Nucleic Acids Research, 1993, Vol. 21, No. 14 3197-3203

Antisense oligodeoxynucleotides: synthesis, biophysicaland biological evaluation of oligodeoxynucleotidescontaining modified pyrimidines

Yogesh S.Sanghvi*, Glenn D.Hoke+, Susan M.Freier, Maryann C.Zounes, Carolyn Gonzalez,Lendell Cummins, Henri Sasmor and P.Dan CookISIS Pharmaceuticals, 2280 Faraday Avenue, Carlsbad, CA 92008, USA

Received April 16, 1993; Revised and Accepted May 21, 1993

ABSTRACT6-Azathymidine, 6-aza-2'-deoxycytidine, 6-methyl-2'-deoxyuridine, and 5,6-dimethyl-2'-deoxyuridine nucleo-sides have been converted to phosphoramidite syn-thons and incorporated into oligodeoxynucleotides(ODNs). ODNs containing from 1 to 5 of these modifiedpyrimidines were compared with known 2'-deoxy-uridine, 5-iodo-2'-deoxyuridine, 5-bromo-2'-deoxy-uridine, 5-fluoro-2'-deoxyuridine, 5-bromo-2'-deoxy-cytidine, and 5-methyl-2'-deoxycytidine nucleosidemodifications. Stability in 10% heat inactivated fetalcalf serum, binding affinities to RNA and DNA comple-ments, and ability to support RNase H degradation oftargeted RNA in DNA - RNA heteroduplexes weremeasured to determine structure-activity relationships.6-Azathymidine capped ODNs show an enhancedstability in serum (7- to 12-fold increase over un-modified ODN) while maintaining hybridization proper-ties similar to the unmodified ODNs. A 22-mer ODNhaving its eight thymine bases replaced by eight6-azathymines or 5-bromouracils hybridized to a targetRNA and did not inhibit RNase H mediated degradation.

INTRODUCTIONA variety of modified and unmodified oligonucleotides (ODNs)have been examined as antisense inhibitors of gene expression(1). Although progress has been made in this drug discoveryapproach, certain pharmacokinetic factors may limit the use ofunmodified antisense ODNs in vivo (2). In particular, degradationby ubiquitous nucleases will most likely render unmodified ODNsunsuitable as antisense drugs. Approaches to overcome thisproblem have primarily been directed towards modifications atthe phosphorus atom of the phosphodiester linkage, for example,phosphorothioates, methylphosphonates, and phosphoramidates(3). Though significant resistance to nuclease degradation resultsfrom phosphorus modifications, other antisense properties suchas binding affinity and the ability to support RNase H mediateddegradation of the RNA are compromised (4). Interestingly,

modification of the heterocyclic bases in ODNs to obtain nucleaseresistance has received very little attention (5). The lack ofsignificant work in the heterocycle part of ODNs most likelyreflects the thought that this type of modification will precludeWatson-Crick base-pair hydrogen bonding essential for theformation of stable heteroduplexes.

In an effort to modulate antisense ODNs properties, we studiedthe nuclease sensitivity and hybridization properties of ODNscontaining pyrimidines modified at their 5- and/or 6-positions.Modifications at these positions lie in the major groove and shouldnot directly affect the pyrimidine hydrogen bonding. However,the same modification might alter the syn/anti -conformationabout the glycosyl bond (X), particularly with a bulky peripheralmodification in the 6-position, and could indirectly affectWatson-Crick base pairing at a specific position.We postulate that the 6-position of the pyrimidine ring of

cytosine and thymine may serve as a binding site for certainnucleases (6). If the 6-position of pyrimidines is required fornuclease binding, then removal of one or more or theseelectropositive centers in antisense ODNs may provide nucleaseresistance. Placement of a ring-nitrogen in the 6-position ofcytosine and thymine (for example: 6-azapyrimidines) mightreverse the electronic effect by providing an electronegativecenter, thus, preventing putative nuclease binding. A6-azapyrimidine modification would also have the advantage ofnot providing steric bulk which may interfere with the glycosylbond conformation. Placing a substituent at the 6-positions ofcytosine bases and thymine bases within oligonucleotides mayalso interfere with nuclease recognition of the oligonucleotide.However, hybridization properties may be compromised due tothe perturbation of the syn/anti-conformation.

Sugar-modified ODNs such as 2'-O-methyl (7), 2'-deoxy-2'-fluoro (8), ca- ODNs (9) and carbocyclic ODNs (10) failedto support RNase H mediated degradation in RNA-DNAduplexes. Terminating essential RNA processing by this modeof action may be an important event in the activity of antisenseODNs (11) and points to the design of modified ODNs whichform heteroduplexes that support RNase H activity. We suggest

* To whom correspondence should be addressed

+ Present address: Synthecell Corporation, 7101 Riverwood Dr., Columbia, MD 21046, USA

3198 Nucleic Acids Research, 1993, Vol. 21, No. 14

that heterocycic modification ofODNs as described in this paper,may only have minimal effects on heteroduplex structure andthus, support RNase H degradation of targeted RNA.

Herein, we describe the synthesis of several pyrimidinenucleosides phosphoroamidite synthons and their incorporationinto an antisense ODN sequence by automated DNA synthesis.Structure-activity relationships (SAR) were determined byreplacing one or more thymine/cytosine bases in 16-mer ODNsequence with modified pyrimidines. Modified ODNs were

examined for their nuclease stability in 10% fetal calf serum

(FCS), binding affmnity to DNA and RNA complements, and theability of these modified ODN to support RNase H mediatedcleavage.

RESULTSSynthesis of modified blocked nucleosidesOur route to the synthesis of 6-azathymidine (6), 6-aza-2'-deoxy-cytidine (10), 6-methyl-2'-deoxyuridine (15) and 5,6-dimethyl-2'-deoxyuridine (20) is illustrated in Scheme 1. The blockednucleosides 23-28 were commercially available (12). The recentlydescribed procedure for the synthesis of 2'-deoxynucleosides viaCul catalyst (13) was found suitable to prepare appropriate

O-TMS

N=N

I R"szHH2 R" = CH,

NHN,R

4 N.

R0J

R'O9 -13

O-TUS

HTCN O-TS SI

is

3

4 - 8

R R'

vi 21 Tol. Tol.?1 H

II I1 DMT H

v 12 DMT HIV(13 DMT Amidite

R R' R"

4 Tol. Tol. H6i(S Tol. Tol. CHSil H H CH3'7 DMT H CHS

Iv (* DMT Amidit. CH3

0

HSC...(NH

H~~~~HR@@ R O/

H R'OH 14 - 17

BzBz _ R R'

VI4 .TIPS-

>5 H H

vi6 DMT H

Iv C17 DMT Amidite

R R'

191 Tol. Tol.i20 H H

I2 DMT H

2v2DMT Amidite

19 - 22

R

23 124 Br25 F26 H

23 - 26

R

27 Br2 8 CH3

27, 28

Scheme 1. Reagents and conditions: (i) (a): HMDS, TMSCI, reflux, 1 h, (b):CHCI3, CuI, RT, 3 h; (ii) MeOH/NaOMe, RT, 1 h, dowex H+; (iii) Pyridine,DMTCI, DMAP, RT, 6 h; (iv) THF, DIPEA, DPPC, RT, 2 h; (v) pyridine,TMSCI, BzCI, RT, 2 h, NH40H; see experimental details for abbreviationsused.

quantities of nucleosides 6, 10 and 20. Thus, glycosylation ofsilylated 6-azauracil (1) with 1-chloro-2-deoxy-3,5-di-O-p-toluoyl-a-D-erythro-pentofuranose (3) (14) furnished blockednucleoside 4 in 70% yield as the major isolated product afterchromatography. The blocked nucleoside 4 was further convertedinto 6-aza-2'-deoxycytidine (10) by a reported procedure (15).The latter compound 10 on dimethoxytritylation gave 11 (91%),which on selective N-benzoylation furnished 12 (67%). Phos-phitylation of 12 yielded the desired protected 13 (79%) as aglassy material. 1H and 31P NMR of compound 13 (Table 1)indicated the desired structure of high purity. Following thegeneral glycosylation procedure, silylated 2 was coupled with3 to furnish 5 in 75% isolated yield. The latter compound 5 wasconveniently deblocked by NaOMe/MeOH to furnish 6 inquantitative yield. 6-Azathymidine (6) on dimethoxytritylationfollowed by phosphitylation gave 8 as white foam in 66% overallyield. The structure of compound 8 was established by IH and31P NMR data (Table 1).

Deoxyribosylation of silylated 18 in the presence of CuIfurnished two products in a 7:3 ratio; purification by silica gelchromatography afforded 19 (50%) as crystalline product. Theminor product was characterized as N-3 positional isomer of 19.The structure of 19 was supported by UV and 1H NMR data.Compound 19 was deblocked to furnish nucleoside 20 (95%) asfine needles. Dimethoxytritylation and phosphitylation of 20furnished blocked 22 (>68% for two steps) as a white foam.The structure of 2 was established by spectral data (Table 1).Compound 14 was conveniently prepared by a known procedure(16) and deblocked to furnish 15 in quantitative yield.Dimethoxytritylation and phosphitylation of nucleoside 15 yieldedblocked 17 (> 64% for two steps) as white foam. The structureof 17 was supported by NMR data (Table 1).

Oligonucleotide synthesis, purification, and analysisModified ODNs listed in Tables 2 and 3 were synthesized bystandard solid-support phosphoramidite chemistries on anautomated DNA synthesizer (17). We chose a 16-mer sequence[5'-d(TCCAGGTGTCGGCATC)] from the 5-lipoxygenase5'-untranslated region having a reasonable representation ofpyrimidine bases and lacking discernible hairpin structures as onetarget molecule. Modified ODNs were synthesized by couplingthe desired modified amidite (8, 13, 17, 22-28) at the preselectedpositions. The effectiveness of the coupling was considerablyimproved by introducing a longer wait step ( - 6 min) during thesynthesis. The stepwise coupling efficiency was calculated bymeasuring released DMT cation for each synthesis (1 ismol) andfound to be greater than 98%. The ODNs were removed fromthe CPG-support by standard NH4OH treatment and purified ona reverse phase HPLC column as the 5'-DMT species.Detritylation of each ODNs was carried out under standardconditions to furnish ODNs. The purity of the final detritylatedODNs products was determined by means of denaturingpolyacrylamide gel electrophoresis and were found to be greaterthan 90% full length ODN. 6-Aza T linked to the CPG support,prepared via standard methods (18), was utilized to incorporatethe modified oligonucleotide at the 3'-terminal end of ODNs fornuclease stability studies.The integrity and stability of sequences containing 6-aza T

residues was demonstrated by examining the relative nucleosideratios of a 22-mer 5'-AAA TAG TGT TGC TGA TCT TGAC, wherein T represents the incorporation of 6-aza T. Thismolecule was enzymatically degraded, and nucleotides

Nucleic Acids Research, 1993, Vol. 21, No. 14 3199

dephosphorylated to the corresponding nucleosides and analyzedby HPLC. The calculated relative ratios of the nucleosides werein agreement with the experimental ratios indicating that 6-azaT was successfully incorporated using conventionalphosphoramidite chemistry (data not shown).

Hybridization of oligonucleotidesThe melting temperatures (Tm) for duplexes of ODNs withcomplementary RNA or DNA are listed in Table 2 for severalmodified sequences (19). Table 2 also lists ATm per substitutionaveraged over the sequences listed.

Modification of the 6-position of thymidine with an isostericring nitrogen atom or the more bulky methyl group destabilizedthe duplex formed with RNA or DNA complement.Destabilization was greater for 5, 6-DiMe dU than for 6-aza T.The 6-aza dC modification exhibited destabilizations somewhatgreater than those for 6-aza T.

Substitution of the 5-Me group of T with a halogen (23-25)has essentially no effect on duplex stability In contrast, removalof the 5-Me group ofT (dU, 26) destabilizes the duplex slightly.Addition of a 5-Me or 5-Br group to dC (27, 28), increases duplexTm 0.5 to 1.2°C per substitution. Other investigators havereported similar results with 5-Me pyrimidines (20).1H NMR (NOE) experiments and molecular modeling results

demonstrate 6-Me dU and 5,6-DiMe dU adopt the syn-

conformation (21); the anti-conformation is prevented due tosteric bulk of the 6-methyl group. Our hybridization results are

consistent with such a structure for 6-Me dU and 5,6-DiMe dUcontaining ODNs. Duplex destabilization caused by thesemodifications may be due to inability of 6-methyl pyrimidinesto adopt the anti-conformation in the modified ODNs requiredfor appropriate Watson-Crick hydrogen bonding in an A- orB-form duplex.

Nucleolytic degradationODNs containing a 3'-terminal cap comprised of one to threeT or C modified nucleosides (6-aza T, 6-aza dC, 5-Br dU,5,6-DiMe dU, and 6-Me dU, Table 3) were examined fornuclease sensitivity in DMEM media supplemented with 10%heat inactivated (55°C for 30 min) FCS. The predominantnucleolytic degradation in sera systems of this nature has beenestablished to be 3' to 5'-exonuclease (22). A variety of 3' cappingstrategies and phosphate-backbone modifications, which preventdegradation by serum nucleases, have been reported (23).

The results of the nuclease stability experiments are

summarized in Table 3. The control ODN A containing threeunmodified thymidines at the n- 1, n-2 and n-3 positions wasrapidly degraded with an approximate half life of 0.5 h forhydrolysis of the first nucleotide from the 3'-end of the ODN.Introduction of 6-aza T at the n-1, n-2 and n-3 positions(ODN B) increased the half life of the n-2 oligomer (cleavageof n-I to n-2) - 7-fold (3.5 vs. 0.5 h) over control ODN A.Introduction of a methyl group in the 6-position of T (5,6-DiMedU) and incorporation at two or three sites (ODN K and L)provided a 48- and - 10-fold increase in half-life vs. control ODNA. A cap of three 6-aza Ts at the n, n-1, and n-2 positions(ODN D) afforded roughly 12-fold increase in half-life (3.7 vs.

0.3 h) over control ODN C. One or two 6-aza T modificationsat the 3'-end provided about 3 and 7-fold increases in half-lives(ODN F and E) compared to control ODN C. Three 6-Aza Cat the n-1, n-2 and n-3 positions (ODN H) provides about10-fold increase in half life of n-2 (10 vs. 1 h) over controlODN G. A 5-Br dU capped oligomer J was only twice as stableas the control ODN I in FCS.The structural identity of ODNs containing 6-aza T was

confirmed by HPLC analysis of the enzymatic degradation of[5'-d(AAA TAG TGT TGC TGA TCT TGA C)]. HPLCanalysis after incubation at 37°C for 63 h exhibited fournucleosides, corresponding to dC, dG, 6-aza T and dA. Theunmodified control sequence was completely converted to dC,dG, T, and dA within 0.5 h. Incubation of the oligomer for upto 14 h was insufficient to complete the digestion. At these shorterincubation times, peaks other than the expected nucleosides wereobserved. As the incubation time increased, these peaksdiminished and the 6-aza T peak increased. These results suggestthat incorporation of 6-aza T nucleosides impart substantialnuclease resistance to the modified ODN.

Ribonuclease H activitySince the mechanism of action of antisense oligonucleotides ismost often considered to be cleavage of targeted mRNA by RNaseH (2), we were interested in determining if ODNs containingheterocylic modification in the pyrimidine ring would supportthis activity. This led us to incorporate 6-aza T or 5-Br dU intoan antisense ODN to human 5-lipoxygenase DNA [(5'-d(AAATAG TGT TGC TGA TCT TGA C) wherein T indicates 6-azaT or 5-Br dU] and determine whether the modified ODNs-complementary RNA duplexes were substrates for RNase H. The

Table 1. Summary of NMR data (a) in CDC13

Compd H-l' H-2',2" H-3' H4' H-5'," H-5 or NH or Aromatic Others(m) (m) (m) (m) C5-CH3 NH2 H or

(s) (br S) (m) (31P)

7 6.56 (t) 2.25, 2.60 4.60 4.08 3.25, 3.40 2.02 8.6 T6.8-7.5 -81 6.65 (m) 2.42, 2.63 4.60 4.21 3.2-3.65 1.98, 2.01 8.56 6.8-7.44 3.78 (s, OMe); 148.8, 149.010 6.40 (t) 2.0, 2.35 4.21 3.67 3.2-3.5 7.47 7.84, 7.95 - 5.12 (d), 4.63 (t); OH112 6.45 (t) 2.05, 2.35 4.24 3.8 2.86 - 7.81, 8.0 6.8-7.38 3.7 (s,OCH3), 5.2 (d,OH)12 6.65 (t) 2.3, 255 4.58 4.03 3.22, 3.36 - 8.58 6.8-8.0 3.76, 3.77 (2s, OMe)13' 6.70 (m) 2.4, 2.6 4.2 4.17 3.2-3.6 - 8.75 6.8-8.0 3.75 (m, OCH3); 147.5, 147.716 6.0 (t) 2.2, 2.83 4.6 3.82 3.35, 3.5 5.51 8.5 6.8-7.4 3.78 (s, OCH3)17' 6.15 (t) 2.45, 2.85 4.8 4.05 3.25-3.5 5.55 8.6 6.78-7.45 3.78 (s, OCH3); 147.8, 148.0202 6.17 (t) 1.95, 2.63 4.65 4.25 3.6 1.83 11.2 - 5.1 (d, OH)21 6.05 (t) 2.1, 3.1 4.8 4.0 3.65-3.9 1.954 8.38 6.8-7.35 3.8 (s, OCH3)221 6.20 (m) 2.2, 2.8 4.65 4.05 3.35-3.9 1.955 8.05 6.75-7.45 3.77 (s, OCH3); 147.3, 147.6

1 In DMSO-d6; 2 Mixture of distereoisomers; 3 C6-CH3 at 2.25; 4 C6-CH3 at 2.33; 5 C6-CH3 at 2.35.Abbreviations: br s = broad singlet; d = doublet; m = multiplet; t = triplet; s = singlet.

3200 Nucleic Acids Research, 1993, Vol. 21, No. 14

Table 2. Melting temperature (Tm) and average ATm/modification data

5' T C C A G G T G T C G G C A T C1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

3,

Compound2 Site(s) of TM ( c)4(Amidite used) Modification3 DNA5 RNA6

natural bases6-Aza T

(8)

average ATm/mod.6Aza dC

(13)

average ATm/mod.6-Me dU

(17)

average ATm/mod.5,6-DiMe dU

(22)

average ATm/mod.5-I dU(23)

average ATm/mod.5-Br dU

(24)

average ATm/mod.5-F dU(25)

average ATm/mod.dU(26)

average ATm/mod.5-Br dC

(27)average ATm/mod.

5-Me dC(28)

average ATm/mod.

none

1157,9

1,7,9,15

22,310,11

115

7,91,7,9,15

157,9

1,7,9,15

157,9

1,7,9,15

115

7,91,7,9,15

115

7,91,7,9,15

157,9

1,7,9,15

10,112,3,10,11,13

10,112,3,10,11,13

62.9762.862.058.254.5nd61.058.956.3-2.462.260.954.150.9-2.362.161.255.052.7-2.462.762.562.361.7-0.161.861.963.463.3+0.062.762.363.762.7+0.262.461.962.161.7-0.364.666.8+0.963.966.0+0.7

62.38

56.5-1.4

58.553.7-3.1

46.5-3.9

49.2-3.3

61.9-0.162.462.162.163.6-0.0

61.7-0.262.161.860.560.1-0.665.767.5+1.263.663.2+0.5

lAntisense DNA was modified each time with appropriate pqyriimidine base as

specified in the table; 2see figure 1 for chemical structure; the number in thetable refers to the position in parent antisense DNA wherein 1-5 bases were

replaced at a time as specified; 4see experimental details for Tm measurements;5Tm value with a complementary DNA; 6T value with a complementary RNA;7Tm of a natural DNA with antisense DNA; Tmof a natral RNA with antisense

DNA.

E. coli RNase H requires at least 2 sequential ribonucleotidespaired with two ribodeoxynucleotides within a duplex to cleavein the ribonucleotide portion (4) and mammian RNase H (HeLacells) may require at least a five base pair duplex (11). Themodified ODNs sequence studied, has eight 6-aza T or 5-Br dUmodifications, separated by no more than two natural bases, andshould support cleavage by E. coli RNase H but may not supportRNase H from HeLa cell extracts. Our results indicated thatE. coli RNase H was able to recognize and cleave mRNA bound

Table 3. Nucleolytic degradation data in 10% FCS

ODN sequence T1/2 (hr) Tmd (5'-3') in FCSa (MC)c

A X TTTC 0.5 59.2B X TTTC 3.5b 55.4C X TTT 0.3 59.0D X TTT 3.7 44.1E X TTIT 2.2 ndF X TTT 1.0 ndG XCCCC 1.0 69.5H XCCCC 10.6b 60.3I X UUUC 0.7 58.7J x uuu'c 1.5b 57.3K X UUU2C 24.0b 43.9L X UUU2C 5.0" ndM X UUU3C 2.2b 53.6

X = TCC AGG TGT CCG; T = 6-aza T; C =6 aza dC; UUU1 = three 5-BrdU; UUU2 = three 5,6-DiMe dU; U2 = two 5,6-DiMe dU; UUU3 = three6-Me dU; nd = not determined; a = same lot of FCS was used for allexperiments; b = T1/2 represents the values for full length 15-mer, after thecleavage of first 3'-unmodified nucleoside;C = Tm value with a complementaryRNA.

to either natural or 6-aza T modified DNA (data not shown).Similar cleavage results were found when the 5-Br dU modifiedDNA-RNA complement was treated with HeLa cell extracts.

CONCLUSIONSOur primary objective in this work was to examine if the 5 and6-positions of the pyrimidine ring could be used to modulate theessential antisense properties of ODNs. One must consider thatany structural perturbation of the pyrimidines devised to obtainnuclease resistance must be compatible with an appropriate levelof specificity and binding affinity to the targeted RNA to provideuseful antisense activity. Would the pyrimidine modified ODNssupport RNase H mediated cleavage of the targeted RNA isanother design feature to consider?

3'-Capping of an ODN with 6-aza T or 6-aza dC enhancednuclease resistance compared to the unmodified sequence (7- to12-fold for caps of three substituted pyrimidines); even greaternuclease resistance was obtained by placing a methyl group inthe 6-position ofT (48-fold increase for ODN K). However, thesedata are not sufficient to allow us to conclude that FCSexonuclease recognition of single-stranded DNA is through the6-position of pyrimidines and requires further studies.The 6-aza modified ODNs, in addition to their enhanced

nuclease resistance, demonstrate reasonable binding affinities(destabilization of only -1.4 to - 1.8°C/modification). This levelof destabilization is in the range of methoxyethylamidates (31)and methylphosphonates (32), which also provide enhancednuclease stability to antisense ODNs. Furthermore, we find thatthe pyrimidine modified oligonucleotides form heteroduplexeswith RNA that support RNase H cleavage. Our data demonstratedthat a ring nitrogen in the 6-position of T or a bromo group inthe 5-position ofdU provided modified ODNs that support RNaseH cleavage.Oligonucleotides with backbone modifications suchas methylphosphonates and amidates or with sugar modificationssuch as 2'-0-methyls or a-sugars do not form heteroduplexesthat support RNase H activity (2). These types of modified ODNsmay form heteroduplexs that do not allow conformations

Nucleic Acids Research, 1993, Vol. 21, No. 14 3201

necessary for the formation of the enzyme-substrate complex.Further studies are in progress to understand the interactions ofpyrimidines with cellular nucleases.

EXPERIMENTAL

The 1H NMR spectra were recorded on a Varian 400 MHzinstrument. Chemical shifts were reported (6 ppm) relative toDMSO or TMS. The 31P NMR spectra were recorded on a

Varian 400 MHz instrument at 161.9 MHz. Chemical shifts werereported relative to H3PO4 used as external standard. Thin-layerchromatography (tic) was performed on Merk precoated silicagel 60F254 plates. Silica gel (ICN 32/63;/60 A) was used forflash chromatography. All solvents used were sure-seal (AldrichCo.) reagent grade. Evaporation's were conducted underdiminished pressure with the bath temperature below 30°C.

Abbreviations: 6-aza T: 6-azathymidine; 6-aza dC: 6-aza-2'deoxycytidine; 6-Me dU: 6-methyl-2'-deoxyuridine; 5, 6-DiMedU: 5,6-dimethyl-2'-deoxyuridine; 5-I dU: 5-iodo-2'deoxy-uridine; 5-Br dU: 5-bromo-2'-deoxyuridine; 5-F dU: 5-fluoro-2'-deoxyuridine; 5-Br dU: 5-bromo-2'-deoxcytidine; 5-Me dC:5-methyl-2'-deoxycytidine; DMTCI: 4,4'-dimethoxytriphenyl-methyl chloride; CuI: copper iodide; NaOMe: sodium methoxide;DIPEA: N,N-diisopropyl ethylamine; THF: tetrahydrofuran;DPPC: N, N-diisopropylmethylphosphonamidic chloride;HMDS: 1,1,1,3,3,3-hexamethyldisilazane; TMSCI: chloro-trimethylsilane; RT: room temperature; CC: column chromato-graphy; DMAP: 4-dimethylaminopyridine; EDTA: ethylene-diaminetetraacetic acid; DMEM: Dulbecco's modified essentialmedium; FCS: 10% fetal calf serum.

General methods for (i) silylation/glycosylation, (ii)detoluoylation, (iii) tritylation, (iv) phosphoramidationMethod (i) Silylation/glycosylation (e.g., 1-4, and 18-19)(2-5: a representative example). A mixture of 6-azathymine(Aldrich Co.)(5.0 g, 39.4 mmol), HMDS (15 ml), and TMSC1(0.5 ml) was heated to reflux (150°C) for 1 h. The excess ofHMDS/TMSC1 was removed under vacuum. The residual oilcrystallized on drying under vacuum to furnish 6.57 g (61%)of 2; mp 43°C. A mixture of 2 (4.0 g, 14.7 mmol) and 3 (4.8g, 12.4 mmol) in dry CHC13 (300 ml) was stirred at RT underArgon. Cul (2.4 g, 12.4 mmol) was added to the solution andslurry stirred for 3 h at RT. The reaction was quenched byaddition of sat. aq. NaHCO3 (200 ml), stirred for 15 min andfiltered through a pad of celite. The CHC13 layer was separated,washed (sat. NaCl, 200 ml), dried (MgSO4) and concentrated.The residue was purified by short silica gel CC. Elution withEtOAc/hexanes (1:1, v/v) furnished the product, which on

crystallization gave 4.37 g (in two crops, 75.6%) of analyticallypure 5, mp 170°C [lit. (24) 173-175°C].

Method (ii) Detoluoylation (e.g., 9-10, 19-20) (5-6: a

representative example). To a stirred solution of 5 (0.48 g, 1mmol) in MeOH (15 ml) was added NaOMe (50 mg) at RT.After 1 h, the solution was neutralized (Dowex-50H+) andfiltered. The resin was washed with MeOH (2x25 ml). Thecombined filtrates were concentrated under vacuum and theresidue dissolved in water (3 ml), washed with CHC13 (3 x5ml). The aq. layer was freeze dried to yield 6; 0.23 g(96%)(compound 6 was identical in all respects e.g. UV, 1HNMR and tlc to the values in reference 24).

Method (iii) Tritylation (e.g., 10-11, 15-16, 20-21) (6-7:a representative example). To a stirred solution of 6 (2.43 g, 10mmol) in dry pyridine (100 ml) was added DMTC1 (4.06 g, 12mmol) and DMAP (0.12 g, 1 mmol) at RT. The solution wasstirred for 6 h and reaction quenched by adding MeOH (10 ml).The reaction mixture was concentrated under vacuum and theresidue dissolved in CH2CI2 (200 ml), washed (sat. NaHCO3,100 ml and sat. NaCl, 100 ml) and dried (MgSO4). The solutionwas concentrated under vacuum and the residue purified by silicagel CC (pre wet with EtOAc/TEA, 99:1, v/v, until basic andeluted with same solvent). Appropriate fractions were pooled andevaporated to furnish 3.82 g (70%) of 7 as white foam.Homogenous by tic (CHCl3/MeOH, 9:1, v/v) Rf = 0.46.

Method (iv) Phosphoramidation (e.g., 12-13, 16-17, 21-22)(7-8: A representative example). To a stirred solution of 7 (1.63g, 3 mmol) in dry THF (50 ml) was added DIPEA (1.56 ml,9 mmol) and solution cooled to 0°C. DPPC (1.42 ml, 6 mmol)was added to the cold solution over a period of 15 min. Thereaction mixture was then stirred at RT for 2 h. EtOAc (100ml, containing 1% TEA) was added and the solution washed (sat.NaCl, 2 x 100 ml), dried (MgSO4) and solvent removed underreduced pressure. The residue was purified by silica gel CC (sameas in method iii). Pooling and concentration of appropriatefractions furnished 1.41 g (66%) of 8 as white foam.Homogenous by tic (CHCl3/MeOH, 9:1,v/v) Rf = 0.65.

5,6-Dimethyl-2'-deoxyuridine (20). Silylation of 5,6-dimethyl-uracil (25)(5 g, 35.71 mmol) followed by glycosylation with 3(13.96 g, 36 mmol) as described in method (i) furnished a mixtureof N-1 and N-2 (7:3 respectively by 'H NMR) glycosylatedproducts in 50% yield. The desired compound 19 was easilyseparated from minor positional isomer by silica gel CC. TheN-I nucleoside 19 was deblocked by method (ii) in quantitativeyield to furnish 20; mp 170- 171°C.

ODN synthesis. All ODNs were synthesized on an ABI model380 B DNA synthesizer using standard chemistry (17). The 6-azaT was coupled to a controlled-pore glass (LCAA-CPG) via arecently reported method (18). The 6-aza T attached CPG servedas the starting material for introducing the modification at the3'-end of an ODN.

Determination of thermal stabilityAbsorbance vs. temperature curves were measured at 260 nmusing a Gilford 260 spectrophotometer interfaced to an IBM PCcomputer and a Gilford Response II spectrophotometer. Thebuffer contained 100 mM Na+, 10 mM phosphate and 0.1 mMEDTA, pH 7.0. Oligonucleotide concentration was 4 mM eachstrand determined from the absorbance at 85°C and extinctioncoefficients calculated according to Puglisi and Tinoco (26).Tm's of duplex formation were obtained from fits of data to atwo state model with linear sloping baselines (27). Reportedparameters are averages of at least three experiments.

Enzymatic digestion of oligonucleotidesODNs (66 mg/ml, - 13 mM) were incubated at 370C in DMEMsupplemented with 10% heat inactivated (55°C for 1 h) FCS for24 h. At various time points 15 ,gl of the ODN/media + FCSmixture were removed and added to.20 tl of 9M urea in 2 xTBE(178 mM Tris, 178 mM borate and 4 mM EDTA, pH 8.0),mixed and stored at -20°C. The samples were then analyzed

3202 Nucleic Acids Research, 1993, Vol. 21, No. 14

by polyacrylamide electrophoresis (20% PA/7 M urea denaturinggels) and visualized by stains all reagent (l-ethyl-2-[3-(1-ethyl-naptho[ 1 ,2-d]thiazolin-2-ylidene)-2-methylpropenyl]napthol[1,2-d]thiazolium bromide). Following electrophoresis and stain-ing, degradation products were quantitated using laser densitometry(Molecular Dynamics Densitometer, Molecular Dynamics,Sunnyvale, CA). Rates of degradation were determined for theloss of the first 6-aza T from the 3'-end of the ODN (see ref.28 for experimental details). The data was graphed and thegraphic estimations for the half-life of the modified ODNs weremade by comparison to the unmodified ODN A.

Modified oligonucleotide compositional analysisEnzymatic degradation was carried out using 0.1 units of snakevenom phosphodiesterase (Pharmacia, Piscataway, NJ), 23 unitsnuclease P1 (Gibco BRL, Gaithersburg, MD), 24 units calfintestinal phosphatase (Boehringer Mannheim, Indianapolis, IN),and 10 nanomoles oligomer in 50 mM Tris-HCl pH 8.5, 14mM MgCl2, and 72 mM NaCl in a total volume of 20 tAl.Reactions were incubated at 37°C for 30 min. HPLC analysiswas carried out using a reverse phase column (Alltechnucleoside/nucleotide; 4.6 x250 mm; Alltech Associates, Inc.,Deerfield, IL). All analyses were performed at room temperature.The solvents used were A: water and B: acetonitrile. Separationof 6-aza T from other nucleosides was accomplished with thefollowing gradient: 0-5 minutes, 2% B (isocratic); 5-20minutes, 2% B to 10% B (linear); 20-40 minutes, 10% B to50% B. The integrated area per nanomole was determined usinga nucleoside standards. Relative nucleoside ratios were calculatedby converting integrated areas to molar values and comparingall values to thymidine, which was set at its expected value foreach oligomer.

RNase H mediated cleavageE.coli RNase H cleavage: The 5-lipoxygenase mRNA (2.5 Kb)was synthesized in vitro for this experiment. 2.5 lAg mRNA (3.2pM) was suspended in 10 yl 10 mM Tris-HCI, pH 7.8, andheated to 60°C for 5 min and then rapidly cooled to 4°C on ice.A 3-fold molar excess of oligonucleotide [5'd(AAA TAGTGTTGC TGA TCT TGA C)], where T is 6-aza T (9.6 pM) wasadded and the solution adjusted to 3 mM MgCl2, 0.5 mM ATP,and 20 mM creatine phosphate, final volume 20 ,ul. Mixtureswere incubated at 37°C for 10 min and then 0.8 Units E.coliRNase H was added and the reaction allowed to continue for15 min at 37°C. The samples were then heated to 90°C for 10min in the presence of formamide gel loading buffer. Thesesamples were analyzed for cleavage using 1.2% agarose-formaldehyde gel electrophoresis.Mammalian RNAse H cleavage: Modified DNA [5'd(AAATAGTGT TGC TGA TCT TGA C)], where T is 5-Br dU] andcomplementary synthetic RNA was used to assess in vitrocleavage by HeLa RNase H (29). The synthetic RNA was endlabeled using gamma [32P] ATP and gel purified. The oligoDNA and labeled RNA were then prehybridized at 37°C for 15min. and slow cooled to room temperature in 20mM Tris-Cl pH7.5, l10mM KCI, 10mM MgCl2, lmg/ml tRNA (BRL cat#5401SA), 40U/reaction RNasin (Promega cat # N251B) at lnMRNA: 10nM DNA. Purified HeLa nuclear extract (30) at 2,ugtotal protein was then added to the duplexed reactions andincubated at 37°C for 10 min. The reactions were quenched in2 xTBE/9M urea then loaded directly onto 20% PAG/7M urea.

ACKNOWLEDGMENTS

We thank Dr Joel Martin and Patrick Wheeler for the NMRstudies and Hedy Chan for the technical assistance.

REFERENCES1. (a) Erickson, R.P. and Izant, J.G. (1992) Gene Regulation: Biology of

Antisense RNA and DNA, Raven Press, New York, NY; (b) Wickstrom,E. (1991) Prospects for Antisense Nucleic Acid Therapy of Cancer and AIDS,Wiley-Liss, New York, NY; (c) Murray, A.H. (1992) Antisense RNA andDNA, Wiley-Liss, New York, NY; (d) Crooke, S.T. and Lebleu, B. (1993)Antisense Research and Applications, CRC Press, Florida, FL; (e) Propst,C.L. and Perun, T.J. (1992) Nucleic Acid Targeted Drug Design, MarcelDekker, Inc., New York, NY.

2. Crooke, S.T. (1992) Annu. Rev. Pharmacol. Toxicol., 32, 329-376; Cohen,J.S. (1991) Antiviral Res., 16, 121-133; Cook, P.D. (1991) AnticancerDrug Design, 6, 585-608; Mirabelli, C.K., Bennett, C.F., Anderson, K.and Crooke, S.T. (1991)AnticancerDrug Design, 6, 647-661; Stein, C.A.(1992) Leukemia, 6, 967-974.

3. Uhlamann, E. and Peyman, A. (1990) C7enm Rev., 90, 543-584; Goodchild,J (1990) Bioconjugate Chem., 1, 165-186 and references cited therein.

4. Dagle, J.M., Weeks, D.L and Walder, J.A. (1991)Antisense Res. Develop.,1, 11-20.

5. Sanghvi, Y.S. Heterocyclic Base Modifications in Nucleic Acids and TheirApplications in Antisense Oligonucleotides in book reference Id above;Sanghvi, Y.S., Hoke, G.D., Zounes, M.C., Freier, S.M., Martin, J.F.,Chan, H., Acevedo, O.L., Ecker, D.J. Mirabelli, C.K., Crooke, S.T. andCook, P.D. (1991) Nucleosides Nucleotides, 10, 345-346.

6. Ivanteich, K.M. and Santi, D.V. (1992) Prog. NucleicAcids Res. Mol. Biol.,42, 127-156.

7. Inoue, H., Hayase, Y., Imura, A., Iwai, S., Miura, K. and Ohtsuka, E.(1987) NucleicAcids Res., 15, 6131-6148; Sproat, B.S., and Lamond, A.I. 2'-O-alkyloligoribonucleotides in book reference ld above.

8. Kawasali, A.M., Casper, M.D., Freier, S.M., Lesnik, E.A., Zounes, M.C.,Cumniins, L.L., Gonzalez, C. and Cook, P.D. (1993) J. Med. Chem., 36,in press.

9. Morvan, F., Rayner, B. and Imbach, J.-L. (1991)AnticancerDrug Design,6, 521-530.

10. Moser, H. personal communication.11. Eder, P.S. and Walder, J.A. (1991) J. Biol. Chem, 266,6472-6479; Monia,

B.P., Lesnik, E.A., Gonzalez, C., Lima, W.F., McGee, D., Guinosso, C.J.,Kawasaki, A.M., Cook, P.D. and Freier, S.M. (1993) J. Biol. Chem., inpress.

12. Purchased from ChemGenes, Needham, MA.13. Freskos, J.N. (1989) Nucleosides Nucleotides, 8, 549-556 and 1075-1076.14. Hoffer, M. (1960) Chem. Ber., 93, 2777-2783 .

15. Pliml, J. and Sorm, F. (1963) Coll. Czech. Chem. Comm., 28, 546-549;Drasar, P., Hein, L. and Beranek, J. (1976) Coll. Czech. Chem. Comm.,41, 2110-2123.

16. Tanaka, H., Hayakawa, H., lijima, S., Haraguchi, K. and Miyasaka, T.(1985) Tetrahedron, 41, 861-866; Holy, A. (1973) Coll. Czech. Chem.Comm., 38, 3912-3931.

17. Standard protocol using an ABI 380B DNA synthesizer.18. Damha, M.J., Giannaris, P.A. and Zabarylo, S.V. (1990) Nucleic Acids

Res., 18, 3813-3821.19. Freier, S.M., Lima, W.F., Sanghvi, Y.S., Vickers, T., Zounes, M., Cook,

P.D. and Ecker, D.J. Themodynamics of Antisense Oligonucleotidehybridization In book reference la above.

20. Sowers, L.C., Shaw, B.R. and Sedwick, W.D. (1987) Biochem. Biophys.Res. Comm., 148, 790-794; Sagi, J., Szemzo, A., Ebinger, K., Szabolcs,A., Sagi, G., Ruff, E. and Otvo3, L. (1993) Tet. eits., 34, 2191-2194.

21. Martin, J., Grieffy, R., Cook, P.D. and Sanghvi, Y.S. unpublished data.22. Walder, R.Y. and Walder, J.A. (1988) Proc. Natl. Acad. Sci. USA, 85,

5011-5015.23. Quartin, R.S., Brakel, C.L. and Wetmur, J.G. (1989) Nucleic Acids Res.,

18, 7253-7262; Shaw, J.-P., Kent, K., Bird, J., Fishback, J. and Froehler,B. (1991)NucleicAcidsRes., 19, 747-750; Gamper, H.B., Reed, M., Cox,T., Virosco, S., Adams, A.D., Gall, A.A., Scholler, J.K. and Meyer, Jr.R.B. (1993) Nucleic Acids Res., 21, 145-150.

24. Prystas, M. and Sorm, F. in 'Synthetic procedures in nucleic acid chemistry'Eds. Zorbach, W.W. and Tipson, R.S., John Wiley & Sons, Inc. 1968,350-354.

Nucleic Acids Research, 1993, Vol. 21, No. 14 3203

25. Vorbruggen, H., Krolikiewicz, K. and Bennua, B. (1981) Chem.Ber., 114,1234-1255.

26. Puglisi, J.D. and Tinoco, I. Jr. (1989) Methods Enzymol., 180, 304-324.27. Petersheim, M. and Turner, D.H. (1983) Biochemistry, 22,256-263.28. Hoke, G.D., Draper, K., Freier, S.M., Gonzalez, C., Driver, V.B., Zounes,

M.C. and Ecker, D.J. (1991) Nucleic Acids Res., 20, 5743-5748.29. Agrawal, S., Mayrand, S.H., Zamecnik, P.C. and Pederson, T. (1989) Proc.

Natl. Acad. Sci. USA, 87, 1401-1405.30. Dignam, J.D., Lebovitz, R.M. and Roeder, R.G. (1983) NucleicAcids Res.,

11, 1475-1489.31. Dagle, J.M., Andracki, M.E., Devine, R.J. and Walder, J.A. (1991) Nucleic

Acids Res., 19, 1805-1810.32. Spiller, D.G. and Tidd, D.M. (1992) Anti-Cancer Drug Design, 7, 115 -129;

Giles, R.V. and Tidd, D.M. (1992) Nucleic Acids Res., 20, 763 -770.