Synthesis and structure determination of kahalalide F

Article abstract of DOI:10.1021/ja0116728

Kahalalide F, the only member of the family of peptides called kahalalides, isolated from the sacoglossan mollusc Elysia rufescens and the green alga Bryopsis sp., with important bioactivity, is in clinical trials for treatment of prostate cancer. An efficient solid-phase synthetic approach is reported. Kahalalide F presents several synthetic difficulties: (i) An ester bond between two Β-branched and sterically hindered amino acids; (ii) a didehydroamino acid; and (iii) a rather hydrophobic sequence with two fragments containing several Β-branched amino acids in a row, one of them terminated with a saturated aliphatic acid. The cornerstones of our strategy were (i) a quasiorthogonal protecting system with allyl, tert-butyl, fluorenyl, and trityl-based groups, (ii) azabenzotriazole coupling reagents, (iii) formation of the didehydroamino acid residue on the solid phase, and (iv) cyclization and final purification in solution. HPLC, high-field NMR, and biological activity studies showed that the correct stereochemistry of the natural product is that proposed by Rinehart et al. whereas the stereochemistry proposed by Scheuer et al. is tha of a biologically less active diastereoisomer.

Full text of DOI:10.1021/ja0116728

11398
J. Am. Chem. Soc. 2001, 123, 11398-11401
Synthesis and Structure Determination of Kahalalide F1,2
AÅ ngel Lo´pez-Macia`, Jose Carlos Jime´nez, Miriam Royo, Ernest Giralt,* and
Fernando Albericio*
Contribution from the Department of Organic Chemistry, UniVersity of Barcelona, 08028 Barcelona, Spain
ReceiVed July 9, 2001
Abstract: Kahalalide F, the only member of the family of peptides called kahalalides, isolated from the
sacoglossan mollusc Elysia rufescens and the green alga Bryopsis sp., with important bioactivity, is in clinical
trials for treatment of prostate cancer. An efficient solid-phase synthetic approach is reported. Kahalalide F
presents several synthetic difficulties: (i) an ester bond between two â-branched and sterically hindered amino
acids; (ii) a didehydroamino acid; and (iii) a rather hydrophobic sequence with two fragments containing
several â-branched amino acids in a row, one of them terminated with a saturated aliphatic acid. The cornerstones
of our strategy were (i) a quasiorthogonal protecting system with allyl, tert-butyl, fluorenyl, and trityl-based
groups, (ii) azabenzotriazole coupling reagents, (iii) formation of the didehydroamino acid residue on the
solid phase, and (iv) cyclization and final purification in solution. HPLC, high-field NMR, and biological
activity studies showed that the correct stereochemistry of the natural product is that proposed by Rinehart et
al. whereas the stereochemistry proposed by Scheuer et al. is that of a biologically less active diastereoisomer.
Introduction
When this work was started, the absolute stereochemistry of
kahalalide F was unknown.1 Further investigations carried out
independently by Rinehart et al.9 (1a) and Scheuer et al.10 (1b)
allowed them to propose the stereochemistry of kahalalide F.
However, these results were not identical, as they differed in
the stereochemistry of Val-3 and Val-4.
A family of peptides called kahalalides3-6 has been isolated
from the sacoglossan mollusc Elysia rufescens and the green
alga Bryopsis sp. on which it feeds. Two of them, kahalalide F
and its acyclic analogue kahalalide G, incorporate the rare
Z-didehydroaminobutyric acid (Dhb) moiety. Each peptide also
has a saturated aliphatic acid at the N-terminus. Among these
peptides, the largest, kahalalide F, is a cyclic depsipeptide that
contains 13 amino acids and 5-methylhexanoic acid at the
N-terminus, which recently exhibited very interesting antitumor
activity,1,7 and is in clinical trials for treatment of prostate
cancer.8
(1) Abbreviations used for amino acids and the designations of peptides
follow the rules of the IUPAC-IUB Commission of Biochemical Nomen-
clature in: J. Biol. Chem. 1972, 247, 977-983. The following additional
abbreviations are used: ACH, R-cyano-4-hydroxycinnamic acid; Alloc,
allyloxycarbonyl; Boc, tert-butyloxycarbonyl; t-Bu, tert-butyl; Cl-TrtCl-
resin, 2-chlorotrityl chloride-resin; DHB, 2,5-dihydroxybenzoic acid; Z-Dhb,
R,â-didehydro-R-aminobutyric acid; DIEA, N,N-diisopropylethylamine;
DMF, N,N-dimethylformanide; EDC, 1-ethyl-3-(3′-dimethylaminopropyl)-
carbodiimide hydrochloride; ESMS, electrospray mass spectrometry; FABMS,
fast atom bombardment mass spectrometry; Fmoc, 9-fluorenylmethoxycar-
bonyl; HATU, N-[(dimethylamino)-1H-1,2,3-triazolo-[4,5-b]pyridin-1-yl-
methylene]-N-methylmethanaminium hexafluorophosphate N-oxide; HBTU,
N-[(1H-benzotriazol-1-yl)(dimethylamino)methylene]-N-methylmethanamin-
ium hexafluorophosphate N-oxide; HOAt, 1-hydroxy-7-azabenzotriazole (3-
hydroxy-3H-1,2,3-triazolo-[4,5-b]pyridine); HOBt, 1-hydroxybenzotriazole;
5-MeHex-OH, 5-methylhexanoic acid; MeOH, methanol; NMM, N-meth-
ylmorpholine; PyAOP, 7-azabenzotriazol-1-yl-N-oxy-tris(pyrrolidino)phos-
phonium hexafluorophosphate; PyBOP, benzotriazol-1-yl-N-oxy-tris-
(pyrrolidino)phosphonium hexafluorophosphate; SPS, solid-phase synthesis;
TFA, trifluoroacetic acid; amino acid symbols denote the L-configuration
unless stated otherwise. All solvent ratios are volume/volume unless stated
otherwise.
Results and Discussion
Although due to its complexity kahalalide F represents a real
synthetic challenge, the development of an efficient synthetic
approach should allow the doubt over the stereochemistry to
be settled as well as making possible the synthesis of large
amounts of the compound and its analogues for further
pharmacological investigations. Besides being a cyclic com-
pound, kahalalide F presents several synthetic difficulties:
(2) Taken in part from the Ph.D. Thesis of AÅ ngel Lo´pez-Macia`,
University of Barcelona, Spain, December 2000.
(3) Hamann, M. T.; Scheuer, P. J. J. Am. Chem. Soc. 1993, 115, 5825-
5826.
(4) Hamann, M. T.; Otto, C. S.; Scheuer, P. J.; Dunbar, D. C. J. Org.
Chem. 1996, 61, 6594-6600. Hamann, M. T.; Otto, C. S.; Scheuer, P. J.;
Dunbar, D. C. J. Org. Chem. 1998, 63, 4856.
(5) Goetz, G.; Nakao, Y.; Scheuer, P. J. J. Nat. Prod. 1997, 60, 562-
567.
(6) Horgen, F. D.; Delossantos, D. B.; Goetz, G.; Sakamoto, B.; Kan,
Y.; Nagai, H.; Scheuer, P. J. J. Nat. Prod. 2000, 63, 152-154.
(7) Garc´ıa-Rocha, M.; Bonay, P.; Avila, J. Cancer Lett. 1996, 99, 43-
50.
(8) Faircloth, G. Personal communication.
(9) Bonnard, I.; Manzanares, I.; Rinehart, K. L. J. Nat. Prod. In press.
(10) Goetz, G.; Yoshida, W. Y.; Scheuer, P. J. Tetrahedron 1999, 55,
7739-7746.
10.1021/ja0116728 CCC: $20.00 © 2001 American Chemical Society
Published on Web 10/27/2001

Synthesis and Structure Determination of Kahalalide F
J. Am. Chem. Soc., Vol. 123, No. 46, 2001 11399
Scheme 1. Solid-Phase Strategies Investigated for the Preparation of Kahalalide Fa
a Cl attached to the shaded circle represents Cl-TrtCl-resin, 2-chlorotrityl chloride-resin.
Figure 1. HPLC chromatograms of peptides 1a (a) and 1b (b).
Figure 2. HPLC chromatograms of co-injection of peptide 1a and
Reverse-phase C-18 columns were used for the analysis with elution
by a linear gradient over 30 min of 0.036% TFA in ACN and 0.045%
natural kahalalide F (a) and co-injection of peptide 1b and natural
kahalalide F (b). Reverse-phase C-18 columns were used for the analysis
with elution by a linear gradient over 30 min of 0.036% TFA in ACN
TFA in H2O from 45:55 to 90:10, flow rate 1.0 mL min-1
.
and 0.045% TFA in H2O from 45:55 to 6:4, flow rate 1.0 mL min-1
.
(i) an ester bond between two â-branched and sterically hindered
amino acids (Val and D-alloThr); (ii) the didehydroamino acid
(Z-Dhb); and (iii) a rather hydrophobic sequence with two
fragments containing several â-branched amino acids in a row,
one of them terminated with the saturated aliphatic acid. Taking
these difficulties into account, three different strategies (Scheme
1, a-c) based on the solid-phase approach11 were studied for
the synthesis of 1a and 1b, the two proposed structures of
kahalalide F, to establish unequivocally the correct stereochem-
istry of the natural product.
All of these strategies involve the elongation of the synthetic
chain on the solid phase, cleavage of the protected peptide from
the resin, subsequent cyclization, and final deprotection in
solution. Although the macrocyclization can be carried out in
any position except that between Phe and Z-Dhb because the
formation of the double bond is carried out after the amide
between these two residues is formed,12 doing so between Val
and Phe looked to be the most favorable option, since the other
(11) Lloyd-Williams, P.; Albericio, F.; Giralt, E. Chemical Approaches
to the Synthesis of Peptides and Proteins; CRC Press: Boca Raton, FL,
1997.
(12) Attempts to perfom the cyclization step before the formation of the
Z-Dhb failed, presumably because Z-Dhb facilitates a favorable conformation
for the reaction.

11400 J. Am. Chem. Soc., Vol. 123, No. 46, 2001
Lo´pez-Macia` et al.
Figure 3. Expansion of the aromatic and amide/aliphatic regio of (a) the TOCSY (500 MHz, 72 ms) spectrum of 1a, (b) the TOCSY (500 MHz,
72 ms) spectrum of 1b, and (c) the ROESY (500 MHz, 200 ms) spectrum of 1a. All the spectra were registered in DMSO-d6 at 298 K.
positions involve two â-branched amino acids or the poorly
reactive carboxyl or amino function of Z-Dhb.
CuCl (60 equiv) in DMF-CH2Cl2 (5:1) for 6 days. This reaction
led, as demonstrated in a model peptide,17 exclusively to the Z
isomer that is thermodynamically more stable. Alternatively in
strategy c, the dipeptide Alloc-Phe-(Z)-Dhb-OH (5 equiv), which
was prepared in solution from Alloc-Phe-OH and H-Thr-OtBu
with EDC and subsequent dehydration with EDC (6.5 equiv)
in the presence of CuCl (2.7 equiv) and treatment with TFA,
was coupled with HATU-DIEA (5:10) for 16 h with a further
recoupling for 3 h.21
The linear sequence was synthesized on a 2-chlorotrityl
chloride-resin (ClTrt-Cl-resin),13 which allowed the cleavage
of the peptide under very mild acid conditions and in the
presence of other acid-labile protecting groups. The limited
incorporation of Fmoc-D-Val-OH (0.2 equiv) was performed
in the presence of DIEA (2 equiv).14,15 The elongation of the
peptide chain was carried out by using the Fmoc/tBu strategy.
The D-alloThr and the Thr precursor of the Z-Dhb were both
introduced without protection of the hydroxyl function. HATU-
DIEA16 was used for all the amide formations (in all cases single
coupling for 90 min with 5 equiv of Fmoc-amino acid and
HATU and 10 equiv of base in DMF gave a ninhydrin negative
test).
Before the cleavage of the protected peptide from the resin
[TFA-CH2Cl2 (1:99), 5 × 0.5 min, 65-87% yield)], the Alloc
group was removed as described above.22 The cyclization step
was performed with PyBOP-DIEA (3:6 equiv) in DMF for 1
h.23 Final deprotection was carried out with TFA-H2O (95:5)
for 1 h.
In strategy a, the incorporation of Fmoc-Val-OH did not take
place in good yield (30% with DIPCDI-DMAP), presumably
due to the hydrophobicity of the peptide chain, which favors
interchain aggregation. Thus, protected Val should be incorpo-
rated in a previous step (strategies b and c), requiring the use
of a protecting group for the R-amino function, such as the Alloc
group, which is orthogonal to the Fmoc and the acid-labile
protecting groups.17 Removal of the Alloc group was carried
out with Pd(PPh3)4 (0.1 equiv) in the presence of PhSiH3 (10
equiv) under an atmosphere of Ar.18 The double bond of the
didehydroamino acid was formed on the solid phase through a
â-elimination reaction (strategy b) with use of a method
developed recently in our laboratory,19 with the modification
described by Fukase et al.20 that uses EDC (100 equiv) as the
activating reagent of the hydroxyl function in the presence of
Although strategies b and c both led to the correct product
in the case of both stereoisomers of kahalalide F, the HPLC
quality of the crude products obtained by strategy b was in both
cases clearly superior. Crude 1a and 1b were purified by
medium-pressure chromatography24 to give the title products
[10-14% overall yield (synthesis and purification)], which both
showed high purity by HPLC (Figure 1) and a correct MALDI-
TOF-MS.
Product 1a coeluted by HPLC with an authentic sample of
kahalalide F (Figure 2a), while 1b showed a longer retention
time than the natural compound (Figure 2b).
1
Furthermore, the H NMR spectra [(500 MHz, DMSO-d6)
1H, TOCSY (72 ms), ROESY (200 ms)] of 1a were identical
1
with those of the natural peptide, while the H NMR spectra
for 1b showed a chemical shift pattern different from that of
natural kahalalide F as well as the presence of two conformations
due to a cis-trans equilibrium around the L-Val-D-Pro peptide
bond, an equilibrium that was not observed in the natural product
(13) Barlos, K.; Gatos, D.; Scha¨fer, W. Angew. Chem., Int. Ed. Engl.
1991, 30, 590-593.
(14) The use of ClTrt-Cl-resins of high loading (>0.5 mmol/g) for the
preparation of hydrophobic peptides such as kahalalide F usually leads to
impure peptide. Chiva, C.; Vilaseca, M.; Giralt, E.; Albericio, F. J. Pept.
Sci. 1999, 5, 131-140.
(15) Unreacted reactive sites were capped with MeOH-DIEA.
(16) Carpino, L. A.; El-Faham, A.; Minor, C. A.; Albericio, F. J. Chem.
Soc., Chem. Commun. 1994, 201-203.
(17) The incorporation of Alloc-Val-OH (7 equiv) was carried with an
equimolar amount of DIPCDI and 0.7 equiv of DMAP for 2 h. This protocol
was repeated twice and the yield by amino acid analysis was quantitative.
Other methods based in the use of Fmoc-Val-Cl in the presence of base
gave less satisfactory results.
(18) Thieriet, N.; Alsina, J.; Giralt, E.; Guibe´, F.; Albericio, F.
Tetrahedron Lett. 1997, 38, 7275-7278.
(19) Royo, M.; Jime´nez, J. C.; Lo´pez-Macia`, A.; Giralt, E.; Albericio,
F. Eur. J. Org. Chem. 2001, 45-48.
(20) Fukase, K.; Kitazawa, M.; Sano, A.; Shimbo, K.; Horimoto, S.;
Fujita, H.; Kubo, A.; Wakamiya, T.; Shibe, A. Bull. Chem. Soc. Jpn. 1992,
65, 2227-2240.
(21) Other coupling reagents based on HOBt, such as HBTU or DIPCDI-
HOBt, led to incomplete incorporations of the dipeptide.
(22) The R-amino function of the Phe also can be protected with Fmoc,
because this protecting group is totally stable to the dehydration reaction
conditions.
(23) Kates, S. A.; Sole´, N. A.; Johnson, C. R.; Hudson, D.; Barany, G.;
Albericio, F. Tetrahedron Lett. 1993, 34, 1549-1552.
(24) Vydac C18 (15-20 µm, 300 Å, 240 × 24 mm), linear gradient from
20 to 60% of acetonitrile (+0.05% TFA) in water (+0.05% TFA) in 5 h
(300 mL each solvent), 120 mL/h, detection at 220 nm.

Synthesis and Structure Determination of Kahalalide F
J. Am. Chem. Soc., Vol. 123, No. 46, 2001 11401
(Tables 1-3, Supporting Information).25 Finally, while the
biological activity of 1a was in the same range as that of natural
kahalalide F, 1b was more than 10 times less active.26 In
conclusion, this first synthesis of kahalalide F and a diastereomer
unequivocally established the stereochemistry of the natural
compound to be that proposed by Rinehart et al.6 whereas the
stereochemistry proposed by Goetz et al.7 corresponds to a
biologically less active diastereomer. The synthetic method
reported herein may find general application for the preparation
of other peptides containing didehydroamino acids. All reactions
except cyclization and the final deprotection take place on the
solid phase.
(25) The 1H NMR chemical shifts in DMF-d7 containing 1 equiv of TFA
of natural kahalalide F were slightly different from those shown in Table
1 of the Supporting Information. However, when natural kahalalide F was
treated in a manner similar to the synthetic kahalalides (solution in TFA,
evaporation under reduced pressure, and lyophilization), all chemical shifts
were identical with those for synthetic kahalalide F (1a).
Acknowledgment. We thank Dr. I. Manzanares (Pharma
Mar, s.a) for his helpful discussions. This work was partially
supported by Pharma Mar, s.a., CICYT (BIO1099-0484 and
BQU2000-0235), and Generalitat de Catalunya [Grup Consolidat
(1999SGR 00042) and Centre de Refere`ncia en Biotecnologia].
Supporting Information Available: Experimental data
corresponding to the synthesis of kahalalide F (1a) and 1b via
both dehydration on solid-phase and incorporation of the
dehydrated dipeptide; Tables 1-3 giving 1H NMR (500 MHz,
DMSO-d6) data for kahalalide F (1a) and 1b (both isomers);
1
Figures 4-12 showing H NMR spectra (500 MHz) (5 mM),
TOCSY (72 ms), and ROESY (200 ms) for kahalalide F (1a),
1b, and natural kahalalide in DMSO-d6 at 298 K (PDF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
(26) The biological test was performed in Pharma Mar s.a.
JA0116728