Dynamic cyclic thiodepsipeptide libraries from thiol-thioester exchange

Article abstract of DOI:10.1021/ol1004752

Thiol-thioester exchange was found to readily generate libraries of cyclic thiodepsipeptides under thermodynamic control, which will enable their use in a variety of dynamic combinatorial chemistry assays. The kinetic determinants of macrocycle formation and the role of amino acid structure on the reaction dynamics are discussed.

Full text of DOI:10.1021/ol1004752

ORGANIC
LETTERS
2010
Vol. 12, No. 8
1860-1863
Dynamic Cyclic Thiodepsipeptide
Libraries From Thiol-Thioester Exchange
Soumyadip Ghosh, Lindsey A. Ingerman, Aaron G. Frye, Stephen J. Lee,†
Michel R. Gagne´,* and Marcey L. Waters*
Department of Chemistry, CB 3290, UniVersity of North Carolina, Chapel Hill, North
Carolina 27599, and U.S. Army Research Office, P.O. Box 12211, Research Triangle
Park, North Carolina 27709
mgagne@unc.edu; mlwaters@unc.edu
Received February 25, 2010
ABSTRACT
Thiol-thioester exchange was found to readily generate libraries of cyclic thiodepsipeptides under thermodynamic control, which will enable
their use in a variety of dynamic combinatorial chemistry assays. The kinetic determinants of macrocycle formation and the role of amino acid
structure on the reaction dynamics are discussed.
Cyclic peptides have been described as “privileged
structures” for drug design because so many natural and
synthetic cyclic peptides exhibit biological acitivity.1-3
For example, cyclosporin is a cyclic nonribosomal peptide
that is widely used as an immunosuppressant, revolutionizing
organ transplants.1 As another example, both natural and
synthetic cyclic peptides have recently been identified as
HDAC inhibitors, which hold promise for treatment of
cancer.4 As drug scaffolds, cyclic peptides are advantageous
because they mimic native protein structure, exhibit enhanced
metabolic stability, and are structurally preorganized, which
reduces the entropic cost of binding.
Despite their therapeutic potential, the options for syn-
thesizing structurally diverse libraries of cyclic peptides in
a high-throughput format are limiting.3 For example, phage
display of disulfide-linked cyclic peptides has been used to
screen large libraries of potential inhibitors,5 but disulfides
are not stable within the cellular environment and there is
no straightforward synthetic replacement for a disulfide
linkage. Split-and-pool synthesis is another popular method
of generating libraries of cyclic peptides but requires
orthogonal protecting group strategies of on-bead cyclization
and can result in mixtures of cyclic and acyclic peptides.3
Herein, we describe a strategy to use thioester exchange
for generating solution-phase cyclic thiodepsipeptide libraries
via dynamic combinatorial chemistry (DCC).6 DCC is a
method for the in situ generation of a complex mixture of
macrocycles from smaller building blocks via reversible
covalent bond formation. In DCC, the members of an
equilibrating library compete for an added template and
† U.S. Army Research Office.
(1) Craik, D. J. Science 2006, 311, 1563–1564
(2) Horswill, A. R.; Benkovic, S. J. Cell Cycle 2005, 4, 552–555
(3) Horton, D. A.; Bourne, G. T.; Smythe, M. L. J. Comput. Aided Mol.
Des. 2002, 16, 415–430
.
.
.
(4) (a) Minucci, S.; Pelicci, P. G. Nat. ReV. Cancer 2006, 6, 38–51. (b)
Meinke, P. T.; Liberator, P. Curr. Med. Chem. 2001, 8, 211–235. (c) Shute,
R. E.; Kawai, M.; Rich, D. H. Tetrahedron 1988, 44, 685–95. (d) Montero,
A.; Beierle, J. M.; Olsen, C. A.; Ghadiri, M. R. J. Am. Chem. Soc. 2009,
131, 3033–3041. (e) Olsen, C. A.; Ghadiri, M. R. J. Med. Chem. 2009, 52,
7836–7846.
(5) Sidhu, S. S.; Fairbrother, W. J.; Deshayes, K. ChemBioChem 2003,
4, 14–25.
(6) Corbett, P. T.; Leclaire, J.; Vial, L.; West, K. R.; Wietor, J. L.;
Sanders, J. K.; Otto, S. Chem. ReV. 2006, 106, 3652–3711.
10.1021/ol1004752  2010 American Chemical Society
Published on Web 03/23/2010

Table 1. Monomers, Half-Life of Reactions, And Conversions
of 1-15 (100 mM NH4Ac Buffer, pH 6.75, 25 °C)
compd
sequence
t1/2
product
1
2
3
4
5
6
7
8
Ac-ECHypG-COSR′ ∼12 min dimer
Ac-ECPH-COSR′
∼12 min dimer, trimer
Ac-ECHypF-COSR′ ∼1.5 h
dimer
dimer
Ac-ECPF-COSR′
∼1.5 h
Ac-ECHypQ-COSR′ ∼1.5 h
Ac-ECHypE-COSR′ ∼5.5 h
Ac-ECHypV-COSR′ NDa
dimer, trimer, tetramer
dimer
oligodimer
dimer
Ac-ECWE-COSR′
∼3.5 h
9
Ac-EC(D-P)Q-COSR′ ∼10 min dimer, trimer
10
11
12
13
14
15
Ac-KCPR-COSR′
Ac-RCPK-COSR′
Ac-KCPK-COSR′
Ac-KCPQ-COSR′
Ac-KCWR-COSR′
Ac-KCPV-COSR′
∼15 min dimer
∼15 min dimer
∼15 min dimer
∼20 min dimer, trimer
∼15 min dimer
∼10 h
dimer, trimer
Figure 1. (a) General design of monomers and major products. (b)
Sequence of steps toward dimeric cyclic thiodepsipeptide.
a Not determined. Monomers were hydrolyzed before complete reaction.
respond to amplify the best binder if one exists. Recent
reports suggest that thiol-thioester exchange is a promising
reversible reaction for DCC since it is rapid in aqueous
solution at neutral pH7,8 and provides a native-like linkage
that is subsequently replaceable by more robust amide or
ester functionalities. However, all reported cases of thioester
exchange in DCC used thioesters that were unsubstituted at
the R-position, which would significantly limit the diversity
of structures possible in a library of cyclic peptides. Thus,
we have investigated the reactivity of peptidic thiol-thioester
exchange to determine the scope and limitations for its
application to DCC.
Figure 1a shows the general structure of the monomers
used in this study. Each monomer was a four-residue peptide
containing a thioester at the C-terminus and a thiol at AA2
(Cys). Structural variations were then made in AA1, AA3,
and AA4 (Table 1); by design, AA2-AA3-AA4 forms the
macrocycle, and AA1 remains exocyclic. Charged amino
acids (Lys, Arg, Glu) were incorporated into position AA1
to enhance water solubility and prevent aggregation. Proline
(or hydroxyproline, Hyp, to improve solubility) was initially
included in AA3 as a turn residue that favors macrocycles.9
Various amino acids were incorporated into AA4 to study
their effect on macrocyclization, including positively charged
amino acids (Lys, Arg), a negatively charged amino acid
(Glu), a hydrogen-bonding and neutral amino acid (Gln), and
hydrophobic and sterically bulky amino acids (Phe, Val).
The monomers were synthesized using a modification of
the method reported by Gellman and co-workers (Supporting
Information),8 which provided monomers 1-15 in 70-80%
Figure 2. Overlay of HPLC traces showing the consumption of 5
and the appearance of macrocycles over time.
yield with no observable epimerization. Upon dissolving
1-15 in potassium phosphate buffer (pH 7), the monomers
underwent thiol-thioester exchange to form a mixture of
macrocycles, with the dimeric 20-atom macrocyclic hexapep-
tides as the major product (Figure 1a). Based on HPLC
analysis of the reaction mixture over time (Figure 2), we
propose the simple two-step mechanism shown in Figure 1b.
First, two monomers react to form an oligodimer followed
by the ring closing intramolecular trans thioesterification
reaction. The oligodimer intermediate was not observed to
accumulate, indicating that the intermolecular step is rate
limiting.10
Not surprisingly, the rate of macrocycle formation was
dependent on the amino acid sequence. Since the first step
in Figure 1b is rate determining, the overall reaction rate
was determined by monitoring the disappearance of monomer
over time, which in turn was measured by manual integration
(7) (a) Leclaire, J.; Vial, L.; Otto, S.; Sanders, J. K. M. Chem. Commun.
2005, 1959–1961. (b) Larsson, R.; Pei, Z.; Ramstro¨m, O. Angew. Chem.,
Int. Ed. 2004, 43, 3716–3718. (c) Ura, Y.; Beierle, J. M.; Leman, L. J.;
Orgel, L. E.; Ghadiri, M. R. Science 2009, 325, 73–77
.
(8) Woll, M. G.; Gellman, S. H. J. Am. Chem. Soc. 2004, 126, 11172–
11174
.
(9) Cousins, G. R. L.; Poulsen, S.-A.; Sanders, J. K. M. Chem. Commun.
1999, 1575–1576.
(10) The macrocyclic thiodepsipeptides were found to hydrolyze over
extended periods, but not before reaching equilibrium.
Org. Lett., Vol. 12, No. 8, 2010
1861

Figure 4. Comparison of the disappearance of monomers 7 and
15 over time (100 mM NH4Ac buffer, pH 6.75, 25 °C).
Figure 3. Plot of the disappearance of monomers 1-3 and 5-7
over time (100 mM NH4Ac buffer, pH 6.75, 25 °C).
the OH group on hydroxyproline does not affect the rate of
reaction. We also investigated the replacement of Hyp with
the more flexible Trp (monomer 8) to determine if a turn is
necessary for rapid cyclization. In fact, we found that
monomer 8 reacts faster than the Hyp analogue. Stereo-
chemistry was also a factor in the reaction rates as 9 (D-Pro)
reacted considerably faster than 5 (Hyp). This result does
not reflect conformational preferences for cyclization but
instead an influence of chirality on the accessibility of the
thioester or thiol.
of the area under the monomer peak in the HPLC-UV trace.
Since several cases were too fast to measure at pH 7, the
reactions were followed at pH 6.75 for comparison. Impor-
tantly, with the exception of monomers containing Val at
the C-terminus, equilibrium was always reached prior to the
detection of hydrolysis, and no acylation of Lys was
observed.
Initially, the C-terminal amino acid (AA4) was varied while
holding the rest of the molecule constant (Table 1, com-
pounds 1-7). The observed reactivity trend, 1 ∼ 2 > 3 ∼ 4
∼ 5 > 6 > 7, corresponds generally to differences in steric
effects at the C-terminal amino acid, but in some cases, the
differences in rate are more subtle (Figure 3, Table 1). This
trend is generally consistent with the reactivity trends
observed in the native chemical ligation of peptides and
sugar-assisted ligation of glycopeptides, which initiate with
the same first step.11,12 We find that His reacts as rapidly as
Gly (Table 1, 1 vs 2), despite its larger size, as was observed
by Dawson,11 suggesting that the imidazole group of the
C-terminal His stabilizes the transition state of the trans
thioesterification reaction. In contrast, ꢀ-branched amino
acids such as Val react significantly more slowly (Table 1,
entry 7, and Figure 3), resulting in measurable hydrolysis
before cyclization was complete. Monomer 6, which contains
Glu at position 4, reacts more slowly than its Gln analogue
and forms multiple mass degenerate isomers with different
retention times in the HPLC/MS. On the basis of the mass
spectrometric analysis of individual peaks and literature
precedent, it appears that the γ-carboxyl cyclizes onto the
thioester to form anhydride and ultimately 6-isom (see the
Supporting Information).13
We also investigated the effect of positively charged amino
acids (Lys and Arg) at positions AA1 and AA4. Interestingly,
monomers 10-15, which contain Lys or Arg at AA1 and/or
AA4, react significantly faster than their negatively charged
analogues (Supporting Information). This rate enhancement
suggests that the positively charged Arg and Lys residues
stabilize the buildup of negative charge in the transition state
through hydrogen bonding and/or electrostatic interactions.
To further explore the influence of positively charged amino
acids on reactivity, we replaced the Glu at AA1 of monomer
7 with Lys to give monomer 15. Although hydrolysis was
still observed before a complete cyclization, 15 reacts almost
five times faster than 7 (Figure 4).
When we analyzed binary mixtures of monomers 1-7 and
10-15 by HPLC, we observed the formation of a mixture
of macrocyclic thiodepsipeptides with homo- and het-
erodimers as the major products. To confirm thermodynamic
control over the library speciation, experiments were per-
formed adding 1 to the preformed homodimer 32 and 3 to
the preformed homodimer 12. Nearly superimposable HPLC
traces were obtained after 18 h at pH 6.75, indicating that
the equilibrium position was reached from both directions.
Figure 5 demonstrates that high chemical diversity is
achievable upon simple mixing of monomers, in this case
1 h after mixing 1, 3, and 5-7. It is worth noting that 7 was
well incorporated in the library despite its slow reaction with
itself. In addition to the cyclodimers, cyclo-tripeptides were
present. The library speciation was readily deconvoluted
using LCMS, which showed extensive cross-reactivity of
monomers and a general lack of self-sorting.14
We varied several other positions to further explore
reactivity. Replacement of Hyp with Pro resulted in identical
HPLC traces and rate profiles for 3 versus 4, indicating that
(11) Hackeng, T. M.; Griffin, J. H.; Dawson, P. E. Proc. Natl. Acad.
Sci. U.S.A. 1999, 96, 10068–10073
(12) Brik, A.; Yang, Y. Y.; Ficht, S.; Wong, C. H. J. Am. Chem. Soc.
2006, 128, 5626–5627
(13) Villain, M.; Gaertner, H.; Botti, P. Eur. J. Org. Chem. 2003, 3267,
3272.
.
.
In summary, we have demonstrated an efficient methodol-
ogy to rapidly generate a complex library of macrocyclic
1862
Org. Lett., Vol. 12, No. 8, 2010

influence through hydrogen-bonding and/or electrostatic
interactions on the tetrahedral intermediate. Interestingly,
positively charged amino acids were also found to increase
reaction rates when incorporated at the N-terminus, suggest-
ing that the transition state can be stabilized when flanked
by a positively charged amino acid on either side. In all cases,
the dimer is the major product, but the degree of formation
of larger macrocycles is dependent on the peptide sequence.
Upon mixing two or more monomers, we have shown that
a complex library of cyclic thiodepsipeptides can be gener-
ated in situ under thermodynamic control.
This study demonstrates the feasibility of this methodology
for generating a large number of cyclic thiodepsipeptides
that can be screened against a particular target efficiently
within hours. Since thioesters have transient stability in vivo,
a “hit” in the high-throughput screening can be resynthesized
as a more stable analogue by replacing of the thioester group
with an amide or ester. Moreover, upon mixing different
monomers containing Cys located at different positions,
macrocycles of different sizes can in theory be generated,
such that the outcome is not limited to only cyclic hexapep-
tides. In the future, we aim to apply this high-throughput
screening method for inhibitor development, disruption of
protein-protein interactions, and binding of nucleic acids.
Figure 5. HPLC trace of a mixture of 1, 3, and 5-7 after
equilibration for 1 h (100 mM NH4Ac buffer, pH 6.75, rt).
Underlined species indicate linear oligomers.
thiodepsipeptides at neutral pH for high-throughput screen-
ing. Structure-function studies indicate that the thiol-
thioester exchange is tolerant of a variety of amino acids at
the C-terminal position, with the exception of ꢀ-branched
amino acids such as Val, which retard the rate of ho-
modimerization, and carboxylic acid side chains, which can
isomerize through anhydride intermediates. Positively charged
amino acids and His at the C-terminus were found to enhance
the rate of thiol-thioester exchange, suggesting a stabilizing
Acknowledgment. We thank the Defense Threat Reduc-
tion Agency (DTRA, W911NF-06-1-0169) for support.
Supporting Information Available: Synthetic procedure,
characterization data, and HPLC traces. This material is
available free of charge via the Internet at http://pubs.acs.org.
(14) (a) Rowan, S. J.; Hamilton, D. G.; Brady, P. A.; Sanders, J. K. M.
J. Am. Chem. Soc. 1997, 119, 2578–2579. (b) Rowan, S. J.; Reynolds, D. J.;
Sanders, J. K. M. J. Org. Chem. 1999, 64, 5804–5814. (c) Nitschke, J. R.
Acc. Chem. Res. 2007, 40, 103–112.
OL1004752
Org. Lett., Vol. 12, No. 8, 2010
1863