Convergent Synthesis and Discovery of a Natural Product-Inspired Paralog-Selective Hsp90 Inhibitor

Article abstract of DOI:10.1021/ol2019828

A convergent synthesis of benzoquinone ansamycin analogs is described that proceeds by a sequence of metallacycle-mediated alkyne-alkyne coupling, followed by site- and stereoselective dihydroxylation and global carbamate formation. These studies have led

Full text of DOI:10.1021/ol2019828

ORGANIC
LETTERS
2011
Vol. 13, No. 19
5108–5111
Convergent Synthesis and Discovery
of a Natural Product-Inspired Paralog-
Selective Hsp90 Inhibitor
Valer Jeso,† Lisa Cherry,‡ Todd K. Macklin,† Subhas Chandra Pan,†
Philip V. LoGrasso,*,‡ and Glenn C. Micalizio*,†
Department of Chemistry, Department of Molecular Therapeutics, and the Translational
Research Institute, The Scripps Research Institute, Jupiter, Florida 33458, United States
micalizio@scripps.edu; lograsso@scripps.edu
Received July 22, 2011
ABSTRACT
A convergent synthesis of benzoquinone ansamycin analogs is described that proceeds by a sequence of metallacycle-mediated alkyneÀalkyne
coupling, followed by site- and stereoselective dihydroxylation and global carbamate formation. These studies have led to (1) validation of
alkyneÀalkyne coupling to produce geldanamycin analogs that lack the problematic quinone, (2) the discovery that C6ÀC7 bis-carbamate
functionality is compatible with Hsp90 inhibition, and (3) the identification of 1 as a nonquinone geldanamycin-inspired paralog-selective Hsp90
inhibitor.
Benzoquinone ansamycins are polyketide-derived natural
products that display a broad range of antitumor, anti-
bacterial, antifungal, and antiprotozoal activities (Figure
1A).1 While efforts in chemical synthesis have targeted this
natural product class for over 30 years, with the first total
synthesis appearing in 1989,2 the discovery that benzoqui-
none ansamycins are potent inhibitors of Hsp90 has led to
a substantial increase in activity in this area with the goal
of developing novel anticancer chemotherapeutic agents.3
Hsp90 is widely held as a promising therapeutic target
for cancer, as it plays a central role in controlling the
post-translational conformational maturation and activation
of a large number of oncogenic client proteins in a highly
regulated and ATP-fueled manner (i.e., Bcr-Abl, Raf-1,
HER2, Cdk4, Akt, ErbB2, and HIF-1R).4 Additionally, a
number of mutant oncoproteins require Hsp90 function,
whereas their wild-type counterparts are either not depen-
dent or only weakly dependent upon this machinery
(v-SRc, EGFR, B-Raf).5 Further enhancing the potential
value of Hsp90 as a therapeutic target is the fact that it is
constitutively expressed at 2- to 10-fold higher levels in
tumor cells compared to their “normal” counterparts.6
Overall, inhibition of Hsp90 results in the proteasome-
mediated depletion of a number of proteins involved in cell
signaling, proliferation, survival, immortalization, invasion,
† Department of Chemistry.
‡ Department of Molecular Therapeutics and the Translational Research
Institute.
(1) For recent reviews on the biological activity of benzoquinone
ansamycins, see: (a) Xu, W.; Neckers, L. Clin. Cancer Res. 2007, 13,
1625–1629. (b) Miyata, Y. Curr. Pharm. Des. 2005, 11, 1131–1138.
(2) (a) Baker, R.; Castro, J. L. J. Chem. Soc., Chem. Commun. 1989,
378–381. (b) Baker, R.; Castro, J. L. J. Chem. Soc., Perkin Trans. 1 1990,
47–65.
(4) Neckers, L. Trends Mol. Med. 2002, 8, S55–S61.
(5) Chandarlapaty, S.; Sawai, A.; Ye, Q.; Scott, A.; Silinski, M.;
Huang, K.; Fadden, P.; Partridge, J.; Hall, S.; Steed, P.; Norton, L.;
Rosen, N.; Solit, D. B. Clin. Cancer Res. 2008, 14, 240–248.
(6) Ferrarini, M.; Heltai, S.; Zocchi, M. R.; Rugarli, C. Int. J. Cancer
1992, 51, 613–619.
(3) Whitesell, L.; Mimnaugh, E. G.; De Costa, B.; Myers, C. E.;
Neckers, L. M. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 8324–8328.
r
10.1021/ol2019828
Published on Web 08/25/2011
2011 American Chemical Society

metastasis, and angiogenesis. This promising profile is
further enhanced by a fundamental difference in the nature
of Hsp90 in cancer cells, where it is maintained in an
activated multichaperone complex that binds 100 times more
tightly to Hsp90 inhibitors than Hsp90 from normal cells.7
The ready availability of geldanamycin from fermenta-
tion promptly placed it at the center of early studies aiming
to advance a clinically relevant chemotherapeutic agent.
Early studies revealed a number of problems with this
natural product that included poor solubility, dose-limit-
ing toxicity, cellular instability, and nearly equipotent
activity for inhibition of GRP94;an endoplasmic reticu-
lum paralog of the Hsp90 family whose clients include
Toll-like receptors, integrins, and immunoglobulins (IgGs).
While derivatives have been advanced with enhanced
solubility profiles (17-AAG, IPI-504, and 17-DMAG;
Figure 1B) and paralog selectivity, these semisynthetic
derivatives retain the metabolic liability associated with
the quinone/hydroquinone moiety central to the natural
benzoquinone ansamycin skeleton.8 Recently, related nat-
ural products that lack the problematic quinone have been
discovered, synthesized, and demonstrated to also have
potentHsp90-inhibitoryprofiles(i.e., reblastatinandauto-
lytimycin, Figure 1C), albeit harboring molecular features
that likely impart similarly suboptimal solubility profiles.9
We have initiated a project to secure a robust synthesis
pathway to this class of natural products to enable a search
for natural product-like Hsp90 inhibitors that is unbound
by constraints associated with biosynthesis, thereby pro-
viding maximum flexibility to discover superior candidates
for development. Here we describe our first-generation
chemical strategy, the synthesis of a focused panel of
synthetic ansamycins, and the identification of a natural
product-like lead as a paralog-selective Hsp90 inhibitor.
Our initial foray in to the benzoquinone ansamycins
resulted in a concise (15-step longest linear) synthesis of
macbecin I (Figure 2A).10 While defining a successful
chemical synthesis, this study suffered from a technical
challengesthatincluded (1)difficulties establishing the C15
stereocenter with a high degree of control, (2) the establish-
ment of the quinone proceeded in low-moderate yield, (3)
the Ti-mediated alkyneÀaldehyde coupling provided a
mixture of stereoisomers atC7, the minorproductofwhich
was not advanced to the natural product, and (4) the poly-
unsaturated C1ÀC7 fragment employed was quite sensi-
tive and difficult to work with. While these issues did not
Figure 1. Introduction to benzoquinone ansamycin Hsp90 in-
hibitors and related natural products.
interfere with a successful synthesis of the natural product,
they were perceived to be liabilities for the current pursuit.
As depicted in Figure 2B, we targeted a chemical path-
way to the ansamycin skeleton that would afford deriva-
tives that (1) lack stereochemical information at C15 (as in
geldanamycin), (2) possess a saturated C4ÀC5 linkage (as
in reblastatin and autolytimycin), (3) lack the central
quinone,11 and (4) harbor distinct C6 ether functionality.
These design criterea led to the synthesis strategy sum-
marized in Figure 2C. Here, the C7ÀC8 linkage is targeted
by metallacycle-mediated reductive coupling between
alkynes,12 a feature that was anticipated to resolve our
previous issues with diastereoselection encountered in the
alkyneÀaldehyde coupling process (i.e., Figure 2A). With
this basic synthesis strategy in mind, we required simple
C1ÀC7-containing terminal alkyne coupling partners
(2À4), a few suitably functionalized anilines to incorporate
as the aromatic core (5À7), and sufficient quantities of the
stereodefined C8ÀC15 alkyne-containing subunit (8) to
enable late-stage coupling sequences for analog genera-
tion (Figure 3).13 The rationale for selecting these coupling
(7) Kamal, A.; Thao, L.; Sensintaffar, J.; Zhang, L.; Boehm, M. F.;
Fritz, L. C.; Burrows, F. J. Nature 2003, 425, 407–410.
(8) (a) Snader, K. M.; Neckers, L. M.; Wishnuvajjala, B. R.; Sausville,
E. A.; Xu, W.; Rosser, M. R.; Nicchitta, C.; Isaacs, J. A. (The Government
of the United States of America; National Institutes of Health). Geldana-
mycin derivatives useful for the treatment of cancer. WO application
200236574, May 10, 2002. For structures of natural products bound to Hsp90
and GRP94, see: (b) Immormino, R. M.; Metzger, L. E. IV; Reardon, P. N.;
Dollins, D. E.; Blagg, B. S. J.; Gewirth, D. T. J. Mol. Biol. 2009, 388, 1033–
1042.
(9) Wrona, I. E.; Gozman, A.; Taldone, T.; Chiosis, G.; Panek, J. S.
J. Org. Chem. 2010, 75, 2820–2835.
(10) (a) Belardi, J. K.; Micalizio, G. C. Angew. Chem., Int. Ed. 2008,
47, 4005–4008. See also: (b) Belardi, J. K.; Micalizio, G. C. Org. Lett.
2006, 8, 2409–2412.
(11) For an early example of using engineered biosynthesis for the
preparation of geldnamaycin analogs lacking a central quinone, see:
Patel, K.; Piagentini, M.; Rascher, A.; Tian, Z.-Q.; Buchanan, G. O.;
Regentin, R.; Hu, Z.; Hutchinson, C. R.; McDaniel, R. Chem. Biol.
2004, 11, 1625–1633.
(12) (a) Shimp, H. L.; Micalizio, G. C. Org. Lett. 2005, 7, 5111–5114.
(b) Perez, L. J.; Shimp, H. L.; Micalizio, G. C. J. Org. Chem. 2009, 74,
7211–7219. For examples in natural product synthesis, see: (c) Reichard,
H. A.; Rieger, J. C.; Micalizio, G. C. Angew. Chem., Int. Ed. 2008, 47,
7837–7840. (d) Shimp, H. L.; Micalizio, G. C. Tetrahedron 2009, 65,
5908–5915. (e) Wu, J.; Panek, J. S. Angew. Chem., Int. Ed. 2010, 49,
6165–6168.
Org. Lett., Vol. 13, No. 19, 2011
5109

in the formation of the functionalized coupling partner
10 in 68% yield. Formation of a TiÀalkyne complex
(Ti(O-i-Pr)4, c-C5H9MgCl, Et2O, À78 to À30 °C), fol-
lowed by exposure to the C1ÀC7 terminal alkyne
coupling partner 2 led to formation of the fully func-
tionalized triene 11 in 47% yield (the remainder of the
starting alkyne 10 could be isolated as its corresponding
reduction product, the Z-alkene). Here, the C7ÀC8
bond was forged in concert with the establishment of
each stereodefined alkene of the 1,3-diene motif
(stereoselectivity g20:1; regioselectivity = 5:1). This
demonstration of the utility of Ti-mediated reductive
cross-coupling of alkynes is significant, as bimolecular
CÀC bond formation between the preformed TiÀ
alkyne complex and terminal alkyne proceeds faster
than reaction with the R,β-unsaturated ester group in 2,
an electron-deficient functionality of potential sensitivity in
reactions with preformed organometallic intermediates.
Moving on, saponification of the ethyl ester (LiOH,
THF, MeOH, H2O) and macrolactamization (BOPCl,
i-Pr2NEt, PhMe, 80 °C) delivered the macrocyclic triene
12 in 59% yield. Site and stereoselective dihydroxyla-
tion was then accomplished under the conditions of
Sharpless16 and generated the stereodefined diol 13
in 60% yield as a single isomer. Finally, formation of
the bis-carbamate proceeded by sequential treatment
with trichloroacetyl isocyanate and i-Pr2NEt/MeOH to
generate the fully functionalized analog 1 in nine steps
from the coupling partners.17 While this final func-
tionalization reaction reproducibly proceeded in poor
yield,18 the process was sufficient to generate material
suitable for biochemical evaluation. Efforts to optimize
this final step will be addressed in future studies.
Figure 2. From a synthesis of macbecin I to a general strategy to
nonbenzoquinone ansamycin derivatives.
partners was based on a few simple considerations: (1)
exploring the impact of conformational restriction in the
C1ÀC5 region (i.e., use of A-1,3 strain;14 3, 4) and (2)
assessing the impact of aromatic motifs that contain
H-bond acceptors (5, 6), as well as providing a bias to
support the cis-amide geometry of relevance in binding
Hsp90 (7).
Beginning with efforts targeting the methoxy-substi-
tuted analogs derived from 5, LiÀhalogen exchange
(t-BuLi, Et2O, À78 to 20 °C) followed by addition of
aldehyde 8 led to the production of the benzylic alcohol
9 in 71% yield (dr = 1:1). Conversion to the corre-
sponding xanthate (NaHMDS, CS2, then MeI), fol-
lowed by Barton deoxygenation (Et3B, Bu3SnH,
PhMe, 20 °C)15 and reductive deprotection of the ani-
line (Pd(PPh3)4, N,N-dimethylbarbituric acid) resulted
Using the general aspects of the synthesis pathway de-
picted in Figure 4, a collection of benzoquinone ansamycin
derivatives were prepared (Figure 5A). These derivatives
vary with respect to (1) the nature of the aromatic core,
(2) the substitution and stereochemistry at C4, and (3)
the nature of the functionality at C6ÀC7. It is well-
known that the C7-carbamate of the benzoquinone
ansamycins plays an important role in binding to
Hsp90, and as such, we targeted derivatives that uni-
formly retain this functionality. Surprisingly, during
the course of our studies, we observed that the final
hydrolysis reaction for generating the carbamate at
C7 (of the resulting N-trichloroacetyl carbamate) was
(13) The C8ÀC15 subunit was available using our previously de-
scribed synthesis strategy, while the other subunits were prepared by
well-established methods. See the Supporting Information for details.
(14) Hoffmann, R. W. Angew. Chem., Int. Ed. 1992, 31, 1124–1134.
(15) Barton, D. H. R.; McCombie, S. W. J. Chem. Soc., Perkin Trans.
1 1975, 1574–1585.
(16) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem.
Rev. 1994, 94, 2483–2547.
(17) (a) Kocovsky, P. Tetrahedron Lett. 1986, 27, 5521. (b) Carter,
K. D.; Panek, J. S. Org. Lett. 2004, 6, 55–57.
(18) Investigation of alternative reagents and reaction conditions to
address the low yield encountered in this step is currently underway. The
installation of this carbamate in total syntheses has been accomplished
with NaOCN, TFA (see ref 2 for an early example).
Figure 3. Coupling partners employed.
5110
Org. Lett., Vol. 13, No. 19, 2011

this compound class withrespect tochanges in substitution
at C4 and the aromatic nucleus.
Figure 4. Synthesis pathway employed for ansamycin ana-
logs.
unpredictable, generating in some cases the corre-
sponding imides (18À21).19
Figure 5. Benzoquinone ansamycin analogs prepared. (a) EC50
values are from the average of 10 experiments for geldanamycin
(each read in duplicate), two experiments ( standard deviation
for 1 (each read in triplicate), and three experiments for 14, 17,
and 19 (each read in triplicate).
To evaluate the Hsp90 inhibitory potential of the
natural product-like derivatives prepared, we utilized
fluorescence polarization (FP) competition binding as-
says for Hsp90R, Hsp90β, and GRP94, the endoplasmic
reticulum paralog of Hsp90.20 Consistent with previous
reports, geldanamycin was found to be a potent binder to
all three proteins with an apparent EC50 near 70 nM.
Perhaps not surprisingly, all imide-containing derivatives
(18À21) lacked affinity for any of the biological targets
explored (EC50 g 10 μM). Interestingly, for the remaining
series (1, 14À17), only compound 1 showed substantial
affinity for Hsp90R and Hsp90β with EC50’s of 1.7 μM and
980 nM, respectively (Figure 5B), yet having no affinity for
GPR94 up to 10 μM. These results reveal that C6 carba-
mate functionality is compatible with Hsp90 inhibition
and offers a means to accomplish paralog selectivity
between Hsp90 and GRP94. Further, the lack of affinity
associated with 14À17 demonstrates the sensitive nature of
Overall, we report the establishment of a chemical
synthesis pathway to unnatural derivatives of benzo-
quinone ansamycins that proceeds by metallacycle-
mediated alkyneÀalkyne coupling. This synthesis path-
way was employed to generate a preliminary panel of
natural product-like analogs and led to the discovery
of a paralog-selective Hsp90 inhibitor (1). While affi-
nities of 1 to Hsp90 are still more than 1 order of
magnitude lower than geldanamycin, the selectivity profile
is unique, and the molecular skeleton lacks the problematic
quinone central to geldanamycin and clinically rele-
vant analogs. These discoveries solidify the relevance of
pursuing Hsp90 inhbitors based on the C6,C7-bis carba-
mate functionality resident in 1, and establish the signifi-
cance of alkyneÀalkyne coupling for the synthesis of
related natural product-like analogs. We look forward to
future developments that build on these intitial findings.
(19) These compounds are most likely generated via the haloform-
like expulsion of CCl3 during the collapse of the tetrahedral inter-
À
mediate during final methanolysis. We suspect that this anomalous
reaction pathway will be supressed by conducting the hydrolysis under
aqueous conditions. This was not pursued in the current study but will be
addressed in future work.
(20) (a) Kim, J.; Felts, S.; Llauger, L.; He, H.; Huezo, H.; Rosen, N.;
Chiosis, G. J. Biomol. Screen. 2004, 9, 375–381. (b) Ge, J.; Mormant, E.;
Porter, J. R.; Ali, J. A.; Dembski, M. S.; Gao, Y.; Georges, A. T.;
Grenier, L.; Pak, R. H.; Patterson, J.; Sydor, J. R.; Tibbitts, T. T.; Tong,
J. T.; Adams, J.; Palombella, V. J. J. Med. Chem. 2006, 49, 4606–4615.
(c) Gopalsamy, A.; Shi, M.; Golas, J.; Vogan, E.; Jacob, J.; Johnson, M.;
Lee, F.; Nilakantan, R.; Petersen, R.; Svenson, K.; Chopra, R.; Tam,
M. S.; Wen, Y.; Ellingboe, J.; Arndt, K.; Boschelli, F. J. Med. Chem.
2008, 51, 373–375.
Acknowledgment. We gratefully acknowledge support
of this program by the James & Esther King Biomedical
Research Program (10KG-09).
Supporting Information Available. Experimental pro-
cedures and tabulated spectroscopic data for new com-
pounds. This material is available free of charge via the
Internet at http://pubs.acs.org.
Org. Lett., Vol. 13, No. 19, 2011
5111