A synthesis-driven structure revision of berkelic acid methyl ester

Article abstract of DOI:10.1002/anie.200803339

A subtle difference: A single step suffices to transform a linear precursor into the chromane spiroketal core of the metalloproteinase-3 inhibitor berkelic acid by an acid-catalyzed deprotection/Michael addition/acetalization cascade. This efficient route

Full text of DOI:10.1002/anie.200803339

Communications
DOI: 10.1002/anie.200803339
Natural Products
A Synthesis-Driven Structure Revision of Berkelic Acid Methyl
Ester**
Philipp Buchgraber, Thomas N. Snaddon, Conny Wirtz, Richard Mynott, Richard Goddard, and
Alois Fꢀrstner*
Matrix metalloproteinases (MMPs) are a family of zinc-
containing endopeptidases involved in homeostasis of the
extracellular matrix.[1] Abnormal activity of such enzymes is
implicated in pathological processes that result in osteo-
arthritis, rheumathoid arthritis, and multiple sclerosis, and
also plays a decisive role in tumor metastasis. Small-molecule
inhibitors of the individual MMPs are therefore highly
interesting as prospective complements to the current chemo-
therapeutic regimens used in clinical settings.[2]
A promising lead in this context is berkelic acid, which
was isolated from a Penicillium species collected in the very
hostile environment of Berkeley Pit Lake, a flooded former
copper mine in Butte, Montana.[3] This particular extremo-
phile has adapted to the waters of this lake, which are highly
acidic (pH ꢀ 2.5) and contain a cocktail of heavy-metal salts in
remarkably high concentrations. Bioassay-guided fractiona-
tion of the CHCl3 extracts of the fungus showed berkelic acid
to be the major metabolite responsible for the pronounced
inhibition of MMP-3 (GI50 = 1.87 mm).[3] Moreover, the com-
pound exhibits selective and potent activity against the
ovarian cancer cell line OVCAR-3 (GI50 = 91 nm).[3] As
MMP-3 is upregulated in OVCAR-3 but not in other ovarian
cancer cell lines, these preliminary activity and selectivity data
are highly encouraging and suggest that berkelic acid and
derivatives thereof deserve more intense scrutiny.[3,4]
Scheme 1. Retrosynthetic analysis of the structure 1 attributed to
berkelic acid.
The remarkable structural attributes of berkelic acid add
further to the appeal of this compound. It was assigned the
constitution and relative configuration 1 (Scheme 1) mainly
on the basis of NMR experiments.[3] Recent model studies
directed towards synthesizing berkelic acid appear to corrob-
orate this proposed structure, even though they reached
contradictory conclusions as to whether the acetalization that
produces the conspicuous chromane spiroketal core is ther-
modynamically or kinetically controlled.[5–7] We now report
our own investigations on berkelic acid methyl ester (2) which
not only resolve this open question but also suggest that the
original structure assignment needs to be revised.
Since the configuration at the lateral quaternary stereo-
center C22 of berkelic acid is unknown, a convergent
approach was adopted that should allow both possible
isomers to be prepared by incorporating either enantiomer
of synthon A at a late stage (Scheme 1). This route utilizes a
Michael addition/spiroacetalization cascade intended to con-
vert a linear precursor of type C into the tetracyclic core B of
the target in one step. Compound C, in turn, should arise from
an aldol condensation between the aromatic nucleus D and
the polyketide segment E.
The preparation of the required building block D
(Scheme 2) commenced with the copper-catalyzed opening
of (R)-(+)-2-pentyloxirane (> 99% ee)[8] by the Grignard
reagent derived from 3,5-bis(benzyloxy)-1-bromobenzene[9]
to give 3. Hydrogenolysis of the benzyl ethers followed by a
regioselective Kolbe–Schmitt carboxylation of the resulting
phenol 4 furnished acid 5,[9] which was esterified by treatment
with diazomethane prior to conversion into bis-TBS ether 8
by exhaustive silylation and selective mono-desilylation.
Since the attempted direct formylation of this product was
unrewarding,[10] we chose to introduce the required formyl
[*] Dr. P. Buchgraber, Dr. T. N. Snaddon, C. Wirtz, Dr. R. Mynott,
Dr. R. Goddard, Prof. A. Fꢀrstner
Max-Planck-Institut fꢀr Kohlenforschung
45470 Mꢀlheim an der Ruhr (Germany)
Fax: (+49)208-306-2994
E-mail: fuerstner@mpi-muelheim.mpg.de
[**] Generous financial support by the MPG and the Fonds der
Chemischen Industrie is gratefully acknowledged. We thank Prof.
A. A. Stierle and Prof. D. B. Stierle, Montana Tech of the University
of Montana, for providing copies of the original spectra, Dr. C. W.
Lehmann for solving the X-ray structure of 27, the ANKA
Angstroemquelle Karlsruhe for the provision of beamtime, and A.
Deege and his team for expert HPLC support.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200803339.
8450
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 8450 –8454

Angewandte
Chemie
minor syn isomer was removed by flash chromatography
before the pure anti-configured compound was converted into
methyl ketone 16 by treatment with MeMgBr at low temper-
ature.
The kinetic enolate of 16 was treated with aldehyde 10 to
give enone 18 (Scheme 4). This transformation exploits the
propensity of the acetyl group in intermediate 17 to migrate
from the phenolic site to the more basic alkoxide generated in
the aldol step. This transfer facilitates the subsequent
elimination by turning this site into a good leaving group
and deprotects the phenolic hydroxy group that is needed to
participate in the subsequent Michael addition/spirocycliza-
tion cascade. Much to our surprise, however, treatment of 18
with HCl in MeOH did not deliver a single (or at least a
major) spiroketal product; rather, an almost statistical
mixture of four isomeric compounds (19–22, d.r. =
0.9:1:0.8:1.3) was formed, which could be separated by
preparative HPLC. In contrast to the results of a previous
yet simpler model study,[5] attempts to equilibrate the crude
mixture by exposure to different acids were in vain. Likewise,
treatment of any of the individual isomers with pyridinium
p-toluenesulfonate (PPTS) rapidly regenerated the original
product distribution.
Extensive NMR investigations allowed the stereostruc-
ture of each isomer to be established beyond doubt (for
details see the Supporting Information). The assignment was
independently confirmed when crystals of 19 were grown that
were suitable for X-ray structure analysis (Figure 1).[16] Our
fully consistent NMR data set also allowed for a highly
informative comparison with the reported spectroscopic
properties of berkelic acid,[3] even though 19–22 differ from
the natural product in the lateral chain and by the presence of
a methyl ester moiety. While an in-depth discussion must
await a future full paper, a few comments are necessary. First,
compound 19 — which features the structure attributed to
berkelic acid — is the only isomer in which the 13C NMR
signal of the methyl branch C25 (d = 14.38 ppm) deviates
considerably from the chemical shift of this group in the
natural product (d = 11.9 ppm; Scheme 4). NOESY data as
well as the solid-state structure of this compound (Figure 1)
show an unfavorable syn-periplanar arrangement between
the methyl substituent and the C16 methylene group of the
adjacent tetrahydropyran ring (torsion angle C25-C18-C17-
C16: 138). All other isomers avoid such an eclipsed situation
by orienting the methyl group away from the benzopyran
unit. This is possible by either adopting the opposite config-
uration at the spiroacetal C17 (20 and 22) or, alternatively, by
an isomerization of the stereocenter C18 which carries the
methyl branch during the acetalization process (21). Collec-
tively, these structural data suggest that the positioning of the
methyl group has a strong impact on the stability of the
core.[17] Moreover, the recorded 13C NMR spectra show that
the signal for at least one C atom in each of the isomers
deviates significantly from the corresponding signal of
berkelic acid (see Scheme 4 and the Supporting Information).
Further important information can be deduced from NOE
data. Compound 22 is the only isomer in which H9 and H15
are trans-configured, and therefore cannot correspond to the
natural product. Moreover, Stierle et al. reported that “irra-
Scheme 2. Reagents and conditions: a) 1. Mg, THF, reflux; 2. [CuCl-
(cod)] (10 mol%), (R)-2-pentyloxirane, ꢁ508C!RT, 74%; b) H2, Pd/C
(10% w/w), MeOH, quant.; c) CO2 (1 atm), KHCO3, glycerine, 1508C;
d) TMSCHN2, MeOH, 66–77% (over both steps); e) TBSCl, imidazole,
CH2Cl2, 96%; f) K2CO3, MeOH, 458C, 72%; g) N-iodosuccinimide,
CH2Cl2, 98%; h) 1. MeLi, Et2O, ꢁ788C; 2. tBuLi, ꢁ1058C; 3. DMF,
ꢁ105!ꢁ358C; 4. AcCl, ꢁ55!ꢁ258C, 71%. Bn=benzyl, cod=1,5-
cyclooctadiene, TMS=trimethylsilyl, TBS=tert-butyldimethylsilyl.
group by a high-yielding sequence of regioselective iodination
(8!9), metal–halogen exchange, and trapping of the resulting
organolithium reagent with DMF. The fact that the unpro-
tected phenolate site in the resulting primary product could
be acylated in situ to give acetate 10 turned out to be highly
advantageous for the ensuing fragment coupling (see below).
An Ireland–Claisen rearrangement[11] of ester 13, result-
ing from the lactate-derived alcohol 12[12] and propionyl
chloride, furnished the second building block (Scheme 3). The
reaction was best performed with KHMDS in toluene,[13] and
delivered the required anti-configured ester 15 in good yield
and high diastereoselectivity (d.r. = 10.2:1, 91% ee).[14] After
transformation into the corresponding Weinreb amide,[15] the
Scheme 3. Reagents and conditions: a) Ref. [12]; b) propionyl chloride,
pyridine, CH2Cl2, ꢁ208C, 98% (98% ee); c) KHMDS (1.5 equiv),
TMSCl (2 equiv), toluene, ꢁ788C!RT; d) TMSCHN2, MeOH, 77%
(over both steps, anti/syn=10.2:1, 91% ee); e) 1. Me(MeO)NH·HCl,
iPrMgCl, THF, ꢁ188C, 78%; 2. MeMgBr, ꢁ18!08C, 93%.
KHMDS=potassium hexamethyldisilazide.
Angew. Chem. Int. Ed. 2008, 47, 8450 –8454
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8451

Communications
Scheme 4. Reagents and conditions: a) LDA, THF, ꢁ788C, then 10, ꢁ788C!RT, 85%; b) acetyl chloride, MeOH/CH2Cl2, 08C!RT, 91% (19/20/
21/22=0.9:1:0.8:1.3). LDA=lithium diisopropylamide. The DdC values refer to the differences in the chemical shifts between the indicated
C atoms in the individual isomers and the corresponding resonances in berkelic acid (for the full data set, see the Supporting Information). The
blue arrows indicate strong and characteristic NOE interactions.
clear that not only is the originally proposed stereostructure
incorrect, but that none of the isomers produced in the acid-
catalyzed cyclization of precursor 18 matches the core of the
natural product (Scheme 4).
The conclusions drawn from this analysis let us envisage
that structure 23 or its mirror image ent-23 represents berkelic
acid (Scheme 5). Inspection of a molecular model shows that
the C25 methyl branch in these compounds is oriented away
from the crowded core, and the NOE interactions are
expected to be consistent with the reported ones. Spurred
on by the prospect of finally solving this puzzle, we opted for
the preparation of enantiomer 23 because its synthesis only
requires readily accessible ent-3, which can then be elabo-
rated into the cyclization precursor 24 by following the
established route (Scheme 6). It was gratifying to see that 24
indeed cleanly converted into one major product on exposure
Figure 1. Structure of 19 in the solid state (for additional information,
to HCl/MeOH, whereas its 9-epi-analogue 18 had given an
see the Supporting Information).[16]
almost statistical mixture of four isomers under the same
conditions. As expected, the NMR spectra of spiroketal 25
showed that this compound features the acetal configuration
diation of methyl H-25 resulted in enhancement of H-16a and
H-20”.[3] However, when this experiment was repeated with
19, the isomer corresponding to the supposed structure,
unmistakable enhancements of not only H16a but also H16b
signals were observed. Together with the analysis of the 13C
NMR data outlined above, this inconsistent pattern makes it
which allows the C25 methyl branch to reside in the
unencumbered periphery; importantly, all NOE interactions
are in accord with the characteristic pattern of berkelic acid.
The structure assignment was corroborated by the X-ray
analysis of iodide 27 (Figure 2) derived from 25, which was
prepared for the final fragment coupling.
8452
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 8450 –8454

Angewandte
Chemie
Scheme 5. Proposed structure revision for berkelic acid: depending on
whether 23 or its enantiomer ent-23 turns out to be the natural
product, the differences relative to the original structure either
originate from a misassignment of the configuration at C9 (which then
formally propagates through the structure by the observed NOE
interactions), or are caused by a misassignment of C18 and C19. The
resulting deviations from the previously proposed structure 1 are
highlighted in red.
To this end, iodide 27 was subjected to a metal–halogen
exchange at low temperature, followed by addition of the
resulting polyfunctional lithium species to aldehyde (S)-28;
both enantiomers of this coupling partner can be obtained in
high purity from malic acid by following a literature
procedure (Scheme 6).[18] Oxidation of the resulting diaste-
reomeric alcohols 29 furnished (22S)-23b. Even though this
final oxidation turned out to be difficult and was plagued by
formation of the corresponding methylthiomethyl adduct 30
(Scheme 7),[19] the final product could be obtained in suffi-
cient quantity to allow for an unambiguous analysis. The
match between the recorded data and the reported spectrum
of berkelic acid methyl ester in CDCl3 is excellent, and leaves
no room for interpretation (see the Supporting Information).
Therefore, we confidently reassign the relative configuration
of this promising bioactive metabolite to that of 23 (or
ent-23).[20,21]
Scheme 6. Reagents and conditions: a) 1. Mg, THF, reflux; 2. [CuCl-
(cod)] (10 mol%), (S)-(ꢁ)-2-pentyloxirane, ꢁ508C, 74%; b) see
Scheme 2 and step (a) in Scheme 4; c) acetyl chloride, MeOH/CH2Cl2,
08C!RT, 94% (d.r.ꢂ12.5:1); d) OsO4 (2 mol%), N-methylmorpho-
line-N-oxide, acetone; e) Pb(OAc)4, CH2Cl2; f) NaBH4, MeOH, 08C,
59% (over three steps); g) I2, PPh3, imidazole, Et2O/MeCN, 85%;
h) 1. MeLi, Et2O, ꢁ1058C; 2. tBuLi, then 28; i) (COCl)2, DMSO,
CH2Cl2, then Et3N, ꢁ78!08C, 69% (23b/30=1:1).
Scheme 7. Structure of the by-product formed in the final Swern
oxidation reaction.
The 22R-configured product was prepared analogously by
1
using (R)-28 as the reaction partner. The H NMR spectrum
of (22R)-23b in CD3OD recorded at 600 MHz is subtly
different from that of (22S)-23b, but a confident assignment
of the configuration of the lateral quaternary center at C22
mandates direct comparison with an authentic sample.[22] We
are now in the process of refining the first generation total
Figure 2. Structure of 27 in the solid state.[16]
Angew. Chem. Int. Ed. 2008, 47, 8450 –8454
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8453

Communications
C. McFarland, M. C. McIntosh, Tetrahedron 2002, 58, 2905 –
2928.
[12] Y. Ichikawa, K. Tsuboi, M. Isobe, J. Chem. Soc. Perkin Trans. 1
1994, 2791 – 2796.
[13] a) A. Franꢃais, O. Bedel, W. Picoul, A. Meddour, J. Courtieu, A.
Haudrechy, Tetrahedron: Asymmetry 2005, 16, 1141 – 1155;
b) W. Picoul, R. Urchegui, A. Haudrechy, Y. Langlois, Tetrahe-
dron Lett. 1999, 40, 4797 – 4800.
[14] The relative configuration was unambiguously assigned after
conversion of the esters into the corresponding lactones and was
later confirmed by the X-ray structure of compounds 19 and 27.
Details will be reported in a forthcoming full paper.
[15] S. Nahm, S. M. Weinreb, Tetrahedron Lett. 1981, 22, 3815 – 3818.
[16] Anisotropic displacement parameters are drawn at the 50%
probability level and hydrogen atoms are omitted for clarity.
CCDC 694490 (19) and 693829 (27) contain the supplementary
crystallographic data for this paper. These data can be obtained
free of charge from The Cambridge Crystallographic Data
Centre via www.ccdc.cam.ac.uk/data_request/cif. Short summa-
ries of the crystallographic data can also be found in the
Supporting Information.
synthesis outlined above and adjusting the protecting groups
used such that free berkelic acid itself can also be reached.[23]
Received: July 9, 2008
Published online: September 24, 2008
Keywords: anticancer agents · enzyme inhibitors ·
.
natural products · spiroacetals · structure elucidation
[1] C. E. Brinckerhoff, L. M. Matrisian, Nat. Rev. Mol. Cell Biol.
2002, 3, 207 – 214.
[2] a) “Inhibition of Matrix Metalloproteinases: Therapeutic Ap-
plications”: Annales of the New York Academy of Science,
Vol. 878 (Eds.: R. A. Greenwald, S. Zucker, L. M. Golub),
Blackwell, Oxford, 1999; b) H. Matter, M. Schudok, Curr. Opin.
Drug Discovery Dev. 2004, 7, 513 – 535; c) M. Whittaker, C. D.
Floyd, P. Brown, A. J. H. Gearing, Chem. Rev. 1999, 99, 2735 –
2776; d) C. M. Overall, O. Kleifeld, Nat. Rev. Cancer 2006, 6,
227 – 239.
[3] A. A. Stierle, D. B. Stierle, K. Kelly, J. Org. Chem. 2006, 71,
5357 – 5360.
[17] It is important to note that the model compounds reported by
Zhou and Snider lacked this decisive methyl branch, which may
explain the misleading outcome of the study; see Ref. [5].
[18] P. Renaud, M. Hꢀrzeler, D. Seebach, Helv. Chim. Acta 1987, 70,
292 – 298.
[19] Other oxidants were found to degrade the electron-rich arene
ring. For precedence for the ortho-methylthiomethylation of a
phenol with DMSO and acid chlorides, see: K. Sato, S. Inoue, K.
Ozawa, M. Tazaki, J. Chem. Soc. Perkin Trans. 1 1984, 2715 –
2719.
[20] For other total syntheses from our research group which resulted
in the revision/clarification of the originally proposed structures,
see: a) A. Fꢀrstner, K. Radkowski, C. Wirtz, R. Goddard, C. W.
Lehmann, R. Mynott, J. Am. Chem. Soc. 2002, 124, 7061 – 7069;
b) A. Fꢀrstner, M. Albert, J. Mlynarski, M. Matheu, E.
DeClercq, J. Am. Chem. Soc. 2003, 125, 13132 – 13142; c) A.
Fꢀrstner, J. Ruiz-Caro, H. Prinz, H. Waldmann, J. Org. Chem.
2004, 69, 459 – 467; d) A. Fꢀrstner, K. Radkowski, H. Peters,
Angew. Chem. 2005, 117, 2837 – 2841; Angew. Chem. Int. Ed.
2005, 44, 2777 – 2781; e) A. Fꢀrstner, K. Radkowski, H. Peters,
G. Seidel, C. Wirtz, R. Mynott, C. W. Lehmann, Chem. Eur. J.
2007, 13, 1929 – 1945; f) A. Fꢀrstner, E. K. Heilmann, P. W.
Davies, Angew. Chem. 2007, 119, 4844 – 4847; Angew. Chem. Int.
Ed. 2007, 46, 4760 – 4763; g) A. Fꢀrstner, O. Larionov, S. Flꢀgge,
Angew. Chem. 2007, 119, 5641 – 5644; Angew. Chem. Int. Ed.
2007, 46, 5545 – 5548.
[4] Moreover, berkelic acid was reported to show an appreciable
inhibitory effect against caspase-1 (GI50 = 0.098 mm), a cysteine
protease playing a key mediator role in inflammation; for a
review see: T. Ghayur, S. Banerjee, M. Hugunin, D. Butler, L.
Herzog, A. Carter, L. Quintal, L. Sekut, R. Talanian, M. Paskind,
W. Wong, R. Kamen, D. Tracey, H. Allen, Nature 1997, 386, 619 –
623.
[5] J. Zhou, B. B. Snider, Org. Lett. 2007, 9, 2071 – 2074.
[6] a) M. A. Marsini, Y. Huang, C. C. Lindsey, K.-L. Wu, T. R. R.
Pettus, Org. Lett. 2008, 10, 1477 – 1480; b) Y. Huang, T. R. R.
Pettus, Synlett 2008, 1353 – 1356.
[7] For general reviews on spiroacetals, see: a) F. Perron, K. F.
Albizati, Chem. Rev. 1989, 89, 1617 – 1661; b) J. E. Aho, P. M.
Pihko, T. K. Rissa, Chem. Rev. 2005, 105, 4406 – 4440; c) for
recent studies on spiroketals from our research group, see: A.
Fꢀrstner, M. D. B. Fenster, B. Fasching, C. Godbout, K. Rad-
kowski, Angew. Chem. 2006, 118, 5632 – 5636; Angew. Chem. Int.
Ed. 2006, 45, 5506 – 5510; d) A. Fꢀrstner, M. D. B. Fenster, B.
Fasching, C. Godbout, K. Radkowski, Angew. Chem. 2006, 118,
5636 – 5641; Angew. Chem. Int. Ed. 2006, 45, 5510 – 5515.
[8] Conveniently prepared by hydrolytic kinetic resolution (HKR);
see: S. E. Schaus, B. D. Brandes, J. F. Larrow, M. Tokunaga, K. B.
Hansen, A. E. Gould, M. E. Furrow, E. N. Jacobsen, J. Am.
Chem. Soc. 2002, 124, 1307 – 1315.
[9] T. Rꢁdel, H. Gerlach, Liebigs Ann. Chem. 1997, 213 – 216. The
large-scale preparation of 3,5-bis(benzyloxy)-1-bromobenzene is
described in the Supporting Information.
[10] T. V. Hansen, L. Skattebøl, Org. Synth. 2005, 82, 64 – 67.
[11] a) R. E. Ireland, R. H. Mueller, J. Am. Chem. Soc. 1972, 94,
5897 – 5898; b) R. E. Ireland, P. Wipf, J.-N. Xiang, J. Org. Chem.
1991, 56, 3572 – 3582; c) reviews: A. M. Martꢂn Castro, Chem.
Rev. 2004, 104, 2939 – 3002; d) Y. Chai, S. Hong, H. A. Lindsay,
[21] For a review on natural products of mistaken identity, see: K. C.
Nicolaou, S. A. Snyder, Angew. Chem. 2005, 117, 1036 – 1069;
Angew. Chem. Int. Ed. 2005, 44, 1012 – 1044.
1
[22] Only the H NMR spectrum in CDCl3 has been recorded; we
were informed that there is very little of the authentic sample
left; supplying material is therefore currently not possible.
[23] Saponification of 23b is unselective, with both methyl esters
cleaved at similar rates.
8454
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 8450 –8454