effectively in most cases but inconsistencies in the assay
results became apparent during simultaneous study of
multiple 2′-OMe sites in yeast rRNA. The inconsistencies
were likely caused by two drawbacks associated with the
step of enzymatically generating AppDNA. The first draw-
back is that T4 DNA ligase requires a donor-template duplex
of 11-bp in length for the adenylation step to be fully
efficient.5 The requirement for a donor-template duplex was
also observed in the preparation of AppRNA by T4 DNA
ligase.7 This drawback is difficult to overcome when using
biological RNAs, which contain varying degrees of structure
that interfere with DNA substrate hybridization. The second
drawback is the dependence of adenylation efficiency on the
sequence of the DNA donor substrate. Some DNA donor
substrates are adenylated efficiently while others react poorly,
such that the outcome for any new substrate is generally
unpredictable.8
be monitored conveniently by 31P NMR, we first studied the
coupling reactions of two activated 5′-phosphorylated ade-
nylate derivatives with the adenosine 5′-monophosphate
derivative 5 and then extended the conditions to the solid
phase synthesis of AppDNA.
Phosphorimidzaolides have been used widely as phospho-
rylation reagents for pyrophosphate bond formation and are
prepared readily from the corresponding 5′-H-phospho-
nates.15 Therefore, we first studied the solution coupling of
5 with adenosine 5′-phosphorimidazolidate 4 (Scheme 1).
Scheme 1.
Synthesis of A5′-p-p-5′A by Coupling of 4 and 5
To determine whether the inconsistencies in the ligation
reaction with rRNA arise primarily from the generation of
the adenylated DNA, we conducted ligation reactions using
purified AppDNA substrates. First, we generated AppDNA
substrates enzymatically according to Chiuman and Li’s
procedure. After purification by gel electrophoresis we used
them in ligation reactions. Using these independently gener-
ated AppDNA substrates, we found the ligation results were
reproducible. However, this procedure is extremely labor
intensive and impractical for high-throughput analysis. For
example, to analyze the 55 2′-OMe sites in yeast rRNA, we
would have to generate enzymatically and isolate 55 App-
DNA substrates, possibly at vastly different efficiencies.
Another way to make AppDNA is through chemical syn-
thesis. However, solution synthesis of AppDNA usually gives
unsatisfactory yield.9 In this work, we establish conditions
for efficient 5′ adenylation during solid phase oligonucleotide
synthesis, thereby making it possible to reliably and conve-
niently generate 5′ adenylated DNA oligonucleotides. We
demonstrate that the chemically synthesized AppDNA
products efficiently support ligation-based detection of RNA
modifications.
The key step in the chemical synthesis of AppDNA
involves pyrophosphate bond formation, which entails cou-
pling of an activated 5′-phosphorylated adenylate derivative
with an immobilized DNA-5′-monophosphate. Several types
of activated phosphoester derivatives have been reported to
react with monophosphates in solution to form the pyro-
phosphate bond, including phosphorimidazolidate,10 phosphor-
N-methylimidazolidate,11 phosphoromorpholidates,12 sub-
stituted benzotriazol-1-yl phosphorothioates,13 and phosphate
derivatives.14 As formation of the pyrophosphate bond can
For two reasons, we chose to protect the 2′, 3′-OH, and N6-
NH2 groups by benzoylation: First, after the coupling reaction
the benzoyl groups can be removed readily by ammonia
treatment. Second, benzoyl groups have greater hydrophobic
character than acetyl groups, which facilitates the workup
and chromatographic purification of 3 and 5 relative to the
corresponding acetyl derivatives.
Thus, commercially available 1 was converted to inter-
mediate 2 by benzoyl chloride followed by removal of the
trityl group. Treatment of 2 with diphenyl phosphite followed
by hydrolysis in the presence of Et3N16 afforded 5′-H-
phosphonate 3 as pale foam, which is quite stable and can
be stored at rt for months. We also tried to synthesize the
corresponding 2′, 3′, N6-triacetyl 5′-H-phosphonate deriva-
tive; however, aqueous phase solubility during extraction
rendered the product difficult to purify.
5′-H-phosphonate 3 was then activated to phosphorimi-
dazolidate 4 in quantitative yield (as indicated by 31P NMR,
δ changed from 3.3 to -9.4 ppm in DMF) by treatment with
(trimethylsilyl)imidzaole in the presence of Et3N in CH3CN-
(12) (a) Wittmann.; Wong, C.-H. J. Org. Chem. 1997, 62, 2144–2147.
(b) Zhao, Y.; Thorson, J. S. J. Org. Chem. 1998, 63, 7568–7572. (c) Zhang,
Q.; Liu, H.-W. J. Am. Chem. Soc. 2000, 122, 9065.
(6) Saikia, M.; Dai, Q.; Decatur, W. A.; Fournier, M. J.; Piccirilli, J. A.;
Pan, T. RNA 2006, 12, 2025–2033.
(7) Wang, Y.; Silverman, S. K. RNA 2006, 12, 1142–1146.
(8) Patel, M. P.; Baum, D. A.; Silverman, S. K. Bioorg. Chem. 2008,
36, 46–56.
(13) (a) Sekine, M.; Hata, T. Curr. Org. Chem. 1998, 32–66. (b) Sekine,
M.; Ushioda, M.; Wada, T.; Seio, K. Tetrahedron Lett. 2003, 44, 1703–7.
(14) (a) Ohkubo, A.; Aoki, K.; Seio, K.; Sekine, M. Tetrahedron Lett.
2004, 45, 979–982. (b) Kadokura, M.; Wada, T.; Seio, K.; Moriguchi, T.;
Huber, J.; Luhrmann, R.; Sekine, M. Tetrahedron Lett. 2001, 42, 8853–
8856.
(9) (a) Lohrmann, R.; Orgel, L. E. Tetrahedron 1978, 34, 853. (b) Unraw,
P. J.; Bartel, D. P. Nature 1998, 395, 260. (c) Lau, N. C.; Lim, L. P.;
Weinstein, E. G.; Bartel, D. P. Science 2001, 294, 858.
(10) (a) Tsvetkov, Y. E.; Nikolaev, A. V. J. Chem. Soc., Perkin Trans.,
I 2000, 889–891. (b) Shimazu, M.; Shinozuka, K.; Sawai, H. Tetrahedron
Lett. 1990, 31, 235–238.
(15) Kadokura, M.; Wada, T.; Urasjima, C.; Sekine, M. Tetrahedron
Lett. 1997, 38, 8359–8362.
(16) Jankowska, J.; Sobkowski, M.; Stawinski, J.; Kraszewski, A.
Tetrahedron Lett. 1994, 35, 3355–3358.
(11) Marlow, A. L.; Kiessling, L. L. Org. Lett. 2001, 3, 2517–2519.
1068
Org. Lett., Vol. 11, No. 5, 2009