Convergent assembly of aldol linkages. A study of the stereoselective nitrile oxide cycloaddition reaction

Article abstract of DOI:10.1016/S0040-4039(00)00981-3

The aldol equivalency of a stereoselective nitrile oxide cycloaddition is exploited in the development of a general synthetic strategy to the polyene macrolide antibiotics. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4039(00)00981-3

Tetrahedron Letters 41 (2000) 6011±6015
Convergent assembly of aldol linkages. A study of the
stereoselective nitrile oxide cycloaddition reaction
Glenn J. McGarvey,* Je€rey A. Mathys and Kenneth J. Wilson
Department of Chemistry, University of Virginia, Charlottesville, Virginia 22901, USA
Received 1 May 2000; accepted 30 May 2000
Abstract
The aldol equivalency of a stereoselective nitrile oxide cycloaddition is exploited in the development of a
general synthetic strategy to the polyene macrolide antibiotics. # 2000 Elsevier Science Ltd. All rights
reserved.
Keywords: nitrile oxide cycloaddition; amphotericin B; polyol synthesis.
Within the polyene macrolide antibiotics, the heptaene/pseudoheptaene subfamily has gained
prominence as a consequence of their clinically important antifungal activity.1 We have recently
reported on our successful e€orts to develop a synthetic strategy that may be generally applicable
to this structural sub-group.2 As illustrated (Fig. 1), signi®cant structural homology exists among
these compounds, with most of the variation residing in the C1±C10 segment of the macrocycle.
These structural relationships guided the design of our synthetic approach to amphotericin B. In
order to realize generality in our strategy, a convergent approach was adopted that featured
fusion of the conserved and variable segments of the macrocycle. As a result of these considera-
tions, polyol segment 1 was identi®ed as a key intermediate that would allow the fusion of the
conserved and variable carbon segments.
Successful execution of our strategy hinged on an e€ective means of forming the C13±C15
aldol linkage. As the shaded region of segment 1 highlights, standard aldol construction of the C15
alcohol is not viable as this would require a condensation reaction with an enolizable b-dicarbonyl
component. In a search for alternative approaches, we were attracted to the aldol equivalency of
the isoxazoline adducts 2 resulting from nitrile oxide cycloadditions.3 Application of this method
to the assembly of polyol segment 1 requires that this cycloaddition reaction proceed in both an
ecient and stereoselective manner to a€ord the desired C15 stereocenter without undue loss of
valuable synthetic precurors. With these concerns in mind, we embarked upon model studies
to optimize these reaction parameters.
* Corresponding author.
0040-4039/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(00)00981-3

6012
Figure 1.
A survey of reported stereoselective nitrile oxide cycloaddition reactions clearly indicates that
the dipolarophile component dominants steric course of the reaction.4 Armed with this observa-
tion, we embarked upon studies to identify an asymmetric dipolarophile that would be generally
applicable to the assembly of the (psuedo)heptaene macrocrolactones. A C14±C19 dipolarophile
was eciently prepared in the manner shown in Scheme 1. An Evans' aldol condensation5
establishes the desired C16 and C17 stereochemistry (3!4a/b) in an adduct (5) which, following
selective protecting group manipulation, a€ords lactone 6. It was interesting to observe the
sensitivity of this product to base-induced epimerization. In fact, while attempting to obtain a
deprotected dipolarophile for study, complete isomerization to 7 resulted, presumably as a con-
sequence of a preferred boat-like conformation of the lactone ring.
n
Scheme 1. (a) Bu2BOTf, Et3N, CH2Cl2; (b) ROCH2CH2CHO, 95% ds, R=TBS, 78%, R=pMBꢀ, 71%; (c) TBSOTf,
ꢀ
t
.
.
Et3N, CH2Cl2, 0 C; (d) HF pyridine, pyridine, rt, 64% (two steps); (e) BuMgBr, THF, ^78!^25 C; (f) HF pyridine,
pyridine, rt
While not appropriate for the synthesis of amphotericin B, cycloaddition studies using lactone
7 were instructive. We chose aldoximes derived from d- or l-malic acid as model dipole pre-
cursors,6 anticipating that asymmetry in this component would exert a small in¯uence on
cycloaddition stereoselection. Several sets of standard conditions for the generation of nitrile
oxides were applied to the R-oxime to give the results shown (Scheme 2). When basic dehydro-
halogenation conditions were employed, the desired product 8 was isolated in poor yields with the
major product being a spirocyclic adduct 9, resulting from double bond isomerization in the
dipolarophile. Fortunately, the use of (nBu3Sn)2O to promote dehydrohalogenation7 led to clean
cycloaddition with no isomerized adduct. While acceptably ecient, this cycloaddition showed
unacceptable stereoselectivity. It was also disappointing to ®nd that use of the dipolarophile

6013
having the proper stereochemistry (6) gave similarly modest levels of stereoselectivity. Never-
theless, the aldol equivalency of these cycloaddition reactions was demonstrated by the ecient
Mo(CO)6-promoted cleavage/hydrolysis8 of the isoxazoline adduct to the corresponding b-hydroxy
ketone 10.
Scheme 2.
An investigation of other dipolarophiles was initiated in an e€ort to improve the diastereo-
selection of this cycloaddition process. Fortunately, it was found that the Evans' aldol product
4b could be used, even without protection of the putative C17 alcohol. The advantages of this
dipolarophile were immediately apparent. First, the double bond in this substrate was far less
prone to isomerization, thus allowing the use of Et3N as a base for nitrile oxide formation. More
importantly, stereoselectivity was signi®cantly improved (see Table 1). The use of halosuccinimides
with Et3N or (nBu3Sn)2O to e€ect dehydrohalogenation gave similar levels of diastereoselection
but with modest yields re¯ecting diculties in removing succinimide from the product (entries 1±4).
It bears noting that the yield of this cycloaddition is strongly e€ected by the order of addition of
t
the dipolarophile with the halosuccinimide (entries 3 and 4). Use of BuOCl as the chlorinating
Table 1

6014
agent led to signi®cant improvements in both yield and stereoselectivity (entry 5). It was ultimately
found that good yields and diastereoselectivity could be obtained using approximately stoichio-
metric amounts of the reagents and a modest excess of the dipolarophile (entries 6 and 7).
With an e€ective protocol identi®ed for the cycloaddition, it remained to establish the stereo-
chemistry of the cycloadduct. This was accomplished as depicted in Scheme 3. The isomeric
b-hydroxy ketones could be unmasked then trapped as their cyclic mixed acetal through the
agency of (MeO)3CH. The resultant conformationally biased diastereomers were examined by
various NMR methods to establish stereochemistry.9 With the acetal methoxy group occupying
the steric/stereoelectronically favored axial position, two distinct substitution patterns could be
identi®ed. Notably, the major isomer exhibited a clear NOE e€ect between HA±HD and HB±HC,
as well as J-coupling values consistent with equatorial substitution about the ring. As a result, the
major diastereomer of the nitrile oxide cycloaddition could be assigned to the desired S-stereo-
Scheme 3. (a) Mo(CO)6, 10% H2O/MeCN, 80ꢀC; (b) PPTs, MeOH:(MeO)3CH (1:1), rt
chemistry.
With these encouraging results in hand, the optimized cycloaddition conditions were applied to
the synthetic C1±C13 fragment of amphotericin B2 (11, Scheme 4). In the event, cycloaddition
using 3 equivalents of dipolarophile 4b10 proceeded in an excellent 88% yield with diastereo-
selectivity identical to the model studies. This latter result lends credence to our anticipation that
the dipolarophile would govern the stereochemical course of this bond-forming reaction. Cleavage/
hydrolysis of the isoxazoline adduct followed by cyclization/acetalization11 proceeded smoothly
to a€ord the intact C1±C19 fragment of amphotericin B. As we previously reported, this material
Scheme 4. (a) Sequential addition: (i) (nBu3Sn)2O; (ii) ole®n 4b (3 equiv.); (iii) tBuOCl; (b) Mo(CO)6, 10% H2O/MeCN,
70ꢀC; (c) MeC(OMe)3, PPTs, MeOH, rt
could be incorporated into the polyene macrocycle to complete the total synthesis.2,12
This study has demonstrated a remarkably e€ective means of linking highly functionalized
carbon frameworks as part of a strategy for the synthesis of complex natural products. In the present
case, the nitrile oxide cycloaddition contributes to a general strategy to the polyene macrolide
antibiotics. Given the stereodirecting in¯uences of the dipolarophile 4b, it can be anticipated that
other dipole precursors may be used to allow synthetic access to other members of this class of
compounds (see Fig. 1).

6015
Acknowledgements
We gratefully acknowledge the National Institutes of Health and the University of Virginia for
partial support of this work.
References
1. Omura, S.; Tanaka, H. In Macrolide Antibiotics: Chemistry, Biology, and Practice; Omura, S., Ed.; Academic
Press: New York, 1984; pp. 351±404.
2. McGarvey, G. J.; Mathys, J. A.; Wilson, K. J. J. Org, Chem. 1996, 61, 5704±5705.
3. For a recent review, see: Easton, C. J.; Hughes, C. M. M.; Savage, G. P.; Simpson, G. W. Adv. Heterocycl. Chem.
1994, 60, 261±327.
4. Gothelf, K. V.; Jùrgensen, K. A. Chem. Rev. 1998, 98, 863±909.
5. Evans, D. A.; Sjogren, E. B.; Bartroli, J.; Dow, R. L. Tetrahedron Lett. 1986, 27, 4957±4960.
6. The dipole precursors were prepared using the methods described in: (a) Hanessian, S.; Ugolini, A.; Therien, M.
J. Org. Chem. 1983, 48, 4427±4430. (b) Barco, A.; Benetti, S.; De Risi, C.; Pollini, G.; Romagnoli, R.; Zanirato, V.
Tetrahedron 1994, 50, 10491±10496.
7. (a) Moriya, O.; Urata, Y.; Endo, T. J. Chem. Soc., Chem. Commun. 1991, 17±18. (b) Moriya, O.; Takenaka, H.;
Iyoda, M.; Urata, Y.; Endo, T. J. Chem. Soc., Perkin Trans. 1 1994, 413±417.
8. Baraldi, P.; Barco, A.; Benetti, S.; Manfredini, S.; Simoni, D. Synthesis 1987, 276±278.
1
9. Both diastereomers were analyzed using 1D H, COSY, and NOESY experiments to obtain vicinal J-coupling
values, proton connectivity, and spatial relationships, respectively.
10. An examination of molar dipolarophile:dipole ratios from 1:1 to >5:1 revealed that the yield was optimized at 3:1.
11. Mixed acetal formation was signi®cantly improved when CH3C(OMe)3 was used in place of HC(OMe)3.
12. Physical data for selected new compounds. Compound 4a: [ꢀ]D=+20.82 (c 3.08, CCl4); IR (®lm) 3400, 2940, 2920,
1
1775, 1680, 1450, 1340, 1250, 1190, 1080 cm^1; H NMR (CDCl3, 300 MHz) ꢁ 0.02 (s, 4H), 0.82 (d, 3H, J=6.3
Hz), 0.85 (s, 9H), 1.88 (m, 2H), 3.81 (m, 3H), 4.21 (m, 1H), 4.50 (dd, 1H, J=4.2, 8.7 Hz), 4.76 (m, 1H), 5.27 (m,
2H), 5.64 (d, 1H, J=7.2 Hz), 6.01 (m, 1H), 7.23±7.36 (m, 5H); 13C (CDCl3, 75 MHz) ꢁ ^5.03, 14.67, 18.65, 26.34,
35.60, 36.58, 53.40, 55.33, 59.66, 62.09, 71.74, 79.30, 100.25, 120.81, 126.06, 129.14, 129.23, 132.26, 133.60, 153.06,
173.57. Compound 6: [a]d=+33.87ꢀ (c 1.42, CCl4); IR (®lm) 2975, 2960, 2822, 1722, 1508, 1299, 1252, 1212, 1160,
1
1070 cm^1; H NMR (CDCl3, 300 MHz) ꢁ 0.06 (s, 6H), 0.88 (s, 9H), 1.77 (m, 1H), 2.14 (m, 1H), 3.20 (dd, 1H,
J=5.8, 7.2 Hz), 4.03 (dd, 1H, J=4.8, 8.4 Hz), 4.29 (ddd, 1H, J=4.5, 8.7, 12.4 Hz), 4.55 (dt, 1H, J=3.9, 10.8 Hz),
5.26 (m, 2H), 5.77 (m, 1H); 13C NMR (CDCl3, 75 MHz) ꢁ ^4.40, ^4.25, 18.32, 26.06, 29.96, 55.02, 65.47, 68.92,
120.11, 133.84, 171.43; anal. calcd for C13H24O3Si: C, 61.13, H, 9.08; found: 61.32; H, 9.25. Compound 7:
[ꢀ]D=^60.82 (c 1.00, CHCl3); mp=44^46ꢀ C; IR (CCl4) 3005, 2980, 2960, 1720, 1505, 1465, 1410, 1200, 1050
1
cm^1; H NMR (CDCl3, 300 MHz) ꢁ 0.05 (s, 6H), 0.80 (s, 9H), 1.90 (m, 1H), 2.12 (m, 1H), 3.10 (dd, 1H, J=3.3,
8.1 Hz), 4.31 (m, 2H), 4.60 (dt, 1H, J=10.2, 11.1 Hz), 5.24 (dd, 2H, J=10.2, 17.4 Hz), 6.02 (m, 1H); 13C NMR
(CDCl3, 75 MHz) ꢁ ^4.75, ^4.25, 18.42, 26.09, 31.63, 53.04, 65.78, 69.75, 119.60, 133.90, 171.85; anal. calcd for
C13H24O3Si: C, 61.13, H, 9.08; found: C, 61.00, H, 9.13. S-product (Table 1): 1H NMR (CDCl3, 300 MHz) ꢁ 0.91
(d, 3H, J=6.6 Hz), 1.32 (s, 3H), 1.41 (s, 3H), 1.78 (m, 2H), 2.01 (s, 1H), 2.57 (m, 2H), 3.09 (m, 2H), 3.64 (m, 6H),
3.75 (s, 3H), 4.10 (m, 1H), 4.27 (m, 1H), 4.33 (M, 1H), 4.72 (S, 2H), 4.65 (dd, 1H, J=6.6, 9.6 Hz), 4.75 (m, 1H),
4.92 (dd, 1H, J=7.2 Hz), 6.85 (d, 2H, J=8.4 Hz), 7.22±7.42 (m, 7H); 13C NMR (CDCl3, 75 MHz) ꢁ 14.45, 14.59,
25.78, 25.93, 27.31, 27.39, 32.71, 34.49, 44.22, 53.31, 55.53, 68.67, 71.34, 73.49, 73.67,79.16, 80.69, 100.65, 114.30,
126.01, 126.28, 129.00, 129.23, 129.51, 129.56, 129.74, 129.95, 130.25, 133.64, 153.29, 157.06, 172.43; HRMS
(M+1): anal. calcd for C32H40N2O8: 596.2734; found: 596.2781. Major mixed acetal (Scheme 3): 1H NMR (CDCl3,
300 MHz) ꢁ 0.90 (d, 3H, J=6.6 Hz), 1.35±1.45 (s, 6H), 1.6±2.1 (m, 5H), 2.19 (dd, 1H, J=4.59, 12.65 Hz), 2.52 (d,
1H, J=10.2 Hz), 3.14 (s, 3H), 3.5±3.8 (m, 7H), 3.9±4.3 (m, 4H), 4.40 (d, 2H, J=3.9 Hz), 4.65 (m, 1H), 5.17 (d, 1H,
J=7.2 Hz), 6.84 (d, 2H, J=9.0 Hz), 7.21 and 7.40 (m, 7H); 13C NMR (CDCl3, 75 MHz) ꢁ 14.72, 26.20, 27.28,
30.10, 30.12, 33.56, 40.13, 43.27, 48.01, 52.89, 55.57, 55.89, 66.47, 68.74, 69.66, 70.86, 72.65, 73.33, 100.10, 108.92,
114.17, 126.06, 129.11, 129.22, 129.97, 130.74, 133.32, 154.69, 159.67, 173.48; anal. calcd for C33H43NO10:
613.2785; found: 613.284.