Stereochemistry of the allylation and crotylation reactions of α-methyl-β-hydroxy aldehydes with allyl- and crotyltrifluorosilanes. Synthesis of anti, anti-dipropionate stereotriads and stereodivergent pathways for the reactions with 2,3-anti- and 2,3-syn-α-methyl-β-hydroxy aldehydes

Article abstract of DOI:10.1021/jo0267908

A new method for the stereoselective synthesis of the anti,anti-dipropionate stereotriad via the reaction of α-methyl-β-hydroxy aldehydes with (Z)- crotyltrifluorosilane (24) is described. These reactions were designed to occur through bicyclic transition states (e.g., 31) in which the silane reagent is covalently bound to the β-hydroxyl group of the aldehyde and the crotyl group is transferred intramolecularly. This methodology was used to synthesize the C(7)-C(16) segment (58) of zincophorin, which contains a synthetically challenging all-anti stereopentad unit. Surprisingly, 2,3-anti- and 2,3-syn-β-methyl-β-hydroxy aldehydes react in a stereodivergent manner with 24: 2,3-anti-β-hydroxy aldehydes give the targeted anti,anti-dipropionate adducts with high selectivity, but the reactions of 2,3-syn-β-hydroxy aldehydes are poorly selective. The stereodivergent behavior of 2,3-syn- vs 2,3-anti-α-methyl-β-hydroxy aldehydes is also exhibited in their reactions with the allyl- (68) and (E)-crotyltrifluorosilanes (27). Competition experiments performed with β-hydroxy aldehydes 37a (anti) and the corresponding p-methoxybenzyl (PMB) ether 48, and between aldehyde 39 (syn) and the PMB ether 90, established that the 2,3-anti-β-hydroxy aldehydes react predominantly through bicyclic transition states while the 2,3-syn aldehydes react predominantly through conventional Zimmerman-Traxler transition states. NMR studies established that both the 2,3-syn and the 2,3-anti aldehydes form stable, pentavalent silicate intermediates (98 and 100) with PhSiF3, but chelated structures 99 and 101 could not be detected. The activation energies for the competing bicyclic and conventional Zimmerman-Traxler transition states were calculated by using semiemperical methods (MNDO/d). These calculations indicate that the stereodivergent behavior of the 2,3-syn-β-hydroxy aldehydes and the 2,3-anti-β-hydroxy aldehydes is due to differences in nonbonded interactions in the bicyclic transition states. Specifically, nonbonded interactions in the bicyclic transition states for the allylation/crotylation reactions of the 2,3-syn-β-hydroxy aldehydes permits the traditional Zimmerman-Traxler transition states to be preferentially utilized.

Full text of DOI:10.1021/jo0267908

Ster eoch em istr y of th e Allyla tion a n d Cr otyla tion Rea ction s of
r-Meth yl-â-h yd r oxy Ald eh yd es w ith Allyl- a n d
Cr otyltr iflu or osila n es. Syn th esis of a n ti,a n ti-Dip r op ion a te
Ster eotr ia d s a n d Ster eod iver gen t P a th w a ys for th e Rea ction s w ith
2,3-a n ti- a n d 2,3-syn -r-Meth yl-â-h yd r oxy Ald eh yd es
Sherry R. Chemler1 and William R. Roush*
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109
roush@umich.edu
Received December 3, 2002
A new method for the stereoselective synthesis of the anti,anti-dipropionate stereotriad via the
reaction of R-methyl-â-hydroxy aldehydes with (Z)-crotyltrifluorosilane (24) is described. These
reactions were designed to occur through bicyclic transition states (e.g., 31) in which the silane
reagent is covalently bound to the â-hydroxyl group of the aldehyde and the crotyl group is
transferred intramolecularly. This methodology was used to synthesize the C(7)-C(16) segment
(58) of zincophorin, which contains a synthetically challenging all-anti stereopentad unit.
Surprisingly, 2,3-anti- and 2,3-syn-R-methyl-â-hydroxy aldehydes react in a stereodivergent manner
with 24: 2,3-anti-â-hydroxy aldehydes give the targeted anti,anti-dipropionate adducts with high
selectivity, but the reactions of 2,3-syn-â-hydroxy aldehydes are poorly selective. The stereodivergent
behavior of 2,3-syn- vs 2,3-anti-R-methyl-â-hydroxy aldehydes is also exhibited in their reactions
with the allyl- (68) and (E)-crotyltrifluorosilanes (27). Competition experiments performed with
â-hydroxy aldehydes 37a (anti) and the corresponding p-methoxybenzyl (PMB) ether 48, and
between aldehyde 39 (syn) and the PMB ether 90, established that the 2,3-anti-â-hydroxy aldehydes
react predominantly through bicyclic transition states while the 2,3-syn aldehydes react predomi-
nantly through conventional Zimmerman-Traxler transition states. NMR studies established that
both the 2,3-syn and the 2,3-anti aldehydes form stable, pentavalent silicate intermediates (98
and 100) with PhSiF3, but chelated structures 99 and 101 could not be detected. The activation
energies for the competing bicyclic and conventional Zimmerman-Traxler transition states were
calculated by using semiemperical methods (MNDO/d). These calculations indicate that the
stereodivergent behavior of the 2,3-syn-â-hydroxy aldehydes and the 2,3-anti-â-hydroxy aldehydes
is due to differences in nonbonded interactions in the bicyclic transition states. Specifically,
nonbonded interactions in the bicyclic transition states for the allylation/crotylation reactions of
the 2,3-syn-â-hydroxy aldehydes permits the traditional Zimmerman-Traxler transition states to
be preferentially utilized.
In tr od u ction
Much effort in the area of acyclic stereocontrol has been
focused on the development of highly stereoselective
methods for synthesis of the four “dipropionate” stereo-
triads 2, 3, 4, and 5, which are commonly found subunits
in polyketide natural products.2 Of these stereotriads, the
anti,anti-isomer 3 has historically been the most chal-
lenging to synthesize.2,3 In many cases, indirect methods
involving multistep sequences have been employed for
the synthesis of stereotriad 3.3
The stereoselective synthesis of the anti,anti-dipropi-
onate 3 via the reaction of 1 with an (E)-crotylmetal
reagent 6 or an (E)-(O)-enolate 7 is inherently problem-
atic as it must arise from a disfavored transition state
12 where the reagent adds to the chiral aldehyde in an
anti-Felkin manner.4-8 This is illustrated below for the
reaction of 1 with (E)-crotylmetal reagent 6.9 The 3,4-
anti-4,5-syn homoallylic alcohol 11, which arises through
(1) Current address: Department of Chemistry, SUNY Buffalo,
Buffalo, NY 14260. E-mail: schemler@buffalo.edu.
(2) Hoffmann, R. W. Angew. Chem., Int. Ed. Engl. 1987, 26, 489.
(3) Hoffmann, R. W.; Dahmann, G.; Andersen, M. W. Synthesis 1994,
629.
10.1021/jo0267908 CCC: $25.00 © 2003 American Chemical Society
Published on Web 01/24/2003
J . Org. Chem. 2003, 68, 1319-1333
1319

Chemler and Roush
the Felkin-Anh transition state 10,10-12 is favored in
these reactions, and as the size of R increases relative to
Me, the selectivity for 11 increases proportionately since
the competing anti-Felkin transition state, 12, becomes
increasingly destabilized by the gauche interactions
between the reagent γ-methyl and the aldehyde R
group.4-7
ing ca. 9:1 mixture favoring the undesired diastereomer
17.20
The highly enantioselective enol borane and crotyl-
boronate reagents developed by Masamune21,22 and Hoff-
mann3,23,24 have been used successfully in challenging
examples of mismatched double asymmetric reactions
leading to the anti,anti-dipropionate stereotriad. How-
ever, the complex, multistep sequences required for
synthesis of these reagents have restricted their wide-
spread use in organic synthesis.
Marshall25 and Panek26,27 have demonstrated that
chelation-controlled addition of chiral Type II9 allenyl-
stannane and crotylsilane reagents to R-methyl-â-ben-
zyloxy substituted aldehydes provide anti,anti-dipropi-
onate adducts selectively.7,28 However, the Marshall and
Panek reagents also require multistep preparations, and
this method for the synthesis of the anti,anti stereotriad
is limited to chiral aldehydes that possess â-alkoxy ether
protecting groups that are capable of participating in
chelate-controlled carbonyl addition reactions.29-31
Our inability to achieve selective access to 16 by using
the highly enantioselective chiral (E)-crotyboronate 1519,20
led us to explore a conceptually different strategy that
uses the intrinsic diastereofacial selectivity preference
of the chiral aldehyde to direct the formation of the
subsequent stereocenters. We envisiaged that this could
be accomplished via the reaction of a R-methyl-â-hydroxy
aldehyde with a (Z)-crotylmetal reagent 8 that proceeds
by way of the bicyclic transition state 19 in which the
â-hydroxyl group of 18 is engaged in a chelate with the
(Z)-crotylmetal reagent. Bond formation should then
occur anti to the aldehyde R-methyl group, thus favoring
the 3,4-anti-4,5-anti-dipropionate 20 without recourse to
expensive chiral reagents. Several examples of stereose-
lective intramolecular allylation reactions occurring
through R- or â-hydroxyl chelation had been reported
prior to the initiation of our work.32-38
In principle, anti,anti-dipropionate synthons 3 or 13
can be prepared directly from 1 by using chiral crotyl-
metal reagents or chiral enolates. However, the intrinsic
diastereofacial bias of 1 requires that 3 be produced via
mismatched double asymmetric reactions.13 For example,
the often used, easily synthesized tartrate ester-derived
(E)-crotylboronate reagent developed in our laboratory6,14
and the (E)-crotyldiisopinocampheylborane developed by
Brown15,16 have demonstrated proficiency in overcoming
the diastereofacial bias of chiral aldehydes in cases when
the aldehyde R group is relatively nonsterically demand-
ing.4,7 In more challenging cases involving aldehyde
substrates in which the R substituent (see 1) is quite
sterically demanding, however, these reagents fail to give
synthetically useful levels of selectivity for the anti,anti-
dipropionate.6,17,18 For example, attempts to achieve a
reagent-controlled mismatched crotylboration of 14 with
use of (E)-crotylboronate 1519 containing our most highly
enantioselective tartramide auxiliary gave a disappoint-
(4) Roush, W. R. In Comprehensive Organic Synthesis; Trost, B. M.,
Ed.; Pergamon Press: Oxford, UK, 1991; Vol. 2, p 1.
(5) Roush, W. R. J . Org. Chem. 1991, 56, 4151.
(6) Roush, W. R.; Palkowitz, A. D.; Ando, K. J . Am. Chem. Soc. 1990,
112, 6348.
(7) Chemler, S. R.; Roush, W. R. In Modern Carbonyl Chemistry;
Otera, J ., Ed.; Wiley-VCH: Weinheim, Germany, 2000; p 403.
(8) Denmark, S. E.; Almstead, N. G. In Modern Carbonyl Chemistry;
Otera, J ., Ed.; Wiley-VCH: Weinheim, Germany, 2000; p 299.
(9) For the definition of Type I-Type III allylorganometallic re-
agents, see: Denmark, S. E.; Weber, E. J . Helv. Chim. Acta 1983, 66,
1655-1660.
(10) Cherest, M.; Felkin, H.; Prudent, N. Tetrahedron Lett. 1968,
2199.
(11) Anh, N. T.; Eisenstein, O. Nouv. J . Chim. 1977, 1, 61.
(12) Anh, N. T. Top. Curr. Chem. 1980, 88, 145.
(13) Masamune, S.; Choy, W.; Petersen, J . S.; Sita, L. R. Angew.
Chem., Int. Ed. Engl. 1985, 24, 1.
(14) Roush, W. R.; Ando, K.; Powers, D. B.; Palkowitz, A. D.;
Halterman, R. L. J . Am. Chem. Soc. 1990, 112, 6339.
(15) Brown, H. C.; Bhat, K. S.; Randad, R. S. J . Org. Chem. 1987,
52, 3701.
(16) Brown, H. C.; Bhat, K. S.; Randad, R. S. J . Org. Chem. 1989,
54, 1570.
Initial attempts to obtain the anti,anti-dipropionate
adduct via the reaction of â-hydroxy aldehyde 21 with
(17) Zheng, W.; DeMattei, J . A.; Wu, J .-P.; Duan, J . J .-W.; Cook, L.
R.; Oinuma, H.; Kishi, Y. J . Am. Chem. Soc. 1996, 118, 7946.
(18) Ogawa, A. K.; Armstrong, R. W. J . Am. Chem. Soc. 1998, 120,
12435.
(20) Grover, P. T. Ph.D. Thesis, Indiana University, 1994.
(21) Short, R. P.; Masamune, S. Tetrahedron Lett. 1987, 28, 2841.
(22) Tanimoto, N.; Gerritz, S. W.; Sawabe, A.; Noda, T.; Filla, S. A.;
Masamune, S. Angew. Chem., Int. Ed. Engl. 1994, 33, 673. See, in
particular, the conversion of 13 to 15 therein.
(19) Roush, W. R.; Grover, P. T. J . Org. Chem. 1995, 60, 3806.
1320 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
(Z)-crotyldiisopropylboronate36-38 were unsuccessful.39
This reaction provided a mixture of 22 and 23, the
products of normal Zimmerman-Traxler transition path-
ways.4 Assuming that our lack of success with this
experiment arose from the reluctance of the diisopropyl
(Z)-crotylboronate to esterify with the â-hydroxyl group
of 21, as crotylboronates do not require nucleophilic
activation for normal, nonchelated carbonyl addition to
occur, we turned our attention toward the use of a (Z)-
crotylmetal reagent that requires nucleophilic activation
for reaction with carbonyl electrophiles.
γ-carbon of the crotyl unit without substantially decreas-
ing the Lewis acidity of the silicon center,47,48 which then
engages with carbonyl substrates via Zimmerman-
Traxler allylation transition states 25 and 28 to give the
homoallylic alcohol products with excellent stereoselec-
tivity. Several reports of enantioselective allylation
and crotylation reactions of achiral aldehydes and tri-
chloroallylsilanes catalyzed by chiral nucleophiles have
appeared.49-60 In addition, several applications of allyl-
trifluorosilanes for allylation reactions of R-hydroxy
ketones, R-diketones, and salicaldehyde derivatives have
also been reported, where the reagent is activated by
chelation to the substrate.33-35
Allyl- and crotylsilanes that contain electronegative
substituents (e.g., -OMe, -Cl, or -F) at silicon are
known to undergo nucleophile-promoted addition reac-
tions with carbonyl compounds via the usual Type I
allylation pathway.40-46 For example, the reaction of
benzaldehyde with (Z)-crotyltrifluorosilane (24) in the
presence of CsF gives the syn homoallylic alcohol 26 with
99:1 selectivity, while the (E)-crotylsilane reagent 27
gives the anti homoallylic alcohol 29, also with 99:1
selectivity.42 Formation of a pentavalent silicate species
by addition of a nucleophile to the silicon center of 24
and 27 serves to increase the nucleophilicity of the
Accordingly, we envisioned that the â-hydroxyl group
of aldehyde 18 would add to the silicon center of (Z)-
crotyltrifluorosilane 24 to generate the pentacoordinate
crotylsilicate intermediate 30, which would then undergo
crotyl transfer via the bicyclic transition state 31, thereby
affording the anti,anti-dipropionate 20. We report here
the full details of our development of this reaction,
including our observation that the success of this process
is highly dependent on the stereochemistry of the C(3)
stereocenter of 18. Preliminary accounts of this work
have appeared previously.61,62
(23) Andersen, M. W.; Hilderbrandt, B.; Hoffmann, R. W. Angew.
Chem., Int. Ed. Engl. 1991, 30, 97-99.
(24) Andersen, M. W.; Hildebrandt, B.; Dahmann, G.; Hoffmann,
R. W. Chem. Ber. 1991, 124, 2127.
(25) Marshall, J . A.; Perkins, J . F.; Wolf, M. A. J . Org. Chem. 1995,
60, 5556.
(26) J ain, N. F.; Takenaka, N.; Panek, J . S. J . Am. Chem. Soc. 1996,
118, 12475.
(27) Panek, J . S.; J ain, N. F. J . Org. Chem. 1998, 63, 4572.
(28) Marshall, J . A. Chem. Rev. 2000, 100, 3163.
(29) Reetz, M. T. Angew. Chem., Int. Ed. Engl. 1984, 23, 556.
(30) Reetz, M. T. Acc. Chem. Res. 1993, 26, 462-468.
(31) Mengel, A.; Reiser, O. Chem. Rev. 1999, 99, 1191.
(32) Reetz, M.; J ung, A. J . Am. Chem. Soc. 1983, 105, 4833.
(33) Kira, M.; Sato, K.; Sekimoto, K.; Gewald, R.; Sakurai, H. Chem.
Lett. 1995, 281.
(34) Sato, K.; Kira, M.; Sakurai, H. J . Am. Chem. Soc. 1989, 111,
6429.
(35) Gewald, R.; Kira, M.; Sakurai, H. Synthesis 1996, 111.
(36) Wang, Z.; Meng, X.-J .; Kabalka, G. W. Tetrahedron Lett. 1991,
32, 1945.
(37) Kabalka, G. W.; Narayana, C.; Kesavulu, R. Tetrahedron Lett.
1996, 37, 2181.
(38) Brzezinski, L. J .; Leahy, J . W. Tetrahedron Lett. 1998, 39, 2039.
(39) Roush, W. R.; Grover, P. T. Unpublished research.
(40) Kira, M.; Kobayashi, M.; Sakurai, H. Tetrahedron Lett. 1987,
28, 4081.
(41) Hayashi, T.; Matsumoto, Y.; Kiyoi, T.; Ito, Y.; Kohra, S.;
Tominaga, Y.; Hosomi, A. Tetrahedron Lett. 1988, 29, 5667.
(42) Kira, M.; Hino, T.; Sakurai, H. Tetrahedron Lett. 1989, 30, 1099.
(43) Hosomi, A.; Kohra, S.; Ogata, K.; Yanagi, T.; Tominaga, Y. J .
Org. Chem. 1990, 55, 2415 and literature cited therein.
(44) Kira, M.; Sato, K.; Sakurai, H. J . Am. Chem. Soc. 1990, 112,
257.
Resu lts a n d Discu ssion
Develop m en t of th e Ch ela te-Con tr olled (Z)-Cr ot-
ylsilyla tion Rea ction . (Z)-Crotyltrifluorosilane 24 with
(47) Deiters, J . A.; Holmes, R. R. J . Am. Chem. Soc. 1990, 112, 7197.
(48) Kira, M.; Sato, K.; Sakurai, H.; Hada, M.; Izawa, M.; Ushio, J .
Chem. Lett. 1991, 387.
(49) Iseki, K.; Kuroki, Y.; Takahashi, M.; Kobayashi, Y. Tetrahedron
Lett. 1996, 37, 5149.
(50) Denmark, S. E.; Coe, D. M.; Pratt, N. E.; Griedel, B. D. J . Org.
Chem. 1994, 59, 6161.
(51) Denmark, S. E.; Fu, J . J . Am. Chem. Soc. 2000, 122, 12021.
(52) Nakajima, M.; Saito, M.; Shiro, M.; Hashimoto, S.-i. J . Am.
Chem. Soc. 1998, 120, 6419.
(45) Kobayashi, S.; Nishio, K. Tetrahedron Lett. 1993, 34, 3453.
(46) Chuit, C.; Corriu, R. J . P.; Reye, C.; Young, J . C. Chem. Rev.
1993, 93, 1371.
(53) Iseki, K.; Kuroki, Y.; Takahashi, M.; Kishimoto, S.; Kobayashi,
Y. Tetrahedron 1997, 3513.
J . Org. Chem, Vol. 68, No. 4, 2003 1321

Chemler and Roush
TABLE 1. Cr otyla tion of â-Hyd r oxya ld eh yd e 38a a n d (Z)-Cr otyltr iflu or osila n e (24): Op tim iza tion of Rea ction
Con d ition s
entry
conditionsa
yield (%)
43a :44a
1b
2
3
4
5
1-7 equiv of 24, 1-5 equiv of i-Pr2NEt or Et3N, CH2Cl2, 0 °C, 24-36 h
1.5 equiv of 24, 1.5 equiv of i-Pr2NEt or 2 equiv of H2O, CH2Cl2, 0 °C, 24-36 h
5 equiv of 24, 5 equiv of i-Pr2NEt, 4 Å molecular sieves, CH2Cl2, -10 °C, 36 h
3 equiv of 24, 3 equiv of i-Pr2NEt, 4 Å molecular sieves, CH2Cl2, 0 °C, 36 h
3 equiv of 24, 3 equiv of i-Pr2NEt, 4 Å molecular sieves, toluene, 0 °C, 36 h
3 equiv of 24, 3 equiv of i-Pr2NEt, 4 Å molecular sieves, THF, 0 °C, 36 h
40-75
24
36
75
n.d.
n.d.
75:25 to 95:5d
10:90
95:5
95:5
84:16c
76:24c
6
a
b
All reactions were run at 0.08 M concentration with respect to 38a . Reactions run at several different concentrations. c Sum of two
d
minor diastereomers. Variable from run to run.
an isomeric purity of 98:2 was prepared in two steps from
1,3-butadiene by using the procedure reported by Kira
(82% overall yield).42 R-Methyl-â-hydroxy aldehydes 37-
41 were prepared from the homoallylic alcohol precursors
32-366 by a two-step procedure involving olefin dihy-
droxylation63 (OsO4) and oxidative diol cleavage (NaIO4).
The crude â-hydroxy aldehydes so prepared were very
clean by H NMR analysis and were used directly in the
crotylation reactions without purification. In some cases,
the aldehydes were prepared by ozonolysis of the ho-
moallylic alcohol precursors with pyridine as an additive
to decompose the ozonide.64 The â-hydroxy aldehydes
were unstable to chromatographic purification, which
invariably provided dehydration products and/or epimers
at the R-center. The â-hydroxy aldehydes were also
unstable with respect to dimerization (cf., 38a to 42)
when stored neat at 0 or 23 °C.65 Therefore, 37-41 were
used immediately after preparation in the subsequent
crotylation reactions.
Initial studies of the (Z)-crotylation reaction of alde-
hyde 38a and (Z)-crotyltrifluorosilane (24) in the presence
of a tertiary amine base revealed that the desired
anti,anti-adduct 43a was formed as the major product.
However, product yields at the outset of these investiga-
tions were low (ca. 40%) and the stereoselectivity was
variable, with the ratio of 43a :44a ranging from 75:25
to 95:5 (Table 1, entry 1). Adventitious water was
identified as being responsible for the variable stereo-
selectivity in these initial reactions. Indeed, when the
reaction of 38a and 24 was performed in the presence of
added water (2 equiv), the 3,4-syn-4,5-anti dipropionate
44a was produced as the major product (d.s. 90:10), with
only minor amounts of the 3,4-anti-4,5-anti dipropionate
43a formed (Table 1, entry 2). It is known that water
can act as a nucleophile to catalyze the reactions of
aldehydes with allyl- and crotyltrifluorosilanes.66 The 3,4-
syn-4,5-anti stereochemistry of 44a indicates that the
1
(54) Iseki, K.; Mizuno, S.; Kuroki, Y.; Kobayashi, Y. Tetrahedron
Lett. 1998, 39, 2767.
(55) Iseki, K.; Mizuno, S.; Kuroki, Y.; Kobayashi, Y. Tetrahedron
1999, 55, 977.
(56) Cozzi, P. G.; Tagliavini, E.; Umani-Ronchi, A. Gazz. Chim. Ital.
1997, 127, 247.
(57) Denmark, S. E.; Fu, J . Org. Lett. 2002, 4, 1951.
(58) Denmark, S. E.; Fu, J . J . Am. Chem. Soc. 2001, 123, 9488.
(59) Denmark, S. E.; Wynn, T. J . Am. Chem. Soc. 2001, 123, 6199.
(60) Shimada, T.; Kina, A.; Ikeda, S.; Hayasi, T. Org. Lett. 2002, 4,
ASAP.
(61) Chemler, S. R.; Roush, W. R. J . Org. Chem. 1998, 63, 3800.
(62) Chemler, S. R.; Roush, W. R. Tetrahedron Lett. 1999, 40, 4643.
(63) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem.
Rev. 1994, 94, 2483.
(64) Hung, D. T.; Nerenberg, J . B.; Schreiber, S. L. J . Am. Chem.
Soc. 1996, 118, 11054.
(65) Rychnovsky, S. D.; Skalitzky, D. J . J . Org. Chem. 1992, 57,
4336.
(66) Kira, M.; Sato, K.; Sakurai, H. J . Am. Chem. Soc. 1988, 110,
4599.
1322 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
water-facilitated reaction occurs predominantly through
the normal chairlike Zimmerman-Traxler transition
state 45,4,5 and not the internally chelated, bicyclic
transition state 46. Fortunately, the deleterious effect of
water on the reaction stereoselectivity can be minimized
by rigorously drying the aldehyde substrate in vacuo (ca.
0.1 atm) for 1 h prior to use and by treating a solution of
the aldehyde in CH2Cl2 with 4 Å molecular sieves prior
to the addition of the crotylsilane reagent and the amine
base. Typically, we have used an amount of sieves
equivalent (by weight) to the amount of â-hydroxy
aldehyde employed in the reaction. Under these condi-
tions, a 95:5 mixture of diastereomers 43a and 44a was
obtained consistently for reactions performed in CH2Cl2
(entries 3 and 4). Use of 4 Å molecular sieves also had a
beneficial effect on the yield of homoallylic alcohol,
possibly by decreasing competitive aldehyde dimer (e.g.,
42) formation if the reaction mixture is wet.
hydrolyze the intermediate silylene ketals (e.g., 47) which
are observed in these reactions.
In cases where the â-hydroxyl group is protected, the
reaction proceeds very slowly and preferentially provides
the 3,4-syn homoallylic alcohol product. For example, the
reaction of (Z)-crotylsilane 24 and aldehyde 48 with the
â-hydroxyl group protected as a p-methoxybenzyl (PMB)
ether proceeded to only 17% conversion after 36 h and
gave a 67:17:8:8 mixture of products, among which the
3,4-syn-4,5-anti adduct 49 predominated. The stereo-
chemistry of 49 is consistent with its formation through
a conventional anti-Felkin Zimmerman-Traxler transi-
tion state analogous to 45.
The reaction concentration also proved be an important
factor in obtaining the crotylation products in good yield.
A concentration of 0.08 M with respect to the aldehyde
was optimal. Poor yields were obtained at concentrations
greater than 0.1 M, an observation likely due to competi-
tive formation of aldehyde dimer, e.g. 42, at higher
reaction concentrations.
Results of reactions of (Z)-crotyltrifluorosilane 24 and
R-methyl-â-hydroxy aldehydes 37-41 are summarized in
Table 2. The crotylation reactions of the 2,3-anti alde-
hydes 37, 38, and 41 were generally quite selective for
the 3,4-anti-4,5-anti dipropionate products 43, 50, and
52 (entries 1-5). The reaction diastereoselectivity and
yield were affected only slightly with changes in the
aldehyde δ-alkoxy protecting group (entries 2-4) and
γ-carbon stereochemistry (entries 1 and 3). When the
γ-carbon of the aldehyde substrate is unsubstituted, as
with 41, the diastereoselectivity is somewhat diminished
(e.g., compare entries 3 and 5).
Surprisingly, a different pattern of stereoselectivity
emerged in the reactions of the 2,3-syn-R-methyl-â-
hydroxy aldehydes 39 and 40. In these cases, the reac-
tions were much less selective and the major products
54 and 56 possess 3,4-syn-4,5-anti stereochemistry (Table
2, entries 6 and 7). All attempts to obtain the desired
anti,anti-dipropionate products 55 and 57 from 39 and
40 by changing reaction conditions were unsuccessful.
The stereochemistry of the two major products (entries
6 and 7) in these cases is consistent with reactions
proceeding by way of the usual Zimmerman-Traxler
transition state analogous to 45,4,5 and not the bicyclic
transition states (like 46) that we had targeted in these
experiments. Additional studies designed to probe the
surprising divergence of stereoselectivity of the (Z)-
crotylation reactions of the 2,3-anti and 2,3-syn aldehydes
are presented subsequently.
Syn th esis of th e C(7)-C(16) F r a gm en t of Zin co-
p h or in . The anti,anti,anti,anti-stereopentad unit span-
ning the C(8)-C(12) fragment of zincophorin67,68 provided
an ideal target for demonstrating the utility of our new
methodology for the synthesis of anti,anti-dipropionates.
We targeted 58, which had been previously synthesized
by Danishefsky and co-workers en route to their total
The stoichiometry of the (Z)-crotyltrifluorosilane rela-
tive to the â-hydroxyaldehyde affected the reaction
efficiency, but not the stereoselectivity. In general, 3
equiv of 24 was necessary to obtain optimal yields (Table
1, entry 4). The reaction temperature also influenced the
efficiency, as a considerably lower yield of homoallylic
alcohols was obtained at -10 °C compared to reactions
performed at 0 °C (Table 1, entries 3 and 4). Of the three
solvents examined, the reaction stereoselectvity was
much better in CH2Cl2 (selectivity ) 95:5, Table 1 entry
4) than in toluene (84:16, entry 5) or THF (76:24, entry
6).
A sequential acidic (1 N HCl, 15 min) and then basic
(3:1 THF:1 N NaOH, 1 h) workup was required to
(67) Brooks, H. A.; Gardner, D.; Poyser, J . P.; King, T. J . J . Antibiot.
1984, 37, 1501.
(68) Schade, W.; Grafe, U.; Radics, L.; Incze, M.; Ujszaszy, K. J .
Antibiot. 1984, 37, 836.
J . Org. Chem, Vol. 68, No. 4, 2003 1323

Chemler and Roush
TABLE 2. Rea ction s of Ch ir a l â-Hyd r oxy-r-m eth yl Ald eh yd es 37-41 a n d (Z)-Cr otyltr iflu or osila n e
a
All reactions were conducted with 3 equiv of 24 and 3 equiv of i-Pr2NEt in CH2Cl2 at 0 °C in the presence of 4 Å molecular sieves for
36 h, unless noted otherwise. Yields are reported for the mixture of homoallylic alchol products. c Product ratios were determined by H
NMR analysis (400 or 500 MHz) of the crude reaction mixture. In all cases, products were separable by chromatographic methods. Product
b
1
d
ratio refers to the two major products whose structures are shown; the third figure is the sum of the other two product diastereomers
from each reaction. e Stereochemical assignments for all homoallylic alcohol products are summarized in the Supporting Information.f 7
g
equiv of 24 and 5 equiv of i-Pr2NEt were used. The reaction time was 72 h in this case.
synthesis of the natural product.69 After our work on this
problem was completed, the same intermediate was
synthesized by Marshall via double asymmetric reactions
of chiral allenyltin reagents.70
as the starting material for the synthesis of 58. Protection
of the 1,3-diol unit as an acetonide followed by olefin
dihydroxylation afforded an inconsequential mixture of
diastereomeric diols (94% yield, selectivity ) 87:13) that
was subsequently treated with NaIO4, thereby providing
the sensitive aldehyde 59 in 90% yield. We had hoped
originally that the synthesis of 58 would be easily
accomplished by treatment of 59 with Grignard reagent
63.71 However, in our hands this reaction proceeded in
very poor yield with 1-3 equiv of 63. For example,
treatment of 59 with 3 equiv of 63 and 3 equiv of CeCl3
in THF from -78 to 23 °C provided the targeted carbonyl
addition product in only 4% yield. After our work,
Marshall accomplished this conversion by using a very
large excess of the Grignard reagent.70 Interestingly, the
reaction of 59 with allyltributylstannane and BF3‚Et2O
in toluene at -78 °C provided an 83:17 mixture of
products (94% yield) among which the anti-Felkin dias-
tereomer predominated! At this juncture, we used the
(69) Danishefsky, S. J .; Selnick, H. G.; Zelle, R. E.; DeNinno, M. P.
J . Am. Chem. Soc. 1988, 110, 4368.
(70) Marshall, J . A.; Palovich, M. R. J . Org. Chem. 1998, 63, 3701.
(71) Bu¨chi, G.; Wu¨est, H. J . Org. Chem. 1969, 34, 1122.
Diol 50a , which we had already prepared with 93:6:1
selectivity via the (Z)-crotylation of 37a (Table 2), served
1324 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
vinyllithium reagent derived from vinylstannane 60.
Thus, treatment of 60 with n-BuLi in THF at -90 °C
followed by addition of 59 afforded an 86:14 mixture of
61 (Felkin) and 62 in a combined yield of 73%. The
sensitive allylic alcohol 61 was hydrogenated over Pd/C
in benzene. Protection of the hydroxyl group as a benz-
yloxymethoxy (BOM) ether then completed our synthesis
of the C(7)-C(16) segment 58 of zincophorin. The ster-
eochemistry of our sample was assigned by comparison
of the spectroscopic data for 58 kindly provided by Prof.
Danishefsky. This synthesis provides unequivocal con-
firmation of the stereochemistry of 50a .
divergent behavior of the 2,3-anti- vs 2,3-syn-â-hydroxy
aldehydes, we examined the reactions of R-methyl-â-
hydroxy aldehydes 37a , 38a , 39, and 40 with the allyl-
and (E)-crotyltrifluorosilane reagents 68 and 27. On the
basis of the results of the reactions of 2,3-anti aldehydes
with (Z)-crotyltrifluorosilane 24 (vide supra), we pre-
dicted that the 4,5-anti-adducts 72 and the 3,4-syn-4,5-
anti adducts 73 would emerge as the major products of
allylation of 2,3-anti aldehydes of type 69.
Ster eod iver gen t Beh a vior of 2,3-a n ti a n d 2,3-syn -
r-Meth yl-â-h yd r oxy Ald eh yd es in Rea ction s w ith
Allyl- a n d (E)-Cr otyltr iflu or osila n es. The (Z)-croty-
lation reactions of the 2,3-anti-R-methyl-â-hydroxy alde-
hydes 37a with (Z)-crotyltrifluorosilane 24 provides the
anti,anti-dipropionate stereotriad 50a with high selectiv-
ity (Table 2). The formation of the 3,4-anti-4,5-anti
stereochemistry of adduct 50a can be rationalized by
invoking the bicyclic transition state 64 where C-C bond
formation occurs anti to the aldehyde R-methyl group.
However, the (Z)-crotylation reactions of 2,3-syn-R-
methyl-â-hydroxy aldehyde 39 under comparable condi-
tions gave much more complex product mixtures. The
major products (54 and 55) of this reaction possess 3,4-
syn stereochemistry, suggesting that the crotylation of
39 (and of other 2,3-syn-â-hydroxy aldehydes such as 40)
proceeds via the normal Zimmerman-Traxler transition
states 65 and 66,4,5 rather than the anticipated bicyclic
transition state 67. Undefined at present is the identity
of the nucleophilic promotor, X, of the reactions of the
2,3-syn aldehydes that proceed by way of 65 and 66, but
as will be shown subsequently the free alcohol unit of 39
is a potential candidate, as is adventitious water or
fluoride ion deriving from the reagent or a reaction
intermediate.
Allyl- and (E)-crotyltrifluorosilanes 68 and 27 are
readily available from allyltrichlorosilane precursors by
silyl chloride to silyl fluoride exchange, using antimony
trifluoride.42,72 Allyltrichlorosilane is commercially avail-
able (Aldrich), while isomerically pure (E)-crotyltrichlo-
rosilane was prepared from (E)-crotyl chloride following
Kira’s procedure.42
Reactions of â-hydroxy aldehydes 37a , 38a , 39, and 40
with the allyl- and (E)-crotyltrifluorosilanes 68 and 27
To probe the generality of this allylation process and
also to gain additional insight into the striking stereo-
(72) Mironov, V. F. Bull. Acad. Sci. USSR (Engl. Trans.) 1962, 1797.
J . Org. Chem, Vol. 68, No. 4, 2003 1325

Chemler and Roush
TABLE 3. Rea ction s of Ch ir a l â-Hyd r oxy-r-Meth yl Ald eh yd es w ith Allyltr iflu or osila n e (68) a n d
(E)-Cr otyltr iflu or osila n e (27)a ,b
a
All reactions were conducted with 3 equiv of 68 or 27 and 3 equiv of i-Pr2NEt in CH2Cl2 at 0 °C in the presence of 4 Å molecular
sieves for 36 h, unless noted otherwise. b Yields are reported for the mixture of homoallylic alchol products. c Product ratios were determined
by 1H NMR analysis (400 or 500 MHz) of the crude reaction mixture. In most cases, products were separable by chromatographic methods.
d
Product ratio refers to the two major products whose structures are shown; the third figure is the sum of the other two product
diastereomers from each reaction. e Stereochemical assignments for all homoallylic alcohol products are summarized in the Supporting
Information. f The reaction time in this case was 48 h. This reaction was performed for 7 equiv of 27 and 5 equiv of i-Pr2NEt for 72 h.
g
were performed with the conditions determined to be
optimal for the (Z)-crotylation reactions of these sub-
strates (Table 3). The reactions of 2,3-anti aldehydes 37a
and 38a with reagents 68 and 27 were quite selective
(90:10 to 95:5 d.s.) for the 4,5-anti adducts 74 and 51a
from 37a (entries 1 and 2), and 77 and 44a from 38a
(entries 3 and 4). The stereochemistry of the major
products of these reactions is consistent with pathways
involving bicyclic transition states 70 and 71.
Because we had previously observed that adventitious
water could easily divert the (Z)-crotylation reaction of
the 2,3-anti aldehyde 38a with 24 to produce the 3,4-
syn-4,5-anti diastereomer 44a in preference to the desired
3,4-anti-4,5-anti 43a (Table 1), we suspected that the 2,3-
syn aldehydes might be more sensitive to the presence
of an added nucleophile, much more so than the reactions
of the 2,3-anti aldehydes. However, attempts to improve
the stereoselectivity of the (Z)- or (E)-crotylation reactions
of the 2,3-syn aldehydes by using scrupulously purified
reagents and solvents were unproductive.
These observations led us to consider the possibility
that the aberrant behavior of the 2,3-syn aldehydes 39
and 40 in these allylation reactions could be explained
by assuming that the hydroxyl group of one molecule of
39 and 40 activates the reagent, and that the resulting
pentacoordinate allylsilicate intermediate then reacts
with a second molecule of 39 and 40 via an intermolecu-
lar transition state (e.g., 65, 66, or 89, where -X in these
structures is a second molecule of the â-hydroxy aldehyde
substrate).
In contrast to the excellent results in the allylation
reaction of aldehydes 37a and 38a , allylation reactions
of 2,3-syn aldehydes 39 and 40 with allyltrifluorosilane
68 were virtually nonselective (Table 3, entries 5 and 7).
The (E)-crotylations of 39 and 40 were reasonably selec-
tive for 82 and 86 (entries 6 and 8). However, the 3,4-
anti stereochemistry of 82 and 86 suggests that they arise
via Felkin-selective Zimmerman-Traxler transition state
89,4-6 and not by way of the bicyclic transition structure
88.
This hypothesis was probed by performing a series of
competition experiments.73,74 In one experiment, a 1:1
mixture of the 2,3-anti-â-hydroxy aldehyde 37a and PMB
ether 48 (1 equiv of each) was treated with 1 equiv of
(73) Frye, S. V.; Eliel, E. L.; Cloux, R. J . Am. Chem. Soc. 1987, 109,
1862.
(74) Paquette, L. A.; Mitzel, T. M. J . Am. Chem. Soc. 1996, 118,
1931.
1326 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
(Z)-crotyltrifluorosilane 24 under standard conditions.
This provided a 94:6 mixture of 50a and 51a in 59% yield.
Only a 5% yield of products was obtained from the
crotylation of the PMB ether-protected aldehyde 48, of
which 49 possessing 3,4-syn-4,5-anti stereochemistry
predominated (9:1 d.s.). It is striking that the selectivity
of the products deriving from 37a was the same under
the conditions of this competition experiment as in the
experiments summarized in Table 2, again implicating
the involvement of t.s. 64 in the major pathway leading
to 50a . Moreover, the ratio of the two major products
deriving from 48 is comparable to that obtained in
experiments performed with 48 alone (vide supra). The
3,4-syn-4,5-anti stereochemistry of 49 again implicates
the involvement of the conventional anti-Felkin Zimmer-
man-Traxler transiton state 65. Importantly, this ex-
periment reveals that 37a reacts considerably faster than
48, and that the presence of the other component in the
reaction mixture had little, if any, effect on the product
distribution.
of 90 with 24. This is illustrated in the following diagram,
using the generalized â-hydroxy aldehyde 92 as the
substrate. Complexation of 92 to the crotyltrifluorosilane
24 provides intermediate 93. The data suggest that
reaction by way of t.s. 67 with internal chelation of the
aldehyde carbonyl and intramolecular crotyl transfer is
quite slow, since 94 is a minor product of these experi-
ments. Alternatively, if the silicon center of 93 prefer-
entially coordinates with a second equivalent of the
aldehyde substrate, the opportunity then exists for a
conventional crotylation to proceed by way of 65. We
would expect that aldehydes 39 and 90 would be com-
parably reactive with 93, as is borne out by the experi-
mental results. This hypothesis is also consistent with
knowledge that external nucleophiles such as CsF, ROH,
phosphonamides, etc., are viable nucleophilic catalysts
for aldehyde allylation reactions with allyltrifluoro-
silanes.33,34,40,42,45,49,50,52-56,66
Strikingly different results were obtained when a 1:1
mixture of the 2,3-syn-â-hydroxy aldehyde 39 and the
corresponding PMB ether 90 was treated with 1 equiv
of (Z)-crotyltrifluorosilane 24. This reaction provided a
ca. 50:20:20:10 mixture of the four diastereomers deriving
from 39 in 26% yield, among which 54 predominated. In
addition, four products (ratio ) 55:31:10:4, of which 91
predominated) deriving from 90 were obtained in com-
bined 22% yield. The 3,4-syn-4,5-anti stereochemistry of
the major adducts 54 and 91 indicates that both alde-
hydes react substantially through the anti-Felkin Zim-
merman-Traxler transition state 65.
On th e Or igin of th e Ster eod iver gen t Beh a vior
of 2,3-a n ti- a n d 2,3-syn -r-Meth yl-â-h yd r oxy Ad e-
h yd es in Th eir Allyla tion Rea ction s w ith Allyl- a n d
Cr ot ylt r iflu or osila n es. The data summarized in the
preceding sections of this paper are consistent with the
proposal that the 2,3-syn aldehydes 39 and 40 react with
That both aldehydes, 39 and 90, react at comparable
rates under these conditions is consistent with the thesis
that 39 promotes the reactions of 39 and 24, as well as
J . Org. Chem, Vol. 68, No. 4, 2003 1327

Chemler and Roush
TABLE 4. NMR Da ta for Ald eh yd e-P h SiF 3 Com p lexes
1H NMR, δ
13C NMR, δ
carbonyl
sample
formyl H
Ha
Hb
J
Ha-Hb, Hz
19F NMRb
2,3-anti 37a
complex 98a
2,3-syn 39
9.80
9.73
9.64
9.65
2.59
2.82
2.58
2.65
3.76
4.42
4.11
4.78
5.1
5.9
5.3
4.5
205.5
203.4
204.4
202.9
-117.7c
complex 100a
-117.6c
a
b
Generated by treatment of 37a or 39 with PhSiF3 (1 equiv), i-Pr2NEt (1 equiv), 4 Å molecular sieves in CD2Cl2, 23 °C. The 19F
resonance of PhSiF3 in CD2Cl2 appears at -141.0 ppm. c Fluorine chemical shift (δ) relative to CFCl3.
(Z)-24 and (E)-27 preferentially by way of the conven-
tional Zimmerman-Traxler transition structures 65 and
89, respectively, whereas the 2,3-anti-â-hydroxy alde-
hydes 37a and 38a react with (Z)-24 and (E)-27 by way
of the bicyclic transition states 64 and 71. However, it is
not obvious a priori why the bicyclic transition states 67
and 88 for the crotylation reactions of the 2,3-syn-â-
hydroxy aldehydes are so much less favorable than 64
and 71 for the 2,3-anti-â-hydroxy aldeyhydes. It is
conceivable that the stereodivergent behavior of the 2,3-
anti and 2,3-syn-â-hydroxy aldehydes is due to the
inability of the 2,3-syn-â-hydroxy aldehydes to form the
required six-centered chelate, 96, owing to the require-
ment that one alkyl substituent (Me or R′) must be axial
in the chelate. However, it is also possible that the
problem resides with nonbonded interactions that de-
velop in the bicyclic transition states 67 or 88.75 We
considered the latter to be most probable, since it is likely
that the C-C bond forming event is rate determining.
Nevertheless, we decided to probe these possibilities by
performing NMR studies of model aldehyde-fluorosilane
complexes, and by molecular modeling studies of the
reaction transition states.
comparable to that of the allyl- and crotyltrifluorosilanes
24, 27, and 68, but will not react with the aldehydes.
Key NMR data for aldehydes 37a and 39 and their
complexes with PhSiF3 are summarized in Table 4. On
the basis of these data, we conclude that the aldehyde-
PhSiF3 chelates 99 and 101 are not observable by NMR
spectroscopy. Significantly, the 13C chemical shifts of the
carbonyl carbons of the complexed species appear within
2 ppm of those of the starting aldehydes, indicating that
the aldehyde is not complexed to the silane reagent (one
would expect a significant downfield shift of the aldehyde
carbonyl carbon if complexation occurred).76-80 Moreover,
the J a,b coupling constants of the complexed species are
inconsistent with cyclic structures. This is especially
apparent for the complex formed between PhSiF3 and the
2,3-anti aldehyde 37a , where a large coupling constant
(9-12 Hz) would be expected for a structure such as 99;
J a,b ) 5.9 Hz was observed instead. The 1H and 13C NMR
spectra of these complexes showed little temperature
dependence over a 23 to -60 °C range.
The 19F chemical shifts of the aldehyde-PhSiF3 com-
plexes appear as broad singlets at -117.7 and -117.6
ppm for the 2,3-anti and 2,3-syn aldehyde-PhSiF3 com-
plexes, respectively. This is a substantial change relative
to the 19F resonance for PhSiF3, which is a sharp singlet
at -141.0 ppm, and appears as a broad singlet at -140.8
ppm in the presence of i-Pr2NEt (19F δ relative to CFCl3).
The 19F signals at ca. -117 ppm are consistent with the
formation of hypervalent silicate species. For example,
the 19F resonance of the pentacoordinate silicate [(C3H7)]-
[SiF4C4H9] appears at -116.6 ppm in CH2Cl2.82 We could
not ascertain from these spectra if the observed species
(98 and 100) are pentacoordinate or hexacoordiante
silicates as the 19F signals of hexacoordinate silicates are
also reported to appear in this region at 23 °C.83,84
Additionally, we were unable to observe F-F coupling
NMR Stu d ies. Reports by Keck76-78 and Denmark79,80
suggested that it might be possible to determine if
chelates 99 or 101 are stable, observable intermediates.
1
We examined the H, 13C, and 19F NMR spectra of the
complexes formed between the 2,3-anti and 2,3-syn
aldehydes 37a and 39 with phenyltrifluorosilane (Ph-
SiF3)81 in the presence of i-Pr2NEt. Phenyltrifluorosilane
was used in this study since its Lewis acidity should be
(75) We thank a referee of our inital communication for suggesting
that we attempt to differentiate between these possibilities.
(76) Keck, G. E.; Castellino, S. J . Am. Chem. Soc. 1986, 108, 3847.
(77) Keck, G. E.; Castellino, S.; Wiley: M. R. J . Org. Chem. 1986,
51, 5478.
(78) Keck, G. E.; Castellino, S. Tetrahedron Lett. 1987, 28, 281.
(79) Denmark, S. E.; Almstead, N. G. Tetrahedron 1992, 48, 5565.
(80) Denmark, S. E.; Almstead, N. G. J . Am. Chem. Soc. 1993, 115,
3133.
(82) Klanberg, F.; Muetterties, E. L. Inorg. Chem. 1968, 7, 155.
(83) Marat, R. K.; J anzen, A. F. Can. J . Chem. 1977, 55, 1167.
(84) Marat, R. K.; J anzen, A. F. J . Chem. Soc., Chem. Commun.
1977, 671.
(81) Swamy, K. C. K.; Chandrasekhar, V.; Harland, J . J .; Holmes,
J . M.; Day, R. O.; Holmes, R. R. J . Am. Chem. Soc. 1990, 112, 2341.
1328 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
at temperatures ranging from 23 to -60 °C, which could
have been a useful indicator of the geometry at the silicon
center; presumably inter- or intramolecular fluoride
exchange processes are rapid under the observation
conditions.83,84
ethane conformations are also pointed out when ob-
served, as these interactions generally cost a molecular
conformation ca. 3 kcal/mol.
Molecu la r Mod elin g Stu d ies. We employed semiem-
pirical molecular modeling calculations to aid in our
analysis. The calculations were performed to compare the
relative energies required for formation of the 2,3-syn-
and 2,3-anti-â-hydroxy aldehyde chelates, should they be
discrete reaction intermediates, and to investigate the
relative contributions that nonbonded interactions at the
transition states make to the stereodivergent behavior
of the 2,3-syn and 2,3-anti aldehydes in the crotylsilyla-
tion reactions.
The MNDO/d semiempirical basis set85,86 in the Spar-
tan87 molecular modeling software package was used for
these calculations because it contains parameters for
hypervalent silicon atoms that take into account the
participation of d orbitals. Although the absolute activa-
tion energies calculated by semiempirical methods are
not expected to be meaningful, comparisons of the
calculated activation energies of several related processes
(i.e., determination of relative activation energies) should
be more reliable due to cancellation of systematic errors.
The lengths of the six bonds involved in each of the
transition state cores were fixed (according to Houk’s
“fixed core method”),88 using bond lengths obtained by
Kira et al.,48 who performed ab initio molecular orbital
calculations with the Gaussian 8289 and HONDO90
programs using the MIDI-4(*)91,92 basis set to locate the
transition state for the reaction of tetrafluoroallylsilicate
with formaldehyde. Kira et al. did not include the
counterion in their calculated transition structure, and
likewise, we also ignored the counterion (i-Pr2NEtH+) in
our calculations, with the assumption that the counterion
would affect each transition state in a similar manner,
and thus would be canceled out when we compared
relative transition state activation energies.
The transition states were approximated with the
Linear Synchronous Transit method, which interpolates
geometry between that of the reactant and that of the
product. The transition states were then minimized by
fixing the core, using the bond lengths derived from
Kira’s transition state calculation48 and minimizing the
geometry about the core. The bond angles in the transi-
tion state core were allowed to vary in the calculations.
The lowest energy transition state for each case was
located by calculating the energies of a family of transi-
tion states containing different aldehyde side-chain rota-
mers. In the following transition state structures, HsH
distances less than 2.3 Å are highlighted to point out
destabilizing van der Waals interactions, and eclipsed
The allylation reactions of 2,3-anti and 2,3-syn-â-
hydroxy aldehydes were modeled for the formation of the
desired 4,5-anti adducts, 106 and 111, using the struc-
turally simplified aldehydes 102 and 107. The silicate
intermediates 103 and 108 served as the ground states
for these reactions (as supported by our NMR studies,
which show that these species are stable intermediates).
The activation energies were calculated from the
differences in energy between the pentacoordinate silicate
intermediates 103 and 108 and the intramolecular,
bicyclic transition states 105 and 110 (vide infra). Ad-
ditionally, the difference in energy between the penta-
coordinate silicates 103 and 108 and the chelate inter-
mediates 104 and 109 gave us the energies required for
chelate formation.
The lowest energy conformations for the 2,3-anti and
2,3-syn aldehyde ground states 103 and 108 are shown
in Figure 1. In these structures, axial placement of the
aldehyde fragment (coordinated to silicon via the hy-
droxyl group) and equatorial placement of the allyl group
about the trigonal bipyramidal silicate center provided
the lowest energy arrangement. The calculated energies
for the ground state conformations of 103 and 108 are
within 0.2 kcal/mol of each another.
(85) Thiel, W.; Voityuk, A. A. Theor. Chim. Acta 1992, 81, 391.
(86) Thiel, W.; Voityuk, A. A. Int. J . Quantum Chem. 1992, 44, 807.
(87) Wavefunction, Inc.: Irvine, CA 92612, 1993-1997.
(88) Houk, K. N.; Paddon-Row, M. N.; Rondan, N. G.; Wu, Y.-D.;
Brown, F. K.; Spellmeyer, D. C.; Metz, J . T.; Li, Y.; Loncharich, R. J .
Science 1986, 231, 1108.
(89) Binkley, J . S.; Frisch, M. J .; Whiteside, R. A.; DeFrees, D. J .;
Rahgavachari, K.; Whiteside, R. A.; Schlegel, H. B.; Fluder, E. M.;
Pople, J . A. GAUSSIAN 82; Carnegie-Mellon University: Pittsburgh,
PA.
Results of calculations of the chelated intermediates
and transition states involved in the allylation reactions
of the 2,3-anti aldehyde 102 and the 2,3-syn aldehyde
107 are shown in Figure 2.
In transition state 105 for the 2,3-anti aldehyde, which
leads directly to adduct 106, the only potentially desta-
(90) Dupuis, M.; Watts, J . D.; Villar, H. O.; Hurst, G. J . B. HONDO,
7.0 ed.; IBM Corporation, Scientific Engineering Computations: King-
ston, NY 12401; QCPE No. 544, Indiana University: Bloomington, IN.
J . Org. Chem, Vol. 68, No. 4, 2003 1329

Chemler and Roush
tion. The corresponding boatlike transition state, which
is higher in energy for most aldol and allylation reac-
tions,88,93,94 was calculated to be 3.0 kcal/mol higher in
energy than 105. However, the chelate ring bends away
from the allyl group in this transition state, and adopts
an overall boatlike conformation. This trend is consistent
for all of the bicyclic allylation transition states that we
calculated.
The energy required for chelate formation with the 2,3-
anti aldehyde was calculated as 26.6 kcal/mol by taking
the difference in energy between chelate 104 and ground
state 103. In chelate 104 (and in all other of the chelate
models that we calculated) the silicon-carbonyl oxygen
distance was fixed at 2.3 Å to simulate a loose association
(for reference, the Si-O silyl ether bond length in 104 is
1.88 Å). The Si-O distance had to be fixed because the
modeling program did not accept direct bonding between
these atoms.48
F IGURE 1. Minimized structures of 103 and 108.
In the bicyclic transition state 110 for the reaction of
the 2,3-syn-â-hydroxy aldehyde 107 with allyltrifluoro-
silane, which led to the 4,5-anti adduct 111, the chelate
ring must bend away from the allyl group to avoid steric
interactions with the allyl group of the allylsilane, thus
bringing the C(2) methyl and the C(3) isopropyl groups
closer together. To relieve steric interactions, the C(3)
isopropyl group rotates, causing the C(3)-C(4) bond to
become eclipsed. The C(4) hydrogen is within 2.17 Å of
one of the hydrogens of the C(2) methyl group. Addition-
ally, the allyl group bends away from the chelate ring,
bringing the formyl hydrogen and one of the hydrogens
on the allylsilane γ-C within 2.26 Å of one another. The
activation energy for this reaction was calculated to be
70.1 kcal/mol. The energy required for the 2,3-syn alde-
hyde to adopt the chelate structure 109 was calculated
as 29.0 kcal/mol.
A comparison of the calculated activation energies for
the allylation reactions proceeding by way of bicyclic
transition structures 105 and 110 reveals that the
activation energy for the allylation reaction of the 2,3-
syn aldehyde is 6.0 kcal/mol greater than that of the 2,3-
anti aldehyde. In comparing the calculated energies of
chelation of the two aldehydes, we find that the chelate
formation with the 2,3-syn aldehyde 107 requires 2.1
kcal/mol more energy than does chelate formation for the
2,3-anti aldehyde 102.
In an analogous manner, the chelation and the activa-
tion energies for the reactions of aldehydes 102 and 107
with the (E)- and (Z)-crotyltrifluorosilane reagents were
calculated. The results of these calculations closely
parallel the reactions of these aldehydes with the allyl-
trifluorosilane reagent (see Table 5). A more complete
discussion of the calculations of the transition states
involving the (E)- and (Z)-crotyltrifluorosilane reagents
is provided in the Supporting Information.
These calculations are consistent with the conclusion
that the stereodivergent behavior of the two aldehyde
classes is the result of the different levels of destabilizing
nonbonded interactions experienced in the bicyclic tran-
sition states. However, because the stereochemical out-
F IGURE 2. (a) Intramolecular reaction of allyltrifluorosilane
68 with a 2,3-anti-â-hydroxy aldehyde 102. (b) Intramolecular
reaction of allyltrifluorosilane 68 with 2,3-syn-â-hydroxy al-
dehyde 107.
bilizing interaction observed is between the aldehyde
formyl hydrogen and one of the allylsilane Cγ-hydrogens
(2.29 Å separation in this structure). The energy of
activation in this reaction was calculated to be 64.1 kcal/
mol (Eact ) E105 - E103). The atoms involved in bond
formation and bond breaking adopt a chairlike orienta-
(91) Tatewaki, H.; Huzinaga, S. J . Comput. Chem. 1980, 1, 205.
(92) Sakai, Y.; Tatewaki, H.; Huzinaga, S. J . Comput. Chem. 1981,
2, 100.
(93) Li, Y.; Houk, K. N. J . Am. Chem. Soc. 1989, 111, 1236.
(94) Spellmeyer, D. C.; Houk, K. N. J . Org. Chem. 1987, 52, 959.
1330 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
TABLE 5. Com p a r ison of th e Ca lcu la ted Ch ela tion a n d
Activa tion En er gies (k ca l/m ol) for th e Allyla tion a n d
Cr otyla tion Rea ction s of â-Hyd r oxy Ald eh yd es via
Ch ela ted Bicyclic Tr a n sition Sta tes
reagent
68, allyl
27, (E)-crotyl
24, (Z)-crotyl
∆Echelate(anti - syn)
∆Eact(anti - syn)
-2.1
-1.9
-1.9
-6.0
-3.7
-6.4
come of the allylation/crotylation reactions is determined
by competition between the bicyclic and conventional
Zimmerman-Traxler pathways, we also used semi-
emperical calculations to estimate the activation energies
for the latter processes for both aldehyde classes. Again,
it must be noted that these calculations are qualitative
in nature, as they are based on the fixed core of Kira’s
allylation transiton state.48 These calculations only ac-
count for differences in enthalpy. The loss of entropy in
a bimolecular reaction would be another factor that could
disfavor the conventional Zimmerman-Traxler path-
ways.
The Zimmerman-Traxler allylation reactions of 2,3-
anti and 2,3-syn aldehydes with allyltrifluorosilane were
modeled by the reactions of the â-methoxy aldehydes 112
and 113 with the methoxyallyltrifluorosilicate 114. In
reality, these reactions probably take place between two
equivalents of the silicate-aldehyde complex, e.g. 103,
one to serve as the aldehyde, the other to serve as the
allylating reagent. For ease of molecular modeling of the
transition states, however, the simplified â-methoxy
aldehydes 112 and 113 and the simplified methoxy
allylsilane 114 were used. The transition states leading
to the Felkin adducts 115 (derived from 112) and 116
(derived from 113) were calculated. The activation ener-
gies were calculated from the difference between the
Zimmerman-Traxler transition states and the sum of
the ground states for the two reactants (e.g., Eg.s. ) E114
+ E112). The ground-state energy for the 2,3-anti-â-
methoxy aldehyde 112 was calculated to be 1.6 kcal/mol
lower than that calculated for the 2,3-syn aldehyde 113.
F IGURE 3. (a) Reaction of allyltrifluorosilane with 2,3-anti-
â-methoxy aldehyde 112. (b) Reaction of allyltrifluorosilane
with 2,3-syn-â-methoxy aldehyde 113.
Results from the modeling of the Zimmerman-Traxler
transition states are shown in Figure 3. No destabilizing
van der Waals or eclipsed interactions were observed in
the Felkin transition state 117 for the 2,3-anti aldehyde
112, and the activation energy for this reaction was
calculated to be 70.0 kcal/mol [Eact ) E117 - (E112 + E114)].
Also, no destabilizing interactions were observed in the
Felkin transition state 118 for the 2,3-syn-â-methoxy
aldehyde 113. The activation energy for this reaction was
calculated to be 69.5 kcal/mol [Eact ) E118 - (E113 + E118)],
which is only 0.5 kcal/mol lower than the activation
energy calculated for the analogous reaction for the 2,3-
anti-â-methoxy aldehyde 112.
In an analogous manner, the activation energies for
the reactions of the 2,3-anti- and 2,3-syn-â-methoxy
aldehydes 112 and 113 with the (E)- and (Z)-methoxy-
trifluorocrotylsilicates 119 and 120 were calculated (for
additional discussion complete with figures, see Support-
ing Information). A summary of the activation energies
for the traditional allylation reactions of the 2,3-anti and
2,3-syn aldehydes with allyl- and (E)- and (Z)-methoxy-
trifluorosilicates is provided in Table 6.
These computational results indicate that the small
energetic difference in the Zimmerman-Traxler medi-
ated allylation reactions of the two aldehydes 112 and
113 is not a significant contributor toward their stereo-
divergent behavior that was observed.
Several general trends have appeared in these calcula-
tions. By comparing the activation energies of the bicyclic,
internally chelated allylation/crotylation reactions of the
2,3-anti- and 2,3-syn-â-hydroxy aldehydes, we find that
in each case the 2,3-anti aldehydes have lower activation
energies than do the 2,3-syn aldehydes with energetic
differences ranging from 3.7 to 6.4 kcal/mol (Table 5). On
J . Org. Chem, Vol. 68, No. 4, 2003 1331

Chemler and Roush
TABLE 6. Com p a r ison of th e Activa tion En er gies
(k ca l/m ol) for th e Zim m er m a n -Tr a xler Med ia ted
Allyla tion Rea ction s
from this exercise is that the 2,3-anti-â-hydroxy alde-
hydes are excellent substrates for this reaction, but that
the 2,3-syn-â-hydroxy aldehydes are not. Our computa-
tional analysis suggests that in the productive bicyclic
transition state, one ring will prefer to adopt a chairlike
orientation while the other ring will adopt a boatlike
conformation to avoid transannular nonbonded interac-
tions between them. The boatlike conformation of the
chelate ring brings the C(2) and C(3) aldehyde substit-
uents closer together, a situation that is more easily
accommodated by the 2,3-anti aldehydes than by the 2,3-
syn-â-hydroxy aldehydes. The latter interaction destabliz-
es the chelated bicyclic transition state for the allylation/
crotylation reactions of the 2,3-syn-â-hydroxy aldehydes,
thereby allowing the traditional Zimmerman-Traxler-
type transition states to be competitive in such cases.
Ideally, a study that would separate the intrinsic
selectivity of the R- and â-stereocenters of the aldehyde
could assist in an analysis of the influence of each
stereocenter on these bicyclic transition states. Although
we did not study substrates with a single stereocenter
at these positions, analogous examples are available in
the literature. For example, the effect of the R-methyl
stereocenter on the course of the reaction is demonstrated
in an analogous reaction where an R-chiral titanium (Z)-
(O)-enolate reacts with isobutyraldehyde to generate the
aldol adduct 122 as the predominant isomer. The ster-
eochemistry of 122 is best rationalized as having been
created through the bicyclic transition state 123.96
silicate
∆Eact(anti - syn)
114, allyl
119, (E)-crotyl
212, (Z)-crotyl
0.5
1.6
3.0
the other hand, comparison of the calculated activation
energies of the Zimmerman-Traxler mediated reactions
of the 2,3-anti and 2,3-syn aldehydes with the allyl-
(crotyl)trifluorosilane reagents shows that the reactions
of the 2,3-syn aldehydes have lower activation energies
than the 2,3-anti aldehydes, but with a smaller energetic
preference of 0.5-3.0 kcal/mol (Table 6). Additionally, we
have found that the 2,3-anti aldehydes more easily adopt
chelate structures, e.g. 104, than do the 2,3-syn aldehydes
(e.g., 109), with chelate formation being 1.9-2.1 kcal/mol
less endothermic in the latter cases. However, compari-
son of the energies required for formation of the chelates
(e.g., 26.6 kcal/mol to form chelate 104 from ground state
103) vs the energies of activation for the overall reaction
(e.g., 64.1 kcal/mol to transition state 105 from ground
state 103) reveals that, according to these calculations,
chelation is not the rate determining step, and therefore
differences in the rates of chelate formation of the 2,3-
syn vs 2,3-anti aldehydes cannot account for the stereo-
divergent behavior of the two aldehyde classes. It thus
appears that the most significant factor in the stereodi-
vergent behavior of 2,3-anti vs 2,3-syn aldehydes in their
reactions with allyl- and crotyltrifluorosilanes arises from
the different degrees of strain or nonbonded interactions
experienced by the two aldehydes in the bicyclic allylation
transition states.
These models indicate that in a 6,6-bicyclic transition
state (e.g. 105 and 110, Figure 2), one ring will prefer to
adopt a chairlike orientation while the other ring will
adopt a boatlike conformation to avoid transannular
nonbonded interactions between them. The boatlike
conformation of the chelate ring brings the C(2) and C(3)
aldehyde substituents closer together, a situation that
is more easily accommodated by the 2,3-anti aldehydes
than by the 2,3-syn aldehydes. These models may be
predictive of other designed 6,6-bicyclic transition states,
e.g. other chelate-controlled allylation or aldol reactions.95
The intrinsic preference of the â-stereocenter of our
aldehydes may be interpolated from the analogous reac-
tions of â-hydroxy ketones with allyl- and crotylboronic
acids.38 As illustrated in the equation below, the major
product of these reactions contains the 1,3-anti diol
relationship, which is consistent with having emerged
from the bicyclic transition state 128.
Con clu sion
The initial objectives of this study focused on the
development of a concise, efficient solution to the problem
posed by the anti,anti-dipropionate stereotriad (cf., 16).
As the science progressed, we were led to examine in
detail the fundamental properties of the proposed bicyclic
transition state 64. The general rule we have gleaned
(95) The transition states are likely to be very sensitive to the
aldehyde substitution pattern. Most of the reactions reported herein
and all of the calculations were performed with â-branched aldehydes.
It is possible that unbranched aldehydes will experience fewer non-
bonded interactions in the transition states, with different selectivities
than the examples discussed here.
(96) Luke, G. P.; Morris, J . J . Org. Chem. 1995, 60, 3013.
1332 J . Org. Chem., Vol. 68, No. 4, 2003

Stereoselective Synthesis of anti,anti-Dipropionate Stereotriads
Taken together, these examples indicate that from a
simple, linear analysis (if one were to add the effects of
the R- and â-stereocenters on the product outcome), the
2,3-syn aldehydes should be more amenable to the bicyclic
transition state than the 2,3-anti aldehydes. But this is
not the case. This phenomena has also been observed in
a recent example involving syn- and anti-R-methyl-â-
hydroxy titanium enolates.96 Although the authors were
unable to offer a mechanistic explanation for their
results, they observed a stereodivergence between these
two enolate classes: the enolate generated from the 2,3-
anti ethyl ketone 129 selectively reacted with isobutryral-
dehyde to provide the aldol adduct 130, while the enolate
derived from the 2,3-syn ketone 132 gave a mixture of
the four aldol adducts 133-136.96 We can analyze these
results in light of our data if we assume that the key
steric interactions felt in these bicyclic transition states
are valid within a small range of bond lengths and bond
angles as these factors will vary depending upon the type
of atoms invoved in the transition state core. Transition
state 131 for the 2,3-anti enolate derived from 129 should
be favored for the creation of aldol adduct 130, whereas
an analogous transition state for the 2,3-syn enolate,
which would lead to adduct 134, is unfavorable due to
the nonbonded interactions that would exist between the
R- and â-substitutents of the enolate in the bicyclic
transition state.
Finally, the analysis presented herein is also relevant
to the results of tandem aldol-allylation reactions, using
strained silacycles recently reported by Leighton.97
Ack n ow led gm en t. Financial support provided by
the National Institute of General Medical Sciences (GM
38436) is gratefully acknowledged. We thank Dr. Marty
Pagel (Indiana University) for assisting with the MO-
PAC calculations, Prof. Ted Widlanski (Indiana Uni-
versity) for helpful discussions, and Prof. Mark Banaszak-
Holl (University of Michigan) for use of his Spartan
program. We also thank Roxanne Kunz and Dustin
Mergot for assistance with the preparation of this
manuscript.
Su p p or tin g In for m a tion Ava ila ble: Complete experi-
mental details, stereostructure proofs for allylation products,
details of MOPAC calculations, and copies of 1H NMR spectra.
This material is available free of charge via the Internet at
http://pubs.acs.org.
J O0267908
(97) Wang, X.; Meng, Q.; Nation, A. J .; Leighton, J . L. J . Am. Chem.
Soc. 2002, 124, 10672.
J . Org. Chem, Vol. 68, No. 4, 2003 1333