Pentopyranosyl oligonucleotide systems: Communication No. 13. The α-L-arabinopyranosyl-(4′ → 2′)-oligonucleotide system: Synthesis and pairing properties

Article abstract of DOI:10.1002/hlca.200390118

Among the members of a family of diastereoisomeric penlopyranosyl-(4′ → 2′)-oligonucleotide systems derived from D-ribose, D-xylose, L-lyxose, and L-arabinose, α-arabinopyranosyl system shows by far the strongest Watson-Crick base pairing. The system is,

Full text of DOI:10.1002/hlca.200390118

Helvetica Chimica Acta Vol. 86 (2003)
1259
Pentopyranosyl Oligonucleotide Systems
Communication No. 131)
The a-l-Arabinopyranosyl-(4' ! 2')-oligonucleotide System: Synthesis and Pairing
Properties1)
by Oliver Jungmann2a), Markus Beier2b), Anatol Luther2c), Hoan K. Huynh2d), Marc-Olivier Ebert2e),
Bernhard Jaun2e), Ramanarayanan Krishnamurthy*3), and Albert Eschenmoser*3)
The Skaggs Institute for Chemical Biology at The Scripps Research Institute (TSRI) 10550 North Torrey Pines
Road, La Jolla, CA-92037, USA
and
Laboratory of Organic Chemistry, Swiss Federal Institute of Technology, Hˆnggerberg, HCI-H309,
CH-8093 Z¸rich
Dedicated to Jack Dunitz, teacher, colleague, and friend, on the occasion of his 80th birthday
Among the members of a family of diastereoisomeric pentopyranosyl-(4' ! 2')-oligonucleotide systems
derived from d-ribose, d-xylose, l-lyxose, and l-arabinose, the a-arabinopyranosyl system shows by far the
strongest WatsonÀCrick base pairing. The system is, in fact, one of the strongest oligonucleotide-type base-
pairing systems known. It undergoes efficient cross-pairing with all the other members of the pentopyranosyl
family, but not with RNA and DNA. The paper describes the synthesis and pairing of the properties of a-l-
arabinopyranosyl-(4' ! 2')-oligonucleotides.
1. Introduction. Among the six pentopyranosyl oligonucleotide systems studied in
our laboratories [1][3 6], the a-l-arabinopyranosyl-(4' ! 2')-oligonucleotide system
turned out to be the most surprising of all: here, we encountered the largest gap
between expectation and observation with respect to a system×s base-pairing proper-
ties. The qualitative conformational pairing criteria that had been conspicuously
successful in predicting the base-pairing capabilities of homo-DNA [7] and p-RNA [4a]
had led us to expect that an arabino analog of p-RNAwould be a weaker pairing system
than pyranosyl-RNA [4a]. Fact is that a-l-arabinopyranosyl-(4' ! 2')-oligonucleotides
constitute the strongest base-pairing system among all potentially natural nucleic acid
alternatives studied so far, surpassing the pairing strengths of both natural RNA and
pyranosyl-RNA by a large margin [4]. Recently, Jaun and co-workers [8] determined
the structure of an octamer duplex of the a-l-arabinopyranosyl-oligonucleotide series
by NMR spectroscopy and provided therewith the basis for a rationalization of
remarkable properties of the arabino system. Here, we report experimental aspects of
1
)
)
)
Communication No. 12: [1]. The paper is also communication No. 38 in the series −Chemistry of a-
Aminonitriles×. For communication No. 37, see [2].
a) TSRI October 1998 June 2000; b) TSRI September 1996 December 1998; c) TSRI October 1998
0
3
e
October 2000; d) TSRI October 1998 January 2001; ) ETH-Hˆnggerberg.
E-mail: eschenmoser@org.chem.ethz.ch; rkrishna@scripps.edu.

1260
Helvetica Chimica Acta Vol. 86 (2003)
the synthesis and the pairing behavior of a-d- and a-l-arabinopyranosyl-(4' ! 2')-
oligonucleotides, complementing the two preliminary publications on the topic
[3a][3b] and present a discussion of the system in comparison with the other members
of the pentopyranosyl-(4' ! 2')-oligonucleotide family.
Scheme 1 illustrates the selection problem that arose when choosing the nucleoside
configuration for each of the pentopyranosyl nucleoside members for a comparative
study of four of the family×s eight members. For the ribopyranosyl member [4], the first
studied experimentally, the choice had been straightforward, since, of the two possible
nucleoside d-diastereoisomers4), only the b-d-epimer could be expected to exist in a
well-defined single-type of conformation. In this isomer, the nucleobase would
unambiguously occupy an equatorial position at the pyranosyl chair. Furthermore, the
b-d-nucleoside epimer promised to be the more accessible one, since, in this epimer, the
nucleobase and the substituents at C(2') are trans to each other. In choosing the other
members of the family, the equatorial position of the nucleobase in the more-stable
pyranose-chair conformation became the predominant selection criteria by reasons
that were both chemical and etiological in nature. According to this criterion, the
xylopyranosyl member had obviously to be the b-d-diastereoisomer [1]. Less certain
was the choice in the cases of the remaining lyxo- and arabinopyranosyl members,
since, for both, it could be expected that, in each of the two epimers, the nucleobase
would occupy an equatorial position. In the case of lyxopyranose, the a-l-lyxopyr-
anosyl series was chosen because its nucleosides were expected to be readily accessible
(nucleobase trans to substituent at C(2')), and, furthermore, the sense of chirality at the
nucleobase-bearing anomeric center is the one corresponding to that in the b-d-ribo-
and b-d-xylopyranosyl series. In the arabinose case, we originally chose to synthesize b-
d-arabinopyranosyl-nucleosides to be able to study a quartet of pentopyranosyl-(4' !
2')-oligonucleotides in which the configurational differences between its members
would follow a consistent and transparent pattern, allowing a meaningful comparison
and interpretation of pairing properties. This pattern would have involved having the
phosphodiester group 2',4'-diequatorial in the b-d-ribo and b-d-xylo systems, 2'-
equatorial/4'-axial in the a-l-lyxo system, 2'-axial/4'-equatorial in the b-d-arabino
system, and the nucleobases equatorial in all four of them, with the anomeric center in
all cases possessing the same sense of local chirality (1'R).
Serious experimental difficulties in attempts to prepare the thymine nucleoside of
the b-d-arabinopyranosyl series with its cis-arrangement of nucleobase and the
substituent at C(2') made us reconsider the choice of target. Since the epimeric a-d-
arabinopyranosyl nucleoside persistently dominated the product mixtures from such
nucleosidation experiments, we eventually made this epimeric nucleoside (and its
adenine analog) our new and much more conveniently accessible target. This switch in
the configuration at the anomeric center demanded switching at the same time from the
d- to the l-arabinopyranosyl series (Scheme 2). For practical reasons, we developed the
methodology for the synthesis of a-arabinopyranosyl-(4' ! 2')-oligonucleotides in the
(chirally wrong) d-series and made the switch to the (chirally correct) a-l-
arabinopyranosyl series in a later phase of our work. After this change, the pattern
4
)
a/b-Nomenclature: a-diastereoisomer in the pentopyranose series: nucleobase at C(1) is cis to the OH
group at C(4), whereas these two substituents are trans in the b-diastereoisomer [9].

Helvetica Chimica Acta Vol. 86 (2003)
1261
Scheme 1. Configuration and (Idealized) Conformation of the Eight Diastereoisomeric Repeating Units of the
Family of Pentopyranosyl-(4' ! 2')-Oligonucleotides. Pyranose chairs of the b-d-series are shaded. The set of
the four members that were studied experimentally is framed. B ˆ nucleobase.
of configurational differences within the family of the four pentopyranosyl-oligonu-
cleotide systems becomes one in which both the nucleobase and the substituent at C(2')
are equatorial in all four systems and in which the configurational differences refer only
to the substituents at C(3') and C(4'). In all four systems, the local sense of chirality at
the anomeric center is the same; this is essential for intersystem cross-pairing studies.
The formulae of the four oligonucleotide systems studied experimentally are framed in
Scheme 1.
2. Synthesis. 2.1. Preparation of a-d-Arabinopyranosyl Nucleosides Containing
Adenine (A) and Thymine (T), and of a-l-Arabinopyranosyl Nucleoside Building
Blocks Containing Adenine (A), Thymine (T), Cytosine (C), and Guanine (G). The

1262
Helvetica Chimica Acta Vol. 86 (2003)
Scheme 2. Constitution, Configuration, and Idealized Pairing Conformation of a-l-Arabinopyranosyl-(4' !
2')-oligonucleotides. B ˆ nucleobase.
synthesis of a-arabinopyranosyl-(4' ! 2')-oligonucleotides to be described in Sect. 2.2
was executed in two parts: the initial approach made use of a Fmoc-protecting-group
strategy, resulting in the first preparations of adenine and thymine containing the a-
arabinopyranosyl oligonucleotides in the d-series. The approach was then extended to
the preparation of adenine, thymine, cytosine, and guanine oligonucleotides in the l-
series. Eventually, a follow-up synthesis of a-arabinopyranosyl oligonucleotides
containing all four canonical nucleobases in the l-arabinose series by a DMT-
protecting-group strategy paralleling the syntheses of oligonucleotides in the ribo-,
xylo-, and lyxopyranosyl series [1][4][5] was developed. This latter development was
induced by the demands of material for the recently published NMR-structure analysis
of the pa[(CGAATTCG)]2 duplex [8].
Although a part of the work on a-arabinopyranosyl oligonucleotides described here
had been initially carried out in the d-arabinose series (see above), all reactions
reported in this paper in Schemes 2 5 are uniformly formulated in terms of the l-
series. Special mention will be made whenever data are presented that were collected in
the d-series only. There are reports in the literature where arabinopyranosyl nucleo-
sides have been prepared in the a-l-, a-d-, and b-d-series, but in a different context
[10].
2.1.1. Synthesis of a-d- and a-l-Arabinopyranosyl Nucleoside Building Blocks
Containing the Fmoc Protecting Group. Nucleosidation of the readily accessible 1,2,3,4-
tetra-O-benzoyl-b-l-arabinopyranose (1) [11] with the appropriately protected nucle-
obases N6-benzoyladenine (ABz), thymine (T), N4-benzoylcytosine (CBz), and N2-
isobutyryl guanine (Gi-bu) under the Vorbr¸ggenÀHilbertÀJohnson conditions [12]
furnished the nucleoside derivatives 2a 2d in modest (44%) to high (84%) yields
(Scheme 3). Reaction with N2-isobutyrylguanine proved to be low-yielding, with
interference from the N7-isomer 2e necessitating time-consuming purification by

Helvetica Chimica Acta Vol. 86 (2003)
1263
column chromatography (CC)5). The constitutional assignment of N9- vs. N7-guanine
regioisomers was accomplished by NMR-spectral-data comparisons between 2d and 2e
as documented in Table 1, and UV-spectral data comparisons of the completely
deprotected guanine derivatives (Table 1 [13]), obtained by hydrazinolysis (NH2NH2 ¥
H2O, EtOH, 12 h, r.t.) of 2d and 2e. Selective deprotection of the BzO groups without
touching acyl protection of nucleobases was brought about by basic treatment (with
either NaOH or Na/MeOH) to afford the triols 3a 3d with yields in the 80% range6).
At this stage, configurational assignments of compounds 3a 3d were accomplished by
1H-NMR analysis: typically large coupling constants (J ˆ 9 12 Hz) between the
vicinal HÀC(1') and HÀC(2') as documented in Table 2 confirmed that these
nucleosides 3a 3d have the a-configuration, and that the preferred conformation in
solution ((D6)DMSO) is the one depicted in their respective formulas of Scheme 3. For
the thymine derivative 3b in the d-arabinose series, an X-ray analysis provided the final
confirmation (Fig. 1)7).
Table 1. Constitutional Assignments of N7- and N9-(a-l-Arabinopyranosyl)guanine-Regioisomers Based on
Comparison of Selected NMR Data of 2d and 2e, and UV Data of the Corresponding Free Triols
Compounda)
UVb)
lmax (e) [nm]
1H-NMRc)
((D6)DMSO)
d [ppm]
13C-NMRc)
((D6)DMSO)
d [ppm]
N7-Isomer 2e
6.75 (br. s, HÀC(1'))
8.60 (s, HÀC(8))
84.63 (C(1'))
7-(a-l-Arabinopyranosyl)guanine
N9-Isomer 2d
285 (5300)
6.02 (d, J ˆ 9.5, HÀC(1'))
8.49 (s, HÀC(8))
81.31 (C(1'))
9-(a-l-Arabinopyranosyl)guanine
252 (11400)
a) Free triols were obtained by hydrazinolysis (NH2NH2 ¥ H2O, EtOH, 12 h, r.t.) of 2e and 2d, respectively.
b) Measured in H2O, pH 7, c ꢀ 1 mg/100 ml. Assignment based on the consistent behavior that N9-guanine
nucleosides exhibit a maxima around 250 nm, while, in the corresponding N7-guanine isomers, it is shifted
bathochromically to ca. 285 nm [13a]; also cf. Table 1 on page 1428 of [13c]. c) Assignment based on the fact that
the HÀC(1'), HÀC(8), and C(1') chemical shifts are consistently downfield for the N7-guanine nucleosides as
compared to the N9-derivatives [13].
Exploratory experiments in the d-series with free nucleosides (mostly 3b) and
corresponding 2'-acylated derivatives of 3a 3d with reagents such as BzCl and DMT-
chloride led to the observation that the relative reactivities of the three OH groups
follows the order HOÀC(2') ꢀ HOÀC(3') ꢁ HOÀC(4')8). It was decided, therefore,
that the least-reactive 4'-OH function should carry the phosphoramidite group, while
5
)
)
)
Further improvements using 2-amino-6-chloropurine as a surrogate for guanine in the nucleosidation
reaction were promising, but difficulties with purification led to their abandonment.
6
7
In the case of cytosine nucleoside 2c, it was necessary to monitor the reaction carefully by TLC and stop
the reaction before the loss of the N4-benzoyl protecting group.
Carried out by Raj K. Chadha, TSRI. Crystallographic data (excluding structural factors) for the structure
reported in this paper has been deposited with the Cambridge Crystallographic Data Centre as deposition
No. CCDC 201895. Copies of the data can be obtained, free of charge, on application to the CCDC, 12
Union Road, Cambridge CB12 1EZ, UK (fax: ‡ 44 (1233) 336 0333; e-mail: deposit@ccdc.cam.ac.uk).
For example, tritylation of the 2'-acetate derivative of arabinopyranosyl thymine 3b led to the tritylation at
C(3') in 38% yield with no tritylation occurring at the 4'-OH.
8
)

1264
Helvetica Chimica Acta Vol. 86 (2003)
Table 2. 1H-NMR J(HÀC(1'),HÀC(2')) Values [Hz] for Selected Compounds (in (D6)DMSO) Depicted in
Scheme 3
Compound
a
b
c
d
(A)
(T)
(C)
(G)
2
3
4
5
6
7
9.1
9.2
9.6
9.7
9.3
9.3
9.5
9.3
9.5
9.7
9.3
9.3
12.4
9.7
9.8
9.3
8.0
9.0
9.3
8.5 : 8.8 (exo/endo)
9.0
Fig. 1. X-Ray structures of 2b, 6b, and 11a. Nucleosidic torsion angle (max Æ are as follows):
OÀC(1')ÀN(1)ÀC(2): 2b: 86.28; 6b: 125.48. OÀC(1')ÀN(9)ÀC(4): 11a: À102.18 (see Footnotes 7, 10, and 14).

Helvetica Chimica Acta Vol. 86 (2003)
1265
Scheme 3. Preparation of the A, T, C, and
G Nucleoside Building Blocks for the Synthesis of a-l-
Arabinopyranosyl-(4' ! 2')-oligonucleotides via the Fmoc-Protecting-Group Strategy

1266
Helvetica Chimica Acta Vol. 86 (2003)
the more-reactive 2'-OH would act as the nucleophile on the crucial chain elongation
on the CPG-solid-support synthesis; such an approach is the opposite of that
undertaken in the synthesis of oligonucleotides in the ribo- [4], xylo- [1], and the
lyxopyranosyl series [5]. A regioselective introduction of a protecting group at the 2'-
OH position is to be achieved via a transient regioselective blockade of the cis-vicinal
3'-OH and 4'-OH groups by, e.g., cyclic isopropylidine acetal group, easily introduced or
removed under mildly acidic conditions. The choice of an acid-labile isopropylidine
acetal protecting group for the 3',4'-positions precluded the use of DMT- as a 2'-O-
protecting group and thus dictated the choice of the Fmoc group as an acid resistant,
base labile protecting group for the 2'-OH position.
Reaction of triols 3a, 3b, and 3d with 2-methoxyprop-1-ene in DMF with HCl or
TsOH as catalyst afforded the corresponding 3',4'-isopropylidene derivatives 4a, 4b,
and 4d, respectively, in over 80% yields, as expected with no signs of a competing
formation of a (trans!) 2',3'-isopropylidene derivative. For triol 3c, the use of 4-
anisaldehyde (4-methoxybenzaldehyde) dimethyl acetal in place of 2-methoxyprop-1-
ene under similar reaction conditions (DMF, TsOH) afforded a mixture of
diastereoisomeric 3',4'-benzylidene acetals 4c, in 89% yield, as the sole product9).
Introduction of Fmoc group at C(2') was achieved by treatment of 4a 4d with Fmoc-
Cl in pyridine at 48 to give 5a 5d in good yields (73 88%). Subsequent acid-catalyzed
cleavage of the 3',4'-acetal protecting groups by treatment with 80% aq. AcOH for 5a
and 5d, with 4m HCl in dioxane for 5b, and with 0.7m TFA in MeOH/THF 2 :1 for 5d,
generated the 2'-Fmoc derivatives 6a 6d in yields ranging from 70 99%10). As
pointed to before, the lack of reactivity of the 4'-OH group towards electrophiles came
in handy for the selective introduction of the Bz group at 3'-O-position by simply
treating 6a 6c with ca. 1 equiv. of BzCl at À20 48 to afford the 3'-O-Bz derivatives
7a 7d in 70 83% yields11). No additional benzoylation at the nucleobases under
these conditions were observed. The appropriately protected 3'-O-Bz-2'-Fmoc
derivatives 7a 7d were the starting materials for the preparation of building blocks
suitable for automated solid-support synthesis.
The phosphitylations of 7a 7d were conducted under standard conditions [14] with
(2-cyanoethoxy)bis(diisopropylamino)phosphine and with 1H-tetrazole as a catalyst
(Scheme 4). A full equivalent of 1H-tetrazole was used to prevent the loss of the Fmoc
9
)
2-Methoxyprop-1-ene, in fact, could be used in the case of 3c with no difficulty for the formation of 3',4'-
isopropylidene acetal derivative; but, after formation of 2'-Fmoc derivative, problems (side-product
formation) were encountered in the subsequent deprotection of 3',4'-isopropylidene acetal group to
generate 6c.
In the case of the thymine derivative 6b in the d-arabinose series, X-ray analysis (carried out by Raj K.
Chadha, TSRI) confirmed the constitution and configuration (Fig. 1). Crystallographic data (excluding
structural factors) for the structure reported in this paper has been deposited with the Cambridge
Crystallographic Data Centre as deposition No. CCDC 201894. Copies of the data can be obtained, free of
charge, on application to the CCDC, 12 Union Road, Cambridge CB12 1EZ, UK (fax: ‡ 44 (1233) 336
0333; e-mail: deposit@ccdc.cam.ac.uk).
10
)
11
)
Among the benzoylating reagents tried BzCl, BzCN, and Bz2O, BzCl was found to be the reagent of
choice. While Bz2O (in the presence of DMAP or 1-methyl-1H-imidazole) was unreactive and resulted in
the cleavage of the base-sensitive Fmoc group, use of BzCN in the case of 6b led to mixtures of 3'-,4'-, and
3',4'-benzoylated products under the reaction conditions employed.

Helvetica Chimica Acta Vol. 86 (2003)
1267
group by the (i-Pr)2NH generated during the reaction12). The purification and isolation
of the phosphoroamidites were carried out by very quick (10 15 min.) CC over silica
gel at room temperature (silica gel pretreated with toluene containing 1% H¸nig×s
base, followed by careful washing with toluene to remove all of the base) to afford 8a
8d in yields of up to 95%. Such precautions were necessitated by the observations that
these phosphoramidites readily decompose under normal chromatographic isolation
procedures.
The sensitivity of the Fmoc group to base interfered, once again, with the loading of
the solid support CPG with nucleobases 7a 7d; compromises were nevertheless found,
which led to satisfactory production of suitably loaded solid supports 10a 10d. As the
first step, the 4'-OH group was derivatized as the 4'-O-succinate by reacting 7a 7d with
2.0 equiv. of succinic anhydride in the presence of 0.1 0.5 equiv. of DMAP in CH2Cl2
as solvent over a 5 7 day period at room temperature [15]. Minimizing the amount of
DMAP (and monitoring the reaction by TLC) was found to eliminate the unwanted
1
loss of the Fmoc group. The crude 4'-O-succinates 9a 9b (confirmed by H-NMR
spectroscopy and mass spectrometry) were directly used in the next step of attachment
to the specially derivatized LCAMA-CPG-solid support, which has been shown to be
stable to the conditions of base induced eliminative deprotection of the Fmoc group
[16]. Commercially available glyceryl-modified CPG was reacted with N,N'-dimethyl-
hexane-1,6-diamine in the presence of 1,1'-carbonyldi[1H-imidazole] to generate the
LCAMA-CPG (which has the advantages of a long spacer arm and the large pore size
of 500 ä, characteristics considered to be advantageous for larger loading values and
synthesizing long oligonucleotides). Reaction of the LCAMA-CPG with 4'-O-
succinylated nucleosides 9a 9d in the presence of TOTU [17] as a condensing agent,
followed by capping with excess Ac2O in the presence of 1-methyl-1H-imidazole, led to
suitably derivatized LCAMA-CPGs 10a 10d. The loading capacities were determined
to range from 12 19 mmol/g by treatment of 5 10 mg of the CPG with 5 ml of 0.1m
soln. of DBU in MeCN, and by measuring the absorbance of the solution at 305 nm with
a molar absorptivity value (e) for methylidene-fluorene of 9100 (see Exper. Part, Sect. 2
for details).
With derivatives 8a 8d and 10a 10d, the requisite building blocks, suitable for
proceeding to the synthesis of a-d- and a-l-arabinopyranosyl-(4' ! 2')-oligonucleo-
tides by automated solid-phase synthesis and the Fmoc strategy, were in hand.
2.1.2. Synthesis of a-l-Arabinopyranosyl Nucleoside Building Blocks Containing the
DMT Protecting Group. While the above Fmoc-protecting-group approach was useful
for furnishing enough phosphoramidites for the synthesis of a-d-arabinopyranosyl-
(4' ! 2')-oligonucleotides in a small-scale (0.5 1 mmol) range (see Sect. 2.2), the low
loading capacity of the CPG solid support and the problems presented by the Fmoc
group due to its inherent instability under basic conditions made this approach
12
)
These conditions were chosen over the usual −basic× reaction conditions, i.e., phosphitylation with
(chloro)(diisopropylamino)phosphine with either H¸nig×s base or with collidine and N-methyl-1H-
imidazole, because the sensitivity of the Fmoc group to the base used or produced in the phosphitylation
reaction. These −basic× reaction conditions were found to yield only ca. 60% of the phosphitylated
products, compounded by complications due to loss of Fmoc group and the incomplete removal of the
unreacted phosphitylating reagent by CC.

1268
Helvetica Chimica Acta Vol. 86 (2003)
Scheme 4. Preparation of the Phosphoroamidite and CPG Derivatives for the Automated Synthesis of a-l-
Arabinopyranosyl-(4' ! 2')-oligonucleotides via the Fmoc-Protecting-Group Strategy
0.03M (2 equiv.) succinic anhydride
0.05M (0.1–0.5 equiv.) 4-DMAP
CH2Cl2 / pyridine 50:1 / r.t. / 3–4 days
OH
0.1–0.9M (1–6 equiv.) (2-cyanoethoxy)-
bis(diisopropylamino)phosphine
O
BzO
B
0.4–1.2M (3–7 equiv.) 1H-tetrazole
FMOCO
7a B = ABz
7b B = T
CH2Cl2/ r.t. / 1 h
7c B = CBz
7d B = Gi-Bu
O
O(CH2)2CN
HO
(i-Pr)2N
P
O
O
O
O
O
BzO
B
BzO
B
FMOCO
FMOCO
9a B = ABz
(67%)
8a B = ABz
8b B = T
8c B = CBz
8d B = Gi-Bu
(80%)
(68%)
(72%)
(95%)
9b B = T
(84%)
(84%)
(67%)
9c B = CBz
9d B = Gi-Bu
(a) 500 mg LCAMA-CPG
0.06M (2 equiv.) TOTU
0.1M (3 equiv.) N-methylmorpholine
NHBz
O
N
NHBz
N
N
N
HN
N
ABz
=
T =
CBz
=
MeCN/ r.t. / 1.5 h
N
O
O
N
(b) 20 equiv. N-methyl-1H-imidazole,
xs. Ac2O, THF / pyridine / r.t. / 20 min.
O
O
O
N
NH
NH
O
Gi-Bu
=
C
Bz =
LCAMA-CPG
O
N
N
O
O
BzO
B
O
FMOCO
O
FMOC =
10a B = ABz
19 µmol/g
18 µmol/g
15 µmol/g
12 µmol/g
10b B = T
10c B = CBz
10d B = Gi-Bu
1,1'-carbonyldi[1H-imidazole]
CH2Cl2 / r.t. / 24 h
HO
HO
Si
Si
O
O
H3C
NH
O
OH
O
LCAMA-CPG
N
CH3
Glyceryl-CPG
(H2C)6
NH
(H2C)6
NH
H3C
H3C

Helvetica Chimica Acta Vol. 86 (2003)
1269
unsuitable for large-scale synthesis. Yet, the need for such a −large-scale× production
(ca. 10 15 mg) of pa(CGAATTCG) came up in the project to determine the structure
of a-arabinopyranosyl-(4' ! 2')-oligonucleotide duplex by NMR spectroscopy [8].
Therefore, we resumed our efforts to develop an approach based on the use of DMTas
a protecting group with the sole purpose of having access to large quantities of a-L-
arabinopyranosyl-(4' ! 2')-oligonucleotides.
The first requirement of this approach was to find a way to produce the 2'-O-trityl
derivatives with high selectivity and in high yields. Toward this goal, exploratory studies
on the tritylation of triols 3a and 3b were conducted. Treatment of triol 3a with
Me3SiNEt2 (DMF/ À 40 to À558) followed by tritylation (DMTCl/AgOTf/collidine/
MeCN/room temperature/3 d) furnished the 4'-O-DMT derivative, in low yields
(31%). Attempts to achieve a regioselective benzoylation with BzCN via a cyclic
tributyl stannane cis-diol derivative [18], furnished the 3'-O-benzoylated derivatives of
3a and 3b, subsequent tritylation (DMTCl/AgOTf/collidine) provided the 3'-O-
benzoyl-4'-O-trityl derivative in 53% yield in the case of 3a, while a mixture of the 3'-
O-benzoyl-4'-O-trityl and the 3'-O-benzoyl-2'-O-trityl derivatives was obtained starting
from 3b. Once again, the formation of product mixtures and low yields in the tritylation
step prompted us to abandon these approaches.
While searching for a selective protection group for the cis-vicinal 3',4'-diol
function, which could be removable again in the presence of a O-DMT group, we came
across a recent paper by R. A. Jones et al. [19] describing the use of the reagent N,N-
dimethylformamide dimethyl acetal for the reversible protection of vicinal diols. Cyclic
(dimethylamino)methyleneacetals of cis-diols were found to be cleaved with simple
protic solvents such as MeOH. The use of this reagent13) solved our problem for a
selective and efficient tritylation of the 2'-OH group.
Another improvement we wanted to realize while shifting from the Fmoc- to the
DMT-based strategy concerned the problems encountered in the nucleobase-protec-
tion step of the oligonucleotide synthesis (see Sect. 2.2.1). Those problems were to be
overcome by replacing the NH2 protecting groups for adenine (benzoyl), guanine
(isobutyryl), and cytosine (benzoyl) by phenoxyacetyl (−PAC×, for adenine as well as
guanine [20]) and isobutyryl (for cytosine [20]), the former for the sake of being more
easily removable, the latter for being more stable (see Footnote 6).
Scheme 5 summarizes the preparation of the four nucleoside building blocks 15a
15d that were required for the DMT-based approach to a-arabinopyranosyl-(4' ! 2')-
oligonucleotides, work that was done now exclusively in the l-arabinose series. Central
intermediates were the corresponding trihydroxy nucleosides 12a, 3b, 12c, and 12d, of
which the thymine member was available from the earlier work (see Scheme 3),
whereas the cytosine member was made from scratch by nucleosidation of l-
arabinopyranosyl-tetrabenzoate 1 with N-isobutyrylcytosine to 11c, followed by
hydrolytic cleavage of all protecting groups and selective reprotection of the cytosine
nucleus×s primary amino group with isobutyric anhydride in DMF. The trihydroxy-
purine nucleosides 12a and 12d, on the other hand, were made by exchange of the N-
protecting groups in the nucleosides 2a and 2d used earlier. This was accomplished by
13
)
Initiated by A. L. The publication [19] appeared while our work was in progress.

1270
Helvetica Chimica Acta Vol. 86 (2003)
complete deprotection by treatment with 2m NH3/MeOH to give 11a14) and 11b,
followed by introduction of the new PAC protection group into the purine nuclei by a
three-operation procedure consisting of O-silylation [21], N-acylation [20], and O-
desilylation, producing the nucleosides 12a and 12d in high yields.
With the four trihydroxy nucleosides 12a, 3b, 12c, and 12d in hand, we were ready
for the critical introduction of the trityl group the 2'-O-position via transient protection
of the vicinal 3',4'-diol groups utilizing the chemistry described by R. A. Jones et al. [19].
We were relieved to find that a one-pot procedure of 3',4'-amide-acetal formation with
nucleosides 12a, 3b, 12c, and 12d (1.5 equiv. N,N-dimethylformamide dimethyl acetal,
pyridine, collidine, DMF, room temperature, 1 3 h) followed by tritylation (1.2 1.5
equiv. DMT-Cl, collidine, CH2Cl2, room temperature, 6 12 h) and subsequent careful
methanolysis15) of the 3',4'-amide-acetal group worked reliably and effectively
furnished the 3',4'-dihydroxy-2'-O-tritylated nucleosides 13a 13d in yields of 71
61% after chromatographic purification16).
In our search for a solution to the last protection problem, that involving the 3'-OH
function, we eventually had to accept a compromise. In deciding to stick to benzoate as
the protection group, the strategy we had successfully used in the ribopyranosyl series
[4], we first had to learn that, in this DMT-protected arabino series, it is not the
equatorial 3'-OH group that can regioselectively be benzoylated but rather the axial 4'-
OH group. The behavior is the opposite of what we had observed previously while
exploring the 2'-Fmoc-protected 3',4'-dihydroxy nucleosides 6a 6d (see Scheme 3); it
is clearly due to steric hindrance to 3'-benzoylation exerted by the trityl group. In
welcome analogy to our experience in the synthesis of ribopyranosyl-(4' ! 2')-
oligonucleotide [4], where the observation of a (2' ! 3')-migration of the O-Bz group
had been crucial, we showed that the (under mild conditions17)) regioselectively
accessible 4'-monobenzoates 14a 14d undergo a clean equilibration in CH2Cl2 solution
and the presence of Et3N at room temperature producing a ca. 1:1 mixture of 4'- and 3'-
monobenzoates from which the pure 3'-monobenzoates 15a 15d could be isolated by
chromatography in yields of 35 38% (with respect to 14a 14d). This we accepted as a
compromise for the preparation of the desired nucleosides 15a 15d, taking into
account that the recovered 4'-benzoates could be (and were) recycled.
Phosphitylation of 15a 15d with the commercially available chloro(2-cyanoethoxy)-
(diisopropylamino)phosphane in the presence of (i-Pr)2NH gave phosphoroamidites
16a 16d in 79 85% yields (Scheme 6). For the preparation of the CPG solid support,
14
)
The structure was also confirmed by X-ray analysis (Fig. 1; carried out by Raj K. Chadha, TSRI).
Crystallographic data (excluding structural factors) for the structure reported in this paper has been
deposited with the Cambridge Crystallographic Data Centre as deposition No. CCDC 201893. Copies of
the data can be obtained, free of charge, on application to the CCDC, 12 Union Road, Cambridge CB12
1EZ, UK (fax: ‡ 44 (1233) 336 0333; e-mail: deposit@ccdc.cam.ac.uk).
15
)
Continuous TLC monitoring of the hydrolytic workup is important due to the high lability of PAC
protecting group. After addition of MeOH, a spot with lower Rf (corresponding to presumed hemi-acetal)
appears within minutes, which, then, over a period of 30 min converts to a more-polar spot corresponding
to the desired triols.
16
17
)
)
Chromatographic isolation in the case of 13a and 13d was performed with i-PrOH instead of MeOH to
minimize the loss of PAC protecting groups.
Benzoylation with BzCl/pyridine in CH2Cl2 at À308 to À408; the guanine derivative 13d required special
conditions (Bz2O/pyridine, DMPA, r.t.).

Helvetica Chimica Acta Vol. 86 (2003)
1271
Scheme 5. Preparation of the A, T, C, and
G Nucleoside Building Blocks for the Synthesis of a-l-
Arabinopyranosyl-(4' ! 2')-oligonucleotides via the DMT-Protecting-Group Strategy

1272
Helvetica Chimica Acta Vol. 86 (2003)
the standard one-pot protocol used previously for the pyranosyl ribo, xylo, and lyxo
series [1][4][5], (formation of the 4'-O-succinic acid derivative by treatment with
succinic anhydride in the presence of DMAP, followed by CPG derivatization with
TOTU [17] as an activator and subsequent capping with Ac2O), was found to work with
15c to give 17c without any problem. The loading capacity [13c] was determined to be
53 mmol/g, what we considered to be high. The cytosine-containing CPG solid support
was the only one needed, since pa(4'-CGAATTCG-2') was the only arabinopyranosyl-
oligonucleotide strand we synthesized with the DMT-based strategy.
Scheme 6. Preparation of the Phosphoroamidite and CPG Derivatives for the Automated Synthesis of a-l-
Arabinopyranosyl-(4' ! 2')-oligonucleotides via the DMT-Protecting-Group Strategy
2.2. Synthesis of a-Arabinopyranosyl-(4' ! 2')-oligonucleotides. Table 3 lists all a-
arabinopyranosyl sequences synthesized in the course of our study. By reasons
mentioned in the Introduction, sequences made in the first phase of the project
contained the d-arabinopyranose unit, while those prepared at a later stage belong to
the l-arabinopyranosyl series. With the exception of the self-complementary sequence
pa(CGAATTCG), which was prepared in −large× amounts (ca. 10 mg) for the NMR
structure analysis [8], the amount made in the case of all other sequences was in the
range of 1 mmol. The methodology followed was in essence the one previously used in
the ribo- [4], xylo- [1], and lyxopyranosyl [5] series on an automated DNA synthesizer
(Perseptive×s Expedite), taking into account, as detailed in Sect. 2.2.1 and 2.2.2 below,
the specific aspects and demands of the two different protection strategies. In these two

Helvetica Chimica Acta Vol. 86 (2003)
1273
strategies, the solid-phase synthesis of a-arabinopyranosyl oligonucleotide strands
proceeds in opposite directions, namely, the 3'(phosphityl) ! 2'(hydroxy) direction in
the Fmoc-protection strategy and the 2'(phosphityl) ! 3'(hydroxy) direction in the
DMT-protection strategy. If, in a future preparation of a-arabinopyranosyl-(4' ! 2')-
oligonucleotides, one had to choose between the two strategies with respect to the
criterion of expediency and yield, the DMT-protection strategy would clearly be
preferred.
Table 3. HPLC and MS Data of a-Arabinopyranosyl-(4' ! 2')-oligonucleotides
Entry Base sequencesa)
(all 4' ! 2')
Deprotectionb) O.D. 260 nm Anal. HPLC
MALDI-TOF-MSd)
[M ‡ H]
ion exchangec)
‡
Method
(yield)
gradient, tR [min]
obs.
calc.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
d-pa(A8)
l-pa(A8)
A
A
A
B
A
A
A
B
B
B
B
B
A
A
C
B
A
A
A
A
A
A
D
D
4.4 (5%) 10 50% in 30 min, tR 14.8 2573
4.8 (9%) 10 50% in 30 min, tR 14.8 2571
4.9 (5%) 10 50% in 30 min, tR 20.2 3890
20 80% in 30 min, tR 14.1 1223 (Na ) 1224 (Na )
4.2 (16%) 0 100% in 30 min, tR 21.0 2501
8.8 (18%) 0 100% in 30 min, tR 21.0 2501
2572
2572
3889
l-pa(A12
d-pa(T4)
d-pa(T8)
l-pa(t8)
)
‡
‡
2500
2500
3780
1546
1876
2205
2892
1566
1557
2214
2534
2534
2536
2536
2527
2545
2009
1770
2538
2538
2538
l-pa(T12
)
4.8 (7%) 0 100% in 30 min, tR 22.9 3780
d-pa(T4A)
d-pa(T4A2)
d-pa(T4A3)
d-pa(TA8)
d-pa(( ATATA )
d-pa( TATAT )
d-pa( ATATATA )
d-pa(A4T4)
d-pa(T4A4)
d-pa((AT)4
10 50% in 30 min, tR 11.5** 1547
10 50% in 30 min, tR 15.2** 1876
10 50% in 30 min, tR 25.9** 2206
2891
8.5 (31%) 10 50% in 30 min, tR 17.5 1564
20 80% in 30 min, tR 10.3 1557
20 80% in 30 min, tR 12.7 2214
7.6 (13%) 10 50% in 30 min, tR 28.2 2534
7.6 (13%) 10 50% in 30 min, tR 28.4** 2535
20 80% in 30 min, tR 14.4 2536
20 80% in 30 min, tR 15.1 2536
7.4 (16%) 0 100% in 30 min, tR 19.0** 2527
13.1 (25%) 0 100% in 30 min, tR 16.2 2544
d-pa(TA)4
l-pa(TATTTTAA)
l-pa(TTAAAATA)
l-pa(G8)
l-pa(C8)
l-pa(ATTCAGCG)
l-pa(CGCTGAAT)
2.7 (9%)
0
100% in 30 min, tR 26.6 2010
1771
4.3 (16%) 10 50% in 30 min, tR 9.9
4.3 (7%)
2.2 (6%)
446 (11%)
0
0
0
100% in 30 min, tR 19.1 2538
100% in 30 min, tR 18.9 2536
100% in 30 min, tR 21.5 2537
l-pa(CGAATTCG)* E
a) All sequences were synthesized by the Fmoc-protection strategy with the exception of the sequence 25,
labeled *, which was synthesized by the DMT-protection strategy. b) Method A: 0.15m MeONH2 in 25% aq.
NH3/EtOH 3 : 1 at r.t. for 10 h (24 h for mixed AT and homo-G sequences); Method B: 25% aq. NH3 at 48, 9
24 h; Method C: 25% aq. NH3 at 48, 21 h followed by 25% aq. NH2NH2, r.t., 10 min; Method D:
ethylenediamine/EtOH 1 : 1, r.t., 2 h [22]; Method E: pyridine/Et3N 4 : 1, r.t., 5 h; 40% aq. MeNH2, r.t., 1 h.
c) All oligonucleotides were purified by ion-exchange chromatography on a Mono Q HR 5/5 column (58 Â
6.0 mm, Pharmacia); elution with 10 mm Na2HPO4 in H2O and a linear gradient of 1m NaCl with a flow of 1 ml/
min. pH 10.5 11.5; followed by desalting on Sep-Pak cartridges. peak purity (260 nm) 95 99%. ** Elution with
25 mm Tris and 1.5m urea in H2O, and a linear gradient of 1m NH4Cl, pH 8.0. d) Matrix-assisted laser-desorption
ionization time-of-flight mass spectroscopy; matrix: 3-hydropicolinic acid or a-cyanohydroxycinnamic acid or
2,4,6-trihydroxyacetophenone and ammonium citrate buffer.
2.2.1. Oligonucleotide Synthesis with Fmoc-Protected Building Blocks. Since MeCN
solutions of Fmoc-phosphoramidites 8a 8d (stored with 4-ä molecular sieves) were

1274
Helvetica Chimica Acta Vol. 86 (2003)
found to decompose slowly (over periods of days), for their use in the solid-support
synthesis, all solutions were freshly prepared and used within 24 h18). The syntheses
were performed with ca. 30 mg of solid-support derivatives 10a 10d with the following
modifications to the protocols used earlier: a) coupling of phosphoroamidites over a
period of 15 min with either 0.5m pyridine hydrochloride or 0.5m 1H-tetrazole in
MeCN and b) cleavage of Fmoc with 0.1m DBU in MeCN over a 1.5-min period.
Coupling efficiencies were monitored by measurement of UV absorbance of fulvene
resulting from the cleavage of Fmoc group after each coupling (in analogy to Trityl
assay). They ranged from 95 to 98% for T-, C-, and G-containing residues, while around
90% for A-containing residues19). Coupling yields were found, sporadically, to drop to
ca. 80% in the synthesis of longer oligonucleotide strands.
The Fmoc groups of the last residue were removed before detaching the
oligonucleotide strand from the solid support. Detachment and concomitant removal
of all remaining groups posed a major problem. Under the base treatments normally
used (such as aq. NH3), some of the a-arabinopyranosyl-(4' ! 2')-oligonucleotides
were found to be extremely sensitive towards strand scission. Therefore, various
conditions had to be explored and fine-tuned for the specific sequence under
consideration, these conditions were
depending on the specific oligonucleotide
sequence either treatment of the dried CPG material with 0.15m MeONH2 ¥ HCl in
25% aq. NH3/EtOH 3 :1 at room temperature for 10 24 h (Method A, Table 3), or
with 25% aq. NH3 at 48 for 9 24 h (Method B, Table 3), or with 25% aq. NH3 at 48 for
21 h, followed by 25% aq. NH2NH2 ¥ H2O at room temperature for 10 min (Method C,
see Table 3), or with ethylenediamine/EtOH 1:1 at room temperature for 2 h (Method
D, Table 3). All these treatments led to the deprotection of the Bz groups on the sugar,
the 2-cyanoethyl group on the phosphate, as well as the acylamino groups of adenine,
cytosine, and guanine with the concomitant detachment of the oligonucleotides from
the CPG solid support. The homo-adenine sequences (Entries 1 3, Table 3) did not
fare well under the standard aqueous NH3 deprotection conditions, yielding lots of
strand-scission products; milder treatment with MeONH2 ¥ HCl in combination with aq.
NH3 (Method A) resulted in the least strand scission. For the homo-guanine-containing
sequence l-pa(G8) (Entry 21), it was observed that the N2-isobutyryl groups were not
fully deprotected within 10 h; therefore, the deprotection time was prolonged, and a
balance had to found between complete deprotection and the amount of strand-scission
products. Thus, the deprotection was stopped after 24 h, and the peak corresponding to
l-pa(G8) was isolated20). In the case of l-pa(ATTCAGCG) and its complementary
sequence where decomposition was most serious, treatment with ethylenediamine/
EtOH 1:1 at room temperature for 2 h [22] led to a marked improvement in
minimizing strand scission. In all cases, deprotections were monitored by ion-exchange
HPLC for optimal deprotection time. While almost all of the oligonucleotide sequences
18
)
)
Decomposition of Fmoc-phosphoramidites as seen by TLC. The problem was most severe in the case of
guanine derivative 7d.
In a typical 0.5-mmol-scale synthesis, each Fmoc-derived fulvene eluate was diluted to 5 ml with MeCN,
and the absorbance at 305 nm was measured against a 0.1m soln. of DBU in MeCN as a reference. The
value obtained for each fulvene eluate was compared as a percentage to that of the eluate of the previous
step.
19
20
)
Also isolated were l-pa(G6) and l-pa(G7) from this deprotection procedure as strand-scission products.

Helvetica Chimica Acta Vol. 86 (2003)
1275
could be purified by ion-exchange HPLC (target purity 95%), on Mono Q HR 5/5
column (Pharmacia) with 10 mm NaH2PO4 buffer at pH 10.5 11.5 with a linear
gradient of 1m NaCl (see Exper. Part, Sect. 6), some sequences containing thymine and
adenine (Entries 8 10, 16, and 19) remained still unresolved; these sequences were
purified with 25 mm Tris ¥ HCl buffer at pH 8.0 containing 1.5m urea with a linear
gradient of 1m NH4Cl. Product peaks (monitored at 260 nm) were collected in vials
containing 20 ml of a 1m aq. AcOH (to prevent further strand scission) and desalted (see
Exper. Part, Sect. 7), and the desalted oligonucleotides dissolved in 1 ml of H2O. The
concentrations of such purified oligonucleotide stock solutions were determined by UV
(at ca. 808) with the following molar extinction coefficients at 260 nm at pH 7:
e(pa(A)) ˆ 15000, e(pa(T)) ˆ 10000, e(pa(C)) ˆ 8400, e(pa(G)) ˆ 11900. The purified
oligonucleotides were checked for the correct mass by matrix-assisted laser-desorption
ionization time-of-flight (MALDI-TOF) mass spectrometry [23]. For the oligonucleo-
tides synthesized by this procedure and for their HPLC and MSdata, see Table 3.
2.2.2. Synthesis of a-l-Arabinopyranosyl-(4'-CGAATTCG-2') with DMT-Protected
Building Blocks. This synthesis followed the previously established protocols [1][4][5]
on the automated DNA synthesizers (Perseptive×s Expedite) on a 5-mmol scale, with the
following modifications to the standard program: a) increase of the coupling time to
27 min with a 0.07m solution of phosphoramidite in MeCN (this modification allowed
the parsimonious use of only ca. 5-fold excess of phosphoramidite), and b) use of 0.35m
solution of 5-(ethylthio)-1H-tetrazole in MeCN as activator. Average coupling
efficiencies were ca. 93%, as judged by trityl assay (e(500 nm) ˆ 66000). A total of
nine such synthesis operations, carried out in the −Trityl-Off× mode, were performed to
reach an overall 45-mmol scale.
The detachment from the solid support and the deprotection of the oligonucleotide
was effected in two stages: first, treatment of the dried CPG solid support with
pyridine/Et3N 4 :1 for 5 h at room temperature [24] (removal of the 2-cyanoethyl
protecting group converting the phosphotriester group into the more-stable phospho-
diester group) and, second, exposure to 40% aqueous MeNH2, effecting the
simultaneous deprotection of the sugar and nucleobase protecting groups along with
detachment of the oligonucleotide from the solid support. The crude oligonucleotide
was purified (target purity > 95%) by ion-exchange HPLC on Mono Q HR 5/5 column
(Pharmacia) with 10 mm NaH2PO4 buffer at pH 11.3 with a linear gradient of 1m NaCl
(see Exper. Part, Sect. 6)21). The product peak (monitored at 260 nm) was collected in a
vial containing 0.1m phosphate buffer at pH 7.0 to prevent any strand scission. The
oligonucleotide was desalted (see Exper. Part, Sect. 7), and its integrity was checked by
MALDI-TOF mass spectroscopy (Entry 25, Table 3). The counter ion was exchanged
‡
‡
to Na by treatment with Na form of DOWEX (50 WÂ 8) to make it suitable for
NMR measurements. The concentration of the oligonucleotide measured with e(260
nm) ˆ 90600, gave a total of 4.92 mmol of the oligomer (15.9 mg isolated; 11% overall
yield).
21
)
This high pH was found to be more satisfactory in preventing the self-association of this self-
complementary oligonucleotide than the usual pH of 10.5.

1276
Helvetica Chimica Acta Vol. 86 (2003)
3. Base-Pairing Properties. The pairing properties of a-d- and a-l-arabinopyr-
anosyl-(4' ! 2')-oligonucleotides were characterized by methods used in earlier studies
in the ribo- [4], xylo- [1], and lyxopyranosyl oligonucleotide series [5]. These methods
are: temperature-dependent UV spectroscopy (for Tm measurements [25]), concen-
tration-dependent Tm measurement by UV spectroscopy (for determination of
thermodynamic data [26]), molar-ratio-dependent UV spectroscopy (for checking
stoichiometry of complex formation [27]), and temperature-dependent CD spectro-
scopy. All measurements were made in −phosphate buffer× (10 mm aq. NaH2PO4 buffer
containing 0.1 mm Na2(EDTA), 150 mm (or 1m) NaCl at pH 7.0) with a total
oligonucleotide concentration of ca. 10 mm, unless otherwise stated. Tables 4 and 5 list
the Tm and thermodynamic data collected for the intra- and intersystem pairing duplexes
formed by the a-l-arabinopyranosyl-(4' ! 2')-oligon ucleotides, while Figs. 3 6 docu-
ment pairing behavior, complementing information published in preliminary commu-
nications [3]. The data of Table 4 indicate whether a given observation refers to the d- or
l-series22); all data in Table 5 refer to the l-arabinopyranosyl series.
The Tm values listed in Tables 4 and 5 document the dominant characteristic of a-
arabinopyranosyl-(4' ! 2')-oligonucleotides, namely, that they form more-stable du-
plexes among all members of the pentopyranosyl-oligonucleotide family studied thus
far. Duplexes of pa-octamers have Tm values on the average of ca. 208 higher than
corresponding pl-octamers, their nearest corresponding competitors (Table 5). This
property tends to extend to the phenomenon of self-association of non-self-comple-
mentary strands, both in homo-oligomer (Table 4, Entries 2 4) and hetero-oligomeric
strands (Entries 8, 11, and 12). While self-association is weak for (A8) and (A12) oligomers
(Tm < 58 and 7.88, resp.), it is remarkably pronounced in (T8) and (T12) series (Tm ˆ 26.98
and 52.28, resp., see Fig. 2). This behavior is also reflected in the behavior of the
antiparallel-complementary, but not self-complementary, sequences (4'-TATTTTAA-2')
and (4'-TTAAAATA-2'); mixed together, they form a duplex that is stronger than the
self-association of the individual strands (Fig. 3,d, and Table 4, Entries 8 10). The
sequences (G6) and (C6), remarkably, do not exhibit discernable self-pairing; yet, mixing
of these two sequences produces a very strong duplex (Tm ˆ 72.38 in 0.15m NaCl, Entry 7;
Fig. 3). The tendency for self-pairing in a sequence with mixed A and T nucleobases is
diminished by the introduction of G and C nucleobases. This is demonstrated by the low
Tm values of self-association exhibited by the non-self-complementary sequence (4'-
ATTCAGCG-2') and its complementary sequence (4'-CGCTGAAT-2') (Entries 11 and
12). Once again, the duplex formed after mixing these two sequences has high stability
(Tm ˆ 80.58 in 0.15m NaCl, Entry 13; Fig. 4). Testifying to the superiority of the a-arabino-
oligonucleotide sequences in terms of the strength of base-pairing, the self-complemen-
tary sequence (4'-CGAATTCG-2') melts at a Tm ˆ 78.1 (1 mm, 0.15m NaCl, Entry 13;
Fig. 4,c), while the Tm value of the corresponding p-RNA sequence is 51823)!
The CD spectra of the a-l-arabino-oligonucleotide sequences are strikingly similar
to those observed in the a-l-lyxo series, but markedly different from the b-d-ribo and b-
22
)
)
In one case, pa(A8) ‡ pa(T8), UV-Tm measurements in both the d-a- and l-a-series were made, observed
Tm values were: 71.18 in the d-a-case (0.15m NaCl, 0.01m Tris buffer) and 68.38 in the l-a-case (0.15m
NaCl, 0.01m phosphate buffer).
23
Calculated from the thermodynamic data reported in [4b].

Helvetica Chimica Acta Vol. 86 (2003)
1277
Ref.
Table 4. Tm Values and Thermodynamic Data of a-Arabinopyranosyl-(4' ! 2')-duplexes
Entry Oligonucleotide
System (l-series)
Tm [8]a)
DG
DH
TDS
Duplexes, c ꢀ 10 mm (258)
[kcal/mol] (258)
1m NaCl 0.15m NaCl 0.15m NaCla)b)
1
2
3
4
5
6
7
8
9
10
11
12
13
pa(A8)
< 5
pa(A12
pa(T8)
)
7.8c)
26.9c)
52.2c)
pa(T12)
d-pa(A8) ‡ d-pa(T8)
78.7d) 71.1d)
> 95
À 15.7d) À 60.6d)
44.9d) [3a]
[3b]
pa(A12) ‡ pa(T12
pa(G6) ‡ pa(C6)
pa(TTAAAATA)
pa(TATTTTAA)
)
74.3e)
72.3
À 12.1
À 8.2
À 30.7
À 31.1
À 53.2
À 66.1
À 18.6
48.1d)
48.9d)
74.6
À 22.9
À 43.8
À 50.8
À 9.4
pa(TATTTTAA) ‡ pa(TTAAAATA)
65.6
80.5
À 15.3
[3b]
[3b]
pa(ATTCAGCG)
< 5
pa(CGCTGAAT)
18.6d)
pa(ATTCAGCG) ‡ pa(CGCTGAAT) > 88
Self-complementary sequences
d-pa(A4T4)
d-pa(T4A4)
d-pa(T4A3)
d-pa(ATATA)
d-pa(TATAT)
d-pa(ATATATA)
d-pa(ATATATAT)
d-pa(TATATATA)
pa(CGAATTCG)
À 17.6
À 63.0
À 45.4
14
15
16
17
18
19
20
21
22
61.2d)
À 13.5d) À 59.9d)
À 14.5d) À 57.6d)
À 46.4d) [3a]
À 43.0d) [3a]
69.4d)
47.5d)
29.2d)
21.0d)
57.4d)
60.0d)
60.8d)
78.1 (1 mm)
[3a]
[3a]
a) Measurements were made in 0.01m NaH2PO4, 0.1 mm Na2(EDTA) buffer, pH 7.0, unless otherwise indicated.
b
Estimated error of Tm determination Æ 0.58. À : not measured. ) Thermodynamic data from plots of TÀm1 vs. ln
c; experimental error estimated in DH values Æ 5%. c) Self-pairing. d) Tm measured in 0.01m Tris ¥ HCl buffer,
pH 7.0. e) Tm measured with no NaCl.
d-xylo series (Fig. 5; also Fig. 2 in [3b]). This parallels the conformational relationships
that are presumed to prevail within the pentopyranosyl family (see Scheme 1). The
parallelism reappears in intersystem cross-pairing studies (Table 5 and Fig. 6). The Tm
values of the duplexes formed between the pa and pl sequences are consistently higher
than any duplex formed from any sequence combinations of pa- with either pr- or px-
oligomers (Table 5). This intersystem cross-pairing behavior pervades the entire family
[3] (Table 5). Not surprisingly, there is no intersystem cross-pairing between a-l-
arabinopyranosyl-(4' ! 2')-oligonucleotides and the corresponding complementary
RNA and DNA sequences; this is similar to the behavior exhibited by the other three
pentopyranosyl-(4' ! 2')-oligonucleotide members [1][3 5].
All observations made both in the inter- and intrasystem pairing studies are
consistent with the WatsonÀCrick base-pairing mode in antiparallel strand orientation.
This is confirmed for the a-l-arabinopyranosyl duplex formed by the self-comple-
mentary strand pa(4'-CGAATTCG-2') by NMR structure analysis [8].
4. Discussion. Among the four experimentally studied members of the family of
pentopyranosyl-(4' ! 2')-oligonucleotide systems, the a-arabinopyranosyl member is,

1278
Helvetica Chimica Acta Vol. 86 (2003)
Table 5. Tm Values of Duplexes Formed by Inter- and Intrasystem Base Pairing of a-l-Arabinopyranosyl-(4' !
2')-oligonucleotides
Entry
Oligonucleotide systems
Tm [8]
For comparison:
Intersystem duplexes
Tm of Intrasystem duplexes
1m NaCl (c ꢀ 5 ‡ 5 mm)a)
1m NaCl (c ꢀ 5 ‡ 5 mm)a)b)
(A8) ‡ (T8)
pa ‡ pr
pr ‡ pa
pa ‡ px
px ‡ pa
pa ‡ pl
1
2
3
4
5
6
44.7
59.4
42.5
54.7
66.9
67.8
pa ‡ pa
pl ‡ pl
px ‡ px
pr ‡ pr
r ‡ r
78.7
51.0
47.3*
45.5
23.0
pl ‡ pa
(A12) ‡ (T12
pa ‡ pr
pr ‡ pa
pa ‡ px
px ‡ pa
pa ‡ pl
)
7
8
9
10
11
12
67.8
83.0
69.5
83.0
89.3
> 90
pa ‡ pa
pl ‡ pl
px ‡ px
pr ‡ pr
r ‡ r
> 95
74.3
73.2
67.6
53.5
pl ‡ pa
(TATTTTAA) ‡ (TTAAAATA)
13
14
15
16
17
18
pa ‡ pr
45.3
53.6
53.1
45.8
61.2
62.3
pa ‡ pa
pl ‡ pl
px ‡ px
pr ‡ pr
r ‡ r
74.6
46.4*
44.4
45.9
17.2
pr ‡ pa
pa ‡ px
px ‡ pa
pa ‡ pl
pl ‡ pa
(ATTCAGCG) ‡ (CGCTGAAT)
19
20
21
22
pa ‡ pr
pr ‡ pa
pa ‡ pl
pl ‡ pa
70.2
66.8
73.3
73.6
pa ‡ pa
pl ‡ pl
px ‡ px
pr ‡ pr
r ‡ r
> 88
61.9
67.6
52.0
a) Measurements were performed in 0.01m NaH2PO4, 0.1 mm Na2(EDTA) buffer, pH 7.0, unless otherwise
indicated. b) For Tm data and thermodynamic data of pa/pa, pl/pl, px/px, pr/pr, and r/r duplexes determined in
0.15m NaCl, see Table 2 in ref. [3b]. Estimated error of Tm determination Æ 0.58. All Tm values taken are from
[3b]. prˆ b-d-Ribopyranosyl, px ˆ b-d-xylopyranosyl, pl ˆ a-l-lyxopyranosyl, pa ˆ a-l-arabinopyranosyl, rˆ
RNA. Values with (*) were measured in 0.01m Tris ¥ HCl buffer, pH 7.0.
in a way, the most informative one when it comes to comparatively analyzing the
family×s base-pairing properties. Some of the structural factors deemed to be
responsible for these remarkable properties express themselves most clearly in the
structure of the a-arabinopyranosyl system. To these factors belong the conformational
unambiguity and rigidity of the pyranosyl chairs, the fact that, in each member, the
(relatively) small set of least-strained (idealized) conformations accessible to the repeating
unit of a single strand contains at least one repetitive [3][7] pairing conformation, and,
finally, that there is in each of these pairing conformations a large inclination between the
(averaged) strand axis and the (averaged) nucleobase pairing axis. The first and second
factors are expected to contribute to the conformational preorganization of pentopyr-
anosyl-(4'! 2')-oligonucleotide single strands towards their duplex conformation. The
third factor is responsible for the two most-characteristic features by which pentopyr-
anosyl-(4'! 2')-oligonucleotide duplexes structurally distinguish themselves from the

Helvetica Chimica Acta Vol. 86 (2003)
1279
Fig. 2. UV and CD data documenting the pairing behavior of A,T-containing (4' ! 2')-a-l-arabinopyranosyl
sequences. a) Self-pairing of the homobasic strands pa(A8), pa (A12), pa(T8), and pa(T12) as demonstrated by
UV-spectroscopic Tm curves (heating). b) UV-Spectroscopic Tm curves (heating) of selected duplex formation.
c) Temperature-dependent CD curves of pa(T8) demonstrating T/T self-pairing (temperature range: 08 ! 558).
d) UV-Spectroscopic Tm curves of the duplex formed by pa(4'-TTAAAATA-2') with its antiparallel
complement pa(4'-TATTTTAA-2') and self-pairing Tm curves of the individual strands. All measurements
were made in 10 mm aq. NaH2PO4 buffer containing 0.1 mm Na2(EDTA), 1m NaCl at pH 7.0, unless otherwise
indicated. Total oligonucleotide concentrations in all measurements were ca. 10 mm. All Tm curves were fully
reversible (no hysteresis). Tm Values are calculated from the maxima of the first derivative curve using
kaleidagraph software program.
natural nucleic acids, namely, their quasi-linear (as against helical) duplex conformation
and their exclusive adoption of an interstrand (as against intrastrand) base-stacking mode.
A large inclination between backbone and nucleobase axes allows the nucleobases to reach
optimal stacking distances without the need for strand helicalization24), implying that base
stacking within the duplex has to be interstrand25).
24
)
)
Strand constriction by either helicalization or inclination are the two major pathways available to a 1,3-
phosphodiester-based (6-bond) oligonucleotide strand to lower the distance between vicinal nucleobases
and reach optimal stacking distances (see, e.g., [28]).
To the extent that nucleobase stacking can be a source of stabilization due to attractive forces between the
stacking partners p-systems, interstrand stacking in a duplex is expected to act as a glue between the
strands, whereas intrastrand stacking is not, at least to the extent that it can persist in a single strand after
de-duplexation.
25

1280
Helvetica Chimica Acta Vol. 86 (2003)
Fig. 3. UV-Tm, Thermodynamic, and −mixing-curve× data documenting the pairing behavior of G,C-containing
(4' ! 2')-a-l-arabinopyranosyl sequences. a) UV-Spectroscopic Tm melting curves for pairing between pa(G6)
and pa(C6), and corresponding curves of individual strands. For conditions of measurements, see caption of
Fig. 2. Tm Curves fully reversible (no hysteresis). b) Thermodynamic data of the duplex pa(G6)/pa(C6) deduced
from concentration dependence of UV-spectroscopically determined Tm values. Tm Values measured in 10 mm
aq. NaH2PO4 containing 0.1 mm Na2(EDTA), 150 mm NaCl at pH 7.0; for method, see [26]. c) Molar-
ratiodependence of UV absorption (275 nm, 208) (−mixing curve×) considered to indicate a 1:1 duplex
stoichiometry for the pairing between pa(G6) and pa(C6). Data measured in 10 mm aq. NaH2PO4 containing
0.1 mm Na2(EDTA), 150 mm NaCl at pH 7.0 with a total oligonucleotide concentration of 10 mm; for method,
see [27].
As delineated in the Introduction, there are two conformational subgroups to be
distinguished among the four (4' ! 2')-pentopyranosyl systems investigated
(Scheme 1); they differ in the conformation of their 4',2'-phosphodiester links; in one

Helvetica Chimica Acta Vol. 86 (2003)
1281
Fig. 4. UV and CD data documenting the pairing behavior of indicated A,T,C,G-containing octamer sequences.
a) UV-Spectroscopic Tm curves and b) Temperature-dependent CD curves (temperature range: 558 ! 958). For
conditions of measurements, see caption of Fig. 2. Tm Curves fully reversible (no hysteresis). c) UV-
Spectroscopic Tm curves of duplex formed by the self-complementary pa(4'-ATTCAGCG-2') sequence (c ꢀ
1 mm, 0.15m NaCl, 10 mm aq. NaH2PO4 containing 0.1 mm Na2(EDTA), 150 mm NaCl at pH 7.0) [8].
group (b-d-ribo and b-d-xylo), this link is 4',2'-diequatorial, and 4'-axial/2'-equatorial
in the other (a-l-lyxo and a-l-arabino). For each of these two groups, a representative
NMR structure analysis has been carried out by Jaun and co-workers, with the duplex

1282
Helvetica Chimica Acta Vol. 86 (2003)
Fig. 5. Temperature-dependent CD curves documenting intra- and intersystem pairing between: a) pa(A8) and
pa(T8), b) pa(A8) and pl(T8), c) pa(A8) and pr(T8), and d) pa(A8) and px(T8). Measurements in 10 mm aq.
NaH2PO4 containing 0.1 mm Na2(EDTA), 1m NaCl at pH 7.0. Total oligonucleotide concentrations were ca. 10 mm.
Fig. 6. UV-Spectroscopic Tm curves documenting intra- and intersystem pairing between pa(A8) and pa(T8),
pa(A8) and pl(T8), pa(A8) and pr(T8), and pa(A8) and px(T8). For conditions of measurement, see caption of
Fig. 5.

Helvetica Chimica Acta Vol. 86 (2003)
1283
of the self-complementary octamer sequence 4'-CGAATTCG-2' in the b-d-ribopyr-
anosyl [4b] and the a-l-arabinopyranosyl series [8]. The latter study was described and
compared with the former in a recent paper [8], so that we can restrict ourselves here to
recapitulate those aspects of the a-l-arabinopyranosyl structure that are considered to
have a direct bearing on the system×s pairing properties.
The axial attachment of the phosphodiester bridge at C(4') in the a-arabino
structure compared to its equatorial attachment in the b-ribo series removes a specific
source of steric encumberment in duplexes of the latter, namely, the intrastrand steric
contact between the equatorial H-atom of the CH2(5') group of the pyranose chairs and
the p-face of the neighboring nucleobase upstream26). The steric confrontation
between these two groups is a direct consequence of the di-equatorial positioning of the
phosphodiester bridge at C(4') and C(2'). It disappears, as Fig. 7 indicates, with the
change of the phosphodiester attachment at C(4') from equatorial to axial. The change
1
clearly reflects itself in chemical shifts of the critical CH2 H-atom in the H-NMR
spectra of the a-arabinopyranosyl vs. the b-ribopyranosyl octamer duplex: in the
spectrum of the latter the critical H-atoms appear shielded by the ring current of a
neighboring purine base, whereas the corresponding H-atoms in the a-arabinopyr-
anosyl duplex are not (see Fig. 7).
Base pairing in a-l-arabinopyranosyl (4'-OP axial) duplexes is found to be always
much stronger than in corresponding b-d-xylopyranosyl (4'-OP equatorial) duplexes
(Table 5 and [1]), and, furthermore, a-l-lyxopyranosyl duplexes tend to be more stable
than b-d-ribopyranosyl and b-d-xylopyranosyl duplexes (Table 5 and [5]). This pattern
of relative strength of base pairing is consistent with the assumption that the intrastrand
steric contact between pyranose rings and neighboring nucleobases in b-ribo- and b-
xylopyranosyl duplexes does indeed act as steric repulsion that lowers duplex stability,
and that it is the absence of this destabilization in the a-arabino- and a-lyxopyranosyl
series that, at least in part, makes their duplexes more stable. The question, however,
why it is that a-arabinopyranosyl duplexes are so much more stable than their a-
lyxopyranosyl analogs, remains.
Fig. 8 points to a conformational divergence between a-arabino- and a-lyxopyr-
anosyl duplexes that can be expected to contribute to this difference in pairing
strengths. The figure indicates, within the simplifying constraints of idealized
conformations, that the diversity of accessible conformers around the axial 4'-
attachment of the phosphodiester bridge in a a-lyxopyranosyl single strand is larger
than in a a-arabinopyranosyl single strand. The equatorial position of the 3'-OH group
in the latter precludes a population of the family of conformers in which the 4'-OÀP
bond of the phosphodiester group is anti-periplanar to the arabinopyranose ring×s
endocyclic C(4')ÀC(5') bond, in contrast to the a-lyxopyranosyl series, where the
population of this family of conformations is unfettered (Fig. 8,b vs. a, see also e). In
both systems, on the other hand, the family of conformers that includes the actual
pairing conformations27) is sterically unencumbered in both single and duplex strands
of the arabino- and lyxopyranosyl series (Fig. 8,d and c). This means that, at least with
respect to this C(4')-domain of the conformational space of pentopyranosyl-(4' ! 2')-
26
)
)
−Upstream× ˆ in the (2' ! 4')-direction as against the (downstream) (4' ! 2')-direction.
4'OÀP Bond anti-periplanar to endocyclic C(4')ÀC(3') bond.
27

1284
Helvetica Chimica Acta Vol. 86 (2003)
Fig. 7. 1H-NMR Chemical shifts of equatorial and axial H-atoms at C(5') in the (4'-CGAATTC-2')2 duplex in the
b-d-ribopyranosyl-(4' ! 2')- and the a-l-arabinopyranosyl-(4' ! 2')-series. The data are taken from [4b] and [8]
and refer to the nucleoside-units T5 and T6, respectively.

Helvetica Chimica Acta Vol. 86 (2003)
1285
oligonucleotide strands, the single strand of a-arabinopyranosyl nucleotides is more
pre-organized toward the pairing conformation than the single strand of a-lyxopyr-
anosyl nucleotides. This constraint in the population of single-strand conformers in the
arabinopyranosyl series would be expected to contribute to higher duplex stability.
Fig. 8. Qualitative conformational reasoning referring to the phosphodiester conformations in the C(4')- and
C(2')-domains for both the a-d-lyxopyranosyl and a-l-arabinopyranosyl series. The sign ‡ denotes allowed and
À disallowed conformational populations according to steric and stereoelectronic criteria. Bold arrow points to
region of strong steric repulsion in (idealized) pairing conformation. For the rules followed in this
conformational analysis, see [4a][7].
Appealing as the above argument may appear, it should not detract from
recognizing the complexity of the problem of rationalizing relative base-pairing
strengths, even in the case of structurally rather closely related oligonucleotide systems,
as long as experimental structural information about the single strands involved is
absent. One is quickly reminded of this complexity by inspecting b in Fig. 8, by which it

1286
Helvetica Chimica Acta Vol. 86 (2003)
becomes clear that there is no way to reliably predict by qualitative conformational
reasoning the type of conformation the phosphodiester group will assume in the C(2')-
domain in either a single- or a double-stranded a-arabinopyranosyl-(4' ! 2')-oligonu-
cleotide28). In fact, it is in this domain that the NMR structure analysis [8] in the
arabino series uncovered a large deviation of the phosphodiester torsion angles from
the values belonging to the corresponding (formally highly strained) idealized pairing
conformation [8]. It was rather surprising to find, in this analysis, that the type of
phosphodiester conformation adopted in the C(2')-domain of the arabinopyranosyl
duplex is the same as in the corresponding ribopyranosyl duplex. This can only mean
that a pentopyranosyl-(4' ! 2')-oligonucleotide strand has, in this domain, no
significantly more-stable alternative conformation available; if it had, and that
alternative would be a pairing conformation, the NMR structure analysis would have
revealed it; if, however, that alternative were not a pairing conformation, the actually
observed base-pairing strengths in a-arabinopyranosyl-(4' ! 2')-oligonucleotide du-
plexes could not possibly be as high as observed. We are led to conclude, therefore, that,
not only with respect to the domain of the C(4')-phosphodiester attachment, but also
with respect to the C(2')-domain, a-arabinopyranosyl-(4' ! 2')-oligonucleotide single
strands are pre-organized toward duplexation in spite of the pairing conformation in
the latter domain being highly strained. Such reasoning would lead to the expectation
that arabinopyranosyl duplexes irrespective of being more stable relative to their
respective single strands would have to be thermodynamically less stable against, e.g.,
strand scission compared to duplexes of the b-ribopyranosyl series. Indeed, a much
higher propensity of a-arabinopyranosyl strands toward strand scission under the basic
conditions of protection-group removal after solid-support synthesis has been one of
the specific observations made in the a-arabinopyranosyl-(4' ! 2')-oligonucleotide
series29).
The CD spectra reproduced in Fig. 5 refer to intersystem cross-pairing within the
pentopyranosyl-(4' ! 2')-oligonucleotide family. These spectra convincingly support
the assignments of pairing conformations for those two members of the family for
which no NMR structure analysis is available (Scheme 1). In full harmony with the
relationships expressed by these spectra, it is found that the combination of a-
arabinopyranosyl strands and complementary a-lyxopyranosyl strands in intersystem
cross-pairing leads to duplexes that are consistently more stable than the corresponding
combination between a-arabinopyranosyl strands and b-ribo- or b-xylopyranosyl
strands (Table 5, Fig. 6, and [3b]). In such intersystem cross-pairings between
homobasic purine and homobasic pyrimidine strands duplex stabilities can strongly
depend on which backbone bears the pyrimidine, and which bears the purine sequence.
This remarkable phenomenon has been discussed at some length in a previous
communication [3b].
Exceptionally strong base pairing such as that seen in a-arabinopyranosyl
oligonucleotides can be accompanied by the occurrence of strong self-pairing of
28
)
)
Quite in contrast to the a-lyxopyranosyl series where the (C(2')ÀO) torsion angles of both single and
double strands are (qualitatively) clearly predictable (Fig. 8,a).
Not surprisingly, comparable lability toward strand scission has been observed in the b-xylopyranosyl
series [1].
29

Helvetica Chimica Acta Vol. 86 (2003)
1287
nucleobases. Such self-pairing may be quite selective for a given purine or pyrimidine
base and be specific for a given type of backbone (see Table 1 in [3b]). For example,
whereas homo-adenine sequences undergo strong (Hoogsteen) self-pairing in the
homo-DNA series [28], no such pairing is observed in pyranosyl-RNA [4a], and almost
none in the a-arabinopyranosyl series (Table 4). The situation is more or less inverse
with regard to the self-pairing of thymine. The homo-pyrimidine sequence pa(T12)
shows the remarkable Tm value of 528 (10 mm, 1m NaCl), yet (fortunately) still
surpassed by the value (> 958) for the corresponding canonical A T pairing in pa(A12)/
pa(T12) (Table 4). Understandably, no such self-pairing seems to occur with cyto-
sine30). The possible self-pairing mode for thymine has been discussed in a previous
communication (see Footnote 12 in [3b]).
An ominous case of noncanonical self-pairing of non-self-complementary A,T-
containing strands was encountered in the behavior of the two antiparallel-comple-
mentary octamer sequence pa(TTAAAATA) and pa(TATTTTAA). Both undergo
strong self-association to duplexes, which have Tm values of ca. 488 under conditions
where the canonical cross-pairing duplex melts around 75831). This case may be taken
as a reminder of at least two questions that are of specific interest in the etiological
context, namely, the one concerning the relationship between base-pairing strength and
performance as an informational base-pairing system, and the other referring to the
limitations of an informational system that would have to operate with two instead of
four nucleobases. We have discussed the first question previously in the context of our
work on template-directed sequence copying in the ribopyranosyl-(4' ! 2')-oligonu-
cleotide series [4c][4e]. With regard to the two-instead-of-four nucleobase question
(see e.g., [29]), the case of self-association of the two-base arabinopyranosyl strands
mentioned above is just echoing a general experience we made in the course of our
extensive studies on pentopyranosyl-(4' ! 2')-oligonucleotides, namely: the higher the
base-pairing strengths in an oligonucleotide system, the more difficult it becomes to
construct (with only two nucleobases) a sequence that does not show self-pairing, even
when the sequence is very short (e.g., octamers). The task of constructing non-self-
associating sequences of corresponding lengths with four instead of only two
nucleobases is obviously much easier. This difference is assumed to be an important
aspect in any conceptual and/or experimental evaluation of potentially primordial
informational oligomers.
The work was supported by the Skaggs Foundation (TSRI). O. J., M. B., and A. L. thank the Deutsche
Forschungsgemeinschaft for fellowship support. O. J., M. B., A. L., and H. K. H were Skaggs postdoctoral
fellows. We thank Dr. A. Shivaniuk for help in the ORTEP drawings in Fig. 1.
30
)
)
Neither pa(C6) nor pa(G6) show discernable self-pairing under identical conditions (c ꢀ 10 mm in 10 mm
Na2HPO4 buffer, 0.1 mm Na2(EDTA), pH 7.0 with 1m NaCl). The self-pairing mode conceivable for
cytosine would require parallel strand orientation (forbidden in systems with large backbone inclination
[28]).
The duplex strand formed by the A-rich sequence is presumably held together by four A/T base pairs, as
well as four A/T and A/A interstrand base stacks.
31

1288
Helvetica Chimica Acta Vol. 86 (2003)
Experimental Part
General. Solvents for extraction: ACS grade. Solvents for reaction: reagent grade. Reagents: unless otherwise
noted, from Acros, Fluka, or Aldrich, highest quality available. Chloro(2-cyanoethoxy)(diisopropylamino)phos-
phine (97%) was purchased from CheM Impex Inc., Wood Dale, IL, USA. TLC: silica gel 60 F254 aluminum
plates (Whatman, Type Al Sil G/UV, 250 mm layer); visualization by UVabsorption and/or A) by dipping in a soln.
of 4-anisaldehyde /H2SO4/AcOH/EtOH 5 :5 :1:18 or B) cerium(IV) sulfate (3 mm)/ammonium molybdate
(250 mm) in aq. H2SO4 (10%), followed by heating. Flash column chromatography (CC) was performed on silica
gel 60 (0.40 0.63 mm, 230 440 mesh; EM Science) at low pressure (max. 2 bar). In case of acid-sensitive
compounds, the silica gel was pretreated with solvents containing ca. 0.5% Et3N. Melting points (uncorrected)
were measured with MEL-TEMP II (Laboratory Devices Inc., US A). NMR:1H: d values in ppm (TMSas
1
internal standard); J [Hz], assignments of H resonances were in some cases based on 2D experiments (1H,1H
COSY); 13C: d values in ppm (TMSas internal standard); J [Hz]; assignments and multiplicities were based on 2D
‡
experiments (1H,13C-COSY); 31P: d values in ppm (85% H3PO4 as external standard). FAB -MS(matrix-soln.):
m/z (intensity in %); performed in the positive-ion mode, with 3-nitrobenzyl alcohol (3-NBA) as matrix, on a VG
ZAB-VSE double-focusing high-resolution mass spectrometer equipped with a cesium ion gun. Matrix-assisted
laser-desorption-ionization time-of-flight mass spectrometry (MALDI-TOF-MS) was performed on a Voyager-
Elite mass spectrometer (Perseptive Biosystems) with delayed extraction with THAP or DHB as the matrix with
ammonium citrate added to the sample. Elemental analysis was performed with Perkin-Elmer PE2400 CHN
analyzer. Oligonucleotides were synthesized on an Expedite 8909 Nucleic Acid Synthesis system (Perseptive
Biosystems). HPLC: Ion exchange (IE) HPLC was performed on Pharmacia ækta Purifier (900) controlled by
UNICORN system. Columns: Mono Q HR 5/5 (Pharmacia) or SAX 1000-8 (Macherey & Nagel); Buffer A:
10 mm Na2HPO4 in H2O, pH 11.5; Buffer B: 10 mm Na2HPO4 in H2O, 1m NaCl, pH 11.5. UV Spectra: on a Cary 1
C spectrophotometer (Varian). Melting-point (Tm) measurements of oligonucleotides were determined with Cary
1 Bio spectrophotometer (Varian). CD Spectrum was measured on an AVIV 61 DS CD spectropolarimeter. All
measurements were made with the −phosphate buffer×, 10 mm aq. NaH2PO4 buffer containing 0.1 mm Na2EDTA,
150 mm (or 1m) NaCl at pH 7.0, with a total oligonucleotide concentration of ca. 10 mm, unless indicated otherwise,
and the samples were thoroughly degassed, either by heating or by vacuum and ultrasonication. Concentrations of
oligonucleotide solns. were calculated from the UV absorbance of the solns. at 260 nm (pH 7) at ca. 808 with the
following molar extinction coefficients: e(pa(A))ˆ 15000, e(pa(T))ˆ 10000, e(pa(C))ˆ 8400, e(pa(G))ˆ 11900.
Abbreviations: Ag-Tf ˆ silver trifluoromethanesulfonate, BOP: (benzotriazol-1-yl)oxytris(dimethylamino)phos-
phonium hexafluorophosphate; BSAˆ bis(trimethylsilyl)acetamide, CPGˆ −controlled-pore glass×, DHB ˆ 2,5-
dihydroxybenzoic acid, DIBAL-Hˆ diisobutylaluminium hydride, DMAP: 4-(dimethylamino)pyridine, DMT:
4,4'-dimethoxytrityl, Fmoc ˆ [(9H-Fluoren-9-yl)methoxy]carbonyl, 1-HOBTˆ 1-Hydroxy-1H-benzotriazole,
LCAA-CPG: long chain aminoalkyl-CPG (500 ä); LCAMA-CPG: long-chain aminomethylalkyl-CPG
(500 ä); PAC ˆ phenoxyacetyl, TMS-Tfˆ trimethylsilyl trifluoromethanesulfonate, THAPˆ 2,4,6-trihydroxy-
acetophenone, TOTU: O-{[(2-cyanoethoxycarbonyl)methylidene]amino}-1,1,3,3-tetramethyluronium tetrafluo-
roborate, TSOHˆ toluene-4-sulfonic acid, UF H2O ˆ ultra-filtered H2O.
1. Experiments Referring to Scheme 3. 1',2',3',4'-Tetra-O-benzoyl-a/b-l-arabinopyranose (1) [11]. To an
ice-cooled mixture of 600 ml of dry pyridine and 500 ml of dry CH2Cl2, over a period of 15 min, were added
200 ml (1.7 mol) of BzCl. To this mixture, 50 g (0.33 mol) of (‡)-l-arabinose was added in small portions over
5 h. The mixture was allowed to warm slowly to r.t., and stirred overnight. The syrupy mixture was poured into
1000 ml of ice H2O, the org. layer was separated and washed twice with 400 ml of sat. aq. NaHCO3 soln. The org.
layer was dried (Na2SO4) and evaporated, and the pale yellow oil was co-evaporated twice with 100 ml of
toluene. The residue was crystallized from 200 ml of hot EtOH. After storage at 48 overnight, the crystals were
filtered, washed with cold EtOH, Et2O and dried under high vacuum (0.03 Torr/r.t., 12 h) to yield 161.1 g (88 %)
1
of 1. Colorless crystals. M.p. 174 1768. TLC (hexane/AcOEt 3 :1): Rf 0.36. H-NMR (600 MHz, CDCl3): 4.20
(dd, J ˆ 2.0, 13.4, HÀC(5)); 4.42 (dd, J ˆ 0.8, 13.4, HÀC(5)); 5.92 (m, HÀC(4)); 6.0 (m, HÀC(3), HÀC(2));
6.88 (d, J ˆ 1.6, HÀC(1)); 7.32 (m, 4 arom. H); 7.49 (m, 2 arom. H); 7.54 (m, 4 arom. H); 7.66 (m, 2 arom. H);
7.90 (m, 4 arom. H); 8.15 (m, 4 arom. H). 13C-NMR (150 MHz, CDCl3): 63.42 (t, C(5')); 68.15, 68.51 (2d, C(3'),
C(2')); 69.88 (d, C(4')); 91.47 (d, C(1')); 128.81, 128.83, 129.04, 129.16, 129.30, 129.50, 129.72, 130.15, 130.31,
130.31, 133.83, 133.99, 134.26 (arom. C); 165.11, 165.99, 166.15, 166.17 (4s, CO of Bz). ESI-MS (pos., MeOH):
‡
‡
‡
‡
605 (5, [M ‡ K] ), 589 (100, [M ‡ Na] ), 467 (4, [M ‡ Na À OBz] ), 445 (19, [M À OBz] ), 323 (6,
‡
[M À 2OBz] ). Anal. calc. for C33H26O9: C 69.96, H 4.63; found: C 70.17, H 4.39.
N6-Benzoyl-9-(2',3',4'-tri-O-benzoyl-a-l-arabinopyranosyl)adenine (2a). To
a suspension of 24.9 g
(0.10 mol) of N6-benzoyladenine and 63.7 g (0.10 mol) of 1 in 450 ml of dry MeCN, 50 ml (0.2 mol, 2 equiv.)