Synthesis of pyrophosphates for in vitro selection of catalytic RNA molecules

Article abstract of DOI:10.1080/15257770701571867

Reported here are synthetic routes to pyrophosphates linking riboflavin with various nucleosides. The focus is on a flavin-uracil dinucleotide having a biotin tag on the uracil, a molecule that has potential value in the selection of RNA enzymes that catalyze the template-directed polymerization of RNA in the 3′-to-5′ direction, which is the direction opposite that catalyzed by standard protein polymerases. Two detailed procedures are presented to prepare this new compound, as well as one procedure to prepare the new flavin-2,6-diaminopurine dinucleotide. Copyright Taylor & Francis Group, LLC.

Full text of DOI:10.1080/15257770701571867

Nucleosides, Nucleotides, and Nucleic Acids, 27:43–56, 2008
Copyright ꢀC Taylor & Francis Group, LLC
ISSN: 1525-7770 print / 1532-2335 online
DOI: 10.1080/15257770701571867
SYNTHESIS OF PYROPHOSPHATES FOR IN VITRO SELECTION
OF CATALYTIC RNA MOLECULES
Hyo-Joong Kim, Myong Jung Kim, Nilesh Karalkar, Daniel Hutter, and
Steven A. Benner
Foundation for Applied Molecular Evolution, Gainesville, Florida, USA
Reported here are synthetic routes to pyrophosphates linking riboflavin with various nucleosides.
2
The focus is on a flavin-uracil dinucleotide having a biotin tag on the uracil, a molecule that has
potential value in the selection of RNA enzymes that catalyze the template-directed polymerization
of RNA in the 3ꢁ-to-5ꢁ direction, which is the direction opposite that catalyzed by standard protein
polymerases. Two detailed procedures are presented to prepare this new compound, as well as one
procedure to prepare the new flavin-2,6-diaminopurine dinucleotide.
Keywords Dinucleotides; flavin; origin of life; ribozymes
Following the discovery of RNA molecules that have catalytic activity as
phosphotransesterases,[1,2] and the proposal that an episode of early life
on Earth may have used RNA as its sole genetically encoded component,[3,4]
many workers have considered the possibility of selecting for RNA molecules
that catalyze the template-directed synthesis of RNA. Obtaining such cata-
lysts would be a step toward realizing a “synthetic biology,”[5] and would also
add support to an RNA world hypothesis for the origin of life.
RNA polymerases known in natural biology synthesize phosphodiester
bonds via a process where the nucleophilic reactant (the 3ꢁ-OH group
of the 3ꢁ-terminal ribonucleotide) is delivered by a primer, while the
electrophilic reactant, a phosphorus atom having a pyrophosphate unit as
the leaving group, is delivered by an incoming monomer. This means that
the nucleophilic component in the reaction is tightly held to the template
Received 2 January 2007; accepted 29 June 2007.
This work was supported by the John Templeton Foundation, which seeks to address life’s big
questions, the National Aeronautics and Space Administration under its Exobiology program, and the
National Science Foundation under its Chemical Bonding Program in a collaboration between the
Foundation for Applied Molecular Evolution, the Harvard Medical School, the Howard Hughes Medical
Institute, and The Scripps Research Institute (La Jolla).
Address correspondence to Steven A. Benner, Foundation for Applied Molecular Evolution, PO
Box 13174, Gainesville, FL, 32604, USA. E-mail: sbenner@ffame.org
43

44
H.-J. Kim et al.
by Watson-Crick pairing that extends the length of the primer–template
duplex, while the electrophilic component (the monomer) is held in the
reacting complex by only (two or three) hydrogen bonds. Thus, in protein
polymerases from the modern world, much of the binding energy holding
the incoming monomer in the proper orientation must come from binding
interactions between the triphosphate and the protein catalyst.
This places multiple demands on the structure of any RNA molecule
that might catalyze the polymerization of RNA, demands that may be found
only infrequently in a library of random RNA sequences. For example, it
may be difficult to uncover an RNA molecule (a polyanion) that binds to
a pyrophosphate leaving group (also a polyanion). Interactions between
the incoming triphosphate and the RNA catalyst may be difficult to find
that do not discriminate between the four standard nucleobases. While we
cannot today calculate the size of a library that is needed to contain (with,
for example, 50% probability) a single RNA molecule that solves all of these
problems, estimates can be made based on the types of RNA catalysts that
have emerged from random libraries,[6] and the size appears to be quite
large, perhaps as large as 1030 random RNA sequences.
With these thoughts in mind, we considered an architecture that
reverses the direction of RNA synthesis, from 5ꢁ-to-3ꢁ to 3ꢁ-to-5ꢁ, and changes
the leaving group (Figure 1). In this architecture, the leaving group is
not pyrophosphate, but rather the phosphate ester of an alcohol. The
electrophilicity of phosphorus is similar for leaving groups of the type
O-PO2-O-R and O-PO2-O-PO3= (the standard triphosphate); the free energy
of hydrolysis of a pyrophosphate bond is about the same as the free energy of
hydrolysis of a triphosphate. In this architecture, however, R can be chosen
to be a better “handle” for the desired RNA catalyst to bind.
In choosing the R group “handle,” we began by considering small
molecules for which RNA molecules had already been developed to bind.
Flavin was a particularly interesting ligand in this respect, as RNA molecules
that bind flavin have been discovered by selection among random RNA
libraries,[7] and have been optimized by deliberate structural change.[8]
This allows the starting point for catalyst selection to contain two
flavin-binding RNA aptamers and a reversed direction of oligonucleotide
synthesis. This has several hypothetical advantages. First, the architecture
should allow both the nucleophilic and electrophilic components to be
held by strong, multipoint noncovalent attachments (Figure 1). Thus, the
incoming nucleophile on the monomer is held in the reaction complex by
hydrogen bonding to the template (as in the standard architecture), and to
a flavin. The electrophilic center, now provided by the primer, is held in the
complex by template–primer interactions, as well as to the catalyst through
the second flavin aptamer.
To implement this idea, a set of four flavin-pyrophosphate substrates are
needed. These fall into the general structure of flavin adenine dinucleotide,

Pyrophosphates for In Vitro Selection
45
FIGURE 1 Schematic diagram of a conceptual 3ꢁ-to-5ꢁ reverse polymerization reaction that is the
application of the pyrophosphates prepared here.
which is one of the substrates that would be used in this architecture.
Pyrophosphates of this type have long been known, back to early work
of Khorana,[9] who prepared flavin adenine dinucleotide by the coupling
of flavin mononucleotide with adenosine-5ꢁ-monophosphate that had been
activated as its corresponding morpholidate, with a low level of protection.
In the Moffatt-Khorana procedure, the 2ꢁ- and 3ꢁ-hydroxyl groups of the
FIGURE 2 The chemical structure of the pyrophosphate with a biotin tag.

46
H.-J. Kim et al.
adenosine, the ribitol hydroxyl groups of the riboflavin, and the exocyclic
amine functionalities were all not protected. The coupling was performed
in pyridine over 2 days.
For several of the pyrophosphates needed for this architecture, this
approach was successful in our hands. In particular, the 2,6-diaminopurine-
flavin, guanine-flavin,[10] and the cytosine-flavin[10] dinucleotides were
all readily prepared from their morpholidates. The procedure for 2,6-
diaminopurine-flavin dinucleotide, a new compound, is included in the
experimental section.
Attempts to prepare biotinylated uracil-flavin dinucleotide (Figure
2) unexpectedly encountered problems, however. These focused on the
difficulty of obtaining the precursor biotinylated 5-aminoallyluridine 5ꢁ-
monophosphate, in part due to an irreproducible selectivity in the phospho-
rylation of biotinylated 5-aminoallyluridine using standard conditions (tri-
ethylphosphate as a solvent, POCl3, and a trace amount of water).[11] These
conditions are used for the preparation of nucleoside 5ꢁ-monophosphates.
Further optimization of the coupling reaction, which remains slow, using
the manganese dication is also reported.
Given the potential value of these and related compounds, we present
detailed experimental procedures for the two new species. We focus on the
most difficult pyrophosphate 1, the species where the nucleobase carries a
biotin tag.
RESULTS AND DISCUSSION
First, 2,6-diaminopurine-flavin dinucleotide was synthesized as shown in
Scheme 1. 2-Aminoadenosine 2 was phosphorylated with phosphorus oxy-
chloride in triethyl phosphate to give the monophosphate 3 in 40% yield.
Compound 3 was refluxed with morpholine and dicyclohexylcarbodiimide
in a mixture of H2O and t-butyl alcohol to afford the morpholidate 4 in 79%
yield. The coupling of compound 4 with pyridinium riboflavin-5ꢁ-phosphate
was performed via a slightly modified Moffatt-Khorana procedure to give
flavin 2,6-diaminopurine dinucleotide 5. The reaction was monitored by
31P NMR. Compound 5 was purified by ion-exchange HPLC, followed by
RP-HPLC, and identified by 1H and 31P NMR (δ −10.66 and −11.37).
The first route to the biotinylated intermediate 9 was attempted from
the commercially available uridine-5ꢁ-monophosphate (Scheme 2). This
appeared to be a strategically wise choice, as the route was known in the
literature,[12] and avoided the need to later phosphorylate an advanced
synthetic intermediate having multiple hydroxyl groups.
However, in our hands, the reaction of the mercuric compound 7
with trifluoroacetyl allylamine (Scheme 2) gave only low yields of 8, as

Pyrophosphates for In Vitro Selection
47
SCHEME 1 Synthesis of 2,6-diaminopurine-flavin dinucleotide. Conditions: (a) (EtO)3PO, H2O,
POCl3, 0◦C, 18 hours (40%); (b) morpholine, DCC, t-BuOH, H2O, reflux, 6 hours (79%); (c) FMN-
pyridinium salt, pyridine, Bu3N, 1-tetrazole, RT, 2 days (11%).
SCHEME 2 Attempted synthesis of the biotinylated monophosphate 9. Conditions: (a) Hg(OAc)2, LiCl,
NaOAc, pH 6 buffer; (b) (1) trifluoroacetyl allylamine, K2PdCl4, NaOAc buffer, DMF, 80◦C, 2 hours;
(2) NH4OH, RT; (c) biotin N -hydroxysuccinimide, pyridine, Et3N.

48
H.-J. Kim et al.
SCHEME 3 Synthesis of the monophosphate 9 with partial protection. Conditions: (a) (1) trifluo-
roacetyl allylamine, Na2PdCl4, NaOAc buffer, DMF, 80◦C, 2 hours; (2) TBDMSCl, imidazole, DMF; (3)
NH4OH, MeOH, RT, 17 hours (40% for 3 steps); (b) (1) biotin N -hydroxysuccinimide, pyridine, Et3N,
RT, 2 hours; (2) Et3N 3HF, THF, RT (75% for two steps); (c) triethylphosphate, POCl3, H2O, 0◦C, 18
hours, (45%).
well as several impurities that were not separable by preparative column
chromatography.
In contrast, the Heck reaction of iodouridine 10 with trifluoroacetyl
allylamine in the presence of Na2PdCl4 gave the coupled product in good
yield (Scheme 3).[13] The initial product, a trialcohol, was difficult to
purify. Therefore, the catalyst was removed by purification using column
chromatography, the crude product was converted to the tris-TBDMS ether,
and the trifluoroacetyl group was removed to give the allylamine compound
11 in 40% overall yield.
The amino group of the allylamine compound 11 was then reacted with
the activated biotin derivative biotin N -hydroxysuccinimide. The silyl groups
were then removed with Et3N.3HF to give the biotinylated uridine 12 in 75%
overall yield. Treating with three equivalents of POCl3 in triethylphosphate
in the presence of a small amount of H2O gave regioselectively the desired
monophosphate 9, but only in 45% yield, and only in the best experiment.
The inconsistent regioselective phosphorylation required us to develop
a more reliable method for the phosphorylations. First, the 2ꢁ- and 3ꢁ-
hydroxyl groups were protected following the procedure in Scheme 4. It
was expected that protection of the 2ꢁ, 3ꢁ-diol would give a substrate that was
more consistently phosphorylated by increasing the solubility of the reactant
and removing the possibility of unwanted 2ꢁ- or 3ꢁ-phosphorylation.
After Heck reaction of iodouridine 10 with trifluoroacetyl allylamine,
the product was purified by column chromatography and treated with

Pyrophosphates for In Vitro Selection
49
SCHEME 4 Synthesis of the monophosphate 9 with full protection. Conditions: (a) (1) trifluoroacetyl
allylamine, Na2PdCl4, NaOAc buffer, DMF, 80◦C, 2 hours; (2) 2,2-dimethoxypropane, p-TsOH, acetone,
RT, 1 hour; (3) MeNH2, RT, 3 hours, (67% for 3 steps); (b) biotin N -hydroxysuccinimide, pyridine,
Et3N, RT, 2 hours (72%); (c) triethylphosphate, POCl3, 0◦C, 18 hours, (56%); (d) HCl, H2O, RT,
18 hours (64%).
2,2-dimethoxypropane in acetone in the presence of p-TsOH, followed by
treatment with 40% methylamine solution. This gave the 2ꢁ,3ꢁ-protected
allylamine derivative 13 in 67% yield (Scheme 4). Then, 13 was biotinylated
to give 14 and converted to the monophosphate 15 by treatment of POCl3
in triethylphosphate in 56% yield. Finally, deprotection of isopropylidene of
15 by HCl in water gave the key intermediate 9.
With the monophosphate 9 in hand, pyrophosphate 1 was accessible
using an adaptation of a procedure of Lee et al. (Scheme 5).[14] First,
the monophosphate group in 9 was converted to its morpholidate with
dicyclohexylcarbodiimide to give 16 in 82% yield. The morpholidate 16
was then treated with flavin mononucleotide 17 as its trioctylamine salt
in pyridine and acetonitrile in the presence of tetrazole in the dark. The
progress of the reaction was followed by analytical ion-exchange HPLC; a
photodiode array detector allowed for easy identification of the different
compounds according to their UV spectra; the desired pyrophosphate
shows, in addition to a flavin absorbance, a shoulder at 300 nm, arising
from the uridine heterocycle conjugated with the double bond of the linker
carrying biotin). After 6 days at room temperature, all of the morpholidate
had been consumed. The desired pyrophosphate was recovered from the
reaction mixture by ion-exchange and reverse-phase HPLC to give a yellow
solid 1 in 5% yield.
[15]
By changing the acid catalyst from tetrazole to MnCl2
we were able
to shorten the reaction time to less than one day and double the yield,

50
H.-J. Kim et al.
SCHEME 5 Synthesis of the pyrophosphate 1. Conditions: (a) DCC, morpholine, t-BuOH, H2O, reflux,
4 hours (52%); (b) trioctylamine, pyridine, tetrazole, RT, 6 days (5%) or MnCl2, Na2SO4, formamide,
RT, 18 hours (10%).
which also made the HPLC purification easier, as more of the nearby FMN
peak had been transformed into product. We are now conducting in vitro
selection of catalytic RNA; these results will be reported in due course.
EXPERIMENTAL
2-Aminoadenosine-5-monophosphate (3)
To a cooled mixture of 2-aminoadenosine (500 mg, 1.77 mmol) in
(EtO)3PO (4.5 mL) was added a mixture of (EtO)3PO (4.5 mL), water
(6 µL), and POCl3 (0.5 mL, 5.36 mmol) at 0◦C; reaction mixture was kept
at 0◦C for 18 hours. The reaction mixture was poured into 1 M TEAB (2

Pyrophosphates for In Vitro Selection
51
mL) in water (20 mL) and extracted with EtOAc (2 × 10 mL). The aqueous
layer was rotary evaporated under reduced pressure and the residue was
purified by flash chromatography (silica, i-PrOH:NH4OH:H2O = 6:3:2) to
give 3 (ammonium salt, 270 mg, 0.71 mmol, 40%) as a solid.
2-Aminoadenosine 5ꢁ-Phosphomorpholidate (4)
To a refluxed solution of 3 (270 mg, 0.71 mmol) and morpholine
(0.24 mL, 2.75 mmol) in t-BuOH (7 mL) and H2O (7 mL) was added a
solution of DCC (585 mg, 2.83 mmol) in t-BuOH (10 mL). The reaction
mixture was refluxed for 6 hours, cooled to RT and filtered (washed
with t-BuOH). The filtrate was rotary evaporated under reduced pressure,
dissolved in H2O (20 mL) and washed with ether (10 mL × 3). The
aqueous layer was rotary evaporated under reduced pressure. The residue
was dissolved in MeOH (10 mL) and ether was added. The precipitate was
filtered and washed with ether to give 4 (410 mg, 0.57 mmol, 79%) as a solid.
1H NMR (300 MHz, D2O) δ 7.95 (s, 1H), 5.78 (d, 1H, J = 4.8 Hz), 4.61 (t,
1H, J = 5.3 Hz), 4.36 (t, 1H, J = 4.7 Hz), 4.16 (m, 1H), 3.85–3.88 (m, 2H),
2.77–3.66 (m, 18H), 0.95–1.76 (m, 20H); 31P NMR (D2O) δ 7.45.
Flavin-2-aminoadenosine dinucleotide (5)
Flavin mononucleotide (FMN) (220 mg, 0.41 mmol) and compound 4
(300 mg, 0.41 mmol) were dried at 100◦C under high vacuum for 2 h. A
mixture of FMN and Bu3N (0.1 mL, 0.42 mmol) in pyridine (5 mL) was
concentrated. The mixture was coevaporated with pyridine (3 mL × 2).
Compound 4 was added to this mixture and coevaporated with pyridine
(3 mL × 2). To this mixture were added pyridine (6 mL) and 1-tetrazole
(0.45 M in CH3CN, 2.73 mL) at RT. After being stirred at RT for 2 days,
the reaction mixture was rotary evaporated under reduced pressure and
quenched with sat. NH4HCO3 solution. The mixture was washed with
ether (10 mL), concentrated and purified by ion-exchange HPLC with
a linear gradient of ammonium bicarbonate (10–250 mM, followed by
reverse-phase HPLC with a linear gradient of acetonitrile, 0–60% in 25 mM
TEAA. After lyophilization, compound 5 (45 mg, 11%) was obtained as
1
its triethylammonium salt. H NMR (300 MHz, D2O) δ 7.86 (s, 1H), 7.58
(s, 1H), 7.46 (s, 1H), 5.52 (d, 1H, J = 5.1 Hz), 4.82 (m, 1H), 3.76–4.61
(m, 11H), 3.02 (q, Et3N), 2.30 (s, 3H), 2.25 (s, 3H), 1.11 (t, Et3N); 31P NMR
(D2O) δ −10.66 (d), −11.37 (d). The phosphorus–phosphorus coupling
constant is 22.0 Hz. Mass spectrometry (HPLC/ESI-MS) observed a single
charged [M-H]− at m/e 799.2.

52
H.-J. Kim et al.
5-Aminoallyl-2ꢁ,3ꢁ,5ꢁ-tri-O-tert-butyldimethylsilyl-uridine (11)
A mixture of 5-iodouridine 10 (2 g, 5.4 mmol), Na2PdCl4 (1.43 g,
4.9 mmol) and trifluoroacetyl allylamine (6 mL) in DMF (20 mL) and
NaOAc buffer (10 mL, 0.1 N, pH 5.2) was heated to 80◦C for 2 hours.
The mixture was cooled to RT and rotary evaporated under reduced pres-
sure. The product was purified from the residue by flash chromatography
(silica, gradient CH2Cl2:MeOH = 7:1 to 4:1). The appropriate fraction
(by tlc) was collected, evaporated and dissolved in DMF (40 mL). To this
solution were added imidazole (2.94 g, 43.2 mmol) and TBDMSCl (2.93 g,
19.4 mmol) at RT. After being stirred overnight at RT, the reaction mixture
was poured into water (200 mL) and extracted with ether (200 mL × 2). The
combined organic layers were dried over MgSO4, filtered and concentrated
by rotary evaporation. The residue was purified by flash chromatography
(silica, hexane:EtOAc = 4:1). The appropriate fraction was collected and
evaporated and treated with 28% aq. NH4OH (20 mL) in MeOH (40 mL).
The mixture was stirred at RT for 17 hours and evaporated. The residue was
purified by flash chromatography (silica, CH2Cl2:MeOH = 6:1) to give 11 as
a light yellow solid (1.38 g, 40% for 3 steps). 1H NMR (300 MHz, DMSO-d6)
δ 7.69 (s, 1H), 6.65 (m, 1H), 6.25 (d, 1H, J = 15.9 Hz), 5.84 (d, 1H, J = 6.9
Hz), 4.31 (dd, 1H, J = 4.5, 6.6 Hz), 4.06 (dd, 1H, J = 1.5, 4.5 Hz), 3.92 (m,
1H), 3.83 (dd, 1H, J = 4.8, 11.4 Hz), 3.70 (1H, J = 3.6, 11.4 Hz), 3.35–3.48
(m, 4H), 0.91, 0.89, 0.81 (3s, 27H), 0.11, 0.10, 0.09, 0.00, −0.09 (5s, 18H).
5-Biotinylaminoallyl-uridine (12)
A mixture of compound 11 (467 mg, 0.73 mmol), biotin-N -
hydroxysuccinimide ester (298 mg, 0.87 mmol) and Et3N (0.2 mL,
1.44 mmol) in pyridine (15 mL) was stirred at RT for 2 hours and concen-
trated. The residue was diluted with CH2Cl2 (200 mL) and washed with 0.1
N HCl (20 mL). The organic layer was washed with sat. NaHCO3 solution,
dried over Na2SO4, filtered, and concentrated. The residue was purified by
flash chromatography (silica, CH2Cl2:MeOH = 8:1) to give the protected
intermediate as a white solid (545 mg, 86%). 1H NMR (300 MHz, DMSO-d6)
δ 11.56 (s, 1H), 7.97 (t, 1H, J =5.7 Hz), 7.57 (s, 1H), 6.40 (m, 3H), 6.07
(d, 1H, J = 16.2 Hz), 5.86 (d, 1H, J = 7.2 Hz), 4.26 (m, 2H), 4.09 (m, 1H),
4.04 (m, 1H), 3.93 (m, 1H), 3.70-3.87 (m, 4H), 3.05 (m, 1H), 2.71 (m, 1H),
2.54 (d, 1H, J = 12.3 Hz), 0.05 (t, 1H, J = 7.5 Hz), 1.29–1.69 (m, 6H), 0.90,
0.88, 0.80 (3s, 27H), 0.1, 0.09, 0.07, −0.01, −0.11 (5s, 18H).
A mixture of biotinylated compound (545 mg, 0.63 mmol) and triethy-
lamine trihydrofluoride (Aldrich, 1.02 mL, 6.26 mmol) in THF (10 mL) was
stirred at 50◦C for 16 hours and concentrated. The residue was recrystallized
from MeOH to give 12 as a white solid (286 mg, 87%). 1H NMR (300 MHz,
DMSO-d6) δ 11.44 (s, 1H), 8.08 (s, 1H), 7.96 (t, 1H, J = 5.6 Hz), 6.36-6.45

Pyrophosphates for In Vitro Selection
53
(m, 3H), 6.10 (d, 1H, J = 15.9 Hz), 5.77 (d, 1H, J = 4.8 Hz), 5.40 (br
s, 1H), 5.21 (br s, 1H), 5.08 (br s, 1H), 4.28 (m, 1H), 4.10 (m, 1H), 4.05
(m, 1H), 3.97 (m, 1H), 3.84 (m, 1H), 3.55–3.76 (m, 4H), 3.07 (m, 1H), 2.79
(dd, 1H, J = 5.1, 12.6 Hz), 2.55 (d, 1H, J = 11.7 Hz), 2.07 (t, 2H, J = 15.0
Hz), 1.30–1.63 (m, 6H).
5-Biotinylaminoallyl-uridine 5ꢁ-monophosphate (9)
To a cooled mixture of compound 12 (280 mg, 0.53 mmol) in (EtO)3PO
(2.5 mL) was added a mixture of (EtO)3PO (2 mL), water (5 µL), and
POCl3 (0.15 mL, 1.61 mmol) at 0◦C and reaction mixture was kept at 0◦C
for 18 hours. The reaction mixture was neutralized with NaOH (1 N) and
extracted with EtOAc (2 × 10 mL). The aqueous layer was concentrated
in vacuo and the residue was purified by flash chromatography (silica,
i-PrOH:NH4OH;H2O = 6:2:2) to give 9 as a white solid (ammonium salt,
150 mg, 45 %). 1H NMR (300 MHz, D2O) δ 7.83 (s, 1H), 6.29 (m, 1H), 6.17
(d, 1H, J = 16.2 Hz), 5.84 (d, 1H, J = 5.1 Hz), 4.42 (m, 1H), 4.14–4.28
(m, 4H), 3.98 (m, 2H), 3.76 (m, 2H), 3.14 (m, 1H), 2.80 (dd, 1H, J = 5.1,
13.2 Hz), 2.59 (d, 1H, J = 13.2 Hz), 2.16 (t, 1H, J = 7.1 Hz), 1.24–1.58
(m, 6H); 31P NMR (D2O) δ 0.12. Mass spectrometry (HPLC/ESI-MS)
observed a single charged [M-H]− at m/e 604.3.
5-Aminoallyl-2ꢁ,3ꢁ-O-isopropylidene-uridine (13)
A mixture of 5-iodouridine 10 (2.10 g, 5.7 mmol), Na2PdCl4
(1.43 g, 4.9 mmol) and trifluoroacetyl allylamine (6 mL) in DMF
(20 mL) and NaOAc buffer (10 mL, 0.1 N, pH 5.2) was heated to 80◦C for
2 hours. The mixture was cooled to RT and evaporated and purified by
flash chromatography (silica, gradient CH2Cl2:MeOH = 7:1 to 4:1). The
appropriate fraction was collected and evaporated and treated with acetone
(100 mL), 2,2-dimethoxypropane (5 mL, 40 mmol) and p-TsOH·H2O
(0.80 g, 4.2 mmol). The mixture was stirred at rt for 1 hour and basified
by the addition of NH4OH (2 mL), evaporated, and resolved by flash
chromatography (silica, gradient ethyl acetate:hexane = 1:1 to 2:1 to 3:1).
The appropriate fraction was collected and evaporated and treated with
methylamine (10 mL, 40% in water). The mixture was stirred at RT for
3 hours and evaporated. The residue was purified by flash chromatography
(silica, gradient CH2Cl2:MeOH = 10:1 to 6:1 to 3:1) to give 13 as a light
yellow solid (1.3 g, 67% for 3 steps). 1H NMR (300 MHz, DMSO-d6) δ 11.5
(br s, 1H), 8.05 (s, 1H), 8.0 (br s, 2H), 6.57 (m, 1H), 6.33 (d, 1H, J =
16 Hz), 5.85 (d, 1H, J = 3 Hz), 5.25 (t, 1H, J = 5 Hz), 4.91 (dd, 1H, J =
6, 3 Hz), 4.76 (dd, 1H, J = 6, 3 Hz), 4.11 (dd, 1H, J = 7.5, 3.6 Hz), 3.58
(m, 2H), 3.50 (d, 1.5H, J = 6 Hz), 3.17 (d, 0.5H, J = 2.7 Hz), 1.47 (s, 3H),
1.27 (s, 3H).

54
H.-J. Kim et al.
5-Biotinylaminoallyl-2ꢁ,3ꢁ-O-isopropylidene-uridine (14)
To a mixture of the amine 13 (555 mg, 1.64 mmol) and biotin
N -hydroxysuccinimide (586 mg, 1.71 mmol) in pyridine (10 mL) was added
triethylamine (0.46 mL, 3.28 mmol) and the mixture was stirred at RT for
2 hours. The volatiles were removed by evaporation, and the residue was
resolved by flash chromatography (silica, gradient CH2Cl2:MeOH = 8:1
1
to 6:1 to 4:1 to 3:1) to give 14 as a white solid (670 mg, 72%). H NMR
(300 MHz, DMSO-d6) δ 11.48 (s, 1H), 7.99 (t, 1H, J = 5.4 Hz), 7.94 (s, 1H),
6.3∼6.4 (m, 3H), 6.10 (d, 1H, J = 15.9 Hz), 5.85 (d, 1H, J = 2.7 Hz), 5.18
(t, 1H, J = 5.1 Hz), 4.92 (dd, 1H, J = 6.3, 2.7 Hz), 4.76 (dd, 1H, J = 6.3,
3.3 Hz), 4.28 (m, 1H), 4.09 (m, 1H), 3.73 (m, 1H), 3.59 (m, 1H), 3.0∼3.1
(m, 3H), 2.5∼2.9 (m, 3H), 2.0∼2.1 (m, 2H), 1.4∼1.6 (m, 5H), 1.2∼1.4
(m, 5H), 1.17 (t, 2H, J = 7.5 Hz).
5-Biotinylaminoallyl-2ꢁ,3ꢁ-O-isopropylidene-uridine
5ꢁ-monophosphate (15)
To a suspension of compound 14 (565 mg, 1.00 mmol) in triethylphos-
phate (10 mL) was added POCl3 (270 µL, 3.00 mmol) at 0◦C. The mixture
was stirred at 0◦C for 18 hours, diluted with water (20 mL), neutralized
by 1 N NaOH solution, and extracted with ethyl acetate (2 × 20 mL).
The aqueous phase was evaporated and the residue was purified by flash
chromatography (silica, gradient iPrOH:NH4OH:H2O = 6:1:1 to 5:1:1 to
1
4:1:1) to give 15 as a light tan solid (380 mg, 56%). H NMR (300 MHz,
D2O) δ 7.73 (s, 1H), 6.3 (m, 1H), 6.17 (d, 1H, J = 16.2 Hz), 5.79 (br s, 1H),
4.8∼5.0 (m, 2H), 4.46 (m, 2H), 4.27 (m, 1H), 3.99 (m, 3H), 3.79 (m, 1H),
3.19 (br s, 1H), 2.6∼2.9 (m, 2H), 2.19 (m, 2H), 1.2∼1.7 (12H); 31P NMR
(D2O) δ –0.049 (s).
5-Biotinylaminoallyl-uridine 5’-monophosphate (9)
A solution of compound 15 (380 mg, 0.56 mmol) in H2O (60 mL) and
conc. HCl (0.3 mL) was stirred at RT overnight. The mixture was neutralized
aqueous NaOH (1 N) and evaporated. The residue was purified by flash
chromatography (silica, gradient iPrOH:NH4OH:H2O = 6:1:1 to 5:2:2) to
give 9 as a light tan solid (230 mg, 64%). The analytical data are shown for
compound 9 above.
5-Biotinylaminoallyl-uridine 5ꢁ-phosphomorpholidate (16)
A mixture of compound 9 (230 mg, 0.36 mmol) and morpholine
(0.13 mL, 1.4 mmol) in t-BuOH (5 mL) and H2O (5 mL) was heated to
reflux. To this mixture was added a solution of DCC (309 mg, 1.50 mmol)
in t-BuOH (6 mL). The resulting mixture was refluxed for 4 hours. After

Pyrophosphates for In Vitro Selection
55
cooling to RT, the mixture was filtered through Celite and washed with
t-BuOH. The combined filtrate was evaporated and dissolved in H2O (20
mL) and extracted with ethyl ether (2 × 20 mL). The aqueous phase
was evaporated and the residue was dispersed in methanol (3 mL) and
treated with ethyl ether (30 mL). The ether layer was decanted away and
the resulting solid was dried under high vacuum to give 16 as a light
1
tan solid (180 mg, 52%). H NMR (300 MHz, D2O) δ 7.72 (s, 1H), 6.23
(m, 2H), 5.83 (d, 1H, J = 5.1 Hz), 4.42 (m, 1H), 4.12–4.46 (m, 4H), 3.93
(m, 2H), 3.76 (m, 2H), 2.79–3.67 (m, 18H), 2.59 (d, 1H, J = 12.6 Hz),
2.15 (m, 1H), 1.03–1.78 (m, 26H); 31P NMR (D2O) δ 7.31.
Flavin-5-biotinylaminoallyl-uridine dinucleotide (1)
Method 1
FMN as its pyridinium salt (45 mg, 0.084 mmol) in pyridine (2 mL)
was treated with trioctylamine (0.037 mL, 0.084 mmol) and stirred for
10 minutes and evaporated. The residue was dissolved in pyridine (2 mL)
and evaporated to dryness. To this solid in flask (10 mL) was added
compound 16 (90 mg, 0.093 mmol) and treated with pyridine (2 mL)
and 0.45 M tetrazole in MeCN (0.67 mL). The resulting mixture was
stirred at RT and periodically analyzed by analytical ion-exchange HPLC
(Dionex DNAPac PA-100 4 × 250-mm column, eluent A = 25 mM TEAA,
pH 7, eluent B = 100 mM NaCl in 25 mM TEAA, pH 7, gradient from 0
to 100% B in 30 minutes; morpholidate 16: Rt = 3 minutes; FMN: Rt = 12
minutes, plus small peaks at 13 minutes; pyrophosphate 1: Rt = 14 minutes).
After 6 days, all of the morpholidate 16 had been comsumed. The volatiles
were removed by evaporation and the residue was dissolved in water (10
mL) and filtered through a 0.2-µm filter. The filtrate was purified by ion-
exchange HPLC (GE Healthcare HiPrep 16/10 DEAE FF column, eluent
A = 10 mM NH4HCO3, eluent B = 500 mM NH4HCO3, gradient from 0
to 100% B in 50 minutes, flow rate 3 mL/minutes). The fraction between
28 and 36 minutes was collected and lyophilized and further purified by
reverse-phase HPLC (Waters Nova Pak HR C18 6-µm 19 × 300-mm column,
eluent A = 10% CH3CN in 25 mM TEAA pH 7, eluent B = 60% CH3CN
in 25 mM TEAA pH 7, gradient from 0 to 50% B in 25 minutes, flow rate 5
mL/minutes). The peak at 21 minutes was collected and lyophilized to give
an orange solid (5 mg, 5%). Mass spectrometry (HPLC/ESI-MS) observed
a single charged [M-H]− at m/e 1042.2.
Method 2
A mixture of FMN pyridinium salt (45 mg, 0.084 mmol), morpholidate
16 (90 mg, 0.093) and sodium sulfate (26 mg, 0.183) in 0.2 M MnCl2 in
formamide (0.7 mL, 0.14 mmol) was stirred at RT for 18 h. Analytical ion-
exchange HPLC (conditions as in method 1) showed complete consump-

56
H.-J. Kim et al.
tion of the morpholidate 16. The mixture was diluted with CH3CN (10 mL).
The resulting yellow solid was filtered and dissolved in water (10 mL) and
filtered through a 0.2-µm filter. The filtrate was purified by ion-exchange
HPLC and reverse-phase HPLC (conditions as in method 1) to give an
orange solid (10 mg, 10%) still containing some excess triethylammonium
1
salt. H NMR (300 MHz, D2O) δ 7.74 (s, 2H), 7.63 (s, 1H), 6.11 (m, 1H),
5.96 (d, 1H, J = 15.9 Hz), 5.66 (d, 1H, J = 4.8 Hz), 3.70-4.43 (m, 14H), 3.63
(m, 2H), 3.13 (m, 1H), 3.04 (q, Et3N), 2.76 (dd, 1H, J = 5.1, 13.2 Hz), 2.54
(d, 1H, J = 12.9 Hz), 2.08 (t, 2H, J = 7.1 Hz), 1.32-1.52 (m, 6H) (s, 1H),
1.13 (t, Et3N); 31P NMR (D2O) δ −10.72 (m), −11.46 (m).
REFERENCES AND NOTES
1. Cech, T.R.; Zaug, A.J.; Grabowski, P.J. In vitro splicing of the ribosomal-RNA precursor of
tetrahymena, involvement of a guanosine nucleotide in the excision of the intervening sequence.
Cell 1981, 27, 487–496.
2. Guerrier-Takada, C.; Gardiner, K.; Marsh, T.; Pace, N.; Altman, S. The RNA moiety of ribonuclease-P
is the catalytic subunit of the enzyme. Cell 1983, 35, 849–857.
3. Gilbert, W. The RNA world. Nature 1986, 319, 818.
4. Benner, S.A.; Ellington, A.D.; Tauer, A. Modern metabolism as a palimpsest of the RNA world. Proc.
Natl. Acad. Sci.USA 1989, 86, 7054–7058.
5. Benner, S.A.; Sismour, A.M. Synthetic biology. Nat. Rev. Genet, 2005, 6, 533–543.
6. Ekland, E.H.; Szostak, J.W.; Bartel, D.P. Structurally complex and highly active RNA ligases derived
from random RNA sequences. Science 1995, 269, 364–370.
7. Winkler, W.; Cohen-Chalamish, S.; Breaker, R.R. An mRNA structure that controls gene expression
by binding FMN. Proc. Natl. Acad. Sci. USA 2002, 99, 15908–15913.
8. Anderson, P.C.; Mecozzi, S. Identification of a 14mer RNA that recognizes and binds flavin
mononucleotide with high affinity. Nucleic Acids Res. 2005, 33, 6992–6999.
9. Moffatt, J.G.; Khorana, H.G. Nucleoside polyphosphates, VIII: new and improved syntheses of uri-
dine diphosphate glucose and flavin adenine dinucleotide using nucleoside-5ꢁ phosphoramidates. J.
Am. Chem. Soc. 1958, 80, 3756–3761.
10. McCormick, D.B.; Chassy, B.M.; Tsibris, J.C.M. Coenzyme specificity of image-amino acid oxidase for
the adenylate moiety of FAD. Biochim. Biophys. Acta 1964, 89, 447–452.
11. Zatorski, A.; Watanabe, K.A.; Carr, S.F.; Goldstein, B.M.; Pankiewicz, K.W. Chemical synthesis of
benzamide adenine dinucleotide: inhibition of inosine monophosphate dehydrogenase (type I and
II). J. Med. Chem. 1996, 39, 2422–2426.
12. Langer, P.R.; Waldrop, A.A.; Ward, D.C. Enzymatic synthesis of biotin-labeled polynucleotides: novel
nucleic acid affinity probes. Proc. Natl. Acad. Sci. USA 1981, 78, 6633–6637.
13. Dey, S.; Sheppard, T.L. Ketone-DNA: A versatile postsynthetic DNA decoration platform. Org. Lett.
2001, 3, 3983–3986.
14. Lee, L.V.; Mitchell, M.L.; Huang, S.-J.; Fokin, V.V.; Sharpless, K.B.; Wong, C.-H. A potent and highly
selective inhibitor of human α-1,3-fucosyltransferase via Click chemistry. J. Am. Chem. Soc. 2003, 125,
9588–9589.
15. Wagner, G.K.; Guse, A.H.; Potter, B.V.L. Rapid synthetic route toward structurally modified deriva-
tives of cyclic adenosine 5ꢁ-diphosphate ribose. J. Org. Chem. 2005, 70, 4810–4819.