Synthetic studies related to CP-225,917

811

AWARD LECTURE / CONFÉRÉNCE D’HONNEUR

Synthetic studies related to CP-225,917¹

Derrick L.J. Clive, Paulo W.M. Sgarbi, Xiao He, Shaoyi Sun, Junhu Zhang, and

Ligong Ou

Abstract: Synthetic studies related to CP-225,917 are described, including the preparation of the fully oxygenated

tetracyclic central core 3.

Key words: CP-225,917, siloxy-Cope rearrangement, Cope rearrangement, strain-assisted rearrangement, furan, ruthe-

nium dioxide, bridgehead olefin.

Résumé : On décrit des études synthétiques reliées au CP-225,917 comprenant la préparation du noyau central tétracy-

clique complètement oxygéné 3.

Mots clés : CP-225,917, réarrangement siloxy-Cope, réarrangement de Cope, réarrangement assisté par la tension, fu-

rane, dioxyde de ruthénium, oléfine pontée.

[Traduit par la Rédaction] Clive et al.

824

Introduction

Bredt’s rule. This accommodation is possible mainly

because the double bond is in a nine-membered ring.

When we began work in this area no synthesis had yet

been published, but there was some background information

available in the literature. CP-225,917 is really a hemiacetal

of the structural type 4, and the two related substances 6 and

7 had been prepared many years ago (6) as a mixture, but

the method used to make them — acid catalyzed dehydra-

tion of alcohols 5 — did not seem to be well-suited to the

task of synthesizing the natural product, mainly because

there was little control over the final position of the double

bond. In fact, the bridgehead olefin 7 was the minor product

(Scheme 2).

There are a good many ways in which the synthesis of ke-

tones of type 4 might be attempted, and after some explor-

atory experiments, we decided to look at the possibility of

using an oxy-Cope rearrangement (7).

We had noticed a potential relationship between the CP-

225,917 bridgehead olefinic core and the [2.2.1] bicyclic

structure 8 (Scheme 3).

I am going to describe synthetic work related to the natu-

ral product CP-225,917 (1) (1, 2).³The compound is an in-

hibitor of ras farnesyl transferase, and this fact means that it

might serve as a lead structure for the design of anticancer

drugs (1). It also has another significant biological property —

it inhibits squalene synthase (1) — but the real attraction to

an organic chemist is the structural complexity of the mole-

cule, and a great deal of synthetic work has now been pub-

lished in this area.⁴The unnatural enantiomer of CP-225,917

has been made by the Nicolaou group (3) and racemic mate-

rial by the Danishefsky group (4). Two syntheses of the re-

lated compound CP-263,114 (2) have also been reported (5).

This compound can be generated from 1 by treatment with

methane sulfonic acid (1), and the reverse transformation —

conversion of 2 into 1 — has been achieved under controlled

basic conditions (3).

My own research has led to the synthesis of the fully oxy-

genated core structure 3, which is a crystalline substance

whose dimensions were obtained by single crystal X-ray

analysis (Scheme 1).

CP-225,917 is an unusual molecule in which the bridge-

head double bond is embedded within a framework that can

fairly easily accommodate such a bond without violating

If 8 is subjected to conditions for an anionic oxy-Cope re-

arrangement, then enolate 9 would be formed, and we hoped

it could be trapped by a suitable electrophile, such as

Mander’s reagent, so as to generate a keto ester (9 →10).

When the keto ester is redrawn as 11, its resemblance to the

Received 18 February 2003. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 4 July 2003.

D.L.J. Clive,²P.W.M. Sgarbi, X. He, S. Sun, J. Zhang, and L. Ou. Department of Chemistry, University of Alberta, Edmonton,

AB T6G 2G2, Canada.

¹2002 Alfred Bader Award Lecture. Presented at C.S.C. meeting, Vancouver, 3 June 2002.

²Corresponding author (e-mail: derrick.clive@ualberta.ca).

³Non-systematic numbering is used in diagram 1, and the corresponding numbers are used for all other structures.

⁴For references to the many model studies, see ref. 2f.

Can. J. Chem. 81: 811–824 (2003)

doi: 10.1139/V03-086

812

Can. J. Chem. Vol. 81, 2003

Scheme 1.

Scheme 3.

Scheme 4.

Scheme 2.

Scheme 5.

core of CP-225,917 is obvious. Structures 8–11 include sev-

eral generic groups, R¹–R⁵, in order to show that this ap-

proach should accommodate the early introduction of some

of the substituents that are present in the natural product.

To test our plan, we carried out a model study in which

most of these generic substituents were absent.

The short sequence shown in Scheme 4, which is reported

in the literature (8), was repeated: a Diels–Alder reaction be-

tween tetrachlorocyclopentadienone dimethyl acetal (12) and

vinyl acetate, followed by acetate hydrolysis (13 →14), and

dechlorination (14 → 15) gave us the basic [2.2.1] bicyclic

skeleton. The yields are good, but the starting material 12 is

expensive.

The double bond in 15 was then hydrogenated, and the

hydroxyl was protected as its benzoyl ester (Scheme 5). We

later found that this hydrogenation is unnecessary because, if

Clive et al.

813

Scheme 6.

Scheme 8.

Scheme 9.

Scheme 7.

possibility of fragmentation by a retroaldol mechanism. We

tried to make the tert-butyldimethylsilyl ether or a p-

methoxybenzyl ether, but those experiments were not suc-

cessful.

At this point, to set the stage for an anionic oxy-Cope re-

arrangement, we had to introduce an exocyclic double bond

at the CH₂group adjacent to the carbonyl. Formation of the

double bond was accomplished by a short sequence

(Scheme 7) starting with an aldol condensation with acetal-

dehyde (21 → 22).

The aldol mixture was treated with mesyl chloride and

then with DBU. Those operations gave ketone 24a in rea-

sonable yield, plus a small amount of the Z isomer (24b).

The compounds are easy to separate, and we continued our

experiments using only the major isomer — the E olefin.

Sodium borohydride reduction in the presence of cerium

chloride gave a 1:1 mixture of exo and endo alcohols

(Scheme 8, 25-exo, 25-endo). Each of these was separately

benzylated under standard conditions and then desilylated

(25 → 26 → 27), and we were then ready to test the anionic

oxy-Cope rearrangement.

When the alcohols 27-endo and 27-exo were individually

heated in toluene in the presence of potassium hexamethyl-

disilazide, they were slowly converted into the desired

enolates, and workup gave ketones 28 and 29, respectively,

in good yield (Scheme 9). When we tried the reaction in

an excess of sodium is used in the dechlorination step, as

well as a longer reaction time (1 h instead of 5 min), then

the double bond is also reduced and in better overall yield.

Finally, acid hydrolysis of the acetal released the parent

ketone (16 → 17). When we tried to hydrolyze the acetal

without protecting the hydroxyl, extensive decomposition

occurred, presumably, by a retroaldol process.

Ketone 17 reacted with vinylmagnesium bromide to give

mainly alcohol 18. We did not separate the C(2) isomer at

this stage, but the yield of isomer 18 is at least 53%.

The facial selectivity is presumably because of the fact

that, of the two single bonds “a” and “b” (see 18), the for-

mer is more electron-rich and interacts preferentially with

the developing sigma star orbital at C(2) (9) if the organo-

metallic attacks from the right.

The mixture of hydroxy vinyl benzoates resulting from

the Grignard addition was then hydrolyzed (Scheme 6). Ini-

tially, the hydrolysis was done with lithium hydroxide, but

later, the process of removing the benzoyl group was im-

proved by using LiAlH₄. The diol 19 was easy to purify, and

at this point the C(2) epimers were separated.

The secondary hydroxyl of 19 was now oxidized with the

Dess–Martin reagent, and then the tertiary hydroxyl was

protected by silylation. This was done in order to hinder the

814

Can. J. Chem. Vol. 81, 2003

Scheme 10. (a) LDA, THF, –78°C, then excess BnOCH₂CHO,

–78°C; (b) MsCl, Et₃N; DBU; (c) BnOCH₂SnBu₃, BuLi, i-

PrMgCl, CuBr·SMe₂, BF₃·Et₂O, –78°C; (d) BF₃·OEt₂, –45°C;

PhSeCl, HMPA, –45 to 0°C; 66% for E, 65% for Z-isomer;

(e) H₂O₂, pyridine, CH₂Cl₂, 30–35°C; 84%; (f) LiBH₄,

CeCl₃·7H₂O, MeOH, 0°C; (g) BnBr, NaH, THF, reflux.

Scheme 11.

refluxing THF the process was slower, and addition of 18-

crown-6 did not accelerate the reaction. Although these an-

ionic oxy-Cope rearrangements are slow, the experiments

served the purpose of showing that bridgehead olefins struc-

turally related to CP-225,917 could be generated from

[2.2.1] bicyclic systems.

What we had to do next was to repeat the sequence with

an additional carbon at C(5) in place of the hydrogen (see

structures 27), so that the required C(5) quaternary center

would be generated in the rearrangement.

To that end, ketone 21 was condensed with benzyloxy-

acetaldehyde (Scheme 10), and the hydroxyl in the products

was eliminated by mesylation and base treatment to afford

enones 31. These underwent conjugate addition with [(ben-

zyloxy)methyl]cuprate in the presence of boron trifluoride

and isopropylmagnesium chloride. The latter scavenges

traces of copper(II) that would otherwise cause dimerization

of the cuprate reagent. The purpose of the borontrifluoride is

simply to activate the enone to conjugate addition. The re-

sulting enolate reacted with benzeneselenenyl chloride. We

found that if the enolate from the conjugate addition is

quenched to afford the ketone, then we could not introduce

the selenium by deprotonation with LDA and reaction with

benzeneselenenyl chloride. Evidently, the selenation is de-

pendent on the precise nature of the enolate; we also found

that the presence of HMPA is essential. Finally, selenoxide

elimination (32 → 33) proceeded without incident, bringing

us to ketone 33. Reduction with lithium borohydride in the

presence of ceric chloride gave the corresponding alcohol as

a separable mixture of endo and exo isomers.

The hydroxyl was protected by benzylation (34 → 35),

and we then removed the silyl group by treatment with tetra-

butylammonium fluoride in the usual way (Scheme 11).

With 36 (and the corresponding exo-isomer) in hand, we

were ready to test the oxy-Cope rearrangement. Surprisingly,

this reaction did not work. We tried a variety of conditions

and, besides an anionic process, we also examined purely

thermal reactions, as well as potential catalysis by palladium

or mercuric ion. Scheme 11 shows our observations with an

endo O-benzyl group (see 36), but the exo isomer behaved in

the same way. We also heated the silyl ether 35-endo neat at

230°C. Again no rearrangement occurred.

The fact that our first model, in which the exocyclic dou-

ble bond is only trisubstituted (see Scheme 9), did rearrange

prompted us to make compounds 38 (Scheme 11) in the

hope that the trisubstituted nature of the double bond would

permit oxy-Cope rearrangement. However, even these com-

pounds did not rearrange under anionic conditions or in the

presence of a palladium catalyst.

Clive et al.

815

Scheme 12.

Scheme 13.

1-methyl-2-pyrrolidinone (NMP) is an excellent solvent that

facilitates the desired rearrangement (Scheme 12).

Heating compound 35-endo for a very long time gave

ketone 37 in high yield, if correction is made for recovered

starting material. The initial product is, of course, a silyl

enol ether, but the silyl group is probably transferred to the

solvent, so that on workup we get the corresponding ketone.

We have carried out a number of related reactions in

which we vary the substitution pattern on the exocyclic dou-

ble bond, the orientation (exo or endo) of the oxygen func-

tion on the [2.2.1] bicycle, as well as the nature of the

protecting groups. In all cases, we observe rearrangement

(see Scheme 12, compounds 41, 43–48, and 50). Sometimes,

as in the formation of 45 and 46, which both arise from the

same starting material, some of the silyl enol ether is actu-

ally isolated. The same is true in the case of 47 and 48.

When the phenylthio substrate 49 was heated it gave 50,

as expected. The reaction is faster than the other cases, al-

though the yield is not very good.

1-Methyl-2-pyrrolidinone has been reported before (10,

11) as a solvent for oxy-Cope rearrangement, but only for al-

cohols, where it can become involved in hydrogen bonding;

its use for silyl ethers has not been recognized. All of our re-

arrangements are very clean if degassed 1-methyl-2-

pyrrolidinone is used, and so we now had a method for mak-

ing the simplified bridgehead olefin substructure of CP-

225,917.

While these studies were going on, another member of my

group examined an alternative way of speeding up the

siloxy-Cope rearrangement.

We had noticed from inspection of models that incorpora-

tion of the exocyclic double bond into a lactone ring, as in

structure 51 (Scheme 13), generates some strain in the

[2.2.1] bicyclic system; this should provide a driving force

for rearrangement, and in the event, that is exactly what was

observed. Lactone 51 was prepared as shown in Schemes 14

and 15.

Aldol condensation of ketone 21 with (p-methoxybenzyl-

oxy)acetaldehyde (53) worked very well, and the aldols were

mesylated and treated with DBU. Those experiments al-

lowed us to isolate the desired Z olefin 55 in just under 60%

yield. The p-methoxybenzyl group was removed by the ac-

tion of DDQ, and with this reagent an aldehyde (56) was ob-

tained directly. Oxidation in the standard manner gave the

corresponding acid, and that was trapped as its methyl ester

(56 → 57 → 58).

It was clear that the number as well as the length of the

substituents on the exocyclic double bond are critical factors

that determine the ease of rearrangement, and so we studied

the two truncated models 39 and 40. Again we tried a num-

ber of conditions: anionic for alcohol 39 and catalyzed and

purely thermal for the silyl ether 40. However, only simple

heating of the silyl ether as a neat oil gave the desired prod-

uct 41, although the yield was poor (23%).

In the compounds we had examined so far the C(2) vinyl

group was free to adopt a conformation in which it points

away from the exocyclic double bond, and so we decided to

try to bias the conformational preference by placing a bulky

substituent on the vinyl group. To this end we prepared com-

pounds 42, but even they did not rearrange when heated to

300°C.

Reduction of 58 with sodium borohydride in the presence

of ceric chloride gave largely the desired exo alcohol 59

(Scheme 15). The endo isomer was isolated in a little under

20% yield, but we did not try to recycle the material by oxi-

dation and reduction.

The modest conversion of 40 into 41 caused us to exam-

ine thermal processes more closely, and we soon found that

Ester hydrolysis now gave the hydroxy acid 60, and the

material was cyclized to lactone 51 using N-methyl-2-

816

Can. J. Chem. Vol. 81, 2003

Scheme 14.

Scheme 15.

Scheme 16.

chloropyridinium iodide. When the lactone was heated in

o-dichlorobenzene it rearranged smoothly in the required

manner, and we could isolate the product in almost 80%

yield. We later found that these conditions are harsher than

they need to be, but it was clear that the additional strain as-

sociated with the lactone unit does indeed facilitate the oxy-

Cope process. It is also very fortunate that the additional

strain is not so large as to hinder formation of the lactone or

to make it very sensitive to hydrolysis. The use of a strained

lactone has also been investigated independently by Bio and

Leighton, who have priority of publication (2b, 12).

At this point in our research we had two methods for con-

verting [2.2.1] bicyclic systems into bridgehead olefins that

resemble CP-225,917, but we decided to use only the

strained lactone approach for further work, and we dealt first

of all with the problem of generating the quaternary center.

Ketone 21 was hydroxylated under standard conditions

(Scheme 16), and then oxidation by the Dess–Martin proce-

dure afforded the corresponding α-diketone 62. This was

condensed with the dianion derived from methyl 3-hydroxy-

propanoate (63). That experiment afforded a mixture of two

isomers, which were easily separated, but the stereochem-

istry was not determined. The primary hydroxyl of the major

isomer was silylated (t-BuPh₂SiCl), and dehydration with

thionyl chloride and pyridine served to generate the Z enone

66. The yields in this sequence are good, but there is no

stereocontrol in the aldol condensation; that is a factor that

still has to be dealt with.

Reduction of ketone 66 with the sodium borohydride –

ceric chloride combination gave mainly the exo alcohol 67

(Scheme 17), and again we did not try to recycle the endo

isomer. Demethylation of the ester was accomplished this

time with the lithium salt of propanethiol, and then cycli-

zation, as before, generated the strained lactone 68. When

this was heated in o-dichlorobenzene for 10 min it seemed to

have rearranged completely, and after a further 10 min we

could isolate the siloxy-Cope rearrangement product 69 in

quantitative yield.

That experiment showed that we now had a procedure for

dealing with the C(5) quaternary center, and the next thing

to do was to find out how to construct the anhydride and

how to make the exocyclic double bond of 66 in a stereo-

controlled manner. We dealt first of all with the stereochem-

ical problem.

In our earlier work we had used the dianion derived from

hydroxy ester 63; we now (Scheme 18) used an ester with a

protected hydroxyl (70), and we found that the isomer ratio

Clive et al.

817

Scheme 17.

Scheme 18.

was improved from 4:3 for 64a:64b to 10:1 for 71a:71b. It

is not clear why this happens, but it was certainly a very

welcome fact.

Scheme 19.

When the major isomer (71a) was dehydrated with thionyl

chloride and pyridine (Scheme 19), we were very pleased to

obtain the desired Z olefin 72 (Scheme 19). Reduction of the

ketone carbonyl with sodium borohydride showed only a 3:1

selectivity in favor of the required exo alcohol, but the endo

isomer was easily reconverted into the starting ketone (73-

endo → 72). Once again, the ester was demethylated with

the lithium salt of propanethiol, and the resulting hydroxy

acid was cyclized to the lactone (73-exo → 74). Finally, ther-

mal rearrangement gave us a product (75) with the quater-

nary center and the correct chain length at C(5).

At this point we needed to develop a method for making

the anhydride unit that is characteristic of CP-225,917.

Our initial experiments (Scheme 20) were done with 45

and some related compounds, which were all obtained by

the method using thermolysis in 1-methyl-2-pyrrolidinone.

We discovered that these compounds have a tendency to

enolize towards the bridgehead, and so, in order to introduce

C(15) (which is needed for the eventual anhydride), we had

to block the bridgehead position. That can be done, as

shown in Scheme 20, but such a route is cumbersome, and

so we modified our approach in such a way that C(15) is in-

troduced at a very early stage — in fact, before oxy-Cope re-

arrangement.

Our route (Scheme 21) follows along the lines of our pre-

vious experiments, except that we treat ketone 17 with iso-

propenylmagnesium bromide instead of vinylmagnesium

bromide. Alcohol 78 is formed in acceptable yield, and we

then oxidized what I call C(15), using tert-butylhydroper-

oxide in the presence of a catalytic amount of selenium di-

oxide. This allylic oxidation is quite slow — it takes 3 days

at room temperature. The desired alcohol 79 can be isolated

in 60% yield, but almost 30% of the corresponding aldehyde

80 is also formed. Fortunately, the two compounds are easy

to separate, and the aldehyde can be reduced efficiently to

the alcohol, so that the overall transformation to 79 is quite

efficient (81%).

Next, the primary hydroxyl at C(15) was protected as its

p-anisyloxymethyl (AOM) ether, under phase-transfer condi-

tions (79 → 81, 69%). The selectivity is not as high as I

would have wished, and 16% of the doubly etherified mate-

rial is also formed, but that product can be hydrolyzed in

over 80% yield back to the starting diol 79, and so, after one

recycling, the mono-protected compound 81 can be obtained

in 82% yield.

The remaining tertiary hydroxyl (at C(2)) was then

silylated (Scheme 22, 81 → 82); the benzoate group was re-

moved by treatment with lithium aluminum hydride, and the

818

Can. J. Chem. Vol. 81, 2003

Scheme 20.

Scheme 22.

Scheme 21.

Scheme 23.

alcohol released in that experiment was oxidized with the

Dess–Martin reagent (82 → 83 → 84). The resulting ketone

was subjected to α-hydroxylation and then to oxidation,

again with the Dess–Martin reagent, so as to generate diket-

one 85. All these experiments are very simple, and the

yields are good. We condensed the diketone with our

silylated ester (70) and obtained the condensation product 86

as a single isomer in nearly 80% yield. We did not look for

other isomers in the reaction mixture. Dehydration with

thionyl chloride generated the required Z double bond in

high yield (86 → 87), and once again we were at a stage

where we needed to form a strained lactone and then carry

out the thermal rearrangement.

6 h at room temperature — but the stereoselectivity was

good. The ester was demethylated in the usual way, and

lactonization was again achieved in high yield with the

pyridinium reagent (87 → 88 → 89). When we did the

siloxy-Cope rearrangement (89 → 90), we used chloroben-

zene instead of o-dichlorobenzene. Its boiling point is 55°C

lower, and so the reaction takes longer, but the yield is al-

All those transformations were achieved by the methods

that had served us well in the simpler models, although we

did make some improvements.

The reduction of ketone 87 was done with a very hindered

hydride (Scheme 23). The reaction was quite slow — it took

Clive et al.

819

Scheme 26. (a) Dess–Martin oxidation, CH₂Cl₂, 89%; (b) DBU,

THF, then HCl, 5 min, 97%; (c) Rose Bengal, tungsten light,

DBU, O₂, –78°C, then 0°C; (d) Pr₄NRuO₄, N-methylmorpholine

N-oxide, 86% overall.

Scheme 24. (a) LiBH₄, THF, 93%; (b) CF₃CO₂H, THF, water,

67%; (c) t-BuPh₂SiCl, Et₃N, DMAP, CH₂Cl₂, 99%; (d) t-

BuMe₂SiOSO₂CF₃, 2,6-lutidine, CH₂Cl₂, 95%.

Scheme 25. (a) Tebbe reagent, THF, 77%; (b) bromocatecholbor-

ane, CH₂Cl₂, –78°C, 84%; (c) VO(acac)₂, t-BuOOH, PhH, 70%.

Scheme 27. (a) DIBAL-H, CH₂Cl₂, –78°C, 95%; (b) CF₃CO₂H–

THF–water, 90%; (c) CH(OMe)₃, TsOH·pyridine, 96%.

most the same, and we felt it was advisable to practice with

milder conditions, with a view to eventually using more del-

icate substrates.

In order to construct the anhydride unit, we need to add

one carbon at C(2) (see 90, Scheme 24), and so we had to

first liberate the potential carbonyl at C(2) and protect the

lactone carbonyl.

The lactone ring was opened by reduction with lithium

borohydride (90 → 91), and then the silyl enol ether was hy-

drolyzed (91 → 92) to set the stage for introduction of C(14)

of the anhydride unit.

The primary hydroxyl of 92 was selectively protected as

its tert-butyldiphenylsilyl ether, after which the secondary

hydroxyl was silylated with tert-butyldimethylsilyl triflate.

Both protection steps go in good yield, and the next task was

to introduce a carbon at C(2).

93 with 2 equiv of the Tebbe reagent gave the desired olefin

94 in acceptable yield (77%).

Removal of the AOM group (94 → 95) was carried out

with bromocatechol borane. With a number of related sub-

strates we had found that ceric ammonium nitrate, magnesium

bromide, bromodimethylborane, or N-bromosuccinimide did

not work, but bromocatecholborane was satisfactory.

The homoallylic alcohol 95 was amenable to epoxidation,

which brought the sequence to compound 96, from which

point we could finish construction of the anhydride, as

shown in Scheme 26.

The free hydroxyl group of 96 was oxidized to the corre-

sponding aldehyde with the Dess–Martin reagent. In this

case, the Swern procedure gave a complex mixture. Treat-

ment of the epoxy aldehyde first with DBU to open the

epoxide by deprotonation α to the aldehyde group, followed

by brief treatment with acid to effect dehydration, gave the

In some earlier studies we had found that ketones of simi-

lar structure to 93 did not behave properly with Wittig or Pe-

terson reagents, but that they did react with the Tebbe

reagent, and in the present case (Scheme 25), treatment of

820

Can. J. Chem. Vol. 81, 2003

Scheme 28.

Scheme 29.

expected furan 98. This was then converted by photo-

oxygenation to a mixture of hydroxybutenolides 99 (and the

regioisomers with the OH and C=O interchanged), which

were oxidized with TPAP. Those experiments all work in

good yield, and they afforded anhydride 100.

Having reached a model with the quaternary center and

the anhydride, the next step was to generate the lactone unit

that spans C(10) and C(11).

Our attempts to do that, starting from 100 or its furan pre-

cursor 98, were not at all promising, and so we returned to

the product of the siloxy-Cope rearrangement (90) and elab-

orated it in a different way.

Previously, we had reduced the lactone 90 down to the

diol stage (see Scheme 24); this time we took the lactone

only to the lactol oxidation level (Scheme 27, 90 → 101).

The lactols were then treated with acid in order to hydrolyze

the silyl enol ether. That experiment gave a single lactol

(102) whose stereochemistry was not established. Another

isomer was isolated in 7% yield. The major product (102)

was converted into lactol methyl ether 103, which has the

stereochemistry shown.

The next task was to introduce a carbon at C(2) (see 103).

Use of the Tebbe reagent gave a very low yield, but we

eventually found that a modified Peterson reaction (13)

worked very well.

when ceric ammonium nitrate is used as the oxidant (14).

Scheme 29 shows compound 106 drawn in such a way as to

emphasize the structural similarity between an AOM group

and p-dimethoxybenzene. We found that, although ceric am-

monium nitrate itself causes no reaction with 106, in the

presence of the pyridine oxide diacid we can get almost 80%

yield of the parent alcohol 110. The beneficial effect of the

catalyst appears to be a general one for removal of AOM

groups.

Epoxidation under standard conditions (Scheme 30) took

us to 111, and the corresponding aldehyde was made by

Dess–Martin oxidation. Treatment with base served to open

the epoxide, and then brief treatment with hydrochloric acid

completed assembly of the furan (111 → 112). Once again,

photo-oxygenation and TPAP oxidation converted the furan

into the anhydride (112 → 113).

At this point the remaining transformations were oxida-

tion of the side chain on the quaternary center and introduc-

tion of oxygen at C(10).

We decided to approach these tasks by regenerating the

lactone unit that spans C(5) and C(10). We would then open

the lactone by basic hydrolysis so as to expose the C(10) ox-

ygen as an alcohol, which we would oxidize in the basic me-

dium to the corresponding ketone. The ketone, in turn,

should spontaneously form the desired hemiacetal on acidifi-

cation.

The lithium salt 104 was added to ceric chloride to gener-

ate a reagent that gave alcohol 105 very efficiently

(Scheme 28).

The alcohol is crystalline, and its structure was assigned

by X-ray analysis. Deprotonation of the alcohol with potas-

sium hexamethyldisilazide gave the desired olefin 106.

When we then tried to remove the AOM group, we found

the reaction very troublesome. In fact, treatment with ceric

ammonium nitrate under standard conditions simply did not

work — the AOM group is inert, to our surprise. We tried a

variety of methods and almost abandoned this protecting

group, but, fortunately, we recognized something about the

deprotection that proved to be very helpful.

We realized that the mechanism for deprotection should

be similar to that for converting p-dimethoxybenzene (107)

into p-benzoquinone (109) (Scheme 29), and so we exam-

ined the literature for that transformation and found that the

pyridinedicarboxylic acid N-oxide 108 facilitates reaction

Clive et al.

821

Scheme 30.

Scheme 31.

Scheme 32.

To regenerate the lactone, we needed to replace the

methoxy group of the lactol ether 113 by an OH group. Acid

hydrolysis and trimethylsilyl iodide were unsuitable, but bo-

ron trichloride – dimethyl sulfide complex performed well

(Scheme 31) and gave the desired lactols 114 as a mixture of

two isomers, together with some of the corresponding chlo-

ride. Treatment with aqueous silver nitrate converted the

chlorine byproduct into the lactols, so that the overall yield

is high. Finally, Dess–Martin oxidation regenerated the lac-

tone subunit (114 → 115). Most of our Dess–Martin oxida-

tions are done at room temperature; this one required rather

more vigorous conditions but was still efficient. Lactone 115

is crystalline, and X-ray analysis confirmed the structure.

With the lactone in place, we now had to oxidize the side

chain, and to prepare for that the silicon group was removed

by prolonged exposure — some 50 h — to aqueous tri-

fluoroacetic acid.

The resulting primary alcohol (116) was oxidized first to

the aldehyde and then to the carboxylic acid (Scheme 32), so

as to obtain compound 117. This substance does not have

very convenient chromatographic behavior and was, there-

fore, used crude.

The material was stored for 12 h in 1 M aqueous sodium

hydroxide with the intention of opening both the lactone and

the anhydride. The resulting salt was oxidized by addition of

ruthenium dioxide (15) to the basic solution, which was kept

at 70–80°C for 12 h. After acidification, we were able to iso-

late the complete tetracyclic core of CP-225,917 (3) in 40%

yield from alcohol 116.

The core is crystalline, and X-ray analysis gave us the di-

mensions of the molecule. A noticeable feature of the struc-

ture is the shape of the bridgehead double bond. The four

atoms C(8), C(7), C(6), and C(10) are in a plane — the devi-

ation is less than 1° — but C(5) is 19° out of that plane. The

other dimensions of the molecule are not unusual, although

the lactone carbonyl angle is on the small side — 110°.

During this work we had encountered a large number of

unexpected difficulties, and so we took the precaution of de-

vising an alternative route to the core. This route, which

turned out to be shorter, involves a strain-assisted Cope

rearrangement as opposed to oxy-Cope or siloxy-Cope rear-

rangement, and we also used a different method for making

the anhydride.

The starting point is norbornene (120), which was con-

verted in four simple steps into ester acetal 123

(Scheme 33). These steps are all reported in the literature

(16). A Prins reaction converts norbornene into the bis-

formate 121; Jones oxidation gives the corresponding keto

acid 122, and methylation with diazomethane, followed by

ketalization, provides 123. Although the yield in the oxida-

tion step is low, the experiments are simple, and the ketal es-

ter is easily made on a 20-g scale.

822

Can. J. Chem. Vol. 81, 2003

Scheme 33.

Scheme 34.

Scheme 35.

Deprotonation with LDA and condensation with parafor-

maldehyde affords a mixture of alcohols, and the major iso-

mer (124) can be isolated in 43% yield. The carbon of the

hydroxymethyl group was destined to become one of the

carbonyl carbons of the anhydride, and it has been intro-

duced very early, thereby bypassing some of the difficulties

we had experienced in the first route.

After protection of the hydroxyl, the ester was treated

with methyllithium. That experiment gave ketone 126.

The next task was to introduce the second carbonyl carbon

of the anhydride. To that end, our ketone was first converted

into an enol phosphate.

That was accomplished (Scheme 34) in the usual way, by

deprotonation and treatment with diethyl chlorophosphate

(126 → 127). The product was treated with methylmag-

nesium bromide in the presence of nickel acetylacetonate

(17) in warm dibutyl ether, and the result was introduction

of a methyl group that was destined to become one of the

carbonyl carbons of the anhydride.

Next the ketal was hydrolyzed, and the resulting ketone

(129) was reduced with DIBAL and acetylated to give a

mixture of epimeric acetates (130).

seven steps we had been working with mixtures of epimeric

acetates, and that is why the Schemes do not indicate indi-

vidual yields. The overall transformation can be done in

45% yield from ketone 129, representing an average of 89%

per step.

With ketone 133 in hand we were now in familiar terri-

tory. Both carbons for the anhydride carbonyls were in

place, and we now had to build up the strained lactone sys-

tem.

We used almost the same reactions that had served us well

in the previous experiments. Ketone 133 was hydroxylated

(Scheme 36) by treating the derived enolate with MoOPH

(oxodiperoxymolybdenum(pyridine)(hexamethylphosphoric-

triamide)), and then oxidation gave the symmetrical diketone

To place an oxygen at C(15), the acetates were treated

with a catalytic amount of selenium dioxide in the presence

of tert-butyl hydroperoxide (Scheme 35). The reaction was

very slow and there was some overoxidation to the aldehyde,

but treatment with sodium borohydride in the presence of

ceric chloride served to make the necessary adjustment, and

desilylation in the usual way gave diols 131 in a little under

70% from ketone 129.

The free hydroxyls were protected as MOM ethers

(131 → 132), and we removed the acetyl group with DIBAL.

Finally, Dess–Martin oxidation gave the expected ketone

133 — as a single compound, of course. During the last

Clive et al.

823

Scheme 36. (a) (Me₃Si)₃NK, THF, MoOPH, –23°C, 72%;

(b) Dess–Martin oxidation, CH₂Cl₂, 77%; (c) ester 70, LDA,

THF, –78°C; (d) SOCl₂, pyridine; (e) NaBH₄, CeCl₃·7H₂O, 55%

from 134; (f) PrSLi, HMPA, 83%; (g) 2-chloro-1-methylpyridi-

nium iodide, CH₂Cl₂, reflux, 77%.

Scheme 37.

Scheme 38. (a) NaClO₂, NaH₂PO₄, 2-methyl-2-butene, t-BuOH,

water, 12 h; (b) Pr₄NRuO₄, N-methylmorpholine N-oxide, 50%

from 139; (c) 1 N NaOH, t-BuOH, 70°C, 12 h; catalytic

RuO₂·xH₂O, stoichiometric NaIO₄; pH 3, ≥40%.

134. That was condensed with the enolate derived from the

protected ester 70, and the resulting alcohols were dehy-

drated with the thionyl chloride – pyridine combination.

Finally, reduction with sodium borohydride afforded alcohol

135 (134 → 135). In retrospect, we should have tried a very

hindered reducing agent because it ought to have given

better stereoselectivity, but nonetheless, the overall yield for

the three steps used to convert 134 into 135 was acceptable.

Demethylation of the ester with lithium propanethiolate

and lactonization brought us to the intended substrate for the

Cope rearrangement (135 → 136). That process does not

benefit from the factors that make the oxy-Cope rearrange-

ment especially favorable, and so we wondered if the strain

in the lactone would be sufficient to allow reaction under

reasonably mild conditions.

We heated the lactone in o-dichlorobenzene and were very

pleased to find that rearrangement does indeed occur

(Scheme 37, 136 → 137) and in good yield, although the

rate is lower than for the corresponding siloxy process.

Global deprotection with hydrochloric acid gave triol 138,

and that was subjected to Dess–Martin oxidation to afford

furan aldehyde 139. There are several reactions involved

here, of course. We assume that each of the hydroxyls at

C(14) and C(15) is independently converted into an alde-

hyde, and the remaining hydroxyl — either C(15) or C(14) —

forms a lactol, which, in turn, is dehydrated.

β,β′-disubstituted furan into an hydroxybutenolide under

these conditions seems to be general. Finally, we dissolve

anhydride 142 in aqueous sodium hydroxide, as before. The

base opens up the lactone function to expose an hydroxyl

group at C(10). That hydroxyl was converted into a ketone,

and, after acidic workup, we were able to isolate once more

the complete core of CP-225,917.

Conclusion

Next, we oxidized the aldehyde function to the corre-

sponding acid, using standard conditions (Scheme 38), and

we were very pleased to find that at the same time the furan

was converted into the isomeric hydroxybutenolides 140 and

141, each of which is equally suitable for the next step. That

involves oxidation to the anhydride. The conversion of a

In summary, we have developed two routes to the fully

oxygenated core of CP-225,917. One route is based on a

siloxy-Cope rearrangement and the other on a Cope rear-

rangement. Both processes are driven by release of strain in

the [2.2.1] bicyclic starting materials. The core of CP-

225,917 is crystalline, and X-ray analysis has provided

824

Can. J. Chem. Vol. 81, 2003

6. B.G. Cordiner, M.R. Vegar, and R.J. Wells. Tetrahedron Lett.

38, 2285 (1970).

7. For a recent review, see L.A. Paquette. Tetrahedron 53, 13 971

(1997).

8. L.A.Paquette, K.S. Learn, J.L. Romine, and H.-S. Lin. J. Am.

Chem. Soc. 110, 879 (1988).

structural details of this unusual system. Further work is be-

ing directed to the task of repeating one or both of our ap-

proaches with suitable substituents that can be elaborated

into the C₈side chains of the natural product.

Acknowledgments

9. Cf. (a) A.S. Cieplak. J. Am. Chem. Soc. 103, 4540 (1981);

(b) A.S. Cieplak, B.D. Tait, and C.R. Johnson. J. Am. Chem.

Soc. 111, 8447 (1989); (c) C.K. Cheung, L.T. Tseng, M.-H.

Lin, S. Srivastava, and W.J. le Noble. J. Am. Chem. Soc. 108,

1598 (1986); (d) G. Mehta and F.A. Khan. J. Am. Chem. Soc.

112, 6140 (1990); (e) G. Mehta, F.A. Khan, B. Ganguly, and J.

Chandrasekhar. J. Chem. Soc. Perkin Trans. 2, 2275 (1994);

(f) Y.-D. Wu, J.A. Tucker, and K.N. Houk. J. Am. Chem. Soc.

113, 5018 (1991).

10. (a) T. Onishi, Y. Fujita, and T. Nishida. Synthesis, 651 (1980)

and references therein; (b) Y. Fujita, T. Onishi, and T. Nishida.

Synthesis, 934 (1978); (c) M.A. Battiste, J.R. Rocca, and R.L.

Wydra. Tetrahedron Lett. 29, 6565 (1988); (d) A. Utagawa, H.

Hirota, S. Ohno, and T. Takahashi. Bull. Chem. Soc. Jpn. 61,

1207 (1988).

11. Use of DMF with O-methyl ethers: D. Backhaus and L.A.

Paquette. Tetrahedron Lett. 8, 29 (1997).

12. M.M. Bio and J.L. Leighton. J. Am. Chem. Soc. 121, 890

(1999).

13. C.R. Johnson and B.D. Tait. J. Org. Chem. 52, 281 (1987).

14. L. Syper, K. Kloc, J. Mlochowski, and Z. Szulc. Synthesis,

521 (1979).

15. (a) G. Green, W.P. Griffith, D.M. Hollinshead, S.V. Ley, and

M. Schröder. J. Chem. Soc. Perkin Trans 1, 681 (1984);

(b) M. Masakatsu and N. Watanabe. J. Org. Chem. 49, 3435

(1984).

We acknowledge financial support from the Natural Sci-

ences and Engineering Research Council of Canada

(NSERC) and Merck Frosst, and we thank Dr. V. Gagliardini

for the experiments with compounds 49 and 50.

References

1. T.T. Dabrah, T. Kaneko, W. Massefski, Jr., and E.B. Whipple.

J. Am. Chem. Soc. 119, 1594 (1997).

2. (a) P.W.M. Sgarbi and D.L.J. Clive. J. Chem. Soc. Chem.

Commun. 2157 (1997); (b) D.L.J. Clive, S. Sun, X. He, J.

Zhang, and V. Gagliardini. Tetrahedron Lett. 40, 4605 (1999);

(c) D.L.J. Clive and J. Zhang. Tetrahedron 55, 12 059 (1999);

(d) D.L.J. Clive, S. Sun, V. Gagliardini, and M.K. Sano. Tetra-

hedron Lett. 41, 6259 (2000); (e) D.L.J. Clive and S. Sun. Tet-

rahedron Lett. 42, 6267 (2001); (f) D.L.J. Clive and L. Ou.

Tetrahedron Lett. 43, 4559 (2002).

3. (a) K.C. Nicolaou, P.S. Baran, Y.-L. Zhong, H.-S. Choi, W.H.

Yoon, Y. He, and K.C. Fong. Angew. Chem. Int. Ed. Engl. 38,

1669 (1999); (b) K.C. Nicolaou, P.S. Baran, Y.-L. Zhong, K.C.

Fong, Y. He, W.H. Yoon, and H.-S. Choi. Angew. Chem. Int.

Ed. Engl. 38, 1676 (1999); (c) K.C. Nicolaou, J.-K. Jung,

W.H. Yoon, Y. He, Y.-L. Zhong, and P.S. Baran. Angew.

Chem. Int. Ed. Engl. 39, 1829 (2000); (d) K.C. Nicolaou, Y.-L.

Zhong, P.S. Baran, J. Jung, H.-S. Choi, and W.H. Yoon. J. Am.

Chem. Soc. 124, 2202 (2002).

16. (a) P. Flury and C.A. Grob. Helv. Chim. Acta 66, 1981 (1983);

(b) P. Kramp, G. Helmchen, and A.B. Holmes. J. Chem. Soc.

Chem. Commun. 551 (1993).

4. Q. Tan and S.J. Danishefsky. Angew. Chem. Int. Ed. Engl. 39,

4509 (2000).

17. (a) T. Hayashi, T. Fujiwa, Y. Okamoto, Y. Katsuro, and M.

Kumada. Synthesis, 1001 (1981); (b) A.S.E. Karlström, K.

Itami, and J.-E. Bäckvall. J. Org. Chem. 64, 1745 (1999).

Karlström, K. Itami, and J.-E. Bäckvall. J. Org. Chem. 64,

1745 (1999).

5. Synthesis of CP-263,114: ref. 3 and (a) N. Waizumi, T. Itoh,

and T. Fukuyama. J. Am. Chem. Soc. 122, 7825 (2000); (b) C.

Chen, M.E. Layton, S.M Sheehan, and M.D. Shair. J. Am.

Chem. Soc. 122, 7424 (2000); (c) Synthesis of naturally occur-

ring isomers of CP-molecules: D. Meng, Q. Tan, and S.J.

Danishefsky. Angew. Chem. Int. Ed. Engl. 38, 3197 (1999).

Article Doi

Synthetic studies related to CP-225,917

DOI: 10.1139/v03-086

Source and publish data:

Authors:

Article abstract of DOI:10.1139/v03-086

Full text of DOI:10.1139/v03-086

Products guided by the article

R&D Labs maybe for 112460-48-7

Relevant to this article

Hot Product