Proc. Nati. Acad. Sci. USAVol. 89, pp. 3761-3764, May 1992Biochemistry

Oligonucleotide-mediated triple helix formation using anN3-protonated deoxycytidine analog exhibiting pH-independentbinding within the physiological rangeSTEVEN H. KRAWCZYK, JOHN F. MILLIGAN, SHALINI WADWANI, COURTNEY MOULDS, BRIAN C. FROEHLER,AND MARK D. MATrEUCCIGilead Sciences, Inc., 346 Lakeside Drive, Foster City, CA 94404

Communicated by Peter B. Dervan, January 15, 1992 (received for review November 5, 1991)

ABSTRACT Triple helix formation with pyrimidine de-oxyoligonucleotides for the sequence-specific recognition ofDNA duplex targets suffers from a decrease in affinity as the pHof the medium increases to that of physiological fluids. Asolution to this problem has been identified and entails thesubstitution of N'6-methyl-8-oxo-2'-deoxyadenosine (M) for the5-methyl-deoxycytosine base residues. The triple helix formingability of an oligonudeotide consisting of thymidine and Mresidues is pH independent in the physiological range. Fur-thermore, M has been found to be superior to the previouslyused 5-methyldeoxycytidine and deoxyguanosine in conferringincreased affinity for duplex DNA under physiological saltconditions.

In recent years the development of methodologies for thesequence-specific recognition and chemical modification ofDNA have created the opportunity to modulate gene expres-sion in vivo (1-3). The most promising, currently availabletechnology for the sequence-specific recognition of DNA istriple helix formation via Hoogsteen hydrogen bond interac-tions within T-A-T and C+-G-C triplets. Although the forma-tion of triple helical RNA structures in homopolymers hasbeen known for over three decades (4, 5), this phenomenonhas only recently been exploited for the recognition of mixedsequences by deoxyoligonucleotides (6). Such triple-stranded interactions have been used to inhibit the binding oftranscription factors and restriction endonucleases to theirDNA targets (7).A severe limitation of this recognition motif is the strong

pH dependence of triple strand formation (8). This pHdependence is due to the requirement that the N3 of cytidine,of the third strand, be protonated in order to form the C+*G-Ctriplet (Fig. 1A) (9). Thus, the binding of a mixed polypurineduplex utilizing deoxycytidine- and thymidine-containing oli-gonucleotides requires conditions considerably more acidicthan the intracellular pH range of 7.1-7.6 (6).Three approaches have been attempted to circumvent this

shortcoming. The first involved the substitution of 5-meth-yldeoxycytidine for deoxycytidine, a strategy based upon thefact that homopolymers of 5-methyldeoxycytidine and de-oxyguanosine can form triple-stranded structures at neutralpH under certain, nonphysiological, salt conditions (10). Thisobservation has been extended to the recognition of mixedsequences (11). The second approach taken was the substi-tution of deoxyguanosine for deoxycytidine (12-14), to yieldan alternative binding motif in which the third strand binds inan antiparallel orientation with respect to the purine strand ofthe duplex (Fig. 1B). However, this latter modificationpertains only to DNA targets containing a relatively largeproportion of G-C base pairs. Recently, a third approach has

involved the use of a pseudoisocytidine derivative that con-tains the appropriate pattern of hydrogen bond donors in aneutral species (15).An evaluation of the first two methodologies has led us to

the conclusion that the 5-methyldeoxycytidine and deoxy-guanosine substitution schemes only marginally enhancebinding under physiological salt and pH conditions. There-fore, we searched for a neutral heterocyclic mimetic ofprotonated cytosine that would enable the synthesis of oligo-mers having enhanced specific affinity for duplex DNA underconditions that mimic the intracellular environment.The conversion of a deoxyguanosine residue into an 8-

oxodeoxyguanosine residue is accompanied by a GC to T-Amutation upon replication in Escherichia coli (16). The ex-planation offered was that the introduction of the keto groupin the 8-position forces the deoxyguanosine residue to adopta syn conformation such that the N7 proton and the 06 mimicthe base-pairing face of thymidine. Since 8-substituted aden-osine analogs are known to exist predominantly in the synconformation (17), we reasoned that an 8-oxodeoxyadeno-sine derivative could behave as a mimetic of the protonateddeoxycytidine species implicated in triplex formation. Thehydrogens present on the 6-amino group and the N7 of8-oxodeoxyadenosine would take the place of the two hy-drogens located on the 4-amino group and N3 of the proto-nated deoxycytidine residue as shown in Fig. 1C.

MATERIALS AND METHODSSynthesis ofM Synthon. The synthesis of compound 3 was

based upon the methodology reported by Ikehara andKaneko (18) as shown in Fig. 2.A solution of 20 g of N6-methyl-2'-deoxyadenosine (19) in

300 ml of water containing 15 g of sodium acetate was treatedwith 500 ml of saturated bromine water. After 6 hr of stirringat room temperature, the mixture was stored at 4°C for 10 hr.The pH was then adjusted to 7 with 2 M sodium hydroxideand sodium bisulfite was added to the stirred suspension untilit became light yellow in color. The mixture was heated untila clear solution was obtained and then allowed to cool. Afterthe mixture had been kept at 4°C for 18 hr, the precipitate wascollected by filtration to afford 13.7 g of solid (51%).Nineteen grams of material obtained by this method was

suspended in 150 ml of acetic anhydride containing 19 g ofsodium acetate. The mixture was kept at 130°C for 48 hr andthen poured onto 600 cm3 of crushed ice. The mixture wastreated with 600 ml of ethyl acetate and allowed to stir for 6hr. The biphasic mixture was filtered through a fritted glassfilter, and the layers were separated. The organic layer waswashed with water, saturated sodium bicarbonate solution,and brine. The organic solution was filtered through a 10-cm-wide by a 4-cm-deep silica gel bed, and the bed was

Abbreviation: M, N6-methyl-8-oxo-2'-deoxyadenosine.

3761

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.

3762 Biochemistry: Krawczyk et al.

MeH H

dR-N NN N,,Ct

H., H% H t

HN NNH

dR H

A

dR

dR

B

o RH

*N N' 0

N No~<N HI

dR H

CFIG. 1. Putative binding schemes for C+G-C triplet (A), G-G-C triplet (B), and M-G-C triplet (C), where C+ represents N3-protonated

5-methyldeoxycytidine, G represents deoxyguanosine, and M represents N6-methyl-8-oxo-2'-deoxyadenosine.

washed with 500 ml of ethyl acetate. The combined filtrateswere evaporated and coevaporated three times with 100-mlportions of methanol. The residue was dissolved in 100 ml ofa 1 M sodium methoxide solution in methanol, and themixture was kept at room temperature for 96 hr. After thistime, the pH was adjusted to 7 with acetic acid and thesolution was evaporated. The residue was chromatographedon a silica gel column using methylene chloride/methanol,4:1 (vol/vol), containing 2.5% triethylamine. The product-containing fractions were evaporated, and the residue wascrystallized from ethanol to afford 3.1 g of product (18%). mp182-1840C; 1H NMR(dimethyl sulfoxide-d6): 10.27 (bs, 1H,H-7), 8.09 (s, 1H, H-2), 6.49 (q, 1H, H-N6), 6.14 (dd, 1H,J1',2'a = J1',2'b = 6.6 Hz), 5.17 (2m, 2H,3'-OH and 5'-OH),4.38 (m, 1H, H-3'), 3.80 (m, 1H, H-4'), 3.60 and 3.43 (2m, 2H,H-5'ab), 2.97 and 2.0 (2m, 2H, H-2'ab); high-resolutionMS(FAB) 282.120000 (M + 1); Calcd for C11H16N504 =282.120229.Conversion of the free nucleoside into the 5'-dimethoxy-

trityl-3'-H-phosphonate, 4, was then accomplished usingestablished methodologies (20).

Me ,H

0

L~sa \>

HO

OH

Me% IH

a N r

0HO

OH

I 2

Oligonucleotides. Oligonucleotides were synthesized usingH-phosphonate methodology (20) and purified by denaturingacrylamide gel electrophoresis followed by Sephadex G-25gel filtration. A key feature to note is that the N6 exocyclicamine of M does not require protection for oligonucleotidesynthesis as assessed by nucleoside composition analysis ofthe resulting deoxyoligonucleotides (21). The nucleosidecomposition of each of the oligomers was confirmed bydigestion of the oligomers to nucleosides and analysis byreverse-phase HPLC. The oligomers were converted to nu-cleosides by incubating 1 OD of oligomer with 3 units of P1nuclease and 2 units of calf intestinal alkaline phosphatase(Boehringer Mannheim) in 30 mM NaOAc, pH 5.2/1 mMZnSO4 overnight at 37°C. The digested material was theninjected directly onto a C18 column (Amicon, 100-A pores),and the nucleosides were separated by an acetonitrile gradi-ent buffered with 50 mM potassium phosphate (pH 4.2).DNA Duplex Target. The duplex DNA target used was the

259-base-pair Pst I-Xma I fragment of the HER2 promoterlabeled at the 3' end of the purine-rich strand by filling in theXma I site with [a-32P]dCTP and Klenow fragment (22).

MeN'OHH

b, c N' N

K N0

HO

OH

3

d, e, f

MeN, H

H

N N

0:0- Et3NH+

4

FIG. 2. Synthesis ofM phosphonate monomer. Reaction a, Br2, NaOAc, H20; reaction b, NaOAc, Ac2O, 130°C; reaction c, NaOMe, MeOH;reaction d, 4,4'-dimethoxytrityl chloride, pyridine; reaction e, 2-chloro-4H-1,3,2-benzodioxaphosphorin-4-one, pyridine, CH2C12; reaction f,Et3NH' HCO3- (aqueous), pH 7.6. DMT, 4,4'-dimethoxytrityl.

Proc. Natl. Acad. Sci. USA 89 (1992)

Proc. Natl. Acad. Sci. USA 89 (1992) 3763

Hybridization Conditions. Triple-strand hybridizationswere performed at 370C for 1 hr with 1 nM radiolabeled DNAtarget (-50,000 cpm) and increasing concentrations (0.1, 1,10 ,uM) of the designated oligomer in the presence of 1 mMspermine, 140 mM KCl, 10 mM NaCi, and 1 mM MgCl2 in 20mM Mops at pH 7.2 or pH 7.6. For the hybridizations at pH6.4, 20 mM Pipes was substituted for the Mops buffer. Theoligonucleotide concentrations were determined by measur-ing A260 of the solutions. For quantifications of oligomerscontaining M, an extinction coefficient of 12900 M-1 per Mresidue was used.

Footprint. Enzymatic footprinting was carried out by treat-ment of the hybridization mixtures with 1 unit of DNase I at370C for 1 min. To ensure reproducible digestion patterns, 1,.g of salmon sperm DNA was added to each reaction mixturejust prior to the addition of the DNase I. The reactions werethen terminated by the addition of 20 mM EDTA and placingthe mixture on ice. The samples were precipitated with 1 ,ugof carrier tRNA, 0.3 M sodium acetate, and 2 volumes ofethanol, washed with 70% ethanol, and vacuum evaporated.The samples were resuspended in 80% formamide containingdyes and heated to 950C for 5 min prior to loading. Digestionproducts were analyzed by 5% denaturing polyacrylamide gelelectrophoresis and visualized by autoradiography.

RESULTS AND DISCUSSIONTo test the hypothesis that an 8-oxo-2'-deoxyadenosine couldbehave as a protonated deoxycytidine analog the nucleosideanalog 3 (M) was substituted for 5-methyldeoxycytidineresidues in triple helix-forming oligonucleotides.The DNA duplex target chosen was the promoter region of

HER2, a protooncogene implicated in human breast and ova-rian cancers (22). The 15-nucleotide-long sequences shown inFig. 3 were synthesized and their binding was assessed byDNAfootprinting using DNase I digestion (23). The presence of thebound third strand inhibits digestion by DNase I.

Buffer conditions were selected to approximate knownintracellular pH and mono- and multivalent cation concen-trations (24, 25). We performed the footprint experiment atpH 7.2 and 37°C in the presence of 1 mM spermine, 140 mMKCl, 10mM NaCl, and 1 mM MgCl2 in 20mM Mops with theresults shown in Fig. 4. Under these conditions, oligomer A,with M substitution, afforded partial protection at 1 ,uM andfull protection at 10 ,uM. Neither the control oligomer B northe antiparallel deoxyguanosine-substituted oligomer C gavea partial footprint at 10 ,uM. It should be noted that Beal andDervan (13) reported that an iron-EDTA-derivatized G- andT-containing 19-mer could efficiently cleave a target se-quence at a concentration of 0.5 ,uM, whereas we do notobserve any protection with oligomer C. This difference inaffinities could be due to several factors, including thedifference in length, effect of context and the ratio of G-C toA*T base pairs comprising the binding site, and differentmono- and divalent cation concentrations used in the assay.We also tested the effect of the incorporation ofM on the

specificity of binding of oligomers to their target sequences.

U.:I

....?

pH 7.2

A B C Dm

0 0.1 I 10 0.1 I 10 0.1 I 10 0.1 1 10 PM

5AGGAGAAGGAGGAGG3,

do,.-.

FIG. 4. DNase I footprint ofoligomers A, B, C, and D at indicatedconcentrations (ILM) at pH 7.2. U, untreated Pst I-Xma I fragment;G, dimethyl sulfate ladder to indicate the positions of the guanosineresidues (26); 0, no oligomer, DNase digestion control. The positionand sequence of the targeted region are shown on the left.

Sequence D, with a single M for thymidine mismatch, showsno footprint at 10 ,M. We conclude that the M-G-C interac-tion is highly specific.The above experiment was repeated at pH 6.4 using oligo-

mers A and B. Under these conditions, oligomers A and Bafford a footprint with full protection at 10 ,uM (Fig. 5). Thefootprint experiment was also repeated at pH 7.6, where theM-substituted oligomer A again afforded a full footprint at 10,M (Fig. 5), thus demonstrating the pH independence ofbinding of M-containing oligomers within the intracellular pHrange.

CONCLUSIONThese studies led to the conclusion that M-substituted oligo-mers are superior to 5-methyldeoxycytidine-substitutedoligomers in this sequence in regard to duplex DNA binding

Target Sequence:

5CTCCCATCACA G G A G A A G G A G G A G GTTcA4cALaCV~3-AGGGTAGTGT C C T C T T C C T C C T C CACCTCCTCCTCXGA-6

Oligomers:A) 5'-T M M T M T T MB) 5'-TmCmC TmC T T"CC) 3'-T G G T G T T GD) 5'-T M M T M T M M

M T M M T M M-3'mC TmCTmC TMCmC-3'G T G G T G G-5'M T M M T M M-3'

FIG. 3. Sequences of HER2 target DNA and the targeted oligomers. For the oligomers in bold type, MC signifies 5-methyldeoxycytidine,G signifies deoxyguanosine, and T signifies thymidine. Mismatch position in oligomer D is underlined.

Biochemistry: Krawczyk et al.

3764 Biochemistry: Krawczyk et al.

pH 6.4

"_llL

..

wa_ ...W:

No ...

_.- *-_..*.

__I* M

DH 7.6

A B A

U G 0 0.1 1 l0 0.1 1 l0 0. 1 l ~lOM

8s-- w$S-'sX AXWmAd

M:-AT=.EH40Ask

-

-Som..

_~~~a Mm*

lo VWam.

*-.--._ _ N.Art

4m.

FIG. 5. DNase I footprint of oligomers A and B at the indicatedconcentrations (aM) at pH 6.4 and of oligomer A at pH 7.6. U,untreated Pst I-Xma I fragment; G, dimethyl sulfate ladder toindicate the positions of the guanosine residues (26); 0, no oligomer,DNase digestion control. The bracket on the left indicates theposition of the targeted region.

affinity under intracellular salt and pH conditions. The com-plete substitution ofthe 5-methyldeoxycytidine residues withM confers pH independence of binding for an oligonucleo-tide, within the physiological pH range, without compromis-ing its specificity. Furthermore, M is easily synthesized anddoes not require exocyclic amino group protection duringoligonucleotide synthesis. We have also demonstrated that,for the HER2 target sequence, the parallel motif incorporat-ing M has better binding properties compared to the antipar-allel motif under physiological pH and salt conditions.We have employed M in the construction of oligomers for

the inhibition of transcription elongation in an in vitro'system(27). Thus, M-containing deoxyoligonucleotides have signif-

icant promise in the development of therapeutic agents basedupon triple helix formation in vivo.

We acknowledge the efforts of Sarah McCurdy and Terry Terhorstin synthesis of the oligomers, Dave Sweedler for the large-scalepreparation of M, John Helms for base composition analysis, andSam Butcher for technological assistance. We thank Dr. JohnLatham for helpful discussions and Debbie Surdel and Susan Hub-bard for preparation of the figures and manuscript.

1. Uhlmann, E. & Peyman, A. (1990) Chem. Rev. 4, 543-584.2. Goodchild, J. (1990) Bioconjugate Chem. 3, 165-187.3. van der Krol, A. R., Mol, J. N. M. & Stuitje, A. R. (1988)

BioTechniques 10, 958-976.4. Felsenfeld, G., Davies, D. R. & Rich, A. (1957) J. Am. Chem.

Soc. 79, 2023-2024.5. Riley, M., Maling, B. & Chamberlain, M. J. (1966) J. Mol. Biol.

20, 359-389.6. Moser, H. E. & Dervan, P. B. (1987) Science 238, 645-650.7. Maher, L. J., III, Wold, B. & Dervan, P. B. (1989) Science 245,

725-730.8. Plum, G. E., Park, Y.-W., Singleton, S. F., Dervan, P. B. &

Breslauer, K. J. (1990) Proc. Nati. Acad. Sci. USA 87, 9436-9440.

9. Live, H. D., Radhakrishnan, I., Misra, V. & Patel, J. D. (1991)J. Am. Chem. Soc. 113, 4687-4688.

10. Lee, J. S., Woodsworth, M. L., Latimer, L. J. P. & Morgan,A. R. (1984) Nucleic Acids Res. 16, 6603-6614.

11. Posvic, T. J. & Dervan, P. B. (1989) J. Am. Chem. Soc. 111,3059-3061.

12. Cooney, M., Czernuszewicz, G., Postel, E. H., Flint, S. J. &Hogan, M. E. (1988) Science 241, 456-459.

13. Beal, P. A. & Dervan, P. B. (1991) Science 251, 1360-1363.14. Durland, R. H., Kessler, D. J., Gunnell, S., Duvic, M., Pettitt,

B. M. & Hogan, M. E. (1991) Biochemistry 30, 9246-9255.15. Ono, A., Ts'o, P. 0. P. & Kan, L.-S. (1991) J. Am. Chem. Soc.

113, 4032-4033.16. Wood, M. L., Dizdaroglu, M., Gajewski, E. & Essigman, J. M.

(1990) Biochemistry 29, 7024-7032.17. Giessner-Prettre, C. & Pullman, B. J. (1977) J. Theor. Biol. 65,

189-201.18. Ikehara, M. & Kaneko, M. (1970) Chem. Pharm. Bull. 18,

2441-2446.19. Jones, J. W. & Robins, R. K. (1963) J. Am. Chem. Soc. 85,

193-201.20. Froehler, B., Ng, P. & Matteucci, M. D. (1986) Nucleic Acids

Res. 14, 5399-5407.21. Eadie, J. S., McBride, L. J., Efcavitch, J. W., Hoff, L. B. &

Cathcart, R. (1987) Anal. Biochem. 165, 442-447.22. Tal, M., King, C. R., Kraus, M. H., Ullrich, A., Schlessinger,

J. & Givol, D. (1987) Mol. Cell. Biol. 7, 2597-2601.23. Brenowitz, M., Senear, D. F., Shea, M. A. & Ackers, G. K.

(1986) Proc. NatI. Acad. Sci. USA 83, 8462-8466.24. Hedley, D. W. & Jorgensen, H. B. (1989) Exp. Cell Res. 180,

106-116.25. Albert, B. (1989) Molecular Biology of the Cell (Garland, New

York), p. 301.26. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65,

499-560.27. Young, S. L., Krawczyk, S. H., Matteucci, M. D. & Toole,

J. J. (1991) Proc. NatI. Acad. Sci. USA 88, 10023-10026.

Proc. Natl. Acad. Sci. USA 89 (1992)

::