Direct and facile syntheses of heterocyclic vinyl-C-nucleosides for recognition of inverted base pairs by DNA triple helix formation: First report by direct Wittig route

Article abstract of DOI:10.1021/jo062206+

(Chemical Equation Presented) The ability to recognize specific gene sequences canonically would allow precise means for genetic intervention. However, specific recognition of two of the four possible base pairs by triplex-forming oligonucleotides (TFO) as X·T-A and Y·C-G within a triplex currently remains elusive. A series of C1-vinyl nucleosides have been proposed, and their stability and specificity have been evaluated extensively by molecular dynamics simulation. Because most C-nucleoside syntheses extend through direct substitution at the C1-position, a more convenient strategy for their syntheses via a direct Wittig coupling is presented here.

Full text of DOI:10.1021/jo062206+

Direct and Facile Syntheses of Heterocyclic
Vinyl-C-Nucleosides for Recognition of Inverted
Base Pairs by DNA Triple Helix Formation:
First Report by Direct Wittig Route
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 October 24, 2006
FIGURE 2. T‚A-T triplet and C+‚G-C triplet.
The ability to recognize specific gene sequences canonically
would allow precise means for genetic intervention. How-
ever, specific recognition of two of the four possible base
pairs by triplex-forming oligonucleotides (TFO) as X‚T-A
and Y‚C-G within a triplex currently remains elusive. A
series of C1-vinyl nucleosides have been proposed, and their
stability and specificity have been evaluated extensively by
molecular dynamics simulation. Because most C-nucleoside
syntheses extend through direct substitution at the C1-
position, a more convenient strategy for their syntheses via
a direct Wittig coupling is presented here.
FIGURE 3. X3‚T-A triplet.
base pairs, T-A and C-G, directly by TFO is the major
deterrent in the development of anti-gene methodologies
Proposed here are the syntheses of a prototype 2′-deoxy-C-
vinyl â-D-ribonucleoside analogue, X3 (Figure 3). Because
TFOs that incorporate interspersed R-anomer nucleoside ana-
logues have shown augmented binding stability,1,2 synthesis of
the nucleoside R-anomer is also of interest. The design of these
2′-deoxy-C-vinyl â-D-ribonucleosides is based upon extensive
molecular dynamics evaluations3,4 that display significantly
favorable major groove Hoogsteen H-bonding of the T-A base
pair orientation. Specificity is also preserved, which shows
unfavorable interactions toward other unintended base pair
combinations.5
The growing knowledge of gene sequences has made DNA
a suitable drug target. This demands the development of a
method for sequence-specific recognition of DNA duplex.
Because genes are directly responsible for function and dysfunc-
tion that cause chronic maladies and disease, the ability to
suppress, delete, or modify them directly will lead toward more
direct avenues for therapy. The first step toward developing
synthetic bioorganic devices that operate upon gene sequences
is to have the means for their recognition.
The formation of intermolecular DNA triple helices offers
the possibility of designing compounds with extensive sequence
recognition properties that would be useful as anti-gene 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-homopyri-
midine target sites (Figure 1). Triplex-forming oligonucleotides
(TFO) strands in parallel orientation H-bonds the polypurine
duplex strand of A-T and G-C in the major groove (Hoogsteen
bonding) as T‚A-T and C+‚G-C, respectively (Figure 2). The
current inability to form stable triplexes by recognizing inverted
C-Nucleoside syntheses in the literature have been, most
commonly, for aryl nucleobases, where the synthetic strategies
generally employ an activated deoxyribose C1-position. One
of the more versatile procedures involves treatment of 3,5-di-
(1) Noonberg, S. B.; Praseuth, D.; Helen, C.; et al. Nucleic Acids Res.
1995, 23, 4042-4049.
(2) Debart, F.; Rayner, B.; Degols, G.; Imbach, J.-L. Nucleic Acids Res.
1992, 20, 1193-1200.
(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/jo062206+ CCC: $37.00 © 2007 American Chemical Society
Published on Web 04/14/2007
J. Org. Chem. 2007, 72, 3945-3948
3945

SCHEME 1
O-silyl protected 2-deoxyribonolactone with an aryllithium,
whereby the resulting hemiacetal is reduced with Et3SiH.6 Other
popular methods to produce aryl C-nucleosides include treatment
of Hoffer’s R-chlorosugar (3,5-di-O-toluoyl-1-chloro-2-deoxy-
R-D-ribofuranose) with diaryl cadmium7,8 and Heck coupling
of aryl triflates or iodides to ribofuranoid glycols.9-11
Regarding C-vinyl deoxyribose derivatives, a few synthetic
strategies have been reported. The strategy of Takase et al.12
commences with the addition of alkynyllithium reagents to 3,5-
di-O-benzyl-2-deoxyribofuranose to afford diastereomeric mix-
tures of the corresponding ring-opened alkynyldiols. A cobalt-
mediated cyclization (intramolecular Nicholas reaction) follows
to give C-alkynyl-3,5-di-O-benzyl-2-deoxy-ribofuranosides with
some â-selectivity. These may be further modified to C-vinyl
derivatives. In another example, a one-pot transformation of
unprotected monosaccharides to give styrenyl C-glycosides, via
a Horner-Wadsworth-Emmons ring-closure and a tandem
halogenation/Ramberg-Ba¨cklund sequence, proceeds in reason-
able yield of the C-vinyl deoxyribose with equal R- and
â-anomeric preference. Nonetheless, the availability of the
nucleobase as a sulfonylphosphonate is a requirement.13 A six-
step intramolecular cyclization strategy that allows C-derivati-
zation via Wittig addition at the C1 of a protected 5-iodo
glucofuranose followed by its recyclization to a C-vinyl
2-deoxy-â-ribofuranoside is also available.14 Ring reclosure
appears to be quite sterically hindered and also requires
relatively high temperatures for cyclization; however, yields are
high for the E-methacrylate example.
SCHEME 2
These examples are clever and imaginative strategies; how-
ever, some either may not be the most suitable for the creation
of vinyl C-nucleosides with reactively vulnerable base moieties
or may require protective groups with incompatible deprotection
conditions. Silyl ether protection offers significant advantages
with regard to minimal unreactiveness toward exposure to ylide
and to orthogonal deprotection conditions following attachment
of the nucleobase. Additionally, most of these procedures do
not have effective stereocontrol of nucleobase addition at the
C1 center. This leads to squandering nucleobase, especially if
no corrective epimerization pathway is available.
Examples of Wittig coupling of 2-deoxyribofuranosyl car-
baldehydes with even simple R-ester phosphoranes, let alone
any heterocyclic phosphoranes, are surprisingly limited. One
notable case is the Wittig coupling of a methoxycarbonyl-
methylene phosphorane to a 3-O-TBDMS-5-O-benzyl-2-deoxy-
ribofuranosyl carbaldehyde.15 For this particular synthesis and
others, preparation of C1-carbaldehydes14 has been commonly
achieved through Hoffer’s R-chlorosugar via substitution with
NaCN to form 2-deoxy-â-D-ribofuranosyl cyanide. This is
followed by reduction with sodium hypophosphate/Raney nickel
to form the 2-deoxy-â-D-ribofuranosyl carbaldehyde that is then
captured as the N,N′-diphenylethyleneimine. Treatment with
TsOH liberates the carbaldehyde.16 Many aspects of this strategy
are incompatible with the necessity of 3,5-O-disubstituted silyl
ether protection. A more viable and effective strategy is
illustrated below.
The strategy shown here allows a direct accession of these
derivatives where the vinyl nucleoside base adds to a 3,5-di-
O-silyl protected 2-deoxy-D-ribofuranosyl carbaldehyde via
Wittig coupling at the final phase. The 2-deoxyribofuranosyl
carbaldehydes 7 and 8 are obtained from the nitriles, 5 and 6,
respectively, according to Scheme 1. The phosphonium salt 14
was synthesized according to Scheme 2.
Synthesis of the 2-deoxy-D-ribofuranosyl carbaldehyde com-
mences with 3,5-bis-O-silyl protection of 2-deoxy-D-ribono-1,4-
lactone (1)17 that is available via aqueous bromine oxidation of
2-deoxy-D-ribose.18 Reduction of the lactones 2a,b with DIBALH
provides the 3,5-bis-O-silyl protected 2-deoxy-D-ribose 3a,b in
high yield.17 This is then converted to a mixture of O-acetyl-
2-deoxy-D-ribose anomers 4a,b to allow for subsequent trans-
formation to the nitriles 5a,b and 6a,b (Scheme 2). The R- and
â-anomers, 5a,b and 6a,b, were reduced with DIBALH to their
corresponding aldehydes, 7 and 8, respectively. The 7 and 8
series aldehydes were stable for months when stored at -78
°C (Scheme 1).
(6) Wichai, U.; Woski, S, Org. Lett. 1999, 1, 1173-1176.
(7) Chaudhuri, N. C.; Kool, E. T. Tetrahedron Lett. 1995, 36, 1795-
1798.
(8) Pirrung, M. C.; Zhao, X.-D; Harris, S. V. J. Org. Chem. 2001, 66,
2067-2071.
(9) Hacksell, U.; Daves, G. D. J. Org. Chem. 1983, 48, 2870.
(10) Czernecki, S.; Dechavanne, V. Can. J. Chem. 1983, 61, 533.
(11) Coleman, R. S.; Mortensen, M. A. Tetrahedron Lett. 2003, 44,
1215-1219.
(12) Takase, M.; Morikawa, T.; Abe, H.; Inouye, M. Org. Lett. 2003, 5,
625-628.
(13) Jeanmart, S.; Taylor, R. J. K. Tetrahedron Lett. 2005, 46, 9043-
9048.
(14) Egron, D.; Durand, T.; Roland, A.; Vidal, J.-P; Rossi, J.-C. Synlett
1999, 4, 435-437.
(15) Bradley, G. W.; Thomas, E. J. Synlett 1997, 629-631.
The predominance of the R-anomers 5a,b would be consistent
with the model proposed by Woerpel et al.19 that suggests a
(16) Bergstrom, D. E.; Zhang, P.-M.; Zhou, J. J. Chem. Soc. Perkin Trans.
1994, 20, 3029-3034.
(17) Walker, J. A.; Chen, J.-J.; Wise, D. S.; Townsend, L. B. J. Org.
Chem. 1996, 61, 2219-2221.
(18) Deriaz, R. E.; Overend, W. G.; Stacey, M.; Teece, E. G.; et al. J.
Chem Soc. 1949, 1879-1883.
(19) Larsen, C. H.; Ridgeway, B. H.; Shaw, J. T.; Woerpel, K. A. J.
Am. Chem. Soc. 1999, 121, 12208-9.
3946 J. Org. Chem., Vol. 72, No. 10, 2007

FIGURE 4. Observed NOEs for the R- and â-anomers, 5c and 21
and 6c and 17, respectively; assignments are based upon 1H-1H COSY
data.
nucleophilic “inside”-attack of the five-membered oxocarbenium
ion intermediate. In this particular case, a lower energy
pseudoaxial configuration of the 3,5-bis-O-silyl substituted
oxocarbenium would favor the 1,3 syn addition. Consequently,
treatment of 10 with trimethylsilyl cyanide (TMSCN) and
BF3‚OEt2 at -40 °C yielded a 2:1 R/â anomeric ratio. Treatment
with Et2Al-CN at 0 °C produced a better yield of the â-anomer,
but the R-anomer was still preferred at a 4:3 ratio.
FIGURE 5. 1H NMR of H2′ methylenes for R and â nitriles, 5c and
6c, respectively.
Treatment of Hoffer’s R-chlorosugar with either NaCN14 or
20
TMSCN/BF3‚OEt2 predominantly yields the â-anomer as a
1:3 R/â anomeric ratio. Moreover, with Et2Al-CN, the pre-
dominant preparation of â-C1-nitrile is achieved from a 1:121
to a 1:522 R/â anomeric ratio. Additionally, treatment of 1,3,5-
O-benzoyl-2-deoxy-D-ribofuranose with TMSCN and BF3‚OEt2
also mainly yields the â-C1-nitrile23 in a 3:7 ratio. To elucidate
the mechanisms of these anomeric preferences Narasaka and
co-workers24 have studied the influence of 3-substitution upon
the anomeric product ratio in the C-glycosidation of 1-O-acetyl-
5-O-benzyl-2-deoxy-D-ribofuranose. In accord with Woerpel’s
studies, 3-O-benzyl derivatization favors the R-anomer 82:18
(R/â), while 3-O-CH2(SO)Me derivatization deviates to favor
the â-anomer 32:68 (R/â). The cause of this deviation is
postulated to be due to the sulfoxide oxygen shielding the
R-face, likely through lone pair association with the oxocar-
benium cation. Similarly, the difference in stereoselectivity
between the ether and the ester 3,5-disubstituted 2-deoxyribose
at C1 may likely be due to a more prominent configurational
association of the more proximal 3-ester carbonyl lone pair
toward the R-face of the oxocarbenium cation intermediate than
that of the more distal 5-ester toward the â-face.
The chromatographically separated R-and â-nitrile anomer
pairs, 5a/6a and 5b/6b, are identified by the presence of unique
NOE between H4′ and H1′ protons for the â-anomer and for
the R-anomer by the presence of unique NOE between H1′ and
H5′ protons14,25 (Figure 4). In addition to the NOE results, the
nonequivalence in chemical shift of the H2-methylene protons
of the nitrile â-anomer 6 being generally greater than that of
the R-anomer 5 (Figure 5) is also consistent with what is
typically observed for C1 substituted 2-deoxyribose.26 The same
NOE pattern and H2-methylene chemical shift nonequivalence
correlation pattern (Figure 6) is observed for the R and â C-vinyl
thiazolone derivative anomers, 17 and 21, respectively.
FIGURE 6. 1H NMR of H2′ methylenes for R- and â-vinyl thiazolones,
21 and 17, respectively.
The initial 2-mercapto-4-methyl-5-hydroxy-5-acetoxymethyl
thiazole core (11) of the vinyl nucleoside X3 is synthesized by
the addition of 4-methoxybenzyl dithiocarbamate (created in
situ via CS2 and 4-methoxybenzylamine) to the bromoketone
10, which was obtained by acetylation and subsequent bromi-
nation27 of 1-hydroxy-butan-2-one (9). Desulfurization and
dehydration of 11 is achieved in one step via mercuric
trifluoroacetate to yield N-(4-methoxybenzyl)-4-acetoxymethyl-
5-methyl-thiazolone, 12. Hydrolysis of its ester moiety to the
alcohol 13 was facile, and subsequent halogenation of the
alcohol via CBr4 was achieved by the careful addition of PPh3.
Quaternization of the 4-bromomethyl thiazolone to the phos-
phonium salt 14 with n-Bu3P is achieved in the usual manner
(Scheme 2).
The Wittig coupling employing the R-anomer 13a yields
predominantly the Z-alkene, 18, despite employment of the
Schlosser modification28 or salt and solvent alteration. However,
following deprotection of the thiazolone moiety to yield 19,
olefin photoisomerization toward the more thermodynamically
stable E-alkene, 20, was easily achieved with catalytic I2.
Apparently, attempts at photoisomerization of the 4-methoxy-
benzyl protected thiazolone, 18, were not successful. This may
likely be due to stabilizing electronic effects available through
π-overlap of the 4-methoxybenzyl moiety with the TBDPS
ether moiety. Of the TBDPS-protected 2-deoxy-D-ribofuranosyl
carbaldehyde derivatives, 13a and 14a, only the C-nucleoside
R-anomer 13a was capable of the Wittig coupling. No â-anomer
(20) Jazouli, M.; Guianvarc’h, D.; Soufiaoui, M.; et al. Tetrahedron Lett.
2003, 44, 5807-5810.
(21) Iyer, R. P.; Philips, L. R.; Egan, W. Synth. Commun. 1991, 21,
2053-2063.
(22) Frazer, J. D.; Horner, S. M.; Woski, S. A. Tetrahedron Lett. 1998,
39, 1279-1282.
(23) Togo, H.; Miyagawa, N.; Yokoyama, M. Chem. Lett. 1992, 9, 1673-
1676.
(24) Ichikawa, Y.; Kubota, H.; Fujita, K.; et al. Bull. Chem. Soc. Jpn.
1989, 62, 845-849.
(25) Seitz, G.; Lachmann, J. Z. Naturforsch. 1999, B54, 549-558.
(26) Srivastava, P. C.; Robbins, R. K.; Takusagawa, F. J. Heterocycl.
Chem. 1981, 18, 1659-1662.
(27) Gladiali, S.; Pusino, A.; Rosnati, V.; Saba, A., et al. Gazz. Chim.
Ital. 1977, 107, 535-542.
(28) Schlosser, M.; Schaub, B.; de Oliveira-Neto, J.; Jeganathan, S.
Chimia 1986, 40, 244.
J. Org. Chem, Vol. 72, No. 10, 2007 3947

SCHEME 3
the introduction. The phosphonium salt components are easily
generated and available; epimerization of the carbaldehyde is
negligible under the proposed Wittig conditions. Moreover,
further control of olefin stereochemistry is available through a
photoisomerization route. With regard to some of the more
elaborate strategies, this direct, one-stage assembly and depro-
tection strategy, which utilizes the option of mild Wittig
conditions and silyl ether deprotection conditions, also allows
for the use of a wide range of nucleobases.
Experimental Procedures
4-[1-(3,5-Bis-O-(t-butyldiphenylsilyl)-2-deoxy-R-D-ribofurano-
syl)-2-Z-vinyl]-N-(4-methoxybenzyl)-5-methyl-thiazol-2-one.18
Compound 6 (787 mg, 1.47 mmol) was added to a flame-dried
flask. The flask was then purged with and placed under an
atmosphere of argon. Twenty milliliters of dry dichloromethane
was added to the flask to dissolve the starting material. The solution
was chilled to -78 °C, and n-butyllithium (1.6 M hexanes; 1.00
mL, 1.62 mmol) was added over 5 min to the stirring solution.
After the mixture was stirred for 15 min, 13a (918 mg, 1.47 mmol)
that was dissolved in 5 mL of dry dichloromethane was added over
15 min. The solution was stirred for 2 h, warmed to 0 °C, and
stirred for an additional 5 min. The reaction was quenched/washed
with 20 mL of saturated aqueous ammonium chloride, and the
organic phase was separated, dried over magnesium sulfate, and
evaporated by rotary evaporator to yield a clear yellow oil. The
residue was purified by flash chromatography (Rf ) 0.50) with 3:1
hexanes/ethyl acetate (the sample was loaded with a small, yet
sufficient, amount of dichloromethane to improve solubility) to yield
product from 14a was detected when employing reaction
conditions of up to 60 °C. However, employment of the less
sterically encumbered TBDMS-protected 2-deoxy-â-D-ribo-
furanosyl carbaldehyde 14b allowed the generation of the vinyl
C-nucleoside â-anomer, 15, through the Wittig procedure in
almost entirely the E-configuration. The E and Z configurations
of these nucleosides were identified by their vinyllic 1H NMR
coupling constants, 15 and 12 Hz, respectively. Noticeable
epimerization was not detected for either anomer at the C1
position under Wittig conditions. The preservation of stereo-
chemistry avoids squandering the nucleobase component on the
undesired anomer. Furthermore, attempts at epimerization of
the C1 center via TFA or TsOH29 of the 3,5-di-O-acetyl
protected vinyl thiazolone C-nucleoside merely resulted in the
decomposition of the nucleobase.
Deprotection of the 4-methoxybenzylamides, 15 and 18, was
effected by formation of the benzyl anion with tert-butyllithium
at -78 °C, with subsequent oxidation by O2 where the resulting
hemi-aminal sustains loss of 4-methoxybenzaldehyde.30 Depro-
tection of their silyl ethers by tetra-n-butylammonium fluoride
generates the thiazolone nucleoside R- and â-anomers, 17 and
21, respectively (Scheme 3). This strategy offers a shorter, more
modular direct approach toward the assembly of C-vinyl
ribonucleosides with respect to those available as described in
1
972 mg (78% from 6) as a clear yellow oil. H NMR (300 MHz,
CDCl3) δ 7.66-7.26 (20H, m), 7.15 (d, 2H, J ) 8.7 Hz), 6.82 (d,
2H, J ) 8.7 Hz), 6.25 (dd, 1H, J ) 9.6 Hz, 10.8 Hz), 5.75 (d, 1H,
J ) 10.8 Hz), 4.92 (d, 1H, J ) 15.3 Hz), 4.60 (d, 1H, J ) 15.3
Hz), 4.58 (m, 1H), 4.46 (m, 1H), 4.09 (m, 1H), 3.76, 3.75 (2s,
3H), 13C NMR (75 MHz, CDCl3) δ 171.5, 159.1, 141.7, 136.0,
135.9, 135.7, 135.6, 133.7, 133.6, 133.3, 133.2, 130, 129.9, 129.0,
128.8, 128.0, 127.9, 127.8, 127.5, 117.5, 114.3, 110.4, 87.5, 76.1,
75.6, 64.8, 55.6, 46.9, 41.1, 27.3, 27.1, 19.5, 13.7. (doubling of
some signals due to the presence of TBDPS rotamers). High-
resolution FAB MS MH+ [Found: (MH+), 853.3665; calculated
for C51H59O5NSi2S, (MH+), 853.3653].
Acknowledgment. This work was financially supported
through NSF Grants BES-032197 and IIS-0324845 of MNS.
Supporting Information Available: Detailed experimental
procedures for all compounds and spectroscopic data are available
free of charge via the Internet at http://pubs.acs.org.
(29) Jiang, Y.-L.; Stivers, J. T. Tetrahedron Lett. 2003, 44, 4051-4055.
(30) Williams, R. M.; Kwast, E. Tetrahedron Lett. 1989, 451-454.
JO062206+
3948 J. Org. Chem., Vol. 72, No. 10, 2007