Aldol additions of dihydroxyacetone phosphate to N-Cbz-amino aldehydes catalyzed by L-fuculose-1-phosphate aldolase in emulsion systems: Inversion of stereoselectivity as a function of the acceptor aldehyde

Article abstract of DOI:10.1002/chem.200400648

The potential of L-fuculose-1-phosphate aldolase (FucA) as a catalyst for the asymmetric aldol addition of dihydroxyacetone phosphate (DHAP) to N-protected amino aldehydes has been investigated. First, the reaction was studied in both emulsion systems and

Full text of DOI:10.1002/chem.200400648

Aldol Additions of Dihydroxyacetone Phosphate to N-Cbz-Amino
Aldehydes Catalyzed by l-Fuculose-1-Phosphate Aldolase in Emulsion
Systems: Inversion of Stereoselectivity as a Function of the Acceptor
Aldehyde
Laia Espelt,[a] Jordi Bujons,[a] Teodor Parella,[b] Jordi Calveras,[a] Jesffls Joglar,[a]
Antonio Delgado,[c, d] and Pere ClapØs*[a]
Abstract: The potential of l-fuculose-
1-phosphate aldolase (FucA) as a cata-
lyst for the asymmetric aldol addition
from (S)-N-Cbz-alaninal, which gave a
55:45 mixture of both epimers. From
the stereochemical analysis of the re-
sulting iminocyclitols, it was concluded
that the stereoselectivity of the FucA-
catalyzed reaction depended upon the
structure of the N-Cbz-amino aldehyde
acceptor. Whereas the enzymatic aldol
reaction with both enantiomers of N-
Cbz-alaninal exclusively gave the ex-
pected 3R,4R configuration, the stereo-
chemistry at the C-4 position of the
major aldol adducts produced in the re-
actions with N-Cbz-glycinal and N-
Cbz-3-aminopropanal was inverted to
the 3R,4S configuration. The study of
the FucA-catalyzed addition of DHAP
to phenylacetaldehyde and benzyloxya-
cetaldehyde revealed that the 4R prod-
uct was kinetically favored, but rapidly
disappeared in favor of the 4S diaster-
eoisomer. Computational models were
generated for the situations before and
of
dihydroxyacetone
phosphate
(DHAP) to N-protected amino alde-
hydes has been investigated. First, the
reaction was studied in both emulsion
systems and conventional dimethylform-
amide (DMF)/H2O (1:4 v/v) mixtures.
At 100mm DHAP, compared with the
reactions in the DMF/H2O (1:4) mix-
ture, the use of emulsion systems led to
two- to three-fold improvements in the
conversions of the FucA-catalyzed re-
actions. The N-protected aminopolyols
thus obtained were converted to imino-
cyclitols by reductive amination with
Pd/C. This reaction was highly diaster-
eoselective with the exception of the
reaction of the aldol adduct formed
À
after C C bond formation in the active
site of FucA. Moreover, the lowest-
energy conformations of each pair of
the resulting epimeric adducts were de-
termined. The data show that the prod-
ucts with a 3R,4S configuration were
thermodynamically more stable and,
therefore, the major products formed,
in agreement with the experimental re-
sults.
Keywords:
DHAP-dependent
biotransformations
aldolases
·
·
enzyme catalysis · iminocyclitols ·
lyases
Introduction
[a] Dr. L. Espelt, Dr. J. Bujons, J. Calveras, Dr. J. Joglar, Dr. P. Clapꢀs
Institute for Chemical & Environmental Research-CSIC
Jordi Girona 18–26, 08034-Barcelona (Spain)
Fax : (+34)93-204-5904
À
Stereoselective C C bond formation catalyzed by aldolases
has recently attracted tremendous interest as a powerful
tool in asymmetric synthesis.[1,2] Aldolases catalyse aldol ad-
ditions of aldehydes and ketones with fine control of the ab-
solute configuration of the newly formed stereogenic cen-
ters. In particular, dihydroxyacetone phosphate (DHAP) de-
pendent aldolases[3,4] catalyse the aldol addition of DHAP
to a variety of non-natural aldehyde acceptors to generate
two stereogenic centers and add three carbon atoms to the
carbon backbone.[5] Aldolases react highly stereoselectively
at the C-3 position of DHAP (i.e., the stereocenter arising
from DHAP addition), whereas the stereoselectivity at the
C-4 position (i.e., the one generated from the addition to
E-mail: pcsqbp@iiqab.csic.es
[b] Dr. T. Parella
Servei de Ressonꢁncia Magnꢂtica Nuclear Universitat Autꢃnoma de
Barcelona (Spain)
[c] Dr. A. Delgado
Unit of Medicinal Chemistry, Associated Unit to CSIC Faculty of
Pharmacy, University of Barcelona (Spain)
[d] Dr. A. Delgado
RUBAM Institute for Chemical & Environmental Research-CSIC
(Spain)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
1392
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/chem.200400648
Chem. Eur. J. 2005, 11, 1392 – 1401

FULL PAPER
the aldehyde) depends on the structure and stereochemistry
of the acceptor aldehyde (1–4, see Scheme 1).[3,6,7]
The stereoselective synthesis of aminopolyols and imino-
cyclitols using DHAP aldolases is currently a topic of inter-
est in our laboratory. The key step in our strategy involves
the aldol addition of DHAP to N-benzyloxycarbonyl (Cbz)
amino aldehydes.[7] We recently reported the use of d-fruc-
tose-1,6-diphosphate aldolase from rabbit muscle (RAMA)
and l-rhamnulose-1-phosphate aldolase from E. coli
(RhuA) as catalysts for this reaction. As a continuation of
our research project, herein we report on the reactivity and
stereoselectivity of recombinant l-fuculose-1-phosphate al-
dolase from E. coli (FucA)[8] in the aldol reaction of N-Cbz-
amino aldehydes[7] 1–4 with DHAP (Scheme 1). In this
study we focused on three aspects. Firstly, to assess whether
the advantages of emulsion systems, already found to be
useful in RAMA- and RhuA-catalyzed aldol reactions, ap-
plied to the new enzyme, the influence of the reaction
medium on the conversion was systematically investigated.
Secondly, to determine the stereoselectivity of FucA to-
wards the acceptor aldehydes,
brium values. As shown in Table 1, the aldehydes 1–4 are
suitable acceptors for FucA aldolase, the degree of conver-
sion to aldol adducts depending mainly on the reaction
medium used. At 100mm DHAP, better results were ob-
tained with the emulsion systems than with the DMF/water
mixtures (Table 1, entries 1, 3, 4, and 6), although under
these conditions, aldehydes 1 and 3 only gave moderate con-
versions in the emulsion systems (Table 1, entries 1 and 4).
Interestingly, lowering the DHAP concentration (45mm)
greatly improved the reaction conversions in DMF/H2O
(1:4), as well as for aldehydes 1 and 3 regardless of the reac-
tion medium used (Table 1, entries 2, 5, and 7). Note that at
high substrate concentrations, low conversions were also ob-
served with fructose-1,6-diphosphate aldolase from rabbit
muscle (RAMA) in DMF/H2O (1:4) systems.[7] This behav-
ior appeared to be related to both the limited solubility of
the acceptor aldehyde and the low stability of the enzyme in
the reaction medium. Moreover, the improved conversion at
the lower DHAP concentration also suggests that the aldo-
lase is inhibited by the N-protected amino aldehyde[6,10] or
the structure and stereochemis-
try of the resulting cyclic prod-
ucts, namely iminocyclitols,
were analyzed. Thirdly, to un-
derstand the stereochemical
outcome of the FucA-catalyzed
aldol addition reactions, compu-
tational modeling of the alde-
hyde–DHAP–FucA complexes
was carried out as well as con-
formational searches for each
pair of possible epimeric prod-
ucts at C-4.
Results and Discussion
Scheme 1. Chemo-enzymatic synthesis of linear N-Cbz-aminopolyols 5–8 (disodium salts) and iminocyclitols
9–15 from N-Cbz-amino aldehydes 1–4. For the stereochemistry at C-4 of compounds 5–8, see text. For the
Influence of the reaction
medium:
l-Fuculose-1-phos- stereochemistry at C-2, C-4, and C-5 of compounds 9–15, see Table 3.
phate aldolase-catalyzed aldol
additions of DHAP to alde-
hydes 1–4 (Scheme 1) were per-
formed in both emulsion and
DMF/water (1:4) cosolvent sys-
tems.[7,9] Two DHAP concentra-
tions, 45 and 100mm with
1.8 equivmolÀ1 of N-Cbz amino
aldehyde, were examined. The
data, given as a molar percent-
age of conversion, were the
maximum values obtained and
remained constant up to 12–
24 h, even after the addition of
more enzyme. Therefore it may
be assumed that these conver-
Table 1. FucA-catalyzed aldol reaction between DHAP and N-Cbz-amino aldehydes.
Entry
Acceptor
aldehyde
FucA [UmLÀ1
]
DHAP [mm][a]
Conversion [%][b] (Time [h])
Product
Reaction media[c]
A
B
C
D
1
2
3
4
5
6
7
1
1
2
3
3
4
4
8
6
8
8
6
8
6
100
45
100
100
45
35 (8)
52 (6)
53 (8)
30 (7)
57 (2)
41 (8)
58 (7)
34 (6)
50 (6)
66 (5)
32 (7)
57 (5)
56 (4)
56 (7)
34 (5)
51 (6)
49 (5)
35 (7)
57 (3)
58 (3)
63 (7)
16 (4)
42 (6)
15 (4)
16 (1)
58 (5)
17 (4)
50 (4)
5
5
6
7
7
8
8
100
45
[a] Molar percent conversion to the aldol adduct (5–8) with respect to the starting DHAP concentration, deter-
mined by HPLC from the crude reaction mixture using purified standards. [b] Reaction conditions: A: H2O/
C14E4/tetradecane 90:4:6 wt%; B: H2O/C14E4/hexadecane 90:4:6 wt%; C: H2O/C14E4/squalane 90:4:6 wt%,
where C14E4 is tetra(ethylene glycol) tetradecyl ether, C14H29(OCH2CH2)4OH, with an average of 4 moles of
ethylene oxide per surfactant molecule; D: DMF/H2O (1:4 v/v). [c] 1.8 equiv molÀ1 of acceptor aldehyde; reac-
sions were close to the equili- tion volume 5 mL, T=258C.
Chem. Eur. J. 2005, 11, 1392 – 1401
www.chemeurj.org
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1393

P. Clapꢀs et al.
Table 2. Structures of the iminocyclitols derived from linear compounds 5–8.
by the methylglyoxal which re-
sults from DHAP decomposi-
tion.[3]
Like
l-rhamnulose-1-phos-
phate aldolase,[7] FucA showed
no clear preference for either
of the two enantiomers of N-
Cbz-alaninal (Table 1, entries 5
and 7), although at 100mm
DHAP the conversion to the R
enantiomer was higher than the
conversion to the S enantiomer
(Table 1, entries 4 and 6). In
contrast, high kinetic enantio-
discrimination by FucA has
been reported for a series of
racemic 2-hydroxyaldehyde ac-
ceptors.[11] These differences
may be explained by the fact
that 2-hydroxyaldehydes have a
higher affinity and better orien-
tation in the biocatalyst active
site than the aldehydes 3 and 4
Entry
1
Linear product/
Cyclic products
d.r.
Cyclic product
[a]2D0 (c=1 in MeOH)
or mixture [a]2D0
À6.4
5/À8.9
76:24
(c=0.9 in MeOH)
À21.2
2
3
4
6/À1.2
7/+1.0
8/+4.5
80:20
55:45
100:0
(c=0.8 in MeOH)
À7.1
(c=1.2 in MeOH)
+37.9
(c=1.4 in MeOH)
[a] See ref. [7]. [b] See refs. [20,21]. [c] Not previously described. [d] See refs. [22,23].
which have
methyl group.
a
hydrophobic
Stereochemical characterization: Compounds 5–8 were syn-
thesized (ca. 50–160 mg) and transformed into iminocyclitols
by a previously described procedure (Scheme 1).[7,12] The
relative stereochemistries of the newly formed stereogenic
centers of the iminocyclitols were unequivocally ascertained
by one- and two-dimensional NMR techniques.[13] This al-
lowed us to quantify the diastereoisomers thus formed and
to elucidate the stereoselectivity of FucA in the aldol addi-
tion of DHAP to aldehydes 1–4. For the DHAP-aldolase
catalysis, it is accepted that the absolute configuration at the
C-3 position (i.e. R for FucA) (Scheme 1) is conserved upon
reaction with electrophiles.[14–19] Thus, the identified cyclic
products and the absolute configuration of their stereogenic
centers could be assessed (Table 2).
Similar to the NMR study on the iminocyclitols obtained
with RAMA and RhuA aldolases,[7] and as a complement to
NOE data, the shielding effects were also used as a probe to
assign the relative stereochemistry of 9–15 (Table 3). The
OH(3) and CH2OH(2) groups induced a 0.08–0.26 ppm up-
field shift on the H-2 and H-3 protons of compound 13, with
respect to 14 and 15, whereas a downfield shift was observed
for the C-2, C-3, and C-4 carbon centers of 13 (Table 3).
Moreover, when the OH(4) and Me(5) groups adopted a cis
configuration (e.g. compounds 13 and 14), a strong upfield
13C chemical shift of 3.2–3.5 ppm was induced on the Me(5)
group. Similarly, a cis orientation of the CH2OH(2) and
OH(4) moieties caused an upfield 13C chemical shift on the
C-4 carbon atom.
delivered from the face opposite the hydroxy group at the
C-4 position.[7] This was not found in the reductive amina-
tion of compound 7; here, there was no face selectivity and
about 50% epimerization at C-2 was obtained.
Analysis of the stereochemistry at the C-4 position of the
1
iminocyclitols 9–15 (Table 2) by high-field H NMR spectro-
scopy allowed us to deduce, within the limits of detection,
the stereoselectivity of FucA towards each of the N-Cbz-
protected amino aldehydes. As Table 2 entries 1 and 2
shows, the major cyclic diastereoisomer obtained from alde-
hydes 1 and 2 (i.e. iminocyclitols 9 and 11, respectively) has
an S configuration at the C-4 position. This stereochemistry
arose from attack of the enzyme–DHAP-enediolate com-
plex on the re face of the carbonyl group of the aldehyde.
Interestingly, this diastereofacial selectivity was the inverse
of that found for the natural acceptor, l-lactaldehyde, and
other non-natural aldehydes.[24–26] In these cases, the
DHAP–FucA complex attacks the carbonyl component at
the si face, although the formation of diastereoisomers (3–
30%), epimeric at the C-4 position, has been reported.[14]
On the other hand, FucA was highly stereoselective towards
both enantiomers of Cbz-alaninal with approximately 99%
de of the expected 4R diastereoisomer (iminocyclitols 13–
15) being formed.
We then asked if the stereochemical outcome of FucA-
catalyzed aldol addition reactions was kinetically or thermo-
dynamically controlled. Stereochemical analysis of the ad-
ducts formed at the beginning of the reaction and after long
incubation times may shed light on this point. RP-HPLC
was experimentally the most convenient method for this
purpose, however, the diastereoisomers of the aldol prod-
Inspection of the stereochemistry at the C-2 position re-
vealed that the reductive amination of compounds 5, 6, and
8 with Pd/C was stereoselective, the hydrogen atom being
1394
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1392 – 1401

Aldol Additions of Dihydroxyacetone Phosphate to Amino Aldehydes
FULL PAPER
Table 3. 1H and 13C chemical shifts [ppm] of the iminocyclitols 13, 14, and 15.
Docking simulations of alde-
hydes 16 and 17 in the active
center of FucA, with the alde-
hydes approaching the DHAP-
enediolate from both possible
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’ orientations,
furnished
the
model structures shown in
Figure 2 (panels A and B, re-
spectively). From these models
it appears that both aldehydes
can adopt similar conforma-
tions in the enzyme pocket, ir-
13
14
15
3.51/62.1
3.69/58.4
3.75/61.5
4.19/71.7
4.45/70.4
4.27/70.4
4.22/78.3
4.10/71.5
3.95/76.7
3.65/57.2
3.61/56.9
3.54/56.7
1.30/11.8
1.32/11.5
1.38/15.0
3.87/58.7
3.79/58.6
3.92/58.4
3.76/58.7
3.73/58.6
3.80/58.4
ucts formed from the reactions of the amino aldehydes 1–4
could not be resolved by this technique. Interestingly, the in-
version of stereoselectivity induced by FucA in these reac-
tions was also observed in the aldol addition reactions of
respective of their orientation of attack (si or re), which are
stabilized by a hydrogen bond between their carbonyl
oxygen atom and the phenol group of Tyr113’. A model for
the binding of l-lactaldehyde to DHAP-complexed FucA
has been proposed in which the
high stereoselectivity of the
enzyme was attributed mainly
to analogous interactions be-
tween the carbonyl and hy-
droxy groups of the substrate
and the phenol group of
Scheme 2. FucA-catalyzed aldol addition of DHAP to phenylacetaldehyde (16) and to benzyloxyacetaldehyde
(17).
Tyr113’.[26] At variance with this
l-lactaldehyde binding model,
the absence of an a-hydroxy
group in aldehydes 16 and 17
both phenylacetaldehyde (16) and benzyloxyacetaldehyde
(17) (Scheme 2). In these cases, the diastereoisomers ob-
tained could be resolved by RP-HPLC and quantified.[27]
Hence it was thought that these reactions may be helpful in
the analysis of the stereochemical course of the reaction and
most of the studies that follow were carried out with these
substrates. The progress of the aldol addition reaction, as
depicted in Figure 1, revealed that both (4R)-18 and (4R)-19
were kinetically favored while their epimers, (4S)-18 and
(4S)-19, respectively, were the major products obtained and,
presumably, thermodynamically more stable (approximate
ratio 70:30). Note that the rate of formation of the 4S prod-
ucts was quite fast, being the major adducts after 50–60 min.
favors an orientation in which their aromatic substituents
are directed towards the hydrophobic wall formed by the
side chains of Phe131 and Phe206’. Similar interactions were
observed in the corresponding models for adducts 18 and 19
(see Supporting Information). However, the conformational
differences due to the two possible orientations of attack (si
or re) of the aldehydes or between the two corresponding
epimeric adducts are rather small. Therefore, it seems plau-
sible that the differences between the geometries and, by ex-
tension, the energies of the corresponding transition states
should be small. This could explain the relatively small ki-
netic preference shown by the enzyme in the reactions with
aldehydes 16 and 17.
As stated above, the mobile C-terminal tail of the protein
was initially omitted from these simulations because of the
lack of structural data. Joerger et al.[26] proposed a model for
the induced fit of this tail suggesting that, upon substrate
binding, it undergoes a conformational shift from a less-or-
dered state to one in which it covers the entrance to the
active center of FucA. This allowed some of its residues (i.e.
Tyr209’) to come into contact with the substrates, thus ren-
dering a more packed structure around the active site.
Therefore, we performed further docking simulations that
included the coordinates of the modeled C-terminal tail of
the protein.[30] The results obtained for aldehydes 16 and 17
are shown in panels C and D of Figure 2, while those for the
corresponding adducts 18 and 19 can be found in the Sup-
porting Information.
Computational models: To gain an insight into the mecha-
nisms responsible for the above observations, computational
models pertinent to both enzyme–substrate and enzyme–
product complexes were generated and conformational anal-
ysis of the aldol products was performed.
Computational models were generated for the reaction in-
À
termediates before and after C C bond formation between
DHAP and aldehydes 16 and 17 in the active center of
FucA (I and II, respectively, in Scheme 3), as previously re-
ported for other aldolases.[7,26] The starting point for these
models was the reported crystal structure of wild-type FucA
with the inhibitor phosphoglycolohydroxamate (PGH)
bound to its active site.[28] This structure lacks the coordi-
nates of the mobile C-terminal tail of the protein (residues
207–215), which has been proposed to undergo an induced
fit upon substrate binding (vide infra).[26]
As could be anticipated, the aldehydes that are larger
than the natural substrate, l-lactaldehyde, could sterically
Chem. Eur. J. 2005, 11, 1392 – 1401
www.chemeurj.org
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1395

P. Clapꢀs et al.
interfere with the proposed conformation of the C-terminal
fragment of the protein. Consequently, the resulting mini-
mized conformations of 16 and 17 are directed away from
the Phe131 and Phe206’ residues and towards the inner part
of the enzyme pocket. Again, only small differences could
be observed between the conformations derived from the si
or re approach of the carbonyl group to the enediolate, and
also between the minimized conformations obtained for the
epimers of products 18 and 19. These models provided no
evidence for stabilizing interactions between the substrates
and Tyr209’ but, since the simulations were carried out with
a fixed protein structure, it cannot be ruled out that they
would be established if the structure of the protein was al-
lowed to relax. However, it seems reasonable to think, as
mentioned earlier, that the size of these substrates could
pose a significant steric hindrance that would hamper, or
even prevent, the induced fit of the C-terminal FucA tail.
This would lead to a reduction in the activity of the enzyme
and a modified stereoselectivity towards the aldehyde, both
of which were observed with the substrates used in this
study. Hence, the reduced, but not zero, activities reported
for different C-terminus FucA mutants suggest that this tail
is not essential for enzyme activity.[26] In addition, these mu-
tants gave higher proportions of the l-threo adducts derived
from the “wrong” re approach of the aldehyde molecule to
the DHAP-enediolate.
The time course of the reaction of aldehydes 16 and 17
shown in Figure 1 indicated a thermodynamic control of the
reaction products. To substantiate this assumption, we per-
formed an extensive exploration of the conformational
space of each of the C-4 epimers of adducts 18 and 19, and
this was extended to 5–8. We considered only the linear
forms of these products since our NMR data indicate that
the relative abundance of the corresponding cyclic forms
always accounted for between 5 and 50% of the total ad-
ducts present in the crude reaction products. To this end, we
first ran a systematic conformational search on every mole-
cule (see Experimental Section). The energies were calculat-
ed by using the Merck force field (MMFF94)[31] and the
Born continuum solvation model,[32–34] which have been suc-
cessfully used in the modeling of polar compounds[35] and
carbohydrates.[36,37]
Figure 1. Time–course of the FucA-catalyzed aldol reaction between
DHAP and a) phenylacetaldehyde 16 and b) benzyloxyacetaldehyde 17.
Relative molar percentages of (4R)-18, (4S)-18, (4R)-19 and (4S)-19 for-
mation and molar percentage conversion with respect to the starting
DHAP concentration. Reactions were carried out in H2O/C14E4/hexade-
cane (90:4:6 wt%) emulsion systems: [DHAP]=30 mm, 1.8 equivmolÀ1
of 16 or 17, and 1–2.5 UmLÀ1 FucA; reaction volume=2.5 mL and T=
258C.
Table 4 summarizes the cal-
culated energetic differences
between the lowest energy con-
formations found for each pair
of epimers, while the corre-
sponding geometries and actual
energy values are reported in
the Supporting Information.
Table 4 also shows both the ex-
perimental and predicted ratios
of the 4R and 4S adducts; the
latter were calculated from the
energy differences mentioned
above, assuming that the en-
tropic contributions to DG
Scheme 3. Proposed enzymatic mechanism for the FucA-catalyzed aldol addition reactions.[26,29] The stereo-
chemistry at the C-4 position of the adducts is determined by the face (si or re) of the aldehyde that ap-
proaches the reactive C-3 atom of the DHAP-enediolate.
1396
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1392 – 1401

Aldol Additions of Dihydroxyacetone Phosphate to Amino Aldehydes
FULL PAPER
The experimental and com-
putational results collected in
this work for the formation of
aldol adducts 18 and 19 suggest
the reaction coordinate scheme
depicted in Figure 3 is opera-
tive. This is similar to the one
proposed for N-acetylneuramin-
ic acid aldolase.[38] Hence, alde-
hydes 16 and 17 could approach
the FucA-bound DHAP-ene-
diolate by either of the two pos-
sible reactive orientations, with
the si approach energetically
preferred over the re approach.
However, the energetically pre-
ferred reaction products are
those derived from the attack
of the enediolate on the re face
of the carbonyl group and
therefore these are the major
products of the reaction. This
hypothesis may be extended to
the rest of the N-Cbz-amino al-
dehydes considered in this
study since they have analogous
steric hindrance. To verify this
hypothesis additional quantum
mechanics/molecular mechanics
(QM/MM) modeling studies are
being carried out in order to ac-
curately calculate the energy
Figure 2. Structural models of aldehydes 16 (panels A and C) and 17 (panels B and D) in the active center of
FucA, approaching the Zn2+-bound DHAP-enediolate from their si (green) and re (orange) faces. Panels A
and B: Models generated from the reported crystallographic coordinates of FucA.[28] Panels C and D: Models
generated from a modeled FucA structure that includes coordinates of the proposed induced fit of the mobile barriers of these reaction coor-
C-terminal tail.[26]
dinates.
Table 4. Energy differences (DE4RÀ4S) between the global minima deter-
mined for the 4R and 4S epimers of the aldol adducts 5–8, 18, and 19,
Conclusions
and the predicted and experimental compositions of the reaction mix-
tures.
l-Fuculose-1-phosphate aldolase catalyzed the aldol addi-
Product
DE4RÀ4S
Ratio 4R:4S
(predicted)
Ratio 4R:4S
(experimental)
tion of DHAP to several N-Cbz-amino aldehydes in conver-
sion yields that range from 50 to 60%. At 100mm DHAP
the emulsion systems gave the highest substrate conversion
to aldol adduct compared with those achieved in DMF/H2O
(1:4) mixtures. Together with earlier findings,[7] the high-
water-content emulsions appear to be of general applicabili-
ty, yet easy to prepare, for DHAP-aldolase-catalyzed aldol
addition reactions, especially when dealing with low-water-
soluble aldehyde acceptors.
[kcalmolÀ1
]
5
6
7
8
18
19
1.1
0.4
À0.6
À1.0
0.4
13:87
33:67
73:27
84:16
33:67
18:82
24:76
20:80
~100:0
~100:0
30:70
0.9
30:70
We report here an inversion of stereoselectivity in FucA-
catalyzed aldol addition of DHAP to N-Cbz-glycinal, N-
Cbz-3-aminopropanal, phenylacetaldehyde, and benzyloxya-
cetaldehyde. Time–progress curves for the reactions with
phenylacetaldehyde and benzyloxyacetaldehyde revealed
that the products with the “natural” 4R configuration were
kinetically favored, whereas the inverted 4S ones were the
major products formed, which suggests that the reactions
have a thermodynamic outcome. Computational models of
should be similar. It is clear that there is a good correlation
between the predicted and experimental values for products
5–8, 18, and 19. Thus, the global minima determined for the
4S epimers of adducts 5, 6, 18, and 19 have lower energies
than the corresponding minima determined for the 4R epi-
mers and, accordingly, these are the major reaction products,
while the opposite is true for the diastereoisomers of ad-
ducts 7 and 8.
Chem. Eur. J. 2005, 11, 1392 – 1401
www.chemeurj.org
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1397

P. Clapꢀs et al.
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.
Methods:Molecular modeling: All molecular simulations were conducted
with the MOE program (v. 2003.02, Chemical Computing Group, Mon-
treal) using the MMFF94 force field with its standard atomic charges and
parameters.[31] All energy calculations were carried out by using the Born
continuum solvation model[32–34] as implemented in the MOE program.
For the FucA complexes, a smoothed cut-off between 14 and 15 ꢇ was
used to model the nonbonded interactions. All minimizations were per-
formed up to an RMS gradient of <0.01.
The coordinates for E. coli fuculose-1-phosphate aldolase complexed to
phosphoglycolohydroxamate (PGH)[28] were obtained from the Protein
Data Bank[40] at Brookhaven National Laboratory (entry 4FUA). The
structure includes the coordinates of one FucA monomer with the essen-
tial Zn2+ ion complexed to PGH, one molecule of b-mercaptoethanol co-
valently bound to Cys14, one sulfate anion, and several water molecules.
Nine C-terminal residues (residues 207–215) were missing from the crys-
tal structure, and therefore they were not initially included in the original
simulations although they were considered at a later stage (see text). By
applying the necessary crystallographic symmetry operators to this struc-
ture, the homotetramer that constitutes the biological unit was built.
Since the active center of FucA is located at the interface between each
pair of contiguous monomers, each homotetramer contains four catalytic
centers. However, in order to reduce the calculation time, the simulations
were performed on a FucA dimer that contained just one catalytic site.
In addition, the sulfate and the solvent molecules were removed, the hy-
drogen atoms were added, and the PGH molecule was conveniently
modified to obtain the a-hydroxyketonic adducts 18 and 19 complexed to
the Zn2+ cation in the catalytic active center (state II of Scheme 3). The
conformational space of these ligands was explored to find the low-
energy minima by running a stochastic conformational search keeping
the coordinates of the protein and the essential Zn2+ fixed.
Figure 3. Proposed reaction coordinate for the FucA-catalyzed aldol ad-
dition to the aldehydes considered in this study.
The structure of the ligands was further modified to generate the situa-
À
tion before the C C bond formation, as represented by state I in
Scheme 3, in which the aldehyde molecule is close to the Zn2+-coordinat-
ed DHAP-enediolate. To avoid the exclusion of this aldehyde molecule,
a restraint was imposed between the two carbon atoms that participate in
substrates and products complexed to the active-site cavity
of the enzyme and theoretical calculations of the energy dif-
ferences between the lowest-energy conformers of each pair
of aldol epimers at C-4 substantiate the above experimental
results.
À
the C C bond formation (i.e., the C-3 atom of the enediolate and the C-
1 atom of the aldehyde). This restraint was arbitrarily set so as to keep
the distance between these two carbon atoms close to 2.5 ꢇ to ensure
that both substrates would adopt conformations that resemble those of
the putative transition states. The conformational space of the aldehydes
was then stochastically searched as before, keeping the coordinates of the
rest of the atoms fixed.
The conformational spaces of compounds 5–8, 18, and 19 were exhaus-
tively searched by using the two different strategies available in the
MOE program in order to find the global energy minima. First, starting
from an initially optimized structure, a systematic conformational search
was run in which every non-amide nonterminal single bond was rotated
in 60–1208 steps. The conformations generated for each compound, which
in some cases amounted to several hundred thousands, were then mini-
mized and ranked according to their energy. To confirm the nature of the
lowest energy conformer determined, a subsequent stochastic conforma-
tional search was run using a procedure similar to that reported by Au-
zanneau et al.[36] Briefly, the conformational space of the molecules was
explored by random rotation of bonds and simultaneous Cartesian per-
turbation. The conformations thus generated were minimized and
checked to determine, within an RMS tolerance (0.1 ꢇ), whether they
were duplicates of previously generated conformations. The process was
terminated when the number of failures to find new conformations ex-
ceeded a large enough number (1000) of consecutive attempts.
Experimental Section
Materials: l-Fuculose-1-phosphate aldolase was produced at the Depar-
tament d’Enginyeria Quꢅmica, Universitat Autꢃnoma de Barcelona, from
recombinant E. coli (ATCC no 86984) and purified by affinity chroma-
tography. Acid phosphatase (PA, EC 3.1.3.2, 5.3 UmgÀ1) was from Sigma
(St. Louis, USA). The non-ionic poly(oxyethylene ether) surfactant, tet-
ra(ethylene glycol) tetradecyl ether, C14H29(OCH2CH2)4OH (C14E4), with
an average of 4 mol of ethylene oxide per surfactant molecule (C14E4),
was from Albright & Wilson (Barcelona, Spain). Tetradecane (99%),
hexadecane (99%), and 2,6,10,15,19,23-hexamethyltetracosane (squa-
lane) (99%) were from Sigma. DOWEX (H+) 50ꢆ8 ion-exchange resin
was from Fluka. HPLC isocratic grade acetonitrile was from Merck
(Darmstadt, Germany) and Multisolvent acetonitrile for preparative RP-
HPLC was from Scharlau (Barcelona, Spain). MacroPrep High Q Sup-
port anion-exchange resin was from BioRad (Hercules, USA). Triethyla-
mine (Calbiochem, San Diego, USA) was of buffer grade. Phenylacetal-
dehyde and benzyloxyacetaldehyde were from Aldrich (Milwaukee,
USA). N-Benzyloxycarbonyl amino aldehydes 1–4 were synthesized in
our laboratory by using previously described procedures.[7] The precursor
of dihydroxyacetone phosphate (DHAP), dihydroxyacetone phosphate
dimer bis(ethyl ketal), was synthesized in our laboratory by using a pro-
cedure similar to that described by Jung et al.[39] De-ionized water was
HPLC analyses: HPLC analyses were performed on a Lichrographꢈ
HPLC system (Merck, Darmstadt, Germany) fitted with a RP-HPLC car-
tridge, 250ꢆ4 mm, filled with Lichrosphereꢈ 100, RP-18, 5 mm (Merck).
Samples (50 mg) were withdrawn from the reaction medium, dissolved in
methanol (0.5–1 mL) to stop any enzymatic reaction, and subsequently
analyzed by HPLC. The solvent systems used were: solvent A: 0.1% v/v
1398
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1392 – 1401

Aldol Additions of Dihydroxyacetone Phosphate to Amino Aldehydes
FULL PAPER
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, UV detection 215 nm. The retention factors (k’) for each
aldol condensation product are given below.
NMR analysis: High-field 1H and 13C NMR analyses of the compounds
under study in D2O solutions were carried out at the Servei de Ressonꢁn-
cia Magnꢂtica Nuclear, Universitat Autꢃnoma de Barcelona, using an
AVANCE 500 BRUKER spectrometer. The compounds were fully char-
acterized by typical gradient-enhanced 2D experiments, COSY, NOESY,
HSQC, and HMBC, under routine conditions. When possible, NOE data
was obtained from selective 1D NOESY experiments by using a single
pulsed-field-gradient echo as a selective excitation method and a mixing
time of 500 ms. When necessary, proton and NOESY experiments were
ture of two diastereoisomers along with cyclic species, were consistent
with those previously reported.[7]
(3R,5S)-5-(Benzyloxycarbonylamino)-5,6-dideoxy-1-O-phosphonohex-2-
ulose disodium salt (7): The title compound (180 mg, 46% yield) was pre-
pared by using conditions B (see Table 1) with 45mm DHAP. HPLC: k’=
4.5; m.p. 113–1158C (99.9% pure by HPLC); elemental analysis calcd
(%) for C14H18NO9Na2P·1= H2O·NaCl: C 36.05, H 4.29, N 3.00; found C
2
36.38, H 4.28, N 2.74.
(3S,5R)-5-(Benzyloxycarbonylamino)-5,6-dideoxy-1-O-phosphonohex-2-
ulose disodium salt (8): The title compound (320 mg, 56% yield) was pre-
pared by using conditions C (see Table 1) with 45mm DHAP. HPLC: k’=
4.5; m.p. 119–1238C (99.9% pure by HPLC): elemental analysis calcd
(%) for C14H18NO9Na2P·5= H2O: C 35.99, H 4.92, N 3.00; found: C 35.90,
2
recorded at different temperatures to study the different behavior of the
exchange phenomena and thereby avoiding the presence of false NOE
cross-peaks that make both structural and dynamic studies difficult. 1H
(300 MHz) and 13C NMR (75 MHz) spectra of compounds in [D6]DMSO
and D2O solutions were recorded with a Varian Unity-300 spectrometer
at the Instituto de Investigaciones Quꢅmicas y Ambientales-CSIC.
H 4.77, N 2.95.
Removal of the phosphate group and catalytic hydrogenation: The phos-
phate group of compounds 9–15 was removed by acid phosphatase cata-
lyzed hydrolysis following the procedure described by Bednarski et al.[42]
The resulting products were hydrogenated with 50 psi H2 in the presence
of Pd/C for 24 h as previously described.[7]
Elemental analyses and specific rotations: Elemental analyses were per-
formed by the Servei de Microanꢁlisi Elemental IIQAB-CSIC. Specific
rotations were measured with a Perkin Elmer Model 341 (ꢉberlingen,
Germany) polarimeter.
(2S,3S,4S)-2-(Hydroxymethyl)piperidine-3,4-diol (9) and (2R,3S,4R)-2-
(hydroxymethyl)piperidine-3,4-diol (10): The title compounds were pre-
pared according to the general procedure described above. The 1H and
13C NMR spectra of the major 9 and minor 10 diols were consistent with
those reported previously.[7]
Enzymatic aldol condensations in emulsions (conditions A, B, and C,
Table 1): The emulsion systems consisted of ternary mixtures of water
(90 wt%), technical grade tetra(ethylene glycol) tetradecyl ether surfac-
tant, C14H29(OCH2CH2)4OH (C14E4) (4 wt%), with an average of 4 mol
of ethylene oxide per surfactant molecule, and either tetradecane (C14),
hexadecane (C16), or squalane (C30) (6 wt%). The reactions were carried
out in 10 mL screw-capped test-tubes. The aldehyde (0.4–0.9 mmol), oil
(6% w/w), and the surfactant (4% w/w) were mixed vigorously. The
DHAP solution (0.225–0.500 mmol) at pH 6.9, freshly prepared as de-
scribed by Effenberger and Straub,[41] was then added dropwise to the
surfactant mixture while stirring at 258C with a vortex mixer. The final
reaction volume was 5 mL. Finally, FucA (6–8 UmLÀ1) was added and
the solution mixed again. The test-tubes were placed in a horizontal
shaking bath (100 rpm) maintained at a 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 mixture purified as described previously.[7]
(2S,3S,4S)-2-(Hydroxymethyl)pyrrolidine-3,4-diol (11) and (2R,3S,4R)-2-
(hydroxymethyl)pyrrolidine-3,4-diol (12): The title compounds were pre-
pared according to the general procedure described above. The 1H and
13C NMR spectra of the major 11 and minor 12 diols were consistent
with those reported previously.[7]
(2S,3S,4R,5S)-2-(Hydroxymethyl)-5-methylpyrrolidine-3,4-diol (13) and
(2R,3S,4R,5S)-2-(hydroxymethyl)-5-methylpyrrolidine-3,4-diol (14): The
title compounds were prepared according to the general procedure de-
scribed above. 1H NMR (500 MHz, D2O, 25 8C) major product, 13: d=
4.22 (dd, 3J(H,H)=3.8, 3J(H,H)=8.8 Hz, 1H; H4), 4.19 (t, 3J(H,H)=
3.2 Hz, 1H; H3), 3.87 (dd, 3J(H,H)=3.9, 3J(H,H)=12.3 Hz, 1H; H7),
3
3
3.76 (d, 1H; H7), 3.65 (m, 1H; H5), 3.51 (m, J(H,H)=3.5 Hz, J(H,H)=
3
9.0 Hz, 1H; H2), 1.30 ppm (d, J(H,H)=6.7 Hz, 3H; H6); minor product,
14: d=4.45 (dd, 3J(H,H)=4.5, 3J(H,H)=7 Hz, 1H; H3), 4.10 (dd,
3J(H,H)=4.9, 3J(H,H)=9 Hz, 1H; H4), 3.79 (dd, 3J(H,H)=11.9 Hz, 1H;
H7), 3.73 (dd, 3J(H,H)=3.5, 3J(H,H)=11.9 Hz, 1H; H7), 3.69 (m, 1H;
H2), 3.61 (m, 1H; H5), 1.32 ppm (d, 3J(H,H)=6.7 Hz, 3H; H6); 13C
NMR (125 MHz, D2O, 25 8C) major product, 13: d=78.3 (C4), 71.3 (C3),
62.1 (C2), 58.1 (C7), 57.2 (C5), 11.8 ppm (C6); minor product 14: d=71.5
(C4), 70.4 (C3), 58.4 (C2), 58.6 (C7), 56.9 (C5), 11.5 ppm (C6).
Enzymatic aldol condensations in mixtures of dimethylformamide/water
(1:4) (condition D, Table 1): The reactions were carried out in 10-mL
screw-capped test-tubes. The aldehyde (0.4–0.9 mmol) was dissolved in
DMF (1 mL). Then, the DHAP solution (4 mL, 0.225–0.500 mmol), pre-
pared as described above, was added dropwise while mixing. The rest of
the experimental procedure was identical to that described above for the
reaction in emulsions.
(2R,3S,4R,5R)-2-(Hydroxymethyl)-5-methylpyrrolidine-3,4-diol (15): The
title compound was prepared according to the general procedure de-
scribed above. 1H NMR (500 MHz, D2O, 25 8C): d=4.27 (t, 3J(H,H)=
3.7 Hz, 1H; H3), 3.95 (dd, 3J(H,H)=3.7 3J(H,H)=9 Hz, 1H; H4), 3.92
(dd, 3J(H,H)=4.7, 3J(H,H)=11.8 Hz, 1H; H7), 3.80 (dd, 3J(H,H)=
8.4 Hz, 1H; H7), 3.75 (m, 3J(H,H)=3.7, 3J(H,H)=4.7, 3J(H,H)=8.4 Hz,
Physical data for the linear compounds 5–8: The melting points of the
compounds given below correspond to lyophilized solids rather than crys-
tals. Note that some of them are mixtures of C-4 epimers. The yields cor-
respond to the amounts of product derived from the aldol enzymatic re-
actions at the semipreparative level. The purification procedures were
not optimized.
3
1H; H2), 3.54 (m, 1H; H5), 1.38 ppm (d, J(H,H)=6.7 Hz, 3H; H6); 13C
NMR (125 MHz, D2O, 25 8C): d=76.7 (C4), 70.4 (C3), 61.5 (C2), 58.4
(C7), 56.7 (C5), 15.0 ppm (C6).
(3R)-6-(Benzyloxycarbonylamino)-5,6-dideoxy-1-O-phosphonohex-2-
ulose disodium salt (5): The title compound (160 mg, 35% yield) was pre-
pared by using conditions A (see Table 1) with 45mm DHAP. HPLC:
k’=5.2; m.p. 109–1118C (99.9% pure by HPLC); elemental analysis
calcd (%) for C14H18NO9Na2P·H2O·NaCl: C 33.79, H 4.05, N 2.81; found:
C 33.95, H 3.71, N 2.60. The 1H and 13C NMR spectra of the product, a
mixture of two diastereoisomers, were consistent with those previously
reported.[7]
(3R,4S)-5-Deoxy-5-phenyl-1-O-phosphonopent-2-ulose disodium salt
((4S)-18): The title compound was prepared according to the general pro-
cedure described above. HPLC: k’=3.2. 1H NMR (300 MHz, D2O,
258C): d=7.2 (m, 5H; Ph), 4.5 (dd, 3J(H,H)=7.0 Hz, 3J(H,H)=18.4 Hz,
2H; CH2OP), 4.1 (m, 2H; 2(CHOH)), 2.7 ppm (m, 2H; CH2Ph); 13C
NMR (75 MHz, D2O, 25 8C): d=213.7 (CO), 140.5 (C quat), 131.9 (C ar),
131.1 (C ar), 129.1 (C ar), 79.3 (CHOH), 75.1 (CHOH), 70.3 (CH2OP),
41.1 ppm (CH2).
(3R)-5-(Benzyloxycarbonylamino)-5-deoxy-1-O-phosphonopent-2-ulose
disodium salt (6): The title compound (130 mg, 26% yield) was prepared
by using conditions B (see Table 1) with 100mm DHAP. HPLC: k’=4.1;
m.p. 114–1188C (99.9% pure by HPLC); elemental analysis calcd (%)
(3R,4R)-5-Deoxy-5-phenyl-1-O-phosphonopent-2-ulose disodium salt
((4R)-18): The title compound was prepared according to the general
procedure described above. HPLC: k’=2.9. 1H NMR (300 MHz, D2O,
258C): d=7.2 (m, 5H; Ph), 4.5 (dd, 3J(H,H)=6.0 Hz, 3J(H,H)=18.7 Hz,
2H; CH2OP), 4.3 (d, 3J(H,H)=5.5 Hz, 1H; 2(CHOH)), 4.0 (m,
3J(H,H)=4.0 Hz, 3J(H,H)=5.1 Hz, 3J(H,H)=5.5, 1H; CHOH), 2.8 ppm
for C13H16NO9Na2P·5= H2O·NaCl: C 30.55, H 4.11, N 2.74; found: C
2
30.55, H 4.24, N 2.44. The 1H and 13C NMR spectra of the product, a mix-
Chem. Eur. J. 2005, 11, 1392 – 1401
www.chemeurj.org
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1399

P. Clapꢀs et al.
(m, 2H; CH2Ph); 13C NMR (75 MHz, D2O, 25 8C): d=213.7 (CO), 140.5
(C quat), 131.9 (C ar), 131.1 (C ar), 129.1 (C ar), 79.8 (CHOH), 75.4
(CHOH), 71.0 (CH2OP), 40.1 ppm (CH2).
the reaction media. Secondly, both the deprotection of the amino
group and the subsequent reductive amination were performed in a
one-pot procedure by treatment with H2 at 50 psi in the presence of
10% Pd/C. Aqueous solutions of each of the iminocyclitols in their
free-base form were adjusted to pH 6.5, lyophilized, and submitted
to NMR analysis without any further purification.
(3R,4S)-5-O-Benzyl-1-O-phosphonopent-2-ulose disodium salt ((4S)-19):
The title compound was prepared according to the general procedure de-
scribed above. HPLC: k’=4.6. 1H NMR (300 MHz, D2O, 25 8C): d=7.2
3
[13] In previous work (see ref. [7]), the relative configurations at the C-3
and C-4 positions of the linear structures 6–8 could not be unambig-
uously assigned by NMR analysis owing to overlapping signals re-
sulting from the equilibria between the linear adducts and the corre-
sponding cyclic hemiaminals. This was not the case with compound
5; less than 5% of the hemiaminal was formed and the proportion
of both diastereoisomers of the linear compound could be deter-
mined.
(m, 5H; Ph), 4.5 (m, 2H; CH2OP), 4.4 (s, 2H; CH2Ph), 4.3 (d, J(H,H)=
1.8 Hz, 1H; CHOH), 4.1 (m, 3J(H,H)=2.2 Hz, 3J(H,H)=5.5 Hz, 1H;
CHOH), 3.5 ppm (m, 2H; CH2O); 13C NMR (75 MHz, D2O, 25 8C): d=
212.9 (CO), 139.5 (C quat), 131.1 (C ar), 130.9 (C ar), 130.7 (C ar), 78.1,
75.5, 72.6, 72.1, 70.6 ppm (CH and CH2); 1H NMR (300 MHz,
3
[D6]DMSO, 258C): d=7.2 (m, 5H; Ph), 4.8 (dd, 3J(H,H)=6.9, J(H,H)=
18.3 Hz, 2H; CH2OP), 4.5 (s, 2H; CH2Ph), 4.1 (d, 3J(H,H)=1.8 Hz, 1H;
CHOH), 3.9 ppm (dt, 3J(H,H)=6, 3J(H,H)=9.3 Hz, 1H; CH2O); 13C
NMR (75 MHz, [D6]DMSO, 258C): d=207.8 (CO), 138.5 (C quat), 128.3
(C ar), 127.5 (C ar), 127.4 (C ar), 76.1, 72.2, 70.4, 70.3, 69.1 ppm (CH and
CH2).
[14] 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. Engl. 1991, 30, 555–558.
[15] T. Gefflaut, C. Blonski, J. Perie, M. Willson, Prog. Biophys. Mol.
Biol. 1995, 63, 301–340.
[16] 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. Engl. 1996, 35, 2219–2221.
[17] A. Dalby, Z. Dauter, J. A. Littlechild, Protein Sci. 1999, 8, 291–
297.
[18] D. R. Hall, G. A. Leonard, C. D. Reed, C. I. Watt, A. Berry, W. N.
Hunter, J. Mol. Biol. 1999, 287, 383–394.
(3R,4R)-5-O-Benzyl-1-O-phosphonopent-2-ulose disodium salt ((4R)-
19): The title compound was prepared according to the general procedure
described above. HPLC: k’=4.3. 1H NMR (300 MHz, D2O, 25 8C): d=
7.2 (m, 5H; Ph), 4.5 (m, 2H; CH2OP), 4.4 (s, 2H; CH2Ph), 4.3 (d,
3J(H,H)=1.5 Hz, 1H; CHOH), 4.0 (m, 3J(H,H)=5.1 Hz, 1H; CHOH),
3.4 ppm (m, 2H; CH2O); 13C NMR (75 MHz, D2O, 25 8C): d=212.9
(CO), 139.5 (C quat), 131.1 (C ar), 130.9 (C ar), 130.7 (C ar), 78.1, 75.5,
72.7, 71.7, 71.0 ppm (CH and CH2).
[19] A. R. Plater, S. M. Zgiby, G. J. Thomson, S. Qamar, C. W. Wharton,
A. Berry, J. Mol. Biol. 1999, 285, 843–855.
[20] C.-H. Wong, D. P. Dumas, Y. Ichikawa, K. Koseki, S. J. Danishefsky,
B. W. Weston, J. B. Lowe, J. Am. Chem. Soc. 1992, 114, 7321–7322.
[21] Y. F. Wang, D. P. Dumas, C.-H. Wong, Tetrahedron Lett. 1993, 34,
403–406.
[22] T. Sifferlen, A. Defoin, J. Streith, D. Le Nouen, C. Tarnus, I.
Dosbaa, M.-J. Foglietti, Tetrahedron 2000, 56, 971–978.
[23] A. Defoin, T. Sifferlen, J. Streith, I. Dosbaa, M.-J. Foglietti, Tetrahe-
dron: Asymmetry 1997, 8, 363–366.
[24] C.-H. Wong, R. Alajarin, F. Moris-Varas, O. Blanco, E. Garcia-Jun-
ceda, J. Org. Chem. 1995, 60, 7360–7363.
Acknowledgements
Financial support from the Spanish CICYT (PPQ2002-04625-CO2-01 and
BQU2003-01677) is acknowledged. J.C. acknowledges the CSIC for the
I3P predoctoral scholarship. We thank Prof. Josep Lꢊpez-Santꢅn and the
team at the Chemical Engineering Department, Universitat Autꢃnoma
de Barcelona, for the supply of FucA aldolase. The authors gratefully ac-
knowledge Dr. G. E. Schulz and Dr. A. C. Joerger for providing the coor-
dinates for the FucA model, which include the induced fit of the C-termi-
nal tail.
[25] M. Mitchell, L. Qaio, C.-H. Wong, Adv. Synth. Catal. 2001, 343,
596–599.
[26] A. C. Joerger, C. Gosse, W.-D. Fessner, G. E. Schulz, Biochemistry
2000, 39, 6033–6041.
[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] W.-D. Fessner, C. Walter, Top. Curr. Chem. 1996, 184, 97–194.
[4] H. J. M. Gijsen, L. Qiao, W. Fitz, C.-H. Wong, Chem. Rev. 1996, 96,
443–473.
[5] 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.
[6] R. Schoevaart, F. vanRantwijk, R. A. Sheldon, J. Org. Chem. 2001,
66, 4559–4562.
[27] The stereochemistry of each chromatographic peak was identified
by applying NMR techniques to purified samples of the major dia-
stereoisomers and to diastereomeric mixtures of the minor ones.
[28] M. K. Dreyer, G. E. Schulz, J. Mol. Biol. 1996, 259, 458–466.
[29] A. C. Joerger, C. Mueller-Dieckmann, G. E. Schulz, J. Mol. Biol.
2000, 303, 531–543.
[30] Kindly provided by G. E. Schulz and A. C. Joerger.
[31] T. A. Halgren, J. Comput. Chem. 1996, 17, 490–519.
[32] D. Qui, S. Shenkin, F. P. Hollinger, W. C. Still, J. Phys. Chem. A
1997, 101, 3005–3014.
[33] W. C. Still, A. Tempczyk, R. C. Hawley, T. Hendrickson, J. Am.
Chem. Soc. 1990, 112, 6127–6129.
[34] M. Schaefer, M. Karplus, J. Phys. Chem. 1996, 100, 1578–1599.
[35] P. A. Nielsen, T. Liljefors, J. Comput.-Aided Mol. Des. 2001, 15,
753–763.
[7] L. Espelt, T. Parella, J. Bujons, C. Solans, J. Joglar, A. Delgado, P.
Clapꢀs, Chem. Eur. J. 2003, 9, 4887–4899.
[8] FucA was produced by the Department of Chemical Engineering,
Universitat Autꢃnoma de Barcelona, within the framework of a col-
laborative CYCIT project (PPQ2002-04625-CO2-01).
[9] L. Espelt, P. Clapꢀs, J. Esquena, A. Manich, C. Solans, Langmuir
2003, 19, 1337–1346.
[10] R. L. Pederson, J. Esker, C.-H. Wong, Tetrahedron 1991, 47, 2643–
2648.
[11] W.-D. Fessner, J. Badia, O. Eyrisch, A. Schneider, G. Sinerius, Tetra-
hedron Lett. 1992, 33, 5231–5234.
[12] Iminocyclitols were obtained from compounds 5–8 in two steps
(Scheme 1). Firstly, the phosphate group was removed by treatment
with acid phosphatase followed by desalting by RP-HPLC. No base-
line separation of the putative diastereoisomers was observed by an-
alytical RP-HPLC. Therefore, the purpose of the purification step
was to isolate the aldol adducts from the other species present in
[36] F. I. Auzanneau, E. Sourial, J. M. Schmidt, M. Feher, Can. J. Chem.
2002, 80, 1088–1095.
[37] S. Perez, A. Imberty, S. B. Engelsen, J. Gruza, K. Mazeau, J. Jime-
nez-Barbero, A. Poveda, J. F. Espinosa, B. P. van Eyck, G. Johnson,
A. D. French, M. Louise, C. E. Kouwijzer, P. D. J. Grootenuis, A.
Bernardi, L. Raimondi, H. Senderowitz, V. Durier, G. Vergoten, K.
Rasmussen, Carbohydr. Res. 1998, 314, 141–155.
[38] W. Fitz, J.-R. Schwark, C.-H. Wong, J. Org. Chem. 1995, 60, 3663–
3670.
[39] S.-H. Jung, J.-H. Jeong, P. Miller, C.-H. Wong, J. Org. Chem. 1994,
59, 7182–7184.
1400
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1392 – 1401

Aldol Additions of Dihydroxyacetone Phosphate to Amino Aldehydes
FULL PAPER
[40] 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.
[42] 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.
Received: June 28, 2004
[41] F. Effenberger, A. Straub, Tetrahedron Lett. 1987, 28, 1641–1644.
Published online: January 24, 2005
Chem. Eur. J. 2005, 11, 1392 – 1401
www.chemeurj.org
ꢄ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1401