Kinetics and mechanism of the defluorination of 8-fluoropurine nucleosides in basic and acidic media

Article abstract of DOI:10.1016/j.jfluchem.2006.06.006

For investigating the stability of C(8)-fluorine bond in 8-fluoropurine nucleosides some protected 8-fluoroguanosine, 8-fluoroinosine and 8-fluoroadenosine derivatives were prepared by direct fluorination of acetyl-protected purine nucleosides with elemental fluorine in solvents such as chloroform, acetonitrile and nitromethane. Fluorination reactions conducted in chloroform medium gave better yields of 8-fluoropurines. The fluorination yields were slightly lower when acetonitrile or nitromethane was used as solvent, but the product purification was found to be much easier. When the synthesized, protected fluoronucleosides were subjected to standard basic (NH₃ in methanol or 2-propanol) and acidic (HCl in methanol) deprotection conditions relevant to nucleoside chemistry, an efficient defluorination reaction took place. The kinetics of these defluorination reactions were conveniently followed, under pseudo-first-order reaction conditions, using ¹⁹F NMR spectroscopy. ¹H NMR, LC-MS and mass spectroscopy identified the products of the kinetic reaction mixtures. The defluorination reaction rate constants (k_obs) in basic media depended upon the electron density at C(8) while the k_obs data in acidic medium were determined by the pK_a of N⁷. An addition-elimination based mechanism (S_NAr) has been proposed for the defluorination reactions of these 8-fluoropurine nucleosides.

Full text of DOI:10.1016/j.jfluchem.2006.06.006

Journal of Fluorine Chemistry 127 (2006) 1175–1187
www.elsevier.com/locate/fluor
Kinetics and mechanism of the defluorination of
8-fluoropurine nucleosides in basic and acidic media
*
Jie Liu, Jorge R. Barrio, Nagichettiar Satyamurthy
Department of Molecular and Medical Pharmacology, David Geffen School of Medicine,
University of California, Los Angeles, CA 900950-6948, USA
Received 2 May 2006; received in revised form 2 June 2006; accepted 2 June 2006
Available online 9 June 2006
Abstract
For investigating the stability of C(8)–fluorine bond in 8-fluoropurine nucleosides some protected 8-fluoroguanosine, 8-fluoroinosine and 8-
fluoroadenosine derivatives were prepared by direct fluorination of acetyl-protected purine nucleosides with elemental fluorine in solvents such as
chloroform, acetonitrile and nitromethane. Fluorination reactions conducted in chloroform medium gave better yields of 8-fluoropurines. The
fluorination yields were slightly lower when acetonitrile or nitromethane was used as solvent, but the product purification was found to be much
easier. When the synthesized, protected fluoronucleosides were subjected to standard basic (NH3 in methanol or 2-propanol) and acidic (HCl in
methanol) deprotection conditions relevant to nucleoside chemistry, an efficient defluorination reaction took place. The kinetics of these
defluorination reactions were conveniently followed, under pseudo-first-order reaction conditions, using 19F NMR spectroscopy. 1H NMR, LC–MS
and mass spectroscopy identified the products of the kinetic reaction mixtures. The defluorination reaction rate constants (kobs) in basic media
depended upon the electron density at C(8) while the kobs data in acidic medium were determined by the pKa of N7. An addition–elimination based
mechanism (SNAr) has been proposed for the defluorination reactions of these 8-fluoropurine nucleosides.
# 2006 Elsevier B.V. All rights reserved.
Keywords: Fluorination of purine nucleosides; Defluorination of 8-fluoropurine nucleoside; Stability of 8-fluoropurine nucleoside; Addition–elimination
mechanism and 8-fluoropurine nucleosides
1. Introduction
determination of HSV1-tk mediated gene expression in mice
using positron emission tomography (PET) [6].
Purine nucleosides with C(8) substituents have been known to
exhibit significant biological activities [1]. During the past 25
years, reports have appeared on the pharmacological properties
of 8-chloro- and 8-bromopurine nucleosides [2]. By contrast,
little data are available on the behavior, in biological systems, of
8-fluoropurine nucleosides, mainly because of difficulties
associated with their synthesis [3]. A simple procedure for the
synthesis of 8-fluoropurine nucleosides with elemental fluorine
has previously been developed in our laboratory [4]. Using this
direct fluorination technique with F2 purine nucleosides, both
protected as well as unprotected, are selectively fluorinated at
C(8) position. We have extended this convenient synthesis to the
preparationofseveral18F-labeled8-fluoroguanosinenucleosides
[5] and have utilized them successfully in non-invasive in vivo
Significant defluorination was observed when the protected
8-fluoropurine nucleosides were subjected to the standard
deprotection conditions in basic media [1]. For example,
deacetylation of N2-acetyl-20,30,50-tri-O-acetyl-8-fluoroguano-
sine with methanolic ammonia yielded 8-fluoroguanosine
(50%) along with a defluorinated derivative, namely 8-
methoxyguanosine (50%) [4b]. An analogous deprotection of
20,30,50-tri-O-acetyl-8-fluoroadenosine suffered a complete
defluorination [7,8]. Further, this 8-fluoroadenosine analog is
reported to be unstable in acidic media [8]. In fact, deprotection
of tri-O-acetyl-8-fluoroadenosine without concomitant defluor-
ination could be achieved only via an enzymatic procedure [7].
In view of the potential importance of 8-fluoropurine nucleo-
sides in the field of medicinal chemistry, it would be useful to
understand the limitations of the general deprotection condi-
tions employed for their preparation. In this investigation, we
have synthesized certain protected 8-fluoro analogs of
guanosine, inosine and adenosine, treated them under basic
* Corresponding author. Tel.: +1 310 825 7190; fax: +1 310 825 4517.
E-mail address: nsaty@mednet.ucla.edu (N. Satyamurthy).
0022-1139/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jfluchem.2006.06.006

1176
J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
and acidic deprotection conditions relevant to nucleoside
chemistry, and have evaluated the rates of defluorination and
the factors influencing the stability of the C(8)–fluorine bond,
the results of which are reported herein.
conveniently separated by preparative HPLC. Similarly, the 8-
chloro derivative 3a was also isolated from fluorination of 1a in
chloroform.
As expected, the formation of the 8-chloropurine side
products in the fluorination reactions could be avoided by using
solvents such as acetonitrile or nitromethane. For example,
when the purine derivative 1b in acetonitrile was reacted with
F2, the product 2b was obtained free from any contamination of
the chlorinated analog 3b. In general, conducting the
fluorination reactions in acetonitrile or nitromethane yielded
the 8-fluoro products in lower yields, in comparison with those
obtained when chloroform was used as the reaction medium,
but the overall purification process was found to be easier. The
fluorination of triacetylguanosine analog 1a could not be
conducted in acetonitrile because of its poor solubility.
Interestingly, when chloroform stabilized with 1% EtOH
(Aldrich) was used as the solvent in the fluorination reaction of
20,30,50-tri-O-acetyl-N1-methylinosine (1c), 20,30,50-tri-O-
acetyl-8-ethoxy-N1-methylinosine (7) was produced as the
major product (Scheme 2), as determined by UV, 1H NMR and
mass spectroscopy. The fact that the isolated 8-fluoro product
2c remained stable in EtOH for several hours suggests that the
8-ethoxy product 7 was formed directly from the starting
material 1c during fluorination reaction. Using chloroform
stabilized with amylene (Aldrich) as the solvent for fluorination
avoided the formation of this side product.
2. Results and discussion
2.1. Chemistry
The C(8)–fluoropurine nucleosides used in this study 2a–c
and 5a,b were synthesized (Scheme 1) by following reported
procedures with some modifications [4]. The acetylated
guanosine or inosine derivatives 1a–c and 4a,b [9–12] were
reacted at room temperature with elemental fluorine (1% in
helium). Previously, it was recognized that organic solvents
such as chloroform, acetonitrile or nitromethane, which act also
as radical scavengers, would facilitate electrophilic fluorination
by reducing non-selective radical side reactions [13]. In our
hands, the fluorination reactions using chloroform as the
solvent led to the formation of 8-chloropurine nucleosides as
side products in amounts of 16–30% of the total halogenated
products, as a result probably of purinyl radical abstraction of
chlorine from the solvent [14]. The C(8)-chlorinated products
3c, 6a and 6b were separated from their corresponding 8-fluoro
counterparts 2c, 5a and 5b by semi-preparative HPLC.
Fluorination of 1b in CHCl3 gave an inseparable mixture of
2b and 3b even by several different semi-preparative HPLC
conditions. However, treatment of the product mixture with
HCl in methanol selectively converted the 8-fluoro product 2b
into the corresponding 8-methoxy analog 9b leaving the 8-
chloro derivative 3b intact. This new product mixture was then
2.2. Defluorination studies
The stability of the C(8)–fluorine bond in nucleosides 2a–c,
5a,b and 12 [4] was assessed by monitoring the reaction rates of
Scheme 1. Preparation of 8-fluoronucleosides.

J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
1177
It was deemed probable that a partial or complete
deacetylation of the protected nucleosides 2a–c, 5a,b and 12
would also concurrently occur during the defluorination
reactions under basic conditions. This was found to be the case
and supported by liquid chromatography–mass spectrometry
(LC–MS) analysis of selected dedicated kinetic reaction
mixtures at various time points of the reaction. The data
summarized in Table 2 clearly show that partial deacetylation is a
relatively fast process in methanolic ammonia. Early samples
showed the presence of di- and monoacetylated products but no
attempt was made to distinguish which of the acetyl groups got
cleaved first. The 19F NMR data from the kinetic reactions also
supported the LC–MS results and accordingly exhibited closely
spaced C(8)–fluorine signals corresponding to various acetylated
forms of a given fluoronucleoside. The kobs data given in Table 1,
therefore, constitute an overall defluorination rate as assessed by
the integration of all the C(8)–fluorine signals of different
acetylated forms in the reaction mixture. An overall linear
dependence of defluorination rates with time (R2 > 0.92) for all
the fluoropurines was observed. This finding strongly suggests
that the relative rates of defluorination of all possible acetylated
forms of a given nucleoside are of the same order of magnitude.
The defluorination reaction mixture, after three to four half-
lives of kinetic data collection, was allowed to undergo
complete deacetylation by letting it sit for at least 24 h or by
treatment with 2 M NH3 in methanol. The fully deacetylated
mixture was then subjected to the semi-preparative HPLC
separation and the products (Scheme 3) were isolated and
Scheme 2. Fluorination of nucleoside in chloroform/ethanol solution.
their defluorination at 28 ꢀ 0.1 8C in basic (NH3 in methanol or
2-propanol) and acidic (HCl in methanol) media using 19F
NMR spectroscopy. The 19F NMR signal of C(8)–fluorine in
these fluoronucleosides (d values between ꢁ99 and ꢁ110 ppm,
upfield from CFCl3 standard) was conveniently used to follow
the progress of the reaction. The decrease in the concentration
of the fluoronucleosides with time was determined using the
conventional external standard (4-fluorotoluene in DMSO-d6,
dF = ꢁ117.0 ppm) technique. It has been reported that the
reaction of 8-chloro-9-methylpurine with piperidine or sodium
ethoxide follows a second-order kinetics (first-order with
respect to each reactant) [15]. The kinetics of base catalyzed
defluorination reaction of 8-fluoropurines is likely to be
analogous to the kinetics of dechlorination of 8-chloro-9-
methylpurine with bases. The reaction would obey a second-
order kinetics provided the concentrations of the reagents are
maintained at appropriate stoichiometric levels [16]. To mimic
the general conditions utilized for the deprotection of acetyl
groups in nucleoside chemistry [1], the defluorination kinetic
reactions were conducted with >100 molar excess of ammonia
in methanol or 2-propanol. Under such conditions, the
defluorination reaction would follow a pseudo-first-order
kinetics. The 8-fluoropurines utilized in this investigation were
all found to be stable even in boiling methanol and 2-propanol.
A concentration of 0.5 M NH3 in methanol was chosen for the
kinetic studies because at this concentration a convenient
measurement of the reaction rates by 19F NMR could be
1
identified by UV, H NMR and HRMS. The reaction of 8-
fluoropurines with 0.5 M NH3 in methanol gave the corre-
sponding 8-methoxy and 8-aminopurines as the major and
minor products, respectively. The ratios of these two products
formed in the kinetic reaction mixtures were also determined
1
after complete deacetylation by either H NMR or analytical
HPLC and the data are given in Table 1.
Using the fluoronucleosides 2 as a typical example, a
probable mechanism for the formation of 8-methoxy- and 8-
aminopurines is shown in Scheme 4. The base catalyzed
defluorination of the fluoropurines is presumed to follow a two-
step addition–elimination (SNAr) mechanism similar to the
general mechanism of activated aromatic nucleophilic sub-
stitution reactions involving good leaving groups [17]. It is well
known that C(8) in purine nucleosides is a reactive site [18] and
followed. The pseudo-first-order reaction rate constants (kobs
)
and the corresponding half-life (t1/2) values for the defluorina-
tion of 8-fluoropurines in methanolic ammonia are provided in
Table 1.
Table 1
Kinetic data for defluorination of 8-fluoropurines in 0.5 M NH3/CH3OH, along with 19F and 13C NMR chemical shifts of C(8)–F
kobs (minꢁ1
)
t1/2
Product ratio
a,b
dF (ppm)
a,c
dF (ppm)
d
dC(8) (ppm)
Compound
(9/8, 11/10 and 14/13)
2a
2b
2c
5a
5b
12
(3.26 ꢀ 0.1) ꢂ 10ꢁ4
(8.62 ꢀ 0.38) ꢂ 10ꢁ4
(2.00 ꢀ0.1) ꢂ 10ꢁ2
(4.34 ꢀ 0.46) ꢂ 10ꢁ3
(7.53 ꢀ 0.28) ꢂ 10ꢁ2
(4.06 ꢀ 0.17) ꢂ 10ꢁ2
35.4 h
13.4 h
34.7 min
2.8 h
100/0
100/0
76/24
86/14
65/35
77/23
ꢁ107.7
ꢁ107.5
ꢁ103.9
ꢁ106.7
ꢁ102.9
ꢁ103.7
ꢁ110.4
ꢁ108.2
ꢁ104.2
ꢁ107.1
ꢁ103.3
ꢁ103.3
148.8
147.8
149.8
149.6
150.9
150.8
9.2 min
17 min
a
External reference: 4-fluorotoluene/DMSO-d6 (ꢁ117.0 ppm).
b
c
d
In CH3OH.
In 0.5 M NH3/CH3OH.
In CDCl3.

1178
J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
Table 2
determining step [22], would lead to the formation of the
Meisenheimer complex [23] 15 which could collapse to the
corresponding 8-amino derivatives 8. The highly solvated
tetrahedral intermediates [24] 15 can also undergo methano-
lysis to yield the 8-methodoxy analogs 9. The proportion of the
8-amino product formed from a given 8-fluoropurine depended
upon its observed rate of reaction (Table 1). Faster reacting
purines, e.g. 5b, 12, 2c and 5a, provided higher amounts of 8-
amino products than the slower reacting derivatives like 2b and
2a. Overall, there was preponderance for the formation of the 8-
methoxy analogs over the 8-amino derivatives, suggesting an
efficient solvation of the Meisenheimer complex by methanol
[24].
Deacetylation of selected 8-fluoropurine nucleosides during the kinetic reac-
tions as analyzed by LC–MS
Compound
Time (min)
Acetylated form (%)
Tri-Ac
Di-Ac
Mono-Ac
Fully
deacetylated
(a) In 0.5 M NH3 in methanol
2a
5a
12
0
31
84.5
17.9
0.6
806
2.4
6.7
69.9
67.1
21.6
9.8
9.9
32.3
98
376
737
2177
78.4
90.2
100.0
0
31
60.8
1.0
11.0
4.8
21.8
65.8
55.3
35.6
32.6
13.6
Defluorination rates of the 8-fluoropurines were also
investigated with 0.5 M NH3 in 2-propanol and the data are
provided in Table 3. The rates of defluorination in 2-propanol
were uniformly lower than the corresponding values in
methanol. Both the polarity of the reaction medium and the
extent of solvation of the Meisenheimer complex 15 can have a
major effect on SNAr rate constants. It is known in SNAr
reactions that amine nucleophiles tend to react faster in more
polar solvents [17b]. Thus, in the relatively less polar 2-
propanol [ET(30) value: 48.4] the 8-fluoropurines would react
slower than in methanol [ET(30) value: 55.4] [25]. Additionally,
the Meisenheimer complex 15 may be more efficiently solvated
by methanol than 2-propanol due to lesser steric demands,
leading to a faster rate of reaction. Similar solvent effects in
SNAr mechanism based reactions have previously been
reported [26]. It is interesting to note an exclusive formation
of 8-aminopurines as the product of defluorination with
ammonia in 2-propanol (Table 3) in contrast to the 8-
alkoxypurines as the major product in the case of ammonia
in methanol. This observation also strongly suggests the
solvation of the tetrahedral intermediate 15 by 2-propanol is
quite restricted due to its steric size preventing the formation of
8-isopropoxypurine derivatives. The role of steric hindrance in
SNAr reactions is also known [15b,17b]. Under the condition of
0.5 M ammonia in 2-propanol, the competing deacetylation
reactions also became relatively slower. The selected LC–MS
data (Table 2) suggest that the defluorination reaction happened
essentially in the triacetyl form under these conditions.
The kobs data presented in Tables 1 and 3 show some
interesting correlations. For example, the 8-fluoroadenosine
derivative 12 was defluorinated, both in NH3-methanol and 2-
propanol media, faster than the guanosine analogs 2a,b. In
SNAr reactions, the attack by the nucleophile is enhanced at the
reaction site with decreased electron density [19]. It would be
intriguing to compare the kobs data for the fluoropurines with
their electron density at the site of reaction, namely C(8), but
unfortunately the electron density data for these fluoropurines
are not known yet. However, such data for C(8) in adenine
(0.728) and adenosine (0.928) are known to be lower than those
of the corresponding carbons in guanine (0.991) and guanosine
(0.991) [27,28]. The higher kobs value for 8-fluoroadenosine
analog 12 relative to those of 8-fluoroguanosine derivatives
2a,b (Table 1) thus suggests the trend of electron density at C(8)
in adenosine and guanosine may also be extended to their
27.4
44.7
64.4
67.4
86.4
63
94
126
290
0
31
63
77.2
12.7
3.8
11.9
4.9
2.7
73.5
77.5
9.0
18.7
(b) In 0.5 M NH3 in 2-propanol
2c
0
15
100.0
100.0
100.0
95.7
56
84
4.3
8.0
177
236
92.0
88.6
11.4
12
0
14
100.0
100.0
100.0
95.6
41
69
4.4
7.6
192
92.4
(c) In 0.01 M HCl in methanol
2a
0
18
35
70
100.0
100.0
100.0
100.0
2c
12
0
18
35
100.0
100.0
100.0
0
129
100.0
95.7
89.4
88.2
87.0
86.0
80.3
4.3
10.6
11.8
13.0
14.0
19.7
606
745
884
1385
3071
fluorine because of its high electronegativity can induce a
positive charge on C(8) making it amenable to SNAr type
reactions [19]. This is supported by the observation that
reactions of 2-fluoroimidazole, a model analog of 8-fluoropur-
ines with nucleophiles, follow an addition–elimination
mechanism [20]. Interestingly, reactions of certain substituted
chloropurines have also been reasoned to occur via SNAr
mechanisms [21]. Thus, in the SNAr reaction shown in Scheme
4 the addition of NH3 to the 8-fluoropurines 2, a likely rate-

J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
1179
Scheme 3. Defluorination reactions of 8-fluoronucleosides in basic and acidic media.

1180
J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
Fig. 1. Correlations between ln t1/2 values of 8-fluoropurine nucleosides (2a-b,
5a,b and 12) and the 19F NMR chemical shifts in 0.5 M NH3 solutions.
nucleophile at C(8) (Scheme 4), a concept consistent with a
SNAr mechanism [19].
A closer look into the data given in Tables 1 and 3 also
reveals that a significant correlation exists between the rate
constants and 19F and C(8) NMR chemical shifts. Fig. 1, for
example, illustrates the relationship between t1/2 values and the
19F NMR chemical shifts in 0.5 M NH3/CH3OH and 0.5 M
NH3/2-PrOH. Fig. 2 represents the correlation of defluorination
half-lives with the 13C NMR chemical shifts for C(8) in CDCl3.
Unlike the 19F NMR data, the 13C NMR spectra could not be
obtained in 0.5 M NH3/CH3OH or 2-propanol due to the short
half-lives of some of the fluoropurines in basic media (Table 1)
and the relatively long acquisition times required for NMR data
collection. As expected, substitution of CDCl3 as the solvent for
13C NMR data collection had a profound effect on the C(8)
chemical shift correlation of the fluoroguanosine analog 2a
with its kobs value in 0.5 M NH3/CH3OH, most likely due to the
deprotonation of N1-hydrogen [29] under the basic defluorina-
tion reaction condition [30]. This deprotonation is also
supported by 19F NMR data. In methanolic ammonia the 19F
NMR chemical shift for the derivative 2a moved upfield by
2.7 ppm in comparison with that in methanol (Table 1),
Scheme 4. Mechanism of base catalyzed defluorination of 8-fluoronucleosides.
respective 8-fluoro analogs. More direct evidence in this regard
was obtained from 13C and 19F NMR data. Both the 13C
chemical shift for C(8) and the 19F chemical shift in the
adenosine derivative 12 were shifted downfield when compared
with the corresponding values for the guanosine analogs 2a,b
(Table 1), attesting to the electron density correlations. In
general, substrates with more deshielded carbon-8 and fluorine
tend to react faster because of the facilitated attack of the
Table 3
Defluorination kinetic data for 8-fluoropurines in 0.5 M NH3/2-PrOH and 19F
NMR chemical shifts
kobs (minꢁ1
)
t1/2
a,b
dF (ppm)
Compound
2a
2b
2c
5a
5b
12
Too slow to measure
(1.65 ꢀ 0.00) ꢂ 10ꢁ5
(6.27 ꢀ 0.11) ꢂ 10ꢁ3
(5.73 ꢀ 0.16) ꢂ 10ꢁ4
(4.04 ꢀ 0.16) ꢂ 10ꢁ2
(1.61 ꢀ 0.04) ꢂ 10ꢁ2
29.2 d
ꢁ108.2
ꢁ104.1
ꢁ107.0
ꢁ103.4
ꢁ104.0
1.8 h
20.2 h
Fig. 2. Correlations between ln t1/2 values of 8-fluoropurines nucleosides (2b,c,
5a,b and 12) in 0.5 M NH3 solutions and 13C NMR the chemical shifts of C(8) in
CDCl3. Data for the guanosine analog 2a is not included because of deprotona-
tion of N1 in basic media which profoundly affects the 13C chemical shift of
carbon-8 as described in Section 2.2 (Discussion).
17.0 min
43.0 min
a
External reference: 4-fluorotoluene/DMSO-d6 (ꢁ117.0 ppm).
b
In 0.5 M NH3/2-PrOH.

J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
1181
Scheme 5. Mechanism of acid catalyzed defluorination of 8-fluoronucleosides.
presumably due to the conjugative distribution of the resultant
negative charge into the imidazole ring. In contrast, the
corresponding 19F NMR signal for the N1-blocked analog 2b in
methanolic ammonia moved upfield by only 0.7 ppm. Thus, in
basic medium the guanosine derivative 2a had the most
shielded fluorine and the negative charge built up in the
imidazole ring would induce an electrostatic repulsion between
the approaching nucleophile (NH3) and the reaction center C(8)
(Scheme 4). In agreement with this rationalization, the rate of
defluorination of the guanosine analog 2a in methanolic
ammonia was found to be the slowest among all the
fluoropurines studied (Table 1). More dramatically, its
defluorination rate in 0.5 M NH3 in 2-propanol was too slow
to measure by 19F NMR (Table 3).
deficient pyrimidine ring [33] enabling a facile approach of the
nucleophile to C(8) (Scheme 4). Overall, the defluorination
rates in methanol and 2-propanol media exhibited similar
trends.
The acid catalyzed defluorination rates of 8-fluoropurine
nucleosides2a–c, 5a,b and12were alsoexamined underpseudo-
first-order reaction conditions in 0.01 M dry HCl in methanol at
28 ꢀ 0.1 8C (Schemes 3 and 5). The rate constants (kobs), the
half-life (t1/2) data and the 19F NMR chemical shifts in 0.01 M
methanolicHClaregiven inTable4. Product analysis byLC–MS
as well as 1H NMR spectroscopy showed that all 8-fluoropurines
exclusively yielded their corresponding 8-methoxy derivatives
9a–c, 11a,b and 14. Further, during all the kinetic runs no
deacetylation was observed for the guanosine and inosine
derivatives while the adenosine analog 12 defluorinated mainly
in the triacetyl form along with a smaller contribution from its
diacetyl derivative (Table 2). LC–MS also indicated that no
deglycosylation of any of 8-fluoropurine nucleosides took place
under the kinetic reaction condition.
As expected, other structural characteristics of the fluor-
opurine derivatives also had considerable effects on the
defluorination rates in 0.5 M NH3/CH3OH or 2-propanol.
Chief among them is the presence of an electron donating N2-
amino group [31,32] in the fluoropurines 2a,b and 5a. The 19
F
NMR signals in these derivatives were shifted upfield when
compared with that of the other analogs; a similar trend was
also observed for C(8) 13C NMR chemical shifts (Table 1). In
methanolic ammonia the inosine analog 2c (dF = ꢁ104.2 ppm;
Several interesting trends in the defluorination rates of 8-
fluoropurines in methanolic HCl (Table 4) were observed when
Table 4
d
C(8) = 149.8 ppm) reacted more than twenty times faster than
Data for kinetics of defluorination of 8-fluoropurines in 0.01 M HCl/CH3OH
and 19F NMR chemical shifts
the guanosine derivative 2b (dF = ꢁ108.2 ppm; dC(8)
=
147.8 ppm). Similarly, the rate of defluorination of the O6-
methyl analog 5b (dF = ꢁ103.3 ppm; dC(8) = 150.9 ppm) with a
deactivated N2-amino function was nearly twenty times faster
than its counterpart 5a (dF = ꢁ107.1 ppm; dC(8) = 149.6 ppm)
with a free amino group (Table 1).
a,b
dF (ppm)
Compound
kobs (minꢁ1
)
t1/2
2a
2b
2c
5a
5b
12
(1.03 ꢀ 0.02) ꢂ 10ꢁ1
(1.84 ꢀ 0.04) ꢂ 10ꢁ1
(7.37 ꢀ 0.04) ꢂ 10ꢁ2
(1.67 ꢀ 0.02) ꢂ 10ꢁ1
(2.45 ꢀ 0.03) ꢂ 10ꢁ1
(3.43 ꢀ 0.08) ꢂ 10ꢁ4
6.7 min
ꢁ107.5
ꢁ107.7
ꢁ104.0
ꢁ104.3
ꢁ103.0
ꢁ99.2
3.8 min
9.4 min
4.1 min
2.8 min
33.7 h
In general, the fully aromatic fluoropurines 5a,b and 12
underwent defluorination faster than the non-aromatic analogs,
most likely due to some delocalization of the electron density
from the p-electron rich imidazole ring into the p-electron
a
External reference: 4-fluorotoluene/DMSO-d6 (ꢁ117.0 ppm).
b
In 0.01 M HCl/CH3OH.

1182
J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
compared with the corresponding rates in basic media (Tables 1
and 3). For instance, the 8-fluororadenosine analog 12, one of
the fastest reacting substrates in methanolic ammonia, reacted
more than a hundred times slower in methanolic HCl. In sharp
contrast, the guanosine derivative 2a, with a half-life of 35.4 h
in methanolic ammonia and whose defluorination rate in 2-
propanolic ammonia was too slow to measure, had a t1/2 value
of only 6.7 min in 0.01 M HCl in methanol. In methanolic HCl
all the guanosine and inosine derivatives reacted 102 to 103
times faster than the adenosine analog 12. Unlike in methanolic
ammonia the 19F NMR chemical shifts of the fluoropurines in
methanolic HCl did not show a linear correlation with the t1/2
data. For example, derivatives 2b and 5a exhibited close kobs
values in spite of a difference of 3.4 ppm in their 19F NMR
chemical shifts (Table 4). All these observations imply that, in
acid medium, defluorination rates may not be as closely related
to the electron densities at C(8) as found in basic media.
Based on these data, a defensible reaction mechanism for the
defluorination reaction in acid medium using the guanosine
analog 2 as a typical example is provided in Scheme 5. The first
step of the reaction is the equilibrium protonation of the purine
nucleoside. The principal site of protonation in purine
nucleosides has been well documented [31,32,34,35]. For
example, using 15N NMR, it has been shown that N7 in
guanosine and inosine is the primary site of protonation while
N1 is the corresponding site in adenosine [31,32,34b].
Similarly, an analogous trend of protonation has been observed
in the case of several 8-substituted guanosine, inosine and
adenosine derivatives [32,34d,36]. Interestingly, the pKa
measurement of the fluoroadenosine analog 12 also indicates
N1 as the primary site of protonation [7]. Thus, it is quite
probable the primary sites of protonation in the 8-fluoropurines
discussed herein parallel those of the corresponding non-
fluorinated parent nucleosides (i.e., guanosine, inosine, and
adenosine). The inductive effect due to the fluorine atom in 8-
fluoropurines will, however, decrease the basicity of N7 and
hence its protonation will not be very facile. In methanolic HCl
a solvated, protonated derivative 17 (Scheme 5) for the
guanosine and inosine cases would lead to the formation of the
intermediate 18 which would then eliminate fluoride ion
forming the 8-methoxy product 9. Elimination of fluoride ion
from a tetrahedral intermediate such as 18 is a fast step in SNAr
reactions [19] and hence could not be the rate-determining step.
Thus, the equilibrium protonation of N7 in the fluoropurines
followed by a rate-limiting addition of methanol to the
protonated substrate 17 would determine their acid catalyzed
Fig. 3. Correlation between ln t1/2 values of 8-fluoropurine nucleosides (2a–c,
5a,b) in 0.01 M HCl/CH3OH solution and N7 pKa values of their corresponding
parent nucleosides (Table 5). Data for 8-fluoroadenosine analog 12 is not
included in this plot since it has been reported [7] that 8-fluoroadenosine
(pKa = 2.95) is protonated at N1 in 0.01 M HCl and hence a second protonation
at N7 is more difficult. An analogous trend of protonation has previously been
observed with adenosine [31,32,34b].
defluorination rate constants (Scheme 5) and hence the pKa of
N7 in the 8-fluoropurines would be a suitable parameter to
reflect the propensity of its protonation. Unfortunately, the short
t1/2 values of the other 8-fluoropurines in acid medium (2.8–
9.4 min, Table 4) precluded the measurement of their basicities
with the facilities described. Hence, for an explanation of the
kinetic data presented in Table 4, it was reasoned that the N7
pKa values of the readily available non-fluorinated parent
guanosine, inosine and adenosine derivatives (Table 5) could
logically be utilized because the effect of fluorine substitution
would be relatively similar when inter-comparisons of kobs data
are made. This rationalization is indeed shown to be valid when
the pKa data shown in Table 5 were plotted against lnt1/2 values
provided in Table 4 and yielded a straight line correlation
(Fig. 3, R2 = 0.94).
Among the parent nucleosides (Table 5) the compound with
a higher pKa would be more readily protonated. The higher
basicity of N1-methylguanosine (pKa = 2.6) than that of N1-
methylinosine (pKa = 1.15) is well reflected in the kobs data of
the corresponding 8-fluoroanalogs. The acid catalyzed defluor-
ination rate constant for the guanosine analog 2b is thus 2.5
times higher than that for the inosine derivative 2c (Table 4).
Similarly, the relatively high stability of the adenosine analog
12 in acid medium (t1/2 = 33.7 h) amongst all the 8-
fluoronucleosides can clearly be correlated to the low N7
basicity (pKa = ꢁ1.50, Table 5) of adenosine, resulting from its
facile N1 protonation (pKa = 3.2) [7].
Table 5
Basicities of purine nucleosides
pKa for N7
Reference
3. Conclusions
Compound
Guanosine
2.11 ꢀ 0.04
2.6
[37]
19F NMR spectroscopy is a convenient tool for assessing the
defluorination rates of 8-fluoropurine nucleosides. In general,
the electron density at C(8), as determined by 19F and 13C
NMR, is identified as the underlying factor that determines
their stability in basic media. In acid medium the stability
of 8-fluoropurines is related to their N7 pKa values. More
N1-Methylguanosine
N1-Methylinosine
O6-Methylguanosine
N2-Acetyl-O6-methylguanosine
Adenosine
[38]
1.15 ꢀ 0.03
2.4
[39]
[38]
This worka
2.69 ꢀ 0.02
ꢁ1.50 ꢀ 0.15
[36]
a
The pKa was measured by standard pH titration method.

J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
1183
importantly, the results of this investigation indicate acidic
conditions are incompatible with 8-fluoroguanosine and 8-
fluoroinosine analogs while basic conditions should be avoided
with 8-fluoroadenosine derivatives.
min). Specific solvents for fluorination and conditions for silica
gel column chromatography as well as analytical HPLC are
given below for each compound. Analytical and semi-
preparative HPLC eluents were monitored with UV detectors
set at a wavelength of 254 nm.
4. Experimental
4.3.2. 8-Fluoro-20,30,50-tri-O-acetylguanosine (2a)
4.1. General
Method A—Reaction solvent: CH3NO2. Product Rf 0.37
(TLC, CH2Cl2/CH3OH 95/5). Silica gelcolumnchromatography
eluent: EtOAc/EtOH (98/2 and 95/5). Semi-preparative HPLC
eluent: CH2Cl2/CH3OH (97.5/2.5), tR = 75 min. Yield: 5.2%,
semi-solid. UV(CH3OH) lmax = 242 and 276 nm; 1H NMR
(CD3CN) d (ppm) 2.01 (s, 3H), 2.07 (s, 3H), 2.10 (s, 3H), 4.21
(dd, J = 12.1 and 5.1 Hz, 1H), 4.35 (m, 1H), 4.48 (m, dd, J = 12.1
and 3.6 Hz, 1H), 5.61 (broad s, exchangeable, 2H), 5.69 (pseudo-
t, 1H), 5.87 (d, J = 5.2 Hz, 1H), 5.82 (pseudo-t, 1H), 9.48 (s,
exchangeable, 1H); 13C NMR (CD3CN) d (ppm) 20.4, 20.5, 20.7,
63.0, 70.5, 71.9, 80.0, 85.1, 110.3 (JC,F = 15.2 Hz), 148.8
(JC,F = 246.3 Hz), 150.1, 154.2, 158.1, 169.4, 169.6 and 170.7;
19F NMR (CD3OD) d = ꢁ107.7 ppm; HRMS calcd. for
C16H18N5O8F 427.1139, found 427.1120.
The 1H, 13C and 19F NMR spectra were recorded at 360, 125
and 338.9 MHz, respectively. The chemical shifts (d) are
expressed in parts per million (ppm) downfield from internal
tetramethylsilane (TMS) for H and 13C and from external
1
trichlorofluoromethane for 19F NMR. Ultraviolet spectra were
recorded in methanol solutions. The on-line HPLC-UV spectra
were recorded with a diode array detector. High resolution mass
spectra were obtained using either fast atom bombardment
ionization technique (FAB HRMS) or electron impact
ionization technique (HRMS). Column chromatography was
run on silica gel columns (0.063–0.200 mm), which employed a
stepwise solvent polarity gradient that was correlated with TLC
mobility. TLC analyses were performed on silica gel plates and
the spots were visualized with UV light (254 nm).
Method B—Reaction solvent: CHCl3 (stabilized with
amylene). Product 2a was contaminated with 15% of 3a as
shown by analytical HPLC [Alltech Partisil Silica column,
5 mm, 4.6 mm ꢂ 250 mm; eluent: CH2Cl2/CH3OH (95/5); flow
rate: 1.0 mL/min]. The product mixture was reacted with
0.01 M HCl/CH3OH at room temperature for 2 h. Semi-
preparative HPLC purification [eluent: CH2Cl2/CH3OH (95/5)]
gave 8-chloro-20,30,50-tri-O-acetylguanosine (3a), tR = 11 min.
Yield: 2%, colorless film. UV(CH3OH) lmax = 260 and 275 (sh)
nm; 1H NMR (CDCl3) d (ppm) 2.04 (s, 3H, CH3), 2.16 (s, 3H,
CH3), 2.17 (s, 3H, CH3), 4.36 (m, 2H), 4.50 (dd, J = 10.6 and
2.5 Hz, 1H), 5.98 (d, J = 6.1 Hz, 1H), 6.00 (pseudo-t, 1H), 6.29
(pseudo-t, 1H), 6.39 (broad s, exchangeable, 2H, NH2), 11.93
(s, 1H, NH). FAB HRMS calcd. for C16H18N5O8ClNa (M + Na)
466.0744, found 466.0736.
4.2. Synthesis of starting materials (1a–c, 4a,b)
Guanosine was acetylated with acetic anhydride in pyridine-
DMF at room temperature for 50 h to give 20,30,50-tri-O-
acetylguanosine (1a) [9]. Guanosine and inosine were
methylated with methyl iodide or methyl 4-toluenesulfonate
in the presence of potassium carbonate [11] followed by
acetylation of the sugar moiety to give 20,30,50-tri-O-acetyl-N1-
methylguanosine (1b) and 20,30,50-tri-O-acetyl-N1-methylino-
sine (1c), respectively. For the synthesis of N2-acetyl-20,30,50-
tri-O-acetyl-O6-methylguanosine (4b), guanosine was first
acetylated with acetic anhydride in pyridine at 50 8C in the
presence of DMAP and triethylamine [10] to give N2-acetyl-
20,30,50-tri-O-acetylguanosine, which upon O6-methylation via
its O6-(N-methylpyrrolidinium) derivative [12] afforded 4b.
Compound 4b dissolved in methanol was completely deace-
tylated with 25–28% NH4OH at room temperature [12] and
evaporated to dryness. The residue was dried overnight over
P2O5 under vacuum and reacetylation of the sugar moiety
yielded 20,30,50-tri-O-acetyl-O6-methylguanosine (4a).
4.3.3. 8-Fluoro-20,30,50-tri-O-acetyl-N1-methylguanosine
(2b)
Method A—Reaction solvent: CH3CN. Product Rf 0.45
(TLC, CH2Cl2/CH3OH 95/5). Silica gel column chromato-
graphy eluent: EtOAc/CH3OH (from 98/2 to 90/10). Semi-
preparative HPLC eluent: CH2Cl2/CH3OH (97.5/2.5),
tR = 24 min. Yield: 10%, semi-solid. UV(CH3OH) lmax = 244
1
and 276 nm; H NMR (CDCl3) d (ppm) 2.06 (s, 3H), 2.12 (s,
4.3. Fluorination
3H), 2.14 (s, 3H), 3.51 (s, 3H), 4.32 (dd, J = 11.0 and 6.1 Hz,
1H), 4.39 (m, 1H), 4.50 (dd, J = 11.0 and 3.6 Hz, 1H), 5.73
(broad s, exchangeable, 2H), 5.82 (m, 2H), 6.05 (pseudo-t, 1H);
13C NMR (CDCl3): d (ppm) 20.5, 20.6, 20.7, 28.3, 62.9, 70.2,
71.8, 79.5, 84.6, 110.3 (JC,F = 12.8 Hz), 146.8 (JC,F = 3.7 Hz),
147.8 (JC,F = 249.3 Hz), 153.8 (JC,F = 2.6 Hz), 156.3
(JC,F = 2.4 Hz), 169.6, 169.8 and 170.9; 19F NMR(CD3OD)
d = ꢁ107.5 ppm; FAB HRMS calcd. for C17H21N5O8F (M + 1)
442.1374, found 442.1371.
4.3.1. General procedure [4]
To a solution of the starting material (1a–c or 4a,b) in
anhydrous solvents (e.g., CH3CN, CH3NO2 or CHCl3)
(0.1 mmol/1.5 mL) was bubbled with 1% F2 in argon or
helium (1.5 mol equiv.) at room temperature over a period of 2–
5 h. The reaction mixture was evaporated to dryness under
vacuum. The product mixture was chromatographed on 200–
400 times its weight of silica gel and the isolated products were
further purified by semi-preparative HPLC (Whatman Partisil
Silica column, 10 mm, 9 mm ꢂ 500 mm; flow rate 6.0 mL/
Method B—Reaction solvent: CH3NO2. Yield: 5.7%.
Physical data were identical to the product 2b obtained from
the reaction in CH3CN.

1184
J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
Method C—Reaction solvent: CHCl3 (stabilized with
chromatography eluent: CH2Cl2/CH3OH (99/1 followed by
95/5). Semi-preparative HPLC eluent: CH2Cl2/CH3OH (99.3/
0.7), tR = 55 min. Yield: 3.2%, colorless film. UV(CH3OH)
amylene). Product 2b was found to be contaminated with
20% of 3b as shown by 1H NMR and analytical HPLC [Alltech
Econosil Silica column, 5 mm, 4.6 mm ꢂ 250 mm; linear
gradient elution with CH2Cl2/CH3OH (98.2/1.5 to 97.5/2.5)
during 120 min; flow rate: 1.0 mL/min]. The product mixture
was treated with 0.01 M HCl/CH3OH at room temperature for
2 h. Semi-preparative HPLC purification [eluent CH2Cl2/
CH3OH (97.4/2.6)] gave 8-chloro-20,30,50-tri-O-acetyl-N1-
methylguanosine (3b), tR = 25 min. Yield: 3.9%, colorless film.
UV(CH3OH) lmax = 262 and 280 nm (sh); 1H NMR (CDCl3) d
(ppm) 2.04 (s, 3H), 2.13 (s, 3H), 2.14 (s, 3H), 3.50 (s, 3H), 4.33
(dd, J = 11.8 nd 6.1 Hz, 1H), 4.40 (m, 1H), 4.50 (dd, J = 11.8 and
3.8 Hz, 1H), 5.17 (broad s, exchangeable, 2H), 6.01 (d,
J = 3.8 Hz, 1H), 6.05 (pseudo-t, 1H), 6.20 (pseudo-t, 1H);
HRMS calcd. for C17H20N5O8Cl 457.1000, found 457.0999.
1
lmax = 244 and 278 nm; H NMR(CD3CN) d (ppm) 2.06 (s,
3H), 2.09 (s, 3H), 2.18 (s, 3H), 3.98 (s, 3H), 4.20 (dd, J = 12.1
and 5.9 Hz, 1H), 4.33 (m, 1H), 4.47 (dd, J = 12.1 and 3.6 Hz,
1H), 5.40 (broad s, exchangeable, 2H), 5.76 (pseudo-t, 1H),
5.91 (d, J = 4.6 Hz, 1H), 5.98 (pseudo-t, 1H); 13C NMR
(CDCl3) d (ppm) 20.4, 20.5, 20.6, 54.2, 62.8, 70.4, 72.0, 79.8,
85.0, 109.2 (JC,F = 15.4 Hz), 149.6 (JC,F = 250.9 Hz), 151.7
(JC,F = 3.5 Hz), 158.8 (JC,F = 3.3 Hz), 160.9 (JC,F = 3.6 Hz),
169.3, 169.5, 170.6; 19F NMR (CD3OD) d = ꢁ106.7 ppm; FAB
HRMS calcd. for C17H21N5O8F (M + 1) 442.1374, found
442.1379. 8-Chloro-20,30,50-tri-O-acetyl-O6-methylguanosine
(6a), tR = 66 min. Yield: 1.3%, colorless film. UV(CH3OH)
l
max = 254 and 284 nm; 1H NMR(CDCl3) d (ppm) 2.02 (s, 3H),
2.12 (s, 3H), 2.15 (s, 3H), 4.04 (s, 3H), 4.37 (m, 2H), 4.49 (dd,
J = 10.9 and 2.4 Hz, 1H), 5.01 (broad s, exchangeable, 2H),
6.03 (d, J = 4.1 Hz, 1H), 6.09 (pseudo-t, 1H), 6.26 (pseudo-t,
1H); HRMS calcd. for C17H20N5O8Cl 457.1000, found
457.0987.
4.3.4. 8-Fluoro-20,30,50-tri-O-acetyl-N1-methylinosine (2c)
MethodA—Reactionsolvent: CH3CN. ProductRf 0.24(TLC,
EtOAc). Silica gel column chromatography eluent: CH2Cl2/
CH3OH(99/1followedby95/5). Semi-preparative HPLCeluent:
CH2Cl2/CH3OH (98.3/1.7); tR = 27 min. Yield: 16.5%, white
foam. UV(CH3OH) lmax = 238 and 268 nm; 1H NMR (CDCl3) d
(ppm) 2.09 (s, 2ꢂ 3H), 2.14 (s, 3H), 3.63 (s, 3H), 4.27 (dd,
J = 11.7 and 5.3 Hz, 1H), 4.38 (m, 1H), 4.44 (dd, J = 11.7 and
3.6 Hz, 1H), 5.64 (pseudo-t, 1H), 5.97 (d, J = 5.2 Hz, 1H), 6.03
(pseudo-t, 1H), 8.02 (s, 1H); 13C NMR (CDCl3) d (ppm) 20.3,
20.5, 20.6, 34.6, 62.9, 70.3, 71.8, 80.2, 85.1, 119.2
(JC,F = 13.4 Hz), 145.6 (JC,F = 3.3 Hz), 147.8 (JC,F = 3.3 Hz),
149.8 (JC,F = 252.1 Hz), 155.9 (JC,F = 2.2 Hz), 169.3, 169.5,
170.4; 19F NMR (CD3OD) d = ꢁ103.9 ppm; FAB HRMS calcd.
for C17H20N4O8F (M + 1) 427.1265, found 427.1257.
4.3.6. 8-Fluoro-N2-acetyl-20,30,50-tri-O-acetyl-O6-
methylguanosine (5b)
Reaction solvent: CHCl3 (stabilized with amylene). Product
Rf 0.41 (TLC, EtOAc). Silica gel column chromatography
eluent: EtOAc/Hexane (10/90) followed by EtOAc. Semi-
preparative HPLC eluent: CH2Cl2/CH3OH (99/1). 5b,
tR = 30 min. Yield: 29.5%, colorless film. UV(CH3OH)
l
1
max = 261 nm; H NMR (CDCl3) d (ppm) 2.10 (s, 3H), 2.12
(s, 3H), 2.16 (s, 3H), 2.51 (s, 3H), 4.12 (s, 3H), 4.32 (dd,
J = 11.2 and 5.7 Hz, 1H), 4.41 (m, 1H), 4.51 (dd, J = 11.2 and
3.3 Hz, 1H), 5.83 (pseudo-t 1H), 5.95 (d, J = 4.7 Hz, 1H), 6.04
(pseudo-t, 1H), 8.05 (s, 1H, exchangeable); 13C NMR (CDCl3)
d (ppm) 20.4, 20.5, 20.6, 25.2, 54.7, 63.0, 70.4, 72.2, 80.0, 85.4,
112.3 (JC,F = 13.5 Hz), 150.8, 150.9 (JC,F = 253.1 Hz), 151.9,
160.6, 169.4, 169.5, 170.5, 170.7; 19F NMR (CD3OD)
d = ꢁ102.9 ppm; HRMS calcd. for C19H22N5O9F 484.1480,
found 484.1475. 8-Chloro-N2-acetyl-20,30,50-tri-O-acetyl-O6-
methylguanosine (6b), tR = 39 min. Yield: 1.8%, white foam.
UV (CH3OH) lmax = 268 nm. 1H NMR (CDCl3) d (ppm) 2.01
(s, 3H), 2.11 (s, 3H), 2.16 (s, 3H), 2.49 (s, 3H), 4.14 (s, 3H),
4.34 (dd, J = 11.2 and 6.1 Hz, 1H), 4.41 (m, 1H), 4.54 (dd,
J = 11.2 and 3.6 Hz, 1H), 6.01 (pseudo-t, 1H), 6.06 (d,
J = 3.6 Hz, 1H), 6.21 (pseudo-t, 1H), 8.12 (s, exchangeable,
1H); HRMS calcd. for C19H22N5O9Cl 499.1106, found
499.1113.
Method B—Reaction solvent: CHCl3 (stabilized with
amylene). Semi-preparative HPLC eluent: CH2Cl2/CH3OH
(99/1). 2c, yield: 16.2%, tR = 64 min. 8-Chloro-20,30,50-tri-O-
acetyl-N1-methylinosine(3c),tR = 74 min.Yield:7.2%,colorless
film.UV(CH3OH)lmax = 248and270(sh)nm;1HNMR(CDCl3)
d (ppm) 2.08 (s, 3H), 2.11 (s, 3H), 2.16 (s, 3H), 3.64 (s, 3H), 4.31
(dd,J = 11.0and6.1 Hz,1H),4.39(m,1H),4.46(dd,J = 11.0and
3.7 Hz, 1H), 5.76 (pseudo-t, 1H), 6.09 (d, J = 5.9 Hz, 1H), 6.21
(pseudo-t, 1H), 8.02 (s, 1H, H-2); FAB HRMS calcd. for
C17H20N4O8Cl (M + 1) 443.0970, found 443.0975.
Method C—Reaction solvent: CHCl3 (stabilized with 1%
EtOH). Silica gel column chromatography eluent: CH2Cl2/
CH3OH (98/2 followed by 95/5). 8-Ethoxy-20,30,50-tri-O-acetyl-
N1-methylinosine (7), Rf 0.13 (TLC, EtOAc). Yield: 2.8%, semi-
solid. UV(CH3OH) lmax = 250 and 276 nm; 1H NMR (CDCl3) d
(ppm) 1.49 (t, J = 7.2 Hz, 3H), 2.07 (s, 3H), 2.10 (s, 3H), 2.13 (s,
3H), 3.62 (s, 3H), 4.22 (dd, J = 12.2 and 6.1 Hz, 1H), 4.32 (m,
1H), 4.43 (dd, J = 12.2 and 3.6 Hz, 1H), 4.65 (q, J = 7.2 Hz, 2H),
5.69 (pseudo-t, 1H), 6.05 (m, 2H), 7.92 (s, 1H); FAB HRMS
calcd. for C19H25N4O9 (M + 1) 253.1630, found 253.1622.
4.4. Basic and acidic solutions for kinetic determinations
Solutions were prepared as follows: (A) 0.5 M NH3/CH3OH
by diluting commercial 2 M NH3/CH3OH (Aldrich) with
anhydrous methanol (Aldrich); (B) 0.5 M NH3/2-propanol by
diluting commercial 2 M NH3/2-propanol (Aldrich) with
anhydrous 2-propanol (Aldrich); (C) 0.01 M HCl/CH3OH by
diluting commercial 10% dry HCl/CH3OH solution (TCI
America) with anhydrous methanol (Aldrich). The exact
4.3.5. 8-Fluoro-20,30,50-tri-O-acetyl-O6-methylguanosine
(5a)
Reaction solvent: CHCl3 (stabilized with amylene). Product
Rf 0.65 (TLC, EtOAc/CH3OH 95/5). Silica gel column

J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
1185
concentrations of all the prepared solutions were determined by
standard titration methods. All the standardized solutions were
protected from moisture and kept refrigerated in individual
vials sealed with Teflon septa.
NH3/CH3OH) CH3CN/0.02 M NH4OAc linear gradient: from
1:9 to 2:8 over a period of 8 min, to 3:7 in 20 min and held at 3:7
to 28 min; for compound 2a (kinetic reaction condition: 0.01 M
HCl/CH3OH) CH3OH/H2O (2:8 isocratic); and for compounds
5a (kinetic reaction condition: 0.5 M NH3/2-PrOH), 2c (kinetic
reaction conditions: 0.5 M NH3/2-PrOH and 0.01 M HCl/
CH3OH) and 12 (kinetic reaction conditions: 0.5 M NH3/2-
PrOH and 0.01 M HCl/CH3OH) CH3CN/0.02 M NH4OAc (3:7,
isocratic).
4.5. Kinetic measurements
The defluorination reactions were followed by 19F NMR
spectroscopy at 28 ꢀ 0.1 8C. All kinetic determinations were
conducted under pseudo-first-order reaction conditions. The
samples were prepared by dissolving 0.5 mg of 8-fluoropurine
nucleosides (2a–c, 5a,b and 12) in 0.3 mL of 0.5 M NH3/
CH3OH, NH3/2-propanol or 0.01 M HCl/CH3OH solution and
quickly transferred by syringe to a NMR tube equipped with a
capillary tube for external reference and tightly capped with a
screw-cap to prevent evaporation of volatile reagents or
solvents during data collection. The capillary tube was filled
with a solution of 4-fluorotoluene in DMSO-d6 (25 mL/0.5 mL)
as the external reference (d = ꢁ117.0 ppm from CFCl3). The
19F NMR data were collected at a sweep width of 27,800 Hz
and with a relaxation delay of 2.0 s. Typically 20 scans were
collected for each time point and the Fourier transformed
signals yielded spectra with excellent S/N ratio. From the
relative intensity of 19F signals (ꢁ100 to ꢁ110 ppm) due to 8-
fluoronucleoside and the reference, the concentration of the
fluoronucleoside could be determined at any time point. For
each sample, more than 12 data points were collected at
intervals until more than 70% of the product was defluorinated.
For slower reacting compounds (i.e., t1/2 > 5 h) the samples in
NMR tubes were kept at 28 ꢀ 0.1 8C in a thermostatic bath in-
between successive time points for data collection. The kinetic
data were analyzed using the pseudo-first-order rate equation:
ln Ct/C0 = ꢁkobst where C0 is the initial concentration of the 8-
fluoronucleoside and Ct is the concentration at time t. A plot of
the logarithmic concentration of the 8-fluoronucleosides versus
time yielded straight line graphs with correlation coefficients
(R2) ranging between 0.94 and 0.99. The observed rate
constants (kobs) were calculated from the slopes of the straight
lines. The half-life values were calculated using the equation
4.6. Product analysis of kinetic reaction mixtures
4.6.1. Products from 0.5 M NH3/CH3OH
Samples were all left at room temperature for 24–48 h in the
NMR tubes, after kinetic data collection, to allow for a
complete O-deacetylation. The reaction mixtures were evapo-
rated under vacuum and the residues were then analyzed by 1H
NMR, mass spectroscopy and reverse phase analytical HPLC
[(Alltech Econosil C18 column, 5 mm, 4.6 mm ꢂ 250 mm);
eluent: 10–20% 0.02 M NH4OAc/CH3CN; flow rate: 0.5 mL/
min]. On-line UV spectra were recorded for each eluting peak.
The ratios of 8-methoxy to 8-amino products, namely 9a/8a,
9b/8b and 9c/8c were judged by the peak areas in analytical
HPLC, while those of the others (11a/10a, 14/13 and 11b/10b)
1
were evaluated by H NMR by the integrated ratios of the
proton signals indicated below. These product ratios are
provided in Table 1.
4.6.1.1. 8-Methoxyguanosine (9a) [40]. HPLC eluent:
CH3CN/0.02 M NH4OAc (1/9). tR = 4.0 min. lmax = 244.5
and 285 nm; FAB MS (M + 1) 314. Unreacted 8-fluoroguano-
sine (less than 10%): tR = 3.2 min. lmax = 242.5 and 277.5 nm.
4.6.1.2. 8-Methoxy-N1-methylguanosine (9b). HPLC eluent:
CH3CN/0.02 M NH4OAc (1/9). tR = 5.9 min. lmax = 251 and
283 nm; 1H NMR (CD3CN) of the residue d (ppm) 3.35 (s, 3H),
3.60–3.75 (m, 3H), 4.02 (m, 1H), 4.04 (s, 3H), 4.27 (m, 1H),
4.82 (dd, J = 6.3 and 5.7 Hz, 1H), 4.88 (broad s, 1H,
exchangeable), 5.41 (s, 2H, exchangeable), 5.63 (d,
J = 6.3 Hz, 1H); FAB HRMS calcd. for C12H18N5O6 (M + 1)
328.1257, found 328.1257.
t
1/2 = 0.693/kobs. At least two or three samples were examined
for each nucleoside and the averaged kobs values are provided in
Tables 1, 3 and 4.
Defluorination/deacetylation reactions monitored and t1/2
LC–MS were carried out in capped autosampler vials at
24.0 ꢀ 0.1 8C and the analyses were performed on an Agilent
1100 LC–MS instrument. The system was coupled to a diode
array detector (detection wavelength: 254 nm) and the mass
spectrometer was operated under API-ES ionization mode
(capillary voltage 4 kV and chamber temperature 350 8C). The
HPLC for LC–MS were run on an Alltech Adsorbsphere C18
column (5 mm, 4.6 mm ꢂ 250 mm), eluted at 1 mL/min with
the following mobile phases—for compound 2a (kinetic
reaction condition: 0.5 M NH3/CH3OH) CH3OH/H2O linear
gradient: 2:8 to 4:6 over a period of 20 min; for compound 5a
(kinetic reaction condition: 0.5 M NH3/CH3OH) CH3CN/
0.02 M NH4OAc linear gradient: 1:9 to 6:4 over a period of
25 min; for compound 12 (kinetic reaction condition: 0.5 M
4.6.1.3. 8-Methoxy-N1-methylinosine (9c) and 8-amino-N1-
methylinosine (8c). HPLC eluent: CH3CN/0.02 M NH4OAc
(15/85). 9c: tR = 2.8 min. lmax = 253 and 275 nm (sh). 8c:
tR = 2.3 min. lmax = 264 and 292 nm (sh). 9c/8c = 76:24.
Product 9c and 8c were also isolated by semi-preparative
C18
HPLC
[(Alltech
Econosil
column,
10 mm,
10 mm ꢂ 500 mm); eluent: CH3OH/H2O (15/85) at 4 mL/
min]. 9c: UV (CH3OH) lmax = 250 nm (e = 9.200) and 278 nm
1
(e = 5.200); H NMR (CD3CN) d (ppm) 3.49 (s, 3H), 3.60 (s,
exchangeable, 1H), 3.63 (dd, J = 12.3 and 3.4 Hz, 1H), 3.74
(dd, J = 12.3 and 2.8 Hz, 1H), 3.87 (broad s, exchangeable,
1H), 4.05 (m, 1H), 4.11 (s, 3H), 4.27 (pseudo-t, 1H), 4.40
(broad s, 1H, exchangeable), 4.79 (pseudo-t, 1H), 5.73 (d,
J = 7.2 Hz, 1H), 8.02 (s, 1H); FAB HRMS calcd for
C12H17N4O6 (M + 1) 313.1148, found 313.1149. 8c: UV

1186
J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
(CH3OH): lmax = 262 nm (e = 7.600) and 292 nm (e = 3.800);
1H NMR (CD3CN) d (ppm) 3.48 (s, 3H), 3.75 (m, 2H), 4.07 (m,
1H), 4.10 (broad s, exchangeable, 1H), 4.25 (m, 1H), 4.45
(broad s, 1H, exchangeable), 4.65 (pseudo-t, 1H), 5.55 (s, 2H,
exchangeable), 5.83 (d, J = 7.3 Hz, 1H), 7.91 (s, 1H); HRMS
calcd. for C11H16N5O5 298.1151, found 298.1144.
deacetylation. After evaporation, the residues were analyzed
by HPLC (the same conditions as in experiment A) and high
resolution mass spectroscopy. In all cases only the 8-methoxy
derivatives were observed.
Acknowledgment
4.6.1.4. 8-Methoxy-O6-methylguanosine (11a) and 8-amino-
O6-methylguanosine (10a). HPLC eluent: CH3CN/0.02 M
NH4OAc (2/8). 11a: tR = 3.0 min. UV (CH3OH) lmax = 247.5
and 283 nm; FAB HRMS calcd. for C12H18N5O6 (M + 1)
328.1257, found 328.1248. 10a: tR = 2.2 min. lmax = 256 and
293 nm; FAB HRMS calcd. for C11H17N6O5 (M + 1) 313.1260,
found 313.1265. 11a:10a = 86:14 (dH-2 = 4.86 ppm/dH-
2 = 4.78 ppm).
This work was supported in part by the Department of
Energy Grant DE- FC0387-ER60615.
References
[1] P.C. Srivastava, R.K. Robins, R.B. Meyer Jr., in: L.B. Townsend (Ed.),
Chemistry of Nucleosides and Nucleotides, vol. 1, Plenum Press, New
York, 1988 , pp. 113–281 (Chapter 2).
4.6.1.5. N2-Acetyl-8-methoxy-O6-methylguanosine (11b) and
N2-acetyl-8-amino-O6-methylguanosine (10b). HPLC eluent:
CH3CN/0.02 M NH4OAc (15/85). 11b: tR = 5.3 min.
[2] (a) G.I. Birnbaum, P. Lassota, D. Shugar, Biochemistry 23 (1984) 5048–
5053;
(b) S. Ally, G. Tortora, T. Clair, D. Grieco, G. Merlo, D. Katsaros, D.
Ogreid, S.O. Doskeland, T. Jahnsen, Y.S. Cho-Chung, Proc. Natl. Acad.
Sci. U.S.A. 85 (1988) 6319–6322;
l
max = 268 nm. 10b: tR = 3.5 min. lmax = 283 nm; FAB HRMS
calcd. for C13H19N6O6 (M + 1) 355.1366, found 355.1351.
11b:10b = 65:35 (dH-2 = 4.90 ppm/dH-2 = 4.76 ppm). Some
11b was also isolated by reverse phase semi-preparative HPLC
(Alltech Econosil C18 10 mm, 10 mm ꢂ 500 mm), eluent:
CH3OH/H2O (12/88) at 4 mL/min. UV (CH3OH)
(c) A. Pinto, D. Aldinucci, V. Gattei, V. Zagonel, G. Tortora, A. Budillon,
Y.S. Cho-Chung, Proc. Natl. Acad. Sci. U.S.A. 89 (1992) 8884–8888;
(d) S.P. Lagdon, A.A. Ritchie, M. Muir, M. Dodds, A.F. Howie, R.C.F.
Leonard, P.K. Stockman, W.R. Miller, Eur. J. Cancer 34 (1998) 384–388;
(e) M. Masuda, T. Suzuki, M.D. Friesen, J.-L. Ravanat, J. Cadet, B.
Pignatelli, H. Nishino, H. Ohshima, J. Biol. Chem. 276 (2001) 40486–
40496;
l
max = 268 nm (e = 19.100); 1H NMR (CD3CN) d (ppm)
2.29 (s, 3H), 3.63 (m, 1H), 3.73 (dd, J = 12.2 and
J = 3.5 Hz, 1H), 3.89 (broad s, exchangeable, 1H), 3.98 (m,
1H), 4.05 (s, 3H), 4.43 (pseudo-t, 1H), 4.91 (d, pseudo-t, 1H),
5.74 (d, J = 5.9 Hz, 1H), 8.46 (broad s, 1H, exchangeable); FAB
HRMS calcd. for C14H20N5O7 (M + 1) 370.1362, found
370.1348.
(f) G.H. Park, J. Choe, H.-J. Choo, Y.G. Park, J. Sohn, M.-K. Kim, Exp.
Mol. Med. 34 (2002) 184–193;
(g) S.A. Lum Min, R. Tabrizchi, Eur. J. Pharmacol. 318 (1996) 97–100;
(h) J. Choi, J.M. Farley, J. Pharmacol, Exp. Ther. 285 (1998) 588–594;
(i) D.J. Proctor, E. Kierzek, R. Kienzek, P.C. Bevilacqua, J. Am. Chem.
Soc. 125 (2003) 2390–2391.
[3] (a) P.C. Ratsep, R.K. Robins, M.M. Vaghefi, Nucleos. Nucleot. 9 (1990)
197–204, and the references cited therein;
(b) J.H. Lister, M.D. Fenn, The Purines, Supplement 1, John Wiley &
Sons, Inc., New York, 1996, pp. 99–132.
4.6.1.6. 8-Methoxyadenosine (14) [41] and 8-aminoadenosine
(13) [41]. HPLC eluent: CH3CN/0.02 M NH4OAc (15/85).
14: tR = 3.9 min. lmax = 260 nm; FAB MS (M + 1) 298. 13:
[4] (a) J.R. Barrio, M. Namavari, M.E. Phelps, N. Satyamurthy, J. Org. Chem.
61 (1996) 6084–6085;
tR = 2.5 min.
l
max = 273 nm; FAB MS (M + 1) 283.
(b) J.R. Barrio, M. Namavari, M.E. Phelps, N. Satyamurthy, J. Am. Chem.
Soc. 118 (1996) 10408–10411.
14:13 = 81:19 (dH-2 = 8.08 ppm/dH-2 = 8.00 ppm).
[5] M. Namavari, J.R. Barrio, T. Toyokuni, S.S. Gambhir, S.R. Cherry, H.R.
Herschman, M.E. Phelps, N. Satyamurthy, Nucl. Med. Biol. 27 (2000)
157–162.
4.6.2. Products from 0.5 M NH3/2-propanol
After completion of the defluorination (6 weeks for
compound 2b, 26 h for product 5a, 8 h for the purine 2c, 6 h
for the derivative 12 and 4 h for the compound 5b), the products
were treated with 2 M NH3/CH3OH at room temperature for
24 h for a complete deacetylation. The samples were then
evaporated and the residues were analyzed by HPLC (the same
conditions as above) and high resolution mass spectroscopy.
On-line UV spectra were also recorded for each peak. All the
products (8b,c, 10a,b and 13) were identical to the 8-amino
derivatives obtained in experiment A. In all cases only the 8-
amino products were observed.
[6] (a) S.S. Gambhir, J.R. Barrio, L. Wu, M. Iyer, M. Namavari, N. Satya-
murthy, E. Bauer, C. Parrish, D.C. MacLaren, A.R. Borghei, L.A. Green,
S. Sharfstein, A.J. Berk, S.R. Cherry, M.E. Phelps, H.R. Herschman, J.
Nucl. Med. 39 (1998) 2003–2011;
(b) S.S. Gambhir, J.R. Barrio, M.E. Phelps, M. Iyer, M. Namavari, N.
Satyamurthy, L. Wu, L.A. Green, E. Bauer, D.C. MacLaren, K. Nguyen,
A.J. Berk, S.R. Cherry, H.R. Herschman, Proc. Natl. Acad. Sci. U.S.A. 96
(1999) 2333–2338;
(c) S.S. Gambhir, H.R. Herschman, S.R. Cherry, J.R. Barrio, N. Satya-
murthy, T. Toyokuni, M.E. Phelps, S.M. Larson, J. Balatoni, R. Finn, M.
Sadelain, J. Tjuvajev, R. Blasberg, Neoplasia 2 (2000) 118–138.
[7] J.R. Barrio, M. Namavari, R.E. Keen, N. Satyamurthy, Tetrahedron Lett.
39 (1998) 7231–7234.
[8] Y. Kobayashi, I. Kumadaki, A. Ohsawa, S.-I. Murakami, J. Chem. Soc.,
Chem. Commun. (1976) 403–431.
4.6.3. Products from 0.01 M HCl/CH3OH
[9] B.F.L. Li, P.F. Swann, Biochemistry 28 (1989) 5779–5786.
[10] B.L. Gaffney, L.A. Marky, R.A. Jones, Biochemistry 23 (1984) 5686–
5691.
When defluorination was completed (2 h for guanosine and
inosine derivatives 2a–c, 5a,b and 5 days for adenosine
derivative 12), the reaction mixtures were treated with 2 M
NH3/CH3OH at room temperature for 7 h for complete
[11] A.D. Broom, L.B. Townsend, J.W. Jones, R.K. Robins, Biochemistry 3
(1964) 494–500.

J. Liu et al. / Journal of Fluorine Chemistry 127 (2006) 1175–1187
1187
[12] E. Mounetou, E. Debiton, C. Buchdahl, D. Gardette, J.-C. Gramain, J.-C.
Maurizis, A. Veyre, J.-C. Madelmont, J. Med. Chem. 40 (1997) 2902–
2909.
London, 1971 , pp. 121–182 (Chapter 3) (B. Haigh Translator from
Russian).
[29] P.O.P. Ts’o (Ed.), Basic Principles in Nucleic Acid Chemistry, vol. I,
Academic Press, New York, 1974 , pp. 453–584 (Chapter 6).
[30] To demonstrate that a guanosine derivative can be deprotonated at N1
position in 0.5 M NH3 in methanol and that the formed anion will
delocalize the negative charge to C(8) through conjugation, 13C NMR
data were collected at 125 MHz for guanosine and N1-methylguanosine as
model compounds in CD3OD and 0.5 M NH3/CD3OD solution. The
chemical shift for C(8) in guanosine in CD3OD was 138.68 and
137.80 ppm in 0.5 M NH3 in CD3OD. The corresponding chemical shift
values for N1-methylguanosine were 138.57 and 138.54 ppm, respectively.
Thus, in methanolic ammonia C(8) resonance in guanosine is shifted
upfield by ꢃ0.9 ppm. These results suggest that in methanolic ammonia
there is a formation of an anion at N1 and a negative charge distribution to
C(8) in the case of a guanosine while the methyl group in N1-methylgua-
nosine blocks the anion formation.
[13] T. Tsushima, K. Kawada, T. Tsuji, S. Misaki, J. Org. Chem. 47 (1982)
1107–1110.
[14] V. Nair, S.G. Richardson, J. Org. Chem. 45 (1980) 3969–3974.
[15] (a) G.B. Barlin, N.B. Chapman, J. Chem. Soc. (1965) 3017–3022;
(b) G.B. Barlin, J. Chem. Soc. (B) (1967) 954–958.
[16] K.J. Laidler, Chemical Kinetics, Harper Collins Publishers, Inc., New
York, 1987, pp. 18–53 (Chapter 2).
[17] (a) J.F. Bunnett, R.E. Zahler, Chem. Rev. 49 (1951) 273–412;
(b) J.F. Bunnett, Quart. Rev. 12 (1958) 1–16;
(c) E. Buncel, A.R. Norris, K.E. Russell, Quart. Rev. 22 (1968) 123–146;
(d) C.F. Bernasconi, in: H. Zollinger (Ed.), Aromatic Compounds: MTP
International Review of Science, Series 1, vol. 3, Butterworths, London,
1973 , pp. 33–63 (Chapter 2);
(e) C.F. Bernasconi, Acc. Chem. Res. 11 (1978) 147–152;
(f) V.M. Vlasov, J. Fluorine Chem. 61 (1993) 193–216.
[18] J.H. Lister, M.D. Fenn, The Purines, Supplement 1, John Wiley & Sons,
Inc., New York, 1996, pp. 13–16.
[31] V. Markowski, G.R. Sullivan, J.D. Roberts, J. Am. Chem. Soc. 99 (1977)
714–718.
[32] G. Remaud, X.-X. Zhou, C.J. Welch, J. Chattopadhyaya, Tetrahedron 42
(1986) 4057–4067.
[19] J. March, Advanced Organic Chemistry, 4th ed., John Wiley & Sons, New
York, 1992, pp. 641–644 (Chapter 13).
[33] Z. Neiman, Experientia 31 (1975) 996–997.
[20] K.L. Kirk, W. Nagai, L.A. Cohen, J. Am. Chem. Soc. 95 (1973) 8389–
8392.
[34] (a) H.T. Miles, F.B. Howard, J. Frazier, Science 142 (1963) 1458–1463;
(b) G.W. Buchanan, M.J. Bell, Can. J. Chem. 61 (1983) 2445–2448;
(c) G.W. Buchanan, Tetrahedron 45 (1989) 581–604;
[21] H.J. Whitfield, R.J. Griffin, I.R. Hardcastle, A. Henderson, J. Meneyrol, V.
Mesguiche, K.L. Sayle, B.T. Golding, Chem. Commun. (2003) 2802–
2803.
(d) J. Hovinen, C. Glemarec, A. Sandstrom, C. Sund, J. Chattopadhyaya,
Tetrahedron 47 (1991) 4693–4708;
[22] J.F. Bunnett, E.W. Garbisch Jr., K.M. Pruitt, J. Am. Chem. Soc. 79 (1957)
385–391.
(e) R. Marek, V. Sklenar, in: G.A. Webb (Ed.), Annual Reports on NMR
Spectroscopy, vol. 54, Elsevier, Amsterdam, 2005, pp. 201–242.
[35] J.H. Lister, M.D. Fenn, The Purines, Supplement 1, John Wiley & Sons,
Inc., New York, 1996, pp. 372–373.
[23] F. Terrier, Chem. Rev. 82 (1982) 77–152.
[24] R. Alexander, E.C.F. Ko, A.J. Parker, T.J. Broxton, J. Am. Chem. Soc. 90
(1968) 5049–5069.
[36] M. Novak, M. Ruenz, S. Kazerani, K. Toth, T.-M. Nguyen, B. Heinrich, J.
Org. Chem. 67 (2002) 2303–2308.
[25] J. March, Advanced Organic Chemistry, 4th ed., John Wiley & Sons, New
York, 1992, 361 pp. (Chapter 13).
[37] G. Kampf, L.E. Kapinos, R. Griesser, B. Lippert, H. Sigel, J. Chem. Soc.,
Perkin Trans. 2 (2002) 1320–1327.
[26] J. Miller, A.J. Parker, J. Am. Chem. Soc. 83 (1961) 117–123.
[27] D.V. Subbaiah, M.S. Sastry, V.R. Murthy, J. Mol. Struct. 319 (1994) 41–
45.
[38] B. Singer, Biochemistry 11 (1972) 3939–3947.
[39] R.V. Wolfenden, J. Mol. Biol. 40 (1969) 307–310.
[28] N.K. Kochetkov, E.I. Budovskii, in: B. Haigh, D.M. Lord Todd, Brown
(Eds.), Organic Chemistry of Nucleic Acids, Part A, Plenum Press,
[40] M. Ikehara, K. Muneyama, Chem. Pharm. Bull. 14 (1966) 46–49.
[41] R.E. Holmes, R.K. Robins, J. Am. Chem. Soc. 87 (1965) 1772–1776.