Progress toward the syntheses of (+)-GB 13, (+)-himgaline, and himandridine. New insights into intramolecular imine/Enamine aldol cyclizations

Article abstract of DOI:10.1021/ja3001776

A full account of our total synthesis of the galbulimima alkaloids GB 13 and himgaline is provided. Using a strategy adapted from the proposed biosynthesis of the GB alkaloid family, a linear precursor underwent successive intramolecular Diels-Alder, Mich

Full text of DOI:10.1021/ja3001776

Article
pubs.acs.org/JACS
Progress toward the Syntheses of (+)-GB 13, (+)-Himgaline, and
Himandridine. New Insights into Intramolecular Imine/Enamine Aldol
Cyclizations
David A. Evans,* Drew J. Adams, and Eugene E. Kwan
Department of Chemistry & Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, United States
S
* Supporting Information
ABSTRACT: A full account of our total synthesis of the galbulimima
alkaloids GB 13 and himgaline is provided. Using a strategy adapted from
the proposed biosynthesis of the GB alkaloid family, a linear precursor
underwent successive intramolecular Diels−Alder, Michael, and imine
aldol cyclizations to form the polycyclic alkaloid core. We now show that
modification of this strategy can also deliver an advanced intermediate en
route to the related alkaloid himandridine. The success of the key imine
aldol cyclization is acutely sensitive to substrate structure and solvent, including a case in which cyclization was spontaneous in
protic solvents. A detailed computational investigation of the course of the reaction closely correlates with, and suggests a
rationale for, the observed patterns of imine aldol reactivity.
also described the conversion of GB 13 to the related Type III
alkaloid himgaline. These syntheses led to a revision of the
absolute configuration of the Type III alkaloids. More recently,
an X-ray crystallographic analysis reported by Mander
supported revision of the Type II alkaloids.8 The first synthesis
of a Type II alkaloid (himandrine) was recently reported by
Movassaghi.9 Additional approaches to the syntheses of Type
III alkaloids have also been reported by Sarpong7d and Ma.7e
In this article, we provide a full account of our synthesis of
ent-galbulimima alkaloid GB 13 and ent-himgaline (Scheme
1).10 We also provide new experimental and computational
insights regarding the pivotal imine aldol addition step gained
during out current efforts directed toward the Type II alkaloid
himandridine. Notably, the desired aldol process was found to
be spontaneous in protic solvents, conditions congruent with
the proposed biosynthesis of these alkaloids. Computational
modeling of the thermochemistry of these and related imine
aldol additions indicates that these additions are nearly
thermoneutral.
INTRODUCTION
Background. The galbulimima alkaloids comprise 28
polycyclic isolates of Galbulimima belgraveana, a rare rain
forest tree found in Papua New Guinea and northern Australia
(Figure 1).1 Native tribes have used bark extracts from this tree
■
Subsequent analysis of the GB alkaloid polycycle reveals an
unexpected conformational preference that suggests a steric
rationale for why certain imine substrates readily cyclize while
others are unreactive.
Figure 1. Representative GB alkaloids.
for medicinal and spiritual purposes,2 and the Type I GB
alkaloid himbacine has been demonstrated to have potent,
selective antimuscarinic activity.3 Largely as a result of its
potentially therapeutic biological properties,4 multiple syn-
theses of himbacine5 and its simplified analogs6 have appeared.
The Type II and Type III GB alkaloids have also attracted
the attention of several research groups as a result of their
complex polycyclic architectures. In addition to the racemic
synthesis of GB 13 reported by Mander prior to the outset of
this work,7a both Movassaghi7b and Chackalamannil7c have
reported approaches to GB 13. Chackalamannil’s investigation
TYPE III ALKALOIDS
■
Synthesis Plan. Our strategy for the synthesis of the GB
alkaloids was influenced by the Ritchie/Taylor1i biosynthetic
hypothesis that has subsequently been elaborated by others.11
In this proposal, a monocyclic polyketide precursor undergoes a
succession of intramolecular bond constructions: Diels−Alder
Received: January 6, 2012
Published: April 25, 2012
© 2012 American Chemical Society
8162
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
Scheme 1. Proposed Synthesis Plan for Type III GB Alkaloids
cycloaddition, Michael, and imine aldol addition to construct
the polycyclic skeleton shared by the Type II and Type III
alkaloids. The current plan has integrated each of these bond
constructions (Scheme 1). With respect to the Type III
alkaloids, we anticipated that himgaline might arise from GB 13
after N-conjugate addition and stereoselective reduction at C16.
The bicyclo[3.2.1]octane himgaline core was envisioned to
arise from a keto-imine, which might undergo an imine aldol
addition, and iminium ion reduction to provide the desired
polycyclic skeleton. An intramolecular Michael addition was
envisioned to construct the necessary cyclopentanone moiety.
The Michael precursor might then be derived from the
elaboration of an appropriately functionalized decalin, which
could be obtained in enantiopure form via auxiliary-controlled
intramolecular Diels−Alder (IMDA) cycloaddition.
of the chiral auxiliary to provide the derived thioester was
followed by semireduction to provide aldehyde 7.
Olefination of 7 with phosphonate 85d afforded the desired
enone for the anticipated intramolecular Michael addition
(Scheme 3). Efforts to desilylate 9 in preparation for
introduction of the needed β-ketoester were complicated by
irreversible O-conjugate addition to provide tetrahydrofuran 10.
A single product diastereomer was produced, with the
configuration at the newly formed stereocenter matching that
desired for the planned intramolecular Michael addition (NOE
analysis). Nonselective ketone reduction of 9 followed by silyl
deprotection provided diol 11 while avoiding the undesired
conjugate addition process. N-Benzylation15 to provide 12 was
followed by simultaneous oxidation of both alcohols with Dess-
Martin periodinane.16
A Roskamp reaction with allyl diazoacetate17 and the keto-
aldehyde derived from 12 incorporated the necessary β-
ketoester that, after purification or storage at −20 °C, was
obtained primarily as 13, the product of β-ketoester enol
addition. As anticipated, enol addition proved to be reversible
under basic conditions: a combination of lithium methoxide
and lithium perchlorate in ether, conditions reported to provide
chelated β-ketoester anions,18 provided the desired Michael
adduct 14 in 62% yield over the three steps.
RESULTS AND DISCUSSION
■
Synthesis of ent-GB 13 and ent-Himgaline. Synthesis of
ent-himgaline began with 6,6-dimethoxyhexanal (1)12 which
underwent a vinylogous Horner−Wadsworth−Emmons
(HWE) olefination with triethyl 4-phosphonocrotonate
(Scheme 2). Following ester reduction and alcohol protection,
a
Scheme 2. Synthesis of Functionalized Decalin Subunit
The diastereoselectivity of this process could not be directly
assessed due to the carboalkoxy substituent; however,
palladium-catalyzed decarboxylation of the allyl ester resulted
in the isolation of a single cyclopentanone diastereomer 15,
leading to the conclusion that the Michael addition had
proceeded with high selectivity. Most likely a favored extended
conformation of the enone acceptor explains the observed
outcome. However, when the Michael addition was promoted
by cesium carbonate in ethanol, the stereoselectivity eroded
(3:1). Such conditions are reported to provide dipole
minimized β-ketoester enolates,18 suggesting that the β-
ketoester enolate geometry may play a secondary role in
determining the diastereoselectivity in this Michael addition.
Although our synthesis plan called for an imine/enamine
aldol addition, it is also evident that an enolate aldol addition
followed by amine deprotection and condensation might also
be considered in the construction of the requisite
bicyclo[3.2.1]octane synthon (Scheme 4).19 Unfortunately,
under a variety of enolization conditions across a variety of
substrates, no enolate aldol addition products were observed
with substrates related to either 16b or 18 (Scheme 3). The
overriding reactivity of the cyclopentanone moiety toward
enolization proved to be the obstacle in all instances.20 These
results reinforced the commitment to the intramolecular
enamine aldol approach.
a
Conditions: (a) See Supporting Information; (b) LiBF4, MeCN,
H2O; (c) 3, LiCl, i-Pr2NEt, MeCN; (d) Me2AlCl, PhMe, −30 °C; (e)
K2OsO4, NMO, acetone/pH 7 buffer; (f) dimethoxypropane, TsOH,
acetone; (g) LiSEt, THF, 0 °C; (h) DIBAL-H, PhMe, −90 °C.
acetal 2 was cleaved to the derived aldehyde. A second HWE
olefination with phosphonate 3 then incorporated the
illustrated chiral auxiliary.13 The desired IMDA cycloaddition
was effected using Me2AlCl to provide decalin 5 in 81% yield as
a single diastereomer.14 The necessary oxygenation at C16 was
incorporated by a diastereoselective dihydroxylation, which was
followed by diol protection as its derived acetonide 6. Removal
8163
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
a
Scheme 3. Intramolecular Conjugate Additions and Elaboration to Substrates for Aldol Cyclization
a
Conditions: (a) i-Pr2NEt, LiCl, MeCN; (b) TBAF, HOAc, THF; (c) DIBAL, PhMe, −90 °C; (d) TBAF, HOAc, THF; (e) see Supporting
Information; (f) DMP, NaHCO3, CH2Cl2; (g) allyldiazoacetate, SnCl2; (h) LiOMe, LiClO4, Et2O, 0−23 °C; (i) Pd(PPh3)4, morpholine, THF; (j)
DBU, PhH; (k) Pd(OH)2, H2, THF.
changes to the CO and CN resonances (160−220 ppm)
Scheme 4. Complementary Aldol Constructions
provided clear indications of reaction progress at each stage.
Acylation of iminium ion 24 with TFAA provided enamide 25
as the principal product, confirming successful aldolization. As
25 proved resistant to a variety of transition metal catalyzed and
ionic hydrogenation conditions, reduction of the iminium ion
prior to acylation was pursued. Sodium cyanoborohydride was
chosen to minimize unwanted reduction of the cyclohexanone
moiety; however, under all conditions investigated both the
iminium ion and cyclohexanone were reduced to provide amino
alcohol 26. Although full diastereocontrol was observed at the
three stereocenters formed during the aldol addition and
iminium ion reduction, the ketone reduction proceeded to give
Ent-Himgaline and Ent-GB 13. Scheme 3 summarizes the
construction of advanced intermediates for both the himgaline
a 2:1 mixture of axial and equatorial alcohol epimers. Initially,
and GB 13 syntheses. Several opportunities for the
this mixture was peracylated with TFAA to provide chromato-
implementation of the imine aldol construction are possible
graphically tractable material. Saponification of the trifluor-
from these intermediates. These options were then evaluated.
We were unable to effect aldol addition with imine 17. In
another study, imine 20 was subjected to a broad survey of
reaction conditions; however, no evidence for the desired aldol
addition was obtained (eq 1). Again, 21 also proved resistant to
aldol addition (eq 2). These cases illustrate the tenuous nature
of the aldol addition process.
oacetates provided separable diols 27a and 27b; as 27a
spectroscopically matched an intermediate encountered during
Mander’s synthesis of GB 13, a formal total synthesis of GB 13
had been completed.
In subsequent iterations, the mixture of C16 alcohol epimers
26 was reoxidized using DMP immediately following iminium
ion reduction. The high chemoselectivity observed in this
oxidation is likely derived from the highly hindered environ-
ment surrounding the secondary amine. Acylation of keto-
amine 28 using benzyl chloroformate facilitated purification of
carbamate 29, which was isolated in 39% yield over six steps.
No unwanted diastereomers were isolated from this process.
The necessary unsaturation was introduced using IBX7b,21
(90%); TMSI then cleaved the benzyl carbamate to deliver ent-
GB 13.7b Synthetic GB 13 matched a natural sample
spectroscopically and gave an optical rotation of similar
magnitude but opposite sign (ent-GB 13, [α]20 +71.6 (c 0.2,
D
CHCl3); GB 13, [α]20 −76 (c 1.0, CHCl3)).
Triketone 19 (Scheme 3) ultimately proved to be a tractable
D
The conversion of (+)-GB 13 to (+)-himgaline required
conjugate addition of the piperidine nitrogen to the pendant
enone and stereoselective reduction of the C16 ketone. During
isolation and purification of GB 13, it became clear that even
trace quantities of acid promoted the desired N-conjugate
addition to an appreciable extent (10−15%). Stirring (+)-GB
substrate for the requisite aldol transformation (Scheme 5).
Carbamate deprotection and dehydration of the resulting
hemiaminal 22 afforded imine 23, which gratifyingly provided
the desired aldol adduct 24 upon treatment with acetic acid in
THF (Scheme 5). In situ 13C NMR spectroscopy proved
especially useful during this sequence of transformations, as
8164
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
a
Scheme 5. Ent-GB 13 and Ent-Himgaline Syntheses
a
Conditions: (a) 20% TFA/CH2Cl2, 0 °C, aq. NaHCO3 workup; (b) 4 Å mol. sieves, PhH; (c) HOAc, THF, 0−23 °C; (d) TFAA, NEt3, CH2Cl2, 0
°C; (e) NaBH3CN, EtOH, 0−23 °C; (f) LiOMe, THF; (g) DMP, CH2Cl2; (h) benzyl chloroformate (i) IBX, TsOH·H2O, DMSO/PhH, 65 °C; (j)
TMSI, CH2Cl, 0 °C; HCl; NaOH, 23 °C; (k) HOAc, MeCN, 30 min; NaBH(OAc)3.
13 in a 1:1 mixture of acetic acid and acetonitrile fully
promoted the desired conjugate addition within minutes;22
subsequent addition of sodium triacetoxyborohydride initiated
directed hydride delivery to furnish ent-himgaline in 90% yield.
Synthetic himgaline was spectroscopically identical to a natural
sample and gave an optical rotation of similar magnitude but
isolated in 57% yield. Elaboration of this synthon using
standard conditions allowed assembly of polyene 33, a substrate
for cycloaddition. In contrast to our himgaline approach
(Scheme 2), diastereoselectivity was predicted to be controlled
by 1,3-allylic strain at C14.24 Gratifyingly, under thermal
activation polyene 34 cyclized to the desired Diels−Alder
adduct in high yield and diastereoselection (82%, 9:1 dr).
Subsequent installation of the enone Michael acceptor and β-
keto ester led to 35, a substrate for the planned intramolecular
Michael cylization. Under conditions established during the
synthesis of himgaline (LiClO4, LiOMe, Et2O; Scheme 3),
Michael addition was accompanied by olefin isomerization to
allylic bromide 36. Use of milder soft enolization conditions
(LiBr, NEt3, EtOAc) suppressed formation of side products but
did not prevent olefin migration. The unstable C16 allylic
bromide subsequently underwent oxidation to the derived
ketone in the presence of silver(I) trifluoroacetate and air.25
Allyl ester removal and stereoselective olefin reduction (Zn,
HOAc) led to triketone 38, a substrate for imine aldol
cyclization in analogy to our himgaline precedent (Scheme 5).
Imine Aldol Additions: Himandridine. The synthesis
plan for himandridine was based on the himgaline precedent
for the analogous aldol construction (Scheme 7).
opposite sign (ent-himgaline, [α]20 +80.0 (c 0.1, CHCl3);
D
himgaline, [α]20 −84 (c 1.0, CHCl3)).
D
Progress toward the Synthesis of Himandridine. We
next sought to extend our previously disclosed synthesis of
himgaline to the type II alkaloid himandridine, which features
additional oxygenated stereocenters at C13 and C14 as well as
attachment of nitrogen at C9 rather than C19 (Figure 1).
Although our synthesis plan remained similar to our himgaline
approach (Scheme 1), the structure of himandridine ultimately
required that each of our key intramolecular cyclization steps be
revised. To address the addition of oxygenation at C13 and
C14, we relied on previously developed Sn(II)-promoted
antiselective glycolate aldol methodology (Scheme 6).23 Good
diastereoselectivity was observed in this process (3 to 1, desired
to sum of undesired isomers), and the desired adduct was
The synthesis of imine 39 is outlined in Scheme 8.
Surprisingly, this substrate did not undergo the anticipated
aldol addition under conditions identical to those used during
the himgaline synthesis (eq 3 vs eq 4). These observations
prompted a study of solvent effects on the aldol process. After
a
Scheme 6. Key Steps toward the Synthesis of Himandridine
a
Conditions: (a) Sn(OTf)2, NEt3, TMEDA, CH2Cl2, −78 °C; (b) PhMe, BHT, reflux, 48 h; (c) LiBr, iPr2NEt, EtOAc, 0 °C; (d) Ag(I)
trifluoroacetate, DMF, air; (e) Pd(PPh3)4, morpholine, THF; (f) Zn, HOAc.
8165
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
sodium cyanoborohydride spared the C16 ketone; following N-
acylation, the desired polycyclic adduct 42 (Scheme 8) was
obtained in 45% yield over four steps.26 As before, no unwanted
diastereomers were obtained from this sequence.
Scheme 7. Comparative Enamine Aldol Additions
1
When aldol cyclization of 39 was monitored by H NMR
spectroscopy (d4-MeOH), maximal product formation was
achieved within 45 min of dissolution. Neither prolonged
reaction times (24 h) nor addition of acetic acid or
triethylamine to the reaction mixture altered the product
distribution. As cyclization of 39 was spontaneous in methanol
but not in chloroform, benzene, or THF, a methanol solution
of aldol adduct 40 was concentrated to dryness and redissolved
in CDCl3. Over a period of 16 h, complete retro-aldolization
was observed with apparent concomitant formation of iminium
ion 43.27 This result confirmed the previously observed lack of
reactivity in aprotic solvents, even under acidic conditions. This
highlights the fact that both aldol addition and retro-aldolization in
this system can be dynamic processes. Table 1 highlights the
effectiveness of a variety of protic media in the aldolization with
methanol being the solvent of choice.
Scheme 8. Enamine Aldol Addition Directed toward
Himandridine
a
Computations: Aldol Addition Thermochemistry. As
the factors determining enamine aldol reactivity in these
systems were not apparent, a computational study was
undertaken to elucidate reaction energetics and conformational
preferences of reactants, transition states, plausible intermedi-
ates, and products. Since our experimental data document that
these reactions can be reversible at room temperature, we first
focused on the overall thermodynamics of these imine aldol
additions in comparison with their aldol counterparts (eqs 5−
7).
a
Conditions: (a) TFA, CH2Cl2; (b) 4 Å mol. sieves, PhH; (c) MeOH,
45 min; (d) same pot: HOAc, NaBH3CN, 0 °C; (e) benzyl
chloroformate, Na2CO3, H2O, CH2Cl2; (f) remove MeOH; add
CDCl3, 16 h.
further investigation, it was found that the dissolution of 39 in
methanol was sufficient to induce formation of the desired aldol
adduct as a ca. 2:1 mixture with starting material (Scheme 8,
Table 1). In further contrast to our experience during the
himgaline synthesis, reduction of the resultant iminium ion with
Table 1. Solvent Dependence on Aldol Addition
The energetics of aldolization can be viewed as a nearly
balanced competition between a favorable reaction enthalpy
and the unfavorable entropy inherent to bimolecular reactions
(ca. 12 kcal/mol).28 For example, aldol dimerization of acetone
to provide diacetone alcohol is minimal at equilibrium,29,30
while aldehyde electrophiles are computed to be substantially
more favorable (eq 5).31 Analogous calculations predict the
aldol addition of simple acetone-derived imines to acetone to
be disfavored by 5−7 kcal/mol (eq 6), similar to the
dimerization of acetone. When imine geometry precludes
hydrogen bonding between the product imine and tertiary
hydroxyl, as in the GB alkaloid cases, the reaction is additionally
disfavored (eq 7). Imine aldol additions to ketones are thus
likely to be thermodynamically favorable only in an intra-
molecular setting where the entropic penalty for reaction is
significantly diminished.
a
b
1
Ratio of 41 to 44 estimated from H NMR spectra. Substantial
c
To determine whether the imine aldol reactions encountered
in our synthesis studies are consistent with thermodynamic
incorporation of deuterium was observed. Unidentified byproducts
observed.
8166
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
control, we employed conformational searching and density
functional theory methods to identify the global conformational
minimum for each imine reactant and product (M06-2X/6-31
g(d); see Supporting Information).32,33 SMD continuum
solvent modeling allowed investigation of the role of solvent
effects as noted experimentally.31b
An initial stereochemical issue addressed in this analysis
involves the two possible C5 aldol product diastereomers (A,
B) that may be formed through addition from either enamine
diastereoface (eq 8). It is presumed that these diastereomers
eq 9). By contrast, cyclizations of iminium ions that did not
undergo aldol addition under identical conditions are predicted
to be energetically more costly by ΔΔG = +1.5−5.4 kcal/mol
(eqs 10−12). Additionally, cyclization of imine 39 is predicted
to be 2 kcal/mol more energetically favorable in methanol than
in THF using our standard conditions. This outcome is
consistent with the observed cyclization of this substrate in
MeOH (eq 13) but not THF (eq 12) or other aprotic solvents.
Since the implicit solvation methods used do not account for
H-bonding effects, other parameters, including the smaller
molecular product surface area, are likely responsible for the
altered energetics. In summary, these imine aldol additions may be
run under neutral conditions as long as H-bonding solvents such
methanol are employed.
The computational findings also correlate well with
experimental results when applied to substrates reported by
Movassaghi to undergo analogous enamine aldol additions,34
including those encountered in their total syntheses of the GB
alkaloids GB 137b and himandrine (see Supporting Informa-
tion).9 As their conditions are ambiguous as to whether the
important aldol addition takes place immediately following an
initial cuprate addition in THF or during a subsequent reaction
in ethanol/water, both THF and ethanol were modeled
(implicit solvation). Once again, the reactions are predicted
to be more favorable (by 1.7−2.8 kcal/mol) in ethanol than
THF. Calculated reaction energies in ethanol range from −0.8
kcal/mol to +1.2 kcal/mol, similar to our successful cases and
substantially lower than our unsuccessful cases. The trends in
computed energies of reaction thus correctly predict the
observed reactivities across all cases modeled, with the five
successful reactions shown to be on average 3.9 kcal/mol more
energetically favorable than the three cases in which cyclization
was not observed.
As the origin of the energy differences across our reactions
remained unclear, we next undertook an analysis of the energy-
minimized conformations of each of our reactants and
products. Although the lowest-energy conformers of the
starting materials are nearly identical and feature the expected
chair−chair conformation in the decalin ring system, the low
energy conformers of the bicyclo[3.2.1]octane products
uniformly display a B ring boat conformation (Figure 3).
This conformational preference was also observed for all-
carbon tricyclic ring system 45, in which a B ring boat
conformation was computed (M06-2X/6-31 g(d)) to be
energetically favored over the nearest conformer (an
approximate chair) by 3.7 kcal/mol (Figure 2). Likewise, a
search of the Cambridge Crystallographic Database revealed
two highly analogous ring systems in which the cis-fused
cyclohexane also adopts a boat geometry.35 These examples
include an intermediate accessed during Mander’s synthesis of
might be readily interconvertible through imine-enamine
tautomerism. Both product structures were modeled. In each
case, the naturally occurring C5 isomer A was found to be 3.7−
4.9 kcal/mol more favorable than its epimer B, consistent with
our isolation of only the naturally occurring diastereomer A.
Formation of epimer B substantially alters the bicyclo[3.2.1]-
octane conformation and places the C6 and hydroxyl
substituents in an eclipsing orientation, which may explain
the computed energy difference (eq 8 and Supporting
Information).
The computed reaction energies for these additions correlate
with the experimental data (Scheme 9). Although the reaction
free energies appear to be generally overestimated, successful
reactions are substantially more favorable than those that resist
cyclization. For example, in THF, the aldolization carried out in
the synthesis of himgaline (23−24) is predicted to be
energetically disfavored by ΔG = +1.0 kcal/mol (Scheme 9,
Scheme 9. Computational and Experimental Results
Figure 2. Computed conformations of an alkane ring system.
8167
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
Figure 3. Energy minimized enamine aldol addition products.
GB 13,7 highlighting the ability of our computational approach
to capture known features of the GB ring system. Taken
together, these data suggest the unanticipated preference for
the B ring boat conformer is dictated by the conformational
constraints of the rigid bicyclo[3.2.1]octane system. The boat
conformation also appears to maximize interatomic distances
between hydrogen atoms that are within van der Waals contact
or nearly so (Figure 2).
of promoting the desired cyclization, dissolution in methanol
and stoichiometric acetic acid in THF.
Computations indicate the reaction takes a very different
course in methanol than in tetrahydrofuran. In methanol, the
lowest energy transition state obtained for the formation of 40
(Scheme 7) was zwitterion-like and late, with a forming bond
length of 1.97 Å (Figure 4, 48-TS). The computed activation
This counterintuitive product conformation may provide a
rationale for the experimentally observed patterns of reactivity
based on product stability if the reactions are reversible under
the experimental conditions (as was observed for 39/40). Aldol
adduct 46, which was never obtained, would encounter
unfavorable eclipsing interactions at C15 and C16 as a result
of the favored product boat conformation (Figure 3). By
contrast, aldol adduct 47, which formed readily in HOAc/THF,
instead accommodates the ketone carbonyl in a more favorable
bisected conformation.
Additional oxygenation at C13 and C14, as required for
himandridine, gives rise to aldol adduct 48, which did not form
in HOAc/THF but did proceed to ca. 67% conversion in
methanol. Due to the favored product boat conformation, the
C14 substituent creates a destabilizing syn-pentane-type
interaction with the C16 carbonyl oxygen that likely impedes
the desired cyclization (Figure 3). The energy-minimized
structure of 48 deviates noticeably from an ideal boat geometry,
likely to mitigate the impact of this destabilizing interaction.
Computations: Enamine Aldol Transition States. To
complement the above thermodynamic analysis, we next sought
to identify transition states and plausible intermediates en route
to the observed cyclized enamine aldol products. Although
transition states for enamine aldol additions, including an
intramolecular cyclization, have been studied,36 the GB alkaloid
case differs from previously analyzed systems. Specifically, the
geometric constraints of the forming bicyclo[3.2.1]octane ring
system preclude the closed transition states known to be
favorable in less restricted imine aldol reactions. Therefore, we
modeled the geometrically feasible open transition state
orientations under both sets of conditions shown to be capable
Figure 4. Computed transition states.
energy, 21.7 kcal/mol, together with our computed free energy
of reaction (+0.5 kcal/mol) is consistent with our experimental
observation of a facile but incomplete reaction at room
temperature.37 Additionally, as previously observed,36
a
zwitterion intermediate was also located at an energy of
+17.1 kcal/mol. Both the lowest-energy transition state and
zwitterion intermediate displayed a B-ring boat conformation
(Figure 4), indicating the importance of this conformational
change during bond formation as well as in the ground state
(Figure 3). Transition states possessing a half-chair B-ring
conformation were higher in energy by at least 0.9 kcal/mol.37
By contrast, in THF, neither an analogous transition state nor a
zwitterion intermediate could be located. Structures with the
newly formed aldol bond length constrained to 1.67 Å, a typical
distance for zwitterion intermediates in methanol, were severely
8168
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
Notes
disfavored relative to the methanol case (ΔΔE ca. 19 kcal/
mol). The role of methanol (but not nonpolar solvents) in
promoting the spontaneous formation of 40 thus appears two-
fold: in addition to lowering the free energy of reaction relative
to nonpolar solvents, methanol also provides substantial
stabilization to zwitterion-like transition states, enabling bond
formation to occur at room temperature.
We next examined transition states in THF with an explicit
acetic acid molecule, mirroring conditions used in the
successful cyclization to provide 47. Multiple orientations of
the acetic acid molecule were examined, with two modes giving
rise to transition states of reasonable energy (see Supporting
Information). In the first (ΔG‡ = +22.2 kcal/mol), acetic acid
serves as a hydrogen bond donor to the cyclopentanone oxygen
during bond formation, providing superior delocalization of the
forming anion relative to THF alone. In the second and lowest-
energy transition state (ΔG‡ = +20.3 kcal/mol), acetic acid
both donates a hydrogen bond to the cyclopentanone oxygen
and accepts a hydrogen bond from the vinylic proton on the
reacting enamine carbon (Figure 4, 47-TS).
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Prof. Lewis N. Mander for generous donations of
naturally isolated GB 13 and himgaline. This work was
supported by the National Institutes of Health (GM-33328-
20) and Amgen.
REFERENCES
■
(1) (a) Ritchie, E.; Taylor, W. C. In The Alkaloids; Manske, R. H. F.,
Ed.; Academic Press: New York, 1967; Vol. 9, Chapter 14, p 529.
(b) Collins, D. J.; Culvenor, C. C. J.; Lamberton, J. A.; Loder, J. W.;
Price, J. R. Plants for Medicines. A Chemical and Pharmacological Survey
of Plants in the Australian Region; CSIRO: Melbourne, 1990; Chapter
1. (c) Pinhey, J. T.; Ritchie, E.; Taylor, W. C. J. Aust. Chem. 1961, 14,
106−126. (d) Binns, S. V.; Dunstan, P. J.; Guise, G. B.; Holder, G. M.;
Hollis, A. F.; McCredie, R. S.; Pinhey, J. T.; Prager, R. H.; Rasmussen,
M.; Ritchie, E.; Taylor, W. C. J. Aust. Chem. 1965, 18, 569−573.
(e) Mander, L. N.; Ritchie, E.; Taylor, W. C. J. Aust. Chem. 1967, 20,
981−1019. (f) Mander, L. N.; Ritchie, E.; Taylor, W. C. J. Aust. Chem.
1967, 20, 1021−1027. (g) Guise, G. B.; Mander, L. N.; Prager, R. H.;
Rasmussen, M.; Ritchie, E.; Taylor, W. C. J. Aust. Chem. 1967, 20,
1029−1035. (h) Mander, L. N.; Prager, R. H.; Rasmussen, M.; Ritchie,
E.; Taylor, W. C. J. Aust. Chem. 1967, 20, 1473−1491. (i) Mander, L.
N.; Prager, R. H.; Rasmussen, M.; Ritchie, E.; Taylor, W. C. J. Aust.
Chem. 1967, 20, 1705−1718.
CONCLUSIONS
■
The proposed strategy of sequential intramolecular Diels−
Alder, Michael, and imine aldol cyclizations to construct the
complex architectures of the GB alkaloids has delivered total
syntheses of GB 13 and himgaline as well as an advanced
intermediate directed toward the type II alkaloid himandridine.
The pivotal imine aldol addition common to these syntheses
was found to be highly sensitive to subtle changes in both
substrate structure and reaction conditions. Although kinetic
control in these cyclizations cannot be rigorously excluded, the
observation of reversibility in one case together with computa-
tional predictions of near-thermoneutrality and accessible
activation energies lead us to favor a model in which these
reactions are under thermodynamic control. Calculated free
energies of reaction correlate well with observed reactivity, and
energy-minimized structures predict an unexpected boat
cyclohexane conformation previously observed by X-ray
crystallographic analysis of a closely related structure. Further
conformational analysis suggests a rationale for the observed
reactivity in which the aldol products of unsuccessful substrates,
but not successful substrates, face destabilizing steric inter-
actions as a direct result of the strong bias for a B-ring boat
conformation. At present, two strategies to facilitate the
production of the desired aldol product have been demon-
strated: incorporation of sp2 hybridization at C16 and the use
of a protic solvent. Interestingly, both manipulations likely
bring our system closer to the environment of the biosynthesis
of these alkaloids. These studies highlight the unexpected
complexities of natural products synthesis and the power of
computational methods to provide insight into challenging
reactivities.
(2) Thomas, B. Eleusis: J. Psychoact. Plants Compd. 1999, 3, 82−90.
(3) (a) Kozikowski, A. P.; Fauq, A. H.; Miller, J. H.; McKinney, M.
Bioorg. Med. Chem. Lett. 1992, 2, 797−802. (b) Malaska, M. J.; Fauq,
A. H.; Kozikowski, A. P.; Aagaard, P. J.; McKinney, M. Bioorg. Med.
Chem. Lett. 1995, 5, 61−66.
(4) (a) Clasby, M. C.; Chackalamannil, S.; Czarniecki, M.; Doller, D.;
Eagen, K.; Greenlee, W. J.; Lin, Y.; Tsai, H.; Xia, Y.; Ahn, H.-S.; Agans-
Fantuzzi, J.; Boykow, G.; Chintala, M.; Foster, C.; Bryant, M.; Lau, J.
Bioorg. Med. Chem. Lett. 2006, 16, 1544−1548. (b) Chackalamannil, S.;
Xia, Y. Expert Opin. Ther. Pat. 2006, 16, 493−505. (c) Chackalamannil,
S.; Wang, Y.; Greenlee, W. J.; Hu, Z.; Xia, Y.; Ahn, H.-S.; Boykow, G.;
Hsieh, Y.; Palomanda, J.; Agans-Fantuzzi, J.; Kurowski, S.; Graziano,
M.; Chintala, M. J. Med. Chem. 2008, 51, 3061−3064.
(5) (a) Hart, D. J.; Wu, W.-L.; Kozikowski, A. P. J. Am. Chem. Soc.
1995, 117, 9369−9370. (b) Chackalamannil, S.; Davies, R. J.; Aserom,
T.; Doller, D.; Leone, D. J. Am. Chem. Soc. 1996, 118, 9212−9213.
(c) Takadoi, M.; Katoh, T.; Ishiwata, A.; Terashima, S. Tetrahedron
Lett. 1999, 40, 3399−3402. (d) Tchabanenko, K.; Adlington, R. M.;
Cowley, A. R.; Baldwin, J. E. Org. Lett. 2005, 7, 585−588.
(e) Tchabanenko, K.; Chesworth, R.; Parker, J. S.; Anand, N. K.;
Russell, A. T.; Adlington, R. M.; Baldwin, J. E. Tetrahedron 2005, 61,
11649−11656.
(6) (a) Chackalamannil, S.; Doller, D.; McQuade, R.; Ruperto, V.
Bioorg. Med. Chem. Lett. 2004, 14, 3967−3970. (b) Doller, D.;
Chackalamannil, S.; Czarniecki, M.; McQuade, R.; Ruperto, V. Bioorg.
Med. Chem. Lett. 1999, 9, 901−906. (c) Takadoi, M.; Katoh, T.;
Ishiwata, A.; Terashima, S. Tetrahedron 2002, 58, 9903−9923.
(7) (a) Mander, L. N.; McLachlan, M. M. J. Am. Chem. Soc. 2003,
125, 2400−2401. (f) McLachlan, M. M. Ph.D. thesis, Research School
of Chemistry, Australian National University, 2002. (b) Movassaghi,
M.; Hunt, D. K.; Tjandra, M. J. Am. Chem. Soc. 2006, 128, 8126−8127.
(c) Shah, U.; Chackalamannil, S.; Ganguly, A. K.; Chelliah, M.;
Kolutuchi, S.; Beuvich, A.; McPhail, A. J. Am. Chem. Soc. 2006, 128,
12654−12655. (d) Larson, K. K.; Sarpong, R. J. Am. Chem. Soc. 2009,
131, 13244−13245. (e) Zi, W.; Yu, S.; Ma, D. Angew. Chem., Int. Ed.
2010, 49, 5887−5890.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental procedures, compound characterization data, and
computational details. This information is available free of
charge via the Internet at http://pubs.acs.org.
(8) Willis, A. C.; O’Connor, P. D.; Taylor, W. C.; Mander, L. N. Aust.
J. Chem. 2006, 59, 629−632. Movassaghi’s synthesis of GB 13,
coupled with the conversion of himandrine to GB 13 during isolation
studies, provided additional evidence for revision of the Type II
alkaloids.
AUTHOR INFORMATION
■
Corresponding Author
evans@chemistry.harvard.edu
8169
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170

Journal of the American Chemical Society
Article
(9) Movassaghi, M.; Tjandra, M.; Qi, J. J. Am. Chem. Soc. 2009, 131,
9648−9650. For a synthesis of the himandrine skeleton related to the
approach to GB 13 described in ref 7a, see: O’Connor, P. D.; Mander,
L. N.; McLachlan, M. M. W. Org. Lett. 2004, 6, 703−706.
(10) For an initial communication of this work, see: Evans, D. A.;
Adams, D. J. J. Am. Chem. Soc. 2007, 129, 1048−1049.
(11) Our view is in general agreement with that presented in ref 7b;
for a biomimetic approach to himbacine as well as an alternative view
on GB alkaloid biosynthesis, see ref 5e.
Information for details. M06-2X has been shown to outperform the
commonly used B3LYP basis set for branched alkanes and for π to σ
bond transitions, in part due to a fortuitous cancelation of errors:
(a) Song, J.-W.; Tsuneda, T.; Sato, T.; Hirao, K. Org. Lett. 2010, 12,
1440−1443. (b) Pieniazek, S. N.; Clemente, F. R.; Houk, K. N. Angew.
Chem., Int. Ed. 2008, 47, 7746−7749. (c) Wheeler, S. E.; Moran, A.;
Pieniazek, S. N.; Houk, K. N. J. Phys. Chem. A 2009, 113, 10376.
(33) Tautomeric enamines and enammonium ions were found to be
6−13 kcal/mol higher in energy and were not considered further.
(34) Movassaghi, M.; Chen, B. Angew. Chem., Int. Ed. 2006, 45, 565−
568.
(35) Howells, D. M.; Barker, S. M.; Watson, F. C.; Light, M. E.;
Hursthouse, M. B.; Kilburn, J. D. Org. Lett. 2004, 6, 1943. See
Supporting Information for comparison with computed structures.
(36) Bahmanyar, S.; Houk, K. N. J. Am. Chem. Soc. 2001, 123,
11273−11283.
(37) Modeling of one explicit methanol molecule in a variety of
orientations provided transition states of comparable energy (see
Supporting Information).
(12) Claus, R. E.; Schreiber, S. L. Organic Syntheses; Wiley & Sons:
New York, 1990; Collect. Vol. 7, pp 168−172.
(13) For preparation of 3 and HWE procedure, see: Evans, D. A.;
Miller, S. J.; Lectka, T.; von Matt, P. J. Am. Chem. Soc. 1999, 121,
7559−7573.
(14) Evans, D. A.; Chapman, K. T.; Bisaha, J. J. Am. Chem. Soc. 1988,
110, 1238−1256.
(15) Although N-benzylation is unnecessary for the completion of
the synthesis, yields were optimized with N-benzylated material.
(16) Dess, D. B; Martin, J. C. J. Am. Chem. Soc. 1983, 48, 4155−4156.
(17) Holmquist, C. R.; Roskamp, E. J. J. Org. Chem. 1989, 54, 3258−
3260.
(18) Raban, M.; Noe, E. A.; Yamamoto, G. J. Am. Chem. Soc. 1977,
99, 6527−6731.
(19) For similar ring systems formed using aldol additions, see:
Heathcock, C. H.; von Geldern, T. W. Heterocycles 1987, 25, 75−78.
Alexakis, A.; Chapdlaine, M. J.; Posner, G. H. Tetrahdron Lett. 1978,
44, 4209−4212. Lohray, B. B.; Zimbiniski, R. Tetrahdron Lett. 1990,
50, 7273−7276.
(20) Although kinetic control was desired in many cases, the
thermodynamic acidity of cyclopentanone is notably greater than ethyl
ketone (pKa 25.8 vs 27.1, DMSO). Bordwell, F. G.; Fried, H. E. J. Org.
Chem. 1991, 56, 4218−4223.
(21) Nicolau, K. C.; Montagnon, T.; Baran, P. S. J. Am. Chem. Soc.
2002, 124, 2245−2258.
(22) Chackalamannil performed the conjugate addition with
scandium triflate and hydrochloric acid prior to directed hydride
reduction (ref 7c).
(23) Evans, D. A.; Gage, J. R.; Leighton, J. L.; Kim, A. S. J. Org. Chem.
1992, 57, 1961−1963.
(24) Hoffmann, R. W. Chem. Rev. 1989, 89, 1841−1860.
(25) These conditions were used previously to effect nucleophilic
displacement of allylic halides. See ref 7c and: Boyd, D. R.; Sharma, N.
D.; Kerley, N. A.; McConville, G.; Allen, C. C. R.; Blacker, A. J.
ARKIVOC 2003, vii, 32−48.
(26) Polycycle 34 differs from an advanced intermediate in
Movassaghi’s himandrine synthesis (ref 9) only by addition of
OTBS at C13.
(27) Protonation likely occurs via trace acid present in d-chloroform.
(28) Page, M. I.; Jencks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1971, 68,
1678−1683.
(29) Maple, S. R.; Allerhand, A. J. Am. Chem. Soc. 1987, 109, 6609−
6614.
(30) An analogous intramolecular aldol addition, cyclization of 2,6-
heptanedione to 3-hydroxy-3-methylcyclohexanone, is substantially
exergonic (ΔG = −2.4 kcal/mol), suggesting that intramolecularity can
increase the favorability of an aldol process by 5−6 kcal/mol. Guthrie,
J. P.; Guo, J. J. Am. Chem. Soc. 1996, 118, 11472−11487.
(31) Energies given for eqs 5−7 were computed using the CBS-QB3
method with SMD solvation (THF). (a) CBS-QB3: Montgomery, J.
A., Jr.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys.
1999, 110, 2822−2827. (b) SMD solvation: Marenich, A. V.; Cramer,
C. J.; Truhlar, D. G. J. Phys. Chem. B. 2009, 113, 6378−6396.
(32) Candidate structures were generated using molecular mechanics
and optimized further at M06-2X/6-31g with SMD implicit solvation.
All energies are given as single-point electronic energies at M06-2X/6-
311+g(d,p) with M06-2X/6-31g(d) geometries and free energy
corrections (298 K). In model systems whose energies could be
accurately estimated by CBS-QB3, M06-2X was shown to perform the
best of a variety of density functional methods. See Supporting
8170
dx.doi.org/10.1021/ja3001776 | J. Am. Chem. Soc. 2012, 134, 8162−8170