Intramolecular participation of amino groups in the cleavage and isomerization of ribonucleoside 3′-phosphodiesters: The role in stabilization of the phosphorane intermediate

Article abstract of DOI:10.1002/chem.201301711

A dinucleoside-3′,5′-phosphodiester model, 5′-amino- 4′-aminomethyl-5′-deoxyuridylyl-3′,5′-thymidine, incorporating two aminomethyl functions in the 4′-position of the 3′-linked nucleoside has been prepared and its hydrolytic reactions studied over a wide pH range. The amino functions were found to accelerate the cleavage and isomerization of the phosphodiester linkage in both protonated and neutral form. When present in protonated form, the cleavage of the 3′,5′-phosphodiester linkage and its isomerization to a 2′,5′-linkage are pH-independent and 50-80 times as fast as the corresponding reactions of uridylyl-3′,5′-uridine (3′,5′-UpU). The cleavage of the resulting 2′,5′-isomer is also accelerated, albeit less than with the 3′,5′-isomer, whereas isomerization back to the 3′,5′-diester is not enhanced. When the amino groups are deprotonated, the cleavage reactions of both isomers are again pH-independent and up to 1000-fold faster than the pH-independent cleavage of UpU. Interestingly, the 2′- to 3′-isomerization is now much faster than its reverse reaction. The mechanisms of these reactions are discussed. The rate accelerations are largely accounted for by electrostatic and hydrogen-bonding interactions of the protonated amino groups with the phosphorane intermediate. Amino accelerator: Intramolecular amino groups promote cleavage and isomerization of ribonucleoside phosphodiester bonds in both protonated and deprotonated forms. The most likely mechanistic explanation is facilitation of the first step of the reaction, that is, formation of the cyclic phosphorane intermediate, by concomitant proton transfer. The magnitude of this facilitation differs greatly between the 2′- and 3′-isomers (see figure). Copyright

Full text of DOI:10.1002/chem.201301711

DOI: 10.1002/chem.201301711
Intramolecular Participation of Amino Groups in the Cleavage and
Isomerization of Ribonucleoside 3’-Phosphodiesters:
The Role in Stabilization of the Phosphorane Intermediate
Luigi Lain, Harri Lçnnberg, and Tuomas Lçnnberg*[a]
Abstract: A dinucleoside-3’,5’-phospho-
diester model, 5’-amino-4’-aminometh-
yl-5’-deoxyuridylyl-3’,5’-thymidine, in-
corporating two aminomethyl functions
in the 4’-position of the 3’-linked nu-
cleoside has been prepared and its hy-
drolytic reactions studied over a wide
pH range. The amino functions were
found to accelerate the cleavage and
isomerization of the phosphodiester
linkage in both protonated and neutral
form. When present in protonated
form, the cleavage of the 3’,5’-phospho-
diester linkage and its isomerization to
a 2’,5’-linkage are pH-independent and
50–80 times as fast as the correspond-
ing reactions of uridylyl-3’,5’-uridine
(3’,5’-UpU). The cleavage of the result-
ing 2’,5’-isomer is also accelerated,
albeit less than with the 3’,5’-isomer,
whereas isomerization back to the 3’,5’-
diester is not enhanced. When the
amino groups are deprotonated, the
cleavage reactions of both isomers are
again pH-independent and up to 1000-
fold faster than the pH-independent
cleavage of UpU. Interestingly, the 2’-
to 3’-isomerization is now much faster
than its reverse reaction. The mecha-
nisms of these reactions are discussed.
The rate accelerations are largely ac-
counted for by electrostatic and hydro-
gen-bonding interactions of the proto-
nated amino groups with the phosphor-
ane intermediate.
Keywords: amines · isomerization ·
kinetics · phosphodiester · reaction
mechanisms · RNA
Introduction
catalytic site of ribonucleases and small ribozymes usually
contains positively charged species, either metal ions[9] or
protonated amino functions,[10] which surround the oxyphos-
phorane transition state (or marginally stable intermediate).
A good example is the Lys41 residue of RNase A, the pro-
tonated e-amino group of which stabilizes the phosphorane-
like transition state either electrostatically or by hydrogen
bonding.[2]
We have previously attempted to find out, by using small
molecular model compounds, to what extent the cleavage of
ribonucleoside 3’-phosphodiesters may be accelerated by
stabilization of the phosphorane intermediate by hydrogen-
bonding interactions. Studies with dinucleoside-3’,3’-mono-
phosphates have shown that a hydrogen-bond donor at C2’,
As generally known, cleavage of RNA phosphodiester
bonds may be accelerated by deprotonation of the 2’-OH
group attacking the phosphorus atom and protonation of
the 5’-O atom departing from the phosphorus.[1] When the
reaction is catalyzed by ribonucleases[2] or small ribozymes,[3]
ꢀ
these two proton transfers, accompanied by P’ O2’ bond-
making and P O5’ bond-breaking, take place in a more or
ꢀ
less concerted manner via a penta-coordinated transition
state. By contrast, less-efficient catalysis by Brçnsted acids
or bases, metal ion complexes or related small molecular en-
tities leads to processes taking place by formation of a
penta-coordinated phosphorane intermediate that has a
finite lifetime.[4] When the intermediate is sufficiently stable,
isomerization to a 2’,5’-phosphodiester may compete with
the cleavage. Stability and lifetime of the phosphorane spe-
cies primarily depends on the state of protonation[5,6] and
metal ion binding,[7] but it is also influenced by electrostatic
effects and hydrogen bonding.[8] The energy level of a
penta-coordinated transition state is affected similarly. The
ꢀ
that is, adjacent to the scissile P O3’ bond, markedly accel-
erates the cleavage, the relative rates obtained with 2’-
+[11]
NH2,[11] 2’-OMe,[12] 2’-OH,[12] and 2’-NH3
being 0.4, 1, 34,
and 1700, respectively. This acceleration obviously consists
of three different components, that is, 1) stabilization of the
phosphorane intermediate electrostatically and/or by hydro-
gen bonding, 2) stabilization of the leaving group by intra-
molecular hydrogen bonding, and 3) inductive reduction of
electron density of the departing 3’-oxygen atom. Studies
with trinucleoside-3’,3’,5’-monophosphates suggest that sta-
bilization of the phosphorane intermediate plays an impor-
tant role, since replacement of 2’-OMe with 2’-OH on the
leaving 3’-linked nucleoside enhances the departure of both
the 3’- and 5’-linked nucleosides almost equally.[13] This
structural change also accelerates isomerization of the
3’,3’,5’-triester to a mixture of 2’,3’,5’- and 2’,2’,5’-triesters, as
[a] L. Lain, Prof. Dr. H. Lçnnberg, Dr. T. Lçnnberg
Department of Chemistry
University of Turku
Vatselankatu 2, 20014 Turku (Finland)
Fax : (+358)2-333-6700
E-mail: tuanlo@utu.fi
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201301711.
12424
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 12424 – 12434

FULL PAPER
exemplified with the more slowly reacting phosphoromono-
thioate analogues.[14] These results are consistent with stabi-
lization of the phosphorane intermediate, although stabiliza-
tion of the leaving group by inductive effect and/or intramo-
lecular hydrogen bonding undoubtedly contributes apprecia-
bly. To eliminate the latter influences, we now report on the
cleavage and isomerization of a structural analogue of uri-
dylyl-3’,5’-thymidine bearing two aminomethyl substituents
on C4’ of the 3’-linked nucleoside (1a). The amino groups,
when protonated, may interact with the non-bridging phos-
phorane oxygens, but less likely with the 5’-oxygen of the
departing nucleoside. Accordingly, kinetic studies with 1a
and its isomerization product 1b shed light on influences
that amino functions exert on reactions of a phosphodiester
linkage, when anchored in the vicinity of this linkage in such
a way that hydrogen bonding and direct field effects are ex-
pectedly more significant than inductive effects transmitted
through covalent bonds.
and the 4’-position of the aldehyde 3 thus obtained was hy-
droxymethylated by treatment with an excess of formalde-
hyde and subsequent reduction with NaBH4, affording the
diol 4 in moderate yield. This compound (4) has previously
been prepared in a comparable yield by Swern (rather than
IBX) oxidation.[18] The somewhat sub-satisfactory hydroxy-
methylation step is mostly due to the susceptibility of the
tert-butyldimethylsilyl (TBDMS) protections to hydrolysis
under the strongly alkaline conditions needed.
Both of the hydroxy functions of the diol 4 were activated
as triflates and the resulting compound 5 was converted to
the diazide 6. As previously described, the TBDMS-protect-
ed 2’-oxygen readily attacks the 4’-triflyloxymethyl carbon,
resulting in an LNA-type cyclic product.[19] Contrary to the
original report, simultaneous conversion of both of the hy-
droxy groups to triflates did not in our hands completely
eliminate this side reaction. For this reason, handling and
chromatographic purification of compound 5 was carried
out at low temperature (38C). Both of the silyl protections
were then removed and the 2’-OH group was protected as a
4,4’-dimethoxytrityl ether, affording compound 8 (attempts
to remove a single TBDMS protection from 6 were unsuc-
cessful but, on the other hand, tritylation of the fully depro-
tected product 7 proved surprisingly 2’-selective, giving no
trace of the 3’-isomer).
Results
Preparation of the model compound: The synthesis of 5’-
amino-4’-aminomethyl-5’-deoxyuridylyl-3’,5’-thymidine (1a),
a
dinucleoside phosphodiester model, is described in
Scheme 1. First, the 2’- and 3’-OH groups of uridine were
protected as tert-butyldimethylsilyl ethers (2) by silylating
all of the sugar OH groups[15] and then selectively deprotect-
ing the 5’-position by acid-catalyzed hydrolysis.[16] The ex-
posed 5’-OH was oxidized with iodoxybenzoic acid (IBX)[17]
Synthesis of the phosphodiester linkage was achieved by
5-benzylthio-1H-tetrazole-promoted coupling of compound
8 with commercially available thymidine-5’-phosphorami-
dite, followed by conventional oxidation with aqueous
iodine and removal of the cyanoethyl protection with trie-
thylamine in pyridine. The
product (9), obtained as a trie-
thylammonium salt, was puri-
fied by silica gel chromatogra-
phy. Finally, Staudinger reduc-
tion of the azido groups to
amino groups,[20] followed by
removal of the dimethoxytrityl
protections,
afforded
the
model compound 1a by parti-
tioning between CH2Cl2 and
water in sufficient purity for
the kinetic studies. For charac-
terization, a small sample of
1a was purified by RP HPLC.
The purification was compli-
cated by the fact that 1a un-
derwent during the course of
purification a side reaction at
its amino groups, possibly with
atmospheric carbon dioxide
(see the discussion below).
Scheme 1. Preparation of the dinucleoside phosphodiester model compound. Reagents and conditions:
a) 1) TBDMSCl, imidazole, DMF; 2) THF, TFA, H2O; b) IBX, MeCN; c) 1) formaldehyde, NaOH, 1,4-diox-
ane, H2O, 2) NaBH4, H2O; d) Tf2O, CH2Cl2, pyridine; e) LiN3, DMF; f) Et3N·3HF, THF; g) DMTrCl, DMAP,
pyridine; h) 1) 3’-(4,4’-dimethoxytrityl)-thymidine-5’-[(2-cyanoethyl)-(N,N-diisopropyl)]phosphoramidite, 5-
benzylthio-1H-tetrazole, MeCN; 2) I2, 2,6-lutidine, THF, H2O; 3) pyridine, Et3N; i) 1) PPh3, pyridine, 2) NH3,
H2O, pyridine; 3) HCl, H2O, MeOH. Abbreviations: DMTrCl=4,4’-dimethoxytrityl chloride; DMAP=4-dime-
thylaminopyridine; TFA=trifluoroacetic acid; Tf2O=trifluoromethanesulfonic anhydride.
Determination of the pKa
values of the model com-
pound: For correlation of the
observed partial reactions with
Chem. Eur. J. 2013, 19, 12424 – 12434
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12425

T. Lçnnberg et al.
specific ionic forms of the model compound 1a (or 1b), pKa
values of the two amino groups were determined by NMR
spectrometric titration. The titrations were carried out in
D2O with 0.10 molLꢀ1 sodium dihydrogen phosphate serving
Product distribution and reaction pathways: The hydrolytic
reactions of 1a were followed over a wide pH range (1 to
11) at 908C and an ionic strength of 0.10 molLꢀ1 by analyz-
ing the composition of aliquots withdrawn from the reaction
mixture at appropriate time intervals by RP HPLC. The ali-
quots were quenched by cooling to 08C and, when necessa-
ry, adjusting their pH with a concentrated formate buffer.
The identity of the products detected was established either
by spiking with authentic samples or by mass spectrometric
analysis (HPLC/ESI-MS). The monomeric products derived
from the 3’-linked nucleoside did not retain enough to allow
a reliable quantification with the HPLC system used. The
other peak areas were converted to relative concentrations
by calibrating the system with uridine and thymidine solu-
tions of known concentrations (the hypochromicity of the
dimers 1a and 1b relative to the free nucleosides was as-
sumed to be negligible).
as
a background buffer. Upon gradual addition of
1.0 molLꢀ1 sodium deuteroxide in D2O, a downfield shift of
the 31P signal of the buffer and an upfield shift of the amino-
methyl signals of 1a were observed. When plotted against
pH, the latter yielded a curve with two inflection points, re-
ferring to the pKa values of the two amino groups (Fig-
ure 1A). Titration under the conditions of the kinetic meas-
urements (i.e., at 908C) would have been impractical be-
cause a significant part of 1a would have reacted during ac-
quisition of the NMR spectra. Instead, the titration was per-
formed at three temperatures (22.5, 34.4 and 54.98C) and
the results were extrapolated to 908C by linear regression
(Figure 1B). The two pKa values thus obtained were 5.8 and
7.2. The respective values of 1b were assumed to be largely
similar.
As previously reported for uridylyluridine (UpU),[21] two
main reaction pathways were observed, that is, cleavage to a
nucleoside-2’,3’-cyclic phosphate with release of the 5’-
linked nucleoside (Scheme 2, Routes A and B) and isomeri-
zation between 3’,5’- and 2’,5’-phosphodiesters (Routes C
and D). In striking contrast to UpU, however, the two iso-
mers reacted at pH 3–8 at markedly different rates: the
cleavage rates differed by up to one and the isomerization
rates by up to two orders of magnitude. In addition, a harm-
ful side reaction occurred. Around neutral pH, the two
amino functions of 1a (and 1b) became bridged by a meth-
ylene group (Scheme 2, Routes E and F). In addition to
HPLC/MS analysis, the identity of this side product (10a)
was established by treating 1a with an excess of formalde-
hyde and spiking with the product thus obtained. The origin
of this one carbon fragment, however, remains obscure (see
discussion below).
The multitude of the parallel, consecutive and opposing
reactions precluded applying an integrated rate law for fit-
ting of the kinetic data. Instead, pseudo-first-order rate con-
stants for the various partial reactions were obtained by nu-
merical fitting of the differential rate equations. The rate
constants obtained were usually almost independent of
buffer concentration. Only in formate buffers, significant
buffer catalysis of the cleavage reactions was observed. In
these cases, buffer-independent rate constants were obtained
by varying the buffer concentration from 10 to 100 mmolLꢀ1
and extrapolating the observed pseudo-first-order rate con-
stants to zero buffer concentration by linear regression.
pH–Rate profiles: The pH–rate profiles for the cleavage of
1a (Route A, Scheme 2) and its 2’,5’-isomer 1b (Route B),
overlain with species distribution curves for the different
ionic forms of the starting material, are presented in Fig-
ure 2A. At pH<2, both reactions are first-order in hydroni-
um ion concentration and proceed at equal rates. Between
pH 2 and 3, cleavage of 1a becomes pH-independent and
remains so over a wide range until pH 5, where it becomes
first-order in hydroxide ion concentration. Cleavage of 1b
exhibits a similar profile except that it remains first-order in
Figure 1. A) The average chemical shift of the aminomethyl protons of
~
&
*
1a as a function of pH at 22.5 ( ), 34.4 ( ), and 54.98C ( ) and B) linear
extrapolation of the pKa values of the two amino groups to 908C.
12426
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 12424 – 12434

The Role of Stabilization of the Phosphorane Intermediate
FULL PAPER
Scheme 2. Hydrolytic reactions of 1a.
hydronium ion concentration until pH 3. In other words, the
pH-independent cleavage of 1b is approximately an order
of magnitude slower than that of 1a and it prevails over a
somewhat narrower pH range. Between pH 7.5 and 9, an-
other pH-independent region is observed for both reactions,
before hydroxide ion catalysis takes over at pH> 9.
The pH–rate profiles for the isomerization of 1a (Route
C) and its 2’,5’-isomer 1b (Route D), overlain with species
distribution curves for the different ionic forms of the start-
ing material, are presented in Figure 2B. Over a narrow
region around pH 1, both reactions are first-order in hydro-
nium ion concentration and equally rapid. At pH>1, how-
ever, the pH–rate profiles diverge greatly: isomerization
from 1a to 1b is pH-independent over a wide range from
pH 2 to 5.5, whereas the reverse reaction remains first-order
in hydroxide ion concentration until pH 3 and experiences
only a narrow pH-independent region around pH 4. Be-
tween pH 5.5 and 7.5, the two reactions exhibit opposite
pH-dependencies, Route C being first-order in hydronium
ion concentration and Route D first-order in hydroxide ion
concentration. At pH 7.5, the latter reaction becomes pH-in-
dependent. The former reaction, on the other hand, appears
to remain first-order in hydronium ion concentration, al-
though the data at hand do not allow reliable assessment of
this point (at pH> 8, all the other reactions are much faster
than isomerization through Route C).
The observed pseudo first-order rate constants for the
cleavage of 1a and 1b may be expressed by Equation (1).
cl
W1
2
cl
W2
cl
OH
½Hþꢁ
k
½Hþꢁ þ k Ka1½Hþꢁ þ kWcl 3Ka1Ka2
k
Kw
cl
obs
k
¼ kcHl ½Hþꢁ þ
þ
½Hþꢁ þ Ka1½Hþꢁ þ Ka1Ka2
2
ð1Þ
in which kcHl and k are the second-order rate constants for
cl
OH
+
the hydronium ion catalyzed cleavage of SH2 and hydrox-
ide ion catalyzed cleavage of Sꢀ, respectively. kW1, kW2, and
kW3 are the first-order rate constants for the pH-independ-
ent cleavage of the diprotonated (SH2+), monoprotonated
(SH), and fully deprotonated (Sꢀ) ionic forms, respectively.
The observed pseudo first-order rate constants for the iso-
merization of 1a and 1b may be expressed by Equation (2).
+
Ka1 and Ka2 are the dissociation constants of SH2 and SH,
respectively, and KW is the ionic product of water under the
experimental conditions (6.2ꢁ10ꢀ13). The rate constants
were obtained by fitting the experimental values to Equa-
tion (1) by a nonlinear least-squares method and are pre-
sented in Table 1. To reduce the number of parameters to
be fitted, the values of Ka1 and Ka2 were constrained to
those obtained by NMR titrations and found to be consis-
tent with the observed pH–rate profiles (the values obtained
for pKa by NMR titrations and by fitting the kinetic data
were 7.2ꢂ0.7 and 6.8ꢂ0.2, respectively).
is
W1
2
is
W2
k
½Hþꢁ þ k Ka1½Hþꢁ þ kWis 3Ka1Ka2
is
obs
¼ kiHs ½Hþꢁ þ
ð2Þ
k
½Hþꢁ þ Ka1½Hþꢁ þ Ka1Ka2
2
Here, kiHs is the second-order rate constant for the hydroni-
is
W1
is
W2
um ion catalyzed isomerization of SH2+, and k , k , and
is
W3
k
are the first-order rate constants for the pH-independent
isomerization of SH2+, SH, and Sꢀ, respectively. The overall
rate constant for isomerization, k1 +kꢀ1, remains approxi-
mately constant at pH>3, with only the isomeric ratio being
Chem. Eur. J. 2013, 19, 12424 – 12434
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12427

T. Lçnnberg et al.
reversed around pKa1. Under conditions in which the fully
deprotonated species predominate, the rapid cleavage reac-
tions make reliable determination of the isomerization rate
difficult, particularly in the case of 1a. The rate constants
were obtained by fitting the experimental values to Equa-
tion (2) by a nonlinear least-squares method and are pre-
sented in Table 1.
Discussion
Hydronium ion-catalyzed cleavage and isomerization of 1a
and 1b: Around pH 2, cleavage and isomerization of 1a and
1b are first-order in hydronium ion concentration and
equally rapid. Under these conditions, the respective reac-
tions of UpU are approximately as fast but exhibit second-
order dependence in hydronium ion concentration, taken as
an indication of pre-equilibrium protonation of both of the
non-bridging oxygen atoms of the monoanionic phospho-
diester linkage.[21] The first-order dependence in the case of
1a and 1b indicates, hence, that protonation of the phospho-
diester linkage is retarded, in all likelihood due to electro-
static repulsion and inductive electron withdrawal by the
two protonated amino groups. For comparison, the phos-
phoromonothioate analogue of UpU, having a less basic in-
ternucleosidic linkage than UpU, also exhibits first-order de-
pendence of rate on acidity under these conditions.[22] Ac-
cordingly, the reactions of 1a and 1b are most likely initiat-
ed by attack of the vicinal hydroxy group on neutral phos-
phate with concomitant proton transfer from the attacking
hydroxyl group to a non-bridging oxygen, as reported for
UpU over a narrow range around pH 3.[23] The pseudorota-
tion barrier for neutral (dihydric) phosphorane is, according
to quantum chemical calculations, much lower than those
[6,24]
ꢀ
for any of the P O bond cleavages.
Accordingly, both
the cleavage and the isomerization reactions proceed by
rate-limiting departure of the respective leaving groups
from the penta-coordinated intermediate, concerted with
water-mediated proton transfer from a non-bridging phos-
phoryl oxygen.[6,24,25] The relative rate constants for the
breakdown of the phosphorane intermediate derived from
1a are 1.0:1.0:1.3 for the departure of 3’-O, 2’-O, and 5’-O,
respectively (Table 1). With UpU the corresponding values
are 1.2:1.0:0.4.[21] The protonated amino groups seem,
hence, to slightly favor the exocyclic cleavage reaction. The
product distribution is actually very similar to that reported
for the hydronium ion catalyzed breakdown of the phos-
phorane intermediate derived from UpU, that is, 1.1, 1.0,
and 1.3.[21] Somehow the protonated amino groups appear to
take the role of hydronium ion, although the overall effects
on the hydronium ion catalyzed reactions otherwise remain
modest.
Figure 2. pH–Rate profiles for the cleavage (A) and isomerization (B) of
1a ( ) and 1b (&); T=908C, I
(NaClO4)=0.10 molLꢀ1. The dashed lines
&
refer to the respective reactions of UpU and the gray lines to the mole
fractions of the different ionic forms (C) of 1a and 1b.
Table 1. Rate and equilibration constants for the partial reactions of
cleavage and isomerization of 1a and 1b; T=908C,
IACTHNGUTERNN(UG NaClO4)=
0.10 mol Lꢀ1
.
Route A
(k2)
Route B
(k3)
Route C
(k1)
Route D
(kꢀ1
N
)
kH [10ꢀ3 Lmolꢀ1 sꢀ1
]
1.2ꢂ0.2
22ꢂ4
1.0ꢂ0.2
3ꢂ1
1.0ꢂ0.2
330ꢂ30
130ꢂ30
1.0ꢂ0.3
6ꢂ4
kW1 [10ꢀ7 sꢀ1
kW2 [10ꢀ7 sꢀ1
kW3 [10ꢀ7 sꢀ1
]
]
[a]
–
5ꢂ4
100ꢂ40
300ꢂ100
–
[a]
360ꢂ70
160ꢂ30
6ꢂ1
–
kOH [10ꢀ2 Lmolꢀ1 sꢀ1
pKa1
pKa2
]
7ꢂ1
–
Spontaneous cleavage and isomerization of diprotonated 1a
and 1b: Between pH 3 and 5, cleavage of the 3’,5’-isomer 1a
(Route A) is pH-independent and 76-fold faster than the
pH-independent hydrolysis of UpU.[21] Isomerization of 1a
5.8ꢂ0.1[b]
7.2ꢂ0.7[b]
5.8ꢂ0.1[b]
5.8ꢂ0.1[b]
5.8ꢂ0.1[b]
7.2ꢂ0.7[b]
7.2ꢂ0.7[b]
7.2ꢂ0.7[b]
[a] Could not be determined reliably. [b] Determined by NMR titrations.
12428
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 12424 – 12434

The Role of Stabilization of the Phosphorane Intermediate
FULL PAPER
Scheme 3. Spontaneous cleavage and isomerization of the diprotonated ionic form of 1a and 1b.
(Route C), in turn, becomes pH-independent already at
pH 2 and is 50-fold faster than the pH-independent isomeri-
zation of UpU. Under these conditions, the predominant
ionic form of 1a (and 1b) is SH2+, having both of the amino
groups protonated and the phosphodiester linkage deproto-
nated (Scheme 3). The rapid isomerization is a compelling
piece of evidence of a pseudorotating intermediate. Within
ꢀ
this intermediate, endocyclic P O bond cleavage is marked-
ly favored over the exocyclic one; pseudorotation followed
ꢀ
by P O3’ bond cleavage is 15 times as fast as the cleavage
ꢀ
of the P O5’ bond, although O5’ may take an apical posi-
tion without pseudorotation. The behavior is in this respect
similar to that of UpU.[21] With UpU, a monoanionic phos-
Figure 3. Free-energy profiles for the spontaneous cleavage and isomeri-
ꢀ
zation of the diprotonated ionic form (SH2+) of 1a and 1b.
phorane is obtained by a concerted process consisting of P
O2’ bond formation and proton transfer from the attacking
hydroxy group to a non-bridging oxygen, either directly[26]
or mediated by a water molecule.[27] Evidently, this is also
the case with 1a (TS1 in Figure 3) but, since both the cleav-
age and isomerization of 1a are faster than with UpU, the
protonated amino groups somehow facilitate formation of
the phosphorane intermediate. Possible explanations for
such facilitation include lowering of the electron density of
the phosphate group inductively, electrostatically, and by hy-
drogen bonding. Bearing in mind that the estimates for the
first pKa values of pentaoxyphosphoranes range from 6.5 to
11,[28] proton transfer from the amino group to a non-bridg-
ing oxygen, giving a dihydric phosphorane intermediate (I1
in Figure 3), appears an attractive alternative. This inter-
mediate may collapse back to the starting material through
TS1, undergo pseudorotation through TS2 or expel O5’ via
TS3. The rapid isomerization to 2’,5’-phosphodiester suggests
that the pseudorotation barrier (TS2) is lower than the barri-
er for exocyclic cleavage (TS3). The fact that both cleavage
and isomerization are accelerated to comparable extents in-
dicates that the pre-equilibrium concentration of I1 is high
compared with the situation with UpU. The slightly greater
acceleration of the cleavage reaction could, in principle, be
ꢀ
attributable to general acid catalysis of the P O5’ bond fis-
sion by a protonated amino group but this seems unlikely
given the long distances between the amino groups and the
Chem. Eur. J. 2013, 19, 12424 – 12434
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12429

T. Lçnnberg et al.
5’-oxygen that has to take an apical position upon departure.
Particularly in the case of the pseudorotamer having O3’ in
an apical position, such a proton transfer does not appear
feasible.
For comparison, a fast pH-independent cleavage has pre-
viously been reported for guanylyl-3’,3’-(2’-amino-2’-deoxy-
uridine),[11] and attributed to the reaction through a minor
tautomer having one of the amino groups protonated and
the attacking hydroxy function deprotonated. A similar
mechanism may be followed in the present case. The mole
fraction of such a zwitterionic species is of the order of 10ꢀ5,
taking the difference between the pKa values of the proto-
nated amino group and the 2’-OH into account. Still, this
ionic form probably is the reactive one, since the 2’-Oꢀ is
106 times better nucleophile than the 2’-OH.[5]
Within the reactive zwitterionic tautomer, the protonated
amino group may facilitate formation of the phosphorane
intermediate by inductive and field effects and by hydrogen
bonding, as discussed above. In the present case, proton
transfer from the protonated amino group to an oxyaninon,
giving a monoanionic phosphorane (I1 and I2 in Figure 4),
appears likely in light of the occurrence of relatively fast iso-
merization. It should be borne in mind that the pseudorota-
tion barriers for dianionic phosphoranes are generally pro-
hibitively high.[1,29] Lowering this barrier by proton transfer,
albeit from a hydroxy rather than a protonated amino
group, has been reported for adenosine-3’- and ꢀ2’-(1-hy-
droxyalkyl)phosphonate model compounds.[31]
The overall isomerization between 1a and 1b hardly ex-
periences any change in rate on passing the pKa values of
the two amino groups but phosphate migration from O2’ to
O3’, rather than from O3’ to O2’, takes over. The fact that
1b undergoes cleavage and isomerization at a comparable
rate whereas with 1a the cleavage overwhelmingly predomi-
nates may tentatively be reasoned as follows. Owing to
longer distance to the protonated amino group, the 2’-oxy-
anion of zwitterionic 1a is a better nucleophile than the 3’-
oxyanion of zwitterionic 1b. Although the attack of 3’-Oꢀ
on the negatively charged phosphodiester linkage requires
assistance by intramolecular general acid catalysis of the
protonated amino group (TS5), giving a monoanionic phos-
phorane (I2) able to pseudorotate, the 2’-Oꢀ is nucleophilic
enough to attack without such assistance by intramolecular
proton transfer (TS6). The latter reaction, hence, results in a
dianionic phosphorane (I3) unable to pseudorotate. In other
words, whereas 1b reacts solely through a monoanionic
phosphorane, compound 1a may additionally utilize a route
through a dianionic phosphorane. Since the dianionic phos-
phorane decomposes only to cleavage products, isomeriza-
tion to 1b is hardly able to compete with the cleavage reac-
tion.
In striking contrast to 1a, the 2’,5’-isomer (1b) does not
undergo isomerization much faster than 2’,5’-UpU. The
equilibrium between 1a and 1b thus lies far on the side of
1b. The cleavage rate of 1b is only 14% of that of 1a but
still 10 times as high as with UpU. Since the cleavage is
markedly accelerated compared with UpU, whereas isomeri-
zation is not, it appears reasonable to conclude that the
phosphorane intermediate I2 is formed more readily than
with UpU, but the pseudorotation barrier (TS2) is now
higher than the barrier for nucleophilic attack (TS5) and,
hence, limits the rate of isomerization.
The questions that remain to be answered are, why the
overall rate of the disappearance 1a (Route A+Route C) is
39-fold compared to the disappearance of 1b (Route B+
Route D), and why the equilibrium between 1a and 1b is so
strongly biased on the side of 1b. As mentioned above, the
positively charged protonated 4’-aminomethyl substituents
reduce the electron density at the 3’-phosphodiester linkage
both inductively and electrostatically, and possibly also by
hydrogen bonding. The influence on the electron density
and, hence, nucleophilicity of the 2’-OH remains smaller,
owing to the longer distance. Accordingly, formation of I1 is
facilitated. In the case of 1b, the situation is opposite. The
electron density of the attacking 3’-OH is now reduced
more markedly than that of the electrophilic phosphodiester
linkage and formation I2 is facilitated less than the forma-
tion of I1. For the same reasons, O3’ is a better leaving
group than O2’. The biased isomeric ratio may, hence, be at-
tributed to more facile formation of I1 compared with I2 and
more facile breakdown of I2 to 1b compared with the break-
down of I1 to 1a.
Spontaneous cleavage and isomerization of monoprotonated
1a and 1b: The ionic form SH of 1a and 1b, having only
one of the two amino groups protonated, predominates over
a very narrow range around pH 6.5 and never accounts for
more than 60% of the total concentration. As a result, de-
termination of the reaction rates of this species is extremely
difficult. Owing to simultaneous occurrence of cleavage and
isomerization of three different ionic forms, the rate con-
stants referring to SH could not be determined reliably.
Spontaneous cleavage and isomerization of deprotonated 1a
and 1b: Between pH 7.5 and 9, the cleavage of 1a (Route
A) and 1b (Route B) exhibit another plateau. Under these
conditions, 1a and 1b exist mostly as the fully deprotonated
diamine (Scheme 4). The spontaneous cleavage of this ionic
form is dramatically faster than the pH-independent hydrol-
ysis of UpU, the acceleration being 1000- and 600-fold for
1a and 1b, respectively. Isomerization of 1b to 1a (Route
D) proceeds at a comparable rate, whereas the reverse reac-
tion (Route C) is much slower.
Hydroxide ion-catalyzed cleavage of 1a and 1b: With both
1a and 1b, hydroxide ion-catalyzed cleavage predominates
at pH> 9. Under these conditions, isomerization is too slow
to be detected. Both cleavage reactions (Routes A and B)
proceed at the same rate as the respective reaction of UpU.
In other words, deprotonated 4’-aminomethyl substituents
have virtually no effect on the hydroxide ion-catalyzed
cleavage.
12430
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 12424 – 12434

The Role of Stabilization of the Phosphorane Intermediate
FULL PAPER
Scheme 4. Spontaneous cleavage and isomerization of the fully deprotonated ionic form of 1a and 1b.
other words, under conditions in which at least one of the
two amino groups is deprotonated, both of the amino
groups react with one carbon electrophile, ultimately result-
ing in the formation of a methylene bridge between them.
Over the pH range where it can be detected, the reaction is
pH-independent but somewhat sensitive to the buffer used,
with the greatest accumulation of 10a and 10b observed in
4-(2-hydroxyethyl)-1-piperazineethanesulfonic
acid
(HEPES) buffers. In triethylammonium buffers, the side re-
action was much less significant. In all cases, compounds
10a and 10b reacted further to thymidine and other mono-
meric products.
The most obvious candidate for the carbon electrophile
that causes the side reaction would be formaldehyde. How-
ever, whereas traces of formaldehyde are present in ambient
atmosphere,[31] the concentrations appear insufficient to ex-
plain the observed formation of 10a and 10b. On the other
hand, formation of formaldehyde in the reaction solution,
for example, by elimination of one of the aminomethyl
Figure 4. Free-energy profiles for the spontaneous cleavage and isomeri-
zation of the fully deprotonated ionic form (Sꢀ) of 1a and 1b.
Formation of the hexahydropyrimidine side products 10a
and 10b: Between pH 5 and 9, the hydrolytic reactions are
accompanied by a side reaction yielding the hexahydropyri-
midine side products 10a and 10b (Routes E and F). In
Chem. Eur. J. 2013, 19, 12424 – 12434
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12431

T. Lçnnberg et al.
referenced to internal TMS and the 31P chemical shifts to external ortho-
phosphoric acid. The reagents were commercial products that were used
as received.
groups as an imine which is then rapidly hydrolyzed, cannot
be discarded.
Carbon dioxide is a much more abundant carbon electro-
phile that upon reaction with 1a would first give a urea in-
termediate. Successive reductions of this intermediate would
then eventually afford the observed side product 10a. Possi-
ble intermediates along this reaction pathway, as well as the
final product 10a were, indeed, detected in the mass spec-
trum of a partially decomposed sample of 1a, lending sup-
port to this idea. The mechanisms of the reduction steps,
however, remain obscure.
Kinetic measurements: The reactions were carried out in sealed tubes im-
mersed in a thermostated water bath (90.0ꢂ0.18C). The hydronium ion
concentration of the reaction solutions was adjusted with perchloric acid
at pH<2, sodium hydroxide at pH>10, formic acid buffers for pH 3–4,
cacodylic acid buffers for pH 5.5–7, glycine buffers for pH 7.5–9, and trie-
thylammonium buffer for pH 8.5. To examine the possibility of buffer
catalysis, three different buffer concentrations were used. The ionic
strength of the solutions was kept constant at 0.10 molLꢀ1 with sodium
perchlorate. The initial substrate (1a) concentration was approximately
0.1 mm. At pH<7.5, the 2’,5’-isomer 1b is accumulated in sufficient
quantities to allow determination of its cleavage rate (Route B). To
follow this reaction under more alkaline conditions, equilibrium between
1a and 1b was first allowed to settle in a small volume of dilute formate
buffer (20 h at pH 4), after which a much larger volume of the appropri-
ate reaction buffer was added.
Conclusion
The results of the present study elucidate the impact that
amino groups in the vicinity of an RNA phosphodiester
linkage may have on the rate and mechanism of its cleavage
and isomerization. For this purpose, the kinetics of these re-
actions have been studied with an analog of uridylyl-3’,5’-
thymidine bearing two aminomethyl groups, instead of one
hydroxymethyl group, at C4’ of the uridine moiety. The ef-
fects of these aminomethyl groups are particularly notewor-
thy at pH 3–9. Interestingly, at pH 3–5 when the amino
groups are almost fully protonated, the cleavage of the 3’,5’-
phosphodiester linkage and its isomerization to a 2’,5’-link-
age are pH-independent and accelerated by a factor of 76
and 50 compared with the reactions 3’,5’-UpU, respectively,
mostly due to stabilization of the phosphorane intermediate.
The cleavage of the resulting 2’,5’-isomer is also accelerated
compared with 2’,5’-UpU, although less than the cleavage of
the 3’,5’-isomer, whereas isomerization is not enhanced com-
pared to 2’,5’-UpU. Accordingly, the thermodynamic equili-
brium of the 2’,5’- and 3’,5’-isomers is strongly biased on the
side of the 2’,5’-isomer.
At pH 7–9, at which the amino groups are largely depro-
tonated, the reactions are again pH-independent, but the be-
havior is otherwise quite different from that observed at
pH 3–5. The cleavage of the 3’,5’-and 2’,5’-isomers is acceler-
ated 1000- and 600-fold, respectively, compared with the
rate of the pH-independent cleavage of UpU. The overall
isomerization rate remains unchanged on passing the pKa
values of the amino groups but the equilibrium is now
biased on the side of the 3’,5’-isomer. The reactions most
likely proceed through a minor zwitterionic tautomer having
the attacking sugar hydroxy group ionized and one of the
amino groups protonated. Evidently, proton transfer from
the protonated amino group to a phosphorane oxyanion
plays an important role leading to concurrent occurrence of
reactions through mono- and dianionic intermediates.
The composition of the samples withdrawn at appropriate time intervals
was analyzed by HPLC on a Hypersil-Keystone Aquasil C18 column (4ꢁ
150 mm, 5 mm) using a mixture of acetonitrile and a 60 mmolLꢀ1 acetate
buffer (pH 4.3) containing 0.1 molLꢀ1 ammonium chloride as an eluent.
For the first 8 min, isocratic elution with 100% buffer was used, followed
by a linear acetonitrile gradient from 0 to 30% during 22 min. A constant
flow rate of 1 mLminꢀ1 and a detection wavelength of 260 nm were em-
ployed. The observed retention times (tR, min) for 1a and its hydrolytic
products were as follows: 4.5 (1b), 6.3 (10b), 13.8 (1a), 15.4 (10a), and
15.8 (thymidine). The monomeric products derived from the 3’-linked nu-
cleoside did not retain long enough to be separated with the column
used. The other products were characterized either by spiking with au-
thentic samples or by mass spectrometric analysis (HPLC/ESI-MS). The
peak areas were converted to relative concentrations by calibrating the
system with uridine and thymidine solutions of known concentrations
(the hypochromicity of the dimers 1a and 1b relative to the free nucleo-
sides was assumed to be negligible).
Calculation of the rate constants: No integrated rate law could be de-
rived for the highly complex system of reactions encountered in the pres-
ent study. For this reason, pseudo-first-order rate constants for the vari-
ous partial reactions were calculated by numerical fitting of the experi-
mental data to the much simpler differential rate equations [Eqs. (3)–
(7)]. In cases where the reaction system was simple enough to allow ana-
lytical fitting using an integrated rate law, the results of the numerical fit-
ting were validated by comparing them with the respective results of the
analytical fitting. The rate constants obtained by the two different meth-
ods were found to be in excellent agreement.
d½1aꢁ
ð3Þ
ð4Þ
ð5Þ
ð6Þ
ð7Þ
¼ kꢀ1½1bꢁ ꢀ ðk1 þ k2 þ k4Þ½1aꢁ
¼ k1½1aꢁ ꢀ ðkꢀ1 þ k3 þ k5Þ½1bꢁ
dt
d½1bꢁ
dt
d½10aꢁ
dt
¼ k4½1aꢁ ꢀ ðk6 þ k7Þ½10aꢁ þ kꢀ6½10bꢁ
¼ k5½1bꢁ ꢀ ðkꢀ6 þ k8Þ½10bꢁ þ k6½10aꢁ
¼ k2½1aꢁ þ k3½1bꢁ þ k7½10aꢁ þ k8½10bꢁ
d½10bꢁ
dt
d½Thdꢁ
dt
In cases in which significant buffer catalysis was detected, that is, in for-
mate buffers, the observed rate constants were extrapolated to buffer
concentration zero. In the absence of buffer catalysis, the average of the
rate constants observed at different buffer concentrations was used in-
stead.
Experimental Section
General methods: NMR spectra were recorded on a Bruker Avance 400,
500, or 600 NMR spectrometer and the mass spectra on a Bruker Micro-
TOF-Q high-resolution mass spectrometer. The 1H chemical shifts were
Determination of the pKa values: The pKa values of the two amino
groups of compound 1a were determined by NMR spectrometric titra-
tions in D2O with 0.10 molLꢀ1 sodium dihydrogen phosphate serving as a
12432
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 12424 – 12434

The Role of Stabilization of the Phosphorane Intermediate
FULL PAPER
background buffer. The experimental data (dH vs. pH, Figure 1A) were
fitted to Equation (8).
3.1 Hz), 4.65 (d, 1H, J=6.2 Hz), 4.56 (dd, 1H, J1 =6.2 Hz, J2 =3.1 Hz),
3.86 (dd, 1H, J1 =12.0 Hz, J2 =3.3 Hz), 3.79 (dd, 1H, J1 =12.1 Hz, J2 =
3.3 Hz), 3.64 (dd, 1H, J1 =12.0 Hz, J2 =7.7 Hz), 3.60 (dd, 1H, J1 =
11.9 Hz, J2 =9.9 Hz), 2.93 (dd, 1H, J1 =9.8 Hz, J2 =3.5 Hz), 2.67 (dd, 1H,
J1 =7.7 Hz, J2 =3.4 Hz), 0.94 (s, 9H), 0.92 (s, 9H), 0.16 (s, 3H), 0.13 (s,
3H), 0.11 (s, 3H), 0.06 ppm (s, 3H); HRMS (ESI+): m/z calcd: 525.2428
[M+Na]+; found: 525.2439.
0:5
0:5
ð8Þ
dH ¼ dH1 þ ðdH2 ꢀ dH1Þ½
þ
ꢁ
1 þ 10ðpKa1ꢀpHÞh1 1 þ 10ðpKa2ꢀpHÞh2
In which dH1 and dH2 are the chemical shifts of the fully deprotonated
and fully protonated ionic forms of 1a, respectively and h1 and h2 the
slopes of chemical shift versus pH at the inflection points pKa1 and pKa2,
respectively. Extrapolation to 908C was carried out by linear regression
(pKa vs. 1/T).
2’,3’-Di-O-(tert-butyldimethylsilyl)-4’-C-(trifluoromethylsulfonyloxymeth-
yl)-5’-O-trifluoromethylsulfonyl)uridine (5): Compound
4
(0.50 g,
1.00 mmol) was dried by co-evaporation from anhydrous pyridine. The
residue was dissolved in a mixture of anhydrous CH2Cl2 (10 mL) and pyr-
idine (0.8 mL) and the resulting mixture cooled to ꢀ508C. Trifluorome-
thanesulfonic anhydride (1.0 mL, 5.95 mmol) was added and the reaction
mixture stirred at ꢀ508C for 3 h. The reaction mixture was then dissolved
in CH2Cl2 (25 mL) and washed with saturated aq. NH4Cl (2ꢁ25 mL).
The organic layer was evaporated to dryness and purified by silica gel
chromatography eluting with neat EtOAc, yielding 0.49 g (65%) of the
desired product 5. The work-up and purification were carried out at low
temperature (38C) to minimize the side reaction leading to a cyclic
LNA-type product. Rf =0.9 (EtOAc); 1H NMR (600 MHz, CDCl3): d=
9.64 (s, 1H), 7.23 (d, 1H, J=8.1 Hz), 5.79 (dd, 1H, J1 =8.1 Hz, J2 =
1.7 Hz), 5.44 (d, 1H, J=1.5 Hz), 5.30 (d, 1H, J=12.2 Hz), 4.78 (d, 1H,
J=5.1 Hz), 4.77 (d, 1H, J=11.0 Hz), 4.68 (dd, 1H, J1 =5.3 Hz, J2 =
1.7 Hz), 4.55 (d, 1H, J=11.0 Hz), 4.52 (d, 1H, J=12.2 Hz), 0.93 (s, 9H),
0.93 (s, 9H), 0.17 (s, 3H), 0.14 (s, 3H), 0.13 (s, 3H), 0.12 ppm (s, 3H);
HRMS (ESI+): m/z calcd: 765.1443 [MꢀH]ꢀ; found: 765.1230.
2’,3’-Di-O-(tert-butyldimethylsilyl)-4’-C-(azidomethyl)-5’-azido-5’-deoxy-
uridine (6): Aqueous LiN3 (20%, 4 mL) was evaporated to dryness and
the residue was co-evaporated from anhydrous DMF. The dried salt was
then dissolved in dry DMF (10 mL) and the resulting solution poured
over compound 5 (0.50 g, 0.65 mmol). The reaction mixture was stirred at
room temperature for 16 h, after which it was dissolved in EtOAc
(50 mL) and washed with water (2ꢁ50 mL). The organic layer was
evaporated to dryness and purified by silica gel chromatography eluting
with neat EtOAc. The desired product 6 and the LNA-type side product
(RF=0.8) elute very close to each other; two successive purifications af-
forded 0.27 g (75%) of pure product 6. Rf =0.85 (EtOAc); 1H NMR
(500 MHz, CDCl3): d=8.75 (b, 1H), 7.55 (d, 1H, J=8.1 Hz), 5.78 (dd,
1H, J1 =8.1 Hz, J2 =2.2 Hz), 5.63 (d, 1H, J=2.5 Hz), 4.44 (dd, 1H, J1 =
5.2 Hz, J2 =2.2 Hz), 4.38 (d, 1H, J=5.2 Hz), 4.11 (d, 1H, J=13.7 Hz),
3.91 (d, 1H, J=13.3 Hz), 3.39 (d, 1H, J=13.3 Hz), 3.20 (d, 1H, J=
13.7 Hz), 0.93 (s, 9H), 0.92 (s, 9H), 0.13 (s, 3H), 0.13 (s, 3H), 0.12 (s,
3H), 0.10 ppm (s, 3H); HRMS (ESIꢀ): m/z calcd: 551.2587 [MꢀH]ꢀ;
found: 551.2590.
2’,3’-Di-O-(tert-butyldimethylsilyl)uridine
(2):
Uridine
(10.3 g,
42.13 mmol), tert-butyldimethylsilyl chloride (30.0 g, 199 mmol), and imi-
dazole (25.0 g, 367 mmol) were dissolved in DMF (100 mL). The reaction
mixture was stirred at room temperature for 48 h, after which it was di-
luted with EtOAc (200 mL) and washed with water (3ꢁ200 mL). The or-
ganic layer was evaporated to dryness, affording crude 2’,3’,5’-tri-O-(tert-
butyldimethylsilyl)uridine. The crude product was then dissolved at 08C
in a mixture of THF (200 mL), water (50 mL), and trifluoroacetic acid
(50 mL). The reaction mixture was stirred at 08C for 6 h, after which it
was dissolved in EtOAc (250 mL) and washed with 250 mL aliquots of
saturated aq. NaHCO3 until the aqueous layer was no longer acidic. The
organic layer was evaporated to dryness and the crude product thus ob-
tained purified by silica gel chromatography eluting with a mixture of
ethyl acetate and hexane (stepwise gradient from 1:2, v/v, to neat ethyl
acetate), yielding 15.5 g (80%) of 2. Rf =0.1 (EtOAc/hexane, 1:2, v/v), 0.4
(EtOAc/CH2Cl2, 1:2, v/v); 1H NMR (600 MHz, CDCl3): d=8.07 (brs,
1H), 7.60 (d, 1H, J=8.1 Hz), 5.74 (dd, 1H, J1 =8.1 Hz, J2 =2.4 Hz), 5.47
(d, 1H, J=5.5 Hz), 4.58 (t, 1H, J=5.0 Hz), 4.19 (dd, 1H, J1 =4.3 Hz, J2 =
3.5 Hz), 4.11 (m, 1H), 3.96 (ddd, 1H, J1 =12.2 Hz, J2 =J3 =2.5 Hz), 3.74
(ddd, 1H, J1 =12.2 Hz, J2 =7.7 Hz, J3 =2.0 Hz), 2.94 (dd, 1H, J1 =7.7 Hz,
J2 =2.8 Hz), 0.93 (s, 9H), 0.90 (s, 9H), 0.11 (s, 3H), 0.10 (s, 3H), 0.08 (s,
3H), 0.05 ppm (s, 3H); HRMS (ESI+): m/z calcd: 495.2323 [M+Na]+;
found: 495.2316.
2’,3’-Di-O-(tert-butyldimethylsilyl)-4’-C-(hydroxymethyl)uridine
(4):
Compound 2 (1.00 g, 2.12 mmol) was dissolved in acetonitrile (80 mL). A
mixture of iodoxybenzoic acid (IBX, 0.72 g, 2.57 mmol) and benzoic acid
(added to commercial IBX as a stabilizer, 0.88 g, 7.21 mmol) was added
and the resulting mixture stirred vigorously with heating at reflux for 3 h
(the progress of the reaction was followed by 1H NMR spectroscopy).
The reaction mixture was then filtered and the filtrate evaporated to dry-
ness. The residue was dissolved in CH2Cl2 (100 mL) and washed with sa-
turated aq. NaHCO3 (2ꢁ100 mL). The organic layer was then evaporat-
ed, affording approximately 0.9 g of the aldehyde 3 (despite washing with
aq. NaHCO3, the crude product remains contaminated with benzoic
acid). After 1H NMR spectrometric quantification of the remaining ben-
zoic acid, the crude product was used in the next step without further pu-
rification. 1H NMR (400 MHz, CDCl3): d=9.84 (s, 1H), 9.59 (b, 1H),
7.73 (d, 1H, J=8.1 Hz), 5.84 (d, 1H, J=8.1 Hz), 5.74 (d, 1H, J=5.4 Hz),
4.58 (d, 1H, J=3.4 Hz), 4.36 (dd, 1H, J1 =5.3 Hz, J2 =4.1 Hz), 4.25 (t,
1H, J=3.8 Hz), 0.94 (s, 9H), 0.89 (s, 9H), 0.14 (s, 3H), 0.13 (s, 3H), 0.07
(s, 3H), 0.03 ppm (s, 3H).
2’-O-(4,4’-Dimethoxytrityl)-4’-C-(azidomethyl)-5’-azido-5’-deoxyuridine
(8): Compound 6 (0.60 g, 0.91 mmol) was dissolved in THF (20 mL). Trie-
thylamine trihydrofluoride (1.0 mL, 6.13 mmol) was added and the result-
ing mixture stirred at 408C for 3 days. The solution was then partially
evaporated and residue passed through a short silica gel column eluting
with a mixture of CH2Cl2 and MeOH (10:1, v/v). Product 7 thus obtained
was then dissolved in anhydrous pyridine (50 mL). 4-Dimethylaminopyri-
dine (20 mg) and 4,4’-dimethoxytrityl chloride (1.80 g, 5.31 mmol) were
added and the resulting mixture stirred at 408C for 24 h. The reaction
mixture was then evaporated and the residue purified by silica gel chro-
matography eluting with a mixture of CH2Cl2, MeOH, and Et3N (100:3:1,
v/v). Repeated purifications yielded 0.34 g (50%) of the desired product
8. The correct site of tritylation (the 2’-position) was established by an
HMBC correlation between the 4ꢂ carbon and the 3’-OH proton. In fact,
no trace of the 3’-O-tritylated isomer was observed in the product mix-
ture. Rf =0.2 (CH2Cl2/MeOH/Et3N, 100:3:1, v/v); 1H NMR (500 MHz,
CDCl3): d=7.42 (m, 2H), 7.34 (m, 4H), 7.26 (m, 4H), 6.83 (m, 4H), 6.21
(d, 1H, J=7.7 Hz), 5.68 (d, 1H, J=8.1 Hz), 4.52 (dd, 1H, J1 =7.6 Hz,
J2 =5.0 Hz), 3.79 (s, 3H), 3.77 (s, 3H), 3.47 (d, 1H, J=13.0 Hz), 3.47 (d,
1H, J=13.0 Hz), 3.38 (d, 1H, J=13.0 Hz), 3.38 (d, 1H, J=13.0 Hz), 2.59
(d, 1H, J=4.9 Hz), 2.48 ppm (b, 1H); HRMS (ESIꢀ): m/z calcd:
625.2165 [MꢀH]ꢀ; found: 625.2150.
The crude product 3 was dissolved in 1,4-dioxane (8.0 mL). 37% aq.
formaldehyde (0.7 mL) and 2m aq. NaOH (1.15 equiv, plus the amount
needed to neutralize the remaining benzoic acid). The reaction mixture
was stirred at room temperature for 4 h, after which it was cooled to 08C
and NaBH4 (90 mg, 2.38 mmol) was added. The cooling bath was re-
moved and the stirring was continued at room temperature for 1 h, after
which the reaction mixture was neutralized with 1m aq. acetic acid
(3 mL). The solution was then dissolved in CH2Cl2 (50 mL) washed twice
with saturated aq. NaHCO3 (2ꢁ50 mL). The organic layer was evaporat-
ed to dryness and the residue purified by silica gel chromatography. The
eluent was a mixture of EtOAc and CH2Cl2 (1:2, v/v) until the first signif-
icant fraction (starting material 2) eluted, after which it was changed to
neat EtOAc, yielding 0.37 g (35%) of the desired product 4. In addition,
0.25 g of the starting material 2 could be recovered. Rf =0.1 (EtOAc/
CH2Cl2, 1:2, v/v); 1H NMR (500 MHz, CDCl3): d=8.93 (s, 1H), 7.47 (d,
1H, J=8.1 Hz), 5.74 (dd, 1H, J1 =8.1 Hz, J2 =2.2 Hz), 5.55 (d, 1H, J=
2’-O-(4,4’-Dimethoxytrityl)-4’-C-(azidomethyl)-5’-azido-5’-deoxyuridylyl-
3’,5’-[3’-O-(4,4’-dimethoxytrityl)thymidine] (9): Compound
8
(220 mg,
Chem. Eur. J. 2013, 19, 12424 – 12434
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12433

T. Lçnnberg et al.
0.351 mmol) was dried by co-evaporation from pyridine. The residue was
dissolved in a 0.25m solution of 5-benzylthio-1H-tetrazole in anhydrous
MeCN (2.30 mL, 0.575 mmol). 3’-(4,4’-Dimethoxytrityl)-thymidine 5’-[(2-
cyanoethyl)-(N,N-diisopropyl)]phosphoramidite (370 mg, 0.497 mmol)
was added and the resulting mixture allowed to react for 5 min. The
phosphite triester formed was then oxidized by dropwise addition of a
0.12m solution of I2 in a mixture of THF, water and 2,6-lutidine (4:2:1, v/
v) until the brown color of elemental iodine remained. The solution was
then diluted with CH2Cl2 (20 mL) and washed with 5% aq. NaHSO3 (2ꢁ
20 mL). The organic layer was evaporated to dryness and the residue dis-
solved in a mixture of pyridine and Et3N (3:1, v/v, 50 mL). The reaction
mixture was stirred at room temperature for 16 h, after which it was
evaporated to dryness. The residue was purified by silica gel chromatog-
raphy eluting with a mixture of CH2Cl2, MeOH, and Et3N (100:10:1, v/v)
to yield 330 mg (75%) of the desired product 9 as a triethylammonium
salt. Rf =0.2 (CH2Cl2/MeOH/Et3N, 100:10:1, v/v); 1H NMR (600 MHz,
CDCl3): d=8.61 (m, 1H), 7.74 (b, 1H), 7.54 (m, 2H), 7.43 (m, 4H), 7.36
(m, 2H), 7.31 (m, 4H), 7.22 (m, 4H), 6.77 (m, 9H), 6.44 (dd, 1H, J1 =
8.9 Hz, J2 =5.8 Hz), 6.36 (d, 1H, J=8.0 Hz), 5.38 (d, 1H, J=8.0 Hz), 5.20
(dd, 1H, J1 =10.1 Hz, J2 =6.6 Hz), 5.01 (dd, 1H, J1 =6.5 Hz, J2 =2.6 Hz),
4.85 (b, 1H), 4.31 (d, 1H, J=4.7 Hz), 4.05 (d, 1H, J=13.6 Hz), 3.93 (m,
1H), 3.79 (b, 1H), 3.72 (s, 3H), 3.71 (s, 3H), 3.71 (s, 3H), 3.70 (s, 3H),
3.52 (d, 1H, J=13.6 Hz), 3.51 (d, 1H, J=13.5 Hz), 1.83 (s, 3H), 1.75 ppm
(m, 2H); 31P NMR (600 MHz, CDCl3): d=ꢀ1.31 ppm; HRMS (ESIꢀ):
m/z calcd: 1231.3932 [MꢀH]ꢀ; found: 1231.3893.
[2] R. T. Raines, Chem. Rev. 1998, 98, 1045–1066.
[3] a) J. C. Cochrane, S. A. Strobel, Acc. Chem. Res. 2008, 41, 1027–
1035; b) P. C. Bevilacqua, R. Yajima, Curr. Opin. Chem. Biol. 2006,
10, 455–464; c) S. A. Strobel, Nat. Chem. Biol. 2005, 1, 5–6.
[4] H. Lçnnberg, Org. Biomol. Chem. 2011, 9, 1687–1703.
[5] G. M. Emilsson, S. Nakamura, A. Roth, R. R. Breaker, RNA 2003,
9, 907–918.
[6] C. S. Lꢃpez, O. N. Faza, A. R. de Lera, D. M. York, Chem. Eur. J.
2005, 11, 2081–2093.
[7] a) N. H. Williams, B. Takasaki, M. Wall, J. Chin, Acc. Chem. Res.
1999, 32, 485–493; b) H. Linjalahti, G. Feng, J. C. Mareque-Rivas, S.
Mikkola, N. H. Williams, J. Am. Chem. Soc. 2008, 130, 4232; H. Ko-
rhonen, S. Mikkola, N. H. Williams, Chem. Eur. J. 2012, 18, 659.
[8] K. N. Dalby, A. J. Kirby, F. Hollfelder, J. Chem. Soc. Perkin Trans. 2
1993, 1269–1281.
[9] a) R. K. O. Sigel, A. M. Pyle, Chem. Rev. 2007, 107, 97–113; b) J.
Weston, Chem. Rev. 2005, 105, 2151–2174.
[10] J. L. Wilcox, A. K. Ahluwalia, P. C. Bevilacqua, Acc. Chem. Res.
2011, 44, 1270–1279.
[11] M. Ora, H. Linjalahti, H. Lçnnberg, J. Am. Chem. Soc. 2005, 127,
1826–1832.
[12] A. Kiviniemi, T. Lçnnberg, M. Ora, J. Am. Chem. Soc. 2004, 126,
11040–11045.
[13] T. Lçnnberg, J. Korhonen, J. Am. Chem. Soc. 2005, 127, 7752–7758.
[14] T. Lçnnberg, M. Ora, S. Virtanen, H. Lonnberg, Chem. Eur. J. 2007,
13, 4614–4627.
4’-C-(Aminomethyl)-5’-amino-5’-deoxyuridylyl-3’,5’-thymidine
(1a):
[15] Y. Hisamatsu, K. Hasada, F. Amano, Y. Tsubota, Y. Wasada-Tsutsui,
N. Shirai, S.-i. Ikeda, K. Odashima, Chem. Eur. J. 2006, 12, 7733–
7741.
[16] X.-F. Zhu, H. J. Williams, A. I. Scott, J. Chem. Soc. Perkin Trans. 1
2000, 2305–2306.
[17] J. D. More, N. S. Finney, Org. Lett. 2002, 4, 3001–3003.
[18] K. D. Nielsen, F. Kirpekar, P. Roepstorff, J. Wengel, Bioorg. Med.
Chem. 1995, 3, 1493–1502.
[19] C.-S. Yu, R.-T. Wang, L.-W. Chiang, M.-H. Lee, Tetrahedron Lett.
2007, 48, 2979–2982.
[20] H. Staudinger, J. Meyer, Helv. Chim. Acta 1919, 2, 635–646.
[21] P. Jarvinen, M. Oivanen, H. Lçnnberg, J. Org. Chem. 1991, 56,
5396–5401.
[22] M. Oivanen, M. Ora, H. Almer, R. Strçmberg, H. Lçnnberg, J. Org.
Chem. 1995, 60, 5620–5627.
[23] H. Lçnnberg in Ribonucleases, Vol. 26 (Ed.: A. W. Nicholson),
Springer, Berlin, 2011, pp. 343–366.
[24] C. S. Lꢃpez, O. N. Faza, B. A. Gregersen, X. Lopez, A. R. de Lera,
D. M. York, ChemPhysChem 2004, 5, 1045–1049.
[25] T. Uchimaru, M. Uebayasi, T. Hirose, S. Tsuzuki, A. Yliniemelꢄ, K.
Tanabe, K. Taira, J. Org. Chem. 1996, 61, 1599–1608.
[26] Y. Yang, Q. Cui, J. Phys. Chem. B 2009, 113, 4930–4939.
[27] M. Boero, K. Terakura, M. Tateno, J. Am. Chem. Soc. 2002, 124,
8949–8957.
[28] a) R. Kluger, F. Covitz, E. Dennis, L. D. Williams, F. H. Westheimer,
J. Am. Chem. Soc. 1969, 91, 6066–6072; b) J. P. Guthrie, J. Am.
Chem. Soc. 1977, 99, 3991–4001; c) J. E. Davies, N. L. Doltsinis,
A. J. Kirby, C. D. Roussev, M. Sprik, J. Am. Chem. Soc. 2002, 124,
6594–6599; d) X. Lopez, M. Schaefer, A. Dejaegere, M. Karplus, J.
Am. Chem. Soc. 2002, 124, 5010–5018.
Compound 9 (50 mg, 0.041 mmol) was dissolved in anhydrous pyridine
(4.0 mL). Triphenylphosphine (0.40 g, 1.53 mmol) was added and the re-
sulting mixture stirred at 508C for 5 h. The reaction mixture was then al-
lowed to cool down to room temperature, pyridine (5.0 mL) and 33% aq.
ammonia (3.0 mL) were added and stirring continued at room tempera-
ture for 16 h. The reaction mixture was then evaporated to dryness, and
the residue dissolved in MeOH (2.0 mL). This solution was poured in a
separatory funnel containing 1 mmolLꢀ1 aq. HCl (50 mL) and CH2Cl2
(100 mL) and the resulting mixture shaken for 5 min. After 1 h, the aque-
ous layer was evaporated, affording approximately 15 mg (60%) of the
desired product 1a in purity adequate for the kinetic experiments. For
characterization and the NMR spectrometric titrations, a small amount
of 1a was purified by RP HPLC on a Hypersil-Keystone Aquasil C18
column (4ꢁ150 mm, 5 mm), eluting with a mixture of acetonitrile and a
50 mmolLꢀ1 ammonium formate buffer. For the first 8 min, isocratic elu-
tion with 100% buffer was used, followed by a linear acetonitrile gradi-
ent from 0 to 30% during 22 min. A constant flow rate of 1 mLminꢀ1
and a detection wavelength of 260 nm were employed. The desired prod-
uct eluted at 22 min. During the HPLC purification, a significant fraction
of 1a was lost to the side reaction discussed above (Scheme 2, Route E),
making the method impractical for larger scale preparations. 1H NMR
(600 MHz, D2O): d=7.54 (d, 1H, J=7.9 Hz), 7.54 (d, 1H, J=1.1 Hz),
6.22 (dd, 1H, J1 =J2 =6.7 Hz), 5.76 (d, 1H, J=8.0 Hz), 5.54 (d, 1H, J=
2.7 Hz), 5.07 (dd, 1H, J1 =9.6 Hz, J2 =6.2 Hz), 4.78 (dd, 1H, J1 =6.2 Hz,
J2 =2.8 Hz), 4.46 (m, 1H), 4.14 (m, 1H), 4.04 (m, 2H), 3.45 (d, 1H, J=
14.3 Hz), 3.41 (d, 1H, J=14.1 Hz), 3.34 (d, 1H, J=14.1 Hz), 3.20 (d, 1H,
J=14.3 Hz), 2.30 (m, 2H), 1.81 ppm (d, 3H, J=1.0 Hz); 31P NMR
(600 MHz, D2O): ꢀ0.63. HRMS (ESIꢀ): m/z calcd: 575.1508 [MꢀH]ꢀ;
found: 575.1476.
[29] R. J. G. Thatcher, R. Kluger in Advances in Physical Organic Chem-
istry, Vol. 25 (Ed.: D. Bethell), Academic Press, 1989, pp. 99–265.
[30] T. Lçnnberg, S. Kralikova, I. Rosenberg, H. Lçnnberg, Coll. Czech.
Chem. Commun. 2006, 71, 859–870.
Acknowledgements
[31] T. Salthammer, Angew. Chem. 2013, 125, 3402–3410; Angew. Chem.
Int. Ed. 2013, 52, 3320–3327.
Financial support from the FP7 Marie Curie Actions is gratefully ac-
knowledged.
[1] a) D. M. Perreault, E. V. Anslyn, Angew. Chem. 1997, 109, 470–490;
Angew. Chem. Int. Ed. Engl. 1997, 36, 432–450; b) M. Oivanen, S.
Kuusela, H. Lçnnberg, Chem. Rev. 1998, 98, 961–990.
Received: May 3, 2013
Published online: July 29, 2013
12434
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 12424 – 12434