Oligonucleotide analogues with integrated bases and backbone: Part 22 - Synthesis and association of thiomethylene-linked self-complementary AUAU and UAUA tetramers

Article abstract of DOI:10.1002/hlca.200900203

The tritylated and silylated self-complementary A*[s]U*[s] A*[s]U* and U*[s]A*[s]U*[s]A* tetramers 18 and 24, linked by thiomethylene groups (abbreviated as [s]) between a nucleobase and C(5′) of the neighbouring nucleoside unit were prepared by a linear

Full text of DOI:10.1002/hlca.200900203

2596
Helvetica Chimica Acta – Vol. 92 (2009)
Oligonucleotide Analogues with Integrated Bases and Backbone
Part 221)
Synthesis and Association of Thiomethylene-Linked Self-Complementary AUAU and
UAUA Tetramers
by Bruno Bernet, Zeena Johar, Anne Ritter, Bernhard Jaun*, and Andrea Vasella*
Laboratory of Organic Chemistry, Department of Chemistry and Applied Biosciences, ETH Zꢀrich,
Wolfgang Pauli-Strasse 10, CH-8093 Zꢀrich (e-mail: vasella@org.chem.ethz.ch)
The tritylated and silylated self-complementary A*[s]U*[s]A*[s]U* and U*[s]A*[s]U*[s]A*
tetramers 18 and 24, linked by thiomethylene groups (abbreviated as [s]) between a nucleobase and
C(5’) of the neighbouring nucleoside unit were prepared by a linear synthesis based on S-alkylation of 5’-
thionucleosides by 6-(chloromethyl)uridines, 7 or 10, or 8-(chloromethyl)adenosines, 12 or 15. The
tetramers 18 and 24 were detritylated to the monoalcohols 19 and 25, and these were desilylated to the
diols 20 and 26, respectively. The association of the tetramers 18 – 21 and 24 – 26 in CDCl3 or in CDCl3/
(D6)DMSO 95 :5 was investigated by the concentration dependence of the chemical shifts for HꢀN(3)
or H2NꢀC(6). The formation of cyclic duplexes connected by four base pairs is favoured by the presence
of one and especially of two OH groups. The diol 20 with the AUAU sequence prefers reverse-Hoogsteen,
and diol 26 with the UAUA sequence Watson – Crick base pairing. The structure of the cyclic duplex of 26
in CDCl3 at 28 was derived by a combination of AMBER* modeling and simulated annealing with NMR-
derived distance and torsion-angle restraints resulting in a Watson – Crick base-paired right-handed
antiparallel helix showing large roll angles, especially between the centre base pairs, leading to a bent
helix axis.
Introduction. – In the course of exploring oligoribonucleotide analogues wherein
the backbone of oligonucleotides is replaced by linking elements between nucleobases
(ONIBs [2]), we investigated the pairing of the self-complementary U*[s]A(*) and
A*[s]U(*) dinucleosides2) 1 – 5 (Fig. 1) [3]. Their pairing, i.e., their association to form
cyclic duplexes, depends on the sequence, and on whether the CH2OH groups of units I
and II are protected or not. Thus, the U*[s]A(*) alcohols 1 and 2 possessing a C(5’/
II)OH or a C(8)CH2OH group form preferentially cyclic duplexes, while the fully
protected analogues 3 form only linear duplexes and higher associates. In the A*[s]U(*)
series, only the alcohol 4 forms (mainly) cyclic duplexes, whereas the dinucleosides 5
lead predominantly to linear duplexes and higher associates, irrespectively of whether
1
)
)
Part 21: [1].
2
Conventions for abbreviated notation: The substitution at C(6) of pyrimidines and C(8) of purines is
denoted by an asterisk (*); for example, U* and A* for hydroxymethylated uridine and adenosine
derivatives, respectively. U(*) and A(*) represent both unsubstituted and hydroxymethylated
nucleobases. The moiety x linking C(6)ꢀCH2 or C(8)ꢀCH2 (of unit II) and C(5’) (of unit I) is
indicated in square brackets, i.e., [c] for a C-atom, [o] for an O-, and [s] for a S-atom.
ꢁ 2009 Verlag Helvetica Chimica Acta AG, Zꢀrich

Helvetica Chimica Acta – Vol. 92 (2009)
2597
HOꢀC(5’/II) is protected or not. The conformations of the cyclic duplexes of these
dinucleosides have been analysed, while the influence of the terminal CH2OH groups
has not been explored in similarly detailed way [3].
Fig. 1. The self-complementary U*[s]A(*) dimers 1 and 2 and the A*[s]U(*) dimer 4 prefer the formation
of cyclic duplexes, whereas the U*[s]A(*) dimers 3 and the A*[s]U(*) dimer 5 form mainly linear duplexes
and higher associates. TDS ¼ (thexyl)(dimethyl)silyl (thexyl ¼ 1,1,2-trimethylpropyl), MMTr¼
(monomethoxy)trityl ¼ (4-methoxyphenyl)diphenylmethyl.
In continuing the exploration of thiomethylene-linked ONIBs it appeared of
interest to analyse the pairing of the self-complementary U*[s]A*[s]U*[s]A* and
A*[s]U*[s]A*[s]U* tetranucleosides to further assess the effect of the terminal OH
groups on pairing, to test for an extrapolation of the previous conformational analysis,
and to gain information about the cyclic duplexes. We describe the synthesis of these
tetranucleosides, the analysis of their association, and the structure of one of the
duplexes.
Results and Discussion.
– 1. Synthesis of the A*[s]U*[s]A*[s]U* and
U*[s]A*[s]U*[s]A* Tetranucleosides. The synthesis is based on the nucleophilic
substitution, by a thiol or thiolate anion, of uridine or adenosine derivatives possessing
a C(6) or C(8) halogenomethyl group, respectively. We opted for a linear rather than a
convergent strategy, as it requires fewer building blocks. Four building blocks
(Scheme 1) are required for the linear synthesis of a tetramer (or a higher oligomer).
Two building blocks corresponding to the terminal unit I (Schemes 2 and 3) possess a
C(5’)SH and a protected hydroxymethyl group at C(6) or C(8), and act exclusively as
nucleophile. Two building blocks corresponding to the internal units II and III of the
tetramer possess a protected thiol and a halogenomethyl group, and act in sequence as
electrophile and as nucleophile. Two further monomers corresponding to unit IV
possess a halogenomethyl and a protected hydroxy group at C(5’), and act solely as
electrophile. Exploratory experiments showed that protection of the thiol group by
acetylation is appropriate, as the thiol group can be regenerated by treatment with

2598
Helvetica Chimica Acta – Vol. 92 (2009)
K2CO3 in MeOH without affecting the AcS group at C(5’) of the electrophilic monomer
under the conditions (K2CO3 in DMF) of the nucleophilic substitution.
Scheme 1
a) MeSO2Cl (MsCl), 4-(dimethylamino)pyridine (DMAP), CH2Cl2; 74% of 7; 91% of 10; 61% of 12;
96% of 15. b) Cl2CHCO2H, Et3SiH, CH2Cl2; 73% of 9; 96% of 14. TDS ¼ thexyldimethylsilyl (thexyl ¼
1,1,2-trimethylpropyl), MMTr¼ (monomethoxy)trityl ¼ (4-methoxyphenyl)diphenylmethyl.
The preparation of the monomers 8, 15, 10, and 12, required for the synthesis of the
A*[s]U*[s]A*[s]U* tetramer, and of the additional building blocks 13 and 7, required
for the U*[s]A*[s]U*[s]A* sequence isomer, is shown in Scheme 1. The monomers 7
and 12 corresponding to units IV were obtained in 74 and 61% yield, respectively, by
treating the known alcohols 6 and 11 [3] with MsCl and 4-(dimethylamino)pyridine
(DMAP) in CH2Cl2.
The tritylated thioacetates 8 and 13 corresponding to unit I are known [3]. Their
detritylation with Cl2CHCOOH and Et3SiH in CH2Cl2, followed by treatment of the
resulting alcohols 9 and 14 with MsCl and DMAP in CH2Cl2, gave the chlorides 10
(66%) and 15 (92%), respectively, corresponding to units II and III of the tetramers.
The 1H- and 13C-NMR data of 7, 9, 10, 12, 14, and 15 are listed in Tables 4 and 5 in
the Exper. Part. The chemical shift for HꢀC(2’) of the uridines 7, 9, and 10 in CDCl3 at
5.25 – 5.26 ppm evidences the syn-orientation of the nucleobase. In the adenosine

Helvetica Chimica Acta – Vol. 92 (2009)
2599
series, only the silyl ether 12 (d(HꢀC(2’)) ¼ 5.83 ppm) adopts completely the syn-
conformation, whereas substantial amounts of the anti-conformer of the thioacetates 14
(5.63 ppm) and 15 (5.74 ppm) are revealed by an upfield shift for HꢀC(2’), in
agreement with earlier observations [3]. A 1:1 equilibrium of the gt- and tg-conformers
of the uridine and adenosine derived thioacetates 9, 10, 14, and 15 is deduced from
J(4,5’a) and J(4,5’b) values of 6.9 – 7.5 Hz, whereas slightly smaller values for the silyl
ether 12 (6.0 and 5.7 Hz) reveal a minor contribution of the gg-conformer. Signal
overlap of the silyl ether 7 prevented the determination of the rotameric equilibrium.
The synthesis of the A*[s]U*[s]A*[s]U* tetramer 21 started by forming dimer 16
(Scheme 2). It was obtained in a yield of 67% by substituting the chloride 15 under
basic conditions with the thiol obtained upon deacetylation of the trityl ether 8.
Similarly, trimer 17 was prepared in a yield of 42% by deacetylating 16 and treating the
resulting thiol with the chloride 10. Deacetylation of 17 and reaction with the chloride
12 resulted in 44% of the tetramer 18. Debenzoylation of 18 with MeONa in MeOH
gave the diamine 19 (65%) that was detritylated to the mono-alcohol 20 (24%).
Desilylation of 20 with (HF)3 · Et3N in THF led to the desired diol 21 (38%). Yields
were not optimised. The products 16 – 20 were purified by flash chromatography. This
resulted in losing up to 30% of the products, on account to their low solubility in
solvents allowing chromatographic separation. HPLC on several types of silica gel and
gel permeation chromatography did not lead to improvements, and 21 was purified by
repeated trituration with hexane, again entailing loss of material.
The sequence-isomeric U*[s]A*[s]U*[s]A* tetramer 26 was synthesized similarly
to 21 (Scheme 3). Substitution of chloride 10 by the thiol derived from 13 gave the
dinucleoside 22 (94%), substitution of chloride 15 by the thiol derived from 22 gave the
trinucleoside 23 (> 98%), and substitution of chloride 7 by the thiol derived from 23
gave the benzoylated tetranucleoside that was directly debenzoylated to the diamine 24
(28%). Detritylation of 24 gave the mono-alcohol 25 (60%) that was desilylated to the
diol 26 (35%). Finally, 26 was deisopropylidenated by the action of aqueous CF3CO2H
in CH2Cl2 to yield 52% of the fully deprotected tetranucleoside 27. Purification of the
diamine 24 by chromatography on silica gel entailed loss of material. The mono-alcohol
25 was purified by trituration with MeOH, and the diol 26 and the polyol 27 by
trituration with acetone.
1H- and 13C-NMR data of 16 – 27 are listed in Tables 6 – 9 in the Exper. Part. In
CDCl3, the dimers 16 and 22 give rise to sharp signals, whereas the trimers 17 and 23
and especially the tetramers 18 – 21 and 24 – 26 give rise to strong line broadening due
both to their poor solubility and to association. Therefore, we recorded spectra of
solutions of the tetramers 20, 21, 24, and 25 in (D6)DMSO. In this solvent, the tetramers
are completely solvated monoplexes, as evidenced by d(HꢀN(3)) of 11.40 – 11.41 ppm
and d(H2NꢀC(6)) of 6.68 – 7.32 ppm, values that are similar to those for HꢀN(3) of
uridine mononucleosides (11.37 – 11.44 ppm [4][5]) and for H2NꢀC(6) of adenosine
1
mononucleosides (7.32 – 7.40 ppm [4][6]). The H-NMR spectrum of 26 in CDCl3 at
258 shows very broad lines, sharpened by lowering the temperature to 28, but not
sufficiently so as to detect couplings of less than 2 Hz. However, the resolution was
sufficient to allow investigating the association by the analysis of DQF-COSY, TOCSY,
and NOESY spectra (see below). The polyol 27 is insoluble in CDCl3, and the NMR
spectra were recorded in (D6)DMSO. The NMR data compiled in Tables 6 – 9 agree

2600
Helvetica Chimica Acta – Vol. 92 (2009)
Scheme 2
a) K2CO3, MeOH. b) Chloro compound, K2CO3, DMF; 67% of 16; 42% of 17; 44% of 18. c) MeONa,
MeOH; 65%. d) Cl2CHCO2H, Et3SiH, CH2Cl2; 24%. e) (HF)3 · Et3N, THF; 38%. TDS ¼ (thexyl)(di-
methyl)silyl, MMTr¼ (monomethoxy)trityl.

Helvetica Chimica Acta – Vol. 92 (2009)
2601
Scheme 3
a) K2CO3, MeOH. b) Chloro compound, K2CO3, DMF; 94% of 22; > 98% of 23. c) MeONa, MeOH;
28% of 24. d) Cl2CHCO2H, Et3SiH, CH2Cl2; 60%. e) (HF)3 · Et3N, THF; 35%. f) CF3CO2H, H2O,
CH2Cl2; 52%. TDS ¼ (thexyl)(dimethyl)silyl, MMTr¼ (monomethoxy)trityl.

2602
Helvetica Chimica Acta – Vol. 92 (2009)
well with the proposed structures. Data that are relevant for the analysis of the
association are discussed in the next chapter.
2. Association of the A*[s]U*[s]A*[s]U* and U*[s]A*[s]U*[s]A* Tetranucleosides.
The self-association of the A*[s]U*[s]A*[s]U* tetramers 19 – 21 and of the
U*[s]A*[s]U*[s]A* tetramers 24 – 26 in CDCl3 and in CDCl3/(D6)DMSO mixtures
was investigated by 1H-NMR spectroscopy. The 1H-NMR spectra of solutions of these
tetramers in CDCl3 at 258 show broad signals, evidencing association kinetics on the
NMR time-scale. Despite the broad signals, we could determine the concentration
dependence of the chemical shifts for HꢀN(3) of both uracil units of 19, 21, and 25,
resulting in ꢂshift concentration curvesꢃ (SCCs; cf. [3]). The tetramers are less well
soluble than the corresponding dimers, and this restricted the analysis of the
concentration dependence to lower concentrations (up to 15 mm) for the mono-
alcohols and the diols. The SCCs of 19 and 20 in CDCl3/(D6)DMSO 95 :5 at ambient
temperature, and of 24 and 25 in CDCl3 at 508 were analysed numerically according the
method of Gutowsky and Saika [7].
2.1. Association of the A*[s]U*[s]A*[s]U* Tetranucleosides. HꢀN(3/a) and
HꢀN(3/b)3) of the dibenzamide 18 (15 mm solution in CDCl3) that was not expected
to form a cyclic duplex resonate at 10.20 and 9.73 ppm (Table 1), similarly as HꢀN(3)
of the analogous U*[s]U* dimer (10.56 and 9.96 ppm [8]). These chemical shifts
evidence an equilibrium between the monoplex and linear U · U duplexes, although a
small contribution of linear and cyclic duplexes possessing Hoogsteen-type base pairing
cannot be excluded. Debenzoylation of 18 to 19 led to a strong downfield shift of the
HꢀN(3/a) and HꢀN(3/b) signals, now resonating at 13.03 and 12.38 ppm, and
evidencing the formation of cyclic duplexes possessing Watson – Crick-type H-bonded
base pairs. A comparison with d(HꢀN(3/I)) of the corresponding tritylated A*[s]U*
dimer (13.02 ppm [3]) may suggest that the HꢀN(3/a) signal of 19 at 13.03 ppm
corresponds to HꢀN(3/I), and the one of HꢀN(3/b) at 12.38 to HꢀN(3/III).
Detritylation of 19 to 20 leads to an upfield shift of both HꢀN(3) signals to 11.75 ppm,
revealing a preference for reverse Hoogsteen base pairing, as it was found for the
corresponding A*[s]U* dimer [3]. In agreement with this change of pairing mode,
HꢀC(2) of 20 resonates upfield to HꢀC(2) of 19 (8.20/8.24 vs. 8.30/8.50 ppm).
Reverse Hoogsteen-type base pairing appears to be favoured by an intramolecular H-
bond of HOCH2ꢀC(6/I) to OꢀC(2’/I). The upfield shift for both HꢀN(3) of diol 21
(11.29 and 11.14 ppm) reveals the exclusive formation of reverse Hoogsteen base
pairs.
Table 1. Chemical Shifts [ppm] of HꢀN(3) of the A*[s]U*[s]A*[s]U* Tetramers 18 – 21 and the
U*[s]A*[s]U*[s]A* Tetramers 24 – 26 for 10 – 15 mm Solutions in CDCl3 and at 258 (values in parenthesis
at ꢀ 208)
18
19
20
21
24
25
26
HꢀN(3/a)
HꢀN(3/b)
10.20
9.73
13.03
12.38
11.75
11.75
11.29
11.14
(12.11)
(12.03)
12.70 (12.86)
11.86 (12.06)
14.11 (HꢀN(3/IV)
13.03 (HꢀN(3/II)
3
)
The two HꢀN(3) signal of 18 – 21, 24, and 25 were not assigned to the individual uridine units. The
more deshielded HꢀN(3) signal is labeled HꢀN(3/a) and the other one HꢀN(3/b).

Helvetica Chimica Acta – Vol. 92 (2009)
2603
The SCCs for the two HꢀN(3) signals of 19 in CDCl3 are shown in Fig. 2,a. The
SCC for HꢀN(3/a) and its downfield shift evidence the presence of cyclic duplexes
already at a low concentration by the absence of a curvature below 10 mm and nearly
constant chemical-shift values above 10 mm. The SCC for HꢀN(3/b) increases slightly
from 12.33 to 12.40 ppm upon increasing the concentration from 7.5 to 32 mm; at
concentrations below 7.5 mm, the HꢀN(3/b) signal is not visible. In contradistinction,
the SCC for HꢀN(3/I) shows decreasing shift values, from 13.10 ppm at 2.2 mm to
12.94 ppm at 32 mm. This decrease suggests a further association of the cyclic duplexes
with increasing concentration (cf. [8]).
Fig. 2. a) SCCs for HꢀN(3/a) and HꢀN(3/b) of 19 in CDCl3 solution. b) Influence of the (D6)DMSO
content upon the chemical shift of HꢀN(3/a) and HꢀN(3/b) of 19 and 20 in CDCl3 solution.
The preference of diol 21 to form cyclic duplexes in CDCl3 is evidenced by a slight
increase of d(HꢀN(3/a)) (from 11.22 to 11.29 ppm) and of d(HꢀN(3/b)) (from 11.03
to 11.14 ppm) with increasing concentration (from 5.5 to 13.9 mm). Like the SCC of 19
(Fig. 2,a), the one of 21 in CDCl3 (not shown) shows no curvature at low
concentrations, preventing a quantitative analysis of the association.
Adding increasing amounts of (D6)DMSO to CDCl3 solutions of self-complemen-
tary dinucleosides shifts the monoplex > duplex equilibrium progressively in favour of
the (solvated) monoplex [9]. Upon addition of 50% of (D6)DMSO, d(HꢀN(3/a)) and
d(HꢀN(3/b)) of 19 decrease from 13.70 to 10.70, and from 12.53 to 10.59 ppm,
respectively. The lines parallel to the two branches of the curves depicting the solvent
dependence of the chemical shift of HꢀN(3a and b) (Fig. 2,b) cross at a CDCl3/
(D6)DMSO ratio of ca. 9 :1, suggesting complete solvation, i.e., the presence
exclusively of the monoplex at a (D6)DMSO content of 10%. As HꢀN(3) of a
reverse Hoogsteen base-paired cyclic duplex and HꢀN(3) of a completely solvated
monoplex in (D6)DMSO resonate at a similar field (ca. 11 ppm), one expects at best a
weak solvent dependence upon adding increasing amounts of (D6)DMSO to solutions
of 20 and 21 in CDCl3. This was indeed observed. The curves depicting the solvent
dependence for d(HꢀN(3)) for 20 (Fig. 2,b) show a weak increase upon adding 10%
of (D6)DMSO (Dd(HꢀN(3)) ¼ 0.12 and 0.04 ppm). Coalescence prevented the
determination of d(HꢀN(3) at a 2 – 7% content of (D6)DMSO. Above a content of
10% (D6)DMSO, we observed a steady decrease of d(HꢀN(3) up to a 50% content of
(D6)DMSO (Dd(HꢀN(3)) ¼ 0.31 and 0.18 ppm). This corresponds to the expected

2604
Helvetica Chimica Acta – Vol. 92 (2009)
shift change for the transition of a Hoogsteen-type base-paired duplex to a solvated
monoplex. Both HꢀN(3) of 21 in CDCl3 resonate as a single very broad singlet at
11.22 ppm. Similarly as for 19, a steady if weak decrease of this chemical shift was
observed upon adding increasing amounts of (D6)DMSO (Dd ¼ 0.14 ppm after
addition of 40% of (D6)DMSO), also in agreement with the dissociation of a
Hoogsteen-type base-paired duplex.
The above experiments show that the monomethoxytrityl ether 19 and the
corresponding alcohol 20 form a monoplex > duplex equilibrium in CDCl3/
(D6)DMSO 95 :5. This mixed solvent appeared well suited for the purpose of
determining the association constants. The numerical analysis [7] of the SCCs depicted
in Fig. 3 led to a similar value of ca. 10.8 ppm for the chemical shift of HꢀN(3/b) ¼
HꢀN(3/III) of 19 and 20, extrapolated to a concentration of 0 mm (Table 2). For
HꢀN(3/a) ¼ HꢀN(3/I) of 19 and 20, there is a difference of 0.3 ppm between the
d(HꢀN(3/I), c ¼ 0 mm) values which may be at least partly due to the different
substitution at C(6/I). A larger contribution of Hoogsteen-type H-bonded duplexes for
20 than for 19 is suggested by the upfield shift for HꢀN(3) of 20 (Dd(HꢀN(3), c ¼1)
ꢁ 0.4 ppm for both NH). The association constants Kass calculated from d(HꢀN(3/I) of
19 and 20 (310 and 251 mꢀ1, resp.) are larger than those calculated from d(HꢀN(3/III)
(126 and 115 mꢀ1, resp.), but the difference is within the large variance of the values.
The SCCs for HꢀN(3/a) and HꢀN(3/b) of the diol 21 in CDCl3/(D6)DMSO 95 :5 show
a small and steady increase of the chemical shift upon increasing the concentration
from 3.2 to 7.7 mm (Dd ꢂ 0.06 ppm), preventing numerical analysis. These differences
evidence a large contribution of (reverse) Hoogsteen H-bonded cyclic duplexes in the
association equilibrium of 21 (cf. [3]).
To further probe the association of these tetramers, we investigated the temperature
dependence of the HꢀN(3) signals of the fully O-protected 19 (15.8 mm in CDCl3;
Fig. 4). At 408, HꢀN(3/I) and HꢀN(3/III) resonate as broad singlets at 12.5 and
Fig. 3. SCCs for both HꢀN(3) of 19 and 20 in CDCl3/(D6)DMSO 95 :5

Helvetica Chimica Acta – Vol. 92 (2009)
2605
Table 2. Numerical Analysis of the SCCs of the A*[s]U*[s]A*[s]U* and U*[s]A*[s]U*[s]A* Tetramers
1
19, 20, 24, and 25 in Fig. 2, b, and Fig. 4: Calculated H-NMR Chemical Shifts [ppm] of HꢀN(3) or of
H2NꢀC(6) of the Monoplex (c ¼ 0 mm) and of the Cyclic Duplexes (c ¼1), and Calculated Association
Constants Kass [mꢀ1
]
19
20
25
24
Solvent
CDCl3/(D6)DMSO 95 : 5 CDCl3/(D6)DMSO 95 : 5 CDCl3
CDCl3
Temperature 258
258
508
508
HꢀN(3/a)
HꢀN(3/a)
11.12 ꢃ 0.07
11.90 ꢃ 0.06
251 ꢃ 118
HꢀN(3/a)a) H2NꢀC(6/c)b)
d (c ¼ 0 mm) 11.42 ꢃ 0.17
7.69 ꢃ 0.06
11.89 ꢃ 0.04
6523 ꢃ 1403
5.85 ꢃ 0.06
7.04 ꢃ 0.08
456 ꢃ 172
d (c ¼1 )
12.44 ꢃ 0.09
310 ꢃ 253
Kass
HꢀN(3/b)
HꢀN(3/b)
HꢀN(3/b)a) H2NꢀC(6/d)
d (c ¼ 0 mm) 10.76 ꢃ 0.13
10.80 ꢃ 0.05
12.07 ꢃ 0.12
115 ꢃ 43
7.70 ꢃ 0.05
11.60 ꢃ 0.04
3630 ꢃ 479
5.85 ꢃ 0.37
7.07 ꢃ 0.14
259 ꢃ 333
d (c ¼1 )
12.41 ꢃ 0.22
Kass
126 ꢃ 74
a) Including a value of 7.70 ppm for 0.0001 mm. b) Including a value of 5.85 ppm for 0.0001 mm.
11.85 ppm, respectively. They coalesce and disappear at 08, and reappear at ꢀ 208 as
two broad singlets at 14.05 and 13.7 ppm, respectively. At ꢀ 408, two rather sharp
singlets at 14.15 and 13.75 ppm evidence a single cyclic duplex possessing Watson –
Crick-type base pairs. The HꢀN(3) signals of the monoplex are expected at a
somewhat lower field than those of HꢀN(3) of the monoplex at room temperature
(7.7 ppm [3]), considering the expected downfield shift due to cooling to ꢀ 408. Five
broad singlets integrating for 5 H appear between 7.9 and 8.5 ppm at temperatures
below the coalescence temperature. They have to be assigned to the associated NH
atoms of the H2NꢀC(6/II and IV) groups of the cyclic duplex, and to HꢀN(3/I and III)
and HꢀC(2/II and IV) of the monoplex, whereas HꢀC(2/II and IV) of the cyclic
duplex resonate at ꢀ 408 as sharp singlets at 8.72 and 8.47 ppm.
Fig. 4. Temperature dependence of the HꢀN(3) signals of 19 for a 15.8 mm solution in CDCl3

2606
Helvetica Chimica Acta – Vol. 92 (2009)
For the mono-alcohol 20, we expected Watson – Crick- and Hoogsteen-type base-
paired duplexes. HꢀN(3/a) and HꢀN(3/b) of 20 resonate as two rather sharp singlets
at 10.82 and 10.47 ppm at 508, and at 208 as one broad singlet at 11.7 ppm. At ꢀ 10 to
ꢀ 408, coalescence was observed. The chemical shift of HꢀN(3/a and b) at 208 agrees
well with an equilibrium of Watson – Crick- and Hoogsteen-type base-paired cyclic
duplexes.
Unfortunately, solutions of 21 in CDCl3 gave rise to broad signals both at 258 and
28; no NOESY cross-peaks between NH and other H signals could be detected. The
structure of the reverse Hoogsteen H-bonded cyclic duplex of 21 was modeled
(AMBER* in Macromodel v. 6.0 [10]) by extending the structure of the cyclic duplex of
the corresponding A*[s]U* dimer [3], and fixing the distances of all H-bonds and the
torsion angles of the linking CH2SCH2 moieties. This led to a rather regular right-
handed A helix with ca. seven residues per turn (Fig. 5). After releasing all constraints,
the AMBER*-modeled structure showed large propeller and buckle twists that result
from avoiding a close contact between O¼C(4) of U and ROCH2ꢀC(8) of A. This may
be an artifact, as suggested by ab initio calculation (Spartan 2004: Hartree-Fock
calculation with the 3-21G basis set [11]) of a reverse Hoogsteen base pair of uracil with
8,9-dimethyladenine that suggests a binding interaction between O¼C(4) of uridine
and H3CꢀC(8) of 8,9-dimethyladenine4). There is (on the basis of modeling),
therefore, no reason to doubt the reverse Hoogsteen base pairing of 21.
Fig. 5. AMBER*-Modelled (using constraints for the H-bond distances and the torsion angles of the
CH2SCH2 linking units) cyclic duplex of the diol 21 connected by reverse-Hoogsteen base pairing (H-
atoms of isopropylidene groups omitted for enhanced clarity; units of the complementary strand marked
with a star)
CꢀH ··· O¼C H-bonds in b-sheets of proteins were evidenced by h3J(C,C) scalar couplings [12].
Their calculated association enthalpy DH298 of ꢀ 3 kcal/mol corresponds roughly to half the size of
the association enthalpy of a NꢀH ··· O¼C H-bond [13].
4
)

Helvetica Chimica Acta – Vol. 92 (2009)
2607
2.2. Association of the U*[s]A*[s]U*[s]A* Tetranucleosides. The HꢀN(3) signals of
the fully O-protected analogue 24 (10 – 15 mm in CDCl3) could not be observed at 258
due to coalescence. They appear at ꢀ 208 at 12.11 and 12.03 ppm (Table 1). The two
HꢀN(3) of the mono-alcohol 25 resonate at ꢀ 208 at 12.86 and 12.06 ppm, and at 258 at
12.70 and 11.86 ppm. At 258, the two HꢀN(3) of the diol 26 resonate at 14.11 and
13.03 ppm revealing Watson – Crick-type H-bonded cyclic duplexes. The upfield shifts
for the HꢀN(3) signals of 24 and 25 (Dd ¼ 1.2 – 2.2 ppm) may be rationalized by an
increasing contribution of Hoogsteen-type associated duplexes, or, similarly as it was
observed for the corresponding U*[s]A* dimers [3], by an increasing contribution of
linear duplexes and of the monoplex. 2D-NMR Experiments (see below) show that the
more deshielded HꢀN(3/a) of 26 corresponds to HꢀN(3/IV), but do not settle the
assignment of the HꢀN(3) signals of 24 and 25. The uncertainty stems from a
comparison of the chemical shifts for HꢀN(3) of 24 – 26 with those for the
corresponding U*[s]A* dimers (12.5 mm CDCl3) where HꢀN(3/II) resonates at 11.1,
11.9, and 12.8 ppm, respectively, suggesting that the more shielded HꢀN(3/b) of 26
corresponds to HꢀN(3/IV).
The HꢀN(3) signals of the mono-alcohol 25 in CDCl3 at 258 are very broad. At 508,
however, the signals are rather sharp and allow determination of the concentration
dependence. The SCC for HꢀN(3/a) of 25 (Fig. 6,a) shows the characteristic shape
denoting an equilibrium between monoplex and one or several cyclic duplexes, i.e., a
strong curvature below 10 mm and a plateau above 15 mm. A weaker curvature of the
SCC for HꢀN(3/b) of 25 suggests the formation also of minor amounts of linear
duplexes. Numerical analysis of the SCCs led to reliable results only upon including a
value of 7.70 ppm/0.0001 mm (cf. [3][14]), resulting in Kass values of 6523 ꢃ 1403 and
3630 ꢃ 479 mꢀ1 (Table 2). Assuming the same relative chemical shifts for HꢀN(3/II)
and HꢀN(3/IV) of 25 as for 26 one may rationalize these results by an equilibrium
between cyclic duplexes with all four units base-paired, and cyclic duplexes where only
units III and IV pair. In such a partially paired duplex, HꢀN(3/a ¼ IV) is completely
involved in a base pair, and HꢀN(3/b ¼ II) only partially so.
Fig. 6. a) SCCs for HꢀN(3/a) and HꢀN(3/b) of 25 in CDCl3 at 508. b) SCCs for H2NꢀC(6/a) and
H2NꢀC(6/b) of 24 in CDCl3 at 508.
1
The temperature dependence of the H-NMR spectra of 25 (12 mm in CDCl3)
between 22 and ꢀ 408 (Fig. 7) evidences the predominant formation of a single

2608
Helvetica Chimica Acta – Vol. 92 (2009)
Watson – Crick base-paired cyclic duplex at ꢀ 408. At 228, HꢀN(3/a ¼ IV?) and
HꢀN(3/b ¼ II?) resonate as broad singlets at 12.70 and 11.86 ppm, whereas the
H2NꢀC(6) signals are hidden due to coalescence. Lowering the temperature to 108 led
to a weak downfield shift for HꢀN(3/a ¼ IV?) and HꢀN(3/b ¼ II?) (Dd ¼ 0.19 and
0.26 ppm, resp.) and to the appearance of four NH2 signals, i.e., two signals at 9.54 and
9.46 ppm of associated HꢀN, and two signals at 7.55 and 6.55 ppm of free HꢀN. All
these signals became sharper upon cooling to ꢀ 1085). While these six NH signals
evidence a single, Watson – Crick base-paired cyclic duplex, additional very weak
HꢀN(3) signals (at 11.8, 12.3, 12.55, and 13.2 ppm, ꢀ 408) suggest a small contribution
of two other Watson – Crick-type base-paired cyclic duplexes.
Fig. 7. Temperature dependence of the HꢀN(3) and HNꢀC(6) signals of 25 between 9 and 13 ppm for a
12-mm solution in CDCl3
A cyclic duplex of 25 (molecular mass 1439.73 g/mole) is further evidenced by the
apparent molecular mass of 2850 ꢃ 100 for a 10 mm solution in CH2Cl2, as determined
by vapour pressure osmometry.
The HꢀN(3) signals of the fully O-protected 24 in CDCl3 show coalescence at 258.
The signal is very broad at 508 and does not allow us to determine the concentration
dependence of the chemical shift. For this reason, we followed the concentration
dependence of the two H2NꢀC(6) signals of the A(I and III) moieties at 508 which
were only hidden between 6.72 and 6.77 ppm by the signals of the MeOC6H4 group
(Fig. 6,b). In agreement with observations for adenosine-derived dimers [3], the two
H2NꢀC(6/c and d) signals (integrating for two H) were assigned to one H2NꢀC(6)
group each of the two adenosine moieties; they represent an average of H-bonded and
free HN. Numerical analysis of the SCC for H2NꢀC(6/d) led to a reliable result, with
d(H2NꢀC(6), c ¼ 0 mm) of 5.85 ꢃ 0.37 ppm and Kass ¼ 259 ꢃ 333 mꢀ1 (Table 2).
Numerical analysis of the SCC for H2NꢀC(6/c) also gave a reliable result upon
5
)
There is a weak influence of the temperature on the chemical shift for the associated NH upon
cooling from 10 to ꢀ 408, i.e., a weak downfield shift for HꢀN(3/a), HꢀN(3/b), and HNꢀC(6/c)
(Dd ꢂ 0.20 ppm) and a weak upfield shift for HNꢀC(6/d) (Dd ¼ 0.13 ppm). The chemical shift of the
free NHs is hardly influenced by the temperature (Dd ꢂ 0.03 ppm).

Helvetica Chimica Acta – Vol. 92 (2009)
2609
including a value of 5.85 ppm/0.0001 mm, with Kass ¼ 456 ꢃ 172 mꢀ1. These approximate
values evidence a weak association of the fully O-protected 24, similarly to the
corresponding U*[s]A* dimer (Kass ¼ 227 mꢀ1 [3]), although the values are based
on signals representing an average of H-bonded and free H of the NH2 groups (cf. [8]).
The HꢀN(3) signals of the diol 26 in CDCl3 at 278 appear at 14.11 and 13.03 ppm,
independently of the concentration (2 – 10 mm). This evidences a strong association
(Kass ꢄ 104 mꢀ1). The broad HꢀN(3) signals of 26 disappear in the noise upon
progressive dilution of a 12 mm solution in CDCl3/(D6)DMSO 95 :5 preventing a
numerical determination of Kass. The strong association suggested analyzing the
1
structure of the duplex, and this required a detailed analysis of the H-NMR spectra.
2.2.1. Qualitative Analysis of the 1H-NMR Data of the Diol 26 in CDCl3 at 28. At 28,
the resolution of the 1H-NMR spectra of a 5 mm solution of 26 in CDCl3 was sufficient
to allow investigating the associated structure by analysis of DQF-COSY, TOCSY, and
NOESY spectra. An unambiguous assignment of the signals was only feasible by a
combined analysis of these spectra.
The signals of the U units II and IV were unambiguously assigned by TOCSY cross-
peaks between HꢀN(3) and HꢀC(5), intra-units ROESY cross-peaks between
HꢀC(5) and CH2ꢀC(6), CH2ꢀC(6) and HꢀC(1’), and HꢀC(1’) and HꢀC(4’), and
TOCSY cross-peaks between HꢀC(1’) and HꢀC(2’), and between HꢀC(4’) and
H2C(5’). Similarly, the CH2ꢀC(8) signals of the A units I and III were correlated by
NOESY and DQF-COSY cross-peaks with the corresponding H2C(5’) signals. The
HꢀC(2) signals were assigned on the basis of weak NOESY cross-peaks with the
corresponding HꢀC(3’) signals.
Only the H2NꢀC(6) signals could not be assigned by intra-unit cross-peaks. They
resonate as two pairs of singlets at 9.15/7.47 and 7.97/5.66 ppm, the non-equivalence of
the H-atoms of both NH2 groups evidencing the formation of a stable cyclic duplex
connected by four base pairs. The pairs of singlets were assigned by DQF-COSY and
TOCSY cross-peaks. The individual signals were assigned with the help of inter-unit
NOESY cross-peaks of HꢀN(3/IV) and HꢀN(3/II). HꢀN(3/IV) shows cross-peaks
with the NH2 signals at 9.15/7.47 ppm and the HꢀC(2/I) signal at 8.38 ppm, and
HꢀN(3/II) shows cross-peaks with the other NH2 group at 7.97/5.66 ppm and the
HꢀC(2/III) signal at 7.94 ppm. In addition, both HꢀN(3) signals show exchange-
NOESY cross-peaks with both OH signals at 5.77 and 4.68 ppm, and with HDO at
1.80 ppm. The cross-peaks with HꢀC(2) evidence Watson – Crick or reverse Watson –
Crick base pairing between units I and IV, and between units II and III. As expected
from the strong downfield shift of both HꢀN(3) signals, there is no evidence for
Hoogsteen-type base pairing (no cross-peaks between HꢀN(3/IV) and CH2ꢀC(8/I)
nor between HꢀN(3/II) and CH2ꢀC(8/III)).
HOꢀC(5’/IV) resonates as a doublet at 4.69 ppm. The downfield shift and the large
coupling of 9.7 Hz is consistent with a rather persistent bifurcated H-bond to O¼C(2/
IV) and OꢀC(4’/IV) (cf. [3] and refs. cit. therein). HOCH2ꢀC(8/I) resonates as a
doublet at 5.77 ppm. The large J(H,OH) value of 11.5 Hz evidences that the OH group
is forming a H-bond, with HꢀO antiperiplanar to one CꢀH of the CH2 group. This
excludes an intramolecular H-bond to N(7/I), as this would require HꢀO to be in the
s-plane of the adenine moiety (torsion angle HꢀCꢀOꢀH ꢃ 1208), while an
intramolecular H-bond to OꢀC(2’/I) agrees well with this coupling.

2610
Helvetica Chimica Acta – Vol. 92 (2009)
The combined analysis of the DQF-COSY, TOCSY, and NOESY spectra allowed
us to unambiguously assign all 1H-NMR signals of 26 in CDCl3 at 28 (see Table 8 in the
Exper. Part). There are several unexpected chemical shifts that must be due to
anisotropy effects in the cyclic duplex. One notes a strong upfield shift of HꢀC(5/II)
(4.77 ppm; Dd ꢁ 1 ppm), an upfield shift for HꢀC(2/III) (7.94 ppm; Dd ꢁ 0.4 ppm), a
strong upfield shift for HꢀC(1’/IV) (5.23 ppm; Dd > 0.6 ppm), a downfield shift for
HꢀC(2’/II) (5.49 ppm; Dd ꢁ 0.3 ppm), upfield shifts for HꢀC(2’/I) (5.05 ppm; Dd >
0.45 ppm) and HꢀC(2’/III) (5.20 ppm; Dd > 0.3 ppm), a downfield shift for HꢀC(3’/
I) (5.41 ppm; Dd ꢁ 0.3 ppm), a downfield shift for HꢀC(4’/I) (4.65 ppm; Dd ꢁ
0.3 ppm), and an upfield shift for HꢀC(4’/IV) (3.90 ppm; Dd ꢁ 0.4 ppm)6). Large
shift differences (0.47 – 0.86 ppm) are found for the geminal H-atoms of CH2ꢀC(8/III),
CH2ꢀC(6/II), CH2ꢀC(6/IV), H2C(5’/I), H2C(5’/III), and H2C(5’/IV), evidencing a
strongly anisotropic environment. The geminal H-atoms of CH2ꢀC(8/I) and H2C(5’/
II), however, show only a weak Dd of 0.11 – 0.13 ppm.
The syn-orientation of all nucleobases in the duplex of 26 is evidenced by NOESY
cross-peaks between HꢀC(1’) and CH2ꢀC(6 or 8), and corroborated by NOESY
cross-peaks between HꢀC(3’) and HꢀC(2) of the adenosine units. Hence, the upfield
shift observed for HꢀC(2’/I) and HꢀC(2’/III) must indeed be due to anisotropy effects
and does not denote an anti-orientation of the adenosine unit. A northern (N)
conformation is evidenced for units IꢀIII by DQF-COSY cross-peaks between
HꢀC(3’) and HꢀC(4’), and by the absence of DQF-COSY cross-peaks between
HꢀC(1’) and HꢀC(2’); line broadening prevents the exact determination of the
J(1’,2’)/J(3’,4’) ratio. Unit IV, however, shows a preference for a southern (S)
conformation, evidenced by DQF-COSY cross-peaks between HꢀC(1’/IV) and
HꢀC(2’/IV). It is likely that this conformation is induced by the intramolecular
bifurcated H-bond of HOꢀC(5’/IV) to O¼C(2/IV) and OꢀC(4’/IV). NOESY Cross-
peaks between the signals of HOCH2ꢀC(8/I) and both HꢀC(1’/I) and HꢀC(2’/I)
confirm the intramolecular H-bond to OꢀC(2’/I).
A gt- or tg-orientation of the sulfanyl substituent of units I – III is evidenced by a
large coupling of 10.2 – 10.8 Hz between HꢀC(4’) and the more strongly deshielded
HaꢀC(5’). The also expected small coupling of HꢀC(4’) with the less strongly
deshielded HbꢀC(5’) was not resolved. The gt- or tg-orientation of the sulfanyl group is
corroborated by strong DQF-COSY cross-peaks between the HꢀC(4’) and HaꢀC(5’)
signals, and the absence of cross-peaks between HꢀC(4’) and HbꢀC(5’). As expected
for an ap orientation of HaꢀC(5’) and HꢀC(4’), and a gauche orientation of HbꢀC(5’)
and HꢀC(4’), the NOESY spectrum shows weaker cross-peaks between the signals of
HaꢀC(5’) and HꢀC(4’) than between those of HbꢀC(5’) and HꢀC(4’) (volume ratio
ca. 1:1.5 – 2.0). However, both the HaꢀC(5’) and HbꢀC(5’) signals show cross-peaks
with the HꢀC(3’) signal of similar intensity (volume ratio ca. 1:1). This is only
compatible with a gt-conformation. Thus, the more strongly deshielded HaꢀC(5’/
6
)
Some of these unexpected shifts may be rationalized on the basis of the calculated duplexes 26A –
26C (see below). The upfield shifts of HꢀC(5/II) and HꢀC(2/III) agree with their localisation in
the shielding cone above A(I) and U(IV), respectively, whereas a downfield shift of HꢀC(2’/II) is
expected from its position in the deshielding cone near the plane through U(II).

Helvetica Chimica Acta – Vol. 92 (2009)
2611
IꢀIII) corresponds to HRe, in agreement with the analogous assignment for HaꢀC(5’/I)
of the corresponding U*[s]A* dimer [3]. The intramolecular H-bond of HOꢀC(5’/IV)
requires a gg-orientation. This conformation could not be corroborated by DQF-COSY
and NOESY cross-peaks due to overlapping HꢀC(4’/IV) and HaꢀC(5’/IV) signals.
In the absence of any coupling constants, the deduction of the approximate torsion
angle about the H2CꢀC(6 or 8) bond of the sulfanyl moiety has to rely on the relative
NOE intensities of the two CH2. In U(IV), both HCꢀC(6/IV) show a strong
intraresidual NOE to HꢀC(1’/IV). This indicates that, in U(IV), both HCꢀC(6) point
towards HꢀC(1’), and that the torsion angle N(1)ꢀC(6)ꢀCH2ꢀS is 180 ꢃ 608. With
this general orientation it was possible to individually assign the two H2CꢀC(6/IV); the
one showing a NOE with HꢀC(2’/IV) was assigned to HSi. For A(III) and U(II), a
specific assignment was only possible via a cyclic process using the structures calculated
with pseudo-atoms for H2CꢀC(8 or 6) to deduce the assignment that is congruent with
the observed NOEs.
2.2.2. AMBER* Modeling of the Cyclic Duplex of 26. A rough calculation of the
solution structure of 26 at 28 in CDCl3 by simulated annealing with NMR-derived
distance and torsion-angle restraints (programme XPLORE-NIH, version 2-0-4 [15])
led to 64 Watson – Crick base-paired structures showing neither NOE-distance
(>0.1 ꢄ) nor torsion-angle violations (> 58) [16]. The 14 structures lowest in energy
had in common the syn-orientation of all nucleobases and the tg-orientation of all
sulfanyl groups, but suffered from inefficient base pairing (evidenced by O ··· HN and
NH ··· N distances of 1.51 – 1.84 and 2.02 – 2.81 ꢄ, resp.), unfavourable propeller and
buckle twists (up to 508), free OH groups, and especially from poor base stacking
(distance of 4.6 – 7.2 ꢄ between the centre base pairs and of 3.5 – 5.6 ꢄ between the
border base-pairs). The conformation of the H2C(5’)ꢀSꢀCH2 linker between units I
and II was similar to that between units II and III, but different from that between units
III and IV. The structure showing the weakest propeller and buckle twists was selected
for optimization with the programme AMBER* implemented in Macromodel v. 6 [10].
First, constraints were set for Watson – Crick base pairing, for the formation of the
intramolecular H-bonds of the OH groups, and for a distance of 3.4 ꢄ between the base
pairs. In one calculation, the conformation of the H2C(5’)ꢀSꢀCH2 linkers was
maintained, and in another calculation the conformation of the linker between units III
and IV was brought in line with that of the other linkers. After this calculation, all
constraints were released for the final optimisation. This led to the duplex stuctures
26A and 26B (5 kcal/mol lower in energy than 26A), differing essentially in the
conformation of the linker between units III and IV (Fig. 8). The front views of 26A
and 26B show more or less parallel border base pairs and strongly inclined centre base
pairs evidenced by roll angles7) of 4.9, 5.1, and 15.08 for 26A, and of twice 5.7 and 15.28
for 26B. Thus, only the border base pairs show base stacking (distance 3.20 – 3.25 ꢄ).
The strong inclination of the centre base pairs leading to a bent helix axis is presumably
the result of backbone strain in these cyclic duplexes. The top views of 26A and 26B
7
)
Determined by measuring the angle between the two main planes through C(2U), C(4U), C(6U),
C(3A), C(6A), and C(8A) of the two base pairs.

2612
Helvetica Chimica Acta – Vol. 92 (2009)
reveal a mean twist angle of ca. 608, i.e., six units per turn in a right-handed A helix
possessing a central hole the size of a phenyl ring.
Fig. 8. AMBER*-Modelled cyclic duplexes 26A and 26B connected by Watson – Crick base pairing (H-
atoms of isopropylidene groups omitted for enhanced clarity; units of the complementary strand marked
with a star)
The structures of the duplexes 26A and 26B were analyzed in more detail (Fig. 8
and Table 3). The strands are connected by four Watson – Crick base pairs (charac-
terized by NH ··· N and NH ··· O distances of 1.75 – 1.76 ꢄ) that show no or only weak

Helvetica Chimica Acta – Vol. 92 (2009)
2613
Table 3. H-Bond Distances [ꢄ] and Selected Torsion Angles [8] of the Cyclic Duplexes 26A, 26B, and 26C
(obtained by AMBER* and XPLOR-NIH calculations, resp.).
H ··· X Bond
H ··· X Distance [ꢄ]
26A
26B
26C
N(3)ꢀH ··· N(1)
C(4)¼O ··· HNꢀC(6)
OH ··· O(2’/I)
1.75, 1.78a)
1.78, 1.71a)
2.00
1.75
1.76 – 1.77
1.98
1.92, 2.05a)
2.16, 1.98a)
OH ··· O¼C(2/IV)
1.71
2.33
1.70
2.43
OH ··· O(5’/IV)
Torsion Angle
Short notation
Torsion angle [8]b)
O(4’)ꢀC(1’)ꢀN(9)ꢀC(4), unit I
O(4’)ꢀC(1’)ꢀN(1)ꢀC(2), unit II
O(4’)ꢀC(1’)ꢀN(9)ꢀC(4), unit III
O(4’)ꢀC(1’)ꢀN(1)ꢀC(2), unit IV
O(4’)ꢀC(4’)ꢀC(5’)ꢀS, unit I
O(4’)ꢀC(4’)ꢀC(5’)ꢀS, unit II
O(4’)ꢀC(4’)ꢀC(5’)ꢀS, unit III
O(4’)ꢀC(4’)ꢀC(5’)ꢀO, unit IV
C(3’)ꢀC(4’)ꢀC(5’)ꢀS, unit I
C(3’)ꢀC(4’)ꢀC(5’)ꢀS, unit II
C(3’)ꢀC(4’)ꢀC(5’)ꢀS, unit III
C(3’)ꢀC(4’)ꢀC(5’)ꢀO, unit IV
C(4’)ꢀC(5’)ꢀSꢀCH2, unit I
C(4’)ꢀC(5’)ꢀSꢀCH2, unit II
C(4’)ꢀC(5’)ꢀSꢀCH2, unit III
C(5’)ꢀSꢀCH2ꢀC(6), unit I
C(5’)ꢀSꢀCH2ꢀC(8), unit II
C(5’)ꢀSꢀCH2ꢀC(6), unit III
SꢀCH2ꢀC(6)ꢀN(1), unit I
c/I
c/II
52.3
50.0
52.7
48.1
90.2
80.2
64.5
51.9
53.3
45.1
54.7
90.7
78.7
77.3
59.5
77.1
67.5
36.0
90.4
90.1
76.0
c/III
c/IV
h1/I
h1/II
h1/III
h1/IV
h2/I
h2/II
h2/III
h2/IV
q/I
q/II
q/III
i/I
c
ꢀ 59.2
ꢀ 152.3
ꢀ 160.2
ꢀ 177.4
59.5
ꢀ 59.7
ꢀ 66.0
ꢀ 147.2
ꢀ 59.8
ꢀ 59.2
64.2
ꢀ 64.0
ꢀ 151.7
ꢀ 161.6
ꢀ 163.1
54.8
ꢀ 61.1
ꢀ 68.9
ꢀ 70.8
ꢀ 61.7
ꢀ 60.5
ꢀ 47.7
ꢀ 61.8
ꢀ 59.0
ꢀ 60.7
)
ꢀ 150.1
ꢀ 150.0
ꢀ 164.7
c
)
ꢀ 78.7
ꢀ 92.5
ꢀ 177.1
ꢀ 52.5
ꢀ 62.0
62.7
ꢀ 73.4
ꢀ 61.8
ꢀ 149.9
i/II
i/III
k/I
k/II
k/III
ꢀ 61.2
ꢀ 59.0
ꢀ 162.6
SꢀCH2ꢀC(8)ꢀN(9), unit II
SꢀCH2ꢀC(6)ꢀN(1), unit III
a) First value: terminal units, second value: central units. b) Mean value of both strands (Daꢂ 0.48).
c) Not determined (programme allows free rotation of the CH2OH group).
propeller and buckle twists. All nucleobases adopt a syn-conformation (c/I – IV 45 –
558). The sulfanyl moieties of units I – III adopt a distorted gt-conformation. The
distortion decreases from unit I (h1/I ca. 908) to units II (h1/II ca. 808) and to units III
(h1/III ¼ 778 for 26B and 64.58 for 26A; see Table 3 for the definition of the angles).
The C(4’)ꢀC(5’)ꢀSꢀCH2ꢀC(6 or 8)ꢀN(1 or 9) fragment of all linkers adopt a gꢀgꢀgꢀ
(torsion angles q, i, and k) conformation, except for the linker between units II and III
of 26Awhich adopts a tgþt-conformation. In agreement with the experimental data, the
3
furanose rings of units II and III of 26A and 26B adopt a northern E-conformation,
while unit I adopts a southern 2E-conformation, with a calculated J(1,2) > 7.0 Hz that is
not in agreement with the experimental data. In the AMBER* calculation, the
southern conformation is a consequence of the intramolecular H-bond to OꢀC(2’),

2614
Helvetica Chimica Acta – Vol. 92 (2009)
while Maruzen modeling suggests that this intramolecular H-bond is also compatible
with a northern conformation8).
HOꢀC(5’/IV) of 26A and 26B forms an intramolecular H-bond to O¼C(2/IV)
rather than a bifurcated H-bond to OꢀC(4’/IV), as indicated by H ··· O distances of
1.71 and 1.70 vs. 2.33 and 2.43 ꢄ, respectively (Table 3), whereas bifurcated (if strongly
asymmetric) H-bonds are found in crystal structures of closely related uridine
monomers (H ··· O distances of 1.71 – 1.83 and 2.15 – 2.37 ꢄ [17 – 19]). This H-bond
leads to a gg-orientation of the CH2OH side chain and an E1 furanose conformation, as
also found in the X-ray structures.
Finally, the inter-proton distances of the minimized structures 26A and 26B were
compared with the NOE-derived distance constraints. For both structures 26A and 26B,
minimized with AMBER*, NOE back-calculations showed a number of large
discrepancies (ꢀ1.2 ꢄ < d <þ 1.0 ꢄ) between the experimental NOE constraints and
the corresponding distances in 26A (13 violations ꢄ 0.5 ꢄ) and 26B (24 violations
ꢄ 0.5 ꢄ). Most of these violations involved the sequential distances across the sulfanyl
linkers, and the relative intra-residual distances between the CH2ꢀC(6/8) and
HꢀC(1’).
2.2.3. Calculation of the Solution Structure of 26 in CDCl3 at 28 by Simulated
Annealing with NMR-Derived Distance and Torsion-Angle Restraints. A total of 288
cross-peaks from the NOESY, tm ¼ 300 ms, spectrum of 26 in CDCl3 at 28 were
integrated, and the volumes converted into distance restraints using the two-spin
approximation (for details, see Exper. Part). The torsion angles C(3’)ꢀC(4’)ꢀC(5’)ꢀS
were restricted to 180 ꢃ 208 based on vicinal couplings between HꢀC(4’) and
HReꢀ(C5’). Because the downfield shift of HꢀN(3) clearly indicated the presence of
four base pairs, Watson – Crick base pairing was enforced by additional distance
restraints across the H-bonds. Starting with a loose and extended antiparallel duplex, a
total of 97 structures were calculated according to the torsion-angle-simulated
annealing molecular-dynamics protocol of Stein et al. with the programme XPLOR-
NIH [15]. A small subset of five NOEs that involved the terminal OH H-atoms (NOEs
between units A(I) and U(IV)) was consistently violated in all initial structure
calculations. Because of possible contributions of chemical exchange between the OH
H-atoms, these NOEs were omitted from the final calculations. It is conceivable that
these weak NOEs originate from end-to-end association of duplexes mediated through
H-bonds involving the terminal OH groups, but, in view of the sparse data available,
this question has to remain open.
Since qualitative analysis of the NMR data did not allow us to definitely decide
between Watson – Crick (WC) and reverse Watson – Crick (rWC) base pairing, all four
C2-symmetric patterns of base pairing with WC or rWC base pairs (WC-WC-WC-WC,
WC-rWC-rWC-WC, rWC-WC-WC-rWC, rWC-rWC-rWC-rWC) were calculated sepa-
rately. However, all three variations with rWC base pairs generated only structures with
severe violations of experimentally derived distance or dihedral angle restraints,
8
)
AM1, but not AMBER* modeling suggested also a conformation in which the CꢀO bond of the
OH is in the p-plane of the adenine moiety, and OꢀH is parallel to the C(8)¼N(7) bond. This
conformer agrees with J(H,OH) ¼ 11.5 Hz, but not with the NOESY cross-peaks between the signal
of HOCH2ꢀC(8/I), and of both HꢀC(1’/I) and HꢀC(2’/I) signals.

Helvetica Chimica Acta – Vol. 92 (2009)
2615
whereas, for the duplex with four WC base pairs, 65 out of the 97 calculated structures
did not show any violations of the distance or torsion-angle contraints. A bundle of all
accepted structures, aligned based on the S-atoms, is shown in Fig. 9,a, and the
structure 26C with the lowest calculated energy in the bundle is depicted in Fig. 9,b – d,
from different view angles.
Fig. 9. Structure of the Watson – Crick base-paired cyclic duplex 26C as calculated with the NMR-derived
contraints for 26 in CDCl3 at 28. a) Bundle of all 67 accepted structures (rmsd ¼ 0.057 ꢄ). b) Single
(lowest calculated energy) structure from the bundle in a view perpendicular to the C2 axis. c) View
showing the strong roll leading to a pronounced helix curvature. The purple spheres are the centres of
each base pair (midpoint of the line connecting the two exocyclic CH2ꢀC(6/8)) of opposite bases. d) Top
view of the helix showing the twists between the base pairs. The purple cylinders connect the two
exocyclic CH2ꢀC(6/8) of opposite bases.

2616
Helvetica Chimica Acta – Vol. 92 (2009)
With such a short duplex of only four base pairs, a standard geometrical analysis of
the helix parameters is impossible, because the helix axis is not well-defined. Hence, the
following description has to remain qualitative. In CDCl3, 26C forms an antiparallel
duplex with four Watson – Crick base pairs in a right-handed, strongly bent helix
fragment (Fig. 9,b). The estimated twist angles are 608 between the terminal and 458
between the centre base pairs (Fig. 9,d)9).
The most striking feature of 26C is the very pronounced roll of the base pairs, in
particular between the two centre ones (roll angle of 39.58)7), whereas the roll angle
between the border base pairs is ca. 228 (Fig. 9,d). This large and unidirectional roll
leads to a strong curvature of the helix axis. A polygon drawn through the centers of the
base pairs (midpoints between the two CH2ꢀC(8/6)) does not approximate a straight
line (as, e.g., in B-type DNA) but exhibits a curvature with a radius of ca. 10 ꢄ (see
Fig. 9,c). Although all neighbouring base pairs approach each other to Van der Waals
distance at one point, the large roll between the central base pairs prevents efficient
base stacking and opens a cleft in the overall shape of the duplex that is best seen from
the CPK model shown in Fig. 10. The strong curvature of the helix axis of 26C predicts
that duplex formation of the homologous self-complementary hexamers and octamers
is not feasible due to disturbing steric interactions of the terminal A units.
Fig. 10. CPK Representation of the Watson – Crick base-paired cyclic duplex 26C showing the
pronounced cleft in the overall shape
The structure of 26C was analyzed in more details (Table 3). It forms Watson –
Crick base pairs with N(3)H ··· N(1) and C(4)¼O ··· HNꢀC(6) distances of 1.92 and
2.15 ꢄ for the border base pairs, and 2.05 and 1.98 ꢄ, respectively, for the centre base
pairs. The structure adopted by 26C is similar to the one of the AMBER* modeled 26A,
possessing a syn-orientation of the nucleobases, a gt-orientation of the sulfanyl
substituents,
a
gꢀgꢀgꢀ (torsion angles q, i, and k) conformation of the
C(4’)ꢀC(5’)ꢀSꢀCH2ꢀC(6 or 8)ꢀN(1 or 9) fragment of the linkers between units I
9
)
In its geometrical definition, the twist as seen in this figure is not identical to the twist of a standard
helix (linear helix axis) as defined for DNA and RNA.

Helvetica Chimica Acta – Vol. 92 (2009)
2617
and II, and units II and III, and a tgþt-conformation of this linking fragment between
units III and IV. Thus, relatively small differences of the angles c, and h – k are mainly
responsible for the different puckering of 26A and 26C. In agreement with the
experimental J(1’,2’) and J(3’,4’) values, the furanose rings of units I – III adopt a
O
northern (3H2 or 3E) and those of unit IV an E conformation.
The diastereotopic HOCH2(5’/IV) and HOCH2ꢀC(8/I) cannot be assigned by the
SPARKY programme [20], due to overlapping HꢀC(4’/IV) and HaꢀC(5’/IV) signals,
and to poor NOESY interactions of HOCH2ꢀC(8/I). Hence, XPLOR-NIH calculation
cannot predict the conformation of the CH2OH groups established by the large
J(H,OH) values (see above), but leads to freely rotating CH2OH groups.
The broadening of all 1H-NMR signals of 26 in CDCl3 at 28 excludes the presence of
a single conformer, but may indicate an equilibrium of the easily interconvertible
duplexes 26A and 26C (but with intramolecular H-bonds of the OH groups) and/or
aggregation of the cyclic duplexes. Indeed, J(5’,OH) ¼ 9.7 Hz suggests a ca. 90%
persistent intramolecular H-bond and would allow the aggregation of cyclic duplexes
by an intermolecular H-bond of HOCH2C(5’/IV) to OCH2ꢀC(8/I). This could
rationalize the five weak ROESY cross-peaks between the units A(I) and U(IV) that
consistently violated all initial duplex structures of the XPLOR-NIH calculation.
2.2.4. No Association of the Polyol 27 in (D6)DMSO. The completely deprotected
polyol 27 proved insoluble in CDCl3 and CDCl3/(D6)DMSO 9 :1. In pure (D6)DMSO,
HꢀN(3) and H2NꢀC(6) resonate at 11.34/11.31 and 7.29 ppm, respectively. These
values are characteristic for U and A monomers in (D6)DMSO [4 – 6] and evidence a
completely solvated monoplex of 27 in (D6)DMSO.
Conclusions. – The analysis of the sequence-isomeric self-complementary tetramers
confirms the results obtained from the investigation of the corresponding dimers. In the
U*[s]A*[s]U*[s]A* and U*[s]A* series, the diol forms exclusively a Watson – Crick
base-paired cyclic duplex, with the tetramer showing the structure of an incipient A-
helix, but possessing a bent helix axis, and, in the A*[s]U*[s]A*[s]U* and A*[s]U*
series, the diols form a reverse Hoogsteen base-paired cyclic duplex. The OH groups at
the terminal units strongly enhance the formation of cyclic duplexes.
We thank the Swiss National Science Foundation, F. Hoffmann-La Roche AG, Basel, and Syngenta
AG, Basel, for generous support. B. J. and Z. J. thank the ETH Zꢀrich for financial support (TH Project
TH-17/02).
Experimental Part
General. See [3].
6-(Chloromethyl)-5’-O-[dimethyl(1,1,2-trimethylpropyl)silyl]-2’,3’-O-isopropylideneuridine (7). Un-
der N2, a soln. of 6 [3] (700 mg, 1.53 mmol) in CH2Cl2 (5 ml) at 08 was treated with DMAP (460 mg,
3.83 mmol) and MsCl (238 ml, 3.06 mmol), stirred for 1 h at 0 and for 2 h at 238, diluted with CH2Cl2
(50 ml), and washed with sat. NH4Cl soln., sat. NaHCO3 soln., and brine. The combined org. phases were
dried (MgSO4) and evaporated. FC (AcOEt/cyclohexane 2 :3) gave 7 (540 mg, 74%). Rf (AcOEt/
cyclohexane 1:1) 0.50. [a]2D5 ¼ ꢀ5.9 (c ¼ 1.0, CHCl3). IR (CHCl3): 3387w, 3189w (br.), 2961m, 2869w,
1
1700s, 1626w, 1461m, 1447m, 1379m, 1268w, 1256w, 1157w, 1131w, 1084m, 980w, 876m, 835m. H-NMR
(300 MHz, CDCl3): see Table 4; additionally, 1.60 (sept., J ¼ 6.9, Me2CH); 1.55, 1.31 (2s, Me2CO2); 0.86
(d, J ¼ 6.9, Me2CH); 0.83 (s, Me2CSi); ꢀ 0.07, ꢀ 0.06 (2s, Me2Si). 13C-NMR (75 MHz, CDCl3): see
Table 5; additionally, 113.55 (s, Me2CO2); 34.08 (d, Me2CH); 27.24, 25.34 (2q, Me2CO2); 25.26 (s,

2618
Helvetica Chimica Acta – Vol. 92 (2009)
Table 4. Selected 1H-NMR Chemical Shifts [ppm] and Coupling Constants [Hz] of the U* and A*
Monomers 7, 9, 10, 12, 14, and 15 in CDCl3
7
9
10
12
8.81
14
8.63
15
8.74
HꢀC(2)
–
–
–
HꢀN(3) or HNꢀC(6)
HꢀC(5)
9.16
5.82
4.44
4.33
5.82
5.26
4.86 – 4.77
4.20 – 4.10
3.84 – 3.71
3.84 – 3.71
9.35
5.73
4.55
4.55
5.81
5.25
4.86
4.14
3.23
3.23
10.45
5.76
4.41
4.32
5.83
5.25
4.86
4.12
3.21
3.21
12.9
< 1.5
6.3
8.94
–
9.53
–
4.98 – 4.85
4.98 – 4.85
6.24
5.63
4.98
4.19
3.10
9.16
–
CHaꢀC(6 or 8)
CHbꢀC(6 or 8)
HꢀC(1’)
4.92
4.86
6.32
5.83
5.13
4.30
3.75
3.62
15.3
2.4
6.3
3.6
6.0
4.85
4.85
6.24
5.74
5.06
4.27
3.16
3.04
HꢀC(2’)
HꢀC(3’)
HꢀC(4’)
HaꢀC(5’)
HbꢀC(5’)
J(CHa,CHb)
J(1’,2’)
2.97
a
a
a
13.2
< 1.5
6.0
)
)
)
1.5
6.3
3.9
7.2
7.2
2.1
6.3
3.6
7.5
6.9
13.8
1.5
6.3
3.3
7.2
6.9
13.8
J(2’,3’)
J(3’,4’)
J(4’,5’a)
J(4’,5’b)
a
)
)
)
)
3.6
7.2
a
a
7.2
5.7
11.9
a
a
a
J(5’a,5’b)
)
)
a) Not assigned.
Table 5. Selected 13C-NMR Chemical Shifts [ppm] of the U* and A* Monomers 7, 9, 10, 12, 14, and 15
7
9
10
12
14
15
Solvent
CDCl3
(D6)DMSO
CDCl3
CDCl3
CDCl3
CDCl3
C(2)
C(4)
C(5)
C(6)
150.57a)
162.94
104.62
150.42a)
–
41.09
91.67
84.20
81.91
89.62
63.83
151.57
163.48
101.67
155.59
–
60.03
91.33
85.16
84.43
87.97
31.69
150.95a)
163.31
105.11
150.76a)
–
41.29
92.26
85.23
84.48
88.69
31.58
152.58
149.77
122.26
152.52
150.06
36.73
90.42
83.24
81.59
152.46
149.19
121.70
152.26
154.81
57.55
89.92
84.04
83.86
86.61
31.26
153.24
150.05
122.49
152.39
150.05
36.64
90.43
84.09
83.95
86.97
31.30
C(8)
CH2ꢀC(6 or 8)
C(1’)
C(2’)
C(3’)
C(4’)
C(5’)
87.65
62.83
a) Assignment may be interchanged.
Me2CSi); 20.38, 20.22 (2q, Me2CH); 18.54, 18.49 (2q, Me2CSi); 3.23 (q, Me2Si). HR-ESI-MS: 499.1823
(34), 498.1888 (24), 497.1839 (100, [M þ Na]þ, C21H35ClN2NaO6Siþ; calc. 497.1851).
5’-S-Acetyl-6-(hydroxymethyl)-2’,3’-O-isopropylidene-5-thiouridine (9). A soln. of 8 [3] (150 mg,
0.23 mmol) in CH2Cl2 (3 ml) under N2 was treated with Cl2CHCO2H (0.3 ml, 3.63 mmol) and Et3SiH
(75 ml, 0.47 mmol), and stirred for 15 min at 238. The mixture was poured into sat. NaHCO3 soln. After
extraction with CH2Cl2, the combined org. phases were washed with brine, dried (MgSO4), and
evaporated. The residue was triturated with cyclohexane. Filtration gave 9 (63 mg, 73%). Colourless
powder. Rf (AcOEt) 0.55. M.p. 202.6 – 203.78. [a]2D5 ¼ þ10.7 (c ¼ 1.0, CHCl3). IR (ATR): 3355w, 3195w,

Helvetica Chimica Acta – Vol. 92 (2009)
2619
3001w, 2987w, 2934w, 1686s, 1664s, 1617w, 1463w, 1421w, 1395m, 1385m, 1308w, 1275w, 1254w, 1238w,
1211w, 1157w, 1138w, 1101s, 1064w, 1042s, 1029m, 1003m, 983w, 972w, 952w, 865m, 855s, 831m, 805m.
1H-NMR (300 MHz, CDCl3): see Table 4; additionally, 3.12 (t, J ¼ 6.0, OH); 2.34 (s, AcS); 1.52, 1.33 (2s,
Me2C). 13C-NMR (75 MHz, (D6)DMSO): see Table 5; additionally, 195.36 (s, SC¼O); 113.34 (s, Me2C);
31.17 (q, MeC¼O); 27.61, 25.61 (2q, Me2C). HR-MALDI-MS: 396.0923 (16), 395.0886 (100, [M þ Na]þ,
C15H20N2NaO7Sþ; calc. 395.0883), 315.0641 (17), 273.1280 (15, MMTrþ, C20H17Oþ; calc. 273.1279). Anal.
calc. for C15H20N2O7S (372.40): C 48.38, H 5.41, N 7.52; found: C 48.42, H 5.56, N 7.41.
5’-S-Acetyl-6-(chloromethyl)-2’,3’-O-isopropylidene-5’-thiouridine (10). Under N2, a soln. of
9
(44 mg, 0.12 mmol) in CH2Cl2 (1 ml) at 08 was treated with DMAP (36 mg, 0.29 mmol) and MsCl (20 ml,
0.24 mmol), stirred for 10 min at 0 and for 2 h at 238, and evaporated. A soln. of the residue in AcOEt was
washed with sat. NH4Cl soln. The combined org. phases were washed with brine, dried (MgSO4), and
evaporated. FC (AcOEt/cyclohexane 1:1) gave 10 (42 mg, 91%). Colourless foam. Rf (AcOEt/
cyclohexane 4 :1) 0.57. [a]2D5 ¼ þ0.5 (c ¼ 1.0, CHCl3). IR (CHCl3): 3386w, 3189w (br.), 3027w, 3014w,
2939w, 1698s (br.), 1627w, 1460w, 1447w, 1383m, 1272w, 1229w (br.), 1157w, 1136w, 1092m, 1063m, 982w,
1
909w, 875w, 836w. H-NMR (300 MHz, CDCl3): see Table 4; additionally, 2.30 (s, AcS); 1.49, 1.30 (2s,
Me2C). 13C-NMR (75 MHz, CDCl3): see Table 5; additionally, 195.05 (s, SC¼O); 114.09 (s, Me2CO2);
30.80 (q, MeC¼O); 27.26, 25.34 (2q, Me2C). HR-MALDI-MS: 415.0511 (12), 413.0539 (38, [M þ Na]þ,
C15H19ClN2NaO6Sþ; calc. 413.0545), 379.0932 (31, [M ꢀ Cl þ H þ Na]þ, C15H20N2NaO6Sþ; calc.
379.0940), 333.0303 (17), 274.0426 (15), 273.1280 (100, MMTrþ, C20H17Oþ; calc. 273.1279).
N6-Benzoyl-8-(chloromethyl)-5’-O-[dimethyl(1,1,2-trimethylpropyl)silyl]-2’,3’-O-isopropylideneade-
nosine (12). Under N2, a soln. of 11 [3] (1.000 g, 1.76 mmol) in CH2Cl2 (5 ml) was cooled to ꢀ 108,
treated with DMAP (640 mg, 4.57 mmol) and MsCl (330 ml, 4.2 mmol), stirred for 10 min at 0 and for 3 h
at 238, diluted with CH2Cl2 (50 ml), and washed with sat. NH4Cl soln., sat. NaHCO3 soln., and brine. The
combined org. phases were dried (MgSO4) and evaporated. FC (AcOEt/cyclohexane 1:10 ! 1:1) gave
12 (0.69 g, 61%). Rf (AcOEt/cyclohexane 1:1) 0.47. [a]2D5 ¼ ꢀ18.3 (c ¼ 1.0, CHCl3). IR (CHCl3): 3408w,
1
3014m, 1709m, 1613s, 1589m, 1463m, 1360m, 1268m, 1089s. H-NMR (300 MHz, CDCl3): see Table 4;
additionally, 8.00 (d, J ¼ 7.8, 2 arom. H); 7.62 – 7.49 (m, 3 arom. H); 1.64, 1.42 (2s, Me2CO2); 1.57 (sept.,
J ¼ 6.9, Me2CH); 0.84 (d, J ¼ 6.9, Me2CH); 0.79, 0.78 (2s, Me2CSi); ꢀ 0.004, ꢀ 0.014 (2s, Me2Si).
13C-NMR (75 MHz, CDCl3): see Table 5; additionally, 164.94 (s, C¼O); 133.57 (s); 133.15 (d); 129.06
(2d); 128.22 (2d); 114.66 (s, Me2CO2); 34.25 (d, Me2CH); 27.43, 25.63 (2q, Me2CO2); 25.47 (s, Me2CSi);
20.46 (q, Me2CH); 18.65 (q, Me2CSi); ꢀ 3.29 (q, Me2Si). HR-MALDI-MS: 624.2381 (50, [M þ Na]þ,
C29H40ClN5NaO5Siþ; calc. 624.2385), 590.2764 (100, [M ꢀ Cl þ H þ Na]þ, C29H41N5NaO5Siþ; calc.
590.2775).
5’-S-Acetyl-N6-benzoyl-8-(hydroxymethyl)-2’,3’-O-isopropylidene-5’-thioadenosine (14). Under N2,
a soln. of 13 [3] (100 mg, 0.13 mmol) in CH2Cl2 (2 ml) was treated with Cl2CHCO2H (0.2 ml, 2.4 mmol)
and Et3SiH (50 ml, 0.31 mmol) and stirred for 15 min at 238. The mixture was poured into sat. NaHCO3
soln. and extracted with CH2Cl2. The combined org. phases were washed with brine, dried (MgSO4), and
evaporated. FC (CH2Cl2/MeOH 100 :0 ! 95 :5) gave 14 (62 mg, 96%). Foam. Rf (AcOEt) 0.28. [a]D25
¼
ꢀ13.2 (c ¼ 1.0, CHCl3). IR (CHCl3): 3405w, 3030w, 2999m, 2940w, 1703s (br.), 1614s, 1590m, 1527w,
1
1480s, 1460m, 1427m, 1376m, 1356m, 1266m (br.), 1158m, 1134m, 1093s. H-NMR (300 MHz, CDCl3):
see Table 4; additionally, 7.95 (d, J ¼ 7.5, 2 arom. H); 7.48 (br. t, J ¼ 7.5, 1 arom. H); 7.38 (d, J ¼ 7.8, 2 arom.
H); 6.05 (br. s, OH); 2.26 (s, AcS); 1.26, 1.37 (2s, Me2C). 13C-NMR (75 MHz, CDCl3): see Table 5;
additionally, 194.64 (s, SC¼O); 165.42 (s, NC¼O); 133.57 (s); 132.89 (d); 128.79 (2d); 128.10 (2d); 114.67
(s, Me2C); 30.73 (q, MeC¼O); 27.24, 25.51 (2q, Me2C). HR-MALDI-MS: 523.1439 (27), 522.1412 (100,
[M þ Na]þ, C23H25N5NaO6Sþ; calc. 522.1418), 500.1589 (16, [M þ H]þ, C23H26N5O6Sþ; calc. 500.1604).
5’-S-Acetyl-N6-benzoyl-8-(chloromethyl)-2’,3’-O-isopropylidene-5’-thioadenosine (15). Under N2, a
soln. of 14 (62 mg, 0.12 mmol) in CH2Cl2 (1 ml) was cooled to 08, treated with DMAP (37 mg,
0.31 mmol) and MsCl (20 ml, 0.24 mmol), and stirred for 1 h at 0 and for 3 h at 238. The mixture was
poured into sat. NH4Cl soln. and extracted with CH2Cl2. The combined org. phases were washed with
brine, dried (MgSO4), and evaporated. FC (CH2Cl2/MeOH 100 :0 ! 95 :5) gave 15 (62 mg, 96%).
Colourless foam. Rf (AcOEt/cyclohexane 4 :1) 0.46. [a]2D5 ¼ þ1.1 (c ¼ 1.0, CHCl3). IR (CHCl3): 3406w,
3014m, 2936w, 1707s (br.), 1613s, 1589m, 1525w, 1478m, 1463m, 1436m, 1424m, 1358m, 1329m, 1268m,
1249m, 1157w, 1134m, 1093s, 909w, 865w. 1H-NMR (300 MHz, CDCl3): see Table 4; additionally, 7.95 (d,

2620
Helvetica Chimica Acta – Vol. 92 (2009)
J ¼ 7.8, 2 arom. H); 7.54 (br. t, J ¼ 7.5, 1 arom. H); 7.45 (d, J ¼ 7.8, 2 arom. H); 2.27 (s, AcS); 1.56, 1.36 (2s,
Me2C). 13C-NMR (75 MHz, CDCl3): see Table 5; additionally, 194.58 (s, SC¼O); 164.91 (s, NC¼O);
133.67 (s); 133.07 (d); 129.05 (2d); 128.13 (2d); 114.70 (s, Me2C); 30.74 (q, MeC¼O); 27.31, 25.54 (2q,
Me2C). HR-MALDI-MS: 540.1073 (23, [M þ Na]þ, C23H24ClN5NaO5Sþ; calc. 540.1079), 506.1481 (100,
[M ꢀ Cl þ H þ Na]þ, C23H25N5NaO5Sþ; calc. 506.1469).
5’-S-Acetyl-N6-benzoyl-2’,3’-O-isopropylidene-5’-thioadenosine-8-methyl-(81 ! 5’-S)-2’,3’-O-isopro-
pylidene-6-{[(4-methoxyphenyl)diphenylmethoxy]methyl}-5’-thiouridine (16). Under N2, a soln. of 8
(600 mg, 0.93 mmol) in O2-free MeOH (2 ml) was treated with K2CO3 (400 mg, 2.79 mmol) and stirred
for 10 min at 238. Sat. NH4Cl soln. (20 ml) was added, and the mixture was extracted with CH2Cl2. The
combined org. phases were washed with brine, dried (MgSO4), and evaporated. Under N2, a soln. of the
residue (500 mg, 0.83 mmol) in O2-free DMF (2 ml) was treated with 15 (430 mg, 0.83 mmol) and K2CO3
(340 mg, 2.5 mmol), and stirred for 2 h at 238. The mixture was poured into sat. NH4Cl soln. and
extracted with CH2Cl2. The combined org. phases were washed with brine, dried (MgSO4), and
evaporated. FC (AcOEt/cyclohexane 3 :2 ! 1:0) yielded 16 (606 mg, 67%). Rf (AcOEt/cyclohexane
7:3) 0.25. [a]D25 ¼ ꢀ19.7 (c ¼ 1.0, CHCl3). IR (CHCl3): 3395w (br.), 3200w (br.), 3015m, 2937w, 1697s
(br.), 1612m, 1589w, 1510w, 1457w, 1423w, 1384m, 1356w, 1326w, 1299w, 1259m (br.), 1092m, 1062m.
1H-NMR (300 MHz, CDCl3): see Table 6; additionally, 10.81 (br. s, HꢀN(3/I)); 9.56 (s, HNꢀC(6/II));
7.92 (d, J ¼ 8.4, 2 arom. H); 7.49 – 7.24 (m, 15 arom. H); 6.79 (d, J ¼ 9.0, 2 arom. H); 3.81 (s, MeO); 2.29 (s,
AcS); 1.57, 1.36, 1.32, 1.16 (4s, 2 Me2C). 13C-NMR (75 MHz, CDCl3): see Table 7; additionally, 194.66 (s,
SC¼O); 165.84 (s, NHC¼O); 159.20, 143.36, 143.26 (3s); 134.38 (br. s, 2 C); 132.91 (d); 130.42 (2d);
128.78 – 128.33 (several d); 127.70 (2d); 114.58, 113.80 (2s, 2 Me2C); 113.64 (2d); 88.33 (s, Ph2C); 55.54 (q,
MeO); 30.74 (q, MeC¼O); 27.33, 27.21, 25.56, 25.38 (4q, 2 Me2C). HR-MALDI-MS: 1108.3442 (15),
1107.3444 (29), 1106.3413 (43, [M þ Na]þ, C56H57N7NaO12Sþ2 ; calc. 1106.3404), 273.1280 (100, MMTrþ,
C20H17Oþ; calc. 273.1279).
5’-S-Acetyl-2’,3’-O-isopropylidene-5’-thiouridine-6-methyl-(61 ! 5’-S)-N6-benzoyl-2’,3’-O-isopropyl-
idene-5’-thioadenosine-8-methyl-(81 ! 5’-S)-2’,3’-O-isopropylidene-6-{[(4-methoxyphenyl)diphenylme-
thoxy]methyl}-5’-thiouridine (17). Under N2, a soln. of 16 (300 mg, 0.28 mmol) in O2-free MeOH (2 ml)
was treated with K2CO3 (115 mg, 0.83 mmol), and stirred for 10 min at 238. After dilution with sat. NH4Cl
soln. (20 ml) under N2, the mixture was extracted with CH2Cl2. The combined org. phases were washed
with brine, dried (MgSO4), and evaporated. Under N2, a soln. of the residue (270 mg, 0.26 mmol) in O2-
free DMF (2 ml) was treated with 10 (100 mg, 0.26 mmol) and K2CO3 (106 mg, 0.77 mmol), and stirred
for 2 h at 238. The mixture was poured into a sat. NH4Cl soln. and extracted with CH2Cl2. The combined
org. phases were washed with brine, dried (MgSO4), and evaporated. FC (CH2Cl2/acetone 1:0 ! 1:1)
gave 17 (150 mg, 42%). Rf (AcOEt/cyclohexane 4 :1) 0.20. [a]2D5 ¼ ꢀ39.1 (c ¼ 1.0, CHCl3). IR (CHCl3):
3390w (br.), 3197w (br.), 3015m, 2936w, 1696s (br.), 1612m, 1588w, 1510w, 1457m, 1422w, 1384m, 1354w,
1326w, 1267w, 1254w (br.), 1157w, 1091m, 1070m, 1035w. 1H-NMR (300 MHz, CDCl3): see Table 6;
additionally, 10.62, 10.39 (2 br. s, HꢀN(3/I), HꢀN(3/III)); 10.01 (s, HNꢀC(6/II); 8.04 (dd, J ¼ 6.6, 3.3, 2
arom. H); 7.47 – 7.24 (m, 15 arom. H); 6.81 (d, J ¼ 8.7, 2 arom. H); 3.79 (s, MeO); 2.29 (s, AcS); 1.61, 1.51,
1.38, 1.32 (2 Me), 1.15 (5s, 3 Me2C). 13C-NMR (75 MHz, CDCl3): see Table 7; additionally, 194.55 (s,
SC¼O); 165.45 (s, NHC¼O); 158.81, 142.97 (2 C), 134.05, 133.28 (4s); 132.52 (d); 130.15 (2d); 128.49 –
128.04 (several d); 127.38 (2d); 114.48, 113.68, 113.55 (3s, 3 Me2C); 113.36 (2d); 88.10 (s, Ph2C); 55.36 (q,
MeO); 30.68 (q, MeC¼O); 27.27 (2 C), 27.16, 25.43, 25.38 (2 C) (4q, 3 Me2C). HR-MALDI-MS:
1421.423 (16), 1420.419 (54), 1419.417 (97), 1418.420 (100, [M þ Na]þ, C69H73NaN9O17S3þ ; calc.
1418.418).
N6-Benzoyl-5’-O-[dimethyl(1,1,2-trimethylpropyl)silyl]-2’,3’-O-isopropylideneadenosine-8-methyl-
(81 ! 5’-S)-2’,3’-O-isopropylidene-5’-thiouridine-6-methyl-(61 ! 5’-S)-N6-benzoyl-2’,3’-O-isopropylidene-
5’-thioadenosine-8-methyl-(81 ! 5’-S)-2’,3’-O-isopropylidene-6-{[(4-methoxyphenyl)diphenylmethoxy]-
methyl}-5’-thiouridine (18). Under N2, a soln. of 17 (150 mg, 0.11 mmol) in O2-free MeOH (2 ml) was
treated with K2CO3 (60 mg, 0.43 mmol) and stirred for 10 min at 238. After dilution with sat. NH4Cl soln.
(20 ml) under N2, the mixture was extracted with CH2Cl2. The combined org. phases were washed with
brine, dried (MgSO4), and evaporated. Under N2, a soln. of the residue (145 mg) in O2-free DMF
(2.5 ml) was treated with 12 (70 mg, 0.11 mmol) and K2CO3 (60 mg, 0.43 mmol), and stirred for 2 h at 238.
The mixture was poured into a sat. NH4Cl soln., and extracted with CH2Cl2. The combined org. phases

Helvetica Chimica Acta – Vol. 92 (2009)
2621
Table 6. Selected 1H-NMR Chemical Shifts [ppm] and Coupling Constants [Hz] of the A*[s]U* Dimer 16,
the U*[s]A*[s]U* Trimer 17, and the A*[s]U*[s]A*[s]U* Tetramers 18 – 21
16a)
17b)
18c)
19d)
20e)
21f)
Solvent
CDCl3
CDCl3
CDCl3
CDCl3
(D6)DMSO (D6)DMSO
Adenosine unit IV
HꢀC(2/IV)
CH2ꢀC(8/IV)
HꢀC(1’/IV)
HꢀC(2’/IV)
HꢀC(3’/IV)
HꢀC(4’/IV)
2 HꢀC(5’/IV)
J(Ha,Hb/IV)
J(1’,2’/IV)
8.70g)
4.18 – 3.95
6.32
8.50g)
8.12g)
4.13/4.09h)
6.22
8.11g)
4.16 – 3.97
6.18
5.61
5.03
4.13/4.00h)
6.37
5.86
5.81
5.11
4.32
3.58/3.47
14.7
1.8
6.3
4.2
5.68
5.15 – 5.06
4.42 – 4.36
3.67/3.57
5.05i)
4.27
3.62 – 3.45
14.5
2.1
6.3
3.0
7.0/7.0
4.26
3.40 – 3.25
m
m
)
)
2.1
6.0
2.4
6.3
3.0
J(2’,3’/IV)
J(3’,4’/IV)
J(4’,5’/IV)
m
)
m
6.0/6.3
7.2/6.3
)
Uridine unit III
HꢀC(5/III)
CH2ꢀC(6/III)
HꢀC(1’/III)
HꢀC(2’/III)
HꢀC(3’/III)
HꢀC(4’/III)
2 HꢀC(5’/III)
J(Ha,Hb/III)
J(1’,2’/III)
5.09
3.58/3.37
5.75
5.22
4.82
3.96
3.21/3.16
15.0
5.85 – 5.70
3.52/3.36
5.72i)
3.46/3.38
5.70
5.18
4.85k)
4.12 – 3.90
3.03 – 2.72
14.7
< 1.5
6.3
5.69
3.62 – 3.45
5.74k)
5.68
3.55 – 3.40
5.73h)
5.16
4.76i)
5.85 – 5.70
5.22 – 5.15
5.02 – 4.94
4.28 – 4.21
2.92 – 2.72
5.17
4.745l)
4.14 – 4.05
2.85 – 2.81
4.06
2.90 – 2.67
m
m
14.5
)
< 1.5
6.2
)
m
m
m
m
< 1.5
6.3
3.9
)
)
)
)
< 1.5
6.3
J(2’,3’/III)
J(3’,4’/III)
J(4’,5’/III)
3.3
3.7
< 1.5
m
m
m
6.6/7.2
)
)
)
Adenosine unit II
HꢀC(2/II)
CH2ꢀC(8/II)
HꢀC(1’/II)
HꢀC(2’/II)
HꢀC(3’/II)
HꢀC(4’/II)
2 HꢀC(5’/II)
J(Ha,Hb/II)
J (1’,2’/II)
8.75
3.67/3.49
6.16
5.82
5.06
4.28
3.19/3.07
12.3
2.1
6.3
8.74
3.90 – 3.65
6.22
5.62
5.21
8.71g)
4.18 – 3.95
6.22
8.33g)
8.15g)
8.14g)
4.16 – 3.97
6.21
5.67
5.03
4.11/4.00h)
6.32
4.07/4.01h)
6.25
5.85 – 5.70
5.15 – 5.06
4.28 – 4.21
2.92 – 2.72
5.96
5.11
4.21
3.03 – 2.72
14.7
1.5
6.3
3.0
5.78
5.03i)
4.41
2.84 – 2.72
4.14 – 4.05
2.85 – 2.81
14.5
4.12
2.90 – 2.67
m
m
m
)
< 1.5
6.0
)
)
< 1.5
1.7
6.2
3.3
1.8
6.3
3.0
m
J (2’,3’/II)
J (3’,4’/II)
J (4’,5’/II)
)
)
)
m
3.0
7.5/6.9
3.9
6.3/6.3
m
m
m
6.6/6.6
)
)
Uridine unit I
HꢀC(5/I)
5.39
4.06/4.00
5.39
4.81
4.68
5.44
4.16 – 4.02
5.53
4.91
4.53
5.85 – 5.70
5.35i)
4.12 – 3.90
5.70
5.18
4.82k)
4.12 – 3.90
5.63
5.62
CH2ꢀC(6/I)
HꢀC(1’/I)
HꢀC(2’/I)
HꢀC(3’/I)
HꢀC(4’/I)
4.18 – 3.95
5.85 – 5.70
5.02 – 4.94
4.68 – 4.60
4.18 – 4.11
4.36/4.32
5.73k)
5.17
4.740l)
4.14 – 4.05
4.38 – 4.28
5.71h)
5.16
4.72i)
4.06
3.99
4.14 – 4.04

2622
Helvetica Chimica Acta – Vol. 92 (2009)
Table 6 (cont.)
16a)
17b)
18c)
19d)
20e)
21f)
Solvent
CDCl3
CDCl3
CDCl3
CDCl3
(D6)DMSO (D6)DMSO
2 HꢀC(5’/I)
J(Ha,Hb/I)
J(1’,2’/I)
J(2’,3’/I)
J(3’,4’/I)
2.82 – 2.65
14.7
1.8
6.3
3.9
2.84 – 2.72
)
2.92 – 2.72
3.03 – 2.72
)
< 1.5
6.3
2.87/2.74 2.90 – 2.67
14.8
m
m
m
m
)
)
)
)
)
m
< 2
6.0
4.2
< 1.5
6.2
< 1.5
6.3
m
m
3.3
3.7
7.7/6.4
< 1.5
)
m
m
m
m
J(4’,5’/I)
8.1/3.9
)
)
)
a) J(5’a,5’b/II) ¼ 13.8 Hz. b) Broad signals for HꢀC(5/III), CH2ꢀC(8/II), and all H-atoms of unit I.
J(5’a,5’b/III) ¼ 13.5 Hz. c) Broad signals. J(5’a,5’b/IV) ¼ 12.6 Hz. d). Broad signals for CH2ꢀC(8/II).
J(5’a,5’b/IV) ¼ 10.5 Hz. e) J(5’a,5’b/I) ¼ 13.7, J(Ha,OH/I) ¼ 6.0, J(Hb,OH/I) ¼ 5.1 Hz. f) Assignments
based on a HSQC and a HMBC spectrum. g) – l) Assignments may be interchanged. m) Not assigned.
were washed with brine, dried (MgSO4), and evaporated. FC (CH2Cl2/MeOH/NH4OH 98 :2 :1 !
90 :10 :1) gave 18 (90 mg, 44%). Rf (CH2Cl2/MeOH 95 :5) 0.38. [a]2D5 ¼ ꢀ44.9 (c ¼ 1.0, CHCl3). IR
(CHCl3): 3396w (br.), 3204w (br.), 3015m, 2959w, 2933w, 2869w, 1698s (br.), 1612s, 1589m, 1511w,
1
1458m, 1423m, 1384m, 1357m, 1330w, 1263m, 1243m, 1157w, 1089s (br.). H-NMR (300 MHz, CDCl3):
see Table 6; additionally, 10.2, 9.73 (2 br. s, HꢀN(3/I), HꢀN(3III)); 9.55 (s, HNꢀC(6/II), HNꢀC(6/
IV)); 8.02 – 7.92 (m, 4 arom. H); 7.53 – 7.24 (m, 18 arom. H); 6.82 (d, J ¼ 9.0, 2 arom. H); 3.79 (s, MeO);
1.60 (2 Me), 1.46, 1.39 (2 Me), 1.33, 1.27, 1.16 (6s, 4 Me2C); 1.50 – 1.40 (m, Me2CH); 0.82 (d, J ¼ 6.9,
Me2CH); 0.78, 0.77 (2s, Me2CSi); ꢀ 0.03, ꢀ 0.04 (2s, Me2Si). 13C-NMR (75 MHz, CDCl3): see Table 7;
additionally, 166.10, 165.86 (2s, 2 NHC¼O); 159.24, 143.38, 143.30 (3s); 134.35 (2s); 134.30 (s); 132.84,
132.72 (2d); 130.55 (2d); 128.86 – 128.34 (several d); 127.67 (2d); 114.79, 114.32, 114.06, 113.92 (4s,
4 Me2C); 113.65 (2d); 88.45 (s, Ph2C); 55.50 (q, MeO); 34.28 (d, Me2CH); 27.41 (br. q, 4 C), 25.64 (q),
25.53 (br. q, 3 C) (4 Me2C); 25.53 (s, Me2CSi); 20.49 (q, Me2CH); 18.67 (q, Me2CSi); ꢀ 3.25 (q, Me2Si).
HR-MALDI-MS: 1943.687 (68), 1942.684 (100), 1941.682 (80, [M þ Na]þ, C96H110N14NaO21S3Siþ; calc.
1941.683).
5’-O-[Dimethyl(1,1,2-trimethylpropyl)silyl]-2’,3’-O-isopropylideneadenosine-8-methyl-(81 ! 5’-S)-
2’,3’-O-isopropylidene-5’-thiouridine-6-methyl-(61 ! 5’-S)-2’,3’-O-isopropylidene-5’-thioadenosine-8-
methyl-(81 ! 5’-S)-2’,3’-O-isopropylidene-6-{[(4-methoxyphenyl)diphenylmethoxy]methyl}-5’-thiouri-
dine (19). A soln. of 18 (90 mg, 47 mmol) in MeOH (2 ml) was treated with MeONa (25 mg, 0.47 mmol)
and stirred for 12 h at 238. The mixture was poured into sat. NH4Cl soln. and extracted with CH2Cl2. The
combined org. phases were washed with brine, dried (MgSO4), and evaporated. FC (AcOEt/acetone/
NH4OH 100 :0 :1 ! 50 :50 :1) gave 19 (52 mg, 65%). Rf (CH2Cl2/MeOH 92 :5) 0.55. [a]2D5 ¼ ꢀ97.9 (c ¼
1.0, CHCl3). IR (CHCl3): 3314w, 3200w, 1698s, 1641m, 1510w, 1459w, 1441w, 1375m, 1330w, 1299w,
1208w, 1157m, 1092s. 1H-NMR (300 MHz, CDCl3): see Table 7; additionally, 13.03, 12.38 (2 br. s, HꢀN(3/
I), HꢀN(3/III)); 7.66 (br. s, NH2); 7.48 – 7.28 (m, 12 arom. H, NH2); 6.85 (d, J ¼ 8.7, 2 arom. H); 3.80 (s,
MeO); 1.62, 1.59, 1.47, 1.42, 1.40, 1.39, 1.28, 1.23 (8s, 4 Me2C); 1.53 (sept., J ¼ 6.9, Me2CH); 0.81 (d, J ¼ 6.9,
Me2CH); 0.76, 0.75 (2s, Me2CSi); ꢀ 0.03, ꢀ 0.06 (2s, Me2Si). 13C-NMR (75 MHz, CDCl3): see Table 7;
additionally, 158.86, 143.18, 142.99, 134.05 (4s); 130.33 (2d); 128.19 (4d); 128.04 (4d); 127.35 (2d); 113.55,
113.47 (br., 3 C) (2s, 4 Me2C); 113.36 (2d); 88.27 (s, Ph2C); 55.33 (q, MeO); 34.27 (d, Me2CH); 27.34,
27.28 (2 C), 27.18, 25.72, 25.59, 25.28 (2 C) (6q, 4 Me2C); 25.28 (s, Me2CSi); 20.40 (q, Me2CH); 18.58 (q,
Me2CSi); ꢀ 3.29 (q, Me2Si). HR-MALDI-MS: 1735.634 (62), 1734.631 (96), 1733.629 (100, [M þ Na]þ;
C82H102N14NaO19S3Siþ: calc. 1733.628).
5’-O-[Dimethyl(1,1,2-trimethylpropyl)silyl]-2’,3’-O-isopropylideneadenosine-8-methyl-(81 ! 5’-S)-
2’,3’-O-isopropylidene-5’-thiouridine-6-methyl-(61 ! 5’-S)-2’,3’-O-isopropylidene-5’-thioadenosine-8-
methyl-(81 ! 5’-S)-6-(hydroxymethyl)-2’,3’-O-isopropylidene-5’-thiouridine (20). Under N2, a soln. of 19
(30 mg, 18 mmol) in CH2Cl2 (1 ml) was treated with Cl2CHCO2H (50 ml, 0.6 mmol) and Et3SiH (10 ml ,

Helvetica Chimica Acta – Vol. 92 (2009)
2623
Table 7. Selected 13C-NMR Chemical Shifts [ppm] of the A*[s]U* Dimer 16, the U*[s]A*[s]U* Trimer 17,
and the A*[s]U*[s]A*[s]U* Tetramers 18 – 21
16
17
18
19
20
21a)
Solvent
CDCl3
CDCl3
CDCl3
CDCl3
(D6)DMSO
(D6)DMSO
Adenosine unit IV
C(2/IV)
C(4/IV)
C(5/IV)
C(6/IV)
152.41
149.89
122.14
152.65
152.18b)
27.90
87.83
83.27
81.87
85.07
153.12b)
150.46c)
118.26d)
155.75e)
149.46c)
29.81
152.25
149.66b)
117.78
155.51c)
148.08
28.89
152.25
149.64b)
117.78
155.50
147.87
27.04
C(8/IV)
CH2ꢀC(8/IV)
C(1’/IV)
89.02f)
83.17
88.89
89.09
C(2’/IV)
C(3’/IV)
C(4’/IV)
82.45d)
81.47
82.22c)
81.17
82.98g)
88.02
85.57
85.53
C(5’/IV)
62.91
62.88
62.58
61.30
Uridine unit III
C(2/III)
C(4/III)
C(5/III)
C(6/III)
151.71b)
162.62
152.06b)
162.24
104.12c)
151.28b)
33.48d)
91.32
83.95
83.95
89.81
33.28d)
152.53h)
)
150.72e)
161.92
103.43
151.14
31.64
150.70d)
161.92
103.42
151.15
31.82
m
103.23c)
151.16b)
33.00d)
91.44
104.06
151.33i)
CH2ꢀC(6/III)
32.90k)
m
C(1’/III)
)
90.44
90.42
C(2’/III)
C(3’/III)
C(4’/III)
C(5’/III)
84.44e)
84.24e)
89.40
85.27l)
84.83
84.18f)
83.70g)
87.45h)
33.53i)
84.18e)
83.35
89.41
34.15k)
87.45f)
33.28
32.62d)
Adenosine unit II
C(2/II)
C(4/II)
C(5/II)
C(6/II)
152.53
149.38
122.32
152.94
151.66b)
31.37
87.96
83.74
84.20
86.87
152.46
149.64
123.22
152.46b)
150.79
31.48
152.41
149.32
122.14
152.65
151.21b)
27.90
88.00
83.95
83.27
152.61b)
150.46c)
118.40d)
155.40e)
150.28c)
29.81
152.46
149.71b)
117.78
155.48c)
147.89
28.89
152.46
149.71b)
117.78
155.50
147.87
27.04
C(8/II)
CH2ꢀC(8/II)
C(1’/II)
88.10
88.53f)
83.17
88.81
88.80
C(2’/II)
C(3’/II)
C(4’/II)
C(5’/II)
83.77e)
84.03e)
87.51
82.64d)
83.73
82.62c)
83.68
82.20g)
88.02
87.83
87.06
86.26
33.28
33.55
33.00d)
33.28d)
32.90k)
32.61i)
Uridine unit I
C(2/I)
C(4/I)
C(5/I)
C(6/I)
b
152.53
163.41
103.71
151.66b)
62.53
92.63
84.82
84.82
90.42
)
152.08b)
163.08
103.86c)
150.95b)
62.34
152.06b)
163.05
103.48c)
151.09b)
62.57
92.14
84.71
84.52
152.53h)
163.05
104.06
151.33i)
62.88
150.79e)
162.64
100.86
154.84
59.25
150.79d)
162.64
100.85
154.83
59.24
CH2ꢀC(6/I)
m
C(1’/I)
92.07
)
90.80
90.81
C(2’/I)
C(3’/I)
C(4’/I)
C(5’/I)
85.01e)
84.44e)
89.66
85.07l)
84.83
84.40f)
83.35g)
87.64h)
33.23i)
84.40e)
83.35
90.14
32.80d)
89.41
34.15k)
87.57f)
33.28
33.55
33.78d)
a) Assignments based on a HSQC and a HMBC spectrum. b) – l) Assignments may be interchanged.
m) Hidden by the noise.

2624
Helvetica Chimica Acta – Vol. 92 (2009)
63 mmol), and stirred for 10 min at 238. The mixture was poured into sat. NaHCO3 soln. and extracted
with CH2Cl2. The combined org. phases were washed with brine, dried (MgSO4), and evaporated.
Trituration of the crude material in cyclohexane, followed by FC (CH2Cl2/MeOH/NH4OH 95 :5 :1 !
90 :10 :1), gave 20 (6 mg, 24%). Rf (CH2Cl2/MeOH 9 :1) 0.30. [a]2D5 ¼ ꢀ86.5 (c ¼ 1.0, CHCl3). UV 263
(31625). IR (CHCl3): 3472w (br.), 3389w (br.), 3327w (br.), 3195w (br.), 2971m, 2931m, 1696s (br.),
1643m, 1604m, 1522w, 1478w, 1425m, 1386m, 1376m, 1331w, 1298w, 1157m, 1089s, 1046m. 1H-NMR
(500 MHz, (D6)DMSO): see Table 6; additionally, 11.40 (br. s, HꢀN(3/I), HꢀN(3/III)); 7.25 (br. s,
2 NH2); 5.84 (t, J ¼ 5.5, HOCH2ꢀC(6/I)); 1.54, 1.53, 1.41, 1.39, 1.32, 1.31, 1.24, 1.23 (8s, 4 Me2CO2); 1.47
(sept., J ¼ 6.8, Me2CH); 0.770, 0.768 (2d, J ¼ 6.8, Me2CH); 0.72, 0.71 (2s, Me2CSi); ꢀ 0.10, ꢀ 0.11 (2s,
Me2Si). 13C-NMR (125 MHz, (D6)DMSO): see Table 7; additionally, 113.27, 112.87, 112.48, 112.47 (4s,
4 Me2C); 33.54 (d, Me2CH); 26.87 (2 C), 26.80, 25.13, 25.06, 24.96, 24.89, 24.60 (7q, 4 Me2C); 25.54 (s,
Me2CSi); 20.03, 20.00 (2q, Me2CH); 18.19, 18.16 (2q, Me2CSi); ꢀ 3.71, ꢀ 3.75 (2q, Me2Si). HR-MALDI-
MS: 1480.474 (34), 1479.481 (62), 1478.483 (86), 1477.476 (100, [M þ K]þ, C62H86KN14O18S3Siþ; calc.
1477.481), 1464.499 (31), 1463.503 (51), 1462.507 (70), 1461.505 (86, [M þ Na]þ, C62H86N14NaO18S3Siþ;
calc. 1461.507).
2’,3’-O-Isopropylideneadenosine-8-methyl-(81 ! 5’-S)-2’,3’-O-isopropylidene-5’-thiouridine-6-meth-
yl-(61 ! 5’-S)-2’,3’-O-isopropylidene-5’-thioadenosine-8-methyl-(81 ! 5’-S)-6-(hydroxymethyl)-2’,3’-O-
isopropylidene-5’-thiouridine (21). Under N2, a soln. of 20 (60 mg, 30 mmol) in THF (1 ml) was treated
with (HF)3 · Et3N (50 ml, 0.3 mmol) and stirred for 3 d at 238. The mixture was poured into sat. NaHCO3
soln. and extracted with CH2Cl2. The combined org. phases were washed with brine, dried (MgSO4), and
evaporated. The crude material was purified by repeated trituration with hexane and AcOEt to yield 21
(15 mg, 38%). Rf (CH2Cl2/MeOH 92 :8) 0.37. [a]2D5 ¼ ꢀ41.8 (c ¼ 1.0, CHCl3). UV 263 (48259). IR
(CHCl3): 3527w, 3472w, 3408w, 3319w, 3200w, 3019s, 2962w, 2928w, 1710m, 1682m, 1644w, 1603w, 1455w,
1
1304w, 1276w, 1262m, 1221s, 1093m, 1011m. H-NMR (300 MHz, (D6)DMSO; assignments based on a
HSQC and a HMBC spectrum): see Table 6; additionally, 11.41 (br. s, HꢀN(3/I and III)); 7.32, 7.29 (2 br.
s, 2 NH2) ; 7.3 – 7.1 ( br. s, HOꢀC(5’/IV)); 5.90 – 5.78 (br. s, HOCH2ꢀC(6/I)); 1.54 (2 Me), 1.39, 1.31
(2 Me), 1.28, 1.23 (2 Me) (5s, 4 Me2C). 13C-NMR (150 MHz, (D6)DMSO; assignments based on a HSQC
and a HMBC spectrum): see Table 7; additionally, 113.25, 113.08, 112.47 (2 C) (3s, 4 Me2C); 26.85, 26.83,
26.49, 26.45, 25.18, 25.04, 24.94, 24.87 (8q, 4 Me2C). HR-MALDI-MS: 1322.392 (16), 1321.391 (38),
1320.392 (67), 1319.389 (100, [M þ Na]þ, C54H68N14NaO18Sþ3 ; calc. 1319.390).
5’-S-Acetyl-2’,3’-O-isopropylidene-5’-thiouridine-6-methyl-(61 ! 5’-S)-N6-benzoyl-2’,3’-O-isopropyl-
idene-8-{[(4-methoxyphenyl)diphenylmethoxy]methyl}-5’-thioadenosine (22). Under N2, a soln. of 13
(410 mg, 0.53 mmol) in O2-free MeOH (2 ml) was treated with freshly powdered K2CO3 (220 mg,
1.55 mmol) and stirred for 10 min at 238. After dilution with sat. NH4Cl soln. (20 ml), the mixture was
extracted with CH2Cl2. The combined org. phases were washed with brine, dried (MgSO4), and
evaporated. Under N2, a soln. of the residue (380 mg) in O2-free DMF (2 ml) was treated with 10
(203 mg, 0.52 mmol) and K2CO3 (118 mg, 0.85 mmol), and stirred for 2 h at 238. The mixture was poured
into sat. NH4Cl soln. and extracted with CH2Cl2. The combined org. phases were washed with brine, dried
(MgSO4), and evaporated to yield 22 (530 mg, 94%) that was directly used for the next step. For analysis,
a sample was purified by FC (AcOEt/cyclohexane 3 :2 ! 1:0). Rf (AcOEt/cyclohexane 4 :1) 0.35. [a]D25
¼
ꢀ34.3 (c ¼ 1.0, CHCl3). IR (CHCl3): 3395w (br.), 3204w (br.), 3016m, 1697s (br.), 1614m, 1588w, 1510m,
1489w, 1448m, 1428m, 1384m, 1372w, 1356w, 1299w, 1259m, 1243m, 1157w, 1090m, 1070m, 1037w.
1H-NMR (300 MHz, CDCl3): see Table 8; additionally, 11.96 (br. s, HꢀN(3/II)); 10.55 (br. s, HNꢀC(6/
I)); 8.15 – 8.01 (m, 2 arom. H); 7.52 – 7.21 (m, 15 arom. H); 6.85 – 6.79 (m, 2 arom. H); 3.75 (s, MeO); 2.25
(s, AcS); 1.57, 1.52, 1.35 (2 Me) (3s, 2 Me2C). 13C-NMR (75 MHz, CDCl3): see Table 9; additionally,
194.49 (s, SC¼O); 165.12 (s, NHC¼O); 158.68, 143.39, 143.09, 134.11, 132.03 (5s); 132.30 (d); 130.35
(2d); 128.56 – 127.98 (several d); 127.23, 127.15 (2d); 114.36, 113.49 (2s, 2 Me2C); 113.36 (2d); 88.15 (s,
Ph2C); 55.23 (q, MeO); 30.57 (q, MeC¼O); 27.37, 27.22, 25.47, 25.33 (4q, 2 Me2C). HR-MALDI-MS:
1108.362 (19), 1107.357 (50), 1106.353 (68, [M þ Na]þ, C56H58N7NaO12S2þ ; calc. 1106.340), 1086.370 (29),
1085.365 (66), 1084.359 (100, [M þ H]þ, C56H59N7O12S2þ ; calc. 1084.358).
5’-S-Acetyl-N6-benzoyl-2’,3’-O-isopropylidene-5’-thioadenosine-8-methyl-(81 ! 5’-S)-2’,3’-O-isopro-
pylidene-5’-thiouridine-6-methyl-(61 ! 5’-S)-N6-benzoyl-2’,3’-O-isopropylidene-8-{[(4-methoxyphenyl)-
diphenylmethoxy]methyl}-5’-thioadenosine (23). Under N2, a soln. of 22 (110 mg, 0.1 mmol) in O2-free

Helvetica Chimica Acta – Vol. 92 (2009)
2625
Table 8. Selected 1H-NMR Chemical Shifts [ppm] and Coupling Constants [Hz] of the U*[s]A* Dimer 22,
the A*[s]U*[s]A* Trimer 23, and the U*[s]A*[s]U*[s]A* Tetramers 24 – 27
22
23a)
24
25
26b)
27c)
Solvent
CDCl3
CDCl3
(D6)DMSO (D6)DMSO CDCl3 (28) (D6)DMSO
Uridine unit IV
HꢀC(5/IV)
CH2ꢀC(6/IV)
HꢀC(1’/IV)
HꢀC(2’/IV)
HꢀC(3’/IV)
HꢀC(4’/IV)
2 HꢀC(5’/IV)
J(Ha,Hb/IV)
J(1’,2’/IV)
5.59d)
5.63d)
5.87
5.64d)
3.79 – 3.60
3.78 – 3.60 3.85/3.11 3.70 – 3.65
5.76e)
5.71e)
5.17
5.23
5.11
5.46
4.50
4.10 – 4.06
3.70
5.18
4.73
4.75 – 4.63 4.92
4.10 – 3.92 3.90
4.08 – 3.98
4.32 – 4.23
4.32 – 4.20 3.91/3.44 3.59/3.43
h
h
h
)
< 1.5
6.3
)
16.8
)
h
< 1.5
6.0
)
3.7
h
J(2’,3’/IV)
J(3’,4’/IV)
J(4’,5’/IV)
6.0
2.4
)
h
3.9
)
)
6.6
3.4/5.4
h
h
h
)
)
Adenosine unit III
HꢀC(2/III)
CH2ꢀC(8/III)
HꢀC(1’/III)
HꢀC(2’/III)
HꢀC(3’/III)
HꢀC(4’/III)
2 HꢀC(5’/III)
J(Ha,Hb/III)
J(1’,2’/III)
8.71
4.07 – 3.94
6.25
5.78
5.06
4.24
8.13f)
3.79 – 3.60
6.20
5.66
5.02
8.13f)
7.94
8.13
3.78 – 3.60 4.07/3.64 4.06 – 4.02
6.20
5.64
5.01g)
6.31
5.20
5.13
5.87
5.08
4.31
4.06 – 4.02
4.24 – 4.15
2.94 – 2.68
4.10 – 3.92 4.48
3.10/3.00
2.72 – 2.64 3.40/2.54 3.02/2.91
h
h
h
h
)
)
)
14.9
)
h
1.5
6.3
3.3
7.5/6.9
1.5
6.3
3.0
1.8
6.3
3.0
)
5.2
h
J(2’,3’/III)
J(3’,4’/III)
J(4’,5’/III)
6.0
)
)
h
h
)
h
h
)
)
10.8/h)
5.1/4.8
5.63d)
Uridine unit II
HꢀC(5/II)
CH2ꢀC(6/II)
HꢀC(1’/II)
HꢀC(2’/II)
HꢀC(3’/II)
HꢀC(4’/II)
2 HꢀC(5’/II)
J(Ha,Hb/II)
J (1’,2’/II)
4.92
3.73/3.32
5.74
5.29
4.84
4.06
3.25 – 3.01
14.7
5.09
3.59/3.34
5.68
5.16
4.87
4.11
2.93 – 2.79
15.0
< 1.5
6.3
5.63d)
5.62d)
4.77
3.79 – 3.60
3.78 – 3.60 4.05/3.48 3.70 – 3.65
5.71e)
5.75e)
5.14
6.06
5.49
5.43
4.56
4.17
3.86
5.15
4.71
4.75 – 4.63 4.87
4.10 – 3.92 4.41
4.08 – 3.98
2.94 – 2.68
2.72 – 2.64 2.58/2.45 2.91/2.79
h
h
h
)
< 1.5
6.6
)
13.2
)
h
< 1.5
6.3
3.9
< 1.5
6.0
)
3.2
h
J (2’,3’/II)
J (3’,4’/II)
J (4’,5’/II)
6.0
)
h
h
4.2
7.2/7.2
3.9
)
)
)
6.3
5.6/7.1
h
h
6.9/6.9
)
10.2/h)
Adenosine unit I
HꢀC(2/I)
8.77
4.79/4.50
6.23
5.37
5.23
8.74
4.65/4.49
6.22
5.44
5.11
8.17f)
4.11/3.97
5.97
5.60
4.85
8.14f)
8.38
8.15
CH2ꢀC(8/I)
HꢀC(1’/I)
HꢀC(2’/I)
HꢀC(3’/I)
HꢀC(4’/I)
4.75 – 4.63 5.07/4.96 4.70/4.65
6.31
5.57
5.00g)
6.63
5.05
5.41
6.02
5.08
4.31
4.06 – 4.02
4.46
4.34
4.24 – 4.15
4.10 – 3.92 4.65