Diastereoselectivity in Organometallic Additions to the Carbonyl Group of Protected Erythrulose Derivatives

Article abstract of DOI:10.1021/jo9716744

We have investigated a number of nucleophillic additions to L-erythrulose derivatives (4-12) bearing protective O-silyl, O-benzyl, and O-trityl groups in various relative positions. The results are discussed in the frame of chelated vs nonchelated transit

Full text of DOI:10.1021/jo9716744

698
J . Org. Chem. 1998, 63, 698-707
Dia ster eoselectivity in Or ga n om eta llic Ad d ition s to th e Ca r bon yl
Gr ou p of P r otected Er yth r u lose Der iva tives
J . Alberto Marco,*,1a Miguel Carda,*,1b Florenci Gonza´lez,1b Santiago Rodr´ıguez,1b
Encarna Castillo,1b and J uan Murga1b
Departamento de Quı´mica Orga´nica, Universidad de Valencia, E-46100 Burjassot, Valencia, Spain, and
Departamento de Quı´mica Inorga´nica y Orga´nica, Universidad J aume I, E-12080 Castello´n, Spain
Received September 9, 1997
We have investigated a number of nucleophillic additions to L-erythrulose derivatives (4-12) bearing
protective O-silyl, O-benzyl, and O-trityl groups in various relative positions. The results are
discussed in the frame of chelated vs nonchelated transition states with additional support of
previously published theoretical calculations. Sound evidence appears to exist for the formation
of R-chelates as the key intermediates in nucleophillic additions to these R,â-dioxygenated ketones.
Since such evidence is still lacking in the case of â-chelates, proposals of their intermediacy have
been relegated in favor of the more solid Felkin-Anh model, which predicts the same stereochemical
result. The behavior of these highly functionalized ketones does not always match that of
structurally similar aldehydes.
In tr od u ction
Carbohydrates, very particularly monosaccharides,
represent one of the most convenient chiral sources in
the synthesis of enantiopure compounds.2 The ke-
totetrose L-(S)-erythrulose and derivatives3 of general
formula 1 (R′, R′′, R′′′ ) protecting groups) are useful
additions to the list of chiral precursors of this type. The
carbonyl group of 1 is a prominent site for the appendage
of additional carbon fragments via nucleophillic addition.
The selection of suitable protecting groups is important,
as they will exert a control on the steric course of
nucleophillic additions.4 The literature contains a great
deal of studies on stereoselective additions of carbon
nucleophiles to polyoxygenated aldehydes.5 However, the
corresponding behavior of highly functionalized ketones
such as 1 is known with much less detail.3d,6 In two
preliminary reports, we have described the stereochem-
ical outcome of organometallic additions to the carbonyl
groups of 1-O-silylated erythrulose 3,4-acetonides7a (1, R′,
R′′ ) CMe2; R′′′ ) silyl) and 3,4-di-O-benzyl derivatives7b
(1, R′, R′′ ) benzyl, Bn; R′′′ ) silyl). More recently, the
diastereoselective additions to erythrulose 1,3-O-eth-
ylidene acetals7c have also been reported. The diaste-
reoselectivity of these reactions proved dependent on the
type of hydroxyl protecting group. In the present paper,
we describe in full7d the results of the diastereoselective
reactions of several nucleophillic reagents with a range
of variously protected L-erythrulose derivatives 1. The
(1) (a) Universidad de Valencia. (b) Universidad J aume I, Castello´n.
(2) (a) Vasella, A. In Modern Synthetic Methods 1980; Scheffold, R.,
Ed.; Salle & Sauerla¨nder Verlag: Frankfurt, 1980; pp 173-267. (b)
Fraser-Reid, B.; Anderson, R. C. Prog. Chem. Org. Nat. Prod. 1980,
39, 1-61. (c) Hanessian, S. Total Synthesis of Natural Products: the
Chiron Approach; Pergamon Press: Oxford, 1983. (d) Inch, T. D.
Tetrahedron 1984, 40, 3161-3213. (e) Lichtenthaler, F. W. In Modern
Synthetic Methods 1992; Scheffold, R., Ed.; VHCA.-VCH: Basel, 1992;
pp 273-376. (f) Ferrier, R. J .; Middleton, S. Chem. Rev. 1993, 93,
2779-2831. (g) Monneret, C.; Florent, J .-C. Synlett 1994, 305-318.
(h) Collins, P.; Ferrier, R. J . Monosaccharides. Their Chemistry and
Their Roles in Natural Products; J ohn Wiley and Sons: New York,
1995. (i) Bols, M. Carbohydrate Building Blocks; J ohn Wiley and
Sons: New York, 1996.
(5) (a) J urczak, J .; Pikul, S.; Bauer, T. Tetrahedron 1986, 42, 447-
488. (b) Huryn, D. M. In Comprehensive Organic Synthesis; Trost, B.
M., Fleming, I., Schreiber, S. L., Eds.; Pergamon Press: Oxford, 1991;
Vol. 1, pp 49-75. (c) Lipshutz, B. H. In Comprehensive Organic
Synthesis; Trost, B. M., Fleming, I., Schreiber, S. L., Eds.; Pergamon
Press: Oxford, 1991; Vol. 1, pp 107-138. (d) Ferreri, C.; Palumbo,
G.; Caputo, R. In Comprehensive Organic Synthesis; Trost, B. M.,
Fleming, I., Schreiber, S. L., Eds.; Pergamon Press: Oxford, 1991; Vol.
1, pp 139-172. (e) Saccomano, N. A. In Comprehensive Organic
Synthesis; Trost, B. M., Fleming, I., Schreiber, S. L., Eds.; Pergamon
Press: Oxford, 1991; Vol. 1, pp 173-209. (f) Knochel, P. In Compre-
hensive Organic Synthesis; Trost, B. M., Fleming, I., Schreiber, S. L.,
Eds.; Pergamon Press: Oxford, 1991; Vol. 1, pp 211-229. (g) Imamoto,
T. In Comprehensive Organic Synthesis; Trost, B. M., Fleming, I.,
Schreiber, S. L., Eds.; Pergamon Press: Oxford, 1991; Vol. 1, pp 231-
250. (h) Shambayati, S.; Schreiber, S. L. In Comprehensive Organic
Synthesis; Trost, B. M., Fleming, I., Schreiber, S. L., Eds.; Pergamon
Press: Oxford, 1991; Vol. 1, pp 283-324. (i) Yamaguchi, M. In
Comprehensive Organic Synthesis; Trost, B. M., Fleming, I., Schreiber,
S. L., Eds.; Pergamon Press: Oxford, 1991; Vol. 1, pp 325-353.
(3) (a) De Wilde, H.; De Clercq, P.; Vandewalle, M.; Ro¨per, H.
Tetrahedron Lett. 1987, 4757-4758. (b) Van der Eycken, E.; De Wilde,
H.; Deprez, L.; Vandewalle, M. Tetrahedron Lett. 1987, 4759-4760.
(c) Marco, J . L. J . Chem. Res. (S) 1988, 276, (M) 2013-2016. (d)
Nagano, H.; Ohno, M.; Miyamae, Y.; Kuno, Y. Bull. Chem. Soc. J pn.
1992, 65, 2814-2820. (e) Dequeker, E.; Compernolle, F.; Toppet, S.;
Hoornaert, G. Tetrahedron 1995, 51, 5877-5890.
(4) Chen, X.; Hortelano, E. R.; Eliel, E. L.; Frye, S. V. J . Am. Chem.
Soc. 1992, 114, 1778-1784.
S0022-3263(97)01674-5 CCC: $15.00 © 1998 American Chemical Society
Published on Web 01/22/1998

Diastereoselectivity in Organometallic Additions
J . Org. Chem., Vol. 63, No. 3, 1998 699
Sch em e 1. Syn th esis of Er yth r u loses 10 a n d 11
Sch em e 2. Ster eoselective Ad d ition of
Or ga n om eta llic Rea gen ts to Er yth r u lose
Der iva tives a n d P ossible Tr a n sition Sta tes
protecting groups were selected as to either favor (Bn)
or to disfavor (tert-butyldiphenylsilyl, TPS, and trityl, Tr)
the formation of metal chelates,4 thus directing the
process toward either a cyclic Cram chelation or to a
nonchelation Felkin-Anh pathway.8 The preparation of
1-O-silylated L-erythrulose derivatives 4-8 has recently
been described.9a Ketones 9-11, which bear a 1-O-trityl
group (see comments below), have been obtained as
reported in the literature3c or as depicted in Scheme 1
from the known compounds 139a and 14.10
Aside from the carbonyl group, molecules 1 contain
three additional, potentially complexing oxygen atoms in
R, R′, and â positions. Our purpose was to determine to
which extent the stereoselectivity of nucleophillic addi-
tions would be controlled by either Cram type chelation
mechanisms involving any of these oxygen atoms or by
transition states of the Felkin-Anh type. If either the
R- or R′-oxygen atom enters complex formation, a five-
membered chelate is formed (see Scheme 2), whereas
involvement of the â-oxygen atom leads to the formation
of a six-membered chelate. The R-chelate is predicted
to react from the less hindered Si side of the carbonyl
group (TS-1), leading predominantly to stereoisomer 1A,
which we here will arbitrarily name the syn stereoiso-
mer.11 The alternative anti stereoisomer 1B, derived
from attack to the Re side, is expected to be the major
product when a â-chelate is the key intermediate (TS-
(6) Chikashita, H.; Nakamura, Y.; Uemura, H.; Itoh, K. Chem. Lett.
1992, 439-442.
(7) (a) Carda, M.; Gonza´lez, F.; Rodr´ıguez, S.; Marco, J . A. Tetra-
hedron: Asymmetry 1992, 3, 1511-1514. (b) Carda, M.; Gonza´lez, F.;
Rodr´ıguez, S.; Marco, J . A. Tetrahedron: Asymmetry 1993, 4, 1799-
1802. (c) Carda, M.; Casabo´, P.; Gonza´lez, F.; Rodr´ıguez, S.; Domingo,
L. R.; Marco, J . A. Tetrahedron: Asymmetry 1997, 8, 559-577. (d)
Gonza´lez, F., Ph.D. Thesis, University J aume I, Castello´n, 1995.
(8) (a) Anh, N. T. Top. Curr. Chem. 1980, 88, 145-162. (b) Reetz,
M. Top. Curr. Chem. 1982, 106, 1-54. (c) Reetz, M. T. Angew. Chem.,
Int. Ed. Engl. 1984, 23, 556-569. (d) Reetz, M. T. Acc. Chem. Res
1993, 26, 462-468. For the importance of chelation in ketone
reductions, see, for example: Oishi, T.; Nakata, T. Acc. Chem. Res 1984,
17, 338-344. Iida, H.; Yamazaki, N.; Kibayashi, C. J . Org. Chem.
1986, 51, 3769-3771. Taniguchi, M.; Fujii, H.; Oshima, K.; Utimoto,
K. Tetrahedron 1993, 49, 11169-11182. Sarko, C. R.; Guch, I. C.;
DiMare, M. J . Org. Chem. 1994, 59, 705-706. Sarko, C. R.; Collibee,
S. E.; Knorr, A. L.; DiMare, M. J . Org. Chem. 1996, 61, 868-873. For
the role of chelation in other synthetically important processes, see:
Kauffmann, T. Synthesis 1995, 745-755.
2). When a Felkin-Anh transition state is traversed,
stereoisomer 1B should also be preferentially formed (TS-
3, Scheme 2).
Resu lts a n d Discu ssion
It was clear from the beginning of the present work
that, if a highly stereoselective addition was desired, the
oxygen atom of the primary alcohol at C-1 should be
prevented from complexation with the metal atom of the
reagent. Examples of low stereoselectivities due to
competitive complexation at both sides of ketone carbonyl
groups are known,5b,12 and in fact, we found that a very
bulky protecting silyl group12b,c at the C-1 hydroxyl was
necessary for having a high stereoselectivity.7 The
(9) (a) Marco, J . A.; Carda, M.; Gonza´lez, F.; Rodr´ıguez, S.; Murga,
J . Liebigs Ann. Chem. 1996, 1801-1810. (b) Part of the present results
have been taken from the projected Ph.D. Thesis of E. Castillo.
(10) Steuer, B.; Wehner, V.; Lieberknecht, A.; J a¨ger, V. Org. Synth.
1997, 74, 1-11.
(11) The scope of this syn/anti nomenclature is limited to the present
work and has no relation to the use of the same prefixes in other
instances: Heathcock, C. H. In Comprehensive Organic Synthesis;
Trost, B. M., Fleming, I., Schreiber, S. L., Eds.; Pergamon Press:
Oxford, 1991; Vol. 2, pp 181-238.
(12) (a) Annunziata, R.; Cinquini, M.; Cozzi, F.; Fuchicello, A.
Tetrahedron 1991, 47, 3853-3868. (b) Frye, S. V.; Eliel, E. L. J . Am.
Chem. Soc. 1988, 110, 484-489. (c) Bai, X.; Eliel, E. L. J . Org. Chem.
1992, 57, 5166-5172.

700 J . Org. Chem., Vol. 63, No. 3, 1998
Marco et al.
Ta ble 1. Ster eoselective Ad d ition s of Nu cleop h illic
best results in terms of yield and stereoselectivity.
Detrimental changes in the former or in the latter were
consistently observed with deviations from these condi-
tions (changes in temperature or solvent). Other changes
such as addition of Lewis acids (BF3, TiCl4, Me3SiCl, with
the aforementioned exceptions),5h,i cation-sequestering
agents (12-crown-4),5i and metal salts (LiCl, LiClO4,
ZnCl2)5i proved in most cases unproductive or even
deleterious,7 as did prior metal exchange of either orga-
nolithium or organomagnesium reagents with trivalent
lanthanide salts.5g,13e
Ketone 4 showed dr’s which ranged from good to
mediocre.7a Among the reagents assayed, only MeLi and
the cuprate Me2CuLi gave good dr’s of 1A (entries 1 and
6). In contrast, magnesium reagents (entries 3-5)
behaved here in an erratic way and displayed unsatisfac-
tory dr’s. The allyl derivatives of lithium and magnesium
displayed stereoselectivities which are opposite to those
of non-allyl counterparts. The titanium reagent MeTi-
(OiPr)3 gave a quite good dr (entry 7) with predominant
formation of 1B. It is worth mentioning here that all
ketones assayed showed a very low reactivity toward this
sterically crowded reagent (reaction times 2-3 d), which
had to be used without solvent in great excess at room
temperature. It is interesting to note that the stereose-
lectivity of 4 in all the aforementioned reactions does not
match that of the structurally close 2,3-O-isopropylidene
glyceraldehyde.5a,14a
Ketone 5 was, as previously reported,7b more diaste-
reoselective than 4 toward organometallic reagents, most
particularly with Grignard derivatives. With these, very
high dr’s were observed, the major product being 1A,
predicted by Cram’s R-chelation model (entries 13-16).
With Me2Zn/TiCl4 (entry 22) the stereoselectivity was also
very high and had the same sense. In contrast, the sense
of stereoselectivity with organolithium reagents corre-
sponds to a non-R-chelate pathway (entries 10 and 12)
and is thus opposite to that observed with 4. Allylmetal
derivatives showed here low stereoselectivities, except for
allyltrimethylsilane/SnCl4 (entry 18), where practically
only one diastereomer is formed. This result is extremely
interesting from the preparative point of view, since the
allyl group can be transformed later into a broad range
of carbon appendages. Contrary to 4, ketone 5 reacted
with MeTi(OiPr)3 to give mainly the syn isomer 1A (entry
20), expected for an R-chelated transition state. As for
4, ketone 5 differs in its behavior from the structurally
close 2,3-di-O-benzylglyceraldehyde.14a
In relation to one of our research projects,9b we also
investigated the reactions of 4 and 5 with the lithium
enolate of tert-butyl acetate (entries 9 and 23) and the
reaction of 5 with the carbanion formed by deprotonation
of acetonitrile (entry 24). Stereoisomer 1B was in all
cases the major product. Chiral R-alkoxy aldehydes have
been shown to react with lithium enolates in a very
unselective way whereas their â-alkoxy counterparts
often show useful levels of stereoselectivity.8c It is
difficult, however, to say whether chelation may be
involved in the present case, as both mechanistic alterna-
tives have been reported.11
Rea gen ts to Er yth r u lose Der iva tives 4-8a
entry compd
RM
solvent T (°C)/t (h) % yield 1A:1Bb
1
2
3
4
5
6
7
8
9
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
6
6
6
6
6
6
6
7
7
7
7
8
8
8
MeLi
allylLi
MeMgBr
Et2O
THF
Et2O
THF
Et2O
Et2O
neat
C6H6
Et2O
Et2O
Et2O
THF
Et2O
THF
THF
Et2O
Et2O
-78/1
-78/1
0/1
-78/1
-78/1
-78/2
25/60
25/1
-78/1
-78/1
0/1
-78/2
-78/1
-78/1
-78/1
0/1
86
85
88
93
90
80
67
80
78
86
69
87
95
94
90
80
86
80
80
48
94
85
95
70
83
72
81
76
90
46
51
88
60
71
52
80
88
73
86:14
22:78
75:25
80:20
20:80
88:12
9:91
60:40
10:90
26:74
81:19
30:70
>95:5
>95:5
>95:5
93:7
67:33
>95:5
88:12
81:19
50:50
>95:5
20:80
24:76
15:85
>95:5
>95:5
>95:5
91:9
57:43
62:38
14:86
58:42
58:42
14:86
16:84
20:80
15:85
vinylMgBr
allylMgBr
Me2CuLi
MeTi(OiPr)3
AlMe3
LiCH2COOtBu
MeLi
MeLi/TiCl4
allylLi
MeMgBr
EtMgBr
vinylMgBr
ethynylMgBr
allylMgBr
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
-78/1
allylSiMe3/SnCl4 CH2Cl2 -78/6
Me2CuLi
MeTi(OiPr)3
AlMe3
Me2Zn/TiCl4
LiCH2COOtBu
LiCH2CN
MeLi
Et2O
neat
C6H6
-78/2
25/48
25/2
CH2Cl2 -78/0.5
Et2O
THF
Et2O
THF
THF
THF
THF
Et2O
neat
Et2O
Et2O
Et2O
neat
Et2O
Et2O
Et2O
-78/1
-78/1
-78/1
-78/1
-78/1
-78/1
0/1
-78/2
25/48
-78/1
-78/1
-78/2
25/48
-78/1
-40/5
-40/5
MeMgBr
EtMgBr
vinylMgBr
ethynylMgBr
Me2CuLi
MeTi(OiPr)3
MeLi
MeMgBr
Me2CuLi
MeTi(OiPr)3
MeLi
MeMgBr
Me2CuLi
a
In most cases, 3 equiv of the organometallic reagent was added
to the appropriate ketone under the indicated conditions (see
b
Experimental Section for more specific details). Determined by
1H and 13C NMR (dr > 95:5 means that NMR signals from the
minor isomer are not visible).
voluminous TIPS and TPS groups gave rise to the highest
diastereoisomeric ratios (dr’s),7a but the latter was pre-
ferred because of several convenient features such as its
UV absorbance, which facilitated TLC analysis, and its
1
more desirable H NMR properties (no blurring multip-
lets around δ 1 ppm). With all these ideas in mind,
erythruloses 4-8, which contain various combinations
of acetonide, TPSO, and BnO groups at C-3/C-4, were
prepared in enantiopure form9 and submitted to reaction
with a series of different nucleophiles, mainly organo-
metallic reagents. Several organolithium,5b,13a organo-
magnesium,5b,13d organocopper5c,13c and organotitanium5d,13b
derivatives were examined for this purpose. In ketones
4 and 5, the organoaluminum reagent AlMe3 and allyl-
trimethylsilane/Lewis acid were also assayed. The or-
ganozinc reagent Me2Zn was found to be reactive only in
the presence of Lewis acids. The main results are
presented in Table 1. The particular reaction conditions
indicated for each reagent are those giving rise to the
(13) (a) Schlosser, M. In Organometallics in Synthesis; Schlosser,
M., Ed.; J ohn Wiley and Sons: New York, 1994; Chapter 1. (b) Reetz,
M. T. In Organometallics in Synthesis; Schlosser, M., Ed.; J ohn Wiley
and Sons: New York, 1994; Chapter 3. (c) Lipshutz, B. H. In
Organometallics in Synthesis; Schlosser, M., Ed.; J ohn Wiley and
Sons: New York, 1994; Chapter 4. (d) Wakefield, B. J . Organomag-
nesium Methods in Organic Synthesis; Academic Press: London, 1995;
Chapter 6. (e) Imamoto, T. Lanthanides in Organic Synthesis;
Academic Press: London, 1994; pp 80-97.
(14) (a) Mead, K.; Macdonald, T. L. J . Org. Chem. 1985, 50, 422-
424. (b) Reetz, M. T.; Kesseler, K. J . Org. Chem. 1985, 50, 5434-
5436. These results should be compared with those observed for a
sterically encumbered glyceraldehyde derivative, where the nonchela-
tion mode appears to be the main mechanistic pathway: Ley, S. V.;
Woods, M.; Zanotti-Gerosa, A. Synthesis 1992, 52-54.

Diastereoselectivity in Organometallic Additions
J . Org. Chem., Vol. 63, No. 3, 1998 701
communications (Scheme 3).7 In the case of the products
derived from 4 and 5, additional evidence of the correct-
ness of our configurational assignments comes from their
conversion into some naturally occurring compounds.16
For ketones 6-8 and 9-11, the configurations of the
addition products have been determined by chemical
correlation with the known products 20, 25, and their
epimers,7,9b which were identified either pure or as
stereoisomeric mixtures by 1H and 13C NMR spectros-
copy.17
Taking the results of Tables 1 and 2 as a whole, some
general conclusions may be drawn. Isomer 1A, predicted
only by Cram’s R-chelation model, is formed with a very
high dr in several cases, mainly with Grignard reagents
and ketones 5 and 6, where R-chelation is not sterically
impeded by O-substitution (entries 13-16 and 26-29).
It is thus likely that R-chelates are formed in these cases.
There is presently a great deal of experimental evidence4,8d
of the real existence of R-chelates as intermediates in
carbonyl additions, particularly with organomagnesium
reagents.5b A further aspect deserves mention. Although
we have not performed kinetic measurements, we have
observed that these highly stereoselective reactions were
also particularly fast. In most cases, all reactions were
complete within 1 h. We observed, however, that the
reactions corresponding to entries 13-15 and 26-28 were
already complete after only 2 min at -78 °C. The
addition of 3 equiv of Grignard reagent was necessary,
however, for the reaction to occur in such a fast way. With
addition of only 1 equiv, 30% of unreacted ketone was
recovered after 1 h. Furthermore, addition of chelation-
disrupting HMPA to the Grignard reaction mixture
caused both a strong decrease in the reaction rate (10%
unreacted ketone after 7 h at 0 °C) and an almost complete
disappearance of the stereoselectivity (dr ∼ 1.1:1). It is
Ta ble 2. Ster eoselective Ad d ition s of Nu cleop h illic
Rea gen ts to Er yth r u lose Der iva tives 9-11a
entry compd
RM
solvent T (°C)/t (h) % yield 1A:1Bb
1
2
3
4
5
6
7
8
9
9
9
9
MeLi
Et2O
THF
THF
THF
THF
THF
THF
-78/1
-78/1
-78/1
-78/1
-78/1
-78/1
-78/1
0/1
84
95
85
92
86
87
95
80
95
90
90
78:22
77:23
78:22
45:55
>95:5
90:10
>95:5
93:7
MeMgBr
vinylMgBr
MeLi
MeMgBr
EtMgBr
vinylMgBr
ethynylMgBr Et2O
allylMgBr
allylLi
10
10
10
10
10
10
10
11
Et2O
Et2O
THF
-78/1
-78/1
-78/1
60:40
40:60
>95:5
10
11
vinylMgBr
a,b
See footnotes to Table 1.
Ketone 6, where the â-OBn group of 5 is replaced by a
TPSO group, was prepared with the idea that its struc-
ture would allow the formation of an R-chelate but,
because of the bulky silyl moiety, not a â-chelate.8c,d In
the event, 6 behaved in essentially the same way as 5
toward lithium and magnesium reagents (entries 25-
29), with Me2CuLi and MeTi(OiPr)3 being, however, much
less stereoselective (entries 30 and 31). In general terms,
this behavior is similar to, but not completely coincident
with, that of the structurally close 2-O-benzyl-3-O-(tert-
butyldimethylsilyl)glyceraldehyde,14b where MeTi(OiPr)3
gave very predominantly the nonchelation-controlled
product.
In comparison with 6, ketone 7 represents the alterna-
tive situation in which the R-chelate pathway has been
blocked via protection of the corresponding hydroxyl with
a bulky silyl group, leaving the possibility of a â-chelation
still open. Indeed, anti isomer 1B was the major isomer
with both MeLi and MeTi(OiPr)3 (entries 32 and 35) but
Me2CuLi reacted with a unexpectedly5c low stereoselec-
tivity (entry 34).
Erythrulose derivative 8 represents a limiting case as
it has its three hydroxyl functions protected with volu-
minous TPS groups. This excludes any mechanistic
pathway involving chelates. In line with this, the anti
stereoisomer 1B was the major product in the cases
essayed (entries 36-38), in agreement with a Felkin-
Anh transition state. The reactions of 8 were more
sluggish than those of the other ketones and required
higher temperatures and/or reaction times to completion.
This obviously reflects the high steric hindrance of the
carbonyl group in this molecule. For instance, the bulky
reagent MeTi(OiPr)3 did not react at all with 8.
(16) (a) Marco, J . A.; Carda, M.; Gonza´lez, F.; Rodr´ıguez, S.; Murga,
J .; Falomir, E. An. Qu´ım. 1995, 91, 103-112. (b) Marco, J . A.; Carda,
M.; Gonza´lez, F.; Rodr´ıguez, S.; Murga, J . J . Chem. Res. (S) 1996, 1,
(M) 201-205.
(17) The configurations of the products formed after addition of
allylmetal derivatives, acetonitrile, and tert-butyl acetate enolates to
ketone 4 were determined by conversion into lactone i and its epimer
ii, as either the pure compounds or mixtures thereof. Addition
products to ketone 5 were subjected to an analogous sequence of
reactions, and the obtained lactones were then converted into i and/or
ii. These results constitute a part of the projected Ph.D. Thesis of E.C.
The somewhat high price of TPS chloride led us to test
alternative protecting groups such as the cheaper trityl
group,15 which displays a comparable steric size. We then
synthesized erythruloses 9,3c 10, and 11 (Scheme 1) and
tested their reactions with several key organometallic
reagents (Table 2). It was very satisfactory to find that
the dr’s of these reactions were quite similar to those of
their 1-O-silylated counterparts. The reactions with 9
gave similar results as those previously found.3d The
higher sensitivity of the trityl group toward acidic
reagents, however, showed up in extensive decomposition
when 9-11 were allowed to react in the presence of Lewis
acids such as TiCl4.
The configurations of the diastereoisomers formed in
these reactions have been established with the aid of
chemical correlations, described in part in our previous
(15) Greene, T. W.; Wuts, P. G. M. Protective Groups in Organic
Synthesis, 2nd ed.; J ohn Wiley and Sons: New York, 1991; pp 60-62.

702 J . Org. Chem., Vol. 63, No. 3, 1998
Marco et al.
Sch em e 3. Ch em ica l Cor r ela tion s betw een th e Ad d ition P r od u cts of Nu cleop h iles to Keton es 4-12
pertinent remembering here the observations of Eliel and
co-workers4 on organometallic additions to ketones bear-
ing R-OR groups. They found that these ketones reacted
not only much faster when R was a small alkyl group
than when R was a bulky silyl group but also with a
higher diastereoselectivity.18 The same authors did not
find rate enhancements in ketones bearing â-OR groups,
which reacted with similar rates as substrates where
chelation was not possible.4 It thus appears that forma-
tion of 1A through the R-chelate represents the reaction
channel with the lowest activation barrier on both
electronic and steric reasons. The fact that 6 gives
essentially the same dr’s as 5 toward Grignard reagents
further indicates that the â-chelate pathway, even if
feasible, does not participate in these cases to a noticeable
extent. It may be concluded therefore that ketones 5 and
6, which bear R-chelating groups, will react with orga-
nomagnesium reagents exclusively through the R-chelate,
whether or not additional chelation points are offered.
MeLi reacts with a lower stereoselectivity and in the
opposite sense to that displayed by magnesium reagents,
a not unprecedented behavior.5b,12b Whether the reasons
of this apparent absence of an R-chelate involving the
lithium cation are of thermodynamic (low stability) or
kinetic (slow formation) nature remains unknown.
The behavior of 4 is surprising. The selectivities of
Grignard reagents were lower than expected and not
always predictable. Allylmagnesium bromide, for in-
stance, gave mainly 1B, opposite to that predicted by the
R-chelation model. This may be due to the fact that many
allyl-type organometallics react in most instances with
allylic inversion through cyclic transition states of the
metallo-ene type, which do not involve chelation.19 MeLi
gave a high dr of the R-chelation isomer 1A, a not
anticipated result with organolithium derivatives.5b These
findings should be considered in the light of the previous
observations on nucleophillic additions of 2,3-O-iso-
propylideneglyceraldehyde5a,14a and its 2,3-di-O-benzyl
analogue.14a In all cases examined, the acetonide was
clearly the less diastereoselective compound, with the
major stereoisomer being opposite to that predicted by
the R-chelation model. To explain this, it has been
proposed14a that glyceraldehyde acetonide reacts mainly
under nonchelation control because of (a) a too high
chelate energy contents due to appreciable ring strain
and to nonbonded interactions of the acetonide geminal
methyl groups with metal ligands (cf. Scheme 4a), and/
or (b) a depressed donor ability of the acetonide oxygens
(19) (a) Roush, W. R. In Comprehensive Organic Synthesis; Trost,
B. M., Fleming, I., Schreiber, S. L., Eds.; Pergamon Press: Oxford,
1991; Vol. 2, pp 1-53. (b) Yamamoto, Y.; Asao, N. Chem. Rev. 1993,
93, 2207-2293. (c) Yamamoto, Y.; Shida, N. In Advances in Detailed
Reaction Mechanisms; Coxon, J . M., Ed.; J AI Press Inc.: Greenwich,
CT, 1994; pp 1-44.
(18) A parallel increase of reactivity and enantioselectivity has also
been observed in some addition reactions: Zhang, H.; Xue, F.; Mak,
T. C. W.; Chan, K. S. J . Org. Chem. 1996, 61, 8002-8003.

Diastereoselectivity in Organometallic Additions
J . Org. Chem., Vol. 63, No. 3, 1998 703
reactions. For instance, Mead and Macdonald14a found
that cuprates prepared by metal exchange from organo-
lithium reagents did not give the same results as those
similarly prepared from Grignard reagents. It is likely
that the lithium and magnesium cations, which consti-
tute a part of the reagent structure21 and display different
chelation abilities, are in these cases the ionic species
actually involved in chelate formation.5c
Sch em e 4. (a ) Str u ctu r e of Ch ela tes of Keton e 4
a n d P ossible Tr a n sition Sta tes. (b) Tem p ta tive
Con for m a tion s of th e P u ta tive â-Ch ela tes of
Er yth r u lose Der iva tive 7
The case of allyltrimethylsilane is especially interest-
ing. In the presence of tin tetrachloride as a Lewis acid,
this reagent added to ketone 5 (entry 18) to yield
essentially pure stereoisomer 1A (dr > 95:5). With TiCl4
the stereoselectivity was equally high but the yield was
lower. This is consistent with TS-1 in Scheme 2, where
either of these bidentate Lewis acids forms the R-chelate
and the nucleophile (CH2dCHCH2SiMe3) adds subse-
quently to the less hindered carbonyl side. Similar
considerations apply for Me2Zn/TiCl4 (entry 22). In the
presence of these Lewis acids, however, ketone 4 gave a
mixture of ill-defined products which still had a carbonyl
group but lacked the acetonide moiety. Most likely, a
Sakurai-type22 reaction took place between the electro-
phillic acetal carbon and the allylsilane. Substitution of
SnCl4 or TiCl4 for BF3 only led to recovery of the starting
ketone, even in the presence of an excess of the Lewis
acid. In all probability, the complex between the ketone
carbonyl oxygen and the monodentate Lewis acid BF3 is
not stable enough for steric reasons.
Methyltitanium trisisopropoxide is a further reagent
which displays contrasting behavior toward ketones 4-7.
Since this weakly Lewis-acidic reagent is not very prone
to form chelates,5d,13b the formation of 1B from 4 (entry
7) is most likely explained through a Felkin-Anh transi-
tion state. However, 1A is the major product with 5 and,
to a lesser extent, with 6 (entries 20 and 31), which
strongly suggests reaction through an R-chelate. This
is not the usual outcome with this reagent but precedent
has been reported.8c,23 In the case of 5, the reagents
combination MeLi/TiCl4 (entry 11) yields essentially the
same stereoisomeric mixture as MeTi(OiPr)3. Most likely,
initial transmetalation of MeLi and TiCl4 to yield MeTiCl3
takes place, with the latter reagent thus reacting via
chelate formation.5d,13b This constitutes the opposite
result to that observed with MeLi alone (entry 10). The
reaction of 4 with the same reagents mixture would have
been meaningful, but unfortunately, the compound de-
composed under these conditions, perhaps due to sensi-
tivity of the acetal ring to strong Lewis acids (see above).
owing to mutual electron-withdrawing inductive effect.20
In the case of 4, however, these proposals do not explain
why some organolithium and organomagnesium reagents
(entries 1 and 3-4) give predominantly 1A (Scheme 4a).
Rational explanations for these findings are still lacking.
No regular trends are recognizable in the behavior of
the Gilman reagent Me2CuLi. It showed a fairly good
stereoselectivity toward 4 and 5, with 1A being predomi-
nantly formed. In contrast, it proved quite unselective
in its reactions with 6 and 7. Much work has been
reported on stereoselective additions to polyoxygenated
aldehydes, most particularly those having R- and â-alkoxy/
silyloxy groups.5c Addition of the nucleophile to the
corresponding cyclic R- or â-chelate has been invoked in
such cases to explain the stereochemical outcome of the
alkoxy derivatives, whereas the silyloxy compounds were
postulated to react through Felkin-Anh transition states.
Unfortunately, there is almost no precedent concerning
structurally related ketones. Furthermore, it appears
that the way of preparing these copper reagents may
have an influence on the stereoselectivity of their addition
Ketone 7, where â-chelation is the only possible che-
lation mode, reacted with an organocopper reagent in an
almost stereorandom way (entry 34). Grignard reagents
were also unselective. It is possible that the â-chelate,
if formed to any extent, has a weak diastereofacial
preference in the present case. This might be due to a
higher flexibility of the six-membered chelate, when
compared with the more compact and conformationally
constrained five-membered R-chelate. Indeed, the struc-
tures of chelates formed between â-alkoxy aldehydes and
Lewis acid such as TiCl4 or MgBr2 have been studied by
(20) There is some experimental basis which lends support to this
proposal. In the reactions of Grignard reagents with 2-acyltetrahy-
drofurans, which only have an oxygen atom in the ring, high stereo-
selectivities are often observed: Eliel, E. L. In Asymmetric Synthesis;
Morrison, J . D., Ed.; Academic Press: New York, 1983; Vol. 2, Chapter
5. For a further example structurally very close to 4, see: Rao, A. V.
R.; Gurjar, M. K.; Devi, T. R.; Kumar, K. R. Tetrahedron Lett. 1993,
1653-1656.
(21) Krause, N.; Gerold, A. Angew. Chem., Int. Ed. Engl. 1997, 36,
187-204.
(22) Weber, W. P. Silicon Reagents for Organic Synthesis; Springer-
Verlag, Berlin, 1983; pp 173-205.
(23) Reetz, M. T.; Hu¨llmann, M. J . Chem. Soc., Chem. Commun.
1986, 1600-1602.

704 J . Org. Chem., Vol. 63, No. 3, 1998
Marco et al.
Ta ble 3. Ster eoselective Ad d ition s of Gr ign a r d Rea gen ts
to 12a
followed by 1,3 magnesium-to-carbon migration of the
nucleophillic methyl group within the chelate. An ener-
getically favored intervention of R-chelates was predicted
for the R-alkoxy carbonyl compounds, with the observed
stereochemical outcome being in a good agreement with
theoretical conclusions. In contrast, the situation with
â-chelates was much less clear-cut. While the bare
existence of â-chelates seems to be confirmed by various
experimental procedures,4,24 their actual intervention in
addition processes is still dubious. Alternative pathways
not involving chelation may likely become competitive
or even faster in such cases. A particularly interesting
conclusion of this theoretical study is that C-C bond
formation must not necessarily be the rate-limiting step
of the whole process.25a
entry
RM
solvent T (°C)/t (h) % yield 1A:1Bb
1
2
3
EtMgBr
vinylMgBr
ethynylMgBr THF
THF
THF
-78/1
-78/1
-78/1
87
85
84
>95:5
>95:5
87:13
a,b
See footnotes of Table 1.
NMR.24 The authors found that 1:1 complexes of 2-methyl-
3-(benzyloxy)propanal with either of these Lewis acids
are conformationally rigid, with the methyl group oc-
cupying a pseudoequatorial position in a flattened half-
chair. However, when these concepts are translated to
ketone 7, important differences immediately appear
(Scheme 4b). One of the half-chair conformations (7a )
may be destabilized by a strong torsional gauche interac-
tion between the bulky R-OTPS and CH2OTPS groups
(this interaction is negligible in aldehydes, where a
hydrogen atom replaces the latter group). In the alter-
native conformation 7b, similar destabilizing interactions
are expected to arise between the OBn group and,
depending of the spatial orientation of the benzyl moiety,
either the pseudoaxial R-OTPS group or one of the
magnesium ligands. The â-chelate therefore may not be
formed at all here by either thermodynamic (insufficient
stability) or kinetic reasons (too slow formation).4,25a
Moreover, if actually formed, it may not have a marked
diastereofacial bias.
Still more recently,25b ab initio studies on the additions
of the organomagnesium species MeMg+ and MeMgCl (1
or 2 equiv) to some R-alkoxy and R,â-dialkoxy carbonyl
compounds have been made. A particularly interesting
example of the latter compound type was 3,4-di-O-
methyl-1-O-(trimethylsilyl)-L-erythrulose, which is a suit-
able model for the compounds under discussion. In
addition, the organometallic species selected as model
reagents corresponded more closely to those we actually
used. As in the previous instance, the authors found that
an initial exothermic R-chelate formation takes place
without any noticeable energy barrier (solvation was not
considered in this study). Under assumption of a 1:1
ketone/Grignard reagent stoichiometry, rate-limiting C-C
bond formation occurs subsequently via magnesium to
carbon 1,3-methyl transfer. Interestingly, the authors
also found that inclusion of a second molecule of the
Grignard reagent into the reactive complex, i.e., inter-
molecular methyl transfer from one molecule of MeMgCl
to a preformed MeMgCl/carbonyl chelate, causes a de-
crease of the predicted energy barrier. This suggests a
second-order kinetics in the organomagnesium reagent
and may be qualitatively in line with the fact that 3 equiv
of the organomagnesium reagent are necessary for a fast
reaction to occur at -78 °C (see above).26 The stereofacial
outcome predicted by this model (attack from the Si side
of the carbonyl group)25 nicely agrees with our experi-
mental observations. This lends an additional support
to our assumption of the intermediacy of R-chelates in
these additions.
In summary, R,â-dioxygenated ketones will undergo
very rapid and stereoselective organometallic additions
to the carbonyl group provided that (a) the R-oxygen atom
is not prevented by its substituent4,24 from forming a five-
membered chelate and (b) the organometallic reagent
contains a Lewis acidic5h metal atom (e.g., magnesium,
titanium) capable of forming sufficiently stable chelates
of this ring size. Under these conditions, attack by the
nucleophile will take place from the less hindered face
of the chelated molecule, leading in most cases to high
stereoselectivities. No conclusive evidence has been
gathered in the present case, however, to substantiate
the participation of six-membered â-chelates, which may
even not be stable enough to be formed here. Despite
the fact that â-chelates have often been invoked to
explain the stereochemical outcome of nucleophillic ad-
ditions in 1,3-difunctional compounds such as â-oxo
MeLi and MeTi(OiPr)3 reacted with 7 to show the same
dr, with 1B being the major isomer. Similar dr values
are also observed in the reactions of ketone 8, where
chelation cannot take place. It is likely that these
reactions take place through a Felkin-Anh transition
state.
The diastereoselectivities of hydride reductions of the
protected 1-deoxyerythrulose 12 have already been
investigated.14a To make mechanistic comparisons with
erythrulose 5, we have now tested some additions of
carbon nucleophiles. Since, in comparison with 5, only
the bulky nonchelating group OR′′′ at C-1 has been
removed, the stereoselectivities should be similar to those
of the latter ketone. This assumption has actually been
borne out in practice in the case of Grignard reagents.
Ketone 12 was allowed to react with ethyl-, vinyl-, and
ethynylmagnesium bromide, to yield 1A with high dr
values than the stereoisomer, expected from addition to
the R-chelate (Table 3). The configurations of the addi-
tion products were established in each case by chemical
correlation with the known product 35 (Scheme 3).7b,d
Some of the questions discussed above have also been
addressed with the aid of computational methods. In a
recently published ab initio study,25a the additions of Me2-
Mg to several R- and â-alkoxy carbonyl compounds have
been investigated. The authors relied upon the afore-
mentioned kinetic measurements of Eliel’s group on the
reactions of Me2Mg with R-alkoxy ketones,4 where an
overall second-order kinetics (first order in both substrate
and Grignard reagent) was established. They then
postulated an initial, fast and exothermic chelation step,
(24) (a) Keck, G.; Castellino, S. J . Am. Chem. Soc. 1986, 108, 3847-
3849. (b) Keck, G.; Castellino, S.; Wiley, M. R. J . Org. Chem. 1986,
51, 5478-5480.
(25) (a) Mori, S.; Nakamura, M.; Nakamura, E.; Koga, N.; Moro-
kuma, K. J . Am. Chem. Soc. 1995, 117, 5055-5065. (b) Safont, V. S.;
Moliner, V.; Oliva, M.; Castillo, R.; Andre´s, J .; Gonza´lez, F.; Carda,
M. J . Org. Chem. 1996, 61, 3467-3475.
(26) It should be remembered, however, that the complex composi-
tion of Grignard reagent mixtures makes the interpretation of kinetic
measurements a very difficult issue: Ashby, E. C.; Laemmle, J .;
Neumann, H. M. Acc. Chem. Res 1974, 7, 272-280.

Diastereoselectivity in Organometallic Additions
J . Org. Chem., Vol. 63, No. 3, 1998 705
sulfoxides,27 â-oxo amides,28 â-oxygenated aldehydes,29
â-hydroxy ketones,30 â-oxo phosphine oxides,31a â-silyloxy
ketones,31b etc., their actual participation in addition
processes of â-oxygenated carbonyl compounds is still an
object of discussion.4,25a It has recently been proposed
that the stereochemical outcome of nucleophillic additions
to polyalkoxy carbonyl compounds may be controlled by
conformational factors in the ground state.32 Of course,
this stereochemical model might be considered here for
those cases where R-chelation may be excluded. How-
ever, it is difficult to understand on this basis how the
same molecule reacts with different reagents under
similar temperature and solvent conditions to yield very
different dr values. Additional kinetic measurements as
well as high-level theoretical studies will thus be neces-
sary for a deeper understanding of this complex mecha-
nistic frame.
temperature for 18 h. Workup (CH2Cl2) and CC (hexane/
EtOAc 9:1) afforded 15 (3.812 g, 70%).
(S)-1-O-Tr ityl-3,4-d i-O-ben zyl-1,3,4-tr ih yd r oxybu ta n -2-
on e (10) by Sw er n Oxid a tion of 15. Dry DMSO (1.70 mL,
24 mmol) was added under Ar at -60 °C to a solution of oxalyl
chloride (1.05 mL, 12 mmol) in dry CH2Cl2 (30 mL). After the
mixture was stirred at this temperature for 2 min, a solution
of 15 (3.268 g, 6 mmol) in dry CH2Cl2 (30 mL) was added
dropwise. The stirring was further continued for 15 min, and
then Et3N (3.4 mL, 24 mmol) was added, with additional
stirring at -60 °C for 15 min. The temperature was then
increased to 25 °C and the reaction mixture was stirred for 1
h. Workup (CH2Cl2) and CC (hexanes/EtOAc 9:1) afforded 10
(2.54 g, 78%).
(2S ,3S )-1,4-D i -O -t r i t y l-2-O -b e n z y lb u t a n e -1,2,3,4-
tetr ol (16) was prepared from 1410 as described above for 15
but using a double amount of TrCl. Workup (CH2Cl2) and CC
(hexane/EtOAc 9:1) afforded 16 in 70% yield.
(S)-1,4-Di-O-tr ityl-3-O-ben zyl-1,3,4-tr ih yd r oxybu ta n -2-
on e (11) was obtained from 16 in 90% yield as described above
for 10.
(S)-3,4-Di-O-ben zyl-3,4-dih ydr oxybu tan -2-on e (12). Diol
139a (3.024 g, 10 mmol) was dissolved in benzene (50 mL) and
treated with lead tetraacetate (8.87 g, 20 mmol). After being
stirred at room temperature for 30 min, saturated aqueous
NaHCO3 (25 mL) was added. The reaction mixture was then
extracted with CH2Cl2 (4 × 10 mL), and the organic layer was
filtered through Celite and evaporated at reduced pressure.
This yielded crude 2,3-di-O-benzyl-L-glyceraldehyde, which
was dried at reduced pressure and used immediately without
purification for the next step.
The product obtained in the previous step was dissolved
under Ar in dry THF (30 mL) and treated at -78 °C with a
1.6 M solution of MeLi in THF (20 mL, 32 mmol). The reaction
mixture was stirred for 1 h at the same temperature. Workup
(CH2Cl2) and solvent removal at reduced pressure afforded a
crude dibenzylated triol (mixture of two epimers), which was
filtered through a pad of silica gel (elution with hexanes/EtOAc
7:3). Evaporation of the solvent at reduced pressure provided
an oily residue (1.174 g, 41% crude yield), which was used
without further purification in the next step.
The mixture of epimeric alcohols obtained previously was
oxidized as above (15 f 10) by the Swern procedure. Workup
(CH2Cl2) and column chromatography on silica gel (elution
with hexanes/EtOAc 8:2) furnished 12 (995 mg, 35% overall
yield from 13).
Gen er a l Exp er im en ta l P r oced u r es for Or ga n om eta llic
Ad d ition s to Keton es 4-12. Substrate, solvent, tempera-
ture, reaction time, and yield are indicated in Tables 1-3.
Careful exclusion of oxygen and moisture is assumed in all
cases.
(a ) F or or ga n olith iu m , Gr ign a r d r ea gen ts, a n d Me3Al.
A solution of the appropriate ketone (1 mmol) in the indicated
solvent (4 mL) was cooled to the indicated temperature. The
required organometallic reagent (3 mmol) was then added
dropwise, and the reaction mixture was stirred for the
indicated time. Workup (Et2O) and column chromatography
(hexane/EtOAc mixtures) yielded the desired product with the
indicated yield and diastereoisomeric composition.
(b) F or Or ga n om eta llic Ad d ition s in th e P r esen ce of
Lew is Acid s (TiCl4, Sn Cl4, etc.). Meth od A. A solution of
the appropriate ketone (1 mmol) in the indicated solvent (4
mL) was cooled to the indicated temperature. The Lewis acid
(1 mmol) was then added dropwise at the same temperature,
and the reaction mixture was stirred for 15 min. After this,
the organometallic reagent (3 mmol) was added dropwise,
followed by stirring for the indicated time. Workup (Et2O or
CH2Cl2) and column chromatography (hexane/EtOAc mixtures)
yielded the desired product with the indicated yield and
diastereoisomeric composition. Meth od B. Titanium tetra-
chloride (1 mmol) and MeLi (1 mmol) were dissolved in dry
Et2O (2 mL) at 0 °C. After 15 min of stirring, a solution of
the ketone (0.5 mmol) in dry Et2O (4 mL) was added dropwise.
The reaction mixture was then stirred for 1 h at the same
temperature. Workup and column chromatography as above.
Exp er im en ta l Section
Gen er a l. Column chromatography (CC) was performed on
silica gel Su¨d-Chemie AG (50-200 µm) with the mixture of
solvents indicated in each case. Experiments which required
an inert atmosphere were carried out under dry argon (Ar) in
a flame-dried glass system. THF and benzene were freshly
distilled from sodium/benzophenone ketyl and sodium wire,
respectively, and were transferred via syringe. Methylene
chloride was distilled from P2O5 and stored over 4 Å molecular
sieves. DMSO was dried and stored on 4 Å molecular sieves.
Triethylamine was distilled from CaH2. Other commercially
available reagents (Aldrich or Fluka) were used as received:
organometallic reagents were used as solutions in Et2O (MeLi,
MeMgBr), THF (other magnesium reagents), benzene (AlMe3),
or toluene (Me2Zn). Allyllithium was prepared immediately
prior to use by reaction of lithium with allyl phenyl ether.13a
If not detailed otherwise, the workup of the reactions was
consistently performed in the following manner: the reaction
mixture was poured into brine and extracted twice with solvent
(Et2O or CH2Cl2), the organic layer was washed with diluted
acid or base (depending on whether the reaction conditions
were basic or acid, respectively) and then washed again with
brine, the organic layer was dried over anhydrous MgSO4 or
Na2SO4, and the solvent was eliminated with a rotary evapo-
rator at aspirator pressure.
(S)-1-O-Tr it yl-3,4-O-isop r op ylid en e-1,3,4-t r ih yd r oxy-
bu ta n -2-on e (9) was prepared according to ref 3c.
(2S ,3S )-4-O -T r it y l-1,2-d i-O -b e n zy lb u t a n e -1,2,3,4-
tetr ol (15). 1,2-Di-O-benzyl-L-threitol (13)9a (3.024 g, 10
mmol) was dissolved in dry CH2Cl2 (40 mL) and treated under
Ar with Et3N (2.1 mL, 15 mmol), DMAP (15 mg), and TrCl
(3.067 g, 11 mmol). The reaction mixture was stirred at room
(27) Bueno, A. B.; Carren˜o, M. C.; Garc´ıa-Ruano, J . L. An. Quı´m.
1994, 90, 442-451. See also: Page, P. C. B.; Purdie, M.; Lathbury,
D. Tetrahedron Lett. 1996, 8929-8932.
(28) (a) Taniguchi, M.; Fujii, H.; Oshima, K.; Utimoto, K. Bull. Chem.
Soc. J pn. 1994, 67, 2514-2521. (b) Taniguchi, M.; Oshima, K.;
Utimoto, K. Bull. Chem. Soc. J pn. 1995, 68, 645-653.
(29) (a) Marshall, J . A.; Perkins, J . F.; Wolf, M. A. J . Org. Chem.
1995, 60, 5556-5559. (b) Banfi, L.; Guanti, G.; Zannetti, M. T. J . Org.
Chem. 1995, 60, 7870-7878. (c) De Kermadec, D.; Prudhomme, M.
New. J . Chem. 1993, 17, 499-503. (d) Paquette, L. A.; Mitzel, T. M.
Tetrahedron Lett. 1995, 6863-6866. Nonchelation has been postulated
in other cases: Braun, M.; Mahler, H. Liebigs Ann. Chem. 1995, 29-
40.
(30) Garc´ıa-Ruano, J . L.; Tito, A.; Culebras, R. Tetrahedron 1996,
52, 2177-2186. Protection of the free hydroxyl group, however, seems
to favor Felkin-Anh transition states in some cases: Guanti, G.; Banfi,
L.; Riva, R. Tetrahedron 1995, 51, 10343-10360.
(31) (a) Bartoli, G.; Bosco, M.; Sambri, L.; Marcantoni, E. Tetrahe-
dron Lett. 1996, 7421-7424. (b) Bartoli, G.; Bosco, M.; Sambri, L.;
Marcantoni, E. Tetrahedron Lett. 1997, 3785-3788.
(32) (a) Evans, D. A.; Dart, M. J .; Duffy, J . L.; Yang, M. G. J . Am.
Chem. Soc. 1996, 118, 4322-4343. (b) Mulzer, J .; Pietschmann, C.;
Buschmann; Luger, P. J . Org. Chem. 1997, 62, 3938-3943.

706 J . Org. Chem., Vol. 63, No. 3, 1998
Marco et al.
(c) F or Me2Cu Li. CuI (228.5 mg, 1.2 mmol) was flame-
dried under Ar until the appearance of a yellowish color. After
the solution was cooled to 0 °C, Et2O (3.5 mL) was added
followed by MeLi (1.6 M in hexanes, 1.57 mL, 2.5 mmol). The
mixture was then cooled to the indicated temperature and
treated dropwise with a solution of the appropriate ketone (0.4
mmol) in Et2O (2 mL). The reaction mixture was stirred at
the same temperature for the indicated time. Workup (Et2O)
and column chromatography (hexane/EtOAc mixtures) yielded
the desired product with the indicated yield and diastereoi-
someric composition.
(d ) F or MeTi(OiP r )3. A solution of ClTi(OiPr)3 (1 M in
hexanes, 7 mL, 7 mmol) was treated at -50 °C with MeLi (1.6
M in Et2O, 4.4 mL, 7 mmol). The cooling bath was removed,
and the mixture was stirred at room temperature for 2 h. After
this, the solution was filtered under Ar with careful exclusion
of moisture and poured directly into the flask containing the
appropriate ketone (0.5 mmol). The solvent was then elimi-
nated at reduced pressure, and the oily mixture was stirred
at room temperature for the indicated time. Workup (Et2O)
and column chromatography (hexane/EtOAc mixtures) af-
forded the desired product with the indicated yield and
diastereoisomeric composition.
(e) Aceton itr ile a n d ter t-Bu tyl Aceta te En ola tes. LDA
was generated by dropwise addition of BuLi (620 µL of a
solution ca. 1.6 M in hexanes, 1 mmol) to a solution of
diisopropylamine (155 µL, 1.1 mmol) in dry Et2O or THF (10
mL) under Ar at -78 °C. After 15 min of stirring at this
temperature, dry tert-butyl acetate (135 µL, 1 mmol) or
acetonitrile (52 µL, 1 mmol) was added dropwise via syringe
at such a rate that the temperature of the mixture remained
below -70 °C. After further stirring for 50 min, the appropri-
ate ketone (0.4 mmol) dissolved in dry THF (3 mL) was slowly
added dropwise via syringe. The reaction mixture was then
stirred for 1 h at -78 °C. Workup (CH2Cl2) and column
chromatography (hexane/EtOAc mixtures) yielded the desired
product with the indicated yield and diastereoisomeric com-
position.
Gen er a l Desilyla tion P r oced u r e. A solution of the
substrate (1 mmol) in dry THF (15 mL) was treated with solid
tetra-n-butylammonium fluoride trihydrate (275 mg, 1.05
mmol) and stirred at room temperature for 30 min. After
addition of water (1 mL), the volatiles were totally eliminated
at reduced pressure. CC of the residue (hexane/EtOAc mix-
tures) provided the desilylation product. For compounds
having two or three silyl groups, the relative proportion of
TBAF was accordingly increased.
Workup (CH2Cl2) and CC (hexanes/EtOAc mixtures) afforded
the desired hydrolysis product.
Gen er a l Detr ityla tion P r oced u r e. Meth od A.33 A 1.8
M solution of trifluoroacetic acid/trifluoroacetic anhydride was
prepared by dissolving these reagents in the appropriate
amount of dry CH2Cl2. The substrate (0.4 mmol) was then
dissolved under Ar in dry CH2Cl2 (1 mL) and treated dropwise
at room temperature with the aforementioned solution (0.65
mL, ca. 3 equiv). The reaction mixture turned yellow and was
then cooled to 0 °C, followed by addition of triethylamine (0.5
mL, 3.6 mmol). After being stirred for 5 min, the reaction
mixture was poured into MeOH (10 mL). Stirring was
continued for 30 min at room temperature. After removal of
all solvents at reduced pressure, the residue was chromato-
graphed (hexane/EtOAc mixtures) to yield the desired detri-
tylation product. Meth od B. The substrate (0.5 mmol) was
dissolved in a 1:1 MeOH/THF mixture (5 mL) and treated with
concentrated HCl (0.1 mL). The mixture was then stirred at
room temperature for 18 h. Workup (Et2O) and CC (hexanes/
EtOAc mixtures) furnished the detritylation product.
(2R,3S)-3,4-O-Isop r op ylid en e-2-m eth ylbu ta n e-1,2,3,4-
tetr ol (17). The addition product of MeLi or Me2CuLi with
ketone 4 was desilylated with TBAF. CC of the crude desily-
lation product (hexane/EtOAc 1:2) followed by crystallization
from hexane/Et2O provided 17 in 50% overall yield.
(2R,3S)-1,2,3,4-Tetr a -O-ben zyl-2-m eth ylbu ta n e-1,2,3,4-
tetr ol (18). Compound 17 was benzylated under the afore-
mentioned conditions. Workup and CC (hexane/EtOAc 19:1)
provided in 95% yield an oily dibenzyl derivative, which was
then submitted to acetonide hydrolytic cleavage. Workup and
CC (hexane/EtOAc 1:1) yielded an oily diol (95%), which was
then benzylated as above. Workup and CC (hexane/EtOAc 9:1)
afforded finally tetrabenzyl derivative 18 in 65% yield. The
same product was obtained by benzylation of the known diol
19.16a
(2R ,3S )-1,3,4-T r i-O -b e n zy l-2-e t h y lb u t a n e -1,2,3,4-
tetr ol (20) was obtained as depicted in Scheme 3 either as a
mixture with its C-2 epimer (from the addition product of 4
with vinylmagnesium bromide, followed by five standard
transformations) or as a stereoisomerically homogeneous
compound from 22 by sequential desilylation and benzylation
(50% overall yield).
(2R,3S)-1-O-(ter t-Bu tyld ip h en ylsilyl)-3,4-d i-O-ben zyl-
2-m eth ylbu ta n e-1,2,3,4-tetr ol (21), (2R,3S)-1-O-(ter t-bu -
t yld ip h en ylsilyl)-3,4-d i-O-b en zyl-2-et h ylb u t a n e-1,2,3,4-
tetr ol (22), (2R,3S)-1-O-(ter t-bu tyld ip h en ylsilyl)-3,4-d i-O-
ben zyl-2-vin ylbu ta n e-1,2,3,4-tetr ol (23), a n d (2R,3S)-1-O-
(ter t-bu tyldiph en ylsilyl)-3,4-di-O-ben zyl-2-eth yn ylbu tan e-
1,2,3,4-tetr ol (24) were obtained by reaction of 5 with,
respectively, methylmagnesium bromide, ethylmagnesium
bromide, vinylmagnesium bromide, and ethynylmagnesium
bromide. Compound 22 was also obtained by catalytic hydro-
genation of 23 or 24.
Gen er a l Ben zyla tion P r oced u r e: An 80% suspension of
NaH in mineral oil (90 mg, ca. 3 mmol of sodium hydride) was
washed three times under Ar with dry hexane. Dry THF (1
mL) was then added, followed by a solution of the substrate
(1 mmol) in dry THF (3 mL). The solution was stirred at reflux
for 30 min. Benzyl bromide (0.3 mL, 2.5 mmol) was then
added dropwise, followed by nBu4N+ I- (55 mg, 0.15 mmol).
The reaction mixture was then heated at reflux for 90 min.
Workup (Et2O) and CC (hexanes/EtOAc mixtures) furnished
the desired benzylation product. Two free hydroxyl groups are
assumed. For compounds having three free hydroxyl groups,
the proportions of NaH and benzyl bromide were increased to
4 and 3.5 equiv, respectively.
(2R ,3S )-1,3,4-Tr i-O-b e n zyl-2-m e t h ylb u t a n e -1,2,3,4-
tetr ol (25) was obtained as a stereoisomerically homogeneous
compound from 21 by sequential desilylation and benzylation
(70% overall yield). The same product, either pure or as a
mixture with its C2-epimer, was also obtained from addition
products of ketones 6-10 (Scheme 3).
(2R,3S)-1,4-Di-O-(ter t-bu tyld ip h en ylsilyl)-3-O-ben zyl-
2-m eth ylbu ta n e-1,2,3,4-tetr ol (26), (2R,3S)-1,4-d i-O-(ter t-
b u t yld ip h e n ylsilyl)-3-O-b e n zyl-2-e t h ylb u t a n e -1,2,3,4-
tetr ol (27), (2R,3S)-1,4-d i-O-(ter t-bu tyld ip h en ylsilyl)-3-O-
ben zyl-2-vin ylbu ta n e-1,2,3,4-tetr ol (28), a n d (2R,3S)-1,4-
di-O-(ter t-bu tyldiph en ylsilyl)-3-O-ben zyl-2-eth yn ylbu tan e-
1,2,3,4-tetr ol (29) were obtained by reaction of 6 with,
respectively, methylmagnesium bromide, ethylmagnesium
bromide, vinylmagnesium bromide, and ethynylmagnesium
bromide. Compound 27 was also obtained by catalytic hydro-
genation of 28 or 29.
Gen er a l Hyd r ogen a tion P r oced u r e. A 5% Pd/C hydro-
genation catalyst (20 mg) was suspended in EtOAc (1 mL) and
stirred for 10 min under an H2 atmosphere. The substrate
(0.5 mmol) was dissolved in EtOAc (10 mL) and added via
syringe to the catalyst suspension. The reaction mixture was
then stirred for 3 h at room temperature. After this time, the
mixture was filtered through Celite, the reaction flask and the
Celite were washed two times with EtOH, and the organic
layers were concentrated at reduced pressure. CC of the oily
residue (hexanes/EtOAc mixtures) furnished the desired hy-
drogenation product.
Gen er a l Aceton id e Hyd r olysis P r oced u r e. The sub-
strate (1 mmol) was dissolved in 80% aqueous HOAc (7 mL).
The solution was then stirred at room temperature for 18 h.
(33) Krainer, E.; Naider, F.; Becker, J . Tetrahedron Lett. 1993,
1713-1716.

Diastereoselectivity in Organometallic Additions
J . Org. Chem., Vol. 63, No. 3, 1998 707
(2R ,3S )-1-O-Tr it yl-3,4-d i-O-b e n zyl-2-m e t h ylb u t a n e -
1,2,3,4-t et r ol (30), (2R,3S)-1-O-t r it yl-3,4-d i-O-b en zyl-2-
eth ylbu ta n e-1,2,3,4-tetr ol (31), (2R,3S)-1-O-tr ityl-3,4-d i-
O-ben zyl-2-vin ylbu ta n e-1,2,3,4-tetr ol (32), a n d (2R,3S)-
1-O -t r it y l-3,4-d i-O -b e n zy l-2-e t h y n y lb u t a n e -1,2,3,4-
tetr ol (33) were obtained by reaction of 10 with, respectively,
methylmagnesium bromide, ethylmagnesium bromide, vinyl-
magnesium bromide, and ethynylmagnesium bromide. Com-
pound 31 was also obtained by catalytic hydrogenation of 32
or 33.
(2R,3S)-1,4-Di-O-tr ityl-3-O-ben zyl-2-vin ylbu tan e-1,2,3,4-
tetr ol (34) was obtained by reaction of 11 with vinylmagne-
sium bromide. Detritylation of 34 (method B) and subsequent
silylation (for reaction conditions, see ref 9a) afforded 28.
(2S,3S)-1,2-Di-O-ben zyl-3-m eth ylpen tan e-1,2,3-tr iol (35),
(2S,3S)-1,2-di-O-ben zyl-3-m eth ylpen t-4-en e-1,2,3-tr iol (36),
a n d (2S,3S)-1,2-d i-O-b en zyl-3-m et h ylp en t -4-yn e-1,2,3-
tr iol (37) were obtained by reaction of 12 with, respectively,
ethylmagnesium bromide, vinylmagnesium bromide, and ethy-
nylmagnesium bromide. Compound 35 was also obtained by
catalytic hydrogenation of 36 or 37.
Ack n ow led gm en t. Financial support for the project
was provided through the Spanish Ministry of Educa-
tion and Science (DGICYT projects PB92-0745 and
PB95-1089) and by BANCAJ A (project P1B95-21). E.C.
thanks the Conseller´ıa de Cultura, Educacio` i Ciencia
de la Generalitat Valenciana, for a predoctoral fellow-
ship.
Su p p or tin g In for m a tion Ava ila ble: Analytical and tabu-
lated spectral data of compounds 10-12, 15-18, and 20-37
(10 pages). This material is contained in libraries on micro-
fiche, immediately follows this article in the microfilm version
of the journal, and can be ordered from the ACS; see any
current masthead page for ordering information.
J O9716744