Oxidation, Reduction, and Deoxygenation
2.2
199
dihydroxylation immediately after the Wittig reaction. In this way, lactone 47 is obtained in
a better overall yield from glucose than by isolating the intermediate ester 46. Similar yields
and diastereoselectivities are observed when galactose, arabinose, and xylose are subjected
to the one-pot Wittig-dihydroxylation reaction [134]. Mannose, ribose, and lyxose, on the
other hand, give a lower diastereoselectivity in the dihydroxylation [134]. In all cases, the
stereochemistry of the major product is in accordance with Kishi’s empirical rule [135]. This
predicts that for dihydroxylation of acyclic allylic alcohols (or protected alcohols), the relative
stereochemistry between the preexisting hydroxy group (or protected hydroxy group) and the
adjacent, newly introduced hydroxy group in the major product is erythro. Kishi’s rule applies
to dihydroxylation of a variety of carbohydrate allylic systems [136]. The selectivity in the
dihydroxylation can be improved or inverted by using a chiral ligand for osmium [137]. This
is demonstrated very effectively in a recent study where all six L-hexoses are prepared from
L-ascorbic acid [138]. The latter is converted into L-erythrose and L-threose which are both
subjected to a Wittig reaction to afford either the (E)- or the (Z)-unsaturated ester. Each Wittig
adduct is then dihydroxylated into either the 3,4-erythro or the 3,4-threo product depending
on the chiral ligand for osmium. Thus, dihydroxylation of (E)-olefin 48 in the presence of
hydroquinine 1,4-phthalazinediyl diether ((DHQ) 2 PHAL) gives exclusively the L-altro product 49 while the reaction with hydroquinidine 1,4-phthalazinediyl diether (DHQD) 2 PHAL)
affords only the opposite L-gluco isomer 50 ( > Scheme 16). Similar results are obtained for
the corresponding (Z)-olefin [138].
Although osmium(VIII) oxide continues to be the most popular reagent for dihydroxylation of
olefins, it does have two major drawbacks: it is very expensive and very toxic. As a result, other
reagents have been investigated. Potassium permanganate has been successfully applied for
dihydroxylation of the electron-deficient double bond in 2,3-unsaturated aldonolactones and –
lactams [139]. Ruthenium(VIII) oxide generated from catalytic ruthenium(III) chloride and
stoichiometric sodium periodate has been applied for the dihydroxylation of various protected
carbohydrates with very short reaction times [140]. Although good yields can be obtained with
both potassium permanganate and ruthenium(VIII) oxide, these reagents are more powerful
oxidants than osmium(VIII) oxide and less chemoselective. As a result, byproducts arising
from overoxidation and oxidative fission are more common.
Osmium(VIII) oxide also catalyzes the aminohydroxylation of olefins using chloramine-T as
a stoichiometric oxidant and nitrogen source. Aminohydroxylation has been performed on several unsaturated carbohydrates including glycals, hex-2-eno- and hex-3-enopyranosides [141].
The yields are typically in the 60–80% range, the major byproduct being the diol. However,
the reaction suffers badly from poor regioselectivity and often large amounts of both isomers
are obtained.
Azidonitration represents a special aminohydroxylation reaction because the azido group often
serves as a masked amino group. Azidonitration is possible only on glycals where the azido
group is introduced at C2 and the nitrate at C1. The reaction is typically carried out with
1.5 equiv. of sodium azide and 3 equiv. of ceric ammonium nitrate (CAN) in acetonitrile
at about −15 °C [142,143,144]. In this way, a number of acetylated glycals 51a–e undergo fairly selective azidonitration to give azidonitrates 52a–e ( > Table 8). The reaction works
particularly well on derivatives of D-galactal [142,145,146]. In contrast, azidonitration of triacetyl D-glucal is not very selective under these standard conditions, but gives an almost equal
amount of the corresponding 2-azido-D-glucose and –D-mannose products [142,147]. Howev-
2.2
199
dihydroxylation immediately after the Wittig reaction. In this way, lactone 47 is obtained in
a better overall yield from glucose than by isolating the intermediate ester 46. Similar yields
and diastereoselectivities are observed when galactose, arabinose, and xylose are subjected
to the one-pot Wittig-dihydroxylation reaction [134]. Mannose, ribose, and lyxose, on the
other hand, give a lower diastereoselectivity in the dihydroxylation [134]. In all cases, the
stereochemistry of the major product is in accordance with Kishi’s empirical rule [135]. This
predicts that for dihydroxylation of acyclic allylic alcohols (or protected alcohols), the relative
stereochemistry between the preexisting hydroxy group (or protected hydroxy group) and the
adjacent, newly introduced hydroxy group in the major product is erythro. Kishi’s rule applies
to dihydroxylation of a variety of carbohydrate allylic systems [136]. The selectivity in the
dihydroxylation can be improved or inverted by using a chiral ligand for osmium [137]. This
is demonstrated very effectively in a recent study where all six L-hexoses are prepared from
L-ascorbic acid [138]. The latter is converted into L-erythrose and L-threose which are both
subjected to a Wittig reaction to afford either the (E)- or the (Z)-unsaturated ester. Each Wittig
adduct is then dihydroxylated into either the 3,4-erythro or the 3,4-threo product depending
on the chiral ligand for osmium. Thus, dihydroxylation of (E)-olefin 48 in the presence of
hydroquinine 1,4-phthalazinediyl diether ((DHQ) 2 PHAL) gives exclusively the L-altro product 49 while the reaction with hydroquinidine 1,4-phthalazinediyl diether (DHQD) 2 PHAL)
affords only the opposite L-gluco isomer 50 ( > Scheme 16). Similar results are obtained for
the corresponding (Z)-olefin [138].
Although osmium(VIII) oxide continues to be the most popular reagent for dihydroxylation of
olefins, it does have two major drawbacks: it is very expensive and very toxic. As a result, other
reagents have been investigated. Potassium permanganate has been successfully applied for
dihydroxylation of the electron-deficient double bond in 2,3-unsaturated aldonolactones and –
lactams [139]. Ruthenium(VIII) oxide generated from catalytic ruthenium(III) chloride and
stoichiometric sodium periodate has been applied for the dihydroxylation of various protected
carbohydrates with very short reaction times [140]. Although good yields can be obtained with
both potassium permanganate and ruthenium(VIII) oxide, these reagents are more powerful
oxidants than osmium(VIII) oxide and less chemoselective. As a result, byproducts arising
from overoxidation and oxidative fission are more common.
Osmium(VIII) oxide also catalyzes the aminohydroxylation of olefins using chloramine-T as
a stoichiometric oxidant and nitrogen source. Aminohydroxylation has been performed on several unsaturated carbohydrates including glycals, hex-2-eno- and hex-3-enopyranosides [141].
The yields are typically in the 60–80% range, the major byproduct being the diol. However,
the reaction suffers badly from poor regioselectivity and often large amounts of both isomers
are obtained.
Azidonitration represents a special aminohydroxylation reaction because the azido group often
serves as a masked amino group. Azidonitration is possible only on glycals where the azido
group is introduced at C2 and the nitrate at C1. The reaction is typically carried out with
1.5 equiv. of sodium azide and 3 equiv. of ceric ammonium nitrate (CAN) in acetonitrile
at about −15 °C [142,143,144]. In this way, a number of acetylated glycals 51a–e undergo fairly selective azidonitration to give azidonitrates 52a–e ( > Table 8). The reaction works
particularly well on derivatives of D-galactal [142,145,146]. In contrast, azidonitration of triacetyl D-glucal is not very selective under these standard conditions, but gives an almost equal
amount of the corresponding 2-azido-D-glucose and –D-mannose products [142,147]. Howev-