoralen conjugatewas crosslinked to its target site. The OLO-psoralen conju-gate was much more efficient than the corresponding an-tisense oligonucleotide-psoralen conjugate.

Antisense oligonucleotides have been successfully used tobind complementary sequences of messenger or viral RNAs,thereby inhibiting the biological function of the latter (1). Itis possible to form triple helices involving an oligopurineRNA sequence as one of the Watson-Crick strands (22) andto bind an oligopyrimidine RNA sequence to a double helix(23, 24). Formation of a triple-stranded structure on a single-stranded template might prove to be more efficient in block-ing translation, splicing, and reverse transcription than du-plex formation via an antisense oligomer.

We thank Dr G. Duval-Valentin for helpful discussions. This workwas supported in part by Rh6ne-Poulenc-Rorer, by the Ligue Na-tionale Francaise contre le Cancer, and by the Agence Nationale deRecherche contre le SIDA.

1. Helene, C. & Toulme, J. J. (1990) Biochim. Biophys. Acta 1049,99-125.

2. Helene, C. (1991) Anti-Cancer Drug Design 6, 569-584.3. Giovannangeli, C., Montenay-Garestier, T., Rougee, M., Chas-

signol, M., Thuong, N. T. & Helene, C. (1991) J. Am. Chem.Soc. 113, 7775-7777.

4. Xodo, L. E., Manzani, G. & Quadrifoglio, F. (1990) NucleicAcids Res. 18, 3557-3564.

5. Kool, E. T. (1991) J. Am. Chem. Soc. 113, 6265-6266.6. Prakash, G. & Kool, E. T. (1992) J. Am. Chem. Soc. 114,

3523-3527.7. Durand, M., Chevrier, M., Chassignol, M., Thuong, N. T. &

Maurizot, J. C. (1990) Nucleic Acids Res. 18, 6353-6359.8. Takasugi, M., Guendouz, A., Chassignol, M., Decout, J. L.,

Lhomme, J., Thuong, N. T. & Helene, C. (1991) Proc. Natl.Acad. Sci. USA 88, 5602-5606.

9. Wain-Hobson, S., Sonigo, P., Danos, O., Cole, S. & Alizon, M.(1985) Cell 40, 9-17.

10. Sun, J. S., Franqois, J. C., Montenay-Garestier, T., Saison-Behmoaras, T., Roig, V., Thuong, N. T. & Helene, C. (1989)Proc. Natl. Acad. Sci. USA 86, 9198-9202.

11. Giovannangeli, C., Thuong, N. T. & Helne, C. (1992) NucleicAcids Res. 16, 4275-4281.

12. Sun, J. S., Giovannangeli, C., Franqois, J. C., Kurfurst, R.,Montenay-Garestier, T., Asseline, U., Saison-Behmoaras, T.,Thuong, N. T. & Helene, C. (1991) Proc. Natl. Acad. Sci. USA88, 6023-6027.

13. Kulka, M., Smith, C. C., Aurelian, L., Meade, K., Miller, P.& Ts'o, P. 0. P. (1989) Proc. Natl. Acad. Sci. USA 86,6868-6872.

14. Chang, E. H., Miller, P. S., Cushman, C., Devadas, K., Pirolo,K. F., Ts'o, P. 0. P. & Yu, Z. P. (1991) Biochemistry 30,8283-8286.

15. Shi, Y., Gamper, H. & Hearst, J. E. (1987) Nucleic Acids Res.15, 6843-6854.

16. Shi, Y., Gamper, H., Van Houten, B. & Hearst, J. E. (1988) J.Mol. Biol. 199, 277-293.

17. Sastry, S. S. & Hearst, J. E. (1991) J. Mol. Biol. 221, 1091-1110.

18. Young, S. L., Krawczyc, S. H., Matteucci, M. D. & Toole,J. J. (1991) Proc. Natl. Acad. Sci. USA 88, 10023-10026.

19. Duval-Valentin, G., Thuong, N. T. & Helene, C. (1992) Proc.Natl. Acad. Sci. USA 89, 504-508.

20. Ollis, D. L., Brick, P., Hamlin, R., Xuong, N. G. & Steitz,T. A. (1985) Nature (London) 313, 762-766.

21. Joyce, C. M., Ollis, D. L., Rush, J., Steitz, T. A., Konigsberg,W. H. & Grindley, N. D. F. (1986) UCLA Symposia on Mo-lecular and Cellular Biology, ed. Oxender, D. (Liss, NewYork), pp. 197-205.

22. Roberts, R. W. & Crothers, D. (1992) Science 258, 1463-1466.23. Shimizu, M., Konishi, A., Shimada, Y., Inoue, H. & Ohtsuka,

E. (1992) FEBS Lett. 302, 155-158.24. Escude, C., Sun, J. S., Rougee, M., Montenay-Garestier, T. &

HdlEne, C. (1992) C.R. Acad. Sci. Paris Serie III 315, 521-525.

Biochemistry: Giovannangeli et al.