minor amounts (1-3%) of each of the other isomers and
residual rapamycin. The major product was crystallized from
methanol/water, and the structure was determined by X-ray
crystallography to be 28-epirapamycin (2). The identities of
the three minor isomers were shown by extensive NMR
analysis to be 29-epirapamycin (3), 28,29-bisepirapamycin
(4), and the known 28-secorapamycin (5).6
Secorapamycin (5) appears within seconds following the
addition of Ti(OiPr)4 and remains at a constant level (∼3-
5%) throughout the course of the reaction. Within the first
2 min of the reaction, more than half of the rapamycin is
consumed and roughly equal amounts of the rapamycin
diastereomers 2-4 are formed (1:2:3:4 ≈ 2:1:1:1). The
amounts of diastereomers 3 and 4 peak at 2 and 4 min,
respectively. After 4 min, 3 and 4, along with rapamycin,
decrease steadily while the amount of 2 increases. After 30
min, the ratio of diastereomers effectively arrives at a steady
state, which consists of 1:2:3:4 ≈ 1:7:1:1.
cannot be considered a true intermediate. Presumably, the
Ti(OiPr)4-mediated retroaldol produces a ring-opened reac-
tive titanium intermediate (a C29-C30 enolate or its
equivalent) which is in equilibrium with the ring-closed
rapamycin isomers. This reactive intermediate must be
present in small proportions relative to that of the closed
macrocycle, and thus only a small amount of secorapamycin
is observed following an aqueous quench. The reactive
titanium species is apparently not formed directly from
secorapamycin under the same conditions but is only
produced from the retroaldol of the â-hydroxyketone.9
We were unable to alkylate the reactive species with
methyl iodide or methyl triflate; however, we could “trap”
the intermediate in a ring-opened form using benzaldehyde
as a competing aldol substrate (Scheme 1). The addition of
Scheme 1
1 molar equiv of benzaldehyde at a 20 mM concentration
resulted in a 1.4:1 ratio of rapamycin diastereomers (pre-
dominantly 28-epi) to benzaldehyde adducts 6 (as a mixture
of four diastereomers). The product ratio was shifted to 10:1
(7) The oxepane tautomer of 2 is also present in the reaction mixture.
This compound is found to the extent of about 1-3% during the first few
minutes of reaction and increases to as much as 20% after 60 min and to
80% (with concomitant loss of 2) following overnight treatment with
Ti(OiPr)4. Formation of the oxepane is dependent on Ti(OiPr)4 as 2 is stable
in dichloromethane. The oxepane tautomer can be isolated but rapidly
equilibrates to a 9:1 mixture of pyran:oxepane in methanol/water. Treatment
of FK506 with Ti(OiPr)4 similarly and more rapidly yields its oxepane
tautomer which reverts to pyran during chromatography (Wu Yang,
unpublished). Hughes has reported the preparation and isolation of the
oxepane tautomer of rapamycin: Hughes, P.; Musser, J.; Conklin, M.; Russo,
R. Tetrahedron Lett. 1992, 33, 4739-4742.
Our observations suggest the establishment of an equilib-
rium between all four C28/C29 diastereomers, with 28-
epirapamycin (2) predominating as the thermodynamically
preferred.7 Indeed, independent subjection of purified 2, 3,
or 4 to the same reaction conditions results in a similar
distribution of products. A plausible mechanism for this
equilibration, consistent with a low steady-state level of
secorapamycin, involves a reversible retroaldol/aldol reactions
cleavage of C28-C29 bond with concomitant loss of
stereochemical integrity at C28 and C29sfollowed by
reclosure of the macrocycle to provide the mixture of four
rapamycin stereoisomers.8
(8) Ti(OiPr)4 has been utilized in macrolide total synthesis to facilitate
ring size equilibration through transesterification; Kigoshi, H.; Suenaga,
K.; Mutou, T.; Ishigaki, T.; Atsumi, T.; Ishiwata, H.; Sakakura, A.; Ogawa,
T.; Ojika, M.; Yamada, K. J. Org. Chem. 1996, 61, 5326-5351, Paterson,
I.; Watson, C.; Yeung, K.-S.; Wallace, P. A.; Ward, R. A. J. Org. Chem.
1997, 62, 452-453.
(9) Ti(OiPr)4 has been reported to promote intermolecular aldol con-
densation of enolizable aldehydes and ketones; however, the reactions are
carried out at elevated temperatures and invariably result in dehydrated R,â-
unsaturated products. Mahrwald, R.; Schick, H. Synthesis 1990, 592-595.
Curiously, secorapamycin (5) is unreactive under the
isomerization conditions, even for extended times, and thus
(5) Several numbering schemes are in use for rapamycin. We have opted
to maintain the scheme previously employed by us (L.W.R.; D.A.H.) and
originated by the group responsible for the structure elucidation of
rapamycin. Findlay, J. A.; Radics, L. Can. J. Chem. 1980, 58, 579-590.
(6) Luengo, J. I.; Konialian, A. L.; Holt, D. A. Tetrahedron Lett. 1993,
34, 991-994.
2034
Org. Lett., Vol. 1, No. 12, 1999