Observations in the chemistry and biology of cyclic enediyne antibiotics: Total syntheses of calicheamicin γ1I and dynemicin A

16

J . Org. Chem. 1996, 61, 16-44

Perspective

Obser va tion s in th e Ch em istr y a n d Biology of Cyclic En ed iyn e

I

An tibiotics: Tota l Syn th eses of Ca lich ea m icin γ₁a n d Dyn em icin A

Samuel J . Danishefsky*^,†,‡and Matthew D. Shair^†,‡

Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for Cancer Research, 1275 York Avenue,

Box 106, New York, New York 10021, and Department of Chemistry, Havemeyer Hall,

Columbia University, New York, New York 10027

Received August 25, 1995

For sheer structural audacity, complexity, and chemi-

compounds are among the most potent prospective

anticancer drugs from the natural products domain

known to date.⁶There was certainly a perception, though

not yet realized, that such substances, or analogs thereof,

might be useful in oncology. Clearly, the goal of selec-

tively directing these small molecule cell killers to

diseased cells would pose no small challenge to the

science of pharmacology.⁷

cal novelty, few classes of natural products can rival the

cyclic enediynes.¹While there might have been a ten-

dency to dismiss compounds such as neocarzinostatin

choromophore (1),²esperamicin (2),³and calicheamicin

(3)⁴as bizarre secondary metabolites (Scheme 1), arising

Sch em e 1. En ed iyn e An tibiotics

Further drawing many in organic chemistry to enlist

to do combat in the “enediyne theater” was the mecha-

nism of the DNA cleaving properties of esperamicin and

calicheamicin. In a highly original insight following early

observations of Sondheimer⁸and of Masamune,⁹Robert

Bergman¹⁰had perceived and demonstrated the feasibil-

ity of electronic reorganization of a cis related 1,5-diyn-

Sch em e 2. Mod e of Action of Ca lich ea m icin (3)

a n d Esp er a m icin (2)

from biosynthetic aberrations, their high double-stranded

DNA cleaving ability (in some instances with significant

sequence selectivity) ensured that they would receive

serious attention.⁵Moreover, interest was augmented

by reports of the cytotoxicity of such agents. In fact, these

^†Sloan-Kettering Institute for Cancer Research. Tel: 212-639-5502.

Fax: 212-772-869.

^‡Columbia University. Tel: 212-854-6195. Fax: 212-854-7142.

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 17

Sch em e 3. Recen tly Discover ed Mem ber s of th e

En ed iyn e Cla ss of An tibiotics

3-ene to provide a 1,4-aryl diyl. Accordingly, when the

extraordinary cytotoxicity and DNA cleaving abilities of

esperamicin and calicheamicin were appreciated (even

before the structures were fully formulated), a mecha-

nism accounting for their activity could already be

surmised. Thus, for either drug, cleavage of the trisulfide

(Scheme 2) could set the stage for conjugate addition of

a thiyl species to the “anti-Bredt” double bond. This

accomplished, the setting would be in place for an

enediyne f diyl tranformation (see 4 f 5). It is diyl 5

which is presumed to be the active DNA cleaving agent

starting with hydrogen abstraction from 2-deoxyribose

residues. This simple, inspired concept of a “bio-Berg-

man” rearrangement to account for the mechanism of

action of 2 and 3 was supported by voluminous backup

studies. These included nonexperimental molecular

modeling,¹¹chemical modeling via syntheses of simple

congeners,¹²and theoretical argumentations as to the

relationship of diyl formation with precise molecular

structure. There thus arose an intellectually pleasing

framework in which chemists could contribute in a

detailed way.

Another highly inspired insight, this time by Andrew

Myers, served to include the older DNA cleaving agent

neocarzinostatin chromophore (1) in a more generalized

catechism.¹³A unifying theme emerges wherein a “cu-

mulene drug” (which exists in nature in pro-drug form)

upon suitable priming suffers transformation to a diyl

which, through the medium of hydrogen atom transfer,

functions as a DNA cleaving agent leading to cell death.

The repertoire of such “enediyne drugs” was further

broadened by the discovery of dynemicin A (6)¹⁴as

depicted in Figure 1. The combination of highly limited

availability, lability, and difficult solubility considerably

dims the prospects for dynemicin itself to emerge as a

useful drug.

There are many vantage points from which novel and

biologically potent chemical agents can be investigated.

The quest for total synthesis can be one of the most

illuminating. The scientific group undertaking a pro-

spective total synthesis must concern itself with the

complete “molecular personality” of its target. While, at

any particular point in time, the focus is likely to be

directed to a particular locale, or even a specific bond of

the target, these efforts must be choreographed within

the context of a comprehensive view of the problem.

Insights garnered form such “total molecule” undertak-

ings can be quite revealing. However, quite aside from

bioorganic and medicinal (analog synthesis) motivations,

members of our laboratory were enticed by the sheer

novelty and, even, beauty of the structures and, yes, by

the captivating challenge associated with the prospect

of the total synthesis of such natural products.

Below, we relate an account of the activities of our

laboratory which led to total synthesis of both cali-

cheamicin (2)^18-21and dynemicin (6).^22,23This essay is

not intended as a review of the enediyne or cumulene

drug field. Many excellent coverages of the enediynes,

including total synthesis in the case of 2, have been

provided.²⁴Rather, what we present is a personalized

reminiscence of our involvement, and the involvement

of our colleagues, in two rather special research projects.

Our calicheamicin effort started first. We began by

investigating the synthesis of a “core-structure” construct

of an aglycon version of calicheamicin (we called the fully

functional, but then unknown, aglycon calicheamicinone

(9, Scheme 4). In the opening phase, we were willing to

deal with a solution in which the urethan was deleted.

We would, however, set as a requirement that our core

construct contain the bridgehead double bond and that

it have sound provision for the eventual ketone at C₁₁

and for the two hydroxyl groups at carbons C₁and C₈,

and a credible modality for eventual installation of the

F igu r e 1. Dynemicin A.

At the level of chemical understanding, Semmelhack

contributed the basic idea around which much of the

dynemicin bioorganic investigations have been centered.¹⁵

He argued, quite sensibly, that the quinone substructure

provides the stabilizing element which protects dynemi-

cin from a virtually spontaneous self-destruct pathway.

This will be discussed later. For the moment, suffice it

to say that Semmelhack’s insight prompted the search

for a variety of other ways to store and to unleash

dynemicin-like activity. More recently, the list of ene-

diyne drugs has been expanded with some further

interesting additions (see 7¹⁶and 8¹⁷in Scheme 3).

The background discussion provided above is certainly

not presented in the spirit of a general review of cumu-

lene or enediyne drugs. It is offered to provide an

understanding of the stimulation and excitement which

these agents provided to organic chemists, including

those in our laboratory.

18 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

allylic trisulfide at C₁₃. Keeping in mind the need for a

“global molecular solution”, we were wary of alluring

short term successes which upon subsequent analysis

lacked plausibility for evolving into a fully comprehensive

program of total synthesis.

problems posed by the approach of 10 + 11 f 12. Thus,

it was hoped that in a spiro dienone of the type 15 the

future C₁₃functionality could be stored as an epoxide

while an enol ether would guard the future C₁₁ketone.

It was even hoped, though not realized in practice (vide

infra), that the methylene group of the spiroepoxide in

15 could be exploited to serve as a member of the two-

carbon appendage that must eventually be added to C₁₃.

The journey began with diester 16 (Scheme 5) which,

upon reduction, gave rise to diol 17. Becker-Adler

oxidation of 17 served to differentiate the hydroxymethyl

groups ortho and meta to the phenolic hydroxyl, thereby

affording 18 and then, after Dess Martin oxidation, 19.

Sch em e 4. In itia l Syn th etic P er cep tion s of a

Ca lich ea m icin on e (9) Syn th esis

Sch em e 5. Syn th esis of Keto Ep oxid e (19)^a

a

(a) DIBAH, THF; (b) NaIO₄, THF-H₂O, 75% overall; (c) Dess-

Martin periodinane, CH₂Cl₂, 70%.

Dennis Yamashita, who initiated this project, first

learned how to prepare and condense the dilithio ene-

diyne 10 with excess benzaldehyde.²⁶He also found it

possible to deliver the dilithio agent to the aldehyde

function of 19 (Scheme 6). However, there was no sign

A concept which was captivating in its formal simplic-

ity virtually imposed itself for putting together a core

structure which meets the specifications outlined above.

Such a core might be assembled by interpolating a (Z)-

hex-3-ene-1,4-diyne unit (10) as a “double nucleophile”

between “ketonic” and “aldehydic” centers in a domain

of the type 11. At this stage, we were not necessarily

committed to the delivery of 10 as a single six-carbon

unit. Perhaps, the overall unit would be delivered in

modular form (see dotted lines). Even if the diynene

segment were to be presented as a six-carbon unit, we

maintained an open mind as to the issue of timing. The

two nucleophilic attacks could be delivered in a single

strike or staged in staccato-like steps, as appropriate.

Closer analysis of hypothetical construct 11 helped to

sharpen the specificity of the thought process. There was

obvious concern about equipping 11 with functionality

implements which would serve the needs of the project

over the long haul. The groups X and Z must not be

competitive as electrophiles with the ketonic and alde-

hydo centers designated for the bridging strategy. Two

perceptions of the nature of X were considered. In the

most obvious view, “X” represents a type of ketal protect-

ing group. Alternatively, X might represent an enol ether

function. Again, the most obvious view of Z would be

that of an acetal type of linkage. There was concern that

prototype structures of the type 13 or 14 might be

difficultly accessible and chemically unwieldy, particu-

larly in regard to differential exposure of functionality

at future carbons 11 and 13.

Sch em e 6. Ad d ition of Dilith io En ed iyn e 10 to

Keto Ald eh yd e 19

of cyclization to the ketone. All attempts to accomplish

the cyclization of 20 to 21 in a separate step were

unsuccessful. Although confirmatory products were never

characterized, we began to suspect the chemical lability

of the proton at C₈of adduct 20 (see circle).

It seemed that the Becker-Adler reaction²⁵might

provide the basis for a solution to several of the interactive

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 19

Sch em e 8. Syn th esis of Cyclic En ed iyn e 31^a

We wondered about the possibility of conducting the

cyclization (Scheme 7) in the reverse sense (cf. 22 f 23).

The logic was that in the case at hand the intramolecular

step (i.e., the cyclization) may actually be more demand-

ing than the first coupling. In that case, it would be

better to reserve the aldehydo electrophile for the most

problematic phase. This line of thinking first led us to

explore addition to reduced versions of the future alde-

hyde group (cf. 24 and 25). Addition was indeed achieved,

but the lability of 26 and 27 was such that we were never

able to make our way to enediyne aldehyde 22 to attempt

cyclization.

Sch em e 7. Attem p ted Syn th esis of Ald eh yd e 22

a

(a) LiNMePh, THF; (b) 10; (c) H₃O⁺, 40% overall; (d) TMS(O-

Given these disappointing results, the only recourse

for reaching a system of the type 22 called for delivery

of 10 to the ketone of keto aldehyde 19. At this stage

Dr. Nathan Mantlo had joined the project. Indeed,

following the chemistry of the in situ protection of

aldehydes (originally developed by Comins for the pur-

pose of lithiation of aryl bromides in the presence of

resident “aldehydes” ²⁷) Mantlo accomplished (Scheme 8)

this goal (see presumed intermediates 28 and 29 and

product 22). Moreover, the addition of 10 was quite

stereoselective. The resultant 22 corresponds to the

acetylide having added to the ketone syn to the oxygen

face of the spiroepoxide. After nontrivial protection of

the tertiary alcohol, silyl ether 30 was in hand. In the

event, deprotonation of 30 with lithium hexamethyldi-

silazide and warming to rt led to a 40-60% yield of core

system 31. When the cyclization was conducted under

the strictly defined conditions shown, this compound was

obtained with high stereoselectivity. It is the product

which would formally arise from addition of the acetylide

to the s-trans version of the enal.

COCF₃), Et₃N, CH₂Cl₂, 72%; (e) KHMDS, PhCH₃, 39%.

nucleophiles failed at the level of the ring-opening

reaction. This was certainly one of the low points of the

project.

Sch em e 9^a

Having achieved the first synthesis of a core structure

of an enediyne drug,²⁸attentions were directed to elabo-

rating C₁₃by opening of the spiroepoxide with a carbon-

centered nucleophile. Experiments were conducted on

the core system containing the enol ether linkage present

in the enone 32, or on the ketal 33 (Scheme 9). While

we were able to open the epoxide linkage with cyanide

as the nucleophile, the â-cyanohydrin thus obtained did

not lend itself to dehydration. Difficulties ranged from

general decomposition to nonreactivity. A range of other

a

(a) (CO₂H)₂, 95%; (b) (CH₂OH)₂, PPTS.

In the 11th hour of his stay at Yale, Mantlo (who was

Merck bound) discovered that enol ether epoxide 31

(Scheme 10) could be subjected to high-yielding acetolysis

to afford 34. The diol function in the derived 35 could

20 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

be cleaved to produce a ketone (cf. 36). While the hope

of exploiting the methylene carbon of the spiroepoxide

had been frustrated, the Becker-Adler chemistry in

conjunction with the enediyne interpolation concept (after

significant on-site modification) had proved to be work-

able. At this stage, Dr. J ohn Haseltine (currently at

Georgia Tech) replaced Dr. Mantlo on the project.

11). His elegant intramolecular Emmons reaction af-

forded 37 which, after two stage reduction, gave rise to

38.

In attempting to set the stage for installation of the

allylic trisulfide, a very serious setback in terms of yield

was sustained. All attempts to carry out displacement

of the primary allylic alcohol (cf. 38) with thiyl-based

nucleophiles were severely complicated by formation of

a cyclic ether in competition with the desired inter-

molecular displacement product. The formation of 39

and 40 under the conditions shown is illustrative. Not

withstanding this setback, Haseltine showed how thio-

acetate 40 could be cleaved reductively (Scheme 12) to

Sch em e 10. In sta lla tion of th e C₁₃Keton e of 36^a

Sch em e 12. Syn th esis of

Desca r ba m oylca lich ea m icin on e 43^a

a

(a) KOAc, HOAc, DMSO, 95%; (b) (CH₂OH)₂, CSA, 80%; (c)

NH₃, MeOH; (d) HIO₄, THF, 84%.

Yamashita began our first explorations directed toward

incorporating the urethane linkage, looking toward the

synthesis of a fully equipped aglycon. Haseltine under-

took the not inconsiderable challenge of charting a line

of chemistry to proceed from 36 to reach descarbamoyl-

Sch em e 11. Syn th esis of Th ioa ceta te 40^a

a

(a) DIBAH, CH₂Cl₂; (b) CSA, THF-H₂O, 100%.

generate thiol 41. The latter was converted to trisulfide

42,²⁹which, remarkably, could be readily deprotected to

provide descarbmoylcalicheamicinone 43.³⁰

Following these critical experiments, the campaign to

reach calicheamicinone was intensified. With Haseltine

leaving for Georgia Tech, the cause was taken on by Dr.

Maria Paz Cabal (currently at SUNY-Buffalo) and Dr.

Robert Coleman (currently at the University of South

Carolina). The thought was to include provision for the

eventual urethane linkage in our aromatic precursor

(Scheme 13) and then conversion to a spiroepoxide

similar to 19 (Scheme 5). Two possibilities were enter-

tained at the experimental level. In one version, the

interim goal system would be 45, which contains either

the intact urethane, some form of an amino group, or,

less likely, a nitro precursor. The other route, with

system 49 as its interim target, involved placing a

bromine atom where the urethane would eventually

appear. These possibilities were investigated concur-

rently.

Unfortunately, no pathway leading to any version of

the system corresponding to 45 was validated in practice.

Attempts in this direction broke down even at the stage

of projected early aromatic intermediates en route to the

Becker-Adler reaction. The problem of installing and

maintaining various amino permutations (or even a

a

(a) DCC, (EtO)₂P(O)CH₂CO₂H; (b) LiBr, Et₃N, THF, 76%; (c)

DIBAH, CH₂Cl₂; (d) NaBH₄, MeOH, 69% overall; (e) DIAD, PPh₃,

AcSH, 52%.

calicheaminone. After continuing trial, error, and ret-

rofitting, Haseltine worked out a viable route (Scheme

Perspective

Sch em e 13. Syn th esis of Br om id e 49^a

J . Org. Chem., Vol. 61, No. 1, 1996 21

Sch em e 14. Syn th esis of En ed iyn e 51^a

a

(a) LiNMePh, THF; (b) enediyne 10; (c) TMS(OCOCF₃); (d)

KHMDS, PhCH₃, 60%.

resultant vinyllithium reagent would be carboxylated,

thus, setting into motion a possible Curtius degradation

en route to the desired urethane. Surprisingly, some

progress in this regard was realized in that the lithiation

and carbomethoxylation steps could actually be ac-

complished on compound 51 in this multifunctional

setting, though in poor yield. However, several attempts

at Curtius and related degradations failed.

a

(a) NaH, then DIBAH, THF; (b) NaIO₄, THF-H₂O; (c) Dess-

Martin periodinane, CH₂Cl₂, 40% overall.

precursor nitro group) through the steps required to

reach diol 44 did not prove to be manageable in an

acceptable time frame.

More workable, though far from simple, was a route

centered around a bromine handle. This goal was

eventually achieved starting with the bromo ester alde-

hyde 46. The inclusion of the bromine caused several

serious technical problems. Particularly frustrating was

the matter of retaining the bromine atom while reducing

the ester aldehyde (46 f 47). The use of lithium

aluminum hydride led to extensive debromination in

addition to the desired reduction. This problem could be

circumvented through the agency of diisobutylaluminum

hydride as the reducing agent. Unfortunately, the

workup protocols necessitated by this reducing agent

proved to be difficultly manageable on a large scale, and

yields of desired bromo diol suffered. The Becker-Adler

oxidation of crude 47 was successful and provided the

desired spiroepoxy dienone 48 containing the bromine.

Again, Dess-Martin oxidation gave rise to 49.

Here again, the yield of addition (Scheme 14) of

compound 10 to aldehyde 49 seems to suffer from the

presence of the bromine group. Once more, the hydroxyl

group of the resultant and unstable 50 had to be

protected as its silyl ether. In the event, cyclization of

this compound gave the desired 51. In this case, in

contrast to the desbromo series, the intramolecular

addition of the acetylide to the aldehyde seemed to be

stereospecific, producing only 51. This finding could be

rationalized. The presence of the ortho bromo function

likely favors the s-trans enal form required to produce

the desired S configuration at C₈. Thus, we had in hand

51 (a modified version of 31), where the bromine would

serve as the eventual access point for introducing the

urethane.

The alternative possibility was to find a point in the

progression wherein the vinyl bromide might be displaced

by an azido nucleophile. The idea was to take advantage

of a niche in the synthesis in which the electronic

character at C₁₃would favor addition by azide ion and

displacement of bromide ion. At an appropriate stage,

the azide would be reduced to an amine which, following

carbomethoxylation, would provide the urethane.

We proceeded as follows (Scheme 15). The spiro-

epoxide 51 was converted to hydroxy acetate 53 in

accordance with precedent in the descarbomyl series. (We

shall return to this compound shortly in dealing with the

synthesis of the optically pure aglycon). Compound 53

was converted to diol 54 and then to ketone 55. In this

substrate, it was possible to achieve addition of azide and

elimination of bromide (see compound 56). We processed

compound 56 onward (Scheme 16), reaching unsaturated

lactone 57 by intramolecular Emmons reaction. In this

substrate, the azide function underwent reduction to a

vinyl amine (stabilized no doubt by vinylogous conjuga-

tion to the lactonic carbonyl center). Two-stage acylation

of the crude amine via “diphosgene” led to 58. Thus was

the urethane installed.

Reduction of the unsaturated lactone, through the

previously described two-stage protocol, afforded 59.

Once again, we were not spared from the ravages of the

low-yield Mitsonobu-type conversion of 59 to 60. Com-

pound 60 did, indeed, respond, albeit in lower yield, to

the same protocol used in the model descarbamoylcali-

cheamicinone series to provide the aglycon calicheami-

cinone (9).²⁰

Two routes were now pursued for this conversion. In

one route, the vinyl bromide would be lithiated and the

Before returning to the total synthesis project, it is

worth recalling some insights gathered into the mode of

22 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Sch em e 15. Syn th esis of Azid e 56^a

action of calicheamicin which were rendered possible by

the exercises described above. Thus, an early demon-

stration proceeded via enedione 61 (Scheme 17) obtained

from 36. The latter was reduced to the dihydro com-

pound 62, as shown. This compound would, in principle,

be a candidate for spontaneous Bergman cyclization. As

it turned out, Bergman cyclization occurred only upon

thermolysis. Following a precedent of Magnus,³¹62 was

reduced to a diol. The latter underwent virtually spon-

taneous Bergman cyclization in the presence of 1,4-

cyclohexadiene to produce 64.³²Thus, in a fully elabo-

rated series, a new triggering mechanism for diol

formation was established via reduction of the bridgehead

double bond. Furthermore, in keeping with the prece-

dent of Magnus, the role of the hybridization at C₁₃was

verified in the more fully equipped system.

Sch em e 17. Exp lor a tion of Va r iou s Tr igger in g

Mech a n ism s for En ed ion e 61

a

(a) (CH₂OH)₂, THF, CSA, 89%; (b) KOAc, HOAc, DMSO, 88%;

(c) NH₃, MeOH; (d) NaIO₄, acetone-H₂O, 80% overall; (e) NaN₃,

MeOH, H₂O, 82%.

Sch em e 16. Syn th esis of Ca lich ea m icin on e (9)^a

Moreover, allylic alcohol 38 (Scheme 18) provided

another opportunity for triggering a Bergman reaction.

Cleavage of the ketal linkage led to a ketone which, in

the presence of 1,4-cyclohexadiene, afforded 65. We had

shown for the first time that an allyllic alcohol Michael

addition reaction will inaugurate the bioactivation cas-

cade. Similarly, thioacetate 60, upon deketalization, gave

rise to 66 (Scheme 19). Exposure of 66 to diethylamine

sets into motion a Bergman-like process leading to 67.^20b

These demonstrations, in the context of fully elaborated

structures, provided more persuasive, though still sug-

gestive, evidence to their pertinence to the drug itself.

Another important observation involved the capacity

of the thioacetate and the trisulfide, both in the urethane-

containing series and the descarbamoyl series, to effect

cleavage of DNA. It was found that the aglycons them-

selves are quite cytotoxic and do induce cleavage of

a

(a) ClC(O)CH₂PO(OEt)₂, pyridine, THF, 64%; (b) LiBr, Et₃N,

THF, 92%; (c) H₂S, piperidine, MeOH, 85%; (d) (Cl₃C)₂CO, pyri-

dine, then MeOH, 82%; (e) DIBAH, CH₂Cl₂; (f) NaBH₄, MeOH,

H₂O, 65%; (g) DIAD, (3-OMe)Ph₃P, AcSH, THF, 60%; (h) DIBAH,

CH₂Cl₂; (i) Harpp reagent, CH₂Cl₂, 46%; (j) CSA, THF, H₂O, (65%).

Perspective

Sch em e 18. Mich a el Ad d ition of 38 a s a

J . Org. Chem., Vol. 61, No. 1, 1996 23

aglycon lacks any semblance of sequence selectivity. This

study utilizing synthetic aglycon was the first to pinpoint

the critical importance of the carbohydrate domain in

mediating DNA cutting properties of the drug.³³

Tr igger in g Device

These findings, once again, underscored the unique role

of synthesis in providing systems which are not available

from the natural product. All previous attempts to

retrieve aglycon from deglycosylation of natural cali-

cheamicin were unsuccessful. Indeed, such hydrolysis

was also not a viable way of producing the carbohydrate

domain. That, too, has to be fashioned by total synthesis

(vide infra).

Another important step in moving toward a total

synthesis of calicheamicin itself was that of obtaining the

aglycon in optically pure form. A key step in this regard

was accomplished by Dr. Vincent Rocco (currently at Eli

Lilly) via an enzymatically specific acetylation, following

protocols of Wong conducted on intermediate 54 (vide

infra).³⁴This gave rise, eventually, to the two antipodes

of calicheamicinone ketal and calicheamicinone itself. It

was instructive to compare the DNA cleaving capacity

of the two antipodes.³⁵Surprisingly, the unnatural

antipode exhibited greater double-stranded cleaving

capacity than did the natural one. In the unnatural

series, approximately 25% double-stranded cleavage was

noted. Once again, the unique role of this sugar in

modulating drug-DNA recognition of the drug had been

illustrated through the medium of synthesis.

Sch em e 19. Activa tion by Clea va ge of

Th ioa ceta te 66

F igu r e 2.

Naturally, as the importance of the carbohydrate

domains was becoming apparent, and as success in the

synthesis of calicheamicinone was becoming more prob-

able, our horizons began to shift toward its carbohydrate

sector and that of esperamicin. There were several

considerations involved in taking on this challenge.

First, it was emerging that only synthesis could provide

the fully operational carbohydrate entity.¹⁹All attempts

to wean this intact domain from the drug were unsuc-

cessful. One of the sources of this vulnerability was to

emerge as a major issue in our synthetic work (vide

infra).

Insofar as we were interested in the biological proper-

ties of the carbohydrate sector itself, which in fact turned

out to be fascinating, we would have to assemble it.

Furthermore, this goal was, in itself, a synthetic chal-

lenge of some scope. Part of our agenda was its encour-

agement to extend our synthetic concepts in glycal

assembly to some new structural types.³⁶In addition,

oligonucleotide constructs. The cleavage included a

residual double-stranded cleaving capacity of about 10-

15% in the racemate, which is considerably less than that

of the drug itself, with suitably chosen oligonucleotide

targets.

Indeed, it was found that DNA cleavage is more readily

realized from the thioacetates than via the trisulfide. This

trend was subsequently confirmed by Nicolaou and co-

workers in the context of the fully synthetic drug analog

and drug.²⁴

Moreover, the aglycon retained, to only a limited

extent, the capacity to participate in direct hydrogen

atom transfer from the oligonucleotide. This was estab-

lished by conducting reactions in a deuterated medium.

An experimentally reproducible and meaningful amount

of aromatized aglycon containing proteo functions at

carbons 3 and 4 was obtained by Nobuharu Iwasawa.

However, once again, the efficiency of this transfer was

far less than that of the native drug. These studies

taught us for the first time that the bulk of the molecular

recognition which leads to double-stranded cleavage

capacity and hydrogen atom transfer resides in the

carbohydrate entity rather than in the aglycon itself. This

point was further corroborated by the fact that the

we began to think about the possibility of achieving a

I 21

total synthesis of calicheamicin γ₁

itself. If the car-

bohydrate domain could be assembled, perhaps it could

be delivered in a suitable form to a workable version of

calicheamicinone.

Assuming the chemical issues could be negotiated,

several questions could be addressed. For instance, it

would be necessary to ascertain the stability limits of the

24 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Sch em e 21. Syn th esis of Th iosu ga r 80b^a

carbohydrate domain of either calicheamicin or espe-

ramicin if the “reducing end” terminates in a free hy-

droxyl. What would be the minimum protection required

to stabilize a molecule with a free reducing end? As-

suming such protection were possible, could a viable

glycosylation donor function be installed from such a free

anomeric hydroxyl? Assuming this were achieved, could

glycosylation with aglycon actually be achieved? Finally,

assuming this were all feasible, could the coupling be

executed at a sufficiently advanced stage such that

deprotection and retrieval of the final product could be

managed?

The conceptual framework for synthesizing the carbo-

hydrate domains (both for esperamicin and calicheamicin

I

γ₁) was that of glycal assembly. A review of the overall

logic and practice of glycal assembly is beyond the scope

of this report. Moreover, various facets of glycal assembly

have been reviewed elsewhere.³⁷Thus, here we confine

ourselves to the sequences which were successfully

implemented to attain the goals enumerated above.

Randall Halcomb (currently at the University of Colo-

rado, Boulder), Dr. Mark Wittman (currently at Bristol

Myers), and Steven Olson first probed the issue of the

stability of the aglycon-free carbohydrate domain in the

less elaborate esperamicin series. Epoxidation of rham-

nal derivative 67a (Scheme 20), followed by methanolysis,

afforded 68. Compound 68 was used as the glycosyl

acceptor in Thiem-like³⁸iodoglycosylation with glycal 70

(prepared form the previously known phenyl thioglyco-

side 69) to give 71. De-iodination gave 72 and eventually

triflate 73.

a

(a) PhSH, SnCl₄, CH₂Cl₂, 96%; (b) NaOMe, MeOH; (c) TsCl,

CHCl₃, TBAB, 86%; (d) LAH, THF, 91%; (e) MsCl, Et₃N; (f) KSAc,

DMF, 95%; (g) LAH, THF; (h) DNP-F, 89%; (i) m-CBA, CH₂Cl₂,

then Et₂NH, THF, 76%; (j) TBSOTf, pyridine, CH₂Cl₂.

series. However, in our hands, the projected thio-Ferrier

rearrangement failed with 6-deoxy substrates. Accord-

ingly, it was necessary to conduct this reaction in the

galactose series bearing an oxygen function at C₆. Sub-

sequent deoxygenation at C₆was followed by installation

of an R-methylthio group at C₄(see 78 f 79a ). Inversion

of the triflate was accomplished with thioacetate to afford

79a and then the 2,4-dinitrophenylthiolate 79b.

It was at this stage that the rearrangement of the

sulfoxide corresponding to 79b was carried out. The axial

alcohol of the resultant 80a was protected as its silyl

ether (see compound 80b).

Sch em e 20. Syn th esis of Tr ifla te 73^a

At approximately the time these investigations were

progressing, Falck had described a useful direct conver-

sion of glycals to 2-deoxyglycosides.³⁹The Falck logic

proved applicable to the case at hand. Treatment of 80b

with

N-(trimethylsilylethoxycarbonyl)hydroxylamine

(TEOC-NHOH, Scheme 22) in the presence of Ph₃P‚HBr

Sch em e 22. Syn th esis of Hyd r oxyla m in e 82c^a

a

(a) 2,2-Dimethyldioxirane, CH₂Cl₂; (b) MeOH, 68%; (c) m-

CPBA, CH₂Cl₂; (d) benzene, reflux; (e) I⁺ClO₄(sym-collidine)₂,

-

CH₂Cl₂, 49%; (f) Ph₃SnH, AIBN, benzene, reflux, 84%; (g) Tf₂O,

pyridine.

The preparation of the coupling partner (80b) for the

triflate required a fairly lengthy sequence. It started

with ^D-galactal triacetate (74), employing a phenylthio

version of the Ferrier rearrangement (Scheme 21). The

plan was to install the C_3ahydroxyl group by 2,3-

sigmatropic rearrangement of a derived pseudo glycal

R-sulfoxide. Indeed, it was for the esperamicin carbo-

hydrate that this chemistry for reaching glycals with

3-axial alcohols was developed (vide infra).

a

(a) TEOC-NHOH, Ph₃P-HBr, CH₂Cl₂, 57%; (b) EtSH, K₂CO₃,

MeOH; (c) MeI, DBU, benzene, 87%.

afforded only 2-deoxy-â-glycoside product 82a . Given

Falck’s less than favorable precedents at the level of

stereochemistry, this stereospecificity presumably reflects

the substantial hindrance provided by the 3-R (axial)

It would of course have been much preferable to

conduct the desired sequence directly in the ^D-fucal

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 25

OTBS group to glycoside formation. Unfortunately,

however, the addition to 80b produced a 3:2 mixture of

desired 82a and N-glycosylated 81. The next phase

involved conversion of the 2,4-dinitrophenyl protecting

group on sulfur to the required S-methyl function (see

compound 82c, R ) CH₃).

Coupling of deprotonated 82c and 73 under the condi-

tions worked out by Kahne and co-workers afforded 83⁴⁰

(Scheme 23). Hydrazinolytic cleavage of the N-phthaloyl

action of p-methoxybenzyl alcohol. Once again, the

epoxide opened to give 86. Unfortunately, in this in-

stance, with p-methoxybenzyl alcohol as the acceptor,

significant erosion of stereoselectivity occurred and 86

was accompanied by a substantial amount of R-glycoside

87. As will be seen in the calicheamicin series, each of

these two compounds could be advanced equally well in

the synthesis, but the need for partitioning, so as to

operate with homogenous materials, was certainly a large

inconvenience. In this orienting phase of the esperamicin

work, we only operated with the â-glycoside 86.

Sch em e 23. Syn th esis of Esp er a m icin

Tr isa cch a r id e 84^a

Sch em e 24. Syn th esis of Disa cch a r id e 89^a

a

-

(a) 2,2-Dimethyldioxirane, CH₂Cl₂; (b) PMB-OH; (c) I⁺ClO₄

-

(sym-collidine)₂, CH₂Cl₂; (d) Ph₃SnH, AIBN, benzene, reflux.

As before, in the methyl glycoside series, compound 86

reacted with the phthalimido glycal 71 to give, in this

instance, 88 and then 89 (Scheme 24). Following closely

the protocols described above, triflate 90 (Scheme 25) was

Sch em e 25. Syn th esis of Tr isa cch a r id e 91^a

a

(a) NaH, DMF, 78%; (b) N₂H₄, EtOH; (c) acetone, NaCNBH₃,

iPrOH, 85%; (d) DDQ, CH₂Cl₂-H₂O, 99%; (e) TBAF, THF, 92%.

function and introduction of the isopropyl unit on nitro-

gen was followed by cleavage of the O-silyl protecting

groups to afford 84. This compound corresponds to the

full trisaccharide carbohydrate sector of esperamicin with

the reducing end blocked as a methyl glycoside. Pleasing

as was this demonstration, it left unaddressed the matter

of fashioning this type of domain to serve as a competent

glycosyl donor. Since we had reached the aglycon of

calicheamicin by total synthesis, we were, of course,

aiming at solving this problem with the calicheamicin

carbohydrate sector. However, it seemed that the some-

what simpler esperamicin domain would serve as a

convenient model. The specific question to be faced was

the protection levels needed throughout the domain as

the donor element was being installed via a construct

bearing a free (OH) reducing end. In particular, we had

learned from our collaborators at Bristol-Myers about an

apparent incompatiblity in a fully unprotected construct,

i.e., 85 (vide infra).

a

(a) Tf₂O, pyridine; (b) NaH, DMF.

To probe this area, we started with the same rhamnal

derivative 68 (Scheme 24). This time, after epoxidation

with dimethyldioxirane, the product was subjected to the

produced and coupled with the previously described 82c.

This gave rise to the fully protected system (91) in which

a p-methoxybenzyl group protects the latent anomeric

26 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Sch em e 26. Syn th esis of 97^a

a

(a) N₂H₄, EtOH; (b) acetone, NaCNBH₃, iPrOH; (c) DDQ, CH₂Cl₂-H₂O; (d) TBAF, THF; (e) MeOH, AcOH.

hydroxyl while a TEOC function guards the hydroxy-

lamine “linker”. Again, we first proceeded with cleavage

of the phthalimide group (Scheme 26) and installation

of an N-isopropyl function (cf. 93). This was followed by

full deprotection, including the TEOC group, leading to

the formation of 96. Once again, as with the case of

methyl glycoside 84, capping of the reducing end stabi-

lizes the otherwise fully deprotected system in unrear-

ranged form (vide infra). Similarly, cleavage of the two

p-methoxybenzyl groups resulted in the formation of 94

where the domain terminates in a free anomeric hydroxyl

on the A ring sugar. The substance was well character-

ized and appeared to be a stable entity. However, upon

treatment of 91 with tetra-N-butylammonium flouride

(for purposes of cleavage of the TEOC function), there

was isolated not 95 but, in its stead, the pyrrolidine

derivative 96, best characterized as its methyl glycoside

97. This compound proved to be identical with a speci-

men obtained from the Bristol-Myers Company. Hence,

we had demonstrated two important features of the

tricyclic esperamicin domain. With the reducing end

capped, the system was quite stable even in the absence

of protection of the hydroxylamine spacer. Conversely,

with the hydroxylamine spacer in place in blocked

(TEOC) form, the reducing end could be liberated in the

form of a free hydroxyl. What did not appear to be

feasible was retention of the core carbohydrate domain,

in the face of simultaneous deprotection of the N-Teoc

and anomeric p-methoxybenzyl groups. This combination

(see hypothetical compound 95) leads to very rapid

rearrangement, presumably by ring opening and reclo-

sure to the N-hydroxypyrrolidine type of valence isomer

(see 96).^41,42

findings in the esperamicin area. The AEB system, at

the fully protected stage, was in fact identical with the

one used in the esperamicin series (see triflate 90).

For calicheamicin, only a modest modification was

necessary in the construction of the thiosugar. Thus, it

would be presented in the form of the free thiol derivative

82b. The CD system, however, had to be constructed de

novo since it has no counterpart in the esperamicin

domain. Actually, some strikingly new chemistry was

employed in the construction of the B ring of this

subdomain by Steven Olson. His synthesis started with

the commercially available quinone derivative 98 (Scheme

27) which was converted, according to precedents of

Evans, to cyanohydrin 99. Treatment of 99 with sa-

marium diiodide produced 100. This chemistry has been

used as a general synthesis of extensively substituted

p-hydroxybenzonitrile groups, but here we confine our-

selves to the calicheamicin connection. Iodination of the

single available site was carried out with iodonium

chloride to give 101.⁴³

Sch em e 27. Syn th esis of th e Ar yl P or tion of

Ca lich ea m icin ^a

These lessons were not lost upon us as we embarked

on the calicheamicin carbohydrate domain. This depar-

ture seemed to be preferable to continuing in the espe-

ramicin series, in that we had learned to assemble the

calicheamicin aglycon but not the esperamicin aglycon

bearing the additional hydroxyl group. Furthermore,

from a biological standpoint, the calicheamicin system

which exhibits such striking sequence selectivity in its

double-stranded DNA cleavage constituted the more

exciting synthetic target. Fortunately, the calicheamicin

problem was significantly simplified by our previous

a

(a) TMSCN, KCN, 18-C-6; (b) SmI₂, THF, MeOH, 82%; (c) ICl,

MeCN, 93%.

There remained, also, the need to construct the car-

bohydrate E ring which also has no counterpart in the

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 27

Sch em e 29. Syn th esis of Ar yl Sa cch a r id e 113^a

esperamicin system. Randall Halcomb and Serge Boyer

took up this problem in earnest. Their synthesis started

with ^L-rhamnal diacetate 102 (Scheme 28) which was

Sch em e 28. Syn th esis of Tr ich lor oa cetim id a te

108^a

a

(a) BF₃‚OEt₂, CH₂Cl₂, 95%; (b) NaOMe, MeOH; (c) TBSOTf,

pyridine, CH₂Cl₂, 87%; (d) DIBAH, CH₂Cl₂, 77%; (e) NaClO₂, H₂O,

tBuOH, NaHPO₄; (f) (COCl)₂.

coupling a calicheamicin aglycon with an advanced

carbohydrate model. We turned to the use of a trichlo-

roacetimidate as the donor function, using the very

powerful Schmidt protocol. It seemed that the trichlo-

roacetimidate method would give us the best combination

of a proven donor type which could be fashioned under

mild conditions and which came with a long track record

of couplings with rather hindered acceptors. The accep-

tor, in this instance, would be aglycon 56. Before

proceeding to such coupling steps, it would be necessary

to produce the aglycon as a single enantiomer.

a

(a) BnOH, CH₂Cl₂, BF₃‚OEt₂; (b) NaOMe, MeOH; (c) TBSCl,

imidazole, CH₂Cl₂, 91%; (d) OsO₄, NMO, acetone-H₂O, 96%; (e)

Bu₂SnO, MeOH, then MeI, TBAB, benzene; (f) Ac₂O, pyridine; (g)

H₂, Pd(OH)₂, MeOH; (h) Cl₃CCN, NaH, CH₂Cl₂.

Fortunately this series of compounds was easily ac-

cessed by going back to the total synthesis of racemic

calicheamicinone at the stage of compound 54 (Scheme

31). Treatment of this compound under the Wong

conditions with vinyl acetate and lipase PS-30 indeed led

to high enantioselectivity in the acetylation of the

primary alcohol. Elsewhere we have described this

process in greater detail.³²Suffice it to say here that.

building on the key Wong method, we could gain access

to both the calicheamicinone and the ent-calicheamici-

none series.^18bAfter some initial confusion in the proper

absolute configurational assignment of these aglycons,

the matter was clarified and each of these series could

be converted to their final glycosyl acceptor versions, i.e.,

the calicheamicinone ketal and the ent-calicheamicinone

ketal (vide infra). It was our hope that there would be

sufficient chemical distinction between the secondary and

tertiary alcohols that it would not be necessary to protect

the latter as a prelude to glycosylation.

In a major advance in the area (Scheme 32), it was

shown by Randall Halcomb for the first time that it was

indeed possible by trichloroacetimidate coupling to join

an advanced, fully synthetic, aglycon construct to an

advanced fully synthetic carbohydrate domain.⁴⁵While

it was gratifying to demonstrate that such a coupling

could be conducted, the sense of triumph was tinged with

considerable concern in regard to total synthesis. For

instance, the first step in developing the principal

â-product 118 would involve cleavage of the phthalolyl

group. Unfortunately, it soon became clear that sub-

strates bearing the enediyne functionality, including 118,

were unstable to various conditions attempted in

converted to 103a by benzyloxy Ferrier rearrangement

and protected via free alcohol 103b as the TBS derivative

103c. Reaction of 103c with osmium tetraoxide gave,

substantially, 104, which could be selectively methylated

at the equatorial center via stannylation (see compound

105). Upon acetylation, 105 was converted to 106.

Cleavage of the anomeric benzyl glycoside was ac-

complished through hydrogenolysis. The anomeric hy-

droxyl group of 107 thus unveiled was activated as its

trichloroacetimidate 108 through the action of sodium

hydride and trichloroacetonitrile. Coupling of 108 with

101 (Scheme 29) was accomplished via the Schmidt

methodology to produce 109. This substance was, in

turn, converted to bis-TBS derivative 111 via the mono-

alcohol 110. In model systems related to 111, it had been

shown that the nitrile group (or even a methyl ester) was

very sluggish with respect to hydrolysis. Accordingly, we

proceeded via reduction of the nitrile function in 111 to

aldehyde 112 followed by oxidation to the free carboxylic

acid and activation as its acid chloride 113. At this stage,

we could couple the CD segment to the aforementioned

B ring thiol 82b to give rise to the tricyclic system 114

(Scheme 30). The stage was then set for coupling CD

substructure 114 with the previously described AE

system 90. Once again, Kahne coupling was successful

and the full “penta domain” was in hand (see compound

115).⁴⁴

At this stage, we could selectively deprotect the ano-

meric p-methoxybenzyl function to give rise to 116. We

were now in a position to explore the possibility of

28 J . Org. Chem., Vol. 61, No. 1, 1996

Sch em e 30. Syn th esis of Ca lich ea m icin P en ta sa cch a r id e 116^a

Perspective

a

(a) Et₃N, DMAP, CH₂Cl₂, 85%; (b) NaH, DMF, then 90, 80%; (c) DDQ, CH₂Cl₂, H₂O.

Sch em e 31. En zym a tic Resolu tion of 54^a

a

(a) Lipase PS-30, DME, vinyl acetate; (b) 7 N NH₃, MeOH.

dephthaloylation. Thus, it became obligatory that the

carbohydrate domain be introduced at an even more

advanced level than is represented by compound 118. We

took this to mean that it would be necessary to advance

the E ring nitrogen of the donor structure to the point

where it had already incorporated its ethyl group and

was protected with a readily dischargable function such

as an Fmoc function.

It also seemed unlikely that the steps necessary to

traverse the gap between the C₁₃ketone in the acceptor

which was used and the final vinyl trisulfide would

probably be too extensive to be feasible in a structure of

this complexity. Accordingly, it would also be necessary

to introduce the acceptor at a more advanced stage.

Some early efforts using the benzoate 59a (natural

enantiomer) as the glycosyl acceptor were explored by

Margaret Chu-Moyer with good indications of feasibility.

However, in our series, we were somewhat skeptical

about the possibility of conducting the advancement of

the allylic benzoate to the allylic thioacetate with survival

of the Fmoc function alluded to above. Hence, Dr.

Stephen Hitchcock opted to examine the bolder possibility

where the thioacetate would already be included in the

structure of the acceptor molecule undergoing glycosy-

lation (see compound 60, Scheme 31).

Returning again to the total synthesis of calicheami-

cinone, this time carried out on the natural (as well as

the unnatural) enantiomer, we examined glycosylation

of compound 60 (Scheme 33). We also effected changes

in the donor system. Our in-house research had indi-

cated that the cleavage of the TBS protecting groups post-

glycosylation would be highly problematic in the setting

of the full panoply of calicheamicin functionality.

A

particular source of vulnerability would be the slowness

of the removal of the TBS function from the D ring sugar.

This deprotection was possible in the methyl glycoside

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 29

Sch em e 32. F ir st Glycosyla tion of a Ca lich ea m icin on e Con gen er w ith a F u ll P en ta Dom a in ^a

a

(a) DBU, Cl₃CCN, CH₂Cl₂; (b) BF₃‚OEt₂, CH₂Cl₂, -78 °C, 28%.

Sch em e 33. Syn th esis of P r otected Ca lich ea m icin Th ioa ceta te 121^a

a

(a) Cs₂CO₃, Cl₃CCN, CH₂Cl₂, 91%; (b) AgOTf, 4 Å MS, CH₂Cl₂, 58%.

series, upon prolonged treatment with TBAF. Unfortu-

nately, it was demonstrated by model studies that such

multihour exposure would not be feasible in the context

of the fully developed, advanced, precalicheamicinone

30 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

I

a

Sch em e 34. Syn th esis of Ca lich ea m icin γ₁

functionality. In fact, our studies indicated that the

tolerance of such advanced constructs toward TBAF

would be feasible only if exposure would be confined to a

few minutes. While recourse to room temperature ex-

posure to TBAF would shorten the time necessary for

deprotection, even the carbohydrate domain did not

tolerate these conditions well. Attempts to carry out such

deprotections led to loss of the entire D ring from the

domain by cleavage of the aryl glycoside bond. Accord-

ingly, it became necessary to modify the carbohydrate

domain such that the hydroxyl groups were protected as

triethylsilyl (TES) ethers. Here, it seemed likely that

deprotection could be accomplished under much milder

conditions which might be compatible with survival of

the sensitive resident functionality (vide infra). The

schemes which had initially been used in construction

of the per-TBS-protected glycosyl donor were now recon-

stituted using triethylsilyl protection groups for the

carbohydrate domain hydroxyls. The coupling reaction

types described above were carried out using triethylsilyl

protecting groups, and the E ring nitrogen was protected

as its N-ethyl Fmoc carbamoyl group. The feasibility of

using this group had been earlier demonstrated by

Nicolaou in his total synthesis in the context of an oxime

ether linking strategy.¹⁹We were, of course, most

thankful to have a documented precedent albeit on a

different construct. The adaptation of our earlier chem-

istry for synthesizing donor 120 is shown in Scheme 33.

In the event, coupling at the stage of optically pure

thioacetate 60 with trichloroacetimidate 120 was suc-

cessful under the conditions shown. A 50% yield of

glycoside 121 was obtained. Removal of the triethylsilyl

and TEOC protecting groups was accomplished (Scheme

34) with HF‚pyridine, and the ketal function was cleaved

with aqueous acid to afford 122. The fully deprotected

thioacetate had also been prepared during the Nicolaou

synthesis,²¹and upon receipt of an NMR spectrum of an

authentic sample from the La J olla workers, it was clear

that our totally synthetic product 122 was correctly

formulated. Therefore, the total synthesis of the thio-

acetate, itself a biologically active substance, had been

accomplished by a highly convergent strategy.

a

(a) DIBAH, CH₂Cl₂, -78 °C, then N-(methyldithio)phthalimide

(54%); (b) CSA, THF-H₂O, then TBAF, THF, R ) Ac (40% overall),

R ) SSMe (37% overall).

Sch em e 35. Syn th esis of 125^a

We returned to the protected compound 121 (Scheme

34). Cleavage of the acetyl group was effected via

diisobutylaluminum hydride. Installation of the trisul-

fide linkage was carried out under modified Harpp

conditions to give 123. After cleavage of the ketal with

aqueous acetic acid, a reaction which also led to some

silyl deprotection, the molecule was subjected to short

term treatment with anhydrous TBAF. This led to

complete deprotection of all silyl groups as well as the

TEOC and the Fmoc functions, giving rise to fully

synthetic calicheamicin γ¹_I(3) identical with an authentic

sample provide by the Lederle group.

The ratio of â:R glycosylation in the natural series was

approximately 3:1. The minor R product 124 could be

processed (Scheme 35), as above, to give rise to the fully

synthetic “allo” calicheamicin thioacetate construct 125,

differing from 122 only in the stereochemistry of the

anomeric bond joining the carbohydrate domain to the

C₈hydroxyl of the aglycon thioacetate sector. We also

coupled aglycon acceptor (ent-60) to glycosyl donor 120

(Scheme 36). Remarkably, in this case there was a

striking reversal in the sense of glycosylation. The ratio

of glycosides 126 and 127 produced in this reaction was

18:1 favoring the R-glycoside. We take this result as a

striking reaffirmation of the concept that glycosidation

a

(a) CSA, THF-H₂O, then TBAF, THF, 40%.

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 31

Sch em e 36. Syn th esis of 127^a

Sch em e 37. Syn th esis of 128^a

a

(a) CSA, THF-H₂O, then TBAF, THF, 40%.

Sch em e 38. A High ly Con ver gen t Syn th esis of

I

Ca lich ea m icin γ₁(tw o step s p ost-glycosyla tion )^a

a

(a) AgOTf, 4 Å MS, CH₂Cl₂, 38%.

ratios exhibited by a particular glycosyl donor are not

an independent property of the leaving group and pro-

tecting group ensemble but also reflect a high order of

interactivity with the acceptor system. We would argue

that in the case of acceptor ent-60 the formation of a

â-equatorial glycoside giving rise to 127 cannot be

accomplished without considerable abutments between

donor and acceptor in the transition state of the emerging

glycoside. However, R-glycosidation with formation of an

axial bond allows for a much greater autonomy of the

two sectors, even at the transition state level of their bond

formation (see R-glycoside 126). Full deprotection was

carried out only on the prevalent R-glycoside to give the

ent (aglycon domain) and epimeric thioacetate 128

(Scheme 37).

Having accomplished glycosidation at a rather ad-

vanced level, both in terms of glycosyl donor and glycosyl

acceptor, we were unable to resist the temptation to

explore the possibility of achieving maximal convergence

in the synthesis. For that purpose the glycosyl acceptor

would have to include the full trisulfide functionality.

Accordingly, the optically pure natural calicheamicinone

series was advanced to the point of calicheamicinone

ketal 60a itself, following chemistry previously carried

out in the racemic series. In the event, coupling of donor

120 (Scheme 38) with calicheamicinone ketal 60a (natu-

ral antipode), under very specified conditions developed

by Dr. Stephen Hitchcock (silver triflate, 4 Å molecular

a

(a) AgOTf, 4 Å MS, CH₂Cl₂, 34%; (b) CSA, THF-H₂O, then

TBAF, THF, 32%.

32 J . Org. Chem., Vol. 61, No. 1, 1996

Sch em e 39. Mod e of Action of Dyn em icin A (6)

Perspective

sieves, methylene chloride, room temperature), gave rise

to a glycoside (129) as the only detectable product, albeit

in only 34% yield. We attribute the somewhat disap-

pointing yield to the relatively small amounts of acceptor

which were available to us at this phase of the study.

Nonetheless, it was remarkable that sufficiently mild

glycosylation conditions had been developed such that it

was possible to join the aglycon and donor entities of the

fully functionalized calicheamicin system. A two-tep

deprotection protocol involved camphorsulfonic acid which

cleaved the ketal as well as some of the silyl groups and

brief exposure to TBAF which led to completion of the

silyl deprotection as well as cleavage of the TEOC and

Fmoc functions. A maximally convergent total synthesis

of calicheamicin (two steps post coupling!) had in fact been

accomplished.¹⁸

Even while the calicheamicin project was occupying our

attentions, there appeared in the literature another

enediyne natural product, dynemicin A (see 6, figure 1),

isolated from micromonospora chersina.¹⁴Its structure

was established via crystallographic means. Dynemicin

A exhibits potent cytotoxicity and double-strand DNA

cleavage capacity although lacking the sequence specific-

ity of calicheamicin.⁴⁶

In comparing dynemicin with calicheamicin, an obvious

similarity in the presence of a cyclic enediyne in each

structure is perceived as is a major difference. Thus,

dynemicin lacks the carbohydrate domain which is

central in establishing a molecular rapport between the

enediyne and DNA. The anthraquinone sector of dyne-

micin presumably provides some of the DNA contacts

which are furnished by the relatively lipophilic carbohy-

drate domain of calicheamicin.⁴⁴

In a very early, but still influential position paper

concerning the bioorganic chemistry of dynemicin, Sem-

melhack proposed that the bioactivation steps for dyne-

micin involves reduction of quinone to hydroquinone

(Scheme 39).¹⁵At this point, the lone pair on the

quinoline nitrogen, no longer part of a vinylogus amide

linkage as in dynemicin, is more capable of manifesting

internal nucleophilic character in participating in sol-

volytic displacement of the epoxide (see arrows: 130 f

131). Without specifying the nature of the nucleophile

(Nuc:) in detail (hydroxyl, cystine-based thiol, or amino

functions), it is seen that overall front side displacement

would give rise to 132. The latter structure type emerges

as a likely substrate for Bergman-type rearrangement

to a 1,4-diyl (133) which inaugurates the DNA cleavage

and cytotoxicity cascades.

This Semmelhack formulation, while unsupported by

any chemical experiments, seemed so intuitively reason-

able as to be persuasive and was also of great teaching

value at the level of synthesis. From it followed the

perception that, during the course of building dynemicin,

once the epoxy enediyne system was in place it would be

necessary to downregulate the nucleophilic character of

the quinoline nitrogen either through acylation, through

electronic connectivity to an anthraquinone moiety, or

to an equivalent thereof.

Indeed, it was compellingly clear that the opportunities

and challenges posed by the novel metabolite dynemicin

could not be overlooked or bypassed by our laboratory.

As is our custom, we would become engaged at the level

of total synthesis.^47,48The interesting chemical issues

posed by dynemicin were reason enough to institute an

exploratory effort in synthesis. However, we further

hoped that out of such synthetic studies might follow

opportunities to garner some new insights regarding drug

design in this area.⁴⁹

Given our involvements in calicheamicinone and cali-

cheamicin, we naturally came to consider the possibility

of closing the enediyne ring by coupling of acetylide ion

to a suitable electrophilic center. Clearly, two possibili-

ties can be entertained for such a closure (Scheme 40).

In one broad category, the enediyne is encased in the

quinoline sector, presumably at the dihydro or tetrahydro

stages (X-Y may correspond to an epoxide, to the second

bond of a double bond, or to sp³-bound functional groups

which might become, through some sequence, an ep-

oxide). The electrophile E would be positioned at C₇, and

bond formation with the acetylide would occur. The

electrophile might be the carbonyl carbon of a ketone or

an epoxide joining C₇and C₆(cf. 135 f 136). Alterna-

tively, the enediyne might be moored to C₇. Bond

formation might involve attack of the acetylide on carbon

2. The electrophilic center might arise from an inter-

molecular variant of a Reissert-like reaction.⁵⁰

To pursue the Reissert analogy further, the nitrogen

Perspective

Sch em e 40. In itia l Syn th etic P er cep tion s for

J . Org. Chem., Vol. 61, No. 1, 1996 33

nature of X, Y and A, B in the cyclization precursor. We

refer to the interpolative approach as case II.

Cyclic En ed iyn e F or m a tion (Ca se Ia a n d Ib)

Sch em e 41. Syn th etic P er cep tion s of a n

In ter p ola tive Ap p r oa ch to th e Cyclic En ed iyn e

(Ca se II)

It will be noted that, from a stereochemical perspective,

case II represents the greater challenge in preassembly.

In either version of case I, a single cis relationship

between the secondary methyl center at C₄and an

enediyne arm of a suitable seco precursor must be

fashioned. In essence, the cyclization (in whichever

order) implements the second cis acetylide-methyl re-

lationship. This is in contrast to case II where two cis

methyl-acetylide relationships governing C₂, C₄, and C₇

must in place before interpolative cyclization between bis-

alkyne and olefin can be considered.

We elected to pursue concurrently case I_band case II.

In each instance, a need for fashioning a cis relationship

between the secondary methyl at C₄and the propynyl

arm at C₇would be necessary. Obviously, the solution

for this requirement must be melded into a comprehen-

sive synthetic program. As in calicheamicin, our aim

with dynemicin was the full natural product in all its

structural detail. Solutions which seemed to solve re-

gional issues without the promise of extendibility to a

total molecular solution were of less interest. It is with

the matter of a solution to the cis C₄-C₇relationship in

a realistic setting appropriate for reaching dynemicin

that our explorations began. For establishing this rela-

tionship, Tae-young Yoon turned to one of the most

reliable precepts in organic chemistry, i.e., the suprafacial

governance of the Diels-Alder cycloaddition reaction

(vide infra).

Commercially available aldehyde 144 (Scheme 42) was

O-alkylated with sorbyl bromide to afford 145. Through

sound chemical steps, 145 was converted to 147 and then

to trienal 148. The asterisks at the butadienyl carbons

(cf. 148, Scheme 43) direct attention to the cisoid rela-

tionship of the methyl and methylene groups projecting

from prochiral centers corresponding respectively to C₄

and C₇of the future dynemicin target. Under catalysis

by zinc chloride, a highly stereoselective Diels-Alder

reaction occurred, affording 149a , the product of endo

addition. The next steps involved treatment of 149a with

ceric ammonium nitrate (CAN). A great deal of prece-

dent could be marshaled for the use of this reagent to

achieve oxidative dealkylation of diethers of hydroquino-

nes. Ordinarily both alkyl carbons constitute C₁or C₂

reaction “throwaways”. In the case at hand, the role of

the methoxy carbon is indeed relegated to that of a

“throwaway” carbon. However, the methylene etheral

carbon had undergone conversion to a hydroxymethyl

might be suffering acetylation as the ring is being closed

(cf. 137 f 138). In this way, in keeping with the

Semmelhack teaching discussed above, amide-like char-

acter is introduced even as the enediyne ring system is

being fashioned. Alternatively, one could readily imagine

a potential carbinolamide-based leaving group being

displaced by the nucleophilic acetylide center of the

enediyne (cf. 139 f 140). Bond formation would thus

correspond to amidoalkylation. Again, amide-like char-

acter of the nitrogen coincides with establishment of the

cyclic enediyne systems.

Of the two acetylide-electrophile closures, the case I_b

was, for us, the more interesting. Case I_a, while certainly

far from trivial in leading to an actual synthesis of

dynemicin, represented at the conceptual level a rela-

tively modest extension of the calicheamicinone approach.

Happily, many other groups were now practicing this

chemistry throughout the enediyne field. In fact, with

the notable exception of the Schreiber approach to

dynemicin (which led to a synthesis of methylated

versions of dynemicin via transannular Diels-Alder

reaction)⁴⁶every route to either calicheamicin or dyne-

micin was utilizing, in one form or another, an acetylide-

electrophile bond connection for fashioning the cyclic

enediyne. While this motif also pervaded attempts to

close a bond to C₂, the challenges associated with

organizing the requisite functionality and the likelihood

that they would bring in their wake greater opportunities

for chemical innovation served to tilt us in the direction

of case I_b.

We also wanted to examine an unprecedented ap-

proach (Scheme 41), i.e., closure of the enediyne by

interpolating a generic cis-disubstituted ethylene (142)

between two acetylides as shown in the structure (141

+ 142 f 143). Once again, we leave unspecified the

34 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Sch em e 42. Syn th esis of Tr ien e Ald eh yd e 148^a

Treatment of 150 with ammonium acetate in acetic

acid at 100 °C afforded 151 and, following bis silylation,

152 in which the quinoline structure had been installed.⁵¹

This matrix in which the necessary C₄-C₇cis relation-

ship had been put into place was to serve both cases I_b

and II. The latter would lead to dynemicin A (vide

infra).²²

Before detailing the progression which did bring suc-

cess, we digress to describe the highlights of our efforts,

albeit unsuccessful, to realize an intramolecular Reissert

reaction.⁵²To pursue this possibility, compound 152

(Scheme 44) was subjected to osmylation to produce diol

Sch em e 44. Syn th esis of Seco En ed iyn e 158^a

a

(a) Sorbol bromide, K₂CO₃, acetone, 96%; (b) (EtO)₂P(O)-

CH₂CO₂Et, NaH, THF, 95%; (c) DIBAH, CH₂Cl₂, 99%; (d) Swern

oxidation, 91%.

group during the same oxidative dealkylation. This cis

relationship of the hydroxymethyl group and the alde-

hyde need not have been critical to the total synthesis

plan but turned out to be important. It allowed for

containment of these groups in the form of a hemiacetal

linkage (see compound 150) which stabilizes the func-

tionality to the oxidative conditions. We note parentheti-

cally that the exo IMDA product 149b formed as a minor

product in the strictly thermal cyclization of 148 suffered

virtually complete destruction upon treatment with CAN.

This difference of outcomes presumably reflects the

crucial stabilization role of the hemiacetal linkage.

Sch em e 43. Syn th esis of Dih yd r op h en a n th r a d in e

152^a

a

(a) Catalytic OsO₄, NMO, THF-H₂O, 90%; (b) nBu₃SnH,

ClCO₂Me; (c) Me₂C(OMe)₂, catalytic pTsOH, 95% (two steps); (d)

K₂CO3, MeOH; (e) K₂CO₃, Me₂SO4; (f) TBAF, THF, then Swern

oxidation, 81% (three steps); (g) PPh₃, CBr₄, CH₂Cl₂, 79%; (h)

nBuLi, PhCH₃, -78 °C, 77%; (i) 157, Pd(PPh₃)₄, CuI, BuNH₂, PhH,

93%.

153 as shown. The resultant 153 was exposed to reduc-

tive carbomethyoxylation to deliver 154 following forma-

tion of the acetonide. At this stage, functional and

protecting group differentiation afforded 155. The de-

rived aldehyde 155, when taken through a Corey-Fuchs

protocol,⁵³afforded terminal acetylide 156. Elongation

of the latter by Castro-Stephens coupling⁵⁴to 157 gave

rise to 158. The scenario for attempting intermolecular

Reissert bond closure was then set (Scheme 45) by

oxidative decarbomethoxylation of 158 with DDQ fol-

lowed by desilylation with TBAF to afford enediyne 159.

Cyclization of 159 was attempted with lithium hexa-

methyldisilazide. The reaction afforded the â-eliminated

product 160. Clearly, reaction had occurred through

deprotonation of the C₇-H bond with ejection of acetone.

The acidity of the C₇center and its undermining effect

on our proposed cyclization had not been properly an-

ticipated. Other experiments pointed to the same prob-

lem. Thus, attempted desilylation of 158 (Scheme 46)

with TBAF led instead to 162. Another sequence also

a

(a) ZnCl₂, CH₂Cl₂, 60%; (b) Ce(NH₄)₂(NO₃)₆, MeCN, H₂O, 90%;

(c) NH₄OAc, HOAc, 100 °C, 89%; (d) TBSCl, IMID, CH₂Cl₂, 98%.

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 35

Sch em e 45. Syn th esis of Qu in olin e 159 a n d

Sch em e 47. Attem p ted Cycliza tion of 159^a

Attem p ted Cycliza tion ^a

a

(a) DDQ, TMSCN, then TBAF, 68%; (b) LHMDS, THF, -78

(a) iPrMgBr, ClCO₂Me.

°C.

as follows (Scheme 48). Compound 153 was successfully

subjected to reductive allyloxycarbomylation to afford

Sch em e 46. Attem p ted En ed iyn e Cycliza tion s^a

Sch em e 48. Syn th esis of Im in o Ep oxid e 169 a n d

Attem p ted En ed iyn e F or m a tion ^a

a

(a) TBAF; (b) DDQ, TMSCN; (c) ClCO₂Ph, CsF.

supported this trend. Treatment of 158 with DDQ in the

presence of TMSCN afforded 161. We then hoped to

achieve carbamoylative cyclization through joint treat-

ment of 161 with phenyl chloroformate and cesium

fluoride (for disylylation). Once again, these conditions

led to compound 163. We then returned to compound 159

in which the acetylenic silyl group had already been

cleaved (Scheme 47).

The thought was to deprotonate the acetylene while

carbamoylating the quinoline nitrogen as a protocol to

favor cyclization. We hoped to bring this result about

by treatment of 159 with isopropylmagnesium bromide

in the presence of methyl chloroformate. Unfortunately,

this reaction afforded not the cyclic enediyne 165 but 164,

in which the same type of elimination, by deprotonation

from C₇, had occurred.

a

(a) nBu₃SnH, ClCO₂Allyl; (b) 2-methoxypropene, PPTS; (c)

TBAF, THF, 96%; (d) NaH, TBSCl, 79%; (e) Swern oxidation; (f)

PPh₃, CBr₄, CH₂Cl₂; (g) nBuLi, PhCH₃, -78 °C, 74% (three steps);

(h) aqueous TFA, CH₂Cl₂; (i) VO(acac)₂, tBuOOH, PhH, 83% (two

steps); (j) Ac₂O, Et₃N, DMAP; (k) Pd(PPh₃)₄, nBu₃SnH, then DDQ,

CH₂Cl₂, 56%.

At this stage, we despaired of achieving cyclization in

the intended sense, unless the problem of the C₇acidity

had been countered. One progressive way of dealing with

this problem was to install the epoxide at the C₈-C₉

double bond prior to projected cyclization. We proceeded

166. We hoped that the Alloc group (allyl carbamate)

could be cleaved under very mild conditions after instal-

lation of the C₈-C₉epoxide. It seemed likely that such

an epoxide would itself be quite labile. In the event, a

36 J . Org. Chem., Vol. 61, No. 1, 1996

Sch em e 49. Syn th esis of An th r a qu in on e 176 a n d Attem p ted En ed iyn e F or m a tion ^a

Perspective

a

(a) K₂CO₃, MeOH, then PhI(OAc)₂, MeOH; (b) Ac₂O, Et₃N, DMAP, 72%; (c) 172, tBuOLi, LiCl, 37%; (d) Me₂SO₄, K₂CO₃, 84%; (e)

Pd(PPh₃)₄, morpholine, CH₂Cl₂, 96%; (f) DDQ, MeOH, CH₂Cl₂, 87%; (g) 157, Pd(PPh₃)₄, CuI, BuNH₂, PhH. In one instance, we were able

to reach a promising pentacyclic domain, containing a fully extended enediyne side chain at C7.

Sch em e 50. Syn th esis of P en ta cyclic Seco En ed iyn e 185^a

a

(a) NaOAc, HOAc, 100 °C. (b) NH₄OAc, HOAc, 100 °C, then K₂CO₃, Me₂SO₄, then Ac₂O, Et₃N, 25% overall; (c) OsO₄, NMO, THF; (d)

2-methoxypropene, catalytic pTsOH, then K₂CO3, MeOH; (e) Swern oxidation; (f) PPh₃, CBr₄, then nBuLi, THF; (g) 157, Pd(PPh₃)₄, CuI,

BuNH₂, PhH.

significant part of the scheme was accomplished. Com-

pound 166 could be advanced to 167 and, remarkably,

epoxidized, as shown to afford 168. Even more remark-

able was the feasibility of transforming 168 f 169.

Here, however, our good fortune ran out. All attempts

to carry out Castro-Stephens-like elongation of imine

epoxide 169 to gain access to enediyne 170 were unsuc-

cessful. We attribute this noncorrectable difficulty to the

instability of the imine epoxide in the presence of the

palladium(0) reagents necessary for extension of the

propynyl group to the enediyne.

cleaved and the resultant phenol was oxidized with

phenyliodosobenzene diaceteate to afford compound 171.

It then proved to be possible to annulate the system with

the dimethoxycyanophthalide 172, as shown,⁵⁵in a

Swenton-like reaction.^56,57Curiously, only one stereo-

isomer of 171 functioned in this reaction (probably the

epimer with the â-methoxy group). The other methoxy

aminal was impervious to the Swenton reaction under a

variety of conditions. While this steric dependence was

certainly a serious incovenience in moving large amounts

of material forward, we did reach compound 173 and then

to 174 as shown. Presumably, the viability of the epoxide

owes much to the carbamoyl group and to the quinone.

Such conjugation attenuates the internal nucleophilicity

We returned to compound 168 and investigated the

possibility of building up a pentacyclic domain prior to

the acetylide activated imine cyclization step (Scheme

49). Following bisacetylation, the TBS ether group was

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 37

of the tetrahydroquinoline type of nitrogen, thereby

lessening solvolytic threats to the epoxide.

acidity of the C₇proton (see compounds 159, 161). With

the epoxide in place, the C₇acidity problem would

probably not pertain. However, in such intermediates

(see 169 and 176) we could not elaborate the enediyne.

In pentacyclic intermediates, where the enediyne could

be installed, acylation of the azomethine linkage as a first

step in the Reissert sequence failed (cf. 185-187).

Of course, in any undertaking of this level of complex-

ity, one can always point to possibilities which had not

yet been excluded. However, the accumulated and seri-

ous setbacks which we had suffered served to focus our

attentions on the prospects of implementing case II. For

this purpose, we returned once again to compound 152.

Our goal structure (Scheme 52) emerged as compound

188 or, more specifically, diyne 189 (vide infra). With

such a substrate in hand, attempt an interpolative

cyclization by insertion of a Z-ethylene linkage inserted

between the cis-related acetylenes would be under-

taken.⁵⁸As an interim goal, we sought to gain access to

a generic structure of the type 190a . The problem was

obvious. How could one direct introduction of a group

at C₂from the â-face of the azomethine linkage in the

presence of the two â-resident functions (190a f 190b)?

It seemed likely that intermolecular addition of a nu-

cleophile in a Reissert-like reaction to a structure of the

type 190a would occur from the R-face, given the â-dis-

posed resident groups at C₄and C₇.

Indeed, quinone substitution itself suffices for this

purpose. Thus, the Alloc group was cleaved with tet-

rakis(triphenylphosphine)palladium(0) to afford 175. Fur-

thermore, oxidation of 175 led to the sensitive, but still

viable, 176. However, attempts to chain extend 176 in

the direction of 177 failed. It seemed that there was an

incompatibility between the presence of the quinone (and

possibly the epoxide) with the insertion chemistry neces-

sary to bring about the Castro-Stephens chemistry

needed for elongation to the seco enediyne.

For this purpose, we went back to the virtual beginning

of the synthesis. Thus, treatment of compound 150

(Scheme 50) with presumed isobenzofuran 178 afforded

179. The latter, on aromatization (cf. 179 f 180)

provided the labile, poorly characterized 180 which was

converted to 181 via aminolysis. Once again, the oxida-

tion of the isolated double bond and protection of the R

diol could be accomplished (see compound 182). Oxida-

tion of 182 afforded aldehyde 183 which lent itself to

conversion to acetylene 184 and then to enediyne 185.

Unfortunately, this system could not be cyclized by

acylative (Reissert) activation of the azomethine linkage

(Scheme 51). Indeed, even a much simpler model,

Sch em e 51. Attem p ted In ter - a n d In tr a m olecu la r

Reisser t Rea ction s of P en ta cyclic Com p ou n d s^a

Sch em e 52. Syn th etic P er cep tion s for Ca se II

Ap p r oa ch

a

(a) base, ClCO₂Me; (b) EtMgBr, ClCO₂Me; (c) ethynylmagne-

sium bromide, ClCO₂Me.

compound 186, failed to cyclize in the Reissert mode.

Moreover, substrate 187, lacking the enediyne append-

age, failed to afford even an intermolecular addition of

ethynyl Grignard under Riessert conditions. We can only

conclude that the steric hindrance of the pendant naph-

thalene system prevents the acylative activation phase

of the Riessert-like sequence (see depiction of bay region

type of interference in 187). At this point, we were

confronted with a major dilemma. Tricyclic intermedi-

ates lacking the putative C₈-C₉epoxide but containing

the fully developed enediyne side chain could be synthe-

sized but were interdicted en route to cyclization by the

At this point, a bold possibility surfaced. If the double

bond of 190a could be substituted on its R-face with

sterically demanding groups (cf. 191), nucleophilic attack

38 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Sch em e 53. Syn th esis of Alk yn e 196^a

a

(a) Ph₂C(OMe)₂, H₂SO₄, CH₂Cl₂, 40 °C, 83%; (b) (TIPS)ethynylmagnesium bromide, ClCO₂Me, THF, -20 °C, 95%; (c) (TIPS)ethyn-

ylmagnesium bromide, allyl chloroformate, THF, -20 °C, 75%.

Sch em e 54. Syn th esis of Diyn e 203^a

on the azomethine linkage might occur from the â-face

(see 191 f 192). This was to be our guiding paradigm.

Compound 153 (Scheme 53), which we had prepared

earlier, appeared to us to be an excellent starting point

for the examination of the above-discussed theory. The

two hydroxyls of 153 were engaged in the form of a

benzophenone acetal through an exchange reaction with

its dimethyl acetal. There was thus obtained an 83%

yield of 193. Happily, treatment of 193 with the Grig-

nard reagent derived from TIPS acetylene in the presence

of methyl chloroformate gave rise to 194 and 195 in a

ratio of 9:1 favoring the desired â-isomer. The structure

of 194 was proven crystallographically.

In a similar vein, anticipating future needs (vide infra),

the acylating agent was changed to allyl chloroformate

and compound 196 bearing the â-vinyl linkage was in

hand. Thus, an important milestone had been passed

in that the three stereogenic centers at positions 2, 4,

and 7 had been properly emplaced.

The time was now at hand to advance the hydroxy-

methyl group at C₇toward the acetylene linkage, pro-

jected to appear in compound 207 (vide infra). This goal

was smoothly accomplished.

The sequence started (Scheme 54) with selective cleav-

age of the aliphatic TBS group (without concomitant loss

of the phenolic TBS group and without forfeiture of the

benzophenol ketal). Compound 198 upon oxidation and

Corey-Fuchs elongation was converted to acetylide 199.

Finally, discharge of the TIPS group was achieved

through the action of TBAF to produce the free diyne 200.

The benzophenone ketal function was cleaved with

methanolic HCl to give rise to 201. A TBS group was

reinstalled on the phenolic hydroxyl through the agency

a

(a) Concentrated HCl, THF, 86%; (b) Swern oxidation; (c) PPh₃,

of sodium hydride and TBS chloride to afford 202 and

CBr₄, CH₂Cl₂, 84%; (d) nBuLi, PhCH₃, -78 °C, 74%; (e) TBAF,

then diacetate 203.

THF; (f) HCl, MeOH, 74%; (g) NaH, TBSCl, THF, 85%; (h) Ac₂O,

Et₃N, DMAP, 95%.

It became appropriate to direct our attentions to the

precise nature of the carbamoyl protecting device. What

we would need, of course, is that the urethane linkage,

in fact, shield the nitrogen from undesirable chemical

changes which might occur through the steps which

would be required to finish the synthesis. Furthermore,

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 39

Sch em e 56. Attem p ted Son oga sh ir a Cycliza tion of

208^a

we required the carbamoyl group to be cleavable at a

strategic point. Finally, it is a group whose stability must

be consistent with the anticipated use of Pd(0) in the

cyclization step. With the issue of the removal of the

carbon-bound silyl functionality having been established,

it was no longer necessary that the urethane linkage be

stable to desilylating conditions. If all went well, the only

desilylation which would be necessary would be that of

the phenolic TBS function. What could almost certainly

not be tolerated was the continued use of the Alloc

protecting device. It was critical that the protective

arrangement survive the action of Pd(0) which we

perceived to be necessary to mediate cyclization (vide

infra), although the precise form of the interpolation had

not even been decided. Accordingly, the Alloc group of

203 was discharged (Scheme 55) through the agency of

a

(a) 210, Pd(PPh₃)₄, CuI, nBuNH₂, PhH.

moiety between two acetylenes. Returning to compound

210a (Scheme 57), the two ethynyl groups were oxidized

with N-iodosuccinimide in the presence of silver nitrate

to give the bis-iodo system 211.⁵⁹

Sch em e 55. Syn th esis of Ep oxy Diyn e 207^a

Sch em e 57. Attem p ted Cycliza tion of

Bisiod oa lk yn e 211^a

a

(a) AgNO₃, NIS, THF, 75%; (b) 212, Pd(PPh₃)₄, DMF, 70 °C.

(a) Pd(PPh₃)₄, morpholine, THF, 0 °C; (b) TEOCCl, NaH, THF;

(c) NH₃, MeOH, 90% (two steps); (d) m-CPBA, CH₂Cl₂, 87%.

Once again, this compound (211) failed to undergo

interpolative cyclization with the (Z)-1,2-bis(trimethyl-

stannyl)ethylene (212).⁶⁰It was hoped, however, that

cyclization could be accomplished if the epoxide were in

place. The presence of the C₈-C₉epoxido linkage would

serve to bring the two ethynyl functions closer together

and to remove some of the steric strains associated with

the projected cyclization. In the event (Scheme 58),

treatment of compound 215 with the bis-stannyl ethylene

agent 212 did indeed give rise to the epoxy enediyne

216!⁶¹

tetrakis(triphenylphosphine)palladium(0) in the presence

of morpholine. This reaction undoubtedly passed through

compound 204, which was immediately protected as the

TEOC urethane 205, and following deacetylation with

ammonia, diol 206 was in hand. A large step forward

was accomplished when it was shown that epoxidation

of the double bond, even in the presence of the ethynyl

linkages, could be conducted. Anticipating a result

proven later, this oxidation was assumed to have pro-

duced R-epoxide 207.

Several attempts to conduct a conventional Castro-

Stephens reaction (Scheme 56) for insertion of the cis-

dichloroethylene (210) which were conducted in the

related methyl carbamate series (see compound 208)

were unsuccessful (208 N 209) and could never be

accomplished in our hands. At that point, we resorted

to the possibility of a Stille-like coupling, with the caveat

that it would accomplish interpolation of an ethylenic

We returned to the TEOC-protected epoxy diacetylene

207 (Scheme 59). This compound was iodinated to give

217. In the defining step of the synthesis, 217 cyclized

in the presence of tetrakis(triphenylphosphine)palladium-

(0) (5 mol %) and delivered an 81% yield of the epoxy

enediyne 218.

At this point, a clear strategy to reach dynemicin itself

surfaced. Many detours and failed initiatives still awaited

40 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Sch em e 58. In ter p ola tive Cycliza tion of 215 To

Deliver Cyclic En ed iyn e 216^a

capacity to differentiate the two hydroxyl groups of 218

by triflation at the equatorial site. This success was

followed by oxidation of the axial alcohol to the ketone

(see compound 220) and reductive excision of the trifyloxy

function (see ketone 221).

Sch em e 60. Con ver sion of Diol 218 to Vin ylogou s

Ca r bon a te 223^a

a

(a) AgNO₃, NIS, THF, 98%; (b) 212, Pd(PPh₃)₄, DMF, 70 °C,

78%.

Sch em e 59. In ter p ola tive Cycliza tion of 217 To

Deliver Cyclic En ed iyn e 218^a

a

(a) Tf₂O, Pyr, CH₂Cl₂, -20 °C, 95%; (b) Dess-Martin perio-

dinate, CH₂Cl₂, 95%; (c) CrCl₂, THF, 75%; (d) MgBr₂, CO₂, Et₃N,

MeCN; (e) MOMCl, iPr₂NEt, THF, 61% (over two steps).

At this point, we thought it would be appropriate to

introduce the carboxyl group at carbon 5. A variety of

possibilities were surveyed with a view toward enoliza-

tion of the C₆ketone. Apparently, enolates derived from

such full scale deprotonation are quite unstable. We

were, however, able to achieve the same result through

a pathway which had been developed for the carboxyla-

tion of ketones by Rathke and associates.⁶²In the event,

compound 221, upon treatment with magnesium bromide

under an atmosphere of carbon dioxide, gave a rather

unstable â-keto acid which on treatment with MOMCl

in the presence of Hunig’s base, under carefully controlled

conditions, progressed to the MOM protected enol version

of the â-keto ester (222) in 61% yield. The latter could

be converted to the required methyl enol ether 223

through the agency of diazomethane. Model reactions

indicated (see compound 223) that the cleavage of the

MOM function would produce the required carboxyl

group and that it should prove to be feasible at a rather

late stage of the synthesis.⁶³

a

(a) AgNO₃, NIS, THF, 95%; (b) 212, Pd(PPh₃)₄, DMF, 70 °C,

81%.

us. Here we relate the route which did indeed lead to

dynemicin A.

The first stage of that progression involved the proper

development of the vinylogous carbonate functionality

(Scheme 60). An important step in this regard was the

Returning, then, to compound 223, it was possible to

cleave the TEOC function with TBAF (Scheme 61). This,

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 41

in turn gave rise to an unstable and ill-characterized

intermediate (224) in which the full vulnerability of the

epoxide would be manifest. However, product 224 was

immediately oxidized to give quinone imine 225 which

was in fact a stable entity and turned out to be a critical

intermediate in the eventual total synthesis.^23b,64

Finally, the elusive anthraquinone was installed by

exposure of the quinone imine 225 (Scheme 63) to the

enolate generated from deprotonation of 229. There was

thus generated an adduct which we believe corresponds

to structure 230. In the formation of 230 a formal

cycloaddition followed by decarboxylation would have

occurred.⁶⁷Compound 230 would, in itself, be likely to

be an unstable compound, and its formation was followed

immediately by oxidation with bis(trifluoroacetoxy)iodo-

benzene leading to 231 in which the quinone imine ring

had been restored, thereby stabilizing, at least to some

extent, the C₈-C₉epoxide against solvolytic attack for

reasons discussed above. The structure of 231 has not

been fully established, but it is consistent with spectral

data provided elsewhere.⁶⁵Exposure of 231 to the action

of air, under daylight conditions, produced quinone

system 232 which in fact corresponds to tris-MOM-

protected dynemicin A. In the concluding step of the

synthesis, the MOM functions were cleaved through the

action of magnesium bromide in ether. There was thus

isolated, albeit in rather low yield, the long sought after

dynemicin A (15% from structure 225).²²We attribute

this disappointing yield not so much to the chemical

inefficiency of the transformations themselves but to the

instability of dynemicin A and to the difficulties in

handling this compound. Similar findings were regis-

tered by Myers using a totally different, but still logical,

annulation scheme. In the Myers annulation, at the end

of the sequence, dynemicin A could be isolated in 14%

yield, also reflecting the extreme difficulty encountered

in dealing with the anthraquinone system.²³In any case,

the total synthesis of dynemicin A has now been ac-

complished in our laboratory.

Sch em e 61. Syn th esis of Im in o Qu in on e 225^a

a

(a) TBAF, THF, 0 °C; (b) PhI(OAc)₂, THF, 0 °C, 60% from 223.

We also note that the quinone imine concept as a

stabilizing device for epoxide solvolysis was used in a new

concept for elaboration of a new kind of dynemicin drug

analog. In this connection, compound 233 (Scheme 64)

has been prepared by a route similar to those shown

above.^23bIt has been shown to be a potent DNA cleaving

agent and to be a compound of high cytotoxicity. In fact,

during in vivo investigations in mice, it has outperformed

standard clinically used anticancer drugs such as mito-

mycin C.^64,68It is presumed that the mechanism of action

of compound 233 involves in vivo reduction of the quinone

imine linkage or in vivo Michael addition of a bionucleo-

phile to the quinone imine linkage. In either case, there

would be generated a p-hydroxytetrahydroquinoline of

the type 234 which would then become a major prospect

for entering the bio-Bergmann manifold discussed at the

outset of this manuscript. Interestingly, a compound

related to 234 had been generated and observed spec-

troscopically by Nicolaou and associates.⁴⁹

Chronicled elsewhere are a variety of approaches which

were used to take advantage of the functionality of the

quinone imine to install the fully deprotected anthraquino-

ne system.⁶⁵Here we content ourselves with that route

which brought success. The coupling partner for a

quinone imine was prepared from bis-MOM-protected

bromo hydroquinone 226 (Scheme 62) which was con-

verted through the action of lithiomalonate in the pres-

ence of LiTMP to the homophthallate 227. Hydrolysis

of the latter followed by cyclization with (trimethylsilyl)-

ethoxyacetylene gave rise to 229.⁶⁶

Sch em e 62. Syn th esis of Hom op h th a lic

An h yd r id e 229^a

Con clu sion s

In the preceding text, we have described the total

synthesis of two complex enediyne drugs, calicheamicin

I

γ₁and dynemicin A. These goals were accomplished

through totally different strategies. During this work

there was learned a great deal about enediynes as well

as about the feasibility of various routes to reach them.

The calicheamicinone synthesis revealed the potentiality

of using enediyne anions for cyclization reactions. (In

one fashion or another, not always with full scholarly

attribution, the chemistry has found wide application in

the field.) The total synthesis of calicheamicin itself also

involved the development of some very new methodology

a

(a) Lithium dimethylmalonate, LiTMP, THF, -78 °C, 71%;

(b) KOH, MeOH, H2O, 95%; (c) (trimethylsilyl)ethoxyacetylene,

CH₂Cl₂, 100%.

42 J . Org. Chem., Vol. 61, No. 1, 1996

Sch em e 63. An n u la tion of An th r a qu in on e a n d Syn th esis of Dyn em icin A (6)^a

Perspective

a

(a) LHMDS, 229, then 225, THF, 0 °C; (b) PhI(OCOCF₃)₂, THF, 0 °C, 5 min; (c) air, daylight, THF, high concentration; (d) MgBr₂,

Et₂O, 24 h, 15% (overall from 225).

Sch em e 64. Develop m en t of En ed iyn e Qu in on e

Im in es (233) a s Biologica lly Active Dyn em icin

Con gen er s

oligonucleotides has recently been exploited in exerting

transcriptional control.⁶⁹This concept may yet have

further extensions as a device in mediating DNA-protein

interactions. Such interactions are of course critical for

protein expression.

In the total synthesis of dynemicin, a new paradigm

for enediyne synthesis has been demonstrated. Already,

it has found application in another laboratory.⁷⁰Also,

while following quite a related line of reasoning, the

dynemicin effort has led to a new type of enediyne drug

stabilized as a quinone imine linkage (see compound

233). The clinical usefulness of such an agent awaits

definition following quite favorable preliminary in vivo

experiments in animals.

As first hand witnesses and participants in these

exciting explorations we can attest to the captivating

nature of the problems and to the special motivations

which these different quests provided for finding solu-

tions. We are confident that the two expeditions were

fruitful in developing new chemistry and even hopeful

that the insights thus garnered will find application in

future drug design. Ultimately, the goal must be to

improve upon these fascinating metabolites.

Why do Micromonospora echinospora spp. calichensis

and M. chersina manufacture, respectively, calicheamicin

γ₁I and dynemicin A? For those with a positivist view of

life, it seems not unlikely that these compounds serve a

useful purpose in promoting the well being of their

biological inventors. Our task as chemists is to draw from

the genius of these biological inventions and to apply it

to the much more demanding needs of human applica-

tion. We remain confident that as time goes on, the role

of synthesis in fostering productive coalitions between

chemistry and biology in such grand pursuits will become

even more apparent.

in glycal assembly of rare carbohydrate types. The

eventual synthetic mastery over the carbohydrate do-

mains of esperamicin and calicheamicin, in conjunction

with appropriate binding and spectroscopic investiga-

tions, led to a much enhanced perception of the critical

role of the carbohydrate domain in the drugs. This

domain is what seems to provide the basic recognition

motif in mediating the encounter between the drug and

DNA. While much modeling and interpretation is still

necessary, it is clear that the sequence selectivity of

calicheamicin γ₁^Iarises from drug-carbohydrate interac-

tions of a type which had not been fully appreciated.

We note parenthetically that the concept of using

relatively hydrophobic sugars to bind to stretches of

Ack n ow led gm en t. This work was supported by

NIH Grant CA 28824. A predoctoral fellowship from

Perspective

J . Org. Chem., Vol. 61, No. 1, 1996 43

S. J .; Schulte, G. K. J . Am. Chem. Soc. 1991, 113, 3850. For the

first total synthesis of (-)-calicheamicinone, see: (c) Smith, A.

L.; Hwang, C.-K.; Pitsinos, E. N.; Scarlato, G. R.; Nicolaou, K.

C. J . Am. Chem. Soc. 1992, 114, 3134. (d) Smith, A. L.; Pitsinos,

E. N.; Hwang, C.-K.; Mizuno, Y.; Saimoto, H.; Scarlato, G. R.;

Memorial Sloan-Kettering Cancer Center to M.D.S. is

gratefully acknowledged. It is a pleasure to acknowl-

edge the dedication of our graduate student and post-

doctoral colleagues, mentioned throughout the text,

whose work provided the basis for this review. In

particular, we thank Dr. Taeyoung Yoon (currently at

Cal Tech), Dr. Stephen A. Hitchcock (currently at Eli

Lilly), Dr. Margaret Chu-Moyer (currently at Pfizer), Dr.

Serge H. Boyer (currently at Gensia), Dr. Stephen H.

Olson (currently at U.C. Berkeley), who were involved

in a critical way in the very demanding “end-games” of

these syntheses.

Suzuki, T.; Nicolaou, K. C. J . Am. Chem. Soc. 1993, 115, 7612.

I

,

see: (a)

(21) For the first total synthesis of calicheamicin γ₁

Nicolaou, K. C.; Hummel, C. W.; Pitsinos, E. N.; Nakada, M.;

Smith, A. L.; Shibayama, K.; Saimoto, H. J . Am. Chem. Soc.

1992, 114, 10082. (b) Nicolaou, K. C.; Hummel, C. W.; Nakada,

M.; Shibayama, K.; Pitsinos, E. N.; Saimoto, H.; Mizuno, Y.;

Baldenius, K.-U.; Smith, A. L. J . Am. Chem. Soc. 1993, 115,

7625.

(22) Shair, M. D.; Yoon, T. Y.; Danishefsky, S. J . Angew. Chem., Int.

Ed. Engl. 1995, 34, 1721.

(23) For the first total synthesis of (+)-dynemicin A, see: (a) Myers,

A. G.; Fraley, M. E.; Tom, N. J .; Cohen, S. B.; Madar, D. J . Chem.

Biol. 1995, 2, 33. (b) Myers, A. G.; Fraley, M. E.; Tom, N. J . J .

Am. Chem. Soc. 1994, 116, 11556. For the synthesis of various

methylated versions of dynemicin A, see: (c) Taunton, J .; Wood,

J . L.; Schreiber, S. L. J . Am. Chem. Soc. 1993, 115, 10378.

(24) Nicolaou, K. C. Angew. Chem., Int. Ed. Engl. 1993, 32, 1377.

(25) (a) Adler, E.; Brasen, S.; Miyake, H. Acta. Chem. Scand. 1971,

25, 2055, (b) Becker, H.-D.; Bremholt, T.; Adler, E. Tetrahedron

Lett. 1972, 13, 4205.

(26) Yamashita, D. S.; Mantlo, N. B.; Danishefsky, S. J . Tetrahedron

Lett. 1988, 29, 4681.

(27) Comins, D. L.; Brown, J . D. Tetrahedron Lett. 1981, 22, 4213.

(28) Mantlo, N. B.; Yamashita, D. S.; Schulte, G.; Danishefsky, S. J .

J . Am. Chem. Soc. 1988, 110, 6890. At virtually the same time

Refer en ces

(1) For recent reviews of the biology and chemistry of the enediyne

antibiotics, see: (a) Nicolaou, K. C.; Dai W.-M. Angew. Chem.

1991, 103, 1453; Angew. Chem., Int. Ed. Engl. 1991, 30, 1387.

(b) Enediyne Antibiotics as Antitumor Agents; Doyle, T. W.,

Borders, D. B., Eds.; Marcel-Dekker: New York, 1994.

(2) Edo, K.; Mizagaki, M.; Koide, Y.; Seto, H.; Furihata, K.; Otake,

N.; Ishida, N. Tetrahedron Lett. 1985, 26, 331.

(3) (a) Golik, J .; Clardy, J .; Dubay, G.; Groenwold, G.; Kawaguchi,

H.; Konishi, M.; Krishnan, B.; Ohkum, H.; Saitoh, K.; Doyle, T.

W. J . Am. Chem. Soc. 1987, 109, 3461. (b) Golik, J .; Clardy, J .;

Dubay, G.; Groenwold, G.; Kawaguchi, H.; Konishi, M.; Krish-

nan, B.; Ohkuma, H.; Saitoh, K.; Doyle, T. W. Ibid. 1987, 109,

3462.

(4) (a) Lee, M. D.; Dunne, T. S.; Siegel, M. M.; Chang, C. C.; Morton,

G. O.; Borders, D. B. J . Am. Chem. Soc. 1987, 109, 3464. (b)

Lee, M. D.; Dunne, T. S.; Siegel, M. M.; Chang, C. C.; Morton,

G. O.; Borders, D. B. Ibid. 1987, 109, 3466.

(5) (a) Neocarzinostatin: Goldberg, I. H. Acc. Chem. Res. 1991, 24,

191. (b) Calicheamicin: Zein, N.; Sinha, A.; McGahren, W. J .;

Ellestad, G. A. Science 1988, 240, 1198. (c) Esperamicin: Long,

B. H.; Golik, J . ; Forenza, S.; Ward, B.; Rehfuss, R.; Dabrowiak,

J . C.; Catino, J . J .; Musial, S. T.; Brookshire, K. W.; Doyle, T.

W. Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 2. Kishikawa, H.;

J iang, Y. -P.; Goodisman, J . ; Dabrowiak, J . C. J . Am. Chem.

Soc. 1991, 113, 5434.

(6) Casazza, A. M.; Kelley, S. L. Biological Properties of Esperamicin

and Other Enediyne Antibiotics. In Enediyne Antibiotics as

Antitumor Agents; Doyle, T. W., Borders, D. B., Eds.; Marcel-

Dekker: New York, 1994; pp 283-299.

(7) For a discussion of calicheamicin-monoclonal antibody conju-

gates, see: Hinman, L. M.; Hamann, P. R.; Upeslacis, J .

Preparation of Conjugates to Monoclonal Antibodies, in ref 6,

pp 87-105.

(8) (a) Mayer, J .; Sondheimer, F. J . Am. Chem. Soc. 1966, 88, 602.

(b) Wong, H. N. C.; Sondheimer, F. Tetrahedron Lett. 1980, 21,

217.

Kende and associates disclosed the construction of

a less

functionalized core structure through the use of acetylide aldol-

like ring closure. Kende, A. S.; Smith, C. A. Tetrahedron Lett.

1988, 29, 4217.

(29) For the first installation of an allylic trisulfide on a calicheami-

cinone model system, see: Magnus, P. ; Lewis, R. T.; Bennet, F.

J . Chem. Soc., Chem. Commun. 1989, 916.

(30) Haseltine, J . N.; Schulte, G.; Danishefsky, S. J . J . Am. Chem.

Soc. 1989, 111, 7638.

(31) Magnus, P.; Fortt, S.; Pitterna, T.; Snyder, J . P. J . Am. Chem.

Soc. 1990, 112, 4986.

(32) Mantlo, N. B.; Danishefsky, S. J . J . Org. Chem. 1989, 54, 2781.

(33) (a) Drak, J .; Iwasawa, N.; Danishefsky, S. J .; Crothers, D. Proc.

Nat. Acad. Sci. U.S.A. 1991, 88, 7464. (b) Aiyar, J .; Danishefsky,

S. J .; Crothers, D. M. J . Am. Chem. Soc. 1992, 114, 7552.

(34) Rocco, V. P.; Danishefsky, S. J . Tetrahedron Lett. 1991, 32, 6671.

(35) Aiyar, J .; Hitchcock, S. A.; Denhart, D.; Liu, K. K. C.; Danishef-

sky, S. J .; Crothers, D. M. Angew. Chem., Int. Ed. Engl. 1994,

33, 855.

(36) Halcomb, R. L.; Danishefsky, S. J . J . Am. Chem. Soc. 1989, 111,

6661.

(37) Bilodeau, M. T.; Danishefsky, S. J . Coupling of Glycals: A New

Strategy for the Rapid Assembly of Oligosaccharides. In Modern

Methods in Carbohydrate Synthesis; Harwood Academic Pub-

lishers. In press.

(38) Thiem, J .; Karl, H.; Schwenter, J . Synthesis 1978, 7, 696.

(39) Bolitt, V.; Mioskowski, C.; Lee, S.-G.; Falck, J . R. J . Org. Chem.

1990, 55, 5812.

(40) Yang, D.; Kim, S.-H.; Kahne, D. J . Am. Chem. Soc. 1991, 113,

4715.

(41) Halcomb, R. L.; Wittman, M. D.; Olson, S. H.; Danishefsky, S.

J .; Golik, J .; Wong, H.; Vyas, D. J . Am. Chem. Soc. 1991, 113,

5080.

(42) For an additional synthesis of the core trisaccharide of cali-

cheamicin, see: Kahne, D.; Yang, D.; Lee, M. D. Tetrahedron

Lett. 1990, 31, 21.

(43) (a) Olson, S. H.; Danishefsky, S. J . Tetrahedron Lett. 1994, 35,

7901. (b) Evans, D. A.; Hoffman, J . M.; Truesdale, L. K. J . Am.

Chem. Soc. 1973, 95, 5822. (c) For previous syntheses of the

aromatic region of calicheamicin, see: (d) Nicolaou, K. C.; Li,

T.; Masahisa, N; Hummel, C. W.; Hiatt, A.; Wrasildo, W. Angew.

Chem., Int. Ed. Engl. 1988, 27, 1097. (e) Laak, K.; Scharf, H.-

D. Tetrahedron 1989, 45, 5511.

(44) Another synthesis of the calicheamicin oligosaccharide has

recently been published, see: Kim, S.-H.; Augeri, D.; Yang, D.;

Kahne, D. J . Am. Chem. Soc. 1994, 116, 1766.

(9) Darby, N.; Kim, C. U.; Salaun, J . A.; Shelton, K. W.; Takada,

S.; Masamune, S. Chem. Commun. 1971, 1516.

(10) Bergman, R. G. Acc. Chem. Res. 1973, 6, 25.

(11) Snyder, J . P. J . Am. Chem. Soc. 1989, 111, 7630.

(12) Nicolaou, K. C.; Zuccarello, G.; Ogawa, Y. ; Schweiger, E. J .;

Kumazawa, T. J . Am. Chem. Soc. 1989, 111, 4866.

(13) (a) Myers, A. G. Tetrahedron Lett. 1987, 28, 4493, b) Myers, A.

G.; Kuo, E. Y.; Finney, N. S. J . Am. Chem. Soc. 1989, 111, 8057.

(14) (a) Konishi, M.; Ohkuma, H.; Tsuno, T.; Oki, T.; VanDuyne, G.

D.; Clardy, J . J . Am. Chem. Soc. 1990, 112, 3715. (b) Konishi,

M.; Ohkuma, H.; Matsumoto, K.; Tsuno, T.; Kamei, H.; Miyaki,

T.; Oki, T.; Kawaguchi, H.; VanDuyne, G. D.; Clardy, J . J .

Antibiot. 1989, 42, 1449.

(15) Semmelhack, M. F.; Gallagher, J .; Cohen, D. Tetrahedron Lett.

1990, 31, 1521.

(16) Lee, J . E.; Schroeder, D. R.; Langley, D. R.; Colson, K. L.; Huang,

S.; Klohr, S. E.; Lee, M. S.; Golik, J .; Hofstead, S. J .; Doyle, T.

W.; Matson, J . A. J . Am. Chem. Soc. 1993, 115, 8432.

(17) Minami, Y.; Yoshida, K.; Azuma, R.; Saeki, M.; Otani, T.

Tetrahedron Lett. 1993, 34, 2633.

(18) (a) Hitchcock, S. A.; Boyer, S. H.; Chu-Moyer, M. Y.; Olson, S.

H.; Danishefsky, S. J . Angew. Chem., Int. Ed. Engl. 1994, 33,

858. (b) Hitchcock, S. A.; Boyer, S. H.; Chu-Moyer, M. Y. ; Olson,

S. H.; Danishefsky, S. J . J . Am. Chem. Soc. 1995, 117, 5750.

(19) For the first synthesis of the calicheamicin aryl tetrasaccharide,

see: (a) Nicolaou, K. C.; Groneberg, R. D.; Miyazaki, T. ;

Stylianides, N. A.; Schultze, T. J .; Stahl, W. J . Am. Chem. Soc.

1990, 112, 8193. (b) Groneberg, R. D.; Miyazaki, T.; Stylianides,

N. A.; Schultze, T. J .; Stahl, W.; Schreiner, E. P.; Suzuki, T.;

Iwabuchi, Y.; Smith, A. L.; Nicolaou, K. C. J . Am. Chem. Soc.

1993, 115, 7593.

(45) Halcomb, R. L.; Boyer, S. H.; Danishefsky, S. J . Angew. Chem.,

Int. Ed. Engl. 1992, 31, 338.

(46) (a) Sugiura, Y.; Shiraki, T.; Konishi, M.; Oki, T. Proc. Nat. Acad.

Sci. U.S.A. 1990, 87, 3831. (b) Langley, D. R.; Doyle, T. W.;

Beveridge, D. L. J . Am. Chem. Soc. 1991, 113, 4395. (c) Wender,

P. A.; Kelly, R. C.; Beckham, S.; Miller, B. L. Proc. Natl. Acad.

Sci. U.S.A. 1991, 88, 8835. (d) For the first report of a dynemi-

cin-DNA binding constant as well as an elegant study to provide

insight into the dynamics of dynemicin-DNA interaction/

cleavage, see: Myers, A. G.; Cohen, S. B.; Tom, N. J .; Madar, D.

J .; Fraley, M. E. J . Am. Chem. Soc. 1995, 117, 7574.

(47) For syntheses of various dynemicin A model systems, see: (a)

Nicolaou, K. C.; Dai, W.-M. J . Am. Chem. Soc. 1992, 114, 8908

and references cited therein. (b) Wender, P. A.; Zercher, C. K.;

(20) For the first total synthesis of (()-calicheamicinone, see: (a) Paz

Cabal, M.; Coleman, R. S.; Danishefsky, S. J . J . Am. Chem. Soc.

1990, 112, 3253. (b) Haseltine, J . N.; Paz Cabal, M.; Mantlo, N.

B.; Iwasawa, N.; Yamashita, D. S.; Coleman, R. S.; Danishefsky,

44 J . Org. Chem., Vol. 61, No. 1, 1996

Perspective

Beckham, S.; Haubold, E.-M. J . Org. Chem. 1993, 58, 5867 and

references cited therein. (c) Nishikawa, T.; Isobe, M.; Goto, T.

Synlett 1991, 393. (d) Magnus, P.; Fortt, S. M. J . Chem. Soc.,

Chem. Commun. 1991, 544. (e) Takahashi, T.; Sakamoto, Y.;

Yamada, H.; Usui, S.; Fukazawa, Y. Angew. Chem., Int. Ed.

Engl. 1995, 34, 1345.

(58) For the only example of this type of construction to produce a

cyclic enediyne in the context of an annulene synthesis, see:

Huynh, C.; Linstrumelle, G. Tetrahedron 1988, 44, 6337.

(59) Hofmeister, H.; Annen, K.; Laurent, H.; Wiechert, R. Angew.

Chem., Int. Ed. Engl. 1984, 23, 727.

(60) Mitchell, T. N.; Amamria, A.; Killing, H.; Rutschow, D. J .

Organomet. Chem. 1986, 304, 257.

(61) Shair, M. D.; Yoon, T. Y.; Danishefsky, S. J . J . Org. Chem. 1994,

59, 3755.

(48) For the first synthesis of the ABC ring system of dynemicin A,

see: (a) Wood, J . L.; Porco J . A., J r.; Taunton, J .; Lee, A. Y.;

Clardy, J .; Schreiber, S. L. J . Am. Chem. Soc. 1992, 114, 5898.

(b) Porco, J . A., J r.; Schoenen, F. J .; Stout, T. J .; Clardy, J .;

Schreiber, S. L. J . Am. Chem. Soc. 1990, 112, 7410.

(49) For the biological evaluation of dynemicin model systems, see:

Nicolaou, K. C.; Dai, W.-M.; Tsay, S. C.; Estevez, V. A.; Wrasildo,

W. Science 1992, 256, 1172.

(50) For the addition of cyanide anion to a variety of quinoline

structures, see: (a) Popp, F. D. Chem. Heterocycl. Comp. 1982,

32, 353. For the addition of alkynyl Grignard reagents to

acylpyridinum ion intermediates, see: (b) Yamaguchi, R.; Na-

kazono, Y.; Matsuki, T.; Hata, E.; Kawanisi, M. Bull. Chem. Soc.

J pn. 1987, 60, 215 and references cited therein.

(62) Tirpak, R. E.; Olsen, R. S.; Rathke, M. W. J . Org. Chem. 1985,

50, 4877.

(63) Previously, we had demonstrated that a SEM-protected vinylo-

gous carbonate could be efficiently deprotected with MgBr₂in

the presence of the epoxy enediyne system, see ref 61. For the

first disclosure that MgBr₂could be used to deprotect MOM/

SEM esters, see: Kim, S.; Park, Y. H.; Kee, I. S. Tetrahedron

Lett. 1991, 32, 3099.

(64) Shair, M. D.; Yoon, T. Y.; Chou, D.; Danishefsky, S. J . Angew.

Chem., Int. Ed. Engl. 1994, 33, 2477.

(65) Ph.D. Thesis of M. D. Shair, Columbia University, 1995.

(66) Kita, Y.; Akai, S.; Ajimura, N.; Yoshigi, M.; Yasuda, H.; Tamura,

Y. J . Org. Chem. 1986, 51, 4150.

(51) Yoon, T. Y.; Shair, M. D.; Danishefsky, S. J .; Schulte, G. K. J .

Org. Chem. 1994, 59, 3752.

(52) Taken from the Ph.D. Thesis of T. Y. Yoon, Yale University,

1994.

(53) Corey, E. J .; Fuchs, P. L. Tetrahedron Lett. 1972, 13, 3769.

(54) (a) Stephens, R. D.; Castro, C. E. J . Org. Chem. 1963, 28, 3313.

(b) Sonogashira, K.; Tohda, Y.; Hagihara, N. Tetrahedron Lett.

1975, 16, 4467.

(55) Yoon, T. Y.; Shair, M. D.; Danishefsky, S. J . Tetrahedron Lett.

1994, 35, 6259.

(67) For the use of homophthalic anhydrides in annulative sequences

with various quinones, see: Tamura, Y.; Fukata, F.; Sasho, M.;

Tsugoshi, T.; Kita, Y. J . Org. Chem. 1985, 50, 2273 and

references cited therein.

(68) (a) Shair, M. D.; Yoon, T. Y.; Chou, D.; Danishefsky, S. J . 08/

347, 952. (b) Chou, T. C.; Shair, M. D.; Yoon, T. Y.; Zheng, Y.

H.; Danishefsky, S. J . Proc. Am. Assoc. Cancer Res. 1995, 36,

384.

(69) Ho, S. N.; Boyer, S. H.; Schreiber, S. L.; Danishefsky, S. J .;

Crabtree, G. R. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 9203.

(70) Takahashi, T.; Sakamoto, Y.; Yamada, H.; Usui, S.; Fukazawa,

Y. Angew. Chem., Int. Ed. Engl. 1995, 34, 1345.

(56) Morrow, G. W.; Swenton, G. S.; Filppi, J . A.; Wolgemuth, R. L.

J . Org. Chem. 1987, 52, 713.

(57) For other syntheses of the anthraquinone portion of dynemicin,

see: (a) Chikashita, H.; Porco, J . A., J r.; Stout, T. J .; Clardy, J .;

Schreiber, S. L. J . Org. Chem. 1991, 56, 1692. (b) Nicolaou, K.

C.; Gross, J . L.; Kerr, M. A.; Lemus, R. H.; Ikeda, K.; Ohe, K.

Angew. Chem., Int. Ed. Engl. 1994, 33, 781. (c) Magnus, P.;

Eisenbeis, S. A.; Magnus, N. A. J . Chem. Soc., Chem. Commun.

1994, 1545.

J O951560X

Article Doi

Observations in the chemistry and biology of cyclic enediyne antibiotics: Total syntheses of calicheamicin γ1I and dynemicin A

DOI: 10.1021/jo951560x

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jo951560x

Full text of DOI:10.1021/jo951560x

Products guided by the article

R&D Labs maybe for 168421-51-0

Relevant to this article

Hot Product