Protonation studies of modified adenine and adenine nucleotides by theoretical calculations and 15N NMR

Article abstract of DOI:10.1021/jo0107554

The acid/base character of nucleobases affects phenomena such as self-association, interaction with metal ions, molecular recognition by proteins, and nucleic acid base-pairing. Therefore, the investigation of proton-transfer equilibria of natural and synthetic nucleos(t)ides is of great importance to obtain a deeper understanding of these phenomena. For this purpose, a set of ATP prototypes was investigated using ¹⁵N NMR spectroscopy, and the corresponding adenine bases were investigated by theoretical calculations. ¹⁵N NMR measurements provided not only acidity constants but also information on the protonation site(s) on the adenine ring and regarding the ratio of the singly protonated species in equilibrium. Substituents of different nature and position on the adenine ring did not change the preferred protonation site, which remained N1. However, for 2-thioether-ATP derivatives a mixed population of N1 and N7 singly protonated species was observed. Reduction of basicity of 0.4-1 pK_a units relative to ATP was also observed for all evaluated ATP derivatives, except for 2-Cl-ATP, for which K_a was ca. 10,000-fold lower. To explain the substitution-dependent variations in the experimental pK_a values of the ATP analogues, gas-phase proton affinities (PA), ΔΔG^hyd, and pK_a values of the corresponding adenine bases were calculated using quantum mechanical methods. The computed PA and ΔΔG^hyd values successfully explained the experimental pK_a values. A computational procedure for the prediction of accurate pK_a values was developed using density functional theory and polarizable continuum model calculations. In this procedure, we developed a set of parameters for the polarizable continuum model that was fitted to reproduce experimental pK_a values of nitrogen heterocycles. This method is proposed for the prediction of pK_a values and protonation site(s) of purine analogues that have not been synthesized or analyzed.

Full text of DOI:10.1021/jo0107554

7
90
J . Org. Chem. 2002, 67, 790-802
P r oton a tion Stu d ies of Mod ified Ad en in e a n d Ad en in e Nu cleotid es
by Th eor etica l Ca lcu la tion s a n d 15N NMR
Dan T. Major, Avital Laxer, and Bilha Fischer*
Department of Chemistry, Gonda-Goldschmied Medical Research Center, Bar-Ilan University,
Ramat-Gan 52900, Israel
bfischer@mail.biu.ac.il
Received J uly 25, 2001
The acid/base character of nucleobases affects phenomena such as self-association, interaction with
metal ions, molecular recognition by proteins, and nucleic acid base-pairing. Therefore, the
investigation of proton-transfer equilibria of natural and synthetic nucleos(t)ides is of great
importance to obtain a deeper understanding of these phenomena. For this purpose, a set of ATP
1
5
prototypes was investigated using N NMR spectroscopy, and the corresponding adenine bases
were investigated by theoretical calculations. 15N NMR measurements provided not only acidity
constants but also information on the protonation site(s) on the adenine ring and regarding the
ratio of the singly protonated species in equilibrium. Substituents of different nature and position
on the adenine ring did not change the preferred protonation site, which remained N1. However,
for 2-thioether-ATP derivatives a mixed population of N1 and N7 singly protonated species was
observed. Reduction of basicity of 0.4-1 pK
ATP derivatives, except for 2-Cl-ATP, for which K
substitution-dependent variations in the experimental pK
a
units relative to ATP was also observed for all evaluated
was ca. 10,000-fold lower. To explain the
values of the ATP analogues, gas-phase
a
values of the corresponding adenine bases were calculated
a
a
hyd
proton affinities (PA), ∆∆G , and pK
using quantum mechanical methods. The computed PA and ∆∆G values successfully explained
the experimental pK values. A computational procedure for the prediction of accurate pK values
hyd
a
a
was developed using density functional theory and polarizable continuum model calculations. In
this procedure, we developed a set of parameters for the polarizable continuum model that was
fitted to reproduce experimental pK
a
values of nitrogen heterocycles. This method is proposed for
the prediction of pK values and protonation site(s) of purine analogues that have not been
a
synthesized or analyzed.
In tr od u ction
The most basic position on the adenine ring is N1.3,4
Protonation occurs on the pyrimidine ring of the purine
base even though pyrimidine itself is a weaker base than
imidazole. This protonation is due to the imidazole NH
nitrogen, which is strongly electron-donating to the
Natural and synthetic adenine nucleosides and nucle-
otides comprise a class of biologically important mol-
ecules. Intracellular ATP plays a fundamental role in
energy metabolism, nucleic acid synthesis, pump activi-
ties, and enzyme regulation. Additionally, there is now
widespread evidence that extracellular adenine nucleos-
4b
6
pyrimidine ring. Furthermore, the exocyclic N -amine
in adenine is an efficient electron donor that increases
electron density on N1, as reflected by the 100-fold
(
t)ides mediate signals via membrane-associated adenos-
5
increase in basicity of adenine vs purine. On the other
1
ine and ATP receptors.
hand, the affinity of N7 in adenine nucleotides for proton
6
An important factor in the functionality of adenine
nucleos(t)ides is their ability to participate in acid-base
equilibria. The acidity of natural and synthetic adenine
nucleos(t)ides affects phenomena such as self-association;
interaction with metal ions; interaction with receptors
and enzymes; and base-pairing in nucleic acids, which
influences their structure, stability, and function.2
There are several possible ionizable sites in adenine
nucleos(t)ides: the phosphate, base, or sugar moieties.
is very low. For instance, the pK of N7 of 5′-AMP is
a
(2) (a) Legault, P.; Pardi, A. J . Am. Chem. Soc. 1997, 119, 6621-
6628. (b) Gforer, A.; Schnetter, M.; Wolfrum, J .; Greulich, K. D.
Ber. Bunsen-Ges. Phys. Chem. 1989, 93, 300-304. (c) Cali, R.; Musu-
meci, S.; Rigano, C.; Sammartano, S. Inorg. Chim. Acta 1981, 56, L11-
L13.
(3) (a) Saenger, W. Principles of Nucleic Acid Structure; Springer-
Verlag: New York, 1984; pp 107-110. (b) Sysoeva, S. G.; Utyanskaya,
E. Z.; Vinnik, M. I. Izv. Akad. Nauk, Ser. Khim. 1983, 7, 1505-1511.
(c) Raszka, M. Biochemistry 1974, 13, 4616-4622. (d) Tribolet, R.; Sigel,
H. Eur. J . Biochem. 1987, 163, 353-363.
(4) (a) Sigel, H.; Scheller, K. H.; Milburn, R. M. Inorg. Chem. 1984,
*
To whom correspondence should be addressed: Fax: 972-3-
351250. Tel: 972-3-5318303.
1) (a) Fredholm, B. B.; Arslan, G.; Halldner, L.; Kull, B.; Schulte,
23, 1933. (b) Lister, J . H. Fused Pyrimidines; Interscience: New York,
1971; p. 444. (c) Buchner, P.; Maurer, W.; Ruterjans, H. J . Magn.
Reson. 1978, 29, 45-63.
5
(
G.; Wasserman, W. Naunyn-Schmiedeberg’s Arch. Pharmacol. 2000,
(5) Gonella, N. C.; Roberts, J . D. J . Am. Chem. Soc. 1982, 104, 3162-
3
1
3
62, 364-374. (b) Bhagwat, S. S.; Williams, M. Eur. J . Med. Chem.
997, 32, 183-193. (c) Fischer, B. Expert Opin. Ther. Patents 1999, 9,
85-399.
3164.
(6) Utyanskaya, E. Z.; Lezina, V. P.; Stepanyants, A. U. Izv. Akad.
Nauk, ser. Khim. 1986, 12, 2691-2696.
1
0.1021/jo0107554 CCC: $22.00 © 2002 American Chemical Society
Published on Web 01/10/2002

Protonation Studies of Nucleos(t)ides
1.6; this adenine nitrogen is protonated only after
J . Org. Chem., Vol. 67, No. 3, 2002 791
-
Sch em e 1
7
complete protonation of N1.
A multitude of modified adenines and adenine nucleos-
(
t)ides have been synthesized as inhibitors for enzymes
1
c
or as agonists/antagonists to target receptors. Modifica-
tions, such as substitution of the adenine ring with
electron-donating (ED) or electron-withdrawing (EW)
substituents, can substantially alter the electronic prop-
erties of the purine ring. Such a change in the charge
distribution of the nucleobase is expected to influence the
acid-base properties of the molecule and possibly change
the site of protonation. Therefore, a systematic evaluation
of the basicity of substituted adenines and adenine
nucleos(t)ides is important and may shed light on the
chemical and biochemical properties of these nucleos(t)-
ides.
acidity, or because of low solubility in water. Therefore,
a theoretical approach serves as a possible solution for
such cases. Moreover, a theoretical approach enables the
In general, ionization constants are directly measured
by pH titration monitored by a variety of spectroscopic
techniques. Acidity constants for nucleos(t)ides have been
prediction of pK values of derivatives that have not been
a
synthesized.
1
8
9
10
obtained by H NMR, IR, potentiometric titrations, etc.
However, most of these methods do not provide informa-
tion on the site of protonation in the nucleobase (e.g., N1
or N7). This lack of information is a severe drawback,
especially since the preferred site of protonation may be
altered upon substitution on the nucleobase and several
singly protonated species may coexist.
In this paper, we report pK
a
values and protonation
sites of several prototypic ATP derivatives estimated by
1
5
N NMR. In addition, we propose a theoretical method
for the prediction of basicity and protonation sites of
adenine bases based on Density Functional Theory (DFT)
calculations. The effects of ED and EW substituents on
the pK values of adenine derivatives are discussed in
a
An alternative spectroscopic method is 15N NMR, which
allows unambiguous determination of the protonation
site. This determination is due to the direct measurement
of the nitrogens undergoing protonation and the great
sensitivity of this spectroscopic method to protonation.
For example, protonation of the N1-position in ATP
causes an upfield shift of ca. 70 ppm.11
light of the present results.
Resu lts
Selection of ATP Der iva tives P r ototyp es. The pK
values and protonation sites of synthetic ATP prototypes
a
14,15
used in our previous biochemical studies
were deter-
15
mined by N NMR.
Indeed, 15N NMR has been utilized for studies of
The set of the investigated adenine-modified ATP
derivatives is shown in Scheme 1. The structural changes
included substituents of different electronic nature (ED
or EW) and of different volume substituted at different
protonation in nucleosides, nucleotides, nucleic acids, and
protein-bound nucleotides.1
1a,12
However, as a result of
the limited solubility of nucleobases and nucleosides in
water, various reports describe their protonation in
6
positions on the adenine ring (C2, N , C8).
1
3
DMSO solutions, with trifluoroacetic acid. Therefore,
Syn th esis. Compounds 2 and 3 were prepared from
1
5
11a,13
no pK
a
values are provided based on N NMR.
16
2
-thiol-adenosine by selective S-alkylation of the thi-
1
5
Nevertheless, N NMR may be a suitable tool for pK
a
olate salt of 7 with alkyl bromide/iodide and subsequent
one-pot 5′-triphosphorylation
adenosine, 11, was prepared in 73% yield from 2,6-
dichloro-9â-(2′,3′,5′-tri-O-acetyl)-D-ribofuranosylpurine,
10,17 upon heating to 100° C for 24 h in a sealed ampule
determination for nucleotides, which are highly water
soluble.
14c
(Scheme 2). 2-Chloro-
a
Although the experimental determination of pK values
is usually the most reliable option, theoretical calcula-
tions can be used to supplement and rationalize experi-
mental data. Sometimes, experimental determination of
containing 2 M NH /EtOH (Scheme 3). This method is
3
superior to the common preparation of 2-chloro-adenos-
ine.18 Only NH /EtOH is useful in our procedure and not
pK
a
values of compounds may be complex or even
3
impossible, as a result of either a very high or very low
NH /MeOH, because in the latter case 2-chloro-6-meth-
3
(
(
7) Martin, R. B. Acc. Chem. Res. 1985, 18, 32-38.
(14) (a) Fischer, B.; Boyer, J . L.; Hoyle, C.; Ziganshin, A.; Ralevic,
V.; Knight, G.; Brizzolara, A.; Zimmet, J .; Burnstock, G.; Harden, T.
K.; J acobson, K. A. J . Med. Chem. 1993, 36, 3937-3946. (b) Burnstock,
G.; Fischer, B.; Hoyle, C. H. V.; Maillard, M.; Ziganshin, A. U.;
Brizzolara, A. L.; von Isakovics, A. K.; Boyer, J . L.; Harden, T. K.;
J acobson, K. A. Drug Dev. Res. 1994, 31, 206-219. (c) Halbfinger, E.;
Major, D. T.; Ritzmann, M.; Ubl, J .; Reiser, G.; Boyer, J . L.; Harden,
K. T.; Fischer, B. J . Med. Chem. 1999, 42, 5325-5337 (d) Major,
D. T.; Halbfinger, E.; Fischer, B. J . Med. Chem. 1999, 42, 5338-
5347.
(15) Gendron, F.-P.; Halbfinger, E.; Fischer, B.; Duval, M.; D’Orleans-
J uste, P.; Beaudoin, A. R. J . Med. Chem. 2000, 43, 2239-2247.
(16) (a) Kikugawa, K.; Suehiro, H.; Ichino, M. J . Med. Chem. 1973,
16, 1381-1388. (b) Kikugawa, K.; Suehiro, H.; Yanase, R.; Aoki, A.
Chem. Pharm. Bull. 1977, 25, 1959-1969 and references therein.
(17) (a) Robins, M. J .; Uznanski, B. Can. J . Chem. 1981, 59, 2601-
2606. (b) Nair, V.; Richardson, S. G. Synthesis 1982, 670-672.
(18) Montgomery, J . A.; Hewson, K. J . Heterocycl. Chem. 1964, 1,
213-214.
8) Wang, P.; Izatt, R. M.; Oscarson, J . L.; Gillespie, S. E. J . Phys.
Chem. 1996, 100, 9556-9560.
9) Hofmann, K. P.; Zundel, G. Z. Naturforsch., C: Biosci. 1974, 29,
9-28.
10) Tribolet, R.; Malini-Balakrishnan, R.; Sigel, H. J . Chem. Soc.,
Dalton Trans. 1985, 2291-2303.
11) (a) Markowski, V.; Sullivan, G. R.; Roberts, J . D. J . Am. Chem.
Soc. 1977, 99, 714-718. (b) Buchanan, G. W. Tetrahedron 1989, 45,
81-604.
12) (a) Hovinen, J .; Glemarec, C.; Sandstr o¨ m, A.; Sund, C.; Chat-
(
1
(
(
5
(
topadhyaya, J . Tetrahedron 1991, 47, 4693-4708. (b) Gonella, N. C.;
Nakanishi, H.; Holtwick, J . B.; Horowitz, D. S.; Kanamori, K.; Leonard,
N. J .; Roberts, J . D. J . Am. Chem. Soc. 1983, 105, 2050-2055. (c) Rhee,
Y.; Wang, C.; Gaffney, B. L.; Roger, R. A. J . Am. Chem. Soc. 1993,
15, 8742-8746. (d) Drohat, A. C.; Stivers, J . T. J . Am. Chem. Soc.
000, 122, 1840-1841.
(
1
2
13) Buchanan, G. W.; Bell, M. J . Can. J . Chem. 1983, 2445-
2
448.

7
92 J . Org. Chem., Vol. 67, No. 3, 2002
Major et al.
Sch em e 2a
substituted adenines, described below, are discussed in
the light of changes in π-electron density due to substitu-
tion.
The electron-withdrawing effect of chlorine at the C2
position in 2-Cl-ATP is clearly observed by the downfield
shift of N1 as compared to the shift of N1 of ATP (∆δ
+6.5 ppm), Table 1, and also in the chemical shift of the
exocyclic amine in 2-Cl-ATP (∆δ +5 ppm). These results
are consistent with our previous theoretical findings on
electron density on 2-Cl-adenine nitrogens.1 The deshield-
ing effect of Br substituent in 8-Br-ATP is more pro-
nounced at the endocyclic nitrogens, especially on N7 (∆δ
4d
+
7.6 ppm). Ν9, on the other hand, undergoes an upfield
6
shift (∆δ -3.4 ppm). 2-Alkylthio- and N -methyl substi-
tutions have only a minor effect on the exo- and endocy-
clic nitrogens, except for N3 where an upfield shift (∆δ
-
5.4 and -4.4 ppm, respectively) is observed. This find-
ing indicates higher electron density due to mesomeric
contributions. In all of the ATP derivatives, with the
exception of 2-MeS-ATP, N1 is deshielded relative to
N1 in ATP (∆δ 1.6-6.5 ppm), indicating a lower π-elec-
tron density electron density due to adenine substitu-
tion.
15N NMR Deter m in a tion of p K
tives 1-6. Protonation equilibria of several nucleobases
Va lu es of Der iva -
a
a
Conditions: (a) (1) NaOH, isolation, (2) CH3I/DMF for 1 or
(
1) 0.25 M NaOH/MeOH, isolation, (2) BrCH2C(CH3)3/DMF
for 2; (b) (1) POCl3, PO(OCH3)3, (2) (Bu3NH )2H2P2O7 , (3)
hydrolysis.
+
2-
1
13
8,20
have previously been elucidated by H and C NMR.
15
Yet, N NMR is a direct probe for protonation studies of
the nucleos(t)ide bases4
c,20b
and allows unambiguous
Sch em e 3a
determination of the protonation site. Furthermore, the
effects on 15N shifts are much larger than H shifts and
are therefore less influenced by environmental factors
1
other than base protonation.2 In addition, unlike
a
1
H
NMR measurements that are performed in D
2
O and
provide pD values, 15N NMR measurements are per-
formed in H
relevant.
2
O and provide values that are biologically
On the other hand, a major disadvantage of this
1
5
spectroscopy is the extremely low sensitivity of N at the
-
6
natural abundance level, which is 3.8 × 10 of that of a
proton at constant magnetic field.21 Therefore, N NMR
15
a
determination of pK values of adenine nucleotides at
natural abundance requires concentrated solutions, typi-
cally 0.5-1 M. At this concentration range, phenomena
that are connected with high concentration such as self-
association22 and high ionic strength should be consid-
ered.
a
Conditions: (a) POCl3, CH3CN; (b) (CH3)2CHCH2CH2ONO,
+
CCl4; (c) NH3/EtOH; (d) (1) POCl3, PO(OCH3)3, (2) (Bu3NH )2-
H2P2O7 , (3) hydrolysis.
15
5
We have reported that a 23-fold increase of ( N) -
2
-
adenine concentration (from 35 mM to 0.8 M) resulted
1
5
only in a marginal shift of the nitrogens’ N chemical
oxy-purine riboside was also obtained in approximately
0% yield. 2-Chloro-adenosine was then 5′-triphospho-
rylated to give derivative 5.
15N NMR Mea su r em en ts. 15
2
3
shifts (i.e., ∆δ of 0.4-1 ppm). The same observation was
6
1
5
made with ( N)
4
-purine riboside; a 75-fold increase of
purine riboside concentration (from 13 mM to 0.97 M)
N NMR spectra of triso-
dium salts of ATP derivatives 1-6 were measured at
natural abundance in H O solutions at pH 5.2-5.4. The
results are summarized in Table 1.
(
19) Sierzputowska-Gracz, H.; Wiewiorowski, M.; Kozerski, L.; von
2
Philipsborn, W. Nucleic Acids Res. 1984, 12, 6247-6258 and references
therein.
1
5
(20) (a) Liu, M.; Farrant, R. D.; Lindon, J . C.; Barraclough, P.
Spectrosc. Lett. 1991, 24, 1115-1121. (b) Lister, J . H. The Purines,
Supplement 1; J ohn Wiley: New York, 1996; p 372.
(21) (a) Kanamori, K.; Roberts, J . D. Acc. Chem. Res. 1983, 16, 35-
A major parameter determining the N chemical
shift of a heterocyclic nitrogen is its π-electron density.19
For instance, N9 in the purine ring system has a high
π-electron density since it provides two electrons to the
delocalized π-orbital system, whereas N1, N3, and N7
4
1. (b) Berger, S.; Braun, S.; Kalinowski, H.-O. NMR Spectroscopy of
the Nonmetallic Elements; J ohn Wiley: New York, 1996, Chapter 4,
pp 229-237.
have a lower π-electron density since they provide
(22) (a) Scheller, K. H.; Hofstetter, F.; Mitchell, P. R.; Prijs, B.; Sigel,
H. J . Am. Chem. Soc. 1981, 103, 247-260. (b) Scheller, K. H.; Sigel,
H. J . Am. Chem. Soc. 1983, 105, 5891-5900.
only one electron. Therefore, 15N chemical shift of N9 is
shifted 40-60 ppm upfield relative to N1, N3, and N7
(23) Laxer, A.; Major, D. T.; Gottlieb, H.; Fischer, B. J . Org. Chem.
(
Table 1). Changes of nitrogens’ chemical shifts in
2001, 66, 5463-5481.

Protonation Studies of Nucleos(t)ides
J . Org. Chem., Vol. 67, No. 3, 2002 793
Ta ble 1. 15N NMR Ch em ica l Sh ifts of ATP Der iva tives 1-6
compound
derivative
pH
N1
N3
N7
N9
NH2
1
2
3
4
5
6
ATP
2-MeS-ATP
5.24
5.32
5.36
5.31
5.41
5.15
-164.5
-164.7
-160.8
-162.9
-158.0
-161.7
-164.2
-169.6
-169.1
-168.6
-166.9
-160.6
-148.9
-151.2
-150.7
-150.1
-149.0
-141.3
-211.0
-213.0
-212.5
-211.8
-211.0
-214.4
-302.7
-304.0
-303.7
-302.4
-297.7
-301.3
2-(CH3)3CCH2S-ATP
6
N -Me-ATP
2-Cl-ATP
8-Br-ATP
3
Ta ble 2. ∆δ of Nitr ogen s Mea su r ed a t th e Begin n in g a n d
ppm, indicating an increase of its sp character due to
En d of Titr a tion of ATP Der iva tives
protonation (Table 3). A downfield shift of 9-10 ppm of
the exocyclic amine is attributed to its increased sp
character. This is readily seen in the mesomeric struc-
tures stabilizing the formed cationic nucleotide species.
This effect is most pronounced (∆δ 12 ppm) for N -Me-
ATP. The positive inductive effect of the methyl group
renders the lone electron pair of N -amine more available
to mesomeric stabilization of the protonated adenine ring.
The positive inductive effect of the methyl group is also
reflected in a significant N3 upfield shift with ∆δ of 8.3
ppm.
δobs
N1 N3 N7 N9 NH2
2
∆
compd
derivative
ATP
2-MeS-ATP
pH range
1
2
3
2.05-5.99 -66.4 5.3 4.8 6.9 9.8
2.14-5.94 -51.4 1.5 -2.3 5.9 9.9
2-(CH3)3CCH2S - 1.90-5.90 -52.2 1.1 -5.0 6.1 9.2
6
6
ATP
6
4
5
6
N -Me-ATP
2-Cl-ATP
8-Br-ATP
1.77-6.36 -67.3 8.3 4.3 7.1 11.8
a
-0.15-5.41 -8.2 2.7
b
6.3 7.9
1.92-6.09 -62.4 4.8 3.4 7.5 9.0
a
Titration was stopped at pH -0.15. b The chemical shift of N7
was not observed in the acidic range.
Plotting the chemical shift of N1 vs pH provides a
sigmoid curve (e.g., Figure 2, for derivative 6). The pK
a
Ta ble 3. F itted P a r a m eter s fr om Titr a tion Cu r ves
value is obtained from the inflection point, which is
determined by the second derivative of the fitted sigmoid
compd
derivative
∆δ(N1)extrp a
pKab
1
2
3
4
5
6
ATP
-70
-73
4.16
3.22
3.54
3.88
-0.2
3.83
function (Figure 3). In this way, pK
for derivatives 1-6 (Table 3). The observed pK
ATP was 4.16. This is in a good agreement with reported
a
values were obtained
2-MeS-ATP
2-(CH3)3CCH2S-ATP
N -Me-ATP
2-Cl-ATP
8-Br-ATP
a
value for
6
-73
-67
2
5
26
pK
4.03.4
pH titration of most of the samples resulted in a drastic
a
values obtained by other methods: 4.15, 4.10, and
a
a
Full ∆δ for N1 were obtained from the fitted asymptotic values
of the sigmoid titration curve (e.g., Figure 2). Fitting of the
upfield shift in the N1 chemical shift, whereas the signals
experimental data points to a sigmoid curve was achieved by eq
6
of other adenine nitrogens (N3, N7, N9, N ) are slightly
b
7
.
pKa is the inflection point of the fitted curve (e.g., Figure 2).
moved downfield (Table 2), reflecting the electron poor
nature of the protonated ring system. For derivatives 2
and 3 we observed an upfield shift also for N7, although
less than that for N1 (Table 2). This finding indicates
the existence of an N7 singly protonated species in
The inflection point was determined by calculating the second
derivative of the fitted curve. The estimated error for pH meter
reading is (0.02, except for reading below pH 0 (i.e., last
measurement at pH -0.15 for compound 5).
had a negligible effect on the corresponding 15N NMR
spectra (i.e., ∆δ of 0.7-0.8 ppm).24 These minor changes
of chemical shifts indicate that base stacking and H-
bonding are limited at 0.8-1 M nucleoside concentrations
and/or do not influence 15N NMR chemical shifts. On the
other hand, protonation of N1 leads to a huge upfield shift
of ca. 70 ppm. In any case, the overwhelmingly large ∆δ
of N1 masks environmental factors that are about 2
orders of magnitude smaller.
In addition, Sigel et al. have shown that self-associa-
tion in nucleotides decreases considerably by going
from adenosine to ATP.4 Therefore, we intended to
deduce pK
in a facile way in H
+
addition to the N1-H form. Strong electron-donating
groups at C8 are known to induce protonation of the
adenine ring at N7 in addition to N1.1 Here, we find
that strong electron-donating groups at C2 have practi-
cally the same effect.
2a
The 15N NMR method is suitable for the determination
of acidity constants of nucleoside-triphosphate deriva-
tives but not of the corresponding monophosphate ana-
logues, because of their low solubility in acidic solutions.
AMP, for instance, precipitates out of solution at pH 4.
Clearly, this method is not applicable to nucleosides as
a result of their extremely low solubility in aqueous
solutions.
,22
values of adenine nucleotides with 15N NMR
a
2
O at natural abundance, expecting
Because of the low sensitivity of natural abundance
only a marginal error under these experimental condi-
tions.
Solutions of trisodium salts of the nucleotide deriva-
1
5
N NMR, one must use 0.5-1 M solutions of nucleotides
in water for the determination of pK values. This
a
requirement makes this method impractical for cases
where such concentrations are not readily available.
Therefore, we developed a simple theoretical procedure
tives 1-6, at pH 6, were titrated with dilute hydrochloric
1
5
acid at 25 °C, and their N NMR spectra were recorded
at 60.8 MHz. Changes of 15N NMR spectra as a function
of the pH were monitored, and a gradual upfield shift of
N1 was noted (Figure 1).
for the prediction of pK
Once this method reproduces the substituent-dependent
variations in experimentally determined pK values of
nucleotides, we can apply this method to the prediction
a
values of modified adenine bases.
a
The increased acidity of the solution caused changes
in all nitrogens’ chemical shifts (Table 2). The most
pronounced change was an upfield shift of N1 of ca. 70
(25) Watanabe, S.; Evenson, L.; Gulz, I. J . Biol. Chem. 1963, 238,
3
24-327.
(24) Laxer, A.; Fischer, B. Unpublished results.
(26) Tribolet, R.; Sigel, H. Eur. J . Biochem. 1988, 170, 617-626.

7
94 J . Org. Chem., Vol. 67, No. 3, 2002
Major et al.
F igu r e 1. Changes of 15N NMR spectra of 2-(2,2-dimethylpropyl)thio-ATP, 3, as a function of pH (N1 is marked by an arrow),
recorded at 60.8 MHz, 0.91 M, 25 °C.
27p
of pK
a
values of novel chemically modified nucleobase
acids,27j uracil derivatives,
and model amino acid
residues.2 Other models for the calculation of pK
values were developed for transition states and interme-
diates, as in the enzymatic methylation of cytosine,
or pK values of acetic acid during R-proton abstrac-
7d
derivatives. Furthermore, this method should enable the
prediction of the preferential protonation site(s) (e.g., N1/
N7/N3) and may help rationalize adenine substituent
a
2
7k
effects on pK
Th eor etica l Ca lcu la tion of p K
t u t ed Ad en in e Ba ses. Several theoretical studies on
the determination of pK of various organic compounds
have been reported.27 These studies include the cal-
culation of the pK of individual molecules such as
a
.
a
7l
tion.2
a
Va lu es of Su bsti-
The free energy of protonation in solution is related to
the pK
a
by the following expression:
1
a
prot
a
pK )
∆Gaq
(1)
a
amines,2 nitriles, hydroxybenzoic acids, carboxylic
7g
27h
27i
2.303RT

Protonation Studies of Nucleos(t)ides
J . Org. Chem., Vol. 67, No. 3, 2002 795
The computation of accurate, absolute pK values is a
a
demanding task as a result of errors both in the calcu-
lated gas-phase free energy of protonation and free
energy of solvation.28 An error of 1.36 kcal/mol in the
calculated free energy of protonation leads to a pK
of about one pK unit.
However, J ang and Goddard III et al. recently reported
first principles calculation of pK values for 5-substituted
uracils.2 This work presented an important step toward
the prediction of accurate first principles pK values.
a
shift
a
a
7p
a
Nonetheless, the free energy of solvation of the proton,
for which several experimental values exist, was explic-
itly fitted to reproduce the experimental pK values for
a
the 5-substituted uracil series. Moreover, the dielectric
constant used to represent the quantum mechanical
region within the Poisson-Boltzmann continuum model
F igu r e 2. Changes of N1 15N chemical shift vs pH in 8-Br-
was fitted to reproduce the experimental pK
Thus, it is questionable whether this model would
correctly predict pK values for other heterocycles unless
a
values.
ATP, 6. A 0.9 M solution of 8-Br-ATP at pH 6.09 was titrated
with dilute HCl at 25 °C, and changes of 1 N NMR spectra (at
5
a
6
0.8 MHz) were monitored as a function of pH.
new parameters are developed.
An alternative approach involves the calculation of
semiempirical pK
a
values.2 This can be performed by
7k,l
calculating the free energy of protonation in aqueous
solution (eq 2) and pK values (eq 1) for a range of
compounds with known experimental pK values. The
calculated pK values, which contain systematic errors
a
a
a
inherent to the computational methods, may be fitted to
the experimental ones by a linear regression analysis
leading to the following correlation equation:
F igu r e 3. Determination of pK
second derivative of the fitted sigmoid curve (shown in Figure
).
a
for 8-Br-ATP based on the
exp
calc
pKa ) a‚pKa + b
(3)
2
The free energy of protonation in aqueous solution,
Accurate, “experimental-like” pK values of new com-
pounds may then be predicted using the linear equation
obtained.
a
prot
∆
G
, may be calculated using a thermodynamic cycle:
aq
The goal of this theoretical study is to determine and
rationalize the accurate, experimental-like pK
nitrogen heterocycles of biological importance. In par-
ticular, the pK values of derivatives of adenine (Scheme
) were calculated, and the substituent-dependent varia-
tions in these pK values were compared with the
a
values of
a
4
a
experimental results for the corresponding ATP deriva-
tives (Scheme 1). Moreover, the influence of ED and EW
substituents on the pK of adenine derivatives 12-22 was
a
prot
aq
prot
hyd
hyd
hyd
∆
G
) ∆Ggas + ∆GBH
+
- ∆GB - ∆GH
+
)
investigated, as well as the localization of the protonation
site(s) on the molecule and the identification of the major
protonated species. This approach is complementary to
the experimental data presented above for ATP deriva-
prot
hyd
∆
G
+ ∆∆G
(2)
gas
The gas-phase protonation free energy is usually obtained
from quantum mechanical calculations, while the free
energy of hydration may be calculated by discrete solvent
models or by continuum models. Most theoretical ap-
proaches differ concerning solvent treatment. For this
treatment several solvation models exist, each having its
strengths and weaknesses. Some of the discrete solvation
(
27) (a) Warshel, A. Biochemistry 1981, 20, 3167-3177. (b) J or-
gensen, W. L.; Briggs, J . M.; Gao, J . J . Am. Chem. Soc. 1987, 109,
6857-6858. (c) J orgensen, W. L.; Briggs, J . M. J . Am. Chem. Soc. 1989,
1
1
1
11, 4190-4197. (d) Lim, C.; Bashford, D.; Karplus, M. J . Phys. Chem.
991, 95, 5610-5620. (e) Gao, J .; Pavelites, J . J . J . Am. Chem. Soc.
992, 114, 1912-1914. (f) Yang, B.; Wright, J .; Eldefrawi, M. E.;
Sovitj, P.; MacKerell, A. D., J r. J . Am. Chem. Soc. 1994, 116, 8722-
8
2
732. (g) Kallies, B.; Mitzner, R. J . Phys. Chem. B 1997, 101, 2959-
a
methods used in pK calculations include free energy
967. (h) Colominas, C.; Orozco, M.; Luque, F. J .; Borrel, J . I.; Teixid o´ ,
2
7b,c,e
perturbation simulations in explicit water
and the
J . J . Org. Chem. 1998, 63, 4947-4953. (i) Shapley, W. A.; Bacskay, G.
B.; Warr, G. G. J . Phys. Chem. B 1998, 102, 1938-1944. (j) Sch u¨ u¨ r-
mann, G.; Cossi, M.; Barone, V.; Tomasi, J . J . Phys. Chem. A 1998,
Langevin dipole model.2 On the other hand, the com-
7a
monly used continuum models are based on the Born
1
02, 6706-6712. (k) Per a¨ kyl a¨ , M. J . Am. Chem. Soc. 1998, 120,
12895-12902. (l) Per a¨ kyl a¨ , M. Phys. Chem. Chem. Phys. 1999, 1,
643-5647. (m) Gargallo, R.; Sotriffer, C. A.; Liedl, K. R.; Rode, B. M.
model,2 the Poisson-Boltzmann (PB) equation,
7f
27d,p
and
various implementations of the polarizable continuum
5
J . Comput.-Aided Mol. Des. 1999, 13, 611-623. (n) G u¨ ven, A.; Yekeler,
H.; O¨ zkan, R. J . Mol. Struct. (THEOCHEM) 2000, 499, 13-19. (o)
Alkorta I.; Gonzales, E.; J agerovic, N.; Elguero, J .; Flammang, R.
J . Phys. Org. Chem. 2000, 13, 372-381. (p) J ang, Y. H.; Sowers,
L. C.; C¸ a gˇ in. T.; Goddard, W. A., III. J . Phys. Chem. A 2001, 105, 274-
model (PCM).2
7g-l
The advantage of continuum models
over discrete methods is that the solvent reaction field
may be incorporated into the quantum mechanical Hamil-
tonian, and thus an accurate description of the solvated
solute may be achieved.
2
80.
(28) Cramer, C. J .; Truhlar, D. G. Chem. Rev. 1999, 99, 2161-2200.

7
96 J . Org. Chem., Vol. 67, No. 3, 2002
Major et al.
Ta ble 4. Com p u ted a n d Exp er im en ta l Ga s-P h a se P r oton
Affin ities (P A, k ca l/m ol) a n d p Ka Va lu es of Va r iou s
Heter ocycles Used To Cr ea te th e Lin ea r P A a n d p Ka
Cor r ela tion Equ a tion sa
compd
derivative
adenine
purine
cytosine
uracil
thymine
guanine
imidazole (Im)
4-Cl-1-Me-Im
4-Br-Im
2-Br-Im
thiazole
PAcalc b PAexp c pKacalc d pKaexp e
1
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
2
224.3
220.6
226.6
206.4
208.0
227.2
224.2
ndf
225.3
219.9
227.0
207.1
209.0
227.4
225.3
ndf
3.95
2.56
4.47
9.39
9.72
3.31
6.37
2.63
3.22
3.31
2.54
5.48
1.77
3.89
0.88
5.24
7.10
6.40
9.10
6.04
5.72
5.95
17.65
4.12
2.3
4.5
9.5
9.9
3.3
6.95
3.1
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
f
f
nd
nd
3.6
ndf
ndf
3.85
3.44
5.36
1.3
3.45
0.65
5.25
6.82
6
9.11
5.97
5.68
6.02
f
f
nd
nd
ndf
211.7
ndf
209.6
222.0
226.4
228.1
234.2
226.8
225.5
226.4
ndf
2- NH2-thiazole
pyrimidine
ndf
212.8
2-NH2-pyrimidine ndf
pyrazine
pyridine
210.3
222.1
226.6
228.2
234.0
226.3
225.4
226.4
ndf
2-NH2-pyridine
3-NH2-pyridine
4-NH2-pyridine
2-Me-pyridine
3-Me-pyridine
4-Me-pyridine
pyrrole
F igu r e 4. Schematic representation of the computational
strategy adopted to explain the substituent-dependent varia-
17.5
tion in pK
a
of the investigated ATP prototypes.
corr
coeff
0.994
0.993
g
slopeg
interceptg
rms
0.89
24.16
0.625
0.81
-0.19
0.29
Sch em e 4
errorh
a
Figures 5 and 6. b PA computed according to eqs 4-6. The
electronic energy was calculated at the B3LYP/aug-cc-pVTZ(-f)//
B3LYP/6-31G(d,p) level. The thermal corrections were computed
at the B3LYP/6-31G(d,p) level. The computed PAs were obtained
exp
calc
from a linear regression equation of the form PA ) a‚PA
+
b. Experimental data was taken from ref 33. d pKa computed
according to eqs 1-3 at the B3LYP/aug-cc-pVTZ(-f)-PCM-pKa level.
The computed pKa values were obtained from a linear regression
c
tives 1-6. This approach can also provide predicted
pK values for adenine analogues where it is impossible
to obtain these values by experiment and provide pK
exp
calc
e
equation of the form pKa
) a‚pKa
+ b. Experimental data
was taken from ref 36. Not determined. g Data for linear regres-
sion equations. h Root-mean-square error (kcal/mol and pKa units)
in calculated values.
f
a
a
values of novel derivatives that have not been synthe-
sized.
These results were, in turn, utilized to explain the
Com p u ta tion a l Str a tegy. The current approach,
trend in the substitution-dependent experimental pK
a
combining DFT and PCM29 calculations for the prediction
values for the corresponding ATP derivatives (1-6). The
results of steps A and C are presented in Table 4 and
Figures 5 and 6, while the results of steps B and D are
presented in Table 5.
a
and interpretation of adenine pK values, involved the
following four steps (Figure 4): (A) Calculating the gas-
phase proton affinity (PA) for a set of 16 heterocycles (12,
2
3-28, 34, 36-43, Table 4) for which experimental PAs
(
A) Ca lcu la tion of Ga s-P h a se P r oton Affin ity for
are available. These heterocycles are structurally related
to the target adenine derivatives (12-22). The computed
PAs were correlated with experimental values. (B) Pre-
dicting accurate, experimental-like PAs for the target
adenine derivatives (12-22) on the basis of results
a Ca libr a tion Set of Heter ocycles. A prerequisite to
rationalize the basicity of nucleic acid bases in a given
medium is to understand their inherent basicity in the
gas phase. The basicity of a given base, B, in the gas
phase may be described by its PA. The PA is defined as
the negative of the enthalpy of protonation of B:
obtained in A. (C) Calculating ∆∆Ghydand pK
values for
a
a set of 23 heterocycles (12, 23-44, Table 4) related to
the target adenine derivatives (12-22). These computed
prot
+
∆Hgas
+
pK
eq 3). In this approach, the atomic radii used to build
up the molecular cavity within PCM were fitted to
a
values were correlated with experimental values
B(g) + H (g)
BH (g)
8
(
prot
PA ) -∆Hgas (T)
(4)
reproduce the experimental pK
accurate, experimental-like pK
a
values. (D) Predicting
values of the adenine
a
prot
prot
prot
prot
5
2
derivatives (12-22), using the correlation equation ob-
tained in C.
∆Hgas (T) ) ∆Eelec + ∆Ezpe + ∆Evib (T) + RT (5)
prot
The gas-phase protonation enthalpy, ∆Hgas (T), is de-
fined as the difference in electronic energy, ∆Eelec, zero-
(
29) Miertus, S.; Scrocco, E.; Tomasi, J . Chem. Phys. 1981, 55, 117-
prot
1
29.

Protonation Studies of Nucleos(t)ides
J . Org. Chem., Vol. 67, No. 3, 2002 797
Ta ble 5. P r ed icted P A, GB, ∆∆Gh yd (a ll k ca l/m ol), a n d
p Ka Va lu es for N1 P r oton a tion of Va r iou s Ad en in e
Der iva tives
compd
derivative
PAa
GBb
∆∆Ghyd c
pKad
1
1
1
1
1
1
1
1
2
2
2
2
3
4
5
6
7
8
9
0
1
2
adenine (Ad)
2-MeS-Ad
224.3 218.5
228.9 221.8
211.45
216.02
nd
3.95
3.25
nd
3.71
-0.50
3.93
5.55
6.06
3.71
3.89
e
e
2-(CH3)3CCH2S-Ad 231.3 224.7
6
N -Me-Ad
2-Cl-Ad
228.8 222.6
221.4 212.3
223.3 217.3
229.2 224.0
231.2 227.1
231.1 224.3
223.8 216.2
227.7 222.2
215.97
212.83
210.36
214.27
216.6
217.74
209.33
215.10
8-Br-Ad
2-MeNH-Ad
8-NH2-Ad
6
N -diMe-Ad
8-Cl-Ad
8-MeS-Ad
4.02
a
PA computed according to eqs 4-6. The electronic energy was
calculated at the B3LYP/aug-cc-pVTZ(-f)//B3LYP/6-31G(d,p) level.
The thermal corrections were computed at the B3LYP/6-31G(d,p)
level. The computed PAs were obtained from a linear regression
F igu r e 5. Correlation between the calculated and the corre-
sponding experimental gas-phase proton affinities (PA) for
heterocycles 12, 23-28, 34, and 36-43 (Table 4). PA was
computed according to eqs 4-6.
exp
calc
equation of the form PA
) a‚PA
+ b, with the coefficients
b
found in Table 4. The gas-phase basicity (GB) was defined as
prot
c
-
∆Ggas
.
Free energy of solvation was calculated using PCM-
d
pKa-B3LYP/6-31G(d,p). pKa computed according to eqs 1-3 at
the B3LYP/aug-cc-pVTZ(-f)-PCM-pKa level. The computed pKa
values were obtained from a linear regression equation of the form
exp
calc
e
pKa
) a‚pKa
+ b, with the coefficients found in Table 4. Not
determined.
minimal influence on the calculated PA.30 The results are
presented in Table 4.
The absolute PAs obtained with the B3LYP/aug-cc-
pVTZ(-f) functional agree more closely with the experi-
mental values than B3LYP/cc-pVTZ(-f). The root-mean-
square errors were 1.24 and 2.16 kcal/mol, respectively.
However, these errors are partly due to systematic errors
in the calculated PA values. The results may be improved
if relative PAs are calculated instead of absolute values.
Thus, a linear regression analysis was performed (Figure
F igu r e 6. Correlation between the calculated and the corre-
5
) where the experimental and calculated PAs were
sponding experimental pK
a
values for heterocycles 12 and 23-
related by the following equation:
4
4 (Table 4). pK was computed according to eqs 1-3. The
a
PCM atomic radii were fitted to reproduce experimental pK
a
exp
calc
values (Table 6).
PA ) a‚PA + b
(6)
prot
prot
2
point energy, ∆Ezpe , and vibrational energy, ∆Evib (T),
The R values obtained were 0.994 and 0.986 with and
without diffuse functions, respectively. The root-mean-
square errors were 0.63 and 0.88 kcal/mol, respectively.
The values obtained are well within the estimated
between the protonated and unprotonated species. The
/ RT term includes the translation energy of the proton
2
and the ∆(PV) term. Analogous to the PA, the gas-phase
basicity (GB) is defined as -∆Ggas
In a recent comprehensive post-Hartree-Fock (HF)
study, accurate PA values for nucleic acid bases were
calculated.30 However, post-HF methods are still expen-
sive from a computational point of view. It has previously
been shown that the B3LYP functional yields reason-
able PAs for nucleic acid bases.32 Thus, we decided to use
the far more economical B3LYP DFT method.
5
prot
33
.
experimental error of 2 kcal/mol.
(
B) Ca lcu la tion of Ga s-P h a se P r oton Affin ity for
Su bstitu ted Ad en in e Der iva tives. From the linear
regression analysis equation obtained (Figure 5), it is
possible to predict the expected experimental PAs of the
investigated adenine derivatives (12-22, Table 5).
For all cases, only the N -amino adenine tautomer was
considered since it is far more stable than the imino
3
1
6
14d
tautomers. Likewise, only the N9-H adenine tautomer
The ability of the B3LYP functional to reproduce the
experimental PA values was examined for a set of 16
heterocycles (12, 23-28, 34, 36-43) for which experi-
mental PA values are available (Table 4).33 These het-
erocycles are structurally related to the target adenine
analogues. The B3LYP functional was chosen together
with the cc-pVTZ(-f) and aug-cc-pVTZ(-f) basis sets.34 The
latter basis set includes diffuse functions on all atoms.
The selected set of heterocycles span a range of PAs from
(30) Podolyan, Y.; Gorb, L.; Leszczynski, J . J . Phys. Chem. A 2000,
1
04, 7346-7352.
31) (a) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B: Condens. Matter
988, 37, 785-789. (b) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H.
(
1
Chem. Phys. Lett. 1989, 157, 200-206. (c) Becke, A. D. J . Chem. Phys.
993, 98, 5648-5652.
32) (a) Russo, N.; Toscano, M.; Grand, A.; J olibois, F. J . Comput.
1
(
Chem. 1998, 19, 989-1000. (b) Wolken, J . K.; Ture e` ek, F. J . Am. Soc.
Mass. Spectrom. 2000, 11, 1065-1071.
(33) Hunter, E. P. L.; Lias, S. G. J . Phys. Chem. Ref. Data 1998,
2
7, 413-656.
2
07.1 to 234.2 kcal/mol. For the five natural nucleic acid
(34) (a) Dunning, T. H. J . Chem. Phys. 1989, 90, 1007-1023. (b)
Kendall, R. A.; Dunning, T. H.; Harrison, R. J . J . Chem. Phys. 1992,
bases (12, 24-27) only the most stable gas-phase tau-
tomer was considered, as inclusion of rare tautomers has
9
6, 6796-6806. (c) Woon, D. E.; Dunning, T. H. J . Phys. Chem. 1993,
98, 1358-1371.

7
98 J . Org. Chem., Vol. 67, No. 3, 2002
Major et al.
was considered since this is the major tautomer in
equilibrium in solution.23 In addition, only pK
calculated for the N9-H tautomer are relevant when
analyzing effects on acidity constants of adenine nucle-
otides, where the N9 position bears a phosphorylated
ribose.
Ta ble 6. Sca lin g F a ctor s for th e P a u lin g vd W Ra d ii Set
F itted To Rep r od u ce Exp er im en ta l p Ka Va lu es of
Nitr ogen Heter ocycles (12, 23-46)
a
values
atom
vdW radii (Å)
scaling factor
H
1.2
1.2
1.5
1.5
1.2
1.0
1.35
1.0
1.0
1.0
1.2
1.2
H (polar)
C
N1 was found to be the favored site of protonation in
the gas phase for all the derivatives (results not shown).1
Table 5 presents the results obtained using B3LYP/aug-
cc-pVTZ(-f). The PA of adenine, which is included in the
calibration set, is predicted to be 224.3 kcal/mol, close to
the experimental value with an error of 1 kcal/mol. The
PA of 2-MeS-adenine is increased by 4.6 kcal/mol relative
to adenine, clearly revealing the strong ED character of
N
O
S
Cl
Br
4d
1.4
1.85
1.80
1.95
spheres centered on the solute nuclei, and the radii of
each sphere is usually taken as the van der Waals (vdW)
radii. In the calculation of the electrostatic part of the
solvation energy, each atomic radius is multiplied by a
3 3 2
the thioether group. In 2-(CH ) CCH S-adenine, this
effect is even more pronounced as is expected by the
greater ED effect of the 2,2-dimethylpropyl group. A
strong ED effect is also observed for 2-MeNH-adenine,
where the PA for N1 protonation is almost 5 kcal/mol
greater than that of the parent compound. On the other
hand, the 2-Cl-adenine derivative is less basic than
adenine by more than 3 kcal/mol as a result of the EW
effect of the chlorine atom. The ED effect of the meth-
ylthio group is reduced when the substitution is at the
C8-position rather than at C2. In this case, the PA is
increased by only 3.4 kcal/mol compared to adenine. If
3
5
scaling factor so that it mimics the first hydration shell.
An advantage of using the scaling factor is that it is used
only in the calculation of the electrostatic part of the
solvation energy. Thus, it may be changed in order to
fine-tune the electrostatic interaction between solvent
and solute.
Usually the atomic radii or the scaling factors are fitted
to reproduce experimental solvation energies. For neutral
species, this leads to very accurate results.35 However,
for ions, the experimental solvation energies are not very
accurate because of the large absolute values. Experi-
adenine is substituted at the C8 position with an NH
2
mental pK
obtained with high accuracy. Thus, calculated pK
be greatly improved by fitting the PCM atomic radii to
reproduce experimental pK values (denoted PCM-pK ).
a
values, on the other hand, are usually
group the PA is increased by 6.9 kcal/mol. An EW effect
is observed in 8-Cl-adenine, albeit smaller than for 2-Cl-
adenine because of the more remote position of the
a
may
6
a
a
chlorine atom. The N -Me-adenine derivative shows
The scaling factors for the relevant atoms in deriva-
tives 12-22 were optimized to reproduce experimental
a
pK values, and the results are shown in Table 6. A
description of the fitting process of the atomic radii
appears in Methods.
increased basicity of N1 as a result of an ED effect. The
PA is further increased if an additional methyl is added
6
to N .
(
a
C) Cr ea tin g th e Lin ea r pK Equ a tion . The free
energy of protonation in aqueous solution was calculated
according to eq 2. The gas-phase part of the protonation
free energy was calculated using the B3LYP functional
with the aug-cc-pVTZ(-f) basis set. The difference in free
energy of hydration was calculated using the polarizable
continuum model. The free energy of solvation in con-
tinuum solvation models may be written
A set of 23 heterocycles with known pK
) were included in the calibration set to assemble a
linear relationship between the experimental and the
calculated pK values (Figure 6). The selection of the
a
values36 (Table
4
a
calibration set was based on the following criteria: (A)
Heterocycles that are structurally related to the target
molecules of this study (i.e., adenine derivatives), namely,
pyridines, pyrimidines, imidazoles, and purines. Two
thiazoles were also included to allow the parametrization
solv
es
cav
dis-rep
∆
G
) ∆G + ∆G + ∆G
(7)
for sulfur. (B) Heterocycles that span a wide range of pK
i.e., 0.65-17.5). (C) Even distribution of pK values
within the range of 0.65-17.5.
a
where the superscripts stand for electrostatic, cavitation,
and dispersion-repulsion. The electrostatic contribution
to the solvation free energy is represented quantum
mechanically and accounts for the mutual polarization
of the solute and solvent. The polarization of the solvent
due to the solute is reproduced by a set of point charges
distributed on the molecular cavity (i.e., the reaction
field). This reaction field induces a polarization of the
solute charge distribution, and the solute in turn polar-
izes the solvent. This process is continued until self-
consistency is achieved. The solute-solvent interaction
is dependent upon the shape and size of the molecular
cavity. A cavity that is too small will overestimate the
solute-solvent interaction, while too-large a cavity will
underestimate this interaction. Thus, a crucial parameter
in continuum models is the definition of the molecular
boundary. In PCM, the solute is embedded in a cavity of
molecular shape surrounded by a continuous dielectric
medium. The cavity is formed by a set of interlocking
(
a
A study by Per a¨ kyl a¨ showed that different functional
groups have inherent systematic errors in the calculated
27k,l
energies.
These systematic errors may be due to errors
in the calculated hydration energies. Marten et al.
suggested that systematic errors in solvation energies are
due to nonelectrostatic hydrogen bonding properties and
are related to the acid/base properties of hydrogen bond
37
donors and acceptors. Thus, the best results may be
(
35) Bachs, M.; Luque, F. J .; Orozco, M. J . Comput. Chem. 1994,
1
5, 446-454.
(36) (a) CRC Handbook of Chemistry and Physics, 8th ed.; Lide, D.
R., Ed.; CRC Press: Boca Raton, FL, 1999-2000. (b) Grimmett, M. R.
In Comprehensive Heterocyclic Chemistry. The Structure, Reactions,
Synthesis and Uses of Heterocyclic Compounds; Katritzky, A. R., Rees,
C. W., Eds.; Pergamon Press Ltd.: Oxford, 1984; Vol. 5, p 384.
(37) Marten, B.; Kim, K.; Cortis, C.; Friesner, R. A.; Murphy, R. B.;
Ringnalda, M. N.; Sitkoff, D.; Honig, B. J . Phys. Chem. 1996, 100,
11775-11788.

Protonation Studies of Nucleos(t)ides
J . Org. Chem., Vol. 67, No. 3, 2002 799
obtained by choosing a set of compounds that have
related structures, where the systematic errors are
similar.
(D) P r ed icted pK
The pK values of 11 adenine derivatives (Scheme 4, 12-
22) were predicted with the linear pK equation (eqs 1-3)
using the PCM-pK scaling factors in the solvation
calculations (Table 5).
a
Va lu es for Ad en in e Der iva tives.
a
a
A range of pK
to allow wider application of the linear pK
a
values, as wide as possible, was sought
equation.
a
a
As mentioned above, in all of the neutral adenine
derivatives, the N9-H amino tautomer is the major and
Hence, the protonation of both neutral species (e.g.,
adenine) and anions (e.g., uracil) was investigated. In
addition, molecules with several tautomeric structures
a
relevant one for comparison of pK values with adenine
nucleos(t)ides. Additionally, the 15N NMR spectra show
that protonation occurs at the N1 position, except for
(e.g., purine, adenine, guanine, cytosine, uracil, and
thymine) were included, to investigate the robustness of
the equation. In these cases, the most stable tautomer
in solution was considered. For purine and adenine the
2
-MeS- and 2-(CH
also occurs. It has also been shown for 8-NH
that the most basic site on the adenine ring is N7.
2
Thus, with the exception of 2-MeS-adenine and 8-NH -
3
)
3
CCH
2
S-ATP, where N7 protonation
2
-adenosine
1
2a
N9-H tautomer was used.38a In the case of guanine and
cytosine, the amino-oxo tautomer was considered,38a,b
adenine, only the N9-H amino tautomer and its N1
protonated conjugated acid will be discussed.
and for uracil and thymine the di-oxo tautomer was
considered.38b However, no molecules with conformational
The calculated pK
calibration set, is very close to the experimental value
with an error of only 0.17 pK units. Substitution at the
C2 position of the adenine ring with a chlorine atom leads
to a dramatic decrease in pK , with a predicted value of
0.5. The pK of 2-MeNH-adenine is increased relative
to adenine by 1.6 pK units. Surprisingly, the pK of
-MeS-adenine is lowered relative to adenine. Replacing
a
of adenine, which is included in the
flexibility (i.e., nucleosides, nucleotides) were included,
as this would greatly complicate the creation of the pK
a
a
equation.
Table 4 and Figure 6 present the calculated pK
obtained with the PCM-pK parametrization. Using
, an R value of 0.993 was found and the rms
error was 0.29 pK units, which corresponds to 0.40 kcal/
a
values
a
a
-
a
2
PCM-pK
a
a
a
a
2
mol. The largest and most systematic errors for the
calibration set are found for the imidazole series, where
6
the N -hydrogens with one or two methyl groups de-
creases the basicity of N1. The substitutions at the C8-
the pK
a
a
is overestimated by 0.38-0.59 pK units. This
position with chlorine and bromine led to calculated pK
a
indicates that the atomic radii scaling factors may be
improved for imidazoles.
values similar to those of adenine, with values of 3.89
and 3.93, respectively. The 8-MeS-adenine is predicted
The importance of choosing appropriate atomic radii
scaling factors is clearly seen when comparing the results
obtained with the PCM-pK parametrization with results
a
to have a pK
its PA is greater than that of adenine. On the other hand,
the N7 nitrogen in 8-NH -adenine is found to be sub-
a
very similar to that of adenine, although
2
obtained using two other models: PCM-UAHF39 at the
HF/6-31G(d,p) level, and PCM with B3LYP/6-31G(d,p)
and the Pauling atomic radii set with standard scaling
factors (1.2 for all atoms except for polar hydrogens where
a scaling factor of 1.0 is used). PCM-UAHF provided
stantially more basic than N1 in adenine. N7 protonation
in 8-NH -adenine is found to be slightly more favorable
2
than N1 protonation; the difference between N1 and N7
protonation being less than 0.5 kcal/mol. That is, for
8-NH -adenine a mixed population of monoprotonated
2
excellent pK
a
values for carboxylic acids in a study by
species is expected.
Sch u¨ u¨ rmann et al. with correlation coefficients of 0.97
Comparison of the calculated pK values for the above-
mentioned adenine derivatives with corresponding ex-
a
units.2 Therefore, the
7j
and standard error of 0.24 pK
a
PCM-UAHF model was tested in this study to evaluate
its performance for heterocycles. However, PCM-UAHF
leads to a correlation coefficient of only 0.879 and a large
perimental values is not feasible because of the low
solubility of these compounds in water. However, it may
be instructive to compare the relative pK values of the
a
rms error of 1.22 pK
with standard scaling factors, a correlation coefficient of
.922 and rms error of 1.04 pK units were found. It is
a
units. Using the Pauling radii set
adenine derivatives with those obtained experimentally
for the corresponding ATP derivatives 1-6 (Table 3). The
a
predicted trend in pK values show excellent agreement
0
a
thus clear that the fitting of the scaling factors, used in
the creation of the molecular cavity to reproduce experi-
mental pK values, greatly improves the results. Reduc-
a
ing the scaling factor for the hydrogen-bonding atoms (N,
O, H, S) to about 1.0 compared to the standard scaling
with the experimental results, and the conclusions drawn
regarding the substituent-dependent adenine basicity
should also be valid for the corresponding ATP deriva-
tives.
factor of 1.2 led to results closest to experimental pK
a
Discu ssion
values (Tables 5 and 6). Reduced scaling factors cor-
respond to a stronger interaction between solute and
solvent. This may be necessary to compensate for non-
electrostatic effects in solvation, such as charge transfer,
which continuum models cannot take into account.
The effect of the nature and position of adenine
substituents on the acidity constant of ATP derivatives
cannot be analyzed solely on the basis of structural
considerations, such as inductive or resonance effect,
H-bonding, steric effect, or hybridization. This is due to
Moreover, the PCM-pK
a
parametrization is simple, and
solvent effects that also contribute to the measured pK
a
uses a single scaling factor for each atom type.
value. Specifically, the effects of ED or EW groups are
much weakened in aqueous solutions and may even be
reversed.
The most dramatic effect on N1 basicity was observed
for 2-Cl-ATP, with more than 10,000-fold reduction of
(38) (a) Shcherbakova, I.; Elguero, J .; Katritzky, A. R. Adv. Hetero-
cycl. Chem. 2000, 77, 51-113. (b) Brown, D. J . In Comprehensive
Heterocyclic Chemistry. The Structure, Reactions, Synthesis and Uses
of Heterocyclic Compounds. Katritzky, A. R., Rees, C. W., Eds.;
Pergamon Press Ltd.: Oxford, 1984; Vol. 3, pp 67-68.
basicity compared with ATP. The pK
2-Cl-ATP is based on extrapolation (Tables 2 and 3)
a
value of -0.2 for
(
39) Barone, V.; Cossi, M.; Tomasi, J . J . Chem. Phys. 1997, 107,
3
210-3221.

8
00 J . Org. Chem., Vol. 67, No. 3, 2002
Major et al.
because of the instability of nucleotides and inaccuracy
of pH measurements at highly acidic solutions. This low
value implies that the chloro substituent reduces con-
siderably the electron density on the neighboring N1, as
indeed reflected by 15N chemical shift of N1 (Table 1).
Although 8-NH
study, it is of interest to compare the basicity of 8-NH
adenine to that of adenine. Interestingly, for 8-NH
adenine, 19, N1 protonation is clearly favored in the gas
phase. However, hydration makes the N7 protonation
favored in the aqueous phase by 0.5 kcal/mol (Table 5)
2
-ATP was not investigated in this
2
-
-
2
a
This result was supported by the calculated pK for 2-Cl-
+
+
adenine, 16 (Table 5). The strong EW nature of Cl is
clearly seen from the gas phase PA and GB (Table 5),
and solvation does not significantly change the basicity
of N1.
a
with pK values of 6.06 and 6.43 for N1H and N7H
singly protonated species. Chattopadhyaya et al. dem-
onstrated by 15N NMR and semiempirical calculations
that 8-amino and alkylamino substituents influence the
properties of the resultant adenine ring system, including
the relative basicity of N1 and N7 positions. The extent
of N7 vs N1 protonation by TFA in DMSO was controlled
by the nature of the substituent at the 8-amino group.
For instance, 8-amino-adenosine was 66% N1 and 34%
N7 protonated.1 This finding fits our results considering
that DMSO is a less polar solvent than water.
However, an EW group (Br) at C8 has only a minor
effect on the nucleotide basicity, reducing the pK value
a
of 8-Br-ATP to 3.83. 8-Br minimally destabilizes the
protonated species as a result of the remote position, and
a
this was confirmed by the calculated pK value for 8-Br-
adenine. The gas-phase PA and GB of 8-Br- and 8-Cl-
adenine, 17 and 21, showed a slight decrease in proton
affinity relative to adenine (Table 5). Thus, solvation
slightly increases the basicity of 8-Br- and 8-Cl-adenine.
2a
6
A slight reduction in N1 basicity for N -Me-ATP,
reflected in a pK value of 3.88, was rather unexpected
a
on the basis of the high proton affinity of the correspond-
ing adenine derivative, 15, (228.8 kcal/mol). This PA
value is due to a positive inductive effect of N -Me group.
Yet, unfavorable hydration (∆∆G
verses the ED effect and the predicted pK
Me-adenine, 15, (3.71) vs adenine (3.95) is in a good
agreement with the variations of the measured pK of
ATP derivative 4 vs 1 (3.88 vs 4.16). The less favorable
solvation may be attributed to increased delocalization
of the protonated species, as in the case of the 2-thioether
derivatives. Moreover, steric hindrance reduces the sol-
On the other hand, for 2-MeS-ATP, a pK
than that of ATP may be expected as a result of a
mesomeric effect of the ED group. Yet, it is one pK unit
a
value higher
6
a
solv
216 kcal/mol) re-
lower than that of ATP. Apparently, the hydration effect
reduces the expected basicity. This is probably due to a
less favorable solvation of the protonated nucleotide as
6
a
value of N -
a
indicated by the quantum mechanical calculations. Analy-
prot
aq
sis of the two terms comprising the ∆G
of 2-MeS-
adenine in eq 2, namely, ∆Ggas and ∆∆Ghyd, may give
an insight into the origin of the shift in pK . The PA and
prot
a
the GB of 2-MeS-adenine, 13, showed a clear ED effect
+
vent-accessible surface of the N1H group and restricts
rotations of water molecules H-bonded to N1H , leading
compared to that of adenine (Table 5). However, it is
+
hyd
found that the ∆∆G of 2-MeS-adenine is greater than
to an unfavorable entropy of solvation as seen in the
that for adenine. This finding is due to the less favorable
solvation of the N1-protonated 2-MeS-adenine compared
to that of the N1-protonated adenine. The reduced
solvation of the cationic 2-MeS-adenine is probably a
result of the conjugation between the sulfur 3p (normal)
lone pair and the adenine ring that delocalizes the
hyd
6
6
∆
∆G term (Table 5). Indeed, N ,N -diMe-adenosine is
1
2a
a
even less basic, with an experimental pK of 3.50. For
the latter derivative, minor protonation was also observed
1
2a
at N3, although no protonation at this site was detected
6
for N -Me-ATP.
+
positive charge. Moreover, the NH moiety is not as
Con clu sion s
solvent-accessible as in ATP because of the bulky thiom-
ethyl group. Thus, the basicity of 2-MeS-adenine is
lowered in a highly polar environment, although this does
not mean that its inherent basicity (i.e., PA or GB) is
lower than that of adenine.
In this study the acid/base character of a set of ATP
prototypes was investigated using N NMR spectroscopy,
15
and the corresponding set of adenine bases was investi-
gated by theoretical calculations. The prototypes studied
are ATP analogues with ED or EW substituents at the
When comparing 2-MeS-ATP with 2-(CH
3 3 2
) CCH S-
6
ATP a change of pK from 3.22 to 3.54 is observed. This
a
C2-, C8-, or N -position of the adenine ring.
change is due to the field contribution of the larger alkyl
group, as observed from the calculated PA of 231.3 vs
15
N NMR was utilized for the determination of acidity
constants for ATP derivatives. The acidity constant
2
28.9 kcal/mol for (CH
3
)
3
CCH
2
S-adenine and 2-MeS-
15
2
obtained by N NMR in H O solution for ATP is in good
adenine, 14 and 13, respectively.
agreement with values obtained by various techniques.
Furthermore, 15N NMR measurements provided not only
acidity constants for ATP derivatives 1-6 but also
information on the protonation site(s) in the adenine ring
and on the ratio between the different monoprotonated
species existing in equilibrium. This is important infor-
mation that cannot be easily obtained by most other
titration techniques.
In both 2-thioether-ATP derivatives investigated (2, 3),
a small extent of N7-protonation was observed in addition
to N1 protonation, resulting in a mixed population of N1
+
+
and N7 singly protonated species. Both N1H and N7H
species are formed at the same pH range, and the
chemical shift of each of these nitrogens is equally
effected by protonation. We therefore conclude, on the
basis of ∆δ values in Table 2, that for compounds 2 and
To supplement the experimental results, DFT calcula-
tions combined with PCM solvation calculations were
performed, using PCM parameters that were fitted to
+
+
3
9
the ratio of equilibrating N1H and N7H species are
5:5, and 90:10, respectively. This conclusion is supported
by our calculations that predict the N7 protonation to be
less favorable by only 0.5 kcal/mol, with pK values of
reproduce experimental pK
cycles (PCM-pK ). The quantum mechanical calculations
(i.e., PA, GB, and ∆∆G ) provided sound explanations
for the variations in pK of the ATP derivatives due to
adenine ring substitutions. Therefore, this method is
a
values of nitrogen hetero-
a
a
+
+
hyd
3
.25 for 2-MeS-adenine N1H species and 2.85 for N7H
species. These findings indicate that both monoproto-
nated species may exist in solution.
a

Protonation Studies of Nucleos(t)ides
J . Org. Chem., Vol. 67, No. 3, 2002 801
). 13C NMR (CD
-
3
proposed for the prediction of acidity and protonation
site(s) of purine analogues that have not been synthesized
or analyzed.
(d, J ) 1.5 Hz, 2H, SCH
2
), 1.05 (s, 9H, CH
3
OD, 300 MHz) δ 167.2 (s, C-2), 156.7 (s, C-6), 151.2 (s, C-4),
40.6 (d, C-8), 121.1 (s, C-5), 90.7 (d, C-1′), 87.4 (d, C-4′), 75.3
d, C-2′), 72.3 (d, C-3′), 63.3 (t, C-5′), 45.3 (t, SCH ), 29.2 (q,
S 370.1549, found 370.1562.
-[(2′,2′-Dim et h ylp r op yl)t h ioet h er ] Ad en osin e 5′-T-
1
(
2
We found that substituents of different nature and
position on the adenine ring in ATP derivatives 1-6 did
not change the preferential protonation site, namely, the
N1 position. For the 2-thioether-ATP derivatives we
observed a mixed population of N1 and N7 monoproto-
nated species. The ratio between these species was 9:1
for compound 3. This is a rather unexpected finding
because of the remote position of the substituent.
Despite the different nature and position of substitu-
ents on the adenine moiety, for most evaluated ATP
derivatives (2-5) the basicity of N1 was reduced by 0.4-1
3 24 5 4
CH ). HRMS calcd for C15H N O
2
1
4c
r ip h osp h a te (3). Obtained according to literature procedure
in 71% yield. 1H NMR (D
2
O, 200 MHz) δ 8.21 (s, 1H, H-8),
.01 (d, J ) 4.5 Hz, 1H, H-1′), (H-2′ is hidden by water peak),
4.50 (t, J ) 4.50 Hz, 1H, H-3′), 4.31 (br.s, 1H, H-4′), 4.20 (br.s,
6
3
1
2H, H-5′), 3.00 (s, 2H, SCH
00 MHz) δ -6.67 (d), -10.26 (d), -20.87 (t). High-resolution
FAB calcd for C15 S 608.0382, found 608.0380.
-Ch lor o-a d en osin e (11). A sealed ampule containing 2,6-
2 3 2
), 0.85 (s, 9H, CH ). P NMR (D O,
2
25 5 13 3
H N O P
2
1
7
dichloro-9â-(2′,3′,5′-tri-O-acetyl)-D-ribofuranosylpurine (650
mg, 1.45 mmol) and 2 M NH /EtOH (60 mL) was heated to
00° C for 24 h. Stirring was continued for an additional 24 h
3
a
pK units, relative to ATP. The most dramatic effect was
observed for 2-Cl-ATP, where the acidity constant was
ca. 10,000-fold lower.
1
at room temperature, and then the solvent was removed under
reduced pressure. The residue was purified on silica gel
3
chromatography (elution with CHCl /MeOH; 9:1). The product
was obtained as a yellowish solid (322 mg, 73% yield). Spectral
data is consistent with the literature.
Exp er im en ta l Section
Ap p a r a tu s a n d Mea su r em en ts. NMR spectra were mea-
sured on a Bruker AC-200 instrument (200.2 and 80.3 MHz
Meth od s of Th eor etica l Ca lcu la tion s. Gas-phase calcu-
31
lations were performed using the B3LYP functional. The
1
31
for H and P, respectively) or on a Bruker DPX-300 (300.1
geometries of all molecules were optimized with B3LYP and
1
13
15
and 75.5 MHz for H and C, respectively). N NMR spectra
were recorded on a Bruker DMX-600 instrument (60.8 MHz
42
the 6-31G(d,p) basis set, and the nature of the stationary
points was investigated. In cases were local maxima were
found, the structures were reoptimized with small geometry
perturbations, stricter convergence criteria, more accurate
grids, or accurate Hessian matrix. Zero-point energies and
thermal corrections at 1 atm and 298 K were included. The
computed zero-point energies were scaled by a factor of 0.9806,
1
5
for N) and measured with nitromethane (δ ) 0 ppm) as an
external standard. Negative chemical shifts are upfield from
nitromethane. Determination of apparent pH values was made
with a Hanna Instruments pH meter (HI 8521) equipped with
an Orion microcombination pH electrode (9802) or Hanna
instruments electrode (FC200). Neutral solutions of trisodium
salts of the nucleotide derivatives at 0.9-1.2 M concentration
range were titrated with dilute hydrochloric acid. 15N NMR
chemical shift of N1 was monitored as a function of the pH.
A five-parameter sigmoid function was fitted to the data using
SigmaPlot 2000 (SPSS, Inc.):
and the enthalpy and entropy contributions were scaled by
43
0
.9989 and 1.0015, respectively. Single-point calculations
were then performed using B3LYP with the cc-pVTZ(-f) basis
34
set augmented with diffuse functions on all atoms (B3LYP/
aug-cc-pVTZ(-f)//B3LYP/6-31G(d,p)). The cc-pVTZ(-f) basis set
is the cc-pVTZ basis set of Dunning et al., but with the outmost
polarization and diffuse functions deleted. In the case of
bromine derivatives, the effective core potential LAVCP**
a
δ ) δ +
(7)
0
44
-
((pH-pH0)/b) c
basis set was used for the bromine instead of 6-31G(d,p) and
(1 + e
)
f-functions were kept in the cc-pVTZ basis set. For molecules
containing anions, diffuse functions were added to the 6-31G-
Full ∆δ values for N1 (and other adenine nitrogens) were
obtained from the fitted asymptotic values of the sigmoid
(d,p) basis set on all heavy atoms, both in the gas phase and
in solution.
titration curves (Table 3). The pK
point, was determined by the second derivative of the fitted
sigmoid function. Ionic strength near the pK value is 2.8-
a
value, namely, the inflection
Solvent effects were included using PCM29 with the B3LYP/
3
9
6
6
-31G(d,p) functional, and those for PCM-UAHF with HF/
-31G(d,p). All compounds were optimized in aqueous solution
a
3
.3 M.
Ma ter ia ls. 8-Bromo-adenosine 5′-triphosphate and N6-
methyl-adenosine were purchased from Sigma Chemical Co.
-Methylthio-adenosine was prepared according to a literature
using the ICOMP)2 keyword, while subsequent single point
calculations used the more sophisticated ICOMP)4 option.
Within the PCM approach, both the standard Pauling atomic
radii with standard scaling factors were used (Gaussian 98
keywords: Pauling, AlphaH)1.0, Alpha)1.2) and the Pauling
2
4
1
procedure. 5′-Phosphorylated adenosine derivatives were
prepared and purified according to our previous report.14c
set with a set of scaling factors fitted to reproduce pK
(PCM-pK ).
a
values
2
-[(2′,2′-Dim eth yl-p r op yl)-th ioeth er ] Ad en osin e (8). A
1
6
a
suspension of 2-thiol-adenosine (700 mg, 3.34 mmol in 45
mL of MeOH) was dissolved in 0.25 M NaOH (10.3 mL) and
stirred at room temperature for 1.5 h. The solvent was
evaporated under high vacuum. The obtained thiolate sodium
salt was then dissolved in dry DMF (40 mL), and 1-bromo-2,
In the fitting process, the Pauling atomic radii set was used
and the scaling factor was varied for each atom type (N, C, H,
H(polar), S, Cl, Br) until the best fit with the experimental
a
pK values was obtained. The scaling factor was initially
2
-dimethylpropyl (0.32 mL, 2.54 mmol) was added. The reac-
tion mixture was stirred for 48 h at 80° C. TLC (CHCl /MeOH;
:5) indicated that the entire starting material was consumed,
defined as f ) 0.9 + 0.1n where n ) 0, 1, 2, 3, 4, 5. After finding
an optimal scaling factor for each atom, changes of (0.05 were
also tested. The optimal scaling factors are presented in Table
6, together with the atomic radii as defined in the Pauling set.
3
4
f
and a new product was formed (R 0.33). The solvent was
evaporated under reduced pressure, and the residue was
separated on a silica gel column (CHCl /MeOH; 9:1). The
product was obtained as a yellowish solid (445 mg, 52%). H
NMR (CD OD, 300 MHz) δ 8.17 (s, 1H, H-8), 5.93 (d, J ) 6
3
(
42) (a) Ditchfield, R.; Hehre, W. J .; Pople, J . A. J . Chem. Phys. 1971,
1
5
4, 724-728. (b) Hehre, W. J .; Ditchfield, R.; Pople, J . A. J . Chem.
3
Phys. 1972, 56, 2257-2261. (c) Hariharan, P. C.; Pople, J . A. Mol. Phys.
Hz, 1H, H-1′), 4.73 (t, J ) 5.5 Hz, 1H, H-2′), 4.34 (dd, J ) 5,
1974, 27, 209-214. (d) Gordon, M. S. Chem. Phys. Lett. 1980, 76, 163-
1
2
68. (e) Hariharan, P. C.; Pople, J . A. Theo. Chem. Acta 1973, 28, 213-
3
1
.5 Hz, 1H, H-3′), 4.14 (q, J ) 3 Hz, 1H, H-4′), 3.89 (dd, J )
2.5, 3 Hz, 1H, H-5′), 3.75 (dd, J ) 12.5, 3 Hz, 1H, H-5′), 3.24
22. (f) Francl, M. M.; Pietro, W. J .; Hehre, W. J .; Binkley, J . S.;
Gordon, M. S.; DeFrees, D. J .; Pople, J . A. J . Chem. Phys. 1982, 77,
3
654-3665.
(
40) (a) Hall, G. G. Proc. R. Soc. London, Ser. A 1951, 205, 541-
52. (b) Roothaan, C. C. J . Rev. Mod. Phys. 1951, 23, 69-89.
41) Macfarlane, D. E. Methods Enzymol. 1992, 215, 137-142.
(43) Scott, A. P.; Radom, L. J . Phys. Chem. 1996, 100, 16502-16513.
(44) (a) Hay, P. J .; Wadt, W. R. J . Chem. Phys. 1985, 82, 270-283.
5
(
(b) Wadt, W. R.; Hay, P. J . J . Chem. Phys. 1985, 82, 284-298.

8
02 J . Org. Chem., Vol. 67, No. 3, 2002
Major et al.
4
5
All calculations were performed using the Gaussian 98 and
J aguar 4.0 programs.
Ack n ow led gm en t. This work was supported in part
by the Marcus Center for Medicinal Chemistry. Dan T.
Major thanks the Israeli Ministry of Science for financial
support.
4
6
(
45) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J . R.; Zakrzewski, V. G.; Montgomery, J . A.;
Stratmann, R. E.; Burant, J . C.; Dapprich, S.; Millam, J . M.; Daniels,
A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J .; Barone, V.;
Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford,
S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma,
K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J . B.;
Cioslowski, J .; Ortiz, J . V.; Stefanov, B. B.; Liu, G.; Liashenko, A.;
Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J .; Keith,
T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.;
Challacombe, M.; Gill, P. M. W.; J ohnson, B. G.; Chen, W.; Wong, M.
W.; Andres, J . L.; Head-Gordon, M.; Replogle, E. S.; Pople, J . A.
Gaussian 98, Revision A.7; Gaussian, Inc.: Pittsburgh, PA, 1998.
Su p p or tin g In for m a tion Ava ila ble: Z-matrices with the
computed total energies for adenine derivatives 12-22 and
heterocycles 23-44. This material is available free of charge
via the Internet at http://pubs.acs.org.
J O0107554
(46) J aguar 4.0; Schr o¨ dinger, Inc.: Pasadena, CA, 2000.