Binding of an Acetic Acid Ligand to Adenosine: A Low-Temperature NMR Study

Article abstract of DOI:10.1021/ja038630z

Binding of an acetic acid (HAc) ligand to adenosine (A) was studied by ¹H NMR spectroscopic techniques. Using a low-melting deuterated Freon mixture as solvent, liquid-state measurements could be performed in the slow exchange regime and allowe

Full text of DOI:10.1021/ja038630z

Published on Web 01/29/2004
Binding of an Acetic Acid Ligand to Adenosine: A
Low-Temperature NMR Study
Eline M. Bas´ılio Janke, Hans-Heinrich Limbach, and Klaus Weisz*,†
Contribution from the Institut fu¨r Chemie, Freie UniVersita¨t Berlin,
Takustrasse 3, D-14195 Berlin, Germany
Received September 20, 2003; E-mail: weisz@uni-greifswald.de
1
Abstract: Binding of an acetic acid (HAc) ligand to adenosine (A) was studied by H NMR spectroscopic
techniques. Using a low-melting deuterated Freon mixture as solvent, liquid-state measurements could be
performed in the slow exchange regime and allowed a detailed characterization of the formed associates.
Thus, at 128 K, trimolecular complexes A‚HAc2 and A2‚HAc with both Watson-Crick and Hoogsteen sites
of the central adenine base occupied coexist in various amounts depending on the adenosine:acetic acid
molar ratio. Whereas the carboxylic acid OH proton is located closer to the acid for all hydrogen bonds
formed, a more deshielded proton at the Watson-Crick site is evidence for a stronger hydrogen bond as
compared to the Hoogsteen interaction. For the binding of acetic acid to an adenosine-thymidine base
pair in either a Watson-Crick or a Hoogsteen configuration, hydrogen bonds to the available adenine
binding site are strengthened as compared to the corresponding hydrogen bonds in the A‚HAc2 complex.
Introduction
strength and specificity of adenine-carboxylic acid interactions
has previously been investigated in aprotic solvents by NMR
The potential of nucleobases to form well-defined hydrogen-
bonded base pairs is not only a major determinant of nucleic
acid structure but is also fundamental to important biological
processes such as replication and transcription. In addition,
hydrogen-bond interactions between nucleobases and amino acid
side chains are believed to play a crucial role in the recognition
of specific nucleotide sequences by DNA-binding proteins.1a,b
Thus, whereas Watson-Crick pairing between the adenine (A)
and the uracil (U) or thymine (T) base constitutes a structural
key element in double-stranded RNA or DNA, multiple proton
donor and acceptor sites at the adenine base enable its interaction
with suitable functional groups in various geometries. Clearly,
understanding the structural basis of ligand binding to adenine
is not only important to DNA-protein interactions but also to
the development of synthetic receptors for adenine derivatives.2a-i
Acetic acid, a simple carboxylic acid, can not only serve as
a bidentate adenine receptor molecule by use of its carboxylic
acid functionality, but also as a model for aspartic or glutamic
acid side chains involved in DNA-protein contacts. The
spectroscopic techniques.3-5 However, due to the fast exchange
of individual species under the solution conditions employed,
these studies failed to provide unambiguous information on the
specificity and strength of individual hydrogen bonds formed.
In particular, the preference of ligands to bind in either a
Hoogsteen or a Watson-Crick geometry remains elusive. To
circumvent averaging effects and to characterize hydrogen-
bonded complexes by liquid state NMR in more detail, we have
recently performed NMR measurements at very low tempera-
tures by employing a deuterated Freon mixture CDClF2/CDF3
as solvent. With measurements in the liquid state as low as 113
K, individual hydrogen-bonded complexes of nucleobases in
slow exchange could be observed and characterized in detail
for the first time in solution.6a-d We now present NMR studies
on the association of adenosine with acetic acid at low and
ambient temperatures to explore the strength and preference of
hydrogen bonding in the formed complexes.
Experimental Section
† Present address: Institut fu¨r Chemie und Biochemie, Ernst-Moritz-
Arndt-Universita¨t Greifswald, Soldmannstrasse 16, D-17487 Greifswald,
Germany.
(1) (a) Pabo, C. O.; Sauer, R. T. Annu. ReV. Biochem. 1992, 61, 1053-1095.
(b) Luscombe, N. M.; Laskowski, R. A.; Thornton, J. M. Nucleic Acids
Res. 2001, 29, 2860-2874.
Methods. NMR experiments were performed on a Bruker AMX500
spectrometer. Temperatures were adjusted by a Eurotherm Variable
1
Temperature Unit to an accuracy of (1.0 °C. H chemical shifts in
chloroform at 293 K were referenced relative to CHCl3 (δH ) 7.24
ppm) and in a Freon mixture relative to CHClF2 (δH ) 7.13 ppm).
(2) (a) Goswami, S.; Hamilton, A. D. J. Am. Chem. Soc. 1989, 111, 3425-
3426. (b) Zimmermann, S. C.; Wu, W. J. Am. Chem. Soc. 1989, 111, 8054-
8055. (c) Adrian, J. C.; Wilcox, C. S. J. Am. Chem. Soc. 1989, 111, 8055-
8057. (d) Rebek, J., Jr. Acc. Chem. Res. 1990, 23, 399-404. (e) Conn, M.
M.; Deslongchamp, G.; De Mendoza, J.; Rebek, J., Jr. J. Am. Chem. Soc.
1993, 115, 3548-3557. (f) Chen, H.; Ogo, S.; Fish, R. H. J. Am. Chem.
Soc. 1996, 118, 4993-5001. (g) Spivak, D.; Gilmore, M. A.; Shea, K. J.
J. Am. Chem. Soc. 1997, 119, 4388-4393. (h) Alkorta, I.; Elguero, J.;
Goswami, S.; Mukherjee, R. J. Chem. Soc., Perkin Trans. 2 2002, 894-
898. (i) Castellano, R. K.; Gramlich, V.; Diederich, F. Chem.-Eur. J. 2002,
8, 118-129.
(3) Lancelot, G. J. Am. Chem. Soc. 1977, 99, 7037-7042.
(4) Pistolis, G.; Paleos, C. M.; Malliaris, A. J. Phys. Chem. 1995, 99, 8896-
8902.
(5) Rao, P.; Ghosh, S.; Maitra, U. J. Phys. Chem. B 1999, 103, 4528-4533.
(6) (a) Weisz, K.; Ja¨hnchen, J.; Limbach, H.-H. J. Am. Chem. Soc. 1997, 119,
6436-6437. (b) Dunger, A.; Limbach, H.-H.; Weisz, K. Chem.-Eur. J.
1998, 4, 621-628. (c) Dunger, A.; Limbach, H.-H.; Weisz, K. J. Am. Chem.
Soc. 2000, 122, 10109-10114. (d) Bas´ılio Janke, E. M.; Dunger, A.;
Limbach, H.-H.; Weisz, K. Magn. Reson. Chem. 2001, 39, 177-182.
9
10.1021/ja038630z CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 2135-2141
2135

A R T I C L E S
Bas´ılio Janke et al.
Concentration-dependent chemical shifts were fitted with an appropriate
equation by employing the Marquardt-Levenberg algorithm. Ab initio
calculations were performed using SPARTAN, Version 4.1.1.
0.3) as eluent. The appropriate fractions were combined, and the solvent
was removed to provide 0.91 g (1.166 mmol, 65.3%) of pure product.
1H NMR (250 MHz, CDCl3): δ (ppm) -0.043 (s, 3H, CH3Si), -0.024
(s, 3H, CH3Si), 0.14 (s, 3H, CH3Si), 0.13 (s, 3H, CH3Si), 0.12 (s, 3H,
CH3Si), 0.11 (s, 3H, CH3Si), 0.78 (s, 9H, (CH3)3C), 0.94 (d, 9H
(CH3)3C), 0.95 (d, 9H, (CH3)3C), 3.72 (dd 1H, H5′), 3.91 (dd, 1H, H5′′),
4.08 (q, 1H, H4′), 4.25 (t, 1H, H3′), 4.48 (t, 1H, H2′), 5.29 (d, 2H,
CH2Ph), 5.85 (d, 1H, H1′), 7.30 (m, 5H, Ph), 7.69 (s, 1H, H8), 7.90 (s,
1H, H2).
Materials. Reagents of the highest quality available were purchased
from Sigma-Aldrich, Deisenhofen, Germany. 15NH4Cl was purchased
either from Chemotrade, Leipzig (95% of label), or from Deutero
GmbH, Kastellaun (99% of label). All reactions were controlled by
TLC on silica gel plates (Merck silica gel 60 F254). If necessary, solvents
were dried by standard procedures prior to use. The deuterated Freon
mixture CDClF2/CDF3 was prepared as described7 and handled on a
vacuum line which was also used for the sample preparation. 3′,5′-
Diacetyl-3-15N-2′-deoxythymidine was prepared from unlabeled 2′-
deoxythymidine in analogy to procedures published previously.6b,8 3′,5′-
Di-O-(triisopropylsilyl)-7-15N-2′-deoxyadenosine was synthesized
according to literature procedures9,10 as described.6c
(d) 2′,3′,5′-Tri-O-(tert-butyldimethylsilyl)-6-N-benzyl-1-15N-ad-
enosine. Addition of 63 mL of methanolic dimethylamine (1:1) to 0.96
g (1.230 mmol) of 2′,3′,5′-tri-O-(tert-butyldimethylsilyl)-1-benzyl-6-
15N-adenosine resulted in a two-phase system, which was vigorously
stirred at room temperature for 65 h. The reaction was followed by
TLC on silica gel plates. After completion, the solvent was evaporated,
and the dry residue was purified by chromatography on silica gel using
dichloromethane-methanol (10:0.2) as eluent. The appropriate fractions
were combined to provide after evaporation of the solvent 0.81 g (1.156
mmol, 94%) of pure 2′,3′,5′-tri-O-(tert-butyldimethylsilyl)-6-N-benzyl-
The synthesis of 2′,3′,5′-tri-O-(tert-butyldimethylsilyl)-1-15N-ad-
enosine was performed with some modifications based on a published
procedure.11
(a) 6-15N-Adenosine. After the addition of 0.38 g (6.97 mmol) of
15NH4Cl and 1.05 g (10.48 mmol) of KHCO3 to 1 g (3.49 mmol) of
6-chloroadenosine dissolved in 5 mL of dimethyl sulfoxide, the reaction
mixture was left for 3 days at 80 °C. It was subsequently diluted with
8.8 mL of distilled water, and the pH was adjusted to 7 using
concentrated acetic acid. The solution was evaporated to dryness, and
the product was isolated via column chromatography (SiO2, ethyl
acetate-methanol 1:1). The purification afforded 750 mg (2.79 mmol,
80%) of 6-15N-adenosine. 1H NMR (500 MHz, DMSO): δ (ppm) 3.54
(d, 1H, H5′), 3.65 (d, 1H, H5′′), 3.98 (m, 1H, H4′), 4.15 (m, 2H, OH,
H3′), 4.59 (m, 2H, OH, H2′), 5.88 (d, 1H, H1′), 7.39 (d, 2H, 15NH2),
8.13 (s, 1H, H8), 8.37 (s, 1H, H2).
1
1-15N-adenosine. H NMR (250 MHz, CDCl3): δ (ppm) -0.043 (s,
3H, CH3Si), -0.024 (s, 3H, CH3Si), 0.14 (s, 3H, CH3Si), 0.13 (s, 3H,
CH3Si), 0.12 (s, 3H, CH3Si), 0.11 (s, 3H, CH3Si), 0.78 (s, 9H, (CH3)3C),
0.94 (d, 9H (CH3)3C), 0.95 (d, 9H, (CH3)3C), 3.75 (dd 1H, H5′), 4.01
(dd, 1H, H5′′), 4.11 (q, 1H, H4′), 4.32 (t, 1H, H3′), 4.70 (t, 1H, H2′),
4.87 (d, 2H, CH2Ph), 5.99 (d, 1H, H1′), 6.10 (t, br, NH), 7.32 (m, 5H,
phenyl), 8.05 (s, 1H, H8), 8.37 (d, 1H, H2).
(e) 2′,3′,5′-Tri-O-(tert-butyldimethylsilyl)-1-15N-adenosine. To 0.81
g (1.156 mmol) of 2′,3′,5′-tri-O-(tert-butyldimethylsilyl)-6-N-benzyl-
1-15N-adenosine dissolved in a mixture of 8 mL of dichloromethane, 8
mL of acetonitrile, and 12 mL of distilled water were added 1 g (4.74
mmol) of NaIO4 and 6.77 mg (0.05 mmol) of RuO2‚xH2O. The dark
colored solution was stirred for 23 h at room temperature and filtered.
The filtrate was treated with 50 mL of 6% aqueous sodium hydrogen
carbonate and extracted with three 100 mL portions of dichloromethane.
The organic extracts were combined, concentrated, and treated with
11.6 mL of methanol saturated with ammonia for 37 h. Subsequently,
the solvent was removed and the dry residue was purified by
chromatography on silica gel using dichloromethane-methanol (10:
0.3). The appropriate fractions were pooled and evaporated to dryness
to provide 0.545 g (0.894 mmol, 77%) of crystalline 2′,3′,5′-tri-O-(tert-
butyldimethylsilyl)-1-15N-adenosine. M(C28H55N415NO4Si3) ) 610 g/mol.
1H NMR (250 MHz, CD2Cl2): δ (ppm) -0.043 (s, 3H, CH3Si), -0.024
(s, 3H, CH3Si), 0.14 (s, 3H, CH3Si), 0.13 (s, 3H, CH3Si), 0.12 (s, 3H,
CH3Si), 0.11 (s, 3H, CH3Si), 0.78 (s, 9H, (CH3)3C), 0.94 (d, 9H,
(CH3)3C), 0.95 (d, 9H, (CH3)3C), 3.79 (dd 1H, H5′), 4.01 (dd, 1H, H5′′),
4.11 (q, 1H, H4′), 4.33 (t, 1H, H2′), 4.67 (t, 1H, H2′), 5.73 (br, NH2),
5.99 (d, 1H, H1′), 8.14 (s, 1H, H8), 8.28 (d, 1H, H2).
(b) 2′,3′,5′-Tri-O-(tert-butyldimethylsilyl)-6-15N-adenosine. To a
solution of 0.75 g of 6-15N-adenosine (2.8 mmol) and 1.68 g of
imidazole (24.64 mmol) in 2.8 mL of dry dimethylformamide (DMF)
was added 1.86 g (12.32 mmol) of tert-butyldimethylsilylchlorosilane.
The solution was stirred at room temperature for 20 h under an Ar
atmosphere while being followed by TLC on silica gel plates. After
removal of the solvent under reduced pressure, the syrup-like residue
was dissolved in 50 mL of dichloromethane and mixed with 75 mL of
6% aqueous sodium hydrogen carbonate. The aqueous phase was
extracted with 230-mL portions of CH2Cl2. The combined organic phase
was dried over sodium hydrogen carbonate, filtered, and concentrated
by rotary evaporation to yield a gummy product. This product was
purified by column chromatography (silica gel) using dichloromethane-
methanol (10:0.2) as eluent to provide 1.11 g (1.82 mmol, 65%) of
1
crystalline 2′,3′,5′-tri-O-(tert-butyldimethylsilyl)-6-15N-adenosine. H
NMR (250 MHz, CD2Cl2): δ (ppm) -0.043 (s, 3H, CH3Si), -0.024
(s, 3H, CH3Si), 0.14 (s, 3H, CH3Si), 0.13 (s, 3H, CH3Si), 0.12 (s, 3H,
CH3Si), 0.11 (s, 3H, CH3Si), 0.78 (s, 9H, (CH3)3C), 0.94 (d, 9H
(CH3)3C), 0.95 (d, 9H, (CH3)3C), 3.79 (dd 1H, H5′), 4.01 (dd, 1H, H5′′),
4.11 (q, 1H, H4′), 4.33 (t, 1H, H3′), 4.67 (t, 1H, H2′), 5.69 (d, 15NH2),
5.99 (d, 1H, H1′), 8.15 (s, 1H, H8), 8.3 (s, 1H, H2).
(c) 2′,3′,5′-Tri-O-(tert-butyldimethylsilyl)-1-N-benzyl-6-15N-ad-
enosine. To 1.09 g (1.786 mmol) of 2′,3′,5′-tri-O-(tert-butyldimethyl-
silyl)-6-15N-adenosine dissolved in 11 mL of dry DMF was added 0.42
mL (3.572 mmol) of freshly distilled benzylbromide. The reaction
mixture was stirred at 40 °C for 2 days under an Ar atmosphere. The
solvent was evaporated in vacuo, and the residue was purified by
chromatography on silica gel using dichloromethane-methanol (10:
Results
Low-Temperature Measurements on Adenosine-Acetic
Acid Association. One-dimensional 1H NMR spectra showing
the carboxylic acid OH and adenosine base proton resonances
for a mixture of 3′,5′-di-O-(triisopropylsilyl)-7-15N-2′-deoxy-
adenosine (7A) and acetic acid (HAc) in a Freon solvent are
plotted as a function of temperature in Figure 1. Upon cooling,
the two amino proton signals (broadened at 273 K at around 7
ppm) shift downfield as a result of their increased participation
in hydrogen-bond formation. A significant downfield shift is
also observed for the adenine H8 proton, whereas adenine H2
is slightly upfield shifted at decreasing temperatures. H8 and
H2 protons are easily assigned at higher temperatures because
H8 is split into a doublet due to its scalar coupling 2JNH ) 11.2
Hz with the labeled endocyclic nitrogen-7 of 7A. No resonances
of carboxylic acid OH protons are observed above 193 K, but
(7) Golubev, N. S.; Smirnov, S. N.; Gindin, V. A.; Denisov, G. S.; Benedict,
H.; Limbach, H.-H. J. Am. Chem. Soc. 1994, 116, 12055-12056.
(8) Ariza, X.; Bou, V.; Vilarrasa, J. J. Am. Chem. Soc. 1995, 117, 3665-
3673.
(9) Gaffney, B. L.; Kung, P.-P.; Jones, R. A. J. Am. Chem. Soc. 1990, 112,
6748-6749.
(10) Ogilvie, K. K.; Thompson, E. A.; Quilliam, M. A.; Westmore, J. B.
Tetrahedron Lett. 1974, 33, 2865-2868.
(11) Gao, X.; Jones, R. A. J. Am. Chem. Soc. 1987, 109, 1275-1278.
9
2136 J. AM. CHEM. SOC. VOL. 126, NO. 7, 2004

Binding of an Acetic Acid Ligand to Adenosine
A R T I C L E S
Figure 1. Temperature-dependent 1H NMR spectra for a mixture of 3′,5′-
di-O-(triisopropylsilyl)-7-15N-2′-deoxyadenosine and acetic acid in Freon
showing the OH (left) and base proton (right) spectral region.
they appear at lower temperatures. At 133 K, three carboxylic
acid OH protons in slow exchange, Ha, Hb, and Hc, resonate at
chemical shifts between 16.3 and 13 ppm. Due to their different
solubility in Freon at very low temperatures, relative concentra-
tions of A and acetic acid were reexamined by signal integration,
yielding an adenosine:acetic acid molar ratio of 1:4 at T e 133
K.
Figure 2. (a) Geometries of acetic acid dimer 1 and A‚HAc2 complex 2
with arrows indicating NOE contacts observed in a Freon solution at 128
K. (b) Portion of a 2D NOE spectrum of 3′,5′-di-O-(triisopropylsilyl)-7-
15N-2′-deoxyadenosine and acetic acid (1:4) in Freon showing cross-peaks
between OH and adenine base protons. The connectivities are indicated by
solid lines. The spectrum was acquired at 128 K with an 80 ms mixing
time.
1H-1H NOE experiments under slow exchange conditions
at 128 K provide a wealth of information about nucleoside
conformation and A-HAc complex geometries (Figure 2). Thus,
an intramolecular cross-peak observed between adenine H8 and
the H2′ sugar proton is stronger when compared to the cross-
peak between H8 and H3′ (spectral region not shown).
Moreover, a cross-peak observed for H8 but missing for H2
with the 5′-triisopropylsilyl (TIPS) protecting group clearly
establishes an anti orientation of the base relative to the sugar
moiety. Consequently, the TIPS-protected 2′-deoxyadenosine
adopts a C2′-endo sugar pucker with an anti glycosidic torsion
angle.12
Likewise, carboxylic acid OH protons are easily assigned to
different binding modes through 1H-1H NOE contacts. As
shown in Figure 2, the two OH protons at 14.90 ppm (Hb) and
16.30 ppm (Ha) having equal intensity show NOE contacts to
adenosine H8 and H2 protons, respectively. Clearly, Hb and Ha
are associated with Hoogsteen and Watson-Crick bound acetic
acid molecules. The Hoogsteen Hb signal also shows NOE
contacts to the adenosine 5′-TIPS protecting group, further
substantiating its binding at the Hoogsteen site of A with anti
conformation (not shown). In addition, cross-peaks of Ha and
Hb to the same two amino protons with reversed relative
intensity indicate their binding to the same adenine base in a
7
single A‚HAc2 complex 2 (Figure 2). No additional sets of
resonances are seen for adenine H2, H8, and amino protons,
confirming the presence of adenosine in a single 1:2 complex.
The remaining Hc proton resonance at 13.04 ppm exhibits no
NOE contact to adenine base protons and can be unambiguously
assigned to an acetic acid dimer 1 by comparison with a low-
temperature measurement on a pure acetic acid solution.
We could not detect any resolved scalar coupling of the
Hoogsteen bound Hb resonance to the labeled adenine N-7,
indicating localization of the proton closer to the oxygen of
acetic acid in a neutral complex. To explore the possibility of
trans-hydrogen scalar coupling in the case of the Watson-Crick
bound proton and N1 of adenosine, 2′,3′,5′-tri-O-(tert-butyldi-
methylsilyl)-adenosine was selectively 15N-labeled at N-1 and
used for the association studies with acetic acid in Freon. tert-
(12) Wu¨thrich, K. NMR of Proteins and Nucleic Acids; Wiley-Interscience: New
York, 1986; pp 205-214.
9
J. AM. CHEM. SOC. VOL. 126, NO. 7, 2004 2137

A R T I C L E S
Bas´ılio Janke et al.
1
Figure 3. Temperature-dependent H NMR spectra showing the OH and
base proton spectral region for a mixture of 2′,3′,5′-tri-O-(tert-butyldi-
methylsilyl)-1-15N-adenosine and acetic acid (1:0.7) in Freon.
Butyldimethylsilyl (TBDMS)-protected adenosine (1A) has been
found to possess favorable solubility properties in Freon
especially at lower temperatures, while at the same time NOE
contacts between H8 and H2′ (strong) and the 5′-TBDMS
protecting group show that it adopts an anti conformation like
the TIPS-protected deoxyribonucleoside. In Figure 3, the OH
and base proton spectral regions for a mixture of 1-15N-
1
adenosine and acetic acid with a molar ratio A:HAc of >1 in
Freon are plotted as a function of temperature. Due to the 15N
label at position 1 of the adenine base, H2 protons are easily
2
assigned on the basis of their scalar coupling of JNH ) 15.6
Hz to N1. Again, a downfield shift is observed for the adenine
H8 proton, whereas adenine H2 moves upfield upon lowering
the temperature. However, in contrast to the 7A-HAc mixture,
both H2 and H8 resonances of 1A split into at least two different
signals below 173 K. In particular, two H2 doublets at 8.11
and 8.27 ppm are clearly identified in the 1D spectra (see Figure
3), while two different H8 resonances with chemical shifts of
8.53 and 8.60 ppm can be assigned by their corresponding NOE
contacts to the sugar protons in a 2D NOE spectrum at 128 K
(not shown). It has to be noted that at this temperature the
stoichiometry of adenosine and acetic acid was found to be about
1:0.7 based on signal integration.
Again, the acid OH signals are only observed at low
temperatures. At 153 K, the OH resonances are considerably
broadened and finally split into three predominant signals, Hd,
Ha, and Hb, at 133 K with chemical shifts of δ ) 17.76 ppm,
δ ) 17.10 ppm, and δ ) 15.11 ppm. Obviously, these
resonances must be attributed to different coexisting species in
slow exchange, and their assignment makes use again of 2D
NOE experiments under slow exchange conditions at 128 K.
As seen from the 2D NOE spectrum in Figure 4a, Ha and Hb
resonances are easily assigned to Watson-Crick and Hoogsteen
Figure 4. (a) Portion of a 2D NOE spectrum of 2′,3′,5′-tri-O-(tert-
butyldimethylsilyl)-1-15N-adenosine and acetic acid (1:0.7) acquired in Freon
at 128 K with an 80 ms mixing time showing base-base (top) and OH-
base cross-peaks (bottom). The connectivities are indicated by solid and
dashed lines. (b) Geometries of complexes 2 and 3 formed between A and
HAc with arrows indicating NOE contacts observed in a Freon solution at
1
128 K. Also given are H chemical shifts δ of base and acid OH protons.
supra). Again, no resolvable scalar coupling between Ha and
the labeled nitrogen-1 is observed, in agreement with localization
of the proton closer to the oxygen of acetic acid. The additional
broad and most downfield shifted Hd signal at 17.76 ppm
exhibits cross-peaks with an adenine H2 at 8.12 ppm, slightly
1
bound acetic acid in a A‚HAc2 complex 2 based on cross-
peaks to adenine H2, H8, and the same two amino protons (vide
9
2138 J. AM. CHEM. SOC. VOL. 126, NO. 7, 2004

Binding of an Acetic Acid Ligand to Adenosine
A R T I C L E S
Figure 5. Dependence of 1H NMR spectra on the adenosine:acetic acid
molar ratio in a Freon solvent at 123 K.
Figure 6. Concentration-dependent adenine amino proton chemical shift
in a 1:1 mixture of 3′,5′-di-O-(triisopropylsilyl)-2′-deoxyadenosine and acetic
acid in CDCl3 at 293 K. The line represents the least-squares fit.
downfield shifted with respect to H2 in 2 at 8.11 ppm (Figure
4a). Moreover, weak but noticeable cross-peaks are observed
between Hd and amino protons at 8.74 ppm (stronger, i.e.,
hydrogen bonded to acetic acid) and 9.72 ppm (weak). These
in turn have NOE contacts to the H2 proton at 8.27 ppm, which
has no additional contacts to OH acetic acid protons. Taken
together, this pattern of NOE connectivities suggests the
formation of a 2:1 1A2‚HAc complex 3 with acetic acid bound
at the Watson-Crick site and a second adenine base at the
Hoogsteen site.
The analysis of signal intensities not only supports the
coexistence of complexes 3 and 2 but also indicates a molar
ratio 3:2 of about 2:1: (i) with an approximate stoichiometry
of adenosine and acetic acid of 1:0.7, relative signal intensities
for Hd, Ha, and Hb are found to be 2:1:1; (ii) the relative
intensities of the H2 proton signals at 8.27 ppm and 8.11/8.12
ppm are about 2:3; and (iii) coexistence of 2 and 3 accounts
for one H8 resonance of the Hoogsteen bound adenosine in 3
at 8.53 ppm and for the two other H8 signals being isochronous
at 8.60 ppm.
Moreover, on the basis of its downfield chemical shift, the
amino proton with δ ) 9.72 ppm is expected to be involved in
a NHN rather than a NHO hydrogen bridge as inferred from
homoassociation studies of adenosine and 6-methyl-adenosine
(K. Weisz, unpublished results). It should also be mentioned
that OH protons of acetic acid show weak NOE contacts to
their own methyl group (not shown). The methyl resonance is
split into at least 2-3 components for all stoichiometries,
reflecting the different environment of the acetic acid molecule
in the complexes.
1H NMR spectra showing the OH proton region of adeno-
sine-acetic acid complexes are plotted in Figure 5 as a function
of the A:HAc molar ratio. With acetic acid present in excess,
Watson-Crick and Hoogsteen sites of adenine are completely
occupied by acid molecules, giving rise to A‚HAc2 complexes
2 and additional formation of acetic acid dimers 1 having a
more upfield shifted OH resonance (Figure 5, top). Upon a
lowering of the acid concentration, acetic acid dimers disappear
at stoichiometric ratios of A:HAc ∼0.5 and complexes 2 coexist
with a new trimeric A2‚HAc complex 3 that is increasingly
formed at ratios >0.5 and characterized by a broad downfield
shifted OH resonance at about 17.8 ppm (Figure 5, bottom).
No effort was made to assign additional OH proton signals of
minor species due to their low intensity. Note that differences
observed in OH chemical shifts for solutions of different molar
ratios and especially apparent for the Watson-Crick bound OH
signal in the top spectrum arise from Freon mixtures of slightly
different composition. Because the proton transfer in a neutral
hydrogen-bonded complex is associated with a charge separa-
tion, the hydrogen-bond geometry and hence the proton chemical
shift are very sensitive to the temperature- and composition-
dependent dieletric constant ꢀ of the Freonic solvent.13a,b
Concentration-Dependent Measurements on Adenosine-
Acetic Acid Association. Considering the higher-order com-
plexes 2 and 3 formed at low temperatures, there seems to be
no unambiguous and easy way to determine an association
constant between adenosine and acetic acid. Surprisingly,
however, a good fit of the concentration-dependent amino proton
chemical shift is achieved by employing a simple 1:1 association
model between 3′,5′-di-O-triisopropylsilyl-2′-deoxyadenosine
and acetic acid in chloroform at 293 K (Figure 6).
Due to the broadening of carboxylic acid OH resonances
under these conditions, concentration-dependent chemical shifts
of the adenine amino protons (in fast exchange) were used for
the determination of the association constant. Values obtained
from the least-squares fit amount to 296 ( 43 M-1 for the
association constant as well as to 5.56 and 7.08 ppm for the
limiting chemical shift of monomer and complex, respectively.
With an association constant of 45 M-1 determined for adenine
complexation with thymidine under the same experimental
conditions (not shown), the A‚Hac dimer is more stable as
compared to the AT base pair. The good fit with a 1:1 model
may be indicative of predominant formation of 1:1 complexes
at ambient temperatures. Nevertheless, the determined associa-
tion constant only represents an average over the various
association modes, and the limiting chemical shift at infinite
(13) (a) Golubev, N. S.; Denisov, G. S.; Smirnov, S. N.; Shchepkin, D. N.;
Limbach, H.-H. Z. Phys. Chem. 1996, 196, 73-84. (b) Shenderovich, I.
G.; Burtsev, A. P.; Denisov, G. S.; Golubev, N. S.; Limbach, H.-H. Magn.
Reson. Chem. 2001, 39, S91-S99.
9
J. AM. CHEM. SOC. VOL. 126, NO. 7, 2004 2139

A R T I C L E S
Bas´ılio Janke et al.
resonances characteristic of thymidine homodimers appear at
around 12.1 ppm and are still exchange broadened at 133 K.
The two OH resonances at 16.99 and 15.06 ppm of about equal
intensity must be attributed to Watson-Crick and Hoogsteen
bound acetic acid molecules in a ternary A‚(HAc)2 complex 2
(vide supra). Further information can be obtained from a 2D
NOE spectrum at 133 K (not shown). At this temperature, very
weak exchange cross-peaks are still observed between the two
downfield shifted thymidine imino protons and the imino protons
in the homodimers. As expected for the OH resonances in 2,
the OH signals at 16.99 and 15.06 ppm exhibit a NOE contact
to adenine H2 and H8 (identified by a cross-peak to adenosine
sugar protons), respectively. Moreover, the most downfield
shifted OH proton at 17.30 ppm is connected to adenine H2
and thus Watson-Crick bound to the adenine base.
The assignment can be completed by further considering
signal intensities. Thus, about equal intensities are found for
the OH resonance at 17.30 ppm and the TNH signal at 14.15
ppm. The same applies to the OH and NH resonances at 15.23
and 14.91 ppm, respectively. Such a situation is compatible with
the formation of two different ternary A‚T‚HAc complexes 4
and 5 (Figure 7). Obviously, the former two protons are
positioned in 4 with acetic acid bound in a Watson-Crick and
thymidine bound in a Hoogsteen fashion. The latter two protons
are associated with complex 5, where HAc and T ligands have
changed binding sites. Note that, in contrast to T homodimers,
no acetic acid dimers are observed in the spectra. By increasing
the acetic acid concentration, signal intensities of the T
containing ternary complexes 4 and 5 decrease, and A‚(HAc)2
complexes 2 clearly predominate (Figure 7, bottom).
Discussion
Strength of Hydrogen Bonds in Adenosine-Acetic Acid
Complexes. Having unambiguously assigned proton resonances
to particular complex structures in slow exchange allows the
extraction of important information regarding hydrogen-bond
contacts. Thus, the proton chemical shift can be used as a
sensitive indicator for the relative strength of hydrogen bonds.
A more deshielded proton in the O-H‚‚‚N hydrogen bridge
originates from a displacement of the proton toward the accep-
tor atom and is associated with a shortening of the hydro-
gen bond.6c,14a,b Experimentally, a stronger Watson-Crick
O-H‚‚‚N1 hydrogen bond is clearly indicated by its more
downfield shifted OH resonance. However, for all complexes
observed, the acetic acid OH proton is still localized closer to
the oxygen of acetic acid in a nonzwitterionic complex. Due to
the absence of 1:1 complexes, experimental data related to the
hydrogen-bond strength in A‚HAc associates are only available
for the simultaneous binding of two carboxylic acid ligands in
a ternary complex 2. Although low-level ab initio calculations
point to a slight weakening of hydrogen bonds upon binding of
a second carboxylic acid ligand, the relative strength of hydrogen
bonds within a 1:2 complex should closely follow the situation
in the corresponding 1:1 dimers.
Figure 7. 1H NMR spectra of a mixture of adenosine, 3-15N-2′-deoxythy-
midine, and acetic acid in Freon at 133 K showing the acetic acid OH and
thymidine imino proton spectral region with different acetic acid concentra-
tions. The proton resonances are assigned to the complexes indicated on
top.
concentration is actually an average over all coexisting species
in solution. Comparing the value of 7.08 ppm to the value of
7.98 ppm, obtained in a separate titration experiment for
adenosine homodimers, indicates that amino protons engaged
in a N-H‚‚‚O hydrogen bond with the acid are actually more
upfield shifted as compared to amino protons in a N-H‚‚‚N
hydrogen bond formed in the adenosine self-associates.
Measurements on Ternary Mixtures of Adenosine, Thy-
midine, and Acetic Acid. Because in nucleic acid structures
the adenine base is often hydrogen bonded to a thymine or uracil
base to form Watson-Crick or in some cases Hoogsteen base
pairs, we also examined the association in a ternary mixture of
free adenosine, 2′-deoxythymidine, and acetic acid. To simplify
the analysis of the spectra, we decided to employ specifically
3-15N-labeled 2′-deoxythymidine (T), which provides for a
convenient discrimination between thymine imino and acid OH
protons (vide infra). As shown in Figure 7, the 1H NMR
spectrum of a 1:1:0.8 adenosine-thymidine-acetic acid mixture
in Freon at 133 K displays several T imino and acid OH protons
at low field. The T imino resonances at 14.91 and 14.15 ppm
are easily identified by their one-bond scalar coupling to 15N,
giving rise to a doublet with 1JNH of about 90 Hz and a central
resonance due to residual 14N isotopomers (∼37%). Also, imino
Stronger acetic acid binding is observed when substituting
one acetic acid molecule in the A‚(HAc)2 complex by a second
adenine or a thymine as evidenced by a downfield shift of the
(14) (a) Smirnov, S. N.; Benedict, H.; Golubev, N. S.; Denisov, G. S.; Kreevoy,
M. M.; Schowen, R. L.; Limbach, H.-H. Can. J. Chem. 1999, 77, 943-
949. (b) Dingley, A. J.; Masse, J. E.; Peterson, R. D.; Barfield, M.; Feigon,
J.; Grzesiek, S. J. Am. Chem. Soc. 1999, 121, 6019-6027.
9
2140 J. AM. CHEM. SOC. VOL. 126, NO. 7, 2004

Binding of an Acetic Acid Ligand to Adenosine
A R T I C L E S
OH resonance. Thus, a shorter hydrogen bond between acetic
acid and A N1 or A N7 is observed when thymidine is
simultaneously base-paired to A. Clearly, this situation bears
special relevance when ligand binding to AT base pairs, for
example, in nucleic acid structures, is of major interest.
Structure of Complexes. For the association of adenosine
with acetic acid at low temperatures, trimolecular complexes
are formed. Thus, in the presence of acetic acid, adenosine forms
various amounts of 1:2 A‚HAc2 and 2:1 A2‚HAc complexes
depending on the molar ratio of nucleoside to ligand. With an
excess of acetic acid, both the Watson-Crick and the Hoogsteen
binding site of the adenine base are occupied by acid. Upon an
increase in the adenosine:acetic acid molar ratio to >0.5,
increasing amounts of A2‚HAc complexes are formed at the
expense of the 1:2 associates. It is interesting to note that even
at a molar ratio of 1:1 no binary 1:1 adenosine-acetic acid
complexes have been detected but rather 2:1 and 1:2 complexes
coexist in equal amounts. Obviously, the enthalpy gain of such
a situation (more hydrogen-bond interactions are formed)
outweighs the entropic penalty in a trimolecular complex, at
least at low temperatures.
H8 signals upon formation of the 1:2 complex 2 with Watson-
Crick and Hoogsteen bound acid. When compared to free
1
adenosine, differences in the H chemical shift for H2 and H8
in 2 at low temperatures amount to about -0.2 and +0.8 ppm,
respectively (see also Figure 1).
On the basis of the hydrogen-bond strength, Watson-Crick
binding should be preferred by acetic acid, but solvation effects
also contribute to the free enthalpy of association and hence to
the extent of specific complex formation. Whereas at low
temperatures no preferential adenine binding site for acetic acid
can be assigned due to the lack of any 1:1 complex, the acid
preferentially occupies the Watson-Crick site in the A2‚HAc
complex 3. However, a broad low-intensity peak downfield from
the Hoogsteen OH resonance also points to a small population
of Hoogsteen bound acid in such 2:1 associates (see Figure 5).
Broadening of carboxylic acid protons in the latter complexes
may be attributed to exchange processes, for example, rapid
dissociation of the second weakly bound adenine base and
possibly a concomitant exchange between its N1 and N7 atoms
as a hydrogen-bond acceptor.
In addition to Watson-Crick and Hoogsteen binding, there
is also the possibility of acetic acid binding to adenine N3
through only one hydrogen bond. This binding mode, although
only populated to a small extent, was considered in adenine-
carboxylic acid associations at ambient temperatures.3,4 How-
ever, on the basis of the present studies on adenosine-acetic
acid mixtures, such a binding can be excluded.
In contrast to our findings, adenine binding studies with
various carboxylic acids at ambient temperatures have mostly
been interpreted in terms of bimolecular associates, although
evidence for the formation of 1:2 adenine-acid complexes was
provided at higher concentrations for some acids.5,15 Clearly,
whereas lower temperatures will increasingly favor aggregation
to larger clusters, 1:1 complexes are likely to predominate at
higher temperatures especially at low concentrations. This is
also indicated here by the concentration-dependent chemical
shift data of adenine amino protons in the presence of an
equimolar amount of acetic acid in chloroform at 293 K that fit
to a simple 1:1 association model (Figure 6). The corresponding
Conclusion
The study of low-molecular weight host-guest interactions
by NMR is often hampered by the fast exchange of species
coexisting in solution. To overcome this problem, we have
performed measurements at temperatures low enough to be in
the slow exchange regime, thus allowing for the unambiguous
characterization of the structures and interactions present. The
results obtained here for the complexation of adenosine or
adenosine-thymidine base pairs with acetic acid provide new
insight into the strength and specificity of binding. Clearly, this
information is not only of relevance to protein-nucleic acid
interactions which may include acidic amino acid side chains
of aspartic and glutamic acid. It is also important for the
development of potent receptors for nucleic acid bases, which
rely on the specificity and affinity for binding their target.
association constant of Kass ) 296 M-1 is in line with a Kass
)
160 M-1 that has been previously determined for 9-ethylad-
enine-butyric acid association in chloroform at 303 K.3
In general, both Watson-Crick and Hoogsteen binding modes
have been reported in the past to be operative depending on
the particular ligand. NMR studies on adenine binding sites
either have employed NOE experiments or have followed the
chemical shift of adenine H2 and H8 protons in the presence
of the ligand. Thus, by adding butyric acid to 9-ethyladenine, a
downfield shift of the adenine H8 resonance and an upfield shift
of the H2 signal was observed by Lancelot.3 By taking
downfield and upfield shifted H8 and H2 resonances as evidence
for a Hoogsteen and Watson-Crick bound acid, respectively,
aromatic carboxylic acids have indicated a distinct preference
for the Hoogsteen site, whereas aliphatic acids were found to
prefer the Watson-Crick site for binding adenine.5 Indeed, we
observe corresponding upfield and downfield shifts of H2 and
Acknowledgment. We thank Dr. N. S. Golubev, V.A. Fock
Institute of Physics, St. Petersburg University, for introducing
us to the low-temperature NMR technique. Furthermore, we
thank the Deutsche Forschungsgemeinschaft, Bonn-Bad Godes-
berg, and the Fonds der Chemischen Industrie, Frankfurt, for
financial support.
(15) Lancelot, G. Biochimie 1977, 59, 587-596.
JA038630Z
9
J. AM. CHEM. SOC. VOL. 126, NO. 7, 2004 2141