Stereoselective Aldol Additions Catalyzed by Dihydroxyacetone Phosphate-Dependent Aldolases in Emulsion Systems: Preparation and Structural Characterization of Linear and Cyclic Iminopolyols from Aminoaldehydes

Article abstract of DOI:10.1002/chem.200304966

The potential of dihydroxyacetone phosphate (DHAP)-dependent aldolases to catalyze stereoselective aldol additions is, in many instances, limited by the solubility of the acceptor aldehyde in aqueous/co-solvent mixtures. Herein, we demonstrate the efficiency of emulsion systems as reaction media for the class I fructose-1,6-b isphosphate aldolase (RAMA) and class II recombinant rhamnulose-1-phosphate aldolase from E. coli (RhuA)catalyzed aldol addition between DHAP and N-benzyloxycarbonyl (N-Cbz) aminoaldehydes. The use of emulsions improved the RAMA-catalyzed aldol conversions by three to tenfold relative to those in conventional DMF/ water mixtures. RhuA was more reactive than RAMA towards the N-Cbz aminoaldehydes regardless of the reaction medium. With (S)- or (R)-Cbz-alaninal, RAMA exhibited preference for the R enantiomer, while RhuA had no enantiomeric discrimination. The linear N-Cbz aminopolyols thus obtained were submitted to catalytic intramolecular reductive amination to afford the corresponding iminocyclitols. This reaction was diastereoselective in all cases examined; the face selectivity was controlled by the stereochemistry of the newly formed hydroxyl group originating from the aldehyde. Characterization of the resulting iminocyclitols allowed the assessment of the diastereoselectivity of the enzymatic aldol reactions with respect to the N-protected aminoaldehyde. RAMA formed single diastereoisomers from N-Cbz-glycinal and from both enantiomers of N-Cbz-alaninal, while 14% of the epimeric product was observed from N-Cbz-3-aminopropanal. Diastereoselectivity from RhuA was lower than that observed from RAMA. Interestingly, a single diastereoisomer was formed from (S)-Cbz-alaninal, whereas only a 34% diastereomeric excess was observed from its enantiomer (i.e., (R)-Cbz-alaninal).

Full text of DOI:10.1002/chem.200304966

FULL PAPER
Stereoselective Aldol Additions Catalyzed by Dihydroxyacetone Phosphate-
Dependent Aldolases in Emulsion Systems: Preparation and Structural
Characterization of Linear and Cyclic Iminopolyols from Aminoaldehydes
Laia Espelt,[a,c] Teodor Parella,[e] Jordi Bujons,[b] Conxita Solans,[c] Jesus Joglar,[d]
¬
Antonio Delgado,[d,f] and Pere Clape¬s*[a]
Abstract: The potential of dihydroxya-
cetone phosphate (DHAP)-dependent
aldolases to catalyze stereoselective al-
dol additions is, in many instances,
limited by the solubility of the acceptor
aldehyde in aqueous/co-solvent mix-
tures. Herein, we demonstrate the effi-
ciency of emulsion systems as reaction
media for the class I fructose-1,6-bi-
sphosphate aldolase (RAMA) and class
II recombinant rhamnulose-1-phos-
phate aldolase from E. coli (RhuA)-
catalyzed aldol addition between
DHAP and N-benzyloxycarbonyl (N-
Cbz) aminoaldehydes. The use of emul-
sions improved the RAMA-catalyzed
aldol conversions by three to tenfold
relative to those in conventional DMF/
water mixtures. RhuAwas more reactive
than RAMA towards the N-Cbz amino-
aldehydes regardless of the reaction
medium. With (S)- or (R)-Cbz-alaninal,
RAMA exhibited preference for the R
enantiomer, while RhuA had no enan-
tiomeric discrimination. The linear N-
Cbz aminopolyols thus obtained were
submitted to catalytic intramolecular
reductive amination to afford the corre-
sponding iminocyclitols. This reaction
was diastereoselective in all cases exam-
ined; the face selectivity was controlled
by the stereochemistry of the newly
formed hydroxyl group originating from
the aldehyde. Characterization of the
resulting iminocyclitols allowed the as-
sessment of the diastereoselectivity of
the enzymatic aldol reactions with re-
spect to the N-protected aminoalde-
hyde. RAMA formed single diaster-
eoisomers from N-Cbz-glycinal and
from both enantiomers of N-Cbz-alani-
nal, while 14% of the epimeric product
was observed from N-Cbz-3-aminopro-
panal. Diastereoselectivity from RhuA
was lower than that observed from
RAMA. Interestingly, a single diaster-
eoisomer was formed from (S)-Cbz-
alaninal, whereas only a 34% diastereo-
meric excess was observed from its
enantiomer (i.e., (R)-Cbz-alaninal).
Keywords: biotransformations
¥
colloids ¥ DHAP-dependent aldo-
lases ¥ iminocyclitols ¥ lyases
conventional and uncommon carbohydrates as well as other
complex hydroxylated products. DHAP aldolases catalyze the
aldol addition of the DHAP donor substrate to a variety of
non-natural aldehyde acceptors and, in many cases, the
stereoselectivity of the reaction is controlled by the enzyme,
regardless of the structure and stereochemistry of the
substrate acceptor.[3]
Introduction
The stereoselective carbon carbon bond formation catalyzed
by aldolases has attracted tremendous interest in recent
years.[1, 2] In particular, dihydroxyacetone phosphate
(DHAP)-dependent aldolases have been demonstrated to
be extremely useful tools in the asymmetric synthesis of
¬
[a] Dr. P. Clapes, Dr. L. Espelt
[d] Dr. J. Joglar, Dr. A. Delgado
Department of Peptide and Protein Chemistry
Institute for Chemical and Environmental Research-CSIC
Jordi Girona 18 26, 08034 Barcelona (Spain)
Fax : (‡34)93-4006112
Research Unit on Bioorganic Molecules (RUBAM)
Department of Biological Organic Chemistry
Institute for Chemical and Environmental Research-CSIC
Jordi Girona 18 26, 08034 Barcelona (Spain)
E-mail: pcsqbp@iiqab.csic.es
[e] Dr. T. Parella
¡
¡
¡
[b] Dr. J. Bujons
Servei de Ressonancia magnetica Nuclear Universitat Autonoma de
Department of Biological Organic Chemistry (DBOC)
Institute for Chemical and Environmental Research-CSIC
Jordi Girona 18 26, 08034 Barcelona (Spain)
Barcelona, Bellaterra (Spain)
[f] Dr. A. Delgado
¡
Unitat de QuÌmica Farmaceutica
[c] Dr. L. Espelt, Dr. C. Solans
Facultat de Farmacia, Universitat de Barcelona (Spain)
Department of Surfactant Technology
Institute for Chemical and Environmental Research-CSIC
Jordi Girona 18 26, 08034 Barcelona (Spain)
Supporting information for this article is available on the WWW under
http://www.wiley-vch.de/home/chemeurj.org/ or from the author.
Chem. Eur. J. 2003, 9, 4887 4899
DOI: 10.1002/chem.200304966
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4887

¬
P. Cla pes et al.
FULL PAPER
DHAP aldolases have been exploited as catalysts for the
solubilization of the aldehyde and an effective contact with
the enzyme. For the sake of comparison, a reaction medium
based on co-solvent mixtures such as dimethylformamide
(DMF)/water 1:4 was also assayed. Furthermore, another goal
was to elucidate the stereochemistry of both the linear
adducts and the corresponding cyclic products (iminocycli-
tols) to assess the stereoselectivity of the aldolases used
towards the newly formed carbon carbon bond. We selected
the following acceptor aldehydes: benzyl 3-oxopropylcarba-
mate (5), benzyl 3-oxoethylcarbamate (Cbz-glycinal) (6), (S)-
benzyl 1-methyl-2-oxoethylcarbamate ((S)-Cbz-alaninal) (7),
and (R)-benzyl 1-methyl-2-oxoethylcarbamate ((R)-Cbz-ala-
ninal) (8). Type I d-fructose-1,6-diphosphate aldolase from
rabbit muscle (RAMA) and type II recombinant l-rhamnu-
lose-1-phosphate aldolase (RhuA) from E. coli were the
biocatalysts.
synthesis of glycoprocessing enzyme inhibitors such as poly-
7]
hydroxy-N-heterocycles (iminocyclitols).[4
These com-
pounds have been largely investigated as therapeutic targets
for the design of new antibiotics, antimetastatic, antihyper-
10]
glycemic, or immunostimulatory agents.[8 The key step of
the chemoenzymatic synthesis of these products was the
stereoselective carbon carbon bond formation between
DHAP and synthetic equivalents of aminoaldehydes. It has
been shown that azido aldehydes or N-formamido aldehydes
are the best substrates for fructose 1,6-diphosphate aldolase at
synthetically practical rates.[4, 11] N-Formyl aminoaldehydes
have the tremendous additional advantage that they can be
obtained from the wide structural variety of readily available
optically pure a- or b-amino acids and their derivatives.[12, 13]
Nevertheless, the use of these substrates may be complicated
by the racemization of the a stereogenic center observed with
some derivatives,[14] which is probably mediated by the N-
formyl group. Moreover, N-formyl deprotection conditions
are not always compatible with sensitive compounds.[11]
Use of the well-known benzyloxycarbonyl (Cbz) or tert-
butyloxycarbonyl (Boc) N-protecting groups for aminoalde-
hydes can be advantageous for two main reasons: 1) they
constitute an excellent complement for orthogonal protection
schemes in particular when the aldol adducts obtained are to
be used as chiral building blocks, and 2) the urethane-type
protecting groups may be less prone to spontaneous racemi-
zation of the stereogenic center(s). N-Cbz aminoaldehydes
were found to be poor substrates for fructose 1,6-diphosphate
Results and Discussion
In previous work on aldolase-catalyzed reactions in water/
organic solvent (W/O) gel emulsions, we concluded that the
partition coefficient of the acceptor aldehyde between both
the continuous (i.e., water) and dispersed phase (i.e., aliphatic
hydrocarbon also referred to as the oil) as well as the water
oil interfacial tension were the parameters that control both
the enzymatic activity and product yield.[21] These parameters
were strongly affected by the nature of the aliphatic hydro-
carbon and the acceptor aldehydes, while no influence was
observed by either the DHAP or the enzyme. Thus, the best
reaction systems were W/O gel emulsions formulated with
water (90 wt%), technical-grade tetraethylene glycol tetra-
decyl ether surfactant (C14H29(OCH2CH2)4OH) (4 wt%), with
an average of 4 moles of ethylene oxide per surfactant
molecule, abbreviated as C14E4, and an aliphatic hydrocarbon
(6 wt%) such as tetradecane (C14) or hexadecane (C16).
Interestingly, the stability of the fructose-1,6-diphosphate
aldolase from rabbit muscle (RAMA) in the presence of
50 mm phenylacetaldehyde, a low water-soluble aldehyde, was
25-fold higher in gel emulsions than in a DMF/water 1:4 (v/v)
mixture. In this work, we used these emulsion systems to
perform the enzymatic aldol additions between DHAP and
N-Cbz aminoaldehydes.
17]
aldolase.[4, 11, 15 Only the products formed from N-(Cbz)a-
minoacetaldehyde (i.e., Cbz-glycinal), N-(Cbz)-3-aminopro-
panal, and DHAP have been characterized although these
were obtained in low yields.[15, 17] Some authors suggested that
either the size of the N-substituent or their poor water
solubility might account for the low reactivity ob-
18]
served.[4, 11, 16
Moreover, increasing the percentage of
organic co-solvent in the medium to make the aldehyde
soluble may lead to either a dramatic enzyme deactivation[19]
or an insolubilization of the DHAP sodium salt. As a result,
no reaction or insufficient product yields are often obtained.
In the present study we endeavored to gain insight into the
reactivity of N-(Cbz)-protected aminoaldehydes as substrates
for the aldolase-catalyzed stereoselective synthesis of poly-
hydroxylated amines. To this end, we have recently developed
new reaction systems based on colloidal dispersions,[20, 21]
namely highly concentrated water-in-oil (gel) emulsions,
which overcome most of the disadvantages of the aqueous/
co-solvent mixtures such as inactivation of the aldolase and
incomplete aldehyde solubilization in the medium. These
reaction systems consist of ternary mixtures of water (e.g.,
90% w/w), a dispersant agent of the poly(oxyethyleneglycol
alkyl ether)-type (e.g., 4% w/w), and an aliphatic hydro-
carbon (e.g., 6% w/w).[22]
Preparation of N-protected aminoaldehydes: The N-Cbz
aminoaldehydes were synthesized from the corresponding
amino alcohols. These were first protected as Cbz derivatives
followed by oxidation of the alcohol function (Scheme 1). For
the protected amino alcohols 1 and 2, the Swern oxidation was
the method of choice, whereas for both enantiopure 3[23] and 4
the methodology using 2-iodoxybenzoic acid (IBX)[24] was
preferred.[25] Both methods gave 90 95% oxidation yields. In
most instances, a simple workup followed by a rough
crystallization or precipitation provided the aminoaldehydes
in a form pure enough to be used in the enzymatic aldol
condensation. An exception was the N-protected aminoalde-
hyde 6. In this case, the product did not crystallize and
purification by reversed-phase HPLC was required.
The goal of the present work was to explore whether the
solubility or structural properties of the N-(Cbz)-protected
aminoaldehydes were responsible for their poor reactivity as
acceptor substrates in aldolase-catalyzed condensations. For
this purpose, the reactions were carried out in the aforemen-
tioned high water content emulsions to ensure both a good
4888
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4887 4899

Stereoselective Aldol Additions
4887 4899
of the acceptor aldehyde[4, 17] and the stability of the enzyme.
The solubility of 5 in the DMF/water 1:4 mixture was
measured experimentally and was around 144 mm. Therefore,
the effective molar ratio of aldehyde/DHAP (144:120) was
only 1:1.2. Hence, the lower substrate molar ratio caused by
the limited solubility of acceptor aldehyde is one factor that
might explain the poor conversions achieved in the DMF/
water 1:4 mixture. To demonstrate this, we examined the
influence of the number of molar equivalents (equivmolÀ1) of
aldehyde acceptor used on the reaction conversion to 9. To
this end, experiments were conducted in emulsion media to
ensure good solubilization of the aldehyde. On increasing the
aldehyde concentration up to 3 equivmolÀ1 the conversion
improved slightly (see Table 1, entries 2 and 4). In contrast, at
equimolar aldehyde/DHAP ratio the conversion dropped to
approximately 40%.
A second factor that may reduce the effectiveness of the
DMF/water mixtures was the enzyme stability. Thus, the
enzyme activity may be extinguished before the reaction
reached equilibrium. However, in this case, by increasing the
enzyme concentration up to threefold (i.e., from 12 to
35 UmLÀ1) the conversion increased from 21% (Table 1,
entry 6) to only a modest 38%, which was also the maximum
found for an equimolar aldehyde/DHAP ratio (see above).
An important feature to consider was the evolution of the
aldol condensation product with the reaction time. Analysis of
the reaction progress indicated a fast conversion to 9 reaching
a maximum followed by a gradual decreased for prolonged
incubation times (Figure 1) (10% conversion to 9 after 120 h
of incubation). It is worth stressing that DHAP is not stable,
especially at elevated pH values, and decomposes into
methylglyoxal.[28, 29] Moreover, the aldehyde is also in equili-
brium with its hydrate form and with other polymeric forms
which can vary with the reaction progress.[30] Although a
satisfactory explanation has not yet been found, the observed
consumption of 9 was presumably due to the spontaneous
decomposition of the DHAP, the aldehyde, or both. Con-
sequently, the reaction equilibrium shifted to the reactants in
a retro-aldol-type reaction. As depicted in Figure 1, this effect
was much more evident in emulsion media than in the DMF/
water 1:4 mixture. This was attributed to the enzyme activity,
since the enzyme half-life was 25-fold higher in emulsions
Scheme 1. Synthesis of acceptor aldehydes 5 8: a) Cbz-OSu dioxane/H2O
4/1; b) Swern or IBX oxidation.
Enzymatic aldol reactions: First, we chose the RAMA-
catalyzed aldol condensation of DHAP with the acceptor
aminoaldehyde 5 to test the emulsion media (see Table 1). It is
worth stressing that the reaction performance in gel emulsions
at 258C with C14E4 as surfactant may depend on both the
physicochemical characteristics of the acceptor aldehyde
(e.g., hydrophobicity and solubility) and the aliphatic hydro-
carbon used to formulate the emulsion.[21] Thus, as mentioned
before, the reactions were conducted in the best emulsions of
the ternary water/C14E4/oil (90/4/6 wt%) systems, in which the
oil component was tetradecane, hexadecane, or squalane
(C30).[26] The reaction conversion to 9 was also examined as a
function of DHAP concentration (30 mm and 120 mm) with
1.7 equivmolÀ1 of N-aminoaldehyde. For the sake of compar-
ison, the reactions were also conducted in a conventional
DMF/water 1:4 (v/v) co-solvent mixture.[27]
The maximum reaction conversions to 9 obtained in
emulsion media and the DMF/water 1:4 mixture are sum-
marized in Table 1. At 30 mm DHAP and 1.7 equivmolÀ1 of
acceptor aldehyde, the best yields were achieved in emulsions
formulated with either tetradecane or hexadecane and in the
DMF/water 1:4 (v/v) mixture. This indicates that N-Cbz-
aminoaldehyde 5 was accepted by RAMA at rates good
enough for preparative-scale reactions. Interestingly, at
120 mm DHAP the reaction conversion to 9 achieved in
emulsion media was threefold higher than that in the DMF/
water 1:4 mixture (Table 1, entries 1, 3, and 6). In the latter
medium, the reaction turned out to be impractical for
preparative purposes. These differences in the reaction
efficiency appeared to be related to both the limited solubility
Table 1. RAMA-catalyzed aldol addition between DHAP and 5.
Conversion [%][a] (Time [h])
Entry
Reaction medium[b]
Aldehyde [equivmolÀ1
]
30 mm DHAP[c]
120 mm DHAP[d]
1
2
3
4
5
6
water/C14E4/tetradecane
water/C14E4/tetradecane
water/C14E4/hexadecane
water/C14E4/hexadecane
water/C14E4/squalane
DMF/water 1:4 v/v
1.7
3
1.7
3
1.7
1.7
60(4.5)
60(1)
65(1)
60(1)
62(1)
53(1)
21(24)
[e]
64(4.5)
[e]
51(1)
64(4.5)
[a] Molar percentage conversion to 9 (with respect to the starting DHAP concentration) determined by HPLC from the crude reaction mixture using purified
standard. [b] Reaction volume 2.5 mL, Tˆ 258C, 90/4/6 wt%, respectively, in emulsion. [c] [RAMA] ˆ 5 UmLÀ1. [d] [RAMA] ˆ 12 UmLÀ1. [e] Not
determined.
Chem. Eur. J. 2003, 9, 4887 4899
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4889

¬
P. Cla pes et al.
FULL PAPER
both the reaction time and the amount of enzyme must be
controlled to achieve the maximum yields and avoid the retro-
aldol reaction. In this work, each reaction was monitored by
HPLC and the maximum conversions of product are given.
Synthetic applications
Influence of the acceptor aldehyde and the reaction medium:
The next step was to test the versatility of the novel reaction
medium with some more examples. We chose Cbz-glycinal
(6), (S)-Cbz-alaninal (7), and (R)-Cbz-alaninal (8) as acceptor
aldehydes in addition to benzyl 3-oxopropylcarbamate (5)
using RAMA and recombinant l-rhamnulose-1-phosphate
aldolase (RhuA) from E. coli as catalysts. Reactions were
performed in emulsions and in the DMF/water 1:4 (v/v)
mixture. For the sake of simplicity, we will first discuss the
reactivity of the aldolases as a function of both the N-
aminoaldehyde and the reaction medium; the structural and
stereochemical characterization of the products obtained will
be analyzed in detail later.
Reaction conversions obtained are summarized in Table 2.
It is clear that the N-Cbz aminoaldehydes tested were
accepted as substrates for RAMA and RhuA aldolases.
Overall, RhuA was more active than RAMA; the reactions
catalyzed by the latter were influenced much more by the type
of reaction media (Table 2, entries 1 3). This result could be
explained partially by the observed higher affinity of RhuA
Figure 1. RAMA-catalyzed aldol reaction between DHAP and 5. Reac-
tion time-course of product 9 formation in water/C14E4/oil 90:4:6 wt%
!
emulsion systems in which the oil component was tetradecane ( ),
~
&
*
hexadecane ( ), squalane ( ), and DMF/water 1:4 v/v mixtures ( ).
[DHAP] ˆ 30 mm, 1.7 equivmolÀ1 of 5, reaction volume 2.5 mL, and
[RAMA] ˆ 5 UmLÀ1. Reactions were carried out as described in the
Experimental Section
than in the DMF/water 1:4 mixture (100 h versus 4 h),[21]
which therefore allowed secondary reactions to occur.
The extension of this phenomenon was difficult to general-
ize, since it may also depend on the reaction product. Hence,
Table 2. RAMA- and RhuA-catalyzed aldol addition between DHAP and N-Cbz-aminoaldehydes.
Conversion [%][a] (Time [h])
Reaction conditions[b]
B[d] C[e]
Entry
1
Acceptor aldehyde
Aldolase [U mLÀ1
RAMA (10)
]
Product[g]
A[c]
D[f]
6
76(24)
82(24)
83(24)
27(24)
2
3
4
5
6
7
7
8
5
6
7
8
RAMA (23)
RAMA (21)
RhuA (1)
20(46)
40(24)
50(4)
84(1)
66(2)
67(2)
18(46)
40(24)
48(2)
96(1)
59(2)
74(4)
18(46)
56(24)
47(4)
80(3)
66(5)
67(4)
4(46)
5(24)
58(7)
80(3)
60(2)
61(4)
RhuA (1)
RhuA (1)
RhuA (1.3)
[a] Molar percentage conversion to the aldol adduct (10 16) (with respect to the starting DHAP concentration) determined by HPLC from the crude
reaction mixture using purified standards. [b] The initial concentration of the DHAP was in the range 90 120 mm, reaction volume 5 mL, Tˆ 258C. [c] H2O/
C14EO4/tetradecane 90/4/6 wt%. [d] H2O/C14EO4/hexadecane 90/4/6 wt%. [e] H2O/C14EO4/squalane 90/4/6 wt%. [f] DMF/H2O 1:4 v/v. [g] For C-4
stereochemistry, see text.
4890
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4887 4899

Stereoselective Aldol Additions
4887 4899
for both nonpolar and sterically hindered substrates relative
to that of RAMA.[28, 31]
RhuA showed no preference for any of the enantiomers of
Cbz-alaninal (Table 2, entries 6 and 7). In contrast, high
kinetic enantiodiscrimination by this aldolase has been
reported with a series of racemic 2-hydroxyaldehydes accept-
ors.[32]
To explain the observed preference of RAMA towards the
R isomer of Cbz-alaninal, a computational model was
constructed. Although a detailed study of this kind is beyond
the scope of this work, the preliminary model presented may
help to elucidate the different binding modes to the protein of
both isomers of Cbz-alaninal and those of their reaction
products with DHAP. Based on the known mechanisms of
class I and II DHAP-dependent aldolases[33, 34], we looked at
the states before and after the carbon carbon bond forma-
tion (I and II in Scheme 2).
Figure 2A shows the compared minimized structures
obtained for the R and S enantiomers of Cbz-alaninal in
state I. These structures were generated by imposing a
restraint during the minimization process to keep the distance
between C-3 of the bound DHAP and C-1 of Cbz-alaninal
close to 2.5 ä. The purpose of this arbitrarily chosen restraint
was to ensure that the aldehyde molecule would adopt a
conformation close to the pre-reaction state. In this situation,
both enantiomers of the aldehyde show very similar con-
formations stabilized by hydrogen bonds with several amino
acids that are in the catalytic site. In particular, the one
between the aldehyde oxygen atom and the protonated
carboxylic group of Asp33 fixes the orientation of the
aldehyde group to approach from its si face to the reactive
enamine derived from the DHAP moiety. This model does
not exclude the possibility that other residues located by the
C-terminus of RAMA (i.e., Tyr363) could participate in the
enzymatic mechanism by orienting and promoting proton
transfer to the aldehyde, as has been suggested in the
literature.[35, 36] The enamine C-3 hydroxyl group is also kept
in its reactive orientation by hydrogen bonds to Glu187 and
Asp33. Although there are no obvious interactions between
the methyl group at C-2 of the isomers of Cbz-alaninal and the
protein, there is a difference in the energy calculated for both
Besides the substrate preferences, the emulsion media
enhanced the catalytic efficiency of RAMA towards hydro-
phobic substrates such as the N-Cbz aminoaldehydes tested
by three-, five-, and even tenfold, thereby allowing the
synthesis of the corresponding products at preparative level
in moderate to good yields. In contrast, in the DMF/H2O 1:4
(v/v) mixture, the N-Cbz aminoaldehydes were poor sub-
strates for RAMA. It is worth mentioning that in DMF/H2O
1:4 (v/v), the conversion to 10 improved up to 60% upon
increasing the RAMA concentration by 3.5-fold (i.e., up to
35 UmLÀ1). On the other hand, with aldehydes 7 and 8 no
improvement in the conversion was observed when raising the
concentration of aldolase. These results reinforce the fact that
both the solubility of the aldehyde and the stability of the
aldolase may play a major role in the efficacy of the DMF/
H2O system (see previous section).
These results indicate that the reaction performance may be
affected not only by structure of the aminoaldehyde but also
by the reaction medium. However, this was also a function of
the aldolase used. Thus, reactions with RhuA, a type II
aldolase, gave good results regardless of the reaction media
(Table 2, entries 4 7). This behavior may be related to both
its higher tolerance towards hydrophobic substrates and its
high stability in the presence of organic polar solvents.[31] It is
noteworthy that the reaction profiles with aldehydes 5 and 6
showed, for both enzymes, the phenomenon of the retro-aldol
reaction described above. Hence, conversion to the product
decreased by 10 30% after 24 h depending on both the
reaction conditions and the aldehyde. Interestingly, this
phenomenon was not observed for the reactions with the
acceptor aldehydes 7 and 8.
We found that RAMA exhibited a moderate preference
towards the (R)-Cbz-alaninal (8) over the (S)-Cbz-alaninal
(7) (Table 2, entries 2 and 3). The reverse situation was
reported by Hung et al.,[11] who found that RAMA preferred
the S enantiomer of a series of N-a-amino protected-b-
hydroxypropanals as aldehyde acceptors. On the other hand,
À
Scheme 2. Formation of the C C bond in the reactions catalyzed by a class I aldolase (RAMA) and a class II aldolase (RhuA).
Chem. Eur. J. 2003, 9, 4887 4899
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4891

¬
P. Cla pes et al.
FULL PAPER
Figure 2. Structure of the active site of RAMA (A) and RhuA (B) complexed to DHAP and both isomers of Cbz-alaninal (state I and III, respectively). The
conformations of the active site residues (sticks) and of the substrates (ball-and-stick) are represented in different colors: the complexes with R-Cbz-alaninal
are represented using gray for the carbon atoms, blue for the nitrogen atoms, red for the oxygen atoms, while the complexes with the S isomer are represented
with all atoms in orange. The distance between the two carbon atoms that participate in bond formation is shown for each case.
isomers complexed to the protein; these are lower for the R
isomer (DERÀS ˆ À2.2 kcalmolÀ1). This difference is kept
even if the restraint imposed during the minimization is
removed and the aldehyde is allowed to relax to a new
energetic minimum about 3.5 ä away from the DHAP
moiety. These energy differences arise from a sum of
small contributions from different interactions between the
substrates and the protein bound to DHAP. Similarly,
the structures (not shown) obtained for each adduct in
state II (Scheme 2) show a small difference in energy that
Purification and structural and stereochemical characteriza-
tion of the aldol adducts: Compounds 9 16 were synthesized
on the milligram scale (50 160 mg) and purified by a
combination of ion exchange and reverse-phase (RP) HPLC
methodologies. The RP-HPLC was mostly a desalting step
and, as observed by analytical RP-HPLC, separation of the
putative diastereoisomers was not accomplished under these
conditions. This two-step purification procedure provided
product recoveries, without any optimisation, ranging from 55
to 75%. The purity was assessed by both HPLC and elemental
analysis.
The structural and stereochemical characterization of the
products was ascertained by one- and two-dimensional NMR
techniques. Assignments of relative configurations of acyclic
structures by NMR-based methods are elusive in most
instances, since their rotational freedom leads to multiple
conformers.[37] We therefore studied the stereochemical
structure of both linear and the corresponding cyclic com-
pounds, namely iminocyclitols or iminosugars. The latter are
important biologically active compounds and are among the
most effective glycosidase inhibitors.
To obtain more simple NMR spectra, in some instances the
phosphate group of compounds 9 16 was removed by
treatment with acid phosphatase followed by desalting by
RP-HPLC. The structural features of these compounds were
found to be independent of the presence of the phosphate
moiety. The NMR data from 9 and 13 revealed that one major
conformer with H3/H4-syn orientation was present (i.e., small
values of 3J(H-3,H-4), 3J(C-2,H-4), 3J(C-5,H-3) and NOE (H-
4,H-3) strong) (Scheme 3).
Hence, the relative stereochemistry at C-3 and C-4 carbons
is consistent with that proposed for RAMA and RhuA
catalysis, respectively.[1, 31, 38] Based on mechanistic consider-
ations of the DHAP aldolases[31, 34, 39 41] it can be assumed that
the absolute configuration at C-3 is independent on the
favors the one derived from R-Cbz-alaninal (DERÀS
À1.6 kcalmolÀ1).
ˆ
These differences suggest that the R isomer of Cbz-alaninal
forms a more stable complex with the DHAP-modified
protein than the S isomer. Similarly, the adduct formed with
R configuration at C-5 seems to be more stable than the one
with the S configuration. However, these differences do not
necessarily explain our experimental observations, since we
have only looked at one of the steps of the multistep
enzymatic mechanism that is proposed for this aldolase.
The same type of study for the complexes of R- and S-Cbz-
alaninal with DHAP-ligated RhuA (state III, Scheme 2) gave
the results shown in Figure 2B. Similarly, the orientation of
the aldehyde group is fixed by a hydrogen bond interaction
with the protonated carboxylic group of Glu117, which also
interacts with the hydroxyl group on C-3 of the DHAP
molecule complexed to the Zn2‡ ion. This interaction forces
the approach of the aldehyde to the DHAP-enolate from its re
face. Energetically, the R isomer of Cbz-alaninal is again
favored over the S isomer (DERÀS ˆ À9.0 kcalmolÀ1) and the
same applies for the complex once the carbon carbon bond
between the aldehyde and DHAP is formed (DERÀS
ˆ
À8.6 kcalmolÀ1). Again, these differences cannot be attrib-
uted to a specific interaction of any of the groups present in
the organic ligands with the protein.
4892
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4887 4899

Stereoselective Aldol Additions
4887 4899
group to the ketone carbonyl without formation of the
corresponding enamine (Scheme 4).
For compounds 10 12 and 14 16 even more complex
NMR spectra were obtained, which made it extremely
difficult to unequivocally assign the relative configuration at
C-3 and C-4 as well as to identify and quantify the putative
diastereoisomers. By analogy with compounds 9 and 13,
Scheme 3. Main conformer of compounds 9 and 13.
acceptor used in the reaction. Moreover, optical rotations of 9
and 13 have opposite signs (Table 3, entries 1 and 5). Thus, the
absolute configurations 3S,4R and 3R,4S can be assigned for 9
and 13, respectively. In addition to the expected products 9
and 13, minor signals (8%) corresponding to a cyclic hemi-
aminal, as confirmed from the NOESY and HMBC spectra,
were detected.[42] This compound resulted from the reversible
spontaneous intramolecular addition of the secondary amino
Scheme 4. Spontaneous intramolecular cyclization of 9.
Table 3. Structure of the cyclic iminocyclitols derived from linear compounds 9 16.
Entry
1
Acceptor aldehyde/
aldolase
Linear product/
Cyclic products
C-3, C-4 Stereochemistry
[trans/cis ratio]
[a]2D0 Cyclic product or mixture
‡ 8.2 (c ˆ 1.4 in MeOH)
[a]2D0 (c ˆ 1 in MeOH)
5 RAMA
9/ ‡ 16.2
86/14
2
3
4
6 RAMA
7 RAMA
8 RAMA
10/ ‡ 1.1
11/ À 9.2
12/ À 1.8
trans only
trans only
trans only
‡ 18.7 (c ˆ 1 in MeOH)
‡ 16.0 (c ˆ 0.4 in MeOH)
‡ 45.0 (c ˆ 1.4 in MeOH)
5
5 RhuA
13/ À 11.6
80/20
À 6.0 (c ˆ 1.5 in MeOH)
6
7
8
6 RhuA
7 RhuA
8 RhuA
14/ À 0.2
15/ À 1.4
16/ ‡ 7.8
90/10
À 18.0 (c ˆ 1.5 in MeOH)
À 33.0 (c ˆ 1.2 in MeOH)
À 11.0 (c ˆ 1 in MeOH)
trans only
67/33
[a] References [58 62]. [b] References [60, 61, 63 66]. [c] References [67, 68]. [d] References [47, 67, 69 71]. [e] References [60, 61, 72 78]. [f] Refer-
ences [58, 77, 79 81]. [g] References [47, 70]. [h] References [82].
Chem. Eur. J. 2003, 9, 4887 4899
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4893

¬
P. Cla pes et al.
FULL PAPER
variable-temperature NMR experiments and exchange peaks
observed in NOESY experiments of 10 (or its enantiomer 14)
revealed that both linear and cyclic structures were in
equilibrium, but in this case at equimolar concentrations
(Scheme 5). The major conformer of the linear product (10a)
was similar to the one found for compound 9 and it was
terized for the first time in this work. Compound 27 had been
isolated from the bark of Angylocalyx pynaertii (Legumino-
sae)[47] and its synthesis is presented here for the first time.
In addition to NOE data, the shielding effects observed on
some proton and carbon chemical shift values due to the
spatial effect of the adjacent hydroxyl groups have been used
as a probe to determine and
assign the relative stereochem-
istry in these families of com-
pounds, especially for the five-
membered rings (Tables 4 and
5). As a general trend, a de-
shielding effect was observed
on the chemical shift of both
proton and carbon depending
on the stereochemistry of the
contiguous hydroxyl substitu-
ents. The hydroxyl groups
OH(3) and OH(4) in com-
pounds 19 and 24 induced
Scheme 5. Main conformer and spontaneous intramolecular cyclization of 10 with or without the phosphate
group.
0.2 0.3 ppm upfield shift on the H-2, H-3, and H-4 protons
compared with those of 25 due to their relative proximity, and
therefore, cis stereochemistry. On the other hand, the reverse
effect is observed on the C-2, C-3, and C-4 carbon centers, for
which a 3 6 ppm upfield effect was observed. The anisotropic
effect on the methyl carbon resonance is also important in
compounds 28 and 27 (or its enantiomer 20). In the case that
OH(4) and Me(6) show a cis stereochemistry, a strong upfield
effect of about 4 5 ppm was observed. All these predictions
were further confirmed by NOE data.
consistent with the stereochemistry of RAMA catalysis.
Moreover, the cyclic compound 10b also showed another
additional equilibrium between two other equally populated
conformations (10b(I) and 10b(II)), presumably due to the
pseudorotation commonly observed in five-membered nitro-
gen-containing rings (Scheme 5).[43] The spontaneous forma-
tion of the cyclic compounds and their conformational
equilibrium was also confirmed by observing the H NMR
and two-dimensional NOESY spectra at different temper-
atures.[44]
1
An inspection of the stereochemistry at C-2 for the
iminocyclitols described in Table 3 revealed that the reductive
amination with Pd/C was highly diastereoselective, as report-
ed previously by other authors.[4, 6, 15, 17, 38, 46] Interestingly, we
have found that in both six- and five-membered ring systems,
hydrogenation took place from the face opposite to the C-4
The presence of linear and cyclic species in equilibrium may
explain the low and unexpected values of the optical rotations
found for some enantiomeric products (Table 3, entries 2 4,
and 6 8). The spontaneous cyclization would also explain the
chromatographic behavior of these products on reversed-
phase HPLC analysis: either broad peaks or two peaks
without baseline recovery between them were observed,
suggesting a slow equilibrium between two or more species.
As indicated above, the cyclic compounds, namely imino-
cyclitols, were also obtained and structurally characterized. To
this end, the phosphate group of compounds 9 16 was
removed and the resulting products were hydrogenated
following the procedure described in the Experimental
Section.[45] During this step, we assumed, according to the
literature data,[6, 17, 46] that the stereochemistry at C-3 and C-4
was conserved and that no epimerization occurred. Aqueous
solutions of the iminocyclitol in its free-base form were
adjusted to pH 6.5, lyophilized, and submitted to NMR
analysis without any further purification process.
This allowed us to identify unequivocally and quantify the
diastereoisomers thus formed and therefore to elucidate the
stereoselectivity of the enzymatic aldol condensation towards
the aldehyde acceptor. The identified cyclic products from the
corresponding linear compounds are depicted in Table 3.
Most of them are well described in the literature (Table 3) and
have been either isolated from natural sources or synthesized
by chemical or chemoenzymatic methodologies. Remarkably,
to the best of our knowledge, iminocyclitols 22 and 26 and the
minor products 18 and 23 have been obtained and charac-
Table 4. 1H and 13C chemical shifts [ppm] of the five-membered iminosugars 19,
24, and 25.
Product d(H/C)2 d(H/C)3 d(H/C)4 d(H/C)5 d(H/C)5' d(H/C)6 d(H/C)6'
19/24
3.3/66.6 3.9/76.5 4.2/75.1 3.4/50.2 3.1/50.2 3.8/59.7 3.7/59.7
3.6/63.2 4.2/70.5 4.4/70.7 3.4/47.9 3.1/47.9 3.9/58.4 3.8/58.4
25
Table 5. 1H and 13C chemical shifts [ppm] of the five-membered iminosugars 20,
21, and 26 28.
Product d(H/C)2 d(H/C)3 d(H/C)4 d(H/C)5 d(H/C)6 d(H/C)7 d(H/C)7'
21/26
28
27/20
3.3/62.4 3.3/75.8 3.7/80.3 3.3/57.3 1.3/15.4 3.8/59.6 3.7/59.6
3.7/61.5 4.3/70.4 3.9/76.7 3.5/56.7 1.3/15.0 3.9/58.4 3.8/58.4
3.5/67.5 4.0/76.3 4.1/76.3 3.7/58.3 1.3/10.7 3.8/59.6 3.7/59.6
4894
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4887 4899

Stereoselective Aldol Additions
4887 4899
hydroxyl group, regardless of the relative stereochemistry of
the other substituents (Scheme 6). Hence, the stereochemistry
observed at C-2 was controlled exclusively by the configu-
ration at C-4.
strated that, in most instances, the low yields obtained are not
exclusively due to the structural features of the acceptor
aldehyde, but to its solubility as well. With the use of
emulsion-based reaction systems it is possible to increase the
efficiency of the aldolase-catalyzed carbon carbon bond
formation with insoluble substrates. Besides the reaction
media, RhuA had higher reactivity towards the N-Cbz
aminoaldehydes than RAMA, whereas the stereoselectivity
of the aldol reaction was higher for the latter. With the N-Cbz
aminoaldehydes tested, the stereoselectivity of RhuA de-
pended on both the structure and stereochemistry of the
acceptor aldehyde.
Scheme 6. Diastereoselectivity of the reductive amination for 9 16.
Experimental Section
Materials: Fructose-1,6-diphosphate aldolase from rabbit muscle (RAMA;
EC 4.1.2.13, crystallized, lyophilized powder, 19.5 UmgÀ1) was from Fluka
(Buchs, Switzerland). Rhamnulose-1-phosphate aldolase (RhuA;
EC 4.1.2.19, 100 UmLÀ1) was kindly donated by Boehringer Mannhein
(Mannheim, Germany). Acid phosphatase (PA, EC 3.1.3.2, 5.3 UmgÀ1) was
from Sigma (St. Louis, USA). Nonionic polyoxyethylene ether surfactant
with an average of 4 moles of ethylene oxide per surfactant molecule
(C14E4) was from Albright and Wilson (Barcelona, Spain). MacroPrep
High Q Support anion-exchange resin was from BioRad (Hercules, USA).
The precursor of dihydroxyacetone phosphate (DHAP), dihydroxyacetone
phosphate dimer bis(ethyl ketal), was synthesized in our lab by a procedure
described by Jung et al.[48] with slight modifications. Deionized water was
used for preparative HPLC and Milli-Q-grade water for both analytical
HPLC and gel emulsion formation. All other solvents and chemicals used
in this work were of analytical grade.
Analysis of the stereochemistry at C-4 of iminocyclitols
17 27 (Table 3) can be used to infer, within the limits of
1
detection by high-field H NMR, the stereoselectivity of the
aldolases towards each of the N-protected aminoaldehydes.
As can be seen in Table 3 (entries 2 4), RAMA exhibited
high stereoselectivity forming single cyclic diastereoisomers
from aldehydes 6 8. Condensation with aldehyde 5 afforded
14% of a minor diastereoisomer (Table 3, entry 1), arising
from the re-face attack of the DHAP RAMA complex on
the aldehyde. This finding has also been observed with a series
pyridine carbaldehydes and diethylaminoacetaldehyde.[28]
Class II RhuA was less diastereoselective than RAMA for
the substrates assayed. Formation of diastereoisomers, epi-
meric at C-4, has also been reported with a series of nonpolar
aliphatic aldehydes.[31] In the present study, the diastereose-
lectivity observed depended on both the structure and the
stereochemistry of the N-Cbz aminoaldehyde. Aldehydes 5, 6,
and 8 gave 10 33% of adducts arising from the reverse si-
face attack on the aldehyde during the enzymatic aldol
condensation (Table 3, entries 5, 6, and 8). Interestingly, the
reaction with (S)-Cbz-alaninal (7) gave exclusively the
expected stereochemistry (Table 3, entry 7), whereas its
HPLC analyses: HPLC analyses were performed on an RP-HPLC
¾
cartridge, 250 Â 4 mm filled with Lichrosphere 100, RP-18, 5 mm from
Merck (Darmstadt, Germany). Samples (50 mg) were withdrawn from the
reaction medium, dissolved with methanol to stop any enzymatic reaction,
and analyzed subsequently by HPLC. The solvent systems were the
following: solvent A: 0.1% (v/v) trifluoroacetic acid (TFA) in H2O;
solvent B: 0.095% (v/v) TFA in H2O:CH3CN 1:4; gradient elution from
10% to 70%B in 30 min, flow rate 1 mLminÀ1
, detection 215 nm.
Retention factor (k') for each acceptor aldehyde and condensation product
are given below.
NMR analysis: High-field 1H and 13C nuclear magnetic resonance (NMR)
¡
¡
analyses were carried out at the Servei de Ressonancia Magnetica Nuclear,
enantiomer
8 gave the lowest diastereoisomeric ratio
¡
Universitat Autonoma de Barcelona on an Avance 500 Bruker spectrom-
(34% de)(Table 3, entry 8).
eter for [D6]DMSO, [D]CCl3, and D2O solutions. Full characterization of
the described compounds was performed using typical gradient-enhanced
two-dimensional experiments (COSY, NOESY, HSQC, and HMBC)
recorded under routine conditions. When possible, NOE data was obtained
from selective one-dimensional NOESY versions using a single pulsed-
field-gradient echo as a selective excitation method and a mixing time of
500 ms. When necessary, proton and NOESYexperiments were recorded at
different temperatures in order to study the different behavior of the
exchange phenomena to avoid the presence of false NOE cross-peaks that
otherwise complicate both structural and dynamic studies. 1H (300 MHz)
The major iminocyclitols obtained from RAMA- and
RhuA-catalyzed reactions are enantiomers, in agreement
with the optical rotations of the pure compounds and the
diastereoisomeric mixtures (Table 3). It is also worth men-
tioning that the reaction media did not influence the stereo-
selectivity of the reactions.
and 13C NMR (75 MHz) spectra were observed with
a Unity-300
spectrometer from Varian (Palo Alto, California, USA) for [D6]DMSO
and [D]CCl3 solutions and were carried out at the Instituto de Inves-
tigaciones QuÌmicas y Ambientales-CSIC.
Conclusion
In summary, the physicochemical characteristics of the
reaction medium are of paramount importance for the
enzymatic activity and reaction yield in aldolase-catalyzed
stereoselective carbon carbon bond formation. In this sense,
emulsion-based reaction systems are a promising approach to
overcome the problems facing the conventional aqueous/co-
solvent mixtures when using water-insoluble substrates. In the
particular case of N-Cbz aminoaldehydes, we have demon-
Elemental analyses and specific rotations: Elemental analyses were
¡
performed by the Servei de Microanalisi Elemental IIQAB-CSIC. Specific
rotations were measured with a Perkin Elmer Model 341 (‹berlingen,
Germany) Polarimeter.
Molecular modeling: All molecular simulations were conducted with the
program MOE (v. 2002.03, Chemical Computing Group, Montreal) using
the implemented MMFF94 force field with its standard atomic charges and
parameters. All energy calculations were carried out with the default
parameters provided with the program, a distance-dependent dielectric
Chem. Eur. J. 2003, 9, 4887 4899
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4895

¬
P. Cla pes et al.
FULL PAPER
model for the electrostatic interactions, and a smoothed cut-off between 8
and 12 ä to model the nonbonded interactions. All minimizations were
carried up to an RMS gradient <0.01.
general procedure. M.p. 82 848C; the m.p. and 1H and 13C NMR spectra of
53]
these products were consistent with those reported in the literature.[51
(R)-Benzyl-2-hydroxy-1-methylethylcarbamate ((R)-Cbz-d-alaninol) (4):
The title compound (2.0 g) was obtained in 86% yield by using the above
general procedure. M.p. 82 848C; 1H NMR (200 MHz, [D]CCl3, 258C):
d ˆ 7.2 (m, 5H; Ph), 5.1 (s, 2H; CH2Ph), 5.0 (br, 1H; NH), 3.9 (m, 1H;
NHCH), 3.7 (m, 1H; CH2OH), 3.5 (m, 1H; CH2OH), 1.2 ppm (d, 3H;
CH3); 13C NMR: d ˆ 136.0 (quat.C), 128.5, 128.1, 128.0 (CH, Ph), 156.5
(OCONH), 66.8 (CH2OH ‡ CH2Ph), 48.9 (CH), 17.2 ppm (CH3).
Synthesis of N-protected aminoaldehydes: The synthesis of N-protected
aminoaldehydes was achieved by oxidation of the corresponding N-
protected aminoalcohols. Two oxidation procedures were used as described
below.
The coordinates for rabbit fructose-1,6-diphosphate aldolase covalently
bound to reduced DHAP (rDHAP)[33] and for E. coli rhamnulose-1-
phosphate aldolase complexed with phosphoglycolohydroxamic acid
(PGH)[49] were obtained from the Protein Data Bank[50] at Brookhaven
National Laboratory (entries 1J4E and 1GT7, respectively). The 1J4E
structure is a homotetramer while the 1GT7 structure contains five
homotetramers in the asymmetric unit. In addition, the 1J4E structure
contains four Ala residues at positions 72, 239, 289, and 338 instead of the
Cys residues found in the wild-type enzyme.[33] The simulations on RAMA
were performed with chain A of 1J4E, after removing all water molecules,
adding all the hydrogen atoms, and mutating the mentioned four Ala
residues back to cysteine. The conformation of the side chains of these Cys
residues was determined by minimization keeping the backbone coordi-
nates fixed. Residues 1 3 and 345 363 were missing in the crystal
structure and, therefore, they were not included in the simulations. The
covalently bound rDHAP moiety was modified within MOE to convert it
into the ligand shown in state II of Scheme 2. The conformational space of
this intermediate was explored to find low-energy minima by running a
stochastic conformational search keeping the coordinates of the protein
fixed, except for the side chain of Lys229, and residues Asp33 and Lys229
charged as shown in Scheme 2. All conformations within 10 kcalmolÀ1 of
the global minimum were kept and further minimized; this allowed motion
of all protein residues that contained at least one atom within 12 ä from the
bound ligand. The lowest-energy minimum found in this way was
considered the global minimum. The structure of this ligand was further
Swern oxidation with slight modifications according to Konradi et al.:[54]
A
solution of dimethylsulfoxide (7.7 18 mmol) in dried dichloromethane
(2 4 mL) was added dropwise under argon to a cooled solution (À658C)
of oxalyl chloride (7.7 18 mmol) in dried methylene chloride (2 4 mL).
After stirring for 10 min, a solution of the Cbz-aminoalcohol (5.2
12 mmol) in dried methylene chloride (4 mL) was added slowly whilst
maintaining the temperature at À658C. The solution was stirred for 45 min
at À658C and then triethylamine (51 mmol) was added dropwise. After
being stirred for an additional 10 min at À658C, cold ethyl acetate (50
100 mL) and cold brine solution (25 50 mL) were added. The organic
layer was washed with brine solution (3 Â 50 100 mL), dried over Na2SO4,
and evaporated to dryness under reduced pressure. The residue was further
purified by crystallization or by preparative HPLC methodology.
Benzyl 3-oxoproylcarbamate (5):[52] The title compound (2.1 g, 84% yield)
was prepared according to the general procedure above. HPLC: k' ˆ 8.7;
m.p. 55 578C; 1H NMR (300 MHz, [D]CCl3, 258C): d ˆ 9.7 (s, 1H; CHO),
7.2 (m, 5H; Ph), 5.2 (s, 1H; NH), 5.0 (s, 2H; CH2O), 3.3 (q, 2H; NHCH2),
2.6 ppm (t, 2H; CH2CHO); 13C NMR (75 MHz, CDCl3, 258C): d ˆ 201.0
(CHO), 157.0 (CONH), 136.0 (quat.C), 128.5, 128.1, 128.0 (CH, Ph), 66.7
(CH2Ph), 44.0 (NHCH2), 34.5 ppm (CH2CHO).
À
modified to generate the situation previous to the C C bond formation,
represented by state I in Scheme 2, in which the Cbz-alaninal molecule is
close to the enamine bound to Lys229. To avoid that, the aldehyde molecule
was removed away from this enamine and a restraint was imposed between
À
the two carbon atoms that participate in the C C bond formation
(designated with an asterisk in state I of Scheme 2). This restraint was
arbitrarily set to keep the distance between those two carbon atoms close to
2.5 ä. The conformational space of Cbz-alaninal was then stochastically
searched as before, keeping the coordinates of the rest of the atoms fixed.
Again, the conformations obtained were further minimized and the
geometry of the global lowest-energy minimum determined.
Benzyl 2-oxoethylcarbamate (Cbz-glycinal) (6): The reaction workup
provided an oil (1.9 g, 95% yield, 85% pure by HPLC), which was further
purified by reversed-phase HPLC. The oil (1.9 g) was dissolved in DMSO,
loaded onto the preparative PrePack cartridge (47 Â 300 mm) filled with a
Bondapack C18, 300 ä, 15 20 mm stationary phase, and eluted with a
CH3CN gradient (0 to 20% in 30 min) in plain water. The flow rate was
100 mLminÀ1 and the products were detected at either 215 or 225 nm.
Analysis of the fractions was accomplished by analytical RP chromatog-
raphy using the same elution conditions, flow rate, and detection as
previously described. Pure fractions were pooled and lyophilized to yield 6
(0.65 g, 33% overall yield, 99.9% pure by HPLC) as an oil. 1H NMR data
are consistent with those reported in the literature;[55] HPLC: k' ˆ 7.6 ;
13C NMR (75 MHz, [D6]DMSO, 258C): d ˆ 200.2 (CHO), 156.6 (CONH),
136.9 (quat.C), 128.4, 128.3, 127.8 (CH, Ph), 65.5 (CH2O), 50.6 ppm
(CH2CHO).
For the simulations on RhuA we used chains A and B of 1GT7, since the
active center of RhuA is located at the interface between two monomeric
units of the protein. The PGH molecule was modified conveniently to
obtain the a-hydroxyketonic ligand complexed to the Zn2‡ ion as shown in
state IV of Scheme 2. After finding the global minimum in the same way as
before, this bound ligand was modified to generate the state before the
À
C C bond formation (state III, Scheme 2). The conformational space was
searched again imposing a restraint to keep the distance between the two
reactive carbon atoms close to 2.5 ä.
In all cases the RMSD calculated for the residues of the protein that were
allowed to relax during the minimization was between 0.6 and 1 ä relative
to the conformation of the same residues in the crystallographic structures.
(S)-Benzyl 1-methyl-2-oxoethylcarbamate ((S)-Cbz-alaninal) (7): The title
compound (0.45 g, 90% yield) was obtained as a pale yellow oil by using
the general procedure above. HPLC: k' ˆ 8.8; [a]2D0 ˆ À16.3 (c ˆ 1 in
MeOH); 1H NMR (300 MHz, [D6]DMSO, 258C): d ˆ 9.5 (1H, s, CHO), 5.0
(2H, s; CH2O), 4.0 (1H, m; CH), 1.2 ppm (3H, d; CH3); 13C NMR
(75 MHz, [D6]DMSO, 258C): d ˆ 201 (CHO), 156.1 (CONH), 65.7 (CH2O),
55.1 (CH), 13.8 ppm (CH3).
Synthesis of N-protected amino alcohols
Cbz-amino alcohols: Cbz-OSu (11 56 mmol) in dioxane/water 4:1 (10
50 mL) was added to a solution of the aminoalcohol (13 67 mmol) in
dioxane/water 4:1 (50 150 mL). After stirring for 12 h, the mixture was
evaporated to dryness under reduced pressure. The residue was dissolved
with ethyl acetate and washed successively with citric acid 5% (w/v) (3 Â
50 75 mL), NaHCO3 10% (w/v) (3 Â 50 75 mL), and brine (3 Â 50
75 mL). After drying over Na2SO4, the organic layer was evaporated
under reduced pressure. NMR spectrum and HPLC analysis of the residue
showed only one product.
Oxidation with 2-iodoxybenzoic acid (IBX): 2-Iodoxybenzoic acid was
prepared as described in the literature.[24, 56] Caution! IBX has been
reported to detonate upon heavy impact and/or heating over 2008C.
(S)-Benzyl 1-methyl-2-oxoethylcarbamate ((S)-Cbz-alaninal) (7): IBX (24-
48 mmol) was added to a solution of (S)-Cbz-alaninol (10 19 mmol) in
DMSO (60 120 mL). The reaction was monitored by HPLC until no
alcohol was detected. At this point, the reaction mixture was diluted with
water (30 60 mL) and the mixture was extracted with ethyl acetate (3 Â
75 100 mL). The organic layers were pooled, washed with NaHCO3 5%
(w/w) (3 Â 100 mL) and brine (3 Â 100 mL), dried over Na2SO4, and
evaporated under reduced pressure to give the title compound (2.5 g, 63%
yield) as a colorless oil. HPLC: k' ˆ 8.8; [a]2D0 ˆ À19.9 (c ˆ 1 in MeOH);
1H NMR and 13C NMR were identical to those described above.
Benzyl 3-hydroxypropylcarbamate (1): The title compound (9.6 g) was
obtained in 82% yield by using the above general procedure. M.p. 48
528C.
Benzyl 2-hydroxyethylcarbamate (Cbz-glycinol) (2): The title compound
(8.9 g) was obtained in 89% yield by using the above general procedure.
M.p. 57 638C.
(S)-Benzyl-2-hydroxy-1-methylethylcarbamate ((S)-Cbz-l-alaninol) (3):
(R)-Benzyl 1-methyl-2-oxoethylcarbamate ((R)-Cbz-alaninal) (8): The
The title compound (7.9 g) was obtained in 90% yield by using the above
title compound (1.4 g, 74% yield) was obtained as a colorless oil by using
4896
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4887 4899

Stereoselective Aldol Additions
4887 4899
the procedure described above except starting from (R)-Cbz-alaninol.
HPLC: k' ˆ 8.8; [a]2D0 ˆ 18.7 (c ˆ 1 in MeOH); 1H NMR (300 MHz,
[D6]DMSO, 258C): d ˆ 9.5 (s, 1H; CHO), 5.0 (s, 2H; CH2O), 4.0 (m, 1H;
CH), 1.2 ppm (d, 3H; CH3); 13C NMR (75 MHz, [D6]DMSO, 258C): d ˆ
201.0 (CHO), 156.1 (CONH), 65.7 (CH2O), 55.1 (CH), 13.8 ppm (CH3).
C14H18NO9Na2P ¥ 2H2O: C 36.77, H 4.85, N 3.06; found: C 36.59, H 4.94, N
3.04.
(3S)-5-{[(Benzyloxy)carbonyl]amino}-5-deoxy-1-O-phosphonopent-2-
ulose sodium salt (10): The title compound (416 mg) was obtained in 50%
1
yield. HPLC: k' ˆ 4.1; m.p. 114 1188C; H NMR (500 MHz, D2O, 12 8C)
linear product: d ˆ 7.2 (m, 5H; Ph), 4.9 (s, 2H; PhCH2), 4.5 (dd, 3J(H,H) ˆ
Enzymatic reactions in emulsions: Reactions were carried out in 10-mL test
tubes with screw caps. The aldehyde (0.125, 0.5, or 0.9 mmol), the oil (6%
w/w), and the surfactant (4% w/w) were mixed vigorously. Then, the
DHAP solution (0.075 0.3 mmol) at pH 6.9, freshly prepared as described
by Effenberger et al.,[57] was added dropwise while stirring at 258C with a
vortex mixer. A reaction volume of 2.5 mL or 5 mL was employed for
analytical or preparative scale, respectively. Finally, RAMA (0.65 4 mg
giving 0.26 1.6 mggÀ1 in the final reaction mixture, 19.5 UmgÀ1) or RhuA
3
6.3 Hz, 3J(H,H) ˆ 18.7 Hz, 2H; CH2OP), 4.3 (d, J3,4(H,H) ˆ 1.8, 1H;
CHOH), 4.0 (dt, 3J4,5(H,H) ˆ 5.8 Hz, 3J4,5'(H,H) ˆ 7.7 Hz, 1H; CHOH), 3.1
3
(dd, 1H, J5,5'(H,H) ˆ 13.8 Hz, 1H; NHCH2), 3.0 ppm (dd, 1H; NHCH2);
cyclic species, conformer I: d ˆ 7.2 (m, 5H; Ph), 4.9 (s, 2H; PhCH2), 4.1 (d,
1H; CHOH), 4.1 (d, 1H; CH2OP), 3.9 (dd, 1H; CHOH), 3.8 (dd, 1H;
NHCH2), 3.6 (d, 1H; CH2OP), 2.9 ppm (t, 1H; NHCH2); cyclic species,
conformer II: d ˆ 5.0 (s, 2H; PhCH2), 4.1 (d, 1H; CHOH), 3.9 (d, 1H;
CH2OP), 3.9 (dd, 1H; CHOH), 3.8 (dd, 1H; NHCH2), 3.5 (d, 1H; CH2OP),
2.8 ppm (t, 1H; NHCH2); 13C NMR (125 MHz, D2O, 12 8C): d ˆ 210.8 (C2),
158.7 (C6), 136.4 (Ar), 127.9 (Ar), 75.6 (C3), 69.4 (C4), 67.4 (C7), 67.6 (C1),
41.9 ppm (C5); elemental analysis calcd (%) for C13H16NO9Na2P ¥ 2H2O: C
35.23, H 4.55, N 3.16; found: C 35.24, H 4.47, N 3.07.
(20 40 mL giving 4 16 mLmLÀ1 in the final reaction mixture, 100 UmLÀ1
)
was added and mixed again. The test tubes were placed on a horizontal
shaking bath (100 rpm) at constant temperature (258C). The reactions were
followed by HPLC until the peak of the product reached a maximum. The
enzymatic reactions were stopped by addition of MeOH, and the final
crude purified as described above. The reactions at semipreparative level
were scale up proportionally to a total reaction volume of 5 mL.
(3S,5S)-5-{[(Benzyloxy)carbonyl]amino}-5,6-dideoxy-1-O-phosphonohex-
2-ulose sodium salt (11): The title compound (33 mg) was obtained in 10%
yield. HPLC: k' ˆ 4.5; m.p. 117 1208C; elemental analysis calcd (%) for
C14H18NO9Na2P ¥ 3³2 H2O: C 37.48, H 4.68, N 3.12; found: C 37.41, H 4.82, N
3.06.
Enzymatic reactions in mixtures dimethylformamide/water 1:4: Reactions
were carried out in 10-mL test tubes with screw caps. The aldehyde (0.125
0.36 mmol) was dissolved in DMF 20% (v/v). Then, the DHAP solution
(0.075 0.225 mmol), prepared as described above, was added dropwise
while mixing. The rest of the experimental procedure was identical to that
described for the reaction in emulsions.
(3S,5R)-5-{[(Benzyloxy)carbonyl]amino}-5,6-dideoxy-1-O-phosphonohex-
2-ulose sodium salt (12): The title compound (230 mg) was obtained in
46% yield. HPLC: k' ˆ 4.5, m.p. 119 1238C; elemental analysis calcd (%)
for C14H18NO9Na2P ¥ 1³2 H2O ¥ NaCl: C 36.05, H 4.29, N 3.00; found: C 36.15,
H 4.61; N 2.98.
Purification procedures: The condensation products were purified by ion-
exchange chromatography on an FPLC system taking advantage of the
phosphate group. Bulk stationary phase MacroPrep High Q Support,
50 mm strong anion-exchange resin in neutral form, ClÀ as counter ion, was
packed into a glass Pharmacia FPLC (Uppsala, Sweden) C26/40-type
(3R)-6-{[(Benzyloxy)carbonyl]amino}-5,6-dideoxy-1-O-phosphonohex-2-
ulose sodium salt (13): The title compound (330 mg) was obtained in 40%
yield. HPLC: k' ˆ 5.2, m.p. 109 1118C; elemental analysis calcd (%) for
C14H18NO9Na2P ¥ 3H2O: C 35.38, H 5.09, N 2.95; found: C 35.55, H 5.04, N
3.01.
column to a final bed volume of 40 mL. The flow rate was 3 mLminÀ1
,
0.1 MPa. The resin was equilibrated initially with EtOH/H2O (3:2).
Aliquots of the crude reaction mixture were loaded onto the column and
the excess aldehyde and neutral and cationic impurities were eliminated by
washing with EtOH/H2O (3:2). Then the target product was eluted with
250 mm NaCl in EtOH/H2O (3:2). The operation was repeated until the
whole crude was consumed. Fractions containing the pure product were
pooled and evaporated to dryness under reduced pressure. The residue was
dissolved in water and desalted by reversed-phase HPLC on a Perkin
Elmer semipreparative 250 Â 25 mm column, filled with C18, 5 mm-type
stationary phase by using the following procedure: first, the column was
equilibrated with 0.1% (v/v) TFA in water. The sample was then adjusted
to pH 3 4 with TFA and loaded onto the column. First, the salt was
eliminated by washing with 0.1% (v/v) TFA in water (100 mL) and then
with plain water (50 mL) to eliminate the acid. Then, the product was
eluted with a gradient elution of CH3CN (0 to 48% v/v in 30 min) in plain
water. The flow rate was 10 mLminÀ1 and the products were detected at
225 nm. The pure fractions were pooled, the excess CH3CN was evaporated
under reduced pressure, the pH of the resulting aqueous solution was
adjusted to 7 with NaOH and finally lyophilized.
(3R)-5-{[(Benzyloxy)carbonyl]amino}-5-deoxy-1-O-phosphonopent-2-
ulose sodium salt (14): The title compound (229 mg) was obtained in 53%
yield. HPLC: k' ˆ 4.1, m.p. 114 1188C; elemental analysis calcd (%) for:
C13H16NO9Na2P ¥ 3³2 H2O: C 35.93, H 4.38, N 3.22; found: C 35.95, H 4.24, N
3.20.
(3R,5S)-5-{[(Benzyloxy)carbonyl]amino}-5,6-dideoxy-1-O-phosphonohex-
2-ulose sodium salt (15): The title compound (158 mg) was obtained in
35% yield. HPLC: k' ˆ 4.9, m.p. 132 1368C; elemental analysis calcd (%)
for: C14H18NO9Na2P ¥ 3H2O: C 35.38, H 5.09, N 2.95; found: C 35.13, H
4.95, N 2.95.
(3R,5R)-5-{[(Benzyloxy)carbonyl]amino}-5,6-dideoxy-1-O-phosphono-
hex-2-ulose sodium salt (16): The title compound (255 mg) was obtained in
45% yield. HPLC: k' ˆ 4.5, m.p. 130 1328C; elemental analysis calcd (%)
for: C14H18NO9Na2P ¥ 3³2 H2O: C 37.48, H 4.68, N 3.12; found: C 37.34, H
4.70, N 3.03.
Removal of phosphate groups and catalytic hydrogenation: The phosphate
group of compounds 9 16 was removed by hydrolysis catalyzed by acid
phosphatase following the procedure described by Bednarski et al.[27] The
reaction was followed by HPLC until no starting material was detected.
Then the crude was desalted by HPLC and lyophilized. The residue was
dissolved in methanol and the solution hydrogenated under 50 psi H2 for
24 h with Pd/C as catalyst. The reaction mixture was then filtered through
neutral alumina, the solvent removed under reduced pressure, the residue
dissolved in plain water, adjusted to pH 6.5 and lyophilized.
Physical and spectral data of the linear compounds linear compounds 9
16: The melting points of the compounds given below correspond to
lyophilized solids rather than crystals. Note that some of them are mixtures
of epimers at C-4. The yields correspond to the amounts obtained from the
aldol enzymatic reactions at semipreparative level. The purification
procedures were not optimized.
(3S)-6-{[(Benzyloxy)carbonyl]amino}-5,6-dideoxy-1-O-phosphonohex-2-
ulose sodium salt (9): The title compound (540 mg) was obtained in 40%
(2R,3R,4R)-2-(Hydroxymethyl)piperidine-3,4-diol (17) and (2S,3R,4S)-2-
(hydroxymethyl)piperidine-3,4-diol (18): The title compounds were pre-
pared according to the general procedure described above. 1H NMR
(500 MHz, D2O, 25 8C) major 17: d ˆ 3.84 (dd, 3J(H,H) ˆ 3.2 Hz, 3J(H,H) ˆ
12.6 Hz, 1H; H7), 3.77 (dd, 3J(H,H) ˆ 5.7 Hz, 3J(H,H) ˆ 12.7 Hz, 1H; H7),
3.6 (dt, 3J(H,H) ˆ 4.9 Hz,3J(H,H) ˆ 9.5 Hz, 3J(H,H) ˆ 11.6, 1H; H4), 3.4 (t,
1H; H3), 3.3 (br, 1H; H6), 3.0 (m, 1H; H2), 2.95 (dt, 3J(H,H) ˆ 3.2 Hz,
3J(H,H) ˆ 13.1 Hz, 1H; H6), 2.1 (br, 1H, H5), 1.6 ppm (dq, 3J(H,H) ˆ
4.3 Hz, 1H; H5); minor 18: d ˆ 4.0 (1H; H3), 3.80 (1H; H4), 3.2 (1H;
H2), 1.95 (1H; H5), 1.85 (1H; H5), 3.76 (dd, 1H; H7), 3.72 (dd, 1H; H7),
2.95 (1H; H6), 3.0 ppm (1H; H6); 13C NMR (125 MHz, D2O, 25 8C) major
17: d ˆ 70.5 (C4), 69.7 (C3), 60.0 (C2), 57.9 (C7), 41.7 (C6), 28.8 ppm (C5);
1
yield. HPLC: k' ˆ 5.2; m.p. 109 1118C; H NMR (500 MHz, D2O, 25 8C)
major diastereoisomer: d ˆ 7.2 (m, 5H; Ph), 4.9 (s, 2H; PhCH2), 4.6 (dd,
3J(H,H) ˆ 6.3 Hz, 3J(H,H) ˆ 18.6 Hz, 1H; CH2OP), 4.5 (dd, 3J(H,H) ˆ
3
7.0 Hz, J(H,H) ˆ 18.6 Hz, 1H; CH2OP), 4.25 (s, 1H; CHOH), 4.0 (t, 1H;
CHOH), 3.1 (m, 2H; NHCH2), 1.6 ppm (m, 2H; CH2CHOH); minor
diastereoisomer: d ˆ 7.2 (m, 5H, Ph), 5.0 (s, 2H; PhCH2), 4.27 (1H;
CHOH), 3.82 (1H; CHOH), 3.1 (m, 2H; NHCH2), 1.5 ppm (m, 2H;
CH2CHOH); cyclic product: d ˆ 3.8 (1H; CH2OP), 3.65 ppm (1H;
CH2OP); 13C NMR (125 MHz, D2O, 25 8C) major diastereoisomer: d ˆ
210.8, (C2), 158.2 (C7), 136.4, 128.0 (Ar), 78.4 (C3), 69.1 (C4), 68.4 (C1),
67.2 (C8), 37.4 (C6), 33.0 ppm (C5); elemental analysis calcd (%) for
Chem. Eur. J. 2003, 9, 4887 4899
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4897

¬
P. Cla pes et al.
FULL PAPER
minor 18: d ˆ 67.6 (C4), 66.2 (C3), 60.3 (C2), 59.9 (C7), 42.2 (C6), 24.5 ppm
3H; H6); 13C NMR (125 MHz, D2O, 25 8C): d ˆ 80.3 (C4), 75.8 (C3), 62.4
(C5).
(C2), 59.6 (C7), 57.3 (C5), 15.4 ppm (C6).
(2S,3S,4S,5R)-2-Hydroxymethyl-5-methylpyrrolidine-3,4-diol (27) and
(2R,3S,4R,5R)-2-hydroxymethyl-5-methylpyrrolidine-3,4-diol (28): The ti-
tle compounds were prepared according to the general procedure described
above. 1H NMR (500 MHz, D2O, 25 8C) major 27: d ˆ 4.1 (dd, 3J(H,H) ˆ
1.4 Hz, 3J(H,H) ˆ 3.2 Hz, 1H; H4), 4.0 (dd, 3J(H,H) ˆ 1.8 Hz, 3J(H,H) ˆ
3.2 Hz, 1H; H3), 3.8 (dd, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ 11.9 Hz, 1H; H7),
(2R,3R,4R)-2-(Hydroxymethyl)pyrrolidine-3,4-diol (19): The title com-
pound was prepared according to the general procedure described above.
3
3
1H NMR (500 MHz, D2O, 25 8C): d ˆ 4.2 (m, J(H,H) ˆ 2.8 Hz, J(H,H) ˆ
3
3
3.9 Hz, J(H,H) ˆ 5.0 Hz, 1H; H4), 3.9 (t, J(H,H) ˆ 3.9 Hz, 1H; H3), 3.8
(dd, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ 11.9 Hz, 1H; H6), 3.7 (dd, 3J(H,H) ˆ
7.0 Hz, 3J(H,H) ˆ 11.9 Hz, 1H; H6), 3.4 (dd, 3J(H,H) ˆ 5.0 Hz, 3J(H,H) ˆ
12.3 Hz, 1H; H5), 3.3 (m, 3J(H,H) ˆ 3.9 Hz, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ
7.3 Hz, 1H; H2), 3.1 ppm (dd, 3J(H,H) ˆ 2.8 Hz, 3J(H,H) ˆ 12.3 Hz, 1H;
H5); 13C NMR (125 MHz, D2O): d ˆ 76.5 (C3), 75.1 (C4), 66.6 (C2), 59.7
(C6), 50.2 ppm (C5).
3
3
3
3.75 (dq, J(H,H) ˆ 3.5 Hz, J(H,H) ˆ 6.5 Hz, 1H; H5), 3.7 (dd, J(H,H) ˆ
8.8 Hz, 3J(H,H) ˆ 12.3 Hz, 1H; H7), 3.5 (dt, 3J(H,H) ˆ 3.3 Hz, 3J(H,H) ˆ
5.1 Hz, 3J(H,H) ˆ 8.2 Hz, 1H; H2), 1.3 ppm (d, 3J(H,H) ˆ 6.7 Hz, 3H; H6);
minor 28: d ˆ 4.3 (t, 3J(H,H) ˆ 3.7 Hz, 1H; H3), 3.95 (dd, 3J(H,H) ˆ 3.7 Hz,
3J(H,H) ˆ 9 Hz, 1H; H4), 3.9 (dd, 3J(H,H) ˆ 4.7 Hz, 3J(H,H) ˆ 11.8 Hz,
3
3
(2R,3R,4R,5S)-2-Hydroxymethyl-5-methylpyrrolidine-3,4-diol (20): The
title compound was prepared according to the general procedure described
above. 1H NMR (500 MHz, D2O, 25 8C): d ˆ 4.1 (dd, 3J(H,H) ˆ 1.4 Hz,
3J(H,H) ˆ 3.2 Hz, 1H; H4), 4.0 (dd, 3J(H,H) ˆ 1.8, 3J(H,H) ˆ 3.2, 1H; H3),
3.8 (dd, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ 11.9 Hz, 1H; H7), 3.75 (dq, 3J(H,H) ˆ
3.5 Hz, 3J(H,H) ˆ 6.5 Hz, 1H; H5), 3.7 (dd, 3J(H,H) ˆ 8.8 Hz, 3J(H,H) ˆ
12.3 Hz, 1H; H7), 3.5 (dt, 3J(H,H) ˆ 3.3 Hz, 3J(H,H) ˆ 5.1 Hz, 3J(H,H) ˆ
8.2 Hz, 1H; H2), 1.3 ppm (d, 3J(H,H) ˆ 6.7 Hz, 3H; H6); 13C NMR
(125 MHz, D2O, 25 8C): d ˆ 76.3 (C4), 76.3 (C3), 67.5 (C2), 59.6 (C7), 58.3
(C5), 10.7 ppm (C6).
1H; H7), 3.8 (dd, J(H,H) ˆ 8.4 Hz, J(H,H) ˆ 11.8 Hz, 1H; H7), 3.75 (m,
3J(H,H) ˆ 3.7 Hz, 3J(H,H) ˆ 4.7 Hz, 3J(H,H) ˆ 8.4 Hz, 1H; H2), 3.5 (m,
1H; H5), 1.3 ppm (d, 3J(H,H) ˆ 6.7 Hz, 3H; H6); 13C NMR (125 MHz,
D2O, 25 8C) major 27: d ˆ 76.3 (C4), 76.3 (C3), 67.5 (C2), 59.6 (C7), 58.3
(C5), 10.7 ppm (C6); minor 28: d ˆ 76.7 (C4), 70.4 (C3), 61.5 (C2), 58.4
(C7), 56.7 (C5), 15.0 ppm (C6).
Acknowledgement
Financial support from the Spanish M.C.Y.T. Plan Nacional (grants QUI99-
0997CO2-01, BIO99-1219-C02-02 and PPQ2002-04625-C02-01) and from
Generalitat de Catalunya (grant 2000SGR-00357) is acknowledged.
(2R,3R,4R,5R)-2-Hydroxymethyl-5-methylpyrrolidine-3,4-diol (21): The
title compound was prepared according to the general procedure described
above. 1H NMR (500 MHz, D2O, 25 8C): d ˆ 3.9 (t, 3J(H,H) ˆ 7 Hz, 1H;
H3), 3.8 (dd, 3J(H,H) ˆ 4.2 Hz, 3J(H,H) ˆ 12.0 Hz, 1H; H7), 3.75 (t,
3J(H,H) ˆ 7.5 Hz, 1H; H4), 3.7 (dd, 3J(H,H) ˆ 6.4 Hz, 3J(H,H) ˆ 12.0 Hz,
1H; H7), 3.35 (dt, 3J(H,H) ˆ 4.2 Hz, 3J(H,H) ˆ 6.4 Hz, 3J(H,H) ˆ 7.0 Hz,
1H; H2), 3.3 (dq, 3J(H,H) ˆ 7.0 Hz, 1H; H5), 1.3 ppm (d, 3J(H,H) ˆ 7.0 Hz,
3H; H6); 13C NMR (75 MHz, D2O, 25 8C): d ˆ 80.3 (C4), 75.8 (C3), 62.4
(C2), 59.6 (C7), 57.3 (C5), 15.4 ppm (C6).
[1] T. D. Machajewski, C. H. Wong, Angew. Chem. 2000, 112, 1406 1430;
Angew. Chem. Int. Ed. 2000, 39, 1353 1374.
[2] W. D. Fessner, V. Helaine, Curr. Opin. Biotechnol. 2001, 12, 574 586.
[3] C.-H. Wong, R. L. Halcomb, Y. Ichikawa, T. Kajimoto, Angew. Chem.
1995, 107, 453 474; Angew. Chem. Int. Ed. Engl. 1995, 34, 412 432.
[4] G. C. Look, C. H. Fotsch, C. H. Wong, Acc. Chem. Res. 1993, 26, 182
190.
[5] C. H. Wong, Trends Biotechnol. 1992, 10, 337 341.
[6] S. Takayama, R. Martin, J. Wu, K. Laslo, G. Siuzdak, C.-H. Wong, J.
Am. Chem. Soc. 1997, 119, 8146 8151.
[7] M. H. Fechter, A. E. Stutz, A. Tauss, Curr. Org. Chem. 1999, 3, 269
285.
[8] C.-H. Wong, L. Provencher, J. A. Porco, Jr., S.-H. Jung, Y.-F. Wang, L.
Chen, R. Wang, D. H. Steensma, J. Org. Chem. 1995, 60, 1492 1501.
[9] A. E. Stutz, Iminosugars as Glycosidase Inhibitors. Nojirimicin and
Beyond, Wiley-VCH, Weinheim, 1999.
(2S,3S,4S)-2-(Hydroxymethyl)piperidine-3,4-diol (22) and (2R,3S,4R)-2-
(hydroxymethyl)piperidine-3,4-diol (23): The title compounds were pre-
pared according to the general procedure described above. 1H NMR
(500 MHz, D2O, 25 8C) major 22: d ˆ 3.84 (dd, 3J(H,H) ˆ 3.2, 3J(H,H) ˆ
12.6 Hz, 1H; H7), 3.77 (dd, 3J(H,H) ˆ 5.7 Hz, 3J(H,H) ˆ 12.7 Hz, 1H; H7),
3.6 (dt, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ 9.5 Hz, 3J(H,H) ˆ 11.6 Hz, 1H; H4),
3.4 (t, 1H; H3), 3.3 (br, 1H; H6), 3.0 (m, 1H; H2), 2.95 (dt, 3J(H,H) ˆ
3.2 Hz, 3J(H,H) ˆ 13.1 Hz, 1H; H6), 2.1 (br, 1H; H5), 1.6 ppm (dq,
3J(H,H) ˆ 4.3 Hz, 1H; H5); minor 23: d ˆ 4.0 (1H; H3), 3.80 (1H; H4), 3.2
(1H; H2), 1.95 (1H; H5), 1.85 (1H; H5), 3.76 (dd, 1H; H7), 3.72 (dd, 1H;
H7), 2.95 (1H; H6), 3.0 ppm (1H; H6); 13C NMR (125 MHz, D2O, 25 8C)
major 22: d ˆ 70.5 (C4), 69.7 (C3), 60.0 (C2), 57.9 (C7), 41.7 (C6), 28.8 ppm
(C5); minor 23: d ˆ 67.6 (C4), 66.2 (C3), 60.3 (C2), 59.9 (C7), 42.2 (C6),
24.5 ppm (C5).
[10] V. H. Lillelund, H. H. Jensen, X. Liang, M. Bols, Chem. Rev. 2002, 102,
515 553.
[11] R. R. Hung, J. A. Straub, G. M. Whitesides, J. Org. Chem. 1991, 56,
3849 3855.
[12] J. Jurczak, A. Golebiowski, Chem. Rev. 1989, 89, 149 164.
[13] J. Jurczak, D. Gryko, E. Kobrzycka, H. Gruza, P. Prokopowicz,
Tetrahedron 1998, 54, 6051 6064.
[14] M. Schuster, Bioorg. Med. Chem. Lett. 1999, 9, 615 618.
[15] C. H. Von der Osten, A. J. Sinskey, C. F. Barbas, III, R. L. Pederson,
Y. F. Wang, C. H. Wong, J. Am. Chem. Soc. 1989, 111, 3924 3927.
[16] A. Romero, C. H. Wong, J. Org. Chem. 2000, 65, 8264 8268.
[17] R. L. Pederson, C. H. Wong, Heterocycles 1989, 28, 477 480.
[18] L. Azema, F. Bringaud, C. Blonski, J. Perie, Bioorg. Med. Chem. 2000,
8, 717 722.
(2S,3S,4S)-2-(Hydroxymethyl)pyrrolidine-3,4-diol (24) and (2R,3S,4R)-2-
(hydroxymethyl)pyrrolidine-3,4-diol (25): The title compounds were pre-
pared according to the general procedure described above. 1H NMR
(500 MHz, D2O, 25 8C) major 24: d ˆ 4.2 (m, 3J(H,H) ˆ 2.8, 3J(H,H) ˆ
3
3
3.9 Hz, J(H,H) ˆ 5.0 Hz, 1H; H4), 3.9 (t, J(H,H) ˆ 3.9 Hz, 1H; H3), 3.8
(dd, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ 11.9 Hz, 1H; H6), 3.7 (dd, 3J(H,H) ˆ
7.0 Hz, 3J(H,H) ˆ 11.9 Hz, 1H; H6), 3.4 (dd, 3J(H,H) ˆ 5.0 Hz, 3J(H,H) ˆ
12.3 Hz, 1H; H5), 3.3 (m, 3J(H,H) ˆ 3.9 Hz, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ
7.3 Hz, 1H; H2), 3.1 ppm (dd, 3J(H,H) ˆ 2.8 Hz, 3J(H,H) ˆ 12.3 Hz, 1H;
3
3
3
H5); minor 25: d ˆ 4.4 (dt, J(H,H) ˆ 3.8 Hz, J(H,H) ˆ 4.3 Hz, J(H,H) ˆ
7.3 Hz, 1H; H4), 4.2 (t, 1H; H3), 3.9 (dd, 3J(H,H) ˆ 4.9 Hz, 3J(H,H) ˆ
3
3
12.0 Hz, 1H; H6), 3.8 (dd, J(H,H) ˆ 6.6 Hz, J(H,H) ˆ 12.3 Hz, 1H; H6),
3.6 (m, 1H; H2), 3.3 (dd, 3J(H,H) ˆ 4.3 Hz, 3J(H,H) ˆ 12.0 Hz, 1H; H5),
3.4 ppm (dd, 3J(H,H) ˆ 7.3 Hz, 3J(H,H) ˆ 12.3 Hz, 1H; H5); 13C NMR
(125 MHz, D2O, 25 8C) major 24: d ˆ 76.5 (C3), 75.1 (C4), 66.6 (C2), 59.7
(C6), 50.2 ppm (C5); minor 25: d ˆ 70.5 (C3), 70.7 (C4), 63.2 (C2), 58.4
(C6), 47.9 ppm (C5).
[19] S. B. Sobolov, A. Bartoszko-Malik, T. R. Oeschger, M. M. Montelba-
no, Tetrahedron Lett. 1994, 35, 7751 7754.
¬
[20] P. Clapes, L. Espelt, M. A. Navarro, C. Solans, J. Chem. Soc. Perkin
Trans. 2 2001, 1394 1399.
¬
[21] L. Espelt, P. Clapes, J. Esquena, A. Manich, C. Solans, Langmuir 2003,
19, 1337 1346.
(2S,3S,4S,5S)-2-Hydroxymethyl-5-methylpyrrolidine-3,4-diol (26): The ti-
[22] C. Solans, R. Pons, H. Kunieda, in Modern Aspects of Emulsion
Science (Ed.: B. P. Binks), The Royal Society of Chemistry, Cam-
bridge, 1998, pp. 367 394.
[23] Swern oxidation of alcohol 3 furnished 7 with lower optical purity
than that obtained with IBX.
tle compound was prepared according to the general procedure described
1
3
above. H NMR (500 MHz, D2O, 25 8C): d ˆ 3.9 (t, J(H,H) ˆ 7.0 Hz, 1H;
H3), 3.8 (dd, 3J(H,H) ˆ 4.2 Hz, 3J(H,H) ˆ 12.0 Hz, 1H; H7), 3.75 (t,
3J(H,H) ˆ 7.5 Hz, 1H; H4), 3.7 (dd, 3J(H,H) ˆ 6.4 Hz, 3J(H,H) ˆ 12.0 Hz,
1H; H7), 3.35 (dt, 3J(H,H) ˆ 4.2 Hz, 3J(H,H) ˆ 6.4 Hz, 3J(H,H) ˆ 7.0 Hz,
1H; H2), 3.3 (dq, 3J(H,H) ˆ 7.0 Hz, 1H; H5), 1.3 ppm (d, 3J(H,H) ˆ 7.0 Hz,
[24] M. Frigerio, M. Santagostino, S. Sputore, J. Org. Chem. 1999, 64,
4537 4538.
4898
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 4887 4899

Stereoselective Aldol Additions
4887 4899
[25] A. G. Myers, B. Zhong, M. Movassaghi, D. W. Kung, B. A. Lanman, S.
Kwon, Tetrahedron Lett. 2000, 41, 1359 1362.
[49] M. Kroemer, G. E. Schulz, Acta Crystallogr. D Biol. Crystallogr. 2002,
58, 824 832.
[26] Observation of the emulsions with optical microscopy prior to the
reaction (i.e., immediately before addition of the enzyme) revealed
the following features: at 51 mm aldehyde (30 mm DHAP) the
emulsions formulated with tetradecane and hexadecane were highly
concentrated water-in-oil-type emulsions while the one formulated
with squalane was a diluted oil-in-water emulsion; at 340 mm aldehyde
(120 mm DHAP) all emulsions were diluted oil-in-water emulsions.
During the progress of the enzymatic reaction, the highly concen-
trated water-in-oil emulsions became more fluid, changing progres-
sively to diluted oil-in-water emulsions near the end of the reaction.
From these preliminary observations it appears that the type of
emulsion did not influence the reaction performance.
[50] H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H.
Weissig, I. N. Shindyalov, P. E. Bourne, Nucleic Acids Res. 2000, 28,
235 242.
[51] H. Harada, O. Asano, T. Kawata, T. Inoue, T. Horizoe, N. Yasuda, K.
Nagata, M. Murakami, J. Nagaoka, S. Kobayashi, Bioorg. Med. Chem.
2001, 9, 2709 2726.
[52] A. J. Geall, I. S. Blagbrough, Tetrahedron 2000, 56, 2449 2460.
[53] S. Fernandez, R. Brieva, F. Rebolledo, V. Gotor, J. Chem. Soc. Perkin
Trans. 1, 1992, 2885 2889.
[54] A. W. Konradi, S. J. Kemp, S. F. Pedersen, J. Am. Chem. Soc. 1994,
116, 1316 1323.
[55] N. Bischofberger, H. Waldmann, T. Saito, E. S. Simon, W. Lees, M. D.
Bednarski, G. M. Whitesides, J. Org. Chem. 1988, 53, 3457 3465.
[56] M. Frigerio, M. Santagostino, S. Sputore, G. Palmisano, J. Org. Chem.
1995, 60, 7272 7276.
[27] M. D. Bednarski, E. S. Simon, N. Bischofberger, W. D. Fessner, M. J.
Kim, W. Lees, T. Saito, H. Waldmann, G. M. Whitesides, J. Am. Chem.
Soc. 1989, 111, 627 635.
[28] W.-D. Fessner, C. Walter, Top. Curr. Chem. 1996, 184, 97 194.
[29] The stability of DHAP in emulsions was measured without (i.e., blank
experiment) and in the presence of either RAMA (35 UmLÀ1) or
aminoaldehyde 5 (50 mm) at pH 6.9. In the blank experiment, a 15%
degradation of DHAP was observed after 24 h. In the presence of
either RAMA or aldehyde, 30% degradation was detected after 24 h.
[30] W. J. Lees, G. M. Whitesides, J. Org. Chem. 1993, 58, 1887 1894.
[31] W. D. Fessner, G. Sinerius, A. Schneider, M. Dreyer, G. E. Schulz, J.
Badia, J. Aguilar, Angew. Chem. 1991, 103, 596 599; Angew. Chem.
Int. Ed. 1991, 30, 555 558.
[32] W. D. Fessner, J. Badia, O. Eyrisch, A. Schneider, G. Sinerius,
Tetrahedron Lett. 1992, 33, 5231 5234.
[33] K. H. Choi, J. Shi, C. E. Hopkins, D. R. Tolan, K. N. Allen, Biochem-
istry 2001, 40, 13868 13875.
[34] W.-D. Fessner, A. Schneider, H. Held, G. Sinerius, C. Walter, M.
Hixon, J. V. Schloss, Angew. Chem. 1996, 108, 2366 2369; Angew.
Chem. Int. Ed. 1996, 35, 2219 2221.
[35] J. A. Littlechild, H. C. Watson, Trends Biochem. Sci. 1993, 18, 36 39.
[36] A. C. Joerger, C. Gosse, W.-D. Fessner, G. E. Schulz, Biochemistry
2000, 39, 6033 6041.
[57] F. Effenberger, A. Straub, Tetrahedron Lett. 1987, 28, 1641 1644.
[58] G. W. J. Fleet, P. W. Smith, Tetrahedron 1986, 42, 5685 5692.
[59] G. W. J. Fleet, L. E. Fellows, P. W. Smith, Tetrahedron 1987, 43, 979
990.
[60] G. W. J. Fleet, D. R. Witty, Tetrahedron: Asymmetry 1990, 1, 119 136.
[61] A. M. Scofield, L. E. Fellows, R. J. Nash, G. W. J. Fleet, Life Sci. 1986,
39, 645 650.
[62] Y. Banba, C. Abe, H. Nemoto, A. Kato, I. Adachi, H. Takahata,
Tetrahedron: Asymmetry 2001, 12, 817 819.
[63] N. Asano, E. Tomioka, H. Kizu, K. Matsui, Carbohydr. Res. 1994, 253,
235 245.
[64] N. Asano, H. Kizu, K. Oseki, E. Tomioka, K. Matsui, M. Okamoto, M.
Baba, J. Med. Chem. 1995, 38, 2349 2356.
[65] N. Asano, T. Yamashita, K. Yasuda, K. Ikeda, H. Kizu, Y. Kameda, A.
Kato, R. J. Nash, H. S. Lee, K. S. Ryu, J. Agric. Food Chem. 2001, 49,
4208 4213.
[66] A. N. Hulme, C. H. Montgomery, D. K. Henderson, J. Chem. Soc.
Perkin Trans. 1 2000, 1837 1841.
[67] Y. F. Wang, D. P. Dumas, C. H. Wong, Tetrahedron Lett. 1993, 34, 403
406.
[37] N. Matsumori, D. Kaneno, M. Murata, H. Nakamura, K. Tachibana, J.
Org. Chem. 1999, 64, 866 876.
[38] K. K. C. Liu, T. Kajimoto, L. Chen, Z. Zhong, Y. Ichikawa, C. H.
Wong, J. Org. Chem. 1991, 56, 6280 6289.
[39] T. Gefflaut, C. Blonski, J. Perie, M. Willson, Prog. Biophys. Mol. Biol.
1995, 63, 301 340.
[68] A. M. Palmer, V. Jager, Eur. J. Org. Chem. 2001, 2547 2558.
[69] R. J. Molyneux, Y. T. Pan, J. E. Tropea, A. D. Elbein, C. H. Lawyer,
D. J. Hughes, G. W. J. Fleet, J. Nat. Prod. 1993, 56, 1356 1364.
[70] K. Yasuda, H. Kizu, T. Yamashita, Y. Kameda, A. Kato, R. J. Nash,
G. W. J. Fleet, R. J. Molyneux, N. Asano, J. Nat. Prod. 2002, 65, 198
202.
[40] A. Dalby, Z. Dauter, J. A. Littlechild, Protein Sci. 1999, 8, 291 297.
[41] D. R. Hall, G. A. Leonard, C. D. Reed, C. I. Watt, A. Berry, W. N.
Hunter, J. Mol. Biol. 1999, 287, 383 394.
[42] A possible diastereoisomer of 9 or 13 (15%) was detected but its
relative configuration could not be identified unequivocally due to
overlapping signals in the NMR spectra.
[43] F. G. Riddell in Cyclic Organonitrogen Stereodynamics (Eds.: J. B.
Lambert, Y. Takeuchi), VCH, Weinheim, 1992, pp. 159 165.
[44] At low temperatures (285 K), distinctive signals were observed in the
high-field proton NMR spectrum for the corresponding linear 10 and
the two five-membered ring conformers. Unfortunately, it was not
possible to freeze the equilibrium between the two cyclic conformers
and therefore the stereochemistry at C-2 could not be determined
unequivocally. Moreover, the NMR spectra revealed the existence of
only one of the two possible cyclic diastereoisomers, indicating that
attack of the secondary amine on the ketone carbonyl exhibited face
selectivity.
[71] P. Bisseret, C. Bouix-Peter, O. Jacques, S. Henriot, J. Eustache, Org.
Lett. 1999, 1, 1181 1182.
[72] P. A. M. Van der Klein, W. Filemon, H. J. G. Broxterman, G. A.
Van der Marel, J. H. Van Boom, Synth. Commun. 1992, 22, 1763
1771.
[73] J. J. Naleway, C. R. H. Raetz, L. Anderson, Carbohydr. Res. 1988, 179,
199 209.
[74] Y. Huang, D. R. Dalton, P. J. Carroll, J. Org. Chem. 1997, 62, 372 376.
[75] M. Lombardo, S. Fabbroni, C. Trombini, J. Org. Chem. 2001, 66, 1264
1268.
[76] G. W. J. Fleet, A. Karpas, R. A. Dwek, L. E. Fellows, A. S. Tyms, S.
Petursson, S. K. Namgoong, N. G. Ramsden, P. W. Smith, J. C. Son,
FEBS Lett. 1988, 237, 128 132.
[77] G. W. J. Fleet, S. J. Nicholas, P. W. Smith, S. V. Evans, L. E. Fellows,
R. J. Nash, Tetrahedron Lett. 1985, 26, 3127 3130.
[78] D. Wyn, C. Jones, R. J. Nash, E. A. Bell, J. M. Williams, Tetrahedron
Lett. 1985, 26, 3125 3126.
[45] Hydrogenation at atmospheric pressure for 2 h afforded the corre-
sponding cyclic imine sugars.
[46] T. Kajimoto, L. Chen, K. K. C. Liu, C. H. Wong, J. Am. Chem. Soc.
1991, 113, 6678 6680.
[47] N. Asano, K. Yasuda, H. Kizu, A. Kato, J.-Q. Fan, R. J. Nash, G. W. J.
Fleet, R. J. Molyneux, Eur. J. Biochem. 2001, 268, 35 41.
[48] S.-H. Jung, J.-H. Jeong, P. Miller, C.-H. Wong, J. Org. Chem. 1994, 59,
7182 7184.
[79] N. Ikota, A. Hanaki, Chem. Pharm. Bull. 1987, 35, 2140 2143.
[80] S. Takano, M. Moriya, K. Ogasawara, Tetrahedron: Asymmetry 1992,
3, 681 684.
[81] T. Sifferlen, A. Defoin, J. Streith, D. Le Nouen, C. Tarnus, I. Dosbaa,
M.-J. Foglietti, Tetrahedron 2000, 56, 971 978.
[82] A. Defoin, T. Sifferlen, J. Streith, I. Dosbaa, M.-J. Foglietti,
Tetrahedron: Asymmetry 1997, 8, 363 366.
Received: March 19, 2003 [F4966]
Chem. Eur. J. 2003, 9, 4887 4899
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4899