The total synthesis of coleophomones B, C, and D

Published on Web 05/27/2005

The Total Synthesis of Coleophomones B, C, and D

K. C. Nicolaou,* Tamsyn Montagnon,^†Georgios Vassilikogiannakis,^†and

Casey J. N. Mathison

Contribution from the Department of Chemistry and The Skaggs Institute for Chemical Biology,

The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, and

Department of Chemistry and Biochemistry, UniVersity of California, San Diego, 9500 Gilman

DriVe, La Jolla, California 92093

Received February 16, 2005; E-mail: kcn@scripps.edu

Abstract: Members of the coleophomone family of natural products all possess several intriguing and

challenging architectural features, as well as exhibit unusual biological activity. They, therefore, constitute

attractive targets for synthesis. In this Article, we describe the total synthesis of coleophomones B (2), C

(3), and D (4). The highly strained and congested 11-membered macrocycle of coleophomones B (2) and

C (3) was constructed using an impressive olefin metathesis reaction. Furthermore, both of the requisite

geometric isomers of the ∆^16,17within the macrocycle could be accessed from a common precursor,

facilitating a divergence that lent the coleophomone B (2)/C (3) synthesis an unusually high degree of

efficiency. The synthesis of coleophomone D (4) confirmed that it exists as a dynamic mixture of isomeric

forms with a different aromatic substitution pattern from the other family members.

Introduction

In 1998, a Japanese patent¹was issued on behalf of the

Shionogi Pharmaceutical Co., in which the structures of three

new and unique secondary metabolites (1-3, Figure 1) were

disclosed. These naturally occurring diterpenes, at this stage

assigned only alphanumeric codes to identify them, had been

isolated from a Stachybotrys cylindrospora fungal broth. They

had become the subject of a patent due to the interesting

biological activities that they possessed, including antifungal

action and the ability to inhibit the serine protease enzyme, heart

chymase, which is responsible for converting angiotensin I to

angiotensin II.²This latter property endowed the compounds

with significant potential as leads for development programs

targeting drugs to treat hypertension and congestive heart failure.

One year later, a second patent³appeared from the same

company revealing that there was a fourth sibling, a compound

(4) which existed in a state of flux between different isomeric

forms (Scheme 1). This last member of the family to be

identified had been isolated from another Stachybotrys broth

(Stachybotrys parVispora Hughes), and it showed biological

activity analogous to that of the other congeners.

Figure 1. Structures of coleophomones A-D (1-4).

Later, in 2000, details regarding some of these interesting

fungal metabolites reached a broader audience when a drug

discovery team from Merck published an account in a main-

stream chemical journal of their isolation of coleophomones A

(1) and B (2).⁴This team, apparently unaware of the existing

patent coverage of these compounds, had isolated them from a

Coleophoma sp fungal broth produced using samples collected

in the Sierra Villuercas in Spain, hence, the rationale in naming

the isolates the coleophomones. In this case, the compounds

had been identified as the result of an antibacterial assay, the

origin of their weak antibacterial activity being traced to the

inhibition of a crucial bacterial transglycosylase enzyme. For

^†Current address: Department of Chemistry, University of Crete, L.

Knossou 300, 71409 Heraklion, Crete, Greece.

(1) Kamigakinai, T.; Nakashima, M.; Tani, H. Japanese patent JP 10101666

A2, 1998 [CAN 129:589 AN 1998:236773].

(2) Urata, H.; Kinoshita, A.; Misono, K. S.; Bumpus, F. M.; Husain, A. J.

Biol. Chem. 1990, 265, 22348-22357.

(3) Kamigaichi, T.; Nakashima, M.; Tani, H. Japanese patent JP 11158109

A2 1999 [CAN 131:72775 AN 1999: 378444.

(4) Wilson, K. E.; Tsou, N. N.; Guan, Z.; Ruby, C. L.; Pelaez, F.; Gorro-

chategui, J.; Vicente, F.; Onishi, H. R. Tetrahedron Lett. 2000, 41, 8705-

8709.

9

8872

J. AM. CHEM. SOC. 2005, 127, 8872-8888

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 1. Dynamic Isomerism Exhibited by Coleophomone D

(4a-d)

Figure 2. Retrosynthetic analysis for coleophomones B (2) and C (3).

Strategies A-D.

go a long way toward unraveling these apparently confusing

issues by confirming coleophomone D’s unusual substitution

pattern and by facilitating investigation into its unique structural

features. Thus, we set forth on a program aimed at achieving

the total syntheses of all of the known members of the

coleophomone class. Our successful accomplishment of this goal

for coleophomones B-D (2-4),^6,7using newly developed acyl

cyanide coupling technologies and a pleasing olefin metathesis

reaction, is related herein.

the sake of clarity, we would later baptize the two unnamed

Shionogi compounds 3 and 4, coleophomones C and D,

respectively.

Retrosynthetic Analysis and the Development of a

Blueprint for the Total Synthesis

Our first observation, when considering how to approach the

synthesis of this class of compounds, was that the 11-membered

macrocycle of the coleophomones A-C (1-3) should be seen

as the critical feature, for it undoubtedly presented the most

challenging test for our synthetic acumen. For reasons already

delineated above, this unusual ring is highly strained and

compacted, and, as such, it could be anticipated that it would

show some reluctance to snap shut. A particularly powerful ring

closing reaction would, therefore, need to be found capable of

rising to such a formidable challenge. Because it had been

reported that coleophomones A (1) and B (2) could be

interconverted at will,⁴it did not matter which of these two

compounds we targeted initially; however, we did hope to hit

upon a macrocycle closure strategy that would allow us

relatively easy access to both geometric isomers of the macro-

cyclic ∆^16,17double bond,⁸for therein lay the only difference

between the two targets, coleophomones B (2) and C (3).

Independent of our choice of position at which to attempt

macrocycle closure, a decision that would change over the

course of the investigation according to developing circum-

stances, three pivotal disconnections were identified, which

severed the molecule into three key fragments. These discon-

nections, not the subject of change, except in their order of

execution, as the macrocycle closure strategy evolved, were

O-alkylation, C-acylation, and C-alkylation (disconnections

A-C, respectively, Figure 2). Finally, from our initial cursory

survey, we surmised that any technologies developed to unite

these fragments in the forward sense when targeting coleoph-

omones A-C (1-3) could be concomitantly applied to the rapid

assembly of coleophomone D (4), albeit after simple alterations

in the substitution pattern of the aromatic ring portion.

Our interest was sparked in the coleophomone family of

compounds by a number of captivating features that they

possessed. First, the extremely compact and condensed nature

of the carbon framework, consisting of multiple interconnected

ring systems, gifted the coleophomones with many synthetic

challenges. The strain of an 11-membered macrocycle, exac-

erbated by the presence of 6 sp²carbon atoms, an ethereal

oxygen, a fused aromatic ring, and a bridging six-membered

carbocycle complete with a quaternary center, as found in

coleophomones A-C (1-3), posed a particularly strenuous

synthetic obstacle. Continuing with the synthetic theme, co-

leophomones B (2), C (3), and D (4) also all bear a truculent

tricarbonyl unit with a highly labile central proton (C-9). In

coleophomone A (1), this tricarbonyl unit has been fused to

the proximal aldehyde present in its precursor, coleophomone

B (2), by means of an aldol reaction, thus forming a delicate

spirocycle which endows coleophomone A (1) with yet higher

degrees of tension and strain in comparison to its siblings.

Despite this added strain, the Merck paper suggested that

coleophomones A (1) and B (2) could be interconverted at will,

an observation that was set to become pivotal to our investiga-

tions.⁴The final thread to our developing interest in the

coleophomone family arose from the structural incongruity so

blatantly exhibited by coleophomone D (4). In this interesting

compound, which also lacks the molecular tie-up of a macro-

cycle, the substitution pattern on the aromatic ring (at C-2 and

C-7) has been switched from that present in all its other siblings.

This feature seemed to be at odds with all of the obvious theories

about the biogenesis of the coleophomone family.⁵Furthermore,

coleophomone D (4), as mentioned above, exists as a complex

mixture of isomers making spectral investigation and elucidation

somewhat complicated; indeed, coleophomone D’s structure had

originally been assigned on the basis of information garnered

from synthetic derivatives not from the compound itself.³We

felt that a laboratory synthesis of coleophomone D (4) might

(6) Nicolaou, K. C.; Vassilikogiannakis, G.; Montagnon, T. Angew. Chem.,

Int. Ed. 2002, 41, 3276-3281.

(7) Nicolaou, K. C.; Montagnon, T.; Vassilikogiannakis, G. Chem. Commun.

2002, 2478-2479.

(8) For clarity, we have opted to use the same numbering system for the

coleophomones and their precursors, throughout this paper, as was used in

the last isolation paper (ref 4).

(5) Bode, J. W.; Suzuki, K. Tetrahedron Lett. 2003, 44, 3559-3563.

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8873

A R T I C L E S

Nicolaou et al.

Scheme 2. Synthesis of Mono- and Disubstituted

1,3-Cyclohexadiones 7a,b and 8a,b^a

Our first in-depth proposal involved disconnection of the

macrocycle at the ethereal linkage (Figure 2, first generation

approach: macrocycle formation using O-alkylation - discon-

nection A). This scission offered the possibility of closing the

macrocycle either by using a traditional alkylation reaction,

wherein a phenoxide nucleophile would attack an electrophilic

allylic halide thereby forming the requisite new carbon-oxygen

bond, or by a newer Tsuji-Trost-type reaction (already applied

to similar macrocyclization tasks)⁹between the previously

envisioned phenoxide, and, in this instance, an allyl carbonate-

derived palladium π-allyl complex as electrophile. We envi-

sioned that a second alkylation at C-13 (Figure 2, disconnection

C), to precede this ring closure, was also feasible and would

allow us a facile means to vary the C-15 to C-18 unit of the

molecule and, hence, access both coleophomones A/B (1/2) and

C (3). However, to successfully employ this strategy, the means

to construct the two requisite variations of the 1,2,3-trisubstituted

aromatic ring (one for coleophomones A-C (1-3) and one for

coleophomone D (4)), the highly unsaturated six-membered ring,

and the tricarbonyl structural motif centered around C-9 would

have to be found. It was at the latter of these additional synthetic

hurdles (Figure 2, disconnection B) that this first generation

approach fell, but not before synthesis of some 1,3-cyclohexa-

diones, also suitable for use in our forthcoming forays, had been

accomplished and optimized.

^aReagents and conditions: (a) concentrated H₂SO₄(cat.), MeOH, 65

°C, 12 h, 85%; (b) LiHMDS (1.05 equiv), THF, -78 °C, 1 h; then allyl-

or prenyl-Br (1.1 equiv), -78 to 0 °C, 3 h, 80-85%; (c) LDA (1.1 equiv),

THF, slow addition of a solution of 6a or 6b in THF:HMPA (7:1), -78

°C, 1 h; then prenyl-Br (2.0 equiv), -78 to 20 °C, 12 h, 89%; (d) 1 M

HCl:THF (1:10), 25 °C, 14 h, 95-98%. LiHMDS ) lithium bis(trimeth-

ylsilyl)amide; LDA ) lithium diisopropylamide; HMPA ) hexameth-

ylphosphoramide.

to history’s discard pile alongside its two unsuccessful predeces-

sors.

By now, we understood better than ever that the coleoph-

omone macrocycle’s closure demanded a special type of reaction

with the power to bind together two reluctant partners to make

a highly congested and strained ring. It was from this context

that our fourth and final retrosynthetic strategy emerged (Figure

2, fourth generation approach: macrocycle formation using

olefin metathesis - disconnection D). In this retrosynthetic

blueprint, we proposed severing the macrocyclic ∆^16,17double

bond in the hope that the forward reaction using olefin

metathesis might constitute the auspicious macrocycle closing

procedure that had been so elusive thus far. Retrosynthetically,

after this disconnection was made, we stuck to familiar territory

by proposing disconnections A-C (Figure 2) to break the

molecule down into more digestible portions. Upon its applica-

tion, this design came to fruition, having at its climax an amazing

sequence of reactions, led by a powerful olefin metathesis

reaction, which was able to supply both coleophomones B and

C from a single common precursor. Before we unravel the

intricate secrets of this conquest, which leaves the olefin

metathesis reaction with a sterling set of credentials, we shall

detail the important lessons learned, and technologies developed,

from the earlier strategies that ended prematurely without the

hope for success.

In moving on to the next generation approach, we upgraded

the problem of how to assemble the recalcitrant tricarbonyl

moiety to a priority slot. We considered that this key C-acylation

reaction (indicated as disconnection B) might benefit from being

intramolecular rather than intermolecular. Thus, we were

prompted to investigate whether this reaction could be used to

close the macrocycle itself (Figure 2, second generation ap-

proach: macrocycle formation using C-acylation - disconnec-

tion B). However, this approach was also set to flounder as the

C-acylation continued to prove its intransigence. However, as

of yet unsuccessful in reaching our goal, our investigations had

allowed us to garner much useful reconnaissance information,

which we judiciously applied to the charting of our next route

forward. In our continuing search for viable C-acylation

methods, we had come across several crucial leads in the

literature, discussed in detail later in the text. Ideas from these

studies, published by a number of different groups, led us to

eventually discover and develop the reaction in which aromatic

acyl cyanides could be united with a suitable 1,3-cyclohexadi-

one. Unfortunately, however, when attempted intramolecularly,

as a way of closing the macrocycle, this method proved not to

have the required brawn and, thus, also failed to close the

macrocycle.

Hence, it was with that failure that we moved on to the third

generation approach to the macrocycle closure (Figure 2, third

generation approach: macrocycle formation using C-alkylation

- disconnection C), having already solved two important

problems, first, how to synthesize an appropriate 1,3-cyclo-

hexadione, and, second, how to unite it with a suitable 1,2,3-

substituted aromatic unit to form the coleophomone’s tricarbonyl

motif. Unfortunately, macrocycle closure by alkylation at C-13

would also soon prove to be a dead-end, as, once again, the

proposed ring-closure reaction was incapable of meeting the

onerous criteria. This strategy was, therefore, quickly consigned

The Total Synthesis of Coleophomones B (2) and C (3)

First Generation Strategy: Attempted Macrocycle For-

mation Using O-Alkylation and Attempted Synthesis of

Coleophomone D (4). Our first, and relatively simple, task was

to find a means of synthesizing the requisite 1,3-cyclohexadiones

(8) that had been envisioned as key fragments for the synthesis

of the coleophomones A-D (1-4). This assignment was

successfully accomplished using the short synthetic sequence

shown in Scheme 2. Thus, the commercially available, C_s-

symmetrical 5-methyl-1,3-cyclohexadione (5) was desymme-

trized upon its protection as the corresponding methyl vinylo-

gous ester. The vinylogous ester formation was bought about

(9) Wei, Q.; Harran, S.; Harran, P. G. Tetrahedron 2003, 59, 8947-8954.

9

8874 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 3. Abortive Atempts To Construct the Precursors for the

First Generation Approach to the Macrocycle-Bearing

Coleophomones A-C (1-3) and for the Synthesis of

Coleophomone D Using the Regioisomeric Compounds 10 and

16: O- versus C-Alkylation^a

by refluxing 1,3-cyclohexadione 5 in methanol with catalytic

amounts of concentrated H₂SO₄, in a protocol taken directly

from the literature.¹⁰Alkylation adjacent to the carbonyl group

was then readily achieved by means of a LiHMDS-induced

enolate formation, followed by a quench with prenyl-Br (or allyl-

Br), to afford 6a or b in high yield (80-85%), and as 4:1

mixture of trans:cis isomers. The second alkylation proved to

be much less trivial to achieve. Actualization of the desired

transformation required the careful maintenance of a set of

rigorously derived reaction conditions before the pertinent

bisalkylated congener could be attained. It required that LDA

be used to form the enolate by very slow addition of a solution

of 6 in THF:HMPA (7:1) to the preformed base; this part of

the procedure was then followed by the slow addition of excess

prenyl-Br to the newly formed enolate, and, finally, the reaction

mixture was left to warm from -78 to 20 °C over the course

of a 12 h period. When this protocol was stringently adhered

to, the bisalkylated congeners could be attained in high yield

(89%). Deprotection of the methyl vinylogous ester to finally

reveal the 1,3-cyclohexadiones 8a or 8b was accomplished in

high yield (95-98%) using 1 M HCl in THF at ambient

temperature. This reliable and concise sequence of reactions

would be used in all subsequent synthetic forays. For the time

being, however, the focus of our attention shifted to the

investigation of how 8b might be attached to a suitable 1,2,3-

trisubstituted aromatic unit to gain rapid access to the simplest

(i.e., bearing no macrocycle) target, coleophomone D (4).

Commercially available 2-methoxy-6-methyl benzoic acid

ethyl ester (9) was chosen as a potentially apt starting point for

the construction of all of the aromatic fragments required for

the synthesis of each of the various coleophomones (Scheme

3). Initially, as part of our efforts toward the synthesis of

coleophomone D (4), ester 9 was first brominated at its free

benzylic position (NBS, benzoyl peroxide, CCl₄) in high yield

(98%) and then reduced to the corresponding aldehyde (10)

using a two-step reduction-partial reoxidation sequence (Dibal-

H, followed by PCC; in yields of 93% and 78%, respectively).

C-Alkylation of the 1,3-cyclohexadione 8b with this newly made

benzylic bromide (10) was then attempted using K₂CO₃as base.

Unfortunately, the only products isolated from this reaction were

those obtained through O-alkylation of the 1,3-cyclohexadione

8b by 10. Both of the possible regioisomers, 12b and 13b, were

attained in yields of 55% and 19%, respectively. Variation of

the base or the use of additives, such as a crown ether (15-

crown-5), failed to divert this reaction from its disappointing

course. The same reaction, attempted using the unsubstituted

1,3-cyclohexadione 5 instead of 8b, still afforded mostly the

product of O-alkylation, 12a (61%); however, on this occasion

a small amount of the desired C-alkylation product, 11a (1.5:1

inseparable mixture of stereoisomers, 22%), was also recovered

from the reaction mixture. It should be noted that 11a was

isolated and existed entirely in the closed spirocyclic form,

reminiscent of coleophomone A (1), rather than as the open

chain aldehyde/1,3-dione arrangement, which shares more

similarities with coleophomones B and C (2 and 3). Neither

changes to the base employed, nor the addition of 15-crown-5,

altered the distribution of products for this reaction. In an attempt

to investigate the possibility of building up the carbon frame-

^aReagents and conditions: (a) NBS (1.1 equiv), benzoyl peroxide (0.1

equiv), CCl₄, 77 °C, 3 h, 98%; (b) Dibal-H (2.5 equiv), toluene, -78 to 25

°C, 2 h, 93%; (c) PCC (1.5 equiv), CH₂Cl₂, 25 °C, 3.5 h, 78%; (d) 5 or 8b

(1.0 equiv), K₂CO₃(2.0 equiv), acetone, 25 °C, 0.5 h; then 10 (1.0 equiv),

56 °C, 0.5 h, 61% 12a plus 22% 11a for R ) H; 55% of 12b plus 19% of

13b for R ) prenyl; (e) PCC (3.0 equiv), CH₂Cl₂, 25 °C, 3 h, 65%; (f)

NBS (1.0 equiv), benzoyl peroxide (0.1 equiv), CCl₄, 77 °C, 5 h, 97%; (g)

BH₃‚SMe₂(2.0 equiv), THF, 25 °C, 12 h, 91%; (h) PCC (1.5 equiv), CH₂Cl₂,

25 °C, 2 h, 71%; (i) 8b (1.0 equiv), K₂CO₃(2.0 equiv), acetone, 25 °C, 0.5

h; then 16 (1.0 equiv), 56 °C, 0.5 h, 67% 17 plus 15% 18. NBS )

N-bromosuccinimide; Dibal-H ) diisobutylaluminum hydride; PCC )

pyridinium chlorochromate.

work of coleophomones A-C (1-3) from the spirocycle 11a,

the benzylic alcohol of 11a was oxidized using PCC to furnish

the spirocycle 14a (inseparable mixture of stereoisomers 1.5:1,

65%). However, this problematic avenue to our investigations

was finally abandoned when all attempts (using IBX,¹¹or the

Saegusa method¹²) to introduce the requisite R,â-unsaturation

to the six-membered ring failed.

Switching the substituent’s position on the aromatic ring (i.e.,

using benzylic bromide 16, obtained from commercially avail-

able 15 by a series of standard reactions, rather than its analogue

(11) Nicolaou, K. C.; Gray, D. L. F.; Montagnon, T.; Harrison, S. T. Angew.

Chem., Int. Ed. 2002, 41, 996-1000.

(12) Ito, Y.; Hirao, T.; Saegusa, T. J. Org. Chem. 1978, 43, 1011-1013.

(10) Constantino, M. G.; Beltrame, M., Jr.; Faria de Medeiros, E.; Jose da Silva,

G.-V. Synth. Commun. 1992, 19, 2859-2864.

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8875

A R T I C L E S

Nicolaou et al.

Scheme 4. Second Generation Approach: Abortive Attempts for

The difficult alkylation, uniting the previously synthesized

vinylogous ester 6b with this newly made bromide 20, was then

realized in 60% yield using the conditions (LDA/HMPA) that

had been so stringently derived at the beginning of our

investigations (vide supra). Hydrolysis of the vinylogous ester

moiety using 1 M HCl afforded 21 (as a 1.5:1 mixture of

stereoisomers) in high yield (92%). With the key intermediate

21 now synthesized, it was time to attempt the C-8/C-9 bond

formation using an intramolecular Claisen-type condensation.

Unfortunately, no conditions could be found to encourage the

desired merger, despite a wide ranging trawl through the possible

base options (K₂CO₃, Et₃N, LiHMDS) and extensive experi-

mentation with other reaction variables (e.g., temperature,

solvent, concentration). To investigate whether an aldol reaction

might prove to be a better way of forming the requisite carbon-

carbon bond, the ester 21 was reduced using Dibal-H. This

reduction was plagued by the concomitant reduction of the 1,3-

cyclohexadione, but the situation was made even worse if the

preceding vinylogous ester was used instead of its deprotected

congener 21. The desired benzylic alcohol could be attained

(albeit in a modest yield of 53%) and oxidized to aldehyde 23

in preparation for the aldol reaction attempts. Unfortunately,

the desired aldol macrocyclization could not be accomplished

either.

Intramolecular Claisen and Aldol Reactions^a

Given the poor results obtained from our studies thus far,

we knew that our next plan had to offer convincing improve-

ments in circumstances if we were to hit upon a suitable reaction

for forming the carbon-carbon bond between C-8 and C-9 of

the coleophomone skeleton, rather than simply making more

unwanted O-alkylated/acylated analogues of the 1,3-dione.

After a comprehensive survey of the literature looking into

known methods for acylation of 1,3-dicarbonyls, we uncovered

a series of interesting precedents.^14-19The vast majority of

existing examples for the C-acylation of a 1,3-dicarbonyl

involved initial O-acylation followed by a catalyzed (mediated

by heat, Lewis acid, or 4-DMAP) rearrangement to the C-

acylated congener. Where one, or both, of the 1,3-dicarbonyl’s

functionalities was an ester, the rearrangement was relatively

trivial to accomplish.^14,15,18However, where both of the car-

bonyls were ketones, the rearrangement was far from facile.^16-19

Due to the ease with which we could prepare the requisite

benzoic acid precursor 25²⁰(Scheme 5), we chose to start our

own model studies using a 1993 protocol¹⁸in which a carboxylic

acid could be coupled to a 1,3-dicarbonyl system using DCC

and 4-DMAP. This reaction was reported to furnish the

O-acylated product initially, but upon elevation of the reaction

temperature this intermediate could be coaxed into rearranging

to its C-acylated isomer. Unfortunately, when these conditions

were applied to 25, no such rearrangement of the initially formed

O-acylated benzoic acid 26 was observed; most likely, this

failure was due to severe steric hindrance arising from the

presence of two ortho-substituents on the aromatic ring.

^aReagents and conditions: (a) BBr₃(2.4 equiv), CH₂Cl₂, -78 °C, 0.5

h, 99%; (b) K₂CO₃(3.0 equiv), 19 (1.5 equiv), acetone, 56 °C, 14 h, 84%;

(c) K₂CO₃(1.0 equiv), MeOH:H₂O (4:1), 25 °C, 20 min, 98%; (d) PPh₃

(2.4 equiv), CBr₄(2.4 equiv), CH₃CN, 0 to 25 °C, 2 h, 86%; (e) LDA (1.2

equiv), THF, slow addition of a solution of 6b in THF:HMPA (7:1), -78

°C, 1 h; then 20 (1.2 equiv), -78 to 20 °C, 12 h, 60%; (f) 1 M HCl:THF

(1:10), 25 °C, 14 h, 92%; (g) Dibal-H (2.5 equiv), toluene, -78 to 25 °C,

2 h, 53% plus 30% reduction of 1,3-cyclohexadione moiety; (h) PCC (2.0

equiv), CH₂Cl₂, 25 °C, 2 h, 67%.

10) had no effect on the outcome of the alkylation reaction (16

+ 8b f 17 and 18, Scheme 3). Once again, only the undesired

O-alkylation was observed with no significant change in the

regiochemical outcome occurring either [17 (67%) and 18

(15%)]. Thus, we were forced to rethink our strategy, casting

the net wider in search of new technologies capable of uniting

an appropriate aromatic fragment to a suitable 1,3-cyclohexa-

dione to progress with our syntheses of the various coleoph-

omones A-D (1-4).

Second Generation Strategy: Model Studies and Attempts

at Macrocycle Formation Using Claisen, Aldol, or C-

Acylation Reactions. Because C-acylation of a suitable 1,3-

cyclohexadione represented the transformation that would give

rise to products most closely resembling the target structures,

we chose to move away from C-alkylation and focus on the

feasibility of this reaction instead. To test this updated strategy,

a means of rapidly accessing compounds such as 21 and 23

had to be found (Scheme 4). To this end, 9 was demethylated

to reveal the free phenol in high yield (99%) using boron

tribromide at -78 °C. The phenol was then alkylated with the

known allylic bromoacetate 19,¹³using potassium carbonate as

base, in good yield (84%). Acetate deprotection, mediated by

methanolic potassium carbonate, followed by conversion of the

resultant allylic alcohol to its corresponding bromide using the

PPh₃/CBr₄complex, afforded key bromide 20 (84%, two steps).

(14) (a) Ukita, T.; Nojima, S.; Matsumoto, M. J. Am. Chem. Soc. 1950, 72,

5143-5144. (b) Rogers, N. A. J.; Smith, H. J. Chem. Soc. 1955, 341-

346.

(15) Oikawa, Y.; Sugano, K.; Yonemitsu, O. J. Org. Chem. 1978, 43, 2087-

2088.

(16) Akhrem, A. A.; Lakhvich, F. A.; Budai, S. I.; Khlebnicova, T. S.;

Petrusevich, I. I. Synthesis 1978, 925-927.

(17) Oliver, J. E.; Lusby, W. R. Tetrahedron 1988, 44, 1591-1596.

(18) Tabuchi, H.; Hamamoto, T.; Ichihara, A. Synlett 1993, 651-652.

(19) Ali, A. S.; Liepa, A. J.; Thomas, M. D.; Wilkie, J. S.; Winzenberg, K. N.

Aust. J. Chem. 1994, 47, 1205-1210.

(13) (a) Babler, J. H.; Buttner, W. J. Tetrahedron Lett. 1976, 4, 239-242. (b)

Van Nguyen, T.; De Kimpe, N. Tetrahedron 2000, 56, 7969-7973.

(20) Coutts, S. J.; Wallace, T. W. Tetrahedron 1994, 50, 11755-11780.

9

8876 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 5. Second Generation Approach: Chemoselective C-

versus O-Acylation of 1,3-Cyclohexadiones Using Acyl Cyanides

as Acylating Agents^a

and, furthermore, if the temperature was elevated in an attempt

to coax the reluctant partners into union, only decomposition

of the acyl cyanide 28 was observed. It was therefore clear that

the two partners most relevant to a synthesis of the coleoph-

omones (28 and 8b) would stretch the scope of this reaction to

its limits, if indeed, they could be encouraged to participate at

all by further adaptation of the conditions. This recalcitrance

was observed because one partner, 28, was aromatic (and was

thus deactivated) and had two ortho-substituents increasing

steric hindrance in the reaction vicinity, and the other, 8b, bore

a proximal quaternary center that endowed it with similarly

detrimental space constraints. In an attempt to tip the balance

in our favor, we decided to investigate the intramolecular

reaction of an appropriate acyl cyanide with a suitable 1,3-

cyclohexadione rather than the intermolecular variant. Of course,

this ambitious plan would also use the reaction to close the

coleophomone macrocycle and, as such, would constitute a very

elegant solution to the problem of how to construct this 11-

membered ring. An added advantage to this approach was that,

if successful, it would furnish an adduct that was only a few

steps shy of coleophomone B (2).

To access the key acyl cyanide coupling compound 36,

commercially available phenol 30 was first transformed into

the acetonide/primary alcohol 31 using a reaction sequence taken

directly from the literature (Scheme 6).²⁴The primary alcohol

31 was then protected as its para-bromobenzoate ester, this

protecting group having been chosen because it offered the best

possibility for obtaining crystalline intermediates, suitable for

X-ray analysis, later on in the synthesis. Acetal deprotection

using p-TsOH, followed by MnO₂-mediated benzylic oxidation,

gave the phenolic aldehyde 32 (72%, three steps). Phenol 32

was then alkylated with bromoacetate 19¹³to furnish the acetate

33 in good yield (74%). The acetate functionality of 33 was

then deprotected under acidic conditions (cat. H₂SO₄), and the

resultant primary allylic alcohol was converted to bromide 34

using the PPh₃/CBr₄complex (53%, two steps). Bromide 34

was then employed as the alkylating agent for reaction with

the LDA-derived enolate of vinylogous ester 6b, to furnish the

crucial intermediate 35 in 64% yield and as a 1.6:1 mixture of

stereoisomers. The benzylic aldehyde group of 35 was then

transformed to the corresponding acyl cyanide moiety using the

conditions we had developed previously (60%, two steps). After

deprotection of the vinylogous ester (1 M HCl), the synthesis

of the pivotal acyl cyanide coupling substrate 36 (as a 1.5:1

mixture of stereoisomers) had been completed. However, the

macrocyclization coupling reaction did not work (36f37) under

any of the conditions we tested (variable temperature, solvent,

etc.).

^aReagents and conditions: (a) NaH (4.0 equiv), MeI (6.0 equiv), THF,

1 h at 25 °C plus 2 h at 67 °C, 98%; (b) see ref 18; (c) 5 (1.0 equiv), DCC

(1.1 equiv), 4-DMAP (0.1 equiv), toluene, 110 °C, 72 h, 43%; (d) BH₃‚SMe₂

(2.2 equiv), THF, 25 °C, 15 h, 98%; (e) PCC (1.5 equiv), CH₂Cl₂, 25 °C,

2 h, 78%; (f) Et₂AlCN (2.0 equiv), toluene, 0 °C, 1 h, 96%; (g) PCC (4.0

equiv), CH₂Cl₂, 40 °C, 12 h, 62%; (h) 5 or 7b (1.0 equiv), Et₃N (1.5 equiv),

THF, 25 °C, 12 h, 84% of 29a and 71% of 29b. DCC ) 1,3-

dicyclohexylcarbodiimide; 4-DMAP ) 4-(dimethylamino)pyridine.

It had been known for some time that cyanide ions are

catalysts able to facilitate the rearrangement of O-acylated 1,3-

diones to furnish their C-acylated variants^21,22via the interme-

diacy of an acyl cyanide. More recently, a report appeared in

which acyl cyanides had been employed directly to C-acylate

1,3-dicarbonyl compounds under mild conditions.²³No O-

acylation was detected when this protocol, using triethylamine

as base, was employed. Attracted by the mild conditions of this

transformation, we wanted to test its scope within systems useful

to our synthesis of the coleophomones. Thus, benzoic acid 25²⁰

was reduced to the corresponding aldehyde 27 using a two-

step reduction-partial reoxidation procedure employing BH₃‚

SMe₂and PCC (overall yield 76%, Scheme 5). The resulting

aldehyde 27 was reacted with Nagata’s reagent (Et₂AlCN) to

afford the cyanohydrin, which could then be oxidized to the

acyl cyanide 28 using PCC (60%, two steps).

With the acyl cyanide 28 now in hand, we began to examine

its reaction with some of the 1,3-cyclohexadiones that we had

synthesized previously. In the presence of triethylamine, acyl

cyanide 28 reacted smoothly with both 1,3-cyclohexadiones 5

and 7b (furnishing 29a or 29b in yields of 84% and 71%,

respectively, Scheme 5). The adduct 29b arising from the

reaction between 28 and 7b was formed as a mixture of four

isomers, two stereo- and two atropisomers (6:4:4:1). To our

dismay, however, no reaction was observed with the disubsti-

tuted 1,3-cyclohexadione 8b, under the established conditions,

Our newly developed acyl cyanide coupling reaction had, at

last, shown us an attractive way to make the fragile tricarbonyl

unit of the coleophomones; however, we had also discovered

that, as it stood, it had limits that prevented its use in our desired

scenarios. It could only be employed with a carefully selected

pair of less sterically encumbered substrates. Therefore, on the

basis of our initial model studies, we rationalized that using

1,3-cyclohexadione 7b (bearing no quaternary center) we could

succeed in the coupling reaction even with a suitable, but

hindered, 1,2,3-substituted aromatic fragment. Successful ac-

(21) Oliver, J. E.; Wilzer, K. R.; Waters, R. M. Synthesis 1990, 1117-1119.

(22) Flores Montes, I.; Burger, U. Tetrahedron Lett. 1996, 37, 1007-1010.

(23) Tang, Q.; Sen, S. E. Tetrahedron Lett. 1998, 39, 2249-2252.

(24) Suzuki, M.; Sugiyama, T.; Watanabe, M.; Murayama, T.; Yamashita, K.

Agric. Biol. Chem. 1987, 51, 1121-1127.

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8877

A R T I C L E S

Nicolaou et al.

Scheme 6. Second Generation Approach: Attempts at

Intramolecular C-Acylation of the 1,3-Cyclohexadione Moiety Using

an Acyl Cyanide as the Coupling Partner^a

Scheme 7. Third Generation Approach: Attempts at an

Intramolecular Alkylation^a

^aReagents and conditions: (a) Et₂AlCN (1.5 equiv), toluene, 0 to 25

°C, 1 h, 97%; (b) PCC (4.0 equiv), CH₂Cl₂, 40 °C, 10 h, 62%; (c) 7b (1.0

equiv), Et₃N (2.0 equiv), THF, 25 °C, 12 h, 76%; (d) H₂SO₄(cat.), MeOH,

25 °C, 10 h, 74%; (e) PPh₃(1.5 equiv), CBr₄(1.5 equiv), CH₃CN, 0 to 25

°C, 3 h, 90%; (f) excess of CH₂N₂, Et₂O, 0 °C, 0.5 h, 78%.

be noted here that the standard basic conditions for acetate

hydrolysis were rejected due to the strong acidity, and, thus

lability, of the tricarbonyl C-9 proton. Upon treatment of any

analogue bearing this tricarbonyl unit with even mild base, a

water-soluble salt was formed, manipulation became fraught

with problems, and TLC analysis was rendered almost useless

unless each sample was acidified prior to spotting, and, even

then, streaking plagued the procedure. After the acetate depro-

tection, the resultant primary allylic alcohol was transformed

to the bromide using the PPh₃/CBr₄complex (90%). In

preparation for the pivotal alkylation step, the tricarbonyl group

was then protected by converting it into a mixture of all its

possible regioisomeric methyl vinylogous esters. This protection

reaction affording allylic bromide 40 in high yield (overall 78%,

Scheme 7), as a complex mixture of stereo- and atropisomers

(as well as the aforementioned regioisomers) was facilitated by

diazomethane as the methylating agent. Disappointingly, the

macrocyclization-alkylation reaction failed, even under the

specially devised conditions that had previously been so

successful in generating this quaternary center in analogous

compounds. At this stage, we needed no further evidence that

closure of the coleophomone macrocycle was extremely intran-

sigent and hard to accomplish; an entirely new approach was

needed for this key bond-forming reaction.

Fourth Generation Strategy: Model Studies and the

Successful Syntheses of Coleophomones B (2) and C (3)

Using Olefin Metathesis To Form the Macrocycle. Due to

the acute awareness we had now developed of the challenges

involved in forming the coleophomone macrocycle, we decided

that proof of principle must be obtained for our new olefin

metathesis macrocyclization strategy, alongside the answers to

many probing questions relating to the scope of such a

metathesis reaction, before we expended too much energy on

the synthesis of key precursors to the real target systems. The

^aReagents and conditions: (a) Et₃N (1.5 equiv), p-BrBzCl (1.05 equiv),

4-DMAP (0.05 equiv), CH₂Cl₂, 25 °C, 20 min, 97%; (b) p-TsOH‚H₂O (0.3

equiv), THF:H₂O (3:2), 75 °C, 10 h, 85%; (c) MnO₂(10 equiv), EtOAc,

25 °C, 1.5 h, 87%; (d) K₂CO₃(2.0 equiv), 19 (1.5 equiv), acetone, 56 °C,

2 h, 74%; (e) H₂SO₄(cat.), MeOH, 25 °C, 10 h, 72%; (f) PPh₃(2.0 equiv),

CBr₄(2.0 equiv), CH₃CN, 0 to 25 °C, 1.5 h, 73%; (g) LDA (1.2 equiv),

THF, slow addition of a solution of 6b in THF:HMPA (7:1), -78 °C, 0.5

h; then 34 (1.2 equiv), -78 to 20 °C, 12 h, 64%; (h) Et₂AlCN (1.1 equiv),

toluene, 0 °C, 1 h, 96%; (i) PCC (4.0 equiv), CH₂Cl₂, 40 °C, 10 h, 62%;

(j) 1 M HCl:THF (1:10), 25 °C, 18 h, 83%. p-BrBzCl ) 4-bromobenzoyl

chloride; p-TsOH ) p-toluenesulfonic acid.

complishment of this task would then leave us only the hurdle

of creating the quaternary center in a macrocyclization step.

Thus, it was that the key concepts of our third generation

approach were born.

Third Generation Strategy: Attempted Macrocycle For-

mation Using C-Alkylation at C-13. Having previously

synthesized aldehyde 33, it was convenient to begin our third

synthetic foray with this advanced intermediate. The aldehyde

33 was converted to the corresponding acyl cyanide 38 (Scheme

7) using our established set of conditions (60%, two steps). As

we had predicted, this compound could then be successfully

coupled to 1,3-cyclohexadione 7b at ambient temperature,

furnishing 39 (as a mixture of two stereoisomers each of which

existed as two atropisomers) in a good yield (76%). Satisfied

by the fact that our molecule was smoothly and rapidly gaining

in its resemblance to the targeted coleophomone skeleton with

each new step chartered, we set about converting the acetate to

the corresponding bromide, as was required for the key

macrocyclization-alkylation event. Acidic conditions (cat. H₂-

SO₄) were employed to hydrolyze the acetate (74%). It should

9

8878 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 8. Fourth Generation Approach: (A) Stereospecific

Formation of (Z)-11-Membered Macrocycle 43 via Olefin

Metathesis; (B) 4-DMAP-Mediated Coupling of Acyl Cyanide 41

with Disubstituted 1,3-Cyclohexadiones 8a,b and Olefin Metathesis

of 44a^a

Figure 3. ORTEP representation of 43.

examine whether olefin metathesis was capable of closing such

a strained macrocycle in the first place. To resolve this issue,

the model compound 42b was synthesized as follows (Scheme

8A). The previously synthesized phenol 32 was alkylated with

3-bromo-2-methylpropene (K₂CO₃, refluxing acetone) in high

yield (91%). The resulting aldehyde was converted into its acyl

cyanide congener 41 using our established conditions (Nagata’s

reagent followed by PCC oxidation) in good yield (60%, two

steps). The acyl cyanide 41 was then coupled with 1,3-

cyclohexadione 7a, in the presence of triethylamine, to afford

the crucial metathesis precursor 42b in high yield (91%). When

42b was subjected to the first generation Grubbs catalyst²⁶under

its standard operating conditions, no macrocyclization was

observed; fortunately for us, the second generation Grubbs

catalyst (A)²⁷triumphed where its predecessor had failed by

succeeding in its task of inducing ring closure. Thus, when 42b

was treated with 20 mol % of Grubbs’ catalyst A, in gently

refluxing dichloromethane, for a period of 5 h, the macrocycle

43 was obtained in 60% yield as the sole product of the reaction.

The newly formed macrocyclic double bond of 43 (∆^16,17

coleophomone numbering) was present exclusively in its

Z-configuration. As a result of having incorporated the para-

bromobenzoyl protecting group into our molecule, we were able

to obtain a crystal of 43 that was suitable for X-ray analysis.

This analysis²⁸(see Figure 3) confirmed not only the molecule’s

complete structure, but also the macrocycle double bond

geometry, and it revealed the arrangement of the enol within

the tricarbonyl motif. Olefin metathesis had proved itself a viable

ring-closing tool on a testing ground where so many other

methods had already fallen foul of the devious coleophomone

skeleton.

^aReagents and conditions: (a) K₂CO₃(2.0 equiv), 3-bromo-2-methyl-

propene (1.5 equiv), acetone, 56 °C, 2 h, 91%; (b) Et₂AlCN (1.1 equiv),

toluene, 0 to 25 °C, 1 h, 82%; (c) PCC (4.0 equiv), CH₂Cl₂, 40 °C, 12 h,

73%; (d) 5, 7a, or 7b (1.2 equiv), Et₃N (2.0 equiv), THF, 25 °C, 6-12 h,

91-98%; (e) cat. A (0.2 equiv), CH₂Cl₂, 40 °C, 5 h, 60% for 42b; cat. A

(0.3 equiv), CH₂Cl₂, 40 °C, 18 h, 30% for 42c; (f) 8a,b (1.2 equiv), Et₃N

(2.0 equiv), 4-DMAP (1.0 equiv), THF, 25 °C, 72-96 h, 83% of 44a and

86% of 44b; (g) cat. A (0.1 equiv), CH₂Cl₂, 40 °C, 1 h, 85%. Cy )

cyclohexyl; Mes ) mesityl.

(25) For selected examples of olefin metathesis being used to form 11-membered

rings, see: (a) Wipf, P.; Stephenson, C. R. J.; Walczak, M. A. A. Org.

Lett. 2004, 6, 3009-3012. (b) Ronsheim, M. D.; Zercher, C. K. J. Org.

Chem. 2003, 68, 1878-1885. (c) Arisawa, M.; Kato, C.; Kaneko, H.;

Nishida, A.; Nakagawa, M. J. Chem. Soc., Perkin Trans. 1 2000, 1873-

1876. (d) Hoye, T. R.; Promo, M. A. Tetrahedron Lett. 1999, 40, 1429-

1432. (e) Winkler, J. D.; Holland, J. M.; Kasparec, J.; Axelsen, P. H.

Tetrahedron 1999, 55, 8199-8214. (f) El Sukkari, H.; Gesson, J.-P.;

Renoux, B. Tetrahedron Lett. 1998, 39, 4043-4046. For insightful

discussion, see also: (a) Fu¨rstner, A.; Radkowski, K.; Wirtz, C.; Goddard,

R.; Lehmann, C. W.; Mynott, R. J. Am. Chem. Soc. 2002, 124, 7061-

7069. (b) Deiters, A.; Martin, S. F. Chem. ReV. 2004, 104, 2199-2238.

(26) (a) Schwab, P.; France, M. B.; Ziller, J. W.; Grubbs, R. H. Angew. Chem.,

Int. Ed. Engl. 1995, 34, 2039-2041. (b) Schwab, P.; Grubbs, R. H.; Ziller,

J. W. J. Am. Chem. Soc. 1996, 118, 100-110.

challenges of constructing medium-sized rings by any method

are well known; in rings having 10-12 members, strain and

congestion can easily frustrate attempts at their formation, even

with much simpler exemplars. Olefin metathesis had already

proved its value in this arena, the synthesis of 11-membered

rings, in just a few instances when our investigations were at

the planning stages.²⁵However, all of the literature examples

involved situations distinctly simpler and more favorable than

the coleophomone’s difficult skeleton. We knew, therefore, that

success, if it came to us, would require this reaction to be pushed

far beyond its existing boundaries at the time.

(27) Scholl, M.; Ding, S.; Lee, C. W.; Grubbs, R. H. Org. Lett. 1999, 1, 953-

956.

(28) CCDC-187042 contains the supplementary crystallographic data for

compound 43. These data can be obtained free of charge via www.ccdc.

cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic

Data Center, 12 Union Road, Cambridge, CB21 EZ, UK; fax: (+44)

1223-336-033; or deposit@ccdc.cam.ac.uk).

Model Studies Undertaken To Validate the Olefin Me-

tathesis Approach to Macrocylization. The first and most

pressing query of the current investigation required us to

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8879

A R T I C L E S

Nicolaou et al.

This beautiful olefin metathesis macrocyclization was not the

only pivotal victory that we were to notch up during this

important phase of this adventure. Concomitantly, we had been

continuing our search for acyl cyanide coupling reaction

conditions that would allow for the participation of bisalkylated

1,3-cyclohexadiones in the union, and, not just, their monoalkyl-

ated analogues. We had already discovered that heat could not

be of assistance, because elevated temperatures led to the

decomposition of the acyl cyanide. We rationalized that the

addition of the well-known acylation activator, 4-DMAP, might,

however, provide the requisite mild assistance to cajole the two

partners together. Sure enough, after some experimentation, we

found that the addition of one equivalent of 4-DMAP to the

reaction mixture facilitated the coupling of suitable hindered

bis-ortho-substituted aromatic acyl cyanides with bisalkylated

1,3-cyclohexadiones, albeit after prolonged reaction times (72-

96 h, Scheme 8B).

ably, the desired macrocycle 43 was isolated from this reaction,

albeit in a much reduced yield (30%), and, as a result, our

knowledge of the scope of olefin metathesis, as a means to

cyclize medium-sized rings, was now well and truly expanding.

Indeed, few of those involved in its discovery and development

could have predicted the true power of this synthetic tool³⁰when

it first emerged some decades ago.³¹

Despite being gratified by the successes we had achieved with

these olefin metathesis reactions, we were becoming more and

more perturbed by the problems arising from the presence of

the tricarbonyl unit in our substrates. As mentioned earlier, this

structural motif leaves a lot to be desired in terms of stability

and ease of manipulation, and we felt that it might be a possible

cause of the low yields we were obtaining for some of the

metathesis reactions. We, therefore, opted to examine the

protection of this fragile functionality as our next priority. To

our surprise, this seemingly tangential move would pay con-

siderable dividends in that it was set to reveal to us how we

could control the access to both geometric isomers of the ∆^16,17

macrocylic double bond.

We chose to investigate protecting the tricarbonyl as its

methyl vinylogous esters using diazomethane as we already had

some experience in this approach, and, in addition, it was almost

impossible to find other suitable protecting groups that would

work in this very sterically encumbered environment. Thus,

tricarbonyl compound 42a (for the synthesis of 42a, see Scheme

8A) was treated with diazomethane in diethyl ether at 0 °C,

resulting in the formation of vinylogous ester 47, which

precipitated out of solution in 48% yield, and its sibling

vinylogous ester 46 that was readily isolated from the residual

solution in 48% yield (Scheme 9). The tricarbonyl unit in all

of our previously synthesized substrates had always existed, for

the most part, as the enol isomer(s) wherein the enol moiety

was situated within the six-membered ring. This arrangement

is most similar to the newly protected variant, methyl vinylogous

ester 47. We were curious to see if relocating this double bond

to the exocyclic position, as in 46, would have a bearing on the

olefin metathesis reaction as it had changed the conformation

of the molecule considerably, and so it was with vinylogous

ester 46 that we decided to pursue our studies. Curiosity may

have “killed the cat”, but in our case it turned out to be the

most valuable virtue which led to us solving the problem of

how to efficiently synthesize both coleophomones B (2) and C

(3). Alkylation of vinylogous ester 46 proceeded smoothly, using

LiHMDS, HMPA, and prenyl bromide, to furnish the olefin

metathesis precursor 48 (63%) as an inseparable mixture of ∆^8,9

geometric isomers, each of which also existed as a 1:1 mixture

of atropisomers (as revealed by ¹H NMR). Next, 48 was

subjected to the action of Grubbs’ catalyst A (20 mol %) in

refluxing dichloromethane for a period of 20 h. Macrocyclization

occurred (48f49, 30%), but this was not the main cause of

our jubilation, for the newly formed macrocycle ∆^16,17double

bond existed in the product 49 as a single isomer; however,

this time it tantalizingly bore the E-configuration (see NOE

relationships marked on structure 49, Scheme 9). With this result

in hand, we had now accessed both geometric isomers of the

macrocycle ∆^16,17double bond and had thus gained significant

insight into how we might specifically obtain both coleoph-

Armed with this vital new piece of knowledge, we set forth

on our attempt to test the macrocyclization reaction on the

simplest of the possible bisalkylated congeners, 44a (Scheme

8B). This olefin metathesis precursor was the product of the

coupling of acyl cyanide 41 and 1,3-cyclohexadione 8a, a

reaction that proceeded slowly, but with a remarkably high yield

(83%) given the difficulty we had in defining reaction conditions

that would work at all. Disappointingly, when 44a was treated

with Grubbs’ catalyst A (10 mol %), in refluxing dichloro-

methane, for 1 h, the spirocycle 45 (Scheme 8b) was rapidly

formed as the only product of the reaction (in 85% yield), in

preference to the desired macrocycle. This latest result was quite

interesting in itself, as this work⁶constituted a very early

example of an olefin, which was geminally disubstituted at its

terminus, participating in a metathesis cyclization reaction.

Indeed, this reaction class remains exceptionally rare and is

usually confined to the simplest substrates,²⁹so to observe a

prenyl group willingly participating in the metathesis reaction

of such a complex substrate was truly a pleasing if not

groundbreaking result.

Following the important discovery that a prenyl group could

participate in a simple olefin metathesis cyclization event, we

were anxious to see whether this new paradigm would stretch

as far as a prenyl group also willingly engaging in the key and,

much more challenging, macrocyclization reaction. We rational-

ized that if this approach was successful and could be employed,

it might relocate the metathesis initiation site to the less

substituted olefin (that appended to the aromatic portion of the

molecule) and, thus, encourage macrocyclization to occur in

preference to the alternative and undesired spirocycle formation.

To begin testing this premise, monoprenylated 42c was subjected

to our standard olefin metathesis conditions (30 mol % catalyst

A/refluxing dichloromethane) for 18 h (Scheme 8A). Remark-

(29) For examples of olefins that are geminally disubstituted at the terminus

participating in olefin metathesis reactions reported prior to our work,⁶

see: (a) Nugent, W. A.; Feldman, J.; Calabrese, J. C. J. Am. Chem. Soc.

1995, 117, 8992-8998. (b) Fu¨rstner, A.; Thiel, O. R.; Blanda, G. Org.

Lett. 2000, 2, 3731-3734. (c) Braddock, D. C.; Wildsmith, A. J.

Tetrahedron Lett. 2001, 42, 3239-3242. (d) Fu¨rstner, A.; Dierkes, T.; Thiel,

O. R.; Blanda, G. Chem.-Eur. J. 2001, 7, 5286-5298. (e) Braddock, D.

C.; Matsuno, A. Tetrahedron Lett. 2002, 43, 3305-3308. For examples of

olefins that are geminally disubstituted at the terminus participating in olefin

metathesis reactions reported after our work,⁶see: (f) Donohoe, T. J.;

Blades, K.; Moore, P. R.; Waring, M. J.; Winter, J. J. G.; Helliwell, M.;

Newcombe, N. J.; Stemp, G. J. Org. Chem. 2002, 67, 7946-7956. (g)

Garcia-Fandin˜o, R.; Codesido, E. M.; Sobarzo-Sa´nchez, E.; Castedo, L.;

Granja, J. R. Org. Lett. 2004, 6, 193-196.

(30) Nicolaou, K. C.; Bulger, P. G.; Sarlah, D. Angew. Chem., Int. Ed., in press.

(31) Nicolaou, K. C.; Snyder, S. A. Classics in Total Synthesis II; Wiley-

VCH: Weinheim, 2003; pp 161-206.

9

8880 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 9. Fourth Generation Approach: Stereospecific

Formation of (E)-11-Membered Macrocycle 49 via Olefin

Metathesis^a

reaction conditions (Scheme 10). From this protection reaction,

we were able to isolate and, significantly, separate each of the

three possible regioisomers of the methyl vinylogous ester

product (50, 51, and 52), in high overall yield (96%). Both of

the vinylogous ester regioisomers situated within the six-

membered ring, 50 and 52 (obtained with a yield of 32% and

16%, respectively), existed in a single form, whereas regioisomer

51 (obtained in 48% yield) proved to be an inseparable mixture

of ∆^8,9geometric isomers (1.3:1) as expected, and each of these

last two isomers also existed as a mixture of atropisomers (ca.

1:1 by ¹H NMR spectroscopy). Understandably, the endocyclic

vinylogous ester 52 was the minor regioisomer isolated from

this reaction due to its greater steric congestion around the

quaternary center of the six-membered ring.

Upon exposure of all three regioisomers (50, 51, and 52),

independently, to Grubbs’ catalyst A (10 mol %), each one

remarkably succumbed to a regio- and stereospecific ring-closing

metathesis reaction, to furnish macrocycles 53, 54, and 55,

respectively, in high yields (53, 80%; 54, 86%; 55, 67%, Scheme

10). The products, 53 and 55, arising from the metathesis of

the endocyclic vinylogous ester regioisomers, 50 and 52, both

possessed exclusively the Z-configuration at the newly formed

macrocyclic ∆^16,17double bond (see NOE interactions marked

in Scheme 10). The product 55 of the more congested endocyclic

vinylogous ester 52 existed as a single isomer with no

spectroscopic evidence of atropisomerism. By contrast, mac-

rocycle 53 existed as a 4:1 mixture of atropisomers in CDCl₃

(by ¹H NMR spectroscopy); however, it too collapsed to a single

isomer when dissolved in CD₃CN (by ¹H NMR spectroscopy).

The configurational details of macrocycle 53 were confirmed,

and its precise solid-state structure was determined, by X-ray

crystallographic analysis of a suitable crystal (see Figure 4).³³

The NMR spectra obtained for olefin metathesis product 54

bore many more degrees of complexity in comparison to its

siblings, 53 and 55, and, as a result, some additional deduction

was required to deconvolute many of its structural character-

^aReagents and conditions: (a) excess CH₂N₂, Et₂O, 0 °C, 30 min, 48%

of 46 plus 48% of 47; (b) LiHMDS (1.05 equiv), THF/HMPA (10:1), -78

°C, 1 h; then prenyl-Br (2.0 equiv), -78 to 0 °C, 5 h, 63%; (c) cat. A (0.2

equiv), CH₂Cl₂, 40 °C, 20 h, 30% of 49 plus 35% recovered starting material

48.

omones B (2) and C (3) using a route that diverged from a single

precursor at a late stage. The transformation of 48 into 49

possessed one other beguiling feature related to the geometric

isomers of the ∆^8,9double bond. The product 49 was the ∆^8,9

E-isomer exclusively; in addition, a substantial amount of

starting material was recovered (30%) that had been enriched

in the ∆^8,9Z-isomer. We suspected that only the ∆^8,9E-isomer

participated in the olefin metathesis macrocyclization reaction,

although, at this stage, we had no definitive proof of this

hypothesis because we could not separate the ∆^8,9geometric

isomers of 48. However, a subsequent investigation³²looking

at a series of similar compounds, where such a separation was

possible, would prove that this assertion was indeed correct.

1

istics. The H NMR spectrum clearly revealed that 54 existed

as two inseparable isomers (1:1). Based on the two facts that,

first, the starting material existed as a 1.3:1 mixture of ∆^8,9

geometric isomers, and, second, the reaction yield was 86%,

we surmised that the two isomers seen in the ¹H NMR spectrum

were the ∆^8,9geometric isomers. This result was somewhat

surprising in that, in contrast to the monoprenylated model

system (48f49, Scheme 9), it indicated that in this case both

the ∆^8,9E- and Z-isomers participated in the olefin metathesis

reaction. Following on from this analysis, we deduced that the

newly formed macrocyclic ∆^16,17double bond was present in

just one isomeric form. Based on the result of our model study

employing an exocyclic methyl vinylogous ester (48f49,

Scheme 9), we hoped that this ∆^16,17double bond was the

E-configured variant. At this stage, however, we had only a

few scant, but tantalizing, clues garnered from ¹H NMR

comparisons and inconclusive NOE studies that this might

indeed be the case.

Olefin Metathesis and the Formation of the ∆^16,17E-

and Z-Macrocycles of Coleophomones B and C

With so much new information, and the power of the olefin

metathesis reaction, now revealed to us through our reconnais-

sance, we felt that we were in a particularly strong and

exploitable position from which to tackle the synthesis of the

macrocycle bearing coleophomones A-C (1-3). To this end,

the fully substituted precursor 44b, whose synthesis had only

recently become feasible following the discovery of 4-DMAP’s

crucial role as an additive in the acyl cyanide coupling reaction

with bisalkylated 1,3-cyclohexadiones (Scheme 8B), was pro-

tected using diazomethane under our previously established

The olefin metathesis had been a resounding success, not only

because of its exquisite specificity, which had seemingly

(33) CCDC-187043 contains the supplementary crystallographic data for

compound 53. These data can be obtained free of charge via www.ccdc.

cam.ac.uk/conts/retrieving.html (or from the Cambridge Crystallographic

Data Center, 12 Union Road, Cambridge, CB21 EZ, UK; fax: (+44)

1223-336-033; or deposit@ccdc.cam.ac.uk).

(32) Vassilikogiannakis, G.; Margaros, I.; Tofi, M. Org. Lett. 2004, 6, 205-

208.

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8881

A R T I C L E S

Nicolaou et al.

Scheme 10. Fourth Generation Approach toward the Coleophomones: Stereospecific Key Olefin Metathesis of Regioisomeric Vinylogous

Esters 50, 51, and 52^a

^aReagents and conditions: (a) excess CH₂N₂, Et₂O, 0 °C, 1 h, 32% of 50, 48% of 51 plus 16% of 52; (b) cat. A (0.1 equiv), CH₂Cl₂, 40 °C, 3 h, 80%;

(c) cat. A (0.1 equiv), CH₂Cl₂, 40 °C, 3 h, 86%; (d) cat. A (0.1 equiv), CH₂Cl₂, 40 °C, 4.5 h, 67%.

adopted by the precursors bearing the second prenyl group (50-

52) positioned the two reacting double bonds in close promixity

to one another with less freedom to move and adopt new

conformations where this proximity effect was not in play. We

assumed that the metathesis reaction was irreversible due to

the fact that we did not observe the formation of any spirocyclic

products (analogous to compound 45, Scheme 8).

Our fascination with this puissant reaction did not end here,

for it had yet another captivating feature; the reaction was

completely diastereoselective, only the prenyl group occupying

the position cis to the adjacent methyl (at C-12) participates in

Figure 4. ORTEP representation of 53.

the olefin metathesis reaction. This preference presumably arises

delivered into our hands both of the requisite ∆^16,17macrocycle

from the fact that the observed ring closure places the remaining

variants (corresponding to the two different coleophomone

prenyl group trans to the C-12 methyl substituent in the

macrocycle skeletons), but also because the reaction yields were,

macrocyclic product, thus allowing both substituents to occupy

without exception, much higher than we had obtained in the

the energetically more favorable equatorial positions of the

preceding model studies. Usually, one may expect reaction

yields to decrease on going from a transformation involving

cyclohexyl ring.

Finally, we were able to incontrovertibly prove our earlier

hypothesis that the presence of an unprotected tricarbonyl motif

was detrimental in the olefin metathesis reaction, by subjecting

the unprotected precursor 44b to Grubbs’ catalyst A. The

macrocyclization of this substrate (44b) occurred in only very

modest yield (<20%), an exceedingly poor performance when

compared to the momentous one of its protected derivatives

(50-52) subjected to identical conditions.

The Final Lap - Completing the Synthesis of Coleoph-

omones B (2) and C (3). In operatic terms, the drama had

climaxed, and now it was time for the denouement; following

just a few simple (at least on paper) synthetic operations, we

simplified model compounds to one employing more complex

real systems; however, in this case, the opposite appeared to

be true. We rationalized that the presence of a second prenyl

group forced an additional element of rigidity onto the meta-

thesis precursors (50-52), making the loss in entropy upon

macrocyclization less costly.³⁴Furthermore, our modeling

suggested that the most favorable molecular conformation

(34) For insightful discussion into the effects of entropy in the olefin metathesis

reaction and examples, see: (a) Fu¨rstner, A.; Langemann, K. J. Org. Chem.

1996, 61, 3942-3943. (b) Fu¨rstner, A.; Langemann, K. Synthesis 1997,

792-803. (c) Fu¨rstner, A.; Seidel, G.; Kindler, N. Tetrahedron 1999, 55,

8215-8230. (d) Galli, C.; Mandolini, L. Eur. J. Org. Chem. 2000, 3117-

3125.

9

8882 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 11. Fourth Generation Approach: Completion of the

Total Synthesis of Coleophomones B (2) and C (3)^a

compound, while the exocyclic vinylogous ester 57 remained

as a mixture of ∆^8,9geometric isomers. However, the ratio of

8,9

∆

geometric isomers in 57 had become enriched in favor of

the E-variant, it now being 3:1 (E:Z) having started as a 1:1

mixture in 54. Furthermore, the starting material 54 recovered

from this reaction (15%) contained a higher proportion of the

∆

^8,9Z-isomer. Molecular modeling combined with NOE studies

led us to postulate that this result was due to the additional steric

encumbrance experienced by the C-11 methylene group in the

Z-isomer of 54, caused by the direct positioning of the C-24

methyl group above it.

Brief exposure at ambient temperature of 56 and 57,

independently, to a methanolic potassium carbonate solution led

to the global deprotection of these compounds. Vinylogous ester

56 took longer to complete the desired reaction set (3 h) and

furnish 58 (90% yield) than did vinylogous ester 57, which took

just 30 min to afford 59 (96% yield). Conveniently, both

products (58 and 59) had now coalesced into a single isomeric

state, the structural intricacies of which yielded much more

readily to thorough examination than those of some of the

preceding isomeric mixtures. At this juncture we were, therefore,

able to unequivocally confirm our earlier distinction between

the ∆^16,17macrocyclic double bond E- and Z-geometric isomers,

because 58 and 59 were not identical, and the only remaining

possible difference between them was the ∆^16,17double bond

geometry. As we had expected, the deprotected tricarbonyl

compound 59 bore the E-configuration at the ∆^16,17macrocyclic

double bond (for NOE interactions in 58 and 59, see Scheme

11). For the final flourish, compound 59 was oxidized using

MnO₂in refluxing ether to afford coleophomone B (2) in good

yield (73%). Compound 58, however, reacted sluggishly with

MnO₂and was, therefore, oxidized to coleophomone C (3) using

freshly prepared Collins reagent instead (the yield of this latter

reaction being 81%). The icing on the cake for the last two

steps of this conquest had been the spectroscopic purity of the

products, 58, 59, 2, and 3, obtained from the specially devised

acid-base workup, which rendered column chromatography

superfluous and unnecessary. The full spectroscopic data of both

synthetic coleophomone B (2) and C (3) were in every way

identical to those reported for their naturally occurring coun-

terparts.¹

^aReagents and conditions: (a) LiHMDS (1.3 equiv), THF, -78 to -10

°C, 1 h; then PhSeCl (1.4 equiv), -78 to 0 °C, 30 min; then sat. NH₄Cl

(excess), 31% aqueous H₂O₂(excess), 25 °C, 1 h, 61%; (b) LiHMDS (1.3

equiv), THF, -78 to 25 °C, 1 h; then PhSeCl (1.4 equiv), -60 to 25 °C,

30 min; then sat. NH₄Cl (excess), 31% aqueous H₂O₂(excess), 25 °C, 1 h,

53%, plus 15% recovered 54; (c) K₂CO₃(3.0 equiv), MeOH, 25 °C, 11 h;

then H₂O (excess), 3 h, 90%; (d) K₂CO₃(3.0 equiv), MeOH, 25 °C, 40

min; then H₂O (excess), 30 min, 96%; (e) py (12 equiv), CrO₃(6.0 equiv),

CH₂Cl₂, 0 to 25 °C, 20 min; then 58, 25 °C, 2 h, 81%; (f) MnO₂(20 equiv),

Et₂O, 36 °C, 8 h, 73%. py ) pyridine.

Attempted Synthesis of Coleophomone A

Attempts To Convert Coleophomone B (2) to Coleoph-

omone A (1). With the synthesis of coleophomone B (2) now

completed, we moved on to the new goal of converting it into

coleophomone A (1), the third of our original four natural

product targets.

The Merck discovery team had originally reported⁴that

“coleophomones A and B exist in equilibrium with each other

under physiological conditions,” a statement possessing few

details, which was, fortunately, qualified in a footnote wherein

they added, “1 and 2 [coleophomones A and B, respectively]

exist in equilibrium in CH₃CN/water mixtures. The equilibrium

strongly favors 2 at pH g 7. The half-life of 1 at pH 7.5 is 5

min. Interconversion does not occur at pH < 3.” We were unable

to find a CD₃CN-D₂O ratio in which dissolved synthetic

coleophomone B (2) showed even traces of conversion to

coleophomone A (1) by ¹H NMR spectroscopy, despite the fact

that we also examined a full range of pH values by making

and then employing a series of buffered D₂O standards.

hoped to have synthetically derived coleophomones B (2) and

C (3) in our hands. From there, we anticipated no problems in

delivering coleophomone A (1) shortly thereafter based on the

literature precedent for the conversion of coleophomone B (2)

into coleophomone A (1).⁴The first task of this home stretch

was to install a double bond into the six-membered ring between

carbons 11 and 12 (Scheme 11). This task was accomplished

in both major series (i.e., starting from 53 and 54) through a

one-pot phenylselenide formation/oxidation/syn-elimination se-

quence, affording the R,â-unsaturated products 56 and 57, in

overall yields of 61% and 53%, respectively. The endocyclic

vinylogous ester product 56 was, unsurprisingly, still a single

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8883

A R T I C L E S

Nicolaou et al.

Scheme 12. Abortive Attempts To Synthesize Coleophomone A^a

group of 44b was hydrolytically removed (K₂CO₃, MeOH) in

a somewhat problematic reaction (yield 64%), the difficulties

arising from the insolubility of the tricarbonyl’s potassium salt

in a variety of solvent systems (K₂CO₃deprotonates the

tricarbonyl system forming the potassium salt). The primary

benzylic alcohol, obtained from this basic hydrolysis of 44b,

was oxidized using manganese dioxide in refluxing ether to

afford a complex mixture of isomeric products 60 (82%). The

complexity of the mixture so-obtained (60a:60b, 1:1.6 in CDCl₃)

was derived from the presence of at least two enol tautomers

of the open isomer 60a, existing alongside all possible stere-

oisomers in the closed isomer 60b in a dynamic equilibrium.

Not only was spectroscopic analysis difficult, but manipulation,

reaction monitoring, and purification of this mixture (i.e.,

removal of byproducts, inability to separate components due to

the aforementioned rapid equilibration) were all exceedingly

troublesome. We decided to press ahead despite the problems,

and we attempted to trap out the closed spirocyclic form 60b

by protection of its secondary benzylic alcohol to furnish

spirocycle(s) 61. We quickly found that no protection reac-

tion using basic conditions could be employed, because under

these conditions the open form of the starting material (60a)

predominated and the reactions merely furnished products

wherein the protecting group had been situated on one of the

possible enol groups, or we recovered starting material and/or

decomposed material. Under nonbasic conditions, modified

versions of standard procedures, a range of protecting groups

were tested: triethylsilyl (TES-OTf:Et₃N, 1.2:1.1 equiv),

methoxymethyl (MOM-Cl:Hu¨nig’s base, 2:1.8 equiv), para-

bromobenzoate (p-BrBz:Et₃N, 1.5:1.4 equiv), acetate (AcCl:

Et₃N, 1.2:1.1), and methyl (Meerwein’s salt BF₄OMe₃:proton

sponge, 1.5:1.3, or excess diazomethane). Degradation and/or

preferential protection of the tricarbonyl unit were seen under

most of these reaction conditions. The triethylsilyl-protection

did succeed, albeit in poor yield, only to furnish an unstable

product where migration of the silyl group led to rapid

decomposition of the material. This migration-degradation

sequence occurred even on TLC (thin-layer chromatography)

plates and, therefore, rendered this protecting group unusable.

Methyl protection, using Meerwein’s salt and proton sponge,

furnished 61 (P ) Me) in 13% yield alongside the products of

the regioisomeric protection of 60 as its vinylogous esters and

the recovery of starting material. However, when this scant

material (61, P ) Me) was subjected to our standard olefin

metathesis conditions, no traces of successful macrocyclization

could be detected whatsoever.

^aReagents and conditions: (a) K₂CO₃(2.2 equiv), MeOH, 25 °C, 12 h;

then H₂O (excess), 30 min, 64%; (b) MnO₂(10 equiv), Et₂O, 36 °C, 15 h,

82%; (c) py (12 equiv), CrO₃(6.0 equiv), CH₂Cl₂, 10 min; then 60, 0 to 10

°C, 5 min, 39%.

Furthermore, we investigated these solutions over a broad time

range (from 1 min to 24 h post preparation). Confused by our

inability to find conditions under which coleophomone B (2)

showed even the slightest propensity to undergo the requisite

aldol reaction to form coleophomone A (1), we appealed to

Shionogi and Merck for samples of the naturally obtained

coleophomones. Both groups were forthcoming, and we were

soon gratefully in possession of small quantities of coleoph-

omones A (1) and B (2). Shionogi¹and Merck⁴had both

previously described the quantitative transformation of coleoph-

omone A (1) into coleophomone B (2) upon treatment of the

former with a base, a procedure we found eminently reproduc-

ible even under the mildest of conditions. However, the reverse

reaction continued to elude us. At this point, we considered the

conversion of coleophomone B (2) into coleophomone A (1)

to be too shrouded in mystery to continue down this path, and

hence we opted to pursue another proposal, the de-novo

synthesis of coleophomone A (1).

Attempts To Synthesize Coleophomone A (1) Directly. The

ease with which coleophomone A (1) could be converted to its

open form congener, coleophomone B (2), through a retro-aldol

reaction suggested to us that the spirocycle of 1 was highly

strained and susceptible to opening. This deduction, combined

with the reluctance of coleophomone B (2) to undergo the

desired aldol reaction to form coleophomone A (1), led us to

propose closing, and, then, protecting the spirocycle moiety of

the molecule prior to macrocycle formation at a stage when

the structure generally had fewer strains and congested sites.

To this end, we returned to the previously synthesized olefin

metathesis precursor 44b (Scheme 12). The para-bromobenzoate

At this point, we wondered whether the spirocycle’s inherent

strains would preclude the desired olefin metathesis reaction

altogether, so we decided to investigate this reaction as a priority.

The obvious substrate for such an examination was compound

62, wherein the spirocycle had been locked shut by an oxidation

reaction. Spirocycle 62 was, therefore, synthesized by oxidation

of 60 using Collins reagent (CrO₃‚2py). As we had come to

expect for the reactions of 60, the oxidation proceeded in poor

yield (39%). Furthermore and unfortunately, despite the testing

of a broad range of olefin metathesis conditions (including

testing several different catalyst systems, varying solvents, and

using microwave or thermal assistance), no macrocylization was

ever observed. Bearing in mind the speed with which coleoph-

omone A (1) unravels to give coleophomone B (2), we felt the

9

8884 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

inherent strain of the former compound was just too great a

hurdle for the desired olefin metathesis reaction, and we

regretfully concluded that the limits of this approach had been

reached. In addition, the difficulty we had encountered in

securing the closed spirocyclic forms of these compounds

(possessing no macrocycle to add to the strain) in reasonable

yields added to the mystique surrounding the Merck team’s

reported⁴conversion of coleophomone B (2) to coleophomone

A (1). What seemed to be beyond doubt, however, was that the

synthesis of coleophomone A (1) would require a completely

redesigned strategy and, therefore, constituted another separate

project for consideration sometime in the future.

that there are now two potential initiation sites (both of the two

2-methylallyl functionalities) as opposed to just one (under our

reaction conditions, we never observed any indication, such as

spirocycle formation in the bisprenylated olefin metathesis

substrates, that initiation could occur at a prenyl group). Our

rationale would have it that it is initiation taking place at the

2-methallyl group on the 1,3-cyclohexadione of 67a that affords

the spirocyclic product 70 directly; however, initiation at the

other 2-methallyl group may also give 70 after a second

metathesis reaction and via the macrocyclic intermediate. The

results obtained from the metathesis of 68a upheld and extended

the inferences we had made up to this point. Vinylogous ester

68a yielded two products, the macrocycle 71³⁵and the spiro-

cycle 72 (10% and 65%, respectively), from its metathesis

reaction. Like 67a, 68a has two possible sites for initiation, and,

once again, we propose that the macrocycle 71 forms prefer-

entially when initiation occurs at the 2-methylallyl group

attached to the aromatic portion of the molecule, while the

spirocycle 72 forms when initiation occurs at the other 2-me-

thylallyl group (or when the macrocycle 71 is opened by a

second metathesis reaction, vide supra). The macrocycle 71

contains the new trisubstituted olefin (∆^16,17), as opposed to the

possible alternative of a tetrasubstituted double bond (∆^16,17).

This result is not at all surprising given the difficulty olefin

metathesis appears to have in forming such tetrasubtituted

double bonds, as evidenced by scant representation in the

literature;³⁶however, it does open the molecule up to an

interesting new reaction path. If this olefin metathesis reaction

was permitted to run for 48 h, rather than just until the

disappearance of the starting material by TLC (as in the previous

cases), none of the macrocycle 71 was isolated; instead

spirocycle 72 was the sole product, obtained with only a

marginal reduction in the yield (50%).

Unveiling Some of the Intricacies of the Coleophomone

Olefin Metathesis Macrocycle Formation Reaction. The more

we examined the results from our investigations described thus

far, the more the olefin metathesis reaction stood out as a

remarkable chemical achievement. It was a reaction with so

many intriguing subtleties that we wanted to understand in

greater detail. For example, we were curious about the origin

of the extraordinary selectivity that was observed wherein only

one geometric isomer of the macrocycle’s new olefin (∆^16,17

)

was formed in each reaction; in addition, what were the factors

that precipitated the clean and complete switch in this selectivity

from E- to Z- on going from one substrate to the next?

Furthermore, central to our strategy had been the use of a

bisprenylated 1,3-cyclohexadione, introduced based on our

hypothesis that this choice of substitution pattern would induce

initiation of the metathesis to occur solely at the 2-methylallyl

olefin attached to the aromatic domain of the molecule and,

thus, lead to preferential macrocyclization over the possible

alternative, spirocyclization. The veracity of this postulate

needed investigation. As a result, we decided to undertake some

further studies aimed at deciphering certain secrets of the olefin

metathesis reaction, within this context, including its key

features, and its scope.

Recalling that in the case of the real coleophomone systems

(53-55, Scheme 10) we had concluded that the macrocycliza-

tion was irreversible because no spirocycle formation was ever

seen even after long periods of exposure to catalyst A, we sought

an explanation for our current observation. Our rationale was

as follows: the trisubstituted olefins formed in these reactions

are, essentially, stable to the reaction conditions; they only show

slight degradation as a consequence of very prolonged exposure

to the catalyst. This assertion breaks down, however, when the

product still retains a possible metathesis initiation site. For

example, macrocycle 71 (Scheme 13) has a pendant 2-methyl-

allyl group, and we suggest it is this group that makes the

macrocycle 71 less stable in comparison to macrocycles 53-

An eye for detail will have led the reader to appreciate that

the coleophomone synthesis was complicated at many stages

by a proliferation of isomers (regio-, atrop-, and stereo-). To

circumvent problems arising from the latter in our ensuing

studies, we opted to exorcise the C-12 methyl group from our

model substrates. Thus, six model substrates, 67a-69a and

67b-69b, were synthesized, starting from 1,3-cyclohexadione

(64) and using our established reaction set, to allow us to

investigate the outcome of different olefin combinations in the

metathesis reaction (Scheme 13). On subjecting these substrates

to the action of Grubbs’ catalyst A in refluxing CH₂Cl₂, and

following isolation of the products of the ensuing reaction, we

were to learn a great deal about olefin metathesis in general.

55, for the olefin metathesis catalyst (now bearing a dC(CH

)

3 2

carbene group rather than the original dCHPh group, because

it has already participated in the metathesis catalytic cycle) can

insert into this 2-methylallyl olefin and from here intramolecu-

larly cleave the macrocycle, and, in so doing, form the

Beginning with the results for the series of compounds where

one prenyl group had been replaced with a 2-methylallyl

substituent (substrates 67a-69a), when the vinylogous ester 67a

was subjected to the standard olefin metathesis conditions for

4 h, a single spirocyclic product, 70, was isolated in 67% yield.

None of the macrocyclic alternative was isolated from this

particular reaction. Based on the later results that we will discuss

below, we felt that the macrocycle may have formed only to

succumb to rapid conversion into the spirocycle 70 via a second

metathesis reaction; however, this explanation remains conjec-

ture. If this metathesis precursor, 67a, is compared to the

corresponding coleophomone substrates (50-52), it can be seen

(35) The ∆^16,17double bond geometry was assigned for compounds 71 and 73

on the basis of the direct comparison of ¹H and ¹³C NMR spectral data

obtained for these compounds with that obtained for the corresponding

coleophomone metathesis products.

(36) For selected examples of olefin metathesis being used to form tetrasub-

stituted double bonds, see: (a) Hoye, T. R.; Jeffrey, C. S.; Tennakon, M.

A.; Wang, J.; Zhao, H. J. Am. Chem. Soc. 2004, 126, 10210-10211. (b)

Yao, Q.; Zhang, Y. J. Am. Chem. Soc. 2004, 126, 74-75. (c) Anastasia,

L.; Dumond, Y. R.; Negishi, E. Eur. J. Org. Chem. 2001, 3039-3043. (d)

Garber, S. B.; Kingsbury, J. S.; Gray, B. L.; Hoveyda, A. H. J. Am. Chem.

Soc. 2000, 122, 8168-8179. (e) Ackermann, L.; Fu¨rstner, A.; Weskamp,

T.; Kohl, F. J.; Herrmann, W. A. Tetrahedron Lett. 1999, 40, 4787-4790.

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8885

A R T I C L E S

Nicolaou et al.

Scheme 13. Olefin Methathesis of Model Compounds 67a,b, 68a,b, and 69a,b: Formation of Spirocycles 70, 72, and 74^a

^aReagents and conditions: (a) concentrated H₂SO₄(cat.), MeOH, 65 °C, 12 h, 88%; (b) LiHMDS (1.05 equiv), THF, -78 °C, 1 h; then prenyl-Br (1.1

equiv), -78 to 0 °C, 3 h, 85%; (c) LDA (1.1 equiv), THF, slow addition of a solution of starting material in THF:HMPA (7:1), -78 °C, 1 h; then 2-methylallyl

or crotyl-Br (2.0 equiv), -78 to 20 °C, 12 h, 78%; (d) 1 M HCl:THF (1:10), 25 °C, 14 h, 92-95%; (e) 41 (1.1 equiv), Et₃N (2.0 equiv), 4-DMAP (1.0

equiv), THF, 25 °C, 96 h, 81% of 66a and 83% of 66b; (f) excess CH₂N₂, Et₂O, 0 °C, 1 h, 23% of 67a plus 43% of 68a plus 23% of 69a while 17% of

67b plus 41% of 68b plus 20% of 69b; (g) cat. A (0.2 equiv), CH₂Cl₂, 40 °C, 4 h, 67%; (h) cat. A (0.2 equiv), CH₂Cl₂, 40 °C, 8 h, 65% of 72 plus 10%

of 71 while only 50% of spirocycle 72 after 48 h at 40 °C; (i) cat. A (0.2 equiv), CH₂Cl₂, 40 °C, 3 h, 50% of 74 plus 21% of 73 while only 41% of spirocycle

74 after 48 h at 40 °C.

thermodynamically more stable spirocycle 72. This hypothesis

provides a feasible explanation for the product distribution

obtained from the metathesis of all of the substrates 67a-69a

(and 67b-69b, see below) and explains why no spirocycle was

ever observed in the metathesis reactions of the coleophomone

substrates 50-52 (Scheme 10). The results obtained for the

metathesis of vinylogous ester 69a mirror those obtained and,

as already discussed, for 68a, with the exception that the other

geometric isomer of the new ∆^16,17double bond (this time the

Z-isomer) is formed in its macrocyclic product 73, in accord

with our coleophomone precedent. The precise reasons why

changing the position of the vinylogous ester from endocyclic

to exocyclic caused a complete switch in selectivity were not

found in these studies, or from the molecular modeling of the

coleophomone systems that we undertook at this time. A later

investigation,³²however, revealed that the use of the para-

bromobenzoate as a protection for the benzylic alcohol lay at

the heart of this issue.

Moving on to the substrates 67b-69b (Scheme 13) where a

prenyl group had been replaced with a crotyl group (predomi-

nantly in the trans form), all of these substrates, when subjected

to our standard metathesis conditions, decomposed producing

a mixture of unidentifiable products. We suggest that this

complete degradation occurs because the crotyl olefin is now

favored over the 2-methylallyl olefin alternative as the site for

initiation. With initiation occurring on a substituent of the six-

membered ring, rather than the aromatic portion of the molecule,

no macrocyclization occurs; instead the spirocycle forms rapidly.

We are confident that we have identified correctly this pathway

as being dominant because the macrocycles formed from these

substrates (67b-69b) would be the coleophomone type mac-

rocycles 53-55, (i.e. Scheme 10), which had previously been

shown to be stable to the reaction conditions. The spirocycles

proposed as initial products for these reactions contain disub-

stituted double bonds within the newly formed five-membered

rings; these bonds are rapidly cleaved by the olefin metathesis

9

8886 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

The Total Synthesis of Coleophomones B, C, and D

A R T I C L E S

Scheme 14. Total Synthesis of Coleophomone D^a

catalyst, and, from there onward, polymeric products begin to

appear as degradation continues apace.

The above work very clearly suggests that the initiation of

olefin metathesis reactions does not occur at prenyl groups

(under the range of conditions we investigated), but a prenyl

group may react in a metathesis reaction if prior initiation has

occurred elsewhere in the molecule. Likewise, endocyclic

trisubstituted double bonds are stable to the reaction conditions

unless the same molecule bears a second appropriate initiation

site. Furthermore, crotyl-type olefins become the initiation site

of preference when the choice is between a 2-methylallyl group

and a crotyl group. These results have, therefore, uncovered a

clear order of preference for the initiation step of an olefin

metathesis reaction using Grubbs’ catalyst A and demonstrate

how this information can be used to orchestrate a metathesis

reaction such that one alternative mode of reaction is favored

over another (e.g., macrocyclization over spirocyclization).^36a

In some cases, this option provides us the opportunity to design

substrates that can override the natural bias for producing a

lower energy product via a lower energy transition state (for

example, one can construct strained macrocycles in preference

to simple five-membered spirocycles).

Total Synthesis of Coleophomone D (4)

We hoped we had now accumulated all of the information

necessary to smoothly synthesize the last remaining coleoph-

omone family member, coleophomone D (4), thus bringing our

coleophomone adventure to an auspicious conclusion. In the

end, this confidence proved, at least for the substantial part, to

be well placed as a successful synthesis was forthcoming.

However, even the seemingly simple coleophomone D (4)

molecule held several thorny traps into which the unwary could

step and be caught. The proposed structure for coleophomone

D (4), of course, bore the unique and opposite pattern of

substitution about the aromatic ring from its sibling coleoph-

omones A-C (1-3), and we were prepared, therefore, to find

its structure had been wrongly assigned at the outset due to its

unusual dynamic isomerism.

Because coleophomone D (4) does not contain a macrocycle,

and, therefore, does not require an olefin metathesis reaction to

complete its synthesis, we wanted to avoid the clumsy sequence

of having to protect the tricarbonyl unit, at a late stage of the

sequence, solely to introduce the ∆^11,12double bond, steps that

would have to be followed by immediate deprotection. It seemed

highly desirable, therefore, to introduce the ∆^11,12double bond

into the six-membered ring while it was protected as the

vinylogous ester for the purpose of the two C-13 alkylation

reactions. To this end, the previously obtained vinylogous ester

75 was treated with LDA in the presence of HMPA as a key

additive to form the γ-enolate, which was, in turn, quenched

with PhSeCl (Scheme 14). Without isolation, the resulting crude

selenides (diastereoisomers) were oxidized in situ by treatment

with aqueous H₂O₂at 45 °C for 1 h. This efficient one-pot

reaction sequence afforded the rather labile vinylogous ester

76 in high overall yield (78%). Acidic hydrolysis of the

vinylogous ester 76, employing various conditions, always led

to the predominant formation of the aromatic product 78,

accompanied by only meager amounts of the desired 1,3-

cyclohexadione 77. This unanticipated aromatization (76f78)

must proceed via the initial protonation of 76 to afford a highly

^aReagents and conditions: (a) LDA (2.0 equiv), addition of a solution

of 75 (1.0 equiv) in THF:HMPA (25:1), -78 to 0 °C, 2 h; then PhSeCl

(1.5 equiv), -78 to 20 °C, 0.5 h; then 30% aqueous H₂O₂(excess), 45 °C,

1 h, 78%; (b) 1 M HCl:THF (1:5), 25 °C, 48 h, 50% of 78 plus 30% of 77;

(c) LiOH‚H₂O (5.0 equiv), MeOH:H₂O (2:1), 80 °C, 12 h, 91%; (d) Et₂AlCN

(1.0 equiv), toluene, 0 °C, 2 h, 59% of 81 plus 15% of 80; (e) PCC (4.0

equiv), CH₂Cl₂, 40 °C, 4 h, 51%; (f) 77 (1.0 equiv), 82 (1.0 equiv), Et₃N

(1.0 equiv), THF, 25 °C, 72 h, 80%; (g) K₂CO₃(3.0 equiv), MeOH, 25 °C,

24 h, 94%; (h) MnO₂(10 equiv), Et₂O, 36 °C, 4 h, 83%.

stabilized allylic tertiary cation (at C-10 T C-12), which is then

subject to the loss of a proton and a molecule of isoprene to

give the phenol 78, as indicated in Scheme 14. We avoided

this first ambush by developing a protocol for an alkaline

hydrolysis of 76, using LiOH in a methanol-water mix at 80

°C, which afforded 77 in high yield (91%).

To synthesize the acyl cyanide coupling partner 82, we

modified our existing set of reactions as is summarized below

9

J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005 8887

A R T I C L E S

Nicolaou et al.

and in Scheme 14. Aldehyde 79 was rapidly accessed starting

from 1,2-dimethyl anisole using a known, two-step literature

procedure.³⁷Treatment of 79 with Nagata’s reagent afforded

the desired cyanohydrin 81 (59% yield) as the major product,

accompanied by some of its regioisomer 80 (15%) wherein the

acetate group had migrated. The mixture (80 and 81) was

oxidized with PCC, and, following chromatographic separation,

the relatively fragile acyl cyanide 82 was isolated in 51% yield.

Acyl cyanide 82 was much more prone to decomposition via

hydrolysis than the corresponding acyl cyanide 41 (Scheme 8),

presumably due to its significantly reduced steric encumbrance.

The fragility of 82 was set to cause some problems in the next

step, the coupling of 82 and 77. When we employed the

conditions in this reaction that we had optimized for our previous

substrates, 8b and 41 (see Scheme 8b), hydrolysis competed

with coupling, such that the yields of the desired tricarbonyl

product were disappointingly poor. In a strange twist of chemical

fate, the answer to this conundrum lay with the additive

4-DMAP. Just as the addition of 4-DMAP had made the

coupling of 8b and 41 possible, so its removal from the protocol

held the key to making the coupling of 77 and 82 viable. Thus,

77 and 82 could be coupled successfully in the presence of one

equivalent of Et₃N only, at ambient temperature, in a yield of

80%.

We were now only two steps away from coleophomone D

(4). First, removal of the acetate protecting group was success-

fully achieved using methanolic K₂CO₃, to furnish 83 as a

mixture of atropisomers in high yield (94%), and, second,

oxidation of 83 with MnO₂gave coleophomone D (4) in 83%.

The ¹H and ¹³C NMR spectra of synthetic coleophomone D (4)

revealed the presence of all of the postulated isomers (D₁-D₄,

4a-4d, Scheme 1) whose signals were in accord exactly with

those reported by the Shionogi group,³thus demolishing the

idea that the structure of coleophomone D (4) had been assigned

wrongly and replacing it with the intriguing question of how

nature synthesizes the coleophomones with their regioisomeric

variation (vide supra). Until further studies are instigated, our

current knowledge allows us only to imagine what the answer

to this last question might be, speculation that we would rather

avoid delineating prematurely at this juncture.

Conclusion

With the notable exception of coleophomone A (1) whose

relationship to coleophomone B (2) and, as a direct result, whose

synthesis had eluded us throughout this investigation, this

concise total synthesis of coleophomone D (4) marked the

completion of our originally defined task, that of completing

total syntheses for the entire coleophomone family. The total

syntheses of coleophomones B (2) and C (3) using a remarkable

olefin metathesis reaction to form their highly strained and

congested macrocycle had pushed the frontiers of this venerable

reaction forward into a new domain. The successful coleoph-

omone macrocycle synthesis reported herein illustrates just how

powerful an ally the olefin metathesis reaction can be in forming

constrained medium-sized (10-12 membered) rings. Further-

more, our ensuing studies had unveiled a number of the

reaction’s intricacies that may be gainfully employed in the

future to design synthetic blueprints toward even more demand-

ing synthetic targets.

Acknowledgment. We thank Dr. T. Tsuri (Shionogi Co.) for

the generous gift of samples of naturally derived coleophomones

A, B, and C and for spectral data, and we thank Dr. K. E. Wilson

(Merck) for samples of naturally derived coleophomones A and

B, and for sharing relevant information and spectral data. We

thank Drs. D. H. Huang, G. Siuzdak, and R. Chadha for NMR

spectroscopic, mass spectrometric, and X-ray crystallographic

assistance, respectively. We thank Dr. K. Namoto for his

translation of the relevant patents from Japanese to English.

Supporting Information Available: Experimental procedures

and compound characterization. This material is available free

of charge via the Internet at http://pubs.acs.org.

(37) Box, V. G. S.; Yiannikouros, G. P. Heterocycles 1990, 31, 1261-1270.

JA0509984

9

8888 J. AM. CHEM. SOC. VOL. 127, NO. 24, 2005

Article Doi

The total synthesis of coleophomones B, C, and D

DOI: 10.1021/ja0509984

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja0509984

Full text of DOI:10.1021/ja0509984

Products guided by the article

R&D Labs maybe for 161358-55-0

Relevant to this article

Hot Product