Catalytic conversions of diazosugars

Article abstract of DOI:10.1016/S0040-4039(02)02407-3

Catalytic diazo decomposition chemistry was explored with novel diazoacetate functionalized carbohydrates. In both glucose and galactose derivatives, aromatic cycloadditions with the benzyl protecting groups proved to be most favorable reactions. Replacing the benzyl groups by methoxy groups in the glucose system led to formation of 'carbene dimers'.

Full text of DOI:10.1016/S0040-4039(02)02407-3

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 43 (2002) 9601–9603
Catalytic conversions of diazosugars
Hilbert M. Branderhorst, Johan Kemmink, Rob M. J. Liskamp and Roland J. Pieters*
Department of Medicinal Chemistry, Utrecht Institute for Pharmaceutical Sciences, Utrecht University PO Box 80082,
3508 TB Utrecht, The Netherlands
Received 20 September 2002; revised 11 October 2002; accepted 24 October 2002
Abstract—Catalytic diazo decomposition chemistry was explored with novel diazoacetate functionalized carbohydrates. In both
glucose and galactose derivatives, aromatic cycloadditions with the benzyl protecting groups proved to be most favorable
reactions. Replacing the benzyl groups by methoxy groups in the glucose system led to formation of ‘carbene dimers’. © 2002
Elsevier Science Ltd. All rights reserved.
Chemistry involving catalytic metal carbene transfor-
mations has proven to be a powerful and versatile
area.1 Many highly selective reactions, using diazocom-
pounds as starting materials and preferentially
dirhodium complexes as catalysts, have been developed.
These include cyclopropanations, ylide generation and
their following reactions and insertion reactions.
Despite the rich toolbox that has been generated very
few applications of this chemistry have been directed
towards carbohydrates.2 Considering that metal car-
bene intermediates can add to or insert into various
bonds present in carbohydrate derivatives, the applica-
tion of carbenoid chemistry may give rise to novel
compounds, not accessible via alternative methods. The
synthesis of novel carbohydrate derivatives is an impor-
tant goal. Due to the expanding knowledge of the roles
that carbohydrates play in important biological pro-
cesses, unnatural carbohydrate derivatives may have
potential biomedical applications due to interference
with these processes.3
hydroxyl group was liberated by treatment with acid to
yield 2. Reaction of 2 with diketene resulted in the
corresponding acetoacetate, required for the subsequent
diazo transfer with MsN3. Deacylation by LiOH led to
the desired diazo compound 3.5
In order to explore its catalytic decomposition chem-
istry, diazo compound 3 was dissolved in CH2Cl2 and
added slowly to a solution containing a catalytic
amount of Rh2(OAc)4. Three new products were
formed, all with the same molecular weight correspond-
ing to the molecular weight of 3 minus dinitrogen (m/z
613 M+Na+) according to the electrospray mass spec-
tra. The major product was identified as cyclohepta-
triene 4 and could be isolated in 45% yield. A detailed
analysis of its COSY spectra confirmed the 1,4 substitu-
tion pattern of the cycloheptatriene ring versus the
1
alternative 1,2 and 1,3 patterns.6 The H NMR spec-
trum suggested that the reaction had taken place with
the C(4) benzyloxy group since for 4 the chemical shifts
In order to explore some of the issues mentioned above
we decided to incorporate a diazoacetate moiety into a
carbohydrate. We chose glucose as the carbohydrate
and the C(6) oxygen for attachment of the diazoacetate
group. Glucose derivative 1, containing the 2-
(trimethylsilyl)ethyl (OSE) group at the anomeric oxy-
gen,4 was the starting material and was treated with
trityl chloride in order to protect selectively the C(6)
hydroxyl group (Scheme 1). Following this, the remain-
ing free hydroxyls were benzyl protected and the C(6)
Scheme 1. Reagents and conditions: (a) Ph3CCl, DMAP, pyr,
80°C, 3 h; (b) BnBr, NaH, DMF, 14 h; (c) AcOH/EtOH 1:1,
80°C, 12 h, 47% (3 steps); (d) diketene, NEt3, THF, 6 h; (e)
MsN3, NEt3, THF, 20 h; (f) LiOH, THF/H2O, 14 h, 43% (3
steps); (g) Rh2(OAc)4, CH2Cl2, 22 h, 45%.
* Corresponding author. Tel.: +31-30-2536944; fax: +31-30-2536655;
e-mail: r.j.pieters@pharm.uu.nl
0040-4039/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(02)02407-3

9602
H. M. Branderhorst et al. / Tetrahedron Letters 43 (2002) 9601–9603
of HꢀC(2) and HꢀC(3) were close to the same value
(within 0.1 ppm) as in 3, however the HꢀC(4) signal
had moved from 3.52 to 3.80 ppm. A NOESY spectrum
confirmed this since the observed NOEs clearly allowed
assignment of the benzyloxy groups at C(2) and C(3),
and also showed signals between the cycloheptatriene
ring and the C(4)-linked OCH2 moiety. The reaction
leading to product 4, was an intramolecular aromatic
cycloaddition that proved to be more favorable than a
CꢀH insertion reaction into one of the available CꢀH
bonds of the pyranose ring. For the aromatic cycload-
dition alternative addition products involving the other
benzyloxy groups, or other positions on the C(4) benzy-
loxy group seemed conceivable as well. Indeed 1H
NMR spectra of the two minor products both showed
the characteristic signals of the cycloheptatriene unit. In
order to attempt to influence the reaction outcome,
reactions were also run in the presence of dirhodium(II)
perfluorobutyrate (Rh2(pfb)4) and dirhodium(II) capro-
lactamate (Rh2(cap)4), which should lead to more and
less reactive intermediate metal carbenes, respectively,
in comparison to Rh2(OAc)4. Use of these alternative
catalysts has been shown in several cases to give rise to
vastly different selectivities.1 In the reaction with 3,
however, all three catalysts gave the same products in
close to the same ratios.
Scheme 2. Reported favorable C–H insertion reaction.
Scheme 3. Reagents and conditions: (a) n-Bu2SnO, p-OMe-
BnCl, PhH, 80°C, 2 h, (76%); (b) BnCl, NaH, DMF, 14 h,
(67%); (c) CAN, CH3CN/H2O, 4.5 h, (82%); (d) diketene,
NEt3, THF, 14 h; (e) MsN3, NEt3, THF, 14 h; (f) LiOH,
THF/H2O, 5 h, 72% (3 steps); (g) Rh2OAc4, PhH, 16 h.
disappearance of the starting material and the appear-
ance of three compounds. No true major product was
observed. The mixture was (partially) purified by
column chromatography and the resulting fractions
were analyzed by electrospray mass spectrometry and
NMR. All samples showed only a single MS signal at
m/z 613 (M+Na+), as was expected for all insertion or
addition products. 1H NMR spectra of the samples
showed the characteristic cycloheptatriene signals
between 5.7 and 6.8 ppm, but the specific substitution
pattern could not be identified. This shows that again
the aromatic cycloaddition to the benzyl protecting
groups was the most favorable reaction and no signifi-
cant amounts of products derived from CꢀH insertion
could be detected. Reactions with the alternative cata-
lysts Rh2(pfb)4, and Rh2(cap)4 gave similar reaction
profiles. The ease of cycloheptatriene formation could
also be illustrated by running the Rh2(OAc)4 catalyzed
reaction in benzene, which gave cycloheptatriene 10 as
the only product.
Although the aromatic cycloaddition is known to be a
good reaction, especially with Rh2(pfb)4 as a catalyst,7
CꢀH insertion reactions into HꢀC(4), HꢀC(5) or
HꢀC(6) of 3 and also the Stevens Rearrangement8
seemed possible as well. In an attempt to create a
geometry as favorable as possible for a CꢀH insertion
reaction, a reported successful CꢀH insertion reaction
was used in the design of the next diazosugar. The
reaction involved was the high yield cyclization of 5, a
diazocompound based on glycerol. The cyclization of
this compound gave in good yield the 5-membered
lactone 6 with catalysis by dirhodium(II) carboxamide
catalysts (Scheme 2).9
In order to keep important aspects such as the expected
ring size and electronic activation (by donating sub-
stituents such as OBn) the same, galactose derivative 9
with its diazoacetate moiety linked to the C(3) oxygen
was selected for synthesis (Scheme 3). Despite the simi-
larities between 5 and 9 obvious differences are the
conformational restrictions in 9 and the fact that the
targeted insertion sites have a higher degree of alkyl
substitution. The latter aspect actually makes the sites
more reactive for insertion.1 In the synthesis of 9,
galactose 74 was regioselectively protected with a p-
methoxybenzyl group at the C(3) oxygen via a dibutyl-
stannylene acetal,10 protection of the other hydroxyls
with benzyl groups was followed by selective cleavage
of the p-methoxybenzyl group using cerium(IV) ammo-
nium nitrate10 to yield 8. As before, the liberated
hydroxyl was elaborated into the diazoacetate unit of 9
using the three-step protocol with diketene, MsN3 and
LiOH, respectively. Catalytic diazo decomposition of 9
in the presence of Rh2(OAc)4 in CH2Cl2 resulted in the
The previous results showed that CꢀH insertion into
the pyranose CꢀH bonds was not a favorable reaction
in comparison to the aromatic cycloaddition with the
slightly more remote benzyl protecting groups. In order
to take these benzyl groups out of the equation we
turned to methoxy protecting groups, since these
groups should be inert to the generated metal carbene,
thus giving the targeted insertion reaction a higher
probability. Diazocompound 12 was prepared for this
purpose via a route similar to that used for 3 (Scheme
4). Exposure of this diazocompound to a catalytic
amount of Rh2(OAc)4 in CH2Cl2 resulted in two dis-
tinct major products on TLC. Both NMR and MS
analysis of the two compounds indicated them to be the
two ‘carbene dimers’ 13 and 14 obtained in a close to
1:1 ratio. Again no products from CꢀH insertion reac-
tions could be identified.

H. M. Branderhorst et al. / Tetrahedron Letters 43 (2002) 9601–9603
9603
interesting entry into a class of tricyclic constrained
carbohydrates with increased lipophilicity and will be
further explored. The carbene dimer formation, as
shown for 12, may be an interesting reaction as well, if
used intramolecularly,11 in the construction of con-
strained carbohydrates by making bridges between
sugar residues in an oligosaccharide.
Acknowledgements
Scheme 4. Reagents and conditions: (a) diketene, NEt3, THF,
62 h; (b) MsN3, NEt3, THF, 4 h; (c) LiOH, THF/H2O, 4 h,
60% (3 steps); (d) Rh2OAc4, CH2Cl2, 5 h, 63%.
The Royal Netherlands Academy of Arts and Sciences
(KNAW) is gratefully acknowledged for support.
References
For glucose and galactose derivatives 3 and 9, CꢀH
insertion reactions apparently were disfavored relative
to aromatic cycloadditions. This happened despite the
fact that several factors seemed to favor a CꢀH inser-
tion, e.g. into HꢀC(4) of 9. A favorable factor for
CH-insertion is the fact that the ring formed would be
a 5-membered one. Also favorable is the fact that the
insertion would take place into a tertiary CꢀH bond
and that the carbon C(4) contains an activating oxygen
substituent. The fact that no insertion was observed
indicates that the geometrical constraints of the ring
system disfavor the required geometries towards the
transition state in favor of the aromatic cycloaddition.
In the unrestrained 5, the CꢀH insertion is more favor-
able than the aromatic cycloaddition. The balance may
be a delicate one, considering also that CꢀH insertion
did occur in a constrained furanose sugar leading to a
5-membered ring.2b For these reasons, alternative
geometries should be explored for insertion, which need
not be limited to CꢀH bonds but could also include the
OꢀH or NꢀH bonds. The reaction with the glucose
derivative 12, giving the ‘carbene dimers’, indicates that
this compound had no favorable intramolecular path-
ways available for reaction. This also explains why
compound 3, with the same orientation of the diazoac-
etate group as 12, ‘found’ only the intramolecular
cycloaddition involving the benzyl protecting groups.
The aromatic cycloaddition product 4 provides an
1. Doyle, M. P.; Forbes, D. C. Chem. Rev. 1998, 98, 911.
2. (a) For a synthesis via an oxonium ylide see: Sarabia, F.;
Lo´pez Herrera, F. J. Tetrahedron Lett. 2001, 42, 8805; (b)
For a C–H insertion in a furanose sugar see: Berndt, D.
F.; Norris, P. Tetrahedron Lett. 2002, 43, 3961.
3. Glycobiology reviews: (a) Williams, S. J.; Davies, G. J.
Trends Biotechnol. 2001, 19, 356; (b) Mahon, B. P.;
Moore, A; Johnson, P. A.; Mills, K. H. G. Crit. Rev.
Biotechnol. 1998. 18, 257.
4. Jansson, K.; Ahlfors, S.; Frejd, T.; Kihlberg, J.; Magnus-
son, G.; Dahme´n, J.; Noori, G.; Stenvall, K. J. Org.
Chem. 1988, 53, 5629.
5. The one-step Corey–Myers protocol was also employed
with similar results, Corey, E. J.; Myers, A. G. Tetra-
hedron Lett. 1984, 25, 3559.
6. The NMR spectra showed no sign of the presence of two
diastereomers. The configuration of the newly formed
stereocenter is unknown.
7. Doyle, M. P.; Protopopova, M. N.; Peterson, C. S.;
Vitale, J. P. J. Am. Chem. Soc. 1996, 118, 7865.
8. Doyle, M. P.; Ene, D. G.; Forbes, D. G.; Tedrow, J. S.
Tetrahedron Lett. 1997, 38, 4367.
9. Doyle, M. P.; Dyatkin, A. B.; Tedrow, J. S. Tetrahedron
Lett. 1994, 35, 3853.
10. Kiyoi, T.; Nakai, Y.; Kondo, H.; Ishida, H.; Kiso, M.;
Hasgawa, A. Bioorg. Med. Chem. 1996, 4, 1167.
11. Doyle, M. P.; Hu, W.; Phillips, I. M.; Wee, A. G. H. Org.
Lett. 2000, 2, 1777.