Solid-Phase Oligodeoxynucleotide Synthesis: A Two-Step Cycle Using Peroxy Anion Deprotection

Article abstract of DOI:10.1021/ja030376n

A novel solid-phase phosphoramidite based oligodeoxynucleotide two-step synthesis method has been developed. Keys to this method are replacement of the 5′-dimethoxytrityl blocking group with an aryloxycarbonyl and the use of N-dimethoxytrityl protection for the exocyclic amines of adenine and cytosine. With these modifications, coupling of each 2′-deoxynucleoside 3′-phosphoramidite to the growing oligodeoxynucleotide on the solid support can be followed by treatment with an aqueous mixture of peroxy anions buffered at pH 9.6. This reagent effectively removes the carbonate protecting group and simultaneously oxidizes the phosphite internucleotide linkage. As a consequence a new two-step synthesis cycle is possible. Oligodeoxynucleotides synthesized using this approach are identical to authentic samples when tested by a variety of analytical techniques.

Full text of DOI:10.1021/ja030376n

Published on Web 10/09/2003
Solid-Phase Oligodeoxynucleotide Synthesis: A Two-Step
Cycle Using Peroxy Anion Deprotection
Agnieszka B. Sierzchala, Douglas J. Dellinger,‡ Jason R. Betley,†,§
Tadeusz K. Wyrzykiewicz,§ Christina M. Yamada,† and Marvin H. Caruthers*,†
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Colorado,
Boulder, Colorado 80309, and Agilent Laboratories, 3500 Deer Creek Road,
Palo Alto, California 94304
Received June 23, 2003; E-mail: marvin.caruthers@colorado.edu; doug_dellinger@agilent.com
Abstract: A novel solid-phase phosphoramidite based oligodeoxynucleotide two-step synthesis method
has been developed. Keys to this method are replacement of the 5′-dimethoxytrityl blocking group with an
aryloxycarbonyl and the use of N-dimethoxytrityl protection for the exocyclic amines of adenine and cytosine.
With these modifications, coupling of each 2′-deoxynucleoside 3′-phosphoramidite to the growing
oligodeoxynucleotide on the solid support can be followed by treatment with an aqueous mixture of peroxy
anions buffered at pH 9.6. This reagent effectively removes the carbonate protecting group and
simultaneously oxidizes the phosphite internucleotide linkage. As a consequence a new two-step synthesis
cycle is possible. Oligodeoxynucleotides synthesized using this approach are identical to authentic samples
when tested by a variety of analytical techniques.
Over the past 20 years, the method of choice for the chemical
synthesis of oligodeoxynucleotides (ODNs1) has been the
phosphoramidite four-step process which utilizes the reaction
of deoxynucleoside phosphoramidites with a solid-phase tethered
deoxynucleoside or oligodeoxynucleotide (Scheme 1).2-4 Ini-
tially the 5′-O-dimethoxytrityl (DMT) group is removed from
a deoxynucleoside linked to the polymer support. Step 2,
elongation of a growing oligodeoxynucleotide, occurs via the
initial formation of a phosphite triester internucleotide bond.
This reaction product is first treated with a capping agent
designed to esterify failure sequences and cleave phosphite
reaction products on the heterocyclic bases. The nascent
phosphite internucleotide linkage is then oxidized to the
corresponding phosphotriester. In the final step of each cycle,
the DMT group is removed from the growing oligodeoxynucle-
otide using a large excess of a weak acid, trichloroacetic acid
(TCA), in an organic solvent. Further repetitions of this four-
step process generate the ODN of desired length and sequence.
The final product is cleaved from the solid phase and obtained
free of base and the â-cyanoethyl phosphate5,6 protecting groups
by treatment of the support with concentrated ammonium
hydroxide.4 ODNs synthesized with this chemistry continue to
be of satisfactory quality for most biological uses such as DNA
sequencing, PCR applications, and site-specific mutagenesis.
In recent years, an impetus to develop additional strategies
for the synthesis of ODNs has emerged due to several new,
highly specialized uses of synthetic DNA. For example, a variety
of applications in the fields of genomics and high throughput
screening have fueled the demand for highly parallel, microscale
synthesis of DNA and for DNA sequences attached to planar
glass surfaces. An application which exemplifies this trend is
DNA microarrays.7,8 Demand has also increased for the large-
scale synthesis of modified DNA useful for therapeutic
applications9-11 and for the synthesis of viruses12 or genes.13
Although many genes or gene fragments can be obtained by
the use of various biological (cloning) or biochemical (PCR)
procedures, there are increasing numbers of examples where
this is difficult or impossible. These include the assembly of
‡ Agilent Laboratories, 4380 Ziegler Road, MS-16, Fort Collins, CO
80525.
† University of Colorado.
§ Current Address: JRB: Emmbrook, The Street, Lackford, Suffolk IP28
6HW, UK. TKW: Isis Pharmaceuticals, 2292 Faraday Ave., Carlsbad, CA
92008.
(5) Ogilvie, K. K.; Theriault, N. Y.; Seifert, J.-M.; Pon, R. T.; Nemer, J. J.
Can. J. Chem. 1980, 58, 2686-2693.
(1) Abbreviations: B, appropriately protected purine or pyrimidine base; CPG,
controlled pore glass; DMT, 4,4′-dimethoxytrityl; TCA, trichloroacetic acid;
ODN, oligodeoxynucleotide; ARCO, aryloxycarbonyl; MCPBA, m-chlo-
roperbenzoic acid; TLC, thin-layer chromatography; TIPS, tetraisopropy-
ldisiloxane-1,3-diyl; DCM, dichloromethane; TEMED, N,N,N′,N′-tetram-
ethylethylenediamine; DMSO, dimethyl sulfoxide; Bz, benzoyl; Fmoc,
9-fluorenylmethoxycarbonyl; iBu, isobutyryl; DPC, N,N-diphenylcarbamoyl;
N-Me-Imid, N-methylimidazole; PAP, 4-(phenylazo)phenyl; Fmol, 9-fluo-
renylmethyl; PAGE, polyacrylamide gel electrophoresis; EDTA, ethylene-
diaminetetraacetic acid.
(6) Sinha, N. D.; Biernat, J.; Koster, H. Tetrahedron Lett. 1983, 24, 5843-
5846.
(7) Fodor, S. A. Science 1997, 277, 393-395.
(8) Lipshutz, R. J.; Fodor, S. P. A.; Gingeras, T. R.; Lockhart, D. J. Nature
Genet. Microarray Suppl. 1999, 21, 20-24.
(9) Bennett, C. F. Biochem. Pharmacol. 1998, 55, 9-19.
(10) Yuen, A.; Halsey, J.; Fisher, G.; Advani, R.; Moore, M.; Saleh, M.; Ritch,
P.; Harker, G.; Ahmed, F.; Jones, C.; Polikoff, P.; Keiser, W.; Kwoh, J.;
Holmlund, J.; Dorr, A.; Sikic, B. American Society of Clinical Oncology’s
37th Annual Meeting, San Francisco, CA, May 12, 2001.
(11) Lacy, J.; Loomis, R.; Cheng, Y.-C. Proc. Assoc. Cancer Res. 2003, 44,
Abstract 6438.
(2) Letsinger, R. L.; Lunsford, W. B. J. Am. Chem. Soc. 1976, 98, 3655-
3661.
(3) Beaucage, S. L.; Caruthers, M. H. Tetrahedron Lett. 1981, 22, 1859-1862.
(4) Matteucci, M. D.; Caruthers, M. H. J. Am. Chem. Soc. 1981, 103, 3186-
3191.
(12) Cello, J.; Paul, A. V.; Wimmer, E. Science 2002, 297, 1016-1018.
(13) Marshall, W. S.; Boymel, J. L. Drug DiscoVery Today 1998, 3, 34-42.
9
10.1021/ja030376n CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 13427-13441
13427

A R T I C L E S
Sierzchala et al.
Scheme 1
nonnatural, biologically active genes14 and genes designed for
rapid, extensive modifications (cassette mutagenesis) which are
useful in studying structure-function relationships.15,16
Each of these applications places ever more stringent require-
ments on this four-step chemical synthesis process. Side-
reactions, which are known to occur at detectable but acceptable
amounts during routine synthesis, can rise to unacceptable levels
under the conditions required for these expanded applications.
One example of these side reactions is repetitive exposure of
the growing ODN chain to a protic acid which results in
depurinated residues from deoxyadenosine and deoxy-
guanosine.17-19 In addition to depurination, another related
problem associated with acid detritylation is the reversible
formation of dimethoxytrityl carbocation. To obtain quantitative
deprotection of the DMT group, this carbocation must be flushed
from the surface. Otherwise a series of failure sequences develop
(retritylation of the 5′-deoxynucleoside hydroxyl due to residual
carbocation) that continue to grow in number and size with each
synthesis cycle. To ensure purity of synthesized oligodeoxy-
nucleotides free of these side products, rigorous column
chromatography is required to minimize mutagenesis from
cloned, synthetic DNA.15,20
excess of acid to deprotect and flush the trityl group from the
solid-phase matrix. However complete removal of the dimethox-
ytrityl carbocation under conditions of both microscale, on
planar glass surfaces, and macroscale syntheses requires even
larger volume-equivalents of the protic acid solution and
increased reaction times compared to those used during typical
laboratory-scale synthesis on porous glass surfaces.21-23 This
excessive exposure to protic acid can cause reductions in product
yield and quality primarily due to acid depurination. The
presence of depurinated side products can in turn effect the
performance of the ODN product in these specialized applica-
tions.24 An alternative light-catalyzed detritylation procedure has
been proposed for DNA microarray synthesis.25,26 However with
this approach, only a single exposure to acid detritylation is
possible which would maximize the reversible formation of
residual dimethoxytrityl containing oligodeoxynucleotides in
each cycle.
As these and other applications of phosphoramidite ODN
synthesis continue to evolve, further modifications of the
synthesis cycle are inevitable. These may include new protecting
groups, solvent changes for various steps, different surface
materials, and even different synthesis cycles. The work
described herein outlines the lattersan alternative synthesis
approach that is potentially useful for both macroscale and
microscale DNA synthesis. The key to this new approach is
Under the standard synthesis conditions (Scheme 1), the
carbocation reversibility problem is minimized by using a large
(14) Caruthers, M. H. In Protein Engineering; Oxender, D. L., Fox, C. F., Eds.;
Alan R. Liss: New York, 1987; pp 65-70.
(21) Paul, C. H.; Royappa, A. T. Nucleic Acids Res. 1996, 24, 3048-3052.
(22) Septak, M. Nucleic Acids Res. 1996, 24, 3053-3058.
(23) McGall, G.; Labadie, J.; Brock, P.; Wallraff, G.; Nguyen, T.; Hinsberg,
W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13555-13560.
(24) Temsamani, J.; Kubert, M.; Agrawal, S. Nucleic Acids Res. 1995, 23, 1841-
1844.
(15) Ferretti, L.; Karnik, S. S.; Khorana, H. G.; Nassal, M.; Oprian, D. D. Proc.
Natl. Acad. Sci. U.S.A. 1986, 83, 599-603.
(16) Yu, H.; Kono, M.; McKee, T. D.; Oprian, D. D. Biochemistry 1995, 34,
14963-14969.
(17) Schaller, H.; Weimann, G.; Lorch, B.; Khorana, H. G. J. Am. Chem. Soc.
1963, 85, 3821-3827.
(25) Gao, X.; LeProust, E.; Zhang, H.; Srivannavit, O.; Gulari, E.; Yu, P.;
Nishiguchi, C.; Xiang, Q.; Zhou, Z. Nucleic Acids Res. 2001, 29, 4744-
4750.
(18) Efcavitch, J. W.; Heiner, C. Nucleosides Nucleotides 1985, 4, 267.
(19) Caruthers, M. H. Acc. Chem. Res. 1991, 24, 278-284.
(20) Fritz, H. J.; Belagaje, R.; Brown, E.; Fritz, R. H.; Jones, R.; Lees, R. G.;
Khorana, H. G. Biochemistry 1978, 17, 1257-1267.
(26) Serafinowski, P. J.; Garland, P. B. J. Am. Chem. Soc. 2003, 125, 962-
965.
9
13428 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
Table 1. 5′-Carbonate Protected 2′-Deoxythymidine Derivativesa
molecular weight
measured (ESI-MS)
R
yield %
calculated
2:4-chlorophenyl
3:9-fluorenylmethyl
4:2-nitrophenyl
5:4-(phenylazo)phenyl
6:phenyl
7:3-(trifluoromethyl)phenyl
8:2-chloroethyl
9:1-chloroethyl
10:4-nitrophenyl
11:4-fluorophenyl
12:adamantyl
13:2-chlorophenyl
14:2,2,2-trichloroethyl
15:2-methyl-1-chloropropyl
88
90
35
50
60
70
70
54
60
77
42
64
76
59
396
464
407
466
362
431
348
348
407
380
420
396
416
376
397 (M + 1), 419 (M + Na), 431 (M + Cl)
465 (M + 1)
408 (M + 1)
467 (M + 1)
363 (M + 1)
431 (M + 1)
349 (M + 1), 371 (M + Na), 383 (M + Cl)
349 (M + 1)
408 (M + 1), 430 (M + Na), 442 (M + Cl)
381 (M + 1), 403 (M + Na)
421 (M + 1), 443 (M + Na)
397 (M + 1)
417 (M + 1), 439 (M + Na)
377 (M + 1)
a See Scheme 2 for a definition of R.
Scheme 2
the use of aqueous peroxy anions buffered under mildly basic
conditions (pH < 10) to remove an aryloxycarbonyl (ARCO)
group which is substituted for DMT in the synthesis cycle. This
peroxy anion solution is both strongly nucleophilic and mildly
oxidizing. The strong nucleophilicity of this reagent permits
irreversible removal of the 5′-O-carbonate while its mild
oxidative potential quantitatively oxidizes the internucleotide
phosphite triester without detectable oxidation of the heterocyclic
bases. This combination of desirable chemical properties into
one reagent generates a new, two-step cycle which uses fewer
solvents and eliminates acid-catalyzed detritylation from the
synthesis procedures. As a consequence, depurination and
reversibility of the detritylation step cease to exist as problems
in DNA synthesis.
Table 2. Reaction Times for Complete Removal of Carbonate
Protecting Groups with Peroxy Anion Buffersa
carbonate (R)
A
B
C
D
E
2:4-chlorophenyl
3:Fmol
4:2-nitrophenyl
5:PAP
6:phenyl
<1 min <1 min <1 min <1 min >12 h
>60 min
-
>3 h
-
-
-
-
<1 min <1 min <1 min
<1 min <1 min <1 min <1 min >12 h
<1 min <1 min <1 min <2 min >12 h
Results
The basis for this new, two-step approach is to use oxygen
nucleophiles that have positive deviations from the Brønsted-
type nucleophilicity plot. These molecules are said to exhibit
an R-effectsa term describing a nucleophile alpha to an atom
having a lone pair of electrons.27,28 The most extensive class of
R-effect nucleophiles are peroxy anions that include perborates,
tert-butylperoxide (t-BuOOH), m-chloroperbenzoic acid (MCP-
BA), and various hydrogen peroxide salts. Our objective is to
take advantage of this increased nucleophilicity to deprotect
labile carbonates with hydroperoxides that have dissociation
constants within the pH range 7-12.
To identify an appropriate carbonate having the desired
chemical properties for DNA synthesis, a series of alkyl and
aryl 5′-O-carbonates (2-15) of 2′-deoxythymidine were pre-
pared from the corresponding chloroformates and 2′-deoxy-
thymidine (Scheme 2). In all cases, the best yield for the 5′-
protected 2′-deoxynucleoside (Table 1) was obtained from
reactions carried out in pyridine using a slight excess of the
chloroformate (1.05-1.1 equiv). Under these conditions, regi-
oselectivity for the 5′-hydroxyl was observed with most chlo-
roformates.29 However, the isolated yields varied greatly and
were less than for the protection of 2′-deoxynucleosides with
DMT. Letsinger30 and Tittensor31 previously reported lower
a Buffers: A ) 3.1% LiOH/H2O (10 mL), 1.5 M 2-amino-2-methyl-1-
propanol pH 10.3 (15 mL), dioxane (50 mL), 30% H2O2 (12 mL), pH 12.0.
B ) As for A except substitute DMSO for dioxane, pH 12.0. C ) As for
A with MCPBA (1.78 g), pH 9.6. D ) H2O (10 mL), dioxane (50 mL), 2.5
M Tris (15 mL), 30% H2O2 (12 mL), MCPBA (1.78 g), pH 9.0. E ) H2O
(10 mL), dioxane (50 mL), 2.5 M Tris (15 mL), t-BuOOH (0.1 M), pH
9.0.
yields in the synthesis of 2′-deoxynucleoside 5′-carbonates.
These reduced yields were caused by the formation of 2′-
deoxynucleoside alkyl or aryl carbonates followed by intramo-
lecular elimination to form a 2′-deoxynucleoside 3′,5′-cyclic
carbonate. The better leaving groups led to greater cyclic
carbonate formation. Our results generally support these conclu-
sions.
A series of initial deprotection studies (Scheme 3) were
performed on selected carbonates (2-6) to assess their suscep-
tibility to removal with peroxy anion. The protected 2′-
deoxynucleosides were dissolved in various buffered peroxy
anion solutions (Table 2, A-E) and cleavage rates for removal
of the carbonate from 2′-deoxythymidine were monitored by
TLC. As expected, the carbonate cleavage activity of these
buffers was rapid only under pH conditions above the pKa for
peroxy anion formation. For example solutions at pH 9.6 (buffer
C) and pH 9.0 (buffer D), where MCPBA was completely
ionized, effectively cleaved four of the ARCO groups studied
(2, 4, 5, 6). Conversely tert-butylperoxide (buffer E) failed to
remove ARCO groups under reasonably satisfactory reaction
times (5-10 min). Among the carbonates that were rapidly
cleaved by C and D, compound 2 was selected for further study
(27) McIsaac, J. E., Jr.; Subbaraman, L. R.; Subbaraman, J.; Mulhausen, H. A.;
Behrman, E. J. J. Org. Chem. 1972, 37, 1037-1041.
(28) Edwards, J. O.; Pearson, R. G. J. Am. Chem. Soc. 1962, 84, 16-24.
(29) Balgobin, N.; Josephson, S.; Chattopadhyaya, J. B. Tetrahedron Lett. 1981,
22, 3667-3670.
(30) Letsinger, R. L.; Ogilvie, K. K. J. Org. Chem. 1967, 32, 296-300.
(31) Tittensor, J. R. J. Chem. Soc. (C) 1971, 2656-2662.
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13429

A R T I C L E S
Sierzchala et al.
Scheme 3
Scheme 4
as the synthesis yield was relatively high (Table 1, 88%).
Furthermore, compound 5 proved unsatisfactory as a stable side
product was generated under these cleavage conditions (perhaps
the N-oxide of the carbonate).
The next step in assessing the utility of this chemistry was
to prepare the 5′-(4-chlorophenoxycarbonyl) protected 3′-
phosphoramidite of 2′-deoxythymidine (Scheme 4, 16) and test
this synthon for the preparation of oligodeoxythymidine. Initially
an oligodeoxythymidylate tetramer was prepared on CPG using
16 and an automated DNA synthesizer. The standard cycle as
shown in Scheme 1 was modified to use deprotection mixture
D for 7 min in place of the 3% trichloroacetic acid (TCA)
solution. The resulting product was compared to the same
tetramer synthesized using standard 5′-DMT protected 2′-
deoxythymidine-3′-phosphoramidite. These oligomers were
shown to have identical yield and purity by anion-exchange
HPLC and ESI-MS. Because previous research had demon-
strated that MCPBA was an effective oxidant for ODN
synthesis,34 the peroxy anion deprotection solution was tested
for oxidation of internucleotide phosphite triester linkages. This
was accomplished by using 16, deleting the iodine oxidation
step and substituting peroxy anion solution D for TCA (Scheme
1). The results demonstrated that concomitant deprotection and
oxidation produced a tetramer of identical yield and purity
Although extremely basic conditions (buffers A and B with
ionized hydrogen peroxide, pH 12) rapidly removed carbonates,
these reagents were eliminated from further consideration for
several reasons. One was instability of the linkage joining
oligodeoxynucleotide to controlled pore glass (CPG). When 5′-
O-dimethoxytrityl-2′-deoxythymidine attached to CPG through
the standard succinate ester4 was treated with buffers A and B,
complete cleavage of the 2′-deoxynucleoside from the support
was observed in 20 min (analyzed by measuring spectropho-
tometrically at 498 nm the residual dimethoxytrityl carbocation
released from CPG upon treatment with toluenesulfonic acid).
In contrast less than 5% cleavage was found with buffers C
and D after 210 min. These stability studies suggested that
minimal loss of growing oligomers from the support was
possible with peroxy anion reagents at pH 9-10. Another
observation (TLC analysis) was that the standard amides used
to protect 2′-deoxynucleoside exocyclic amines4 were rapidly
removed under the conditions of buffers A and B. This was
undesirable as phosphitylation of these bases then occurs during
oligodeoxynucleotide synthesis. Finally oxidation of unprotected
bases and thymine was expected to be a problem at pH 12.32,33
(32) Subbaraman, L. R.; Subbaraman, J.; Behrman, E. J. J. Org. Chem. 1971,
36, 1256-1259.
(33) Subbaraman, L. R.; Subbaraman, J.; Behrman, E. J. Biochemistry 1969, 8,
3059-3066.
(34) Letsinger, R. L.; Groody, E. P.; Lander, N.; Tanaka, T. Tetrahedron 1984,
40, 137-143.
9
13430 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
Scheme 5
Table 3. Yield of dT10 as a Function of pHa
yield due to cleavage of the 3′-succinyl linkage joining
oligodeoxynucleotide to CPG. A final modification was to divide
buffer C into a two component system (see Experimental
Section). Separating lithium hydroxide from MCPBA and
mixing just prior to use allowed the deprotection reagent to
remain effective for at least 1 day. This two component mixture
was used on an automated DNA synthesizer by either premixing
the components in a single bottle or by placing these solutions
in the capping port bottles where they were combined in the
mixing chamber.
pH
% full length product
% stepwise yield
8.43
9.17
9.31
9.60
24.2
86.1
88.7
92.6
85.4
98.3
98.7
99.0
a Buffer C where the pH was adjusted with aqueous LiOH.
(anion-exchange HPLC and ESI-MS) to the same compound
synthesized using DMT protected 2′-deoxythymidine-3′-phos-
phoramidite and the standard four-step synthesis cycle.
The decomposition of MCPBA to m-chlorobenzoic acid in
the presence of LiOH35 resulted in deprotection mixture D being
useful for only a few hours. This not only reduced the effective
peroxy anion concentration but also lowered the pH which
decreased the nucleophilicity of the remaining reagent. To slow
this decomposition process, it was necessary to replace Tris with
2-amino-2-methyl-1-propanolsa buffer having a pKa near the
optimum (pH 9-10) for deprotection with MCPBA. Further
studies were then completed with this new reagent to maximize
yield as a function of pH (aqueous lithium hydroxide was used
to vary the pH).
In this experiment, 2′-deoxythymidine decanucleotides (dT10)
were prepared on CPG from 16 and various deprotection buffers
using a two-step synthesis cycle (Scheme 5). The first step was
condensation of 16 with 2′-deoxythymidine linked covalently
to CPG. After appropriate washes, the peroxy anion solution
as the second step in this cycle was used to remove the 5′-
ARCO group and oxidize the internucleotide linkage to phos-
phate. The overall yield of dT10, which measures the ability of
each buffer at variabale pH to remove the ARCO group, was
determined by collecting the products from the support under
standard conditions and evaluating the product mixture by anion
exchange HPLC (Table 3). For this buffer the highest stepwise
yields were observed at pH 9.6. These results established buffer
C (Table 2) as the preferred peroxy anion solution which was
used in subsequent experiments. An increase above pH 9.6 did
not lead to further improvements but actually reduced the overall
Following these preliminary experiments, an oligodeoxythy-
midine 12mer was synthesized via the approach outlined in
Scheme 5. Synthesis begins with 2′-deoxythymidine linked
covalently to CPG via standard procedures. The two-step cycle
involved condensation of 16 with 2′-deoxythymidine in the
presence of tetrazole followed by treatment for 2.5 min with
the peroxy anion solution (buffer C) which oxidized the
internucleotide linkage and removed the 4-chlorophenoxycar-
bonate protecting group. Further repetitions of this cycle
followed by cleavage from the support and HPLC analysis of
the total reaction mixture generated the profile shown in Figure
1, Panel A. When compared to dT12 prepared by the standard
four-step cycle (Figure 1, Panel B), there were numerous failure
sequences (oligomers less than 12 in length corresponding to
19.6% of the total UV absorbing material). To reduce the
amount of these short oligomers, carbonates 7-15 (Table 1)
were converted to their corresponding 3′-diisopropylphosphora-
midites and tested for the preparation of dT12. Generally these
carbonates were selected because they were expected to be more
labile than 4-chlorophenoxycarbonate toward R-effect nucleo-
philes. The resulting crude reaction mixtures were analyzed by
HPLC. A comparison showed that the 3-trifluoromethylphenoxy
derivative (compound 7 or 18a) generated dT12 having the least
amount (13.7% of the total UV absorbing material) of failure
sequences (Figure 1, Panel C). When analyzed by MALDI-TOF
spectrometry, this product was found to have the mass (Table
4) expected of d(T12). There were no higher-molecular weight
peaks corresponding to products having oxidized thymine. On
the basis of these results, further experiments were then directed
(35) Ball, D. L.; Edwards, J. O. J. Am. Chem. Soc. 1956, 78, 1125-1129.
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13431

A R T I C L E S
Sierzchala et al.
Figure 1. Reverse phase HPLC profiles from total reaction mixtures of dT12 prepared on CPG. (A) Two-step cycle with 16; (B) four-step cycle with
5′-DMT-2′-deoxythymidine-3′-phosphoramidite; (C) two-step cycle with 19a.
ycarbonyl)-2′-deoxycytidine (17b), and N2-isobutyryl-O6-(N,N-
Table 4. MALDI-TOF Analysis of Oligodeoxynucleotides
diphenyl)carbamoyl-2′-deoxyguanosine (17c) were visually
molecular weight
estimated to be 30 min (17a), 12 h (17b), and 24 h (17c). Other
ODN
calculated
measured (MALDI-TOF)
commonly used amide protecting groups on cytosine (acetyl
and benzoyl) were even less stable. After 24 h exposure to buffer
C, 5′-O-DMT protected 2′-deoxynucleosides were extracted
from the aqueous mixture with DCM and analyzed by FAB
mass spectrometry. Parent ions for the heterocyclic base
protected 17b and 17c and corresponding heterocyclic base
deprotected 5′-DMT 2′-deoxynucleosides were easily detected.
For 17a, only the heterocyclic base deprotected 5′-DMT 2′-
deoxynucleoside was detected. The mass spectra did not contain
unexplained parent ions such as those corresponding to oxidative
side products of the heterocyclic bases. When analyzed under
the same conditions, the parent ion for 5′-O-DMT-2′-deoxythy-
midine was readily observed without detectable ions corre-
sponding to oxidized side products.
d(TTTTTTTTTTTT)
d(CCCCCCCCCT)
d(AAAAAAAAAT)
d(GTGTGTGTGT)
d(CTCTCTCTCT)
d(ATATATATAT)
d(ATGTCAACTCGTCT)
3059.6
2843.5
3059.6
3103.5
2903.5
3023.6
4211.7
3057.7 (M - 1), 3060.2 (M + 1)
2842.3 (M - 1), 2844.5 (M + 1)
3057.7 (M - 1), 3060.2 (M + 1)
3101.9 (M - 1), 3104.3 (M + 1)
2902.3 (M - 1), 2904.4 (M + 1)
3022.7 (M - 1), 3024.7 (M + 1)
4210.8 (M - 1), 4212.8 (M + 1)
toward the synthesis of oligodeoxynucleotides having all four
standard bases.
To investigate the stability of heterocyclic base protecting
groups toward the peroxy anion solution, 5′-O-DMT and base
protected 2′-deoxynucleosides were incubated at room temper-
ature with a large excess of buffer C and monitored by TLC to
evaluate the extent of protecting group cleavage as a function
of time. These stability half-lives for the 5′-O-DMT derivatives
of N6-benzoyl-2′-deoxyadenosine (17a), N4-(9-fluorenylmethox-
On the basis of these observations, the two-step synthesis
cycle (Scheme 5) was used to prepare homopolymers of 2′-
9
13432 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
Scheme 6
Table 5. Summary of Characterization Data for 5′-ARCO-2′-deoxynucleosides and Corresponding 3′-Phosphoramidites
molecular weight
compound
yield %
31P NMR (ppm)
calculated
ESI-MS
HRMS (FAB)
18a
18b
18c
18d
18e
18f
77
68
69
73
69
60
80
76
45
50
61
65
74
70
431.1066 (M + H)
520.1332 (M + H)
638.1750 (M + H)
717.2298 (M + )
544.1444 (M + H)
740.2332 (M - H)
721.2234 (M + H)
631.2145 (M + H)
720.2410 (M + H)
838.2829 (M + H)
917.3377 (M + )
744.2522 (M + H)
941.3489 (M + )
921.3312 (M + H)
431 (M + 1)
520 (M + 1)
638 (M + 1)
718 (M + 1)
544 (M + 1)
742 (M + 1)
721 (M + 1)
631 (M + 1)
720 (M + 1)
838 (M + 1)
918 (M + 1)
744 (M + 1)
942 (M + 1)
921 (M + 1)
431.1086
520.1315
638.1759
717.2321
544.1456
740.2331
721.2218
631.2141
720.2435
838.2808
917.3346
744.2547
941.3510
921.3353
18g
19a
19b
19c
19d
19e
19f
149.44
149.73, 149.44
149.70, 149.40
149.13, 149.03
149.19, 149.16
149.07, 148.98
149.48, 149.37
19g
deoxycytidine (dC12) and 2′-deoxyadenosine (dA12) from 19b
or 19c and 19e, respectively (Scheme 4 and Table 5). Problems,
however, were observed. Specifically, the HPLC profiles were
found to be heterologous with numerous uncharacterized side
products. These side products appeared to be due to removal
of base protecting groups by the peroxy anion solution followed
by phosphitylation of unprotected cytosine and adenine bases.
This conclusion was supported by the observation that higher-
quality oligomers were obtained when a pyridine hydrochloride/
aniline wash (a solution known to reduce phosphitylation of
unprotected bases36) was introduced into this cycle immediately
following phosphitylation but prior to treatment with peroxy
anion. Additionally when oligomers generated from 19c were
deprotected with concentrated ammonium hydroxide, a signifi-
cant amount of side product with a urea modification was
characterized by LC-mass spectrometry. Similar results were
obtained when oligomers were synthesized by the standard four-
step cycle using 5′-DMT-N4-Fmoc-dC-3′-phosphoramidite. These
results suggest that ammonia deprotection is not compatible with
N4-fluorenylmethoxycarbonyl protection of 2′-deoxycytidine.
This is because ammonia attack on the protecting group leads
not only to the preferred cleavage as dibenzofulvene and carbon
dioxide by â-elimination but also, via nucleophilic attack on
the carbonyl, elimination of 9-fluorenemethanol and generation
of the stable N4-ureido derivative (Scheme 6). This observation
means that an additional non-nucleophilic base deprotection step
would be required to use 19c for oligodeoxynucleotide synthesis.
To overcome the problem of amide protecting group instabil-
ity with peroxy anion (cytosine and adenine), other types of
blocking groups were explored. Among those tested, the most
satisfactory results were obtained by using acid labile N-trityl
groups as these derivatives were completely stable toward
peroxy anion solutions.37 After many preliminary experiments,
compounds 18d and 18f proved to be the most useful as the
N-dimethoxytrityl groups could be removed in 2 h (18d) and
15 min (18f) using 3% TCA in dichloromethane (no base
modification). The N-dimethoxytrityl 2′-deoxynucleosides were
prepared from the appropriate 5′,3′-O-(tetraisopropyldisiloxane-
1,3-diyl)-2′-deoxynucleosides by treatment with dimethoxytrityl
chloride followed by removal of the silyl protecting group with
fluoride ion. Alternatively they can be prepared by the transient,
Jones silylation procedure.40 These N-protected 2′-deoxynucleo-
sides were converted first to the 5′-O-[3-(trifluoromethyl)-
phenoxycarbonyl]-2′-deoxynucleosides (18d, 18f) and then,
using standard procedures,41 to the 2′-deoxynucleoside-3′-
phosphoramidites (19d, 19f). Table 5 summarizes yields and
characterization data for these synthons.
Using the two-step cycle, 2′-deoxynucleoside 3′-phosphora-
midites 19d and 19f were tested for synthesis of oligodeoxy-
nucleotides. In each case these synthons were joined repetitively
to 2′-deoxythymidine linked to CPG through the 3′-hydroxyl
to yield essentially homopolymers of 2′-deoxycytidine and 2′-
deoxyadenosine. This strategy is the most rigorous test of any
new synthon as problems with the chemistry will be amplified
either during synthesis or deprotection. For example, when
homopolymers of deoxyadenosine were prepared via the four-
(37) Although N-trityl protection is common in the peptide field,38 it has only
been used sparingly in nucleic acid synthesis.39
(38) Gross, E.; Meinhofer, J. The Peptides: Analysis, Synthesis, Biology;
Protection of Functional Groups in Peptide Synthesis, Vol. 3; Academic
Press: New York, 1987.
(39) Schaller, H.; Khorana, H. G. J. Am. Chem. Soc. 1963, 85, 3828-3835.
(40) Ti, G. S.; Gaffney, B. L.; Jones, R. A. J. Am. Chem. Soc. 1982, 104, 1316-
1319.
(41) Caruthers, M. H.; Barone, A. D.; Beaucage, S. L.; Dodds, D. R.; Fisher,
E. F.; McBride, L. J.; Matteucci, M.; Stabinsky, Z.; Tang, J.-Y. Methods
Enzymol. 1987, 154, 287-313.
(36) Gryaznov, S. M.; Letsinger, R. L. J. Am. Chem. Soc. 1991, 113, 5876-
5877.
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13433

A R T I C L E S
Sierzchala et al.
Figure 2. Reverse phase HPLC profiles from total reaction mixtures prepared on CPG. (A) synthesis of d(C9T) via the two-step cycle using 19d; (B)
synthesis of d(A9T) via the two-step cycle using 19f.
step cycle, HPLC profiles clearly showed a series of peaks
corresponding to depurinated oligomers.42 Similarly when the
standard thiolate deprotection step was omitted during workup
of deoxythymidine homopolymers,43 a series of N-methyl
thymidine derivatives were easily detected via HPLC. The
results presented in Figure 2 indicate that N-dimethoxytrityl
protection on 2′-deoxycytidine and 2′-deoxyadenosine can be
used in this two-step cycle to prepare homopolymers lacking
unexpected side products. For example when d(C9T) was
prepared via this two-step cycle and the total reaction mixture
analyzed by reverse phase HPLC (Panel A), one major product
peak was observed. There were minor product peaks that eluted
both before (failure sequences) and after d(C9T). When analyzed
by MALDI-TOF mass spectrometry (Table 4), the measured
molecular weight of d(C9T) corresponded to the calculated mass,
and there were no peaks having increased mass which would
be expected for oxidized oligodeoxynucleotides. For 2′-deoxy-
adenosine, the use of 19f in the two-step cycle for synthesis of
d(A9T) also generated one major product peak (Panel B) when
the total reaction mixture was analyzed by HPLC. Analysis by
MALDI-TOF mass spectrometry confirmed that this oligode-
oxynucleotide had the molecular weight as calculated for the
expected product (Table 4). As observed with d(C9T), there were
no higher molecular weight mass peaks in this sample. These
results confirmed that oxidation of the adenine base under these
synthesis conditions was not a problem. There was, however,
more material present in the failure sequence portion of the
chromatography pattern than observed during the synthesis of
d(C9T). Recent research has demonstrated that this effect can
be correlated to a slower rate of 5′-carbonate cleavage via peroxy
anion due to steric hindrance associated with the N-dimethox-
ytrityl group. Additionally, trace amounts of material still
retaining the dimethoxytrityl protecting group were present in
both chromatography profiles as can be seen by the series of
small peaks following elution of products.
Analysis of 19g in the two-step synthesis cycle was somewhat
complicated because homopolymers of 2′-deoxyguanosine tend
to aggregate when analyzed by HPLC. As a consequence this
synthon was tested by preparing d(GT)5 where the resulting
oligomer could be readily characterized by HPLC and compared
with results obtained with d(T)12, d(AT)5 and d(CT)5. As shown
in Figure 3, Panel A, d(GT)5 was the major product obtained
from this synthesis (MALDI-TOF analysis, Table 4). The
molecular mass was consistent with a product that had not been
oxidized at any of the thymine or guanine bases. Failure
sequences and trace amounts of incompletely deprotected
oligomers were also present. The peak eluted late from the
column was the protecting group product (N,N-diphenylcar-
bamate) obtained from cleavage with ammonium hydroxide.
Oligomers synthesized as controls, d(CT)5 and d(AT)5, generated
HPLC profiles (Figure 3, Panels B and C) similar to those
observed for d(C)9T and d(A)9T (Figure 2). Characterization
of d(CT)5 and d(AT)5 by MALDI-TOF mass spectrometry gave
the expected results (Table 4) with no indication of base
oxidation.
To test all four synthons simultaneously (19a, 19d, 19f,
and 19g), mixed sequence oligomers were prepared. The HPLC
profile of the total reaction mixture for one example is
shown in Figure 3, Panel D. Again the major peak corresponds
(42) Scaringe, S. A. Design and Development of New Protecting Groups for
RNA Synthesis. Ph.D. Thesis, 1996.
(43) Urdea, M. S.; Ku, L.; Horn, T.; Gee, Y. G.; Warner, B. D. Nucleic Acids
Symp. Ser. 1984, 16, 257-260.
9
13434 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
Figure 3. Reverse phase HPLC profiles from total reaction mixtures prepared on CPG. (A) synthesis of d(GT)5 via the two-step cycle using 19a and 19g;
(B) synthesis of d(CT)5 via the two-step cycle using 19a and 19d; (C) synthesis of d(AT)5 via the two-step cycle using 19a and 19f; (D) synthesis of
d(ATGTCAACTCGTCT) via the two-step cycle using 19a, 19d, 19f, and 19g.
to product having no modified bases as characterized by
MALDI-TOF (Table 4). There were also failure sequences and
trace amounts of oligonucleotidic material eluting imme-
diately after the product peak. Late in the elution profile a
peak corresponding to the N,N-diphenylcarbamate was also
present.
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13435

A R T I C L E S
Sierzchala et al.
Although 2′-deoxynucleoside oxidation via peroxy anion
deprotection was not detected by mass spectrometry, further
experiments designed to test for oxidation were carried out by
HPLC analysis of enzymatically degraded oligomers. Oligode-
oxynucleotides as isolated from the major peaks shown in the
HPLC profiles reproduced in Figures 2 and 3 (Panels A and D)
were treated with snake venom phosphodiesterase and alkaline
phosphatase followed by HPLC analysis using 2′-deoxynucleo-
side standards. The results as shown in Figure 4 confirm that
oxidation has not occurred as only unmodified 2′-deoxynucleo-
sides were obtained.
only needed to block the further growth of failure sequences
and can thus be eliminated for many applications as the
repetitive coupling yields with 2′-deoxynucleoside phos-
phoramidite synthons is so high (>99%).44 This is especially
the case for syntheses with 2′-deoxythymidine and 2′-deoxy-
cytidine as the repetitive yields and the quality of synthetic DNA
is essentially identical to results obtained using the four-step
cycle (compare Figure 1, Panel B to Figure 1, Panel C and
Figure 2, Panel A). With capping in the four-step cycle (Figure
1, Panel B), the small amount of unreacted oligomer per cycle
ceases to grow to longer DNA and thus the total reaction mix-
ture appears remarkably clean. Without capping (Figure 1,
Panel C and Figure 2, Panel A) in the two-step cycle, these
minute amounts of unreacted oligomers continue to grow and
generate visible, longer-length failure sequences. For syn-
theses with protected purines, a considerably larger amount of
material is present in the failure sequence regions of HPLC
profiles (Figure 2, Panel B and Figure 3, Panel A). Unpublished
results clearly demonstrate that these lower repetitive yields are
due to a slower rate of cleavage of the 5′-carbonate, a con-
sequence of the large steric bulk of the dimethoxytrityl and
N,N-diphenylcarbamoyl groups (adenine and guanine, respec-
tively). Complete removal of carbonates must be achieved,
otherwise this two-step cycle will generate a series of mutated
sequences similar in composition to those from the reversi-
bility of carbocation cleavage. Current research is focused on
testing less bulky, acid labile protecting groups such as
amidines.45-47
To further confirm the integrity of the oligodeoxynucleo-
tides generated using this two-step cycle, additional biochem-
ical and biophysical studies have been completed. For example
when the deoxyoligonucleotide whose purification is shown
in Figure 3, Panel D, d(ATGTCAACTCGTCT, forms duplexes
with complementary DNA and RNA, the melting tempera-
tures of the duplexes (45.1 and 40.1 °C, respectively) cor-
respond to those found using a sample of the same oligomer
synthesized by the four-step cycle. Similarly when the hetero-
duplex of this oligomer and complementary RNA is treated
with RNase H1, the RNA degradation profile is identical to
results obtained using DNA generated by the four-step cycle
(Figure 5). These results demonstrate that ODNs synthesized
using the two-step cycle have biochemical and biophysical
properties analogous to results obtained using the four-step
cycle.
Discussion
Several additional features of this new cycle require comment.
By introducing N-dimethoxytrityl blocking groups on cytosine
and adenine, an acid step must be used during deprotection.
Although generally effective, small amounts of partially pro-
tected oligomers still remain when reaction products are
analyzed by HPLC. This was especially the case with N-
dimethoxytrityl protected cytosine (Figure 2, Panel A). Recent
research has focused, with encouraging results, on designing
more acid labile blocking groups for this base (N-trimethox-
ytrityl). Of particular interest as well was the observation that
acid removal of dimethoxytrityl groups did not lead to depu-
rination. Perhaps this is because the product of detritylation is
2′-deoxyadenosine which is much more stable toward acid
depurination than the amide protected derivative. The latter,
which is present throughout synthesis in the standard four-step
cycle, must survive multiple rounds of acid exposure. Finally
the use of peroxy anion does not lead to oxidized bases. This
was shown by mass spectrometry analysis of synthetic ODNs
and the complete degradation of these oligomers to the expected,
unoxidized 2′-deoxynucleosides.
In conclusion we have developed an alternative strategy for
the preparation of oligodeoxynucleotides. The approach utilizes
an R-effect nucleophile (peroxy anion) during each synthetic
cycle to simultaneously remove a 5′-carbonate and oxidize the
internucleotide phosphite triester (Scheme 5). Relative to the
standard approach (Scheme 1), the introduction of this reagent
plus elimination of capping leads to a new two-step DNA
synthesis cycle that will be useful for all major applications of
synthetic DNA. There are several key features that makes this
alternative cycle superior for cloning synthetic DNA, the large-
scale preparation of DNA, and the synthesis of highly parallel,
microscale DNA arrays. First the cyclical removal of the 5′-
protecting group with peroxy anion under mildly basic condi-
tions is essentially nonreversible and quantitative. This procedure
therefore completely eliminates depurination and one series of
failure sequence which are the side reactions most responsible
for the high mutation frequencies in cloned, synthetic DNA and
the reduced quality of ODNs synthesized in large scale or on
micro-arrays. Second the two-step procedure streamlines ODN
synthesis by eliminating several reagents. This should result in
dramatic cost savings for the large scale synthesis of ODNs.
For highly parallel, microscale DNA synthesis, this streamlined
procedure allows for simpler and potentially more robust
automation.
Although these results encourage us to further develop this
two-step cycle, several key goals must still be attained. These
include examining the mutation frequency of cloned, synthetic
DNA prepared by this procedure and testing the approach in
large scale oligomer synthesis. Of particular interest is recent
research on glass slides where we have successfully prepared
dT25 and shown it to be the only significant product peak when
Several chemical steps leading to this new two-step cycle
require further comment. One is elimination of the capping step.
Removal of this procedure is possible because phosphite adducts
on protected bases [N-dimethoxytrityl and O6-(N,N-diphenyl)-
carbamoyl] do not appear to be a problem. Presumably this is
because protection of this type leads to bases that are unreactive
toward phosphitylation. As a consequence, the capping step is
(44) A capping step can always be added for specific applications.
(45) McBride, L. J.; Caruthers, M. H. Tetrahedron Lett. 1983, 24, 2953-2956.
(46) Froehler, B. C.; Matteucci, M. D. Nucleic Acids Res. 1983, 11, 8031-
8036.
(47) McBride, L. J.; Kierzek, R.; Beaucage, S. L.; Caruthers, M. H. J. Am. Chem.
Soc. 1986, 108, 2040-2048.
9
13436 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
Figure 4. Reverse Phase HPLC profiles following treatment of ODNs with snake venom phosphodiesterase and alkaline phosphatase. (A) d(C9T); (B)
d(A9T); (C) d(GT)5; (D) d(ATGTCAACTCGTCT).
Experimental Section
analyzed by HPLC. Current work is directed toward extending
this research on glass slides to the synthesis of ODNs having
all four bases.
General Procedures. Unless otherwise noted, materials were
obtained from commercial suppliers and used without further purifica-
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13437

A R T I C L E S
Sierzchala et al.
at a rate of 1.0 °C/min. Oligodeoxynucleotide samples prepared by the
two-step cycle were separately mixed with complementary RNA and
DNA in a 1 mL cuvette (1 µM duplex, 0.1 M KCl, 0.1 M KH2PO4, pH
7.0) and the Tm’s determined as the maximum of the first derivative of
the melting curve. Control experiments were performed using DNA
prepared by the four-step cycle.
Peroxy Anion Buffer. The peroxy anion buffer was prepared as
two components. Solution A was 3% (w/v) aqueous LiOH (10 mL),
1.5 M 2-amino-2-methyl-1-propanol in water (15 mL), and dioxane
(17.5 mL). Solution B was m-chloroperbenzoic acid (1.78 g), aqueous
30% H2O2 (10 mL), and dioxane (32.5 mL). Equal volumes of solutions
A and B were mixed just prior to DNA synthesis and placed on port
10 of an ABI/PE Biosystems model 394 DNA synthesizer. A fresh
mixture of these two solutions was prepared daily.
Synthesis of 5′,3′-O-(Tetraisopropyldisiloxane-1,3-diyl)-2′-deoxy-
nucleosides. 2′-Deoxynucleoside (10 mmol) was coevaporated three
times with pyridine. Anhydrous pyridine (35 mL) and 1,3-dichloro-
1,1,3,3-tetraisopropyldisiloxane (3.47 g, 11 mmol) were added, and the
mixture was stirred overnight at room temperature. The solution was
concentrated and chromatographed on silica gel with 6% methanol in
CHCl3.
5′,3′-O-(TIPS)-2′-deoxycytidine: yield 99.9%; TLC Rf (A) 0.30; 1H
NMR (DMSO-d6) δ 7.63 (dd, 1H), 7.17 (bd, 2H), 5.97 (t, 1H), 5.69
(dd, 1H), 4.48-4.44 (m, 1H), 4.04-3.91 (m, 2H), 3.73-3.70 (m, 1H),
2.36-2.18 (m, 2H), 1.07-0.90 (m, 28H); 13C NMR (DMSO-d6) δ
165.73, 154.88, 140.49, 93.84, 84.18, 83.87, 69.36, 61.37, 17.30, 17.16,
17.12, 17.08, 16.91, 16.77, 16.68, 12.79, 12.54, 12.30, 11.98; HRMS
(FAB) calcd for C21H40N3O5Si2. (M + H) 470.2507, found 470.2486.
5′,3′-O-(TIPS)-2′-deoxyadenosine: yield 92%; TLC Rf (A) 0.43;
1H NMR (DMSO-d6) δ 8.24 (s, 1H), 8.09 (s, 1H), 7.45 (bs, 2H), 6.29-
6.27 (m, 1H), 5.18-5.16 (m, 1H), 3.90-3.75 (m, 3H), 2.87-2.54 (m,
2H), 1.11-0.97 (m, 28H); 13C NMR (DMSO-d6) δ 156.15, 152.31,
148.48, 139.89, 119.43, 84.34, 82.17, 71.58, 62.66, 38.98, 17.39, 17.23,
17.21, 17.17, 17.07, 16.92, 16.86, 16.81, 12.71, 12.55, 12.16, 12.02;
HRMS (FAB) calcd for C22H40N5O4Si2 (M + H) 494.2619, found
494.2612.
Synthesis of N-Dimethoxytrityl-2′-deoxynucleosides. 5′,3′-O-(Tet-
raisopropyldisiloxane-1,3-diyl)-2′-deoxynucleoside (10 mmol) was co-
evaporated three times with pyridine and then dried in vacuo for 12 h.
Anhydrous pyridine (30 mL) and dimethoxytrityl chloride (3.6 g, 11
mmol) were added The mixture was stirred at room temperature until
TLC (Solvent A) showed complete disappearance of the 2′-deoxy-
nucleoside substrate (16-24 h). The reaction was quenched with water/
ice. Crude product was extracted with DCM, washed with a 5% aqueous
solution of NaHCO3, and dried with anhydrous Na2SO4. After filtration
the organic layer was concentrated and dried in vacuo for 3 h. Hydrogen
fluorideaq (1.4 mL, 35 mmol) was carefully added with vigorous stirring
to an ice-cold solution of TEMED (7.5 mL, 50 mmol) in acetonitrile
(20 mL). The TEMED-HF reagent so formed was then transferred via
Teflon tubing to a flask containing crude 5′,3′-O-(tetraisopropyldisi-
loxane-1,3-diyl)-N-dimethoxytrityl protected deoxynucleoside (10 mmol),
and the mixture was stirred for 2.5 h. The solution was concentrated
in vacuo and the residue coevaporated sequentially with pyridine,
toluene, and ethanol. The crude product was purified by column
chromatography using CHCl3/pyridine (99.9:0.1) with a gradient of
methanol (0-8%).
Figure 5. E. coli RNase H1 degradation of RNA. 5′-32P-labeled-RNA was
allowed to form duplexes with complementary DNAs, d(ATGT-
CAACTCGTCT), which were then treated with E. coli RNase H1 for 1 h
and analyzed by PAGE. Lane 1: four-step cycle synthesized DNA, RNA;
Lane 2: four-step cycle synthesized DNA, RNA, enzyme; Lane 3: two-
step cycle synthesized DNA, RNA; Lane 4: two-step cycle synthesized
DNA, RNA, enzyme.
tion. DCM and pyridine were distilled from calcium hydride. MCPBA
was obtained 55-83% pure from Aldrich and used without further
purification. Aqueous hydrogen fluoride (48%) was obtained from
Mallinckrodt. 2-Amino-2-methyl-1-propanol was obtained as a 1.5 M
aqueous solution (pH 10.3) from Sigma Diagnostics Inc. 2-Cyanoethyl-
N,N,N′,N′-tetraisopropylphosphorodiamidite (ChemGenes Corp.) was
purified by distillation under high vacuum. All other DNA synthesis
reagents were obtained from Glen Research Corp. Oligoribonucleotides
were obtained from Dharmacon Research Inc., Louisville, CO. Medium-
pressure, preparative column chromatography was performed using
230-400 mesh silica gel from EM Sciences. Thin-layer chromatography
was performed on aluminum-backed silica 60 F254 plates from EM
Sciences (Solvent A: 9/1 CHCl3/methanol; Solvent B: 19/1, DCM/
ethanol). Solid-phase ODN synthesis was performed on an ABI 394
automated DNA synthesizer from ABI/PE Biosystems. Reverse phase
HPLC on an ODS-Hypersil column from Agilent Technologies (50
mM aqueous TEAB, pH 8.0, and acetonitrile as eluants) and ion
exchange HPLC on a DNAPac column from Dionex (25 mM Tris
buffer/0.05% acetonitrile, pH 8.0, and 1 M NaCl as eluants) were
performed on an Agilent Technologies 1100 HPLC. NMR data were
recorded on Varian Inova-400 MHz, Varian Inova-500 MHz and Bruker
400 MHz spectrometers. Tetramethylsilane was used as an internal
1
reference for H and 13C NMR. An external capillary containing 85%
H3PO4 was used as a reference for 31P NMR. Downfield chemical shifts
were recorded as positive values for 31P NMR. MALDI-TOF mass data
was obtained on a Voyager DE-STR instrument from Perseptive
Biosystems by comparison to an internal ODN standard. ESI and FAB
mass spectroscopy was performed by the University of Colorado Central
Analytical Laboratories. Elemental analysis was performed by Huffman
Labs, Golden, CO.
N4-(9-Fluorenylmethoxycarbonyl)-2′-deoxycytidine (dCFm°c) was
synthesized according to a published procedure.48 2′-Deoxyadenosine
(dA), 2′-deoxycytidine (dC), 2′-deoxyguanosine (dG), N6-benzoyl-2′-
deoxyadenosine (dABz) and N4-acetyl-2′-deoxycytidine (dCAc), N4-
benzoyl-2′-deoxycytidine (dCBz) and N2-isobutyryl-2′-deoxyguanosine
(dGiBu) were purchased from ChemGenes Corporation.
Melting Temperatures. Melting points (Tm’s) were determined on
a Varian Cary 1E UV-visible spectrometer. The absorbance at 260
nm was measured while the temperature of the sample was increased
N4-Dimethoxytrityl-2′-deoxycytidine: yield 98%; TLC Rf (A) 0.30;
1H NMR (DMSO-d6) δ 8.36 (bs, 1H), 7.69 (d, 1H), 7.28-7.17 (m,
5H), 7.12 (d, 4H), 6.83 (d, 4H), 6.22 (d, 1H), 6.03 (t, 1H), 5.17 (d,
1H), 4.94 (t, 1H), 4.16-4.13 (m, 1H), 3.72 (s, 6H), 3.52-3.48 (m,
2H), 2.05-1.86 (m, 2H); 13C NMR (DMSO-d6) δ 163.30, 157.43,
154.01, 145.13, 139.62, 137.00, 129.92, 128.51, 127.45, 126.09, 112.72,
96.34, 87.17, 84.65, 70.63, 69.31, 61.50, 54.99; HRMS (FAB) calcd
for C30H31N3O6 (M+) 529.2213, found 529.2200.
(48) Koole, L. H.; Moody, H. M.; Broeders, N. L. H. L.; Quaedflieg, P. J. L.
M.; Kuijpers, W. H. A.; van Genderen, M. H. P.; Coenen, A. J. J. M.; van
der Wal, S.; Buck, H. M. J. Org. Chem. 1989, 54, 1657-1664.
N6-Dimethoxytrityl-2′-deoxyadenosine: yield 91%; TLC Rf (A)
1
0.36; H NMR (DMSO-d6) δ 8.42 (s, 1H), 8.32 (s, 1H), 7.28-7.26
9
13438 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
(m, 5H), 7.19 (d, 4H), 6.84 (d, 4H), 6.32 (t, 1H), 5.31 (d, 1H), 5.12 (t,
1H), 3.86-3.84 (m, 1H), 3.71 (s, 6H), 3.61-3.47 (m, 2H), 2.79-2.74
(m, 1H), 2.27-2.22 (m, 1H); 13C NMR (DMSO-d6) δ 157.69, 153.68,
151.16, 148.07, 145.34, 140.37, 137.27, 129.77, 128.39, 127.71, 126.47,
121.05, 113.00, 88.04, 84.12, 70.92, 69.61, 61.83, 54.99; HRMS (FAB)
calcd for C31H31N5O5 (M+) 553.2325, found 553.2309.
167.36, 163.20, 154.40, 152.49, 150.97, 145.12, 133.13, 132.76, 131.03,
130.47, 130.22, 128.47, 125.79, 123.09, 118.73, 96.39, 86.54, 84.01,
70.08, 68.56; HRMS (FAB) calcd for C24H21N3O7F3 (M + H) 520.1332,
found 520.1315.
Compound 18c: yield 69.4%; TLC Rf (A) 0.48; 1H NMR (DMSO-
d6) δ 11.10 (bs, 1H), 8.10 (d, 1H), 7.93-7.33 (m, 12H), 7.06 (d, 1H),
6.18 (t, 1H), 5.55 (s, 1H), 4.49-4.13 (m, 7H), 2.36-2.00 (m, 2H); 13
C
Synthesis of N-Dimethoxytrityl-2′-deoxynucleosides via the Jones
Procedure. 2′-Deoxynucleoside (25 mmol) was coevaporated three
times with pyridine and then dried in vacuo for 12 h. Anhydrous
pyridine (125 mL) and chlorotrimethylsilane (16 mL, 125 mmol) were
added. After stirring this solution for 2 h, dimethoxytrityl chloride (8.59
g, 26.25 mmol) was added and the reaction mixture stirred overnight
at room temperature. Water (150 mL) and concentrated ammonium
hydroxide (5 mL) were added, and the reaction mixture was stirred for
30 min. The crude product was extracted with dichloromethane, and
the organic phase was washed twice with a 5% aqueous solution of
sodium bicarbonate and then dried with anhydrous Na2SO4. After
filtration, the dichloromethane solution was concentrated and the product
obtained by silica gel column chromatography using CHCl3/pyridine
(99.9:0.1) with a gradient of methanol (0-8%). Yields are 98-99%.
Characterization data for the 2′-deoxyadenosine and 2′-deoxycytidine
derivatives are in the previous section.
NMR (DMSO-d6) δ 162.95, 154.23, 153.18, 152.47, 150.95, 144.84,
143.46, 140.76, 131.01, 130.46, 130.20, 127.81, 127.15, 125.76, 125.59,
123.07, 120.16, 118.71, 94.59, 86.41, 83.94, 70.09, 68.57, 66.93, 46.17,
39.96; HRMS (FAB) calcd for C32H27N3O8F3 (M + H) 638.1750, found
638.1759.
Compound 18d: yield 73.1%; TLC Rf (A) 0.43; 1H NMR (DMSO-
d6) δ 8.43 (bs, 1H), 7.77-7.61 (m, 4H), 7.53 (d, 1H), 7.29-7.21 (m,
5H), 7.14 (d, 4H), 6.84 (d, 4H), 6.27 (d, 1H), 6.12 (t, 1H), 5.45 (d,
1H), 4.43-4.22 (m, 3H), 3.97-3.94 (m, 1H), 3.73 (s, 6H), 2.09-1.95
(m, 2H); 13C NMR (DMSO-d6) δ 163.28, 157.45, 153.91, 152.51,
150.96, 145.09, 139.52, 136.94, 131.03, 130.46, 130.20, 129.93, 128.51,
127.44, 126.10, 125.77, 123.08, 118.75, 112.72, 96.77, 84.68, 83.06,
70.31, 69.38, 68.74, 54.97, 39.10; HRMS (FAB) calcd for C38H34N3O8F3
(M+) 717.2298, found 717.2321.
Compound 18e: yield 69.0%; TLC Rf (A) 0.43; 1H NMR (DMSO-
d6) δ 11.23 (bs, 1H), 8.77 (s, 1H), 8.69 (s, 1H), 8.05 (d, 2H), 7.71-
7.55 (m, 7H), 6.55 (t, 1H), 5.65 (d, 1H), 4.64-4.39 (m, 3H), 4.17 (dt,
1H), 2.99-2.94 (m, 1H), 2.49-2.43 (m, 1H); 13C NMR (DMSO-d6) δ
165.66, 152.61, 151.96, 151.70, 150.91, 150.48, 143.28, 133.36, 132.49,
130.99, 130.44, 130.18, 128.54, 128.49, 125.75, 123.06, 122.49, 118.71,
83.90, 83.73, 70.48, 68.57, 38.42; HRMS (FAB) calcd for C25H21N5O6F3
(M + H) 544.1444, found 544.1456.
Synthesis of N2-isobutyryl-O6-(N,N-diphenyl)carbamoyl-2′-deox-
yguanosine. N2-isobutyryl-2′-deoxyguanosine (10 mmol) was coevapo-
rated three times with pyridine and then dried in vacuo for 6 h.
Anhydrous pyridine (330 mL), triethylamine (13 mL), and N,N-
diphenylcarbamoyl chloride (2.32 g, 10 mmol) were added. The mixture
was stirred overnight at room temperature. Water (6 mL) was added
and the mixture evaporated under reduced pressure. The residue was
coevaporated with toluene, ethanol, and DCM and then was purified
by column chromatography using CHCl3 with a gradient of methanol
(0-5%).
Compound 18f: yield 60.6%; TLC Rf (A) 0.58; 1H NMR (DMSO-
d6) δ 8.43 (s, 1H), 7.95 (s, 1H), 7.71-7.55 (m, 4H), 7.29-7.22 (m,
5H), 7.20 (d, 4H), 6.84 (d, 4H), 6.40 (t, 1H), 5.59 (d, 1H), 4.59-4.34
(m, 3H), 4.11 (dt, 1H), 3.72 (s, 6H), 2.93-2.87(m, 1H), 2.40-2.34
(m, 1H); 13C NMR (DMSO-d6) δ 157.70, 153.66, 152.59, 151.37,
150.91, 148.18, 145.35, 140.42, 137.27, 130.97, 130.44, 129.76, 128.37,
127.69, 126.46, 125.74, 123.04, 120.99, 118.70, 112.99, 83.77, 70.50,
69.59, 68.56, 54.97, 38.26; HRMS (FAB) calcd for C39H33N5O7F3 (M
- H) 740.2332, found 740.2331.
1
Yield 90.5%; TLC Rf (A) 0.39; H NMR (DMSO-d6) δ 10.72 (s,
1H), 8.67 (s, 1H), 7.48-7.31 (m, 10H), 6.37 (t, 1H), 5.35 (t, 1H), 4.92
(t, 1H), 4.45 (s, 1H), 3.87 (t, 1H), 3.62-3.50 (m, 2H), 2.82-2.74 (m,
2H), 2.34-2.30 (m, 1H), 1.09 and 1.08 (2xd, 6H); 13C NMR (DMSO-
d6) d 174.97, 155.06, 154.54, 152.29, 150.27, 144.04, 141.65, 129.46,
127.32, 120.54, 88.12, 83.65, 70.67, 61.61, 39.30, 34.61, 19.32, 19.30;
HRMS (FAB) calcd for C27H29N6O6 (M + H) 533.2148, found
533.2150.
Synthesis of 5′-O-[3-(Trifluoromethyl)phenoxy]carbonyl-2′-
deoxynucleosides. N-protected 2′-deoxynucleoside (5 mmol) was
coevaporated three times with pyridine, dried in vacuo for 16 h, and
dissolved in anhydrous pyridine (50 mL). The solution was cooled over
a dry ice/ethanol bath and 3-(trifluoromethyl)phenyl chloroformate (1.19
g, 5.25 mmol) was added. The cooling bath was removed and the
reaction mixture shaken until all the reactant was completely dissolved.
After stirring overnight at room temperature, the reaction mixture was
quenched with water. The product was extracted with DCM, washed
with 5% aqueous solution of NaHCO3, and dried with anhydrous Na2-
SO4. After concentration on a rotary evaporator, the crude product was
purified by chromatography on a silica gel column using initially CHCl3/
benzene (9:1) followed by a gradient of methanol in CHCl3 (for
N-dimethoxytrityl analogues, 1% pyridine was added to the eluting
system).
Compound 18a: yield 77.4%; TLC Rf (A) 0.42; 1H NMR (DMSO-
d6) δ 11.37 (bs, 1H), 7.75-7.51 (m, 5H), 6.25-6.22 (m, 1H), 5.53 (s,
1H), 4.49-4.32 (m, 3H), 4.04-4.00 (m, 1H), 2.24-2.11 (m, 2H), 1.77
(s, 3H); 13C NMR (DMSO-d6) δ 163.74, 152.61, 150.95, 150.51, 136.07,
131.09, 130.49, 130.23, 125.75, 123.11, 118.72, 109.87, 83.95, 83.22,
70.10, 68.55, 38.52, 12.14; HRMS (FAB) calcd for C18H18N2O7F3 (M
+ H) 431.1066, found 431.1086.
Compound 18b: yield 68.0%; TLC Rf (A) 0.46; 1H NMR (DMSO-
d6) δ 11.32 (bs, 1H), 8.19 (d, 1H), 8.01 (d, 2H), 7.78-7.51 (m, 7H),
7.39 (bs, 1H), 6.22-6.19 (m, 1H), 5.57 (s, 1H), 4.53-4.30 (m, 3H),
4.20-4.16 (m, 1H), 2.37-2.12 (m, 2H); 13C NMR (DMSO-d6) δ
Compound 18g: yield 80.1%; TLC Rf (A) 0.55; 1H NMR (DMSO-
d6) δ 10.75 (bs, 1H), 8.63 (s, 1H), 7.72-7.32 (m, 14H), 6.44 (t, 1H),
5.59 (d, 1H), 4.72-4.42 (m, 3H), 4.15 (dt, 1H), 2.97-2.93 (m, 1H),
2.78 (septet, 1H), 2.45-2.39 (m, 1H), 1.09, 1.08 (2xd, 6H); 13C NMR
(DMSO-d6) δ 174.87, 155.13, 154.27, 152.53, 152.21, 150.93, 150.26,
144.58, 141.61, 130.95, 130.17, 129.45, 127.18, 125.73, 123.01, 120.86,
118.70, 84.25, 84.19, 70.48, 68.83, 38.26, 34.67, 19.26, 19.22; HRMS
(FAB) calcd for C35H32N6O8F3 (M + H) 721.2234, found 721.2218.
Synthesis of 5′-Carbonate Protected 2′-Deoxythymidines (2-15).
2′-Deoxythymidine (4.84 g, 20 mmol) was coevaporated three times
with pyridine, dried in vacuo for 16 h, and then dissolved in anhydrous
pyridine (200 mL). The appropriate aryl or alkyl chloroformate (22
mmol) was added and the reaction mixture was stirred overnight. The
crude product was purified by silica gel column chromatography (0-
10% ethanol in DCM).
Compound 2: yield 88.3%, mp 178-180 °C; TLC Rf (B) 0.22; 1H
NMR (DMSO-d6) δ 7.40 (s, 1H), 7.25 (d, 2H), 7.00 (d, 2H), 6.25-
6.20 (m, 1H), 4.40-3.95 (m, 4H), 2.30-2.05 (m, 2H), 1.80 (s, 3H);
13C NMR (DMSO-d6) δ 164.4, 152.6, 150.5, 149.0, 135.5, 131.5, 129.4,
122.0, 110.9, 84.7, 83.5, 70.2, 67.5, 39.8, 12.0; Anal. Calcd for C17H17-
ClN2O7: C, 5l.5; H, 4.3; N, 7.1. Found: C, 51.3; H, 4.5; N, 7.0; ESI-
MS: 397 (M + H).
Yields and molecular weight analysis (ESI-MS) are summarized in
Table 1 for compounds 3-15. These compounds were prepared by the
procedure used to synthesize 2.
Synthesis of 5′-O-[3-(trifluoromethyl)phenoxycarbonyl]-3′-O-[(2-
cyanoethyl)-N,N-diisopropylaminophosphino]-2′-deoxynucleoside
(19a-g). The protected 2′-deoxynucleoside (3 mmol) and tetrazole (210
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13439

A R T I C L E S
Sierzchala et al.
mg, 3 mmol) were dried separately in vacuo for 16 h. The protected
2′-deoxynucleoside was then dissolved in anhydrous DCM (30 mL),
and 2-cyanoethyl-N,N,N′,N′-tetraisopropylphosphane (950 mg, 3.15
mmol) was added in one portion. Tetrazole was added slowly to the
reaction mixture over 1 h. The reaction mixture was then stirred for
another 3 h. Triethylamine (0.5 mL) was added to neutralize the
solution. The solvent was removed in vacuo, and the crude product
was chromatographed with benzene containing 0.1% Et3N followed
by a gradient of ethyl acetate (0-40%) in benzene containing 0.1%
Et3N (no Et3N for 19c).
Compound 19a: yield 76.6%; 31P NMR (CDCl3) δ 149.44; 13C
NMR (CDCl3) δ 163.90, 153.10, 153.08, 151.06, 151.03, 150.59,
150.52, 135.63, 135.54, 132.48, 132.22, 130.51, 128.52, 124.63, 123.39,
118.46, 118.43, 117.89, 111.75, 111.70, 85.15, 85.12, 83.03, 82.98,
73.29, 73.16, 72.96, 72.83, 67.95, 67.87, 58.29, 58.21, 58.14, 58.05,
43.59, 43.56, 43.49, 43.46, 39.54, 39.52, 24.81, 24.76, 24.69, 20.69,
20.64, 20.59, 12.74; HRMS (FAB) calcd for C27H35N4O8F3P (M + H)
631.2145, found 631.2141.
Compoound 19b: yield 45.7%; 31P NMR (CDCl3) δ 149.73, 149.44;
13C NMR (CDCl3) δ 166.64, 162.48, 155.05, 153.11, 151.06, 144.21,
133.41, 132.41, 132.14, 130.46, 129.23, 128.50, 127.72, 124.75, 123.37,
123.34, 118.56, 118.54, 117.86, 96.90, 87.50, 83.98, 72.54, 72.42, 67.69,
58.21, 58.07, 43.54, 43.45, 40.88, 40.85, 24.82, 24.76, 24.69, 20.60,
20.54; HRMS (FAB) calcd for C33H38N5O8F3P (M + H) 720.2410,
found 720.2435.
Compound 19c: yield 50.3%; 31P NMR (CDCl3) δ 149.70, 149.40;
13C NMR (CDCl3) δ 162.55, 154.89, 153.11, 153.07, 152.39, 151.06,
151.03, 144.02, 143.37, 143.32, 141.49, 132.42, 132.16 130.46, 128.52,
128.17, 127.40, 125.13, 124.74, 124.69, 123.35, 120.34, 118.55, 118.52,
117.86, 117.81, 95.10, 95.03, 87.49, 83.92, 83.59, 72.82, 72.69, 72.59,
72.47, 68.23, 67.72, 67.65, 58.40, 58.25, 58.21, 58.05, 46.78, 43.56,
43.54, 43.46, 43.44, 40.79, 24.79, 24.74, 24.67, 20.61, 20.55; HRMS
(FAB) calcd for C41H44N5O9F3P (M + H) 838.2829, found 838.2808.
Compound 19d: yield 61.2%; 31P NMR (CDCl3) δ 149.13, 149.03;
13C NMR (CDCl3) δ 165.56, 158.83, 155.28, 152.98, 152.95, 151.05,
151.01, 144.59, 140.37, 140.31, 136.36, 136.32, 132.07, 130.37, 130.00,
128.64, 128.48, 127.61, 124.68, 124.63, 123.21, 118.44, 118.41, 117.83,
113.72, 95.05, 94.98, 86.84, 86.76, 83.26, 82.97, 82.92, 73.20, 73.06,
72.75, 72.62, 70.27, 67.98, 58.37, 58.22, 58.18, 58.03, 55.38, 43.51,
43.49, 43.41, 43.39, 40.35, 24.78, 24.75, 24.72, 24.69, 20.59, 20.54,
20.49; HRMS (FAB) calcd for C47H51N5O9F3P (M+) 917.3377, found
917.3346.
Compound 19e: yield 65.2%; 31P NMR (CDCl3) δ 149.19, 149.16;
13C NMR (CDCl3) δ 164.85, 153.08, 153.05, 152.84, 151.59, 151.04,
151.01, 149.79, 141.90, 141.85, 133.71, 132.97, 132.28, 132.01, 130.37,
129.00, 128.02, 124.75, 124.72, 123.75, 123.22, 118.52, 118.49, 117.83,
117.79, 84.93, 83.95, 83.68, 83.63, 73.66, 73.53, 73.21, 73.08, 68.01,
67.88, 58.44, 58.35, 58.28, 43.56, 43.52, 43.46, 43.42, 39.35, 39.31,
39.27, 24.83, 24.76, 24.70, 20.67, 20.62, 20.57; HRMS (FAB) calcd
for C34H38N7O7F3P (M + H) 744.2522, found 744.2547.
Compound 19f: yield 74.1%; 31P NMR (CDCl3) δ 149.07, 148.98;
13C NMR (CDCl3) δ 158.45, 154.39, 153.17, 153.13, 152.57, 151.13,
151.10, 148.61, 148.57, 145.58, 138.73, 138.68, 137.61, 132.35, 132.09,
130.37, 130.27, 128.96, 128.52, 128.07, 127.00, 124.84, 124.80, 123.21,
121.70, 118.63, 118.61, 117.82, 117.78, 113.32, 84.73, 83.78, 83.48,
83.43, 73.86, 73.72, 73.34, 73.21, 70.78, 68.25, 68.15, 58.48, 58.35,
55.40, 43.57, 43.52, 43.47, 43.42, 39.03, 24.86, 24.81, 24.78, 24.76,
24.72, 24.70, 20.68, 20.62, 20.58; HRMS (FAB) calcd for C48H51N7O8F3P
(M+) 941.3489, found 941.3510.
Compound 19g: yield 70.4%; 31P NMR (CDCl3) δ 149.48, 149.37;
13C NMR (CDCl3) δ 175.20, 156.36, 156.33, 154.51, 154.46, 153.16,
153.10, 152.12, 152.10, 151.13, 151.11, 150.60, 143.12, 141.91, 132.26,
131.99, 130.36, 129.33, 128.51, 127.85, 124.85, 124.80, 123.15, 123.13,
122.06, 118.56, 118.53, 117.89, 117.86, 85.53, 85.39, 84.07, 83.90,
83.85, 74.02, 73.88, 73.57, 73.45, 68.32, 68.17, 58.43, 58.28, 43.59,
43.56, 43.49, 43.46, 38.93, 36.35, 24.83, 24.78, 24.71, 20.63, 20.60,
20.57, 20.54, 19.41, 19.38; HRMS (FAB) calcd for C44H49N8O9F3P
(M + H) 921.3312, found 921.3353.
Automated Oligodeoxynucleotide Synthesis. All syntheses were
performed on a 0.2 µM scale using CPG columns joined to 5′-DMT
protected 2′-deoxythymidine. The DMT protecting group was removed
using 3% TCA in DCM prior to automated synthesis. 5′-ARCO
modified 2′-deoxynucleoside-3′-phosphoramidites (0.1 M each) in
anhydrous acetonitrile were installed at ports 1-4 on the ABI/PE
Biosystems model 394 DNA synthesizer. Other solutions unique to
this two-step cycle were installed at ports 10 (peroxy anion buffer), 11
(DMF) and 19 (dioxane). Removal of the 5′-carbonate following each
2′-deoxynucleotide addition was completed by treatment with peroxy
anion buffer (Buffer C, Table 2, unless specifically substituted in an
experiment) for variable times: pyrimidines, 2.5 min; purines, 7 min).
After this 5′-carbonate deprotection step, the column was washed with
dioxane, DMF and acetonitrile.
Following completion of all synthesis steps (including removal of
the 5′-carbonate from the final synthon added to the growing ODN),
the product mixture while still joined to CPG was treated with 3%
TCA in DCM at r.t. (2 h) to remove the dimethoxytrityl groups from
the bases. After a wash with acetonitrile, products were removed from
CPG by treatment with concentrated ammonium hydroxide for 2 h at
rt. The reaction mixture free of CPG was transferred to a screw cap
reactivial fitted with a Teflon-silicon seal, and the solution heated at
55 °C overnight to remove the N,N-diphenylcarbamoyl protecting group
from guanine. The vial was then cooled in an ice bath, the reaction
mixture transferred to an eppendorf tube, and the solution evaporated
to dryness. The resulting oligomer was dissolved in water containing
5% acetonitrile and the product purified by reverse phase HPLC using
an ODS-Hypersil (5 µ) column eluted at 1.5 mL/min. The eluant was
0-20% acetonitrile in 50 mM TEAB (linear gradient, 40 min). Product
fractions were concentrated to dryness to remove the volatile buffer
and frozen.
MALDI-TOF Analysis of Oligodeoxynucleotides. Oligodeoxy-
nucleotides were purified by HPLC using 50 mM TEAB buffer (pH
8.0) with a 0-20% acetonitrile gradient and then dissolved in
acetonitrile/ultrapure water (1:1) to give a final concentration of 0.005
A260 units/µL. The matrix solution was prepared by mixing 2′,4′,6′-
trihydroxyacetophenone (THAP, 45 mg) and ammonium citrate (4
crystals) in acetonitrile/ultrapure water (1:1, 500 µL). The suspension
was vortexed, centrifuged, and the clear matrix solution (0.3 µL)
deposited on the probe. The oligodeoxynucleotide solution (0.3 µL)
was then deposited on the same spot, and the plate left to dry at r.t. for
5 min. All mass spectra were acquired in both negative and positive
ion modes with delayed extraction and reflector operation.
Enzymatic Hydrolysis of Oligodeoxynucleotides. In a typical
experiment the reaction mixture (30 µL) containing 0.3 µg of snake
venom phosphodiesterase, 4.02 µg of alkaline phosphatase, buffer (32
mM Tris-Cl, 12 mM MgCl2, pH 7.5), and an oligomer (0.5 A260 units)
was incubated at 37 °C for 4 h. Following heat denaturation of enzymes
(95 °C, 3 min), the reaction mixture was diluted to 200 µL with water,
centrifuged, and analyzed by reverse phase HPLC [ODS-Hypersil (5
µ) column 4.0 × 250, flow 1.5 mL/min, 0-20% acetonitrile in 50 mM
TEAB (linear gradient) in 40 min].
Hydrolysis of RNA Heteroduplexes with Escherichia coli RNase
H1. 5′-32P-labeled RNA 100,000 cpm/reaction), unlabeled RNA (50
pmole) and complementary oligodeoxynucleotide (50 pmol) were added
to a buffer (pH 7.8) containing 20 mM HEPES-KOH, 50 mM KCl,
10 mM MgCl2 and 1 mM dithiothreitol. The duplexes were hybridized
by heating briefly to 95 °C and then incubated at 4 °C for 30 min. E.
coli RNase H1 (Promega) was added (4 units), and the reactions were
allowed to proceed for 1 h at 25 °C (28 µL total reaction volume).
Aliquots of the reaction mixture (3.5 µL) were quenched with 6.5 µL
of 7 M urea and 20 mM EDTA and were stored on ice until analysis
by PAGE (20%, 19:1 cross-link). The developed gels were analyzed
9
13440 J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003

Solid-Phase Oligodeoxynucleotide Synthesis
A R T I C L E S
using a Molecular Dynamics Phosphorimager (Storm 820) and Im-
ageQuant software (version 5.1).
Philosof-Oppenheimer for scientific discussion; Robert Barkley,
Mark Aldridge, Richard Shoemaker, and Mary Noe for technical
assistance. We also thank Steve Laderman, Mel Kronick,
Darlene Solomon, and Barry Willis for additional support.
Acknowledgment. This work was supported by NIH (Grant
GM25680), NSF-CRIF instrument award (CHE-0131003), Agi-
lent Technologies (N39998-01). We thank Geraldine Fulcrand,
Steve Lefkowitz, Michel Perbost, David Sheehan, Rachel
JA030376N
9
J. AM. CHEM. SOC. VOL. 125, NO. 44, 2003 13441