Large cyclic peptides as cores of multivalent ligands: Application to inhibitors of receptor binding by cholera toxin

Article abstract of DOI:10.1021/jo0489770

Large cyclic decapeptides (up to 50-atom ring) were synthesized efficiently on the solid phase with allylester protection of the carboxyl terminus during elongation. Pentavalent ligands, in a "core-linker-finger" modular setup, were assembled by using these cyclic peptide cores to demonstrate large affinity gains for inhibition of surface receptor binding by the cholera toxin B pentamer. The results suggest that the peptide cores retain expanded conformation in solution so that shorter flexible linkers are needed for larger peptide cores to achieve the best inhibitory results.

Full text of DOI:10.1021/jo0489770

structure-based design approaches, it appears that op-
timal affinity is obtained when the size and geometry of
the resulting ligands match to a large extent the binding
site distribution on the target protein. Therefore, ma-
nipulation of the core of the multivalent ligand can have
significant impact on ligand affinity. So far, in most cases
of structure-based design of multivalent ligands, rela-
tively small cores are used with long linkers, with the
exception of a cyclodextrin-bridged bivalent inhibitor of
tryptase13 and three large peptide cores reported by Ohta
et al.18 Here, we report our investigation on whether
large-ring, head-to-tail cyclic peptides can adopt an
expanded conformation and be effectively used as the core
of multivalent ligands to achieve significant affinity
gains.
La r ge Cyclic P ep tid es a s Cor es of
Mu ltiva len t Liga n d s: Ap p lica tion to
In h ibitor s of Recep tor Bin d in g by
Ch oler a Toxin
Zhongsheng Zhang,† J iyun Liu,†,‡
Christophe L. M. J . Verlinde, Wim G. J . Hol,§ and
Erkang Fan*
Biomolecular Structure Center, Department of Biochemistry,
Department of Chemistry, and Howard Hughes Medical
Institute, University of Washington,
Seattle, Washington 98195
erkang@u.washington.edu
Received J une 17, 2004
Head-to-tail cyclic peptides may be excellent cores for
the assembly of multivalent ligands, due to the ease of
incorporating amino acids with various side chains for
attachment of linkers and monovalent ligands. Manipu-
lation of the length and nature of the amino acid
sequence of the cyclic peptide provides opportunities to
design multivalent ligands that are suitable for different
geometric requirements. However, the success of such a
cyclic peptide approach will require that large-ring cyclic
peptides can retain a predictable, expanded ring confor-
mation in solution to support multivalent ligands. Al-
though cyclic octapeptides (24-atom ring) were found to
retain an expanded conformation in the solid state19 and
larger ones (30- to 63-atom ring) were used as scaffolds
for multivalent ligands,18,20 there is no guarantee for
larger cyclic peptides that the expanded conformation will
be predominant in solution. In fact, Ohta and co-workers
found that the R-amino acid sequence of a large-ring
cyclic peptide has a marked influence on the peptide
conformation in solution.18 To gain access to large-ring
cyclic peptides with a predictable and an expanded
conformation, we designed a series of cyclic decapeptides
(with 30-, 40-, and 50-atom rings) as the core for pen-
tavalent ligands inhibiting surface receptor binding by
cholera toxin (CT). The multivalent inhibitors of this
series were compared with inhibitors obtained in our
previous work that used the smaller pentacyclen core
1.10,21 As shown in Figure 1, three cyclic decapeptides
with alternating L-lysine and either glycine (core 2),
γ-aminobutyric acid (Abu) (core 3), or ꢀ-aminohexanoic
acid (Ahx) (core 4) were considered. The key in our design
is that we use flexible amino acids without side chains
(glycine, Abu, and Ahx) both for achieving the desired
ring size variations and for increased likelihood that the
peptides might adopt expanded conformations in solution.
Before the synthesis, we performed computer simula-
tions to investigate the conformational freedom of the
Abstr a ct: Large cyclic decapeptides (up to 50-atom ring)
were synthesized efficiently on the solid phase with allyl-
ester protection of the carboxyl terminus during elongation.
Pentavalent ligands, in a “core-linker-finger” modular setup,
were assembled by using these cyclic peptide cores to
demonstrate large affinity gains for inhibition of surface
receptor binding by the cholera toxin B pentamer. The
results suggest that the peptide cores retain expanded
conformation in solution so that shorter flexible linkers are
needed for larger peptide cores to achieve the best inhibitory
results.
Multivalent design is an attractive strategy for obtain-
ing high-affinity protein ligands.1-7 In recent years,
significant progress has been made in using the informa-
tion of the three-dimensional structure of the target
protein to guide multivalent ligand design.8-18 In such
* To whom correspondence should be addressed. Phone: 206-685-
7048. Fax: 206-685-7002.
† These authors contributed equally to this work.
‡ Department of Chemistry.
§ Howard Hughes Medical Institute.
(1) Mammen, M.; Choi, S. K.; Whitesides, G. M. Angew. Chem., Int.
Ed. 1998, 37, 2755-2794.
(2) Kiessling, L. L.; Strong, L. E.; Gestwicki, J . E. Annu. Rep. Med.
Chem. 2000, 35, 321-330.
(3) Lee, R. T.; Lee, Y. C. Glycoconj. J . 2000, 17, 543-551.
(4) Wright, D.; Usher, L. Curr. Org. Chem. 2001, 5, 1107-1131.
(5) Fan, E.; Merritt, E. A. Curr. Drug Targets: Infect. Disord. 2002,
2, 161-167.
(6) Lindhorst, T. K. Host-Guest Chem. 2002, 218, 201-235.
(7) Roy, R. Trends Glycosci. Glycotechnol. 2003, 15, 291-310.
(8) Kramer, R. H.; Karpen, J . W. Nature 1998, 395, 710-713.
(9) Loidl, G.; Groll, M.; Musiol, H. J .; Huber, R.; Moroder, L. Proc.
Natl. Acad. Sci. U.S.A. 1999, 96, 5418-5422.
(10) Fan, E.; Zhang, Z.; Minke, W. E.; Hou, Z.; Verlinde, C. L. M.
J .; Hol, W. G. J . J . Am. Chem. Soc. 2000, 122, 2663-2664.
(11) Kitov, P. I.; Sadowska, J . M.; Mulvey, G.; Armstrong, G. D.;
Ling, H.; Pannu, N. S.; Read, R. J .; Bundle, D. R. Nature 2000, 403,
669-672.
(12) Loidl, G.; Musiol, H. J .; Groll, M.; Huber, R.; Moroder, L. J .
Peptide Sci. 2000, 6, 36-46.
(13) Schaschke, N.; Matschiner, G.; Zettl, F.; Marquardt, U.; Bergn-
er, A.; Bode, W.; Sommerhoff, C. P.; Moroder, L. Chem. Biol. 2001, 8,
313-327.
(17) Kitov, P. I.; Shimizu, H.; Homans, S. W.; Bundle, D. R. J . Am.
Chem. Soc. 2003, 125, 3284-3294.
(18) Ohta, T.; Miura, N.; Fujitani, N.; Nakajima, F.; Niikura, K.;
Sadamoto, R.; Guo, C.-T.; Suzuki, T.; Suzuki, Y.; Monde, K.; Nishimura,
S.-I. Angew. Chem., Int. Ed. 2003, 42, 5186-5189.
(19) Bong, D. T.; Clark, T. D.; Granja, J . R.; Ghadiri, M. R. Angew.
Chem., Int. Ed. 2001, 40, 988-1011.
(14) Merritt, E. A.; Zhang, Z.; Pickens, J . C.; Ahn, M.; Hol, W. G. J .;
Fan, E. J . Am. Chem. Soc. 2002, 124, 8818-8824.
(15) Zhang, Z.; Merritt, E. A.; Ahn, M.; Roach, C.; Hou, Z.; Verlinde,
C.; Hol, W. G. J .; Fan, E. J . Am. Chem. Soc. 2002, 124, 12991-12998.
(16) Kitov, P. I.; Bundle, D. R. J . Am. Chem. Soc. 2003, 125, 16271-
16284.
(20) Renaudet, O.; Dumy, P. Org. Lett. 2003, 5, 243-246.
(21) Zhang, Z.; Pickens, J . C.; Hol, W. G. J .; Fan, E. Org. Lett. 2004,
6, 1377-1380.
10.1021/jo0489770 CCC: $27.50 © 2004 American Chemical Society
Published on Web 09/25/2004
J . Org. Chem. 2004, 69, 7737-7740
7737

SCHEME 1. Syn th esis of Hea d -to-Ta il Cyclic
Deca p ep tid es
F IGURE 1. Size comparison of the original pentacyclen-based
core 1 with cyclic decapeptides cores 2-4. The distances
indicated by the black bars are the calculated (see main text),
averaged distances between nonadjacent nitrogen atoms of
pentacyclen, or in the case of peptides, between the R-carbon
atoms of nonadjacent lysines. The unit-less number after the
distance is an indicator for the symmetry of each core (see
main text), for which a value “1” corresponds to an isotropic
ring conformation.
designed peptides. Using the software package QXP,22
2000 rounds of torsional Metropolis Monte Carlo at 1000
K were performed for each cyclic peptide core and the
pentacyclen core. Then the 300 conformations with the
lowest energy for each core were analyzed. In the
simulation, the cyclic peptides retained mostly an ex-
panded ring conformation without major twisting or
collapsing of the ring. The averaged ring size for each
peptide core, defined as the average (Boltzmann-weighted
mean) of all pairs of distances between the R-carbon
atoms of nonadjacent lysines in the 300 lowest energy
conformations, ranged from 10 to 16 Å, as summarized
in Figure 1. These sizes are significantly larger than the
pentacyclen-based core 1, where the averaged distance
between nonadjacent nitrogen atoms was about 6 Å. To
assess how well each core conforms to an expanded ring
conformation, we calculated the ratio of the minimum
and the maximum distances between the R-carbon atoms
of nonadjacent lysines (or between nonadjacent N atoms
in the pentacyclen core) in each conformation, and then
averaged the ratio with Boltzmann-weighting for all 300
lowest energy conformations for each core. A value of 1
for this ratio indicates that the core is isotropic in the
expanded ring conformation. As shown in Figure 1, the
peptide-based cores do not deviate very far from an
isotropically expanded ring, as the ratio varied from 0.73
(core 2) to 0.89 (core 4). The smaller pentacyclen core had
a ratio of 0.88. Not surprisingly, core 2, which incorpo-
rates 5 glycines, is the least isotropic on average.
Computer simulations also revealed that, on average,
the lysine side chains on the peptide cores have about
an equal chance to be on either side of the peptide ring,
and that those side chains mostly point outward. Because
we are using long and flexible linkers,10,15,23 whether the
lysine side chains are on the same side of the peptide
core or not should not have a major impact on the ligand’s
affinity. The flexible linkers are expected to easily
compensate for this conformational change in the peptide
cores.
Synthesis of the cyclic peptide cores is shown in
Scheme 1. Following a strategy developed for synthesiz-
ing head-to-tail cyclic peptides using Fmoc chemistry,24
we used the allyl group as the orthogonal protection for
the C-terminus. First, allyl-protected amino acids25
5
were acylated with Fmoc-Lys to obtain the corresponding
dipeptides 6. After removal of the Boc protection of the
lysine side chain, the dipeptides were attached to Wang
resin with nitrophenyl carbonate activated resin. The
resin-supported dipeptides 7 were then subjected to
standard peptide synthesis for linear peptide elongation,
using the appropriate N-R-Fmoc protected glycine, Abu,
or Ahx, and lysine. Once the resin-supported linear
decapeptides 8 were made, the allyl protection was
removed by using the condition developed by Teixido et
al.,26 followed by Fmoc removal. On bead cyclization was
achieved under standard peptide coupling condition with
(23) Zhang, Z.; Fan, E. Methods Enzymol. 2003, 362, 209-218.
(24) Alsina, J .; Rabanal, F.; Giralt, E.; Albericio, F. Tetrahedron Lett.
1994, 35, 9633-9636.
(22) McMartin, C.; Bohacek, R. S. J . Comput.-Aided Mol. Des. 1997,
11, 333-344.
(25) Waldmann, H.; Kunz, H. Liebigs Ann. Chem. 1983, 1712-1725.
7738 J . Org. Chem., Vol. 69, No. 22, 2004

SCHEME 2. Syn th esis of P en ta va len t Liga n d s on
Cyclic P ep tid e Cor es
F IGURE 2. Plot of IC50 of pentavalent ligands versus the
number of linker units in the ligand. The error bars represent
the standard deviation of two independent measurements of
IC50 values, which generally agree within 20%.
finger” strategy.10,23 Successive coupling of squarate
activated linker unit 1010,23 to the peptide cores produced
a series of core-linker assemblies 12-14 with various
linker lengths. Final assembly of the full ligands by
coupling of the squarate activated monovalent ligand
(“finger”) 1110,23 gave a set of 12 pentavalent ligands 15-
17, with linkers ranging from one to four units of 10.
Excellent isolated yields were obtained after HPLC
purification for all intermediates (12-14) and the final
ligands (15-17), as the yields typically were >70% (see
the Experimental Section and the Supporting Informa-
tion) for each squarate mediated coupling step.
These peptide-core-based pentavalent ligands were
assayed for their ability to block CT B pentamer binding
to ganglioside coated plates.21,29 The IC50 values for
ligands 15-17 were plotted against the number of linker
units in the ligand, as shown in Figure 2. For comparison,
IC50 data of a series of analogous pentavalent ligands 1821
based on a pentacyclen core (Scheme 2) are also included.
It is very exciting to observe that first, compared to
pentacyclen-based ligand series 18, shorter linkers are
needed to achieve optimal inhibition in cyclic peptide-
based pentavalent ligands. For pentacyclen-supported
ligands 18 with a core of about 6 Å, we previously
reported that the optimal inhibition is achieved with four
linker units.21 With core 2-based ligands 15, for which
the core size is about 4 Å larger than the pentacyclen
core, only three linker units are needed for optimal
inhibition. Another increase in core size by 4 Å, in the
case of ligand series 16 compared to 15, again shows one
linker less is needed for optimal inhibition of receptor
binding by the CT B pentamer. For the biggest core
4-based ligands 17 where the core size is 2.5 Å larger
than ligand series 16, it appears that the optimal linkage
is between 1 and 2 linker units. Second, in all series of
pentavalent compounds, ligands with longer or shorter
linkers than optimal all exhibit a loss in inhibitory power.
This is consistent with previous observations8,21 that
when a ligand’s effective dimension is not matching that
of its target, there is a reduction in the ligand’s affinity.
These experimental data firmly support the results from
PyBop. Final cleavage and HPLC purification gave the
desired cyclic decapeptides 2-4 in satisfactory yields
(24-30%). Although there have been reports of lower
yields of formed cyclic peptide for large ring systems with
use of the allyl protection strategy,27,28 this does not
appear to be a problem in our hands because the yields
of 2-4 compared favorably with the yields of cyclic
peptides (48- to 87-atom rings) prepared with alternative
strategies.27,28
As shown in Scheme 2, assembly of the full pentavalent
ligands followed our well-developed modular “core-linker-
(26) Teixido, M.; Altamura, M.; Quartara, L.; Giolitti, A.; Maggi, C.
A.; Giralt, E.; Albericio, F. J . Comb. Chem. 2003, 5, 760-768.
(27) Valero, M. L.; Giralt, E.; Andreu, D. J . Peptide Res. 1999, 53,
56-67.
(28) Cudic, M.; Wade, J . D.; Otvos, J .; Laszlo Tetrahedron Lett. 2000,
41, 4527-4531.
(29) Minke, W. E.; Roach, C.; Hol, W. G. J .; Verlinde, C. Biochemistry
1999, 38, 5684-5692.
J . Org. Chem, Vol. 69, No. 22, 2004 7739

(5 equiv), and DIPEA (10 equiv) in DMA for 1 h at room
temperature. The resin was filtered and washed successively
with DMA (5 × 5 mL) and CH2Cl2 (5 × 5 mL). The resin was
capped with acetic anhydride (0.4 mL, 4.2 mmol) and DIPEA
(1.3 mL, 8.43 mmol) in anhydrous CH2Cl2 for 40 min after each
elongation step. After the resin-supported linear decapeptide 8a
was obtained, the allyl group was removed by Pd(PPh3)4 (0.1
equiv) and PhSiH3 (10 equiv) in anhydrous CH2Cl2 (8 mL) with
nitrogen bubbling for 15 min.26 This treatment was repeated
three more times. The last Fmoc group was removed by 2% DBU/
2% piperidine/DMA (3 × 10 min). After the resin was washed
with DMA (6 × 10 mL), cyclization on the solid phase was
performed by shaking the resin-bound 8a , PyBOP (5 equiv),
HOBt (5 equiv), and DIPEA (10 equiv) in anhydrous DMF
overnight. This process was repeated once to obtain the resin-
bound 9a . Cleavage of the cyclic decapeptide from the resin was
performed with TFA/H2O/TIS (94:3:3, v/v/v) for 3 h. The filtrate
was collected and the solvent was removed. Cold ether was
added to precipitate the peptide and the ether was decanted.
The solid residue was dissolved in water/methanol mixture and
purified by reversed-phase HPLC with 0.1% TFA/water and
acetonitrile as solvents. Solutions containing the desired prod-
ucts were combined, and solvents were removed under reduced
pressure and lyophilization to obtain cyclic peptide 2 as TFA
salt.
computational simulations that the large head-to-tail
cyclic peptides 2-4 were able to retain mostly enlarged
ring conformations in solution. Again as we have ob-
served before,10,21 the low to submicromolar IC50 values
for the best ligands 15c, 16b, and 17b represent a more
than 100 000-fold improvement over monovalent galac-
tose, which has an IC50 of ∼100 mM in the same surface
receptor-binding inhibitory assay.21,29
In summary, large head-to-tail cyclic decapeptides
(with up to 50-atom ring) were efficiently synthesized on
solid support with use of allyl ester for orthogonal
protection of the C-terminus. Pentavalent ligands as-
sembled on these cyclic peptide cores were able to exhibit
large affinity gains over the monovalent ligand. For
effective inhibition, shorter linkers are optimal for larger
cyclic peptide-based cores, indicating the nature of an
expanded ring conformation of the head-to-tail cyclic
peptides. These results validate that large-ring cyclic
peptides can be an easily accessible and effective class
of backbone support for multivalent ligand design. It is
especially significant for blocking cell-surface receptor
binding by cholera and related toxins because there is
no restriction on the size of the designed multivalent
ligands due to the fact that no biological membrane is to
be crossed by these ligands.
Cyclic peptide 2 (0.157 g, 29.5%): 1H NMR (D2O) δ 4.28-
4.35 (t, J ) 7.0 Hz, 1 × 5H), 3.92-4.00 (d, 2 × 5H), 2.93-3.01
(t, J ) 7.5 Hz, 2 × 5 H), 1.57-1.82 (m, 4 × 5 H), 1.31-1.43 (m,
2 × 5 H); ESI-MS 926.6 [M + H] +, 464.0 [M + 2H]2+, 309.6 [M
+ 3H]3+
.
P en ta va len t Liga n d Assem bly.10,23 A cyclic peptide (2, 3,
or 4) (1 equiv) was mixed with N-Boc-N′-(2-methoxy-3,4-dioxo-
1-cyclobuten-1-yl)-4,7,10-trioxa-1,13-tridecane-diamine (10)10,23
(6 equiv for each coupling site) in water/methanol (25/75, v/v)
at pH 9 adjusted by NaHCO3 solution. After 2 days of reaction
at room temperature, the solution was acidified with dilute
aqueous TFA solution and purified by reversed-phase HPLC
with 0.1% TFA/water and acetonitrile as solvents (except Boc-
protected 12c, which was purified by size exclusion chromatog-
raphy (Sephadex LH-20) in DMF, and purity checked by LC-
MS to >80% pure). After the solvent was removed, the residue
was treated with 10 mL of TFA/CH2Cl2 (50/50, v/v) for 40 min
to remove five Boc groups. The solvent was removed completely
in high vacuum to obtain 12a , 13a , or 14a as the TFA salt.
Repeating the procedure produced longer linker attached core-
linker assembly.
Exp er im en ta l Section
Typ ica l P r oced u r e for th e Syn th esis of Dip ep tid es 6.
Allyl protected amino acids 5 were synthesized according to ref
25 by refluxing allyl alcohol and the related amino acid in the
presence of TsOH in benzene with Dean-Stark apparatus.
Fmoc-Lys(Boc)-OH (1 equiv), compound 5 (1.3 equiv), PyBOP
(1.1 equiv), HOBt (1 equiv), and DIPEA (5 equiv) in anhydrous
CH2Cl2 were stirred at room temperature overnight. After
sequential washing with 5% NaHCO3, water, 10% citric acid,
and saturated NaCl and drying over Na2SO4, the solvent was
removed. The residue was purified via silica gel chromatography
with 5% methanol/CH2Cl2 to give 6.
For example, 5a (3.45 g, 12 mmol) and Fmoc-Lys(Boc)-OH
(4.32 g, 9.23 mmol) produced 6a (4.8 g, 92.0%): 1H NMR (CDCl3)
δ 7.75-7.78 (d, J ) 7.2 Hz, 2H), 7.58-7.61 (d, J ) 7.5 Hz, 2H),
7.38-7.43 (t, J ) 7.3 Hz, 2H), 7.29-7.34 (t, J ) 7.3 Hz, 2H),
5.83-5.93 (m, 1H), 5.28-5.35 (m, 2H), 4.62-4.67 (m, 4H), 4.45
(b, 1H), 4.19-4.24 (t, J ) 6.8 Hz, 1H), 4.06 (s, 2H), 3.66-3.77
(m, 1H), 3.09-3.22 (m, 2H), 1.41-1.50 (m, 15H); ESI-MS m/z
588.4 (M + Na)+, 1153.4 (2M + Na)+.
Solid -P h a se P r otocol for Hea d -to-Ta il Cyclic P ep tid e 2.
The Boc group of the lysine side chain in compound 6a (8.2
mmol) was removed in 15 mL of 1:1 TFA:CH2Cl2 for 20 min.
After the solvent was completely removed in high vacuum, the
residue was dissolved in 5 mL of anhydrous CH2Cl2, followed
by slow addition of DIPEA until the solution became basic. This
solution was immediately added into 500 mg of pre-swollen
4-nitrophenyl carbonate Wang resin (loading 0.71 mmol/g) in 5
mL of anhydrous CH2Cl2 with 1.3 mL of DIPEA. After the
mixture was shaken for 2.5 h, the resin was filtered and washed
with CH2Cl2 (5 × 5 mL), DMA (5 × 10 mL), and CH2Cl2 (5 × 5
mL). The resin was capped with acetic anhydride (0.4 mL, 4.2
mmol) with DIPEA (1.3 mL, 8.43 mmol) in anhydrous CH2Cl2
for 40 min to obtain the resin-bound 7a . Linear peptide elonga-
tion was done following standard protocols. Fmoc removal was
achieved by treatment with 20% piperidine in DMA (3 × 10 mL)
for 10 min each. Coupling was done with 5 equiv of amino acids
(Fmoc-Gly-OH or Fmoc-Lys(Boc)-OH), PyBOP (5 equiv), HOBt
For example, core 2 (32 mg, 21.4 µmol) gave compound 12a
(50 mg, 78%): ESI-MS m/z 1209.3 [M + 2H]2+, 806.8 [M + 3H]3+
,
605.6 [M + 4H]4+, 485.0 [M + 5H]5+
.
A core-linker assembly (12-14) was mixed with N-(N′-(2-
methoxy-3,4-dioxo-1-cyclobuten-1-yl)-ꢀ-aminocaproyl)-â-D-galac-
topyranosylamine (11)10 (4 equiv per site) in water/methanol (25/
75, v/v) at pH 9 adjusted by NaHCO3 solution. After overnight
reaction at room temperature, the product was purified by
reversed-phase HPLC, using 0.1%TFA and acetonitrile as
solvents. Final ligands 15-17 were obtained after lyophilization.
For example, 12a (20 mg, 6.7 µmol) gave ligand 15a (26 mg,
91%): ESI-MS m/z 2133.6 [M + 2H]2+, 1423.5 [M + 3H]3+
,
1068.1 [M + 4H]4+
.
Ack n ow led gm en t. This work is supported by the
National Institutes of Health (AI44954 and AI34501).
Su p p or tin g In for m a tion Ava ila ble: Characterization of
new compounds not listed in the main text. This material is
available free of charge via the Internet at http://pubs.acs.org.
J O0489770
7740 J . Org. Chem., Vol. 69, No. 22, 2004