Synthesis of Nucleosides through Direct Glycosylation of Nucleobases with 5-O-Monoprotected or 5-Modified Ribose: Improved Protocol, Scope, and Mechanism

Article abstract of DOI:10.1002/chem.201604955

Simplifying access to synthetic nucleosides is of interest due to their widespread use as biochemical or anticancer and antiviral agents. Herein, a direct stereoselective method to access an expansive range of both natural and synthetic nucleosides up to a gram scale, through direct glycosylation of nucleobases with 5-O-tritylribose and other C5-modified ribose derivatives, is discussed in detail. The reaction proceeds through nucleophilic epoxide ring opening of an in situ formed 1,2-anhydrosugar (termed “anhydrose”) under modified Mitsunobu reaction conditions. The scope of the reaction in the synthesis of diverse nucleosides and other 1-substituted riboside derivatives is described. In addition, a mechanistic insight into the formation of this key glycosyl donor intermediate is provided.

Full text of DOI:10.1002/chem.201604955

A Journal of
Accepted Article
Title: Synthesis of nucleosides through direct glycosylation of
nucleobases with 5-O-mono-protected or 5-modified ribose.
Improved protocol, scope and mechanism
Authors: A. Michael Downey, Radek Pohl, Jana Roithova, and Michal
Hocek
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201604955
Link to VoR: http://dx.doi.org/10.1002/chem.201604955
Supported by

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
Synthesis of nucleosides through direct glycosylation of
nucleobases with 5-O-monoprotected or 5-modified ribose.
Improved protocol, scope, and mechanism
A. Michael Downey,[a] Radek Pohl,[a] Jana Roithová,[b] and Michal Hocek*[a,b]
Abstract: Simplifying access to synthetic nucleosides is of interest
due to their widespread use as biochemical or as anticancer and
antiviral agents. Herein we provide a full and in-depth report on a
direct stereoselective method to access an expansive range of both
natural and synthetic nucleosides up to a gram scale, through direct
glycosylation of nucleobases with 5-O-tritylribose and other C5-
modified ribose derivatives. The reaction proceeds via nucleophilic
epoxide ring opening of an in situ formed 1,2-anhydrosugar (termed
“anhydrose”) using modified Mitsunobu reaction conditions. We
present the scope of the reaction in the synthesis of diverse
nucleosides as well as other 1-substituted riboside derivatives. In
addition, we provide mechanistic insight into the formation of this key
glycosyl donor intermediate.
iterative protection and deprotection sequences, designing
conditions that provide the product with good regio- and
stereoselectively is also required and these extra challenges are
often large ones.
Although several approaches to obtain stereoselective
glycosylation products are known, they suffer from the extensive
utilization of protecting groups (Fig. 1). In the synthesis of -
ribofuranosyl nucleosides, the most commonly used approach is
the Vorbrüggen variant[5] of the silyl-HilbertJohnson reaction
which reacts a perbenzolyated ribofuranoside in the presence of
a Lewis acid usually TMSOTf or SnCl4, thus generating a
benzoxonium ion in situ that blocks the -face of the molecule
from nucleophilic attack, known as a neighboring group effect. A
silylated nucleobase is then added and after extensive heating the
desired protected -nucleoside analog can be isolated (Fig. 1a).[6]
In the synthesis of 2′-deoxynucleosides, the Fischer method[7] is
typically used. The halogen (chloride or bromide) of a carefully
prepared fully protected -halogenose can be displaced in an SN2
fashion by the metal salt of a nucleobase to yield the desired -2′-
deoxynucleoside (Fig. 1b).[8]
Introduction
Increasing efficiency to create organic molecules is of paramount
importance to synthetic chemists as improved routes have
implications that extend into all areas of science including
medicine and biology. Often protecting groups must be employed
to avoid unwanted reactivity at one or more functional groups in a
molecule when a certain transformation is being effected. One
transformation that almost invariably depends on extensive and
iterative protecting group strategies is glycosylation – either of
heterocycles (nucleobases) to form nucleosides[1] or other
carbohydrate molecules to provide di- or oligosaccharides.[2]
Synthetic nucleosides have widespread use as anticancer and
antiviral agents[3] and synthetic oligosaccharides are used in the
study of immunogenicity and vaccine design,[4] therefore,
improved methods to create these molecules may have
tremendous effects in medicine. Besides the obvious burden of
Figure 1. Selected classical diastereoselective glycosylation strategies for the
synthesis of nucleosides that employ extensive protecting groups.
[a]
A. M. Downey, Dr. R. Pohl, Prof. Dr. M. Hocek
Institute of Organic Chemistry and Biochemistry
Czech Academy of Sciences
One commonly known method to access a host of pyranosyl
glycosides, studied extensively by Danishefsky and
collaborators,[9] employs the use of a fully protected intermediate
1,2-anhydrosugar that can be isolated carefully after treatment of
the parent glycal with dimethyl dioxorane (DMDO). Protected
disaccharides can then be generated via epoxide ring opening in
the presence of many Lewis acids or other pyranosides can be
synthesized by nucleophilic ring opening to afford (for example)
fluoro, thiol and cyano pyranosides (Fig. 2).[10]
Flemingovo nám. 2, 16610 Prague 6 (Czech Republic)
hocek@uochb.cas.cz
http://www.uochb.cas.cz/hocekgroup
Prof. Dr. J. Roithová, Prof. Dr. M. Hocek
Department of Organic Chemistry, Faculty of Science
Charles University in Prague
[b]
Hlavova 8, 12843 Prague 2 (Czech Republic)
Supporting information for this article is given via a link at the end of
the document.((Please delete this text if not appropriate))
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
In 2015, our group in collaboration with Mahrwald and
coworkers,[16] expanded upon their findings and also those of
Grynkiewicz[13] to optimize conditions amenable for the synthesis
of -ribosyl nucleosides. In this seminal communication, the
nucleobase was the limiting reagent and was added first and the
reaction was performed in the presence of PBu3 and diisopropyl
azodicarbozylate (DIAD)—solely the -anomer was formed,
however, the undesired thermodynamic pyranosyl regioisomer
presented almost exclusively when completely unprotected ribose
was used. This issue could be circumvented by using the highly
acid-labile 4-methoxytrityl (MMTr) protecting group to provide the
Figure 2. Diastereoselective glycosylation via epoxide ring opening.
Protecting group-free methods to access simple O-glycosides
have been studied and date back to the Fischer glycosylation (Fig.
3a).[11] Modern examples typically employ Lewis acids,[12] but are
limited by the need to use stoichiometric or excessive quantities
of the often toxic acid and long reaction times and high
temperatures are usually required. Furthermore, these conditions
suffer from poor or a complete lack of regio- and stereoselectivity
of the glycoside product. To date very few neutral glycosylation
conditions have been reported. The first neutral glycosylation
conditions subjected glucose in the presence of an excess of
phenol to Mitsunobu conditions to afford the -phenolic glycoside
in moderate yield and reasonable diastereoselectivity (Fig. 3b).[13]
However, the study was limited in scope to only phenolic
acceptors. Much more recently, Mahrwald and coworkers[14]
developed neutral catalytic conditions to provide aliphatic or
benzylic glycosides using several furanoses in good to moderate
yield, however, generally the conditions provided a mixture of
anomers and often the undesired pyranoside regioisomer
prevailing. Furthermore, in all cases the alcohol or acid acceptors
were used in large excess or as solvent and little mechanistic
insight was provided (Fig. 3c). Finally, we note also in a recent
protecting group-free total synthesis of two ellagitannins that the
Mitsunobu reaction was deployed to form a glycopyranosyl -
phenolic ester from the corresponding acid and unprotected D-
glucose in the first step of the constructions (not shown).[15]
desired
-furanosyl
nucleosides
in
a
one-pot
glycosylationdeprotection strategy (Fig. 4). Although a directing
effect of the C2-OH of the ribosyl donor was confirmed, the
mechanism was otherwise undetermined.
Figure 4. Highlights of this work and our previous protecting group-free
glycosylation strategies.
In this full report we significantly expand the scope of the
procedure by tweaking the reaction conditions to form first a 1,2-
anhydrosugar (termed “anhydrose”), followed by nucleophilic
epoxide ring opening by the acceptor thus enabling access not
only to nucleosides, but to other synthetic precursors of biological
value.
Results and Discussion
It was our goal from the outset of the study to design conditions
that would facilitate access to stereoselective glycosylation
products of biologically relevant molecules (i.e. ribofuranosides)
on a semi-preparative to preparative scale. Knowing from the
previous studies[14, 16] that direct access to ribofuranosides from
unprotected ribose is not possible in relevant yield, we began the
study with a 5-O-monoprotected ribose moiety that can be
synthesized and is stable at a molar scale, but is still labile enough
to be cleaved in situ by simple treatment with acid. To our initial
Figure 3. Pioneering works on protecting group-free glycosylation strategies.
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
dismay, the 4-methoxytrityl (MMTr) moiety employed in the
previous study proved not to have a bench life long enough to be
synthesized at a decagram scale and the 5-O-MMTr-ribose
product is oily in nature, further complicating its long term storage
and use. Fortuitously, the cheaper trityl (triphenylmethyl)
protecting group could be substituted while still offering the same
acid lability. 5-O-Tritylribose (1) is a solid, stable indefinitely at
room temperature, and decamole syntheses have been described
(see SI).[17]
slightly lowering the molar equivalents (1.6 equiv. and 1.5 equiv.,
respectfully) of P(nBu)3 and by using 1,1′-(dicarbonyl)dipiperidine
(ADDP) as our electron acceptor (Fig. 5b), nearly quantitative
1
conversion to anhydrose 2 occurred as determined by H NMR
analysis with none of the hydrazine byproduct reacting (Fig 5c).
Also, anhydrose 2 was shown to be stable for at least one month
in situ if left in the absence of moisture.
With these conditions in hand, we next pursued suitable
nucleophiles to be glycosylated at a semi-preparative scale (100
mg of 1) that can act as synthetic precursors to obtain biologically
relevant molecules. To our pleasure, both fluoride 3b and thiol 3c
(potential glycosylation precursors), could be isolated in good
yield (Scheme 1). Initially, attempts at obtaining click precursor 3a
and C-nucleoside precursor 3d directly from NaN3 and NaCN,
respectively, failed (likely due to low solubility). However, by
switching to 1,1,1′,1′-tetramethylguanidinium azide (formed in
situ) and by using freshly dried NEt4CN both azide 3a and cyanide
3d could be provided in good yield. Despite the small size of these
nucleophiles, in all cases, the reaction was perfectly
stereoselective as only the -anomer was formed.
Figure 5. Optimizing the formation of anhydrose 2. a) Formation of the
anhydrose under our classical conditions. Reaction conditions: i) 5-O-trityl-D-
ribose (1, 1.0 equiv.), DIAD (2.1 equiv.), P(nBu)3 (2.0 equiv.), MeCN (0.1 M), 0
oC, 15 min; b) Changing the electron acceptor to ADDP removes any traces of
a side reaction. Reaction conditions: ii) 1, (1.0 equiv.), P(nBu)3 (1.6 equiv.),
ADDP (1.5 equiv.), MeCN (0.06 M), r.t., 15 min; c) The 1H NMR spectrum before
(top) and after (bottom) P(nBu)3 and ADDP were added showing nearly
quantitative conversion to the anhydrose . DIAD = diisopropyl azodicarboxylate;
ADDP = 1,1′-(dicarbonyl)dipiperidine.
Scheme 1. Epoxide ring opening utilizing –N, –S, –C, and –F nucleophiles. a)
Reaction conditions: i) 5-O-trityl-D-ribose (1, 1.0 equiv.), P(nBu)3 (1.6 equiv.),
ADDP (1.5 equiv.), MeCN (0.06 M), r.t., 15 min; ii) Nu: (3.0 or 4.0 equiv.), DMF
(0.75 M or 1 M, resp.), 45 oC, 12 h. b) Examples and isolated yields. Tr = trityl;
ADDP = 1,1′-(dicarbonyl)dipiperidine. *Yield in brackets was isolated at a 1.00
g scale.
In order to expand the scope of our methodology to include
other nucleophile acceptors, we postulated that we should subject
monoprotected compound 1 to our previous conditions in the
absence of nucleobases or other nucleophiles to determine
unequivocally the role of the C2-OH group (Fig. 5). We first
subjected 1 to reaction with P(nBu)3 and DIAD under our classical
conditions in the absence of any nucleobase. To our delight we
We strove next to expand the scope of nucleosides
synthesizable under our improved conditions and to demonstrate
that the monoprotected (Scheme 2) can be isolated at semi-
preparative and preparative scale. By first forming anhydrose 2 in
situ and then adding the deprotonated nucleobase (by using 1.0
or 2.0 equiv. of NaH) to effect the glycosylation, all possible side
reactions between the nucleobase and the Mitsunobu reagents
could be circumvented. In all instances the reaction was perfectly
stereoselective (by 1H NMR analysis of the crude reaction
mixtures) and only the-anomer was formed. We were able to
glycosylate 5-fluorouracil, deazapurines as well as all four major
nucleobases contained in natural RNA. The pyrimidine
1
were able to detect the formation of anhydrose 2 by H NMR,
characterized by the high upfield shift of the H-2 proton,[18] but by
TLC analysis two spots appeared suggesting the presence of an
unwanted side product. Attempts to isolate the epoxide by
column chromatography failed (only ~65 % starting material),
however, the other, unwanted spot, could be isolated (with some
contamination from trinbutylphosphine oxide) in 2030% overall
yield. The mass spectrometry and NMR data confirmed that some
of the hydrazine byproduct formed in the Mitsunobu reaction
actually reacted to form dicarboxylate 2a (Fig. 5a). Undaunted,
we sought an alternative electron acceptor and to our delight by
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
for 100 mg scale, 1.5 equiv. for 1.00 g scale), NaH (1.5 or 3.0 equiv, see SI),
DMF (~0.35 M), r.t., 12 h. b) Examples and isolated yields. Tr = trityl; ADDP =
1,1′-(dicarbonyl)dipiperidine
nucleosides 4a-c were obtained in moderate 39-45% yields. The
glycosylations of purines and deazapurines proceeded with
higher conversions. The 6-chloropurine (4d), 6-chloro-7-
deazapurine (4h), 6-chloro-7-iodo-7-deazapurine (4g) and 7-
deazaguanine (4i) nucleosides were obtained cleanly as single
regioisomers in moderate to good yields. On the other hand, in
the case of adenine and guanine, inseparable mixtures of the
desired N9 nucleosides 4e and 4f with unwanted N3 (in adenine,
4ex) or N7 (in guanine, 4fx) regioisomers were obtained. It was
also crucial to demonstrate that the conditions were compatible
with gram scale synthesis and pleasingly all nucleosides suffered
little decrease in yield when scaled up with the exception of
analog 4c, which we attribute to struggles with purification rather
than glycosylation, as the 1H NMR of the crude reaction mixtures
were very similar.
The poor regiochemical control of our reaction conditions
when utilizing adenine as a substrate led us to investigate the
regioselectivity of the adenine scaffold with the nitrogen atoms
systematically removed from the purine ring (purine numbering
used in the main text for convenience, correct systematic
nomenclature used in the SI). Interestingly, albeit unfortunately,
when glycosylating 7-deazaadenine with anhydrose 2 the
exclusive product was N3 regioisomer 4j (and in very modest
yield) with none of the desired regioisomer being detected.
Glycosylation of 1-deazaadenine also provided the N3
regioisomer (4kx) as the major product (38 % isolated yield),
however, gratefully, the desired N9 could also isolated in very
poor yield (17 %). When 3-deazaadenine was utilized as the
acceptor, we were pleased to discover the desired N9 regioisomer
presented in good isolated yield (49 %) and in large excess over
the undesired N7 regioisomer 4lx (only 8 % isolated yield.
Shortcomings in our initial communication included
glycosylation of cytosine, guanine, 5-fluorouracil, and 7-
deazapurines that were unsubstituted at position 7. Using our
improved procedure in this report these nucleobases are all able
to undergo the glycosylation in moderate yield. We were also able
to show that our conditions are also compatible with other
deazapurines as well, as manifested by the results of our
deazadenine study, further demonstrating the power of our
methodology. As observed previously, the yields of the pyrimidine
nucleosides were still lower than that of purines, and, in the case
of adenine and guanine, some of the undesired N3 (4ex) and N7
(4fx) glycosylated nucleosides as inseparable mixtures,
respectively, presented in both cases.
We next desired to prove that all analogs could be
deprotected at a preparative scale in a one pot process without
any degradation at the anomeric position (Scheme 3). We first
attempted to duplicate our previously reported conditions (1 M
HCl(aq)) however, 7-deazapurine analogs 5h and 5i suffered
from some cleavage at the anomeric position. To our delight, by
simply switching the acid to 90 % trifluoroacetic acid (TFA) in
water and by stirring at room temperature for 12 h, the trityl group
cleavage could be effected while still preserving the integrity of
the nucleoside at the anomeric center. In order to avoid any
problems of contamination of the products by trifluoroacetate salts
and to avoid cleavage of the nucleobase during work-up, a slight
increase in pH of the reaction mixture was completed by using

Dowex 1 × 2 resin in the OH form (to pH ~1). In all cases the
isolated yields were comparable to that of the yields of the
monoprotected products, suggesting the deprotection proceeds
quantitatively and without any destruction of the nucleoside.
Fortuitously, the undesired regioisomers of adenosine 5e and
guanosine 5f could also be separated either through selective
crystallization of the product (adenosine) or by iterative
chromatography (guanosine).
Scheme 2. Epoxide ring opening utilizing nucleobases to form nucleosides. a)
Reaction conditions: i) 5-O-trityl-D-ribose (1, 1.0 equiv.), P(nBu)3 (1.6 equiv.),
ADDP (1.5 equiv.), MeCN (0.06 M), r.t., 15 min; ii) nucleobase NB (2.0 equiv.
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
than at the desired position (N9) on the pyrrole or imidazole ring.
It is also probable that in the instance of 5l the undesired N7
regioisomer (5lx) was also nearly completely destroyed as
attempts to isolate it in pure form failed and was poorly visible in
1
the H NMR spectrum of the crude reaction mixture. Finally, we
also deprotected potential C-nucleoside precursor 3d to afford 6
without any issue using these conditions.
In our previous communication[16] we employed the slightly
more acid-labile 4-methoxytrityl group and the acid used for
deprotection was stronger (aq. HCl). These two factors likely
contributed to a much faster deprotection rate (15 min to 1 h)
whereas using the procedure reported here the deprotection
requires stirring for 12 h. However, the substrate scope has been
substantially expanded. 7-Deazapurine nucleoside analog 5h is a
precursor in the synthesis of the antibiotic tubercidin[19] as well as
a
Suzuki coupling precursor for the installation of small
heterocycles (such as thiophene) to provide cytostatic agents.[20]
Additionally, both deazaadenosine regioisomers 5k and 5l were
not noticeably destroyed using our deprotection conditions which
could have implications in the study and development of
ribozymes.[21] (See the SI, Section 5 for the complete and better
clarified analysis of the deazaadenine study.) As a result of the
vastly broadened substrate scope available under our new and
milder conditions, the longer reaction time is not a significant
factor.
To overcome the poor regioselectivity of the glycosylation
reaction and low yields in the synthesis of adenosine (5e), we
pursued alternate conditions to circumvent the issue. We first
investigated the possibility of a one-pot three step process
whereby first the glycosylation is carried out in 1,4-dioxane as the
solvent and using 6-chloropurine as the acceptor. Then after 12 h
adding NH3(aq) and heating at reflux for 24 h as per standard
heterocycle amination conditions. The trityl group would then be
cleaved as described above. Unfortunately, the glycosylation
does not proceed to a significant extent when 1,4-dioxane is used
as solvent which left us to abandon this strategy. Luckily, by
simply switching back to the reported conditions using
MeCN/DMF and then evaporating the solvent, this problem can
be bypassed. The crude reaction could then be taken up in 1,4-
o
dioxane, NH3(aq) added, and the amination effected at 100 C
Scheme 3. Preparative scale (2.5 mmol) one-pot glycosylationdeprotection
strategy. a) Reaction conditions: i) 5-O-trityl-D-ribose (1, 1.0 equiv.), P(nBu)3
(1.6 equiv.), ADDP (1.5 equiv.), MeCN (0.06 M), r.t., 15 min; ii) nucleobase NB
(1.5 equiv.), NaH (1.5 or 3.0 equiv, see SI), DMF (~0.35 M), r.t., 12 h; iii)
TFA:H2O (9:1), 12 h, r.t. then Dowex 1 × 2 resin (OH form) to pH 1. b) Examples
and isolated yields. *Reaction carried out at 1.3 mmol scale. Tr = trityl. ADDP =
1,1′-(dicarbonyl)dipiperidine. TFA = trifluoroacetic acid.
over 24 h (Scheme 4). The solvents were then once again
evaporated and the trityl group cleaved using TFA as in Scheme
3. After purification, pure adenosine was afforded in 52 % yield.
In our systematic deazaadenine study (compounds 5j5l), we
were very surprised to observe either partial or complete
destruction of the undesired N3 regioisomers upon subjugation of
the nucleosides to the one-pot deprotection conditions.
Regioisomer 4j could not be deprotected (complete
decomposition, 5j not observed) at all and the N3 regioisomer
(5kx) of 5k suffered from a 26 % decrease in overall yield (only
12 % isolated yield. This infers that glycosylation at the N3
position of the pyrimidine ring yields a much less stable product
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
Scheme 4. Improved synthesis of adenosine (5e). Reaction conditions: i) 5-O-
trityl-D-ribose (1, 1.0 equiv.), P(nBu)3 (1.6 equiv.), ADDP (1.5 equiv.), MeCN
(0.06 M), r.t., 15 min; ii) 6-chloropurine (1.5 equiv.), NaH (1.5 equiv), DMF (~0.35
M), r.t., 12 h; iii) NH3 (aq), 1,4-dioxane (0.06 M), 100 °C, 24 h. iv) TFA:H2O (9:1),
MeCN (0.06 M), 12 h, r.t. then Dowex 1 × 2 resin (OH form) to pH 2. ADDP =
1,1′-(dicarbonyl)dipiperidine. TFA = trifluoroacetic acid
Due to the importance of 5-deoxyribose (7) and 5-deoxy-5-
fluororibose (8) (both 4 steps from D-ribose, Scheme S3) and their
nucleosides in medicine and medicinal chemistry, and 5′-O-
dimethoxytrityl (DMT) ribonucleosides as building blocks in
automated solid phase phosphoramidite synthesis we sought to
expand our methodology to include nucleosides bearing these
important sugars (examples include antitumor prodrug
doxifluridine 13a[22] and organic fluoride precursor 5′-deoxy-5′-
fluoroadenosine[23]). It is important to note that 5'-O-tritylated
ribonucleosides are also useful intermediates in the synthesis of
phosphoramidites for automated synthesis of oligonucleotides on
solid support.[24] We synthesized a small series of analogs on a
semipreparative scale using all three sugars and in general we
encountered no issues. Anhydroses 10, 11, and 12 formed nearly
quantitatively and the reaction was still perfectly stereoselective
for the -anomer. It was very encouraging that both purines and
deazapurine nucleosides are still viable and we were particularly
pleased to discover that 5-fluorouracil pro-drug doxifluridine (13a)
was also accessible using our methodology, meaning this drug
can be accessed in only 5 steps from ribose and 5-fluorouracil.
Furthermore, by using N6-benzoyladenine as the acceptor
nucleobase and 5-O-DMTribose (9) (one step from D-ribose,
Scheme S11) as the donor saccharide, the desired adenosine
analog 15b is obtained as the sole product in moderate yield, thus
circumventing the regiochemical problems associated with
adenine discussed above. We were also very pleased that the
highly labile DMT group was not cleaved at all under our
conditions and could be isolated using simple chromatography.
We note, to our slight dismay, once again 7-azaadenine was
glycosylated at the undesired position 3 of the pyrimidine ring to
afford only regioisomer 14c in moderate yield.
Scheme 5. Glycosylation of nucleobases to form nucleosides using C5-
modified medicinally-relevant or solid phase phosphoramidite-amenable sugars.
a) Reaction conditions: i) 7, 8, or 9 (1.0 equiv.), P(nBu)3 (1.6 equiv.), ADDP (1.5
equiv.), MeCN (0.06 M), r.t., 15 min; ii) nucleobase NB (1.5 equiv.), NaH (1.5
equiv), DMF (~0.35 M), r.t., 12 h. b) Examples and isolated yields. ADDP = 1,1′-
(dicarbonyl)dipiperidine.
The unprecedented reactivity described in this study is clear
as to the best of knowledge no such method to access 1,2-
anhydrosugars has ever been described or characterized even
with the other hydroxyl groups protected, with the exception of a
very recent study on the hydroxide-catalyzed hydrolysis of 4-
nitrophenyl -mannopyranoside where the intermediate structure
was calculated (not experimentally observed) to be 1,2--
anhydromannose.[25] Due to the novelty of our chemistry, a
mechanistic study was necessitated. Mitsunobu reactions with
diols are known to proceed through a 5-membered 1,3-
dioxaphospholane intermediate,[26] which then converts into the
epoxide thus extruding phosphine oxide. P(nBu)3 is a highly
reactive phosphine and observation of such a phosphorane has
been reported only twice previously when ethylene glycol was
used as the diol.[27] We postulated that by using NMR experiments
operating under cryogenic conditions (40 oC) we would be able
to increase the lifetime of phosphorane 17 long enough to confirm
its existence and irrefutably determine the stereochemistry at the
anomeric centre before its conversion into anhydrose 2. We were
pleased to observe that by sequential addition of the Mitsunobu
reagents: First P(nBu)3 ( = 33.37 ppm) then ADDP, the
intermediate betaine species (16, Scheme 6) could be observed
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
( = 60.66 ppm) by 31P NMR. If compound 1 was then added
carefully and quickly to the NMR tube, the key 1,3-
dioxaphospholane 17 ( = 14.94 ppm) (See Scheme S1 in the
SI for the overlaying spectra) presented. In fact, at 40 oC
phosphorane 17 was stable indefinitely which allowed us time to
run both 1D and 2D NOE experiments. By irradiating the H-1
proton we observed correlations between both the H-2 proton and
the methylene protons of the butyl groups on the phosphorus
atom, thus confirming the formation of the phosphorane. The cis
stereochemistry of 1,3-dioxaphospholane 16 could therefore be
irrefutably established by the 2D ROESY correlations between
the H-1 and H-2 protons (Scheme 6) (see SI for NOESY spectra).
1
We also observed a distinctive doublet ( = 5.47 ppm) in the H
NMR spectrum containing 17, with a coupling constant of JH,P
17.6 Hz. This was nearly identical to the H-1P coupling constant
(JH,P 17.2 Hz) of known -D-ribofuranosyl 1,2-cyclic
=
=
Figure 6. Mechanism of rearrangement of the phosphorane to the anhydrose
or the tetrahydrofuranone from B3LYP-D3/6-311+G** calculations with
continuum model approximation of MeCN solvation. The selected interatomic
distances at the transition structures are given in Å.
monophosphate[28] thus providing even more stereochemical
evidence. We then warmed the NMR tube to room temperature
and let it stand for two hours and pleasingly, anhydrose 2 was still
formed (see Scheme S1 in the SI for the overlaying spectra).
The rearrangement of dioxaphospholane 17 to anhydrose 2
was modelled by DFT calculations (Fig. 6). For the theoretical
study, simplified substrates were used where a 5-O-methyl group
was substituted for the 5-O-trityl group and methyl groups were
modelled instead of the butyl groups at the phosphorus atom. The
calculations revealed that the elimination of trimethylphosphine
oxide and the closure of the epoxide ring is a concerted process.
The transition structure is stabilized by participation of the non-
bonding electrons of the ring-oxygen atom and hydrogen bonding
between the -OH group at the C-3 position and the oxygen atom
at the C-2 position. The energy barrier for the formation of the
anhydrose is 17.5 kcal mol-1. The alternative reaction pathway
leading to the thermodynamically favoured, but not observed, 3-
tetrahydrofuranone derivative proceeds over the energy barrier of
19.8 kcal mol-1 which explains the excellent selectivity towards the
formation of anhydrose 2.
Scheme 6. Plausible mechanism for the formation of anhydrose
phospholane 17.
2 via
Conclusions
Due to the inherent importance of nucleosides in life, disease,
and medicine, methods to facilitate their creation will always be
welcome. Here we present an expansive report on a mild, neutral,
and operationally simple monoprotection strategy for the
synthesis of ribonucleosides up to a gram scale as well as other
potentially interesting biochemically active precursor molecules.
The reaction of 5-O-tritylribose with nucleobases proceeds under
modified Mitsunobu conditions to form first an anhydrose in situ
where the epoxide is subsequently opened nucleophilically in a
perfectly stereoselective manner with either the heterocycle to
form nucleosides or other nucleophiles. Furthermore, we
demonstrated that the acid-labile trityl group can still be cleaved
essentially quantitatively in a one-pot glycosylationdeprotection
strategy without compromising any integrity of the anomeric
center as reported in our original communication.[16] This
improved one-pot synthesis of ribonucleosides is general and
With these intermediate species confirmed we were able to
posit a plausible mechanism (Scheme 6). The pKa of OH-1 of D-
ribose was recently determined using Raman spectroscopy to be
~11.8,[29] much lower than that of OH-2 and OH-3, thus allowing
for selective deprotonation of OH-1 by the relatively weakly basic
phosphonium betaine. After nucleophilic attack by the alkoxide
ion an oxyphosphosphonium ion is generated that can then be
attacked by OH-2. The reduction to the hydrazine is complete and
phospholane 16 formed, which then converts to anhydrose 2 at
higher temperatures and O=P(nBu)3 is extruded in a concerted
process.
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
(CH-2′); 70.7 (CH-3′); 63.4 (CH2-5′). HR ESIMS: m/z [M+H+] calcd for
C29H26N4O4Cl: 529.16371. Found: 529.16394.
amendable for most nucleobases. By using 5-deoxy- or 5-deoxy-
5-fluororibose, this reaction directly provides 5'-modified
nucleosides, which are highly relevant in medicinal chemistry, in
one step and good yields. Glycosylation of nucleobases with 5-O-
DMTribose gives directly advanced intermediates for building
blocks in solid phase phosphoramidite synthesis. Finally, we
showed both theoretically and experimentally that our
One-pot gram-scale synthesis of 6-chloro-9-(-D-ribofuranosyl)purine (5d)
The reaction of 5-O-trityl-D-ribose (1) (1.00 g, 2.54 mmol, 1.0 equiv.) with
P(nBu)3 (93.5 %, 1.1 mL, 4.06 mmol, 1.6 equiv.) and 1,1′-
(azodicarbonyl)dipiperidine (ADDP) (965 mg, 3.81 mmol, 1.5 equiv.)
followed by reaction with 6-chloropurine (588 mg, 3.81 mmol, 1.5 equiv.)
and NaH (60 % in mineral oil, 160 mg, 3.81 mmol, 1.5 equiv.) was
performed in the same way as above. After stirring for 12 h, the reaction
vessel was then cooled to 0 oC and to it trifluoroacetic acid (TFA)/H2O (9:1
v/v, 10 mL) was added dropwise over 5 min. The reaction was stirred again
for 12 h at r.t. (or as indicated by consumption of the protected nucleoside
by TLC) before increasing the pH to ~1 using Dowex 1 x 2 resin (OH form).
The resin was filtered off and the solvents removed in vacuo coevaporating
with toluene (5 × 20 mL). The crude brown oil was dissolved in a minimal
volume MeOH (~15 mL) and concentrated directly onto silica gel.
Purification of the product using normal phase column chromatography
(silica gel, 25:1 to 9:1 CH2Cl2:MeOH) furnished the product 5d (408 mg,
56 %) as a pale yellow foam that crystallized from water at 4 oC. Rf = 0.32
(9:1 CH2Cl2:MeOH v/v). The analytical data were in agreement with a
commercially available (SigmaAldrich) sample.
unprecedented reaction progresses through
a
cis 1,3-
dioxaphospholane intermediate that collapses into the anhydrose
in a concerted process, which as consistent with previous reports
of diols subjected to Mitsunobu conditions.[26a]
We believe that this key anhydrose intermediate can be
employed in many applications beyond the synthesis of
nucleosides. Should this methodology be amenable to the
formation of more carboncarbon bonds, the implications in
natural product synthesis and C-nucleosides are obvious and is a
future direction in our laboratory that is already underway. We are
also investigating the possibility of synthesizing oligosaccharides
through this pathway as a means to further exploit its potential.
Experimental Section
Acknowledgements
Representative examples (for full experimental section with all procedures
and full characterization data for all new compounds, see Supporting
Information):
This work was supported by the ASCR (RVO: 61388963 and
Praemium Academiae to M. H.), Technology Agency of the Czech
Republic (TE01020028 to A. M. D.), the Czech Science
Foundation (16-001178S to M.H.) and Gilead Sciences, Inc. The
authors thank Dr. H. Hřebabecký (IOCB Prague) for the kind gift
of 1-deazaadenine.
Gram-scale synthesis of 6-chloro-9-(5′-O-trityl--D-ribofuranosyl)purine
(4d)
An argon-purged, dried 250 mL round bottom flask containing a stir bar
was charged with 5-O-trityl-D-ribose (1) (1.00 g, 2.54 mmol, 1.0 equiv.)
and fitted with a rubber septum and argon-filled balloon. To this flask was
added MeCN (40 mL, 0.06 M with respect to the sugar) via syringe. To this
stirring solution was added P(nBu)3 (93.5 %, 1.1 mL, 4.06 mmol, 1.6
equiv.) via syringe followed by 1,1′-(azodicarbonyl)dipiperidine (ADDP)
(965 mg, 3.81 mmol, 1.5 equiv.) at r.t. Over the course of ~15 minutes the
reaction turned from a homogenous orange color to a heterogeneous white
color. The precipitate formed was the reduced ADDP hydrazine product
which indicated conversion to the 1,2-anhydrosugar (pictures in the SI). In
parallel, an argon-purged, dried 50 mL round bottom flask containing a stir
bar was charged with the 6-chloropurine (588 mg, 3.81 mmol, 1.5 equiv.)
and fitted with a rubber septum and argon-filled balloon. To this flask was
added DMF (10 mL, 0.33 M with respect to 6-chloropurine) via syringe. To
this stirring solution was added NaH (60 % in mineral oil, 160 mg, 3.81
mmol, 1.5 equiv.) at r.t. Stirring was continued for 15 min before the
contents of the flask were transferred via syringe to the epoxide. After
stirring at r.t. for 12 h, the reaction was quenched by decreasing the pH to
~7 using 1 M HCl(aq). The solvent was removed in vacuo coevaporating
with toluene (5 × 20 mL). The crude brown oil was dissolved in a minimal
volume of MeOH (~15 mL) and concentrated directly onto silica gel.
Purification of the product using normal phase column chromatography
(silica gel, 1:1 EtOAc:petroleum ether to 100 % EtOAc) furnished the
product 4d (829 mg, 62 %) as a waxy, white solid. Rf = 0.64 (EtOAc); []D:
4.8 (c 0.229, MeOH); 1H NMR (400.0 MHz,CD3OD): 8.65 (s, 1H, H-8);
8.62 (s, 1H, H-2); 7.437.18 (m, 15H, H-Ar Tr); 6.14 (d, 1H, J1′,2′ = 4.9, H-
1′); 4.96 (t, 1H, J2′,1′ = J2′,3′ = 4.9 Hz, H-2′); 4.53 (t, 1H J3′,2′ = J3′,4′ = 5.0 Hz,
H-3′); 4.26 (q, J4′,5a′ = J4′,5b′ = J4′,3′ = 4.3 Hz, H-4′); 3.443.42 (m, 2H, H-5a′,
H-5b′). 13C NMR (100 MHz, CD3OD):  151.6 (CH-2); 151.4 (C-4); 150.1
(C-6); 145.8 (CH-8); 143.7 (C-Ar Tr); 131.7 (C-5); 128.4 (CH-Ar Tr); 127.4
(CH-Ar Tr); 126.8 (CH-Ar Tr) 89.7 (CH-1′); 86.8 (CPh3); 84.2 (CH-4′); 73.6
Keywords: Glycosylation • Nucleosides • Epoxide • Anhydrose •
Mitsunobu reaction
[1]
For an extensive review on glycosylation strategies to form nucleosides
see: G. Romeo, U. Chiacchio, A. Corsaro, P. Merino, Chem. Rev. 2010,
110, 33373370.
[2]
[3]
J. Guo, X.-S. Ye, Molecules 2010, 15, 7235-7265.
a) L. P. Jordheim, D. Durantel, F. Zoulim, C. Dumontet, Nat. Rev. Drug
Discov. 2013, 12, 447464; b) E. De Clercq, Annu. Rev. Pharmacol.
2011, 51, 124.
[4]
[5]
R. D. Astronomo, D. R. Burton, Nat. Rev. Drug. Discov. 2010, 9, 308324.
a) B. Bennua-Skalmowski, K. Krolikiewicz, H. Vorbrüggen, Tetrahedron
Lett. 1995, 36, 78457848; b) U. Niedballa, H. Vorbrüggen, Angew.
Chem. Int. Ed. 1970, 9, 461462.
[6]
a) G. Framski, Z. Gdaniec, M. Gdaniec, J. Boryski, Tetrahedron 2006,
62, 1012310129; b) H. Vorbrüggen, Acta Biochim. Pol. 1996, 43, 2536.
E. Fischer, B. Helferich, Chem. Ber. 1914, 47, 210235.
Z. Kazimierczuk, H. B. Cottam, G. R. Revankar, R. K. Robins, J. Am.
Chem. Soc. 1984, 106, 63796382.
[7]
[8]
[9]
a) R. L. Halcomb, S. J. Danishefsky, J. Am. Chem. Soc. 1989, 111,
66616666; b) K. Chow, S. J. Danishefsky, J. Org. Chem. 1990, 55,
42114214.
[10] F. P. Boulineau, A. Wei, Org. Lett. 2002, 4, 22812283.
[11] a) E. Fischer, Chem. Ber. 1893, 26, 24002412; b) E. Fischer, Chem.
Ber. 1895, 28, 11451167; c) B. Capon, Chem. Rev. 1969, 69, 407498.
[12] a) T. J. Park, M. Weïwer, X. J. Yuan, S. N. Baytas, E. M. Munoz, S.
Murugesan, R. J. Linhardt, Carbohydr. Res. 2007, 342, 614620; b) D.
Joniak, M. Poláková, J. Serb. Chem. Soc. 2001, 66, 8186; c) V.
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
Ferrières, T. Benvegnu, M. Lefeuvre, D. Plusquellec, G. Mackenzie, M.
J. Watson, J. A. Haley, J. W. Goodby, R. Pindak, M. K. Durbin, J. Chem.
Soc., Perkin Trans. 2 1999, 951-959; d) V. Ferrières, J. N. Bertho, D.
Plusquellec, Carbohydr. Res. 1998, 311, 2535; e) R. Velty, T. Benvegnu,
M. Gelin, E. Privat, D. Plusquellec, Carbohydr. Res. 1997, 299, 714; f)
J. N. Bertho, V. Ferrières, D. Plusquellec, J. Chem. Soc., Chem.
Commun. 1995, 13911393; g) V. Ferrières, J. N. Bertho, D. Plusquellec,
Tetrahedron Lett. 1995, 36, 27492752.
[22] C. Avendaño, J. C. Menéndez, Medicinal Chemistry of Anticancer Drugs,
Vol. 2, Elsevier, 2015, pp 4153.
[23] D. O'Hagan, C. Schaffrath, S. L. Cobb, J. T. G. Hamilton, C. D. Murphy,
Nature 2002, 416, 279279.
[24] X. Wei, Tetrahedron 2013, 69, 36153637.
[25] G. Speciale, M. Farren-Dai, F. S. Shidmoossavee, S. J. Williams, A. J.
Bennet, J. Am. Chem. Soc. 2016, 138, 1410214019.
[26] a) N. García-Delgado, A. Riera, X. Verdaguer, Org. Lett. 2007, 9,
635638; b) P. L. Robinson, C. N. Barry, J. W. Kelly, S. A. Evans, J. Am.
Chem. Soc. 1985, 107, 52105219; c) P. L. Robinson, C. N. Barry, S. W.
Bass, S. E. Jarvis, S. A. Evans, J. Org. Chem. 1983, 48, 53965398.
[27] a) B. C. Chang, W. E. Conrad, D. B. Denney, D. Z. Denney, R. Edelman,
R. L. Powell, D. W. White, J. Am. Chem. Soc. 1971, 93, 40044009. b)
D. Camp, I. D. Jenkins, Aust. J. Chem. 1992, 45, 4755.
[13] G. Grynkiewicz, Pol. J. Chem. 1979, 53, 15711579.
[14] S. Schmalisch, R. Mahrwald, Org. Lett. 2013, 15, 58545857.
[15] H. Takeuchi, K. Mishiro, Y. Ueda, Y. Fujimori, T. Furuta, T. Kawabata,
Angew. Chem. Int. Ed. 2015, 54, 61776180.
[16] A. M. Downey, C. Richter, R. Pohl, R. Mahrwald, M. Hocek, Org. Lett.
2015, 17, 46044607.
[17] H. Kristinsson, K. Nebel, A. C. Osullivan, F. Struber, T. Winkler, Y.
Yamaguchi, Tetrahedron 1994, 50, 68256838.
[28] R. Fathi, F. Jordan, J. Org. Chem. 1988, 53, 199720001.
[29] S. Sen, U. Pal, N. C. Maiti, J. Phys. Chem. B 2014, 118, 909914.
[18] Y. G. Du, F. Z. Kong, Carbohydr. Res. 1995, 275, 413420.
[19] H. Rosemeyer, F. Seela, Helv. Chim. Acta 1988, 71, 15731585.
[20] P. Nauš, R. Pohl, I. Votruba, P. Džubák, M. Hajdúch, R. Ameral, G.
Birkuš, T. Wang, A. S. Ray, R. Mackman, T. Cihlar, M. Hocek, J. Med.
Chem. 2010, 53, 460470.
[21] a) E. Mairhofer, E. Fuchs, R. Micura, Beilstein J. Org. Chem. 2016, 12,
25562562. b) M. D. Erlacher, K. Lang, B. Wotzel, R. Rieder, R. Micura,
N. Polacek, J. Am. Chem. Soc. 2006, 128, 44534459.
This article is protected by copyright. All rights reserved.

10.1002/chem.201604955
Chemistry - A European Journal
FULL PAPER
Entry for the Table of Contents (Please choose
one layout)
Layout 1:
FULL PAPER
Text for Table of Contents
Author(s), Corresponding Author(s)*
Page No. – Page No.
Title
Layout 2:
FULL PAPER
Author(s), Corresponding Author(s)*
Page No. – Page No.
Title
Stereoselective one-pot synthesis of nucleosides by
glycosylation of nucleobases with 5'-O-trityl-ribose or 5'-
modified 5'-deoxyriboses.
This article is protected by copyright. All rights reserved.