Proc. Natl. Acad. Sci. USAVol. 93, pp. 3199-3204, April 1996Biochemistry

Sequence-specific arrest of primer extension on single-strandedDNA by an oligonucleotide-minor groove binder conjugate

(hybridization/DNA synthesis inhibition/antisense/antigene)

IRINA AFONINA, IGOR KUTYAVIN, EUGENY LUKHTANOV, RICH B. MEYER, AND HOWARD GAMPER*

Epoch Pharmaceuticals, 1725 220th Street SE, No. 104, Bothell, WA 98021

Communicated by Robert L. Letsinger, Northwestern University, Evanston, IL, December 26, 1995

ABSTRACT A minor groove binder (MGB) derivative(N-3-carbamoyl-1,2-dihydro-3H-pyrrolo [3,2-e] indole-7-carboxylate tripeptide; CDPI3) was covalently linked to the5' or 3' end of several oligodeoxyribonucleotides (ODNs)totally complementary or possessing a single mismatch toM13mpl9 single-stranded DNA. Absorption thermal dena-turation and slot-blot hybridization studies showed thatconjugation of CDPI3 to these ODNs increased both thespecificity and the strength with which they hybridized.Primer extension of the same phage DNA by a modified formof phage T7 DNA polymerase (Sequenase) was physicallyblocked when a complementary 16-mer with a conjugated5'-CDPI3 moiety was hybridized to a downstream site.Approximately 50% of the replicating complexes were ar-rested when the blocking ODN was equimolar to the phageDNA. Inhibition was unaffected by 3'-capping of the ODNwith a hexanol group or by elimination of a preannealingstep. Blockage was abolished when a single mismatch wasintroduced into the ODN or when the MGB was eitherremoved or replaced by a 5'-acridine group. A 16-mer witha 3'-CDPI3 moiety failed to arrest primer extension, as didan unmodified 32-mer. We attribute the exceptional stabilityof hybrids formed by ODNs conjugated to a CDPI3 to thetethered tripeptide binding in the minor groove of thehybrid. When that group is linked to the 5' end of ahybridized ODN, it probably blocks DNA synthesis byinhibiting strand displacement. These ODNs conjugated toCDPI3 offer attractive features as diagnostic probes andantigene agents.

Under conditions where oligodeoxyribonucleotides (ODNs)form stable hybrids, long single-stranded DNA and RNAhave considerable secondary and tertiary structure whichcan interfere with the use of ODNs as diagnostic probes orantisense agents (1-3). Numerous approaches to overcomingthese barriers have been described. Generally, these sort intotwo mutually compatible strategies. The first relies upon thesecondary structure of the target to rationally design ODNswhich either avoid or take advantage of hairpin or higher-order structures. For example, stable hybrids can be formedwhen the ODN targets only one half of a hairpin (4) or thetwo single-stranded branches at the base of a hairpin stem(5). Triplex-forming ODNs can be employed if the hairpinstem contains a homopurine or homopyrimidine run (6).Tethered oligonucleotide segments which target separatesingle-stranded sites in the same molecule can also bind withgood affinity (7). The second strategy involves the design ofODNs that form unusually stable hybrids or triplexes whichcan favorably compete with existing secondary structurefeatures. For example, ODNs which interact by both Watson-Crick and Hoogsteen or reverse Hoogsteen hydrogen bond-ing can form triplexes with a single-stranded homopurine or

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

homopyrimidine run (8). More frequently, however, ODNsare chemically modified to form more stable hybrids. Ex-amples include ODNs conjugated to an intercalator (9),"peptide nucleic acids" (10), C-5-propynyl-containingODNs (11), 2'-O-methyl-ODNs (12), and N3'-P5'-phos-phoramidate ODNs (13).ODNs conjugated to a minor groove binder (MGB) repre-

sent another way to form stabilized hybrids with long single-stranded DNA or RNA. Netropsin and distamycin are twowell-characterized MGB agents (14). These crescent-shapedpeptides bind isohelically to the minor groove of A+T-richdouble-stranded DNA with association constants orders ofmagnitude greater than those exhibited by simple intercalatingagents (i.e., 107-109 M-1 versus 104-106 M-1) (15, 16). CC-1065 is another MGB which binds to and alkylates DNA in theminor groove, and it exhibits potent antitumor and antibioticactivity (17, 18). We have recently described the conjugationof N-methylpyrrole carboxamide (MPC) and N-3-carbamoyl-1,2-dihydro-3H-pyrrolo[3,2-e]indole-7-carboxylate (CDPI)peptides to the ends of oligothymidylate and oligoadenylateODNs (19, 20).t

0

ODN=CDP3 = ODN-Iinke -NN5-linkr = CONH2

3

5'-linker = -(CH2)6NH-3'-linker = -CH2CH(OH)CH2NHCO(CH2)2NH-

Structure I

These synthetic peptides were designed as nonreactive an-alogs of netropsin and CC-1065, and they increased the meltingtemperature (Tm) of 8-bp DNA hybrids by as much as 44°Cwhen covalently linked to one strand. The ODN=CDPI3conjugates also formed stabilized hybrids with complementaryRNA targets (20) and with G+C-rich DNA targets (unpub-lished observations).The impressive stability of hybrids formed by ODN=CDPI3

conjugates encouraged us to investigate the hybridizationspecificity of these agents within the context of a long single-stranded phage DNA and to determine whether the resultanthybrids could arrest primer extension by a modified form of T7DNA polymerase. The results of this study indicate that theseconjugates hybridized with greater specificity and affinity thancontrol ODNs and that the 5' conjugates effectively blocked

Abbreviations: MGB, minor groove binder; ODN, oligodeoxyribo-nucleotide; CDPI, N-3-carbamoyl-1,2-dihydro-3H-pyrrolo[3,2-e]indole-7-carboxylate.*To whom reprint requests should be addressed.tThe tripeptide of CDPI is denoted CDPI3; conjugation of an MGBto the 5', 3', or either end of an ODN is indicated, respectively, by thefollowing conventions: MGB-ODN, ODN-MGB, ODN=MGB.

3199

3200 Biochemistry: Afonina et al.

Table 1. Structures and melting temperatures of ODNs used in the present work

Oligonucleotide Sequence and Modifications Tm,°C4950 4900* *

... TGGGGTCGTCTTCTATTTTGTCTCCACTCCGCCAGTCATAATTGTGGCGGA...ODN8 5'-ATAAAACA-3' 16CDPI3-ODN8 5'-CDPI3-ATAAAACA-3' 49ODN8-CDPI3 5'-ATAAAACA-CDPI3-3' 56ODN16. 5'-ATAAAACAGAGGTGAG-3' 55CDP I3-ODN16, 5'-CDPI3 -ATAAAACAGAGGTGAG-3' 71( m4) ODN16, 5' - ATA&AACAGAGGTGAG- 3' 49CDPI3-(m4)0DN16, 5'-CDPI3-ATA,AACAGAGGTGAG- 3' 62CDPI3-(ml4)0DN16, 5'-CDPI3 -ATAAAACAGAGGTCAG-3' 67CDPI3-psODNi,6 5'-CDPI3 -ATAAAACAGAGGTGAG-3' 60Acr-ODN16, 5'-Acridine-ATAAAACAGAGGTGAG-3' 58ODN16b 5' -CCAGCAGAAGATAAAA-3' 54(ml3)ODNl6b 5'-CCAGCAGAAGATCAAA-3' 50ODN16b-CDPI3 5'-CCAGCAGAAGATAAAA-CDPI3-3' 74(ml3)ODN16b-CDPI3 5'-CCAGCAGAAGATCAAA-CDPI3-3' 67ODN32 5' -ATAAAACAGAGGTGAGGCGGTCAGTATTAACA-3' 71Bases which are mismatched relative to the complementary M13mpl9 (+) DNA strand (22) are doubly

underlined; their position in the ODN is denoted in parentheses by the letter m followed by a number.Italicized bases denote the presence of phosphorothioate linkages. The length of each ODN is indicatedby the number in the subscript, and the letters a and b denote different sequences. Acr, acridine; ps,phosphorothioate backbone; primer = 5'-GAAATGGATTATTTACATTGCAGA; ODN14-CDPI3(nonsense control) = 5'-GTGTGTCATAAATA-CDPI3. Tm values were measured for hybrids formedwith complementary or singly mismatched ODNs that were derived from the M13mpl9 (+) strandsequence.

primer extension. This powerful and targeted inhibition ofprimer extension could have diagnostic and therapeutic appli-cations.

. MATERIALS AND METHODS

Oligonucleotides. DNA synthesis was performed on a Phar-macia automatic synthesizer. For the preparation of ODNswith a 3'-hexanol phosphate modification, synthesis was car-ried out using a derivatized controlled-pore glass (CPG)support as described (21). For preparation of ODNs with a 5'acridine group, 2-cyanoethyl-[5-(9-acridinyl)-l-pentyl]-N,N-diisopropylphosphoramidite was added to the 5' end of theODN in the last automated cycle. The phosphorothioate ODNwas prepared by substituting Beaucage reagent for iodine inthe oxidation steps. CDPI3 was conjugated to 5'- or 3'-amino-tailed ODNs as described by Lukhtanov et al. (20). ODNs werepurified by HPLC on a reverse-phase column eluted by anacetonitrile gradient (usually 0-45%) in 100mM triethylamineacetate (pH 7.5) buffer. Purity of unmodified ODNs wasevaluated by electrophoresis on an 8% polyacrylamide/8 Murea gel, with subsequent visualization by silver staining. Purityof the ODN-CDPI3 conjugates was verified by analyticalHPLC as described above. All ODN preparations had >95%purity. The sequences, with their abbreviations, are shown inTable 1.

Absorption Thermal Denaturation. Hybrids formed be-tween CDPI3-tailed ODN conjugates and ODN replicates ofthe respective complementary M13mpl9 (+) strand sequenceswere melted at a rate of 0.5°C/min in 50 mM NaCl/7 mMMgCl2/24 mM Tris-HCl (pH 7.7) on a Lambda 2 (Perkin-Elmer) spectrophotometer with a PTP-6 automatic multicelltemperature programmer. The final concentration of eachODN in the melting mixtures was 2 ,M. Melting temperatures(Tms) of the hybrids were determined from the first derivativemaxima and are shown in Table 1. Prior to melting, sampleswere denatured at 100°C and then cooled to the startingtemperature over a 10-min period.

Slot-Blot Hybridization. A standard procedures was fol-lowed with minor modifications. Briefly, 10 ng of M13mpl9

tProtocolsfor NucleicAcid Blotting and Hybridization (Amersham, NewYork).

(+) strand DNA was slotted onto a Hybond N+ membrane(Amersham), dried, and prehybridized at room temperaturefor 4 h in a buffer containing 1 M NaCl, 0.05 M sodiumphosphate (pH 7.4), 1 mM EDTA, 30% (vol/vol) formamide,0.1% gelatin, 2% SDS, tRNA at 50 jxg/ml, and 0.1 ,uM(dAp)8-CDPI3 (used only with CDPI3-containing ODNs). Acomplementary or mismatched 5'-[y-32P]ATP-labeled ODNor ODN-CDPI3 conjugate was added to the prehybridizationsolution (-5 x 105 cpm) and hybridized overnight at 22,30,37,42, 47, 52, 56, 65, and 75°C. The slots were washed for 10 minin 2x SSC/0.1% SDS and for 5 min in 0.5x SSC/0.1% SDS(1 x SSC = 150 mM NaCl/15 mM sodium citrate, pH 7.0) andthen analyzed by counting or autoradiography. Since all of theODNs had similar specific activities, results could be directlycompared.Primer Extension. Primer extension reactions were per-

formed as previously described (23) with slight modifications.Briefly, 1 ,tg (9 fmol) of M13mpl9 (+) strand DNAwas mixedwith 1 pmol of [y-32P]ATP-labeled primer (_105 cpm; com-plementary to nucleotides 5123-5147 of M13mpl9) and 10 or192 fmol of blocking ODN (complementary to sites 200 basesdownstream from the primer binding site; see Table 1) in 50

5000-

4000-

3000

E0 2000

1000

0-

20 30 40 50 60 70 80

TemperatureFIG. 1. Slot-blot hybridization of ODN and ODN-CDPI3 conju-

gates to M13mpl9 (+) strand DNA. The following ODNs wereemployed as hybridization probes: ODN16b (-), (m13)ODN16b (A),ODN16b-CDPI3 (0), and (ml3)ODN16b-CDPI3 (v).

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 3201

mM NaCl/40 mM Tris-HCl (pH 7.7). The mixtures (15 ,ul)were heated 3 min at 95°C and 30 min at 45°C to anneal ODNsto the template. A Mg-dNTP mix together with a Sequenase2.0 stock (United States Biochemical) was added to give 6.2mM MgC2, 6.2 mM dithiothreitol, 79 ,uM each dNTP, Seque-nase at 750 units/ml, 31 mM NaCI, and 25 mM Tris-HCl in 24[Ll. The final concentrations of template, primer, and blockingODNs were 0.38 nM, 42 nM, and 0.42 or 8.0 nM, respectively.Primer extension was carried out for 15 min at 45°C and wasstopped by heating at 67°C for 15 min. Reaction products wereprecipitated with alcohol, electrophoresed on an 8% poly-acrylamide/8 M urea gel, and autoradiographed. When block-ing ODN was omitted from the annealing reaction and addedtogether with Sequenase, its final concentration ranged from0.04 to 40 nM. Sequencing reactions were carried out in thepresence of the dideoxynucleotides ddATP, ddCTP, ddGTP,or ddTTP without any blocking ODN.§

RESULTS

Hybridization. A slot-blot hybridization format was used toinvestigate the binding of labeled ODN and ODN-CDPI3conjugates to single-stranded M13mpl9 DNA. Nonspecificadsorption of CDPI3-containing probes onto the membranewas eliminated by prehybridizing with (dAp)8-CDPI3. Theresults in Fig. 1 demonstrate that the presence of a 3' CDPI3group on a 16-mer probe (ODNl6b-CDPI3) improved severalaspects of the hybridization reaction. Compared with anunmodified but otherwise identical 16-mer, the CDPI3-containing probe formed a hybrid 17°C more stable (Tm of50°C versus 33°C).When measured by absorption thermal denaturation, the Tm

of these same ODNs differed by 20°C (see Table 1). The higherannealing temperature of the modified probe more thandoubled the yield of perfectly matched hybrids in the blotassay. When a mismatch was introduced into either probe,stability of the respective hybrids dropped. This drop wasgreater for the CDPI3-containing probes than for the unmod-ified probes, thus providing greater discrimination. The UVmelting data in Table 1 show a 7°C drop for the CDPI3-

§Protocols for DNA Sequencing with Sequenase T7 DNA Polymerase(United States Biochemical).

A 1 2 3 4 5 6 78 9101112 M

stabilized hybrids vs. a 4°C drop for the unmodified hybrids. Inthe slot-blot assay, the modified probes exhibited good dis-crimination between 37 and 50°C, while the control probesprovided moderate discrimination between 22 and 37°C. Atthe temperatures corresponding to maximum discrimination,the ratio of matched to mismatched hybrids was 40:1 for theCDPI3-linked probes and only 3.5:1 for the control probes. Thehybridization specificity is evidence that the ODN-CDPI3conjugates did not interact with regions of secondary structurethrough binding of the MGB moiety alone.Primer Extension. Radiolabeled primer and various block-

ing ODNs were annealed with M13mpl9 (+) DNA. Theblocking ODNs were present in 1.1-fold molar excess over thesingle-stranded template and targeted to a complementaryregion 200 bases downstream from the primer binding site.DNA synthesis was initiated by the addition of T7 DNApolymerase (Sequenase). After a 15-min incubation at 47°C,the reaction was stopped and the labeled DNA product wasanalyzed on a sequencing gel. Primer extension carried out inthe absence of any blocking ODN generated a high molecularweight product which was unresolved on the gel (Fig. 2A, lanes1 and 13). Weak bands corresponding to pause sites or tospontaneous termination events were reproducibly observedin all reaction mixtures. Unmodified 16-mer and 32-merODNs, fully complementary to the target block site, failed toblock primer extension (lanes 2, 5, and 9). This was expected,since Sequenase can catalyze strand displacement synthesis(24). Also without activity were complementary 8-mer and16-mer ODNs to which a CDPI3 was 3'-linked (lanes 3 and 11).Only a fully complementary 16-mer ODN with a 5'-conjugatedCDPI3 group (CDPI3-ODNi6a) arrested primer extension byT7 DNA polymerase (lanes 6 and 14). A complementary 8-merODN with the same 5' modification generated only a traceamount of blocked product (lanes 10 and 15).Two 16-mer ODNs with a single mismatch to the target site,

each with a 5'-linked CDPI3 peptide, were much less inhibitorythan the perfectly matched ODN (lanes 7 and 8). Addition ofthe ODN16,, together with an equimolar amount of free CDPI3had no effect on primer extension (lane 16). Substitution of thebackbone of CDPI3-ODNi6a with phosphorothioate linkagesor replacement of the MGB group with a 5' acridine moietyalso led to loss of inhibitory activity (lanes 17 and 18). Notunexpectedly, the CDPI3-ODNi6a had the highest Tm of all the

B ACGT1

FIG. 2. Effect of ODN=MGB conjugateson primer extension by T7 DNA polymerase.The ODNs described in Table 1 werescreened after a preannealing step for theirability to arrest primer extension at a finalconcentration of 0.42 nM. (A) Lane 1, noblocking ODN; lane 2, ODNI6b; lane 3,ODNl6b-CDPI3; lane 4, (m13)ODN16b-CDPI3; lane 5, ODNi6a; lane 6, CDPI3-ODN6a,; lane 7, CDPI3-(m4)ODNi6,,; lane 8,CDPI3-(m14)ODN,6a; lane 9, ODN32; lane10, CDPI3-ODNs; lane 11, ODNs-CDPI3;lane 12, nonsense ODN14-CDPI3; lane 13, noblocking ODN; lane 14, CDPI3-ODNi6a; lane15, CDPI3-ODN8; lane 16, ODNi6a plus freeCDPI3; lane 17, CDPI3-psODNi6a; and lane18, Acr-ODN16a. Lanes M, Hae III digest ofphage 4X174 DNA. (B) Lane 1, CDPI3-ODNl6a. T, G, C, and A, primer extensioncarried out in the presence of ddTTP,ddGTP, ddCTP, or ddATP.

run-on CAAAATAGAAG

block -

Biochemistry: Afonina et al.

:.......

b."",-:.0

i".:.:;.4w.,,..

3202 Biochemistry: Afonina et al.

modified ODNs tested (see Table 1). We conclude that DNAsynthesis was arrested only when a blocking downstreamhybrid of sufficient stability was formed by a CDPI3-ODNconjugate.The sites at which CDPI3-ODN16a, blocked primer extension

were mapped to base-pair resolution by comparison with addNTP sequencing ladder (Fig. 2B). This analysis showed thatupon reaching the blocking ODN, Sequenase was able todisplace the first three bases of that ODN and extend theprimer accordingly. At lower frequency, the polymeraseaborted synthesis in a stepwise fashion up to three bases earlieror five bases later than the predominant stop site. In shorterexposures, where the unarrested product band was not over-exposed, the presence of CDPI3-ODN,,6 reduced the yield ofthis band by approximately 50%. Since only one round ofprimer extension was expected, the arrested products probablyrepresent strong pauses or absolute blocks to DNA synthesis.The run-on products probably result from primer extension offtemplate DNAs with no blocking ODN hybridized down-stream.

Recognizing that blocking ODNs with an unmodified 3' endcould potentially function as primers, we investigated whetherthe inhibitory activities of several CDPI3-ODNs were reducedby capping each ODN with a 3'-hexanol phosphate ester.When these ODNs were present at a 21-fold molar excess overthe phage DNA, the presence of a capping group did not ablateinhibition of primer extension (Fig. 3). Capped and uncappedversions of the CDPI3-ODN16a were equally effective inhibi-tors (lanes 2 and 3), despite the ability of the uncapped ODNto function as a primer (data not shown). Thus the CDPI3-ODN conjugate itself, and not a longer complementary DNAmade by extension of the conjugate, is blocking primer exten-sion.At the higher concentration of blocking ODNs used in Fig.

3 (i.e., 8 nM), less stable hybrids were able to inhibit primerextension, including those formed by CDPI3-(m4)ODNi6a,

1 2 34 5 6 7 8 910M

run-on

block

FIG. 3. Effect of 3'-end capping on inhibition of primer extensionby MGB-ODN conjugates. Several CDPI3-ODN conjugates with andwithout a 3'-hexanol (Hx) tail were tested after a preannealing stepfor their ability to arrest primer extension at a final concentration of8.0 nM. Lane 1, no blocking ODN; lane 2, CDPI3-ODNi6a-Hex; lane3, CDPI3-ODNt6,,,lane 4, ODNIa; lane 5, CDP3t-(m4)ODNi6a-Hex;lane 6, CDPI3-(m4)ODN6,,; lane 7, (m4)ODNi6a; lane 8, CDPI3-(m14)ODNi6a,-Hex; lane 9, CDPI3-(ml4)ODNi6a; and lane 10, CDPI3-ODNx. Lane M, Hae III digest of «X174 DNA.

CDPI3-(m14)ODNi6a, and CDPI3-ODN8 (lanes 6, 9, and 10).Since a 3'-capped version of the CDPI3-ODN8 was not tested,it is unknown whether the blocking activity of this ODN iscontingent upon it's also acting as a primer. Of the twomismatched 16-mers, the CDPI3-(ml4)ODNi6a was the betterinhibitor. This observation correlates with the differentialstability of the hybrids (67°C versus 62°C; see Table 1) andindicates that a mismatch is more perturbing when it interfereswith the binding of the tethered CDPI3 moiety to the hybrid.Capped versions of these same ODNs were less inhibitory(lanes 5 and 8), suggesting to us that primer extension may havestabilized the mismatched hybrids.

Inhibition of primer extension did not require a preanneal-ing step. When blocking ODNs were added at the same timeas Sequenase, DNA synthesis was inhibited in a concentration-dependent manner (Fig. 4). CDPI3-ODN16a inhibited se-quence specifically at concentrations as low as 0.4 nM (molarratio of blocking ODN to template of 1.1), arresting primerextension at the expected site. When the concentration of thisODN was increased to 40 nM specificity was retained, althoughweak blockage at an unidentified downstream site becamedetectable. By contrast, ODNl6a and CDPI3-(m4)ODN16ainhibited DNA synthesis non-sequence-specifically at concen-trations equal to or greater than 40 nM or 4 nM, respectively.At the highest concentration tested, conjugation of a CDPI3moiety to ODNi6a remarkably improved its specificity ofinhibition. This effect was not observed with the mismatchedCDPI3-(m4)ODNi6a, which generated a product inhibitionpattern very similar to that formed by ODN16l,.

DISCUSSIONOur earlier study of the contribution of the conjugated MGBsto ODN hybrid stability by absorption thermal denaturation(20) demonstrated that ODN=CDPI3 conjugates can formremarkably stable A+T-rich hybrids. In the present study, weadditionally address the specificity of hybrid formation asdetermined by slot-blot hybridization and by inhibition of

4x10"M 4 x 10M 4 x 10 'M 4 x 10 MM1 2 3 4 2 34 2 34 234

run-on

block

FIG. 4. Dose response for inhibition of primer extension by MGB-ODN conjugates in the absence of a preannealing step. ODN6,a (lanes2), CDPI3-ODNi6a (lanes 3), and CDPI3-(m4)ODNi6a (lanes 4) wereadded to primer extension reactions with no preannealing step con-current with Sequenase addition. The final concentrations of blockingODNs are shown above the lanes. Lane 1, no blocking ODN. M, HaeIII digest of «X174 DNA.

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 3203

primer extension. We expected specificity to be primarilydetermined by the ODN portion of the conjugate. We hypoth-esized that hybrids formed by an ODN=CDPI3 conjugatewould not ideally accommodate the CDPI3 moiety if a mis-match existed in that portion of the duplex occupied by theMGB. This effect would be expected to improve the specificityof hybridization by differentially destabilizing a large subset ofmismatched hybrids. Another cause of lack of specificity couldbe binding of the MGB to double-stranded regions of thetarget DNA. Although MGBs have high affinities for A+T-rich double-stranded DNA, we reasoned that conjugation to anegatively charged ODN would reduce these affinities and thatno nonspecific binding of the conjugate to secondary struc-tures in the M13 single-stranded DNA would be observed.The results presented here support these suppositions. At

optimal concentrations of ODN=CDPI3, no nonspecific bind-ing was observed by either technique. The ODN=MGBconjugates retain or improve upon the specificity of bindingexhibited by unmodified ODNs, as also observed with ODNsconjugated to intercalating agents (9, 25). In each case,sequence-specific hybridization creates a high-affinity bindingsite for the attached ligand, which in turn stabilizes the hybrid.When tethered to an ODN, an MGB or intercalator does notbind to double-stranded DNA unless the ODN is involved aswell (for example, as in triplex formation). In both of the assaysemployed here the discrimination between matched and mis-matched complexes was improved by linking the CDPI3 groupto the ODN. When the blot-hybridization assay was carried outunder optimal conditions, this discrimination was 10 timesbetter. In a similar fashion, when a blocking ODN was addedat high concentration without preannealing, the CDPI3-containing version arrested primer extension at a unique site,whereas the unmodified version arrested DNA synthesis to alesser extent at multiple sites. These results make sense ifbinding of the tethered tripeptide to the minor groove of theODN-M13 duplex is impaired by the presence of a mismatch.The Tm values listed in Table 1 for the various matched andmismatched hybrids are consistent with this interpretation.As expected, the ODN=CDPI3 conjugates studied here

formed very stable hybrids. The covalently appended tripep-tide increased the Tm of the 8-bp and 16-bp hybrids by 33-40°Cand 16-20°C, respectively. Because the 16-mer conjugatescould be hybridized at higher stringency than the correspond-ing unmodified ODNs, they were more effective in overcomingthe secondary structure present in a long single-strandedtarget and so hybridized more efficiently. All of theODN=CDPI3 conjugates tested here had an A+T-rich se-quence immediately adjacent to the MGB. Upon hybridizationthese ODNs afforded an ideal binding site for the CDPI3moiety. ODN=CDPI3 conjugates with a flanking G+C-richsequence form less stable hybrids since the CDPI3 groupinteracts less strongly with the minor groove of a G+C-richduplex (unpublished observations).

Inhibition of primer extension served as a high-resolutionmarker for the binding of the CDPI3-ODNi6a conjugate toM13 phage DNA. Interestingly, the 5' but not the 3' 16-merconjugate blocked DNA synthesis. Location of the MGBwithin the obstructing duplex dramatically affected the re-sponse of the polymerase. Sequenase was unable to displacethe ODN when the blocking end of the duplex contained atethered MGB. By contrast, Sequenase readily displaced thesame ODN when the MGB resided at the nonblocking end ofthe duplex. In this case the enzyme probably peeled off the 5'end of the ODN, leaving a destabilized distal hybrid which,despite the presence of an MGB, was unable to block forwardmovement of the polymerase. These observations underscorethe importance of both the ODN and CDPI3 domains inarresting DNA synthesis. Hybridization of the ODN creates ahigh-affinity binding site for the conjugated MGB, which inturn acts as a block to DNA synthesis. In this context

ODN=CDPI3 conjugates can be thought of as sequence-specific nonreactive analogs of CC-1065 which bind to single-stranded instead of double-stranded DNA. The general ab-sence of nonspecific blockage sites with all of theODN=CDPI3 conjugates tested supports the specificity withwhich these ODNs hybridize. The origin of the weak upstreamband obtained in the presence of CDPI3-ODNi6a (see Fig. 2B,lane 1) cannot be readily explained, since this G+ C-rich regionof the template bears no complementarity to the blockingODN. The minimum ODN length required for efficient inhi-bition by the conjugate has not been determined. Interestingly,however, a phosphorothioate version of the active 16-merfailed to block primer extension, possibly because the Tm waslowered from 71 to 60°C (see Table 1).The location of the primary termination site for DNA

synthesis can be explained by a simple structural model. NMRanalysis has shown that bound CC-1065 spans 5-6 bp in theminor groove of DNA (26). By analogy, a conjugated CDPI3group probably occupies a similar length of minor groove, withthe region of interaction displaced 1-2 bp away from the siteof conjugation. Fine mapping of the primary termination siteinduced by CDPI3-ODNi6a showed that synthesis was arrestedopposite the third base of the blocking ODN. This coincideswith the predicted location of the C terminus of the tetheredCDPI3 and suggests that the MGB directly inhibited movementof the enzyme's catalytic site by preventing strand displace-ment. Following alkylation of double-stranded DNA, CC-1065is believed to block DNA synthesis (27). The mechanism ofblockage may be similar to that shown here for CDPI3-ODNI6a.The hybridization properties of ODN=CDPI3 conjugates

warrant their further evaluation as diagnostic probes and aspotential PCR clamping agents (28). The ability of the CDPI3-ODN conjugates to block primer extension also suggests thatthese agents might have therapeutic applications. Others (29-32) have shown that Sequenase-catalyzed DNA synthesis fromdouble-stranded DNA is inhibited by intra- and intermoleculartriplexes. The use of triplexing ODNs to inhibit replication inthe cell is called into question by the recent observation thathelicase readily disrupts triple-stranded DNA (33). It is notknown whether a tethered MGB can inhibit helicase action. IfODN=CDPI3 conjugates form helicase-resistant hybrids, theymight be used to inhibit replication of single-stranded DNAviruses or to block transcription initiation of specific genes bybinding to the open promoter complex.We thank A. David Adams for oligonucleotide synthesis.

1. Gamper, H. B., Cimino, G. D. & Hearst, J. E. (1987) J. Mol. Biol.197, 349-362.

2. Wyatt, J. R., Puglisi, J. D. & Tinoco, I., Jr. (1989) BioEssays 11,100-106.

3. Latham, J. A. & Cech, T. R. (1989) Science 245, 276-282.4. Ecker, D. J., Vickers, T. A., Bruice, T. W., Freier, S. M., Jenison,

R. D., Manoharan, M. & Zounes, M. (1992) Science 257, 958-961.

5. Francois, J.-C., Thuong, N. T. & Helene, C. (1994) Nucleic AcidsRes. 22, 3943-3950.

6. Brossalina, E. & Toulme, J.-J. (1993) J. Am. Chem. Soc. 115,796-797.

7. Richardson, P. L. & Schepartz, A. (1991)J. Am. Chem. Soc. 113,5109-5111.

8. Giovannangeli, C., Montenay-Garestier, T., Rougee, M., Chas-signol, M., Thuong, N. T. & Helene, C. (1991) J. Am. Chem. Soc.113, 7775-7777.

9. Asseline, U., Delarue, M., Lancelot, G., Toulme, F., Thuong,N. T., Montenay-Garestier, T. & Helene, C. (1984) Proc. Natl.Acad. Sci. USA 81, 3297-3301.

10. Nielsen, P. E., Egholm, M. & Buchardt, 0. (1994) BioconjugateChem. 5, 3-7.

11. Wagner, R. W., Matteucci, M. D., Lewis, J. G., Gutierrez, A. J.,Moulds, C. & Froehler, B. C. (1993) Science 260, 1510-1513.

Biochemistry: Afonina et al.

3204 Biochemistry: Afonina et al.

12. Monia, B. P., Johnston, J. F., Ecker, D. J., Zounes, M. A., Lima,W. F. & Freier, S. M. (1992) J. Biol. Chem. 267, 19954-19962.

13. Gryaznov, S. & Chen, J.-K. (1994) J. Am. Chem. Soc. 116,3143-3144.

14. Zimmer, C. & Wahnert, U. (1986) Prog. Biophys. Mol. Biol. 47,31-112.

15. Marky, L. A. & Breslauer, K. J. (1987) Proc. Natl. Acad. Sci. USA84, 4359-4363.

16. Hansen, J. B., Koch, T., Buchardt, O., Nielsen, P. E., Wirth, M.& Norden, B. (1983) Biochemistry 22, 4878-4886.

17. Reynolds, V. L., McGovren, J. P. & Hurley, L. H. (1986) J.Antibiot. 39, 319-334.

18. Boger, D. L. & Johnson, D. S. (1995) Proc. Natl. Acad. Sci. USA92, 3642-3649.

19. Sinyakov, A. N., Lokhov, S. G., Kutyavin, I. V., Gamper, H. B. &Meyer, R. B. (1995) J. Am. Chem. Soc. 117, 4995-4996.

20. Lukhtanov, E. A., Kutyavin, I. V., Gamper, H. B. & Meyer, R. B.(1995) Bioconjugate Chem. 6, 418-426.

21. Gamper, H. B., Reed, M. W., Cox, T., Virosco, J. S., Adams,A. D., Gall, A. A., Scholler, J. K. & Meyer, R. B. (1993) NucleicAcids Res. 21, 145-150.

22. Yanisch-Perron, C., Vieira, J. & Messing, J. (1985) Gene 33,103-119.

Proc. Natl. Acad. Sci. USA 93 (1996)

23. Pfeifer, G. P. & Riggs, A. D. (1993) Methods Mol. Biol. 15,153-168.

24. Tabor, S. & Richardson, C. C. (1989) J. Biol. Chem. 264, 6447-6458.

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

26. Scahill, T. A., Jensen, R. M., Swenson, D. H., Hatzenbuhler,N. T., Petzold, G., Wierenga, W. & Brahme, N. D. (1990) Bio-chemistry 29, 2852-2860.

27. Li, L. H., Swenson, D. H., Schpok, S. L. F., Kuentzel, S. L.,Dayton, B. D. & Krueger, W. C. (1982) CancerRes. 42,999-1004.

28. Orum, H., Nielsen, P. E., Egholm, M., Berg, R. H., Buchardt, O.& Stanley, C. (1993) Nucleic Acids Res. 21, 5332-5336.

29. Samadashwily, G. M., Dayn, A. & Mirkin, S. M. (1993) EMBO J.12, 4975-4983.

30. Giovannangeli, C., Thuong, N. T. & Helene, C. (1993) Proc. Natl.Acad. Sci. USA 90, 10013-10017.

31. Samadashwily, G. M. & Mirkin, S. M. (1994) Gene 149, 127-136.32. Hacia, J. G., Dervan, P. B. & Wold, B. J. (1994) Biochemistry 33,

6192-6200.33. Maine, I. P. & Kodadek, T. (1994) Biochem. Biophys. Res.

Commun. 204, 1119-1124.