Convergent stereospecific total synthesis of monocillin I and radicicol: Some simplifications and improvements

Article abstract of DOI:10.1016/S0040-4039(02)00713-X

A much improved and reliable access to the macrolide 9, key-intermediate in our synthesis of monocillin I and radicicol is reported, via a modification of our first synthesis. The formation of the conjugated E,Z-dienone trans-epoxide is now achieved in a much higher yield, in a stereospecific reaction, by elimination of the methanesulfonate ester of the 6′-OH of the intermediate macrolide. It is also shown that the configuration of the 6′-OH has no significant incidence on all the steps leading to 9, and that consequently the two diastereoisomers 12, epimeric at 6′, can be used for the synthesis of radicicol.

Full text of DOI:10.1016/S0040-4039(02)00713-X

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 43 (2002) 3997–4001
Convergent stereospecific total synthesis of monocillin I and
radicicol: some simplifications and improvements
Isabelle Tichkowsky and Robert Lett*
Unite´ Mixte CNRS-AVENTIS Pharma (UMR 26), 102, route de Noisy, 93235 Romainville, France
Received 5 April 2002; accepted 11 April 2002
Abstract—A much improved and reliable access to the macrolide 9, key-intermediate in our synthesis of monocillin I and radicicol
is reported, via a modification of our first synthesis. The formation of the conjugated E,Z-dienone trans-epoxide is now achieved
in a much higher yield, in a stereospecific reaction, by elimination of the methanesulfonate ester of the 6%-OH of the intermediate
macrolide. It is also shown that the configuration of the 6%-OH has no significant incidence on all the steps leading to 9, and that
consequently the two diastereoisomers 12, epimeric at 6%, can be used for the synthesis of radicicol. © 2002 Elsevier Science Ltd.
All rights reserved.
The recent publication1 of a new asymmetric synthesis
of monocillin I 1 and radicicol 2 by Danishefsky and
co-workers prompts us now to disclose some improve-
ments and simplifications of our first synthesis of these
molecules.2
Our original synthetic approach was achieved accord-
ing to Scheme 1. A difficult step in our synthesis was
the generation of the conjugated dienone epoxide from
the macrolide 3. Our previous results showed it was
necessary to protect the 6%-OH, in order to be able to
Scheme 1.
Keywords: macrolides; isocoumarins; tin and compounds; coupling reactions; dienones; Mitsunobu reactions.
* Corresponding author. Tel.: 33-1 49 91 52 80; fax: 33-1 49 91 38 00; e-mail: robert.lett@aventis.com
0040-4039/02/$ - see front matter © 2002 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(02)00713-X

3998
I. Tichkowsky, R. Lett / Tetrahedron Letters 43 (2002) 3997–4001
with the
E tributylstannylvinyllithio derivative 8
achieve the macrolactonization, in Mitsunobu reaction
conditions, of the intermediate keto hydroxy acid 4o
which moreover exists in solution in the quasi-unique
hemiketal closed form 4c as shown by NMR (200 or
400 MHz, CDCl3, rt) and IR (CHCl3).2d,3 Therefore,
our first synthesis was completed with the 6%-OMOM
protected derivatives, since the protection of the 6%-OH
was very efficient with a preformed MOMCl solution
and the MOMO− anion could also be envisaged as a
reasonable leaving group, this in our initial approach
for the generation of the conjugated dienone epoxide to
obtain 9 from the macrolide 3, in necessarily sufficiently
mild basic and nonnucleophilic conditions in order to
be able to preserve the structures.
(Scheme 3).3
On the other hand, that condensation afforded a 3-4/1
mixture of the two diastereoisomers 12M (major adduct)
and 12m (minor adduct), epimeric at 6% and in a first
approach the completion of our synthesis was initially
achieved only with 6 as the major diastereoisomer
(configuration at C-6% not determined), obtained pure
after chromatography over silica gel.2 Therefore, it was
now of some interest to examine the further steps of the
sequence with the minor diastereoisomer in order to see
if it could be also used successfully for the synthesis of
monocillin I and radicicol or, if not, it was then neces-
sary to improve the stereoselectivity of the condensa-
tion between 7 and 8 or to invert the configuration of
the minor adduct.
However, if the desired macrolide 9 could be obtained
in a stereospecific reaction from 3, it was only in a
25–30% yield, by using anhydrous K2CO3 (1.3 equiv.)
(DME, reflux, 1 h); quite surprisingly, two other
macrolides 10 and 11 were also isolated in 15–25% yield
(in a ca. 1/1 ratio), resulting from the quench by each
regioisomeric enolate of the anhydrous formaldehyde
generated in situ by the elimination of the OMOM
ether, via the equilibrium between CH3OCH2O− and
HCHO plus CH3O− (Scheme 2).2 To our knowledge,
such a reaction has only been previously reported in a
fragmentation–recombination reaction of methoxy-
methylester enolates.4 Elimination of MOMO− most
likely is not concerted to give free formaldehyde and
CH3O−, the equilibrium being highly shifted towards
CH3OCH2O− in the reaction conditions.5 Consistently,
in order to avoid the hydroxymethylation reaction,
attempted elimination of the 6%-OMe derivative in the
same anhydrous conditions (K2CO3, DME, reflux) only
led to degradation products, due to the greater basicity
of CH3O−, thus showing that the MOM ether had a
very subtle role in the elimination reaction.2d,3
Starting the sequence from the minor adduct 12m, it
was of particular relevance to examine the macrolac-
tonization step in Mitsunobu reaction conditions, and
also the formation of the conjugated dienone epoxide
to get the macrolide 9. Those two steps might indeed be
dependent on the conformation of the acyclic precursor
(16M, 16m) or on the conformational constraints of the
formed macrolide (17M, 17m) concerning the macrolac-
tonization step, or on the conformation of the 6%-OMs
macrolide derivative (from 19M or 19m) for the elimina-
tion affording 9, conformations which may be different
for the two epimers at 6% due to the established influ-
ence of the 6%-OR substituent configuration for deter-
mining the privileged conformation of the macrocycle
of resorcylic macrolides.6 In this same communication,
we wish also now to report that the two diastereoiso-
meric adducts 12M and 12m, epimeric at C-6%, can be
employed in our synthesis of monocillin I and radicicol,
the configuration at C-6% having no real significant
incidence on each individual step of the sequence, and
quite unexpectedly on particularly the two aforemen-
tioned key-steps (Scheme 3).3 Consequently, these two
diastereoisomers do not need to be separated and the
whole sequence (12 to 9) can now be achieved on the
mixture of epimers at C-6%.
Further improvements were clearly necessary for the
generation of the conjugated dienone epoxide in order
to obtain 9 in a higher yield, via a reliable reproducible
reaction. The problem was here that a good leaving
group could not be introduced early in the sequence
and, therefore, we had to find a suitable protective
group of the 6%-OH, allowing its efficient deprotection
after the macrolactonization in order to obtain the
macrolide 6%-OH 19 in high yield. Consequently, a
sequence was developed, based on the protection of the
6%-OH as an OMPM ether immediately after the con-
densation of the enantiomerically pure epoxyaldehyde 7
Starting from the adduct 12M or 12m, the protection of
the 6%-OH by MPMCl (1.5 equiv.) required the forma-
tion of the lithium alkoxide and HMPA as a cosolvent
(THF/hexane/HMPA 1/1/1, rt, 48 h) to yield 13M or
13m in 82% yield, and some starting material (15–17%)
was still recovered after chromatography in each case.
Deprotection of the silyl ether (TBAF/THF, rt)
Scheme 2.

I. Tichkowsky, R. Lett / Tetrahedron Letters 43 (2002) 3997–4001
3999
Scheme 3. Reagents and conditions: (a) 8 from E-1,2-bistributylstannylethylene (1.0 equiv.) and nBuLi (1.0 equiv.) in THF–hexane
(4/1), rt, 10 min, then 7 (1.0 equiv.) in THF at −78°C, 40 min, NH4Cl quench; (b) BuLi (1.0 equiv.)/THF–hexane (1/1), −78°C,
then MPMCl (1.5 equiv.), HMPA, 48 h, rt; (c) TBAF 1 M in THF (3 equiv.), rt, overnight; (d) 5 (0.97 equiv.), 3 mol%
PdCl2(CH3CN)2, 5 mol% PPh3, DME, reflux, 2 h 30 min; (e) DIBAH (10 equiv.), THF–hexane (2/1), −78°C, 45 min, then acetone
quench and pH 4 buffer; (f) crude product in tBuOH/H2O/2-methyl 2-butene (1/1/1) NaClO2·H2O (7 equiv.), NaH2PO4·2H2O (18
equiv.), pyridine (6 equiv.), 5 days, rt, pH 3 work-up; (g) 16 (0.005 M) in toluene, PPh3 (2 equiv.), then DEAD (1.4 equiv.), rt,
1 h; (h) TBSCl/iPr2NEt (1.6 equiv.), DMF, rt, 3 h, then pH 3 buffer work-up; (i) DDQ (2.5 equiv.), CH2Cl2–H2O (9/1), rt, 1 h;
(j) CH3SO2Cl (1.5 equiv.), NEt3 (3.25 equiv.), CH2Cl2, 0°C; (k) borax buffer/MeOH (1/1), rt, 24 h, then pH 3 buffer work-up;
(l) 9 (0.004 M) in CH2Cl2/pH 3 buffer (2/1), 0°C, then addition of Ca(OCl)2 0.038 M in H2O until no more starting material
(followed by TLC); (m) crude 20 in THF/MeOH/borax buffer (1/1/1), 40–50°C, 8 h, then pH 3 buffer work-up.
For related compounds, we previously observed that
the isocoumarin cleavage could not be achieved in
aqueous basic conditions, even at room temperature,
since they led immediately only to degradation and
complex mixtures and that NaOH (1.4 equiv.) in excess
afforded 14M or 14m in high yield. Subsequent coupling
with the chloromethyl-isocoumarin 5 (0.97 equiv.), cat-
alyzed by 3 mol% PdCl2(CH3CN)2 with 5 mol% PPh3
(DME, reflux) afforded 15M or 15m in ca. 75% yield
after chromatography.

4000
I. Tichkowsky, R. Lett / Tetrahedron Letters 43 (2002) 3997–4001
30% H2O2 did also not afford the desired keto-acid.2
Therefore, the isocoumarin cleavage was achieved as
previously in two steps to obtain the keto-acid in the
quasi-unique hemiketal form 16M (or 16m) in solution
(as shown by NMR) in 46–49% overall yield for the
three steps (d, e, f), and 12–14% of the bis-OTBS
isocoumarin 15M (or 15m) were also reisolated after
chromatography. It is worth to point out that, for the
DIBAH reduction, quench with excess acetone at
−78°C followed by pH 4 work-up for extraction is
crucial in order to get cleanly the lactol; at pH 3, other
products were formed and at pH above 7 a naphthol is
exclusively formed via an intramolecular aldol of the
keto-aldehyde open form. The equilibrium between the
lactol and the keto-aldehyde can be induced by the
addition of pyridine (6 equiv.), with no naphthol for-
mation, in order to allow the oxidation of the alde-
hyde into a carboxylic acid by NaClO2 in buffered
conditions.2 It is also worth emphasizing that the
reduction of 15M (or 15m) was complete and that no
appreciable desilylation occurred during the reduction,
as shown by the NMR of the crude product.3 On the
other hand, the specific deprotection of the phenol at
the ortho position which is observed in obtaining 16M
(or 16m) is most likely the result of an intramolecular
silyl group migration involving the intermediate car-
boxylate anion which is formed in the oxidation step,
and subsequent hydrolysis of the silyl ester. That spe-
cific deprotection is quite fortunate, since we showed
earlier that the free ortho-phenol had a determining
effect for favoring the macrolactonization with respect
to the competitive isocoumarin formation.2 Under
tection of the 6%-OMPM ether7 was efficient with DDQ
in CH2Cl2–H2O (9/1) (rt, 1 h) to afford 19M and 19m
(83 and 91% yield, respectively). The 6%-OMs derivative
was then formed (CH3SO2Cl/NEt3, CH2Cl2, 0°C to rt)
and its elimination occurred in situ, as soon as formed,
in the presence of an excess of amine in very mild
conditions, in a clean stereospecific reaction affording
only the desired macrolide 9 in high yield (91% from
19M, 83% from 19m). The new sequence3 described
herein gives thus a much improved and reliable access
to the key-macrolide 9, and consequently to monocillin
I and radicicol, as already described in our initial
approach.2 Our synthesis is convergent, stereospecific,
yielding enantiomerically pure compounds, and is quite
flexible for producing related unnatural macrolides. In
order to further illustrate that flexibility, the
macrolides 23 and 24 were obtained from the major
adduct 21M, via the key-macrolide 22. It is worth
pointing out that the diastereoisomeric trans epoxide
here results in very different steric and conformational
constraints and that, quite remarkably, the sequence
could be achieved in quite comparable yields (Scheme
4).3 In the accompanying communication, we also
describe other improvements of our synthesis by a
modification via palladium-catalyzed vinyldisiamyl-
boranes couplings with the chloromethylisocoumarin 5.8
Acknowledgements
We thank the staff of the Analytical Department of
the Research Center at Romainville, Roussel Uclaf
and the CNRS for a Ph.D. grant to I. Tichkowsky,
and the Direction des Recherches Chimiques of Rous-
sel-Uclaf for support of this work.
Mitsunobu reaction conditions (16M or 16m 5×10−3
M
in anhydr. toluene, rt), the macrolides 17M and 17m
were isolated in 66 and 72% yield, respectively, after
chromatography. These macrolactonizations appear to
be quite unique, since being achieved from a precursor
which is not a free hydroxy acid, but which is under
the quasi-exclusive cyclic hemiketal form in solution,
they are also unique with respect to the severe compe-
tition between the 14-membered macrolide and the
isocoumarin formations. At this stage, however, the
o-phenol had to be reprotected as an OTBS ether since
our previous work showed it was necessary not to
have a free o-phenol, which led only to reformation of
the isocoumarin during the attempted elimination of
the OMOM ether followed by subsequent degrada-
tion.2 That protection had to be carried out in condi-
tions avoiding enolization of the ketone at 2%, which
could be achieved by TBSCl/iPr2NEt in DMF at rt, to
produce 18M and 18m in good yield. Subsequent depro-
References
1. (a) Garbaccio, R. M.; Danishefsky, S. J. Org. Lett. 2000,
2, 3127–3129; (b) Garbaccio, R. M.; Stachel, S. J.;
Baeschlin, D. K.; Danishefsky, S. J. J. Am. Chem. Soc.
2001, 123, 10903–10908.
2. (a) Lampilas, M. Ph.D. Thesis, Paris VI University, 1991;
(b) Lampilas, M.; Lett, R. Tetrahedron Lett. 1992, 33,
773–776; (c) Lampilas, M.; Lett, R. Tetrahedron Lett.
1992, 33, 777–780; (d) Lett, R.; Lampilas, M.;
Tichkowsky, I. In Recent Progress in the Chemical Synthe-
sis of Antibiotics and Related Microbial Products; Lukacs,
G., Ed.; Springer Verlag, 1993; Vol. 2, pp. 99–120.
Scheme 4. Reagents and conditions: as for Scheme 3. (a) 55% (74% corr.); (b) 80% (90% corr.); (c) 94%; (d, e, f) 54% overall (three
steps); (g) 62%; (h) 85%; (i) 91%; (j) 90%; (k) 91%; (l) and (m) (48 h, rt) 47% overall (two steps).

I. Tichkowsky, R. Lett / Tetrahedron Letters 43 (2002) 3997–4001
4001
3. Tichkowsky, I. Ph.D. Thesis, Paris VI University,
6. Gelo-Pujic, M.; Antolic, S.; Kojic-Prodic, B.; Sunjic, V.
Tetrahedron 1994, 50, 13753–13764.
1996.
7. Horita, K.; Yoshioka, T.; Tanaka, T.; Oikawa, Y.;
Yonemitsu, O. Tetrahedron 1986, 42, 3021–3028.
8. Tichkowsky, I.; Lett, R. Tetrahedron Lett. 2002, 43,
4003–4007.
4. Schultz, A. G.; Berger, M. H. J. Org. Chem. 1976, 41,
585–586.
5. Young, S. D.; Coblens, K. E.; Ganem, B. Tetrahedron
Lett. 1981, 22, 4887–4888.