Synthesis of proposed oxidation-cyclization-methylation intermediates of the coumarin antibiotic biosynthetic pathway

Article abstract of DOI:10.1021/ol035101r

(Matrix presented) A chemoenzymatic synthesis was described to prepare proposed oxidation-cyclization-methylation intermediates of the coumarin antibiotic biosynthetic pathway. The successful synthesis of these fragile molecules relies heavily on mild enz

Full text of DOI:10.1021/ol035101r

ORGANIC
LETTERS
2003
Vol. 5, No. 18
3233-3236
Synthesis of Proposed
Oxidation−Cyclization−Methylation
Intermediates of the Coumarin Antibiotic
Biosynthetic Pathway
,†
Junhua Tao,†,‡ Shanghui Hu,‡ Michelle Pacholec,† and Christopher T. Walsh*
Department of Biological Chemistry and Molecular Pharmacology, HarVard Medical
School, 240 Longwood AVenue, Boston, Massachusetts 02115, and Pfizer, Inc.,
La Jolla Laboratories, 4215 Sorrento Valley BlVd., San Diego, California 92121
christopher_walsh@hms.harVard.edu
Received June 16, 2003
ABSTRACT
A chemoenzymatic synthesis was described to prepare proposed oxidation−cyclization−methylation intermediates of the coumarin antibiotic
biosynthetic pathway. The successful synthesis of these fragile molecules relies heavily on mild enzymatic deprotection and efficient enzymatic
kinetic resolution to minimize epimerization, decomposition, multiple orthogonal protections, and retro aldol reactions often encountered in
their chemical synthesis.
Coumarin-containing antibiotics such as novobiocin (1,
Figure 1, aminocoumarin core highlighted), clorobiocin (2),
coumermycin A1 (3), and simocyclinone (4), produced by
Streptomyces species, have gained renewed interest since the
discovery that they are potent inhibitors of bacterial DNA
gyrase, which is essential for cell viability.1a,b In addition,
these coumarin antibiotics are potent against methicilin-
resistant strains of Staphylococci species, currently one of
the major concerns in the treatment of bacterial infections.2
Biosynthetically, it is proposed that the conserved ami-
nocoumarin core is produced via an oxidation-cyclization-
methylation sequence from thioester 5 (Scheme 1),3
a
substrate (L-â-OH-Tyr)-carrier protein (T) conjugate via a
phosphopantetheine linker reminiscent of classic nonribo-
somal peptide synthetase (NRPS) systems (squiggly line, see
† Harvard Medical School.
‡ Pfizer, Inc.
(1) (a) Maxwell, A. Trends Microbiol. 1997, 5, 102. (b) Calia, H.;
Hoerman, L.; Schultz, P.; Lebeau, L.; Mallouh, V.; Wigley, D. B.; Wang,
J. C.; Mioskowski, C.; Oudet, P. J. Mol. Biol. 1994, 236, 618.
(2) Boehm, H. J.; Boehringer, M.; Bur, D.; Gmeunder, H.; Huber, W.;
Klaus, W.; Kostrewa, D.; Kuehne, H.; Luebbers, T.; Meunier-Keller, N.;
Mueller, F. J. Med. Chem. 2000, 43, 2664.
Figure 1. Coumarin-containing antibiotics.
box in Scheme 1).4 The first oxidation is predicted to be
catalyzed by enzymes NovJ/K to yield keto derivative 6,
(3) Chen, H.; Walsh, C. T. Chem. Biol. 2001, 8, 301.
10.1021/ol035101r CCC: $25.00 © 2003 American Chemical Society
Published on Web 08/02/2003

hydroxy thioester 10 contains one primary amine and two
hydroxyl groups. We disclose here a chemoenzymatic
approach to the preparation of 10-12. This method can also
be adapted to the synthesis of aminocoumarin 8, the proposed
substrate of NovO (Scheme 1).
Scheme 1
The synthesis started with commercially available N-Boc
diethyl malonate 13 (Scheme 2). Upon selective enzymatic
Scheme 2
followed by another oxidation by a yet unidentified protein
to produce thioester 7.3 At this stage, aminocoumarin 8 could
be released from the carrier protein via intramolecular
cyclization and converted to methyl aminocoumarin 9 by
the methyltransferase NovO in novobiocin or to chloro
aminocoumarin by a halogenase in clorobiocin (not shown,
see 2 in Figure 1).
Currently, however, the evaluation of the proposed ami-
nocoumarin biosynthesis is hampered by the lack of avail-
ability of N-acetyl cysteamine (NAC) thioester (2S,3R)-â-
OH-Tyr-SNAC (10) and its diastereomers (2S,3S)-11 and
(2R,3R)-12 (Figure 2), which are small-molecule surrogates
hydrolysis by pig liver esterase (PLE), monoacid 14 was
obtained in good yields (84%) at gram scale. In the presence
of Et3N and MgCl2, the monoacid was then condensed with
(4-acetoxy)-benzoyl chloride (15) to produce the desired
â-keto ester 16 (86%), which was carried over to the next
step without further purification.6 It should be noted that
vigorous agitation is essential for this reaction to avoid low
yields (20-30%). The acetyl group in 16 was then removed
using Candida rugosa lipase (CRL) to provide the key
intermediate 17 in excellent yields (91%) without hydrolyz-
ing the ethyl ester. In this scheme, both PLE and CRL were
identified by a high-throughput screening protocol reported
recently.7 Nonselective chemical hydrolysis causes extensive
decomposition of the starting material due to the instability
of the resulting â-keto acid (not shown).
To create the desired (2S,3R) stereochemistry in 10, initial
efforts were directed toward the microbial reduction of 17
using yeast libraries.8a,b Preliminary screening showed prom-
ising results with 10∼30% conversions for some strains. To
further optimize the reaction conditions, however, a reference
standard of syn amino alcohol 21 was needed, and an
independent synthesis of 21 was initiated, beginning with
cinnamic ester 19 prepared from 18 by O-benzylation
(>98%) (Scheme 3). Compound 20, whose stereochemistry
is known, was produced as the major regioisomer in 40-
45% yields by Sharpless aminohydroxylation9 and subse-
quently converted to 21 by switching protecting groups. By
this method, the desired product was isolated in high yields
(98%) and dr (syn/anti > 100:1) and good enantiomeric
excess (89% ee by chiral HPLC). At this stage, it appears
that the current route could be used to replace enzymatic
Figure 2. Stereoisomeric thioester precursors for NovJ/K.
of the substrate-protein conjugate 5 and needed to charac-
terize the function of NovJ/K. Despite the apparent structural
simplicity, the range of densely functional groups together
with multiple chiral centers in these inherently unstable
intermediates requires mild conditions for efficient synthesis.5a,b
For example, in general, thioesters are labile toward even
weak nucleophiles such as MeOH, and yet R-amino-â-
(6) Krysan, D. J. Tetrahedron Lett. 1996, 37, 3303.
(7) Yazbeck, D. R.; Tao, J.; Martinez, C. A.; Kline, B. J.; Hu, S. AdV.
(4) Marahiel, M.; Stachelhaus, T.; Mootz, H. D. Chem. ReV. 1997, 97,
2651.
(5) For representative syntheses of thioesters with no amino groups: (a)
Cheung, K.-M.; Coles, S. J.; Hursthouse, M. B.; Johnson, N. J.; Shoolingin-
Jordan, P. M. Angew. Chem., Int. Ed. 2001, 41, 1199. (b) Hunziker, D.;
Wu, N.; Kenoshita, K.; Cane, D. E.; Khosla, C. Tetrahedron Lett. 1999,
40, 635.
Synth. Catal. 2003, 345, 524-532.
(8) For enzymatic reduction on similar substrates, see: (a) Fadnavis, N.
W.; Vadivel, S. K.; Sharfuddin, M.; Bhalerao, U. T. Tetrahedron:
Asymmetry 1997, 8, 4003. (b) Dillon, M. P.; Hayes, M. A.; Simpson, T. J.;
Sweeney, J. B. Biorg. Med. Chem. Lett. 1991, 1, 223.
(9) (a) Dong, L.; Miller, M. J. J. Org. Chem. 2002, 67, 4759. (b) Tao,
B.; Schlingloff, G.; Sharpless, K. B. Tetrahedron Lett. 1998, 39, 2507.
3234
Org. Lett., Vol. 5, No. 18, 2003

Encouraged by this success, we decided to prepare the
other two enantiomeric stereoisomers 11 and 12 through
enzymatic kinetic resolution.10a-d First, a racemic amino
alcohol 24 was prepared from â-keto ester 16 (85%) in high
diastereoselectivity (dr > 10:1) using NaBH4 at -20 °C
(Scheme 5). The key step is an enzymatic kinetic resolution
Scheme 3
Scheme 5
reduction if the enantiomeric purity could be further im-
proved in subsequent synthetic steps toward 10.
Since both the retro aldol reaction and epimerization at
C-2 chiral center were observed when 21 underwent chemical
hydrolysis (syn/anti ) 10:1 vs >100:1), we decided to
remove the ester group enzymatically (Scheme 4). Plate
Scheme 4
where Aspergillus melleus protease (AMP) was identified
from library screening to be able to hydrolytically resolve
the ester precursor 24 to give both the (2S,3S)-acid 25 (49%
yield, 96% ee) and (2R,3R)-ester 26 (35% yield, 94% ee) in
good yields and high optical purity. The anti 2-amino to
3-hydroxy relationship in 26 was unambiguously determined
by comparing it to the syn amino alcohol 21. The acid 25
was subsequently converted to 11 as an HCl salt via the
intermediate 27 with 70% yield over two steps upon coupling
and deprotection. As for the ester 26, it was hydrolyzed and
then turned into 12 via the intermediate 28 (65% over three
steps) by the same coupling and deprotection procedure. The
ees for both 27 (96% ee) and 28 (94% ee) were measured
by chiral HPLC, indicating that no loss of enantiomeric purity
took place during coupling.
With targets 10-12 in hand, we decided to turn our
attention to the preparation of other intermediates in the
biosynthesis. First, â-keto ester 16 could be readily converted
to R-amino â-keto acid salt 30 (Scheme 6), the proposed
product of NovJ/K upon thioester cleavage (Scheme 1),3 by
a two step sequence: removal of the Boc group to give
R-amino â-keto ester 29 (94%) followed by enzymatic
hydrolysis catalyzed by MML (85%). The Boc group in 16
has to be removed before enzymatic hydrolysis since N-Boc
â-keto acid 31 from enzymatic hydrolysis decarboxylates
screening showed that Mucor miehei lipase (MML) was able
to produce acid 22 (69%) at pH 6-7. Interestingly, the
desired (2S,3R)-enantiomer was enriched by chiral HPLC
analysis (95% ee for 22 vs 89% for 21) as a result of kinetic
resolution during hydrolysis. To complete the synthesis of
10, the key step is thioester formation between N-acetyl
cysteamine and 22. Remarkably, in the presence of DCC/
HOBt, both free hydroxyl groups in 22 did not interfere with
the desired coupling to give thioester 23 and the reaction
was finished within 2 h as shown by HPLC. However, the
product is not stable under the reaction conditions and should
be isolated within this time frame before decomposition into
an unknown byproduct. In this fashion, 23 was generated
with excellent ee (95% by chiral HPLC), indicating that no
epimerization occurred during thioester formation. Other
coupling conditions tested were not as successful.5a,b Without
further purification, 23 was deprotected using dry HCl and
converted to 10 as an HCl salt in high yields (70% over two
steps from 22) and good UV purity (90-95%). It should be
noted that this compound has limited stability at -20 °C
(2-3 weeks) and lower temperatures are preferred for
prolonged storage. In addition, the thioester is prone to
alcoholysis in MeOH, and enzyme assays should be per-
formed in weakly nucleophilic buffers, e.g., phosphate over
Tris.
(10) For some recent applications of enzymatic resolution, see: (a) Demir,
A. S.; Sesenoglu, O. Org. Lett. 2002, 4, 2021. (b) Taber, D. F.; Xu, M.;
Hartnett, J. C. J. Am. Chem. Soc. 2002, 124, 13121. (c) Delhaes, L.; Biot,
C.; Berry, L.; Delcourt, P.; Maciejewski, L. A.; Camus, D.; Brocard, J. S.;
Dive, D. ChemBiochem 2002, 3, 418. (d) Luna, A.; Gotor, V. Org. Lett.
2002, 4, 3627.
Org. Lett., Vol. 5, No. 18, 2003
3235

precipitate from the reaction in 93% yield and high purity
(>95% by HPLC and 13C NMR).
Scheme 6
In conclusion, the intrinsically unstable stereoisomers 10-
12 were synthesized as the proposed substrates for NovJ/K
in high optical purity (94-96% ees) over six steps. The
method can be readily adapted to the preparation of the keto
acid salt 30, a product of NovJ/K, and aminocoumarin 8, a
proposed substrate for NovO. While the successful synthesis
of these intermediates may expedite the elucidation of the
coumarin antibiotic biosynthetic pathway, the chemoenzy-
matic approach described here could also be modified to
synthesize other structurally similar thioester intermediates
in many NRPS systems, e.g., chloramphenicol and vanco-
mycin.4 Not only can enzymes facilitate deprotection by
minimizing epimerization, decomposition, and multiple
orthogonal protection-deprotection and retro aldol reactions,
which are often associated with the chemical synthesis of
these fragile molecules, they can also be applied to prepare
two enantiomers convergently with high individual chirality
from one racemate through kinetic resolution. We are
currently using these intermediates to evaluate the function
of heterologously expressed proteins NovJ/K and NovO.
spontaneously to amino ketone 32. It should be noted that
30 should be kept at -80 °C for prolonged storage to
minimize decomposition.
In addition, the aminocoumarin core 8 could be prepared
from â-keto ester precursor 34, a condensed product (85%)
from monoacid 14 and acid chloride 33 (Scheme 7). Upon
Scheme 7
Acknowledgment. Research at HMD was supported by
NIH grant GM20011 (to C.T.W) and a National Defense
Science & Engineering Graduate (NDSEG) fellowship (to
M.P.). J.T. and S.H. thank Kim Albizati and Van Martin for
helpful discussions and Shirley Wang for HRMS support.
Supporting Information Available: Experimental pro-
1
cedures, ESI-MS, HRMS, IR, H and 13C NMR for com-
pounds 8, 10, 16, 21, 22, 24, and 32, and related chiral HPLC
methods. This material is available free of charge via the
Internet at http://pubs.acs.org.
treatment of substrate 34 with 6% NaOH in MeOH, N-Boc
aminocoumarin 35 was isolated (86%) in one pot through
selective hydrolysis and intramolecular cyclization. Without
further purification, the crude product was deprotected to
give the target 8 as an HCl salt, which was collected as a
OL035101R
3236
Org. Lett., Vol. 5, No. 18, 2003