Can Phosphate-Branched RNA Persist under Physiological Conditions?

Article abstract of DOI:10.1002/cbic.201200629

A 2′-O-methyl-RNA oligonucleotide containing a single free 2′-OH group flanking a branching phosphotriester linkage was prepared as a model for phosphate-branched RNA by using an orthogonally protected dimeric phosphoramidite building block in solid-phase

Full text of DOI:10.1002/cbic.201200629

DOI: 10.1002/cbic.201200629
Can Phosphate-Branched RNA Persist under Physiological
Conditions?
Tuomas A. Lçnnberg*[a]
A 2’-O-methyl-RNA oligonucleotide containing a single free 2’-
OH group flanking a branching phosphotriester linkage was
prepared as a model for phosphate-branched RNA by using an
orthogonally protected dimeric phosphoramidite building
block in solid-phase synthesis. The strategy allows the synthe-
sis of phosphate-branched oligonucleotides, the three branch-
es of which may be of any desired sequence. Hydrolytic reac-
tions of the phosphotriester linkages in such oligonucleotides
were studied at physiological pH in the presence (and ab-
sence) of various complementary oligonucleotides. The fully
hybridized oligonucleotide model is an order of magnitude
more stable than its single-stranded counterpart, which, in
turn, is an order of magnitude more stable than its trinucleo-
side phosphotriester core lacking any oligonucleotide arms.
Furthermore, kinked structures obtained by hybridizing the
phosphate-branched oligonucleotide with partially comple-
mentary oligonucleotides are three to five times more stable
than fully double-stranded ones and only approximately three
times less stable than the so-called RNA X structure, which has
been postulated to incorporate an RNA phosphotriester link-
age. The results indicate that when the intrinsically unstable
RNA phosphotriester linkage is embedded in an oligonucleo-
tide of appropriate tertiary structure, its half-life can be at least
several hours.
Introduction
Branched nucleic acids have been shown to be biologically rel-
evant species—lariat RNA, with oligonucleotide chains extend-
ing from all the three sugar hydroxy groups, for example, is
a stable and well-established product of the splicing of eukary-
otic pre-mRNA.[1] In principle, another mode of branching,
through a phosphotriester linkage, would also be available to
nucleic acids. In the case of RNA, such a linkage would be
inherently unstable owing to facile intramolecular attack of
the neighboring 2’-OH group on the neutral phosphate;[2,3]
the half-life of the trinucleoside-3’,3’,5’-monophosphate 1
(Scheme 1A) at 258C and pH 7.4, for example, is only approxi-
mately 3 min.[4]
An RNA phosphotriester linkage embedded within an oligo-
nucleotide structure can, however, be substantially more
stable,[5] even if the oligonucleotide does not form a well-
defined structure such as a double helix (Scheme 1B, com-
pound 2).[6] In fact, such a structure, the so-called RNA X
(Scheme 1C), has been reported as a product of a model reac-
tion of eukaryotic splicing. Whereas in the presence of all the
critical components of the spliceosome—that is, two of the
spliceosomal RNAs and two substrate RNAs resembling the 5’-
splice site and the branch site—a reaction mimicking the first
step of splicing takes place,[7] in the absence of the 5’-splice
site mimic the branch site nucleoside instead attacks one of
the spliceosomal RNAs, resulting in an anomalous X-shaped
oligonucleotide product, the so-called RNA X.[8,9] On the basis
of the iodoethanol cleavage patterns of phosphorothioate-sub-
stituted RNA X and the susceptibility of RNA X to alkaline hy-
drolysis, it was proposed that the two oligonucleotide strands
could be connected through a trinucleoside phosphotriester
linkage.
In the proposed RNA X structure, the phosphotriester link-
age is part of an unusual structural motif, connecting two
double-helical stems of a four-way junction.[8,9] A structure of
[a] Dr. T. A. Lçnnberg
Department of Chemistry, University of Turku
Vatselankatu 2, 20014 Turku (Finland)
E-mail: tuanlo@utu.fi
Scheme 1. A) Trinucleoside-3’,3’,5’-monophosphate model compound 1,
B) short phosphate-branched oligonucleotide model compound 2, and
C) the proposed trinucleoside phosphotriester linkage in RNA X.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201200629.
2690
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 2690 – 2700

Hydrolysis of Phosphate-Branched RNA
this kind could provide some additional stabilization by retard-
ing the formation or the pseudorotation of the pentacoordi-
nate phosphorane intermediate or by shielding the reaction
center from solvent. The reported half-life for the decomposi-
tion of RNA X at 508C and pH 9.0 is approximately 20 min,
almost 3000 times longer than extrapolation from the data ob-
tained for 1 would predict.[4,9]
To study the impact of various structural motifs on the hy-
drolytic stability of an RNA phosphotriester linkage, a phos-
phate-branched 2’-O-methyl-RNA oligonucleotide incorporat-
ing a single 2’-OH function protected with a removable pro-
tecting group (one flanking the scissile phosphotriester link-
age) was synthesized and its hydrolytic reactions were fol-
lowed at physiological pH in the presence and in the absence
of a number of (partly) complementary oligonucleotides. Be-
cause the reactive component of the model system is the
same in all cases, any changes in the reactivity should be due
either to increased entropic penalty for cleaving an internu-
cleosidic bond within a hybridized oligonucleotide or to
changes in the geometry or solvation of the active site.
Results
Scheme 2. Preparation of the branching phosphoramidite building block
8a. a) TBDMSCl, imidazole, DMF; b) TFA, H2O, THF; c) Lev2O, pyridine;
d) Et3N·3HF, THF; e) N,N,N’,N’-tetraisopropylchlorophosphorodiamidite, Et3N,
CH2Cl2; f) 2’-O-(tert-butyldithiomethyl)-5’-O-(4,4’-dimethoxytrityl)uridine,
5-benzylthio-1H-tetrazole, MeCN.
Preparation of the branching phosphoramidite building
block
As previously described for symmetric phosphate-branched oli-
gonucleotides, incorporation of the branching trinucleoside
phosphotriester linkage is conveniently achieved by coupling
of a dimeric phosphoramidite building block to the 5’-OH
group of a CPG-supported oligonucleotide.[10] In the present
case, however, the requirement for all of the oligonucleotide
branches to have unique sequences and the need for remova-
ble 2’-protection of one of the nucleosides flanking the phos-
photriester linkage present additional synthetic challenges. The
5’-OH groups of the two nucleosides have to be protected by
orthogonal protecting groups removable under conditions
under which the linker and the nucleobase protecting groups
are stable. Furthermore, the protection of the 2’-OH group of
one of the nucleosides must be not only orthogonal to both
of the 5’-protecting groups but also rapidly removable because
of the (presumed) high reactivity of the product. For the 5’-
protecting groups, 4,4’-dimethoxytrityl and levulinoyl groups
were chosen, whereas the 2’-position was protected with
a tert-butyldithiomethyl group, removable under reducing con-
ditions.[11]
ly.[11] During coupling in acetonitrile, decomposition of the
product to phosphorothioate-type side products, as reported
by Semenyuk et al.,[11] was observed but with use of the highly
efficient 5-benzylthio-1H-tetrazole as an activator, the coupling
could be performed in a short enough time to minimize this
side reaction. In the absence of acetonitrile, the final product
8a proved stable enough to be purified by silica gel chroma-
tography and stored for several months at ꢀ208C.
By a similar approach (Scheme 3), a 2’-tert-butyldimethylsilyl-
protected version (compound 8b) of the dimeric phosphorami-
dite was also prepared, but the TBDMS protection was found
to be too stable for the kinetic studies (data not shown). How-
ever, this building block was successfully employed for the syn-
The preparation of the branching phosphoramidite building
block is shown in Scheme 2. Firstly, both of the free hydroxy
functions of 2’-O-methyluridine were protected as tert-butyldi-
methylsilyl ethers. The 5’-protecting group was then selectively
removed under acidic conditions and the free 5’-OH group
was protected as a levulinoyl ester. Finally, the 3’-protecting
group was removed by conventional fluoride treatment, the
free 3’-OH group was phosphitylated with N,N,N’,N’-tetraiso-
propylchlorophosphorodiamidite, and the obtained phosphor-
odiamidite 7 was coupled with ’-O-(tert-butyldithiomethyl)-5’-
O-(4,4’-dimethoxytrityl)uridine, prepared as described previous-
Scheme 3. Preparation of the branching phosphoramidite building block
8b. a) N,N,N’,N’-tetraisopropylchlorophosphorodiamidite, Et3N, CH2Cl2; b) 6,
5-benzylthio-1H-tetrazole, MeCN.
ChemBioChem 2012, 13, 2690 – 2700
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
2691

T. A. Lçnnberg
thesis of a reference oligonucleotide incorporating a 3’,3’-phos-
phodiester linkage.
tional aqueous NH3 treatment for deprotection and release of
the oligonucleotide from the support.
For a reference for one of the expected cleavage products,
an oligonucleotide incorporating a 3’,3’-phosphodiester linkage
(compound 11, Scheme 5) was synthesized on a CPG-support-
ed 2-(2-hydroxyethylsulfonyl)ethyl succinate linker (a universal
linker designed for the synthesis of oligonucleotide-3’-mono-
phosphates)[12] with use of the 2’-O-TBDMS-protected branch-
ing phosphoramidite 8b as the first building block. In this
case, the coupling efficiency was 80%, with the subsequent
couplings proceeding with normal efficiency (>99%). Capping
of the first chain and assembly of the second one were then
carried out as described above for the branched oligonucleo-
tide 10a, with similar coupling yields. After deprotection of the
nucleobase(s) and release from the support, the 2’-O-TBDMS
protection was removed with the conventional Et3N·3HF treat-
ment and the product mixture was passed through a short RP
cartridge. In both cases, the final oligonucleotide product (10a
or 11) was purified by anion-exchange HPLC, followed by de-
salting by RP HPLC. Attempts to separate the two diastereo-
mers (RP and SP) of 10a were unsuccessful.
Oligonucleotide synthesis
The synthesis of the phosphate-branched oligonucleotide 10a
is outlined in Scheme 4. Coupling of the branching phosphora-
midite building block 8a to the 5’-OH group of the CPG-sup-
ported oligonucleotide was performed manually with 5-ben-
zylthio-1H-tetrazole as the activator. The coupling yield was
54%. The synthesis was then continued on an oligonucleotide
synthesizer by using the 5’-DMTr-protected nucleoside as the
starting point. The subsequent couplings proceeded with
normal efficiency (>99%). For elongation of the branch, the
5’-OH group of the support-bound main-chain oligonucleotide
was acetylated and the 5’-levulinoyl protection of the branch
site nucleoside was removed with hydrazine acetate. The
branch was then synthesized on an oligonucleotide synthesizer
and by using the free 5’-OH group as the starting point. To
avoid cleavage of the phosphodiester linkage, 40% MeNH2
(90 min at room temperature) was used instead of the conven-
Reaction pathways and
product distribution
The hydrolysis of the phos-
phate-branched oligonucleotide
10a was carried out at 258C
and pH 7.38 in a 50 mmolLꢀ1
tris(2-carboxyethyl)phosphine
(TCEP) buffer. Under these re-
ducing conditions, the disulfide
bond of the 2’-O-tert-butyldi-
thiomethyl protecting group is
rapidly cleaved, yielding the 2’-
thiohemiacetal counterpart 10b
(Scheme 6). The hydrolysis of
this intermediate, in turn, releas-
es the desired oligonucleotide
model compound 10c contain-
ing a free 2’-OH group flanking
the scissile phosphotriester link-
age.
The hydroxide-ion-catalyzed
cleavage of this phosphotriester
was studied by anion-exchange
HPLC with the branched oligo-
nucleotide 10c, either in single-
stranded form or hybridized
with various partially comple-
mentary
oligonucleotides
(Scheme 7). To minimize the hy-
drolysis of the starting material
during analysis, the HPLC runs
were performed at 258C, rather
than at a higher, denaturing
Scheme 4. Preparation of the phosphate-branched oligonucleotide 10a. a) i: 8a, 5-benzylthio-1H-tetrazole, MeCN,
ii: conventional Ac2O capping, iii: I2, 2,6-lutidine, H2O, THF; b) i: conventional phosphoramidite strategy, ii: conven-
tional Ac2O capping, iii: hydrazine, AcOH, pyridine; c) conventional phosphoramidite strategy; d) MeNH2, H2O.
temperature.
Whereas
the
2692
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 2690 – 2700

Hydrolysis of Phosphate-Branched RNA
Scheme 5. Preparation of the reference oligonucleotide 11 incorporating a 3’,3’-phosphodiester linkage. Reagents and conditions: a) i: compound 8b, 5-ben-
zylthio-1H-tetrazole, MeCN, ii: conventional Ac2O capping, iii: I2, 2,6-lutidine, H2O, THF; b) i: conventional phosphoramidite strategy, ii: conventional Ac2O cap-
ping, iii: hydrazine, AcOH, pyridine; c) conventional phosphoramidite strategy; d) NH3, H2O.
Scheme 6. Hydrolytic reactions of the phosphate-branched oligonucleotide 10a. In the oligonucleotide products, the lowercase letters refer to 2’-O-methylat-
ed nucleosides, whereas the uppercase “U” denotes the single nucleoside containing a free 2’-OH in each case.
longer oligonucleotides remained hybridized under these con-
ditions, the shorter products were well resolved, allowing relia-
ble analysis. With the fastest reaction studied (single-stranded
10c), approximately 25% of the starting material decomposed
during a single HPLC run but this loss was uniform for all of
the samples, not constituting a major source of inaccuracy. In
all the other cases, the decomposition of the starting material
during analysis was less than 3%.
products. With all the systems studied, two distinct sets of
shorter oligonucleotide products were indeed observed: 25-
mer and pentamer oligonucleotides (11 and 12, respectively),
resulting from the cleavage of the PꢀO5’ bond of the penta-
coordinate phosphorane intermediate, and 20-mer and deca-
mer oligonucleotides (13 and 14, respectively) resulting from
the cleavage of the PꢀO3’ bond (Scheme 6). Product distribu-
tions for the hydrolysis of 10c within various oligonucleotide
assemblies were determined by comparing the relative con-
centrations of the pentamer and decamer products 12 and 14.
The results, along with those previously obtained for the cor-
responding small phosphotriester model compound 1[4] and
the small chimeric phosphate-branched oligonucleotide 2
(Scheme 1),[6] are presented in Table 1.
From our previous experience with related phosphotriester
models, the reaction is first-order in hydroxide ion concentra-
tion under the conditions used and proceeds via a mono-
anionic
pentacoordinated
phosphorane
intermediate
(Scheme 6).[2–4,6,13] Unlike the dianionic phosphorane species
obtained in the hydroxide-ion-catalyzed cleavage of RNA phos-
phodiester bonds, this intermediate may pseudorotate to yield
both isomerization products and PꢀO3’ and PꢀO5’ cleavage
In contrast to previous results on related systems,[4,6] cleav-
age of the PꢀO5’ bond leading to an unnatural 3’,3’-phospho-
ChemBioChem 2012, 13, 2690 – 2700
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
2693

T. A. Lçnnberg
2694
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 2690 – 2700

Hydrolysis of Phosphate-Branched RNA
material to assume a favorable conformation for direct in-line
displacement of the 5’-linked nucleoside. As previously ob-
served for the small model compound 1,[4] the phosphate-
branched oligonucleotide 10c in all likelihood also undergoes
a very rapid isomerization to the 2’,3’,5’ isomer so that the 20-
mer and 25-mer oligonucleotide products 13 and 11 are also
actually mixtures of two isomers. The mobilities of these oligo-
nucleotides in anion-exchange HPLC were so similar, however,
that only a single peak for each of the products was detected.
In other words, although both the rate of isomerization and
the isomeric ratio probably also reflect changes in the secon-
dary structure of 10c, no data for this reaction could be ob-
tained.
Table 1. Second-order rate constants and product distributions for the
hydroxide-ion-catalyzed cleavage of the phosphotriester linkage of 10c
in the model assemblies of this study [T=258C, pH 7.4, I(NaCl)=
0.22 molLꢀ1]. For comparison, corresponding data for the related small
phosphotriester model compound 1 and the small chimeric phosphate-
branched oligonucleotide 2 (Scheme 1) are also included.
Model
kOH [Lmolꢀ1 sꢀ1
]
% (PꢀO5’)[a]
32[c]
9[d]
63
% (PꢀO3’)[b]
68[c]
8[d]
37
1
2
10c
10c:15
10c:16
10c:17a
10c:17b
10c:18a
10c:18b
7500[c]
620[d]
570ꢁ110
80ꢁ10
44ꢁ9
18ꢁ9
20ꢁ4
20ꢁ2
31ꢁ4
80
20
n.d.[e]
70
n.d.[e]
30
67
69
72
33
31
28
Hydrolytic stability of the phosphotriester linkage in various
oligonucleotide environments
[a] The ratio of PꢀO5’ bond fission to overall hydrolysis, determined as
the mole fraction of the pentamer oligonucleotide product 12. [b] The
ratio of PꢀO3’ bond fission to overall hydrolysis, determined as the mole
fraction of the decamer oligonucleotide product 14. [c] From ref. [4].
[d] From ref. [6]. [e] Could not be determined, due to strong association
of the decamer product oligonucleotide 14 with the complementary 15-
mer oligonucleotide 16.
The rates of cleavage of the phosphate-branched oligonucleo-
tide 10c, either single-stranded or hybridized with various
partly complementary oligonucleotides, at 258C and pH 7.38
were determined by monitoring the formation of the decamer
oligonucleotide product 14 by anion-exchange HPLC. In all
cases, the reactions followed first-order kinetics and the forma-
tion of 14 was accompanied by formation of the 20-mer prod-
uct 13 and the disappearance of 10c. Illustrative examples of
the time-dependent formation of 14 are presented in Figure 1.
As could be expected, the most reactive model system is
that consisting only of the single-stranded phosphate-
branched oligonucleotide 10c. The second-order rate constant
for the hydroxide-ion-catalyzed cleavage of the phosphotriest-
er linkage is essentially identical to that previously obtained
for the related but simpler model compound 2 (Scheme 1)
incorporating only uridine and thymidine residues (Table 1). In
other words, the (presumably) more extensive base-stacking of
10c does not confer any additional stabilization relative to its
all-pyrimidine counterpart 2. Both 10c and 2 are more than an
order of magnitude more stable than the small trinucleoside
monophosphate 1, however. In RNA X, the immediate sur-
roundings of the phosphotriester core are even more purine-
rich, possibly accounting for some of the greater stability of
this species.
diester linkage predominates in all of these model systems.
Somewhat unexpectedly, PꢀO5’ fission is most strongly favored
when the main chain of 10c (containing the sole nucleoside
with a free 2’-OH group) is fully hybridized with a complemen-
tary oligonucleotide (10c:15, Figure 1). In contrast, in the
cleavage of RNA phosphodiester linkages, PꢀO5’ fission is
retarded by a factor of 15 upon hybridization of the substrate
strand.[14] It should be borne in mind, however, that cleavage
of ribonucleoside-3’-phosphotriesters proceeds via relatively
stable monoanionic phosphorane intermediates that might
pseudorotate and, hence, expel any of the alkoxy ligands. The
dianionic intermediates of the cleavage of RNA phosphodiester
linkages, on the other hand, cannot pseudorotate, making the
reaction much more dependent on the ability of the starting
Hybridization of the main chain of 10c with the fully com-
plementary oligonucleotide 15 stabilizes the branching phos-
photriester linkage by an order of magnitude but, unexpected-
ly, cleavage of the PꢀO5’ bond is not specifically retarded. In
fact, the ratio of PꢀO5’ to PꢀO3’ fission is somewhat increased
on going from the single-stranded 10c to the double-stranded
10c:15 model. The PꢀO5’ bond is part of the double-helical
main chain of the 10c:15 assembly, whereas the PꢀO3’ bond
merely connects this double helix to a single-stranded deca-
mer oligonucleotide side arm. Cleavage of the former bond
would hence be expected to disrupt the base-stacking within
10c:15 much more than cleavage of the latter, yet this is not
reflected in the product ratio. When 10c is hybridized with the
shorter oligonucleotide 16, complementary to the decamer
rather than the 15-mer branch, a different model system is ob-
tained. Here the double helix consists solely of 2’-O-methylated
&
Figure 1. Time-dependent formation of 14 from single-stranded 10c ( ) and
&
*
*
from 10c hybridized with 15 ( ), 16 ( ), and 17a ( ). T=258C, pH 7.38,
I(NaCl)=0.22 molLꢀ1
.
ChemBioChem 2012, 13, 2690 – 2700
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
2695

T. A. Lçnnberg
nucleosides and the 2’-OH nucleophile is located in the first
nucleophile of the single-stranded sidearm. In this model, the
phosphotriester linkage is approximately twice as stable as in
10c:15. Unfortunately, because of the tendency of the deca-
mer oligonucleotide product 14 to remain hybridized with 16,
the product ratio for this system could not be determined reli-
ably.
gonucleotides react at essentially identical rates. Furthermore,
in the fully double-stranded model 10c:15, the PꢀO5’ bond
does not experience greater stabilization than the PꢀO3’ bond
(relative to single-stranded 10c), even though the former is
part of the backbone of a double helix and the latter is not.
The greater stabilities of the kinked structures 10c:17a,
10c:17b, 10c:18a, and 10c:18b relative to the fully double-
stranded 10c:15 are also not easily understood in terms of
base stacking alone. Finally, under the conditions used in this
study, the phosphodiester linkages of polyuridylic acid are
cleaved at the same rate as that of UpU, despite the much
more extensive base stacking of the former.[16]
In the proposed structure of RNA X, the putative phospho-
triester linkage is not part of a double helix but joins two
double-helical stem regions in a four-way junction motif.[8,9] To
simulate the effect of such an environment on the stability of
an RNA phosphotriester linkage, the hydrolysis of 10c was
studied in the presence of the partially complementary oligo-
nucleotides 17a, 17b, 18a, and 18b. Hybridization with 17a
or 17b places the scissile phosphotriester linkage in a kinked
position between two double-helical stems, whereas hybridiza-
tion with 18a or 18b results in a bulge containing the phos-
photriester within 10c. Interestingly, in all of these kinked
models, the phosphotriester linkage is more stable than those
in either of the fully double-helical models 10c:15 or 10c:16.
The greatest stabilization was observed with 10c:17a, which is
hydrolyzed nearly five times more slowly than the fully hybrid-
ized 10c:15 and more than 400 times more slowly than the
corresponding small trinucleoside monophosphate model
compound 1.
The quantitative kinetic data for the decomposition of
RNA X were obtained at 508C between pH 9 and 12.[9] Extrapo-
lation from these conditions to those used in this study is not
a straightforward task, due to the potential for disruption of
the secondary structure of RNA at such a high temperature
and pH values. If the dependence of rate on temperature is as-
sumed to be similar to that reported for corresponding phos-
phodiesters,[15] however, a half-life of approximately 2 h would
be obtained for RNA X at 258C and pH 9. In contrast with
simple phosphotriester models,[4,6] the data obtained for RNA X
between pH 9 and 12 would suggest a less than first-order de-
pendence of rate on hydroxide ion concentration and possibly
even a pH-independent reaction below pH 9.[9] Assuming the
reaction to be first-order in [OHꢀ] hence gives an upper limit
of 80 h for the half-life of RNA X at 258C and pH 7.4. Although
subject to marked uncertainty, this value would indicate that
the most stable of the phosphate-branched oligonucleotide
models (10c:17a) is only three times as reactive as RNA X; this
would lend further support to the proposal that the two
strands of RNA X are connected by a phosphotriester linkage.
As discussed above, hydroxide-ion-catalyzed cleavage of ri-
bonucleoside 3’-phosphotriesters proceeds via relatively stable
monoanionic phosphorane intermediates that may pseudoro-
tate. In other words, in-line arrangements of the nucleophile,
the phosphorus atom, and the leaving group in the initial
state are not required, unlike in the corresponding reactions of
RNA phosphodiester linkages. The formation of the mono-
anionic phosphorane intermediates might, however, be hin-
dered by unfavorable initial state geometries around the scis-
sile phosphotriester linkages. In particular, an orientation of
the phosphorus atom away from the flanking 2’-OH group
would be expected to retard the nucleophilic attack yielding
this intermediate significantly. Of the model systems studied,
the most stable one (10c:17a) places the scissile phospho-
triester linkage in a kinked position against a three-nucleotide
bulge, resembling the so-called K-turn motif.[17] In K-turn
motifs, the dinucleotide hinges joining the two double-helical
regions are characterized by unusually large e torsion angles of
up to 2708.[17,18] For comparison, the corresponding average
value for canonical A-type RNA is 2108. An e angle of 2708
translates into nearly the maximum possible distance between
the phosphorus atom of the scissile phosphotriester linkage
and the 2’-OH nucleophile (Figure 2), and this offers a possible
explanation for the observed stabilization, at least in the case
of model 10c:17a (and possibly also 10c:17b). It should be
noted that the branched oligonucleotide 10c used was a mix-
ture of RP and SP diastereomers and, in all likelihood, of 3’,3’,5’-
and 2’,3’,5’-isomers, all of which could react at different rates,
especially when hybridized with another oligonucleotide. Fur-
thermore, in RNA X the branching oligonucleotide is connect-
ed through a 2’-OH group located in the middle of the strand,
rather than a terminal 3’-OH (as is the case with 10c). Clearly,
Discussion
Perhaps the simplest explanation for the stabilization of a ribo-
nucleotide 3’-phosphotriester by an oligonucleotide environ-
ment is that cleavage of the phosphotriester linkage disrupts
base stacking of the oligonucleotide chain, leading to an ener-
getically less favorable reaction. If this were the case, however,
one would expect the longer phosphate-branched oligonu-
cleotide 10c, consisting of a mixture of purine and pyrimidine
bases, to be hydrolytically more stable than its relatively short
homopyrimidine counterpart 2, whereas in fact these two oli-
Figure 2. The phosphotriester branch of 10c: A) in the canonical A-type con-
formation, and B) as part of a K-turn structural motif. In each case only the
SP diastereomer is shown.
2696
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 2690 – 2700

Hydrolysis of Phosphate-Branched RNA
care should be exercised when interpreting these results in
terms of the stability of RNA X. We would like to suggest, how-
ever, that the extra bulk of the fourth oligonucleotide strand
could well serve to hold the phosphotriester linkage away
from the 2’-OH nucleophile, resulting in additional stabilization
relative to 10c.
right as an intermediate in biologically relevant phosphate-
transfer reactions.
Experimental Section
General: The nucleosides, protected nucleoside phosphoramidites,
and other reagents were commercial products that were used as
received. NMR spectra were recorded with Bruker Avance 400 or
500 NMR spectrometers and mass spectra with a Bruker MicroTOF-
Q high-resolution ESI mass spectrometer. The oligonucleotides
were assembled on an Applied Biosystems 3400 DNA synthesizer.
The original RNA X structure was only obtained in the pres-
ence of Mg2+ (or, to a lesser extent, Ca2+).[8,9] Undoubtedly, the
catalytically important metal ions interact closely with the
components forming the unusual phosphotriester structure
and are very likely coordinated to the branching phosphate
itself, as well as the flanking 2’-OH group. Such coordination
has been directly observed in the case of group II introns[19]
and has also been proposed, on the basis of thio effects, in the
case of the spliceosome.[20] Both of these structures promote
splicing reactions, in which external nucleophiles are able to
outcompete the neighboring 2’-OH groups in attacking the
phosphorus atoms of scissile phosphodiester linkages.[21] It
seems likely that the metal ions at the reaction centers play
a role in preventing nucleophilic attack by the flanking 2’-OH
groups either by orienting them away from the scissile phos-
phate groups or by reducing their nucleophilicities. Such con-
tributions by the Mg2+ or Ca2+ ions mandatory for the forma-
tion of RNA X might explain the somewhat higher stability of
this unusual species relative to the phosphate-branched
models presented in this paper.
Kinetic measurements: Hydrolytic reactions of the phosphate-
branched oligonucleotide 10a were carried out in sealed tubes im-
mersed in a water bath, the temperature of which was adjusted to
(25.0ꢁ0.1)8C. Firstly, the starting material and any additional oligo-
nucleotides were dissolved in aq. NaCl (0.10 molLꢀ1) and allowed
to hybridize for 15 min before the start of the reaction. An equal
amount of TCEP buffer (100 mm, pH 7.38, I=0.22 molLꢀ1) was then
added to this solution. The initial concentrations of the oligonucle-
otides were 3.0 mmolLꢀ1. The compositions of samples withdrawn
at appropriate time intervals were analyzed by anion-exchange
HPLC on a Dionex DNA Swift SAX-1S column (5ꢁ150 mm, mono-
lithic), with elution with a linear gradient of NaClO4 (0.33 molLꢀ1
,
10!60% over 25 min, flow rate 1.5 mLminꢀ1) in Tris buffer
(20 mmolLꢀ1, pH 7.0). The observed retention times (tR [min]) for
the hydrolytic products of 10a and the complementary oligonucle-
otides used in this study were as follows: 15.1 (10c), 12.1 (11), 12.1
(17b), 11.0 (17a), 10.9 (13), 10.5 (15), 9.7 (18a), 9.6 (16), 8.7 (18b),
5.5 (14), 2.1 (12). The products were characterized by spiking with
authentic samples and the peak areas were converted to relative
concentrations by using molar absorption coefficients calculated
by an implementation of the nearest-neighbors method. In each
case the ratio of PꢀO3’ to PꢀO5’ bond cleavage was obtained as
the ratio of the relative concentrations of the corresponding deca-
mer and pentamer oligonucleotide products 14 and 12.
Conclusions
An orthogonally protected dimeric phosphoramidite building
block for the synthesis of phosphate-branched oligonucleo-
tides of arbitrary sequence has been prepared and employed
in the synthesis of oligonucleotide models for the so-called
RNA X structure. Embedding the inherently unstable trinucleo-
side phosphotriester as a branching unit within an oligonu-
cleotide structure stabilizes it towards hydrolysis by an order
of magnitude. Hybridization of such a branched oligonucleo-
tide with a fully complementary oligonucleotide results in fur-
ther stabilization, by another order of magnitude. Interestingly,
hybridization with oligonucleotides through which the branch-
ing phosphotriester linkage is placed in a kinked environment,
rather than a double-stranded position, yields structures that
are moderately more stable than their fully double-stranded
counterparts. The half-life of the most stable of such structures
(at 258C and pH 7.38) is more than 24 h, whereas under the
same conditions the half-life of the trinucleoside phospho-
triester branching unit lacking any oligonucleotide arms is only
approximately 3 min. When extrapolated to the conditions
used for the kinetic analysis of the hydrolysis of RNA X (508C
and pH 9), the former value is only approximately three times
shorter than the corresponding value for RNA X (6 and 20 min,
respectively). The proposal that the two oligonucleotide
strands of RNA X are connected by a phosphotriester linkage
would hence seem plausible in view of the elaborate tertiary
structure of this unusual species. On a more general level the
results indicate that, suitably folded, phosphate-branched RNA
can be surprisingly stable and should not be dismissed out-
Under the reducing conditions of the reaction buffer, three consec-
utive reactions take place. Firstly, the disulfide bond of the 2’-O-
tert-butyldithiomethyl protecting group is broken in a reaction that
is too rapid to be followed by HPLC. The hydrolysis of the resulting
2’-thiohemiacetal is a much slower process, with a half-life of ap-
proximately 90 min under the experimental conditions. Finally, the
free 2’-OH group attacks the phosphorus atom in the scissile phos-
photriester linkage, initiating the reaction of interest. With single-
stranded 10c, this reaction is faster than the preceding deprotec-
tion step, so the rate law for consecutive first-order reactions had
to be applied to obtain the desired pseudo first-order rate con-
stant. In all the other cases the cleavage of 10c was sufficiently
slow that the simple first-order rate law could be used instead.
With each system except 10c:16, the pseudo first-order rate con-
stant for the cleavage of 10c was obtained by applying the appro-
priate rate law to the time-dependent increase in the relative con-
centration of the decamer oligonucleotide product 14. With
10c:16, this product remained hybridized to the complementary
oligonucleotide 16, precluding reliable determination of its concen-
tration. In this case, the rate constant was based on the formation
of the pentamer oligonucleotide product 12 instead. From our ear-
lier experience with the reactions of ribonucleoside 3’-phospho-
triesters, isomerization of 10c is in all likelihood a very rapid reac-
tion that reaches equilibrium even before the first sample is with-
drawn.[2,4] Because of the similar retention times and the general
broadness of the HPLC peaks of all the phosphate-branched oligo-
nucleotide species, this reaction could not be detected. For the
ChemBioChem 2012, 13, 2690 – 2700
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
2697

T. A. Lçnnberg
same reasons, the observed rate constants could not be broken
down to individual contributions of RP and SP diastereomers of
10c.
EtOAc/CH2Cl2 (1:79:20, v/v/v) and finally with Et3N/MeOH/CH2Cl2
(1:9:90, v/v/v). Yield: 0.2673 g (25%) as a 1:1 mixture of RP and SP
diastereomers (the absolute configuration at phosphorus was not
established). HRMS (ESI+): m/z calcd: 1188.4045 [M+Na]+; found:
1188.3976.
3’-O-(tert-Butyldimethylsilyl)-5’-O-levulinoyl-2’-O-methyluridine
(5): Levulinic acid (4.68 g, 40.30 mmol) and DCC (4.18 g,
20.26 mmol) were dissolved in anhydrous dioxane and the result-
ing mixture was stirred at room temperature for 2 h, after which it
was filtered, and the filtrate was added to a solution of 3’-O-(tert-
butyldimethylsilyl)-2’-O-methyluridine (1.91 g, 5.13 mmol) in anhy-
drous pyridine (20 mL). The reaction mixture was then stirred at
room temperature for 16 h, after which it was concentrated to dry-
ness. The residue was dissolved in CH2Cl2 (100 mL) and washed
with saturated aq. NaHCO3 (100 mL). The aqueous phase was back-
extracted with CH2Cl2 (100 mL), and the combined organic phases
were dried with Na2SO4 and concentrated to dryness. The crude
product was purified by silica gel chromatography with elution
with a mixture of MeOH and CH2Cl2 (4:96, v/v). Yield: 1.65 g (75%);
1H NMR (CDCl3, 400 MHz): d=9.12 (brs, 1H), 7.72 (d, J=8.2 Hz,
1H), 5.85 (d, J=2.2 Hz, 1H), 5.78 (d, J=8.2 Hz, 1H), 4.42 (dd, J1 =
2.4 Hz, J2 =12.6 Hz, 1H), 4.32 (dd, J1 =3.2 Hz, J2 =12.6 Hz, 1H), 4.20
(m, 1H), 4.15 (dd, J1 =5.0 Hz, J2 =7.6 Hz, 1H), 3.72 (dd, J1 =2.2 Hz,
J2 =4.8 Hz), 3.56 (s, 3H), 2.81 (ddd, J1 =6.2 Hz, J2 =J3 =2.9 Hz, 2H),
2.57 (ddd, J1 =6.2 Hz, J2 =1.5 Hz, J3 =2.1 Hz, 2H), 2.20 (s, 3H), 0.90
(s, 9H), 0.10 (s, 3H), 0.09 ppm (s, 3H); 13C NMR (CDCl3, 100 MHz):
d=206.2, 172.2, 163.2, 150.0, 139.8, 102.3, 88.7, 83.4, 81.2, 69.6,
62.2, 58.5, 37.8, 30.7, 29.7, 25.6, 18.1, ꢀ4.7, ꢀ5.1 ppm; HRMS (ESI+):
m/z calcd: 493.1977 [M+Na]+; found: 493.1952.
Diastereomer A: 1H NMR (CDCl3, 500 MHz): d=7.99 (d, J=8.2 Hz,
1H), 7.55 (d, J=8.2 Hz, 1H), 7.45–7.21 (m, 9H), 6.85 (m, 4H), 6.11
(d, J=4.1 Hz, 1H), 5.83 (d, J=2.4 Hz, 1H), 5.74 (d, J=8.2 Hz, 1H),
5.26 (d, J=8.1 Hz, 1H), 5.07 (d, J=11.2 Hz, 1H), 4.98 (d, J=11.3 Hz,
1H), 4.81 (ddd, J1 =5.0 Hz, J2 =4.7 Hz, J3 =9.1 Hz, 1H), 4.58 (dd, J1 =
4.5 Hz, J2 =4.6 Hz, 1H), 4.38 (m, 1H), 4.31 (m, 1H), 4.16 (m, 1H),
4.14 (m, 1H), 4.09 (ddd, J1 =5.2 Hz, J2 =3.4 Hz, J3 =8.5 Hz, 1H), 4.01
(dd, J1 =2.6 Hz, J2 =4.6 Hz, 1H), 3.80 (s, 3H), 3.80 (s, 3H), 3.56 (m,
2H), 3.51 (m, 2H), 3.45 (s, 3H), 2.77 (m, 2H), 2.52 (m, 2H), 2.18 (s,
3H), 1.31 (s, 9H), 1.20 (d, J=6.6 Hz, 6H), 1.15 ppm (d, J=6.7 Hz,
6H); 13C NMR (CDCl3, 125 MHz): d=206.2, 172.1, 163.7, 163.5,
158.7, 150.3, 150.1, 144.3, 140.5, 139.7, 135.3, 135.0, 130.3, 130.2,
130.2, 128.3, 128.2, 128.1, 128.0, 127.2, 113.4, 113.2, 113.2, 102.4,
102.3, 88.8, 87.0, 83.6, 82.3, 81.2, 80.6, 78.2, 70.4, 68.4, 62.4, 61.7,
58.4, 55.3, 55.2, 47.4, 44.5, 37.7, 29.8, 29.8, 27.7, 24.7, 24.7 ppm;
31P NMR (CDCl3, 202 MHz): d=148.9 ppm.
Diastereomer B: 1H NMR (CDCl3, 500 MHz): d=7.81 (d, J=8.2 Hz,
1H), 7.66 (d, J=8.2 Hz, 1H), 7.45–7.21 (m, 9H), 6.85 (m, 4H), 6.13
(d, J=6.0 Hz, 1H), 6.02 (d, J=4.0 Hz, 1H), 5.91 (d, J=8.2 Hz, 1H),
5.33 (d, J=8.2 Hz, 1H), 4.96 (m, 2H), 4.61 (dd, J1 =5.1 Hz, J2 =
5.7 Hz), 4.46 (m, 3H), 4.36 (m, 1H), 4.19 (m, 1H), 4.16 (m, 1H), 3.88
(dd, J1 =4.5 Hz, J2 =4.8 Hz, 1H), 3.81 (s, 3H), 3.80 (s, 3H), 3.56 (m,
2H), 3.51 (s, 3H), 3.49 (m, 1H), 3.39 (m, 1H), 2.72 (m, 2H), 2.61 (m,
2H), 2.18 (s, 3H), 1.29 (s, 9H), 1.15 (d, J=6.7 Hz, 6H), 1.06 (d, J=
6.7 Hz, 6H); 13C NMR (CDCl3, 125 MHz): d=206.5, 172.4, 163.5,
163.4, 158.7, 150.3, 150.2, 144.0, 140.0, 139.9, 135.1, 134.9, 130.3,
130.2, 130.2, 128.3, 128.2, 128.1, 128.0, 127.2, 113.4, 113.2, 113.2,
103.0, 102.7, 87.9, 86.2, 83.9, 82.4, 81.8, 81.5, 79.0, 71.0, 69.5, 63.3,
62.9, 58.4, 55.3, 55.2, 47.7, 46.0, 43.3, 37.8, 29.8, 29.7, 27.9, 24.6,
24.5 ppm; 31P NMR (CDCl3, 202 MHz): d=150.1 ppm.
5’-O-Levulinoyl-2’-O-methyluridine (6): Compound
5 (1.65 g,
3.51 mmol) was dissolved in THF (41.25 mL). Triethylamine trihy-
drofluoride (2.285 mL, 14.02 mmol) was added, and the resulting
mixture was stirred at room temperature for 64 h, after which it
was concentrated under reduced pressure. The crude product was
purified by silica gel chromatography with elution with MeOH/
CH2Cl2 (3:97, v/v). Yield: 1.12 g (89%); 1H NMR (CDCl3, 500 MHz):
d=8.39 (brs, 1H), 7.73 (d, J=10.2 Hz, 1H), 5.91 (d, J=2.5 Hz, 1H),
5.79 (dd, J1 =10.3 Hz, J2 =2.9 Hz, 1H), 4.46 (brs, 1H), 4.45 (brs, 1H),
4.18 (m, 1H), 4.13 (m, 1H), 3.85 (dd, J1 =6.2 Hz, J2 =2.6 Hz, 1H),
3.65 (s, 3H), 2.83 (ddd, J1 =7.5 Hz, J2 =2.8 Hz, J3 =1.7 Hz, 2H), 2.72
(d, J=9.9 Hz, 1H), 2.56 (ddd, J1 =8.0 Hz, J2 =2.4 Hz, J3 =1.5 Hz, 2H),
2.21 ppm (s, 3H); 13C NMR (CDCl3, 125 MHz): d=206.6, 172.4, 162.7,
149.8, 139.5, 102.5, 87.8, 83.3, 81.7, 68.6, 62.4, 58.8, 37.9, 29.8,
27.8 ppm; HRMS (ESI+): m/z calcd: 379.1112 [M+Na]+; found:
379.1107.
2’-O-(tert-Butyldimethylsilyl)-5’-O-(4,4’-dimethoxytrityl)uridin-3’-
yl 5’-O-levulinoyl-2’-O-methyluridin-3’-yl N,N-diisopropylphos-
phoramidite (8b): 5’-O-(4,4’-Dimethoxytrityl)-2’-O-(tert-butyldime-
thylsilyl)uridine (0.870 g, 1.32 mmol) was dissolved in anhydrous
CH2Cl2 (10 mL). Anhydrous triethylamine (1.0 mL) and N,N,N’,N’-tet-
raisopropylchlorophosphorodiamidite (0.420 g, 1.57 mmol) were
added, and the resulting mixture was stirred at room temperature
for 4 h, after which it was diluted with CH2Cl2 (100 mL) and washed
with saturated aq. NaHCO3 (100 mL). The organic phase was dried
with Na2SO4 and concentrated to dryness. The residue and 5’-O-lev-
ulinoyl-2’-O-methyluridine (6, 0.330 g, 0.926 mmol) were dissolved
under nitrogen in anhydrous MeCN (2.0 mL). A solution of 1H-tetra-
zole in MeCN (0.45m, 2.05 mL, 0.923 mmol) was added, and the re-
sulting mixture was stirred at room temperature for 16 h, after
which it was suspended in saturated aq. NaHCO3 (100 mL). The
aqueous mixture was extracted with CH2Cl2 (2ꢁ100 mL), and the
organic layer was dried with Na2SO4 and concentrated to dryness.
The residue was purified by silica gel chromatography with elution
first with Et3N/EtOAc/CH2Cl2 (1:79:20, v/v/v) and finally with Et3N/
MeOH/CH2Cl2 (1:6:93, v/v/v). Yield: 0.2866 g (27%) as a 4:1 mixture
of RP and SP diastereomers (the absolute configuration at phospho-
rus was not established). HRMS (ESI+): m/z calcd: 1168.4686
[M+Na]+; found: 1168.4540.
2’-O-(tert-Butyldithiomethyl)-5’-O-(4,4’-dimethoxytrityl)uridin-3’-
yl 5’-O-levulinoyl-2’-O-methyluridin-3’-yl N,N-diisopropylphos-
phoramidite (8a): Compound 6 (0.6319 g, 1.773 mmol) was coeva-
porated from anhydrous pyridine (3ꢁ10 mL) and the residue was
dissolved in anhydrous CH2Cl2. Anhydrous triethylamine (1.2 mL)
and N,N,N’,N’-tetraisopropylchlorophosphorodiamidite (0.5257 g,
1.970 mmol) were added, and the resulting mixture was stirred at
room temperature for 4 h, after which it was diluted with CH2Cl2
(100 mL) and washed with saturated aq. NaHCO3 (100 mL). The or-
ganic phase was dried with Na2SO4 and concentrated to dryness.
The residue and 2’-O-(tert-butyldithiomethyl)-5’-O-(4,4’-dimethoxy-
trityl)uridine[11] (0.6025 g, 0.885 mmol) were dissolved under nitro-
gen in a solution of 5-benzylthio-1H-tetrazole in MeCN (0.25m,
3.3 mL). The reaction mixture was stirred at room temperature for
3 h, after which it was diluted with CH2Cl2 (100 mL) and washed
with saturated aq. NaHCO3 (100 mL). The aqueous phase was back-
extracted with CH2Cl2, and the combined organic phases were
dried with Na2SO4 and concentrated to dryness. The residue was
purified by silica gel chromatography with elution first with Et3N/
Major diastereomer: 1H NMR (CDCl3, 500 MHz): d=7.68 (d, J=
8.2 Hz, 1H), 7.55 (d, J=8.2 Hz, 1H), 7.38–7.22 (m, 9H), 6.90 (m, 4H),
5.93 (d, J=5.2 Hz, 1H), 5.77 (d, J=4.4 Hz, 1H), 5.69 (d, J=8.1 Hz,
2698
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 2690 – 2700

Hydrolysis of Phosphate-Branched RNA
1H), 5.34 (d, J=8.1 Hz, 1H), 4.53 (ddd, J1 =4.5 Hz, J2 =5.4 Hz, J3 =
10.0 Hz, 1H), 4.40 (dd, J1 =J2 =5.0 Hz, 1H), 4.35 (m, 1H), 4.25 (dd,
J1 =3.3 Hz, J2 =12.5 Hz, 1H), 4.18 (ddd, J1 =5.3 Hz, J2 =5.0 Hz, J3 =
9.9 Hz, 1H), 4.11 (dd, J1 =4.1 Hz, J2 =12.4 Hz, 1H), 3.98 (m, 1H),
3.93 (dd, J1 =J2 =4.7 Hz, 1H), 3.79 (s, 3H), 3.78 (s, 3H), 3.61 (m, 1H),
3.52 (dd, J1 =4.4 Hz, J2 =11.0 Hz, 1H), 3.44 (dd, J1 =2.2 Hz, J2 =
10.3 Hz, 1H), 3.37 (s, 3H), 2.74 (dd, J1 =6.1 Hz, J2 =6.7 Hz, 2H),
2.56–2.48 (m, 2H), 2.13 (s, 3H), 1.19 (d, J=6.8 Hz, 6H), 1.15 (d, J=
6.8 Hz, 6H), 0.93 (s, 9H), 0.16 (s, 3H), 0.14 ppm (s, 3H); 13C NMR
(CDCl3, 125 MHz): d=206.7, 172.4, 163.0, 162.9, 158.8, 158.7, 150.6,
150.2, 145.1, 140.2, 139.8, 135.5, 135.4, 130.3, 130.1, 128.0, 127.9,
127.0, 113.1, 102.1, 101.9, 88.2, 87.2, 83.5, 81.9, 80.9, 74.9, 73.0,
70.2, 68.4, 63.4, 62.8, 58.0, 54.9, 54.9, 43.1, 37.4, 28.9, 27.6, 25.2,
24.2, 24.1, ꢀ5.2, ꢀ5.5 ppm; 31P NMR (CDCl3, 202 MHz): d=
149.7 ppm.
Synthesis of the reference oligonucleotide 11 incorporating
a 3’,3’-phosphodiester linkage: The branching phosphoramidite
building block 8b (20 mg, 17 mmol) and CPG-supported 2-(2-hy-
droxyethylsulfonyl)ethyl succinate linker with a free OH terminus
(1 mmol, loading=38 mmolgꢀ1) were dissolved in an anhydrous so-
lution of 5-benzylthio-1H-tetrazole in MeCN (0.25 molLꢀ1, 200 mL,
50 mmol). The resulting mixture was shaken for 50 min, after which
the support was transferred to an oligonucleotide synthesizer
column and washed with MeCN. The support-bound oligonucleo-
tide was then capped by conventional Ac2O treatment, washed
with MeCN, oxidized by conventional I2/H2O treatment, and
washed again with MeCN. Assembly of the longer 3’!5’ strand
was completed on an oligonucleotide synthesizer by the standard
phosphoramidite strategy for RNA (600 s coupling time) and the
5’-OH terminus was capped by conventional Ac2O treatment. For
elongation of the shorter 3’!5’ strand, the 5’-O-levulinoyl protec-
tion was first removed by treating the support-bound oligonucleo-
tide with hydrazine hydrate/pyridine/acetic acid (1:32:8, v/v/v) for
30 min. The support was then washed with pyridine, MeOH,
CH2Cl2, and Et2O, and assembly of the shorter strand was complet-
ed on an oligonucleotide synthesizer by the standard phosphora-
midite strategy for RNA. The oligonucleotide was released from
the support, and the nucleobase and phosphate protecting groups
were removed by treating the support with aq. MeNH2 (40%) for
90 min at room temperature. The single 2’-O-TBDMS protection
was removed by conventional Et3N·3HF treatment and the crude
product was purified first with a Glen Research Poly-Pak II RP car-
tridge, followed by anion-exchange HPLC on a Dionex DNA Swift
SAX-1S column (5ꢁ150 mm, monolithic), with elution with a linear
gradient of NaClO4 (0.33 molLꢀ1, 10!100% in 30 min, flow rate=
1.5 mLminꢀ1) in Tris buffer (pH 7.0, 20 mmolLꢀ1). Finally, the puri-
fied product was desalted by RP HPLC on a Thermo ODS Hypersil
column (4.6ꢁ250 mm, 5 mm) with elution with a linear gradient of
MeCN in water (0!70% in 15 min, flow rate=1.0 mLminꢀ1). An
HPLC trace of the crude product mixture is included in the Sup-
Minor diastereomer: 1H NMR (CDCl3, 500 MHz): d=7.69 (d, J=
8.2 Hz, 1H), 7.60 (d, J=8.2 Hz, 1H), 7.38–7.22 (m, 9H), 6.90 (m, 4H),
6.01 (d, J=6.9 Hz, 1H), 5.98 (d, J=5.0 Hz, 1H), 5.82 (d, J=8.2 Hz,
1H), 5.50 (d, J=8.1 Hz, 1H), 4.51–4.44 (m, 2H), 4.35 (m, 1H), 4.29
(dd, J1 =4.3 Hz, J2 =12.3 Hz, 1H), 4.26–4.16 (m, 2H), 3.98 (m, 1H),
3.98, (dd, J1 =J2 =5.0 Hz, 1H), 3.80 (s, 3H), 3.79 (s, 3H), 3.61 (m,
1H), 3,52–3.46 (m, 1H), 3.45 (s, 3H), 3.40–3.33 (m, 1H), 2.77 (m,
2H), 2.57 (m, 2H), 2.13 (s, 1H), 1.21 (d, J=6.9 Hz, 6H), 1.06 (d, J=
6.8 Hz, 6H), 0.91 (s, 9H), 0.15 (s, 3H), 0.10 ppm (s, 3H); 13C NMR: re-
liable assignment of this spectrum was not possible owing to the
small amount of the minor isomer; 31P NMR (CDCl3, 202 MHz): d=
149.4 ppm;
Synthesis of the phosphate-branched oligonucleotide 10a: The
branching phosphoramidite building block 8a (20 mg, 17 mmol)
and the CPG-supported pentamer oligonucleotide with a free 5’-
OH terminus (1 mmol, loading=37 mmolgꢀ1) were dissolved in an
anhydrous solution of 5-benzylthio-1H-tetrazole in MeCN
(0.25 molLꢀ1, 200 mL, 50 mmol). The resulting mixture was shaken
for 50 min, after which the support was transferred to an oligonu-
cleotide synthesizer column and washed with MeCN. The support-
bound oligonucleotide was then capped by conventional Ac2O
treatment, washed with MeCN, oxidized by the conventional I2/H2O
treatment, and washed again with MeCN. Assembly of the main
chain was completed on an oligonucleotide synthesizer by the
standard phosphoramidite strategy for RNA (600 s coupling time)
and the 5’-OH terminus was capped by conventional Ac2O treat-
ment. For elongation of the branching chain, the 5’-O-levulinoyl
protecting group was first removed by treating the support-bound
oligonucleotide with a mixture of hydrazine hydrate, pyridine, and
acetic acid (1:32:8, v/v) for 30 min. The support was then washed
with pyridine, MeOH, CH2Cl2, and Et2O, and assembly of the
branching chain was completed on an oligonucleotide synthesizer
by the standard phosphoramidite strategy for RNA. The oligonu-
cleotide was released from the support, and the nucleobase and
phosphate protecting groups were removed by treating the sup-
port with aq. MeNH2 (40%) at room temperature for 90 min. The
crude product was purified by anion-exchange HPLC on a Dionex
DNA Swift SAX-1S column (5ꢁ150 mm, monolithic), with elution
with a linear gradient of NaClO4 (0.33 molLꢀ1, 10!60% in 25 min,
flow rate=1.5 mLminꢀ1) in Tris buffer (pH 7.0, 20 mmolLꢀ1). Finally,
the purified product was desalted by RP HPLC on a Thermo ODS
Hypersil column (4.6ꢁ250 mm, 5 mm) with elution with a linear
gradient of MeCN in water (0!70% in 15 min, flow rate=
1.0 mLminꢀ1). An HPLC trace of the crude product mixture is
included in the Supporting Information. HRMS (ESIꢀ): m/z calcd:
1010.1836 [Mꢀ10H]10ꢀ; found: 1010.1830.
porting Information. HRMS (ESIꢀ): m/z calcd: 834.6485 [Mꢀ10H]10ꢀ
;
found: 834.8467.
Acknowledgements
The technical assistance of Kiira Rimpilꢀ in the synthesis of the
protected nucleosides is gratefully acknowledged.
Keywords: branched RNA · hydrolysis · mRNA splicing ·
phosphotriesters · RNA structures · RNA X
[1] a) Z. Zhou, L. J. Licklider, S. P. Gygi, R. Reed, Nature 2002, 419, 182–185;
b) C. L. Will, R. Lꢂhrmann, Cold Spring Harbor Perspectives in Biology,
Cold Spring Harbor Laboratory Press, New York, 2010; c) J. Rappsilber,
U. Ryder, A. I. Lamond, M. Mann, Genome Res. 2002, 12, 1231–1245;
d) C. B. Burge, T. H. Tuschl, P. A. Sharp in RNA World II (Eds.: R. F. Geste-
land, T. R. Cech, J. F. Atkins), Cold Spring Harbor Laboratory Press, New
York, 1998, pp. 525–560.
[2] M. Kosonen, R. Seppꢃnen, O. Wichmann, H. Lçnnberg, J. Chem. Soc.
Perkin Trans. 2 1999, 2433–2439.
[3] S. Mikkola, M. Kosonen, H. Lçnnberg, Curr. Org. Chem. 2002, 6, 523–
538.
[4] T. Lçnnberg, J. Kiiski, S. Mikkola, Org. Biomol. Chem. 2005, 3, 1089–
1096.
[5] B. Singer, L. Sun, H. Fraenkel-Conrat, Proc. Natl. Acad. Sci. USA 1975, 72,
2232–2236.
[6] T. Lçnnberg, Nucleosides Nucleotides Nucleic Acids 2006, 25, 315–323.
ChemBioChem 2012, 13, 2690 – 2700
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
2699

T. A. Lçnnberg
[7] S. Valadkhan, A. Mohammadi, C. Wachtel, J. L. Manley, RNA 2007, 13,
2300–2311.
[18] B. Schneider, Z. Morꢄvek, H. M. Berman, Nucleic Acids Res. 2004, 32,
1666–1677.
[8] S. Valadkhan, J. L. Manley, Nature 2001, 413, 701–707.
[9] S. Valadkhan, J. L. Manley, RNA 2003, 9, 892–904.
[10] T. Lçnnberg, K.-M. Kero, Org. Biomol. Chem. 2012, 10, 569–574.
[11] A. Semenyuk, A. Fçldesi, T. Johansson, C. Estmer-Nilsson, P. Blomgren,
M. Brꢃnnvall, L. A. Kirsebom, M. Kwiatkowski, J. Am. Chem. Soc. 2006,
128, 12356–12357.
[12] T. Horn, M. S. Urdea, Tetrahedron Lett. 1986, 27, 4705–4708.
[13] T. Lçnnberg, J. Korhonen, J. Am. Chem. Soc. 2005, 127, 7752–7758.
[14] D. A. Usher, A. H. McHale, Proc. Natl. Acad. Sci. USA 1976, 73, 1149–
1153.
[19] a) N. Toor, K. S. Keating, S. D. Taylor, A. M. Pyle, Science 2008, 320, 77–
82; b) N. Toor, K. Rajashankar, K. S. Keating, A. M. Pyle, Nat. Struct. Mol.
Biol. 2008, 15, 1221–1222; c) N. Toor, K. S. Keating, O. Fedorova, K. Raja-
shankar, J. M. Wang, A. M. Pyle, RNA 2010, 16, 57–69; d) A. M. Pyle, Crit.
Rev. Biochem. Mol. Biol. 2010, 45, 215–232.
[20] P. M. Gordon, E. J. Sontheimer, J. A. Piccirilli, RNA 2000, 6, 199–205.
[21] a) T. Lçnnberg, Chem. Eur. J. 2011, 17, 7140–7153; b) C. L. Peebles, P. S.
Perlman, K. L. Mecklenburg, M. L. Petrillo, J. H. Tabor, K. A. Jarrell, H. L.
Cheng, Cell 1986, 44, 213–223; c) R. van der Veen, A. C. Arnberg, G. van
der Horst, L. Bonen, H. F. Tabak, L. A. Grivell, Cell 1986, 44, 225–234;
d) C. Schmelzer, R. J. Schweyen, Cell 1986, 46, 557–565.
[15] a) P. Jꢃrvinen, M. Oivanen, H. Lçnnberg, J. Org. Chem. 1991, 56, 5396–
5401; b) K. Kawamura, Biochim. Biophys. Acta Gen. Subj. 2003, 1620,
199–210.
[16] S. Kuusela, H. Lçnnberg, J. Chem. Soc. Perkin Trans. 2 1994, 2109–2113.
[17] D. J. Klein, T. M. Schmeing, P. B. Moore, T. A. Steitz, EMBO J. 2001, 20,
4214–4221.
Received: October 1, 2012
Published online on November 26, 2012
2700
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 2690 – 2700