Total synthesis of azumamide A and azumamide E, evaluation as histone deacetylase inhibitors, and design of a more potent analogue

Article abstract of DOI:10.1021/ol070046y

(Chemical Equation Presented) The unprecedented diastereoselective Mannich reaction of a Z-allylsulfoximine was a key step in the total synthesis of the marine natural products azumamide A and E, and an unnatural analogue. Their relative potency as histone deacetylase inhibitors was evaluated and found to correlate with predicted zinc-binding affinity.

Full text of DOI:10.1021/ol070046y

ORGANIC
LETTERS
2007
Vol. 9, No. 6
1105-1108
Total Synthesis of Azumamide A and
Azumamide E, Evaluation as Histone
Deacetylase Inhibitors, and Design of a
More Potent Analogue
Shijun Wen,† Krystle L. Carey,†,‡ Yoichi Nakao,§ Nobuhiro Fusetani,§
Graham Packham,‡ and A. Ganesan*,†
School of Chemistry, UniVersity of Southampton, Southampton
SO17 1BJ, United Kingdom, Cancer Sciences DiVision, School of Medicine,
UniVersity of Southampton, Southampton SO16 6YD, United Kingdom, and
Graduate School of Agricultural and Life Sciences, UniVersity of Tokyo,
Bunkyo-ku, Tokyo 113-8657, Japan
ganesan@soton.ac.uk
Received January 8, 2007
ABSTRACT
The unprecedented diastereoselective Mannich reaction of a Z-allylsulfoximine was a key step in the total synthesis of the marine natural
products azumamide A and E, and an unnatural analogue. Their relative potency as histone deacetylase inhibitors was evaluated and found
to correlate with predicted zinc-binding affinity.
In eukaryotes, DNA is tightly packaged with histone and
non-histone proteins into the higher order structure of
chromatin. Protein posttranslational modifications including
acetylation, methylation, phosphorylation, and ubiquitinyl-
ation constitute a “histone code”1 that ultimately regulates
gene transcription by modulating the unwinding of DNA and
recruitment of binding partners. Among the chromatin
modifying enzymes, zinc metallohydrolase class I/II histone
deacetylases (HDACs)2 have attracted the most attention as
anticancer targets. The two most advanced inhibitors in
development are Merck’s synthetic Vorinostat (Figure 1,
SAHA, suberoylanilide hydroxamic acid), recently receiving
FDA approval, and the natural product FK228 (depsipeptide)
in Phase II clinical trials. These exemplify the classic
pharmacophore for HDAC inhibitors: a “warhead” binding
the active site zinc, linked by a spacer mimicking the
substrate’s acetyl-lysine side chain, and a “cap” protruding
† School of Chemistry, University of Southampton.
‡ School of Medicine, University of Southampton.
§ University of Tokyo.
(1) (a) Biel, M.; Wascholowski, V.; Giannis, A. Angew. Chem., Int. Ed.
2005, 44, 3186-3216. (b) Nightingale, K. P.; O’Neill, L. P.; Turner, B. M.
Curr. Opin. Genet. DeV. 2006, 16, 125-136.
(2) (a) Miller, T. A.; Witter, D. J.; Belvedere, S. J. Med. Chem. 2003,
46, 5097-5116. (b) Moradei, O.; Maroun, C. R.; Paquin, I.; Vaisburg, A.
Curr. Med. Chem.: Anti-Cancer Agents 2005, 5, 529-560. (c) Bolden, J.
E.; Peart, M. J.; Johnstone, R. W. Nat. ReV. Drug DiscoVery 2006, 5, 769-
784.
10.1021/ol070046y CCC: $37.00
© 2007 American Chemical Society
Published on Web 02/21/2007

tors, nevertheless, its reported potency was similar to
azumamide E, despite the carboxylic acid being expected to
be a stronger metal binder; (2) a structural similarity to
FK228, with the depsipeptide replaced by a peptide back-
bone; and (3) A cyclic tetrapeptide scaffold, with all amino
acids in the D-configuration, which would result in unfavor-
able steric congestion,5 relieved in the azumamides by
expansion of one R-amino acid to a â-amino acid.
Retrosynthetically, macrolactamization of a linear precur-
sor was anticipated to be achievable although likely to be
challenging and dependent upon the position of cyclization.
This reduced the problem to an efficient preparation of the
unnatural â-amino acid 4. A total synthesis of azumamides
A and E has recently appeared6 where this moiety was
assembled by a multistep operation from 1,3-propanediol.
Our independent and contemporaneous effort envisioned a
Mannich reaction, using Ellman’s tert-butylsulfinyl auxiliary7
as R*. While the Mannich reaction is precedented8 with
simpler examples, the choice of â,γ-unsaturated imine 5
appeared audacious due to potential for bond isomerization
or migration to conjugated enamine or R,â-unsaturated
isomers. Indeed, the SciFinder database showed no examples
of R-methylene-â,γ-unsaturated sulfinylimines.
Figure 1. SAHA and FK228, with “warhead” highlighted.
beyond the substrate binding channel and contacting the
enzyme’s “rim”. While FK228 has a weaker thiol warhead
compared to a hydroxamic acid, the additional binding
interactions from the macrocyclic cap result in an overall
potency superior to SAHA, and confer selectivity as the
homology betweeen HDAC isoforms is divergent in the rim
region.
In practice, benzyl 4-pentynoate was subjected to Mart´ın’s
four-step sequence9 for conversion of an acetylene to a â,γ-
cis-unsaturated aldehyde (Scheme 1). The fragile aldehyde
Given our interest3 in natural product HDAC inhibitors,
the disclosure4 of the azumamides (Figure 2) was noteworthy
Scheme 1. Synthesis of the â-Amino Acid
Figure 2. Azumamides and retrosynthesis of the â-amino acid.
for several reasons: (1) Azumamide A contains a carbox-
amide warhead, hitherto a rare motif among HDAC inhibi-
(3) (a) Yurek-George, A.; Habens, F.; Brimmell, M.; Packham, G.;
Ganesan, A. J. Am. Chem. Soc. 2004, 126, 1030-1031. (b) Davidson, S.
M.; Townsend, P. A.; Carroll, C.; Yurek-George, A.; Balasubramanyam,
K.; Kundu, T. K.; Stephanou, A.; Packham, G.; Ganesan, A.; Latchman,
D. S. ChemBioChem 2005, 6, 162-170. (c) Doi, T.; Iijima, Y.; Shin-ya,
K.; Ganesan, A.; Takahashi, T. Tetrahedron Lett. 2006, 47, 1177-1180.
(4) (a) Nakao, Y.; Yoshida, S.; Matsunaga, S.; Shindo, N.; Nagai, K.;
Suzuki, K.; van Soest, R. W. M.; Fusetani, N. 46th Symposium on the
Chemistry of Natural Products, Hiroshima, 2004; Abstract 5(B-1). (b)
Nakao, Y.; Yoshida, S.; Matsunaga, S.; Shindoh, N.; Terada, Y.; Nagai,
K.; Yamashita, J. K.; Ganesan, A.; van Soest, R. W. M.; Fusetani, N. Angew.
Chem., Int. Ed. 2006, 45, 7553-7557.
7a was successfully converted to sulfinylimine 8a by CuSO4-
mediated dehydration. The ensuing Mannich reaction with
a propionate ester enolate proceeded with high diastereo-
selectivity to afford â-amino acid 9a. Although the last two
reactions were of modest yield due to the poor stability of
1106
Org. Lett., Vol. 9, No. 6, 2007

poorer diastereoselectivity providing a 5:1 ratio of products.
Removal of the auxiliary and derivatization with Mosher’s
acid chloride (MTPA-Cl) led to (2R,3S) diastereomer 11.
Comparison of HPLC and NMR data with that for the MTPA
amide obtained from the natural product enabled conslusive
assignment of the latter as the (2S,3R) diastereomer.
Scheme 2. Synthesis of the Azumamide Degradation
Fragment
With the structure of the natural products secured, we
returned to the total synthesis. The â-amino acid 9a was
successfully progressed to azumamide A, but alkaline
hydrolysis of the benzyl ester protecting group to provide
azumamide E was sluggish and accompanied by epimeriza-
tion of the â-amino acid. We then devised a second-
generation route using the trichloroethyl ester as a protecting
group cleavable under neutral conditions. Starting from the
trichloroethyl ester analogue 6b, the chemistry in Scheme 1
proceeded uneventfully with the formation of Ellman imine
8b and its Mannich reaction both proceeding in an improved
>50% yield.
Following selective deprotection of the p-methoxybenzyl
ester, 9b was coupled with D-Val-D-Ala-OAll dipeptide to
give 12 (Scheme 3). The tert-butylsulfinyl group having
admirably served its dual purpose as chiral auxiliary and
protecting group, it was now cleaved and the amine ho-
mologated to linear tetrapeptide 13. Following deprotection
at the N- and C-termini, HATU-mediated macrolactamization
furnished the azumamide scaffold 14 in 55-85% yield in
individual experiments. In comparison, this was a difficult
step requiring significant experimentation in the previous6
total synthesis. We believe this underscores the importance
of the position of macrocylization, which was between the
7a and 8a, they provided gratifying support for the overall
strategic disconnection and its potential to deliver the
â-amino acid in a concise manner.
Before proceeding further, it was important to establish
the nature of the â-amino acid within the azumamides. In
the original isolation, this was determined by degradation
and comparison with simpler model compounds. To resolve
any ambiguity regarding the relative and absolute configu-
ration, we prepared imine 10 (Scheme 2) containing the
enantiomeric sulfinylimine to 8a. With this stable saturated
imine, the Mannich reaction occurred in high yield, but in
Scheme 3. Total Synthesis of Azumamide A, Azumamide E, and Hydroxamic Acid Analogue
Org. Lett., Vol. 9, No. 6, 2007
1107

Phe residue and the â-amino acid, unlike our Phe to Ala
cyclization.
FK228. This suggests that any deficiencies in switching from
depsipeptide to peptide backbone can be overcome by
increasing the affinity of the zinc-binding warhead. Interest-
ingly, when a hydroxamic acid was substituted10 for the
epoxyketone warhead in chlamydocin, another cyclopeptide
HDAC inhibitor, an increase in potency was not observed.
In summary, we have achieved a total synthesis of
azumamides A and E. The stereoselective Mannich reaction
of an unusual â,γ-unsaturated sulfinylimine is likely the most
challenging transformation accomplished to date with the
Ellman auxiliary, and enabled a concise route with only 15
steps in the longest linear sequence. Within the series 1-3,
we show that HDAC inhibitory activity is in the order
carboxamide < carboxylic acid < hydroxamic acid, as
predicted by the relative metal binding of these warheads.
We anticipate that variation of the azumamide scaffold will
yield additional synthetic HDAC inhibitors of interest.
The trichloroethyl ester in 14 was reductively removed
without side reactions, and occurred nearly quantitatively to
afford azumamide E (2). Small discrepancies were observed
between our 1H NMR data and that of the naturally isolated
material. The same phenomenon was previously6 noted, and
attributed to pH sensitivity arising from residual TFA used
in the HPLC purification of the natural product. Carbodi-
imide-mediated reaction of azumamide E (2) with ammonia
then completed the total synthesis of azumamide A (1).
Azumamide A and azumamide E inhibited total HDACs
from HeLa cell extracts with IC50s of 5800 ( 1200 and 110
( 33 nM, respectively. Thus, in our hands, the carboxylic
acid warhead is much more potent than the carboxamide.
This is consistent with expected zinc-binding affinity and
we believe the original4 report might have been compromised
by impurities.
Acknowledgment. We thank Cancer Research-UK for
funding and Dr. Richard Brown, University of Southampton,
for advice on Mannich reaction conditions with Ellman
sulfinylimines.
The biological activity of the azumamides is significantly
weaker than FK228 (IC50 15 ( 9 nM in the same assay). To
determine if this was due to the difference in warhead or
the shift from a depsipeptide to peptide scaffold, we
converted azumamide E (2) to the unnatural hydroxamic acid
3. This analogue had an IC50 of 7.0 ( 2.5 nM, similar to
Supporting Information Available: Detailed experi-
mental procedures and NMR spectra for all compounds. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(5) Shute, R. E.; Kawai, M.; Rich, D. H. Tetrahedron 1998, 44, 685-
695.
(6) Izzo, I.; Maulucci, N.; Bifulco, G.; De Riccardis, F. Angew. Chem.,
Int. Ed. 2006, 45, 7557-7760.
(7) Ellman, J. A. Pure Appl. Chem. 2003, 75, 39-46.
(8) Tang, T. P.; Ellman, J. A. J. Org. Chem. 2002, 67, 7819-7832.
(9) D´ıaz, D. D.; Betancort, J. M.; Criso´stomo, F. R. P.; Mart´ın, T.; Mart´ın,
S. V. Tetrahedron 2002, 58, 1913-1919.
OL070046Y
(10) Nishino, N.; Jose, B.; Shinta, R.; Kato, T.; Komatsu, Y.; Yoshida,
M. Bioorg. Med. Chem. 2004, 12, 5777-5784.
1108
Org. Lett., Vol. 9, No. 6, 2007