Biomimetic synthesis of gramicidin s and analogues by enzymatic cyclization of linear precursors on solid support.

Article abstract of DOI:10.1021/ol034437y

[reaction: see text] Gramicidin S is a potent decapeptide antibiotic with high hemolytic activity but is unlikely to provoke microbial resistance. Here we demonstrate that gramicidin thioesterase (GrsB TE) correctly cyclizes immobilized linear decapeptide precursors into head-to-tail products, indicating its suitability for parallel solid-phase synthesis of gramicidin analogues from linear precursors on solid support. This chemoenzymatic method will enable the optimization of the therapeutic index of the natural product to fight microbial resistance.

Full text of DOI:10.1021/ol034437y

ORGANIC
LETTERS
2
003
Vol. 5, No. 10
749-1752
Biomimetic Synthesis of Gramicidin S
and Analogues by Enzymatic Cyclization
of Linear Precursors on Solid Support
1
Xiaoming Wu, Xianzhang Bu, Ka Man Wong, Weili Yan, and Zhihong Guo*
Department of Chemistry and Biotechnology Research Institute,
The Hong Kong UniVersity of Science and Technology,
Clear Water Bay, Kowloon, Hong Kong SAR, China
chguo@ust.hk
Received March 12, 2003
ABSTRACT
Gramicidin S is a potent decapeptide antibiotic with high hemolytic activity but is unlikely to provoke microbial resistance. Here we demonstrate
that gramicidin thioesterase (GrsB TE) correctly cyclizes immobilized linear decapeptide precursors into head-to-tail products, indicating its
suitability for parallel solid-phase synthesis of gramicidin analogues from linear precursors on solid support. This chemoenzymatic method
will enable the optimization of the therapeutic index of the natural product to fight microbial resistance.
4
Widespread resistance of microbial pathogens to currently
available antibiotics has become a serious public health
threat. To contain the resistance, considerable interest has
results in disruption of the barrier functions and no resistance
5
to this antibiotic has been found so far. In addition,
development of resistance to such a compound is unlikely
1
been focused on the amphiphilic peptide antibiotics, including
2
gramicidin S (1, Scheme 1), which has hydrophobic side
Scheme 1. Dimerization and Cyclization of Pentapeptide
chains on one side of its rigid antiparallel â-pleated sheet
3
Precursor in Gramicidin S Biosynthesis
structure and hydrophilic side chains on the other side. This
natural product is an attractive target for new drug discovery
because it acts on cell membranes where its accumulation
(
1) (a) Neu, H. C. Science 1992, 257, 1064. (b) Cohen, M. L. Science
992, 257, 1050.
2) For examples, see: (a) Dathe, M.; Wieprecht, T.; Nikolenko, H.;
1
(
Handel, L.; Maloy, W. L.; MacDonald, K.; Beyermann, M.; Bienert, M.
FEBS Lett. 1997, 403, 208. (b) Oren, Z.; Hong, J.; Shai, Y. J. Biol. Chem.
1
997, 272, 14643. (c) Shai, Y.; Oren, Z. J. Biol. Chem. 1996, 271, 7305.
(d) Oren, Z.; Shai, Y. Biochemistry 1997, 36, 1826. (e) Dathe, M.; Shumann,
M.; Wieprecht, T.; Winkler, A.; Beyermann, M.; Krause, E.; Matsuzaki,
K.; Murase, O.; Bienert, M. Biochemistry 1996, 35, 12612. (f) Kondejewski,
L. H.; Jelokhani-Niaraki, M.; Farmer, S. W.; Bruce, L. Kay, C. M.; Sykes,
B. D.; Hancock, R. E. W.; Hodges, R. S. J. Biol. Chem. 1999, 274, 13181.
1
0.1021/ol034437y CCC: $25.00 © 2003 American Chemical Society
Published on Web 04/16/2003

because it requires significant alteration of the lipid composi-
tion. However, the effort to increase its therapeutic index,
namely, minimizing its high hemolytic activity while main-
taining its high antibiotic activity, has been hampered by
the limited number of accessible analogues of the natural
product through traditional synthetic methods.6
proved by the successful synthesis of tyrocidine A analogues
1
0
by TycC TE, an NRPS thioesterase homologous to GrsB
TE. Here we explore the activity of the gramicidin S
thioesterase toward linear precursor substrates on solid
support and test its suitability for solid-phase parallel
synthesis of the natural product analogues.
One way to circumvent this problem is to take advantage
of enzymes in gramicidin S biosynthesis for chemoenzymatic
analogue synthesis. The natural product is a cyclic decapep-
tide antibiotic composed of two repetitive pentapeptide units
The gramicidin S thioesterase gene was amplified from
B. breVis (ATCC 9999) genomic DNA and expressed in the
11
vector pFAB5c.His. The 34-kD thioesterase with a histidine
tag at the carboxy terminus was purified from an E. coli
host, using metal-chelating affinity chromatography and
anion-exchange column chromatography, in good yield (∼1.0
7
that is produced by Bacillus breVis. It is synthesized by a
thiol-template mechanism catalyzed by modular nonriboso-
8
-1
mal peptide synthases (NRPS) GrsA and GrsB. Of particular
mg L ) and high purity (>95% pure by SDS-PAGE). The
interest is the last functional domain of GrsB, which is
predicted to be a thioesterase (GrsB TE) responsible for
dimerizing the linear pentapeptide precursor generated by
the NRPS and then cyclizing the dimer head-to-tail to release
the product gramincidin S (Scheme 1).
expressed protein was tested active toward NAC-pentapep-
tide and pentapeptide dimer with a comparable kcat and K
as reported.
M
9
The fully deprotected biosynthetic decapeptide precursor
of gramicidin S (4a, Figure 1) was synthesized on TentaGel-
In its putative form, GrsB TE was found to retain the
dimerization and cyclization power toward the pentapep-
tide substrate (2) coupled to a N-acetylcysteaminyl (NAC)
group at carboxy terminus, a mimic of the native phospho-
pantetheinyl spacer linking the linear peptide and the peptidyl
carrier protein (PCP). However, the yields of the dimer and
the cyclic product were very low, whereas the hydrolysis of
9
the thioester dominated the enzyme-catalyzed reaction. The
abortive thioester hydrolysis was avoided when the linear
dimerized decapeptide precursor (3) was used directly as the
substrate for GrsB TE. In homogeneous solution, the enzyme
was further demonstrated to tolerate change in the linear
peptide sequence and be capable of cyclizing peptide
precursors with variable sizes. These investigations show the
potential of the thioester in parallel generation of the cyclic
peptide analogues for the natural product.
Figure 1. Structures of the linear precursors on TentaGel resin
for cyclization by GrsB TE.
The most efficient way to generate analogues of the natural
product is to use the thioesterase to cyclize an array of linear
decapeptide precursors through the combinatorial solid-phase
peptide synthesis (SPPS). Feasibility of this strategy has been
OH resin, using a modified Fmoc deprotection method for
12
solid-phase peptide synthesis. This precursor, similar to that
1
3
for tyrocidine A, was found to cyclize spontaneously into
head-to-tail product (1) in ammonia solution, providing a
convenient way to quantify the total amount of correct linear
decapeptide precursor.
(3) (a) Hodgkin, D. C.; Oughton, B. M. Biochem. J. 1957, 65, 752. (b)
Stern, A.; Gibbons, W. A.; Craig, L. C. Proc. Natl. Acad. Sci. U.S.A. 1968,
6
1
1, 734. (c) Ohnishi, M.; Urry, D. W. Biochim. Biophys. Res. Commun.
969, 36, 194.
(
4) Prenner, E. J.; Lewis, R. N. A. H.; McElhaney, R. N. Biochim.
Biophys. Acta 1999, 1462, 201.
To test the activity of the enzyme toward the solid-
supported substrate, freshly deprotected linear peptide on the
(
5) Hancock, R. E. W. Lancet 1997, 349, 418.
-
1
(6) Waki, M.; Izumiya, N. In Biochemistry of Peptide Antibiotics: Recent
resin (50 mg, loading 0.20 mmol g ) was extensively
washed with methanol and double-deionized water and
AdVances in the Biotechnology of â-Lactams and Microbial BioactiVe
Peptides; Kleinkaug, H., von D o¨ hren, H., Eds.; Walter de Gruyter: Berlin,
-
1
1
990, pp 205-244.
7) (a) Gause, G. G.; Brazhnikova, M. G. Nature 1944, 154, 703. (b)
immersed in 100 µL of the GrsB TE (100 µg mL ) in 25
mM 3-(N-morpholino)propanesulfonate (MOPS) buffer (pH
(
Izumiya, N.; Kato, T.; Aoyagi, H.; Waki, M.; Kondo M. In Synthetic Aspects
of Biologically ActiVe Cyclic Peptides-Gramicidin and Tyrocidines; Kodan-
sha Ltd. and Wiley: Tokyo, 1979.
7
.0) in a 1.5-mL microfuge tube. The suspension was
constantly swirled and incubated at 37 °C for 3 h. After the
incubation, the reaction was quenched by addition of 100
µL of 1.7% trifluoroacetic acid in H O. The resin was
2
separated from the supernatant and washed thrice with 1 mL
(
8) (a) Kr a¨ tzschmar, J.; Krause, M.; Marahiel, M. A. J. Bacteriol. 1989,
1
1
5
71, 5422. (b) Kleinkauf, H.; von D o¨ hren, H. Eur. J. Biochem. 1990, 192,
. (c) Turgay, K.; Krause, M.; Marahiel, M. A. Mol. Micriobiol. 1992, 6,
29. (d) Kleinkauf, H.; von D o¨ hren, H. Eur. J. Biochem. 1996, 236, 335.
(
2
9
e) von D o¨ hren, H.; Keller, U.; Vater, J.; Zocher, R. Chem. ReV. 1997, 97,
675. (f) Marahiel, M. A.; Stachelhaus, T.; Mootz, H. D. Chem. ReV. 1997,
7, 2661.
(10) Kohli, R. M.; Walsh, C. T.; Burkart, M. D. Nature 2002, 418, 658.
(11) Orum, H.; Andersen, P. S.; Oster, A.; Johansen, L. K.; Riise, E.;
Bjørnvad, M.; Svendsen, I.; Engberg, J. Nucleic Acids Res. 1993, 21, 4491.
(12) Bu, X.; Xie, G.; Law, C. W.; Guo, Z. Tetrahedron Lett. 2002, 43,
2419.
(9) (a) Trauger, J. W.; Kohli, R. M.; Mootz, H. D.; Marahiel, M. A.
Walsh, C. T. Nature 2000, 407, 215. (b) Trauger, J. W.; Kohli, R. M.;
Walsh, C. T. Biochemistry 2001, 40, 7092. (c) Kohli, R. M.; Trauger, J.
W.; Schwarzer, D.; Marahiel, M. A.; Walsh, C. T. Biochemistry 2001, 40,
7
099.
(13) Bu, X.; Wu, X.; Xie, G.; Guo, Z. Org. Lett. 2002, 4, 2893.
1750
Org. Lett., Vol. 5, No. 10, 2003

expected cyclic product, gramicidin S, for the enzyme
catalyzed reaction.
Finally, the enzyme treatment of the resin-bound precursor
(
4a) was scaled up, and the product at 32.4 min was isolated
1
by semipreparative HPLC and dried in vacuo. Its H NMR
spectrum was found to be completely consistent with that
of the wild-type gramicidin S. On the basis of these results,
the thioesterase GrsB TE indeed correctly cyclizes the
heterogeneous decapeptide precursor (4a) to form the cyclic
gramicidin S (1).
3b
From the HPLC and mass spectroscopic analyses, appar-
ently gramicidin S formation dominated the enzyme-
catalyzed cyclization reaction, with negligible amount of
hydrolytic product that should appear at about 22 min in the
HPLC elution chromatogram. This conclusion was further
confirmed by LC-MS analysis of the turnover product.
Figure 2. HPLC analysis of the turnover product of gramicidin
thioesterase-catalyzed cyclization of the linear decapeptide precursor
(
4a) on a solid support. (() TE ) gramicidin thioesterase, GrsB
TE; (*) gramicidine S.
However, it was also found that only a fraction of the
linear decapeptide precursor on the resin was converted into
gramicidin S product since treatment of the same amount of
resin in 7 M aqueous ammonia solution produced substan-
tially greater amount of the same product. By comparison
of their HPLC chromatograms and co-injection, it was found
that only 28% of the linear precursor was cyclized by GrsB
TE. Increasing the enzyme concentration (from 100 to 500
of methanol. The solution phase was combined and dried in
vacuo, and the turnover product was dissolved in 500 µL of
methanol. As a negative control, the same amount of
substrate-bearing resin was processed in parallel using pure
buffer instead of the enzyme solution to obtain a control
product.
As shown in Figure 2, HPLC analysis of the turnover
product showed there were two major peaks at retention time
R
(t ) 29.1 and 32.4 min (trace c). Both peaks are associated
with the enzyme treatment of the immobilized linear precur-
sor (4a, Figure 1) because they are absent in the chromato-
gram (trace a, Figure 2) resulting from the MOPS buffer
treatment. The peak at 29.1 min was found to be from the
thioesterase GrsB TE in comparison to the chromatogram
for the pure enzyme (trace b). The other peak at 32.4 min
was determined to be the expected product, gramicidin S
-
1
µg mL ) or the incubation time at 37 °C did not increase
the product yield. These results indicate that the majority of
the linear precursor is hidden in the resin, and only a portion
of it is exposed on the particle surface and accessible for
the enzyme GrsB TE. Although this result contradicts that
from a recent investigation in which TentaGel-supported
substrates were found to be inaccessible to enzymes,14 it is
otherwise consistent with the results of other investigations
of enzyme activity toward substrates on the same solid
1
5
support.
(1), because it showed elution time identical with that of
the authenticated natural product (trace d). This product
identification is further supported by the FAB-MS analysis
Seven other linear decapeptides (4b-4h, Figure 1) with
variations in the wide-type gramicidin S precursor were
synthesized to test GrsB TE activity toward immobilized
unnatural substrates. The enzyme conversion and the turnover
that showed only one molecular ion peak at 1141.9 ([M +
+
1
] ), consistent with the calculated mass of 1140.71 for the
Table 1. Product Characterization of the Cyclization of Linear Decapeptide Precursors 4a-4h catalyzed by GrsB TEa
overall
yieldb
ratio
(cyclic:linear)c
% precursor accessible
by GrsB TEd
precursors
X1
X2
X3
tR (min)
calcd mass
[M + H]+
4
4
4
4
4
4
4
4
a (wt)
DPhe
DPhe
DAla
DAla
DPhe
DPhe
DAla
DAla
Orn
Lys
Orn
Lys
Orn
Lys
Orn
Lys
Val
Val
Val
Val
Phe
Phe
Phe
Phe
32.4
30.0
27.1
26.8
32.4
32.2
29.3
29.0
1140.7
1154.7
1064.7
1078.7
1188.7
1202.7
1112.7
1126.7
1141.7
1155.6
1065.6
1079.4
1189.6
1203.5
1113.5
1127.6
5.5
9.2
13.4
8.8
8:1
10:1
9:1
5:1
4:1
9:1
8:1
10:1
28
28
18
9
29
35
39
37
b
c
d
e
f
7.1
16.8
18.9
9.2
g
h
a
Molecular ion is from mass spectroscopic analysis under FAB mode. Other results are derived from the HPLC analyses and purification performed with
-
1
Waters 600E system with a reverse-phase semipreparative XTerra RP18 column, 7 µm, 7.8 mm × 300 mm. Separation conditions were 3.0 mL min flow
rate, a linear gradient of 80% to 20% A in 25 min, 20% to 0% A in another 10 min, washed with 100% B for 10 min, and then calibrated at 80% A for 15
b
min. Solution A was 0.1% TFA in double-deionized H2O; solution B was 0.1% TFA in acetonitrile. Overall HPLC-isolated yield of the head-to-tail cyclic
c
products based on the original loading value of the resin after 7 M NH3‚H2O treatment. Ratio of the cyclic product peak area to that over 22-25 min on
the HPLC chromatograms where the linear hydrolytic products appear. d Percentage of the linear precursor cyclizable by the enzyme is calculated as the
-
1
ratio of cyclic product peak area on the HPLC chromatogram from GrsB TE (500 µg mL ), after subtraction of that from blank MOPS incubation, to that
from the 7 M NH3‚H2O treatment of identical amount of the precursor-bound resin.
Org. Lett., Vol. 5, No. 10, 2003
1751

products were characterized as summarized in Table 1. The
thioesterase was found to convert the analogous precursors
into cyclic products with high selectivity, similar to the
formation of gramicidin S from its biosynthetic precursor.
spontaneously is an interesting finding that may lead to an
even simpler synthetic method for gramicidin analogues.
In summary, we have demonstrated that the putative
gramicidin thioesterase (GrsB TE) is active toward hetero-
geneous substrate analogues immobilized on solid support
through a poly(ethylene glycol) linker. Considering the high
1
HPLC purification and subsequent H NMR spectroscopic
analyses showed that the resulting cyclic products were
formed through the head-to-tail cyclization of the corre-
sponding precursors.
9
substrate flexibility of the thioesterase, this result indicates
that the enzyme is feasible for generation of diverse cyclic
peptides by cyclizing the linear precursors from combina-
torial solid-phase peptide synthesis. In turn, this chemoen-
zymatic method will provide convenient access to large
amount of gramicidin analogues for lead discovery against
microbial resistance.
Interestingly, these analogous precursors, like the wide-
type one, underwent spontaneous head-to-tail cyclization to
form corresponding gramicidin S analogues in ammonia
solution, enabling quantification of the total amount of the
linear precursors on the resin. Again, it was found that only
a small portion of the linear precursors were accessible to
the enzyme, regardless of the enzyme concentration and
reaction time, leading to lower product yields. Despite this
shortfall, these results nevertheless prove that the thioesterase
GrsB TE is active toward heterogeneous native substrate and
its analogues. In addition, the low availability of linear
precursors to the enzyme can be ameliorated by using the
more accessible solid supports.14 Furthermore, the observa-
tion that the linear precursors can selectively cyclize
Acknowledgment. This work was supported in part by
the Innovation and Technology Fund (ITS/119/00) and the
RGC (DAG01/02.SC09) from the Government of the Hong
Kong Special Administrative Region. We also thank the
Department of Chemistry at the HKUST for providing
postgraduate studentships to X.W. and K.M.W.
Supporting Information Available: Experimental pro-
cedures, LC-ESIMS spectrum of 4a enzyme cyclization
1
(
14) Kress, J.; Zanaletti, R.; Amour, A.; Ladlow, M.; Frey, J. G.; Bradley,
M. Chem. Eur. J. 2002, 8, 3769.
15) (a) Vagner, J.; Barany, G.; Lam, K. S.; Krchnak, V.; Sepetov, N.
F.; Ostrem, J. A.; Strop, P.; Lebl, M. Proc. Natl. Acad. Sci. U.S.A. 1996,
product, and H NMR spectra of cyclization products of 4a-
4
h. This material is available free of charge via the Internet
(
at http://pubs.acs.org.
9
3, 8194. (b) Reents, R.; Jeyaraj, D. A.; Waldmann H. AdV. Synth. Catal.
2
001, 343, 501.
OL034437Y
1752
Org. Lett., Vol. 5, No. 10, 2003