Total syntheses of HMP-Y1, hibarimicinone, and HMP-P1

Article abstract of DOI:10.1021/ja307207q

Total syntheses of HMP-Y1, atrop-HMP-Y1, hibarimicinone, atrop-hibarimicinone, and HMP-P1 are described using a two-directional synthesis strategy. A novel benzyl fluoride Michael-Claisen reaction sequence was developed to construct the complete carbon sk

Full text of DOI:10.1021/ja307207q

Article
pubs.acs.org/JACS
Total Syntheses of HMP-Y1, Hibarimicinone, and HMP-P1
Brian B. Liau, Benjamin C. Milgram, and Matthew D. Shair*
Department of Chemistry and Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, United States
S
* Supporting Information
ABSTRACT: Total syntheses of HMP-Y1, atrop-HMP-Y1, hibarimici-
none, atrop-hibarimicinone, and HMP-P1 are described using a two-
directional synthesis strategy. A novel benzyl fluoride Michael−Claisen
reaction sequence was developed to construct the complete carbon
skeleton of HMP-Y1 and atrop-HMP-Y1 via a symmetrical, two-
directional, double annulation. Through efforts to convert HMP-Y1
derivatives to hibarimicinone and HMP-P1, a biomimetic mono-oxidation
to desymmetrize protected HMP-Y1 was realized. A two-directional
unsymmetrical double annulation and biomimetic etherification was
developed to construct the polycyclic and highly oxidized skeleton of hibarimicinone, atrop-hibarimicinone, and HMP-P1. The
use of a racemic biaryl precursor allowed for the synthesis of both hibarimicinone atropisomers and provides the first
confirmation of the structure of atrop-hibarimicinone. Additionally, this work documents the first reported full characterization of
atrop-hibarimicinone, HMP-Y1, atrop-HMP-Y1, and HMP-P1. Last, a pH-dependent rotational barrier about the C2−C2′ bond
of hibarimicinone was discovered, which provides valuable information necessary to achieve syntheses of the glycosylated
congeners of hibarimicinone.
atropisomer.1i,2 Altogether, the hibarimicins and hibarimicinone
INTRODUCTION
Background. Hibarimicins A−G are complex pseudodi-
meric type-II polyketides isolated from the culture broth of the
rare actinomycete Microbispora rosea subsp. hibaria TP-A0121.1
These metabolites inhibit proliferation and induce differ-
entiation of numerous human cancer cell lines. In particular,
hibarimicin B (1, Figure 1), which is identical to angelmicin
■
(2a) are challenging targets that have resisted total synthesis3
until earlier this year when Tatsuta et al. reported the first total
synthesis of 2a.4 Herein, we report enantioselective total
syntheses of hibarimicinone (2a) and atrop-hibarimicinone
(2b), and the first total syntheses of the biosynthetically related
natural product aglycons HMP-Y1 (3a), atrop-HMP-Y1 (3b),
and HMP-P1 (6) (Scheme 1).
Biosynthesis Hypothesis. Mutagenesis of Microbispora
rosea subsp. hibaria TP-A0121 led to the identification of novel
metabolites, including HMP-Y1 (3a), HMP-P1 (6), and their
glycosylated derivatives (Scheme 1).1f Through 13C-acetate
labeling studies, it was discovered that C2-symmetric 3a is a
precursor to 2a, which is subsequently glycosylated to yield
hibarimicins A−G. Ostensibly, this conversion (3a → 2a)
proceeds by breaking the C2-symmetry of 3a via oxidation of
the B-, C-, and D-rings and demethylation of the C4′−OMe
methyl group. We postulated that a single desymmetrizing
oxidation of the C-ring of 3a to hypothetical quinone 4 would
be sufficient to relay oxidation to the B- and D-rings. This could
be achieved via (1) tautomerization of quinone 4 to C8′-ortho-
quinone methide 5 with subsequent oxy-Michael addition of
the C13′−OH to install the B-ring cyclic ether, (2) reoxidation
of the C-ring, and (3) transposition of the C-ring quinone to
the D-ring with concomitant demethylation to give 2a. HMP-
P1 (6) arises from 2a via cyclization of C1−OH onto C3′ of
the D-ring quinone and subsequent expulsion of methanol.1f
Figure 1. Structure of hibarimicin B (1).
B,1b,c has the most potent antiproliferative activity in HL-60
cells (IC50 = 58 nM).1h The cellular target and biological
mechanism of action of 1 remain undetermined. The
hibarimicins are among the most complex and largest type-II
polyketides known. Hibarimicins A−G share an unprecedented
highly oxidized aglycon, hibarimicinone (2a, Scheme 1). The
C2-symmetry of 2a is broken by oxidation of the B-, C-, and D-
rings relative to the G-, F-, and E-rings, respectively. More
specifically, the B-ring contains a cyclic ether bridging C8′ and
C13′, the C-ring contains a hydroxyl group at C6′, and the D-
ring is a quinone. Furthermore, 2a exhibits axial chirality about
its highly congested C2−C2′ bond and is isolated as a single
Received: July 23, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the feasibility of a biomimetic mono-oxidation to access a
quinone analogous to 4. In contrast to 7, the C2-symmetry of 8
is perturbed by the presence of a benzyl group5 and the C6′−
OH (both highlighted in red), the latter of which would
facilitate chemoselective C-ring oxidation to a quinone. The
most noteworthy feature shared by both 7 and 8 is the
degeneracy of the AB- and HG-ring systems that results from
the retrosynthetic excision of the B-ring cyclic ether bond.
Next, it was envisioned that both octacyclic systems could be
constructed in a single operation via a two-directional double
annulation,6 where the dianion of biaryl 10 would react with
two equivalents of the AB-/HG-enone (+)-9. The use of a
symmetric biaryl annulation donor would lead to 7, whereas the
employment of an unsymmetrical variant, with additional
oxidation at C6′, would result in 8. Both of these strategies are
convergent and circumvent the need to construct the hindered
C2−C2′ bond at a late stage. Efforts by the Roush group to
form the C2−C2′ bond in simple model systems via cross-
coupling were met with difficulty.3d At the outset of our study,
the configuration of the stereochemical axis about the C2−C2′
bond of hibarimicinone (2a) was ambiguous.1i Consequently,
we elected to proceed with racemic biaryl annulation donor 10
to prepare and characterize both atropisomers of HMP-Y1 (3a)
and hibarimicinone (2a).
Scheme 1. Proposed Biosynthetic Conversion of HMP-Y1
(3a) to Hibarimicinone (2a) and HMP-P1 (6)
Synthesis of the AB-/HG-Enone (+)-9. We previously
reported a gram-scale enantiospecific synthesis of the AB-/HG-
ring system corresponding to the unnatural enantiomer of 2a.3i
The AB-/HG-enone (+)-16 (Scheme 3), with stereochemistry
a
Scheme 3. Synthesis of the AB-/HG-enone (+)-9
RESULTS AND DISCUSSION
■
Synthesis Plan. Inspired by our proposed biosynthetic
relay oxidation scheme, we envisioned that a similar set of
biomimetic retrosynthetic disconnections could simplify 2a to
two plausible precursors, C2-symmetric octacycle 7 and pseudo-
C2-symmetric octacycle 8 (Scheme 2). Targeting 7 was
attractive for two reasons: (1) global deprotection would
yield HMP-Y1 (3a), and (2) it would allow direct assessment of
Scheme 2. Biosynthesis-Inspired Retrosynthesis of HMP-Y1
(3a), Hibarimicinone (2a), and HMP-P1 (6)
a
Conditions: (a) AcOH/H2O, 80 °C; (b) PPh3, I2, imidazole, PhMe/
CH2Cl2, room temperature → 45 °C, 76% (two steps); (c) TBSCl,
imidazole, CH2Cl2, 0 °C → room temperature, 99%; (d) PivCl, 4-
DMAP, ClCH2CH2Cl, 50 °C, 94%; (e) DBU, MeCN, 80 °C, 75%; (f)
30 mol % Hg(OCOCF3)2, Me2CO/H2O; (g) MsCl, Et3N, CH2Cl2, 0
°C → room temperature, 74% (two steps); (d) LiHMDS, THF, 0 °C;
then TMSOTf, 0 °C → room temperature, 99%.
corresponding to the natural enantiomer of 2a, was prepared
from key intermediate cyclohexenone (+)-15 following an
analogous series of diastereoselective transformations. Notably,
both enantiomers of (+)-15 were accessed from α-D-methyl-
glucopyranoside by taking advantage of its latent C2-symmetry.
The AB-/HG-enone synthesis commenced with AcOH-
mediated deprotection of benzylidene acetal 11 followed by
selective iodination of the resultant primary hydroxyl group to
provide diol (+)-12. Next, chemoselective monosilylation of
(+)-12 with TBSCl was accomplished by exploiting a subtle
steric difference between its two secondary hydroxyl groups.
The remaining secondary hydroxyl group was then pivaloylated
under forcing conditions to furnish differentially protected
B
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
pyranose (+)-13. Exposure of (+)-13 to DBU promoted
elimination of the primary iodide to generate exocyclic enol
ether (+)-14, which underwent type-II Ferrier rearrangement
upon treatment with catalytic Hg(OCOCF3)2.7 The resultant
β-hydroxy-cyclohexanone was dehydrated to provide (+)-15.
Following our previous procedures,3i (+)-15 was converted to
(+)-16,8 and the tertiary C14−OH was ultimately TMS-
protected to give annulation acceptor (+)-9.
Demonstration of a Biomimetic Etherification on a
Model ABCD-Ring System. A key transformation in our
synthesis plan is a late-stage biomimetic etherification that
retrosynthetically symmetrizes hibarimicinone (2a) and HMP-
P1 (6). Consequently, a model ABCD-ring system was first
investigated to test the viability of this proposed reaction. Kraus
annulation9 of cyanophthalide 1710 with ent-AB-/HG-enone
ent-98 under rigorously oxygen-free conditions gave ABCD-
tetracycle (−)-18 (Scheme 4). Deoxygenation was critical for
lack of robust annulation sequences to generate naphthols (1-
hydroxynaphthalenes rather than 1,4-dihydroxynaphthalenes,
i.e., hydroquinones),12 and thus necessitated a flexible synthesis
of symmetric biaryl annulation donors in which different
substituents at C6/C6' could be introduced. Our biaryl
synthesis began with 5-methylvanillin 2113 (Scheme 5),
which was converted to trialkoxytoluene 22 through a three-
step process involving: (1) O-methylation, (2) Dakin oxidation
followed by in situ formate methanolysis, and (3) MOM
protection of the resultant phenol. Next, regioselective ortho-
lithiation of 22 at C2′ and FeCl3-mediated oxidative
dimerization of the resultant aryllithium species delivered
biaryl ( )-23. Carbomethoxy groups were then installed in a
two-step sequence involving bromination and lithium−halogen
exchange followed by acylation to afford bis-ortho-toluate
( )-24.
The reaction kinetics and ultimate success of Michael−
Claisen reaction sequences14 hinge on numerous factors. These
include, but are not limited to: (1) the stability and
nucleophilicity of the reacting carbanion,15 (2) the stability of
the electrophilic acceptor to the base required to deprotonote
the donor,16 and (3) the steric bulk of the donor (substituents
at C6/C6′ and the ester side chain) and of the acceptor.
Furthermore, the slow step of the tandem reaction sequence
will change based on the particular donor and acceptor used.
We found that ( )-24 could be deprotonated twice by LiTMP
and that the corresponding dianion underwent two-directional
bis-Michael−Claisen reaction sequence with various 2-cyclo-
hexenones, including the AB-/HG-enone ent-9. However, the
slow rate of both the Michael and the Claisen reactions17 of the
sequence with sterically encumbered ent-9 versus simpler 2-
cyclohexenones, coupled with the finite lifetime of the dianion
of ( )-24, led to very low yields (<10%) of the desired
octacyclic dihydronaphthalene products. Attempts to facilitate
the Claisen step of the reaction sequence by utilizing activated
ester analogues (i.e., phenyl and 2,2,2-trifluoroethyl) were
particularly problematic since the bulkier activated esters
slowed the initial Michael reaction and the respective dianions
were more prone to decomposition. Most importantly,
aromatization of the C-/F-rings of the octacyclic products
was met with difficulty and led us to consider alternative
approaches.18 Benzyl sulfoxide19- and benzyl sulfone20-
substituted ortho-toluates were also evaluated to achieve the
desired naphthol annulation with (+)-9 or ent-9, but ultimately
proved unsuccessful in the context of a two-directional double
annulation (vide infra).
We next envisaged a benzyl fluoride Michael−Claisen
reaction sequence to generate naphthalenes after subsequent
dehydrohalogenation.21 Although there was no precedence for
such a strategy, a benzyl fluoride annulation donor was
attractive for several reasons: (1) the electronegative fluorine
atom should stabilize the biaryl dianion, (2) the small atomic
radius of fluorine should provide minimal steric hindrance to
the initial Michael reaction, (3) the strength of C−F bonds
would disfavor α-elimination and SN2 displacement of fluoride,
and (4) despite the strength of C−F bonds, elimination of the
benzyl fluoride under appropriate conditions could lead to C-
and F-ring aromatization.22 The dianion of ( )-24 was
brominated with (BrCF2)2 to yield the bis-benzyl bromide,
which upon heating with TBAT23 afforded bis-benzyl fluoride
( )-25. After significant experimentation, the desired protected
HMP-Y1 derivatives (−)-27a and (+)-27b24 were accessed
from (+)-9 and ( )-25 in a two-step procedure involving: (1) a
Scheme 4. Synthesis of an ABCD-Pentacyclic Model
System
a
a
Conditions: (a) LiHMDS, THF, −78 → 0 °C, 94%; (b) DDQ,
CH2Cl2, −20 °C; (c) HCl, CH2Cl2, 0 °C, 69% (two steps).
the success of the annulation, as adventitious oxygen caused
decomposition. (−)-18 was then oxidized with DDQ to the
corresponding C-ring quinone, which upon exposure to
anhydrous HCl led to clean formation of pentacyclic ether
(+)-20. This transformation presumably occurs through the
intermediacy of ortho-quinone methide 19,11 which is trapped
by the proximal acetonide oxygen with concomitant acetonide
cleavage. Notably, this acid-mediated etherification required
high dilution to avoid formation of (−)-18 and decomposition
products, potentially via intermolecular processes.11c We
propose that acetonide cleavage is triggered by its proximity
to the reactive ortho-quinone methide intermediate due to the
failure to isolate any acetonide cleavage products that lack the
cyclic ether bond. This concept is further validated by later
observations on the dimeric system, where only the acetonide
of the A-ring is cleaved, whereas the G-ring acetonide remains
intact (vide infra). Overall, the model oxidation-etherification
sequence afforded excellent precedence for our subsequent
studies.
Development of a Benzyl Fluoride Michael−Claisen
Reaction Sequence to Achieve the First Total Syntheses
of HMP-Y1 (3a) and atrop-HMP-Y1 (3b). Our synthesis
plan for HMP-Y1 (3a) involved a symmetric two-directional
double annulation to generate C2-symmetric octacycle 7, which
could potentially be desymmetrized through a biomimetic
mono-oxidation to access hibarimicinone (2a) and HMP-P1
(6). Such a double annulation strategy is ambitious due to the
C
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 5. Synthesis of HMP-Y1 (3a) and atrop-HMP-Y1 (3b) via a Benzyl Fluoride Michael−Claisen Reaction Sequence
a
Conditions: (a) Me2SO4, K2CO3, Me2CO, 98%; (b) mCPBA, NaHCO3, CH2Cl2; then Na2CO3, MeOH; (c) NaH, MOMCl, DMF, 0 °C → room
n
temperature, 91% (two steps); (d) BuLi, TMEDA, THF, −78 → 0 °C; then FeCl3, 0 °C → room temperature, 76%; (e) Br2, Py, CH2Cl2, 0 °C,
91%; (f) nBuLi, THF, −78 °C; then ClC(O)OMe, −78 °C → room temperature, 89%; (g) LiTMP, THF, −78 °C; then (BrCF2)2, −78 °C, 62%; (h)
TBAT, MeCN, 82 °C, 75%; (i) (+)-9, LiTMP, THF, −78 °C; then HMDS, −78 → −35 °C; then MgBr2·OEt2, −35 → 0 °C; (j) CF3CH2OH/H2O,
NaHCO3, 80 °C, 55% (two steps); (k) aq HF, MeCN/THF, 50 °C; (l) H2, Pd(OH)2/C, THF; for 3a, 59% (two steps); for 3b, 62% (two steps).
bis-Michael−Claisen reaction sequence promoted by LiTMP
and MgBr2·OEt2 to afford octacycles 26a and 26b and (2) the
formal elimination of HF by heating the unpurified reaction
product in 2,2,2-trifluoroethanol (TFE) to achieve aromatiza-
tion of the C- and F-rings and provide atropisomers (−)-27a
and (+)-27b, which were readily separated and carried forward
independently. Several features of this sequence deserve
comment. As we had hoped, the use of a bis-benzyl fluoride
( )-25 allowed for the initial bis-Michael addition to occur at
−78 °C, thus minimizing decomposition of the dianion
intermediate and of (+)-9. Addition of MgBr2·OEt2 mid
annulation sequence was critical to promote the final
intramolecular Claisen reactions and obviated the need to use
an activated ester analogue. This discovery should help expand
the substrate scope of the Michael−Claisen reaction sequence.
Finally, the unique ability of TFE to promote the desired
elimination is presumably due to its ability to strongly hydrogen
bond with fluorine, and thus activate it for mild solvolysis.
Indeed, use of ethanol in place of TFE only led to trace
elimination. The employment of a benzyl fluoride annulation-
elimination sequence to generate naphthalene derivatives is
without precedence and may prove to be a general method for
the synthesis of naphthols.
Global deprotection of (−)-27a and (+)-27b with aqueous
HF followed by hydrogenolysis afforded HMP-Y1 (3a) and
atrop-HMP-Y1 (3b), respectively.25 Heating 3a or 3b to 90 °C
led to no detectable isomerization about the C2−C2′ bond.8
With no authentic CD spectra for natural 3a available, synthetic
3a and 3b were designated by comparing their CD spectra with
the CD spectrum of the glycosylated derivative of 3a, HMP-
Y6.8 The axial stereochemistry of HMP-Y1 (3a) has not been
rigorously determined, although model studies and the CD-
spectra of HMP-Y6 and hibarimicinone (2a) suggest 3a
possesses the aR configuration by the CD exciton chirality
method.2,26 Additionally, 3a, 2a, and hibarimicin A−G are all
isolated as single atropisomers.26 We show that the axial
stereochemistry of 3a and 2a is not the result of
thermodynamic equilibration (vide infra), and thus their
biosynthetic relationship also argues that they possess the
same relative configuration about the C2−C2′ bond. Therefore,
3a can be assigned the aR configuration since the axial chirality
of 2a was unambiguously determined.4
Biomimetic Mono-Oxidation of Protected HMP-Y1.
With a route to HMP-Y1 (3a) and atrop-HMP-Y1 (3b)
established, we next attempted the biomimetic mono-oxidation
of protected HMP-Y1 derivatives. We discovered that the
desired oxidation of binaphthol ent-27a to the C-ring quinone
(−)-29 could be achieved in low yield with CAN (Scheme 6),
demonstrating the plausibility of our proposed biomimetic
desymmetrizing oxidation. We speculate that the congested
biaryl may sterically occlude the approach of oxidants to the
otherwise easily oxidized D-/E-ring system,27 allowing
oxidation of the more electron-deficient C-/F-rings. Despite
this initial success, our attempts to optimize the CAN oxidation
were met with difficulty due to bis-oxidation and formation of
nitrated byproducts. A survey of other oxidants also proved
fruitless.
Nevertheless, with naphthazarin (−)-29 in hand, we next
investigated the desired biomimetic etherification reaction.
Unfortunately, exposure of (−)-29 to the optimized conditions
developed on our model system led to no observable
etherification but rather only rapid MOM group cleavage. A
screen of various acids and conditions also proved unsuccessful.
The resistance of (−)-29 to undergo the desired etherification
in contrast to the C-ring quinone derivative of (−)-18 was
surprising. Because the major difference between the two
systems is the lability of the MOM groups of (−)-29 relative to
the methyl group of (−)-18, we postulated that a free phenol at
C1′ might disfavor either acetonide decomposition or
formation of the necessary ortho-quinone methide intermediate.
D
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Synthesis of the Unsymmetrical Biaryl Annulation
Donor ( )-34. The unsymmetrical fully substituted biaryl
( )-34 presents unique synthesis challenges that are shared
with the hibarimicins; cross-coupling technology to form such
sterically hindered biaryls from electron-rich arenes is limited.3d
In contrast, dimerization reactions to form hindered biaryls are
robust and reliable (e.g., 22 → ( )-23). Thus, we imagined
that a practical synthesis approach to ( )-34 would necessitate
the desymmetrization of ( )-24. A strategy to monofunction-
alize ( )-24 involving radical bromination would inevitably
result in an inefficient statistical mixture of benzylic bromides.
However, we hypothesized that selective monodeprotonation
of ( )-24 would be feasible since the initial carbanion would
elevate the pKa of the remaining ortho-toluate due to a field
effect. Indeed, we found that selective monodeprotonation of
( )-24 at C6′ could be achieved with 1.25 equiv of LiTMP
(Scheme 7). The resultant anion ( )-30 was then brominated
with (BrCF2)2 to give benzyl bromide ( )-31 in 82% yield.
This single element of asymmetry was sufficient to introduce
the remaining differential functionality of ( )-34. ( )-31 was
oxidized to aldehyde ( )-32,28 which was then converted to
tribenzyl-protected biaryl ( )-33 by (1) acid-promoted
removal of the MOM groups, (2) chemoselective cleavage of
the C4′−OMe methyl group with BCl3, and (3) global
reprotection with BnBr. Treatment of ( )-33 with a controlled
source of hydrogen cyanide afforded a cyanophthalide
intermediate.29 Finally, double deprotonation of this inter-
mediate with LiTMP followed by a short exposure to S-phenyl
benzenethiosulfonate chemoselectively installed the phenyl
sulfide moiety21 at C6 to provide biaryl ( )-34. The observed
chemoselectivity in this reaction is a result of the much higher
reactivity of the ortho-toluate anion relative to the cyanoph-
thalide anion.
Completion of Hibarimicinone (2a) and atrop-Hibar-
imicinone (2b) via an Unsymmetrical Two-Directional
Double Annulation. We anticipated that reaction of the
lithiated cyanophthalide of ( )-34 with (+)-9 would directly
construct the C-ring hydroquinone via a Kraus annulation, and
reaction of the lithiated benzyl phenyl sulfide of ( )-34 with a
second equivalent of (+)-9 would lead to the F-ring via a
Michael−Claisen reaction sequence. We found that the desired
transformations could be achieved by treating a mixture of
( )-34 and (+)-9 with LiHMDS followed by subsequent
addition of KHMDS mid annulation sequence under rigorously
oxygen-free conditions to yield octacycle (−)-35a and (+)-35b
as a ∼1.3:1 mixture of atropisomers (Scheme 8). The addition
of KHMDS was crucial to facilitate the final intramolecular
Claisen reaction to construct the F-ring.30 At this stage,
atropisomers (−)-35a and (+)-35b were separated and carried
forward independently. Elimination of the C6-benzylic phenyl
sulfide was accomplished with dimethyl(methylthio)sulfonium
tetrafluoroborate (DMTSF) to yield binaphthalenes (−)-36a
and (+)-36b. It is worth reiterating at this point that the
corresponding C6-benzyl sulfoxide and sulfone derivatives of
( )-34 ultimately proved unsuccessful,31 highlighting the
difficulty to achieve naphthol annulations in the context of
complex molecule synthesis. To the best of our knowledge, this
is the first example of a benzyl sulfide Michael−Claisen reaction
sequence to generate naphthalenes, and together with the
benzyl fluoride Michael−Claisen reaction sequence reported
herein offers two new alternatives to approach challenging
naphthol annulations.
Scheme 6. Biomimetic Mono-Oxidation of Protected HMP-
Y1
This prompted us to replace the MOM group with a more acid-
stable protecting group.
Additionally, our current biomimetic strategy would
inevitably require a late-stage demethylation of the C4′−OMe
methyl group. Ideally, one would remove the C4′−OMe
methyl group as late in an eventual synthesis of hibarimicin B
(1) as possible to protect the sensitive and stereochemically
labile binaphthyl core (vide infra). However, the acidic
conditions necessary to effect demethylation would be
incompatible with the sensitive 2-deoxy- and 2,3-dideoxyglyco-
sides of 1. The aforementioned reasons prompted us to
investigate our alternative strategy for the synthesis of
hibarimicione (2a) and HMP-P1 (6) utilizing an unsym-
metrical two-directional annulation reaction with biaryl ( )-34
(Scheme 7).
Scheme 7. Synthesis of Unsymmetrical Biaryl ( )-34 via a
a
Selective Mono-Deprotonation of ( )-24
a
Conditions: (a) LiTMP, THF, −78 °C; then (BrCF2)2, 82%; (b)
iPr2NEt, DMSO, 70 °C, 87%; (c) TFA, CH2Cl2, 0 °C → room
temperature; (d) BCl3, CH2Cl2, −78 → 0 °C; (e) BnBr, K2CO3, DMF,
0 → 60 °C, 94% (three steps); (f) Me2C(OH)CN, Et3N, CHCl3, 97%;
(g) LiTMP, THF, −78 °C; then Ph(O)2SSPh, 71%.
E
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 8. Completion of Hibarimicinone (2a), atrop-Hibarimicinone (2b), and HMP-P1 (6)
a
Conditions: (a) LiHMDS, THF, −78 → 0 °C; then KHMDS, 0 °C → room temperature, 50−59%; (b) DMTSF, DTBMP, MeCN, 0 °C → room
temperature; for (−)-35a, 75%; for (+)-35b, 89%; (c) for (−)-36a, DDQ, PhMe, 0 °C; for (+)-36b, DDQ, PhMe, 0 °C → room temperature; (d)
HCl, ClCH2CH2Cl, 5 °C; for (−)-37a, 77% (two steps); for (+)-37b, 86% (two steps); (e) aq HF, MeCN/THF; (f) H2, Pd(OH)2/C, EtOAc; then
HCl, MeOH, air; for 2a, 81% (two steps); for 2b, 60% (two steps); (g) aq pH 7.5 NaH2PO4/NaOH buffer, MeOH, room temperature; from 2a,
84%; from 2b, 84%.
Figure 2. (A) Upon standing in acidic methanol (1 M HCl) at room temperature, hibarimicinone (2a) and atrop-hibarimicinone (2b) underwent
minor interconversion and minimal conversion to HMP-P1 (6) (orange HPLC traces). (B) Exposure of 2a to pH 7.5 aqueous phosphate buffer at
room temperature (blue HPLC traces) or (C) acidic methanol (1 M HCl) at 60 °C (red HPLC traces) resulted in isomerization to 2b and eventual
formation of 6. See the Supporting Information for HPLC time-courses starting with 2b.
Oxidation of (−)-36a and (+)-36b with DDQ produced the
corresponding C-ring quinones. Exposure of the respective
quinones to anhydrous HCl promoted the desired biomimetic
etherification to yield nonacycles (−)-37a and (+)-37b. This
successful etherification of the benzyl protected naphthazarins,
in contrast to MOM-protected (−)-29, confirmed our
suspicion that the C1′-phenol has far-reaching stereoelectronic
effects on this system. With the complete skeletons of 2a and
2b in hand, all that remained to complete the syntheses was
global deprotection and oxidation of the D-ring. Deprotection
of the acid-labile protecting groups was accomplished upon
exposure to aqueous HF. Finally, the benzyl groups were
removed via hydrogenolysis, and after addition of acidic
methanol, filtering, and exposure to air, hibarimicinone (2a)
and atrop-hibarimicinone (2b) were formed. All of the
spectroscopic data for 2a and 2b match those reported4,8,26
and thereby confirmed the structure of 2b.
Discovery of a pH-Dependent Rotational Barrier
about the C2−C2′ Bond in Hibarimicinone (2a) and
atrop-Hibarimicinone (2b). The addition of acidic methanol
prior to aerobic oxidation was crucial to prevent isomerization
between 2a and 2b and formation of HMP-P1 (6).25,32 Upon
prolonged handling or standing at ambient temperatures in
acidic methanol (1 M HCl), we observed only minor
interconversion between 2a and 2b, and minimal formation
of 6 (Figure 2A). However, simple exposure of 2a and 2b,
independently, to pH 7.5 aqueous phosphate buffer in
methanol led to the formation of 6 in 84% yield in both
cases (Scheme 8). Monitoring these transformations by HPLC
revealed that nearly complete isomerization about the C2−C2′
bond (2a ↔ 2b) had occurred within 4 h, while formation of 6
F
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
was almost complete after 24 h (Figure 2B).8 Together, these
observations suggest that the rotational barriers about C2−C2′
for 2a and 2b are pH-dependent.
numerous challenges when conducting naphthol annulation
reactions. Consequently, we developed two valuable Michael−
Claisen reaction sequences to construct complex naphthols that
might find use as general methods. The mild conditions needed
to dehydrohalogenate the benzyl fluoride intermediates are
particularly noteworthy given the strength of C−F bonds.
The plausibility of our proposed biosynthesis was also
validated by the demonstration that a desymmetrizing mono-
oxidation of the C-ring can be conducted on a protected HMP-
Y1 derivative. Overoxidation to the bis-C-/F-ring quinone was
also observed, but natural products corresponding to such a
double oxidation have not been isolated in nature or during
mutagenesis studies. This perhaps suggests that an enzyme
mediates this key biosynthetic transformation, but how 3a is
only mono-oxidized remains unclear.
These findings are particularly interesting due to prior
observations that heating 2a in neutral methanol at 60 °C leads
to nearly complete interconversion to 2b in 30 min and
ultimately complete cyclization to yield 6 after 90 min.1i,2,26
However, we found that heating either 2a or 2b to 60 °C in
acidic methanol (1 M HCl) led to only partial interconversion
between 2a and 2b and minor conversion to 6 after 90 min
(Figure 2C). This suggests that the observed rapid rotation at
60 °C in neutral methanol has less to do with providing the
necessary thermal energy to surpass the intrinsic activation
barrier about C2−C2′ in the uncharged forms of 2a/2b (38a/
38b in Figure 3), but rather enables access to the deprotonated
After the key two-directional annulations, only three and five
steps were needed to complete HMP-Y1 (3a) and
hibarimicinone (2a), respectively. In the case of 2a, these
steps include a biomimetic etherification to install the B-ring
cyclic ether via an ortho-quinone methide intermediate. The
success of this reaction required an acid-stable protecting group
to mask the C1′-phenol owing to subtle yet far-reaching
stereoelectronic effects imparted by the naphthazarin-naph-
thalene system. The peculiarities and sensitivity of this system
are also highlighted by our discovery of the pH-dependent
rotational barrier about the C2−C2′ bond. These particular
observations provide crucial information that will facilitate an
eventual synthesis of hibarimicin B (1).
Last, the intermediate (−)-37a will be highly useful in an
eventual total synthesis of 1; it is suitably protected with
orthogonal protecting groups to allow for the sequential
installation of the 2-deoxy- and 2,3-dideoxyglycosides prior to
deprotection of the sensitive binaphthyl core of the molecule.
Future studies toward the total synthesis of 1 will be reported
in due course.
Figure 3. A proposed model to explain the pH-dependent rotational
barrier about the C2−C2′ bond of 2a and 2b. Only the CDEF-ring
system is depicted for brevity.
form of 2a and 2b (39a/39b in Figure 3) via inter- or
intramolecular proton transfer. Rapid interconversion between
37a and 37b can then follow through a transition state that is
stabilized by π-electron overlap,3b,33 as depicted in cross-
conjugated resonance structures 40a and 40b. Consequently,
variables that affect the equilibrium between 38a and 39a, and
38b and 39b, will affect the rate of isomerization. The addition
of acid to the media inhibits access to species 39a and 39b by
driving the equilibrium toward 38a and 38b, and thus disfavors
isomerization. In contrast, heat (60 °C) should promote
equilibration between the protonation states and thus facilitate
isomerization. Our discovery of the pH-dependent barrier
demonstrates the delicate nature of the C2−C2′ axis, which
must be accommodated in an eventual synthesis of hibarimicin
B (1).32
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental procedures, spectroscopic data, and copies of CD,
1
UV−vis, H, and 13C NMR spectra. This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
shair@chemistry.harvard.edu
Notes
■
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We acknowledge financial support from NIH
(R01GM090068). B.B.L. acknowledges a NSF Predoctoral
Fellowship and Bristol-Myers Squibb. B.C.M. acknowledges Eli
Lilly and AstraZeneca. We thank Profs. Y. Igarashi, H. Hori,
and G. Sulikowski for communication regarding atropisomer-
ism and for providing authentic spectroscopic data. B.B.L.
acknowledges Amy S. Lee for helpful discussions.
CONCLUSION
■
Enantioselective total syntheses of hibarimicinone (2a) and
atrop-hibarimicionone (2b), and the first total syntheses of
HMP-Y1 (3a), atrop-HMP-Y1 (3b), and HMP-P1 (6), have
been accomplished. The complete carbon skeleton of each
natural product was assembled via a convergent two-directional
annulation strategy. The use of a racemic biaryl in conjunction
with the two-directional annulation strategy enabled both
atropisomers of the natural products to be separately
constructed and fully characterized, thus providing the first
reported full characterization of 2b, 3a, 3b, and 6. Additionally,
during the pursuit of this annulation strategy, we encountered
REFERENCES
■
(1) (a) Uehara, Y.; Li, P. M.; Fukazawa, H.; Mizuno, S.; Nihei, Y.;
Nishio, M.; Hanada, M.; Yamamoto, C.; Furumai, T.; Oki, T. J.
Antibiot. 1993, 46, 1306−1308. (b) Hori, H.; Higashi, K.; Ishiyama, T.;
Uramoto, M.; Uehara, Y.; Oki, T. Tetrahedron Lett. 1996, 37, 2785−
2788. (c) Kajiura, T.; Furumai, T.; Igarashi, Y.; Hori, H.; Higashi, K.;
G
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Ishiyama, T.; Uramoto, M.; Uehara, Y.; Oki, T. J. Antibiot. 1998, 51,
394−401. (d) Hori, H.; Igarashi, Y.; Kajiura, T.; Furumai, T.; Higashi,
K.; Ishiyama, T.; Uramoto, M.; Uehara, Y.; Oki, T. J. Antibiot. 1998,
51, 402−417. (e) Hori, H.; Kajiura, T.; Igarashi, Y.; Furumai, T.;
Higashi, K.; Ishiyama, T.; Uramoto, M.; Uehara, Y.; Oki, T. J. Antibiot.
2002, 55, 46−52. (f) Kajiura, T.; Furumai, T.; Igarashi, Y.; Hori, H.;
Higashi, K.; Ishiyama, T.; Uramoto, M.; Uehara, Y.; Oki, T. J. Antibiot.
2002, 55, 53−60. (g) Igarashi, Y.; Kajiura, T.; Furumai, T.; Hori, H.;
Higashi, K.; Ishiyama, T.; Uramoto, M.; Uehara, Y.; Oki, T. J. Antibiot.
2002, 55, 61−70. (h) Cho, S. I.; Fukazawa, H.; Honma, Y.; Kajiura, T.;
Hori, H.; Igarashi, Y.; Furumai, T.; Oki, T.; Uehara, Y. J. Antibiot.
2002, 55, 270−278. (i) Hori, H.; Igarashi, Y.; Kajiura, T.; Sato, S.;
Furumai, T.; Higashi, K.; Ishiyama, T.; Uehara, Y.; Oki, T. Tennen Yuki
Kagobutsu Toronkai Koen Yoshishu 2004, 46, 49−54.
(16) We observed that ent-9 is stable to LiTMP and LDA at −78 °C,
and LiHMDS at 0 °C in THF. At higher respective temperatures for
prolonged reaction times, significant decomposition occurred.
(17) Simple 2-cyclohexenones will undergo the Michael addition
within seconds at −78 °C and eventual Claisen reaction at −10 °C
with the ortho-toluate carbanion corresponding to the D-/E-ring. In
contrast, ent-9 underwent Michael addition after approximately 1 h,
and the Claisen reaction was never driven to completion when the
dianion of ( )-24 was employed.
(18) DDQ or PhSeCl with pyridine could successfully be employed
to aromatize dihydronaphthalenes of simple BCD-ring model systems
but proved unsuccessful on binaphthyl systems.
(19) (a) Hauser, F. M.; Rhee, R. P. J. Org. Chem. 1978, 43, 178−180.
(b) Lee, H. G.; Ahn, J. Y.; Lee, A. S.; Shair, M. D. Chem. Eur. J. 2010,
16, 13058−13062.
(20) (a) Wildeman, J.; Borgen, P. C.; Pluim, H. Tetrahedron Lett.
1978, 25, 2213−2216. (b) Huang, X.; Xue, J. J. Org. Chem. 2007, 72,
3965−3968.
(2) Romaine, I. M.; Hempel, J. E.; Shanmugam, G.; Hori, H.;
Igarashi, Y.; Polavarapu, P. L.; Sulikowski, G. A. Org. Lett. 2011, 13,
4538−4541.
(3) For studies towards the hibarimicins or their related natural
products, see: (a) Lee, C.-S.; Audelo, M. Q.; Reibenpies, J.; Sulikowski,
G. A. Tetrahedron 2002, 58, 4403−4409. (b) Maharoof, U. S.;
Sulikowski, G. A. Tetrahedron Lett. 2003, 44, 9021−9023. (c) Kim, K.;
Mahroof, U. S.; Raushel, J.; Sulikowski, G. A. Org. Lett. 2003, 5, 2777−
2780. (d) Narayan, S.; Roush, W. R. Org. Lett. 2004, 6, 3789−3792.
(e) Lambert, W. T.; Roush, W. R. Org. Lett. 2005, 7, 5501−5504.
(f) Lee, W. D.; Kim, K.; Sulikowski, G. A. Org. Lett. 2005, 7, 1687−
1689. (g) Li, J.; Todaro, L. J.; Mootoo, D. R. Org. Lett. 2008, 10,
1337−1340. (h) Li, J.; Todaro, L.; Mootoo, D. R. Eur. J. Org. Chem.
2011, 6281−6287. (i) Milgram, B. C.; Liau, B. B.; Shair, M. D. Org.
Lett. 2011, 13, 6436−6439.
(4) Tatsuta, K.; Fukuda, T.; Ishimori, T.; Yachi, R.; Yoshida, S.;
Hashimoto, H.; Hosokawa, S. Tetrahedron Lett. 2012, 53, 422−425.
(5) The use of a benzyl group was a strategic concession to protect
the sensitive core of 1 up until the last step of an eventual synthesis.
(6) For other examples of two-directional double annulation
reactions, see: (a) Hauser, F. M.; Gauuan, P. J. Org. Lett. 1999, 1,
671−672. (b) Reference 2. (c) Reference 4 and refs therein.
(7) Ferrier, R. J. J. Chem. Soc., Perkin Trans. 1 1979, 1455−1458.
(8) See the Supporting Information for full details.
(21) Benzyl phenyl sulfide substituted ortho-toluates were con-
currently found to be useful partners for naphthol annulations and
were ultimately employed in our synthesis of 2a and 6 due to the
inability to construct the C6-benzyl fluoride analogue of ( )-34.
(22) Bitha, P.; Hlavka, J. J.; Boothe, J. H. J. Med. Chem. 1970, 13, 89−
92.
(23) Pilcher, A. S.; Ammon, H. L.; DeShong, P. J. Am. Chem. Soc.
1995, 117, 5166−5167.
(24) (a) For brevity, each atropisomer is depicted as a single
structure lacking stereochemistry about the C2−C2′ bond. See the
Supporting Information for full details. (b) The regioisomer of the
enolized 1,3-diketone was undetermined and is arbitrarily depicted.
(25) No NMR or CD spectra for 3a and 6 have been previously
recorded according to ref 26. See the Supporting Information for full
details.
(26) Described in a personal communication with Professor Y.
Igarashi and Professor H. Hori.
(27) (a) Preferential oxidation of the D-ring occurs in simpler BCD-
1
ring model systems. (b) The H NMR signal of the methoxymethyl
groups of ent-27a and ent-27b is shifted over 0.6 ppm upfield relative
to the corresponding monomer, suggesting that they are positioned
over the naphthyl ring systems and subject to anisotropic magnetic
field effects.
(9) Kraus, G. A.; Sugimoto, H. Tetrahedron Lett. 1978, 19, 2263−
2266.
(10) Wendt, J. A.; Gauvreau, P. J.; Bach, R. D. J. Am. Chem. Soc. 1994,
116, 9921−9926.
(28) Kornblum, N.; Jones, W. J.; Anderson, G. J. J. Am. Chem. Soc.
1959, 81, 4113−4114.
(11) For related transformations utilizing ortho-quinone methides,
see: (a) Reference 4. (b) Hart, D. J.; Huang, H.-C. J. Am. Chem. Soc.
1988, 110, 1634−1635. (c) Nicolaou, K. C.; Lim, Y. H.; Piper, J. L.;
Papageorgiou, C. D. J. Am. Chem. Soc. 2007, 129, 4001−4013.
(12) For reviews, see: (a) Mal, D.; Pahari, P. Chem. Rev. 2007, 107,
1892−1918. (b) Rathwell, K.; Brimble, M. A. Synthesis 2007, 5, 643−
662.
(29) Sakulsombat, M.; Angelin, M.; Ramstrom, O. Tetrahedron Lett.
̈
2010, 51, 75−78.
(30) Kummer, D. A.; Li, D.; Dion, A.; Myers, A. G. Chem. Sci. 2011,
2, 1710−1718.
(31) The small coupling constant between the C6′ and C7′ hydrogen
atoms of (−)-35a and (+)-35b suggests a syn relationship of the
hydrogen atoms with respect to the ring system. This relative
stereochemistry would preclude syn-elimination to aromatize the F-
ring. Indeed, one diastereomer of the corresponding sulfoxide of
( )-34 underwent two-directional annulation but failed to eliminate to
aromatize the F-ring.
(13) 5-Methylvanillin (21) was prepared from vanillin in two steps
on multigram scale following a literature procedure: Sinhababu, A. K.;
Borchardt, R. T. Synth. Commun. 1983, 13, 677−683.
(14) (a) For other uses of Michael−Claisen reaction sequences to
construct naphthalene derivatives, see: Sun, C.; Wang, Q.; Brubaker, J.
D.; Wright, P. M.; Lerner, C. D.; Noson, K.; Charest, M.; Siegel, D. R.;
Wang, Y.-M.; Myers, A. G. J. Am. Chem. Soc. 2008, 130, 17913−17927
and references therein. (b) For a related approach to 3a, see ref 2.
(15) (a) ortho-Toluate and related carbanions will suffer from
competitive bimolecular self-condensation reactions with the ester
moiety if the Michael addition is not fast enough. For an interesting
discussion on the stability of ortho-toluate and related carbanions, see:
Brubaker, J. D. Ph.D. Thesis, Harvard University, 2007 and references
therein. (b) In the case of a single annulation process, the instability of
the deprotonated annulation donor can often be circumvented
through the use of excess donor. However, due to the inherent
stoichiometry of the two-directional double annulation, the biaryl
donor is used as the limiting reactant, and thus the stability of its
dianion is critical to the success of the reaction.
(32) The carbohydrates of the hibarimicin natural products are
cleaved with acidic methanol (1 M HCl, 30 °C). These conditions are
similar to those we employ during the benzyl deprotection and
oxidation of (−)-37a and (+)-37b. However, milder acidic conditions
(i.e., aq pH 3.5 phosphate buffer) in methanol may potentially be
substituted during the analogous deprotection and oxidation of 1
because these conditions are employed in the HPLC purification of 1
and hibarimicin related natural products. See ref 1f, g for the
conditions used for carbohydrate cleavage and purification of the
hibarimicin natural products.
(33) Eliel, E. L.; Wilen, S. H. Stereochemistry of Organic Compounds;
John Wiley & Sons, Inc.: New York, 1994.
H
dx.doi.org/10.1021/ja307207q | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX