Anguinomycins and derivatives: Total syntheses, modeling, and biological evaluation of the inhibition of nucleocytoplasmic transport

Article abstract of DOI:10.1021/ja9097093

The preparation of the polyketide natural products anguinomycin C and D is reported based on key steps such as Negishi stereoinversion cross coupling, Jacobsen Cr(III)-catalyzed Hetero Diels - Alder reaction, Evans B-mediated syn-aldol chemistry, and B-alkyl Suzuki - Miyaura cross coupling. The configuration of both natural products was established as (5R,10R,16R,18S,19R, 20S). Biological evaluation demonstrated that these natural products are inhibitors of the nuclear export receptor CRM1, leading to shutdown of CRM1-mediated nuclear protein export at concentrations above 10 nM. Analogues of anguinomycin and leptomycin B (LMB) have been prepared, and the simple α,β-unsaturated lactone analogue 4 with a truncated polyketide chain retains most of the biological activity (inhibition above 25 nM). The structural basis for this inhibition has been demonstrated by modeling the transport inhibitors into X-ray crystal structures, thus highlighting key points for successful and strong biological action of anguinomycin and LMB.

Full text of DOI:10.1021/ja9097093

Published on Web 01/07/2010
Anguinomycins and Derivatives: Total Syntheses, Modeling,
and Biological Evaluation of the Inhibition of
Nucleocytoplasmic Transport
Simone Bonazzi,†,‡ Oliv Eidam,*,§ Stephan Gu¨ttinger,| Jean-Yves Wach,†
Ivo Zemp,| Ulrike Kutay,*,| and Karl Gademann*,†
Chemical Synthesis Laboratory (SB-ISIC-LSYNC), Swiss Federal Institute of Technology (EPFL),
CH-1015 Lausanne, Switzerland, Department of Pharmaceutical Chemistry, UniVersity of
California, San Francisco (UCSF), San Francisco, California 94158, and Institut fu¨r Biochemie,
Swiss Federal Institute of Technology (ETHZ), CH-8093 Zu¨rich, Switzerland
Received November 16, 2009; E-mail: karl.gademann@epfl.ch; oliv.eidam@ucsf.edu;
ulrike.kutay@bc.biol.ethz.ch
Abstract: The preparation of the polyketide natural products anguinomycin C and D is reported based on
key steps such as Negishi stereoinversion cross coupling, Jacobsen Cr(III)-catalyzed Hetero Diels-Alder
reaction, Evans B-mediated syn-aldol chemistry, and B-alkyl Suzuki-Miyaura cross coupling. The
configuration of both natural products was established as (5R,10R,16R,18S,19R,20S). Biological evaluation
demonstrated that these natural products are inhibitors of the nuclear export receptor CRM1, leading to
shutdown of CRM1-mediated nuclear protein export at concentrations above 10 nM. Analogues of
anguinomycin and leptomycin B (LMB) have been prepared, and the simple R,ꢀ-unsaturated lactone
analogue 4 with a truncated polyketide chain retains most of the biological activity (inhibition above 25
nM). The structural basis for this inhibition has been demonstrated by modeling the transport inhibitors into
X-ray crystal structures, thus highlighting key points for successful and strong biological action of
anguinomycin and LMB.
Introduction
Natural products represent a major source of new drugs, and
in particular related to anti-infective and anticancer agents, these
metabolites account for roughly 50% of chemical entities on
the pharmaceutical market.1 In addition, derivatives obtained
through synthetic organic chemistry have allowed one to
leverage the potential of natural products by providing more
effective and selective compounds.2 These favorable properties
of natural products have also led to their use as tool compounds
in biology, as these efficient modulators of biological pathways
enabled the generation of desired phenotypes in a well-controlled
approach (chemical genetics).3
The leptomycin family of natural products features many of
the desired properties regarding their use as clinical candidates
or tool compounds.4 Both leptomycin B (LMB)5 (Figure 1) and
† EPFL.
‡ Present address: Department of Chemistry and Chemical Biology,
Harvard University, Cambridge, MA 02318.
§ UCSF.
| ETHZ.
(1) (a) Newman, D. J.; Cragg, G. M. J. Nat. Prod. 2007, 70, 461–477.
(b) Newman, D. J.; Cragg, G. M.; Snader, K. M. J. Nat. Prod. 2003,
66, 1022–1037.
(2) Newman, D. J.; Cragg, G. M. J. Nat. Prod. 2004, 67, 1216–1238.
(3) (a) Kaiser, M.; Wetzel, S.; Kumar, K.; Waldmann, H. Cell. Mol. Life
Sci. 2008, 65, 1186–1201. (b) Burke, M. D.; Schreiber, S. L. Angew.
Chem., Int. Ed. 2004, 43, 46–58.
Figure 1. Leptomycin B (LMB), anguinomycins C (1), D (2), and
derivatives 3 and 4 featuring the R,ꢀ-unsaturated lactone 5.
callystatin6 were identified and clinically evaluated7 as potent
anticancer agents. LMB was characterized as a potent inhibitor
of the nuclear export of proteins,8 and the mode of action
(4) Kalesse, M.; Christmann, M. Synthesis 2002, 981–1003.
(5) (a) Hamamoto, T.; Gunji, S.; Tsuji, H.; Beppu, T. J. Antibiot. 1983,
36, 639–645. (b) Hamamoto, T.; Seto, H.; Beppu, T. J. Antibiot. 1983,
36, 646–650.
9
1432 J. AM. CHEM. SOC. 2010, 132, 1432–1442
10.1021/ja9097093  2010 American Chemical Society

Syntheses, Modeling, and Biology of the Anguinomycins
A R T I C L E S
involves binding to the chromosome maintenance region 1
(CRM1) exportin through its R,ꢀ-unsaturated lactone moiety.9
This leads to selective inhibition of the protein/protein interac-
tion in the ternary CRM1-RAN-cargo protein complex10 and
thereby to efficient shutdown of CRM1-mediated nuclear protein
export. High specificity and efficiency made LMB a frequently
used tool compound in cell biology with over 1000 publications
reporting its use.11 This selective action also prompted the
synthetic community, and many total syntheses of LMB and
other members of this family have been reported.12 Although
initial studies on LMB were abandoned due to toxicity,7 renewed
clinical interest in LMB focused on its use in combination
therapy (e.g., with gleevec)13 and on the generation of less toxic
analogues with increased therapeutic windows by chemical
modification14 or conjugation to antibodies.15
Anguinomycins A-D were reported as selective agents
targeting immortalized cells,16 as they have been shown to
induce apoptosis in pRB-inactivated tumor cells, while only
inducing growth arrest in normal cells. This astonishing
selectivity caused us to prepare totally synthetic anguinomycin
C, and we were able to show that anguinomycin C is an inhibitor
of CRM1-mediated nuclear export.17 However, major questions
concerning the mode of action of these compounds remain open.
In particular, the role of the polyketide chain is unclear, as, for
example, Kudo et al. suggested either mimicry of the protein
cargo or induction of a conformational change of CRM1 by
LMB.9a,b We wanted to investigate if this chain mimics the
hydrophobic leucine-rich nuclear export signal of the cargo
protein. It has been shown that interactions between this leucine-
rich helical sequence and CRM1 are relatively weak, relying
on Coulomb protein/protein interactions of a single helical
fragment.11 Therefore, we hypothesized that the side chain of
LMB/anguinomycins can be either replaced by a simpler
hydrophobic analogue or completely omitted.
In this study, we investigate these hypotheses by a combina-
tion of synthetic chemistry, biological evaluation, and compu-
tational modeling. We report in full detail the total synthesis
and biological evaluation of anguinomycins C (1) and D (2) as
well as a series of analogues designed to probe these questions
(Figure 1). We show that replacement of the side chain by a
linear terpene is possible without strongly impacting biological
activity (compound 3), and we observed that analogue 4 with a
truncated polyketide chain (Figure 1) almost fully retains
biological activity. The structural basis for the biological activity
of these natural products on CRM1 binding could be rationalized
by atomistic modeling of LMB, anguinomycin, and general
lactones of type 5 to CRM1.
(6) Kobayashi, M.; Higuchi, K.; Murakami, N.; Tajima, H.; Aoki, S.
Tetrahedron Lett. 1997, 38, 2859–2862.
(7) Newlands, E. S.; Rustin, G. J. S.; Brampton, M. H. Br. J. Cancer
1996, 74, 648–649.
(8) (a) Fornerod, M.; Ohno, M.; Yoshida, M.; Mattaj, I. W. Cell 1997,
90, 1051–1060. (b) Fukuda, M.; Asano, S.; Nakamura, T.; Adachi,
M.; Yoshida, M.; Yanagida, M.; Nishida, E. Nature 1997, 390, 308–
311. (c) Wolff, B.; Sanglier, J. J.; Wang, Y. Chem. Biol. 1997, 4,
139–147. (d) Nishi, K.; Yoshida, M.; Fujiwara, D.; Nishikawa, M.;
Horinouchi, S.; Beppu, T. J. Biol. Chem. 1994, 269, 6320–6324.
(9) (a) Kudo, N.; Wolff, B.; Sekimoto, T.; Schreiner, E. P.; Yoneda, Y.;
Yanagida, M.; Horinouchi, S.; Yoshida, M. Exp. Cell Res. 1998, 242,
540–547. (b) Kudo, N.; Matsumori, N.; Taoka, H.; Fujiwara, D.;
Schreiner, E. P.; Wolff, B.; Yoshida, M.; Horinouchi, S. Proc. Natl.
Acad. Sci. U.S.A. 1999, 96, 9112–9117. (c) Daelemans, D.; Afonina,
E.; Nilsson, J.; Werner, G.; Kjems, J.; De Clercq, E.; Pavlakis, G. N.;
Vandamme, A. M. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 14440–
14445. (d) Van Neck, T.; Pannecouque, C.; Vanstreels, E.; Stevens,
M.; Dehaen, W.; Daelemans, D. Bioorg. Med. Chem. 2008, 16, 9487–
9497.
(10) (a) Dong, X.; Biswas, A.; Suel, K. E.; Jackson, L. K.; Martinez, R.;
Gu, H.; Chook, Y. M. Nature 2009, 458, 1136–1141. (b) Dong, X.;
Biswas, A.; Chook, Y. M. Nat. Struct. Mol. Biol. 2009, 16, 558–560.
(c) Monecke, T.; Guttler, T.; Neumann, P.; Dickmanns, A.; Gorlich,
D.; Ficner, R. Science 2009, 324, 1087–1091.
(11) Review: Kutay, U.; Gu¨ttinger, S. Trends Cell Biol. 2005, 15, 121–
124.
Results
(12) For callystatin, see: (a) Murakami, N.; Wang, W.; Aoki, M.; Tsutsui,
Y.; Sugimoto, M.; Kobayashi, M. Tetrahedron Lett. 1998, 39, 2349–
2352. (b) Crimmins, M. T.; King, B. W. J. Am. Chem. Soc. 1998,
120, 9084–9085. (c) Smith, A. B., III; Brandt, B. M. Org. Lett. 2001,
3, 1685–1688. (d) Kalesse, M.; Quitschalle, M.; Khandavalli, C. P.;
Saeed, A. Org. Lett. 2001, 3, 3107–3109. (e) Marshall, J. A.; Bourbeau,
M. P. J. Org. Chem. 2002, 67, 2751–2754. (f) Vicario, J. L.; Job, A.;
Wolberg, M.; Mu¨ller, M.; Enders, D. Org. Lett. 2002, 4, 1023–1026.
(g) Lautens, M.; Stammers, T. A. Synthesis 2002, 1993–2012. (h)
Langille, N. F.; Panek, J. S. Org. Lett. 2004, 6, 3203–3206. (i) Dias,
L. C.; Meira, P. R. R. J. Org. Chem. 2005, 70, 4762–4773. (j) Reichard,
H. A.; Rieger, J. C.; Micalizio, G. C. Angew. Chem., Int. Ed. 2008,
47, 7837–7840. (k) For leptomycin B, see: Kobayashi, M.; Wang,
W.; Tsutsui, Y.; Sugimoto, M.; Murakami, N. Tetrahedron Lett. 1998,
39, 8291–8294. (l) For ratjadone, see: Christmann, M.; Bhatt, U.;
Quitschalle, M.; Claus, E.; Kalesse, M. Angew. Chem., Int. Ed. 2000,
39, 4364–4366. (m) Williams, D. R.; Ihle, D. C.; Plummer, S. V. Org.
Lett. 2001, 3, 1383–1386. (n) For leptofuranin D, see: Marshall, J. A.;
Schaaf, G. M. J. Org. Chem. 2003, 68, 7428–7432. (o) For (-)-
kazusamycin, see: Arai, N.; Chikaraishi, N.; Omura, S.; Kuwajima, I.
Org. Lett. 2004, 6, 2845–2848. (p) For leptostatin, see: Marshall, J. A.;
Mikowski, A. M.; Bourbeau, M. P.; Schaaf, G. M.; Valeriote, F.
Bioorg. Med. Chem. Lett. 2006, 16, 320–323.
Total Syntheses of Anguinomycins C and D. The synthesis
of anguinomycins C (1) and D (2) started with the preparation
of dihydropyran 9 via a hetero Diels-Alder reaction between
known aldehyde 618 and commercially available 1-methoxy-
1,3-butadiene (7) (Scheme 1). The reaction was catalyzed by
the Cr(III) catalyst (8) developed by Jacobsen and co-workers19
and was carried out under solvent-free conditions and in the
presence of 4 Å molecular sieves.20 The dihydropyran 9 was
obtained in high yield (86%), enantioselectivity (96% ee), and
as a 5:1 diastereomeric mixture due to epimerization at the
anomeric center under the reaction conditions. Attempts to
directly use the MeO-protected lactol 9 for the continuation of
the synthesis proved to be problematic. The diastereoisomers
could be easily separated by chromatography, but after depro-
tection of the silyl group with TBAF, the resulting terminal
alkyne 10 was volatile and difficult to handle. Therefore, we
(13) (a) Kancha, R. K.; Von Bubnoff, N.; Miething, C.; Peschel, C.; Go¨tze,
K. S.; Duyster, J. Haematologica 2008, 93, 1718–1722. (b) Aloisi,
A.; Di Gregorio, S.; Stagno, F.; Guglielmo, P.; Mannino, F.; Sormani,
M. P.; Bruzzi, P.; Gambacorti-Passerini, C.; Saglio, G.; Venuta, S.;
Giustolisi, R.; Messina, A.; Vigneri, P. Blood 2006, 107, 1591–1598.
(14) (a) Santi, D. V.; Zhou, Y. WO Patent 2005086944, 2005. (b) Santi,
D. V.; Myles, D. C.; Dong, S.; Brian, H. U.S. Patent 2005272727,
2005. (c) Dong, S.; Santi, D. V. WO Patent 2007033214, 2007. (d)
Dong, S.; Myles, D. C.; Santi, D. V.; Brian, H. CN Patent 1964718,
2008. (e) Mutka, S. C.; Yang, W. Q.; Dong, S. D.; Ward, S. L.; Craig,
D. A.; Timmermans, P. B. M. W. M.; Murli, S. Cancer Res. 2009,
69, 510–517.
(15) Bouchard, H.; Commercon, A.; Chari, R. V. J. E. WO Patent
2007144709, 2007.
(16) (a) Hayakawa, Y.; Sohda, K. Y.; Shin-Ya, K.; Hidaka, T.; Seto, H. J.
Antibiot. 1995, 48, 954–961. (b) Hayakawa, Y.; Adachi, K.; Ko-
meshima, N. J. Antibiot. 1987, 40, 1349–1352.
(17) Bonazzi, S.; Gu¨ttinger, S.; Zemp, I.; Kutay, U.; Gademann, K. Angew.
Chem., Int. Ed. 2007, 46, 8707–8710.
(18) Plater, M. J.; Aiken, S.; Bourhill, G. Tetrahedron 2002, 58, 2415–
2422.
(19) Chavez, D. E.; Jacobsen, E. N. Org. Synth. 2005, 82, 34–42.
(20) Jarvo, E. R.; Lawrence, B. M.; Jacobsen, E. N. Angew. Chem., Int.
Ed. 2005, 44, 6043–6046.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1433

A R T I C L E S
Bonazzi et al.
Scheme 2. Tandem Hydrozirconation-Negishi Cross-Couplinga
Scheme 1. Synthesis of the Terminal Alkyne 11 via a Hetero
Diels-Alder Reactiona
a Reagents and conditions: (a) Cp2ZrHCl, 0 °C f room temperature,
THF, 1 h. (b) ZnCl2, THF, room temperature, 30 min. (c) 15, Pd(PPh3)4 (5
mol %), DIBAH (1.0 M in hexane), room temperature f 40 °C, 13 h,
81%.
a Reagents and conditions: (a) 7, 8 (2.3 mol %), 4 Å MS, room
temperature, 86%; 96% ee. (b) TBAF, THF, 0 °C f room temperature. (c)
PTSA, iPrOH, room temperature, 86%. (d) TBAF (1 M in THF), THF, 0
°C f room temperature, 95%.
Scheme 3. Synthesis of Fragment 17 of Anguinomycin C via
Negishi Cross-Coupling with Inversion of the Configurationa
treated the diastereomeric mixture obtained in the Diels-Alder
reaction under acidic conditions in iPrOH to afford the iPrO-
protected lactol 11 as a single diastereoisomer in the thermo-
dynamically more stable configuration. Deprotection with TBAF
and purification on silicagel afforded the deprotected alkyne 12
as a colorless oil, which was carefully concentrated due to its
volatility (Scheme 1).
a Reagents and conditions: Pd(PPh3)4 (10 mol %), Me2Zn (2.0 M in
toluene), THF, 45 °C, 38 h, 68%, cis/trans > 97:3. Arrows in 17 indicate
NOE correlations.
We wanted to couple the terminal alkyne 12 to the dibromide
15 by a tandem hydrozirconation-Negishi cross-coupling
reaction. Thus, the terminal alkyne 12 was hydrozirconated by
Schwartz’s reagent and the resulting species 13 transmetalated
to the vinyl Zn species 14. Separately, a yellow solution of
Pd(PPh3)4 (5 mol %) in THF was treated with a solution of
DIBAH (10 mol %) to give a dark red solution. After 30 min,
the readily available dibromoolefin 1521 in THF was added, and
the resulting mixture was transferred into the separately prepared
vinylzinc solution of 14. The mixture was stirred 10 h at 40
°C, and after workup and purification, the coupled (E)-product
16 was obtained in 81% yield as a single diastereoisomer
(Scheme 2). 1H NMR and NOE spectroscopic analyses of
product 16 confirmed that the reaction had occurred in a
completely selective fashion, resulting in the (6E,8Z) isomer.
This Negishi cross-coupling between alkyne 12 and dibro-
moolefin 15 afforded the coupled trans Br-olefin 16, but for
the preparation of anguinomycins C and D a cis configuration
of the double bond at C(8)-C(9) was required. To achieve this
goal, we applied the procedure developed by Negishi and co-
workers,22 allowing the installation of the alkyl residue at C(8)
with inversion of the configuration of the double bond. The
nature of the catalyst employed in the Pd-catalyzed alkenylation
of alkenyl halide influences the selectivity of the resulting
product.22 This reaction can occur with retention or inversion
of the configuration depending on the ligands on Pd. As in the
first Negishi cross coupling, the trans product 16 was obtained,
and an inversion of the configuration affording the cis product
(6E,8Z) as present in the anguinomycin structure was required.
In addition, by employing either Me2Zn or Et2Zn in the Negishi
coupling, we would be able to access both anguinomycins C
and D, respectively. We started with the preparation of the
intermediate for the synthesis of anguinomycin C by adding
Me2Zn (2 M in toluene) to a solution of Br-diene 16 and
Pd(PPh3)4 (10 mol %) in THF under argon atmosphere. The
resulting orange-colored solution was stirred for 24 h at 45 °C,
then a second addition of Me2Zn was added, and complete
conversion of the starting material was achieved after an
additional 14 h. As expected, the use of Pd(PPh3)4 afforded the
cis product 17 (assigned by NOEs) as a single isomer in 68%
yield (Scheme 3).
The same Negishi cross coupling with stereoinversion reaction
was adopted for the synthesis of the fragment for anguinomycin
D. In this case, we used Et2Zn (1.5 M in toluene), which was
added to a solution of alkenylhalide 16 and Pd(PPh3)4 (10 mol
%) in THF. However, following workup we obtained a mixture
of unreacted starting material, and the cis and trans products.
Separation of the two isomers by chromatography on SiO2 was
unsuccessful, as the similar polarity of the cis and trans isomer
did not allow the isolation of the pure desired isomer. Attempts
to force the reaction by adding more Et2Zn were unsuccessful
and promoted formation of the undesired trans isomer. Gener-
ally, it was found that varying amounts of O2 present in the
solvents led to mixed results, and degassing the solvent (three
freeze/pump/thaw cycles) gave better results. The cis product
was identified as the major isomer by using Pd(PPh3)4 and Et2Zn,
but conversion of the starting material was not complete and
formation of the trans compound was still observed. At this
point, we chose to screen several different Pd catalysts (5-10
mol %) for the reaction (Table 1). The screened catalysts were
Pd(PPh3)4, PdCl2(dppf), Pd(PtBu3)2, PdCl2(DPEphos), trans-
di(µ-acetato)bis[o-(tolyl-phosphino)benzyl] dipalladium(II), and
allyl[1,3-bis(mesityl)imidazol-2-ylidene] Pd chloride. In agree-
ment with data reported by Negishi and co-workers,22a
(21) Komatsu, K.; Tanino, K.; Miyashita, M. Angew. Chem., Int. Ed. 2004,
43, 4341–4345.
(22) (a) Zeng, X.; Hu, Q.; Qian, M.; Negishi, E. I. J. Am. Chem. Soc. 2003,
125, 13636–13637. (b) Zeng, X.; Qian, M.; Hu, Q.; Negishi, E. I.
Angew. Chem., Int. Ed. 2004, 43, 2259–2263.
9
1434 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Syntheses, Modeling, and Biology of the Anguinomycins
A R T I C L E S
Table 1. Screening of Conditions for the Preparation of Fragment 18 of Anguinomycin D via Negishi Cross-Coupling with Selective
Retention or Inversion of the Configurationa
equivalents
(mol %)
concentration
(M vs 16)
reaction
time (h)
catalyst
ratioa
yield (%)
Pd(PPh3)4
Pd(PPh3)4
PdCl2(dppf)
5
10
10
10
5
10
10
10
0.06
0.1
24
28
20
3.5
14
14
20
20
0.14/1.00/0.38
0.16/1.00/0.17
1.00/1.08/0.66
0/0/1.00
0/1.00/0
0/1.00/0
n.d.
n.d.
n.d.
75%
84%
84%
65%
77%
0.05
0.05
0.08
0.08
0.05
0.05
b
Pd(PtBu3)2
PdCl2(DPEphos)c
PdCl2(DPEphos)c
trans-di(µ-acetato)bis[o-(tolylphosphino)benzyl] dipalladium(II)d
0/0/1.00
0/0/1.00
allyl[1,3-bis(mesityl)imidazol-2-ylidene] palladium chlorideb
a The ratio of 16/18/19 is reported. b During the addition of Et2Zn, the solution turned immediately to a dark-brown color. c During all of the
reactions, the solution maintained an orange-red color. d At ca. 35 °C, the solution turned to a dark-brown color.
Scheme 4. Synthesis of Fragments 20 and 21a
PdCl2(DPEphos) revealed to be the best catalyst to achieve
alkenylation with inversion of configuration. The cis adduct 18,
the intermediate for the synthesis of anguinomycin D, was
obtained in 84% yield as a single isomer. The reaction proved
to be clean, without traces of remaining starting material or the
undesired trans isomer. It is probably the higher thermal stability
of PdCl2(DPEphos) as compared to Pd(PPh3)4 that made the
catalyst efficient over a longer time without any decomposition.
The percentage of the catalyst employed (5 or 10 mol %) did
not influence the yield and selectivity of the reaction, affording
in both cases only the cis isomer. We have found that both
Pd(PPh3)4 and PdCl2(dppf) resulted in mixtures of cis and trans
products. In addition to the previously reported Pd(PtBu3)2, the
trans-di(µ-acetato)bis[o-(tolyl-phosphino)benzyl] dipalladium(II)
and the allyl[1,3-bis(mesityl)imidazol-2-ylidene] Pd chloride
also afforded exclusively the trans adduct 19, via retention of
configuration and moderate to good yields.
By using the Negishi cross coupling with selective inversion
of configuration, a straightforward and effective route to the
trisubstituted C(8)-C(9) olefin in 18 and 19 was identified. It
should be pointed out that the reverse strategy, that is, reacting
the dibromoolefin 15 with Me2Zn or MeB(OH)2 under Negishi
or Suzuki conditions, was evaluated but proved to be trouble-
some as low yields and selectivities were observed. In addition,
the use of the TIPS group for the primary OH group was
necessary, as the TBS group was found to be unstable under
the Negishi conditions.
At this point, the silyl group of both 17 and 18 was removed
by treatment with TBAF in THF, affording the deprotected
products in quantitative yield. Subsequently, both alcohols were
treated with I2, PPh3, and imidazole in toluene to obtain the
alkyl iodide products 20 and 21 in good yields (Scheme 4).
The synthesis of the polyketide chain was based on the classic
Evans syn-aldol strategy,23 but we chose to employ the DIOZ
auxiliary (DIOZ ) 4-isopropyl-5,5-diphenyloxazolidin-2-one)
developed by Seebach and Hintermann.24 The additional Ph
groups on the auxiliary increase the stability against nucleophilic
attack, thus allowing, for example, the formation of the lithiated
a Reagents and conditions: From 17 to 20, (a) TBAF (1.0 M in THF),
THF, 0 °C f room temperature, 2 h, 99%. (b) PPh3, imidazole, I2, toluene/
Et2O, 0 °C f room temperature, 2 h, 75%. From 18 to 21, (c) TBAF (1.0
M in THF), THF, 0 °C f room temperature, 1.5 h, 98%. (d) PPh3,
imidazole, I2, toluene/Et2O, 0 °C, 45 min, 89%.
oxazolidinone by using nBuLi at 0 °C. Moreover, the presence
of the two Ph groups increases the selectivity of the reactions
as well as the crystallinity of the resulting products. This
auxiliary demonstrated its utility in total synthesis when being
used by chemists at Novartis in the synthesis of discoder-
molide.25 We sought the all-syn configuration of the polyketide
chain, due to the structural similarity of anguinomycin to LMB.
In addition, we chose to keep the hydroxy group at C(17)
unprotected for the entire synthesis, based on reports by Dias
and Meira.12i Treatment of the known acylated chiral auxiliary
2224a with LDA generated the lithium enolate, which reacted
via enantioselective alkylation with allylic bromide 2326 to give
the adduct 24 in high yield (92%) and excellent selectivity (dr
> 97:3) as a crystalline white solid. The auxiliary was removed
with LAH in quantitative yield, which allowed for the recycling
of the cleaved oxazolidinone and Swern oxidation of the
intermediate gave the aldehyde 25. A stereoselective B-mediated
aldol reaction was carried out by treatment of ent-22 with
Bu2BOTf following addition of aldehyde 25.27 The aldol 26
was obtained in 77% yield and a diastereomeric ratio of 87:13
(Scheme 5), which is comparable to the selectivities achieved
by other groups using similar substrates.12a The two diastere-
oisomers displayed similar Rf values, but separation by repeated
flash chromatography on SiO2 was possible, affording pure syn-
aldol 26.
(25) (a) Loiseleur, O.; Koch, G.; Wagner, T. Org. Process Res. DeV. 2004,
8, 597–602. (b) Loiseleur, O.; Koch, G.; Cercus, J.; Schu¨rch, F. Org.
Process Res. DeV. 2005, 9, 259–271.
(23) Evans, D. A.; Bartroli, J.; Shih, T. L. J. Am. Chem. Soc. 1981, 103,
2127–2129.
(24) (a) Hintermann, T.; Seebach, D. HelV. Chim. Acta 1998, 81, 2093–
2126. (b) Bull, S. D.; Davies, S. G.; Garner, A. C.; Kruchinin, D.;
Key, M. S.; Roberts, P. M.; Savory, E. D.; Smith, A. D.; Thomson,
J. E. Org. Biomol. Chem. 2006, 4, 2945–2964.
(26) Katzenellenbogen, J. A.; Crumrine, A. L. J. Am. Chem. Soc. 1976,
98, 4925–4935.
(27) For ent-25, see: White, J. D.; Johnson, A. T. J. Org. Chem. 1994, 59,
3347–3358.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1435

A R T I C L E S
Bonazzi et al.
Scheme 5. The Alkylation and the First Aldol Reactiona
quantitative yield. Takai reaction afforded the all-protected vinyl
iodide 36 in 88% yield and in an E/Z ratio of 95:5 (Scheme 6).
Having all of the fragments in hand, we attempted the sp3-sp2
Suzuki cross coupling following Johnson’s conditions30 previ-
ously used by Marshall and co-workers in the synthesis of
callystatin.12e The alkyl iodide 20 was reacted with 9-MeO-9-
BBN and tBuLi at -78 °C, forming the boronate intermediate
37, to which a solution containing vinyl iodide 34, Cs2CO3,
AsPh3, and PdCl2(dppf) in a DMF/water mixture was added.
The reaction proceeded smoothly, and the coupled product 39
featuring the complete skeleton of anguinomycin C was isolated
in 80% yield (Scheme 7).31 The same procedure was applied
to the fully protected vinyl iodide 36 and boronate intermediate
37. However, this reaction proved to be problematic, and the
coupled product 40 was obtained only in low yield (Scheme
7). Having confirmed the superiority, in terms of yield, of
the vinyl iodide 34 with the free hydroxy group at C(17) in the
sp3-sp2 Suzuki cross coupling, we performed the reaction with
compound 34 for the preparation of the anguinomycin D
skeleton. Alkyl iodide 21 was converted to the boronate
intermediate 38, which was coupled to vinyl iodide 34 under
the same conditions. As expected, the reaction proceeded
smoothly, but during purification on SiO2 the appearance of a
side product was observed. Two fractions were collected, one
of which the pure coupled product 41 in 48% yield could be
characterized, and the second fraction appeared to contain
product 41 and a minor side compound (Scheme 7). It was
decided to separately use both fractions in the next steps.
a Reagents and conditions: (a) 22, LDA, -78 °C, 30 min, then 23, THF,
-78 °C f -10 °C, 26 h, 92%, dr > 97:3. (b) LAH, Et2O, 0 °C f room
temperature, 3.5 h, quant. (c) Oxalyl chloride, DMSO, CH2Cl2, -78 °C,
then NEt3, -78 °C f 0 °C, 50 min, 99%. (d) ent-22, Bu2BOTf (1.0 M in
CH2Cl2), NEt3, -5 °C, 1 h, then -78 °C, 25, CH2Cl2, -78 °C f 0 °C, 2 h,
77%, dr > 87:13.
The syn-aldol 26 was transformed to the Weinreb amide 27
in good yield (86%) following standard procedures. Amide 27
could be crystallized in hexane, and the all-syn configuration
was confirmed by X-ray crystallographic analysis (Scheme 6).
Protection of the alcohol with TBS and treatment with DIBAH
gave aldehyde 28 in excellent yield. A second B-mediated aldol
reaction using the same auxiliary, ent-22, and aldehyde 28
afforded the syn-aldol adduct 29 in excellent diastereoselectivity
(dr > 97:3) and 61% yield. Attempts to convert syn-aldol 29 to
the Weinreb amide 30 proved to be difficult, and even under
forcing reaction conditions, the product 30 was only obtained
in 41% yield. Therefore, we sought to exploit a strength of the
DIOZ auxiliary with its sterically protected carbamate CdO
group and directly cleave the imide by using LAH. Thus, the
syn-aldol 29 could be directly converted to the aldehyde 31. A
Wittig reaction between aldehyde 31 and (carbethoxyethylidene)-
triphenylphosphorane afforded the R,ꢀ-unsaturated ester 32 in
excellent yield (99%) as a single isomer. Reduction with DIBAH
to the alcohol and subsequent MnO2 oxidation gave the R,ꢀ-
unsaturated aldehyde 33 in 80% yield over two steps. Aldehyde
33 (mp ) 75-77 °C) was crystallized from hexane, and X-ray
crystallographic analysis allowed for the unambiguous deter-
mination of the relative configuration of the polyketide chain
(Scheme 6). The transformation of the aldehyde 33 to the vinyl
iodide 34 via a Takai reaction was initially expected to be
problematic due to selectivity issues of this reaction when using
an R-substituted R,ꢀ-unsaturated aldehyde.28 For example,
recently Lipshutz and co-workers reported that a similar
aldehyde was found to be troublesome, and chose to carry out
a three-step replacement for the Takai reaction.29 In our case,
the vinyl iodide 34 was smoothly prepared in excellent yield
and selectivity (dr > 97:3) (Scheme 6). At the same time, we
also prepared a derivative of polyketide chain featuring a TMS
protected hydroxy group at C(17) to evaluate the differences
from the unprotected version in the sp3-sp2 Suzuki cross
coupling. Thus, the R,ꢀ-unsaturated ester 32 was protected with
TMSCl in 77% yield, and subsequent reduction of this substrate
with DIBAH gave the corresponding alcohol, which was
oxidized using Swern conditions to afford aldehyde 35 in
Final steps for completion of the total syntheses were then
performed. Thus, treatment of product 39 under acidic conditions
(PPTS) in a mixture of acetone/water cleaved the acetal in 95%
yield to give the corresponding lactol. Surprisingly, attempted
DMP oxidation of the lactol only oxidized the OH group on
the polyketide chain and did not form the lactone from the
starting lactol. A further oxidation step using MnO2 was then
required to furnish lactone 42 in modest yield (47% over two
steps). Finally, the TBS group was removed using HF ·pyridine
buffered with pyridine, which proved to be crucial in avoiding
degradation of the product. After workup, the crude mixture
was directly purified by semipreparative HPLC to afford
anguinomycin C (1) in 82% yield (Scheme 8).
A similar procedure was adopted for the final transformation
of intermediate 41 to anguinomycin D. Acid-catalyzed cleavage
of the acetal in product 41 afforded the corresponding lactol in
good yield. Because of the poor yield obtained during the two-
step oxidation (DMP and MnO2) for the formation of compound
42 in the synthesis of anguinomycin C (1), we chose to use
PCC for the oxidation of both the alcohol at C(17) and the lactol.
The oxidation was successful and afforded lactone 43, which
was directly treated with HF ·pyridine solution buffered with
pyridine for the final deprotection. This time, to avoid the
aqueous workup, we cooled the reaction to 0 °C and added some
SiO2 to the reaction to quench the excess of HF ·pyridine. The
resulting mixture was then loaded directly onto a column of
SiO2 and chromatographed, affording anguinomycin D (2) in
60% yield (over two steps) (Scheme 8).
(30) Johnson, C. R.; Braun, M. P. J. Am. Chem. Soc. 1993, 115, 11014–
11015.
(28) (a) Takai, K.; Nitta, K.; Utimoto, K. J. Am. Chem. Soc. 1986, 108,
7408–7410. (b) Okazoe, T.; Takai, K.; Utimoto, K. J. Am. Chem. Soc.
1987, 109, 951–953.
(31) Unsuccessful results for the cross-coupling of similar substrates using
K3PO3 instead of Cs2CO3 and AsPh3 were reported by Chakraborty
and co-workers: Chakraborty, T. K.; Goswami, R. K.; Sreekanth, M.
Tetrahedron Lett. 2007, 48, 4075–4078.
(29) Lipshutz, B. H.; Amorelli, B. J. Am. Chem. Soc. 2009, 131, 1396–
1397.
9
1436 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Syntheses, Modeling, and Biology of the Anguinomycins
A R T I C L E S
Scheme 6. The Second Aldol and the Takai Reactiona
a Reagents and conditions: (a) MeONHMe ·HCl, AlMe3, CH2Cl2, 0 °C f room temperature, then 26, CH2Cl2, 0 °C f room temperature, 15 h, 86%. (b)
TBSOTf, 2,6-lutidine, -20 °C f 0 °C, 1 h, 99%. (c) DIBAH (1 M in hexane), THF, -78 °C, 1 h, quant. (d) ent-22, Bu2BOTf (1 M in CH2Cl2), NEt3, -5
°C, 45 min, then -78 °C, 28, CH2Cl2, -78 °C f 0 °C, 3 h, 61%, dr > 97:3. (e) MeONHMe ·HCl, AlMe3, CH2Cl2, 0 °C f room temperature, then 29,
CH2Cl2, 0 °C f room temperature, 68 h, 41%. (f) LAH (1 M in Et2O), toluene, -17 °C, 20 min, 83%. (g) 31, toluene, then (carbethoxyethylidene)triph-
enylphosphorane, room temperature f 35 °C, 5 h, 99%, dr > 97:3. (h) DIBAH (1.0 M in hexane), THF, -78 °C f -10 °C, 1.5 h, 93%. (i) MnO2, CH2Cl2,
room temperature, 2.5 h, 86%. (j) CrCl2, CHI3, THF, 0 °C, 2 h, quant., dr > 97:3. (k) TMSCl, DMAP, NEt3, CH2Cl2, 0 °C, 1 h, 77%. (l) DIBAH (1.0 M in
hexane), CH2Cl2, -78 °C, 1 h, quant. (m) MnO2, CH2Cl2, room temperature, 3.5 h, quant. (n) CrCl2, CHI3, THF, 0 °C, 2.5 h, 88%, dr > 95:5.
Scheme 7. The sp3-sp2 9-MeO-9-BBN-Mediated Suzuki Cross-Couplinga
a Reagents and conditions: (a) 20 (respectively 21), 9-MeO-9-BBN (1.0 M in hexane), tBuLi (1.5 M in pentane), Et2O, then THF, -78 °C f room
temperature, 1 h. (b) 34 (respectively 36), Pd(dppf)Cl2 ·CH2Cl2 (5 mol %), AsPh3 (15 mol %), CsCO3, H2O, DMF, then 37, room temperature, overnight,
80% (respectively 34%). (c) 34, Pd(dppf)Cl2 ·CH2Cl2 (5 mol %), AsPh3 (15 mol %), CsCO3, H2O, DMF, then 38, room temperature, 20 h, 48%.
A second reaction batch containing a mixture of product 41
and a minor byproduct was subjected to the same sequence.
Thus, acid-promoted cleavage of the acetal, oxidation using
PCC, and final removal of the silyl protecting group afforded a
mixture of products that were purified by chromatography on
SiO2. Three compounds were isolated, anguinomycin D (2), the
R,ꢀ-unsaturated aldehyde 44, and trace amounts of compound
4 (Scheme 9). Aldehyde 44 results from opening of the lactone
ring, a problem also encountered by Lautens and co-workers
in their synthesis of callystatin.12g Compound 4 probably
originated from degradation of the boronate intermediate in the
Suzuki reaction.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1437

A R T I C L E S
Bonazzi et al.
Scheme 8. Completion of the Total Syntheses of Anguinomycins C
(1) and D (2)a
that for anguinomycin C, the proton of the OH group at C(17)
was not reported in the original isolation report, but was
identified in the 1H NMR spectrum of the synthetic sample (2.40
ppm). It is generally assumed that compounds as complex as
the anguinomycins are unstable due to the R,ꢀ-unsaturated
lactone, the deconjugated diene, or potential epimerization. In
our laboratory, anguinomycin C could be stored at -20 °C for
more than 1 year without decomposition. Moreover, anguino-
mycins C and D are soluble in EtOH and could be stored as a
solution at -20 °C for several months without decomposition
(as assessed by HPLC).
Biological Evaluation as Inhibitors of Nucleocytoplasmic
Transport. Compounds such as LMB selectively inhibit the
CRM1-mediated nucleocytoplasmic transport by blocking the
interaction between CRM1 and the nuclear export signal (NES)
of the cargo.8b To test if anguinomycins C and D and the
prepared derivatives could also inhibit this process, we inves-
tigated how treatment of cells with these products influences
the intracellular localization of the human protein Rio2. This
protein is a cytoplasmic protein kinase, which is exported from
the nucleus to the cytoplasm by CRM1.17,32 Inhibition of CRM1-
mediated transport results in an accumulation of the Rio2 protein
in the nucleus. HeLa cells were incubated with different
concentrations of anguinomycin C, anguinomycin D, and
derivatives for 2 h and then fixed with paraformaldehyde. LMB
was used as a standard reference. After treatment, the localiza-
tion of Rio2 was determined by indirect immunofluorescence
analysis using a specific antibody directed to human Rio2.
Both anguinomycins C and D caused a strong accumulation
of the Rio2 protein in the nucleus and displayed activity similar
to that of the standard reference LMB, whereas in solvent-treated
control cells the Rio2 protein was localized in the cytoplasm
(Figure 2A and C). The results confirmed that both anguino-
mycins C and D are potent inhibitors of CRM1-mediated
nucleocytoplasmic transport. Both anguinomycins C and D
displayed partial inhibition at 5 nM and reached full inhibition
at 10 nM. These values are comparable to those of LMB, which
fully inhibits nucleocytoplasmic transport above 1 nM (Figure
2C).
a Reagents and conditions: For anguinomycin C, (a) PPTS, acetone/water
(3:1), room temperature, 22 h, 95%. (b) (i) DMP, CH2Cl2, room temperature,
4 h; (ii) MnO2, CH2Cl2, room temperature, 14 h, 47%. (e) HF ·pyridine,
pyridine, room temperature, 4.5 days, 82%. For anguinomycin D, (c) PPTS,
acetone/water (5/1), room temperature, 22 h, 91%. (d) PCC, 4 Å MS, AcOH,
CH2Cl2, room temperature, 1.5 h. (f) HF ·pyridine, pyridine, room temper-
ature, 4.5 days, 60% (two steps).
Scheme 9. Compounds Generated by Using a Mixture of
Diastereoisomers at C(17)
Very interesting results were obtained for the R,ꢀ-unsaturated
aldehyde 44 and for lactone 4. The results show that aldehyde
44 completely blocks CRM1-dependent nuclear export at 50
nM, with partial inhibition already observable at 25 nM (Figure
2A). This compound displays the same side chain as anguino-
mycins C and D, but the lactone has been replaced by the R,ꢀ-
unsaturated aldehyde. Even though this replacement results in
a 5-fold loss in activity, compound 44 can still be considered
highly active, as it completely blocks CRM1 at 50 nM
concentration. Even more surprising was the high activity
displayed by product 4, which contains a truncated polyketide
chain. The compound caused the accumulation of the Rio2
protein in the nucleus at 25 nM (Figure 2A). This result
highlights the fundamental role of the R,ꢀ-unsaturated lactone
in the inhibition of CRM1-mediated nucleocytoplasmic trans-
port, supporting the hypothesis that the side chain mainly plays
a role in the molecular recognition and modulation of the
selectivity. Even though compound 4 was a drastic simplification
of the natural anguinomycins C and D, the activity decreased
only marginally. From the synthetic point of view, lactone 4
Both anguinomycins C (1) and D (2) were isolated as a
colorless oils, and their optical rotation values, the UV traces,
and IR spectra of the synthetic products match with those
reported in the literature,16a and the high resolution ESI
confirmed the correct masses. The NMR spectroscopic data of
synthetic anguinomycins C and D were identical to those
reported in the literature16a (see Supporting Information), except
(32) (a) Rouquette, J.; Choesmel, V.; Gleizes, P. E. EMBO J. 2005, 24,
2862–2872. (b) Zemp, I.; Wild, T.; O’Donohue, M.-F.; Wandrey, F.;
Widmann, B.; Gleizes, P.-E.; Kutay, U. J. Cell Biol. 2009, 185, 1167–
1180.
9
1438 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Syntheses, Modeling, and Biology of the Anguinomycins
A R T I C L E S
Figure 2. Anguinomycins C and D and different derivatives block CRM1-mediated nuclear export in vivo. HeLa cells were incubated with the respective
compounds for 2 h at 37 °C. Cells were then fixed in 4% paraformaldehyde, and human Rio2 protein was detected by indirect immunofluorescence analysis
using a Rio2-specific antibody. Pictures were taken using a Leica confocal microscope. (A) Inhibition of CRM1-dependent nuclear export of Rio2 in HeLa
Cells by anguinomycins C (1) and D (2), compounds 4 and 44. (B) Inhibition of CRM1-dependent nuclear export of Rio2 in HeLa Cells by 3. (C) LMB
served as standard reference; incubation with the solvent EtOH (1% fc) served as negative control.
would be much easier to prepare than the natural compounds,
resulting in a gain of time and resources. To understand on an
atomic level the reasons for LMB/anguinomycin binding, we
carried out modeling studies on the recently published structures
of CRM1.10
the inhibitor.33 Current docking programs capture noncovalent
interactions well, but they are largely incapable of estimating
the binding energy associated with the formation of a covalent
bond between the protein and the inhibitor. An additional
challenge is the flexible structure of LMB, which can adopt a
large number of possible conformations. However, covalent
binding also simplifies modeling because it introduces an anchor
point that reduces the number of orientations to be evaluated.
We modeled the LMB binding pose manually and optimized
its conformation by an all-atom energy minimization using
PLOP.34 We found that the shape and the electrostatic properties
of the CRM1 binding pocket and the rigid parts of the inhibitor
restrict the binding mode of LMB to a large extent. Our models
suggest a mechanism for inhibitor binding and help to explain
why only certain derivatives bind to CRM1. They could be
useful for the design of new analogues.
Modeling of the Interaction between LMB and CRM1. In
1999, Kudo and colleagues demonstrated that the natural product
leptomycin B (LMB) inhibits CRM1 through selective alkylation
of a cysteine residue.9a,b Earlier in 2009, two X-ray structures
of full-length mouse and human CRM1 have been published,10a,c
which reveal that the reactive cysteine (Cys 528 in human
CRM1) locates in a hydrophobic groove responsible for binding
of nuclear export signals (NES), as shown in the CRM1-SNUPN
complex (Figure 3). The NES binding site of CRM1 is very
similar in the two structures with a rmsd of 0.31 Å (see
Supporting Information Figure 1). In the following, we inves-
tigated how LMB and other compounds tested above bind to
and interact with CRM1.
We started with the modeling of the LMB-CRM1 complex
at Cys 528, which makes a covalent bond with the lactone ring
of LMB. Our first observation was that, due to steric restraints,
Molecular modeling of covalent inhibitors is considered
challenging because the binding energy results from both
covalent and noncovalent interactions between the protein and
(33) Lindvall, M. K. Curr. Pharm. Des. 2002, 8, 1673–1681.
(34) Kalyanaraman, C.; Bernacki, K.; Jacobson, M. P. Biochemistry 2005,
44, 2059–2071.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1439

A R T I C L E S
Bonazzi et al.
Figure 3. X-ray structure of the CRM1-SNUPN complex.10a (A) CRM1 is depicted as a surface model in white. SNUPN is displayed as ribbon representation
in cyan and magenta. (B) Close-up view on the N-terminal nuclear export signal (NES) of SNUPN (residues 1-16). Leptomycin B binds to Cys 528, which
lies in a hydrophobic groove of CRM1. The position of the sulfur atom of Cys 528 is highlighted in yellow. (C) Hydrophobic residues of the SNUPN NES
(only side chains shown) define five pockets in the CRM1 groove.
Figure 4. Cartoon representations of the CRM1 binding groove and
potential orientations of LMB in the binding pocket. Colored circles indicate
the location of important residues close to the reactive cysteine 528. (A)
The CRM1 groove has five pronounced hydrophobic pockets, displayed
by dashed rings. (B) Orientation 1: LMB occupies pockets one to four. (C)
Orientation 2: LMB occupies pockets four and five, but does not fit entirely
into the binding groove.
Figure 5. The local environment of Cys 528 orients the lactone ring of
LMB. (A) Cys 528 is covalently bonded to the lactone ring in the LMB
adduct model. Glu 529 and Lys 537 are two charged amino acids in close
proximity of Cys 528. Glu 529 appears to be ideally placed to act as a
“general base catalyst” in the Michael reaction, while Lys 537 could stabilize
a negative charge in the lactone ring in the transition state. (B) Comparison
of the crystallographic Cys 528 (depicted in gray) with two possible LMB
adduct stereoisomers. Carbon atoms are displayed in yellow and magenta
for the (S) and (R) configurations.
LMB can fit only in two orientations into the binding groove
of CRM1. CRM1 has five pockets that bind the hydrophobic
SNUPN residues (Figures 3C and 4A).10 Either LMB interacts
with pockets one up to four of CRM1 (orientation 1, Figure
4B), or LMB occupies pockets four and five (orientation 2,
Figure 4C). We favor orientation 1 over 2 because LMB has
not enough space to fit entirely into the groove in orientation
2. In addition, orientation 1 allows clearly more contacts between
LMB and CRM1.
Structural analysis of the local environment of Cys 528
provides further support for orientation 1 (Figure 5A). In the
course of a Michael addition, a negative charge on the lactone
O atom (enolate) is generated. In orientation 2, that enolate
would point toward the carboxylate group of Glu 529 and apolar
Leu 525. Both of these interactions are unfavorable. On the
contrary, in orientation 1, the enolate points toward free space
in pocket five, and there is Lys 537 nearby, which could stabilize
the negative charge in the transition state. Based on proximity,
Glu 529 and Lys 537 are possible candidates as base-catalysts
of the reaction. Both residues are 100% conserved among
nuclear export proteins.
The results after the energy minimization are shown in Figure
5B. To fit the lactone ring in the (R) configuration, the S atom
of cysteine 528 has to move by 1.1 Å respective to the crystal
structure. The lactone ring is easier to accommodate in the (S)
configuration into the CRM1 pocket because only a small
movement of 0.4 Å is necessary. However, this prediction of
the absolute configuration is not absolute because it has been
modeled into a rigid receptor structure.
After the determination of the lactone ring orientation,
modeling of the hydrophobic tail of LMB was straightforward.
In the modeling of the tail, we strived toward burying
hydrophobic LMB entities as much as possible in the groove.
The conjugated diene part of callystatin, a LMB analogue, has
been shown to be essential for binding.35 We concluded
therefore that this group makes specific interactions with the
groove.
The final models for LMB and the small, but still active
compound 4 are presented in Figure 6A and B. Both are shown
in orientation 1 and (S) configuration for carbon C3, respectively.
LMB fills up most of the groove that serves as the molecular
recognition site for NES sequences of CRM1 export substrates/
(CRM1) cargo proteins. Within the limitations of rigid body
Our next finding about the lactone ring concerns stereose-
lectivity of the Michael addition reaction. The thiol of Cys 528
can attack the unsaturated lactone ring of LMB from two
different faces. The LMB-adduct has a new stereogenic center
at carbon C3 (see Figure 6D for atom numbering). To investigate
the configuration of the LMB adduct at C3, we modeled both
the (R) and the (S) stereoisomers into the CRM1 binding pocket.
(35) Murakami, N.; Sugimoto, M.; Nakajima, T.; Kawanishi, M.; Tsutsui,
Y.; Kobayashi, M. Bioorg. Med. Chem. 2000, 8, 2651–2661.
9
1440 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010

Syntheses, Modeling, and Biology of the Anguinomycins
A R T I C L E S
Figure 6. Models of CRM1-inhibitor complexes. CRM1 carbons, oxygens, and nitrogen are in gray, red, and blue, respectively. Inhibitors depicted with
yellow carbon atoms. (A) Sphere model of Leptomycin B adduct. (B) Sphere model of compound 4 adduct. (C) Superposition of LMB adduct with NES
residues of SNUPN. LMB depicted in yellow with hydrogens, SNUPN residues in magenta. (D) Atom numbering of LMB adduct, according to ref 5.
Scheme 10. Preparation of the Anguinomycin/Terpene Hybrida
modeling, we think that our model of LMB may be accurate
up to carbon atom C15: as described above, steric restraints of
the pocket and the arrangement of the charged amino acids Glu
529 and Lys 537 direct the orientation of the lactone and place
it between pockets four and five. After the lactone, LMB
contains two conjugated dienes, which are rigid stretches in the
molecule. Those fit between pockets two and four. Between the
dienes, three rotable bonds between C9 and C12 allow a necessary
turn in the bottleneck of the groove, near Leu 525. The ethyl group
at C8 has been shown to be important for cytotoxicity:35 it
protrudes deeply into pocket four and mimics thereby phenyl-
alanine 12 from the SNUPN NES (Figure 6C). Also, the methyl
groups C29 and C31 mimic hydrophobic SNUPN side chains.
The part of the LMB molecule after the two conjugated dienes
a Reagents and conditions: (a) CBr4, PPh3, 2,6-lutidine, CH2Cl2, 0 °C,
(C16-C24) is sterically less restricted in the groove and
occupies pockets one and two. Seven successive rotable bonds
2 h, 91%. (b) nBuLi, THF, -78 °C to room temperature then NH4Cl. (c)
Cp2ZrHCl, 0 °C f room temperature, THF, 1 h, then -78 °C, I2 (87%
make this part very flexible. We can only speculate that the
keto and hydroxyl groups of LMB may interact with a patch of
polar groups between CRM1 residues Val 518 and Lys 522.
The terminal carboxyl group could make favorable electrostatic
interactions either with His 558 or Lys 514, two positively
charged residues near pocket one. Compound 4 is obviously
much shorter than LMB but is likely to react via a similar
mechanism with CRM1. It retains the ability to block recognition
and binding of NES by occupying pockets three and four.
In conclusion, our manual modeling of LMB was guided by
the geometry of the ligand and the shape of the CRM1 binding
pocket. It yielded plausible and snugly fitting inhibitor confor-
mations rationalizing previous structure-activity experiments.
In addition, structural analysis of the local environment of Cys
528 suggests that Glu 529 and Lys 537 are likely taking part in
the reaction forming the covalent bond.
Synthesis and Evaluation of Anguinomycin/Terpene Hybrid.
At the outset of the research program, we wanted to evaluate
the hypothesis whether the polyketide chain of anguinomycin
mimics the leucine-rich nuclear export signal (NES) of cargo
proteins. This helical peptide NES region features hydrophobic
amino acids and has been postulated to weakly interact with
CRM1.11 This hypothesis is supported by the modeling studies
discussed above, where the tail of LMB/anguinomycin snugly
fits into the binding groove. Therefore, replacement of the
polyketide chain by a simpler hydrophobic moiety was desired,
and the corresponding compound 3, which can be considered a
hybrid between anguinomycin and a terpene, was targeted.
over two steps). (d) 21, 9-MeO-9-BBN (1.0 M in hexane), tBuLi (1.5 M in
pentane), Et2O, then THF, -78 °C f room temperature, 1 h, then 47,
Pd(dppf)Cl2 ·CH2Cl2 (5 mol %), AsPh3 (15 mol %), CsCO3, H2O, DMF,
room temperature, overnight, 87%. (e) PPTS, acetone/water (3/1), room
temperature, 2 h. (f) PCC, 4 Å MS, CH2Cl2, room temperature (53% over
two steps).
Citronellal (45) was converted to the corresponding dibro-
moolefin36 followed by transformation to the alkyne 46 via the
Fritsch-Buttenberg-Wiechell rearrangement.37 The selective
formation of the (E)-iodoolefin 47 was achieved by hydrozir-
conation using Schwartz reagent and quenching with I2.38 All
attempts toward the formation of compound 47 via Takai
olefination28 from citronellal were unsuccessful due to the lack
of selectivity. The coupling of the side-chain to the alkyl iodide
21 was achieved using the previously described sp3-sp2 Suzuki
cross-coupling. Finally, removal of the isopropyl group and
oxidation to the lactone using PCC gave the desired analogue
3 (Scheme 10).
To analyze whether 3 is capable of blocking CRM1-mediated
nuclear export, we incubated HeLa cells with increasing
concentrations of this compound and analyzed the localization
of human Rio2 protein (Figure 2B). Interestingly, 3 displayed
full inhibition of CRM1-dependent nuclear export at 200 nM,
(36) Corey, E. J.; Fuchs, P. L. Tetrahedron Lett. 1972, 13, 3769–3772.
Molander, G. A.; Yokoyama, Y. J. Org. Chem. 2006, 71, 2493–2498.
(37) Minatti, A.; Do¨tz, K. H. J. Org. Chem. 2005, 70, 3745–3748.
(38) Huang, Z.; Negishi, E. Org. Lett. 2006, 8, 3675–3678.
9
J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010 1441

A R T I C L E S
Bonazzi et al.
with partial inhibition of export starting at 25 nM. This result
demonstrates that the polyketide chain of anguinomycins C and
D can be replaced by a hydrophobic moiety without a strong
loss in activity. Moreover, the observed reduction in activity
for 3 might be partially due to an increased capability of the
compound to dissolve into lipid bilayers, as its hydrophobic
moiety strongly resembles the apolar parts of phospholipids.
This might significantly reduce the active concentration of the
compound present within the cell.
first total syntheses of anguinomycins C and D were achieved
in total 29 steps with a longest linear sequence of 18 steps from
(R)-4-isopropyl-5,5-diphenyloxazolidin-2-one and with an over-
all yield of 6.7% and 6.0%, respectively. Noteworthy steps
included: (1) the Cr(III)-catalyzed hetero Diels-Alder giving
straightforward access to the dihydropyran ring in high yield
and selectivity and (2) the Negishi cross-coupling under
stereoinversion that furnished the cis product for both angui-
nomycins C and D from a common starting material. This
exemplifies the potential of this reaction in total synthesis, and
we are confident that more applications will be demonstrated
in the future. The Seebach modification of the Evans auxiliary
demonstrated its utility for the synthesis of polyketide chains
with regard to yield, selectivity, and simplicity. The total
synthesis also definitively established the absolute configuration
of anguinomycins C and D as (5R,10R,16R,18S,19R,20S).
Discussion
The biological results obtained in this work can be compared
to the SAR investigation of other groups on related compounds.
Kobayashi and co-workers reported biological investigations on
callystatin and its derivatives,39 proving the fundamental
importance of the lactone fragment for the activity. The same
studies showed that modifications of the polyketide chain cause
a loss in activity of about 1 order of magnitude as compared to
the natural callystatin. In addition, inversion of configuration
at C(5) caused a 350-fold loss in activity. Kalesse and
co-workers performed similar investigations on ratjadone,40
showing again that the lactone was crucial for activity. Inversion
of configuration at C(10) resulted in a complete loss of activity.
Recently, Mutka and co-workers reported new derivatives of
LMB displaying the same potency as the parent compound, but
with a higher selectivity between normal and cancer cells.14e
Structure-activity relationship studies performed on anguino-
mycins C and D and their derivatives as described above were
in agreement with the literature. The high activity of compound
4 (25 nM) can be explained by the fact that this derivative is a
shortened version of the parent compound, which lacks the
polyketide chain but still contains the important lactone moiety.
This is an interesting finding, as for many natural products,
omission of more than one-half of the C skeleton often leads to
complete loss of bioactivity.
Following the total syntheses of the two natural compounds,
we investigated the biology of these products and more precisely
the mode of action and the structure-activity relationships.
Several derivatives were prepared and submitted for biological
evaluation. The results confirmed the crucial importance of the
lactone ring for the activity and also that the activity can be
modulated by changing the side chain, which mainly plays a
role in the molecular recognition. Both anguinomycins C and
D displayed a strong inhibition of the CRM1-mediated nucleo-
cytoplasmic transport at 10 nM, confirming their powerful
activity. Moreover, the truncated analogue 4 that caused
accumulation of the Rio2 protein in the nucleus at 25 nM was
obtained. This compound represents a drastic simplification of
the parent natural product, and it maintained strong activity
despite omission of more than one-half of the carbon skeleton.
This truncated analogue 4 can be considered a starting point
for the development of similar, short analogues of LMB and
anguinomycins and can thus contribute to the search for more
powerful and selective nucleocytoplasmic transport inhibitors
for cancer treatment.
Conclusion
Acknowledgment. This work is dedicated to Prof. Eric N.
Jacobsen on the occasion of his 50th birthday. K.G. is a European
Young Investigator (EURYI). We thank the SNF for support of
this work (3100AO-118053, PBZHP3-123277, 200021-115918/1,
and PE002-117136/1).
The need to find more selective inhibitors of nucleocytoplas-
mic transport and the unveiled structure of anguinomycins C
and D prompted us to develop a synthesis for these natural
products. With its six unknown stereogenic centers, the lactone
ring, the two diene systems, and the polyketide chain, angui-
nomycins presented a challenging target for total synthesis. The
Supporting Information Available: Experimental procedures,
characterization, spectra, biological, and modeling methods. This
material is available free of charge via the Internet at http://
pubs.acs.org.
(39) Murakami, N.; Sugimoto, M.; Kobayashi, M. Bioorg. Med. Chem.
2001, 9, 57–67.
(40) Kalesse, M.; Christmann, M.; Bhatt, U.; Quitschalle, M.; Claus, E.;
Saeed, A.; Burzlaff, A.; Kasper, C.; Haustedt, L. O.; Hofer, E.; Scheper,
T.; Beil, W. ChemBioChem 2001, 2, 709–714.
JA9097093
9
1442 J. AM. CHEM. SOC. VOL. 132, NO. 4, 2010