Syntheses of heterocyclic ethenyl C-nucleosides for recognition of inverted base pairs within the DNA triple helix by stereoselective intramolecular cyclization and olefin metathesis

Article abstract of DOI:10.1021/jo801910u

Canonical recognition of gene sequences would allow precise means for specific genetic intervention. Unfortunately, specific recognition of two of the four possible base pairs by triplex-forming oligonucleotide (TFO) as X ? T-A and Y ? C-G within a triple

Full text of DOI:10.1021/jo801910u

Syntheses of Heterocyclic Ethenyl C-Nucleosides
for Recognition of Inverted Base Pairs within the
DNA Triple Helix by Stereoselective
Intramolecular Cyclization and Olefin Metathesis
Jeffrey H. Rothman
DiVision of Experimental Therapeutics, Department of
Medicine, Columbia UniVersity Medical Center, 630 West
168th Street, New York, New York 10032
FIGURE 1. DNA triplex; TFO strand parallel or antiparallel.
jhr21@columbia.edu
ReceiVed September 11, 2008
FIGURE 2. T·A-T triplet and 5-MeC+ ·G-C triplet; Hoogsteen nucleo-
side is in parallel 3′-5′ orientation to the purine strand.
Canonical recognition of gene sequences would allow precise
means for specific genetic intervention. Unfortunately,
specific recognition of two of the four possible base pairs
by triplex-forming oligonucleotide (TFO) as X· T-A and
Y · C-G within a triplex currently remains elusive. A series
of C1-ethenyl nucleosides have been devised and evaluated
extensively for stability and specificity by molecular dynam-
ics simulation. A synthesis via olefin metathesis of the
unprotected heterocycle and a conveniently prepared ano-
merically pure C1-vinyl 2-deoxyribofuranose is presented
as a significant improvement over a previously reported
strategy.
FIGURE 3. X·T-A triplet.
groove (Hoogsteen bonding) as T ·A-T and 5-MeC+ ·G-C
triplets, respectively (Figure 2). The 5-methyl derivatization of
cytosine increases the pKa of the H-bonding iminium ion closer
to physiological pH, thus significantly improving triplet stabil-
ity.1 The current inability to form stable triplexes by recognizing
inverted base pairs, T-A and C-G, directly by TFO is a major
deterrent in the development of antigene methodologies. The
X4 Hoogsteen nucleoside has a sufficient advantage over X32
as its heterocyclic π-electron density is closer in dimension to
that of the natural nucleobases.
Proposed here is a more general strategy for ethenyl
heterocyclic C-nucleosides exemplified by the synthesis of
prototype 2′-deoxy-C-ethenyl ꢀ-D-ribonucleoside analogue, X4
(Figure 3). The design of these 2′-deoxy-C-ethenyl ꢀ-D-
ribonucleosides is based upon extensive molecular dynamics
evaluations3-5 which display significantly favorable and specific
The growing interest in gene silencing has made DNA a
suitable drug target, and the ability for their direct suppression,
deletion, or modification will lead toward more direct strategies
for therapy. An essential step toward creating synthetic bioor-
ganic devices that operate upon gene sequences is to have the
means for their sequence-specific recognition.
The ability of TFOs to bind duplex target sequences to form
intermolecular DNA triple helices offers the possibility of
designing compounds with extensive sequence recognition
properties that would be useful as antigene agents or tools in
molecular biology. One remaining major limitation of this
approach is that with natural nucleosides these triplex structures
are generally restricted to homopurine-homopyrimidine target
sites (Figure 1). TFO strands in parallel orientation H-bonds
the polypurine duplex strand of A-T and G-C in the major
(1) Povsic, T. J.; Dervan, P. B. J. Am. Chem. Soc. 1989, 111, 3059–3061.
(2) Rothman, J. H. J. Org. Chem. 2007, 72, 3945–48.
(3) Rothman, J. H.; Richards, W. G. J. Chem. Soc., Chem. Commun. 1995,
1589–90.
(4) Rothman, J. H.; Richards, W. G. Biopolymers 1996, 39, 795–812.
(5) Rothman, J. H.; Richards, W. G. Mol. Simul. 1996, 18, 13–42.
10.1021/jo801910u CCC: $40.75
Published on Web 12/19/2008
2009 American Chemical Society
J. Org. Chem. 2009, 74, 925–928 925

major groove Hoogsteen H-bonding of the T-A base pair
orientation. Since TFOs incorporating interspersed R-anomer
nucleoside analogues have shown augmented binding stability,6,7
its preparation is also of interest.
Aryl nucleobases have been the most frequent C-nucleoside
of interest in the literature. Effective strategies for production
of aryl C-nucleosides include treatement of Hoffer’s R-chloro-
sugar (3,5-di-O-toluoyl-1-chloro-2-deoxy-R-D-ribofuranose) with
a diaryl cadmium8,9 and Heck coupling of aryl triflates or iodides
to ribofuranoid glycals.10-12 One of the more adaptable
procedures involves treatment of 3,5-di-O-silyl-protected 2-deox-
yribonolactone with an aryllithium and subsequent reduction
of the resulting hemiacetal with Et3SiH.13
FIGURE 4. C1-Vinyl-2-deoxyribofuranose produced by the 5-exo-tet
cyclization mechanism.
SCHEME 1
Regarding C-ethenyl deoxyribose derivatives, a few synthetic
strategies have been reported. The strategy of Takase et al.14
commences with addition of alkynyllithium reagents to 3,5-di-
O-benzyl-2-deoxyribofuranose affording diastereomeric mix-
tures of the corresponding ring-opened alkynyldiols. An in-
tramolecular Nicholas reaction follows giving C-alkynyl-3,5-
di-O-benzyl-2-deoxyribofuranosides with some ꢀ-selectivity.
These may be further modified to C-ethenyl derivatives. In
another example, a one-pot transformation of unprotected
monosaccharides to give styrenyl C-glycosides by Horner-
Wadsworth-Emmons ring closure and tandem halogenation/
Ramberg-Ba¨cklund sequence proceeds in reasonable yield of
the C-ethenyl deoxyribose with equal R and ꢀ anomeric
preference.15 Moreover, the availability of the nucleobase as a
sulfonylphosphonate is a requirement. Also available is a six-
step intramolecular cyclization strategy allowing C-derivatization
via Wittig addition at the C-1 of a protected 5-iodo glucofura-
nose followed by its recyclization to a C-ethenyl 2-deoxy-ꢀ-
ribofuranoside.16 Ring reclosure appears quite steric and requires
relatively high temperatures for cyclization; however, yields are
high for the E-methacrylate example.
An olefin metathesis mediated strategy holds the innate
advantages of its functional group tolerance, thermodynamic
preference for the E-configuration, and mild reaction conditions.
For this reason a strategy employing a metathesis of a C-vinyl
ribofuranose and a vinyl heterocycle was pursued.
A more direct synthesis of the C-vinyl ribose component with
respect to available strategies was sought. Even if the C1-vinyl
hemiacetal could be obtained in reasonable yield by addition
of a vinyl carbanion strategy to a 3,5-di-O-silyl-protected
2-deoxyribonolactone, specific reduction to the C1-vinyl ribo-
furanose would be difficult in the presence of an olefin.
Moreover, employment of the HWE-tandem halogenation/
Ramberg-Backlund or alkynyllithium addition-intramolecular
Nicholas rearrangement strategy could be unnecessarily labori-
ous. However, from a recent work concerning the syntheses of
7-membered carbonates,17 by 5-exo-tet elimination, the 3,5-bis-
O-TBDPS protected C-vinyl 2-deoxy-ꢀ-D-ribofuranose, 3a, was
generated stereospecifically as a side product of triphosgene
addition to the (5S) diol, 2a.
The formation of the C1-vinyl 2-deoxyribofuranose byprod-
uct, 3a, is most likely due to initial acylation at the less hindered
alcohol creating the intermediate (Figure 4) that may follow
two possible cyclization pathways. For the circumstances of their
research the desired 7-membered cyclic carbonate (Figure 4) is
achieved by a 7-exo-trig cyclization. However, the byproduct,
3a, is provided by a 5-exo-tet ring closure with inversion of
configuration at the reacting center by means of an SN2-type
pathway.17
Taking advantage of this observation, the 5-exo-tet ring
closure was effected more directly by addition of MsCl to the
(5S) diol 2a at -20 °C in dichloromethane/pyridine (Scheme
1). This yielded 67% C1-vinyl 2-deoxy-D-ribofuranose with near
complete stereoinversion to give the ꢀ-anomer as no R-anomer
could be isolated or seen by NMR. This would imply an SN2-
type mechanism. The NMR data for the isolated product fully
correspond with that reported by Anderson et al.17 for the
ꢀ-anomer, 3a. Analogously, similar treatment of the (5R) 2b
diol diastereomer stereospecifically yields the R-anomer, 3b.
The starting (5S) 2a, and (5R) 2b diols are obtained from 1,
the 3,5-bis-O-TBDPS protected 2-deoxy-D-ribofuranose18 (Scheme
1), by treatment with excess vinyl magnesium bromide and
separated chromatographically.17 Deprotection of the silyl
moieties was effected with tetra-n-butylammonium fluoride
yielding the C1-vinyl 2-deoxyribofuranose, 4a. A mixture of
the bis-O-TBDMS protected diol analogues was also prepared
by the same method; however, efficient chromatographic
separation was not as satisfactory.
(6) Noonberg, S. B.; Praseuth, D.; Helen, C. Nucleic Acids Res. 1995, 23,
4042–4049, et al.
(7) Debart, F.; Rayner, B.; Degols, G.; Imbach, J.-L. Nucleic Acids Res. 1992,
20, 1193–1200.
(8) Chaudhuri, N. C.; Kool, E. T. Tetrahedron Lett. 1995, 36, 1795–1798.
(9) Pirrung, M. C.; Zhao, X.-D.; Harris, S. V. J. Org. Chem. 2001, 66, 2067–
71.
(10) Hacksell, U.; Daves, G. D. J. Org. Chem. 1983, 48, 2870.
(11) Czernecki, S.; Dechavanne, V. Can. J. Chem. 1983, 61, 533.
(12) Coleman, R. S.; Mortensen, M. A. Tetrahedron Lett. 2003, 44, 1215–
1219.
(13) Wichai, U.; Woski, S. Org. Lett. 1999, 1, 1173–1176.
(14) Takase, M.; Morikawa, T.; Abe, H.; Inouye, M. Org. Lett. 2003, 5,
625–628.
(15) Jeanmart, S.; Taylor, R. J. K. Tetrahedron Lett. 2005, 46, 9043–9048.
(16) Egron, D.; Durand, T.; Roland, A.; Vidal, J.-P.; Rossi, J.-C. Synlett 1999,
4, 435–437.
(17) Anderson, E. A.; Davidson, J. E. P.; Harrison, J. R.; et al. Tetrahedron
2002, 58, 1943–71.
(18) Walker, J. A.; Chen, J.-J.; Wise, D. S.; Townsend, L. B. J. Org. Chem.
1996, 61, 2219–2221.
926 J. Org. Chem. Vol. 74, No. 2, 2009

FIGURE 5. 4,5-Dimethylene-2-oxazolidinone preferentially rearranges
to 5-vinyl-2-oxazolone.
FIGURE 6. Structures of Grubbs and Hoveyda-Grubbs metathesis
catalysts.
SCHEME 2
TABLE 1. Initial Catalyst and Solvent Survey for Olefin
Metathesis
C1-vinyl
ribofuranose
catalyst
solvent
T (°C)
yield (%)
3a
3a
3a
3a
3a
3a
3b
3b
3b
3b
4a
4a
4a
4a
HG-I
HG-II
G-I
CH2Cl2
CH2Cl2
CH2Cl2
toluene
CH2Cl2
toluene
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
40
40
40
110
40
110
40
40
40
40
40
63
18
10
5
14
5
67
15
8
12
NR
NR
NR
NR
G-I
G-II
G-II
HG-I
HG-II
G-I
G-II
HG-I
HG-II
G-I
SCHEME 3
40
40
40
G-II
cant over-reduction to the oxazolidinone, have not been entirely
successful. Steric interference at the oxazolone 5-position seems
to decrease over-reduction with DIBAL-H at room tempera-
ture.20 However, synthesis of the 4-hydroxymethylene deriva-
tive, 8, was achieved in reasonable yield with DIBAL-H at -78
°C. Isolation of the 4-carbaldehyde, 9, could not be achieved
under these conditions or with other general transformation
strategies such as LiAlH4-Et2NH or LiAlH(O-t-Bu)3 at 0 °C.
Nonetheless, conversion of the 4-hydroxymethylene, 8, to the
aldehyde, 9, was achieved easily and in high yield with activated
MnO2 in acetonitrile. Yields were relatively modest for this
oxidation with use of TPAP/4-methylmorpholine N-oxide.
Conversion of the aldehyde to the vinyl moiety, 10, was
achieved by Wittig reaction with excess methyltriphenyl phos-
phorane ylide in THF/HMPA at room temperature. The need
for HMPA and relatively high temperature was likely on account
of the general insolubility and betaine stabilization due to the
proximity of the lithiated amide (Scheme 3).
To date, syntheses of 4-vinyl oxazolones, thiazolones, or
azolones have yet to be reported. The closest related reported
structures, N-alkyl/aryl 5-vinyl oxazolones, are obtained initially
as a 4,5-dimethylene-2-oxazolidinone through condensation of
a diacetyl and an alkyl/aryl isocyanate. However, the exocyclic
diene preferentially rearranges in refluxing xylene to the 5-vinyl-
2-oxazolone and its Diels-Alder homoadduct as the major
product (Figure 5).19
The 4-vinyl-2-oxazolone component, 10, reported here is
obtained from the 2-oxazolone-4-carboxylic ester, 7. Through
condensation of 4-aryl/alkyl-3-bromo-2-ketoesters with methyl
carbamate, 5-aryl/alkyl-2-oxazolone-4-carboxylic esters can be
obtained in reasonable yield.20 To obtain the starting 2-keto-
butyric methyl ester, 5, commercially available 2-ketobutyric
acid (Scheme 2) was esterified with methanol in 2,2-dimethoxy-
propane in the presence of catalytic TMSCl, as a convenient
source of anhydrous HCl.21 In accord with the strategy
developed by Hoffmann et al., refluxing the 2-ketoester, 5, in
chloroform/EtOAc with CuBr2 forms the 3-bromo-2-ketoester,
6, which was subsequently purified and condensed with excess
methyl carbamate and silver triflate in refluxing toluene yielding
the 2-oxazolone-4-carboxylic methyl ester, 7. Previous attempts
from literature20 to reduce selectively the ester moiety of the
5-methyl-2-oxazolone-4-carboxylic ethyl ester, without signifi-
The olefin cross-metathesis final step allows efficient coupling
of unprotected nucleoside heterocycle to C1-vinyl 2-deoxyri-
bofuranose. Initially, catalysts (Figure 6) and solvents were
surveyed for cross-metathesis effectiveness of either the 3′,5′-
OH-2-deoxy-C1-vinyl-ꢀ-D-ribofuranose, 4a, or 3′,5′-bis-O-TB-
DPS-2-deoxy-C1-vinyl-ꢀ-D-ribofuranose, 3a, with 4-vinyl-5-
methyl-2-oxazolone, 10 (Table 1). A mixture of 1.0 equiv of
reactants at 0.2 M and catalyst (5 mol %) are refluxed in
dichloromethane or toluene for 8 h under argon. The solvent
was removed and the product purified by preparative silica thin
layer chromatography. As seen from Table 1, for the bis-O-
TBDPS-protected C1-vinyl ribofuranose 3a the best yield of
11a is achieved with Hoveyda-Grubbs-I (HG-I) catalyst at 63%
yield, in comparison to 18% with Grubbs-Hoveyda-II (HG-
II) catalyst. A possible reason for the effectiveness of HG-I over
HG-II could be explained by steric interference of the mesityl
imidazolidinylidene moiety over that of tricyclohexyl phosphine.
However, use of Grubbs-I catalyst (G-I) was also not effective
for this cross-metathesis as was also Grubbs-II catalyst (G-II).
(19) Mandal, A. B.; Gomez, A.; Trujillo, G. J. Org. Chem. 1997, 62, 4105–
4115.
(20) Okonya, J. F.; Hoffman, R. V.; Johnson, C. J. Org. Chem. 2002, 67,
1102–1108.
(21) Rodriguez, A.; Nomen, M.; Spur, B. W.; Godfroid, J. J. Tetrahdron
Lett. 1998, 39, 8563–66.
J. Org. Chem. Vol. 74, No. 2, 2009 927

FIGURE 8. Observed NOEs unique to the proposed configurations.
FIGURE 7. Thermodynamic equilibration by photoinitiated iodine-
cycle and anomerically pure C1-vinyl 2-deoxyribofuranose is
presented. Additionally, a convenient means for preparation of
a useful precursor for C1-nucleosides, diastereomerically pure
R-D and ꢀ-D C1-vinyl-2-deoxyribofuranose, by a stereospecific
cyclization is also described.
catalyzed isomerization of thiazolone analogue.
In all cases, after equilibrium was achieved to yield maximum
product, the only remaining constituents were mainly starting
reactants.
The cross-metathesis reaction is effectively quite selective
in that the product is consistent with E-configuration with no
isomer of Z-configuration able to be isolated. This is indeed an
improvement in comparison to a thermodynamic equilibration
by photoinitiated iodine-catalyzed isomerization of a closely
related thiazolone analogue of 11b (Figure 7) yielding an E/Z
ratio of 3:1.2
Experimental Procedures
4-[1-(3,5-Bis-O-(tert-butyldiphenylsilyl)-2′-deoxy-ꢀ-D-ribofurano-
syl)-2-E-ethenyl]-5-methyloxazol-2-one (11a). In 0.25 mL of dichlo-
romethane both 10 (5.0 mg, 0.046 mmol) and 3a (28 mg, 0.046
mmol) were dissolved. Then Hoveyda-Grubbs-I (HG-I) (2.8 mg,
0.005 mmol) was added. The reaction contents were refluxed under
argon for 8 h, replenishing dichloromethane solvent if necessary.
By preparative thin-layer silica chromatography (Rf 0.35, 6:1
toluene, ethyl acetate) 12 mg of white solid was isolated (63%
yield). 1H NMR (300 MHz, CDCl3) δ 7.92 (s, 1H, N-H),
7.68-7.42 (m, H10, Ar), 7.42-7.23 (m, H10, Ar), 6.26 (d, 1H,
CHdCHsCdC, J ) 15.9 Hz), 5.50 (dd, 1H, C HdCHsCdC, J
) 16.2, 6.3 Hz), 4.80 (m, 1H, H-1), 4.48 (br d, 1H, H-3, J ) 4.8
Hz), 4.03 (m, 1H, H-4), 3.45 (dd, 1H, H-5, J ) 11.1, 4.2 Hz), 3.25
(dd, 1H, H-5, J)11.1, 3.0 Hz), 2.05 (s, 3H, CdCsMe), 2.03-1.95
(m, 1H, H-2), 1.72-1.63 (m, 1H, H-2), 1.07 (s, 9H, tBu), 0.92 (s,
9H, tBu). 13C NMR (75 MHz, CD3OD) δ 156, 152, 135.9, 135.7,
133.8, 133.5, 133.2, 130.0, 129.8, 127.9, 127.8, 121.0, 119.5, 114.6,
88.4, 79.1, 76.0, 64.5, 42.9, 27.3, 27.1, 19.6, 19.5, 14.5, 10.6, 1.45.
High-resolution FAB+ MS found (M) 717.3285 (MH+), 718.3243;
C43H51O5NSi2 requires (M) 717.3306.
The initial rationale regarding the employment of sterically
hindering TBDPS O-protecting groups for the C1-vinyl-2-
deoxyribofuranose reactants is to ensure regioselective cross-
metathesis by avoiding formation of its homodimer. However,
the unprotected C1-vinyl-2-deoxyribofuranose 4a is found
unreactive under metathesis conditions even when refluxing for
48 h. Possibly its vinylidene moiety becomes more durably
associated to the catalyst through diol chelation. Nonetheless,
metathesis of the 4-vinyl-2-oxazolone, 10, with bis-O-TBDPS-
2-deoxy-C1-vinyl-R-D-ribofuranose, 3b, behaves similarly to
that of its epimer, 3a, under these reaction conditions with the
best possible yield of 11b at 67% with Hoveyda-Grubbs-I
catalyst (HG-I). Analogously, with the other catalysts yields of
11b are significantly less, with mainly starting reactants remain-
ing. The bis-O-TBDPS-ribofuranose anomers 11a and 11b were
easily silyl-deprotected with tetra-n-butylammonium fluoride
yielding their corresponding nucleoside analogues, 12a and 12b,
respectively. These nucleosides are stable upon exposure to 3%
TCA in dichloromethane or 50 mM triazole in THF for 8 h.
Irradiation of the anomeric proton produces NOEs consistent
with the orientations of each anomer (Figure 8).
Acknowledgment. This work was financially supported
through NSF Grants BES-032197 and IIS-0324845 of MN
Stojanovic.
Supporting Information Available: Detailed experimental
procedures for all compounds and spectroscopic data. This
material is available free of charge via the Internet at
http://pubs.acs.org.
Currently, no strategies for convenient preparation of such
C1-ethenyl 2′-deoxyribonucleosides are available. A means for
their synthesis via olefin metathesis of the unprotected hetero-
JO801910U
928 J. Org. Chem. Vol. 74, No. 2, 2009