Efficient solid-phase-based total synthesis of the bisintercalator TANDEM

Article abstract of DOI:10.1021/jo050959a

In this article, the first solid-phase-based total synthesis of TANDEM, a synthetic analogue of triostin A, is described. In initial studies, the synthesis incorporated depsipeptide formation, introduction of chromophores, and disulfide bond formation on

Full text of DOI:10.1021/jo050959a

Efficient Solid-Phase-Based Total Synthesis of the Bisintercalator
TANDEM
John P. Malkinson,* Michael K. Anim, Mire Zloh, and Mark Searcey*
Department of Pharmaceutical and Biological Chemistry, School of Pharmacy, University of London,
29-39 Brunswick Square, London WC1N 1AX, U.K.
Andrew J. Hampshire and Keith R. Fox
School of Biological Sciences, University of Southampton, Bassett Crescent East,
Southampton SO16 7PX, U.K.
mark.searcey@ulsop.ac.uk; john.malkinson@ulsop.ac.uk
Received May 13, 2005
In this article, the first solid-phase-based total synthesis of TANDEM, a synthetic analogue of
triostin A, is described. In initial studies, the synthesis incorporated depsipeptide formation,
introduction of chromophores, and disulfide bond formation on the solid phase, prior to a final
solution-phase macrolactamization, to give the target molecule. Although pure TANDEM was
obtained in an overall yield comparable to those for all syntheses to date, the yield of the final
cyclization was low (11%). A more efficient approach involved removal from the solid phase prior
to disulfide bond formation. The resulting linear peptide underwent macrolactamization under mild
conditions and high dilution. Final disulfide bond formation was essentially quantitative and gave
the target molecule, TANDEM, in an overall yield of 18%. The final compound was assessed for its
ability to bind to 5′-TpA sequences on DNA by DNase I footprinting. This efficient synthesis sets
the stage for a study of the structure-activity relationship of TANDEM and the natural product
triostin A, with analogues containing “point mutations” at every site within the cyclic compounds.
Introduction
inserting their chromophores into the duplex on either
side of 5′-CG, forming a two base pair sandwich.6 It has
also been demonstrated that the bases on either side of
the sandwich exert an effect on binding, such that the
natural products effectively recognize a four base pair
sequence.7 Triostin A N-Demethylated (TANDEM 4) is
a synthetic analogue of triostin A in which secondary
amides replace the tertiary amides of the natural prod-
uct.8 This small change results in a change in sequence
selectivity, with TANDEM showing a preference for 5′-
TA steps, particularly 5′-ATAT.9 Such a change mirrors
that seen with the incorporation of imidazole groups into
the hairpin polyamides. However, few, if any, attempts
Echinomycin (1)1 and triostin A2 (2) (Figure 1) are
parent members of a family of antitumor antibiotics that
have, at one stage, advanced into phase I clinical trials.3
Numerous natural product analogues have been de-
scribed, most recently thiocoraline 3,4 which is currently
under preclinical development and has recently been
synthesized by solution-phase methods.5 Both echinomy-
cin and triostin A bind to DNA by bisintercalation,
* Corresponding authors. Phone: 020-7753-5873 or 5862. Fax: 020-
7753-5964.
(1) Waring, M. J. Path. Biol. (Paris) 1992, 1022-1034.
(2) Otsuka, H.; Shoji, J. Tetrahedron 1967, 23, 1535-1542.
(3) Gradishar, W. J.; Vogelzang, N. J.; Kilton, L. J.; Leibach, S. J.;
Rademaker, A. W.; French, S.; Benson, A. B. Invest. New Drugs 1995,
13, 171-174.
(4) Perez Baz, J.; Canedo, L. M.; Fernandez Puentes, J. L.; Silva
Elipe, M. V. J. Antibiot. 1997, 50, 738-741.
(5) Boger, D. L.; Ichikawa, S.; Tse, W. C.; Hedrick, M. P.; Jin, Q. J.
Am. Chem. Soc. 2001, 123, 561-568.
(6) Waring, M. J.; Wakelin, L. P. G. Nature 1974, 252, 653-657.
(7) Van Dyke, M. M.; Dervan, P. B. Science 1984, 225, 1122-1127.
(8) (a) Ciardelli, T. L.; Olsen, R. K. J. Am. Chem. Soc. 1977, 99,
2806-2807. (b) Ciardelli, T. L.; Chakravarty, P. K.; Olsen, R. K. J.
Am. Chem. Soc. 1978, 100, 7684-7690.
(9) Lavesa, M.; Fox, K. R. Anal. Biochem. 2001, 293, 246-250.
10.1021/jo050959a CCC: $30.25 © 2005 American Chemical Society
Published on Web 08/24/2005
7654
J. Org. Chem. 2005, 70, 7654-7661

Total Synthesis of the Bisintercalator TANDEM
FIGURE 2. Potential sites for off-resin cyclization of 4.
Results and Discussion
Design of the Solid-Phase Synthesis. The most
attractive synthesis of TANDEM involved the incorpora-
tion of as many steps as possible on the solid phase prior
to release, maintaining a minimum number of solution-
phase steps to complete the synthesis. Thus, formation
of the depsipeptides, incorporation of the chromophores,
and preferably, closure of the disulfide bridge with the
growing chain immobilized were all considered. The
identification of the point of final ring closure was also
key to the successful achievement of the synthesis of
TANDEM. There are four potential sites for macrocy-
clization (A-D, Figure 2). All were carefully analyzed,
and methodology was compared with published proce-
dures. Olsen and co-workers used NHS-DCC to bring
about macrolactamization at A of the bis-Cbz-protected
cyclic peptide in the original synthesis of the depsipep-
tides, although the isolated yields were only of the order
of 26-43%.8 In our synthesis, this would involve im-
mobilization of alanine and completion of the linear
synthesis at the final cysteine, an attractive proposition
as the starting resin bound Fmoc amino acid is readily
commercially available and the distance from the dep-
sipeptide bond is optimal. The formation of the depsipep-
tide bond (B, Figure 2) as the final step was less
attractive, as the lower nucleophilicity of the alcohol
would not be helpful in the final macrolactonization. The
conditions required for such a ring closure would be a
potential source of significant racemization of the termi-
nal valine and/or dehydration and racemization of the
serine residue. Diederichsen and co-workers have re-
cently described a ring closure using DIC/HOAt between
serine and alanine (C, Figure 2), which gave an excellent
yield of 82% for the bis-Cbz-protected peptide.12 Applica-
tion to a solid-phase method would be difficult. It would
require an orthogonally protected resin-bound D-serine
that could be selectively NR-deprotected to introduce first
the chromophore and then side-chain O-deprotected to
introduce the depsipeptide ester. Wang resin-bound
Fmoc-D-Ser(Trt)-OH would be suitable but is not readily
available in the U.K. Finally, formation of the cysteine-
valine bond (D, Figure 2) is a possibility but would
involve immobilization of an orthogonally protected cys-
teine residue onto the solid support and final macrolac-
tamization via activation of the terminal cysteine car-
boxyl, both of which are known to lead to racemization.13
Bearing these points in mind, we initially focused on a
synthesis involving incorporation of both depsipeptide
bonds, both chromophores and the disulfide bond, fol-
FIGURE 1. Structures of 1-4.
have been made to exploit the effect seen in going from
the natural products to the synthetic agents in the
quinoxaline antibiotics. In part, we believe this stems
from problems encountered in the solution-phase syn-
thesis, which is relatively demanding, even though recent
efforts have been disclosed that focused on developing
methods for solution-phase combinatorial libraries.10 If
a solid-phase approach to these compounds could be
developed, it would allow the rapid synthesis of analogues
with different amino acids at each position. A solid-phase
synthesis of an amide-linked analogue of TANDEM has
recently been disclosed.11 The linear nature of a solid-
phase synthesis would allow the generation of unsym-
metrical derivatives. In all, a parallel synthesis of
analogues would allow the rapid development of a full
structure-activity profile for these compounds. The
investigation of extended analogues will also be straight-
forward and allow the determination of the suitability
of these compounds for targeting longer sequences of
DNA. With this reinvestigation of classical compounds
in the light of new synthetic, high-throughput methods,
we report here the first solid-phase-based total synthesis
of TANDEM and a more efficient second generation
synthesis that gives the highest reported yields of the
target molecule. We also investigated the DNA binding
ability of several of the intermediates in the synthetic
pathway and found that only the final product, TAN-
DEM, binds to DNA with high affinity.
(10) Boger D. L.; Lee, J. K. J. Org. Chem. 2000, 65, 5996-6000.
(11) Dietrich, B.; Diederichsen, U. Eur. J. Org. Chem. 2005, 147-
153.
(12) Lorenz, K. B.; Diederichsen, U. J. Org. Chem. 2004, 69, 3917-
3927.
J. Org. Chem, Vol. 70, No. 19, 2005 7655

Malkinson et al.
SCHEME 1a
a Reagents and conditions: (a) (i) 20% piperidine, DMF, rt, 2 × 10 min, (ii) Fmoc-D-Ser(Trt)-OH, HBTU, HOBt, iPr2NEt, DMF, rt, 2 ×
30 min; (b) (i) 20% piperidine, DMF, rt, 2 × 10 min, (ii) quinoxaline-2-carbonyl chloride, iPr2NEt, DMF, rt, 2 × 30 min; (c) 2% TFA, 5%
iPr3SiH, CH2Cl2, rt, 3 × 5 min then 1 × 20 min, (ii) Fmoc-Val-OH, DIC, DMAP, DMF, rt, 3 × 60 min; (d) (i) 20% piperidine, DMF, rt, 2
× 10 min, (ii) Fmoc-Cys(Acm)-OH, HBTU, HOBt, iPr2NEt, DMF, rt, 2 × 30 min; (e) (i) 20% piperidine, DMF, rt, 2 × 10 min, (ii) Fmoc-
Ala-OH, HBTU, HOBt, iPr2NEt, DMF, rt, 2 × 30 min; (f) I2, DMF, rt, 1 h; (g) TFA/H2O/iPr3SiH (95:2.5:2.5), rt, 2 h; (31% overall yield of
12 from 5); (h) EDC (5 equiv), HOBt (5 equiv), CH2Cl2/DMF 7:1, rt, 24 h, 11% yield.
lowed by release from the resin and solution-phase-based
ring closure.
necessary for formation of the first depsipeptide bond.
Esterification was achieved using a 10-fold excess of NR-
Fmoc-valine in the presence of a 5-fold excess of DIC and
catalytic (0.1 equiv) DMAP. The use of additional base
was avoided to minimize the potential risk of epimeriza-
tion and/or dehydration of the D-serine residue. Resin-
bound tridepsipeptide 8 was then extended, coupling NR-
Fmoc, S-acetamidomethyl-cysteine, and NR-Fmoc-alanine
in turn. Repetition of the sequence described above (up
to and including NR-Fmoc deprotection of the second
cysteine residue) gave the completed linear sequence 11,
immobilized on the solid phase. Simultaneous removal
of the cysteine S-acetamidomethyl protecting groups and
oxidation to form the disulfide bridge was achieved by
treatment with iodine in DMF for 1 h at room temper-
ature. Unlike solution-phase iodine oxidation, which
must be performed under high-dilution conditions, on-
resin oxidation proceeds without the risk of intermolecu-
lar disulfide formation. The disulfide-bridged peptide was
removed from the solid support by acidolysis using 95%
TFA (TFA/H2O/TIS 95:2.5:2.5). The suspension was then
filtered, the filtrate was concentrated in vacuo, and the
First Generation Synthesis. The linear depsipeptide
precursor was assembled using NR-Fmoc-based solid-
phase peptide synthesis. NR-Fmoc-alanine, immobilized
onto a polystyrene solid support via the Wang linker (5),
was deprotected using 20% piperidine (Scheme 1). Sub-
sequent coupling of an NR-Fmoc, side-chain O-trityl-
protected D-serine derivative generated resin-bound dipep-
tide 6. The introduction of this orthogonally protected
trifunctional amino acid was key to the synthesis of the
linear precursor. It allowed selective NR-Fmoc deprotec-
tion, followed by introduction of the quinoxaline-2-
carbonyl chromophore, which was coupled cleanly as its
acyl chloride derivative. The side-chain O-trityl protection
of 7 was then removed by very mild acidolysis (2% TFA,
5% TIS, CH2Cl2) to generate the free â-hydroxyl group
(13) (a) Atherton, E.; Benoiton, N. L.; Brown, E.; Sheppard, R. C.;
Williams, B. J. J. Chem. Soc., Chem. Commun. 1981, 336-337. (b)
Fujiwara, Y.; Akaji, K.; Kiso, Y. Chem. Pharm. Bull. 1994, 42, 724-
726. (c) Kaiser, T.; Nicholson, G. J.; Kohlbau, H. J.; Voelter, W.
Tetrahedron Lett. 1996, 37, 1187-1190. (d) Han, Y.; Albericio, F.;
Barany, G. J. Org. Chem. 1997, 62, 4307-4312.
7656 J. Org. Chem., Vol. 70, No. 19, 2005

Total Synthesis of the Bisintercalator TANDEM
crude peptide 12 (75 mg, 36% based upon resin loading)
was isolated by precipitation with ether. Analytical RP-
HPLC on a Zorbax C18 column revealed one major peak
(Rt 14.6 min) with minor impurities. Preparative HPLC
gave the pure peptide 12 (65 mg, 31% for 21 steps).
HBTU and HOBt in the presence of iPr2NEt (i.e., the
same conditions used for peptide chain elongation) had
previously been found ineffective for esterification, based
on the (indirect) observation of a negative or very weak
Kaiser test after subsequent treatment of a sample of
resin with 20% piperidine (indicating an absence of NR-
Fmoc-protected amino groups and, therefore, a failure of
coupling). Direct monitoring of the presence (or absence)
of hydroxyl groups on the solid phase14 proved insuf-
ficiently sensitive or reliable in our hands. In situ
formation of the OBt active ester (using HOBt in the
presence of equimolar DIC and iPr2NEt) and the use of
a preformed NR-Fmoc-valine pentafluorophenyl ester
(with catalytic DMAP) proved similarly unsuccessful.
Reduction of the molar excess of NR-Fmoc-valine in the
original depsipeptide bond-forming conditions (from 10
equiv to 5 equiv, effectively promoting activation as the
O-acylisourea rather than primarily as the PSA) resulted
in similar to slightly improved yields and purities of
linear precursor 12 on resin cleavage. Increasing the
catalytic amount of DMAP (from 0.1 to 0.5 equiv) resulted
in a slightly reduced yield of crude depsipeptide of similar
purity.
Finally, two other methods for depsipeptide bond
formation were evaluated. The activation of NR-Fmoc-
protected amino acids using 1-(2-mesitylenesulfonyl)-3-
nitro-1H-1,2,4-triazole (MSNT) in the presence of 4-me-
thylimidazole has been reported for the esterification of
hydroxymethyl-derivatized solid supports with minimal
racemization.15 Activation of NR-Fmoc-valine (5 equiv)
using MSNT (5 equiv) and 4-methylimidazole (3.75 equiv)
for the depsipeptide bond-forming steps was successful,
but gave lower crude yields and purities of 12. Mitsunobu
conditions have also been used for loading hydroxym-
ethyl-derivatized resins.16 Esterification using PPh3 and
either DEAD or DIAD in the presence or absence of added
base was unsuccessful (as indirectly monitored as de-
scribed above). Analysis of the mass spectra of the crude
cleavage products after attempted esterification indicated
the presence of both unreacted 7 and the corresponding
side chain dehydrated byproduct, in addition to a small
amount of desired depsipeptide 8.
The alternative convergent approach necessitated the
solution-phase synthesis of a suitably orthogonally pro-
tected didepsipeptide building block. The primary feature
of the solid-phase synthesis of TANDEM is to maximize
the number of synthetic steps performed on the resin,
thus exploiting the advantages of this approach. The
incorporation of a preformed didepsipeptide building
block (prepared in solution), however, serves several
purposes: it removes two of the potentially significant
sources of byproduct formation (i.e., failure of depsipep-
tide bond formation and epimerization/dehydration of the
sensitive D-serine residue), depsipeptide bond formation
is difficult to monitor directly in a satisfactory manner,
and any byproducts resulting from undesirable or incom-
plete reaction(s) are likely to be difficult to separate from
the desired product during final HPLC purification. The
didepsipeptide building block can be prepared relatively
Head-to-tail cyclization of the pure 12 (0.8 mM, CH2-
Cl2/DMF 7:1, 24 h, room temperature (rt)) in solution
using EDC gave a low (11%) yield of the ring-closed
compound TANDEM (4). Efforts to increase the yield,
including longer reaction times and increased tempera-
ture were not successful. Although the starting material
was consumed to generate the active ester, further
product could not be isolated, and it appeared by HPLC
that unreacted active ester and, perhaps, some dimer
peptide were present (although neither was isolated). The
structure of the final product 4 was confirmed by NMR
and mass spectrometry and compared with the literature
data given by Olsen and co-workers.14
Although the yield of the final solution-phase cycliza-
tion was poor, the overall yield of the completed synthesis
was 4.2%, which compares well with the 2.6-6.5%
reported by Olsen for the same synthesis in solution.8
Diederichsen and co-workers recently reported the syn-
thesis of the bis-Cbz-protected cyclic peptide in which a
solution-phase cyclization of the peptide incorporating the
disulfide bond and the depsipeptide bonds resulted in an
82% yield, although their overall yield for this synthesis
was only 3.6%.12 The final cyclization in this case was
between D-Ser and Ala, suggesting, perhaps, that the low
yield in our case may be due to constraints imposed by
the trajectory of attack on the cysteine that is already
involved in the disulfide bond. A solution-phase structure
of peptide 12 suggests that the molecule has a large
hydrophobic core and a hydrogen bond that may also
contribute to hindering the cyclization reaction. However,
this structure was obtained in a solvent (CD3OD) differ-
ent from that of the reaction, as all attempts to obtain
the NMR spectra in CD2Cl2/DMF-d7 (7:1) gave poor
resolution (data not shown), possibly as a consequence
of aggregation and/or limited solubility.
Investigations of Depsipeptide Formation. The
final yield of the peptide 12 represents an average of 94%
for each step of the synthesis. Although this would be
acceptable in a solution synthesis, and a yield of 31% for
21 steps would be remarkable, in the context of a solid-
phase synthesis, yields of more than 99% would be
expected for each peptide coupling. This would suggest
that the introduction of the chromophore and the dep-
sipeptide bond formation may be limiting the final yield.
As a consequence, we briefly investigated the use of other
ester-forming reagents and the assembly of an orthogo-
nally protected didepsipeptide building block in solution
prior to incorporation in the solid-phase synthesis in a
convergent approach.
Depsipeptide bond formation during the original as-
sembly of resin-bound linear precursor 11 was achieved
essentially by activating NR-Fmoc-valine (10 equiv) as the
preformed symmetrical anhydride (PSA), using DIC (5
equiv) and catalytic DMAP. In situ activation using
(15) (a) Blankemeyer-Menge, B.; Nimtz, M.; Frank, R. Tetrahedron
Lett. 1990, 31, 1701-1704. (b) Nielsen, J.; Lyngsø, L. O. Tetrahedron
Lett. 1996, 37, 8439-8442.
(16) Fancelli, D.; Fagnola, M. C.; Severino, D.; Bedeschi, A. Tetra-
hedron Lett. 1997, 38, 2311-2314.
(14) (a) Pomonis, J. G.; Severson, R. F.; Freeman, P. J. J. Chro-
matogr. 1969, 40, 78-84. (b) Attardi, M. E.; Falchi, A.; Taddei, M.
Tetrahedron Lett. 2000, 41, 7395-7399. (c) Kuisle, O.; Lolo, M.; Quinoa,
E.; Riguera, R. Tetrahedron 1999, 55, 14807-14812.
J. Org. Chem, Vol. 70, No. 19, 2005 7657

Malkinson et al.
SCHEME 2a
and solution phase) had demonstrated the stability of
ester-based protecting groups to the conditions of Dde
removal (in some cases up to 10% hydrazine hydrate was
used), and the Dde group is routinely removed in the
solid-phase synthesis of modified peptides without af-
fecting ester-based protecting groups or resin handles.
In addition to the investigations of depsipeptide bond
formation reported earlier in this article, we also briefly
examined the conditions for incorporation of the quinoxa-
line-2-carbonyl chromophores and for disulfide formation
on the solid phase. Acylation with quinoxaline-2-carbonyl
chloride in the presence of HOBt and iPr2NEt, or with
the corresponding quinoxaline-2-carboxylic acid in the
presence of HBTU, HOBt, and iPr2NEt, provided no
significant advantage over the original coupling condi-
tions (quinoxaline-2-carbonyl chloride and iPr2NEt alone).
Furthermore, disulfide formation using Tl(tfa)3 (1.2
equiv) in DMF/anisole (19:1)17 gave disulfide-bridged
depsipeptide 12 in similar crude yields but reduced purity
compared to the use of iodine.
Second Generation, Efficient Synthesis of TAN-
DEM. Bearing in mind the problems encountered with
the alanine-cysteine head-to-tail ring closure reaction, we
reasoned that a more flexible, linear peptide may lead
to higher yields in the macrolactamization step. This
approach removes the conformational constraints im-
posed by the prior formation of the disulfide bridge,
potentially promoting the desired intramolecular cycliza-
tion. We also reasoned that intramolecular disulfide bond
formation would be highly favored (i.e., over competing
intermolecular reactions) in the head-to-tail cyclic bis-
Acm-protected depsipeptide so produced. If so, the final
disulfide-forming step should proceed rapidly and cleanly
with few unwanted byproducts. To this end, we repeated
the solid-phase synthesis and reversed the final two steps
of disulfide bond formation and macrolactamization.
The linear peptide was assembled in a manner similar
to the first synthesis, utilizing NR-Fmoc chemistry (Scheme
3). Depsipeptide bond formation was achieved using a
5-fold excess of NR-Fmoc-valine in the presence of a 5-fold
excess of DIC and catalytic (0.1 equiv) DMAP. Once
assembly on the solid phase was complete, the resin-
bound bis-Acm-protected linear precursor was removed
from the solid support (i.e., without prior disulfide bond
formation). The depsipeptide 18 was isolated in 65% yield
(typically 48-65% yield) in crude form and 30% yield
after purification by HPLC.
a Reagents and conditions: (a) Dde-OH, NEt3, abs. EtOH, reflux,
18 h, 70%; (b) (i) Cs2CO3, MeOH/H2O, 0 °C to rt, 1 h, (ii) BnBr,
DMF, rt, 4 h, 89%; (c) 5% TFA/CH2Cl2, iPr3SiH, rt, 5 min, 85%;
(d) 10, DIC, DMAP, CH2Cl2, rt, 6 h, 85%; (e) 10% Pd(C), THF, H2,
rt, 48 h, 79%.
easily in large scale and with guaranteed diastereomeric
purity. The depsipeptide product thus resulting from a
convergent approach to synthesis would potentially be
cleaner and more easily purified, and the yield (compared
to the original synthesis) would give an indication of the
loss of material attributable to formation of the depsipep-
tide bonds on the solid phase.
Quasi-orthogonal protection of the didepsipeptide build-
ing block was chosen such that the D-serine NR-protection
could first be removed, allowing selective introduction of
the quinoxaline-2-carbonyl chromophore, with subse-
quent deprotection of the L-valine R-amino group allowing
continuation of depsipeptide chain assembly. L-Valine
was protected as its NR-Dde derivative 13 (Scheme 2).
The carboxyl group of commercially available NR-Fmoc,
side-chain O-trityl D-serine was protected as its benzyl
ester derivative 14, with subsequent removal of the
O-trityl protection by mild acidolysis. DIC-mediated
condensation of the carboxyl of the protected L-valine 13
and the free hydroxyl of the protected D-serine residue
15, followed by catalytic hydrogenation, gave the desired
didepsipeptide building block 17 in good overall yield.
Preparative HPLC also allowed the isolation of the two
most significant byproducts present in the crude cleavage
1
mixture. Mass spectrometry, as well as extensive H-,
13C-, and 2D-homo- and heteronuclear NMR, identified
these byproducts as the dipeptide chromophore conjugate
19 (8 mg isolated; 3% of crude product), in which the
D-serine side chain had undergone dehydration to dehy-
droalanine (Dha) and, unexpectedly, a mono-Acm-pro-
tected linear precursor 20 (20 mg isolated; 6% of crude
product). Careful examination of the TOCSY and NOESY
The building block was incorporated smoothly into the
growing depsipeptide chain using the standard coupling
conditions for peptide chain assembly. After NR-Fmoc
deprotection and chromophore coupling, the NR-Dde
group was removed using 2% hydrazine hydrate in DMF
(3 × 3 min) and the depsipeptide assembly continued.
1
Unfortunately, HPLC analysis of the crude peptide
product after removal from the solid support demon-
strated the presence of many uncharacterized byproducts
in addition to the desired product. It is likely that the
depsipeptide bonds were not completely stable to the very
mild conditions used for NR-Dde removal. This was
unexpected since previous work in our laboratory (solid
NMR spectra in conjunction with the assigned H- and
13C-spectra identified the midchain cysteine, rather than
(17) (a) Fujii, N.; Otaka, A.; Funakoshi, S.; Bessho, K.; Watanabe,
T.; Akaji, K.; Yajima, H. Chem. Pharm. Bull. 1987, 35, 2339-2347.
(b) Albericio, F.; Hammer, R. P.; Garcia-Echeverria, C.; Molins, M. A.;
Chang, J. L.; Munson, M. C.; Pons, M.; Giralt, E.; Barany, G. Int. J.
Pept. Protein Res. 1991, 37, 402-413.
7658 J. Org. Chem., Vol. 70, No. 19, 2005

Total Synthesis of the Bisintercalator TANDEM
SCHEME 3a
cyclic depsipeptide 21 in 59% yield (typically 40-59%
yield) after preparative HPLC purification. Increasing the
reaction time (up to 48 h) did not significantly increase
the yield of the macrolactamization step. Head-to-tail
cyclization was also performed using EDC (5 equiv) and
HOAt (5 equiv) under high dilution conditions in THF/
DMF 4:110 for 24 h, but resulted in lower yields (up to
32%) of bis-Acm cyclic depsipeptide 21.
Treatment of the protected cyclic depsipeptide (2 mM
in abs. MeOH, 1 h, rt) with iodine (10 equiv) resulted in
concomitant removal of the Acm protecting groups and
disulfide bond formation affording TANDEM 4 in es-
sentially quantitative yield. This is an overall yield of
18% (based on manufacturer’s resin loading) and repre-
sents the most efficient synthesis of the synthetic bisin-
tercalator reported to date.
DNA Binding and Sequence Selectivity of TAN-
DEM and Intermediates. Unlike the natural products
triostin A and echinomycin, TANDEM 4 has been shown
to bind to DNA with a preference for 5′-TpA sequences,
particularly 5′-ATAT.9 This change in sequence selectiv-
ity on going to a synthetic agent lacking the tertiary
amides of the natural products appears to be unique even
among the cyclic depsipeptides. Thiocoraline, the latest
cyclic depsipeptide to be discovered (actually a thiodep-
sipeptide), has only relatively nonsequence-selective DNA
binding even though it contains similar tertiary amides
to echinomycin.5 The potential to exploit the sequence
selectivity of triostin A and TANDEM in the design of
potential DNA-reading agents remains largely unex-
plored despite extensive studies of the natural products.
It was germane, therefore, to investigate the potential
DNA interactions of the intermediates of the synthesis,
as well as to establish that our synthetic TANDEM is
similar to the previously described agent in its DNA
binding and sequence selectivity. The sequence selectivity
of TANDEM and its close derivatives has previously been
determined by DNase I footprinting,9,19 and representa-
tive footprinting gels showing the interaction with this
synthetic TANDEM are shown in Figure 3. The left-hand
panel shows the interaction with tyrT fragment, used in
the original footprinting studies.19d This shows a single
clear footprint at the sequence 5′-ATAT as indicated, with
a weaker footprint at the 5′-CTAA halfway down the gel,
located at the top of the oligopurine tract. The second
and third panels show the interaction with fragments
MS1 and MS2, which each contain all possible tetra-
nucleotide sequences. The footprinting patterns are
similar to those previously published and show inhibition
of DNase I cleavage at most of the TpA steps, with the
exception of 5′-TTAA as previously noted.9 The final panel
shows the interaction with fragment pAAD1, which
contains different combinations of (A/T)4 sites, separated
by the sequence CGCGCG.20 Again, footprints are ap-
parent at the sequences containing TpA steps (ATAT and
TATA), although there is no interaction with TTAA.9
These results clearly demonstrate that this newly syn-
a Reagents and conditions: (a) TFA/H2O/iPr3SiH (95:2.5:2.5),
rt, 2 h; (b) EDC (5 equiv), HOSu (4 equiv), NMM (1 equiv), THF/
DMF (10:1), 0 °C to rt, 5 h, 59%; (c) I2 (10 equiv), abs. MeOH, rt,
1 h, quant.
the terminal cysteine, as the residue that had undergone
S-Acm deprotection, presumably during TFA-mediated
cleavage.18 The structural assignment was made prima-
rily on the basis of NOEs between the methylene protons
of the Acm group and the NH proton of the valine residue
(Val2) known to be adjacent to the terminal cysteine
(Cys1) and between the Acm methyl group protons and
the âH of Val2.
The linear peptide was then subjected to Olsen’s
methodology for macrolactamization in solution under
high dilution conditions.8 Head-to-tail cyclization of the
pure depsipeptide 18 (1-2 mM, THF/DMF 10:1, 5 h, 0
°C to rt) using EDC (3 equiv) and HOSu (4 equiv) in the
presence of NMM (1 equiv) gave the bis-Acm-protected
(19) (a) Lavesa, M.; Olsen, R. K.; Fox, K. R. Biochem. J. 1993, 289,
605-607. (b) Waterloh, K.; Olsen, R. K.; Fox, K. R. Biochemistry 1992,
31, 6246-6253. (c) Low, C. M. L.; Fox, K. R.; Olsen, R. K.; Waring, M.
J. Nucleic Acids Res. 1986, 14, 2015-2033. (d) Low, C. M. L.; Olsen,
R. K.; Waring, M. J. FEBS Lett. 1984, 176, 414-420.
(20) Abu-Daya, A.; Fox, K. R. Nucleic Acids Res. 1995, 23, 3385-
3392.
(18) Unexpected loss of Cys(Acm) protection during mild acidolytic
cleavage has been previously reported. (a) Singh, P. R.; Rajopadhye,
M.; Clark, S. L.; Williams, N. E. Tetrahedron Lett. 1996, 37, 4117-
4120. (b) Engebretsen, M.; Agner, E.; Sandosham, J.; Fischer, P. M.
J. Pept. Res. 1997, 49, 341-346.
J. Org. Chem, Vol. 70, No. 19, 2005 7659

Malkinson et al.
FIGURE 3. DNase I footprinting patterns showing the interaction of TANDEM 4 with four DNA fragments. The concentration
of 4 (µM) is shown at the top of each gel lane. Tracks labeled “GA” are Maxam-Gilbert markers specific for purines, while “con”
indicates cleavage in the absence of added ligand. The locations of the best binding sites, containing the TpA dinucleotide step,
are indicated by the boxes.
couplings were carried out for 30 min in DMF using a 2.5-fold
excess (over resin loading) of protected amino acid, activated
with an equimolar amount of HBTU and HOBt, in the
presence of a 5-fold excess (over resin loading) of iPr2NEt. Each
coupling was repeated and completion monitored using the
Kaiser test21 for free amines. Unsuccessful couplings were
further repeated until a negative Kaiser test was obtained.
NR-Fmoc protection was removed using 20% (v/v) piperidine
in DMF (2 × 10 min). Introduction of the quinoxaline-2-
carbonyl chromophore was achieved by acylation of the D-
serine R-amino group using a 2.5-fold excess of quinoxaline-
2-carbonyl chloride in the presence of a 5-fold excess of iPr2NEt
(2 × 30 min). D-Serine side-chain O-trityl protection was
removed using 2% (v/v) TFA in dry CH2Cl2 containing 5% (v/
v) iPr3SiH (3 × 5 min; 1 × 20 min). Esterification of the free
D-serine â-hydroxyl was performed in DMF using a 10-fold or
5-fold excess of NR-Fmoc-valine and a 5-fold excess of DIC, in
the presence of a catalytic amount (0.1 equiv with respect to
resin loading) of DMAP (3 × 1 h). Disulfide bond formation
(with simultaneous S-Acm deprotection) was performed on the
solid phase (if required) using a 10-fold excess of iodine in DMF
(1 h), followed by very thorough washing of the resin-bound
depsipeptide with copious DMF and CH2Cl2.
Peptide Cleavage and Isolation. The resin-bound peptide
was washed thoroughly with DMF, CH2Cl2, and then 50% (v/
v) MeOH in CH2Cl2 and dried in vacuo over KOH to constant
weight. The peptide was removed from the solid support by
acidolysis using TFA containing 2.5% (v/v) water and 2.5% (v/
v) iPr3SiH for 2 h at 25 °C. The TFA was removed under
reduced pressure. The crude peptide precipitated, was washed
with cold anhydrous diethyl ether, and then was extracted into
50% (v/v) acetic acid and lyophilized. The crude product was
purified by preparative HPLC (System B). Pure fractions were
pooled and lyophilized.
thesized TANDEM has the same DNA binding proper-
ties, interacting with TpA sites, as previously reported.
In contrast, the bis-Acm-protected cyclic depsipeptide 21
and the disulfide bond-containing linear precursor 12
showed no effect on DNase I cleavage at concentrations
up to 50 µM (data not shown).
Conclusions
The synthesis of TANDEM on solid phase was achieved
by two different routes. Although the first of these routes
involved a final, low-yielding peptide cyclization reaction,
it still represents a route yielding material in amounts
similar to those already published. More gratifyingly, a
second route, giving better yields for the macrolactam-
ization prior to disulfide bond formation, is the most
efficient route to this DNA-binding cyclic peptide de-
scribed to date. The new synthesis gave TANDEM with
DNA binding characteristics essentially identical to those
of the molecule previously synthesized. None of the
intermediate synthetic precursors of TANDEM bind to
DNA with appreciable affinity.
In this article, we have revisited molecules that have
been extensively studied by others but for which struc-
ture-activity studies and analogue structures are few.
The solid-phase-based protocols give us the opportunity
to generate numerous analogues and investigate the
ability of these analogues to bind to DNA and inhibit
either gene or template function on the duplex. The
potential for analogues that may bind to longer sequences
is intriguing, and such new entities are currently under
investigation in our laboratories and will be described
in due course.
Disulfide Bond-Containing Linear Precursor (12).
Disulfide-bridged depsipeptide 12 was synthesized on 0.205
mmol scale (0.50 g of Fmoc-Ala-Wang resin; 0.41 mmol g-1
manufacturer’s substitution). Synthesis, cleavage, isolation,
and HPLC purification (System B) as described earlier afforded
Experimental Section
Peptide Synthesis. Peptide synthesis was accomplished
manually using a stepwise solid-phase procedure. All peptide
(21) Kaiser, E.; Colescott, R. L.; Bossinger, C. D.; Cook, P. I. Anal.
Biochem. 1970, 34, 595.
7660 J. Org. Chem., Vol. 70, No. 19, 2005

Total Synthesis of the Bisintercalator TANDEM
the desired depsipeptide 12 as a white solid (65 mg; 31% based
on manufacturer’s resin loading). ES MS (C46H56N12O13S2)
1048.35 m/z (%): 1049.4 [M + H]+ (100), 1071.5 [M + Na]+
4.80 (dd, 2H, âHb DSer), 4.98 (m, 2H, RH DSer), 7.93 (t, 2H,
Qxc CHAr), 7.99 (t, 2H, Qxc CHAr), 8.14 (d, 2H, Qxc CHAr),
8.18 (d, 2H, CHAr), 8.19 (d, 2H, NH Val), 8.55 (d, 4H, NH Ala
and NH Cys), 8.62 (t, 2H, NH Acm), 9.19 (d, 2H, NH Ser),
9.49 (s, 2H, Qxc H-3).
1
(60). RP-HPLC (System A): Rt ) 14.58 min. H NMR (CD3-
OH): δ 0.89 (d, 3H, γH Val6), 0.92 (d, 3H, γH Val6), 1.05 (d,
3H, γH Val 2), 1.10 (d, 3H, γH Val 2), 1.41 (d, 3H, âH Ala4),
1.45 (d, 3H, âH Ala8), 2.18 (m, 1H, âH Val6), 2.36 (m, 1H, âH
Val2), 3.06 (dd, 1H, âHa Cys5), 3.10 (dd, 1H, âHa Cys1), 3.25
(dd, 1H, âHb Cys5), 3.50 (dd, 1H, âHb Cys1), 4.31 (dd, 1H, RH
Cys1), 4.36 (dd, 1H, RH Ala4), 4.41 (m, 1H, RH Val6), 4.44
(m, 1H, RH Val2), 4.48 (m, 1H, RH Ala8), 4.59 (dd, 1H, âHa
DSer3), 4.65 (d, 2H, âH DSer7), 4.74 (dd, 1H, âHb DSer3), 4.90
(m, 1H, RH Cys5), 5.03 (m, 1H, RH DSer3), 5.11 (m, 1H, RH
DSer7), 7.98-8.03 (m, 4H, Qxc CHAr), 8.04 (d, 1H, NH Val6),
8.20-8.28 (m, 4H, Qxc CHAr), 8.32 (d, 1H, NH Ala4), 8.63 (d,
1H, NH Ala8), 8.67 (d, 1H, NH Val2), 8.70 (d, 1H, NH Cys5),
9.11 (d, 1H, NH DSer3), 9.21 (d, 1H, NH DSer7), 9.56 (s, 1H,
Qxc H-3), 9.58 (s, 1H, Qxc H-3).
TANDEM (4). (Method A). To a solution of disulfide-bridged
depsipeptide 12 (35 mg, 33 µmol) and HOBt (26 mg, 0.17
mmol) in 40 mL of dry CH2Cl2/DMF 7:1 (v/v) was added EDC
(32 mg, 0.17 mmol), and the reaction mixture was stirred for
24 h at 25 °C under N2. The solvents were removed under
reduced pressure, and the residue was taken up in chloroform
(ca. 30 mL) and washed three times with water (ca. 10 mL)
and twice with saturated NaHCO3 solution (ca. 10 mL). The
solution was dried (MgSO4), filtered, and evaporated under
reduced pressure. The crude product was purified by flash
column chromatography on silica gel, using CHCl3/MeOH 19:1
(v/v) as the mobile phase. Pure fractions were pooled and
evaporated, and the residue lyophilized from 60% (v/v) MeCN
in water to afford TANDEM (4) as a white solid (3.9 mg, 11%).
MALDI-TOF MS (C46H54N12O12S2) 1030.34 m/z (%): 1031 [M
+ H]+ (100). RP-HPLC (System A): Rt ) 17.03 min. HRMS
(ES) calcd for C46H55N12O12S2 [M + H]+ 1031.3499, found
Bis-Acm-Protected Linear Precursor (18). Bis-Acm-
protected linear depsipeptide 18 was synthesized on 0.41 mmol
scale (1.00 g of Fmoc-Ala-Wang resin; 0.41 mmol g-1 manu-
facturer’s substitution). Synthesis, cleavage, isolation, and
purification as described above (omitting the solid-phase
disulfide formation step) afforded the desired protected dep-
sipeptide 18 as a white solid (149 mg; 30% based on manu-
facturer’s resin loading). ES MS (C52H68N14O15S2) 1192.44 m/z
(%): 1193.3 [M + H]+ (100), 1215.5 [M + Na]+ (46). RP-HPLC
(System A): Rt ) 13.62 min. 1H NMR (CD3OH): δ 0.83 (d,
3H, γH Val6), 0.85 (d, 3H, γH Val6), 0.93 (d, 3H, γH Val 2),
0.97 (d, 3H, γH Val 2), 1.41 (d, 3H, âH Ala4), 1.45 (d, 3H, âH
Ala8), 1.98 (s, 3H, CH3 Acm5), 2.02 (s, 3H, CH3 Acm1), 2.12
(m, 1H, âH Val6), 2.26 (m, 1H, âH Val2), 2.76 (dd, 1H, âHa
Cys1), 2.80 (dd, 1H, âHa Cys5), 3.07 (dd, 1H, âHb Cys5), 3.17
(dd, 1H, âHb Cys1), 4.09 (dd, 1H, CHa Acm1), 4.23-4.27 (m,
2H, CHa Acm5 and RH Val6), 4.36 (dd, 1H, RH Cys1), 4.40-
4.51 (m, 4H, CHb Acm5, RH Ala8, RH Ala4 and RH Val2),
4.60-4.73 (m, 5H, âH DSer7, âH DSer3 and RH Cys5), 4.77
(dd, 1H, CHb Acm1), 5.02 (m, 1H, RH DSer3), 5.09 (m, 1H, RH
DSer7), 7.92 (d, 1H, NH Val6), 7.95-8.01 (m, 4H, Qxc CHAr),
8.19-8.27 (m, 6H, NH Val2, NH Cys5 and Qxc CHAr), 8.55
(t, 1H, NH Acm5), 8.60 (d, 1H, NH Ala8), 8.68 (d, 1H, NH
Ala4), 8.75 (t, 1H, NH Acm1), 9.20 (d, 1H, NH DSer7), 9.28 (d,
1H, NH DSer3), 9.54 (s, 1H, Qxc H-3), 9.54 (s, 1H, Qxc H-3).
Bis-Acm-Protected Cyclic Depsipeptide (21). To a solu-
tion of bis-Acm-protected linear precursor 18 (24.2 mg; 20.3
µmol) in dry DMF (1 mL) at 0 °C was added N-methylmor-
pholine (NMM, 2.2 µL, 20.3 µmol, 1 equiv) and N-hydroxysuc-
cinimide (HOSu, 9.3 mg, 81.2 µmol, 4 equiv). The reaction
mixture was diluted with dry THF (10 mL) and cooled to 0
°C. EDC (12 mg, 60.9 µmol, 3 equiv) was added, and the
reaction was stirred at 0 °C under N2 for 1 h, then at room
temperature until completion (4 h, by HPLC). The solvents
were removed under reduced pressure (below 40 °C), and the
residue was taken up in ca. 60% aqueous MeCN and lyophi-
lized. Preparative HPLC purification (System B) afforded the
desired cyclic depsipeptide as a white solid (14.0 mg; 59%).
ES MS (C52H66N14O14S2) 1174.43 m/z (%): 1175.2 [M + H]+
(47), 1197.1 [M + Na]+ (100). RP-HPLC (System A): Rt ) 15.63
min. 1H NMR (CD3OH): δ 0.93 (d, 12H, γH Val), 1.51 (d, 6H,
âH Ala), 1.97 (s, 6H, CH3 Acm), 2.35 (m, 2H, âH Val), 3.02
(dd, 2H, âHa Cys), 3.13 (dd, 2H, âHb Cys), 4.33 (dd, 2H, CHa
Acm), 4.39 (dd, 2H, CHb Acm), 4.43 (m, 2H, RH Val), 4.49 (quin,
2H, RH Ala), 4.63 (dd, 2H, âHa DSer), 4.74 (q, 2H, RH Cys),
1
1031.3517. H NMR: δ 1.11 (d, 6H, γH Val), 1.15 (d, 6H, γH
Val), 1.37 (d, 6H, âH Ala), 2.55 (m, 2H, âH Val), 2.92 (d, 4H,
âH Cys), 4.51 (quin, 2H, RH Ala), 4.65 (dd, 2H, âHa DSer), 4.84
(dd, 2H, RH Val), 4.89 (m, 2H, RH DSer), 5.01 (dd, 2H, âHb
DSer), 5.68 (dd, 2H, RH Cys), 6.41 (d, 2H, NH Ala), 7.13 (d,
2H, NH Cys), 7.92 (t, 2H, Qxc H-7), 7.95 (t, 2H, Qxc H-6), 8.12
(d, 2H, Qxc H-5), 8.25 (d, 2H, Qxc H-8), 8.67 (d, 2H, NH Val),
8.84 (d, 2H, NH DSer), 9.66 (s, 2H, Qxc H-3). 13C NMR: δ 17.9
(γC Val), 18.2 (âC Ala), 18.9 (γC Val), 32.2 (âC Val), 46.2 (âC
Cys), 49.6 (RC Ala), 53.4 (RC Cys), 54.7 (RC DSer), 57.6 (RC
Val), 65.1 (âC DSer), 129.0 (Qxc C-8), 129.7 (Qxc C-5), 131.7
(Qxc C-7), 132.8 (Qxc C-6), 140.6 (Qxc C-8a), 142.4 (Qxc C-2),
142.8 (Qxc C-3), 143.2 (Qxc C-4a), 164.0 (Qxc CO), 167.8 (CO
DSer), 170.6 (CO Cys), 171.0 (CO Val), 173.0 (CO Ala).
(Method B). A solution of iodine (35.7 mg, 0.14 mmol, 10
equiv) in absolute MeOH (4.5 mL) was added dropwise over
ca. 60 min to a solution of bis-Acm-protected cyclic depsipeptide
21 (16.5 mg, 14.1 µmol) in absolute MeOH (3 mL). The reaction
was complete (HPLC) after addition of the iodine. The reaction
mixture was cooled to 0 °C, and 1 M aqueous Na2S2O3 solution
was slowly added dropwise until the solution became colorless.
Water was added until no further precipitation was noted (ca.
20 mL), and the precipitate was filtered and washed with
water. The precipitate was extracted into 90% aqueous MeCN,
and the solution was diluted with a little water and lyophilized
to afford TANDEM (4) as an HPLC and NMR pure white
powder (14.5 mg, quantitative).
Acknowledgment. We thank the University of
London for a Maplethorpe Teaching Fellowship (J.P.M.),
ULSOP for funding, and the London School of Pharmacy
for NMR, mass spectral, and elemental analysis facili-
ties.
Supporting Information Available: General experimen-
tal procedures, synthesis, and analytical details for compounds
13-17, analytical details for compounds 19 and 20, and
footprinting protocols. This material is available free of charge
via the Internet at http://pubs.acs.org.
JO050959A
J. Org. Chem, Vol. 70, No. 19, 2005 7661