One-step construction of the pentacyclic skeleton of saframycin A from a "Trimer" of alpha-amino aldehydes.

Article abstract of DOI:10.1021/ol0063398

The entire skeleton of the saframycin antitumor antibiotics is assembled in one remarkable transformation (8 --> 9) from an N-linked oligomer of three alpha-amino aldehyde components, a reaction pathway that may parallel the biosynthetic route to the safr

Full text of DOI:10.1021/ol0063398

ORGANIC
LETTERS
2000
Vol. 2, No. 19
3019-3022
One-Step Construction of the
Pentacyclic Skeleton of Saframycin A
from a “Trimer” of r-Amino Aldehydes
Andrew G. Myers* and Daniel W. Kung
Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138
myers@chemistry.harVard.edu
Received July 17, 2000
ABSTRACT
The entire skeleton of the saframycin antitumor antibiotics is assembled in one remarkable transformation (8 f 9) from an N-linked oligomer
of three r-amino aldehyde components, a reaction pathway that may parallel the biosynthetic route to the saframycins.
Biosynthetic studies have shown that the potent antitumor
alkaloid saframycin A (1) can be assembled from glycine,
alanine, and two molecules of tyrosine.1,2 Five methionine-
derived methyl groups and exogenous hydrogen cyanide
provide all remaining carbon atoms. The path that connects
these precursors to the pentacyclic skeleton of 1 is not known,
nor has any detailed proposal been presented. Gross features
of the connectivity have been established on the basis of
isotopic incorporation studies (Figure 1).1 When the R-amino
acid precursors are mapped onto structure 1 it is evident that
the carboxyl groups of the glycine residue and both tyrosine
residues have undergone reduction during biosynthesis and
rest in the final product at an oxidation state equivalent to
that of an aldehyde group (indicated by the open circles in
structure 1).3 Researchers at Ciba-Geigy have recently
isolated and cloned the genes involved in the biosynthesis
of the closely related antibiotic saframycin Mx1.4 On the
basis of the DNA sequence analysis it was proposed that
the genes isolated code for two multifunctional (nonriboso-
mal) peptide synthetases and a third protein with O-methyl
transferase activity. Each synthetase was proposed to contain
two amino acid activating domains and, unprecedented
among nonribosomal peptide synthetases, a putative reduc-
tase activity. It was proposed that saframycin A is derived
from an oligopeptide precursor, but the stepwise process by
which this might occur was not defined.4
We recently developed a short (eight-step) and enantio-
selective synthetic route to 1 by the directed condensation
of compounds 2 and 3 (N- and C-protected versions,
(1) (a) Mikami, Y.; Takahashi, K.; Yazawa, K.; Arai, T.; Namikoshi,
M.; Iwasaki, S.; Okuda, S. J. Biol. Chem. 1985, 260, 344-348. (b) Arai,
T.; Yazawa, K.; Takahashi, K.; Maeda, A.; Mikami, Y. Antimicrob. Agents
Chemother. 1985, 28, 5-11.
(2) Reviews of the saframycins: (a) Arai, T.; Kubo, A. In The Alkaloids;
Brossi, A., Ed.; Academic Press: New York, 1983; Vol. 21, Chapter 3. (b)
Remers, W. A. In The Chemistry of Antitumor Antibiotics; Wiley-
Interscience: New York, 1988; Vol. 2, Chapter 3.
(3) This assignment of oxidation states is made from the perspective of
synthetic equivalency, where the cyano and benzoquinone substituents are
treated as heteroatom equivalents.
(4) (a) Pospiech, A.; Cluzel, B.; Bietenhader, J.; Schupp, T. Microbiology
1995, 141, 1793-1803. (b) Pospiech, A.; Bietenhader, J.; Schupp, T.
Microbiology 1996, 142, 741-746.
10.1021/ol0063398 CCC: $19.00 © 2000 American Chemical Society
Published on Web 08/23/2000

remarkable transformation from an N-linked oligomer of the
three R-amino aldehyde components 2, 3, and 4, a reaction
that suggests for the first time a viable pathway linking 1
with an oligopeptide precursor and, therefore, a possible
biosynthetic route.
Figure 3. A proposed trimeric R-amino aldehyde precursor to 1.
The specific oligomer that was targeted initially was the
trimeric amino nitrile 6, in which 2, 3, and 4 are linked by
sequential Strecker reactions. The amino nitrile groups serve
to covalently join the three R-amino aldehyde components
and were proposed to function later as precursors to
electrophilic imine or iminium intermediates that would
mediate the three cyclization reactions leading to the safra-
mycin skeleton.7 We had previously demonstrated that
R-amino aldehydes can be coupled using the Strecker
reaction without epimerization of the R-stereocenter.8 Be-
cause amino nitrile formation was anticipated to form two
diastereomeric products in each case (of no consequence in
later C-C bond-forming reactions), we elected to use 13C-
labeled cyanide in the synthesis to facilitate 13C-NMR
analysis of the products. Also, we began the sequence with
a single diastereomer of the C-protected R-amino aldehyde
component 3, bearing a 13C-label on the cyano group.
The order of introduction of R-amino aldehyde compo-
nents was 3 + 2, and then 4, representing C- to N-terminus
directionality in the synthesis. Mixing 3 (1 equiv, 92% ee,
Figure 1. Saframycins A and Mx1, identification of amino acid
precursors and acid to aldehyde oxidation state changes (°).
respectively, of the same complex R-amino aldehyde), and
N-Fmoc glycinal (4, see Figure 2).5,6 Over the course of five
(5) Myers, A. G.; Kung, D. W. J. Am. Chem. Soc. 1999, 121, 10828-
10829.
(6) Other syntheses of saframycins: (()-saframycin B: (a) Fukuyama,
T.; Sachleben, R. A. J. Am. Chem. Soc. 1982, 104, 4957-4958. (b) Kubo,
A.; Saito, N.; Yamato, H.; Masubuchi, K.; Nakamura, M. J. Org. Chem.
1988, 53, 4295-4310. (()-saframcyin A: (c) Fukuyama, T.; Yang, L.;
Ajeck, K. L.; Sachleben, R. A. J. Am. Chem. Soc. 1990, 112, 3712-3713.
(-)-saframycin A: (d) Martinez, E. J.; Corey, E. J. Org. Lett. 1999, 1,
75-77. See also: (e) Zhou, B.; Edmondson, S.; Padron, J.; Danishefsky,
S. J. Tetrahedron Lett. 2000, 41, 2039-2042. (f) Zhou, B.; Guo, J.;
Danishefsky, S. J. Tetrahedron Lett. 2000, 41, 2043-2046.
(7) For selected examples of the use of amino nitriles as imine/iminium
ion precursors in biomimetic systems, see: (a) Overman, L. E.; Jacobsen,
E. J. Tetrahedron Lett. 1982, 23, 2741-2744. (b) Ksander, G.; Bold, G.;
Lattman, R.; Lehmann, C.; Fruh, T.; Xiang, Y.-B.; Inomata, K.; Buser, H.-
P.; Schreiber, J.; Zass, E.; Eschenmoser, A. HelV. Chim. Acta 1987, 70,
1115-1172. (c) Bonin, M.; Grierson, D. S.; Royer, J.; Husson, H.-P. Org.
Synth. 1992, 70, 54-59.
Figure 2. Precursors in the synthesis of saframycin A (1).
steps these components were linked in stepwise fashion, in
a sequence involving two Pictet-Spengler cyclization reac-
tions and an intramolecular Strecker reaction, to form the
pentacyclic saframycin A precursor 5.5 Here, we show that
the entire saframycin skeleton can be assembled in one
(8) Myers, A. G.; Kung, D. W.; Zhong, B.; Movassaghi, M.; Kwon, S.
J. Am. Chem. Soc. 1999, 121, 8401-8402.
3020
Org. Lett., Vol. 2, No. 19, 2000

13C-labeled cyano group) and its N-protected R-amino
aldehyde counterpart 2 (1.05 equiv, 96% ee) in dichloro-
methane with suspended sodium sulfate led to formation of
the corresponding imine, cleanly and without R-epimeriza-
tion, as previously demonstrated.5 In this instance, however,
the imine was captured by Strecker reaction with hydrogen
cyanide in methanol at 23 °C (1.6 equiv of acetic acid, 1.5
equiv of K13CN, Scheme 1), whereas in the prior route the
combined yield). Only two diastereomers were detected
spectroscopically, and these were present in the same ratio
as the starting material 7, suggesting that the cyclic aminal
had been formed with a single stereochemistry, tentatively
assigned as shown in Scheme 1.
Several experiments were conducted to investigate the
proposed rearrangement of 8 to the pentacyclic skeleton of
the saframycins (9). Treatment of 8 with Lewis acid catalysts
typically led to highly complex and difficultly characterized
reaction mixtures, an outcome that was not surprising given
the many reaction manifolds potentially available to this tris
R-amino aldehyde equivalent. However, by sequential treat-
ment of 8 with the Lewis acids lithium bromide (dimethoxy-
ethane, reflux) and then zinc chloride (trifluoroethanol-THF,
23 °C), and assisted by the fact that 8 was fortuitously well
separated chromatographically from all other reaction prod-
ucts, it was possible to isolate the desired pentacyclic
saframycin A precursor 9 from the reaction mixture in pure
form (4%). With further experimentation, conditions were
found to bring about the transformation of 8 to 9 in one step
and in higher yield (Scheme 2); heating a solution of 8 in
tetrahydrofuran at reflux in the presence of magnesium
bromide etherate (20 equiv) afforded 9 in 8.4 and 9.0% yield
in two separate experiments. Importantly, N-acylation of 9
with the enantiomeric Mosher acid chlorides followed by
HPLC analysis of the amide products established that 9 had
been formed without racemization (9 was of 99% ee).
N-Methylation of 9 with formalin and sodium triacetoxy-
borohydride in acetonitrile afforded the pentacyclic safra-
mycin A precursor 5, identical with an authentic sample
prepared by the earlier synthetic route (1H NMR, IR, TLC,
and HPLC analysis), except for the anticipated spectroscopic
differences attributed to the 13C-label. Intermediate 5 can be
transformed into saframycin A in three steps (50% yield).5
Scheme 1. Synthesis of an N-Linked Trimeric R-Amino
Aldehyde Precursor to the Saframycin Skeleton
The one-step conversion of the N-linked oligomer 8 to
the pentacyclic intermediate 9 involves an exceptional
number of individual steps. Three cyclization reactions occur,
and three of the five stereocenters of saframycin A are
established in this step. In theory, each of the five stereogenic
centers of the precursor 5 is epimerizable under the reaction
conditions. A single epimerization event may divert the
course of reaction from 9. In that product which is formed,
the R-amino aldehyde-derived R-centers are preserved. Many
viable sequences can be envisioned to transform 8 into 9;
the pathway shown in Scheme 2 is proposed as that which
actually occurs. In background studies, we have found that
aminals have a greater propensity to form imine or iminium
ion intermediates under mildly acidic conditions than sec-
ondary amino nitriles which, in turn, are more labile than
tertiary amino nitriles.8 For this reason, we propose that
cleavage of the aminal occurs first, followed by trapping of
the resultant imine by Pictet-Spengler cyclization, as
depicted in Scheme 2. Subsequent ionization of the secondary
amino nitrile is proposed to initiate a second Pictet-Spengler
cyclization. Finally, ionization of the tertiary amino nitrile
group leads to internal Strecker reaction to form the
pentacyclic product 9. It is interesting to note that the
ordering of the two Pictet-Spengler reactions in this
imine was cyclized by warming (35 °C) in the presence of
lithium bromide.5 The expected R-amino nitriles 7 (1.1:1
mixture of diastereomers) were obtained in 92% yield after
isolation by flash column chromatography. Sequential re-
moval of the silyl ethers (triethylamine trihydrofluoride, 2.5
equiv, CH3CN, 23 °C) and the N-Fmoc group (30% piperi-
dine-CH2Cl2, 23 °C, 76%, two steps) of 7 afforded the fully
deprotected “dimer” for coupling with the third component,
N-Fmoc glycinal (4). Attempted Strecker coupling of these
components was complicated by internal cyclization of the
glycinaldimine intermediate. Recognizing that such a process
provided an aminal product that was functionally equivalent
to the trimeric R-amino nitrile originally targeted, the
condensation reaction was optimized to form this product
(compound 8, Scheme 1). Thus, addition of 4 (1.1 equiv) to
a solution of the deprotected dimeric R-amino aldehyde (1
equiv) in dichloromethane at 23 °C led to smooth condensa-
tion in the absence of hydrogen cyanide to afford a product
formulated as the cyclic aminals 8. These products were not
stable to chromatography on silica gel, but 1H- and 13C-NMR
analysis showed that they had been formed cleanly (∼90%
Org. Lett., Vol. 2, No. 19, 2000
3021

Scheme 2. Proposed Pathway for the Transformation of the N-Linked Oligomer 8 to the Pentacyclic Saframycin A Precursor 9
proposed sequence is opposite to that of our earlier stepwise
condensation route.5 Both Pictet-Spengler cyclizations are
believed to proceed with cis selectivity, as observed in the
earlier stepwise route, but stereochemical ratios cannot be
assigned in this experiment.
Given the number of steps involved and the many
alternative reaction pathways available to the intermediates
of Scheme 2, the transformation of 8 to 9 is remarkable.
Although the efficiency of the sequence does not rival that
of our earlier route (it is, however, one step shorter), the
primary significance of the transformation is not as a process
improvement, rather, it is the demonstration that an oligo
R-amino aldehyde is a viable synthetic precursor to 1, even
in the absence of an enzymic catalyst. In light of the
implication that 1 is biosynthesized from an oligopeptide
precursor,4 the transformation of 8 to 9 is particularly
significant, for it validates the connectivity implicit in such
a proposal. Moreover, it provides the first detailed stepwise
pathway that has been proposed to link an oligopeptide with
1. Although many alternative sequences of reduction and
cyclization reactions can be imagined, the idea that the three
carboxylate carbons of the oligopeptide precursor are reduced
prior to any cyclization event is now clearly shown to be
one viable pathway, inasmuch as spontaneous cyclization
to the saframycin skeleton has been demonstrated.
Acknowledgment. Financial support from the National
Institutes of Health is gratefully acknowledged. D.W.K. is
grateful to the National Science Foundation for predoctoral
fellowship support.
OL0063398
3022
Org. Lett., Vol. 2, No. 19, 2000