Total synthesis of (+)-herboxidiene/GEX 1A

Article abstract of DOI:10.1039/c7ob00072c

A total synthesis of (+)-herboxidiene/GEX 1A has been accomplished from (R)- and (S)-lactate esters in a highly efficient manner. Key steps of the synthesis involve substrate-controlled titanium-mediated aldol reactions from chiral lactate-derived ethyl ketones, an oxa-Michael cyclization, an Ireland-Claisen rearrangement, and a Suzuki coupling. Furthermore, computational studies of the oxa-Michael reaction have unveiled the dramatic influence of intramolecular hydrogen bonds on the stereochemical outcome of such cyclizations, whereas biological analyses have clearly proved the important cytoxicity of (+)-herboxidiene/GEX 1A.

Full text of DOI:10.1039/c7ob00072c

View Article Online
View Journal
Organic &
Biomol ec ular
Chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: P. Romea, F. Urpí,
A. Gómez-Palomino, M. Pellicena, K. Krämer, G. Aullón and J. M. Padron, Org. Biomol. Chem., 2017, DOI:
1
0.1039/C7OB00072C.
Volume 14 Number
1
7
January 2016 Pages 1–372
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
Organic &
Biomolecular
Chemistry
www.rsc.org/obc
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 1477-0520
COMMUNICATION
Takeharu Haino et al.
Solvent-induced emission of organogels based on tris(phenylisoxazolyl)
benzene
rsc.li/obc

Page 1 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
Total Synthesis of (+)-Herboxidiene/GEX 1A
†
†
†
,†
Alejandro GómezꢀPalomino, Miquel Pellicena, Katrina Krämer, Pedro Romea,*
,†
§
‡
Fèlix Urpí,* Gabriel Aullón, and José M. Padrón
†
Departament de Química Inorgànica i Orgànica, Secció de Química Orgànica, and Institut de
Biomedicina (IBUB), Universitat de Barcelona, Carrer Martí i Franqués 1ꢀ11, 08028 Barcelona,
Catalonia, Spain.
§
Departament de Química Inorgànica i Orgànica, Secció de Química Inorgànica, Universitat de
Barcelona, Carrer Martí i Franqués 1ꢀ11, 08028 Barcelona, Catalonia, Spain.
‡
BioLab, Instituto Universitario de BioꢀOrgánica “Antonio González” (IUBOꢀAG), Centro de
Investigaciones Biomédicas de Canarias (CIBICAN), Universidad de La Laguna, 38206 La Laguna,
Spain.
pedro.romea@ub.edu; felix.urpi@ub.edu
Abstract. A total synthesis of (+)ꢀherboxidiene/GEX 1A has been accomplished from
(
R)ꢀ and (S)ꢀlactate esters in a highly efficient manner. Key steps of the synthesis
involve substrate–controlled titanium–mediated aldol reactions from chiral lactate–
derived ethyl ketones, an oxaꢀMichael cyclization, an IrelandꢀClaisen rearrangement,
and a Suzuki coupling. Furthermore, computational studies of the oxaꢀMichael reaction
have unveiled the dramatic influence of intramolecular hydrogen bonds on the
1

Organic & Biomolecular Chemistry
Page 2 of 64
View Article Online
DOI: 10.1039/C7OB00072C
stereochemical outcome of such cyclizations, whereas biological analyses have clearly
proved the important cytoxicity of (+)ꢀherboxidiene/GEX 1A.
INTRODUCTION
In 1992, the search for novel herbicides derived from Streptomyces chromofuscus
A7847 led to the isolation of (+)ꢀherboxidiene (1 in Figure 1), a secondary metabolite
1
with exceptional phytotoxic activity toward a wide range of broadleaf weeds. Further
studies showed that herboxidiene could also upꢀregulate the gene expression of the lowꢀ
2
densityꢀlipoprotein receptor, which reduces cholesterol in blood, and it was later
identified as the major component, named GEX 1A, of a family of new antitumor
3
,4
antibiotics derived from Streptomyces sp.
The structure of herboxidiene/GEX 1A was established in 1997 by comprehensive Xꢀ
5
ray crystallographic analysis and chemical studies, and was finally confirmed through
6
total synthesis by Kocieński in 1999. As shown in Figure 1, herboxidiene is a mediumꢀ
size polyketide with nine chiral centers in which a conjugated diene brings together a
tetrahydropyran core and a polyoxygenated fragment containing an epoxide, a methyl
ether, and an alcohol.
Figure 1. (+)ꢀHerboxidiene or GEX 1A (1)
The combination of such diverse biological activity, low abundance, and enticing
structural features has aroused considerable interest and has prompted a number of
7
8
synthetic studies, which nowadays involve eight total syntheses, one formal synthesis,
2

Page 3 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
9
and a plethora of related studies. Thus, it is of little surprise that the structure of
herboxidiene/GEX 1A caught our attention as the basis of a possible challenge to the
efficiency of certain stereoselective transformations. At the time we started work, the
6b
7a
7b
7c
syntheses reported by Kocieński, Banwell, Panek, and Forsyth had already
established a successful endꢀgame based on substrateꢀcontrolled epoxidations of
homoallylic alcohols that were prepared by coupling of two fragments of similar size
and complexity. Taking advantage of these precedents, we envisaged that (+)ꢀ
herboxidiene/GEX 1A (1) might be synthesized from the hydroxy ester 2, which could
result from a palladiumꢀmediated crossꢀcoupling reaction of the vinyl iodide 3 and the
alkyne 4 (Scheme 1). In turn, these advanced intermediates could be obtained through
the appropriate manipulation of two aldol adducts, 5 and 6, from the lactateꢀderived
ethyl ketones 7 and 8 respectively. Thus, all the chiral centers would be installed from
highly stereoselective substrateꢀcontrolled reactions, which eventually relied on
1
0,11
titanium–mediated aldol reactions developed by our group.
On these grounds, our
strategy also aimed to adhere to the principles predicated by the economy in synthesis,
which meant that much of the reagents would end up in the final target (atom
1
2
economy), without using protecting groups, and reducing the number of concession
1
3
14
steps (step and redox economy).
According to these principles, we described a total synthesis of herboxidiene a few
7e
years ago. Despite the efficiency of our approach, the construction of the
tetrahydropyran core was puzzling and required harsh conditions, so we have been
working for a better understanding of the stereocontrol of oxaꢀMichael cyclizations and
the improvement of the synthesis of the pyran heterocycle. Both challenges have been
accomplished, so herein we report a full account of our synthetic endeavors, including
3

Organic & Biomolecular Chemistry
Page 4 of 64
View Article Online
DOI: 10.1039/C7OB00072C
theoretical studies of the oxaꢀMichael cyclization of hydroxy esters as well as analyses
of the biological activity of (+)ꢀherboxidiene/GEX 1A and some precursors.
Scheme 1. Retrosynthetic Analysis of (+)ꢀHerboxidiene/GEX 1A (1)
4

Page 5 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
RESULTS AND DISCUSSION
1. Synthesis Plan
The success of the synthesis plan outlined in Scheme 1 relied heavily on the titaniumꢀ
mediated substrate–controlled aldol reactions starting from the chiral ethyl ketones 7
10,11
and 8.
Indeed, the carbon backbone of these simple lactateꢀderived ketones was to
be fully incorporated into herboxidiene/GEX 1A, they embodied the chiral centers from
which the whole stereochemical architecture would be built, and they also contained the
only protecting groups used in the synthesis.
Beyond these preliminary transformations, a few steps demanded special attention.
Since the stereoselective synthesis and reduction of 2,4ꢀantiꢀ4,5ꢀsyn aldol (5) depended
on highly reliable methods, the construction of the cisꢀpyran ring through
alkoxycarbonylation of the resulting diol or cross metathesis followed by an oxaꢀ
Michael cyclization emerged as the most challenging step of the predicted sequence for
the synthesis of the C1–C9 fragment (3). In turn and for similar reasons, we expected
that diol 6 could be easily obtained through a oneꢀpot aldolꢀreduction reaction, whereas
the regioselective conversion of both hydroxyl groups into a methyl ether and an ester,
together with the ensuing IrelandꢀClaisen rearrangement, were anticipated to be major
hurdles in the process leading to the synthesis of the C10–C19 fragment (4). Eventually,
the delicate assembly of such advanced intermediates might be achieved by palladium
coupling of the iodo alkene 3 and an alkenyl intermediate derived from alkyne 4; while
the endꢀgame of the synthesis would involve the vanadiumꢀmediated epoxidation of
hydroxy ester 2 and hydrolysis of the resultant ethyl ester.
5

Organic & Biomolecular Chemistry
Page 6 of 64
View Article Online
DOI: 10.1039/C7OB00072C
2. Synthesis of Fragment C1–C9 (3)
2.1. Initial steps. Stereoselective aldol reaction
The synthesis of fragment C1–C9 (3) began with the preparation of the αꢀbenzyloxy
15
ethyl ketone 7 from commercially available ethyl (S)ꢀlactate. As shown in Scheme 2,
treatment of ethyl (S)ꢀlactate with pyrrolidine without solvent produced the
corresponding αꢀhydroxy propanamide 9, which was converted into αꢀbenzyloxy amide
10 with a 92% yield under phaseꢀtransfer conditions. Further acylation of EtMgBr with
amide 10 provided the enantiomerically pure chiral ketone 7 with a 93% yield at gram
scale. With a ready supply of ketone 7 to hand, the substrateꢀcontrolled Lewis acidꢀ
1
6
mediated aldol addition of its titanium enolate to 3ꢀbutenal was then undertaken. As
expected, the 2,4ꢀantiꢀ4,5ꢀsynꢀaldol adduct 5 was isolated as a single diastereomer (dr ≥
97:3) in 84% yield and was subsequently reduced to 1,3ꢀanti diol 11 using the Evansꢀ
17
ChapmanꢀCarreira protocol
with an excellent overall yield (96%) and
diastereoselectivity (dr 94:6). Thus, the scene was set to tackle the construction of the
tetrahydropyran ring.
Scheme 2. First steps towards C1–C9 fragment
6

Page 7 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
2.2. Synthesis of the tetrahydropyran ring
2.2.1. First approach: alkoxycarbonylation
As mentioned above, we envisaged that the synthesis of the pyran core of fragment C1–
C9 might be achieved by alkoxycarbonylation of diol 11 or oxaꢀMichael cyclization of a
1
8
derived α,βꢀunsaturated ester.
The intramolecular alkoxycarbonylation of δꢀhydroxy alkenes promoted by
palladium(II) complexes has frequently been exploited in the synthesis of cisꢀpyran
rings with a pending carboxylic ester, since the two functional groups can be
simultaneously installed in a single step, with moderate to high yields and excellent
1
9 , 20
stereocontrol.
fragment C1–C9 relied on the intramolecular alkoxycarbonylation of diol 11.
Application of previously reported conditions, based on the catalysis by PdCl and the
use of CuCl as cooxidant under a CO atmosphere, afforded a 94:6 diastereomeric
Therefore, our first approach toward the tetrahydropyran ring in
2
1
2
2
2
2
mixture of the desired ester 12 in 33% yield (Scheme 3). Unfortunately and in spite of
our efforts, we were unable to improve this result. Instead, other cooxidants or solvents
produced complex reaction mixtures from which 12 was isolated in low and non–
reproducible yields. Presumably, the reason of this failure is due to the presence of a
nonꢀprotected alcohol at the homoallylic position.
Scheme 3. Alkoxycarbonylation of δ–hydroxy alkene 11
7

Organic & Biomolecular Chemistry
Page 8 of 64
View Article Online
DOI: 10.1039/C7OB00072C
2.2.2. Second approach: metathesis and oxaꢀMichael cyclization of an α,βꢀunsaturated
ester
The poor results achieved in the direct alkoxycarbonylation reaction led us to explore a
twoꢀstep sequence based on a cross metathesis of diol 11 with ethyl acrylate followed
by an oxaꢀMichael cyclization of the resultant α,βꢀunsaturated ester. This
intramolecular reaction was expected to produce mainly the desired cisꢀpyran ring since
all the substituents lie in equatorial positions. As planned, the initial cross metathesis
promoted by the HoveydaꢀGrubbs ruthenium complex produced the α,βꢀunsaturated
ethyl ester 13 cleanly, with excellent diastereoselectivity and yield (E/Z ≥ 97:3, 90%
yield), but the subsequent intramolecular oxaꢀMichael addition proved troublesome. An
2 2
exploratory attempt under acidic conditions (20 mol % CSA in CH Cl at 35 °C,
overnight) failed and the starting material was recovered unaltered. In contrast, catalytic
amounts of tꢀBuOK triggered the oxaꢀMichael cyclization at 0–20 °C and gave the
apparent kinetic transꢀpyran 14t diastereomer as the major product (entries 1 and 2 in
Table 1). Unexpectedly, higher temperatures and longer reaction times hardly altered
these results and roughly equimolecular mixtures of the 2,6ꢀcisꢀ and the 2,6ꢀtransꢀpyran
diastereomers, 14c and 14t respectively, were isolated in an 80% overall yield (entries 3
and 4 in Table 1). Fuwa faced a similar problem in the synthesis of aspergillides A and
B and observed that diastereomerically pure cisꢀpyran rings could be prepared by
heating the corresponding conjugated hydroxy esters in the presence of a large amount
2
3,24,25
of a nonꢀnucleophilic base such as DBU.
The application of such conditions to our
system did not yield such excellent results, but the 2,6ꢀcisꢀpyran 14c was obtained as
the major diastereomer. Indeed, treatment of 13 with two equivalents of DBU in toluene
at 100 °C gave a 1.3:1 mixture of 14c/14t (entry 5 in Table 1). The diastereoselectivity
was further increased by using ten equivalents of base, while the yield was slightly
8

Page 9 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
reduced over long reaction times (compare entries 5–8 in Table 1). Finally, the use of
xylene and a higher temperature produced a certain improvement of the
diastereoselectivity but the yield slightly decreased (compare entries 8 and 9 in Table
1). Therefore, the most appropriate conditions involved heating of the hydroxy ester 13
in toluene in the presence of ten equivalents of DBU for 5 h, which triggered the oxaꢀ
Michael cyclization and produced an inseparable 1.8:1 mixture of tetrahydropyrans 14
with an 80% overall yield.
Table 1. oxaꢀMichael cyclization of hydroxy ester 13
a
b
entry base
equiv solvent
temperature (°C)
time (h)
dr (14c:14t)
1 : 2
yield (%)
1
2
3
4
5
6
7
8
tꢀBuOK
0.2
0.2
0.2
0.2
2
THF
0
1
76
85
79
81
67
80
78
73
70
tꢀBuOK
tꢀBuOK
tꢀBuOK
DBU
THF
20
15
48
15
2
1 : 1.7
1 : 1.1
1 : 1.1
1.3 : 1
1.8 : 1
1.8 : 1
1.8 : 1
2.1 : 1
THF
50
THF
65
toluene
toluene
toluene
toluene
xylene
100
100
100
100
135
DBU
10
10
10
10
5
DBU
15
24
2
DBU
9
DBU
a
1
Established by H NMR analysis
b
Overall isolated yield
Once the synthesis of the pyran core was completed, a tinꢀfree BartonꢀMcCombie
removal of C5–OH in 14 provided diastereomerically pure 2,6ꢀcis deoxygenated
derivative 15c with a 54% yield and the 2,6ꢀtrans counterpart 15t in 10% yield, which
9

Organic & Biomolecular Chemistry
Page 10 of 64
View Article Online
DOI: 10.1039/C7OB00072C
could be separated by column chromatography (Scheme 4). Remarkably, the 15c/15t
diastereomeric ratio was higher than that of the starting hydroxy pyrans 14c/14t, which
indicated that the reaction conditions caused a certain resolution. Finally, comparison of
7c
the physical and spectroscopic data of pure 15c with those reported in the literature
confirmed the desired 2,6ꢀcis configuration of the tetrahydropyran core.
Scheme 4. Deoxygenation of 14
10

Page 11 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
2.2.3. Theoretical studies on oxaꢀMichael cyclizations
The stereochemical outcome of 6ꢀexoꢀtrig oxaꢀMichael cyclizations of α,βꢀunsaturated
2
6
carbonyl compounds depends on their structure and the reaction conditions. Indeed,
conjugated ketones and thioesters, but no esters, usually participate in Brønsted acidꢀ
2
7,28
mediated cyclizations to provide 2,6ꢀcisꢀsubstituted tetrahydropyran derivatives.
In
turn, parallel baseꢀpromoted kineticallyꢀcontrolled additions in conjugated ketones as
well as esters give 2,6ꢀtransꢀsubstituted tetrahydropyrans, while the corresponding 2,6ꢀ
cisꢀcounterparts are available through
a
retro/recyclization sequence under
thermodynamic conditions.
Therefore, the equatorial position of all the substituents in the cisꢀpyran ring 14c seemed
to warrant it being the main diastereomer at moderate to high temperatures and long
reaction times. Unfortunately, the results shown in Table 1 did not meet these
expectations. Looking for the reasons for such an unexpected result, a close examination
of the literature uncovered a few precedents.
Bates observed during the synthesis of clavosolide A that the success of the oxaꢀ
Michael cyclizations of the conjugated esters 16 depended on the substitution at the
2
9
allylic position. Preliminary studies of the model ester 16a (R: H) exclusively afforded
the desired 2,6ꢀcisꢀtetrahydropyran 17a in 53% yield, whereas the same reaction with
the ester 16b (R: Me) failed entirely (eq 1 in Scheme 5). Thus, the pyran ring had to be
synthesized through an intramolecular oxaꢀMichael reaction of a conjugated alkyl
ketone parent. A similar cyclization was examined by Evans in the total synthesis of
3
0
(
+)ꢀmiyakolide. Treatment of the conjugated ester 18 with tꢀBuOK gave 19 as a 1.5:1
mixture of tetrahydropyrans (eq 2 in Scheme 5). Unexpectedly, this cyclization turned
out to be nonꢀreversible, but the protection of the secondary alcohol followed by
resubmission of the resultant mixture to the former basic conditions produced the cisꢀ
11

Organic & Biomolecular Chemistry
Page 12 of 64
View Article Online
DOI: 10.1039/C7OB00072C
pyran 20 as a single diastereomer (eq 2 in Scheme 5). In turn, Paterson reported that the
oxaꢀMichael cyclization under a range of basic conditions of conjugated ester 21
3
1
provided almost exclusively the 2,6ꢀtrans tetrahydropyran 22 (eq 3 in Scheme 5). All
attempts to generate the desired cisꢀpyran isomer proved unsuccessful. Instead, a
parallel cyclization of conjugated thioester 23 under acidic conditions led to the
selective formation of the targeted 2,6ꢀcis tetrahydropyran 24 (eq 4 in Scheme 5).
2
9
30
31
Scheme 5. oxaꢀMichael cyclizations of dihydroxy unsaturated esters by Bates, Evans, and Paterson
3
2
All together, these and other results show that the structure of δ–hydroxy conjugated
esters has a tremendous influence on such cyclizations. Interestingly, the apparent lack
of reversibility of transꢀpyran 19 and the easy conversion into cisꢀpyran 20 (eq 2 in
Scheme 5) by simple protection of the free hydroxyl suggested that the low
diastereoselectivity achieved in the cyclization of 13 could be due to the presence of a
nonꢀprotected alcohol, as well as the failure to access the thermodynamically favoured
cisꢀpyran from 21 could be also ascribed to the C5 alcohol (eq 3 in Scheme 5).
12

Page 13 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
Keeping this hypothesis in mind, we assessed the relative stability of model 2,6ꢀcis and
,6ꢀtrans tetrahydropyran enolates I and II represented in Figure 2. As expected, via
2
theoretical calculations we found out that both the cis-II and trans-II pyrans lacking a
nonꢀprotected alcohol basically exist as chairꢀlike conformers, since the boatꢀlike
counterparts are too unstable to be considered (see Figure 2b). Meanwhile, the setꢀup is
completely different for the pyrans I (see Figure 2a). The most stable conformation of
2,6ꢀtrans diastereomer trans-I adopts the chairꢀlike form, although the boatꢀlike
conformer is close in energy as it is stabilized by an intramolecular hydrogen bond.
Such an interaction is much more important for 2,6ꢀcis diastereomer cis-I to the extent
that the corresponding boatꢀlike conformer becomes the most stable. All together, these
calculations reveal the dramatic influence of intramolecular hydrogen bonds between
the C5ꢀOH and the oxygen in the heterocyclic ring on the conformational equilibria of
the pyrans I.
13

Organic & Biomolecular Chemistry
Page 14 of 64
View Article Online
DOI: 10.1039/C7OB00072C
Figure 2. Relative stability of 2,6ꢀcis and 2,6ꢀtransꢀtetrahydropyran enolates
Once established the crucial role of intramolecular hydrogen bonds on the
thermodynamic stability of the 2,6ꢀcis and 2,6ꢀtrans pyrans I, cis-I and trans-I
respectively (see Figure 2a), we focused our attention on the cyclization step. Initially,
we located the transition states leading to chairꢀ and boatꢀlike conformations of both
pyrans by lengthening the new C–O bond. Vibrational analysis (first order saddle point)
of the resultant molecular structures confirmed them to be true transition states. Finally,
the structures of the acyclic precursors were also ascertained.
14

Page 15 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
Those calculations indicated that the transition states towards the chairꢀlike conformers
–1
of cis- and trans-I are located more than 9 kcal mol above those of the corresponding
boatꢀlike conformers. Therefore, the oxaꢀMichael cyclization mainly proceeds through a
proꢀboatꢀlike transition state. As represented in Figure 3, the acyclic alkoxide A yield
A
the desired pyran cis-I through a proꢀboat transition state TS ; whereas a related but
more stable transition state TS available from alkoxide B is responsible for the pyran
B
trans-I counterpart. Once again, the strong interaction between the C5–OH and the
anionic oxygen determines the boatꢀlike geometry of both transition states. All together,
these findings suggest that the stereochemical outcome of the oxaꢀMichael cyclization
described in Table 1 can be understood through the CurtinꢀHammett model. Indeed, the
pyran trans-I is obtained via a transition state lower in energy than that leading to the
more stable pyran cis-I (Figure 3), so the experimental conditions govern the
stereochemical outcome of the oxaꢀMichael cyclization of such dihydroxy
α,βꢀ
unsaturated esters: low temperatures and catalytic amounts of base favor the formation
of the 2,6ꢀtrans pyran trans-I (kinetic control); whereas the 2,6ꢀcis pyran cis-I is the
main product at high temperatures and with excess of base (thermodynamic control).
15

Organic & Biomolecular Chemistry
Page 16 of 64
View Article Online
DOI: 10.1039/C7OB00072C
Figure 3. oxaꢀMichael cyclization of a dihydroxy α,βꢀunsaturated ester
2.2.4. Third approach: metathesis and oxaꢀMichael cyclization of an α,βꢀunsaturated
amide
Despite the successful synthesis of the advanced intermediate 15c (see Scheme 4), the
modest selectivity of the baseꢀmediated oxaꢀMichael cyclization of the α,βꢀunsaturated
ester 13 led us to explore an alternative procedure for the preparation of 15c. In this
context, a thorough search for other cis selective methods uncovered the cyclization of
2
8a–c
conjugated thioesters or amides under acidic conditions developed by Fuwa.
Especially, we were attracted by the concise synthesis of 2,6ꢀcisꢀdisubstituted
16

Page 17 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
tetrahydropyrans by means of a domino olefin metathesis and oxaꢀMichael cyclization
3
3
28b,c
from 2,5ꢀdimethylpyrrole amides catalyzed by Brønsted acids,
the desired substrate in a rapid, efficient, and selective sequence.
which might yield
Unfortunately, the direct cross metathesis of the alkenediol 11 and 2,5ꢀdimethylpyrrole
acrylamide 25 proved to be troublesome and the required conjugated amide 26 was
isolated in poor yields (Scheme 6). Considering that the nonꢀprotected 1,3ꢀdiol in 11
could be the cause for such a failure, it was protected as an acetonide. To our pleasure,
the acetonide 27 turned out to be a suitable substrate for the cross metathesis and the
reaction with acrylamide 25 gave the conjugated amide 28 with an 80% yield under
standard conditions (Scheme 6). Then, we assessed the conversion of the conjugated
amide 28 into the ethyl ester 14c, which involved the removal of the acetonide
protecting group, the stereocontrolled oxaꢀMichael cyclization, and the final conversion
of the pyrrole amide into the corresponding ethyl ester in a single step. After testing a
wide range of experimental conditions, it was found that the simple stirring of a solution
of amide 28 and CSA (15 mol %) in 1,2ꢀdichloroethane/EtOH at room temperature
produced the deprotection of the diol and triggered the ring formation. Finally, heating
the resultant solution furnished the desired cis pyran ethyl ester 14c and 2,5ꢀ
dimethylpyrrole amide 29c (19% and 61% yield respectively, see Scheme 6), which was
cleanly converted into 14c by treatment with EtONa in a 92% yield. Importantly, no
traces of the trans pyran ethyl ester 14t were observed, which demonstrates the dramatic
influence of the acid/basic conditions on the steric outcome of the oxaꢀMichael
cyclizations leading to tetrahydropyran heterocycles. In summary, the stereocontrolled
synthesis of cis ethyl ester 14c from alkenediol 11 had been completed in four steps and
a 53% overall yield.
17

Organic & Biomolecular Chemistry
Page 18 of 64
View Article Online
DOI: 10.1039/C7OB00072C
Scheme 6. Metathesis and oxaꢀMichael cyclization of a 2,5ꢀdimethyl pyrrole amide
2.2.5. Fourth approach: oxaꢀMichael cyclization of an α,βꢀunsaturated ester,
deoxygenation of the resultant pyrans, and reequilibration
While it is true that the acidꢀmediated cyclization of pyrroleꢀamide 28 afforded the cis
pyran ring stereoselectively, the need for the acetonide protecting group pushed us to
examine again our second approach. Since the theoretical calculations and our own
experience had identified the C5 alcohol as the reason for the lack of stereocontrol of
the oxaꢀMichael cyclization, we envisioned that the reequilibration of the trans pyran
isomer could be attempted once that C5–OH was removed.
In accordance to this hypothesis, the cyclization of the dihydroxy ester 13 was carried
out in the presence of a 20 mol % of tꢀBuOK (see entry 2 in Table 1). The resultant
mixture of pyrans 14 was not purified by column chromatography and was directly
submitted to the deoxygenation sequence described in Scheme 4. Thus, we were
delighted to observe that the C5–deoxygenated cis pyran ethyl ester 15c was isolated in
a 57% yield and the trans isomer 15t in an 12% yield (Scheme 7). As anticipated, the
trans pyran was easily isomerized into 15c by treatment with tꢀBuOK, which highlights
the crucial role of the C5–OH in the stereochemical outcome of the oxaꢀMichael
18

Page 19 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
cyclization (Scheme 7). Therefore, this new approach turned out to be most efficient
and allowed us to obtain the cis pyran ethyl ester 15c from diol 11 in five steps and an
overall 59% yield under mild conditions and minimizing the number of
chromatographic purifications.
Scheme 7. oxaꢀMichael cyclization and deoxygenation of the dihydroxy ester 13
2.3. Final steps
Once the synthesis of the pyran core was completed, a twoꢀstep sequence consisting of
the removal of the benzyl protecting group of 15c and Swern oxidation of the resultant
alcohol gave the ketone 30 in an 87% yield (Scheme 8). Finally, a Takai olefination of
30 led to the iodo alkene 3 with high stereocontrol of the double bond geometry of the
double bond (E/Z 90:10) in 58% yield (Scheme 8). Hence, the C1–C9 fragment, iodo
alkene 3, had been synthesized in thirteen steps and 20% yield from ethyl (S)ꢀlactate.
19

Organic & Biomolecular Chemistry
Page 20 of 64
View Article Online
DOI: 10.1039/C7OB00072C
Scheme 8. Final steps towards C1–C9 fragment (3)
3. Synthesis of Fragment C10–C19 (4)
3.1. Initial steps. Stereoselective aldol-reduction transformation
The synthesis of fragment C10–C19 (4) hinged on a highly stereoselective oneꢀpot
3
4
aldolꢀreduction sequence from chiral α–silyloxy ketones developed by our group. The
starting αꢀtertꢀbutyldimethylsilyloxy ethyl ketone 8 was prepared from isobutyl (R)ꢀ
lactate; simple stirring of this ester with pyrrolidine produced the corresponding αꢀ
hydroxy propanamide that was protected under standard conditions to give the αꢀ
1
5
silyloxy amide 31 in a 94% yield (Scheme 9). Further acylation of EtLi with 31 gave
enantiomerically pure chiral ketone 8 in an 84% yield at gram scale. Then, the TiCl
mediated aldol addition of 8 to methacrolein and reduction of the resultant sixꢀ
membered chelated aldolate with LiBH gave diastereomerically pure 1,3ꢀdiol 32 in
1% yield (Scheme 9).
4
ꢀ
4
8
20

Page 21 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
Scheme 9. First steps towards C10–C19 fragment
3.2. Regioselective modification of diol 32
Having found a highly efficient route to access the diol 32 containing three of the final
four chiral centers of fragment C10–C19 (4), we next evaluated the regioselective
conversion of C15 and C17 hydroxyl groups to a propanoic ester and a methyl ether
respectively. Since the alcohol at C15 is sterically less hindered than that at C17, a
number of procedures were investigated in order to regioselectively acylate the C15–
OH. Preliminary experiments proved that propanoyl chloride was too reactive to be
selective even at low temperatures, so we focused our attention on the use of propanoic
anhydride under basic conditions. As anticipated, the C15–OH reacted faster than the
C17–OH and the hydroxy ester 33 was obtained as the major product (see entries 1–5 in
Table 2). Nevertheless, the diester 34 was also formed and the regioselective ratio 33/34
turned out to be modest at high conversions (entries 3–5 in Table 2). Looking for a more
selective transformation, we found that the Lewis acidꢀmediated acylation reported by
3
5
Clarke produced the desired hydroxy ester 33 in a highly regioselective manner by
using 10 mol % of CeCl ·7H O in CH Cl at room temperature (see entry 6 in Table 2).
The acylation proved to be rather slow in THF or in the presence of other Lewis acids
such as YbCl ·6H O, whereas higher loadings of the catalyst were also detrimental
compare entries 6–9 in Table 2). Therefore, the regioselective acylation of diol 32 was
3
2
2
2
3
2
(
21

Organic & Biomolecular Chemistry
Page 22 of 64
View Article Online
DOI: 10.1039/C7OB00072C
best accomplished by using propanoic anhydride under CeCl
entry 6 in Table 2).
3
·7H
2
2 2
O catalysis in CH Cl
(
Table 2. Regioselective acylation of diol 32
a
entry
catalyst (mol %)
temperature (°C)
time (h)
2.5
8
solvent
CH Cl
ratio 32/33/34
b
1
2
3
4
5
6
7
8
DMAP (10)
0
37 : 58: 5
2
2
b
DMAP (10)
0
CH
2
Cl
2
2
2
2
2
14 : 76 : 10
3 : 56 : 41
9 : 72 : 19
3 : 70 : 27
3 : 87 :10
59 : 39 : 2
61 : 35 : 4
42 : 55 : 3
b
DMAP (10)
0
15
CH Cl
2
b
DMAP (20)
0
1.5
3.5
48
CH
2
Cl
b
DMAP (20)
0
CH Cl
2
CeCl
3
·7H
2
O (10)
20
20
20
20
CH
THF
CH Cl
CH Cl
2
Cl
CeCl ·7H O (10)
48
3
2
YbCl
3
·6H
2
O (10)
96
2
2
2
9
CeCl ·7H O (20)
24
3
2
2
a
1
Established by H NMR analysis of the crude mixture.
b
3
1.5 Equivalents of Et N were added to the reaction mixture.
With the most appropriate conditions for the regioselective acylation of C15–OH at
hand, treatment of the resultant mixture of 33/34 with Me O·BF produced the ester 35,
3
4
with a 66% overall yield, in which all the stereocenters and the functional groups are
satisfactorily installed (Scheme 10). Therefore, the stage was set to tackle the Irelandꢀ
Claisen rearrangement that should lead to an advanced intermediate toward the
fragment C10–C19 (4).
Scheme 10. Selective functionalization of diol 32
22

Page 23 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
3.3. Ireland-Claisen rearrangement and final steps
The close dependence of the configuration of the new stereocenters on the geometry of
the silyl ketene acetals in IrelandꢀClaisen rearrangements demands quantitative and
highly stereoselective enolization procedures. Wideꢀranging studies of this step have
established that strong bases in THF at low temperatures give access to E enolates,
whereas the simple addition of a cosolvent such as HMPA or DMPU favors the Z
geometry. So, these enolates can be trapped with silyl chloride reagents and further
heating of the resultant silyl ketene acetals triggers the [3,3] sigmatropic rearrangement
3
6
through a sixꢀmembered chairꢀlike transition state.
Despite the apparent simplicity of such a transformation, the literature is plagued by an
excessive variety of experimental conditions that affect the transformation, including
the base, the stoichiometry, the reaction times, the temperatures, and the way the
reagents are mixed. Therefore, a comprehensive analysis of such variables was
undertaken. Following thorough optimization, it was finally found that treatment of the
ester 35 with four equivalents of LiHMDS in THF/DMPU at –78 °C gave the
corresponding Z(O) lithium enolate, which was cleanly trapped with TBSCl to produce
silyl ketene acetal 36 (Scheme 11). Fortunately, this intermediate rearranged at low
temperature to afford the ester 37 with absolute stereocontrol of the configuration of
C12. The resultant mixture turned out to be difficult to purify, so it was reduced with
LiAlH to produce the pure alcohol 38 as a single diastereomer with a 90% overall
4
‡
yield.
23

Organic & Biomolecular Chemistry
Page 24 of 64
View Article Online
DOI: 10.1039/C7OB00072C
Scheme 11. Final steps towards C10–C19 fragment (4)
Having completed the most challenging steps en route to fragment C10–C19, it only
remained for us to introduce the terminal alkyne through the OhiraꢀBestmann
3
7
procedure. Thus, the required aldehyde 39 was prepared by Swern oxidation of the
alcohol 38 (Scheme 11). This aldehyde proved to be very sensitive and easy to
epimerize, so it was not purified further and was immediately converted into the alkyne
4
by treatment with dimethyl (1ꢀdiazoꢀ2ꢀoxopropyl)phosphonate. Unfortunately,
standard conditions (K CO in methanol at room temperature or 0 °C) produced
2
3
epimerization of C12. Thus, we were pleased to observe that the use of MeONa at low
temperature avoided such drawbacks and permitted the isolation of the pure alkyne 4 in
3
8
an excellent 90% overall yield. This culminated the synthesis of the C10–C19
fragment from isobutyl (R)ꢀlactate in ten steps and 34% overall yield.
4. Coupling of fragments 3 and 4
24

Page 25 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
With both fragments to hand, we faced the last steps of the synthesis. A successful
Negishi coupling of an iodo alkene and an alkyne structurally close to our fragments in
3
a total synthesis of milbemycin β encouraged us to try this reaction first to assemble 3
3
9
and 4. Unfortunately and despite intensive efforts, we were unable to find any suitable
conditions and the expected coupling product was only isolated in very low yields.
4
0
Hydrozirconation of alkyne 4 with the Schwartz reagent, transmetallation of the
resultant zirconocene with zinc bromide, and palladiumꢀcatalyzed coupling with the
iodo alkene 3 afforded traces of the desired ester 2 (see Scheme 1). Instead, the terminal
alkene 40 derived from the alkyne 4 was isolated in 94% yield (Scheme 12). Changes of
the zinc salt, longer reactions times, and different reaction conditions did not alter this
result. The required transmetallation Zr–Zn or ZnꢀPd was probably unsuccessful and
thus the assembly failed.
25

Organic & Biomolecular Chemistry
Page 26 of 64
View Article Online
DOI: 10.1039/C7OB00072C
Scheme 12. Attempt on Negishi coupling of fragments 3 and 4
Therefore, it was clear that an alternative coupling method was required. Since Forsyth
had already used a Suzuki coupling to forge the whole backbone of herboxidiene/GEX
7c
1
A, we decided to assay a similar approach using fragments 3 and 4. Nevertheless and
unlike that precedent, we envisaged that the assemblage might proceed at room
temperature, so we decided to use mild conditions to prepare the corresponding boronic
acid and to carry out the palladiumꢀcatalyzed coupling. With this aim, we considered
4
1
that the Arase hydroboration of alkynes based on the reaction with catecholborane
catalyzed by dialkyl boranes was the most appropriate way of preparing the boronic
acid 41 (Scheme 13). The reaction proved to be very reliable, so we were pleased to
observe that treatment of the alkyne 4 with catecholborane and catalytic amounts of
Cy
2
BH at room temperature and subsequent hydrolysis of the resultant boronate
produced 41 quantitatively, which was used in the next step without further purification.
26

Page 27 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
3 4
Then, Suzuki coupling of iodo alkene 3 and boronic acid 41 catalyzed by Pd(PPh )
using thallium(I) carbonate as the base gave the desired ester 2 in an excellent 91%
overall yield after stirring at room temperature overnight (Scheme 13).
Scheme 13. Suzuki coupling of fragments 3 and 4
5. End game
The ester 2 embraced the carbon backbone and the functional groups of (+)ꢀ
herboxidiene/GEX 1A with the exception of epoxide at C14–C15. So, the end game
commenced with the deprotection of the C18–OH and the stereoselective epoxidation of
the resultant bishomoallylic alcohol 42 (Scheme 14). Thereby, removal of the silyl
protecting group of 2 with diluted hydrochloric acid produced the bishomoallylic
alcohol 42 in 91% yield, whose substrateꢀcontrolled oxidation with TBHP catalyzed by
VO(acac)
hydrolysis of epoxy ester 43 was carried out at reflux under basic conditions (K
MeOH/H O) with an 82% yield. To our pleasure, such a reaction could also be
2
afforded epoxy ester 43 with a good 58% yield (Scheme 14). At first,
2
CO in
3
2
27

Organic & Biomolecular Chemistry
Page 28 of 64
View Article Online
DOI: 10.1039/C7OB00072C
completed under milder conditions by using TMSOK at room temperature to produce
+)ꢀherboxidiene/GEX 1A in 93% yield (Scheme 14). Thus, the Suzuki coupling of
(
fragments 3 and 4 afforded the advanced intermediate 2, which was finally converted
into (+)ꢀherboxidiene/GEX 1A in four additional steps and in an overall 44% yield.
In summary, the longest linear sequence leading to (+)ꢀherboxidiene/GEX 1A (1)
involves eighteen steps and an excellent 8% overall yield, the highest yield reported to
date.
Scheme 14. End game steps
6. Biological activity
The antiproliferative activity of synthetic (+)ꢀherboxidiene/GEX 1A (1) was evaluated
against the panel of human solid tumor cell lines HBLꢀ100, HeLa, SW1573, Tꢀ47D and
WiDr. In order to obtain some structureꢀactivity relationships, we selected compounds
2
, 42 and 43, which contain the carbon backbone of 1 but differ in the functional
groups. The results are given in Table 3.
28

Page 29 of 64
Organic & Biomolecular Chemistry
View Article Online
DOI: 10.1039/C7OB00072C
The most potent compound of the series was 1, which showed high potency against all
cell lines, with GI50 values in the range 15ꢀ50 nM. These results are consistent with
3
a
previous studies in other human tumor cell lines. Remarkably, compound 1 exhibited a
superior activity profile in the more drug resistant cell lines Tꢀ47D and WiDr. In
addition, the synthetic hydroxy ester 42 displayed GI50 values in the range 0.14ꢀ3.6 ꢁM,
whilst derivative 2 was inactive.
Recently, novel derivatives of (+)ꢀherboxidiene with structural modifications at the
4
2
epoxide moiety have been reported. The resulting compounds proved inactive against
cancer cell lines, thus demonstrating the role of the epoxide group in the
antiproliferative activity. Our results demonstrate for the first time that the epoxide
group, albeit essential, is not the only structural motif involved in the antiproliferative
activity. Based on the GI values we observe that the free hydroxyl group in 42 seems
5
0
to play a crucial role in the biological activity. Finally, the role of the carboxylic
acid/ester still remains elusive.
a
Table 3. Antiproliferative activity (GI ) against human solid tumor cells.
5
0
Compound
Cell line (tissue)
HBLꢀ100
breast)
HeLa
(cervix)
SW1573
(lung)
Tꢀ47D
(breast)
WiDr
(colon)
(
1
2
0.050 (± 0.005)
0.053 (±0.007)
0.027 (±0.004)
0.024 (±0.001)
0.015 (±0.002)
>100
3.6 (±1.6)
–
>100
2.2 (±0.5)
–
>100
0.91 (±0.16)
–
>100
0.14 (±0.01)
–
>100
0.21 (±0.02)
–
4
2
b
4
3
a
Values are given in ꢁM and are means of two to four experiments; standard deviation is given in
parentheses.
Not tested due to low solubility in DMSO
b
29