The full details of a synthesis of the hetidine framework of the C20-diterpenoid alkaloids and its conversion to the atisine core structure are reported. The application of the hetidine framework to the synthesis of dihydronavirine, which is the formal reduction product of the natural product navirine, is also described. Key to the success of these studies is the use of a Ga(III)-catalyzed cycloisomerization reaction of alkynylindenes to prepare a [6-7-6] framework that was advanced to the hetidine skeleton. Furthermore, a Michael/aldol sequence was developed for the construction of the bicyclo[2.2.2] framework that is characteristic of the hetidines and atisines.
The full details of a synthesis of the hetidine framework of the C20-diterpenoid alkaloids and its conversion to the atisine core structure are reported. The application of the hetidine framework to the synthesis of dihydronavirine, which is the formal reduction product of the natural product navirine, is also described. Key to the success of these studies is the use of a Ga(III)-catalyzed cycloisomerization reaction of alkynylindenes to prepare a [6-7-6] framework that was advanced to the hetidine skeleton. Furthermore, a Michael/aldol sequence was developed for the construction of the bicyclo[2.2.2] framework that is characteristic of the hetidines and atisines.
The diterpenoid alkaloids
(see representative examples in Figure 1) are
a collection of complex alkaloids isolated
from the Ranunculaceae and Rosaceae family of plants.[1] These molecules are among the most architecturally complex
natural products isolated to date. In addition to their imposing skeletal
intricacy, they possess varied, yet dense, arrays of hydroxylation.
To date, over 1100 natural products have been isolated and classified
as diterpenoid alkaloids. On the basis of the number of constituent
carbons, these natural products are described as C18-,
C19-, or C20-diterpenoid alkaloids. The majority
of these metabolites have been isolated from the Aconitum, Consolida, Delphinium, Rumex, and Spiraea genera, and extracts
from these plants have been used in traditional Eastern herbal medicine
for centuries as sedatives and fever reducers. Other pharmacological
effects that have been discovered in these plants, including analgesic
effects and anti-inflammatory, myorelaxant, and antiarrhythmic properties,
are also attributed to their diterpenoid alkaloid constituents. Overall,
this broad spectrum of biological activity likely arises because of
the potent interactions of these alkaloids with voltage-dependent
sodium, potassium, and calcium ion channels. The pharmaceutical potential
of the diterpenoid alkaloids is continuing to emerge, and already,
lappaconitine (allapinin, 1) and guan-fu base A (13, Figure 2; also known as acehytisine
hydrochloride)[2] are marketed as antiarrhythmia
agents.[3][4]
Figure 1
Representative
C18-, C19-, and C20-diterpenoid alkaloids.
Figure 2
Representative C20-diterpenoid alkaloids and
derivatives
in the atisine, hetidine, and hetisine subclass.
Representative
C18-, C19-, and C20-diterpenoid alkaloids.The classification of the C20-diterpenoid alkaloids
by Wang and Liang on the basis of their carbon skeletons into the
atisane, kaurane, rearranged, and bisditerpenoid classes is the most
useful classification from a synthetic strategy perspective.[5] In many ways, this delineation unveils a general
pattern recognition[6] that should facilitate
the synthesis of not only a single member of this family of natural
products but many members from a common intermediate. Our synthetic
studies on these alkaloids began with an interest in the atisane class,
specifically the atisine, hetidine, and hetisine-type skeletons. In
this paper, we report the details of a strategy that led to the synthesis
of an intermediate that provides access to the atisine and hetidine
skeletons (A and B, respectively, Figure 2) and has culminated in a synthesis of dihydronavirine
(8).Representative C20-diterpenoid alkaloids and
derivatives
in the atisine, hetidine, and hetisine subclass.
Background
Despite the isolation of a wide range of diterpenoidalkaloids
over the last half-century across the various skeletal subclasses,
there remains a relatively small number of total syntheses of these
natural products. Early important contributions to the syntheses of
the diterpenoid alkaloids (including a synthesis of talatisamine, 3, and atisine, 9) were made by Wiesner and co-workers,[7] and recently, syntheses of nominine (12, by the groups of Gin[8] and Muratake and
Natsume[9]), neofinaconitine (2, by Gin and co-workers),[10] and lepenine
(4, by Fukuyama, Yokoshima, and co-workers)[11] have appeared.[12]In our approach to the C20-diterpenoid alkaloids, we
envisioned the hetidine framework (see B, Figure 2) as a versatile carbon skeleton that could be converted
at a late stage to the hetisine (i.e., C) and atisine
(i.e., A) skeletons. Key to these conversions would be
a late-stage C6–N bond formation (B → C; by employing a Hoffman–Löffler–Freytag-type
reaction)[13] or a C14–C20 bond cleavage
(B → A; using an oxidative Grob-like
fragmentation)[14] to convert the hetidine
skeleton to the hetisine and atisine skeletons, respectively. Importantly,
Okamoto et al. have already demonstrated the feasibility of converting
the hetidine core to a mixture of hetisine and atisine frameworks
from the corresponding N-chloroamine,[15] and so our studies have focused on the synthesis
of the hetidine framework and its exclusive conversion to the atisine
skeleton.
Results and Discussion
Previously, we reported the
successful synthesis of 14 (Figure 3), which possesses key hydroxy groups
at C11 and C14 that are critical for the synthesis of alkaloids such
as spiramine H (10), kobusine (11), navirine
(5), and guan-fu base A (13).[16] In particular, we viewed the hydroxyl at C14
of 14 as an important functional handle that would facilitate
the desired Grob-like fragmentation to arrive at the atisine skeleton
and natural products such as 9 and 10. Herein,
we report the full details of our studies using 14 as
a key late-stage synthetic intermediate, which have culminated in
the preparation of dihydronavirine (8) as well as our
realization of the conversion of the hetidine framework to the atisine
skeleton by employing a Grob-like fragmentation that is achieved under
oxidation conditions. Our efforts to prepare 15, which
represents the simplest synthetic rendition of the hetidine core,
is also discussed.
Figure 3
Potential synthetic equivalents to the hetidine core.
Potential synthetic equivalents to the hetidine core.
Initial Synthetic Studies on the Hetidine
Core: Attempts To
Prepare 15
From our synthetic analysis of the
C20–diterpenoid alkaloids in the atisine, hetidine
and hetisine skeletal subclasses, two key challenges have emerged.
These are the installation of the [2.2.2] bicycle and the position-selective
introduction of oxygenation, which for many of these molecules is
required at sites distal from other functional groups. Our initial
forays into the synthesis of the hetidine core sought to address the
first of these challenges by preparing 15. From our perspective,
the installation of the [2.2.2] bicycle has been most elegantly addressed
by Gin and Peese, who employed an intramolecular Diels–Alder
(IMDA) reaction to forge the [2.2.2] bicycle in their synthesis of
nominine.[8] Inspired by this powerful complexity
building transformation, we envisioned the hetidine core (see 15, Scheme 1) arising from methoxyarene 16, where in the forward sense, the requisite diene for the
IMDA reaction would be fashioned from the arene using a Birch reduction
as previously demonstrated by Gin and Peese. Piperidine polycycle 16 could arise from amino alcohol 17 through
a C20–N bond formation. In turn, tricycle 17 was
envisioned from benzocycloheptadiene 18, which would
be prepared from alkynylindene 19 using a Ga(III)-catalyzed
cycloisomerization reaction developed in our laboratories.[17]
Scheme 1
Retrosynthetic Analysis of Hetidine Skeleton
Equivalent 15
Our planned synthesis of hexacycle 15 commenced
with
the preparation of alkynylindene 19 (Scheme 2) as previously described.[16] Alkylation of commercially available β-ketoester 20(18) with alkyl iodide 21 (accessed
in racemic form from 3-bromo-1-propanol in nine steps)[19] proceeds in 78% yield to provide the alkylated
β-ketoester adduct (22). Of note, we have also
demonstrated the enantioselective synthesis of alkyl iodide 21 (in 89% ee) using a method developed by Sawamura and Ito.[20] As such, the foundation for an enantioselective
synthesis has been set. However, for the purposes of these studies,
the racemic material was advanced. Saponification and decarboxylation
of 22 using LiOH provides indanone 23 in
98% yield. Chemoselective reduction of the indanone carbonyl group
is achieved in the presence of the nitrile group using NaBH4 as the reducing agent to provide an indanol (not shown; as a mixture
of diastereomers) that was dehydrated using pyridinium p-toluenesulfonate (PPTS) to afford alkynylindene 19.
Scheme 2
Synthesis of Alkynylindene 19
Following a comprehensive optimization study, it was found
that
subjecting rigorously dried alkynylindene 19 to a substoichiometric
amount of GaI3 (25 mol %) in the presence of 4 Å powdered
molecular sieves in toluene at 100 °C for 48 h gives benzannulated
cycloheptadiene 18 in 89% yield (Scheme 3). The temperature and reaction time for the formation of 18 are higher and longer, respectively, compared to conditions
employed previously for the cycloisomerization of other alkynylindene
substrates. The higher temperature and longer reaction time for the
conversion of 19 to 18 is attributed to
several factors including the presence of the cyano group, which likely
interacts with GaI3 and mitigates its Lewis acidity. In
addition, the cyano group, by virtue of its inductive electron-withdrawing
effect, likely decreases the inherent nucleophilicity of the alkyne
group. Finally, in previous alkynylindene cycloisomerization reactions
that we have conducted, highly electron-rich indenes were employed.
As such, it seems reasonable that the less electron-rich indene moiety
in 19 would require more forcing conditions for the desired
transformation. Consistent with this line of reasoning is the observation
that the use of other Lewis acid catalysts such as GaCl3 that had previously been efficient for the syntheses of 24–26, only resulted in the incomplete conversion
of 19 to 18 and low isolated yields of the
desired cycloheptadiene.
Scheme 3
Cycloisomerization of Indenylalkynes
With benzannulated cycloheptadiene 18 in hand, selective
reduction of the disubstituted double bond proceeds in good yield
using diimide (Scheme 4), which is generated
in situ from p-toluenesulfonyl hydrazide and triethylamine.[21] Other conditions for diimide generation (e.g.,
hydrazine, CuSO4·H2O, EtOH)[22] were less effective and led only to recovered
starting material. Next, oxidation to install a carbonyl group at
C20 (hetidine numbering) of 27 was pursued. This oxygenation
proved to be one of the most capricious steps of our synthesis campaign.
An extensive investigation of various oxidants including DDQ, CrO2, CrO3, Ba(MnO4)2, KMnO4, and MnO2 resulted in little to no oxidation,
overoxidation, or nonspecific decomposition. Ultimately, ceric ammonium
nitrate (CAN) adsorbed on silica gel proved to be the most effective
oxidant and provided enone 28 in a workable 52% yield
(0.75 mmol scale). Diastereoselective reduction of the tetrasubstituted
enone double bond is achieved using H2 and Pd/C to provide
a single diastereomer of ketone 29. The structure and
relative stereochemistry of 29 was unambiguously confirmed
by an X-ray crystallographic study of a single crystal (see ORTEP
representation in Scheme 4). A possible rationale
for the excellent diastereoselectivity that is observed in the hydrogenation
of 28 is that hydrogen addition occurs from the face
bearing the methyl group (A = 1.7),[23] which adopts a pseudoequatorial position in 28, whereas the cyano group (A = 0.17) is axially
disposed. As a result, approach from the face opposite the cyano group
is more likely. While this “contra-steric” rationale
is appealing,[24] an electronic argument
may also be advanced. Thus, Woodward has proposed that the facial
selectivity of a reacting olefin group may be determined by the extent
to which that face is electron-depleted.[25] In our case, this would suggest that the face bearing the cyano
group is electron-depleted (by virtue of an interaction between the
π orbital of the tetrasubstituted double bond and the lower-lying
π* orbital of the axially disposed cyano group), and as such,
hydrogenation occurs from the face bearing the methyl substituent.[26]
Scheme 4
Functionalization of Cyloheptadiene 18 to Tricycle 29
With ketone 29 in hand, the next task was
to install
a vinyl group α to the ketone group (Scheme 5). Deprotonation α to the carbonyl group in 29 is readily effected with sodium hydride and if conducted in the
presence of phenyl vinylsulfoxide gives sulfoxide 30 in
87% yield. Sulfoxide extrusion at this stage installs the vinyl group
in 31, which was obtained as a single diastereomer as
confirmed by X-ray analysis of a single crystal of 31 (see ORTEP in Scheme 5). Surprisingly, every
attempt to reduce both the nitrile and carbonyl groups of 31 (in preparation for the formation of a piperidine ring; see 17 to 16, Scheme 1) with
LiAlH4 was frustrated by the formation of a complex mixture,
which included the expected amino alcohol (i.e., 17,
as a mixture of diastereomers) along with unidentifiable side products.
Furthermore, attempts to form the piperidine ring by activation of
the alcohol group of 17 (with SOCl2, oxalyl
chloride, or MsCl) led to a complex mixture where it appeared the
vinyl group was engaged by presumed carbocation intermediates formed
by the activation of the C20 hydroxy group.
Scheme 5
Installation of the
Vinyl Group
We hypothesized that
an allyl group in place of the vinyl group
in 31 would mitigate the unwanted reactivity of the double
bond of the angular substituent upon activation of the hydroxy group.
If this was the case, then the allyl group could be converted to the
vinyl group at a later stage following the formation of the piperidine
ring. As shown in Scheme 6, with an allyl group
in place of the vinyl group, piperidine ring formation is indeed successful.
Thus, generation of the sodium enolate of 29 (with sodium
hydride) in the presence of allyl bromide yielded a mixture of the
C-allylated (32) and O-allylated (not shown) adducts.
The mixture is easily converged to 32 upon heating (μW,
160 °C) through a Claisen rearrangement. At this stage, reduction
of the carbonyl and cyano groups of 32 (with LiAlH4) followed by Boc-protection of the resulting primary amine
provides 33 as a single diastereomer (the stereochemistry
of the C20 center was not determined). Exposure of Boc-protected amino
alcohol 33 to thionyl chloride effects clean cyclization
to forge the piperidine ring to provide 34 in 72% yield.
In contrast, treatment of the free amino alcohol (generated by LiAlH4 reduction of 32) with thionyl chloride led only
to decomposition. The structure of 34 was unambiguously
confirmed by X-ray analysis of the p-nitrobenzoyl
derivative 35.
Scheme 6
Installation of the Allyl Group and
Piperidine Ring formation
At this stage, we next planned the conversion of the angular
allyl
group to a vinyl group, which was accomplished in six steps as illustrated
in Scheme 7. The sequence commenced with an
UpJohn dihydroxylation[27] of 34 followed by periodate cleavage of the resulting diol to afford aldehyde 36. Reduction of the aldehyde group using sodium borohydride
gives alcohol 37. A modified Mitsunobu displacement of
the hydroxyl group of 37 with a phenyl sulfide group
is readily accomplished with tributylphosphine and diphenyl sulfide.
Selective oxidation of the phenyl sulfide group to the corresponding
sulfoxide (38) is effected with H2O2 and Sc(OTf)3.[28] Sulfoxide
extrusion was then accomplished by heating (μW, 180 °C)
to afford 39 in 80% yield.
Scheme 7
Conversion of the
Allyl Substituent to the Vinyl Group
With access to methoxyarene 39, the stage
was set
to attempt the installation of the [2.2.2] bicycle using a sequence
analogous to that employed by Gin and Peese in their synthesis of
nominine.[8] Thus, subjection of arene 39 to dissolving metal reduction conditions (Na, i-PrOH, liquid NH3) followed by hydrolysis of the resulting
methyl enol ether under acidic conditions (with citric or oxalic acid)
gives a mixture of enones (40 and 41) that
proved to be unstable and could not be easily advanced. This is in
marked contrast to the observations of Gin and Peese on the Birch
reduction of 42 to give 43 in high yield
as illustrated in Scheme 8B.[8]
Scheme 8
Reductive Dearomatization Studies
The lengthy sequence required for the installation
of the vinyl
group in 39 from the corresponding allyl compound (34; see Scheme 7) as well as the instability
of the Birch reduction products (40 and 41) caused us to consider an alternative strategy for the preparation
of the hetidine core, especially with regard to installing the [2.2.2]
bicycle.
Revised Approach to the Hetidine Framework
In a revised
approach to the hetidine framework, we sought to address both challenges
inherent in the preparation of the hetidine core. That is, the installation
of the [2.2.2] bicycle (which had been the focal point of our first
approach) as well as the position selective introduction of hydroxylation
to address targets including navirine (5), guan-fu base
A (13), and spiramine H (10). In our new
approach, an alternative synthetic equivalent to the hetidine framework
(14, Scheme 9) was identified.
We envisioned 14 arising from enone aldehyde 44, where in the forward sense the [2.2.2] bicycle would be forged
using a Michael/aldol sequence. In turn, 44 would be
formed from allyl methoxyarene 34 by employing a key
oxidative dearomatization reaction. This strategy would address several
shortcomings of our initial approach. First, the laborious conversion
of the allyl group to a vinyl group (Scheme 7) would be avoided. Second, the programmed installation of hydroxylation
at C11 and C14 (see 14) would be achieved from the allyl
methoxyarene precursor (34) using the oxidative dearomatization
and the aldol synthetic sequences.
Scheme 9
Revised Retrosynthesis to the Hetidine
Framework
The synthesis of 14 (Scheme 10) began with the demethylation
of methoxyarene 34 using
sodium ethanethiolate (μW, 180 °C) to provide phenol 45 in 85% yield. Subjecting phenol 45 to oxidative
dearomatization conditions[29] with phenyliodone
diacetate (PIDA) in a mixture of water and acetonitrile (1:1) yielded
cyclic carbamate 47 as the sole product in 15% yield.
The oxazolidinone moiety in 47 is likely formed by nucleophilic
attack of the Boc carbonyl on the activated phenol (see 46). Although low yielding, the defined trajectory of approach for
the Boc group as a nucleophile controls the newly introduced stereocenter
at C14 in 47. In efforts to optimize this transformation,
the more activated hypervalent iodine reagent, phenyliodine bis(trifluoroacetate)
(PIFA; 1 equiv) was employed in the oxidative dearomatization along
with a buffer (NaHCO3). In addition, trifluoroethanol was
used in place of water as the cosolvent. These modifications increased
the yield of 47 to 54%. Subsequently, we have also found
that iodosobenzene (2 equiv) may be used in place of PIFA to obtain 47 in similar yields.
Scheme 10
Oxidative Dearomatization of Methoxyarene 34 To Obtain 47
With cyclohexanedienone 47 in hand, a dihydroxylation
of the allyl group followed by periodate cleavage leads to aldehyde 48 (Scheme 11). Simply stirring this
aldehyde with silica gel in methylene chloride promotes a Michael
addition to yield 49. A selective reduction of the enone
double bond at this stage affords ketoaldehyde 50, which
upon exposure to mildly basic conditions (K2CO3 in methanol) affords synthetic hetidine framework equivalent 14. The structure of 14 was unambiguously confirmed
by an X-ray crystallographic study (see ORTEP, Scheme 11).
Scheme 11
Synthesis of Hetidine Core Structure 14
Synthesis of Dihydronavirine
In an initial demonstration
of the utility of 14 in the synthesis of diterpenoidalkaloids, we chose to focus on the synthesis of navirine (5) and the related derivative dihydronavirine (8). Navirine
was isolated in 2004 from Aconitum naviculare, a
perennial herb found in the Himalayas of Nepal and Tibet.[30] Navirine is unusual among diterpenoid alkaloids
in that it is a conjugate of the diterpenoid alkaloid core and hordenine
(56, Scheme 13), a phenolic natural
product, which is also coisolated with 5. In 2008, congeners
of navirine that vary in their oxidation level (navirine B, 6, and navirine C, 7)[31] were isolated from the same plant source and would seem to suggest
that like many diterpenoid alkaloid families, additional navirines
of other oxidation levels (e.g., dihydronavirine, 8)
may exist and are yet to be isolated. For our immediate plans, the
hydroxy group at C11 of 14, although useful in the context
of an eventual synthesis of kobusine (11) and guan-fu
base A (13), would have to be removed in order to achieve
a synthesis of dihydronavirine (8) and navirine (5).
Scheme 13
Synthesis
of Dihydronavirine (8)
Our planned synthesis of 5 and 8 commenced with the deoxygenation of 14 at C11,
which was accomplished using the Barton–McCombie protocol[32] to give 51 (Scheme 12) in 65% yield over the two steps. Deprotonation of ketone 51 with LiHMDS followed by the addition of Comins’
reagent (52)[33] to the resulting
enolate gives vinyl triflate 53. At this point, the C17
methylene group of the diterpenoid alkaloid core architecture was
installed using a Stille cross-coupling reaction with tributylstannyl
methanol (54) to yield 55.
Scheme 12
Preparation
of Advanced Hetidine Framework 55
To arrive at the complete framework of dihydronavirine
(8) and navirine (5), the hordenine moiety
had to be appended.
This was accomplished by first activating the C17 allylic alcohol
group of 55 (Scheme 13) toward nucleophilic displacement by converting
it to the corresponding mesylate (not shown). Treatment of this compound
with the sodium salt of hordenine (56) yields the desired
adduct (57). Notably, an attempted Mitsunobu reaction
to link 55 to 56 directly did not result
in any conversion and only returned the starting material. With 57 in hand, hydrolytic cleavage of the cyclic carbamate group
(using barium hydroxide; μW, 160 °C)[34] provided dihydronavirine (8).A conversion of dihydronavirine (8) to navirine
(5) would require a dehydrogenation of the secondary
amine
to the corresponding imine. This type of oxidation has been studied
previously in many contexts, and various oxidants including MnO2 and CrO3 have been employed for this purpose.[35] In a closely related precedent in the diterpenoid
alkaloid series, Benn and co-workers showed during their partial synthesis
studies in the delphinine series that the piperidine moiety of des-N-methyllycaconitine (58, Scheme 14A) is converted to the corresponding imine (anhweidelphinine, 59) by the action of MnO2.[36] However, in our hands, attempts using various oxidants to convert
dihydronavirine (8) to navirine (5) resulted
in nonspecific decomposition in many cases, whereas for a few of the
oxidants (e.g., N-iodosuccinimide and MnO2) trace amounts of the C14–C20 bond cleavage product (60) were isolated. Given that 60 represents the
core of the atisanediterpenoid alkaloids, we investigated the optimization
of the conversion of dihydronavirine to 60. Following
an extensive survey of reagents and reaction conditions, it was found
that exposure of 8 to iodosobenzene (1.5 equiv) in the
presence of NaHCO3 in methylene chloride for 1 h provides 60 in 75% yield. Various thermal and photochemical isomerization
conditions were attempted in order to convert 60 to navirine
(e.g., via an aza-Prins-type process) but led only to nonspecific
decomposition in all cases. Thus, to date, we have not been able to
convert dihydronavirine or related derivatives to navirine.
Scheme 14
Attempted
Dehydrogenation of Dihydronavirine (8) to
Navirine (5)
Synthesis of the Atisine Skeleton
Even though our attempted
conversion of dihydronavirine (8) to navirine (5) was ultimately not successful, the formation of 60 demonstrates that the hetidine framework can be converted to the
atisine skeleton as had been designed at the outset of our studies
(Figure 2).[37] While
the conversion 8 → 60 is in essence
a complexity-minimizing transformation, it achieves an important goal
of installing oxygenation at C11 of the atisine core, which is required
for the synthesis of natural products such as spiramine H (10). As such, we sought to investigate this C–C bond cleaving
transformation on our synthetic equivalent to the hetidine framework
(14). In this regard (Scheme 15), treatment of cyclic carbamate 14 with barium hydroxide
cleaves the oxazolidinone to yield amino diol 61. Subjecting 61 to the optimized conditions identified for the oxidative
Grob-like fragmentation of 8 (PhIO, 1.5 equiv and NaHCO3) leads to 63 in 75% yield. The structure of 63 was unambiguously confirmed by an X-ray crystallographic
study (see ORTEP, Scheme 15). Two possible
intermediates (62A and 62B) may be envisioned
en route to 63. Given the inherently higher nucleophilicity
of secondary amines compared to hydroxyl groups, intermediate 62A would seem more likely, although 62B cannot
be ruled out.
Scheme 15
Conversion of the Hetidine Framework to the Atisine
Skeleton
Conclusion
Our
studies on the C20-diterpenoid alkaloids with a
focus on a general approach to the hetidine and atisine subclasses
have identified a polycylic scaffold (14) that provides
synthetic access to the hetidine framework and demonstrates its conversion
to the atisine framework using an oxidative Grob-like fragmentation.
Furthermore, 14 has been advanced to a synthesis of dihydronavirine
(8), which is formally the hydrogenation product of navirine
(5). These synthetic studies were enabled by a Ga(III)-catalyzed
cycloisomerization reaction of alkynylindenes to build a 6–7–6
carbocyclic tricycle, a Michael/aldol sequence to forge the [2.2.2]
bicycle, and a Stille/etherification synthetic sequence to append
a hordenine fragment. It is our anticipation that these synthetic
studies will set the stage for further studies on the interconversion
of the carbon frameworks of diterpenoid alkaloid congeners and provide
access to unique derivatives of these architecturally fascinating
natural products to aid in future synthetic and biological studies.
Experimental Section
Experimental
procedures and characterization data for the preparation
of 22, 23, 19, 18, 27–29, 32–35, 45, 47–50, and 14 have been reported previously.[16]
Indene 19 was dried by azeotrope with benzene
(3 × 5 mL) and dried under vacuum for 12 h before use. Gallium(III)
iodide (0.80 g, 1.8 mmol, 0.25 equiv) and 4 Å activated powdered
molecular sieves (1.6 g, 2.0 g per gram of GaI3) were added
to a Schlenk flask in a N2 drybox. The Schlenk flask was
removed from the drybox, and a solution of indene 19 (1.9
g, 7.1 mmol, 1 equiv) in freshly distilled toluene (36 mL, 0.2 M)
was added to the flask via cannula. The Schlenk flask was sealed,
and the pale yellow mixture was heated to 100 °C and stirred
vigorously for 48 h. The reaction mixture was cooled room to temperature,
filtered through a pad of Celite, washed with EtOAc (200 mL), and
concentrated. If the reaction mixture turns red in within the first
few minutes of the reaction, it should be stopped immediately in order
to minimize the decomposition of the material. The crude product was
immediately purified by flash chromatography (9:1 hexanes/EtOAc) to
yield 1.7 g (6.3 mmol, 89% yield) of cycloheptadiene 18 as a clear oil: R 0.62
(2:1 hexanes/EtOAc, UV and p-anisaldehyde stain); 1H NMR (600 MHz, CDCl3) δ 7.05 (d, J = 6.8 Hz, 1H), 7.03 (d, J = 3.6 Hz, 1H),
6.89 (dd, J = 8.3, 2.6 Hz, 1H), 6.83 (d, J = 2.5 Hz, 1H), 6.63 (d, J = 11.7 Hz,
1H), 3.80 (s, 3H), 2.98–2.88 (m, 2H), 2.39 (qt, J = 18.8, 6.1 Hz, 2H), 2.08 (ddd, J = 12.3, 9.1,
2.6 Hz, 1H), 1.77 (dtd, J = 14.7, 5.8, 2.4 Hz, 1H),
1.72 (ddd, J = 13.2, 8.8, 2.4 Hz, 1H), 1.69–1.61
(m, 1H), 1.41 (s, 3H); 13C NMR (126 MHz, CDCl3) δ 157.9, 137.3, 136.5, 133.1, 129.6, 128.7, 128.1, 126.9,
124.8, 115.2, 112.1, 55.5, 40.2, 35.5, 35.5, 31.3, 26.7, 19.0; IR
(NaCl, thin film) νmax 3441, 2940, 2229, 1606 cm–1; HRMS (ESI) calcd for [C18H19NONa] ([M + Na]+) m/z 288.1359, found 288.1357.
Sodium hydride (60 wt % dispersion in mineral
oil, 120 mg, 3.1 mmol, 5 equiv) was added to a Schlenk flask followed
by a solution of ketone 29 (170 mg, 0.61 mmol, 1 equiv)
and phenyl vinyl sulfoxide (80 μL, 0.61 mmol, 1 equiv) in dry
DMF (6.0 mL, 0.1 M). The Schlenk flask was sealed and heated to 50
°C for 20 min. The reaction mixture was cooled to room temperature
and quenched by dropwise addition of water until gas evolution ceased
(∼0.5 mL). The reaction mixture was diluted with EtOAc (50
mL), washed with water (3 × 25 mL) and brine (1 × 25 mL),
dried over MgSO4, and concentrated. The crude yellow oil
was purified by flash chromatography (4:1 hexanes/EtOAc to 100% EtOAc)
to give 230 mg (0.53 mmol, 87% yield) of sulfoxide 30 as a clear oil that was immediately carried on to the sulfoxide
extrusion: R 0.07 (2:1
hexanes/EtOAc, UV and p-anisaldehyde stain).Sulfoxide 30 (230 mg, 0.53 mmol, 1 equiv) was dried
by azeotrope with benzene (1 × 10 mL) and transferred to a microwave
vial with freshly distilled toluene (11 mL, 0.05 M). The solution
was sparged with argon for 1 h, and then the reaction mixture was
heated in the microwave to 160 °C for 2 h. The reaction mixture
was diluted with EtOAc (75 mL), washed with water (3 × 50 mL)
and brine (1 × 25 mL), dried over MgSO4, and concentrated.
The crude yellow oil was purified by flash chromatography (4:1 hexanes/EtOAc)
to give 115 mg (0.37 mmol, 70% yield) of ketone 31 as
a white solid. Slow diffusion crystallization from benzene and hexanes
provided X-ray quality crystals: mp 127–128 °C; R 0.47 (2:1 hexanes/EtOAc,
UV and p-anisaldehyde stain); 1H NMR (500
MHz, CDCl3) δ 7.80 (d, J = 8.7 Hz,
1H), 6.80 (dd, J = 8.8, 2.6 Hz, 1H), 6.63 (d, J = 2.6 Hz, 1H), 5.88 (dd, J = 17.6, 10.7
Hz, 1H), 4.93 (d, J = 10.8 Hz, 1H), 4.77 (d, J = 17.6 Hz, 1H), 3.82 (s, 3H), 3.31–3.20 (m, 1H),
3.07 (ddd, J = 16.7, 6.8, 2.3 Hz, 1H), 2.50 (tdd, J = 9.4, 8.2, 5.1 Hz, 1H), 2.21–2.09 (m, 2H), 2.02–1.94
(m, 3H), 1.67 (dddt, J = 13.9, 9.7, 6.0, 2.5 Hz,
1H), 1.56 (t, J = 13.4 Hz, 1H), 1.52–1.45
(m, 1H), 1.42 (s, 3H); 13C NMR (126 MHz, CDCl3) δ 201.8, 162.0, 143.6, 143.1, 133.5, 131.5, 123.6, 114.9,
114.8, 112.2, 58.0, 55.4, 48.6, 38.8, 37.0, 33.6, 29.2, 26.5, 19.8;
IR (NaCl, thin film) νmax 2938, 1662, 1599 cm–1; HRMS (ESI) calcd for [C20H23NO2Na] ([M + Na]+) m/z 332.1621, found 332.1623.
LiAlH4 (330 mg, 8.8 mmol, 5
equiv, from Aldrich) was added to a flame-dried Schlenk flask followed
by a solution of ketone 32 (570 mg, 1.8 mmol, 1 equiv)
in THF (35 mL, 0.05 M) and heated to 65 °C for 4 h. The reaction
mixture was cooled to room temperature, slowly quenched with powdered
Na2SO4·10H2O (1.8 g, 0.2 g per
mmol LiAlH4), and stirred until the gray color disappeared
and the white precipitate clumped to the bottom of flask (∼12
h). The clear solution was filtered through a pad of Celite, washed
with Et2O (75 mL), and concentrated to give an amino alcohol
that was immediately subjected to the Boc-protection conditions: R 0.02 (100% EtOAc, UV and p-anisaldehyde stain).Di-tert-butyl
dicarbonate (Boc2O) (580 mg, 2.6 mmol, 1.5 equiv) and Et3N (0.32 mL, 2.3 mmol, 1.3 equiv) were added to a solution
of the crude amino alcohol (580 mg, 1.8 mmol, 1 equiv) in CH2Cl2 (44 mL, 0.04 M). After 45 min, the solution was concentrated
and the crude clear oil was purified by flash chromatography (4:1
hexanes/EtOAc) to give 700 mg (1.6 mmol, 93% yield over two steps)
of Boc-protected amino alcohol 33 as a clear oil: R 0.47 (2:1 hexanes/EtOAc,
UV and p-anisaldehyde stain); 1H NMR (600
MHz, CDCl3) δ 7.01 (d, J = 8.0 Hz,
1H), 6.66–6.53 (m, 2H), 5.74 (dt, J = 17.1,
8.0 Hz, 1H), 5.00 (dd, J = 9.9, 2.6 Hz, 1H), 4.87
(d, J = 16.9 Hz, 1H), 4.72 (t, J = 6.4 Hz, 1H), 4.38 (s, 1H), 3.76 (s, 4H), 3.63 (d, J = 13.3 Hz, 1H), 3.35–3.18 (m, 2H), 2.68–2.55 (m, 1H),
2.36–1.96 (m, 4H), 1.93–1.69 (m, 5H), 1.43 (s, 9H),
1.10–0.91 (m, 5H).; 13C NMR (151 MHz, CDCl3) δ 159.0, 156.4, 142.9, 135.3, 132.9, 131.9, 117.8, 115.5,
110.3, 78.8, 55.2, 55.2, 50.5, 44.4, 44.0, 38.8, 34.3, 34.1, 31.7,
28.9, 28.6, 28.5, 22.8, 22.7, 18.8, 14.2; IR (NaCl, thin film) νmax 3459, 2930, 2248, 1701, 1608 cm–1; HRMS
(ESI) calcd for [C26H39NO4Na] ([M
+ Na]+) m/z 452.2771,
found 452.2780.
4-Methylmorpholine N-oxide (28 mg, 0.24 mmol, 2 equiv) and OsO4 (2.5
wt % in tert-butyl alcohol, 15 μL, 0.0012 mmol,
0.01 equiv) was added to a stirred solution of 47 (41
mg, 0.12 mmol, 1 equiv) in a 2:1 mixture of THF/H2O (4
mL, 0.03 M). The mixture was stirred at room temperature for 24 h,
at which time the reaction mixture was diluted with EtOAc (20 mL),
washed with water (1 × 10 mL) and brine (1 × 5 mL), dried
over MgSO4, and concentrated. The resulting diol was filtered
through a pad of silica with 9:1 CH2Cl2/MeOH
(10 mL), concentrated, and used without any further purification: R 0.05 (100% EtOAc, UV and
KMnO4 stain).NaIO4 on silica (240 mg,
2.0 g/mmol diol) was added to a stirred solution of the diol (45 mg,
0.12 mmol, 1 equiv) in CH2Cl2 (1.2 mL, 0.1 M).
After 30 min, the mixture was filtered through a pad of Celite, washed
with CH2Cl2 (10 mL), and concentrated. The crude
aldehyde was purified by flash chromatography (1:1 hexanes/EtOAc)
to give 25 mg (0.070 mmol, 58% yield over two steps) of aldehyde 48 as a clear oil: R 0.14 (1:1 hexanes/EtOAc, UV and KMnO4 stain); 1H NMR (500 MHz, CDCl3) δ 9.63 (t, J = 1.5 Hz, 1H), 6.81 (d, J = 10.0 Hz,
1H), 6.33 (dd, J = 10.1, 1.8 Hz, 1H), 6.14 (d, J = 2.0 Hz, 1H), 4.20 (s, 1H), 3.29–3.18 (m, 2H),
2.67–2.55 (m, 2H), 2.55–2.48 (m, 1H), 2.34–2.19
(m, 2H), 1.99 (dq, J = 13.1, 2.7 Hz, 1H), 1.78 (ddt, J = 14.4, 12.6, 3.7 Hz, 1H), 1.72–1.57 (m, 4H), 1.56–1.42
(m, 2H), 1.35–1.22 (m, 2H), 1.11 (s, 3H); 13C NMR
(151 MHz, CDCl3) δ 199.7, 184.8, 158.3, 156.9, 145.4,
131.8, 129.7, 79.8, 65.3, 51.2, 48.9, 46.3, 43.9, 40.7, 38.6, 33.5,
32.3, 28.4, 26.1, 19.4; IR (NaCl, thin film) νmax 2929, 2873, 1755, 1719, 1671, 1636 cm–1; HRMS
(ESI) calcd for [C20H24NO4] ([M +
H]+) m/z 342.1700, found
342.1708.
Triethylamine (28 μL, 0.20 mmol, 2 equiv) was added to a
stirred solution of allylic alcohol 55 (35 mg, 0.10 mmol,
1 equiv) in CH2Cl2 (3.0 mL, 0.03 M) followed
by methanesulfonyl chloride (12 μL, 0.15 mmol, 1.5 equiv). The
reaction mixture was stirred at 0 °C for 15 min and then filtered
through a pad of Celite and washed with 10 mL (1:1 hexanes/EtOAc)
while still cold. The resulting mesylate was concentrated and immediately
carried on to the next reaction. If the reaction was allowed to warm
above 0 °C before filtration through silica the allylic chloride
was obtained as the major product: allylic mesylate R 0.15 (9:1 CH2Cl2/MeOH, 1% aq NH4OH, p-anisaldehyde stain);
allylic chloride R 0.50
(9:1 CH2Cl2/MeOH, 1% aq NH4OH, p-anisaldehyde stain);Sodium hydride (60 wt % dispersion
in mineral oil, 4.0 mg, 0.10 mmol, 1 equiv) was added to a stirred
solution of the mesylate (41 mg, 0.10 mmol, 1 equiv) and hordenine
(56, 16 mg, 0.10 mmol, 1 equiv) in THF (4.0 mL, 0.025
M). After 1 h at room temperature, wet CH2Cl2 (1 mL) was added, and the reaction mixture was concentrated. Flash
chromatography (19:1 CH2Cl2/MeOH, 1% NH4OH) provided 24 mg (0.050 mmol, 50% yield) of 57 as a white solid. The product 57 can also be obtained
from the allylic chloride in diminished yields using the same conditions
reported above except the reaction must be heated to 70 °C for
1 h: mp 132–137 °C; R 0.13 (9:1 CH2Cl2/MeOH, 1% aq NH4OH, p-anisaldehyde stain); 1H
NMR (500 MHz, CDCl3) δ 7.10 (dd, J = 6.7, 1.9 Hz, 2H), 6.82 (dd, J = 6.5, 2.0 Hz,
2H), 5.59 (d, J = 2.0 Hz, 1H), 4.56 (t, J = 1.9 Hz, 2H), 3.56 (d, J = 14.0 Hz, 1H), 3.49
(s, 1H), 2.85 (d, J = 13.9 Hz, 1H), 2.76–2.70
(m, 2H), 2.61–2.57 (m, 1H), 2.57–2.50 (m, 2H), 2.32
(s, 6H), 2.24 (dd, J = 14.1, 9.4 Hz, 1H), 1.94 (dd, J = 12.5, 4.3 Hz, 1H), 1.88–1.81 (m, 2H), 1.80–1.66
(m, 3H), 1.60 (tdd, J = 13.8, 9.4, 4.0 Hz, 1H), 1.45–1.39
(m, 3H), 1.36 (td, J = 10.6, 2.4 Hz, 1H), 1.25 (tdd, J = 13.4, 9.2, 4.4 Hz, 3H), 1.19–1.08 (m, 1H), 1.02
(s, 3H); 13C NMR (126 MHz, CDCl3) δ 159.2,
157.3, 145.5, 132.5, 129.7, 129.5, 114.8, 87.9, 69.9, 68.1, 61.7,
50.8, 48.6, 47.8, 45.8, 45.4, 45.1, 41.6, 41.5, 36.5, 33.8, 33.3,
31.7, 28.8, 28.2, 27.3, 20.8, 20.4; FTIR (NaCl, thin film) νmax, 2925, 1748 cm–1; HRMS (ESI) calcd for
[C31H41O3N2] ([M + H]+) m/z 489.3116, found 489.3112.
Authors: Benjamin G Davis; Michael A T Maughan; Timothy M Chapman; Renaud Villard; Steve Courtney Journal: Org Lett Date: 2002-01-10 Impact factor: 6.005
Authors: Felipe de Jesus Cortez; David Lapointe; Amy M Hamlin; Eric M Simmons; Richmond Sarpong Journal: Tetrahedron Date: 2013-07-08 Impact factor: 2.457
Authors: Kevin G M Kou; Jason J Pflueger; Toshihiro Kiho; Louis C Morrill; Ethan L Fisher; Kyle Clagg; Terry P Lebold; Jessica K Kisunzu; Richmond Sarpong Journal: J Am Chem Soc Date: 2018-06-19 Impact factor: 15.419
Authors: Jason J Pflueger; Louis C Morrill; Justine N deGruyter; Melecio A Perea; Richmond Sarpong Journal: Org Lett Date: 2017-08-18 Impact factor: 6.005
Authors: Kevin G M Kou; Svitlana Kulyk; Christopher J Marth; Jack C Lee; Nicolle A Doering; Beryl X Li; Gary M Gallego; Terry P Lebold; Richmond Sarpong Journal: J Am Chem Soc Date: 2017-09-20 Impact factor: 15.419