Scheme 2 Reagents and conditions: a) PPh3, Br2 (95%); b) (Thienyl)CuCNLi, add NaphtLi, then 8, then 4 (55%); c) H2, Pd(OH)2 (quant.); d) CSA (91%);
e) H2, Pd(OH)2/C (quant.); f) 12, NIS, TfOH, CH2Cl2 (77%); g) CSA (82%); h) H2NNH2, 80 uC (92%); i) H2, Pd black THF–AcOH–H2O (quant.).
epoxide 4 could be prepared in multigram quantities in 43%
overall yield from D-ribose.
intermediate thienyl-based activated copper was generated: reac-
tions were systematically cleaner if NaphtLi was added into a
solution of thienylcyanocuprate instead of the published reverse
addition. Moreover, the yields were always better when the
reaction was carried out with bromide 8 compared with the
corresponding iodide. In summary, the key parameter allowing
the successful coupling of the C30 hopane skeleton to the C5 ribitol
side chain was indeed the quickest formation of an organocopper
derivative directly from a primary bromide, thus confirming our
intuitive assumption that the metallated C-30 position is unstable.
Standard deprotections yielded bacteriohopanetetrol 3 in 3 steps
and 50% overall yield from 7. In all respects, analytical data of
synthetic tetrol 3 and its peracetate corresponded to the natural
ones.9 This approach is the most efficient preparation of natural
tetrol 3 and allowed us to complete the synthesis of the even more
challenging glucosamine derivative 2. Indeed, the quantitative
removal of the benzyl group generated a diol that could be directly
involved in a glycosylation (Scheme 2). As expected, the coupling
of thioglycoside 12{ proceeded regio- and stereoselectively to give
11 in 77% yield. Given the amphiphilic character of the target
molecule, the deprotection strategy was carefully chosen, the
hydrogenolysis step being performed as the very last step.
Therefore the acetonide and the phthalimido groups were removed
first, using classical conditions. After screening both the proper
catalyst and the proper solvent system, we finally found
hydrogenolysis conditions yielding the final pure glycoside 2 in a
quantitative yield. For such a reaction, a ternary solvent system
composed of THF, AcOH and water was absolutely required to
observe a total and clean hydrogenolysis. The structure of
glycolipid 2 was confirmed by its peracetylation and comparison
with the analytical data of the natural molecule’s peracetate.{4,6
In conclusion, this work describes a new biomimetic strategy to
efficiently construct bacteriohopanetetrol and its glucosamine
derivative. This chemical synthesis gives access not only to the
first full characterisation of natural glycoside 2 but also to
interesting tools to study the biosynthesis of ether 1.
The C30 primary alcohol 7 was produced from a natural
hopanone following the procedure developed by Rohmer and
Duvold.9 Bromide 8 was then prepared in sufficient amount to
allow an extensive methodological study (Scheme 2). The first
attempts at metal–halogen exchange followed by nucleophilic
attack on epoxide 4 or on various model epoxides or aldehydes
were very disappointing, the expected product 9 being isolated
only in trace amounts. The usual techniques optimized to generate
organocopper derivatives from 8 were screened12 but the main
product was always that resulting from protonation (compound
10), showing that the activation of the halogenide occurred but not
the nucleophilic attack. For instance, we attempted, as a first
approach, to generate organocuprates through
a transient
organolithium derivative. We screened different lithium–halogen
exchange reactions using metallic lithium, lithium naphthalenide
and LiDBB (lithium di-tert-butylbiphenylide) followed by addition
of CuCN or CuI. These experiments only lead to the protonation
product 10. This poor reactivity might be attributable to the severe
steric hindrance of position C-30 due to the convex shape of this
pentacycle. However we reasoned that the metallated hopane
derivative could also be unstable. Therefore we sought conditions
allowing the fastest formation of the organocopper derivative in
order to expose it to epoxide 4 as quickly as possible. With this
purpose in mind, we took advantage of the procedure established
by Rieke et al., who used 2-thienyl lithium organocuprate and
lithium naphthalenide to generate the so-called zero-valent
organocopper derivatives, directly from primary halogenides.13
Application of this procedure allowed us to isolate the desired
hopanetetrol derivative 9 in moderate yield (,25%). After
optimization, we now have in hand a procedure giving a
reproducible and satisfactory 55% yield. In particular, two
simultaneous modifications have been found to play a critical role
in determining the efficiency of the coupling procedure: the relative
proportions of all reactants and the order of addition of the
required lithium naphthalenide (NaphtLi). Typically, the initial
procedure developed with simple bromides uses large excesses of
NaphtLi (32 equivalents compared to bromide), lithium 2-thienyl-
cyanocuprates (30 eq.) and epoxide (30 eq.).13 We first limited the
excess of epoxide 4 to 2.8 equivalents and quickly realized that
reducing the excess of NaphtLi and thienylcyanocuprates to 4.4
equivalents significantly decreased the amount of the protonation
side-product 10 without affecting the coupling kinetics.{
Surprisingly, the yields were also improved by the way the
This project was funded by the ‘‘Association Franco-Chinoise
pour la Recherche Scientifique et Technique’’ (AFCRST, program
PRA B02-07), the French embassy in China (PhD grant to W. P.)
and the CNRS (Centre National de la Recherche Scientifique).
Weidong Pan,acd Yongmin Zhang,a Guangyi Liang,cd
Ste´phane P. Vincent*ab and Pierre Sinay¨*a
aEcole Normale Supe´rieure, De´partement de Chimie, UMR 8642 du
CNRS, 24 rue Lhomond, 75231 Paris Cedex 05, France.
E-mail: sinay@ens.fr; Fax: +33 (0)1 44 32 33 97
bUniversity of Namur, Laboratoire de Chimie Bio-Organique, rue de
3446 | Chem. Commun., 2005, 3445–3447
This journal is ß The Royal Society of Chemistry 2005
Pan, Weidong
