The reaction of activated RNA species with aqueous fluoride ion: A convenient synthesis of nucleotide 5′-phosphorofluoridates and a note on the mechanism

Article abstract of DOI:10.1016/j.tetlet.2014.01.051

The chemistry of 5′-phosphorimidazolides of ribonucleosides is extended to include their reaction with alkali metal fluorides in aqueous solution. High yields of 5′-phosphorofluoridates are formed, especially with potassium fluoride, but no detectable oligomerization products were formed. A combination of HPLC, mass spectrometry, synthesis, kinetics, and NMR confirms the identities of the products. Judicious control of pH leads to higher yields in shorter reaction times. This new methodology contrasts favorably with other synthetic routes involving non-aqueous chemistry or aqueous chemistry with a nucleotide triphosphate.

Full text of DOI:10.1016/j.tetlet.2014.01.051

Tetrahedron Letters 55 (2014) 1464–1466
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
The reaction of activated RNA species with aqueous fluoride ion:
a convenient synthesis of nucleotide 50-phosphorofluoridates
and a note on the mechanism
a
b
c
Michael F. Aldersley a, , Prakash C. Joshi , Herbert M. Schwartz , Anthony J. Kirby
⇑
a New York Center for Astrobiology and the Department of Chemistry and Chemical Biology, Rensselaer Polytechnic Institute, Troy, NY 12180, USA
b Department of Chemistry and Chemical Biology, Rensselaer Polytechnic Institute, Troy, NY 12180, USA
c University Chemical Laboratory, Cambridge CB2 1EW, UK
a r t i c l e i n f o
a b s t r a c t
The chemistry of 50-phosphorimidazolides of ribonucleosides is extended to include their reaction with
alkali metal fluorides in aqueous solution. High yields of 50-phosphorofluoridates are formed, especially
with potassium fluoride, but no detectable oligomerization products were formed. A combination of
HPLC, mass spectrometry, synthesis, kinetics, and NMR confirms the identities of the products. Judicious
control of pH leads to higher yields in shorter reaction times. This new methodology contrasts favorably
with other synthetic routes involving non-aqueous chemistry or aqueous chemistry with a nucleotide
triphosphate.
Article history:
Received 7 November 2013
Revised 8 January 2014
Accepted 14 January 2014
Available online 20 January 2014
Keywords:
Activated nucleotide
50-Phosphorofluoridate
Alkali metal fluorides
Ó 2014 Elsevier Ltd. All rights reserved.
Phosphorofluoridates have for many years played an important
role as probes in the elucidation of enzyme mechanism.1,2 Their
role as a proxy for phosphates, phosphoramidates, etc., is enhanced
by their unique NMR properties, the interpretation of which can
probe further into the nature of an active site. Additionally, the
reaction between ATP and fluoride has been used as a model for
enzymatic nucleotide fluorolysis.3
Aqueous solutions of ImpN can be hydrolyzed by a relatively
simple process in which the NMP is the major (>99%) product.
Our preliminary experiments showed that in the presence of fluo-
ride, the activated nucleotides did indeed predominantly form the
50-phosphorofluoridates, which were initially characterized by
HPLC and high resolution mass spectrometry (Table 1). Measure-
ments of the rates of reaction in 1 M sodium fluoride (measured
pH 8.7) of ImpU gave identical pseudo first order rate constants
for the loss of activated nucleotide and the formation of UMPF (k1-
(F) = 8.31 ꢀ 10ꢁ7 sꢁ1). This is compared with an extrapolated
k1(H2O) = 1.37 ꢀ 10ꢁ7 sꢁ1 for hydrolysis.8
Literature syntheses of 50-phosphorofluoridates of nucleosides
(NMPF) typically involve either the use of aqueous ATP, for exam-
ple with sodium fluoride, albeit with very low yields (ca. 4%):4 or
the reaction of a nucleotide 50-monophosphate (NMP) with the
Sanger reagent, 2,4-dinitrofluorobenzene (DNFB) in a non-aqueous
medium.5 Since activated ribonucleotides such as the 5-phosphor-
imidazolides (ImpN) are readily synthesized in our laboratories,
with yields in excess of 85% (with 99.5% purity) from the
corresponding NMP,6 we decided to investigate their reaction with
aqueous fluoride solutions (Scheme 1). For comparison purposes,
we repeated the two standard methods for the synthesis of AMPF
referred to above. Other phosphorimidazolides and phosphoroflu-
oridates are conveniently synthesized by the displacement of 2,
4-dinitrophenoxide from 2,4-dinitrophenyl phosphotriesters.7 HF
is a stronger acid than imidazolium by some 4 pKa units, suggesting
that the planned reaction should proceed efficiently.
The uncomplicated kinetics, the reactions being followed by re-
verse phase HPLC, was consistent with a straightforward reaction
with no intermediates. Experiments with ImpA using lower con-
centrations of fluoride ion (0.2 M and 0.5 M NaF) gave, as expected,
reduced yields of AMPF, with a corresponding increase in the com-
peting hydrolysis to AMP (Fig. 1). In the presence of 1 M fluoride,
O
O
O
O
1M MF(aq)
Base
F
N
N
Base
P
P
O
O
-O
-O
X
HO
HO
X
⇑
Corresponding author. Tel.: +1 518 276 4080; fax: +1 518 276 4887.
E-mail address: alderm@rpi.edu (M.F. Aldersley).
Scheme 1. Activated nucleotides ImpN used in this work and the product NMPF.
X = OH, Base = Adenine, Uracil, Guanine, Cytosine; X = H, Base = Adenine.
0040-4039/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tetlet.2014.01.051

M. F. Aldersley et al. / Tetrahedron Letters 55 (2014) 1464–1466
1465
Table 1
Accurate mass dataa for the NMPF synthesized
Compound
Formula
[m/z]+
[m/z]ꢁ
[m/z]+, Sodiated species
Calculated
Found
Calculated
Found
Formula
Calculated
Found
AMPF
UMPF
CMPF
GMPF
dAMPF
C
10H13FN5O6P
350.06602
327.03881
326.05479
366.06094
334.07111
350.06602
327.03880
326.05478
366.06093
334.07113
348.05147
325.02425
324.04024
364.04639
332.05656
348.05145
325.02427
324.04025
364.04637
332.05655
C10H13FN5NaO6P
C9H12FN2NaO8P
C9H13FN3NaO7P
C10H13FN5NaO7P
C10H13FN5NaO5P
371.04797
349.02075
348.03674
388.04288
356.05305
371.04793
349.02071
348.03677
388.04286
356.05310
C9H12FN2O8P
C9H13FN3O7P
C
C
10H13FN5O7P
10H13FN5O5P
a
Obtained via a thermo LTQ orbitrap XL mass spectrometer.
very low. The presence of bADPF could not be assessed using the
HPLC methodology, optimized for NMPF detection and involving a
gradient system (1%/min, 1 mL/min) of 0.2% formic acid (Buffer A)
and 0.2% formic acid/30% acetonitrile (Buffer B), with detection at
260 nm using 1 mM solutions. Larger NMPF samples were col-
lected via semi-preparative reverse phase HPLC from 15 mM
solutions.
The data in Table 2 clearly demonstrate that the reagent of
choice for the chemistry in Scheme 1 is 1 M potassium fluoride,
which gives an average yield from the five syntheses of over 95%.
Noteworthy it is the smooth reaction with ImpG, a compound
which forms a thixotropic gel in aqueous solutions. This inhibits
reaction in clay-catalyzed RNA dimer and oligomer synthesis,6
but here the yield of GMPF is comparable to that from the other
reactions, even in the presence of potassium ions, which are known
to encourage G-quadruplex formation and so enhance gel
formation.9
Figure 1. Progress of the reaction between NaF and ImpA, in water at pH 8.7 and
25 °C.
Table 2
Yields of NMPF produced via Scheme 1
A detailed NMR study (13C, 1H, 31P, 19F) was undertaken using
AMPF to confirm that the products from the reaction of 1 M NaF
with ImpA and the use of the Sanger reagent (Scheme 2), are iden-
tical. In our hands, the product from this latter reaction proved
difficult to separate in all cases save that from adenosine 50-mono-
phosphate. The very large JP–F coupling (930 Hz)3 further con-
firmed the identical outcomes of the two synthetic methods,
with the 1H and 13C spectra being virtually the same as those of
adenosine 50-monophosphate.
The monotonic increase in yield from LiF to NaF to KF with the
activated nucleotides (Table 2) is the reverse of the salt effects ob-
served with activated nucleotides in the presence of clay minerals,
where oligomer formation decreases in that order.10 This decrease
is mainly the result of increased hydrolysis of ImpN, to over 70% in
the case of potassium chloride. Complex formation between the al-
kali metal ion and both the 30-OH and the 50-phosphorimidazolide
groups may explain this: modeling suggests that such complex for-
mation with the alkali metal ions becomes much less favorable as
their size increases.11
Notwithstanding the relatively high pH at which the reaction is
carried out, and the expected ionization of some 10% of the imide
proton from ImpU (pKa = 9.7), the yields of UMPF are comparable
to those from other ImpN. This suggests that the imide anion is
too remote to interfere with the basic reaction of the 50-phosphor-
imidazolide with fluoride ion. However, the rate of formation of
UMPF in 1 M NaF is significantly slower than that of AMPF
(k1(F) = 6.03 ꢀ 10ꢁ6 sꢁ1). A factor of 7 is consistent with the reac-
tion of fluoride with ImpU being slower because of the reduced
reactivity of the ImpU dianion. This special case apart, we consider
that the primary synthetic reaction of Scheme 1 involves nucleo-
philic attack by fluoride on the substrate zwitterion (zw, Scheme 3),
with the displacement of neutral imidazole: since the zwitterion is
known to carry the main spontaneous hydrolysis reaction of ImpN
below pH 10, and fluoride is a much better nucleophile than water
for the POꢁ2 group.12 For example, the measured rate constants at
37 °C for the hydrolysis of the zwitterion and monoanion of ImpG
ImpN
pH
%NMP
%NMPF
LiF
NaF
KF
LiF
NaF
KF
LiF
NaF
KF
LiF
NaF
KF
LiF
NaF
KF
NaF
NaF
NaF
NaF
NaF
ImpA
ImpA
ImpA
ImpU
ImpU
ImpU
ImpC
ImpC
ImpC
ImpG
ImpG
ImpG
ImpdA
ImpdA
ImpdA
ATP
8.6
8.7
8.6
8.6
8.7
8.6
8.6
8.7
8.6
8.6
8.7
8.5
8.7
8.7
8.6
5.5a
8.7b
7.0
5.9
4.4
27.9
8.4
6.5
18.7
4.9
2.9
14.2
4.6
2.1
10.5
4.7
3.2
12.9
5.7
4.7
8.6
0.6
<1%
<1%
<1%
70.7
90.9
93.5
76.3
94.0
96.1
85.4
94.4
97.6
67.2
93.7
95.8
87.1
94.3
95.3
1.8
ATP
0.03
>98%
>98%
>98%
ImpA
ImpA
ImpA
a
Yields of ADP and unreacted ATP after 10 days were 48% and 42%, respectively.
Yields of ADP and unreacted ATP after 10 days were 85% and 14%, respectively.
b
hydrolysis was slower than fluorolysis by almost an order of
magnitude.
Encouraged by these initial results, we examined the reactions
of 1 M alkali metal fluorides with the four activated ribonucleo-
tides and additionally with ImpdA, the 50-phosphorimidazolide
of 20-deoxyadenosine (Scheme 1). The aqueous fluoride reactions
were monitored by reverse phase HPLC over ten days at 25 °C.
The reactions of Scheme 1 gave the product yields shown in
Table 2. Also listed are the yields of AMPF from similar experi-
ments carried out at 25 °C with ATP using 1 M NaF buffered at
pH 5.5 and 8.7. We observed a 60-fold increase in the yield at
the lower pH, although this was accompanied by a considerable
increase in the breakdown of ATP to ADP, the actual yield was still

1466
M. F. Aldersley et al. / Tetrahedron Letters 55 (2014) 1464–1466
F
O
O
HO
HO
F
P
P
O
O
N
N
O2N
Bu3N, DMF
NO2
-O
O
O
NH2
NH2
N
N
N
N
N
N
HO
OH
HO
OH
Scheme 2. AMPF from AMP and 2,4-DNFB.
O
O
O
O
HA
+HN
N
Base
Base
N
N
P
P
O
O
-O
-O
zw
HO
X
HO
X
Scheme 3. Site of protonation of 50-phosphorimidazolide.
Table 3
and Chemical Biology, for their support and interest in this
research. We are also grateful to Dr. Dmitri V. Zagorevskii for mass
spectrometry measurements. This research was supported by
Grant NNA09DA80A from the NASA Astrobiology Institute.
Effect of pH upon yielda and reaction rate for ImpA in 1 M NaF
pH
k1(F), sꢁ1
t1/2, min
Relative rate
%AMPF
4.4b
5.7b
7.0b
8.7
9.72 ꢀ 10ꢁ3c
2.75 ꢀ 10ꢁ3
1.57 ꢀ 10ꢁ3
6.17 ꢀ 10ꢁ6
1.3
4.2
7.4
1873
1576
446
255
1
>98
>98
>98
91
Supplementary data
a
b
c
Supplementary data associated with this article (mass
spectroscopic, NMR, and kinetic methods) can be found, in the
online version, at http://dx.doi.org/10.1016/j.tetlet.2014.01.051.
Yields and reaction times are currently not optimized.
The measurement by HPLC isocratically using 22% buffer B and 1.5 mL/min.
Measured as 9.23 ꢀ 10ꢁ3 sꢁ1 and corrected for the effect of pH upon [Fꢁ].
References and notes
are 1.55 ꢀ 10ꢁ4 sꢁ1 and 8.90 ꢀ 10ꢁ8 sꢁ1, respectively, with a pKa of
the imidazolium group of zw (Scheme 3) of 6.01.13
1. Withers, S. G.; Madsen, N. B. Biochem. Biophys. Res. Commun. 1980, 97, 513–519.
2. Guranowski, A.; Wojdyla, A. M.; Pietrowska-Borek, M.; Bieganowski, P.; Khurs,
E. N.; Cliff, M. J.; Blackburn, G. M.; Blaziak, D.; Stec, W. J. FEBS Lett. 2008, 582,
3152–3158.
3. London, R. E.; Gabel, S. A. Arch. Biochem. Biophys. 1996, 334, 332–340.
4. Alvarez, R.; Moore, T. C.; Vandepeute J. Plant Physiol. 1974, 53, 144–148.
5. Wittmann, R. Angew. Chem., Int. Ed 1962, 74, 214–215.
To test this idea, a series of reactions were run with ꢂ1.5 mM
ImpA at lower pH, using buffered 1 M NaF solutions. Dramatic rate
enhancements and even a further increase in yield were observed
under these conditions, designed to increase the concentration of
the reactive zwitterion without significantly reducing the concen-
tration of fluoride anion (Tables 2 and 3). The results are consistent
with the proposed mechanism.
More strongly acidic 1 M fluoride buffer solutions, for example
with pH <4.4, have the disadvantage of containing a significant
concentration (already ꢂ0.05 M at pH 4.4) of HF.14 Yields and reac-
tion times are yet to be optimized in the light of this factor, but the
simplicity of a system involving 1 M KF ‘self-buffered’ near pH 7,
with no optically significant buffer components during preparative
HPLC purification, looks very promising.
6. Joshi, P. C.; Aldersley, M. F.; Zagorevskii, D. V.; Ferris, J. P. Nucleosides
Nucleotides Nucleic 2012, 31, 536–566.
7. Kirby, A. J.; Younas, M. J. Chem. Soc., Sec. B 1970, 1165–1172. For the fluorolysis
reaction of the monoanion of methyl 2,4-dinitrophenyl phosphate at 39 °C,
k2(F) is 3.2 ꢀ 10ꢁ5 sꢁ1. Compare k0 = 2.95 ꢀ 10ꢁ7 sꢁ1 for the same system.
8. Ruzicka, F. J.; Frey, P. A. Bioorg. Chem. 1993, 21, 238–248.
9. Yu, Y.; Nakamura, D.; DeBoyace, K.; Neisius, A. W.; McGown, L. B. J. Phys. Chem.
B 2008, 112, 1130–1134.
10. Joshi, P. C.; Aldersley, M. F. J. Mol. Evol. 2013, 76, 371–379.
11. Spartan ’10, Wavefunction Inc, Irvine, CA.
12. Lonnberg, T.; Ora, M.; Lönnberg, H. Mini-Rev. Org. Chem. 2010, 7, 33–43. The
extrapolated rate constant for the hydrolysis of ImpU observed in this work, at
pH 8.7 and 25 °C, is consistent with that quoted by these authors for the
hydrolysis of the ImpU zwitterion (9.4 ꢀ 10ꢁ5 sꢁ1 at 27 °C), given the pKa close
to 6.0 for the imidazolium group of zw (Scheme 3).
Acknowledgments
13. Kanavarioti, A.; Bernasconi, C. F.; Doodokyan, D. L.; Alberas, D. J. J. Am. Chem.
Soc. 1989, 111, 7247–7257.
14. Gerken, J. B. J. Fluorine Chem. 2011, 132, 68–70.
The authors are grateful to Professor Douglas Whittet, Director,
New York Center for Astrobiology, Professor James Ferris and
Professor Curt Breneman, both from the Department of Chemistry