Total synthesis of (-)- and ent-(+)-vindoline and related alkaloids

Published on Web 07/22/2006

Total Synthesis of (-)- and ent-(+)-Vindoline and Related

Alkaloids

Hayato Ishikawa, Gregory I. Elliott, Juraj Velcicky,

Younggi Choi, and Dale L. Boger*

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

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

Received February 21, 2006; E-mail: boger@scripps.edu

Abstract: A concise 11-step total synthesis of (-)- and ent-(+)-vindoline (3) is detailed based on a unique

tandem intramolecular [4 + 2]/[3 + 2] cycloaddition cascade of a 1,3,4-oxadiazole inspired by the natural

product structure, in which three rings and four C-C bonds are formed central to the characteristic

pentacyclic ring system setting all six stereocenters and introducing essentially all the functionality found

in the natural product in a single step. As key elements of the scope and stereochemical features of the

reaction were defined, a series of related natural products of increasing complexity were prepared by total

synthesis including both enantiomers of minovine (4), 4-desacetoxy-6,7-dihydrovindorosine (5), 4-des-

acetoxyvindorosine (6), and vindorosine (7) as well as N-methylaspidospermidine (11). Subsequent

extensions of the approach provided both enantiomers of 6,7-dihydrovindoline (8), 4-desacetoxyvindoline

(9), and 4-desacetoxy-6,7-dihydrovindoline (10).

Introduction

Herein, we provide full details of the development of an

unusually concise total synthesis of (-)- and ent-(+)-vindoline^8-10

(3) in which the full pentacyclic skeleton complete with all

substituents and all six stereocenters is created in a single step

enlisting a tandem intramolecular [4 + 2]/[3 + 2] cycloaddition

cascade of a 1,3,4-oxadiazole that was inspired by the natural

product structures.⁸As detailed in ref 11, we extended the scope

of an intermolecular 1,3,4-oxadiazole [4 + 2]/[3 + 2] reaction

cascade beyond that which provides symmetrical 2:1 cycload-

ducts by implementing the reaction in an intramolecular

fashion.¹²The reaction cascade is initiated by a [4 + 2]

cycloaddition reaction of the 1,3,4-oxadiazole with a tethered

dienophile that, for 3, entailed the use of an electron-

Vinblastine (1) and vincristine (2) constitute the most widely

recognized members of a class of bisindole alkaloids as a result

of their clinical use as antitumor drugs (Figure 1).¹Originally

isolated in trace quantities from Cantharanthus roseus (L.) G.

Don,²their biological properties were among the first to be

shown to arise from inhibition of microtubule formation and

mitosis that today is still regarded as one of the more successful

drug targets for the treatment of cancer.³In addition to being

among the first natural products whose structures were deter-

mined by X-ray, they were also among the first for which X-ray

of a heavy atom derivative was used to establish their absolute

configuration.⁴Vindoline (3),^4,5a major alkaloid of Cantha-

ranthus roseus, constitutes the most complex half of vinblastine

and serves as both a biosynthetic²and synthetic⁶precursor to

the natural product.⁷

(8) Choi, Y.; Ishikawa, H.; Velcicky, J.; Elliott, G. I.; Miller, M. M.; Boger,

D. L. Org. Lett. 2005, 7, 4539.

(9) Racemic total syntheses: (a) Ando, M.; Bu¨chi, G.; Ohnuma, T. J. Am.

Chem. Soc. 1975, 97, 6880. (b) Kutney, J. P.; Bunzli-Trepp, U.; Chan, K.

K.; De Souza, J. P.; Fujise, Y.; Honda, T.; Katsube, J.; Klein, F. K.;

Leutwiler, A.; Morehead, S.; Rohr, M.; Worth, B. R. J. Am. Chem. Soc.

1978, 100, 4220. (c) Andriamialisoa, R. Z.; Langlois, N.; Langlois, Y. J.

Org. Chem. 1985, 50, 961. Formal racemic total syntheses: (d) Ban, Y.;

Sekine, Y.; Oishi, T. Tetrahedron Lett. 1978, 2, 151. (e) Takano, S.;

Shishido, K.; Sato, M.; Yuta, K.; Ogasawara, K. J. Chem. Soc., Chem.

Commun. 1978, 943. Takano, S.; Shishido, K.; Matsuzaka, J.; Sato, M.;

Ogasawara, K. Heterocycles 1979, 13, 307. (f) Danieli, B.; Lesma, G.;

Palmisano, G.; Riva, R. J. Chem. Soc., Chem. Commun. 1984, 909. Danieli,

B.; Lesma, G.; Palmisano, G.; Riva, R. J. Chem. Soc., Perkin Trans. 1

1987, 155. (g) Zhou, S.; Bommeziijn, S.; Murphy, J. A. Org. Lett. 2002,

4, 443.

(1) Neuss, N.; Neuss, M. N. In The Alkaloids; Brossi, A., Suffness, M., Eds.;

Academic: San Diego, 1990; Vol. 37, p 229.

(2) (a) Noble, R. L.; Beer, C. T.; Cutts, J. H. Ann. N.Y. Acad. Sci. 1958, 76,

882. (b) Noble, R. L. Lloydia 1964, 27, 280. (c) Svoboda, G. H.; Nuess,

N.; Gorman, M. J. Am. Pharm. Assoc. Sci. Ed. 1959, 48, 659.

(3) (a) Owellen, R. I.; Hartke, C. A.; Dickerson, R. M.; Haines, F. O. Cancer

Res. 1976, 36, 1499. (b) Pearce, H. L. In The Alkaloids; Brossi, A., Suffness,

M., Eds.; Academic: San Diego, 1990; Vol. 37, p 145.

(4) Moncrief, J. W.; Lipscomb, W. N. J. Am. Chem. Soc. 1965, 87, 4963.

(5) Gorman, M.; Neuss, N.; Biemann, K. J. Am. Chem. Soc. 1962, 84, 1058.

(6) (a) Mangeney, P.; Andriamialisoa, R. Z.; Langlois, N.; Langlois, Y.; Potier,

P. J. Am. Chem. Soc. 1979, 101, 2243. (b) Kutney, J. P.; Choi, L. S. L.;

Nakano, J.; Tsukamoto, H.; McHugh, M.; Boulet, C. A. Heterocycles 1988,

27, 1845. (c) Kuehne, M. E.; Matson, P. A.; Bornmann, W. G. J. Org.

Chem. 1991, 56, 513. (d) Magnus, P.; Mendoza, J. S.; Stamford, A.; Ladlow,

M.; Willis, P. J. Am. Chem. Soc. 1992, 114, 10232. (e) Yokoshima, S.;

Ueda, T.; Kobayashi, S.; Sato, A.; Kuboyama, T.; Tokuyama, H.;

Fukuyama, T. J. Am. Chem. Soc. 2002, 124, 2137. Kuboyama, T.;

Yokoshima, S.; Tokuyama, H.; Fukuyama, T. Proc. Natl. Acad. Sci. U.S.A.

2004, 101, 11966.

(10) Enantioselective total syntheses: (a) Feldman, P. L.; Rapoport, H. J. Am.

Chem. Soc. 1987, 109, 1603. (b) Kuehne, M. E.; Podhorez, D. E.; Mulamba,

T.; Bornmann, W. G. J. Org. Chem. 1987, 52, 347. (c) Kobayashi, S.;

Ueda, T.; Fukuyama, T. Synlett 2000, 883. Formal enantioselective total

syntheses: (d) Cardwell, K.; Hewitt, B.; Ladlow, M.; Magnus, P. J. Am.

Chem. Soc. 1988, 110, 2242.

(11) Elliott, G. I.; Fuchs, J. R.; Blagg, B. S. J.; Ishikawa, H.; Yuan, Z.-Q.; Tao,

H.; Boger, D. L. J. Am. Chem. Soc. 2006, 128, 10589.

(12) (a) Wilkie, G. D.; Elliott, G. I.; Blagg, B. S. J.; Wolkenberg, S. E.; Soenen,

D. R.; Miller, M. M.; Pollack, S.; Boger, D. L. J. Am. Chem. Soc. 2002,

124, 11292.

(7) Review: Kuehne, M. E.; Marko, I. In The Alkaloids; Brossi, A., Suffness,

M., Eds.; Academic: San Diego, 1990; Vol. 37, p 77.

9

10596

J. AM. CHEM. SOC. 2006, 128, 10596-10612

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Figure 2. Key cycloaddition cascade.

of minovine (4),¹⁴4-desacetoxy-6,7-dihydrovindorosine (5),

4-desacetoxyvindorosine (6), and vindorosine (7)¹⁵as well as

the extension of the approach to provide N-methylaspidosper-

midine (11) are also provided herein. As a consequence of

insights derived from these studies, an unusually concise

synthesis of 3 was implemented enlisting the oxadiazole

cycloaddition cascade to assemble the fully functionalized

pentacyclic ring system with formation of four C-C bonds and

three rings in a single step with direct introduction of all

substituents and setting all six stereocenters within the central

ring of vindoline including its three quaternary centers. Sub-

sequent extensions of these studies to provide 6,7-dihydrovin-

doline (8), 4-desacetoxyvindoline (9), and 4-desacetoxy-6,7-

dihydrovindoline (10) are also detailed herein.

Figure 1. Natural products and related structures.

Results and Discussion

rich enol ether whose reactivity and regioselectivity were

matched to react with the electron-deficient oxadiazole in an

inverse electron demand Diels-Alder reaction (Figure 2). Loss

of N₂from the initial cycloadduct provides a carbonyl ylide,

which undergoes a subsequent 1,3-dipolar cycloaddition with

a tethered indole.¹³For 3, the diene and dienophile substituents

complement and reinforce the [4 + 2] cycloaddition regio-

selectivity dictated by the linking tether, the intermediate 1,3-

dipole is stabilized by the complementary substitution at the

dipole termini, and the tethered dipolarophile (indole) comple-

ments the [3 + 2] cycloaddition regioselectivity that is set by

the linking tether. The relative stereochemistry of the cascade

cycloadduct is controlled by a combination of the dienophile

geometry and an exclusive endo indole [3 + 2] cycloaddition

dictated by the dipolarophile tether and sterically directed to

the face opposite the newly formed fused lactam.^11-13In the

course of these studies and as key issues regarding the scope

and stereochemical features of this new cycloaddition cascade

were defined, it was implemented in the total synthesis of a

series of related alkaloids, each bearing the characteristic

pentacyclic carbon skeleton of vindoline (3), but of increasing

complexity ultimately culminating with the total synthesis of 3

itself. Thus, full details of the total syntheses of both enantiomers

Minovine and 4-Desacetoxy-6,7-dihydrovindorosine. Mi-

novine (4) is a naturally occurring alkaloid isolated from Vinca

minor L.¹⁶and central to the biosynthetic pathway of many

related alkaloids bearing the pentacyclic vincadifformine skel-

eton. It is much simpler in structure than vindoline, and naturally

occurring 4 was reported as possessing an [R]_D) 0 in alcoholic

solvents,¹⁶an unusual issue that was addressed with our

preparation of optically active material. Prior reports of its

synthesis¹⁷provided racemic material, and no preceding studies

have addressed the optical purity or definitive absolute config-

uration assignment of this alkaloid. 4-Desacetoxy-6,7-dihy-

drovindorosine (5) lacks the 6,7-double bond and C4-acetoxy

(14) Yuan, Z.-Q.; Ishikawa, H.; Boger, D. L. Org. Lett. 2005, 7, 741.

(15) Elliott, G. I.; Velcicky, J.; Ishikawa, H.; Li, Y.; Boger, D. L. Angew. Chem.,

Int. Ed. 2006, 45, 620.

(16) (a) Mokry, J.; Dubravkova, L.; Sefcovic, P. Experientia 1962, 18, 564. (b)

Mokry, J.; Kompis, I.; Dubravkova, L.; Sefcovic, P. Experientia 1963, 19,

311. (c) Zachystalova, D.; Strouf, O.; Trojanek, J. Chem. Ind. 1963, 13,

610. (d) See also: Plat, M.; Le Men, J.; Janot, M.-M. Bull. Soc. Chim. Fr.

1962, 2237.

(17) (a) Kutney, J. P.; Chan, K. K.; Failli, A.; Fromson, J. M.; Gletsos, C.;

Nelson, V. R. J. Am. Chem. Soc. 1968, 90, 3891. (b) Ziegler, F. E.; Spitzner,

E. B. J. Am. Chem. Soc. 1973, 95, 7146; 1970, 92, 3492. (c) Kutney, J. P.;

Chan, K. K.; Failli, A.; Fromson, J. M.; Gletsos, C.; Leutwiler, A.; Nelson,

V. R.; de Souza, J. P. HelV. Chim. Acta 1975, 58, 1648. (d) Kuehne, M.

E.; Huebner, J. A.; Matsko, T. H. J. Org. Chem. 1979, 44, 2477. (e) Kalaus,

G.; Kiss, M.; Kajtar-Peredy, M.; Brlik, J.; Szabo, L.; Szantay, C.

Heterocycles 1985, 23, 2783.

(13) Padwa, A.; Price, A. T. J. Org. Chem. 1998, 63, 556; 1995, 60, 6258.

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10597

A R T I C L E S

Scheme 1

Ishikawa et al.

Scheme 2

diimide hydrochloride) and DMAP to provide the substrate 18

(87%) for the key [4 + 2]/[3 + 2] cycloaddition cascade

(Scheme 1). The tandem cycloaddition reaction proceeded in

excellent yield (74%) upon warming a solution of 18 at 180 °C

in o-dichlorobenzene for 24 h to provide 19 as a single

detectable diastereomer possessing the characteristic alkaloid

pentacyclic skeleton. The stereochemical assignments for 19

were first established by NMR through observation of diagnostic

¹H-¹H NMR NOEs (C5-Et/C14-H) and ultimately confirmed

through X-ray analysis of synthetic 5. The relative stereochem-

istry of 19 is established in the [3 + 2] cycloaddition reaction

and is derived from exclusive endo indole cycloaddition dictated

by the dipolarophile tether and sterically directed to the face of

the carbonyl ylide opposite the newly formed fused lactam.^11-13

Importantly, this sets the crucial relative stereochemistry key

to accessing 3-11 enlisting the oxadiazole cycloaddition

cascade.

Notably, as the dienophile becomes more highly substituted,

the initiating [4 + 2] cycloaddition of the reaction cascade

becomes progressively slower, and this has been described in

detail in ref 11. However, one unanticipated consequence of

this was observed while developing the approach to 4 and 5

that defined the importance of the acyl placement on the

dienophile tether for the preparation of vindoline. That is, in

all cases of a tethered indole, it participates in the reaction

cascade as the dipolarophile, and it has not been observed to

serve as the dienophile initiating the reaction cascade. However,

when we examined the substrate 24 bearing an unactivated and

disubstituted dienophile with the N-acyl group placed in the

dipolarophile tether, good conversions to the monocyclization

product 25 were observed, and only a trace of the tandem

cycloaddition product was detected regardless of the reaction

conditions (180-230 °C, 24-100 h), Scheme 2. Although this

has not yet been examined in detail, it indicates that indole

dienophile initiation of the reaction cascade is possible and that

in the presence of a tethered poor dipolarophile will likely result

only in intercepted products derived from the Diels-Alder

reaction. Notably, this contrasts the behavior of substrates in

which the tethered dienophile olefin is unsubstituted and

correspondingly much more reactive, where the cycloaddition

cascade with substrates bearing the indole acyl tether (e.g., 26)

proceed uneventfully with the alkene, not the indole, participat-

ing in the initiating [4 + 2] cycloaddition. Most importantly,

the observation with 24 clearly defined the N-acyl placement

in the dienophile tether for the di- and trisubstituted dienophiles

required of the alkaloids detailed herein.

group of vindorosine (7) as well as the additional C16 methoxy

group of vindoline (3). Although it has not been isolated as a

naturally occurring substance, it possesses nearly all the key

stereochemical issues that a synthesis of vindoline incorporates.

The starting 2-amino-1,3,4-oxadiazole 16 for the synthesis

of 4-7 was prepared from N-methyl tryptamine (12, Scheme

1). Treatment of 12 with carbonyldiimidazole afforded 13 (89%)

that was converted to the oxadiazole precursor 15 (93%) by

treatment with methyl oxalylhydrazide (14)¹⁸in the presence

of HOAc. Cyclization of 15 to form the corresponding oxadia-

zole 16 was mediated by TsCl and Et₃N (83%).

Coupling of 16 with 4-ethyl-4-pentenoic acid (17)¹⁹was

effected by EDCI (1-[3-(dimethylamino)propyl]-3-ethylcarbo-

(18) Christl, M.; Lanzendoerfer, U.; Groetsch, M. M.; Ditterich, E.; Hegmann,

J. Chem. Ber. 1990, 123, 2031.

(19) Reinheckel, H.; Sonnek, G.; Gensike, R. J. Prakt. Chem. 1975, 317, 273.

9

10598 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Scheme 3

dosperma alkaloids in which the lactam carbonyl is maintained,

as required of some natural products, or removed at a later stage

of the route as outlined in the following section. Significantly,

the simple conversion of 19 to 28 demonstrates that reductive

cleavage of the oxido bridge does not require prior lactam

reduction. The relative stereochemistry in 5 was first established

1

by observation of diagnostic H-¹H NOEs between the C5

ethyl, C6, and C14 protons confirming the stereochemical course

of the oxido bridge reductive ring opening reaction as well as

that of the cycloaddition cascade and was confirmed with a

single-crystal X-ray structure of 5.²³This clean control of all

the relative stereochemistry about the central six-membered ring

of 5 as well as the direct introduction of the C3 substituents

and stereochemistry established the viability of the approach

to vindoline. Treatment of 5 with the Burgess reagent²⁵in CH₃-

CN first provided the sulfamate 22 (91%) which upon isolation

and heating in toluene at 100 °C in the presence of NaH

furnished minovine (4) and its isomer 23. Although not

investigated in detail, attempts to promote a direct single pot

dehydration of 5 to provide 4 with the Burgess reagent were

not successful. Since these reactions proceed at least in part by

aziridinium formation via intramolecular sulfamate displacement

favoring formation of 23, no effort was made to alter or optimize

the 4:23 ratio. It is plausible that 23, whose exclusive formation

from 5 might be possible, could prove useful in the synthesis

of vindoline through syn dihydroxylation of the C3-C4 double

bond, and such intermediates have been utilized in past efforts.^9c

However, such an approach fails to capitalize on the intrinsic

C3 substitution and stereochemistry introduced in the oxadiazole

cycloaddition cascade and, as such, was not pursued.

A separation of the enantiomers of 19 (R ) 1.19) was carried

out on a semipreparative Daicel ChiralCel OD column (2 cm

× 25 cm, 10% i-PrOH-hexane, 10 mL/min flow rate) providing

natural-(+)-19 (t_R) 31.7 min) and ent-(-)-19 (t_R) 26.7 min).

The natural enantiomer series was initially assigned based on

comparison with 5 which has been independently prepared in

optically active form in prior studies²⁰and was unambiguously

established herein with an X-ray structure determination of the

natural enantiomer of 20 bearing a heavy atom (S). Treatment

of 19 with Lawesson’s reagent²¹furnished this thiolactam 20

in 85% yield (Scheme 1, natural enantiomer shown). Desulfu-

rization and opening of the oxido bridge was initially effected

by reduction of 20 with Ra-Ni to provide 21 (90%) followed

by diastereoselective reductive oxido bridge cleavage by

Pt-catalyzed hydrogenation¹³of the intermediate iminium ion

providing 4-desacetoxy-6,7-dihydrovindorosine (5, 73%).²⁰

Alternatively, S-methylation of the thiolactam 20 with Me₃OBF₄

followed by NaBH₄reduction²²in MeOH directly provided

4-desacetoxy-6,7-dihydrovindorosine [5, [R]²⁴+66 (c 0.10,

D

MeOH) vs lit²⁰[R]²⁴+30 (c 0.10, MeOH)] in superb yield

D

(92%) in a single operation (Scheme 3). The disparity in the

magnitude, but not sign, of the optical rotation of 5 raised

concerns over the accuracy of our initial assignment of the

absolute stereochemistry based on this comparison. However,

a subsequent X-ray of (+)-20 disclosed herein²³enlisting sulfur

as the heavy atom unambiguously confirmed the accuracy of

this assignment. The even more direct approach of treating the

lactam 19 with Meerwein’s salt (Me₃OBF₄, CH₂Cl₂) and

subsequent reduction (NaBH₄, MeOH) of the resulting methoxy

iminium salt²⁴also provided 5 but in lower conversions and in

a reaction characterized by several additional byproducts.

Finally, treatment of the lactam 19 with NaCNBH₃in HCl-

MeOH led to in situ acid-catalyzed oxido bridge cleavage and

clean diastereoselective reduction of the intermediate acyl

iminium ion to provide 28 in superb conversions (96%, eq 1).

Although not pursued for the preparation of 3-7, this conversion

permits the development of synthetic approaches to the Aspi-

Naturally occurring minovine (4) was reported with an [R]_D

) 0 in alcoholic solvents.¹⁶This raised the question of whether

the rotation of 4 was simply 0 or whether natural minovine might

suffer a facile racemization that could be envisioned to occur

by retro [4 + 2] cycloaddition providing an intermediate lacking

any stereocenters followed by a diastereoselective, but not

enantioselective, [4 + 2] cycloaddition. With optically active

minovine in hand, we found that it exhibited a remarkable

solvent dependent, but concentration independent, range of

optical rotations: natural 4 [R]²³-17 (c 0.35, CHCl₃), +16

D

(c 0.40, MeOH), and 0 ( 3 (c 0.28, EtOH). Moreover, the

enantiomeric integrity of 4 was maintained not only upon

prolonged storage at 25 °C but also upon deliberate attempts to

promote racemization via a reversible Diels-Alder reaction.

Thus, warming 4 in MeOH (80 °C, 24-48 h), toluene (120 °C,

21 h), or DMF (80-160 °C, 20 h) led to no evidence of

racemization, albeit with variable amounts of decomposition.

The potential racemization was easily monitored by chiral phase

HPLC where the two enantiomers of 4 were readily separable

(t_R) 10.0 min (natural 4) and 8.29 min (ent-4), ChiralCel OD

0.46 cm × 25 cm column, 0.5% i-PrOH-hexane, 1 mL/min

flow rate, R ) 1.21). Thus, naturally occurring minovine exhibits

(20) Optically active 5 has been reported: (a) Calabi, L.; Danieli, B.; Lesma,

G.; Palmisano, G. J. Chem. Soc., Perkin Trans. 1 1982, 1371. (b) Hugel,

G.; Levy, J. Tetrahedron 1984, 40, 1067.

(21) Yde, B.; Yousif, N. M.; Pedersen, U.; Thomsen, I.; Lawesson, S. O.

Tetrahedron 1984, 40, 2047.

(22) Raucher, S.; Klein, P. Tetrahedron Lett. 1980, 21, 4061.

(23) X-ray atomic coordinates for 5 (CCDC 231165), 20 (CCDC 295590), 43

(CCDC 295592), 51 (CCDC 295589), 54 (CCDC 295591), 80 (CCDC

253429), 81 (CCDC 253430), the C7 p-bromobenzoate of 82 (CCDC

255976), the C7 free alcohol of 88 (CCDC 253428), and 110 (CCDC

295588) have been deposited with the Cambridge Crystallographic Data

Center.

(25) (a) Burgess, E. M.; Penton, H. R., Jr.; Taylor, E. A. J. Org. Chem. 1973,

38, 26.

(24) Blowers, J. W.; Saxton, J. E.; Swanson, A. G. Tetrahedron 1986, 42, 6071.

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10599

A R T I C L E S

Ishikawa et al.

Scheme 4

Figure 3. Absolute configuration of natural products.

an unusual optical rotation, whose sign is dependent on the

choice of solvent. In EtOH, its value is 0 as reported,¹⁶but this

is not necessarily because it is racemic.

Although we established this unusual behavior of optically

active minovine, we have not yet been provided the opportunity

to compare our synthetic material with an authentic sample of

the natural product. Consequently, our assignment of the

“natural” and “unnatural” enantiomer series in the discussion

above rests with the correlation made by Ziegler^17benlisting

an ORD comparison of natural minovine with (-)-vincadif-

formine. It is possible that both enantiomers of 4, like those of

vincadifformine,²⁶may be found to occur naturally. Thus, while

the absolute configuration of our synthetic material is unam-

biguously established with the X-ray structures, the designation

of “natural” minovine in the discussion above isolated from

Vinca minor¹⁶is based on the Ziegler assignment^17band is

further supported by the observation that the same plant source

also produces (-)-vincadifformine^16d([R]_D-600 in EtOH) of

this same absolute configuration (Figure 3).²⁶

N-Methylaspidospermidine. The unfunctionalized penta-

cyclic skeleton of 3, aspidospermidine, has been the focus of a

beautiful series of racemic and enantioselective total syntheses

that have characterized efforts directed at this class of alka-

loids.^27,28Remarkably, it possesses a naturally occurring absolute

configuration opposite that found in vindoline and the vinca

alkaloids. A closely related minor alkaloid also identified in

this class is N-methylaspidospermidine (11) which, unlike

aspidospermidine itself, has not been prepared by total synthesis

and whose absolute configuration was assigned by correlation

of its ORD spectrum with that of (+)-aspidospermidine.^29b,30It

has been isolated not only from EVatamia peduncularis,^29c

Vallesia dichotoma,^29aand Aspidosperma quebracho blanco,^29d

but also Vinca minor L.^29bapparently with the same absolute

to our knowledge, the efforts would not only constitute the first

total synthesis^29cbut also serve to unambiguously confirm its

absolute configuration. Moreover, it represented an opportunity

to further explore and highlight the optional order of lactam

carbonyl reduction/oxido bridge opening following the cyclo-

addition cascade.

Thus, treatment of the resolved cycloaddition product 19 with

HCl (3 equiv) in the presence of NaCNBH₃(6 equiv, MeOH,

23 °C, 1.5 h) cleanly provided 28 in superb conversion (96%)

(Scheme 4, natural enantiomer shown). Subsequent treatment

of 28 with LiAlH₄(10 equiv, THF, reflux, 15 h, 99%) provided

29 derived from lactam reduction to the tertiary amine and

concurrent ester reduction to the corresponding alcohol. Initial,

alternative attempts to convert 19 directly to 29 in a single step

with LiAlH₄(15 equiv, THF, reflux, 16 h) provided 29 in lower

conversion (31%) and predominantly afforded products derived

from partial reduction. Given that the two-step reduction

sequence was so efficient (95-96%), this single step conversion

was not examined in further detail but likely could be optimized

to achieve a more satisfactory conversion. Cleavage of diol 29

with NaIO₄(6 equiv, acetone-H₂O, 25 °C, 3 h, 93%) provided

the sensitive ketone 30 in excellent conversion. Storage of 30

(0 °C, 3 d) or its subjection to a short series of classical methods

for reductive removal of a ketone led to significant competitive

decomposition. Consequently and without storage, 30 was

cleanly reduced with NaBH₄(10 equiv, EtOH, 25 °C, 1 h, 99%)

in a reaction that provides the stable axial alcohol 31 as a single

diastereomer derived from equatorial hydride delivery (C2-H;

d, J ) 2.5 Hz). Subsequent conversion of 31 to the methyl

dithiocarbonate 32 (2.5 equiv of LiHMDS, 20 equiv of CS₂,

THF, -10 °C, 4 h followed by 10 equiv of MeI, 25 °C, 12 h,

configuration ([R]²⁴+24 (c 1.25, CHCl₃)^29band +21

D

(CHCl₃)^29a). Thus, following the completion of the total

synthesis of vindoline detailed in a following section, we elected

to further highlight the applicability of 1,3,4-oxadiazole cy-

cloaddition cascade for the synthesis of such simplified Aspi-

dosperma alkaloids with N-methylaspidospermidine (11) where,

(26) For a fascinating discussion of this topic, see: Hesse, M. Alkaloids:

Nature’s Curse or Blessing?; Wiley-VCH: New York, 2002; pp 189-

194.

(27) (a) Saxton, J. E. In The Alkaloids; Cordell, G. A., Ed.; Academic: New

York, 1998; Vol. 51, Chapter 1. (b) Cordell, G. A. In The Alkaloids;

Manske, R. H. F., Rodrigo, R. G. A., Eds.; Academic: New York, 1979;

Vol. 17, Chapter 3.

(28) See: (a) Marino, J. P.; Rubio, M. B.; Cao, G.; de Dios, A. J. Am. Chem.

Soc. 2002, 124, 13398 and references cited therein. (b) Iyengar, R.;

Schildknegt, K.; Morton, M.; Aube, J. J. Org. Chem. 2005, 70, 10645 and

references cited therein.

(29) (a) Walser, A.; Djerassi, C. HelV. Chim. Acta 1965, 48, 391. (b) Mokry,

J.; Kompis, I.; Spiteller, G. Collect. Czech. Chem. Commun. 1967, 32, 2523.

(c) Isolation and partial synthesis from aspidospermidine: Zeches-Hanrot,

M.; Nuzillard, J.-M.; Richard, B.; Schaller, H.; Hadi, H. A.; Sevenet, T.;

LeMen-Olivier, L. Phytochem. 1995, 40, 587. (d) Biemann, K.; Spiteller-

Friedmann, M.; Spiteller, G. J. Am. Chem. Soc. 1963, 85, 631.

(30) Klyne, W.; Swan, R. J.; Bycroft, B. W.; Schumann, D.; Schmid, H. HelV.

Chim. Acta 1965, 48, 443.

9

10600 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Scheme 5

cleavage was established. Efforts to extend these observations

to substrates that permit access to the Aspidosperma alkaloids

have been disappointing to date, but for unexpected reasons. In

contrast to 33 which participates in the cycloaddition cascade

uneventfully, 37 bearing the tethered disubstituted terminal

alkene with the additional ethyl substituent proved problematic.

The best conversions to date (15-26%) fall far short of those

observed with 33 (or 18) and appear to stem from a sensitivity

of the cycloadduct 38 itself to the reaction conditions for the

time required for complete reaction (TIPB, 230 °C, 12 h or

o-Cl₂C₆H₄, 180 °C, 24 h).¹¹Although this has not been carefully

examined, such problems have not been encountered with 18

or other 1,3,4-oxadiazoles bearing a C5 methyl ester. Nonethe-

less and with samples of 38 in hand, its conversion to ketone

40 by acid-catalyzed (5 equiv of HCl) reductive oxido bridge

cleavage (6 equiv of NaCNBH₃, MeOH, 25 °C, 1 h) proved

viable illustrating the potential of such an approach.

4-Desacetoxyvindorosine. In addition to provisions to in-

corporate the C16 methoxy substituent unique to 3, introduction

of the C6-C7 double bond and the C4 acetoxy substituent

would be required to implement the approach for vindoline itself.

The latter was anticipated to arise from an appropriate substitu-

ent on the tethered dienophile,¹¹whereas the former would

require its introduction following the cycloaddition cascade.

Consequently, numerous approaches to the timing and method

for the introduction of the C6-C7 double bond were examined.

Whereas many of these are detailed in the following sections,

the approach that was employed enlisted an R-hydroxylation

of the cycloadduct lactam. Following lactam carbonyl excision

and reductive oxido bridge cleavage, a late stage regioselective

secondary alcohol elimination was anticipated to provide the

∆^6,7-alkene introduction. The latter elimination reaction was well

precedented having been used in the pioneering approach of

Kuehne³¹and later by Fukuyama,^10calbeit on early stage

synthetic intermediates rather than the penultimate intermediate.

In our hands, both these reactions required considerable

optimization over a period of years. The challenges associated

with the R-hydroxylation reaction were addressed while pursuing

vindorosine and vindoline and are accordingly discussed in the

following sections. The unanticipated challenges associated with

the final introduction of the ∆^6,7-alkene via the secondary alcohol

elimination were addressed with 46, providing a total synthesis

of 4-desacetoxyvindorosine (6, aka 2,3-dihydro-3-hydroxy-N-

methyltabersonine),³²a key late stage biosynthetic intermediate

enroute to vindorosine requiring only the final C4-hydroxylation

and subsequent O-acylation. Despite its natural origin as well

as its structural and biosynthetic relationship with vindorosine

(7), we are not aware of a total synthesis of this alkaloid.³³

Thus, as we faced unexpected difficulties in conducting the

final elimination reaction required for both vindorosine and

vindoline, we elected to examine this reaction in detail with

the more readily accessible substrate 46. R-Hydroxylation of

19 under conditions optimized for vindorosine and vindoline

(2 equiv of NaHMDS, 4 equiv of TMSOOTMS, -45 °C, 1 h,

77%) and its Barton-McCombie deoxygenation (Bu₃SnH, cat.

AIBN, toluene, reflux, 1 h, 96%) cleanly provided (+)- and

ent-(-)-N-methylaspidospermidine (11) for which the naturally

occurring enantiomer [lit^29b[R]²⁴+24 (c 1.25, CHCl₃)] was

D

established to possess the absolute configuration shown in

Scheme 4 [synthetic 11 depicted: [R]²⁵_D+26 (c 0.76, CHCl₃)].

Notably, N-methylaspidospermidine and such simplified

Aspidosperma alkaloids lack substitution at the C3 position that

the 1,3,4-oxadiazole cycloaddition cascade so purposely intro-

duces. We initially imagined that this functionality could be

usefully adjusted by enlisting a 1,3,4-oxadiazole bearing a nitrile

substituent which, upon reductive oxido bridge opening, would

release a cyanohydrin that in turn would collapse to a ketone.

This would not only serve to extend the cycloaddition scope

by permitting the direct introduction of alternative useful

functionality (a ketone) following oxido bridge cleavage, but it

would also allow direct access to the Aspidosperma alkaloids

by subsequent removal of the ketone. Consequently, we

examined the cycloaddition reactions of the 1,3,4-oxadiazoles

33 and 37 which bear a C5 nitrile versus methyl ester (Scheme

5). The reaction of 33 bearing the unsubstituted and unactivated

terminal alkene cleanly provided a single diastereomer of the

cascade cycloadduct 34 in good conversions (o-Cl₂C₆H₄, 180

°C, 75%)¹¹requiring the short reaction times (3 h) characteristic

of such substrates. Cleavage of the oxido bridge mediated by

acid catalysis in the presence or absence of NaCNBH₃provided

the stable cyanohydrin 35. Upon standing, 35 was found to

revert back to 34, whereas base treatment of 35 provided the

sensitive ketone 36. Thus, although the intermediate N-acyl

iminium ion was not reductively trapped under the reaction

conditions, the tactical release of ketone upon oxido bridge

(31) (a) Kuehne, M. E.; Okuniewicz, F. J.; Kirkemo, C. L.; Bohnert, J. C. J.

Org. Chem. 1982, 47, 1335. (b) Kuehne, M. E.; Podhorez, D. E. J. Org.

Chem. 1985, 50, 924.

(32) (a) De Carolis, E.; Chan, F.; Balsevich, J.; De Luca, V. Plant Physiol.

1990, 94, 1323. (b) De Luca, V.; Balsevich, J.; Tyler, R. T.; Eilert, U.;

Panchuk, B. D.; Kurz, W. G. W. J. Plant Physiol. 1986, 125, 147.

(33) Partial synthesis: Hugel, G.; Massiot, G.; Levy, J.; Le Men, J. Tetrahedron

1981, 37, 1369.

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10601

A R T I C L E S

Scheme 6

Ishikawa et al.

Scheme 7

competitive chloride nucleophile), by systematic alterations in

the solvent (CH₂Cl₂, ClCH₂CH₂Cl > CH₃CN > THF), with

adjustments to the reaction temperature (70-80 °C > 25 °C),

and with inclusion of a base (py > Et₃N, i-Pr₂NEt > collidine

> DABCO) to promote in situ elimination, the conversions

remained modest (50-55%), challenging to reproduce, and still

plagued by generation of additional byproducts. Moreover, the

conditions (1.3 equiv of Ph₃P, 1.3 equiv of CCl₄, 2.25 equiv of

py, CH₃CN, 70 °C, 2 h, 56% with 28% of a major byproduct;

1.3 equiv of Ph₃P, 10 equiv of CCl₄, 5 equiv of py, ClCH₂-

CH₂Cl, 80 °C, 3 h, 51% with 23% of a major byproduct) did

not immediately generalize to the more highly functionalized

substrates required for vindoline or vindorosine where the

conversions, albeit on smaller scales, were typically lower.

Nonetheless, the detailed study revealed the key features limiting

the reaction. A more effective approach to this elimination would

enlist reagents that do not liberate a nucleophile under the

reaction conditions and, optimally, would stoichiometrically

generate the base required to promote the elimination or

fragment the intermediate aziridinium ion. These requirements

are met by the Mitsunobu reagent³⁴Ph₃P-DEAD which first

activates the secondary alcohol for displacement, releases the

innoculus byproduct Ph₃PO upon elimination or transannular

aziridinium ion formation, and even supplies (generates) the

base (EtO₂CN^--NHCO₂Et) for the reaction. As such, the

conversion of 46 to 6 occurs in high yield under a wide range

of mild conditions even in the presence of excess reagent and

in the absence of added base when conducted enlisting Ph₃P-

DEAD.

51%) followed by silyl ether protection of the isolated alcohol

41 (2 equiv of TIPSOTf, 3 equiv of Et₃N, 25 °C, 0.5 h, 94%)

provided 42 (Scheme 6, only natural enantiomer shown). Like

the observations first made in the following studies, this provided

the equatorial C7 alcohol as the predominant, perhaps exclusive,

product. Although C7-H of 41 was obscured by other signals,

the stereochemistry resulting from the R-hydroxylation reaction

1

was clear upon H NMR analysis of 42 (C7-H; dd, J ) 11.7,

5.8 Hz) and unambiguously established by X-ray of 43.²³This

latter X-ray conducted on the natural enantiomer with a

derivative containing a heavy atom (S) was also of a quality

that supported, but did not unambiguously establish, the absolute

configuration assignment for 6. Thiolactam formation enlisting

Lawesson’s reagent²¹(90%), reductive desulfurization of 43 with

Ra-Ni (90%), diastereoselective reductive cleavage of the oxido

bridge in 44 upon catalytic hydrogenation (H₂, PtO₂, 96%),¹³

and silyl ether deprotection (Bu₄NF, 91%) provided 46.

Subsequent secondary alcohol activation and elimination (2

equiv of Ph₃P, 3 equiv of DEAD, THF, 25 °C, 24 h, 78%)

cleanly provided 6 identical in all respects with properties

reported for naturally derived material [lit³³[R]_D+48 (c 0.7,

23

MeOH); synthetic [R]_D+49 (c 0.3, MeOH)].

Initial efforts to promote the C7 alcohol elimination included

an exhaustive study of the procedures detailed by Kuehne³¹

(Ph₃P-CCl₄, Et₃N) and Fukuyama^10c(Ph₃P-CCl₄) and used

on simpler substrates. In general, the reaction was not easily

implemented and proved sensitive to contaminant moisture,

leading to regeneration of the starting alcohol potentially from

an intermediate aziridinium ion, and intolerant of excess reagent

that results in competitive in situ chloride displacement or

addition to the aziridinium ion (Scheme 7). As first alluded to

by Kuehne,³¹the chloride addition to a putative aziridinium ion

is potentially reversible and can provide two regioisomeric

products whose appearance and ratio are condition and tem-

perature dependent. Although improvements in the reaction were

implemented by ensuring anhydrous reaction conditions (no

water nucleophile to regenerate starting material), by limiting

the reagent stoichiometry (e1.3 equiv Ph₃P-CCl₄, limiting the

One final feature that we examined with 6, which impacts

the potential of developing an asymmetric synthesis, was the

prospect of incorporating the C7-hydroxy group into the

dienophile tether prior to the implementation of the cycloaddition

cascade (eq 2). Not only would this shorten the post cyclo-

addition synthetic sequence by two steps and avoid the

technically challenging lactam R-hydroxylation providing an

(34) (a) Iimori, T.; Ohtsuka, Y.; Oishi, T. Tetrahedron Lett. 1991, 32, 1209. (b)

Mitsunobu, O. Synthesis 1981, 1.

9

10602 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Scheme 8

exquisitely short synthesis, but the use of a chiral center in the

linking tether might be anticipated to define the absolute as well

as relative stereochemical outcome of the cycloaddition cascade.

Simple ground-state modeling of the potential cycloaddition

products, disregarding the oxadiazole endo/exo diastereoselec-

tion which is lost upon elimination of N₂, suggested that

incorporation of a substituent R to the carbonyl in the dienophile

tether could be expected to occupy an equatorial versus axial

position on the tethered six-membered ring adopting a chair-

like conformation in the [4 + 2] cycloaddition transition state.

Adoption of the alternative axial orientation appeared to be

especially destabilized since it entailed a developing OR/Et 1,3-

diaxial interaction (∆E ) 1.9-3.3 kcal/mol). Following an initial

trial examination of a substrate bearing an R-silyloxy substituent

in the dienophile tether which produced at least two cycload-

ducts, we elected to more carefully examine the effect of a

dienophile tether substituent with 49, 52, and 55, bearing an

R-methyl substituent.

The first such substrate examined, 49, bears the amide

carbonyl in the dienophile tether and was prepared by N-

acylation (EDCI, DMAP, CH₂Cl₂, 25 °C, 16 h, 81%) of 16 with

(S)-2-methyl-4-ethyl-4-pentenoic acid.³⁵Warming 49 in either

o-Cl₂C₆H₄(180 °C, 24 h, 5 mM) or TIPB (230 °C, 24 h, 5

mM) provided a disappointing mixture of 51 (53-44%) and

50 (29-32%) in which it was the minor product 50 that

possessed the desired relative and absolute configuration

(Scheme 8). The stereochemical assignment for 50 was estab-

lished by observation of diagnostic NOEs between C7-H/C5-

Et and C5-Et/C14-H and arises from a tethered chair-like

transition state with the methyl substituent adopting an equatorial

position in the fused six-membered ring (C5-Et/C7-Me are

trans in 50). The major product 51, whose structure was

unambiguously established by X-ray,²³possesses the correct

relative stereochemistry for the alkaloids, but the opposite

absolute stereochemistry required for vindoline (at C2, C3, C5,

C12) and the C5-Et/C7-Me substituents are unexpectedly cis.³⁶

Thus, the major product arises from a tethered boat-like

transition state for the fused six-membered ring. This could be

used to install the correct absolute as well as relative stereo-

chemistry simply by utilizing the enantiomer of 49, but the 1.3-

1.6:1 ratio of 51 to 50 indicates that the desired substituent-

induced control of the absolute configuration would be modest.

Thus, the facial selectivity induced by the linking chain

substituent in substrates such as 49 is modest resulting from

competitive transition states of the initiating [4 + 2] cycload-

dition reaction in which the dienophile linking chain adopts both

a chair- (minor) or boat-like (major) conformation (Figure 4).

Notably, the relative stereochemistry that is set in the subsequent

dipolar cycloaddition is unaffected by the tethered dienophile

substitution being directed to the face opposite the newly formed

fused lactam and confined to endo addition by the dipolarophile

tether. As a consequence of these observations, we also

examined the substrates 52 and 55 in which the linking carbonyl

was placed in the dipolarophile, not dienophile, tether (Scheme

8).³⁷Expectations were that this would now favor the adoption

of a chair-like conformation for the dienophile tether in the [4

+ 2] cycloaddition transition state by removing an imbedded

sp²center (Figure 4). Substrate 52, bearing an unsubstituted

terminal alkene as the dienophile provided a 1.7:1 mixture of

(35) Prepared by diastereoselective alkylation (NaHMDS, MeI, THF, -78 to

-48 °C, 79%) of an Evans optically active N-acyloxazolidinone ((S)-3-

(4-ethyl-4-pentenoyl)-4-phenyl-1,3-oxazolidin-2-one) followed by hydroly-

sis (LiOH, H₂O₂, THF-H₂O, 89%).

(36) Resubjecting the products to the reaction conditions (o-Cl₂C₆H₄, 180 °C

or TIPB, 230 °C, 24 h) did not result in epimerization of the C7 center

(Me substituent) or the detection of cycloreversion products derived from

a potentially reversible 1,3-dipolar cycloaddition.

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10603

A R T I C L E S

Ishikawa et al.

Scheme 9

Figure 4. Competing transition states.

53 and 54²³now favoring the expected product 53 in which the

dienophile linking chain adopts the chair-like conformation with

an equatorial methyl substituent that in turn sets the remaining

relative and absolute configuration. Beautifully, the X-ray

structures of 51²³and 54²³reflect these interpretations with the

final lactam of 51 adopting a boat-like conformation with the

C5 ethyl group occupying the axial “flagpole” position and the

cis C7 methyl group adopting a pseudoequatorial position,

whereas the saturated piperidine ring of 54 adopts a chair

conformation with the C7 methyl group adopting an axial

position cis to the axial C5 hydrogen substituent. Aside from

this modest diastereoselection with substrate 52 bearing a C5-H

(vs Et) substituent, the substrate 55 incorporating both the

dipolarophile amide tether and the less reactive disubstituted

dienophile required of the alkaloids expectedly provided

principally the product of an indole initiated [4 + 2] cycload-

dition (57, 48-72%) and a small amount of 56 (7-11%) along

with two other more minor unidentified products. Nonetheless,

56 possessed the requisite relative and absolute configuration

required of vindoline exhibiting diagnostic NOEs between C7-

H/C5-Et and C5-Et/C14-H and confirming the rationale for

the stereochemical observations.

methoxy substituent. As a consequence, it has been the subject

of a series of beautiful and historically important total synthe-

ses.³⁹In conjunction with the efforts on vindoline, we also

developed an unusually concise total synthesis of (-)- and ent-

(+)-vindorosine complementary to these past efforts.¹⁵This

focus on vindorosine arose principally as a result of the greater

accessibility and reduced electrophilic character of the precursor

indole that facilitated a detailed study of the key cycloaddition

reaction. The expectation was that the tethered dienophile would

now bear substitution that would permit the introduction of the

C4 acetoxy group, the remaining functionality necessary to

directly access vindoline through the [4 + 2]/[3 + 2] cyclo-

addition cascade. Thus, coupling of the 1,3,4-oxadiazole 16 with

either (Z)- or (E)-5-benzyloxy-4-ethyl-4-pentenoic acid (58)

provided the key cycloaddition precursors (Z)- or (E)-59 bearing

isomeric trisubstituted electron-rich enol ethers as the tethered

dienophiles (Scheme 9). As detailed in ref 11, trisubstitution of

Vindorosine. Vindorosine (7)³⁸is among the most complex,

highly functionalized, and stereochemically rich natural products

within the family of alkaloids isolated from the Madagascan

periwinkle (Catharanthus roseus (L.) G. Don). It is identical

in structure to vindoline with the exception that it lacks the C16

(37) Without optimization, substrates 52 and 55 were prepared by treatment of

(S)-1-amino-2-methyl-4-butene or (S)-1-amino-4-ethyl-2-methyl-4-butene

with carbonyldiimidazole (85% or 98%) followed by methyl oxalylhy-

drazide (14, 1 equiv, 1 equiv of HOAc, THF, 40 °C, 18 h, >95% or 91%),

subsequent dehydration to the 1,3,4-oxadiazole (1 equiv of TsCl, 2.5 equiv

of Et₃N, CH₂Cl₂, 25 °C, 16 h, 77% or 67%), and N-acylation with 2-(N-

methylindol-3-yl)acetic acid via its mixed anhydride with Me₃CCOCl (1

equiv, 1 equiv of Et₃N, THF, 0 °C, 30 min) employing the lithium anion

of the amino-1,3,4-oxadiazole (1 equiv of n-BuLi; THF, -78 °C, 1.5 h,

46% or 70%).

(39) (a) Buchi, G.; Matsumoto, K. E.; Nishimura, H. J. Am. Chem. Soc. 1971,

93, 3299. (b) Kuehne, M. E.; Podhorez, D. E.; Mulamba, T.; Bornmann,

W. G. J. Org. Chem. 1987, 52, 347. (c) Formal syntheses: Takano, S.;

Shishido, K.; Sato, M.; Ogasawara, K. Heterocycles 1977, 6, 1699. (d)

Veenstra, S. J.; Speckamp, W. N. J. Am. Chem. Soc. 1981, 103, 1645. (e)

Natsume, M.; Utsunomiya, I. Chem. Pharm. Bull. 1984, 32, 2477. (f)

Andriamialisoa, R. Z.; Langlois, N.; Langlois, Y. J. Org. Chem. 1985, 50,

961. (g) Winkler, J. D.; Scott, R. D.; Williard, P. G. J. Am. Chem. Soc.

1990, 112, 8971. (h) Heurenx, N.; Wouters, J.; Marko, I. E. Org. Lett.

2005, 7, 5245.

(38) Mosa, B. K.; Trojanek, J. Collect. Czech. Chem. Commun. 1963, 28, 1427.

9

10604 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Table 1. Cycloaddition Concentration Dependence (59)

concentration on the cascade cycloaddition of both (Z)- and (E)-

59 was retrospectively and systematically examined (Table 1).

Both were found to exhibit a significant dependence on the

reaction concentration with that of (Z)-59 being much more

pronounced suggesting that a competitive bimolecular reaction

may compete with the intramolecular cycloaddition cascade at

the higher reaction concentrations.

% yield

60 or 61

compound

concn (mM)

(Z)-59^a

5

1

0.75

0.50

0.25

0.10

30

39

48

55

63

78

More significantly, we observed that the [4 + 2] cycloaddition

of (Z)-59 is now the fast step in the reaction cascade and that

the subsequent [3 + 2] cycloaddition is the slow step, a reversal

of what is observed with (E)-59 and other typical substrates.¹¹

The net consequence of this observation is that longer reaction

times and/or more vigorous reaction conditions, in conjunction

with the use of dilute reaction concentrations, led to substantial

improvements in the conversion of (Z)-59 to 60. Thus, premature

workup of the cycloaddition reaction resulting from shorter

reaction times or milder reaction conditions provided products

(62; see Supporting Information) derived from a quench of the

intermediate 1,3-dipole or the corresponding cyclobutene ep-

oxide. Notably, the analogous cycloaddition cascade of (E)-59

proceeds not only under milder conditions (180 °C, 24 h) but

also with little or no detection of products derived from quench

of the intermediate 1,3-dipole if the reaction is prematurely

stopped. Although such observations were first misinterpreted

to represent an instability of the intermediate 1,3-dipole derived

from (Z)-59 under the reaction conditions resulting in a search

for milder (not more vigorous) conditions, we ultimately found

that simply extending the reaction times and raising the reaction

temperatures increased the conversions to 60 and the intermedi-

ate-derived byproducts diminished or disappeared altogether.

The origin of this distinction in the slow step of the cycloaddition

cascade of (Z)- versus (E)-59 was not clear. In fact, the 1,3-

dipolar cycloaddition transition state for (E)-59 would appear

to embody the serious steric interactions disfavoring the

observed indole endo approach on the R-face of the 1,3-dipole;

yet it progresses with a greater facility than that of (Z)-59

(Scheme 9). It is possible that the preferred stereochemistry of

the corresponding cyclobutene epoxides or their relative stability,

a potential intermediate and reversible source of the 1,3-dipole,

may dictate the relative ease of the 1,3-dipolar cycloaddition.

However, in examining the relative transition states for the 1,3-

dipolar cycloadditions, it is also possible that (Z)-59 may suffer

a destabilizing electrostatic interaction of its central oxygen with

the OBn substituent that decelerates the reaction or that the OBn

substituent of (E)-59 stabilizes its transition state (transition state

anomeric effect). We are unaware of any studies that might have

previously distinguished such effects, and their magnitude is

surprising.

(E)-59^b

(E)-59^d

50

10

5

1

100

40

47

56 (50-86)^c

76

33

(E)-59^d

50

10

5

52

59

64

1

>95

^a230 °C, TIPB, 60 h. ^b180 °C, o-Cl₂C₆H₄, 24 h. ^cPreparative scale,

multiple runs. ^d230 °C, TIPB, 18 h.

the dienophile typically precludes [4 + 2] cycloaddition

initiation of the reaction cascade. The exception to this

generalization is the use of trisubstituted olefins bearing an

electron-donating substituent to activate the tethered dienophile

for participation in an inverse electron demand Diels-Alder

reaction with the electron-deficient 1,3,4-oxadiazole. Thus, the

electron-rich dienophiles of (Z)- and (E)-59 were ideally suited

for initiation of the cycloaddition cascade enhancing the intrinsic

reactivity of the dienophile, reinforcing the [4 + 2] cycloaddition

regioselectivity dictated by the dienophile tether, and directly

introducing the C4 alkoxy substituent. The distinction in the

two substrates is that (Z)-59 permits the direct introduction of

the naturally occurring C4 acetate â-stereochemistry, whereas

(E)-59 provides the C4 isomer requiring a subsequent inversion

of configuration at this center. Contrary to expectations, the

substrate (Z)-59, which directly provides the preferred cycload-

duct 60 for use in the synthesis of 7, proved to be the more

difficult to implement. As such, these two cycloaddition

cascades were examined in detail. Cyclization of (E)-59

proceeded especially effectively providing a single detectable

diastereomer 61 in superb conversions (86%, o-Cl₂C₆H₄, 24 h),

Scheme 9. In an early survey of this reaction, the conversions

were more modest (30-40%) when conducted at conventional

reaction concentrations (0.2-0.1 M) but increased significantly

when the concentration was reduced to 0.005 M. This initial

observation remained unexplored for a considerable period of

time as the interest in promoting the tandem cycloaddition

cascade of the preferred substrate (Z)-59 under these conditions

was pursued. However, upon revisiting this impact of the

reaction concentration, yields of the cycloadduct 61 from (E)-

59 were found to exceed 95% (TIPB, 230 °C, 18 h, Table 1)

upon further dilution. By contrast, the tandem cycloaddition of

(Z)-59 proved more challenging to implement. After consider-

able exploration as detailed below, (Z)-59 was found to provide

the desired cycloadduct 60 in good yield (78%) as a single

detectable diastereomer when warmed in triisopropylbenzene

(TIPB, 230 °C, 60 h). Initial attempts to promote the cyclo-

addition of (Z)-59 provided 60 in modest conversion (e30-

40% at g5 mM) when conducted under the conditions initially

adopted for (E)-59. However, further dilution of the reaction

led to improved conversions ultimately providing 60 in yields

as high as 78%. As a result, this impact of the reaction

To probe the origin of this effect, (E)- and (Z)-63 bearing a

terminal methyl substituent on the tethered dienophile in place

of the benzyloxy substituent were examined (Scheme 10). The

expectation is that if the electronic nature of the benzyloxy

substituent was responsible for the differences observed in the

slow step of the cycloaddition cascade with (Z)-59 ([3 + 2]

cycloaddition) versus (E)-59 ([4 + 2] cycloaddition), these

differences would disappear in the comparisons of (Z)- and (E)-

63. Thus, the methyl versus benzyloxy substituents would be

expected to mirror or perhaps enhance the steric effects on the

[3 + 2] cycloaddition of the reaction cascade but remove the

electronic effects influencing the relative rate of the [3 + 2]

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10605

A R T I C L E S

Scheme 10

Ishikawa et al.

cycloaddition. Without optimization and much more consistent

with expectations, both (Z)- and (E)-63 participated effectively

in the cycloaddition cascade and did so at essentially identical

rates without detection of products derived from the intermediate

1,3-dipole ([3 + 2] faster than [4 + 2]), Scheme 10. In fact,

when the reaction was followed by ¹H NMR in o-Cl₂C₆D₄, (Z)-

63 participated in the cycloaddition cascade at what might be a

slightly greater rate and with what appears to be cleaner

conversions to the expected product consistent with its less

sterically encumbered [3 + 2] cycloaddition transition state.

Diagnostic NOEs between NMe/C4-H, C4-H/C5-H, and

C5-H/C14-H observed with 64 derived from (Z)-63 and

between N-Me/C4-Me, C4-Me/C5-H, and C5-H/C14-H

for 65 derived from (E)-63 confirmed the stereochemical

assignments of the cycloadducts.

These observations established that it is the electronic

character of the benzyloxy substituent of (Z)-59 that is decel-

erating the typically fast 1,3-dipolar cycloaddition. We attribute

this deceleration of the 1,3-dipolar cycloaddition to a destabiliz-

ing electrostatic interaction of the benzyloxy substituent with

the central oxygen of the carbonyl ylide present only in the [3

+ 2] cycloaddition transition state for (Z)-59 that is absent with

(E)-59 (Scheme 9). An additional, more provocative, explanation

rests with the preferred stereochemistry of the potential inter-

mediate cyclobutene epoxides. Reversible ring closure to the

cyclobutene epoxides by the intermediate 1,3-dipoles derived

from (Z)- and (E)-59, which could serve as a reservoir for the

reactive carbonyl ylides, may well possess preferred epoxide

stereochemistries that influence the ease of the tethered 1,3-

dipolar cycloaddition. Thus, the cyclobutene epoxide derived

from (E)-59 might be expected to possess the â-stereochemistry

indicated in Figure 5 placing the epoxide on the cyclobutane

face opposite the ethyl and benzyloxy substituents and opposite

the incoming indole dipolarophile. This not only allows direct

unfettered access of the indole dipolarophile to the endo face

of the 1,3-dipole as it is reversibly generated but also represents

a strained trans-fused 6,4-ring system destabilizing the cy-

clobutene epoxide. In contrast, the most stable cyclobutene

epoxide derived from (Z)-59 places the epoxide on the R-face,

trans to the electronegative benzyloxy substituent within a more

Figure 5. Cyclobutene epoxides as potential intermediates.

stable cis-fused 6,4-ring system. This not only places the epoxide

on the same face as the approaching indole dipolarophile

potentially blocking its access but also represents a more stable

cyclobutene epoxide. Either, or perhaps both features, may

contribute to the slower [3 + 2] cycloaddition observed with

(Z)-59.

The enantiomers of cycloadduct 60 proved easily separable

by chiral phase chromatography (ChiralCel OD column, 2 cm

× 25 cm, 20% i-PrOH-hexane, 10 mL/min, t_R) 15.2 (natural)

and 21.3 min, R ) 1.40) providing access to either enantiomer

on a preparatively useful scale (150 mg/injection). R-Hydroxyl-

ation of 60 was achieved by treatment of the lactam enolate

(generated with LDA) with TMSO-OTMS⁴⁰and was followed

by a direct quench with triisopropylsilyltriflate (TIPSOTf)

providing silyl ether 69 (60%), Scheme 11 (only natural

enantiomer shown). The R-hydroxylation proceeded in a stereo-

selective manner providing 69 as a single diastereomer bearing

an equatorial substituent (C7-H; dd, J ) 12.3, 6.2 Hz) resulting

from approach of the electrophile from the least hindered convex

face of the enolate. Interestingly, this reaction performed best

(g60%) when the TMSO-OTMS was premixed with the lactam

60 and subsequently treated with base (NaHMDS or LDA) at

low temperature (-78 °C) and provided lower conversions (ca.

20%) if the lactam enolate was first generated and then treated

with the reagent. Typically, LDA provided the highest conver-

(40) (a) Taddei, M.; Ricci, A. Synthesis 1986, 633. (b) Florio, S.; Troisi, L.

Tetrahedron Lett. 1989, 30, 3721.

9

10606 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Scheme 11

Figure 6. Additional tethered dienophiles examined.

Scheme 12

sions if the substrate contained a C4 substituent (e.g., 60, 80,

and 81) but was generally less satisfactory than NaHMDS if

the substrate did not contain a C4 substituent (e.g., 19 or 109).

In these latter instances, amidation of the C3 methyl ester was

observed with the less hindered amide base (LDA) under the

conditions used to drive the reaction to completion. Several

alternative and more conventional R-hydroxylation procedures

were also examined including the use of the Davis oxaziridines

(LDA or NaHMDS, (+)- or (-)-CSO, 30-50%)⁴¹and its many

modifications ((+)-Cl₂CSO, DPO),^42,43O₂,⁴⁴PhNO,⁴⁵or

MoOPH,⁴⁶and each proved less satisfactory, providing sub-

stantial amounts of recovered lactam or, if forced, complete

consumption of starting material without improvements in the

conversions.

60 with alternative electrophiles more commonly enlisted to

provide an unsaturated lactam. Such reactions proceed more

effectively than the R-hydroxylation of 60 and are easily

transformed to the unsaturated lactam (Scheme 12). However,

difficulties arise in the order and manner in which the subsequent

transformations are conducted. Preliminary attempts at lactam

or thiolactam reduction to the tertiary amine without reduction

or migration of the ∆^6,7-alkene did not appear straightforward,

oxido bridge cleavage in the presence of the unmodified

unsaturated lactam resulted in a subsequent conjugate addition

of the liberated C3 tertiary alcohol to provide an extraordinarily

stable ether, and the requisite C4 benzyl ether deprotection and

reductive oxido bridge cleavage were relegated to methods (not

a combined Ra-Ni reduction) that would not competitively

reduce the ∆^6,7-alkene. It was these observations and consid-

erations that led to our commitment to the approach detailed in

Scheme 11 requiring development of satisfactory solutions to

the R-hydroxylation of 60 as well as the late stage regioselective

C7 alcohol elimination.

Finally, we also examined a range of functionalized tethered

dienophiles that potentially could have provided an alternative,

albeit less direct, introduction of the C4 alkoxy group (Figure

6). Notably, none of these performed as effectively as (Z)- or

(E)-59 and were not pursued beyond an initial examination.

Vindoline. The preceding studies set the stage for imple-

mentation of a remarkably concise 11-step total synthesis of

vindoline (3). Treatment of N-methyl-6-methoxytryptamine

(75)⁴⁷with carbonyldiimidazole (CDI, 90%) followed by

The amide 69 was converted to thioamide 70 (93%) with

Lawesson’s reagent.²¹Reduction of the thioamide with Ra-Ni

occurred cleanly in 2 h, but the reaction was allowed to proceed

longer (16 h) leading to subsequent cleavage of the benzyl ether

providing the alcohol 71 (80%) directly in a single step.

Acetylation of secondary alcohol 71 with acetic anhydride

(97%), followed by a diastereoselective reductive oxido bridge

opening of 72 (93%, H₂, PtO₂),¹³and subsequent deprotection

of the TIPS ether 73 with Bu₄NF afforded diol 74 (99%). A

final regioselective elimination of the secondary alcohol 74 using

the newly developed conditions enlisting Ph₃P-DEAD (THF,

23 °C, 24 h, 74%) cleanly provided (-)- and ent-(+)-

vindorosine (7).

It is possible that an earlier stage introduction of the ∆^6,7

-

alkene that would avoid the late stage C7 alcohol elimination

could be implemented in a synthesis of vindorosine, and this

would simply entail reactions of the lactam enolate derived from

(41) Davis, F. A.; Chen, B.-C. Chem. ReV. 1992, 92, 919.

(42) Davis, F. A.; Weismiller, M. C. J. Org. Chem. 1990, 55, 3715.

(43) (a) Boyd, D. R.; Jennings, W. B.; McGuckin, R. M.; Rutherford, M.; Saket,

B. M. J. Chem. Soc., Chem. Commun. 1985, 582. (b) Jennings, W. B.;

Schweppe, A.; Testa, L. M.; Zaballos-Garcia, E.; Sepulveda-Arques, J.

Synlett 2003, 121.

(44) Wasserman, H. H.; Lipshutz, B. H. Tetrahedron Lett. 1975, 16, 1731.

(45) Nomiyama, N.; Yamamoto, H. J. Am. Chem. Soc. 2003, 125, 6038.

(46) Vedejs, E. J. Am. Chem. Soc. 1974, 96, 5944.

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10607

A R T I C L E S

Scheme 13

Ishikawa et al.

Table 2. Cycloaddition Concentration Dependence (79)

% yield

compound

concn (mM)

80 or 81

(Z)-79^a

5

1

0.1

0.05

27-34

43

50

53

(E)-79^b

100

50

10

5

38

58

86

90

99

1

^a230 °C, TIPB, 90 h. ^b230 °C, TIPB, 20 h.

Scheme 14

cm × 25 cm) with a remarkable efficiency (R ) 1.70, t_R

)

15.1 (natural) and 25.6 min, 10 mL/min), providing access to

either enantiomer on a preparatively useful scale (200 mg/

injection). R-Hydroxylation of 80 was achieved best by treat-

ment of the lactam enolate (LDA) with TMSO-OTMS⁴⁰and

was followed by a direct quench with TIPSOTf providing 82

(64%), Scheme 14 (natural enantiomer shown). More common

R-hydroxylation methods including the use of the Davis

oxaziridines ((+)-CSO)⁴¹or MoOPH⁴⁶were not nearly as

effective (30-50%). The stereochemistry of the R-hydroxylation

subsequent reaction of 76 with methyl oxalylhydrazide¹⁸(14)

furnished 77 (79%), Scheme 13. Closure of 77 to the 1,3,4-

oxadiazole 78 (1.0 equiv of TsCl, 2.5 equiv of Et₃N, CH₂Cl₂,

23 °C, 81%) and subsequent N-acylation with isomerically pure

(Z)- or (E)-58 provided the key cycloaddition substrates (Z)-79

(96%) and (E)-79 (92%) now bearing the tethered 6-methoxy-

indole dipolarophile and the stereochemically defined electron-

rich dienophiles. Both (Z)- and (E)-79 underwent the key tandem

[4 + 2]/[3 + 2] cycloaddition cascade to give the pentacyclic

products 80 (53%) and 81 (>95%), respectively, as single

diastereomers whose structures were assigned initially based

on their ¹H NMR spectroscopic properties and later unambigu-

ously established with single-crystal X-ray structures.²³Con-

sistent with the concurrent observations made in the approach

to vindorosine detailed in the preceding section, the reaction of

(E)-79 proceeds more rapidly than (Z)-79, the slow step for the

cascade cycloaddition of the former is the initiating Diels-Alder

reaction whereas that of the latter is the [3 + 2] cycloaddition,

and both exhibit a significant concentration effect (Table 2) that

is most pronounced with (Z)-79.

1

was clear from the H NMR spectrum of 82 (C7-H; dd, J )

5.5, 11.7 Hz) and unambiguously determined with a X-ray

structure determination of the corresponding p-bromobenzoate.²³

Conversion of 82 to thioamide 83 (70%) with Lawesson’s

reagent, reductive desulfurization with Ra-Ni conducted under

conditions (THF, 15 h, 23 °C) that also served to cleave the

benzyl ether provided 84 (91%), and subsequent acetylation of

the resulting secondary alcohol afforded 85 (97%). Diastereo-

selective reductive cleavage of the oxido bridge upon catalytic

hydrogenation (45 psi H₂, PtO₂, 50% MeOH-EtOAc) with

reduction of the intermediate imminium ion from the R-face

provided 86 (98%). Silyl ether cleavage (Bu₄NF, THF, 89%)

and subsequent secondary alcohol activation and regioselective

elimination of 87 (Ph₃P, DEAD, THF, 23 °C) provided either

(-)- or ent-(+)-vindoline (75%) identical in all respects with

authentic material [synthetic natural vindoline [R]²⁰-17 (c

Although unanticipated, the enantiomers of 80 could be easily

separated on a Chiralcel OD column (30% i-PrOH-hexanes, 2

D

0.40, CHCl₃), lit.³⁸[R]²⁰_D-18 (CHCl₃)]. The final conversion

of 87 to 3 enlisting the newly implemented Mitsunobu activation

of the secondary alcohol for elimination proved much more

effective and reproducible than alternative procedures that enlist

(47) Takano, S.; Shishido, K.; Matsuzaka, J.; Sato, M.; Ogasawara, K.

Heterocycles 1979, 13, 307. An improved preparation is detailed in the

Supporting Information.

9

10608 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Scheme 15

Scheme 16

6,7-Dihydrovindoline. Although not a natural product itself,

6,7-dihydrovindoline (8) constitutes a historically important,

characterized derivative of vindoline that assisted in the original

structure determination.⁵Moreover, similar semisynthetic di-

hydro derivatives of vinblastine and its analogues were explored

in early studies, albeit with conclusions that they are active but

less efficacious in vivo.⁴⁹Today this conclusion would be worth

revisiting given the apparent lack of a direct role that the ∆^6,7

-

alkene appears to have on the interaction of vinblastine with

its biological target.⁵⁰Anticipating that the approach detailed

herein may be utilized to provide synthetic vinblastine analogues

containing deep-seated changes in the vindoline subunit, we

elected to extend the studies to a synthesis of 8. Notably,

removing the requirement for introduction of the ∆^6,7-alkene

eliminates 3-4 steps from the route and provides an extra-

ordinarily short synthesis of such vindoline analogues.

reagents (e.g., Ph₃P-CCl₄)^10,31that generate reactive nucleo-

philes (e.g., Cl^-) that can compete with the elimination reaction.

A second total synthesis of vindoline was accomplished

utilizing cycloadduct 81 and was conducted while efforts to

secure 80 were underway (Scheme 15). The advantage is that

the key cycloaddition cascade proceeds with greater facility

under milder conditions and in higher conversions (>95%),

albeit in a route requiring inversion of the C4 stereochemistry.

Thus, R-hydroxylation of 81 (LDA, TMSOOTMS) followed by

in situ treatment with TIPSOTf provided 88 in superb yield

(75%). The stereochemistry of the intermediate alcohol, also

obtained by a less effective reaction of the lactam enolate with

Consequently, a remarkably concise synthesis of 6,7-di-

hydrovindoline (8) was developed enlisting the cascade

cycloadduct 80. Thus, reductive removal of the O-benzyl ether

(H₂, Pd(OH)₂, MeOH, 23 °C, 24 h, 85%) and O-acetylation of

the newly liberated secondary alcohol 97 (Ac₂O, pyr, 23 °C,

16 h, 88%) provided 98 (Scheme 16, natural enantiomer shown).

Treatment of 98 with Lawesson’s reagent (toluene, 90 °C, 2 h,

97%) followed by reductive removal of the thiolactam with Ra-

Ni (THF, 23 °C, 2 h) and a final reductive oxido bridge cleavage

of 100 (H₂, PtO₂, 1:1 MeOH-EtOAc, 65% for two steps)

provided 6,7-dihydrovindoline [synthetic natural enantiomer of

1

Davis’ reagent⁴¹((+)-CSO, 40-50%), was established by H

NMR (C7-H; dd, J ) 5.5, 12.1 Hz) and confirmed by X-ray.²³

C4-Alcohol deprotection (H₂, Pd/C, 90%) and oxidation of 89

with TPAP/NMO (81%) or Dess-Martin periodinane (1.2

equiv, pyr-CH₂Cl₂, 0 °C, 30 min, 86%) provided 90. Thiolac-

tam formation (Lawesson’s reagent, 81%) and Ra-Ni desulfu-

rization of 91 (THF, 25 °C, 78%) provided 92. The alternative

sequence of converting the amide 88 to the corresponding

thioamide with Lawesson’s reagent (toluene, reflux, 60%) and

its reductive removal with Ra-Ni concurrent with O-debenzyl-

ation (THF, reflux, 10 h, 60-80%) was followed by a more

problematic oxidation (TPAP/NMO; SO₃-Py, Et₃N, DMSO)

of the secondary alcohol to ketone 92 in the presence of the

newly liberated tertiary amine (N8). Diastereoselective reductive

oxido bridge cleavage of 92 (H₂, PtO₂, MeOH, H⁺)¹³furnished

93 (82%), TIPS ether cleavage (Bu₄NF, THF, 98%), and

treatment of 94 with Ph₃P-DEAD led to activation and

elimination of the C7 alcohol to provide 95 (62%). Following

the protocol of Buchi,^9adiastereoselective C4 carbonyl reduction

of 95 (Redal-H, AlCl₃) and subsequent acetylation (Ac₂O, 83%)

of deacetylvindoline (96) furnished vindoline (3). Notably, this

latter route proceeded through desacetylvindoline (96), which

itself is a naturally occurring alkaloid being the penultimate

biosynthetic intermediate to vindoline.⁴⁸

8, [R]²³+37 (c 0.75, CHCl₃) and +12 (c 0.4, MeOH)].

D

Authentic 8, ([R]²³+37 (c 0.75, CHCl₃) and +13 (c 0.70,

D

MeOH)) prepared by hydrogenation of a sample of naturally

occurring vindoline (H₂, Pd/C, MeOH, 25 °C, 20 h, 98%),

proved indistinguishable from synthetic (+)-8.

An alternative to this approach was also developed that

employed the more accessible cycloadduct 81 derived from (E)-

79. Thus, O-debenzylation of 81 (H₂, Pd(OH)₂, MeOH, 23 °C,

50 psi, 93%) and oxidation of the liberated alcohol 101 (Dess-

(48) (a) De Carolis, E.; De Luca, V. Plant Cell, Tissue Organ Culture 1994,

38, 281. (b) De Carolis, E.; De Luca, V. J. Biol. Chem. 1993, 268, 5504.

(c) De Carolis, E.; Chan, F.; Balsevich, J.; De Luca, V. Plant Physiol.

1990, 94, 1323. For alternative and earlier proposals on the biosynthesis

of vindoline, see: (d) Fahn, W.; Laussermair, E.; Deus-Neumann, B.;

Stockigt, J. Plant Cell Rep. 1985, 4, 333 and 337. (e) Scott, A. I. Acc.

Chem. Res. 1970, 3, 151.

(49) (a) Neuss, N.; Barnes, A. J.; Huckstep, L. L. Experientia 1975, 31, 18. (b)

Barnett, C. J.; Cullinan, G. J.; Gerzon, K.; Hoying, R. C.; Jones, W. E.;

Nevlon, W. M.; Poore, G. A.; Robison, R. L.; Sweeney, M. J.; Todd, G.

C.; Dyke, R. W.; Nelson, R. L. J. Med. Chem. 1978, 21, 88.

(50) Gigant, B.; Wang, C.; Ravelli, R. B. G.; Roussi, F.; Steinmetz, M. O.;

Curmi, P. A.; Sobel, A.; Knossow, M. Nature 2005, 435, 519.

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10609

A R T I C L E S

Scheme 17

Ishikawa et al.

Scheme 18

Martin periodinane, pyr, CH₂Cl₂, 0-25 °C, 30 min, 92%)

provided ketone 102 (Scheme 17, natural enantiomer shown).

Notably, this oxidation carried out in pyridine proved much more

effective than TPAP (0.1-1.0 equiv, NMO, CH₂Cl₂, 23 °C,

2-24 h, 30-60%) and may benefit from the use of the basic

solvent. Thiolactam formation by treatment of 102 with Lawes-

son’s reagent (toluene, 90 °C, 2h, 70%) followed by its reduction

with Ra-Ni (THF, 23 °C, 1 h, 75%) provided 104. Reductive

oxido bridge cleavage (H₂, PtO₂, MeOH-EtOAc, cat. HCl, 23

°C, 50 psi, 81%) required acid cataysis and provided 105 in

good conversions. The alternative sequence of treatment of 81

with Lawesson’s reagent to provide thioamide 106 (88%), its

reductive removal with Ra-Ni under conditions that also

resulted in O-debenzylation (THF, 65 °C, 60%) and subsequent

oxidation of 107 (Dess-Martin periodinane, 15%) to provide

104, was less straightforward suffering from a problematic

oxidation of the secondary alcohol in the presence of the newly

liberated tertiary amine. Ketone 105 has been converted to 8

previously,^9band it also constitutes a key, late stage intermediate

in the Kutney total synthesis of vindoline.^9b

stereochemistry was apparently inverted^51aproviding 3-epi-4-

desacetoxyvindoline. This lack of preceding work is especially

surprising given the historical importance of desacetoxyvin-

blastine⁴⁹lacking only this vindoline C4 acetoxy group. This

natural product is equipotent or more potent in vitro and exhibits

comparable or enhanced in vivo efficacy compared to vinblas-

tine, but it is present in even lower natural abundance (>10-

fold less) than vinblastine which itself is found in trace amounts

(e0.00025% of dry leaf weight).

Importantly, this extension of the approach to include the total

synthesis of 4-deacetoxyvindoline (9) is straightforward and was

projected to be much easier to implement requiring a simpler,

more reactive disubstituted (vs trisubstituted) tethered dieno-

phile. Thus, N-acylation of 1,3,4-oxadiazole 78 with 4-ethyl-

4-pentenoic acid¹⁹(17, EDCI, DMAP, CH₂Cl₂, 23 °C, 16 h)

provided the cycloaddition substrate 109 (85%). Consistent with

expectations, warming 108 in o-dichlorobenzene (180 °C, 24

h, 74%) or triisopropylbenzene (TIPB, 230 °C, 18 h, 83%)

cleanly provided 109 as a single detectable diastereomer

(Scheme 18, natural enantiomer shown) that was readily

resolved on a semipreparative ChiralCel OD column (2 cm ×

25 cm, 10% EtOH-hexane, 10 mL/min, t_R(natural) ) 21.4

min and t_R(unnatural) ) 29.3 min, R ) 1.37). Without a rotation

of the natural product with which to compare synthetic samples,

the absolute configuration of 109 was established by X-ray

4-Desacetoxyvindoline. Having established the synthetic

approach to vindoline (3) and by virtue of the opportunities this

provides for the preparation of simpler derivatives, we elected

to illustrate this with the preparation of 4-desacetoxyvindoline

(9). This alkaloid is a key naturally occurring⁵¹and late stage

biosynthetic precursor to vindoline requiring only C4 hydroxyl-

ation⁴⁸and subsequent acetylation. Despite this important role

in the biosynthetic pathway to vindoline, we are not aware of

a total synthesis of 9, and to date it has only been characterized

1

by MS and H NMR.⁵¹Moreover, we are aware of only one

report⁵²of its attempted preparation through the degradation of

vindoline, and by virtue of the route used, the modified C3

(51) (a) Balsevich, J.; De Luca, V.; Kurz, W. G. W. Heterocycles 1986, 24,

2415. (b) Balsevich, J.; Hogge, L. R.; Berry, A. J.; Games, D. E.;

Mylchreest, I. C. J. Nat. Prod. 1988, 51, 1173.

(52) Kutney, J. P.; Balsevich, J.; Honda, T.; Liao, P.-H.; Thiellier, H. P. M.;

Worth, B. R. Can. J. Chem. 1978, 56, 2560.

analysis of 110²³([R]²³+131 (c 0.6, CHCl₃), recrystallized

D

from Et₂O, mp 228-231 °C), a heavy atom derivative prepared

9

10610 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

(−

)- and ent-(

+

)-Vindoline and Related Alkaloids

A R T I C L E S

Scheme 19

by treatment of (+)-109 ([R]²³_D+129 (c 0.7, CHCl₃)) with NBS

(2.5 equiv, THF, -30 °C, 30 min, 94%). R-Hydroxylation of

109 (NaHMDS, TMSOOTMS, 62%), protection of the second-

ary alcohol 111 as its TIPS ether 112 (TIPSOTf, Et₃N, 25 °C,

1 h, 91%), conversion to the thioamide 113 (69%) with

Lawesson’s reagent, and its reductive removal with Ra-Ni

(THF, 23 °C, 1 h, 97%) provided 114. Diastereoselective

reductive oxido bridge cleavage of 114 (H₂, PtO₂, EtOAc-

MeOH, 88%) followed by TIPS deprotection of 115 (Bu₄NF,

THF, 23 °C, 30 min, 97%) and regioselective elimination of

the secondary alcohol 116 enlisting Ph₃P-DEAD (71%)

provided 4-deacetoxyvindoline (9) which was fully characterized

for the first time [natural 9: [R]²³+33 (c 0.6, CHCl₃)].

D

4-Desacetoxy-6,7-dihydrovindoline and Its 8-Oxo Deriva-

tive. Two analogues of vindoline that contain deep-seated

structural changes and that we anticipate incorporating into

vinblastine analogues are 4-desacetoxy-6,7-dihydrovindoline

(10) and its 8-oxo derivative 118. As alluded to earlier, the C4-

acetoxy substituent, which lies at the external solvent exposed

face of the vinblastine-tubulin complex,⁵⁰can be removed from

vinblastine without diminishing in vitro potency or in vivo

efficacy.⁴⁹Similarly, the ∆^6,7-alkene, which makes no apparent

key interaction with tubulin,⁵⁰can be removed from vinblastine

resulting in only a 2-fold reduction in in vitro potency⁴⁹and

exhibiting good in vivo efficacy.⁴⁹These observations suggest

that vinblastine analogues containing the modified vindoline 10

and even 118 would be especially timely to examine. Moreover,

their 6-7 step (10) or 5-6 step (118) access using the

cycloaddition cascade described herein highlights the power of

the approach and just how accessible such simplified, synthetic

analogues of vinblastine bearing deep-seated vindoline modi-

fications may now become.

Figure 7. Highlights of the approach.

pentacyclic skeleton complete with all substituents and all six

stereocenters is assembled in a single step tandem [4 + 2]/[3

+ 2] cycloaddition cascade of a 1,3,4-oxadiazole (Figure 7).

The intramolecular reaction cascade is initiated by an inverse

electron demand Diels-Alder reaction of an electron-rich enol

ether in which the diene and dienophile substituents complement

and reinforce the [4 + 2] cycloaddition regioselectivity dictated

by the linking tether. Loss of N₂from the initial cycloadduct

provides a carbonyl ylide that in turn undergoes a 1,3-dipolar

cycloaddition with a tethered indole to provide the vindoline

pentacyclic skeleton. Beautifully, the intermediate 1,3-dipole

is stabilized by the complementary substitution at the dipole

termini derived from the starting oxadizole, and the tethered

indole dipolarophile complements the [3 + 2] cycloaddition

regioselectivity dictated by the linking tether. The stereochem-

istry of the cascade cycloadduct is controlled by a combination

of the dienophile geometry (C4/C5 stereochemistry) and an

exclusive endo indole cycloaddition directed to the dipole face

opposite the newly formed fused lactam dictated by the

dipolarophile tether.⁵³Key issues regarding the scope and

stereochemical features of the new cycloaddition cascade were

defined in the course of the investigations including the

observation of an unusual electronic effect that slows the

typically faster [3 + 2] cycloaddition reaction relative to the

initiating Diels-Alder reaction. Recognition of this unusual

electronic effect and the impact it has on the rate-limiting step

Thus, treatment of the cycloadduct 109 with Lawesson’s

reagent (toluene, 110 °C, 1.5 h, 90%) and its S-methylation

(Me₃OBF₄, CH₂Cl₂) and subsequent reductive removal under

conditions (NaBH₄, MeOH, 23 °C, 2 h) that also result in

reductive oxido bridge cleavage provided 4-desacetoxy-6,7-

dihydrovindoline (10), [R]²³+43 (c 1.8, MeOH), in superb

D

conversions (92%), Scheme 19 (natural enantiomer shown).

Alternatively, reductive oxido bridge cleavage of 109 (NaC-

NBH₃, HCl-MeOH, 25 °C, 1.5 h, 92%) provided the 8-oxo

derivative 118, [R]²³+6 (c 0.7, MeOH), in a single step.

D

Conclusions

(53) Interestingly, intermolecular [3 + 2] variants proceed with indole exo

cycloaddition directed to an analogous R-face; see: Muthusamy, S.;

Gunanathan, C.; Babu, S. A. Tetrahedron Lett. 2001, 42, 523.

Full details of the development of a concise total synthesis

of (-)- and ent-(+)-vindoline are disclosed in which the entire

9

J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006 10611

A R T I C L E S

Ishikawa et al.

of the cycloaddition cascade proved instrumental in implement-

ing conditions that provided direct access to vindoline.

detailed in efforts that highlight the utility of the cycloaddition

cascade for accessing a broader range of alkaloids beyond those

immediately related to vindoline. A continued extension of these

latter studies to additional alkaloid families is in progress as

are efforts directed at vinblastine and key analogues incorporat-

ing deep-seated changes in the vindoline subunit.

As the studies unfolded, the cycloaddition cascade was

enlisted as the key step in the total synthesis of a series of related

alkaloids of increasing complexity. Thus, a total synthesis of

natural and ent-minovine (4) was completed in which the

confusion surrounding the optical rotation, racemization poten-

tial, optical purity, and absolute configuration of the natural

product was resolved. Similarly, the first total syntheses of both

enantiomers of 4-desacetoxy-6,7-dihydrovindorosine (5) and

4-desacetoxyvindorosine (6), a late stage biosynthetic intermedi-

ate for 7, were conducted along with efforts that provided a

concise and effective 11-step total synthesis of (-)- and ent-

(+)-vindorosine (7) complementing the prior two total and six

formal total syntheses of this natural product.³⁹In efforts that

highlight the versatility of the approach and its potential for

accessing vinblastine analogues incorporating deep-seated struc-

tural changes not accessible from the natural product itself, the

cycloaddition cascade was subsequently enlisted in total syn-

theses of both enantiomers of 6,7-dihydrovindoline (8), 4-des-

acetoxyvindoline (9), 4-desacetoxy-6,7-dihydrovindoline (10, 7

steps), and its 8-oxo derivative 118 (6 steps). Notably, the

preparation of (+)-9, constituting its only total synthesis,

provided the first opportunity for complete characterization of

this natural product which constitutes a key late stage biosyn-

thetic intermediate for vindoline. Finally, the total synthesis of

both enantiomers of N-methylaspidospermidine (11) was also

Acknowledgment. We gratefully acknowledge the financial

support of the National Institutes of Health (CA42056 and

CA115526) and the Skaggs Institute for Chemical Biology. We

wish to thank Dr. R. Schaum for initiating these studies, Dr. S.

Pollack for early studies probing the dienophile selection and

stage of ∆^6,7-alkene introduction (Figure 6, Scheme 12) as well

as initial studies with (Z)- and (E)-59 and 79, Dr. Z.-Q. Yuan

for a synthesis of racemic minovine, Dr. M. M. Miller for the

initial synthesis of racemic vindoline from (E)-79 and 81

(Scheme 15) as well as a racemic synthesis of Kutney’s

intermediate 105, and Dr. Y. Li for studies of the lactam

R-hydroxylation reaction of 60, 61, 80, and 81 resulting in the

use of TMSOOTMS and for providing material supplies at an

intermediate stage of studies on vindorosine and vindoline.

Supporting Information Available: Full experimental details

and compound characterizations are provided. This material is

available free of charge via the Internet at http://pubs.acs.org.

JA061256T

9

10612 J. AM. CHEM. SOC. VOL. 128, NO. 32, 2006

Article Doi

Total synthesis of (-)- and ent-(+)-vindoline and related alkaloids

DOI: 10.1021/ja061256t

Source and publish data:

Authors:

Article abstract of DOI:10.1021/ja061256t

Full text of DOI:10.1021/ja061256t

Products guided by the article

R&D Labs maybe for 7649-19-6

Relevant to this article

Hot Product