Solid-phase synthesis of dysidiolide-derived protein phosphatase inhibitors

Article abstract of DOI:10.1021/ja027609f

Biologically active natural products can be regarded as evolutionary selected and biologically validated starting points in structural space for the development of compound libraries. For libraries designed and synthesized around a given natural product,

Full text of DOI:10.1021/ja027609f

Published on Web 10/10/2002
Solid-Phase Synthesis of Dysidiolide-Derived Protein
Phosphatase Inhibitors
Dirk Brohm, Nicolas Philippe, Susanne Metzger, Ajay Bhargava, Oliver Mu¨ller,
Folker Lieb, and Herbert Waldmann*
Contribution from the Max-Planck-Institut fu¨r molekulare Physiologie, Abteilung Chemische
Biologie, Otto-Hahn-Strasse 11, D-44227 Dortmund, Germany and UniVersita¨t Dortmund,
Fachbereich 3, Organische Chemie, Dortmund, Germany
Received July 9, 2002
Abstract: Biologically active natural products can be regarded as evolutionary selected and biologically
validated starting points in structural space for the development of compound libraries. For libraries designed
and synthesized around a given natural product, a higher hit rate and the identification of biologically relevant
hits can be expected, justifying a probably higher investment in the development of the corresponding
syntheses. This approach requires the development of complex multistep reaction sequences on the solid
phase. Employing the protein phosphatase Cdc25 inhibitor dysidiolide as an example, we demonstrate
that this goal can be achieved successfully. The reaction sequences developed led to dysidiolide analogues
in overall 8-12 linear steps with the longest sequence on the solid support amounting to up to 11 sequential
transformations. The desired products were obtained in overall yields ranging from 6% to 27% and in
multimilligram amounts starting from 100 mg of resin. The transformations applied include a variety of very
different reaction types widely used in organic synthesis (i.e., an asymmetric cycloaddition employing a
removable chiral auxiliary, different organometallic transformations, olefination reactions, different oxidation
reactions, acidic hydrolyses, and a nucleophilic substitution). Biological investigation of the eight dysidiolide
analogues synthesized showed that they inhibit Cdc25C in the low micromolar range with the IC50 value
varying by a factor of 20 and that they display considerable and differing biological activities in cytotoxicity
assays employing different cancer cell lines.
Introduction
Paramount to the success of this approach is that efficient
and reliable methods and multistep sequences for the total
The combinatorial synthesis of compound libraries on poly-
meric supports is at the heart of protein ligand and inhibitor
discovery, in particular for research in medicinal chemistry and
chemical biology. To achieve high efficiency in this process,
powerful strategies for the design of compound libraries are of
paramount importance.
A key to the efficient discovery of new ligands and inhibitors
for known and, in particular, for newly discovered proteins is
to identify compound classes already validated as being biologi-
cally relevant and to employ them as starting points in structural
space for library development. Libraries designed and synthe-
sized around the basic structure of such compounds should yield
modulators of protein activity with high hit rates.
The prerequisite of biological prevalidation is fulfilled by
biologically active natural products that can be regarded as
chemical entities evolutionarily selected and validated for
binding to particular protein domains.1 Consequently, the
underlying structural architectures of such natural products may
provide powerful guiding principles for library development.1-3
synthesis of natural products and analogues thereof on polymeric
supports are available. The corresponding transformations must
proceed with a degree of selectivity and robustness typical of
related classical solution-phase transformations, irrespective of
the stringencies and differing demands imposed by the presence
of and the anchoring to the polymeric support. However,
currently this challenge to organic chemistry is nearly unmet.
In a few cases structural variation of natural products via solid-
phase methodologies has been achieved,1,3 mostly in the late
steps of the syntheses and via modification of a core structure
presynthesized in solution. But the feasibility of natural product
and analogue total synthesis in long multistep sequences (e.g.,
10 steps and more) on polymeric supports has only been
achieved in a single case.4
(3) For reviews on compound library synthesis based on natural products,
see: (a) Hall, D. G.; Manku, S.; Wang, F. J. Comb. Chem. 2001, 3, 125-
150. (b) Handbook of Combinatorial Chemistry; Nicolaou, K. C., Hanko,
R., Hartwig, W., Eds.; Wiley-VCH: Weinheim, Germany, 2002. (c) Arya,
P.; Joseph, R.; Chou, D. T. H. Chem. Biol. 2002, 9, 145-156.
(4) (a) Nicolaou, K. C.; Winssinger, N.; Pastor, J.; Ninkovic, S.; Sarabia, F.;
He, Y.; Vourloumis, D.; Yang, Z.; Li, T.; Giannakakou, P.; Hamel, E.
Nature 1997, 387, 268-272. (b) Nicolaou, K. C.; Vourloumis, D.; Li, T.;
Pastor, J.; Winssinger, N.; He, Y.; Ninkovic, S.; Sarabia, F.; Vallberg, H.;
Roschangar, F.; King, N. P.; Finlay, M. R. V.; Giannakakou, P.; Verdier-
pinard, P.; Hamel, E. Angew. Chem. 1997, 109, 2181-2187; Angew. Chem.,
Int. Ed. Engl. 1997, 36, 2097-2103.
* Corresponding
herbert.waldmann@pi-dortmund.mpg.de.
(1) Breinbauer, R.; Vetter, I.; Waldmann, H. Angew. Chem., Int. Ed. 2002,
41, 2878-2890.
(2) Review: Mann, J. Nature ReV. Cancer 2002, 2, 143-148.
author.
Fax:
*49/231/133-2499.
E-mail:
9
10.1021/ja027609f CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 13171-13178
13171

A R T I C L E S
Brohm et al.
induces growth arrest of different cancer cell lines and arrest in
the G1 phase of the cell cycle or apoptosis.13,14
Because of these biological properties and its unique structure,
dysidiolide has been of intense interest to biologists, pharma-
cologists, and organic chemists,15 resulting in the completion
of five successful total syntheses since its discovery.
Results and Discussion
Planning of the Solid-Phase Synthesis. Dysidiolide is a
sesquiterpene with a decalin-type framework that carries a
lipophilic side chain terminating in an olefin and a hydrophilic
side chain incorporating an alcohol and a γ-hydroxybutenolide.
For the planned solid-phase synthesis, we intended to attach
the olefinic side chain to a robust linker that would provide the
terminal alkene structure after cleavage from the solid support
in a mild and traceless manner. The new olefin metathesis
linker4,16 incorporated in 2 was thought to fulfill these require-
ments (Scheme 1). Transition-metal catalyzed olefin metathesis
should liberate the desired compound accompanied by formation
of a polymer bound cyclopentene (step I, Scheme 1). Attachment
to the polymer via an ether bond was thought to provide the
required stability during the planned multistep synthesis. The
γ-hydroxybutenolide moiety should be obtained by addition of
3-lithiofuran to an aldehyde group and subsequent oxidation of
the heterocycle with singlet oxygen (step II, Scheme 1). For
the generation of the bicyclic core structure, it was planned to
apply the Diels-Alder route (step III, Scheme 1) previously
investigated in our laboratories15f and successfully employed
in total syntheses of natural dysidiolide in solution.15b-e The
knowledge gleaned from these studies suggested that this
cycloaddition primarily leads to the 6-epimer of the dysidiolide
framework. However, we chose to accept this deviation from
the goal of synthesizing the parent natural product itself, since
in the concept delineated above this is not necessarily required
to identify new biologically active compounds. It was planned
to synthesize the diene for the Diels-Alder reaction in solution
starting from commercially available chiral ketoester 8 and
attach it to the linker resin 5 carrying an aldehyde group by a
Wittig reaction with ylide 6 (step IV) in a convergent strategy
(4 w 5 + 6).
Figure 1. Activation of Cdk2-cyclin E complex by Cdc25A as essential
step for G1-S transition.
Here we describe the solid-phase synthesis of a close analogue
of the protein phosphatase inhibitor dysidiolide and of a small
library of dysidiolide analogues that rapidly yielded compounds
with significantly improved biological activity.5
The sesquiterpenoid dysidiolide 1 is a naturally occurring
inhibitor of the dual-specificity Cdc25 protein phosphatase
family that plays a crucial role in the regulation of the cell cycle.6
The Cdc25 phosphatases activate cyclin-dependent kinases
(Cdks) and thereby initiate progression of cells through different
phases of the cell cycle. For instance, Cdc25A activates the triply
phosphorylated complex between Cdk2 and cyclin E by
dephosphorylation of Cdk2 at Thr 14 and Tyr 15 (Figure 1).
Subsequently, the cells can progress beyond the G1/S check-
point.7 Similarly, Cdc25B and Cdc25C are thought to be
regulators of the G2/M transition through dephosphorylation and
activation of the Cdk1-cyclin B complex.8,9
The crucial roles of the Cdc25 dual specificity phosphatases
in cell-cycle regulation led to the notion that these enzymes
might be promising targets for the development of new
anticancer drugs.10,11 This prospect spurred studies aiming at
the development of selective Cdc25 inhibitors.12
Aldehyde resin 5 should be accessible from diol 717 by
coupling to Merrifield resin via Williamson ether formation and
subsequent oxidation of the remaining primary alcohol to the
aldehyde (step V, Scheme 1). The diene unit in 6 was traced
back to ketone 8 via addition of a vinyl Grignard to 8 followed
by elimination of the resulting tertiary alcohol. Formation of
the ylide was envisaged to be feasible by means of a nucleophilic
Among these, dysidiolide was identified as particularly
promising, since it inhibits Cdc25A with an IC50 value of 9.4
µM, whereas the phosphatases calcineurin, CD45, and LAR are
not inhibited by the natural product.13 In addition, dysidiolide
(5) Part of this work was published in preliminary form: Brohm, D.; Metzger,
S.; Bhargava, A.; Mu¨ller, O.; Lieb, F.; Waldmann, H. Angew. Chem., Int.
Ed. 2002, 41, 307-311.
(14) Danishefsky, S.; Magnuson, S. R.; Rosen, N. Patent WO 99/40079, 1999.
(15) (a) Corey, E. J.; Roberts, B. E. J. Am. Chem. Soc. 1997, 119, 12425-
12431. (b) Boukouvalas, J.; Cheng, Y.-X.; Robichaud, J. J. Org. Chem.
1998, 63, 228-229. (c) Magnuson, S. R.; Sepp-Lorenzino, L.; Rosen, N.;
Danishefsky, S. J. J. Am. Chem. Soc. 1998, 120, 1615-1616. (d) Takahashi,
M.; Dodo, K.; Hashimoto, Y.; Shirai, R. Tetrahedron Lett. 2000, 41, 2111-
2114. (e) Jung, M. E.; Nishimura, N. Org. Lett. 2001, 3, 2113-2115. (f)
Brohm, D.; Waldmann, H. Tetrahedron Lett. 1998, 39, 3995-3998. (g)
Miyaoka, H.; Kajiwara, Y.; Yamada, Y. Tetrahedron Lett. 2000, 41, 911-
914. (h) Piers, E.; Caille´, S.; Chen, G. Org. Lett. 2000, 2, 2483-2486. (i)
Demeke, D.; Forsyth, C. J. Org. Lett. 2000, 2, 3177-3179. (j) Paczkowski,
R.; Maichle-Mo¨ssmer, C.; Maier, M. E. Org. Lett. 2000, 2, 3967-3969.
(k) Takahashi, M.; Dodo, K.; Sugimoto, Y.; Aoyagi, Y.; Yamada, Y.;
Hashimoto, Y.; Shirai, R. Bioorg. Med. Chem. Lett. 2000, 10, 2571-2574.
(16) For the use of traceless linkers based on olefin metathesis see: (a) Peters,
J.-U.; Blechert, S. Synlett 1997, 348-350. (b) Knerr, L.; Schmidt, R. R.;
Synlett 1999, 1802-1804.
(6) Draetta, G.; Eckstein, J. Biochim. Biophys. Acta 1997, 1332 (2), M53-
M63.
(7) Hoffmann, I.; Draetta, G.; Karsenti, E. EMBO J. 1994, 13, 4302-4310.
(8) Nilsson, I.; Hoffmann, I. Prog. Cell Cycle Res. 2000, 4, 107-114.
(9) Millar, J. B.; Blewitt, J.; Gerace, L.; Sadku, K.; Featherstone, C.; Russell,
P. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 10500-10504.
(10) Eckstein, J. W. Gene Ther. Mol. Biol. 1998, 1, 707-711.
(11) Zhang, Z.-Y. Curr. Opin. Chem. Biol. 2001, 5, 416-423.
(12) (a) Cebula, R. E.; Blanchard, J. L.; Boisclair, M. D.; Pal, K.; Bockovich,
N. Bioorg. Med. Chem. Lett. 1997, 7, 2015-2020. (b) Dodo, K.; Takahashi,
M.; Yamada, Y.; Sugimoto, Y.; Hashimoto, Y.; Shirai, R. Bioorg. Med.
Chem. Lett. 2000, 10, 615-617. (c) Peng, H.; Zalkow, L. H. J. Med. Chem.
1998, 41, 4677-4680. (d) Peng, H.; Xie, W.; Kim, D.-I.; Zalkow, L. H.;
Powis, G.; Otterness, D. M.; Abraham, R. T. Bioorg. Med. Chem. 2000, 8,
299-306. (e) Peng, H.; Otterness, D. M.; Abraham, R. T.; Zalkow, L. H.
Tetrahedron 2001, 57, 1891-1896.
(13) Gunasekera, S. P.; McCarthy, P. J.; Kelly-Borges, M.; Lobkovsky, E.;
Clardy, J. J. Am. Chem. Soc. 1996, 118, 8759-8760.
(17) Samu, E.; Huszthy, P.; Somogyi, L.; Hollo´si, M. Tetrahedron: Asymmetry
1999, 10, 2775-2795.
9
13172 J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002

Synthesis of Protein Phosphatase Inhibitors
A R T I C L E S
Scheme 1. Retrosynthetic Analysis of Dysidiolide and Delineation
of a Solid-Phase Synthesis Strategy Employing an Olefin
Metathesis Linker
Scheme 2. Synthesis of the Dienea
a Key: (a) LAH, THF, 0 °C, 30 min, 100%; (b) TBDPS-Cl, DMAP,
Et3N, dichloromethane, r.t., 17 h; (c) TPAP, NMO, MS 4 Å, dichloro-
methane, r.t., 4 h, 85% (b-c); (d) vinylmagnesium bromide, THF, r.t., 5 h;
(e) CuSO4‚5 H2O, benzene, ∆, 40 h, 58% (d-e); (f) (nBu)4NF (TBAF),
THF, r.t., 3 h, 95%; (g) iodine, imidazole, PPh3, dichloromethane, r.t., 1 h,
85%.
Scheme 3. Synthesis of the Linker and Attachment to the Resina
substitution reaction employing an alkyl iodide obtainable from
ester 8 (step IV, Scheme 1).
Synthesis of the Diene Building Block. Dienyl iodide 13
was synthesized from commercially available ketoester 8 in
seven steps and with an overall yield of 40% (Scheme 2). After
reduction of 8 with LiAlH4, the primary hydroxyl group of 9
was selectively masked as TBDPS-ether, and then the secondary
alcohol was reoxidized with TPAP/NMO to yield masked ketone
10 in high yield. Vinylmagnesium chloride was added to the
carbonyl group, and the resulting tertiary alcohol was converted
into diene 11 by Lewis acid mediated elimination of water. The
best results were obtained with CuSO4 in benzene or BF3‚OEt2
in benzene/THF ) 4:1. Then the primary alcohol was liberated
again and transformed into the iodide required for the subsequent
Wittig reaction by means of iodine, PPh3, and imidazole.
Synthesis and Preliminary Investigation of the Linker
Group. The development of the linker group commenced with
addition of diallyl zinc (generated from allylmagnesium chloride
and ZnBr2) to R,â-unsaturated lactone 14 in the presence of
TMSCl as a Lewis acid and anion scavenger. Lactone 15 was
a
Key: (a) allylmagnesium bromide, ZnBr2, TMS-Cl, THF, -78 °C, 6
h, 52%; (b) LAH, THF, r.t., 1 h, 97%; (c) NaH, (nBu)4NI, benzyl bromide,
DMF, r.t., 12 h, 58%; (d) NaH, (nBu)4NI, Merrifield-Cl resin (1.1 mmol/
g), DMF, r.t., 18 h, 59%; (e) dihydropyrane, Dowex 50 Wx2, toluene, r.t.,
3 h, 95%; (f) NaH, (nBu)4NI, Merrifield-Cl resin (1.1 mmol/g), DMF, r.t.,
18 h; (g) pyridinium-p-toluene sulfonate, EtOH, dichloroethane, ∆, 18 h,
98% (f-g); (h) solution: (nPr)4N RuO4 (TPAP), NMO, MS4 Å, r.t., 30
min, 83%; solid-phase: o-iodoxybenzoic acid (IBX), THF/DMSO ) 1:1,
r.t., 8 h, 92%; (i) EtPPh3I, nBuLi, 13, THF, r.t., 16-20 h, then nBuLi, 0
°C, 1-2 h; (j) (PCy3)2(Cl)2RudCHPh (21) (2 × 10 mol %), dichloro-
methane, r.t., 16 h.
obtained under optimized conditions from the intermediary
formed silylenol ether (identified by GCMS analysis of the crude
reaction mixture) by means of acidic hydrolysis in 51% yield
(Scheme 3). Application of a Cu-catalyzed addition of the
Grignard reagent led to oligomerization of the unsaturated
lactone, and also the Sakurai reaction employing allyltri-
methylsilane and TiCl4 was not successful.
9
J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002 13173

A R T I C L E S
Brohm et al.
Reduction of lactone 15 with LiAlH4 yielded diol 7 in
multigram amounts. To investigate whether the linker group
could be used as planned, diol 7 was converted into monobenzyl
ether 17, and in a second experiment it was attached to
Merrifield polystyrene resin (loading 1.1 mmol/g). In the
solution-phase reaction formation of the mono- and the bis-
benzyl ether was observed. On the solid-phase derivatization
of the introduced primary alcohol groups with Fmoc-Cl,
cleavage of the Fmoc group with piperidine and UV-spectro-
scopic determination of the amount of the fluorenylmethyl
piperidine formed18 indicated that only 59% loading was
achieved. However, gravimetric determination of the loading
indicated quantitative attachment of the linker. Thus, direct
alkylation of diol 7 with Merrifield chloride leads to partial
cross-linking. This is in marked contrast with observations
reported for the attachment of 1,4-butanediol to Merrifield resin,
which proceeds with nearly quantitative yield.19
%) were employed (Scheme 3). Within 1 h conversion was
complete (GCMS control), and cyclopentene 22 and triene 24
were isolated in 99% and 81% yield, respectively.
In contrast to these findings, the desired metathesis product
24 was obtained under the same conditions from resin-bound
intermediate 4 in only 40% yield (for coupling to the support
and release of the metathesis product). Longer reaction times,
elevated temperature, or use of 2 × 10 mol % of the catalyst
resulted in only a moderate increase of yield to 60%. Also
performing the reaction in an ethylene atmosphere22 or with
additional styrene23 to release product from the intermediary
formed ruthenium carbene complex did not result in higher yield.
In addition, the metathesis reaction was performed with the new
Grubbs catalyst (benzylidene-(1,3-dimesityl-4,5-dihydroimid-
azol-2-ylidene)-(tricyclohexylphosphin)-ruthenium dichloride)
that is well-known to have a higher reactivity, especially for
highly substituted double bonds.21b However, the yield was even
lower (38%) as compared to that of the reaction with 21 under
the same conditions.
To avoid this undesired cross-linking, diol 7 was converted
into mono-THP ether 16, which was then attached to the
polymeric support. After cleavage of the THP group with
pyridinium-tosylate (PPTS), linker resin 18 was obtained in
nearly quantitative yield and with loading levels up to 1 mmol/
g. Alcohols 17 and 18 were then oxidized to aldehydes 19 and
5. Oxidation on the solid support proceeded best with IBX.20
Use of the Dess-Martin reagent, the SO3-pyridine complex
in DMSO/THF, or Swern oxidation gave inferior results.
We speculated that these results might be due to partial cross-
linking of neighboring allylic side chains of the linker by cross-
metathesis on the solid support preventing release of the target
compound. To investigate this possibility and to identify possible
dimers formed by cross-metathesis on the resin, the synthesis
was repeated employing a double linker that would allow release
of the dimers under mild conditions. To this end, an acid-
sensitive tetrahydropyranyl linker was introduced (i.e., diol 7
was selectively masked as mono-TBDPS ether) and coupled to
Ellman’s dihydropyran resin24 as shown in Scheme 4.
For quantitative determination of the formed aldehyde a new
assay was developed. A defined amount of the aldehyde resin
was treated with a stock solution of dinitrophenylhydrazine
(DNPH), leading to partial consumption of the hydrazine by
hydrazone formation. DNPH consumption is then readily
determined by UV spectrometric quantitation of the remaining
DNPH and comparison with the concentration of the stock
solution.
After removal of the TBDPS group from 26 the synthesis
sequence described above was carried out, yielding resin 29.
Loading of this polymer with diene was determined by the
release of alcohol 30 from a resin sample under acidic
conditions. Surprisingly, subjection of intermediate 29 to the
ring-closing metathesis conditions described above resulted in
the release of triene 24 in 82% yield (i.e., with nearly the same
result observed for the corresponding reaction in solution (see
above)). Upon subsequent treatment of the resin with PPTS in
dichloroethane-ethanol, cyclopentene 32 was isolated in 84%
yield. In addition, several intermediates that had not been
converted completely were identified; however, products arising
from undesired cross-linking by competing metathesis between
neighboring linker groups could not be detected. These findings
demonstrate that cross-linking most probably is not the reason
for the observed yields of 40-60% recorded in the metathesis
employing resin 4. This conclusion is further supported by the
result that the yield does not increase if a resin with a lower
loading level is used.
Diene 13 was coupled with aldehyde 5 and 19 in a two-step
process (Scheme 3). First, the diene was treated with the ylide
formed from ethyltriphenylphosphonium iodide and n-butyl-
lithium to form phosphonium salt 6. Deprotonation of this
intermediate with a second equivalent of base and the subsequent
Wittig reaction with 5 and 19 yielded alkenes 4 and 20. Soluble
olefin 20 was obtained in 90% yield as an E/Z mixture, and the
IR spectrum of resin 4 did not show the strong CdO bond
characteristic of the linker anymore, thus indicating a high
degree of conversion.
The decisive ring-closing metathesis reaction, including a
triply substituted double bond, proceeded rapidly and in high
yield if soluble diene 20 and the Grubbs catalyst21a 21 (10 mol
Model Synthesis in Solution. To explore whether the planned
synthesis sequence for the construction of the dysidiolide
framework is feasible, it was investigated in solution employing
achiral model diene 33 (Scheme 5). To this end, instead of
ethylaluminum dichloride, which had proven very useful in our
model studies of the central cycloaddition,15f TMS-triflate
introduced by Danishefsky et al. for related transformations15c
(18) Gordeev, M. F.; Luehr, G. W.; Hui, C. H.; Gordon, E. M.; Patel, D. V.
Tetrahedron 1998, 54, 15879-15890.
(19) (a) Nikolaou, K. C.; Winssinger, N.; Vourloumis, D.; Oshima, T.; Kim,
S.; Pfefferkorn, J.; Xu, J.-Y.; Li, T. J. Am. Chem. Soc. 1998, 120, 10814-
10826. (b) Nikolaou, K. C.; Winssinger, N.; Pastor, J.; Ninkovic, S.; Sarabia,
F.; He, Y.; Vourloumis, D.; Yang, Z.; Li, T.; Giannakakou, P.; Hamel, E.
Nature 1997, 387, 268-272. (c) Nikolaou, K. C.; Vourloumis, D.; Li, T.;
Pastor, J.; Winssinger, N.; He, Y.; Ninkovic, S.; Sarabia, F.; Vallberg, H.;
Roschangar, F.; King, N. P.; Ray, M.; Finlay, V.; Giannakakou, P.; Verdier-
Pinard, P.; Hamel, E. Angew. Chem. 1997, 109, 2181-2187; Angew. Chem.,
Int. Ed. Engl. 1997, 36, 2097-2103.
(20) (a) Figerio, M.; Santagostino, M. Tetrahedron Lett. 1994, 35, 8019-8022.
(b) Figerio, M.; Santagostino, M.; Sputore, S.; Palmisano, G. J. Org. Chem.
1995, 60, 7272-7276.
(21) (a) Schwab, P.; France, M. B.; Ziller, J. W.; Grubbs, R. H. Angew. Chem.
1995, 107, 2197-2181. Angew. Chem., Int. Ed. Engl. 1995, 34, 2039-
2041. (b) Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Org. Lett. 1999,
1, 953-956.
(22) Harrity, J. P.; La, D. S.; Cefalo, D. R.; Visser, M. S.; Hoveyda, A. H.; J.
Am. Chem. Soc. 1998, 120, 2343-2351.
(23) Rutjes, F. P. J. T.; Veerman, J. J. N.; van Maarseveen, J. H.; Visser, G.
M.; Kruse, C. G.; Schoemaker, H. E.; Hiemstra, H. Eur. J. Org. Chem.
1998, 2583-2589.
(24) Thompson, L. A.; Ellman, J. A. Tetrahedron Lett. 1994, 35, 9333-9336.
9
13174 J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002

Synthesis of Protein Phosphatase Inhibitors
A R T I C L E S
Scheme 4. Synthesis on Dihydropyran-Resina
required for completion of the reaction. In addition, the endo/
exo selectivity was raised from 91:9 to 96:4 at the lower
temperature.
Next the carbon chain of carbonyl compound 34 was
elongated by the Wittig reaction to give enol ether 35, which
was hydrolyzed under acidic conditions. The latter step turned
out to be crucial because even under weakly acidic conditions
(10% acetic acid) an undesired rearrangement of the carbon
skeleton dominated over the hydrolysis. Finally, use of pyri-
dinium tosylate in refluxing THF/H2O gave aldehyde 36 in high
yield.
The furan ring was introduced by addition of 3-lithiofuran
to the aldehyde, which proceeded with a selectivity of 2:1 and
in 94% yield. Final oxidation of the furan to γ-hydroxybuteno-
lide 38 was achieved by means of in situ generated singlet
oxygen in the presence of diisopropylethylamine.15a,c-e Thus,
the planned reaction sequence was established successfully,
resulting in the identification of reagents and reaction conditions
that should be applicable on the solid phase as well.
Solid-Phase Synthesis of 6-epi-Dysidiolide. For the solid-
phase synthesis of 6-epi-dysidiolide, polymer-bound diene 4
initially was subjected to the Diels-Alder reaction with tiglic
aldehyde 42 in the presence of TMS-triflate (Scheme 6). To
achieve complete conversion within a reasonable time the
reaction temperature had to be raised to -30 °C, and under
these conditions cycloadduct 41 was obtained as a mixture of
four isomers that were formed in the ratio of endo/endo′/exo/
exo′ ) 67:16:16:1, with the desired endo isomer predominating.
1
The isomer ratios were determined by H NMR spectroscopy
a
after release of the cycloadducts from the solid support by olefin
metathesis and comparison with literature data.15d,f Thus, the
cycloaddition proceeded with an endo/exo ratio of 83:17, and
the two endo isomers were formed in a 81:19 ratio.
Key: (a) TBDPS-Cl, Et3N, DMAP, dichloromethane, r.t., 20 h, 39%;
(b) dihydropyran-resin (1 mmol/g), pyridinium-p-toluene sulfonate, di-
chloroethane, 80 °C, 20 h; (c) (nBu)4NF (TBAF), THF, r.t., 19 h; (d)
o-iodoxybenzoic acid (IBX), THF/DMSO ) 1:1, r.t., 12 h; (e) EtPPh3I,
nBuLi, 13, THF, r.t., 12 h, then nBuLi, 0 °C, 2h; (f) pyridinium-p-toluene
sulfonate, dichloroethane:ethanol ) 1:1, 80 °C, 20 h, 92% (b-f); (g)
(PCy3)2(Cl)2RudCHPh (21) (2 × 10 mol %), dichloromethane, r.t., 16 h,
82%; (h) pyridinium-p-toluene sulfonate, dichloroethane:ethanol ) 1:1, 80
°C, 20 h, 84% (g-h).
To increase the stereoselectivity and to achieve a more
efficient stereochemical steering of the cycloaddition, tiglic
aldehyde was converted into quasi-C2-symmetric chiral acetal
39 derived from (R,R)-2,4-pentanediol.25 This more reactive
chiral dienophile underwent the asymmetric Diels-Alder reac-
tion already at -78 °C, and after hydrolytic removal of the chiral
auxiliary group, aldehyde 41 was obtained as a mixture of four
isomers formed in the ratio endo/endo′/exo/exo′ ) 87:4:9:0.1.
Thus the endo/exo ratio was increased to 91:9, the two endo
isomers were formed in the ratio 95:5, and the amount of the
main isomer was raised from 67% to 87%. In particular, the
formation of the second endo isomer was largely suppressed.
Notably, the hydrolysis of the acetal of 40 required optimiza-
tion of the reaction conditions. As mentioned above, a weak
acid had to be used to prevent undesired rearrangement of the
carbon skeleton. The use of toluenesulfonic acid met this
demand, whereas TFA was too strong. Also, the choice of the
solvent was crucial. Thus, in THF/H2O, hydrolysis could not
be affected. Application of acetone to remove the acetal in a
transacetalization in the presence of water proceeded slowly and
was incomplete, since the resin does not swell sufficiently in
this reaction medium. Only after addition of dichloroethane to
guarantee appropriate resin swelling was the aldehyde liberated.
By analogy to the model solution-phase synthesis, it was
planned to elongate the carbon chain of aldehyde 41 by means
of a two-step sequence including a Wittig reaction and hydroly-
Scheme 5. Model Synthesis in Solutiona
a
Key: (a) tiglic aldehyde, TMSOTf, dichloromethane, -100 °C, 1 h,
85%; (b) Ph3PCH2OMeBr, NaNH2, HMDS, r.t., 1 h, 89%; (c) pyridinium-
p-toluene sulfonate, THF/H2O ) 10:1, ∆, 18 h, 91%; (d) 3-bromofuran,
nBuLi, THF, -78 °C, 90 min, 94%; (e) bengal rose, O2, Et(iPr)2N, hν,
dichloromethane, -78 °C, 2 h, then r.t., 10 min, 73%.
was investigated. Thereby, the formation of insoluble aluminum
salts during workup that would not be easily removable from
the resin is prevented . In the presence of 0.1 equiv of TMSOTf,
diene 33 was converted to Diels-Alder adduct 34 at -100 °C
within 1 h. In the presence of EtAlCl2, 2 h at -20 °C were
(25) Sammakia, T.; Berliner, M. A. J. Org. Chem. 1994, 59, 6890-6891.
9
J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002 13175

A R T I C L E S
Brohm et al.
Scheme 6. Solid-Phase Synthesis of 6-epi-Dysidiolidea
reinvestigated. It was finally observed that the THF/H2O/PPTS
mixture formed a two-phase system with the resin being in the
organic phase and most of the acid presumably dissolved in
the aqueous phase. Thereby, the concentration of acid in the
resin obviously is lowered, resulting in only very slow hydroly-
sis. Consequently, the amount of water in the reaction mixture
was reduced to 1%, resulting in the formation of a homogeneous
solution in which the desired hydrolysis proceeded rapidly and
without any undesired rearrangement reaction.
This observation in our opinion is worth noting, since it
demonstrates that reaction conditions preoptimized in solution
cannot necessarily be used on a resin, and this not for chemical
but rather for unexpected reasons that are not related to the
reaction under investigation. It also demonstrates that optimiza-
tion and execution of multistep synthesis in an automated fashion
may not be as straightforward as often claimed.
Introduction of the furan and its oxidative elaboration to the
γ-hydroxybutenolide were carried out as developed in the model
reaction sequence. Nucleophilic addition of furyllithium to a
resin-bound aldehyde was performed twice to increase the yield.
Single addition of the organometallic reagent to enolizable
aldehyde 3 did not result in complete conversion to 44. Sec-
ondary alcohol 44 was formed as a 2:1 mixture of diastereomers
whose configuration was not determined. Subsequently, furan
44 was oxidized to butenolide 45 with singlet oxygen in the
presence of Hu¨nig’s base.
Finally, release of the synthesis products from the polymeric
carrier was achieved as planned in a traceless manner and under
very mild conditions by performing an olefin metathesis reaction
with 10 mol % of Grubbs catalyst twice for 8 h each time. The
product mixture was readily purified by simple flash chroma-
tography and by filtering through a C18 reverse-phase separation
cartridge. 6-epi-Dysidiolide 46 and additional diastereomers
were formed in a total of 11 steps on the solid phase with an
overall yield of 14% (i.e., with an average yield of 84% per
step) based on the Merrifield resin. By using this procedure 48
mg of the desired product can be obtained from 1 g of resin
with a loading of 1.1 mmol/g.
The analytical data recorded for the dysidiolides are in good
agreement with the structures of the compounds and similar to
the data recorded for the natural product. 6-epi-Dysidiolide and
its 4-epimer were formed as the main products. In addition, three
further isomers were identified in small amounts. By means of
semipreparative HPLC on SiO2 the five diastereomers could
be separated.
Solid-Phase Synthesis of Analogues. To demonstrate the
robustness and reliability of the methodology described above,
we embarked on the solid-phase synthesis of dysidiolide
analogues that differ from the natural product in chain length
and substitution pattern.
The solid-phase syntheses of the dysidiolide analogues
commenced with the decisive and crucial Diels-Alder reaction
between chiral polymer-bound diene 4 and tiglic aldehyde 42
leading to immobilized aldehydes 41 and 3 (see above). Use of
achiral aldehyde 42 instead of chiral acetal 39 ultimately will
open up the opportunity to form and biologically investigate
more diastereomers of the corresponding products either as pure
compounds or initially as product mixtures.
a
Key: (a) TMSOTf, dichloromethane, -78 °C, 7 h; (b) p-toluene-
sulfonic acid, acetone, dichloroethane, H2O, ∆, 20 h; (c) TMSOTf,
dichloromethane, -30 °C, 2 h; (d) Ph3PCH2OMeCl, KOtBu, THF, r.t., 4
h, f43; (e) pyridinium-p-toluene sulfonate, THF, 1% H2O, ∆, 16 h; (f)
3-bromofuran, nBuLi, THF, -78 °C, 5 h; (g) bengal rose, O2, Et(iPr)2N,
hν, dichloromethane, -78 °C,
(PCy3)2(Cl)2RudCHPh (21) (2 × 10 mol %), dichloromethane, r.t., 16 h.
5 h, then r.t., 10 min, f45; (h)
sis of the enol ether formed therein. In contrast to the reaction
in solution phase, olefination with a mixture of methoxymethyl
triphenylphosphonium bromide and sodium amide was only
incomplete on the solid phase, as judged by Fourier transform
IR of the resin. However, clean and complete conversion to
the Z/E enol ether 43 was achieved by employing the corre-
sponding phosphonium chloride and KOtBu as base, as judged
by GCMS after cleavage from a resin sample by means of olefin
metathesis.
Acid-mediated hydrolysis of enol ether 43 to aldehyde 3 on
the solid phase turned out to be tricky. The reaction conditions
that had been developed in the course of the model reaction in
solution (see above) could not be employed in the heterogeneous
reaction. Rather, even after 24 h reaction time the use of PPTS
in H2O/THF 1:10 led only to little conversion. After substantial
experimentation and after various alternative reaction conditions
had been explored unsuccessfully, the original conditions were
Dysidiolide analogues 50 and 51 with a shortened carbon
chain were obtained via addition of 3-lithiofuran to aldehyde
9
13176 J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002

Synthesis of Protein Phosphatase Inhibitors
A R T I C L E S
Scheme 7. Solid-Phase Synthesis of Dysidiolide Analogues
50-53a
Scheme 8. Solid-Phase Synthesis of Dysidiolide Analogues 46
and 56-58a
a
Key: (a) 3-bromofuran, nBuLi, THF, -78 °C, 5 h; (b) 3-furylmethyl-
a
Key: (a) 3-bromofuran, nBuLi, THF, -78 °C, 5 h; (b) 3-furylmethyl-
triphenylphosphonium bromide, KOtBu, THF, 60 °C, 30 h; (c) 3-bromo-
methylfuran, Mg, THF, r.t. to 50 °C, 20 h; (d) bengal rose, O2, Et(iPr2)N,
hν, dichloromethane, -78 °C, 5 h, then r.t., 10 min; (e) (PCy3)2Cl2RudCHPh
(21) (2 × 10 mol %), dichloromethane, r.t., 16 h; (f) IBX, DMSO, r.t., 16
h, 80%.
triphenylphosphonium bromide, KOtBu, THF, 60 °C, 30 h; (c) 3-bromo-
methylfuran, Mg, THF, r.t. to 50 °C, 20 h; (d) bengal rose, O2, Et(iPr)2N,
hν, dichloromethane, -78 °C, 5 h, then r.t., 10 min; (e) (PCy3)2Cl2RudCHPh
(21) (2 × 10 mol %), dichloromethane, r.t., 16 h; (f) IBX, DMSO, r.t., 16
h, 75%.
41. After oxidative elaboration of intermediary formed furyl
alcohol 47 with singlet oxygen and subsequent cleavage from
the solid support by means of olefin metathesis, compound 50
was obtained in an overall yield of 26% for a total of eight
steps on the solid phase (Scheme 7).
Subsequent oxidation of the secondary alcohol with o-
iodoxybenzoic acid (IBX) gave the corresponding ketone 51.
Alternatively, oxidation of the secondary alcohol was carried
out on the solid phase; however, in this case the desired carbonyl
compound could only be isolated in trace amounts after release
from the solid support. In both cases oxidation of the hydroxy-
butenolide to the anhydride was not observed.
Application of the same solid-phase reaction sequence to
polymer-bound aldehyde 3 delivered 6-epi-dysidiolide 46 as
expected. Subsequent oxidation of the secondary alcohol yielded
ketone 56 in 75% yield (Scheme 8). In this case oxidation of
the alcohol on the solid phase prior to olefin metathesis also
gave inferior results.
In a second series of experiments dysidiolide analogues 52
and 57, with an olefinic bridge between the anellated ring
systems and the hydroxybutenolide, were synthesized. To this
end, polymer-bound aldehydes 41 and 3 were treated with the
ylide obtained from 3-furylmethyltriphenylphosphonium bro-
mide26 by deprotonation with KOtBu in THF at room temper-
ature (r.t.). However, in both cases the Wittig reactions leading
to immobilized olefins 48 and 54 only proceeded satisfactorily
after the temperature was raised to 50 °C. By analogy to the
synthesis described above, the furan rings were oxidized with
singlet oxygen, and the olefinic dysidiolide analogues 52 and
57 were released from the resin by means of olefin metathesis
in overall yields of 18% for eight steps and 27% for 10 steps,
respectively.
Finally, derivative 53 with the secondary alcohol in a different
position than in dysidiolide and chain-elongated analogue 58
were synthesized employing a Grignard reaction as the key step.
The Grignard reagent was prepared freshly from 3-bromomethyl
furan27 and magnesium in THF and was used immediately for
the reactions with aldehyde resins 41 and 3 to yield addition
products 49 and 55. Notably, the Grignard reactions proceeded
only with low degrees of conversion that could not be improved
by employing up to 5 equiv of reagent, higher temperature (up
to 50 °C), or longer reaction times (18 h). Thus, dysidiolide
analogues 53 and 58 finally were obtained in overall yields of
only 7% for eight steps and 6% for 10 steps on the solid phase,
respectively.
The syntheses of the dysidiolide analogues detailed above
proceed in reaction sequences of overall 8-11 linear steps with
the longest sequence on the solid support amounting up to 10
sequential transformations. The overall yields obtained in these
unoptimized multistep sequences range from 6% to 27%.
Typically, the individual reaction sequences were completed
within 7-10 days, and in all cases the desired products were
obtained after simple flash chromatography in multimilligram
(26) Okabe, M.; Tamagawa, H.; Tada, M. Synth. Commun. 1983, 13, 373-
378.
(27) Mateos, A. F.; Lopez-Barba, A. M. J. Org. Chem. 1995, 60, 3580-3585.
9
J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002 13177

A R T I C L E S
Brohm et al.
Table 1. Results of the Inhibition of Cdc25C and the Cytotoxicity
Tests Performed on the Colon Cancer Cell Lines SW480 and
HCT116, the Prostate Cell Line PC3, and the Breast Cell Line
MDA-MB231
more active than 6-epi-dysidiolide 46. These results indicate
that selectivity between different types of phosphatases and
conceivably among the three Cdc25 family members may be
achieved by means of dysidiolide analogues and compounds
derived therefrom. The data also indicate that a substantial
variation of the precise structural details of the natural product
itself is tolerated and leads to inhibitors with significantly
enhanced potency. Thus, replacement of the hydroxyethyl bridge
between the annelated core ring system and the hydroxy-
butenolide present in compound 46 by an unsaturated three
carbon unit (see 57) or introduction of a keto group (see 51
and 56) leads to more potent Cdc25C inhibitors. The synthetic
dysidiolide analogues also displayed considerable and differing
biological activity in a cytotoxicity assay29 employing the colon
cancer cell line SW480 (Table 1).
Four of the eight compounds investigated showed IC50 values
in the very low micromolar range and pronounced antitumor
activity. In this cellular assay alcohol 50 with a shortened carbon
chain was the most active compound, whereas inhibitors 51 and
56 had shown the lowest IC50 values and compounds 53 and
58, in which the alcohol is positioned differently between the
hydroxybutenolide and the core structure of dysidiolide, were
significantly less active.
This trend also became apparent when ketones 51 and 56 as
well as 6-epi-dysidiolide (46) were subjected to cytotoxicity
assays on the colon cell line HCT116, the prostate cancer cell
line PC3, and the breast cancer cell line MDA-MB231. As
shown in Table 1 ketones 51 and 56 again are substantially
less active than 6-epi-dysidiolide (46), which inhibits cell
proliferation in all three cases with IC50 values in the very low
micromolar range.
Thus, the data indicate that the small library of natural product
analogues already contains potent compounds with significantly
differing biological activity both in vitro and in vivo. The
observation that the order of IC50 values determined in an
enzyme assay does not necessarily parallel the outcome of
cellular assays is not uncommon.
a
c
c
c
Cdc25C
SW480b
IC50 (µM)
HCT116
PC3
MDA-MB231
(µM)
cpd
IC50 (µM)
IC50 (µM)
IC50 (µM)
46
50
51
56
52
57
53
58
38
5.1
16
4
1
20
1.2
1
1.6
1.5
0.8
6.8
2.4
6.1
9
11
15
13
>20
>10
>10
>33
4
2
>33
>33
8
>50
a
For the phosphatase assay, 5 µL of compound dissolved in 100%
DMSO was added to a solution of 0.2 µg of recombinant Cdc25C in 85 µL
of assay buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM DTT, 1
mM EDTA, and 10% DMSO). After incubation for 30 min at 30 °C, the
substrate fluorescein diphosphate (FDP) was added to give a final
concentration of 1 µM. Plates were read after a reaction time of 30 min at
485/535 nm (ex/em). o-Vanadate (IC50 ) 0.1 µM) was used as reference
b
compound. The cells were incubated with the compounds at concentrations
between 1.2 and 100 µM for 3 days. The viability of the cells was measured
with MTT, which is a slightly yellow tetrazolium salt. Living cells reduce
the tetrazolium moiety by a mitochondrial dehydrogenase to a dark violet
c
dye that is detected at 570 nm. The assays were performed using the
CytoTox 96 cytotoxicity assay kit from Promega Corporation, USA. Three
thousand cells were plated per well in a 96-well flat-bottomed plate,
incubated with the compounds at concentrations ranging from 0.033 µM
to 10 µM for 3 days. Cells were lysed, and the cellular lactate dehydrogenase
activity was measured, which quantitatively reflects the number of cells.
amounts starting from only 100 mg of resin (i.e., in sufficient
quantity and in sufficient purity for subsequent biological
evaluation).
These results demonstrate that the multistep total synthesis
of natural products and close analogues thereof on the solid
phase is feasible. The transformations applied in the synthesis
described above include a variety of very different reaction types
widely used in organic synthesis (i.e., an asymmetric cyclo-
addition employing a removable chiral auxiliary, different
organometallic transformations, olefination reactions, different
oxidation reactions, acidic hydrolyses of acetals and enol ethers,
and a nucleophilic substitution).
To determine if the solid-phase synthesis delivered biologi-
cally active natural product analogues with a high frequency,
we investigated the dysidiolide analogues as inhibitors for the
protein phosphatase Cdc25C and in cellular cytotoxicity assays.
From the Cdc25 phosphatase family the Cdc25C protein was
chosen, since 6-epi-dysidiolide 46 has previously been inves-
tigated as an inhibitor of Cdc25A and Cdc25B,28 thereby
allowing for comparison of data.
The results obtained in the phosphatase assay shown in Table
1 demonstrate that all dysidiolide analogues synthesized on solid
support inhibit Cdc25C in the low micromolar range with the
IC50 value varying by a factor of 20. Only analogue 38 that
was synthesized in solution showed a weak inhibition (>50 µM).
The IC50 of 5.1 µM determined for inhibition of Cdc25C by
6-epi-dysidiolide 46 is considerably lower than the values
recorded for the inhibition of Cdc25A (13 µM) and Cdc25B
(18 µM) by this compound. In addition, the most active
compound in this enzyme assay, ketone 56, exhibited an IC50
value in the high nanomolar range (800 nM) and was 6.4-fold
Conclusion
In conclusion we have demonstrated that the synthesis of
natural product derived libraries in long multistep sequences
executed on a polymeric support and employing a variety of
widely differing synthetic transformations is feasible and that
it can deliver potent biologically active compounds with high
frequency. Given this power of organic synthesis, the combi-
natorial synthesis and variation of natural products lies ahead.
Acknowledgment. This research was supported by the
Deutsche Forschungsgemeinschaft, the Fonds der Chemischen
Industrie, and the Bayer AG.
Supporting Information Available: Experimental procedures
for the synthesis and spectroscopic data of all new compounds.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA027609F
(29) Mosman, T. J. Immunol. Methods 1983, 65, 55-63.
(30) Kroeniger, A.; Wheeler, D. M. S. Tetrahedron 1972, 28, 255-266.
(31) Kayser, M. M.; Chen, G.; Stewart, J. D. J. Org. Chem. 1998, 63, 7103-
7106.
(32) Lafontaine, J. A.; Leahy, J. W. Tetrahedron Lett. 1995, 36, 6029-6032.
(33) (a) Bartlett, P. D.; Bank, S.; Crawford, R. J.; Schmid, G. H. J. Am. Chem.
Soc. 1965, 87, 1288-1297. (b) Wilt, J. W.; Massie, S. N.; Dabek, R. B. J.
Org. Chem. 1970, 35, 2803-2806.
(28) Takahashi, M.; Dodo, K.; Sugimoto, Y.; Aoyagi, Y.; Yamada, Y.;
Hashimoto, Y.; Shirai, R. Bioorg. Med. Chem. Lett. 2000, 10, 2571-2574.
9
13178 J. AM. CHEM. SOC. VOL. 124, NO. 44, 2002