Asymmetric synthesis of the aminocyclitol pactamycin, a universal translocation inhibitor

Article abstract of DOI:10.1021/ja409944u

An asymmetric total synthesis of the aminocyclopentitol pactamycin is described. The title compound is delivered in 15 steps from 2,4-pentanedione. Critical to this approach was the exploitation of a complex symmetry-breaking reduction strategy to assemble the C1, C2, and C7 relative stereochemistry within the first four steps of the synthesis. Multiple iterations of this reduction strategy are described, and a thorough analysis of stereochemical outcomes is detailed. In the final case, an asymmetric Mannich reaction was developed to install a protected amine directly at the C2 position. Symmetry-breaking reduction of this material gave way to a remarkable series of stereochemical outcomes leading to the title compound without recourse to nonstrategic downstream manipulations. This synthesis is immediately accommodating to the preparation of structural analogs.

Full text of DOI:10.1021/ja409944u

Article
pubs.acs.org/JACS
Asymmetric Synthesis of the Aminocyclitol Pactamycin, a Universal
Translocation Inhibitor
Robert J. Sharpe, Justin T. Malinowski, and Jeffrey S. Johnson*
Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290, United States
S
* Supporting Information
ABSTRACT: An asymmetric total synthesis of the amino-
cyclopentitol pactamycin is described. The title compound is
delivered in 15 steps from 2,4-pentanedione. Critical to this
approach was the exploitation of a complex symmetry-breaking
reduction strategy to assemble the C1, C2, and C7 relative
stereochemistry within the first four steps of the synthesis.
Multiple iterations of this reduction strategy are described, and
a thorough analysis of stereochemical outcomes is detailed. In
the final case, an asymmetric Mannich reaction was developed to install a protected amine directly at the C2 position. Symmetry-
breaking reduction of this material gave way to a remarkable series of stereochemical outcomes leading to the title compound
without recourse to nonstrategic downstream manipulations. This synthesis is immediately accommodating to the preparation of
structural analogs.
flexible synthesis of 1 is paramount for the success of such
endeavors.
INTRODUCTION
■
Nature continues to test the state of the art in organic synthesis
by providing chemists with both structurally complex and
biologically relevant molecules. Construction of these natural
products often requires the expansion of known synthetic
methods to previously unreported substrate classes or the
development of new approaches for the assembly of natural
frameworks.1 Isolated in 1961 from Streptomyces pactum var.
pactum, pactamycin (1) remains one of the most complex
aminocyclopentitol antibiotics known, bearing a remarkable
array of unique functionality and exceptional bioactivity.2
Pactamycin bears a densely functionalized cyclopentane core
featuring six contiguous stereogenic centers, three of which are
fully substituted.8 Unusual dimethylurea, aniline, and salicylate
functional groups adorn the core structure, presenting
numerous synthetic challenges. These issues have been
addressed in a number of approaches as synthetic interest in
pactamcyin has flourished over the past decade. Hanessian and
co-workers reported the landmark total synthesis in 2011,9,10
and Isobe, Knapp, Looper, Nishikawa, and our group have
disclosed access to advanced core intermediates by varying
methods.11−15 An inspection of these approaches reveals two
common challenges one faces in assembling the core structure:
(i) execution of chemo- and stereoselective reactions in a highly
congested chemical environment and (ii) the method by which
the unusual functionality of 1 is introduced. The Hanessian
group observed numerous side reactions due to functional
group propinquity.9,10 Looper and Haussener also noted the
importance of the order in which functional group manipu-
lations were executed.13 Approaches to introducing the unique
C1 dimethylurea have heretofore relied largely upon the use of
oxazoline or oxazolidinone protecting groups, necessitating
downstream deprotection and chemoselective acylation. In the
development of a synthesis plan, we took note of these issues
and sought to develop a synthesis of 1 that rapidly assembled
the core structure and incorporated all unique functionality in
the absence of nonstrategic redox and protecting group
manipulations.16
Figure 1. Pactamycin (1).
Pactamycin exhibits activity against both Gram-positive and
Gram-negative bacteria and is a powerful antitumor, antiviral,
and antiprotozoal agent.3 By acting as universal inhibitor of
translocation, pactamycin is known to inhibit protein synthesis
via a specific binding event within the 30S ribosomal subunit.4
Cytotoxicity levels as high as IC50 = 53 nM against certain
human cell lines have hindered its medicinal development;5
however, a number of biosynthetically engineered congeners
have been prepared which display diminished toxicity.6 These
data have reignited promise for the investigation of structure−
activity relationships (SAR) of 1 toward the goal of obtaining a
useful drug molecule.7 Thus, the necessity of a practical and
Received: October 1, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
In 2012, we presented an initial report on our efforts toward
a synthesis of 1,15 and earlier this year this work culminated in a
15-step asymmetric total synthesis.17 Herein, we present a full
account of our studies on pactamycin encompassing a
modification of our original route to accommodate early stage
incorporation of the C2 amine functionality. A symmetry-
breaking reduction for rapid access to the C1/C2/C7
stereotriad was developed, and an emphasis was placed on
incorporating pactamycin’s unique functionality in its final form
to obviate downstream functional group adjustment or
protecting group manipulation. This flexible route, we
surmised, would provide access to 1 in a manner amenable
to the synthesis of analogs for biological examination.
Scheme 2. Previously Reported Intermolecular Addition
Approach to 1
Our original retrosynthetic disconnection began with
simplification of 1 to functionalized cyclopentane 2 (Scheme
1). C2 (allylic) functionalization, C4 hydroxylation, and C3
addition mediated by CeCl3, provided the desired C5
stereochemistry with >20:1 diastereoselection. This intermedi-
ate was then elaborated to 15 in three steps. Experiments
directed toward installation of the C2 primary amine from 15
or its derivatives were extensively investigated in parallel with
some of the studies described below, but ultimately failed to
introduce the desired functionality.
Scheme 1. Initial Retrosynthetic Analysis for Pactamycin
RESULTS AND DISCUSSION
■
Desymmetrization Approach. While the above route was
scalable and effective for accessing advanced intermediate 15,
the synthesis of ketone 14 required a number of nonstrategic
redox and protecting group manipulations and lacked
efficiency. As a result, we sought a streamlined approach to
its synthesis. Cognizant of the undesired stereoselectivity
encountered in intermolecular addition to methyl ketone 12,
we envisaged that an intramolecular addition might provide the
opposite facial preference. Delivery of a tethered nucleophile
from the C7 hydroxyl followed by reduction and ring-closing
metathesis would intercept our previous intermediate 15. This
intermediate could be synthesized from our proposed
enantioselective desymmetrization strategy from urea 6, the
diketone analog of 5.
aniline installation might be possible from a C3−C4 alkene in
cyclopentene 3, a compound that was expected to be accessed
by ring-closing metathesis (RCM). The requisite precursor
would be derived from nucleophilic addition to methyl ketone
4. We surmised that this addition could occur either by inter- or
intramolecular nucleophile delivery; the latter facilitated by
attachment to the C7 secondary alcohol. Two approaches to β-
hydroxyketone 4 were envisioned, dependent upon the identity
of the R-substituent. For R = OMe (5), we proposed an
enantioselective Tsuji−Trost allylation of ketoester 7 followed
by diastereoselective ketone reduction and ester → ketone
conversion.15,18 Alternatively, if R = Me (6), we would invoke
an enantioselective, symmetry-breaking diketone monoreduc-
tion, exploiting the hidden symmetry (see outlined region in
Scheme 1) we perceived in the northeast quadrant of 1.19
Critical to our strategy in either case was the early stage
installation of the dimethylurea functionality in its final, native
form, an approach divergent from those previously reported. α-
Ureidodicarbonyls 7 or 8 would serve as our points of origin,
synthesized from commodity chemicals (methyl acetoacetate 9
or 2,4-pentanedione 10, respectively).
The diketone reduction precursor 6 was prepared in three
steps (Scheme 3). The reaction of 2,4-pentanedione 10 with
para-acetamidobenzenesulfonyl azide (pABSA)20 and Et3N
afforded the corresponding diazoketone in nearly quantitative
yield. An N−H insertion reaction analogous to that used in our
previous studies delivered α-ureidodiketone 8 in 67% yield,21
and Tsuji−Trost allylation provided the necessary diketone
precursor 6 in 81% yield. Working first to optimize the racemic
reaction, we began screening reducing agents and conditions
for selectivity. Gratifyingly, LiAl(OtBu)3H (LTBA) emerged
early in our evaluation, providing β-hydroxyketone ( )-17 in
75% yield with >20:1 diastereoselection. The desired stereo-
chemistry was confirmed via the TBS protection of keto alcohol
17 and direct comparison with 12, which had been
independently synthesized via our previous route.15
We speculate that this reduction proceeds via chelated
structure 16 in which steric demand of the dimethylurea
functionality directs hydride addition to the least hindered face
of the enantiotopic ketones, delivering ( )-17 in high
selectivity.
From β-hydroxyketone 17, we began investigating intra-
molecular additions to the C5 ketone. We were encouraged by
the work of Crimmins and co-workers in the use of
organocuprate nucleophiles for initiating intramolecular,
A summary of our initial efforts is outlined in Scheme 2.15
Ester 11, prepared via the strategy outlined in Scheme 1, was
treated with Me3SiCH2Li to afford β-silyloxyketone 12 poised
for nucleophilic addition. From 12, a screen of nucleophiles and
conditions was investigated for access to the requisite C5
alcohol; however, while addition of a model 2-propenylmetal
nucleophile mediated by CeCl3 proceeded in good yield, this
reaction gave consistent preference for the undesired epimeric
C5 configuration (13). This stereochemical outcome neces-
sitated the synthesis of ketone 14, which, upon methide
B
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 3. Intramolecular Addition Approach to 15
Scheme 4. Proposed Routes to C2-Functionalized
Desymmetrization Precursors
C1/C2/C7 stereotriad within the first 4 steps of the synthesis,
from which strategic manipulation of the available functional
handles might give expedient access to 1. Beginning from the
previously synthesized α-ureidodiketone 8, an enantioselective
Tsuji−Trost allylation with difurylidene acetate 21 would install
a 2-furyl group at the C2 center (22); we felt that this group
could function as an amine surrogate via downstream oxidative
cleavage and Curtius rearrangement.24 Since the ideal
functionality at C2 would be the amine itself, a catalytic,
asymmetric Mannich reaction of 8 with a strategically
configured imine such as 23 was projected to deliver diketone
24 with carbamate-protected amine installed directly at C2.25
In both the allylation and Mannich scenarios, enantiose-
lective formation of the C2 asymmetric center would establish
the lone stereochemical element that would be responsible for
all subsequent diastereoselective manipulations. The ensuing
diastereoselective symmetry-breaking monoreduction of a
chiral diketone would be the key for controlling the C1/C7
configurations and would require effective guidance from the
initially installed C2 stereocenter. The identities of the alkene
termini in the generic structures 21 and 23 can be disregarded
since downstream operations would purge these functionalities.
Both of these proposed pathways would deliver the entire core
skeleton of 1 within the first 3 steps. Equipped with these new
hypotheses, we began pursuing each in parallel.
C2-Furan Approach. Our first challenge in realizing the
Tsuji−Trost allylation approach was the synthesis of the allylic
acetate 21, which surprisingly appears to be a new compound
(Scheme 5). To this end, reduction of difurylpropenone 25
afforded the corresponding alcohol 26; however, upon
concentration of the crude mixture this product rapidly
decomposed. The observed decomposition was unexpected
since this compound had been previously reported and no note
had been made regarding its instability.26 A screen of
conditions designed to circumvent this problem revealed that
NaBH4 reduction of 25 followed by immediate acylation using
Et2O as the solvent provided 21 in crude form.27 Acetate 21
was also found to be unstable but could be stored in solution
for up to 30 d at 0 °C.
a
Conditions: (a) pABSA, Et3N, CH3CN; 0 °C to rt; (b) Rh2Oct4 (0.4
mol %), 1,1-dimethylurea, C7H8/DCE (1:1), 80 °C; (c) [allylPdCl]2
t
(0.5 mol %), rac-BINAP (1.1 mol %), BuOK, allyl acetate, C7H8, rt;
(d) LiAl(OtBu)3H, THF, −40 °C; (e) 2-butynoic acid, DMAP (10
mol %), DIC, Et2O, −20 °C to rt; (f) Me2CuLi, Et2O, −78 °C to rt.
alkylative cyclizations and surmised that an appropriately
selected pronucleophile on the C7 hydroxyl could provide
the desired reactivity.22 Thus, acylation of monoalcohol 17 with
2-butynoic acid delivered ynoate 18 in 79% yield.23 This
esterification set the stage for the proposed cyclization.
Me2CuLi emerged from a screen of known conjugate
nucleophiles and conditions to deliver lactone 19 in 53%
yield and 3:1 dr. The desired relative configuration of the C1/
C5/C7 stereotriad was confirmed by nuclear Overhauser effect
spectroscopy (NOESY) analysis.
At this juncture, only reduction remained to provide triol 20;
this intermediate would effectively intercept the synthesis of
cyclopentanone 15 in a highly efficient manner. However, an
exhaustive screen of reducing agents and conditions failed to
provide alcohol 20. Hindered reducing agents such as DIBAL-
H and LTBA displayed no reactivity even at elevated
temperatures, while unhindered reducing agents (LiAlH4,
Super Hydride) resulted only in complex mixtures or reduction
of the dimethylurea functionality. Additionally, attempts at ring-
opening of 19 via transesterification to its corresponding ester
or thioester failed to show any desired reactivity.
Having reached a second impasse, we began to form
conclusions regarding our original and revised strategies.
First, early stage incorporation of the dimethylurea function-
ality, while a strategic risk at the onset of this work, had proven
useful in directing desirable stereochemical outcomes in each of
our initial routes. The impressive diastereoselectivity accessed
from symmetry-breaking reduction of diketone 6 gave us cause
to incorporate this strategy again in future routes to 1; however,
neither our previously reported approach nor the above
strategy addressed a major problem facing the endgame of
our synthesis, namely, the late-stage installation of the primary
amine at C2. In devising a new approach, we envisioned
enantioselective installation of C2 functionality on ureidodike-
tone 8 prior to the symmetry-breaking reduction (Scheme 4).
Monoreduction of this substrate would provide access to the
With acetate 21 in our possession, the reaction of diketone 8
with difurylidene acetate 21 under the previously optimized
allylation conditions afforded C2-functionalized diketone
( )-27 in 80% yield. From this compound, we began to
examine conditions by which we might effect symmetry-
breaking reduction. Monoreduction of the chiral diketone 27
presents a more complicated problem than the reduction of
diketone 6, since four diastereomeric products can result from
the former. Exposure of diketone 27 to our previously
optimized desymmetrization conditions (LTBA, THF, −40
°C) resulted only in retro-aldol decomposition. A screen of
C
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
methylenation with subsequent RCM. As these routes were
pursued, however, it was quickly found that furan 30 was not
compatible with standard oxidative cleavage conditions (O3,
Johnson−Lemieux, RuCl3, etc.), giving only complex mixtures
or starting material decomposition. Turning to the metathesis
strategy, we began investigating α-methylenation protocols.
Using the conditions recently reported by Connell and co-
workers, treatment of 30 with (HCHO)n and diisopropylam-
monium trifluoroacetate30 afforded the undesired Diels−Alder
adduct 31, effectively rendering our RCM approach unfeasible.
These results cast doubt as to whether our proposed C2-furan
approach would provide access to 1. In addition, the problems
encountered in attempted oxidative cleavage of ketoester 30
gave us concern as to whether a late-stage unmasking of C2-
furan could be realized. With these data in hand, we turned our
attention toward developing a route to 1 from an early stage
Mannich reaction.
Scheme 5. Desymmetrization of 2-Furyl Substituted
Diketone Derivative 27 and Advancement toward the
Synthesis of 1
a
C2-Carbamate Approach. Our strategy for direct
installation of a protected amine at C2 was inspired by the
work of Schaus and co-workers wherein cinchona alkaloids
catalyzed 1,3-dicarbonyl Mannich addition to carbamoyl
aldimines. Expansion to our system would involve a new class
of nucleophile possessing α-ureido functionality (Scheme 6).25
Scheme 6. Development of a Mannich Reaction for
Installation of C2 Functionality
a
Conditions: (a) NaBH4, THF:H2O (1:1), rt; (b) Ac2O, NEt3,
DMAP (5 mol %), Et2O, rt; (c) [allylPdCl]2 (2.5 mol %), rac-BINAP
t
(5.28 mol %), BuOK, C7H8, rt; (d) LiAl(OtBu)(iBu)2H, THF, −78
°C; (e) TBSCl, imidazole, CH2Cl2, rt; (f) LDA, THF, −78 °C, then
ethyl cyanoformate, −78 to −20 °C; (g) diisopropylammonium
trifluoroacetate, (HCHO)n, THF, 65 °C.
Also crucial to the success of this method would be selection of
the appropriately protected imine electrophile. In the event, we
proceeded with Cbz-protected cinnamyl imine 32 and began
testing conditions for the Mannich reaction. Working first in
the racemic series, the reaction of α-ureidodiketone 8 with
reducing agents revealed that the reaction of 27 with
LiAl(OtBu)(iBu)2H (LDBBA)28 afforded monoalcohol 28
with moderate diastereoselectivity (4:1 28:Σ other diaster-
eomers). While the reduction of diketone 6 (lacking any C2
substituent) required warmer temperatures and extended
reaction times, reduction of 27 was complete within 10 min
at −78 °C. An X-ray diffraction study of the major diastereomer
confirmed the exact relative configuration needed for
elaboration to 1. While we envision a chelation mode similar
to transition structure 16 might be taking place in this
reduction, the involvement of the C2 furan in directing the C1/
C7 relative configuration and its dramatic effect on the
reactivity are not well understood.
imine 32 in the presence of catalytic quantities of Hunig’s base
̈
delivered Mannich product ( )-33 in 90% yield. With the
feasibility of this bond construction established, focus turned to
finding a suitable chiral catalyst for the reaction. An extensive
screen of known Mannich reaction promotors31 revealed
cinchonidine to be a superior catalyst, providing (+)-33 in
nearly quantitative yield with 84:16 er. Upon trituration,
crystalline racemic 33 could be removed by filtration, leaving
highly enantioenriched (+)-33 in 70% yield and 98:2 er. Unsure
of the enantiomer’s configuration, we proceeded with
optimization of this route with racemic material.
Having accessed this key intermediate, we proceeded to test
conditions for functionalization of the C5 methyl ketone. Based
on our previous work,15 we anticipated potential stereo-
selectivity and reactivity problems associated with nucleophilic
addition to the C5 carbonyl and accordingly decided to invoke
the ability of 28 to participate in enolate chemistry. Silyl
protection of β-hydroxyketone 28 proceeded smoothly to
deliver ketone 29 in 96% yield. The lithium enolate derived
from 29 reacted with ethyl cyanoformate to provide β-ketoester
30 in 80% yield.29 To access the cyclic core of 1 from this
functionality, we proposed two parallel strategies: (i) alkene
oxidative cleavage followed by aldol condensation or (ii) α-
From functionalized dicarbonyl 33, we turned toward
assembly of the C1/C2/C7 stereotriad via symmetry-breaking
reduction (Scheme 7). Referring to our previously optimized
conditions in the C2-unsubstituted case, monoreduction of 33
with LTBA at −35 °C provided β-hydroxy ketone 34 in 72%
yield with high diastereoselectivity (>10:1 34:Σ other
diastereomers).
Efforts to determine the relative configuration of this
monoalcohol were hampered, however, when initial studies
toward accessing a crystalline derivative proved unsuccessful.
Turning to spectroscopic methods, ozonolysis of the styrene
provided lactol derivative 35 from which NOESY analysis
D
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 7. Studies on Symmetry-Breaking Reduction of 33
a
Conditions: (a) LTBA, THF, −35 °C; (b) O3, CH2Cl2, −78; Me2S, rt; (c) I2, NaHCO3, 4 Å MS, 0 °C to rt; (d) LTBA, THF, −10 °C; (e) Zn,
AcOH, Et2O:MeOH (1:1), rt; (f) LDBBA, C7H8, −40 °C; (g) CSA, acetone:dimethoxypropane (1:1), rt; (h) DMP, CH2Cl2, rt; (i) TPAP, CH3CN,
−20 °C.
suggested the relative configuration of the C1/C2/C7 stereo-
triad illustrated in Scheme 7.
engaged as its derived cyclic iodoimidate 36 through the action
of I2 and NaHCO3. Subsequent monoreduction of diketone 36
followed by retrocyclization (mediated by Zn/HOAc) gave the
acyclic hydroxy ketone 37 in 61% yield over two steps. Ketone
37 is a diastereomer different from that accessed via LTBA
reduction of 33. We immediately began work in establishing its
stereochemical identity; however, NOESY analysis by a strategy
analogous to that used for 34 was inconclusive.
Concurrent with these studies, we pursued an alternate
strategy from the perspective of (−)-34. Namely, if the original
monoreduction product could be further reduced to its
corresponding diol (syn or anti), a site selective oxidation
might deliver the desired C1/C2/C7 configuration. To this
end, a screen of conditions revealed that direduction of 33 with
excess LDBBA afforded diol 38 as a 3:1 mixture of separable
diastereomers. The major isomer was revealed to be the 1,3-
trans diol via NOESY and 13C NMR analysis of the derived
acetonide 39.32 Control experiments revealed that this
reduction proceeds via the intermediacy of β-hydroxyketone
34. Consequently, the relative stereochemistry at C2 was
assigned according to that shown in alcohol 34.
With diol 38 in hand, we began evaluating oxidants for
symmetry-breaking oxidation. Treating diol 38 with Dess−
Martin periodinane (DMP) showed complete selectivity for
oxidation of a single site, returning the original hydroxyketone
34. Alternatively, tetrapropylammonium perruthenate (TPAP)
gave preference for the opposite alcohol, delivering mono-
alcohol 37 whose spectral characteristics matched those of the
compound prepared via the iodoimidate. The ability to access
37 from this route enabled us to assign its relative
stereochemistry, which had previously remained ambiguous
via NOESY analysis of its derivatives.
It is germane to emphasize at this juncture that the above
analysis hinged in its entirety on the NOESY analysis of 35,
which was suggestive of the illustrated structure but not
unequivocal. Because it was so easily accessible, we made the
conscious decision to move forward in our synthetic plan with
Stereochemical Analysis. With this tentative structure in
hand, we began to analyze the stereochemical outcome and its
ramifications. For planning purposes, we included both
enantiomeric series in this analysis based on the assumption
that either could be accessible via the catalytic, asymmetric
Mannich addition. β-Hydroxyketone (+)-34 is epimeric at C2
relative to pactamycin (1), a problem for which a solution was
not immediately obvious in light of our projected metathesis-
based synthetic plan. Alternatively, the enantiomeric form
(−)-34 presents C1 and C2 in the correct pactamycin
configuration but is a product resulting from incorrect
diastereotopic ketone site selectivity in the desymmetrization.
Although this was a discouraging initial result, we retained some
measure of confidence in our symmetry-breaking approach to 1
and began pursuing myriad strategies in parallel for the
elaboration of diketone 33 to our desired reduction
diastereomer.
We first pursued an exhaustive screen of reducing agents and
conditions in hopes that reagent control would provide
stereoselectivity different to that observed using LTBA.
Monoreduction with a number of bulky hydride sources (L-
Selectride, LDBBA, Red-Al, DIBAL-H) resulted only in the
formation of stereoisomer 34 in lower yields. Unhindered
hydride sources (LiAlH4, Super Hydride, NaBH4) gave only
minimal amounts of 34 accompanied with retro-aldol
decomposition pathways. Finally, alternative reduction path-
ways (enzymatic reduction, transfer hydrogenation) gave no
promise for delivering diastereoselectivity different to that
observed in LTBA reduction of diketone 34. These
unsuccessful efforts led us to the conclusion that this reduction
was proceeding with virtually complete substrate control, and as
a result, direct reduction strategies of 33 toward the desired
diastereomer were abandoned.
In an effort to alter the apparent conformational bias
associated with the acyclic structure 33, the diketone was
E
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(+)-34 in the interest of evaluating the chemical viability of our
remaining strategy, despite (or because of!) the absence of an
unambiguous stereochemical assignment. It was our hope that
rigorous stereochemical proof would be realized via a suitable
crystalline derivative later in the route and that the chemistry
developed during those studies could be effectively translated to
whatever diastereomer was needed.
Cyclopentane Core. Silyl protection of (+)-34 under
typical conditions proceeded smoothly, delivering β-silyloxy
ketone 40 in 84% yield (Scheme 8). Carboethoxylation of 40
nucleophilic aniline ring-opening sequence was pursued to
access the trans-anilinoalcohol, inspired by a related approach
by Hanessian and co-workers.9,10 We surmised that addition of
a suitable methide nucleophile to the C5 ketone would install
the final stereogenic center. Our experiments revealed that the
order of these steps and the identity of the C6 hydroxy-
methylene protecting group were critical. Thus, treatment of
enone 43 with NaOH/H2O2 delivered epoxy-alcohol 44 in 80%
yield with high diastereoselectivity. As in the case of the
intramolecular aldol condensation (42 → 43), the reaction was
ineffective if the C6 hydroxyl was protected. We then turned
our attention to installation of the C5 methyl group. In the
event, the sterically demanding TBDPS group37 was found to
be necessary to provide the desired stereoselectivity and
withstand the reaction conditions for nucleophilic addition.
Protection of 44 occurred uneventfully to provide TBDPS-
ether 45 in 76% yield, and treating this ketone with MeMgBr at
0 °C provided alcohol 46 (diastereoselection >10:1).
Scheme 8. Access to Cyclopentane Core via Formaldehyde
a
Aldol/Condensation
Stereochemical Outcome. Having arrived at an inter-
mediate bearing all six of pactamycin’s stereocenters, we began
aggressively pursuing a crystalline intermediate to confirm (or
disprove) our earlier stereochemical analyses. Carboxybenzyl
deprotection of 46 occurred under hydrogenolysis conditions
to deliver the corresponding primary amine, which crystallized
readily (Figure 2). X-ray analysis of this derivative confirmed
the desired relative stereochemistry at all six centers.
a
Conditions: (a) TBSOTf, 2,6-lutidine, CH2Cl2, −78 °C; (b) LDA,
THF, −78 °C, then CH2O(g), −45 °C; (c) O3, −78 °C, CH2Cl2;
Me2S, rt; (d) NaOMe, THF:MeOH (4:1), 0 °C; (e) H2O2 (30% aq),
NaOH (20% aq), MeOH:CH2Cl2 (7:1), 0 °C; (f) TBDPSCl, NEt3,
DMAP (10 mol %), CH2Cl2, 0 °C to rt; (g) MeMgBr, THF, 0 °C.
under the previously optimized conditions (cf. 29 → 30)
proceeded uneventfully to deliver the corresponding ketoester
in good yield; however, this intermediate could never be
successfully advanced despite extensive efforts.33 Accordingly,
we began to investigate routes by which we might directly
install the requisite C4 hydroxymethylene in its correct
oxidation state for elaboration to 1. This approach would
deliver a less activated β-hydroxyketone for subsequent
intramolecular condensation. Literature examples for the use
of formaldehyde as an aldol electrophile in complex total
synthesis are limited. Trost and co-workers have demonstrated
its use in their synthesis of corianin.34 The Cao group likewise
has shown the use of CH2O in aldol reactions en route to a
total synthesis of malyngamide U.35 After significant exper-
imentation in our system, we found that bubbling gaseous
CH2O (generated by the pyrolysis of paraformaldehyde)
through a solution of the lithium enolate of 40 at −45 °C
furnished the desired primary alcohol 41 in 70% yield.
Subsequent styrene ozonolysis delivered the corresponding
crude α-carbamoyl aldehyde 42 poised for intramolecular
condensation. NaOMe emerged as a superior promoter from
our screen of conditions, delivering cyclopentenone 43 in 50%
yield from 41. This transformation was rendered completely
ineffective if the C6 hydroxymethylene was protected.36
Fortunately, elimination of H2O strongly favors the formation
of the endocyclic alkene over its constitutional isomer, the α-
methylidene cyclopentanone.
Figure 2. X-ray structure of Cbz-deprotected 46; the silyl groups are
truncated for clarity. Disorder exists in the TBS group.
This surprising confirmation of correct stereochemistry led
us to a number of conclusions regarding the observed results.
With regard to the C5-methylation, nucleophile addition to the
convex surface of similar oxabicyclo[3.1.0]hexanone systems is
well documented; in our system, this trajectory would have
delivered the incorrect C5 configuration (Scheme 9).
Indeed, Hanessian and co-workers witnessed exclusively
convex surface addition of a methide nucleophile to ketone 47
in their total synthesis of 1.9,10 A five-step sequence from 48
provided the correct epoxide stereochemistry for synthesis
completion. Greaney and co-workers likewise observed this
facial preference in the addition of an alkyllithium nucleophile
to ketone 49 in their syntheses of merrilactone A and
anislactone A.38 In our system, however, this inherent
preference was seemingly overridden, delivering the desired
stereochemistry at C5. In the present case, we speculate that
With cyclopentenone 43 in hand, only C5 nucleophilic
addition, C4 hydroxylation, and installation of the C3 aniline
remained to complete the core structure of 1. An epoxidation/
F
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 9. Convex versus Concave Surface Addition of
Carbon Nucleophiles to Oxobicyclo[3.1.0]hexanones
Scheme 10. Deuterium-Labeling Experiments and
Unambiguous Stereochemical Assignment of 34
this selectivity is observed at least in part due to direction by
the C1-dimethylurea, providing additional support to the
decision to incorporate this functionality in its native form
early in the synthesis. Furthermore, the presence of two large
silyl groups on the convex face of epoxide 45 might serve to
block the undesired facial approach.
The presence of the desired C1/C2/C7 stereotriad in 46
seemed at odds with our original stereochemical assignment of
hydroxyketone 34 based on NOESY analysis of lactol derivative
35; however, conscious of the fact that the C2 stereocenter had
potentially become configurationally labile in the form of
aldehyde 42, we questioned whether epimerization had taken
place during base-promoted condensation to deliver the desired
C2 configuration. We devised a deuterium-labeling experiment
to examine the possibility of this pathway (Scheme 10).
Treating α-carbamoyl aldehyde 42 with NaOCD3 in CD3OD
using the optimized conditions afforded enone 43 with
complete incorporation of deuterium at C2. When this
experiment was conducted at −10 °C for the same time
pseudochair conformer 52 gives rise to the observed
monoreduction diastereomer.
Scheme 11. Proposed Stereochemical Model for LTBA
Reduction of 33
1
duration, a complex mixture of products was observed by H
NMR spectroscopy. Resubmission of this unpurified complex
mixture to the reaction conditions at 0 °C afforded enone 43
with complete D-incorporation. Finally, submission of the
product enone 43-d0 to NaOCD3 in CD3OD returned the
starting material with no deuterium incorporation. These
results clearly indicate that the C2 methine was undergoing
proton exchange prior to the condensation; however,
unambiguous confirmation of epimerization required X-ray
analysis of an upstream intermediate. Returning to our previous
attempts at derivatization of hydroxyketone (+)-34 led to the
synthesis of enantioenriched benzoate derivative 51, which
crystallized readily. X-ray analysis of 51 established the sense of
induction in the asymmetric Mannich addition and confirmed
the existence of the incorrect C2 configuration in the
desymmetrization product 34. In light of this result, we
speculate that formation of the observed monoreduction
diastereomer 34 arises via the chelated structure 52, a proposal
modeled after a similar case reported by Davis and co-workers
(Scheme 11).39 Preferential re-face addition of hydride to
From the complete set of experiments relating to the
pactamycin stereochemistry problem, the following conclusions
can be made (Scheme 12): (i) our original stereochemical
assignment of hydroxyketone 34 via NOESY analysis of 35 was
correct; (ii) the enantioselective Mannich addition (8 → 33)
had yielded the incorrect enantiomer nominally required for
elaboration to 1; and (iii) this stereochemical “mistake”,
compulsory for directing the correct C1/C7 stereochemistry in
the symmetry-breaking reduction (33 → 34), was later
corrected via epimerization in the aldol condensation (42 →
43). Incredibly, this series of events had taken place
unbeknownst to us until crystallographic evidence of a much
G
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 12. Completion of the Synthesis of Pactamycin
a
Conditions: (a) LTBA, THF, −35 °C; (b) TBSOTf, 2,6-lutidine, CH2Cl2, −78 °C; (c) LDA, THF, −78 °C, then CH2O(g), −45 °C; (d) O3, −78
°C, CH2Cl2; Me2S, rt; (e) NaOMe, THF:MeOH (4:1), 0 °C; (f) H2O2 (30% aq), NaOH (20% aq), MeOH:CH2Cl2 (7:1), 0 °C; (g) TBDPSCl,
NEt3, DMAP (10 mol %), CH2Cl2, 0 °C to rt; (h) MeMgBr, THF, 0 °C; (i) m-acetylaniline, Sc(OTf)3, C7H8, 60 °C; (j) TBAF, THF, 0 °C; (k) 55,
K2CO3, DMA, rt; (l) H2, Pd(OH)2/C (50% mass), MeOH, rt.
later intermediate led us to suspect the validity of our original
stereochemical analysis.
enantioselective Mannich addition of α-ureidodicarbonyls was
developed via an adaptation of the Schaus conditions.25
A
Completion of Synthesis. Our plan to complete the
synthesis began with the development of a Lewis acid-
promoted aniline epoxide opening to install the required m-
acetylaniline in 1. A similar approach had been employed by
Hanessian and co-workers for introduction of the C3/C4 trans-
anilinoalcohol, whereupon the requisite aniline was incorpo-
rated via its m-propenyl derivative.9,10 The necessary
acetophenone was later revealed via oxidative cleavage of the
olefin. By contrast, we hoped that the required aniline could be
installed in its native form, obviating downstream introduction
of the ketone. After considerable experimentation, we found
that addition of m-acetylaniline to epoxide 46 promoted by
Sc(OTf)3 delivered the desired trans-aminoalcohol 53 in 66%
yield. It is important to note that while this transformation
proceeds in moderate yield, the use of more electron-rich
anilines in the reaction delivers the corresponding epoxide-
opened products in high yield, a valuable result as this step is a
crucial branch point for analog synthesis. Silyl deprotection
proceeded readily upon treatment of 53 with tetrabutylammo-
nium fluoride (TBAF), providing tetraol 54 in 90% yield.
Installation of the salicylate moiety was accomplished via
treatment of alcohol 54 with an in situ generated ketene
electrophile derived from cyanomethylester 55.40 This left only
removal of the C2 protecting group to complete our synthesis.
Cbz-deprotection was effected readily upon hydrogenation of
56 in the presence of Pearlman’s catalyst to deliver pactamycin
(1) in 82% yield.41
symmetry-breaking reduction was employed for rapid delivery
of the C1/C2/C7 stereotriad in 1, and proposed stereo-
chemical models for these reductions are presented herein. In
the case of the C2-carbamate approach, a thorough analysis of
monoreduction stereochemical outcomes is presented. These
studies culminated with the conclusion that selective oxidation
of trans-diol 38 could allow access to a suitable monoreduction
diastereomer of 33 for elaboration to 1. This deduction
directed the decision to move forward in our strategy without
unambiguous stereochemical confirmation of alcohol 34 with
the assumption that the necessary relative configuration of 1
could be realized later in the synthesis. The stereochemical
identity of 34 was later unambiguously determined to be
incorrect at C2, although this “stereochemical error” was
corrected via epimerization during a downstream aldol
condensation. This fortuitous turn of events allowed facile
access to 1 in the absence of nonstrategic stereochemical
manipulations. This route is flexible and immediately amenable
to the synthesis of structural analogs as major functional groups
(aniline, salicylate) are incorporated in a late-stage fashion.
Studies toward the preparation of analogs for analysis of
structure−activity relationships are ongoing in our laboratory
and will be reported in due course.
ASSOCIATED CONTENT
* Supporting Information
■
S
Additional experimental procedures, characterization, and
spectral data for all compounds, and crystallographic data
(CIF). This material is available free of charge via the Internet
at http://pubs.acs.org.
CONCLUSION
■
In summary, we have detailed the entirety of our efforts toward
the synthesis of pactamycin, culminating in an asymmetric, 15-
step total synthesis in 1.9% overall yield from commodity
chemical 2,4-pentanedione. Emphasis was placed on incorpo-
ration of all unique functionality (dimethylurea, aniline,
salicylate) in its native form for minimization of protecting
group manipulations. Revision of our originally published
strategy led to the development of a novel alkylative cyclization
for intramolecular delivery of C5 stereochemistry. A need to
incorporate C2 functionality early stage gave rise to the
synthesis of a new difurylidene acetate reagent which was
employed in a complex Tsuji−Trost allylation, and an
AUTHOR INFORMATION
Corresponding Author
jsj@unc.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The project described was supported by award no. R01
GM084927 from the National Institute of General Medical
H
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(21) Lee, S. H.; Yoshida, K.; Matsushita, H.; Clapham, B.; Koch, G.;
Zimmermann, J.; Janda, K. D. J. Org. Chem. 2004, 69, 8829.
(22) Crimmins, M. T.; Mascarella, S. W.; DeLoach, J. A. J. Org. Chem.
1984, 49, 3033.
(23) 2-Butynoic acid was used due to low yields associated with
esterification of 17 with propynoic acid.
(24) Schmid, G.; Fukuyama, T.; Akasaka, K.; Kishi, Y. J. Am. Chem.
Soc. 1979, 101, 259.
(25) (a) Lou, S.; Taoka, B. M.; Ting, A.; Schaus, S. E. J. Am. Chem.
Soc. 2005, 127, 11256. (b) Ting, A.; Lou, S.; Schaus, S. E. Org. Lett.
2006, 8, 2003.
Sciences. R.J.S acknowledges an NSF Graduate Research
Fellowship. J.T.M. acknowledges an ACS Division of Organic
Chemistry graduate fellowship. X-ray crystallography was
performed by Dr. Peter White (UNC). CCDC 914582,
959045, and 959046 contain the supplementary crystallo-
graphic data for this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Centre via www.
ccdc.cam.ac.uk/data_request/cif.
REFERENCES
■
(26) (a) Paxton, R. J.; Taylor, R. J. K. Synlett. 2007, 4, 633.
(b) Mamaghani, M.; Mahmoodi, N. O.; Moghisseh, A. A.;
Pourmohamad, L. J. Iran. Chem. Soc. 2008, 5, 238.
(27) See the SI.
(1) (a) Nicolaou, K. C.; Sorensen, E. J. Classics in Total Synthesis;
VCH: New York, 1996. (b) Nicolaou, K. C.; Snyder, S. A. Classics in
Total Synthesis II; VCH: New York, 2003.
(2) Argoudelis, A. D.; Jahnke, H. K.; Fox, J. A. Antimicrob. Agents
(28) Kim, M. S.; Choi, Y. M.; An, D. K. Tetrahedron Lett. 2007, 48,
Chemother. 1962, 191.
5061.
(3) (a) Weller, D. D.; Haber, A.; Rinehart, K. L.; Wiley, P. F. J.
Antibiot. 1970, 31, 997. (b) Wiley, P. F.; Jahnke, H. K.; MacKellar, F.
A.; Kelly, R. B.; Argoudelis, A. D. J. Org. Chem. 1970, 35, 1420.
(4) (a) Dinos, G.; Wilson, D. N.; Teraoka, Y.; Szaflarski, W.; Fucini,
P.; Kalpaxis, D.; Nierhaus, K. H. Mol. Cell 2004, 13, 113. (b) Carter, A.
P.; Clemons, W. M., Jr.; Brodersen, D. E.; Morgan-Warren, R. J.;
Wimberly, B. T.; Ramakrishnan, V. Nature 2000, 407, 340.
(5) (a) Otogouro, K.; Iwatsuki, M.; Ishiyama, A.; Namatame, M.;
Nishihara-Tukashima, A.; Shibahara, S.; Kondo, S.; Yamada, H.;
(29) Crabtree, S. R.; Chu, W. L. A.; Mander, L. N. Synlett. 1990, 169.
(30) Bugarin, A.; Jones, K. D.; Connell, B. T. Chem. Commun. 2010,
46, 1715.
(31) See the SI for a complete list of catalysts screened.
(32) Rychnovsky, S. D.; Skalitzky, D. J. Tetrahedron Lett. 1990, 31,
945.
(33) α-Methylenation protocols toward an RCM strategy gave only
starting material decomposition. Ozonolysis provided the correspond-
ing crude α-carbamoyl aldehyde necessary for intramolecular
condensation, but this aldehyde was ultimately too reactive to
participate in productive chemistry. See the SI for experimental details.
(34) Trost, B. M.; Haffner, C. D.; Jebaratnam, D. J.; Krische, M. J.;
Thomas, A. P. J. Am. Chem. Soc. 1999, 121, 6183.
(35) Li, Y.; Feng, J.-P.; Wang, W.-H.; Chen, J.; Cao, X.-P. J. Org.
Chem. 2007, 72, 2344.
Omura, S. J. Antibiot. 2010, 63, 381. (b) Lu, W.; Roongsawang, N.;
̅
Mahmud, T. Chem. Biol. 2011, 18, 425.
(6) (a) Almabruk, K. H.; Lu, W.; Li, Y.; Abugreen, M.; Kelly, J. X.;
Mahmud, T. Org. Lett. 2013, 15, 1678. (b) Iwatsuki, M.; Nishihara-
Tsukashima, A.; Ishiyama, A.; Namatame, M.; Watanabe, Y.;
Handasah, S.; Pranamuda, H.; Marwoto, B.; Matsumoto, A.;
(36) Verlaak, J. M. J.; Klunder, A. J. H.; Zwanenburg, B. Tetrahedron
Lett. 1982, 23, 5463.
Takahashi, Y.; Otoguro, K.; Omura, S. J. Antibiot. 2012, 65, 169.
̅
(c) Lu, W.; Roongsawang, N.; Mahmud, T. Chem. Biol. 2011, 18, 425.
(d) Otogouro, K.; Iwatsuki, M.; Ishiyama, A.; Namatame, M.;
Nishihara-Tukashima, A.; Shibahara, S.; Kondo, S.; Yamada, H.;
(37) Lavallee, P.; Hanessian, S. Can. J. Chem. 1975, 53, 2975.
(38) (a) Shi, L.; Meyer, K.; Greaney, M. F. Angew. Chem., Int. Ed.
2010, 49, 9250. (b) Intermediate diastereoselectivity has also been
observed in the oxabicyclo[3.1.0]hexanone system: Myers, A. G.;
Glatthar, R.; Hammond, M.; Harrington, P. M.; Kuo, E. Y.; Liang, J.;
Schaus, S. E.; Wu, Y.; Xiang, J.-N. J. Am. Chem. Soc. 2002, 124, 5380.
(39) Davis, F. A.; Gaspari, P. M.; Nolt, B. M.; Xu, P. J. Org. Chem.
2008, 73, 9619.
Omura, S. J. Antibiot. 2010, 63, 381. (e) Dobashi, K.; Isshiki, K.; Sawa,
̅
T.; Obata, T.; Hamada, M.; Naganawa, H.; Takita, T.; Takeuchi, T.;
Umezawa, H. J. Antibiot. 1986, 39, 1779.
(7) (a) Hanessian, S.; Vatiki, R. R.; Chattopadhyay, A. K.; Dorich, S.;
Lavallee
́
, C. Bioorg. Med. Chem. 2013, 21, 1775. (b) Tourigny, D. S.;
Fernandez, I. S.; Kelley, A. C.; Vakiti, R. R.; Chattopadhyay, A. K.;
́
(40) Shen, R.; Lin, C. T.; Bowman, E. J.; Bowman, B. J.; Porco, J. A.,
Jr. Org. Lett. 2002, 4, 3103.
(41) Pearlman, W. M. Tetrahedron Lett. 1967, 8, 1663.
Dorich, S.; Hanessian, S.; Ramakrishnan, V. J. Mol. Biol. 2013, 425,
3907.
(8) Duchamp, D. J. Abstracts. In Proceedings from American
Crystallographic Association Winter Meeting, Albuquerque, NM, April
3−7, 1972; American Crystallographic Association: Buffalo, NY, 1972;
p 23.
(9) Hanessian, S.; Vakiti, R. R.; Dorich, S.; Banerjee, S.; Lecomte, F.;
Del Valle, J. R.; Zhang, J.; Deschen
2011, 50, 3497.
(10) Hanessian, S.; Vakiti, R.; Dorich, S.; Banerjee, S.; Deschen
Simard, B. J. Org. Chem. 2012, 77, 9458.
(11) Tsujimoto, T.; Nishikawa, T.; Urabe, D.; Isobe, M. Synlett 2005,
433.
̂
es-Simard, B. Angew. Chem., Int. Ed.
̂
es-
(12) Knapp, S.; Yu, Y. Org. Lett. 2007, 9, 1359.
(13) Haussener, T. J.; Looper, R. E. Org. Lett. 2012, 14, 3632.
(14) Matsumoto, N.; Tsujimoto, T.; Nakazaki, A.; Isobe, M.;
Nishikawa, T. RSC Adv. 2012, 2, 9448.
(15) Malinowski, J. T.; McCarver, S. J.; Johnson, J. S. Org. Lett. 2012,
14, 2878.
(16) Burns, N. Z.; Baran, P. S.; Hoffmann, R. W. Angew. Chem., Int.
Ed. 2009, 48, 2854.
(17) Malinowski, J. T.; Sharpe, R. J.; Johnson, J. S. Science 2013, 340,
180.
(18) Kuwano, R.; Ito, Y. J. Am. Chem. Soc. 1999, 121, 3236.
(19) Magnuson, S. R. Tetrahedron 1995, 51, 2167.
(20) Baum, J. S.; Shook, D. A.; Davies, H. M. L.; Smith, H. D. Synth.
Commun. 1987, 17, 1709.
I
dx.doi.org/10.1021/ja409944u | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX