cADPR Analogues: Effect of an Adenosine 2′- Or 3′-Methoxy Group on Conformation

Article abstract of DOI:10.1021/ol036152r

(Matrix presented) The 2′-OMe-A (2) and 3′-OMe-A (3) analogues of the calcium release agent cADPR (1) were prepared and their solution structures studied by NMR spectroscopy. Compared to 1, 2 shows a shift in its A ring conformation and changes in its R ring N:S and γ^t: γ⁺ ratios, while 3 displays a significant change in the conformation of its A ring γ-bond.

Full text of DOI:10.1021/ol036152r

ORGANIC
LETTERS
2
004
Vol. 6, No. 2
33-236
cADPR Analogues: Effect of an
Adenosine 2′- or 3′-Methoxy Group on
Conformation
2
,†
‡
‡
Steven M. Graham,* Daniel J. Macaya, Raghuvir N. Sengupta, and
†
Kevin B. Turner
Department of Chemistry, St. John’s UniVersity, Jamaica, New York 11439, and
Benjamin N. Cardozo High School, Bayside, New York 11364
grahams@stjohns.edu
Received November 3, 2003
ABSTRACT
The 2′-OMe-A (2) and 3′-OMe-A (3) analogues of the calcium release agent cADPR (1) were prepared and their solution structures studied by
t
+
NMR spectroscopy. Compared to 1, 2 shows a shift in its A ring conformation and changes in its R ring N:S and γ :γ ratios, while 3 displays
a significant change in the conformation of its A ring γ-bond.
Many small molecules are known to have roles in cell
essentially as potent as cADPR toward calcium release,
whereas 3′-dA cADPR was ∼100-fold less potent. A third
analogue, where the 3′-OH was replaced with a 3′-methoxy
group (3′-OMe-A cADPR, 3), did not cause calcium release
but instead blocked the ability of cADPR to release calcium.
It was proposed that perhaps the 3′-OH made a hydrogen
bond contact critical for recognition.
signaling. One such molecule is cyclic adenosine 5′-
+
diphosphate ribose (cADPR, 1), formed in vivo from NAD
in a reaction catalyzed by ADP-ribosyl cyclase (ADPRC).
Produced in response to extracellular signals, cADPR leads
1
to the release of calcium ions from intracellular stores.
cADPR has been found in a variety of organisms and organs,
and is involved in diverse cellular processes such as muscle
contraction, glucose metabolism, and sea urchin egg fertiliza-
tion. However, despite the obvious importance of cADPR
in cell signaling, the precise mechanism by which cADPR
causes calcium release remains poorly understood.
We had recently determined the NMR solution structure
3
of cADPR and were intrigued by an alternate explanation:
that the 2′- and 3′-modifications could be altering the
conformation of the furanose ring. To address this issue, we
prepared the methoxy analogues 2′-OMe-A cADPR (2) and
3′-OMe-A cADPR (3) to determine the effect, if any, of the
methoxy group on cADPR conformation.
The commercial availability and loose substrate specificity
of ADPRC has proved invaluable in the preparation of
cADPR analogues. The Potter group prepared cADPR
The synthesis of cADPR analogues 2 and 3 is shown in
Scheme 1. Briefly, 2′-O-Me- or 3′-O-Me adenosine (4 or 5,
3
2
analogues modified in the adenosine furanose ring. They
found the analogue without a 2′-OH (2′-dA cADPR) was
respectively) were selectively 5′-phosphorylated (POCl /PO-
†
St. John’s University.
Benjamin N. Cardozo High School.
(2) Ashamu, G. A.; Sethi, J. K.; Galione, A.; Potter, B. V. L. Biochemistry
1997, 36, 9509-9517.
(3) Graham, S. M.; Pope, S. C. Nucleosides, Nucleotides, Nucleic Acids
2001, 20, 169-183.
‡
(
1) (a) Lee, H.-C. Annu. ReV. Pharmacol. Toxicol. 2001, 41, 317-345.
(
b) Lee, H.-C. Physiol. ReV. 1997, 77, 1133-1164.
1
0.1021/ol036152r CCC: $27.50 © 2004 American Chemical Society
Published on Web 12/18/2003

Scheme 1. Synthesis of cADPR Analoguesa
a
Reagents and conditions: (i) POCl
iii) Oct N, MeOH; (iv) DPCP, dioxane, DMF, Bu
DMF, pyridine, Bu N; (vi) NH , MeOH; (vii) ADPRC.
3
, TEP; (ii) H
2
O (hydrolysis);
(
3
3
N; (v) Ac
2
NMN,
Figure 1. The defining atoms for the â- (trans conformation shown)
and γ-bonds are C4′-C5′-O5′-P and C3′-C4′-C5′-O5′, respectively.
3
3
1
0
),4 isolated as the free acids, and converted to their
OEt)
3
trioctylammonium salts (6, 7). Activation of 6 or 7 toward
rotamers. The question as to whether the remaining fraction
(
+
-
t
-
is â or â and γ or γ is more complicated. These rotamers
have P (or H4′) gauche to one H but anti to the other H
leading to one small and one large coupling. Identifying
5
5
,
nucleophilic attack by the 5-phosphate of 2′,3′-di-O-acetyl
5
nicotinamide mononucleotide (Ac
2
NMN) was performed by
t
+
rotamers other than â and γ thus requires the stereospecific
assignments of H5′/H5′′. The furanose ring conformations
were determined with the program PSEUROT, whose basic
principles can be understood by examining the furanose
H-H torsion angles in Figure 1. In the S conformer, H1′
and H2′ are approximately antiperiplanar, resulting in a large
using the Michelson procedure (PO(OPh)
2
Cl/Bu
3
N in diox-
ane/DMF).5 Removal of the acetyl groups (MeOH/NH
,6
3
)
11
followed by ion-exchange chromatography afforded the
+
corresponding NAD analogues 8 or 9 in ∼15% overall
+
+
yield. The NAD analogues and NAD itself were cyclized
using ADPRC, purified, and isolated as their sodium salts
7
J
1′2′, whereas the H3′-H4′ torsion angle is ∼90°, yielding a
to give 1-3 in ∼65% yield.
small J3′4′. The situation is reversed in the N conformer and
one observes a small J1′2′ and a large J3′4′. The PSEUROT
program calculates J values for various N and S conformers
(each characterized by P and Φ) and minimizes the differ-
ence between these calculated J values and the observed J
We then focused on determining the NMR solution
structures of 1-3. Furanose rings generally exist in a two-
state equilibrium; the geometry of each ring is described by
8
a phase angle, P, and a puckering amplitude, Φ. Rings with
P ) 0 ( 90° are described as “north” (N, or T
3
2
for P ) 0);
values by adjusting P
sum rules and the principles of pseudorotation, assuming the
needed J Values can be extracted from H NMR data, provide
good estimates of the ring, â-, and γ-bond conformations
and populations.
H NMR spectra for 1-3 are shown in Figure 2 and the
J values are summarized in Table 1. Due to overlaps, not
N
, P
S
, Φ
N S
, Φ , and the N:S ratio. The
those with P ) 180 ( 90° are described as “south” (S, or
2
9
T for P ) 180) (Figure 1). Also shown in Figure 1 are
3
1
the defining conformations of the â- and γ-bonds.
The furanose ring, â-bond, and γ-bond conformations and
1
their relative populations were derived from H NMR data.
1
The â- and γ-bond conformations are reflected in the
1
2
t
coupling of H5′/H5′′ to P and to H4′, respectively. In the â
and γ rotamers, P (or H4′) is gauche to both H5′ and H5′′
all the required J values could be extracted directly from
the normal 1D H spectrum, and we used 1D TOCSY and
+
,
1
13
resulting in small, approximately equal coupling of P (or
4′) to H5′ and to H5′′. Simple “sum rules” are available for
1
31
1
phosphorus-decoupled H ({ P} H) NMR experiments to
extract J values from these crowded regions.
The 1D TOCSY sequence allows magnetization transfer
from a selectively excited proton to potentially all of the
other protons in the spin system, not just adjacent ones. For
H
t
+
calculating the fractional population of the â and γ
(
4) Yoshikawa, M.; Kato, T.; Takenishi, T. Bull. Chem. Soc. Jpn. 1969,
2, 3505-3508.
5) Bailey, V. C.; Sethi, J. K.; Fortt, S. M.; Galione, A.; Potter, B. V. L.
4
(
Chem. Biol. 1997, 4, 51-61.
(
10) (a) γ-bond: fγ+ ) 13.3 - (J4′5′ + J4′5′′)/9.7. Altona, C. Recl. TraV.
(
6) Michelson, A. M. Biochim. Biophys. Acta. 1964, 91, 1-13.
Chim. Pays-Bas 1982, 101, 413-433. (b) â-bond: fât ) 25.5 - (J5′P +
7) Compounds 1-3 gave satisfactory spectral data (UV and H and 31P
1
(
J5′′P)/20.5. Lankhorst, P. P.; Haasnoot, C. A. G.; Erkelens, C.; Altona, C. J.
Biomol. Struct. Dyn. 1984, 1, 1387-1405.
NMR). See the Supporting Information.
(8) The phase angle P is a description of which atoms are above (endo)
(11) van Wijk, J.; Haasnoot, C. A. G.; de Leeuw, F. A. A. M.; Huckriede,
and below (exo) the furanose ring plane; Φ is a measure of the furanose
ring pucker. See: Altona, C.; Sundaralingam, M. J. Am. Chem. Soc. 1973,
B. D.; Westra Hoekzema, A. J. A.; Altona, C. PSEUROT 6.3; Leiden
Institute of Chemistry, Leiden University: Leiden, The Netherlands, 1999.
9
5, 2333-2344.
9) According to the IUPAC/IUB JCBN guidelines, the E/T symbolism
(12) The accuracy of the δ and J values was evaluated through spectral
(
simulation. The rms errors in the simulated versus experimental spectra
were 0.09, 0.19, and 0.21 Hz for 1, 2, and 3, respectively. See the Supporting
Information for simulated spectra and a table of chemical shifts.
is preferable to the C2′-exo-C3′-endo/C2′-endo-C3′-exo descriptions. In the
E/T symbolism, E denotes an envelope and T a twist conformation.
Subscripts and superscripts denote that the indicated atom has the exo or
endo configuration, respectively. See: Eur. J. Biochem. 1983 131, 9-15.
(
13) Kessler, H.; Oschkinat, H.; Griesinger, C.; Bermel, W. J. Magn.
Reson. 1986, 70, 106-133.
234
Org. Lett., Vol. 6, No. 2, 2004

1
31
1
Figure 2. 1D H NMR spectra of (a) 1 (12 °C), (b) 2 (6 °C), and (c) 3 (6 °C) in D
2
O. Inset b1: { P} H NMR spectrum of 2 (20 °C). Insets
c1 and c2: 1D TOCSY NMR spectra (6 °C) of 3 with excitation at R5′′ and at A1′, respectively.
example, 1D TOCSY with selective excitation of A1′ in 3
Figure 2) revealed not only the adjacent A2′ but also that
3′ was a doublet of doublets (dd). Similarly, 1D TOCSY
at the upfield R signal (R5u) clearly showed that R4′ was an
overlapping doublet of quartets. In the normal 1D
spectrum, R4′ and R2′ were overlapped and A3′ was obscured
that was observed. In cADPR itself, only R3′ was sufficiently
(
A
isolated to selectively irradiate, and it too showed an NOE
only to R5u (R5′′) and not to R5d (R5′). All additional 1D
NOESY experiments carried out on 1 and 3 were consistent
5
1
+
t
H
with an R ring γ and A ring γ conformations as the major
equilibrium rotamers; the similarity of the H5′/H5′′ chemical
shifts and J values also supports the stereospecific assign-
ments.
31
1
by three other signals. The { P} H experiment was used to
distinguish J4′5′/J4′5′′ from J5′P/J5′′P and to identify any four-
4
17
bond couplings between H4′ and its 5′-P ( J4′P); couplings
The PSEUROT furanose geometries, N:S ratios, and the
lost upon P irradiation were due to 31P.
3
1
sum rule â-bond and γ-bond rotamer populations are
The last issue was the stereospecific assignment of H5′
and H5′′. The standard method14 is to look for NOEs from
summarized in Table 1. The methoxy group has no effect
on the â-bond; the coupling of P to H5′ and H5′′ in all six
31
H
3′ to H5′ and H5′′, provided H3′ is sufficiently isolated to
rings was always small (2.2-4.0 Hz), and all â-bonds are
t
t
+
allow selective excitation. Fortunately, 2′-OMe analogue 2
had both A3′ and R3′ relatively well separated from other
signals. Using 1D NOESY, we found that selective
thus almost exclusively â . The R ring â -γ conformation
leads to a planar “W” arrangement of R4′ and its 5′-P and
15
4
10a
4
predicts a significant JR4′P
;
1-3 showed JR4′P values of
4
irradiation of A3′ in 2 led to roughly equal enhancements
3.5-4.1 Hz. In contrast to the R rings, no JA4′P couplings
were observed in 1-3, as expected for A ring γ conformers.
t
(1.8% and 2.0%, respectively) of both the downfield A
5
(A5d
,
therefore A5′′) and the A5u (therefore A5′) signals, suggesting
How does the methoxy group drive the differences in
t
16
18,19
an A ring γ conformation. The observation of a large JA4′5′′
and a small JA4′5′ (Table 1) is consistent with this assignment.
The R rings in 1-3 all showed J4′5′ ∼ J4′5′′, consistent only
conformation and population? It is well-established
that
highly electronegative (EN) groups at C2′ and C3′ prefer
pseudoaxial orientations, as this places the EN group gauche
+
with highly populated R ring γ conformers. In this case,
(17) In the pseudorotational treatment, the two-state conformational
irradiation of R3′ should enhance only R5′′. Interestingly, the
downfield/upfield assignments are reversed for the R ring
equilibrium is described by the following five variables: PN, ΦN, PS, ΦS,
and a mole fraction (of either conformer). Ribofuranoses cannot be
completely determined, as there are only three observable values related to
the endocyclic torsion angles (J1′2′, J2′3′, and J3′4′) and therefore assumptions
concerning two of the variables must be made. It has been noted (de Leeuw,
F. A. A. M.; Altona, C. J. Chem. Soc., Perkin Trans. 2 1982, 375-384)
that there is a correlation of low Φ values to high J2′3′ values. In all six
rings in 1-3 we observed relatively high J2′3′ values (5.05 ( 0.15 Hz), and
have thus constrained ΦN and ΦS to the relatively low value of 35°.
Constraining ΦN and ΦS to 33° and 37° gave similar results. The number
of observables can be increased if the J values change significantly with
temperature, but in the limited temperature ranges studied so far cADPR
and its analogues show very little variation with temperature. An extensive
variable-temperature study and full pseudorotation analysis will be presented
in due course.
H5′/H5′′, as in 2 it was an NOE from R3′ to only R5u ()R5′′)
(14) Wijmenga, S. S.; Mooren, M. M. W.; Hilbers, C. W. NMR of
Macromolecules: A Practical Approach; Roberts, G. C. K., Ed.; Oxford
University Press: Oxford, UK, 1993; pp 217-288.
(15) Stott, K.; Stonehouse, J.; Keeler, J.; Hwang, T. L.; Shaka, A. J. J.
Am. Chem. Soc. 1995, 117, 4199-4120. This is the “selnogp.3” pulse
program in Bruker’s XWinNMR 2.6.
(
16) See the Supporting Information. Further confirmation of the A ring
t
the γ conformation was obtained by using 1D NOESY with irradiation at
A2′ (A5d enhanced but not A5u), A5d (A2′ and A3′ enhanced but not A4′), A4′
(A5u enhanced but not A5d), and A5u (A3′ and A4′ enhanced but not A2′).
Org. Lett., Vol. 6, No. 2, 2004
235

Table 1. Coupling Constants (J, in Hz) and Structural Parameters for 1-3a
a
1
All ring J values are (0.1 Hz; all backbone J values are (0.2 Hz. All δ and J values were confirmed by spectral simulation. Values are from 1D H
b
31
1
31
1
c
spectrum except where noted. D2O, 12 °C. JA4′P ) 0.4 Hz used in simulation. JA4′A5′ from { P} H. JA5′5′′ from the 1D spectrum and { P} H. D2O, 6 °C.
d
1
1
31
1
e
D2O, 6 °C. Only JA1′A2′, JR1′R2′, JR2′3′, JR5′P, and JR5′R5′′ are from 1D H spectrum; all others are averaged from 1D H and 1D TOCSY and { P} H. ΦN
and ΦS restrained to 35°. f The precise γ :γ ratio could not be determined; see text.
t
+
ring, instead causes the γ-bond to deviate by ∼30° from an
to the ring oxygen (“gauche effect”). Methoxy is slightly
t
more EN than -OH,19a which predicts, relative to 1, increases
ideal γ angle of ∼180° to a nonideal angle of ∼210°.
t
Though still within the γ range, a bond angle of ∼210° is
in the percent N for 2 and in the percent S for 3. While
precisely these trends were observed in two free nucleo-
t
outside the limited ranges (γ ) 180 ( 8°) over which the
1
0a,b
1
9a,b
γ-bond sum rule
is valid. In the case of 3 this sum rule
sides,
substituent effect of the methoxy group. Compared to 1, the
′-OMe in 2 had virtually no effect on the A ring N:S ratio
2 and 3 to some extent apparently resist the
+
greatly overestimates the amount of γ . That the A ring
γ-bond has a highly populated γ conformation is apparent
t
2
2
from the relatively large JA4′A5′′ (6.4 Hz) observed, demon-
strating that A and A remain largely anti. Our working
4′ 5′′
but rather drove the A ring S conformer from
N conformer from approximately E to E, and the γ
3
E to E, the
t
4
2
hypothesis is that 3 is a cADPR antagonist not because it
has altered A ring or R ring conformations but instead has
an altered backbone conformation. The observed changes
in conformation and population are difficult to rationalize
using the gauche effect. Our results with 2 show that a
conformational change in one furanose ring can drive a
change in the other ring. Further efforts to elucidate the
structure-activity relationships in cADPR, including assess-
ing the activity of 2, full pseudorotation analysis, modeling,
and synthesis of new analogues will be presented in due
course.
population from 59% to 66%. These A ring changes
presumably drive the changes seen in the R ring of 2, where
t
the N and S conformers and the γ population were
unchanged but the percent S dropped from 62% to 51%.
Clearly there are subtle differences between 1 and 2, and it
will be interesting to see if 2 is active. Comparing 3 to 1,
we found them remarkably similar in all aspects of confor-
mation and population except for the A ring γ-bond
conformation. The A5′ signal, a doublet of triplets pattern in
31
1
1
and 2, has collapsed to a dd pattern. The { P} H NMR
experiment unambiguously demonstrated that the lost cou-
pling in 3 was JA4′A5′, suggesting that the torsion angle
between A4′ and A5′, normally ∼60°, had approached ∼90°,
Acknowledgment. This work was funded in part by NIH
grant GM 62150 to S.M.G..
20
reducing JA4′A5′ to ∼0 Hz. It seems that the 3′-OMe group,
rather than altering the conformation or N:S ratio of the A
Supporting Information Available: Experimental details
1
31
31
1
for the synthesis of 1-3; H, P, { P} H, COSY, selected
1D NOESY, and 1D TOCSY NMR spectra. This material
is available free of charge via the Internet at http://pubs.acs.org.
(18) Thibaudeau, C.; Chattopadhyaya, J. Stereoelectronic Effects in
Nucleosides and Nucleotides and their Structural Implications; Uppsala
University Press: Uppsala, Sweden, 1999; pp 19-87.
(19) (a) Thymidine, 64% S; 3′-OMe thymidine, 74% S. Thibaudeau, C.;
OL036152R
Garg, N.; Papchikhin, A.; Chattopadhyaya, J. J. Am. Chem. Soc. 1994, 116,
038-4043. (b) Adenosine, 36% N; 2′-OMe adenosine, 42% N. Uesugi,
4
S.; Miki, H.; Ikehara, M.; Iwahashi, H. Tetrahedron Lett. 1979, 20, 4073-
(20) More accurately, JA4′A5′ has been reduced to the point where it
broadens but does not split the A5′ signal. In the simulation of 3, a value
of 0.4 Hz for JA4′A5′ seemed to broaden the A5′ appropriately.
4
1
076. (c) Guschlbauer, W.; Jankowski, K. Nucleic Acids Res. 1980, 8,
421-1433.
236
Org. Lett., Vol. 6, No. 2, 2004