Total synthesis of (±)-jiadifenin and studies directed to understanding its SAR: Probing mechanistic and stereochemical issues in palladium-mediated allylation of enolate-like structures

Article abstract of DOI:10.1021/ja056980a

The total synthesis of jiadifenin has been accomplished. The synthesis allows us to build an SAR profile which suggests that the jiadifenin skeleton may be less desirable from the standpoint of nominating a potential drug than that of its prerearrangement

Full text of DOI:10.1021/ja056980a

Published on Web 12/20/2005
Total Synthesis of (()-Jiadifenin and Studies Directed to
Understanding Its SAR: Probing Mechanistic and
Stereochemical Issues in Palladium-Mediated Allylation of
Enolate-Like Structures
David A. Carcache,† Young Shin Cho,† Zihao Hua,† Yuan Tian,‡ Yue-Ming Li,‡ and
Samuel J. Danishefsky*,†,§
Contribution from the Laboratory for Bioorganic Chemistry and Laboratory of Biochemistry and
Pharmacology, Sloan-Kettering Institute for Cancer Research, 1275 York AVenue,
New York, New York 10021, and Department of Chemistry, Columbia UniVersity,
HaVemeyer Hall, 3000 Broadway, New York, New York 10027
Received October 12, 2005; E-mail: s-danishefsky@mskcc.org
Abstract: The total synthesis of jiadifenin has been accomplished. The synthesis allows us to build an
SAR profile which suggests that the jiadifenin skeleton may be less desirable from the standpoint of
nominating a potential drug than that of its prerearrangement precursor. The key steps of the jiadifenin
problem involve the construction of two 1,3-related quaternary carbons. The paper describes how the
stereochemistry was managed in this context. The issue was studied in considerable detail at the level of
a then new allyl transfer reaction arising from a palladium-mediated transfer process of an allyl carbonate.
By use of externally deuterated diallyl carbonate, we could probe, for the first time, the stereochemical
relationship between the inter- and intramolecular versions of this process. The existence of concurrent
inter- and intramolecular allylation reactions was demonstrated by deuteration experiments. While in the
particular case at hand, we find very little difference in stereochemical outcome as one partitions between
the inter- and intramolecular pathways, the techniques employed are applicable to other systems.
Introduction
ment of small molecule nonpeptidyl CNS-permeable neurotro-
phins represents an area of growing interest. In this context,
Naturally occurring polypeptidyl neurotrophic factors, such
as NGF, BDNF, and GDNF, play an important role in mediating
neuronal survival, differentiation, growth, and apoptosis.1
Decreased neurotrophic support has been linked to the progres-
sion of a number of neurodegenerative disorders, including
Alzheimer’s disease, Parkinson’s disease, and Huntington’s
disease. Given their potential therapeutic value, it is not
surprising that neurotrophic factors have been the focus of a
considerable amount of interdisciplinary research. Indeed,
several neurotrophic factors have been evaluated in animal
studies. However, due to issues of poor pharmacokinetics and
bioavailability, the prospects for use of these polypeptides as
clinical agents appears to be limited. Consequently, the develop-
our laboratory has launched a research program directed to the
total synthesis of small molecule natural products that mimic
the effects of naturally occurring neurotrophic factors. To date,
we have synthesized several such compounds, including tricy-
cloillicinone,2 merrilactone A,3 scabronine G methyl ester,4 and
NGA0187.5 Our involvement in this field led us to pursue the
total synthesis of the structurally complex small molecule
neurotrophin, jiadifenin. In this paper, we describe how this goal
was accomplished.6 This program led us to investigate some of
the finer mechanistic and stereochemical issues associated with
the creation of a C-allyl quaternary center by decarboxylative
allylation, starting with an enol carbonate. Finally, the begin-
nings of the mapping of a preliminary SAR profile of neu-
rotrophic activity are described.
† Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for
Cancer Research.
Fukuyama’s group recently reported the isolation and struc-
tural identification of a novel majucin-type prezizaane, jiadifenin
(Scheme 1, 1), from the Illicium jiadifengpi species of China.7
The known (2S)-hydroxy-3,4-dehydronemajucin (2) had also
‡ Laboratory of Biochemistry and Pharmacology, Sloan-Kettering Institute
for Cancer Research.
§ Columbia University.
(1) (a) Lodish, H.; Berk, A.; Zipursky, S. L.; Matsudaria, P.; Baltimore, D.;
Darnell, J. Molecullar Cell Biology, 4th ed.; W. H. Freeman & Co.: New
York, 1999; Section 23.8. (b) Kaneko, M.; Saito, Y.; Saito, H.; Matsumoto,
T.; Matsuda, Y.; Vaught, J. L.; Dionne, C. A.; Angeles, T. S.; Glicksman,
M. A.; Neff, N. T.; Rotella, D. P.; Kauer, J. C.; Mallamo, J. P.; Hudkins,
R. L.; Murakata, C. J. Med. Chem. 1997, 40, 1863-1869. (c) Levi-
Montalcini, R.; Hamburge, V. J. Exp. Zool. 1951, 116, 321-362. (d)
Bennett, M. R.; Gibson, W. G.; Lemon, G. Auton. Neurosci. 2002, 95,
1-23. (e) Hefti, F. Annu. ReV. Pharmacol. Toxicol. 1997, 37, 255-257.
(f) Dawbarn, D.; Allen, S. J. Neuropathol. Appl. Neurobiol. 2003, 29, 211-
230.
(2) Pettus, T. R. R.; Chen, X.-T.; Danishefsky, S. J. J. Am. Chem. Soc. 1998,
120, 12684.
(3) Birman, V.; Danishefsky, S. J. J. Am. Chem. Soc. 2002, 124, 2080.
(4) Waters, S.; Tian, Y.; Li, Y.; Danishefsky, S. J. J. Am. Chem. Soc. 2005,
127, 13514.
(5) Hua, Z.; Carcache, D. A.; Tian, Y.; Li, Y.-M.; Danishefsky, S. J. J. Org.
Chem. 2005, 70, 9849.
(6) Cho, Y. S.; Carcache, D. A.; Tian, Y.; Li, Y.-M.; Danishefsky, S. J. J. Am.
Chem. Soc. 2004, 126, 14358.
9
1016
J. AM. CHEM. SOC. 2006, 128, 1016-1022
10.1021/ja056980a CCC: $33.50 © 2006 American Chemical Society

Total Synthesis of Jiadifenin
A R T I C L E S
Scheme 1. Structure of Jiadifenina
a Key: (a) Dess-Martin reagent (ref 9), CH2Cl2, rt, 81%; (b) DBU, benzene, 80 °C, 74%; (c) Dess-Martin reagent, dioxane, rt; (d) MeOH, rt, 9%.
Scheme 2. Synthetic Strategy
been previously isolated from I. jiadifengpi.8 Fukuyama’s group
demonstrated that 2, the absolute configuration of which had
already been determined, could be converted to jiadifenin (1),
as outlined in Scheme 1. This interesting series of manipulations
prompts consideration of 2 as a potential biosynthetic precursor
to jiadifenin. Moreover, since the absolute configuration of 2
is known, the absolute configuration of jiadifenin is securely
established.
optimistic that, with the tricyclic core structure in place,
stereochemical biases could be exploited to achieve adequate
control in the installation of the three remaining stereocenters
at C1, C6, and C7. All indications from Fukuyama’s work
suggested that the hemiacetal at C10 of jiadifenin cannot be
obtained as a single anomer. Thus, a projectable series of
transformations could allow access to 10. On the basis of the
earlier work of Fukuyama’s group. (Scheme 1), it could be
anticipated that, upon oxidation of C10, intermediate 10 would
undergo ring contraction to produce jiadifenin (1).7
It is not uncommon that there are chemical teachings to be
learned from a disciplined pursuit of this type of compact but
functionality-laden target. Moreover, both jiadifenin and 2 have
been shown by Fukuyama’s group to promote neurite outgrowth
in primary cultures of rat cortical neurons at levels as low as
0.1-10 µM.7 While the field of pursuing the neurochemical
foundations and implications of this type of finding is still in
its infancy, it is not inconceivable that, in due course, a
compound of neuronal enhancing characteristics could be used
to advantage. Hence, from both biological and chemical
perspectives, jiadifenin presented itself as a particularly ap-
propriate target for chemical synthesis.
Hoping to circumvent anticipated difficulties associated with
controlled, selective dialkylation at C9, starting with a compound
of the type 6, we first explored the possibility of bypassing this
issue through desymmetrization of a 9,9-diallylcyclohexanone
(cf., 12). In the event, methylation of 5, followed by hydroxy-
methylation10 and protection of the resultant alcohol gave rise
to 11. Sequential diallylation with allyl bromide was followed
by silyl deprotection to afford intermediate 12 (Scheme 3). At
this point, the goal was to differentiate the two C9 allyl
substituents by oxidative etherification based on their differing
stereochemical relation to the C5 hydroxymethyl functionality.
For instance, it was hoped that, upon treatment of 12 with
N-bromosuccinimide, the hydroxyl group of the future C5 would
participate to produce a bromoether across the cyclohexyl ring
system, thereby distinguishing the two allyl functionalities. In
the event, however, treatment of 12 with NBS gave rise to 13,
in which the formal ketone hydrate had facilitated bromination
of both olefin functionalities.11 The failure of the alcohol to
participate in the bromination event may be attributed, in part,
to the configuration of the cyclohexanone ring system, which
requires the hydroxymethyl group to be situated 1,3-syn diaxial.
Having not been successful in achieving diastereocontrol through
Our preliminary synthetic strategy is adumbrated in Scheme
2. It was assumed that compound 6 should be obtainable, albeit
most likely in racemic form, from the commercially available
1,4-cyclohexanedione monoethylene ketal 5. If this were indeed
the case, we planned to install the C9 quaternary carbon center
through sequential alkylation reactions, recognizing, of course,
that there could well be difficulties in achieving diastereocontrol
in such a progression. From intermediate 7, the requisite
cyclopentenone functionality would be installed through an
intramolecular Horner-Wadsworth-Emmons reaction and,
following appropriate functional group management, intermedi-
ate 8 would be in hand. At this stage, we anticipated installation
of the lactone through an intramolecular Claisen-type condensa-
tion with a pendant carbonate moiety (cf., 8 to 9). We were
(9) Dess, D. B.; Martin, J. C. J. Am. Chem. Soc. 1982, 104, 902.
(10) Brieskorn, C. H.; Schwack, W. Chem. Ber. 1981, 114, 1993.
(11) Literature precedents of such a reaction: (a) Heusler, K.; Ueberwasser,
H.; Wieland, P.; Wettstein, A. HelV. Chim. Acta 1957, 40, 787. (b) Ernst,
L.; Gorlitzer, K.; Boventer, K. Arch. Pharm. (Weinheim, Ger.) 1990, 323,
361.
(7) Yokoyama, R.; Huang, J.-M.; Yang, C.-S.; Fukuyama, Y. J. Nat. Prod.
2002, 65, 527.
(8) Kuono, I.; Baba, N.; Hashimoto, M.; Kawano, N.; Takahashi, M.; Kaneto,
H.; Yang, C.-S. Chem. Pharm. Bull. 1990, 38, 422.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 1017

A R T I C L E S
Scheme 3 a
Carcache et al.
a Key: (a) LHMDS, THF, -78 °C, then MeI, -78 °C to room temperature; (b) 10% KOH, MeOH, HCHO(aq), 0 °C; (c) TBSOTf, 2,6-lutidine, CH2Cl2,
0 °C, 64% for three steps; (d) LHMDS, THF, -78 °C, then allyl bromide, -78 °C to room temperature, 73%; (e) LHMDS, THF, -78 °C, then allyl
bromide, -78 °C to room temperature, 95%; (f) TBAF, THF, rt, 76%; (g) NBS, THF, -45 °C, 66%.
Scheme 4 a
a Key: (a) LHMDS, THF, -78 °C, then allyl bromide, -78 °C to room temperature, 73% (14) (99% based on recovered starting material), 51% (18);
(b) LDA, THF, -78 °C to -20 °C, then BrCH2CO2Et, HMPA, -78 °C, 62% (15 + 16), 65% (19 + 20).
Scheme 5 a
a Key: (a) LiCH2P(O)(OMe)2, THF, -78 °C, 81% (99% based on recovered starting material); (b) NaH, THF, reflux, 91%; (c) LDA, THF, -20 °C, then
MeI, -30 °C; (d) 2 N HCl, THF, 85% for two steps; (e) HC(OMe)3, PPA, 105 °C, 39% (24-C1-R) and 19% (24-C1-â).
intramolecular differentiation, we next examined the selectivity
levels attainable through direct sequential dialkylation.
Thus, intermediate 11 was subjected to C9 allylation followed
by C9 carboethoxymethylation to afford compounds 15 and 16
in an approximately 3:1 ratio, as shown (Scheme 4). The
observed product distribution suggests some bias for the second
alkylation event to occur anti to the protected C5 hydroxymethyl
functionality. It was later found (vide infra) that the R,â ratio
in the second alkylation event could be enhanced by increasing
the steric bulk of the hydroxymethyl moiety. Thus, substitution
of the -TBS protecting group with the bulkier -TBDPS moiety
(cf., 17) led to formation of products 19 and 20 in an
approximately 7:1 ratio, as shown.
compound 21 could be readily prepared by alkylation of 11,
attempted base-induced allylation failed (cf., 21 f 15). On the
basis of these and related studies, there was reason to believe
that allylation adjacent to the ester was a complicating factor.
In the above graphic, the following conditions were used:
(a) LDA, THF, -78 °C to -20 °C, then BrCH2CO2Et, -78 °C
to room temperature, 68% (87% brsm); (b) KHMDS, THF, -78
°C, then allylbromide, HMPA, THF, -78 °C.
For the plans we had in mind (vide infra), either of the two
series (15, 16 or 19, 20) could in principle be incorporated into
our proposed progression.
At that time we had, on hand, compound 15, and subsequent
initiatives started with this intermediate. Indeed, conversion of
15 to the corresponding â-ketophosphonate was readily ac-
complished. This was followed by intramolecular Horner-
Wadsworth-Emmons reaction12 to afford intermediate 22
(Scheme 5). At this stage, R-methylation at C1 occurred
primarily from the R-face, anti to the pendant allyl functionality
to afford, upon deprotection, 23, as a 7:1 mixture of isomers.
It was initially sought to interpolate carbon 12 through a
mixed orthoformate ester formation-cyclization sequence.
Indeed, this type of reaction had found excellent application in
our synthesis of tazettine some years ago.13 In the event,
intermediate 23 was first heated with trimethylorthoformate.
Hence, we attempted to reverse the sequence of alkylation
at C9 (in the series carrying the TBS function; i.e., 11). While
(12) Halterman, R. L.; Vollhardt, K. P. C. Organometallics 1988, 7, 883.
9
1018 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Total Synthesis of Jiadifenin
A R T I C L E S
Scheme 6 a
a Key: (a) 2 N HCl, THF, 94%; (b) ClCO2Et, py, DMAP, CH2Cl2, 0 °C to room temperature, 93%; (c) NaH, THF, reflux, 94%; (d) m-CPBA, CH2Cl2,
90%; (e) NaBH4, THF-MeOH (1:1), -78 °C, 93%.
Scheme 7. Completion of the Synthesis of Jiadifenina
a Key: (a) LDA, THF, -40 °C to -15 °C, then MeI, HMPA, -35 °C, 64% (77% based on recovered starting material); (b) O3, sudan 7B red, CH2Cl2-
EtOH (1:1), -78 °C; (c) Jones’ reagent, acetone, 90% for two steps; (d) NaBH4, CeCl3‚7H2O, THF-MeOH (3:1), -65 °C, 88%; (e) NaHMDS, THF, -78
°C, then Davis’ oxaziridine, THF, -78 °C, 42% after one recycle; (f) Jones’ reagent, acetone, MeOH, 28 (29%) and 1 (40%); (g) Jones’ reagent, acetone,
MeOH, 0 °C, 46%.
When the alcohol had been consumed, the reaction was treated
with polyphosphoric acid to afford, upon workup, 24 as only a
2:1 mixture of C1 epimers. Unfortunately, various attempts to
oxidize the glycoside-like linkage (at C12) of 24 to the requisite
lactone were unsuccessful, resulting only in retro-Aldol decom-
position products.
rangement of the R-ketolactone moiety to the hydroxytetrahy-
drofuran carboxylate acetal of jiadifenin. Indeed, upon exposure
of 27 to Jones’ oxidation conditions, we isolated jiadifenin (1,
40%) along with intermediate 28, in which only C2 had been
oxidized (29%). Following separation, the latter could be
converted to jiadifenin upon resubmission to Jones’ oxidation
conditions. The overall yield for the conversion of 27 to
jiadifenin (1) was 53%. The spectral data of the synthetic
material are in complete accord with the data published for the
isolated material.7 We were able to further confirm the identity
of the synthesized material through comparison with the NMR
spectrum of the isolated material, kindly provided by Professor
Fukuyama. As is the case with the naturally derived jiadifenin,
the fully synthetic material exists as an anomeric mixture at
C10. Two separate peaks, corresponding to the jiadifenin
anomers, were observed by HPLC analysis. Each peak was
isolated; however, NMR spectroscopy once more revealed an
anomeric mixture similar in ratio to that observed for the
preseparated material, suggesting rapid C10 equilibration.
Having successfully accomplished the total synthesis of
racemic jiadifenin (1), it was now appropriate to revisit the issue
of the stereoselective installation of the C9 quaternary carbon
center. Our direct dialkylation route had been only moderately
selective, providing, at that time, a 3:1 product ratio of the
desired 15 and the undesired 16 (although we subsequently
learned that this ratio was enhanced when the hydroxymethyl
functionality was protected with the bulkier -TBDPS moiety
(cf., 19:20, 7:1)). Aside from the local jiadifenin priorities, we
were hoping to bring forth a method to install a quaternary
carbon center through allylation of a γ-ketoester substrate. As
noted above, attempts to directly allylate compounds of the type
21 through conventional enolate chemistry had failed to produce
In light of the difficulties encountered in reaching the required
lactone via 24 followed by oxidation, we attempted the one
carbon interpolation with a carbon dioxide rather than ortho-
formate acid equivalent, hoping that this might allow direct
access to the tricyclic system in the required C12 oxidation state.
In addition, it seemed prudent to delay the installation of the
epimerizable C1 methyl group until later in the synthesis.
Accordingly, intermediate 22 was converted to mixed carbonate
ester 8, as shown (Scheme 6). Happily, cyclization proceeded
smoothly to afford the tricyclic core system 9. As expected,
hydroxylation of 9 with mCPBA proceeded exclusively from
the R-face and, following stereoselective reduction, trans diol
25 was in hand. The assignments of the relative configurations
of the newly installed stereocenters were confirmed through
X-ray crystallographic analysis of 25.
Stereoselective methylation of 25 at C1, followed by a two-
step oxidative cyclization sequence, furnished intermediate 10
(Scheme 7). Upon exposure to Luche reduction conditions, 10
was stereoselectively reduced to 26. Incorporation of the C10
hydroxyl group was achieved with the Davis oxaziridine14 to
afford R-hydroxylactone 27 as a 6:1 diastereomeric mixture.
With 27 in hand, we hoped to accomplish concurrent oxidation
of the C2 and C10 hydroxyl groups, thereby prompting rear-
(13) Danishefsky, S. J.; Morris, J.; Mullen, G.; Gammill, R. J. Am. Chem. Soc.
1980, 102, 2838.
(14) Davis, F. A.; Chen, B. C. Chem. ReV. 1992, 92, 919.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 1019

A R T I C L E S
Carcache et al.
Scheme 8. Palladium-Mediated Intramolecular Allylationa
a Key: (a) LDA, BrCH2CO2Et, THF, -78 °C to room temperature, 64-81%; (b) t-BuOK, allylchloroformate, THF, -78 °C, 80% to quantitative; (c)
Pd(PPh3)4, THF, rt (29) or -40 °C (30), 62-65%.
Scheme 9
the desired product, leading instead to a complex mixture of
alkylation products.
gave rise to the precursors 29 (R ) TBS) and 30 (R ) TBDPS).
Following exposure to the conditions known to mediate π-allyl
formation (Pd(PPh3)4 and THF), substrate 29 gave rise to 15
and 16 in a ratio of 1:5.6. Perhaps not surprisingly, the major
product of this transformation was of the opposite relative
stereochemistry to that obtained in the direct dialkylation
sequence. This product distribution can be attributed to the
preference for the allylation event to occur from the R-face of
the molecule, anti to the protected hydroxymethyl functionality.
Similarly, allylation of the TBDPS-protected analogue, 30, led
to formation of 19 and 20 in a more satisfactory ratio of 1:8.7.
The levels of stereoselectivity observed in the allylations of both
29 and 30 were somewhat enhanced in comparison with those
obtained from the analogous dialkylation sequences (vide supra).
In particular, we hoped to address both issues of selectivity
and reactivity through an alternative process involving intramo-
lecular transfer of the required allyl group. To this end, we
turned to a methodology suggested in a different context by
Tsuji and co-workers.15 While our work was in progress, two
more closely related papers appeared from Trost and Xu16 and
Behenna and Stoltz,17 which anticipated some important ele-
ments of our independent findings. However, as will be seen,
some of the questions asked go to some issues which had not
been explored even in these more recent disclosures. At the
time, we proposed to install the allyl group at the quaternary
carbon (C9 in jiadifenin numbering) by palladium-mediated
transfer from an enol carbonate moiety with concurrent exclu-
sion of carbon dioxide (see, for instance, transformations of 29
to 15 and 30 to 19, Scheme 8). Prior to the Trost and Stoltz
disclosures,16,17 no such reaction had been used to create a
quaternary center. In our case, we asked another question; i.e.,
would the diastereoselectivity of such a reaction constructing
two 1,3-related neopentyl groups approximate that of the more
conventional metalloenolate alkylations.
Having developed the means to install, with selectivity, the
C9 quaternary carbon center through the palladium-mediated
allylation reaction, and not then aware of prior findings in other
laboratories,16,17 we sought to learn more about the mechanism
of this transformation. At the extreme, the allylation reaction
could proceed either through a concerted, associative transition
state or through a dissociative transition state involving in
essence allylation of a pallado enolate. Even in the nonconcerted
case, there would still be a question of whether the allyl group
is held at some level in association with the enolate or is fully
dissociated (Scheme 9). To evaluate whether the allyl group of
the allyl carbonate starting material is ever released from the
reacting sphere of the enolate, we sought to conduct the
allylation reaction in the presence of an appropriately labeled
external π-allyl source. Incorporation of external allyl group
In practice, 11 (R ) TBS) and 17 (R ) TBDPS) were
alkylated individually, to afford intermediates 21 and 31. These,
upon treatment with tBuOK and allylchloroformate as shown,
(15) Tsuji, J.; Minami, I. Acc. Chem. Res. 1987, 20, 140. (b) Tsuji, J.; Yamada,
T.; Minami, I.; Yuhara, M.; Nisar, M.; Shimizu, I. J. Org. Chem. 1987,
52, 2988. (c) Tsuji, J.; Ohashi, Y.; Minami, I. Tetrahedron Lett. 1987, 28,
2397.
(16) Trost, B. M.; Xu, J. J. Am. Chem. Soc. 2005, 127, 2846.
(17) Behenna, D. C.; Stoltz, B. M. J. Am. Chem. Soc. 2004, 126, 15044.
9
1020 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Total Synthesis of Jiadifenin
A R T I C L E S
Scheme 10. Mechanistic Evaluationa
a Key: (a) diallyl carbonate (10 equiv), Pd(PPh3)4, THF, rt, 77%; (b) diallyl-d5 carbonate (10 equiv), Pd(PPh3)4, THF, rt, 90%.
into the product would reflect a dissociative pathway. We first
performed a control study in order to determine whether the
reaction could, in fact, be conducted in the presence of an
external π-allyl source. Thus, 30 was exposed to the reaction
conditions shown in the presence of 10 equiv of diallyl carbonate
at room temperature. The product distribution of 19:20 was
observed, as shown in Scheme 10. Of course, this enabling
experiment did not go to the question of inter- or intramolecu-
larity in the allylation event.
Intramolecular aldol condensation of 34 with tBuOK in THF
provided the bicyclic acid 36, presumably via intermediate 35,
with concomitant participation of the ethyl ester functionality
in the aldol cyclization. Intermediate 36 was converted to
terminal olefin 37 through a straightforward series of reactions
involving thioester formation, Fukuyama reduction to the
aldehyde, and Wittig reaction. Compound 37 corresponds to
the -TBDPS-protected analogue of intermediate 22 of the
original synthesis. Thus, we had realized our original expectation
that either diastereomer relating C5 and C9 could be redeemable
in our synthesis (Scheme 11).
Secure in the knowledge that the C-allylation reaction could
indeed be performed in the presence of potential external allyl
source, we exposed compound 30 to the same set of conditions,
using diallyl-d10 carbonate, as shown. The distribution of
products 19 and 20 and the level of deuterium incorporation
were as shown. We found the ratio of 19 to 20 to be essentially
unchanged from the control experiment (1:6.6). Interestingly,
however, both the minor (19) and the major isomer (20)
exhibited approximately 67% allyl-d5 incorporation. This result
is particularly suggestive because the deuterated products (19-
d5 and 20-d5) presumably arise from an intermolecular allyl
transfer reaction with diallyl carbonate, while the nondeuterated
products (19 and 20) arise from the intramolecular allyl transfer
reaction. If the nature of the transition state of the intermolecular
transfer (with diallyl-d5 carbonate) were markedly different from
that of the intramolecular transfer, it would be expected that
the selectivity for the major isomer would be different in the
deutero series. Were this the case, the ratio of 20-d5 to 19-d5
would be expected to be lower than the ratio of 20 to 19. The
fact that there is no apparent change in the distribution of
stereoisomeric products when the allyl source is external (diallyl-
d5 carbonate) could be taken to suggest a similar transition state
for both pathways. We conclude that, at least in this particular
case, the decarboxylative allylation actually occurs through a
tight but dissociative transition state even when actual crossover
had not occurred. Certainly, our study provides no support for
the notion of a concerted decarboxylation-allylation motif.
Having indeed accomplished an improved method for the
stereoselective formation of the C9 quaternary carbon center,
we turned our attention to demonstrating the feasibility of our
proposal that 20 could indeed be merged with a validated
intermediate in our first generation synthesis. Thus, 20 was
subjected to Wacker oxidation to afford methyl ketone 34.
Biological Investigations
At this point, we were in a position to corroborate the claimed
neurotrophic activity of jiadifenin (1) and to begin to establish
an SAR profile. The ability of jiadifenin to promote neurite
outgrowth was measured in both the presence and absence of
nerve growth factor (NGF). In the presence of NGF, jiadifenin
enhanced neurite lengths by 162% (P < 0.05). However, in the
absence of NGF, no neurite outgrowth was observed, indicating
that jiadifenin operates by upregulating the action of NGF rather
than functioning independently.
In addition to jiadifenin itself, we evaluated the in vitro
neurotrophic activity of several synthetic analogues of the natural
product in order to establish an SAR profile (Figure 1). The
most active compound was found to be intermediate 28, which
is the direct precursor to the natural product. This analogue was
found to enhance neurite lengths by 184%. The normethyl
jiadifenin analogue, 38, also exhibits superior activity in
comparison with jiadifenin, enhancing neurite lengths by 181%.
On the basis of the activity observed with 28 and 38, it might
be expected that the unrearranged, normethyl analogue, 39,
would be particularly active. Surprisingly, however, 39 displays
only moderate activity in this assay, suggesting a somewhat
complex SAR profile for this natural product. Interestingly,
intermediate 10, in which C10 is unoxidized, exhibits no neurite
length enhancement. Clearly, additional entries will be necessary
to establish a definitive SAR profile.
In summary, at the chemical level, the total synthesis of
jiadifenin has been accomplished. Among the important chemi-
cal steps are the creation of two 1,3-related quaternary centers
(see transformation of 14 f 15), eventually with reasonable
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 1021

A R T I C L E S
Scheme 11 a
Carcache et al.
a Key: (a) PdCl2, Cu(OAc)2, O2, DMA/H2O, rt, 69%; (b) t-BuOK, THF, 0 °C, 61%; (c) EtSH, DMAP, EDCI‚HCl, CH2Cl2, 0 °C to room temperature,
95%; (d) Et3SiH, Pd/C, CH2Cl2, rt, 67%; (e) MePPh3Br, NaHMDS, THF, rt, 72%.
Figure 1.
stereoselection (see ratio 19:20). Another important step was
the creation of the γ-lactone ring by intramolecular acylation
of a proximal carbonate ester (see transformation 8 f 9). This
having been accomplished, the synthesis could take advantage
of some of the findings registered by the discoverers of the
compound. Thus, hydroxylation of lactone 26 gave rise to 27,
which, upon oxidation followed by rearrangement, leads to
jiadifenin itself. All indications which we have been able to
garner suggest that it will not be possible to separate jiadifenin
into its C10 anomers in that the equilibration of the two anomeric
series is very rapid.
During the course of these studies we also had occasion to
study applicability of enol allyl carbonates as a means of
allylation with improvement of diastereofacial selectivity at C9.
Some improvement was in fact noted; however, the maximal
stereoselectivity that we realized in the allyl carbonate allylation
reaction was not very much greater than that which could be
realized by optimization of the corresponding lithium enolate.
A more careful study of this enol carbonate allylation route
defined that, to a considerable extent, allyl residues from the
initial cyclocarbonate exchange with a pool of external diallyl
carbonate under the alkylation conditions. Interestingly, the
diastereofacial selectivity is unaffected by whether the reaction
had occurred via the initial, unexchanged allyl function or
through the external allyl group. Thus, allyl which is delivered
from the reacting sphere of the starting material and that which
arises from bulk medium provide the same diastereofacial ratio.
Apparently there is nothing in our data which supports the notion
of an associative allyl transfer pathway (see 30 f 32 f 19 or
20).
validated and perhaps more tightly focused. The drawing of an
SAR map has begun, although at a very preliminary and still
confusing level. Already it has been found that C10 must appear
in oxidized form. For instance, compound 10 is inactive. It does
seem that analogue synthesis could well lead to improved
compounds in that products 28 and 38 already seem to exceed
jiadifenin in their neurite-promoting activity. However, the large
question in this area still remains to be addressed: can a
nonpeptidyl small molecule cell-permeable compound of the
type described here find application in clinical settings? This
and other related biological questions are receiving continuing
attention with a view toward identifying candidate structures
for advancement into preclinical and, eventually, clinical
evaluations. These goals tend to be among the central missions
of our laboratory.
Acknowledgment. This work was supported by a Grant from
the National Institutes of Health (AI16943 and CA28824).
D.A.C. gratefully acknowledges the Novartis Foundation and
Roche Research Foundation for a postdoctoral fellowship.
Y.S.C. is the recipient of an Army Prostate Cancer Grant. We
thank Dr. Louis Todaro (Hunter College, New York) for X-ray
structure analyses and Dr. George Sukenick, Ms. Sylvi Rusli,
and Ms. Anna Dudkina (NMR Core Facility, Sloan-Kettering
Institute, CA02848) for mass spectral analyses.
Supporting Information Available: Experimental procedures
and characterization data for compounds 8-22, 25-31, 34, and
36-39 and CIF file for compound 25. This material is available
free of charge via the Internet at http://pubs.acs.org.
At the biological level, we note that the neurotrophic activity
claimed by the discoverers of the compound has been well
JA056980A
9
1022 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006