Fructose 1,6-bisphosphate aldolase from Staphylococcus carnosus: Overexpression, structure prediction, stereoselectivity, and application in the synthesis of bicyclic sugars

Article abstract of DOI:10.1002/(SICI)1521-3765(19990604)5:6<1882::AID-CHEM1882>3.0.CO;2-C

The gene for the fructose 1,6-bisphosphate aldolase from Staphylococcus carnosus (FruA(sca)) was subcloned for overexpression in Escherichia coli using the expression vector pKK223-3. An efficient, single-step purification by DEAE ion-exchange chromatography furnished the recombinant enzyme ready for synthetic applications. Sequence analysis indicated that FruA(sca) shares the overall α/β-barrel structure and most of the active site residues with the structurally well-defined FruA catalyst from rabbit muscle which signaled its functional equivalence for synthetic applications. A preparative study with generic aliphatic and hydroxylated aldehydes indeed confirmed a high level of stereoselectivity for both newly created asymmetric centers, and suggested a kinetic enantioselectivity for anionically charged 3- hydroxyaldehydes. In fact, the monomeric FruA(sca) was found to tolerate even the presence of highly reactive glutardialdehyde derivatives, which otherwise rapidly denature the rabbit muscle enzyme, and to allow their stereoselective conversion to bicyclic sugars.

Full text of DOI:10.1002/(SICI)1521-3765(19990604)5:6<1882::AID-CHEM1882>3.0.CO;2-C

FULL PAPER
Enzymes in Organic Synthesis, Part 14 [= ]
Fructose 1,6-Bisphosphate Aldolase from Staphylococcus carnosus:
Overexpression, Structure Prediction, Stereoselectivity, and Application in
the Synthesis of Bicyclic Sugars
Maria Teresa Zannetti, Christiane Walter, Marion Knorst, and Wolf-Dieter Fessner*[a]
Abstract: The gene for the fructose 1,6-
bisphosphate aldolase from Staphylo-
the structurally well-defined FruA cata-
lyst from rabbit muscle which signaled
its functional equivalence for synthetic
applications. A preparative study with
generic aliphatic and hydroxylated alde-
hydes indeed confirmed a high level of
stereoselectivity for both newly created
asymmetric centers, and suggested a
kinetic enantioselectivity for anionically
charged 3-hydroxyaldehydes. In fact, the
monomeric FruAsca was found to toler-
ate even the presence of highly reactive
glutardialdehyde derivatives, which oth-
erwise rapidly denature the rabbit mus-
cle enzyme, and to allow their stereo-
selective conversion to bicyclic sugars.
coccus carnosus (FruA ) was subcloned
sca
for overexpression in Escherichia coli
using the expression vector pKK223-3.
An efficient, single-step purification by
DEAE ion-exchange chromatography
furnished the recombinant enzyme
ready for synthetic applications. Se-
quence analysis indicated that FruAsca
shares the overall a/b-barrel structure
and most of the active site residues with
Keywords: aldol reactions
´ a/b-
barrel protein ´ asymmetric synthe-
sis ´ carbohydrates ´ enzyme catal-
ysis
Introduction
Mechanistically, activation of nucleophilic substrates by
aldolases is achieved in two different ways.[ Class I enzymes
bind their substrates covalently by Schiff base formation to an
6]
Asymmetric C�C bond formations are the most important
[7]
and most challenging problems in synthetic organic chemistry.
In Nature, such reactions are facilitated by lyases which
catalyze the addition of carbonucleophiles to CˆO bonds in a
active site lysine residue; these catalysts are typically
tetrameric proteins[ composed of subunits of ꢀ40 kDa and
are found usually in higher plants and in mammalian or (as an
exception) in specific microbial organisms. Class II aldolases
8]
manner that is mechanistically classified as an aldol addi-
[
1]
2‡
tion. With respect to synthetic applications, the four stereo-
chemically distinct dihydroxyacetone phosphate (DHAP)
utilize a divalent cation cofactor (usually Zn ) for substrate
coordination;[ these enzymes are usually dimers
9]
[10, 11]
of
[
2]
dependent enzymes are particularly appealing because they
control the creation of two new asymmetric centers at the
termini of the newly formed C�C bond, thus allowing an
ꢀ39 kDa subunits and are commonly found in fungi and
bacteria.
The variant glycolysis aldolases that cleave fructose 1,6-
bisphosphate (FBP) have been pivotal to mechanistic ad-
vancement, structural insight, and synthetic utility.[ In vivo,
fructose 1,6-bisphosphate aldolases (FruA) catalyze the
reversible addition of DHAP to d-glyceraldehyde 3-phos-
phate (GA3P) to give FBP. These enzymes are highly specific
[
3±5]
effective combinatorial synthesis of stereoisomers.
12]
‡]
[
a] Prof. Dr. W.-D. Fessner,[ Dr. M. T. Zannetti, Dr. C. Walter,
Dipl.-Chem. M. Knorst
Institut für Organische Chemie, Rheinisch-Westfälische Technische
Hochschule Aachen
Professor-Pirlet-Strasse 1, D-52056 Aachen (Germany)
for the DHAP nucleophile and tolerate only a few isosteric
replacements in the phosphate moiety,[
13±15]
but they readily
accept a broad range of aldehydes at synthetically useful rates
‡
[
] Current address:
[3, 17]
as the electrophile.
The class I FruA from Oryctolagus
Institut für Organische Chemie, Technische Universität Darmstadt
Petersenstrasse 22, 64287 Darmstadt (Germany)
Fax : (‡ 49)6151-16-6636
cuniculus (rabbit) muscle (FruArab; for the systematic acro-
[3]
nym nomenclature see ) has to be regarded as the classical
reference catalyst since it has long been commercially
available, has tested positive with literally hundreds of
E-mail: fessner@tu-darmstadt.de
[
=] Part 13, see ref. [15].
1
882
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1882 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 6

1
882±1890
aldehydes as substrate analogues in place of the natural
substrate d-glyceraldehyde 3-phosphate, and has proved to be
eminently useful for the stereoselective synthesis of carbohy-
drates as well as related natural and nonnatural compounds.[
However, its main limitations as a practical catalystÐand
tions. We also present a sequence comparison with other
class I FruA proteins which suggests a common three-dimen-
sional fold and thus a similarly broad synthetic applicability.
The extraordinary stability of the enzyme is demonstrated by
the preparation of bicyclic sugars from a glutardialdehyde
derivative.
3]
[
16]
obstacles to any large-scale, industrial utilization Ðremain
in its relatively low stability under the usual reaction
conditions with half-lives of only a few days in buffered
solution, and in the narrow restrictions for application of
Results and Discussion
[
17]
cosolvents.
FruAsca overexpression
In comparison, it has been reported that the prokaryotic
class I aldolases from Staphylococcus aureus and Pseudomo-
nas aerogenes display unusually high heat and pH stability,[
which is most likely due to the fact that these ꢀ33 kDa
enzymes are active as monomers. More recently it was found
that the related, monomeric class I aldolase from Staph-
The S. carnosus gene fda encoding the FruAsca had previously
been cloned as a 5.2 kb PstI fragment of chromosomal DNA
18]
[20]
into the pBluescriptII KS ‡ vector. The resulting construct,
designated pBluescriptII-fda10, was now used for the con-
struction of a suitable overexpression system. The vector
pKK223-3 was chosen because it contains the strong hybrid
tac promoter and rrnB ribosomal termination for controlled
ylococcus carnosus (FruA ) also has a high pH and temper-
sca
[
19]
ature stability, and the encoding gene has been cloned and
[
20]
[23]
the DNA sequence determined. Furthermore, this catalyst
has been shown by a TLC screening to accept a number of
aldehydes as substitutes of the natural substrate GA3P.[
Although the FruAsca enzyme is now offered commercially,[22]
a more detailed investigation of its synthetic utility was
hampered so far by the high cost of the marketed catalyst and
by the efforts required for
protein expression, and because it has proved to be highly
[
24, 25]
successful in the overproduction of other aldolases.
The
21]
fda gene was amplified from pBluescriptII-fda10 by mutating
PCR technique to insert two flanking recognition sequences
for appropriate restriction enzymes using the primers descri-
bed in Scheme 1. The sense primer introduced an XmaI
purification of the wild type
[
19]
enzyme.
Here we report on the con-
struction of a new and highly
efficient heterologous overex-
pression system for FruAsca in
E. coli from which the enzyme
can be obtained by a single-
step purification in a state
ready for synthetic applica-
Scheme 1. Oligonucleotide primers used for the subcloning of FruAsca. Recognition sequences for XmaI and PstI
restriction endonucleases are doubly underlined. Sequence stretches with identity to genetic DNA are indicated
by capital letters; mutating sequences introduced by PCR extension are given by lowercase letters.
restriction site immediately before the geneꢁs natural Shine ±
Abstract in German: Das Strukturgen der Fructose-1,6-bis-
Dalgarno sequence which is located eight base pairs from the
ATG start codon, and the antisense primer incorporated a
PstI restriction site after the stop codon. From the PCR
amplification only a single band with the expected molecular
weight (950 bp) was observed upon DNA agarose gel electro-
phoresis, and no further purification was required. As
summarized by the cloning strategy shown in Figure 1, the
PstI and XmaI digested insert was unidirectionally ligated
into the pKK223-3 vector and the construct, denoted pKKfda,
was transformed into competent E. coli JM105 strain. Out of
six colonies selected from LB-ampicillin plates, four carried
the desired functional insert as determined by enzymatic
assay for FruA activity.[
phosphataldolase aus Staphylococcus carnosus (FruA ) wur-
sca
de für eine Überproduktion in Escherichia coli in den
Expressionsvektor pKK223-3 subkloniert. Mittels Einschritt-
Reinigung an DEAE-Ionenaustauscher konnte das rekombi-
nante Enzym effizient in für Synthesereaktionen ausreichender
Qualität gewonnen werden. Eine Sequenzanalyse deutet darauf
hin, daû die FruAsca die a/b-Barrel-Structure und die meisten
Aminosäurereste des aktiven Zentrums mit dem stukturell gut
charakterisierten FruA-Enzyms aus Kaninchenmuskel teilt,
was seine funktionelle ¾quivalenz für Syntheseanwendungen
nahelegte. Tatsächlich bestätigte eine präparative Studie mit
typischen aliphatischen und hydroxylierten Aldehyden ein
hohes Maû an Stereoselektivität in der Erzeugung beider neuer
Asymmetriezentren und zeigte eine kinetische Enantioselekti-
vität für anionisch geladene 3-Hydroxyaldehyde auf. Die
monomere FruAsca erwies sich selbst gegenüber hochreaktiven
Glutardialdehydderivativen, welche ansonsten das Kaninchen-
muskelenzym rasch denaturieren, als unerwartet stabil, so daû
selbst deren stereoselektive Umsetzung zu bicyclischen Zuk-
kern gelang.
26]
Recombinant cells of E. coli JM105/pKKfda were found to
�
1
overexpress the FruAsca at about 1 kUg wet cell weight (3 ±
kU per liter of cell culture) upon induction of the tac
4
promoter with 0.5mm IPTG which corresponds to roughly a
factor of 50 relative to wild type S. carnosus cells,[ or to a
factor of 20 relative to a previously described autologously
cloned S. carnosus producer.[ Analysis by denaturing SDS
polyacrylamide gel electrophoresis showed that the fully
19]
20]
Chem. Eur. J. 1999, 5, No. 6
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1883 $ 17.50+.50/0
1883

W.-D. Fessner et al.
FULL PAPER
Figure 1. Summary of the subcloning of the FruAsca gene for protein overexpression.
induced recombinant cells contained about 25% of the
FruAsca among total soluble cell protein. Despite the high
levels of foreign gene production, no formation of inclusion
bodies was detected. Because of its strong prevalence, the
recombinant FruAsca could be easily obtained in relatively
high purity (ꢁ90%) and high yield (87%) by a single anion-
exchange chromatography on DEAE sepharose with only one
other major band visible at about 70 kDa, as demonstrated by
SDS-PAGE analysis (Figure 2). In this state, the enzyme was
FruAsca protein structure prediction
The application of enzymic catalysts to asymmetric synthesis
is intimately tied up to the intricacies of their protein
structure. Despite any apparent similarity in primary se-
quence or primary function, the peculiarities of related
catalysts require an empirical scrutiny for their catalyst utility.
Thus, when comparing enzymes from different sources but
having a common catalytic specificity towards the same
natural substrate, each catalyst quality with regard to the
breadth of substrate specificity, such as tolerance for non-
natural substrate analogues, enantioselectivity, and diaster-
eoselectivity, or physical properties, such as heat, solvent, or
pH stability, has to be evaluated individually. In an attempt to
investigate the general synthetic features of the FruAsca with
respect to the established reference Frurab enzyme, we first
tried to probe the structural relationship of the catalysts in
more detail.
Apart from a very narrow sequence comparison of the
FruA enzymes from S. carnosus with that from human tissues,
mays, Drosophila and S. aureus around the active site lysine
(
K212) no structural information for the FruA was avail-
sca
able.[ Particularly, no data on secondary or tertiary structure
20]
elements for the three-dimensional protein folding had been
determined. Conversely, the X-ray structures of the highly
Figure 2. Electrophoretic analysis of the expression of the recombinant
FruAsca in E. coli by SDS-PAGE separation on a 12% gel. Protein bands
were stained by Coomassie blue. A: Molecular weight markers with the size
indicated in kilodaltons to the right; expected subunit size of FruAsca is
indicated in boldface. B: Total cell lysate of strain JM105/pKKfda without
induction. C: Total cell lysate of strain JM105/pKKfda after induction with
[7, 8]
homologous class I FruA proteins from rabbit (1.9 Š)
and
[27]
human muscle (2.0 Š),
and that from Drosophila mela-
nogaster (2.5 Š)[ have been determined at a useful level of
accuracy. Their molecular architecture corresponds to that of
active tetrameric associates of the respective enzyme subunits,
which adopt an eight-stranded a/b- barrel structure motif. The
active site cleft is located in the center of the b- barrel with the
reactive Lys229 projecting towards the C-terminal opening of
the pocket (Figure 3left). A number of additional ionic and
polar residues converge towards the opening of the active site
pocket for substrate binding (Figure 3right).
28]
0.5mm IPTG. D: Purified FruAsca after single-step DEAE chromatography.
found to be quite stable upon storage (95% activity remaining
after 12 months at 48C) and free from activities interfering in
synthetic applications. If desired, practically pure protein
(99%) may be obtained by a further gel permeation
chromatography (GPC) to remove impurities of higher
molecular weight. Also by GPC on Sephadex G150 the
molecular weight of the native FruAsca was determined to be
about 30 kDa which verified the prior analysis of the aldolase
The deduced amino acid sequence of the S. carnosus fda
gene (296 aa)[ was compared to that of the corresponding
rabbit gene for the muscle isozyme (364 aa).[ Sequence
alignment of the core structure (280 aa) for the FruAsca and
FruArab revealed an amino acid identity of less than 30% but
20]
29]
[
19]
to be active in a monomeric state.
1
884
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1884 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 6

Enzymes in Organic Synthesis
1882±1890
Figure 3. Left: Structural view of the fructose 1,6-bisphosphate aldolase, determined by X-ray structure analysis of FruArab, showing the a/b-barrel protein
[
10]
architectural motif. The plot was generated from PDB file 1ADO. b-Strands are shown in cyan and a-helical segments in red. Right: Location of active site
[
8]
residues in the FruArab enzyme as identified by X-ray crystal structure analysis. Labeling of amino acid residues is indicated according to the primary
sequence of FruArab. Residues that are conserved in the S. carnosus aldolase (FruAsca), as determined by sequence alignment, are shown in standard color
code for the elements, and side chains of non-conserved residues are highlighted in yellow. The corresponding substitutions occurring in FruAsca are
Glu34Gln, Arg42Ala, and Thr268Val.
with enough indication for a likely similar overall fold. When
making an allowance for conservative replacements, the
homology rose to a remarkable 45% (Figure 4). By inspection
of this analysis, it becomes evident that there is a strong
clustering of similarity in particular stretches of the primary
sequence which strongly correlate with all eight b-sheet
In addition, out of the total of 17 amino acid residues
contained in the active site, which are situated within the b-
strands or at loops proceeding from the latter, 14 are identical
and only three are not conserved (Figure 3right). Since the
secondary structure elements are thus responsible for the
formation the b-barrel core and also mediating the catalytic
activity, it can thus safely be deduced that the FruAsca shares
with the FruArab not only a common a/b-barrel structure but
segments established for the FruA enzyme (sheets desig-
rab
[
8]
nated a ± h in the crystal structure; almost 80% homology).
Figure 4. Sequence-based alignment of the amino acid sequences of fructose 1,6-bisphosphate aldolases from rabbit muscle (FruArab) and Staphylococcus
carnosus (FruAsca). Red amino acid single-letter codes highlight sequence identity, blue amino acid residues indicate conservative replacements. Black dots
[
8]
above the sequence of FruArab indicate polar active site residues. Elements of secondary structure, as determined by X-ray analysis of FruArab
below the alignment. Arrows represent b-strands (a ± h), and coils represent a-helices (A ± H ); other secondary structure elements, such as random coils and
turns, are represented by a solid line; coloration of the symbols corresponds to that used in Figure 3.
, are drawn
1
2
Chem. Eur. J. 1999, 5, No. 6
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1885 $ 17.50+.50/0
1885

W.-D. Fessner et al.
FULL PAPER
will also very likely have quite similar catalytic properties. A
comparable, if less pronounced, prediction was deduced by
using a recently developed algorithm for sequence-derived
[
30]
fold recognition.
Interestingly, the regions with higher diversity are located
in the a-helical protein segments which make up the
periphery of the three-dimensional structure. The highest
disagreement in fact is found in helices E and F which flank
the b-strand containing the Schiff-base forming lysine (K229
Table 1. Evaluation of FruAsca stereoselectivity.
Aldehyde
R
d-threo l-erythro
Yield
[%]
[%]
[%]
acetaldehyde
CH
3
97
3
2
< 5
< 5
< 5
90
80
90
90
90
in FruArab, K212 in FruA ). The fact that these helices are
propionaldehyde
glycolaldehyde
rac-glyceraldehyde
C H5
98
> 95
OH > 95
sca
2
CH
2
OH
mostly responsible for creating the rigid, nonpolar sur-
face contacts between aggregated subunits of the tetrameric
CHOH�CH
2
succinic semialdehyde (CH
2
)
2
COOH
> 95
[
31]
aldolases makes it clear that profound alterations towards
improved hydrophilicity (Figure 5) are mandatory for
these particular entities to maintain a monomeric state of
isomers was detectable by the distinctive signals of 3-/4-H.
Thus, the FruAsca enzyme has a very high capacity to correctly
bind the CHO group, even in
the FruAsca
.
the most discriminate case of
CH3 versus H differentiation.
Within the limits of detection
by high-field proton and carbon
NMR spectroscopy (ꢂ5%),
only single diastereomers were
formed from 2- or 3-hydroxy-
lated glycolaldehyde and glyc-
eraldehyde, as well as from
succinic acid semialdehyde. In
context with similar observa-
tions this suggests that polar
Figure 5. Plots of hydropathy index against residue number for fructose 1,6-bisphosphate aldolases from
Staphylococcus carnosus (FruAsca) and rabbit muscle (FruArab). The peaks and troughs are filled for an easy
comparison of the profiles. Both plots were generated with an average segment length of five residues. Secondary
structure elements for FruArab are indicated by arrows for b-strands and by capital letters for a-helices, as deduced
from the crystal structure; corresponding structure elements are tentatively assigned for FruAsca as derived from
the sequence alignment shown in Figure 4.
substrates are more strongly
bound by additional hydrogen
bonding or Coulomb attraction.
Thus, substrate analogues com-
prising structural features of the
natural substrate glyceralde-
hyde 3-phosphate seem to be
FruAsca stereoselectivity
excellently suited for complete stereochemical control over
the newly generated vicinal diol unit of an absolute (3S,4R)-
trans configuration (d-threo).
This FruAsca catalyst had been shown earlier to accept a
number of aldehydes as substitutes of the natural substrate
It had been shown previously that the FruArab displays a
pronounced kinetic enantioselectivity for d-configurated
2-hydroxyaldehydes when these bear an anionic group at a
distance similar to that in the natural acceptor.[ In an effort
to study the complementary behavior towards 3-hydroxyal-
dehydes for both of the FruA enzymes, racemic maleic
4-semialdehyde 3 (readily obtainable by a Barbier-type[
chain extension of glyoxylic acid 1, followed by ozonolysis)
was submitted to enzymic aldolization in the presence of
substoichiometric amounts of DHAP (0.3 equivalents;
Scheme 2). NMR spectroscopic analysis of the crude material
revealed that the initially produced diastereomeric open-
chain intermediates 4 and 5 expectedly had cyclized to form
pyranoid rings. As evident from the coupling pattern of the
[
21]
GA3P. However, the simple TLC screening method used to
detect product formation was unsuited to reveal the constitu-
tional identity of the product formed nor its configurational
purity. An evaluation of the stereoselectivity of the aldolase,
however, was particularly desirable as a crucial precondition
for potential applications in asymmetric synthesis.
32]
33]
The stereoselectivity of the purified FruAsca enzyme was
determined by a series of small-scale preparative reactions in
the direction of synthesis, employing two aliphatic, two
hydroxylated, and one carboxyl aldehydes as representative
substrate analogues (Table 1). In each case, reaction rates
were sufficiently fast to maintain the product formation under
kinetic control. At about 90% conversion, crude product
1
mixtures were analyzed by high-field H NMR for diastereo-
meric composition, using authentic reference materials ob-
methylene group, the major component 6 derives from d-3
6
and adopts a b- C
-chair conformation. The minor stereo-
3
tained from parallel reactions catalyzed by FruA or as
available from prior studies.
isomer, produced from l-3, was determined not to be the
rab
[
25]
3
anticipated corresponding C
-chair but rather the derived
6
From nonpolar aliphatic acetaldehyde and propionalde-
intramolecular anomeric lactone 7, probably due to the acidic
hyde, a very small percentage of erythro-configurated stereo-
conditions maintained upon sample preparation. Both cata-
1
886
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1886 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 6

Enzymes in Organic Synthesis
1882±1890
Scheme 3. Two-step synthesis of dialdehyde substrate, and attempted
tandem aldol additions using fructose 1,6-bisphosphate aldolase from
rabbit muscle (FruArab). i) O
5%; ii)
MeOH, � 788C; Me
quantitative. P ˆ PO
3
, MeOH, � 788C; NaBH
4
, room temperature,
S, � 788C !room temperature,
5
ꢀ
O
3
,
2
2
�
3
.
crosslink FruA even at very low concentration,[42] the
derivative 9 was expected to display a moderate reactivity
because of its ability for intramolecular cyclization to form
stable six-membered hemiacetals. In fact, NMR analysis
showed a very complex mixture comprising free aldehydes
Scheme 2. Kinetic enantioselectivity for 2-hydroxy-4-oxobutanoate.
i) Allyl bromide, Sn, EtOH, 45%; ii)
788C !room temperature, ꢀquantitative; iii) FruA, DHAP; ion ex-
change, 6:7 ˆ 85:15 (FruArab) and 90:10 (FruAsca).
O
3
,
MeOH, � 788C; Me
2
S,
�
(broad singlets at d ˆ 9.72 and 9.67), hydrates, and isomeric
lysts showed a similar moderate selectivity for the d-
structures (e.g. 9a,b) to exist in aqueous solution, correspond-
ing to the different possible modes of cyclization. Owing to
the presence of a relatively high portion of free aldehyde,
however, to our dismay a copious precipitate formed within
few minutes when a solution of dialdehyde 9 and DHAP was
incubated with FruArab. No enzymatic activity remained in
solution, no trace of aldolization product was detectable from
this attempt.
configurated aldehyde (FruArab d:l ˆ 85:15, FruA 90:10).
sca
Interestingly, the selectivity was significantly lower for the
corresponding ethyl ester (FruArab d:l ˆ 40:60). The latter
result not only points to the importance of an anionically
charged group but also strongly corroborates the kinetic
nature of the observed selectivity since the all-equatorially
substituted 6 could alternatively have arisen by its higher
[
3]
thermodynamic stability. The fact that from the latter
reaction no ethyl ester remained in the product is in agree-
ment with earlier observations for related substrate types.[
The FruAsca had been reported to show a far superior
stability in comparison to the FruA under typical reaction
34, 35]
rab
conditions.[ Since this is quite likely a consequence of its
monomeric structure, we anticipated also a higher resistance
against crosslinking denaturants. In fact, this supposition
proved correct: when FruAsca was treated with 9 in the
presence of DHAP no precipitate formed but a rapid
consumption of DHAP was monitored by enzymatic assay
(Scheme 4). Analysis by TLC indicated the formation of
monophosphate(s), however, without traces of bisphosphate.
Further conversion of monophosphate intermediates to bi-
sphosphates could also not be detected upon prolonged
incubation. After enzymatic dephosphorylation, the annulat-
ed C -dipyranoses 13b and 14b were obtained as a 3:1
21]
Synthetic applications
Although several attempts at enzymatic addition of DHAP to
simple linear a,w-dialdehydes (glyoxal, glutardialdehyde)
[
36, 21]
have been reported,
in no case had a product been
isolated and characterized. Presumably, this is because the
dialdehydes cause crosslinking of the protein
[
37, 38]
and thus
irreversibly destroy its enzymatic activity. However, we have
recently discovered that a- or b-hydroxylated dialdehydes are
good substrates for a twofold, tandem aldol addition of
[
39, 40]
DHAP.
The contrasting results may be explained by the
10
formation of stable intramolecular hemiacetals in aqueous
solution which mask the reactivity of free dialdehydes.
In this context, the branched chain glutardialdehyde 9 was
of interest as a potential precursor to hydrolytically stable,
one-carbon linked disaccharide mimetics (e.g., 10). The meso-
cyclopentene-3,5-dimethanol 8 was prepared by controlled
ozonolysis of norbornadiene, followed by reduction with
diastereomeric mixture in 26% overall yield from 8. Appa-
rently, the two primary monophosphates 11 and 12 form
rather stable intramolecular acetals 13a and 14a, respectively,
thus effectively precluding a second addition.
To verify the structure as well as to gain insight into the
origin of enantiotopos selectivity, it was desirable to submit
substrate 9 to aldolization with the rhamnulose 1-phosphate
aldolase (RhuA) having a threo stereoselectivity which is
configurationally opposite to that of FruA enzymes.[ Our
experience showed that this class II aldolase from E. coli,
[
41]
NaBH (Scheme 3). Ozonolytic cleavage of the remaining
4
25]
double bond cleanly furnished the corresponding dialdehyde
9. In contrast to glutardialdehyde itself, which is reported to
Chem. Eur. J. 1999, 5, No. 6
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1887 $ 17.50+.50/0
1887

W.-D. Fessner et al.
FULL PAPER
Clearly, because of the contrasting identity of the major
products, the enantiotopos selectivity in the terminus differ-
entiation is (at least partially) of kinetic, and not of
thermodynamic, origin.
Conclusions
We have constructed a new heterologous overexpression
system for the FruAsca enzyme which by far excels the
autologous production system reported previously. The high
productivity and the simplicity of an effective one-step
purification scheme make this a suitable source for large
quantities of the biocatalyst. We have shown that its reported
superior stability is paralleled by a high level of stereo-
selectivity for aldol addition reactions with generic aliphatic
and hydroxylated aldehydes, an experimental outcome which
had been suggested, in fact, by the discovery of high sequence
and structural similarity around the known active site of the
FruArab and the postulated active site environment of the
Scheme 4. Aldol reaction using fructose 1,6-bisphosphate aldolase from S.
carnosus (FruAsca). i) DHAP, FruAsca, pH 7.0; ii) alkaline phosphatase,
2�
pH 8.0; ratio 13:14 ˆ 3:1, 26% from 8. P ˆ PO
3
.
FruAsca
.
The synthetic utility of the enzyme in asymmetric synthesis
was demonstrated by the formation of a new set of unusual
bicyclic sugars from bifunctional precursors of a type that
proved to be impossible to convert by using conventional
FruArab catalysis. Kinetic enantioselectivity could be demon-
strated for an anionically charged 3-hydroxyaldehyde. Thus,
when compared to the FruArab enzyme as the current stand-
ard, the FruAsca is more than functionally equivalent, and
promises to become a cost-efficient and overall effective
replacement, particularly with regard to potential large-scale
applications.[
available commercially or by recombinant overproduction,[25]
also is a rather robust catalyst having a very broad substrate
tolerance. Gratifyingly, when a solution of 9 and DHAP was
treated with RhuA, no protein precipitation occurred and
analysis by TLC indicated monophosphate formation
[
3]
(Scheme 5). Again, no bisphosphates were detectable at
4]
Experimental Section
Materials and methods: Plasmid pBluescriptII-fda10 was provided by
Professor Götz, Tübingen. Oligonucleotide primers, vector pKK223-3,
Escherichia coli strain JM105, and media for protein chromatography were
purchased from Pharmacia Biotech. PCR Core Kit, restriction enzymes,
and T4 DNA ligase were obtained from Boehringer Mannheim and used
according to the manufacturerꢁs instructions. Other routine cloning
[
43]
operations were performed by the standard procedures. The nuclease
Benzonase, grade I) was purchased from Merck, Darmstadt.
(
NMR spectra were recorded on Varian VXR 300 or Unity 500 spectrom-
eters; chemical shifts are referenced to internal TMS or TSP (d ˆ 0.00).
Mass spectra were recorded on a Finnigan MAT 212 system and elemental
analyses were performed on a Heraeus CHN-O-Rapid system. A Fischer
5
02 ozone generator was used for ozonolyses; ultrasound-accelerated
Scheme 5. Aldol reaction using rhamnulose 1-phosphate aldolase from E.
coli (RhuA). i) DHAP, RhuA, pH 7.0; ii) alkaline phosphatase, pH 8.0;
reactions were carried out with the aid of a Bandelin Sonorex TK 52 H
cleaning bath. Column chromatography was performed on Merck 60 silica
gel (0.063 ± 0.200 mesh), and analytical thin-layer chromatography (TLC)
was performed on Merck silica gel plates 60 GF254 using a 1:1 mixture of
saturated ammonia/ethanol for development, and anisaldehyde stain for
detection. Analytical grade ion exchange media (100 ± 200 mesh) were
purchased from Bio-Rad. Fructose 1,6-bisphosphate aldolase from rabbit
muscle (FruA; [EC 4.1.2.13]; type IV) and triose phosphate isomerase from
rabbit muscle ([EC 5.3.1.1]; type I) were purchased from Sigma, and
recombinant rhamnulose 1-phosphate aldolase from E. coli (RhuA; [EC
2
�
ratio ent-13 : ent-14 ˆ 1:3, 32% from 8. P ˆ PO
3
.
longer incubations. After dephosphorylation, a mixture of two
diastereomers was isolated which were identified by NMR
analysis to correspond to products ent-13b and ent-14b in a
ratio of 1:3 (32% overall yield from 8). Interestingly, the
stereoisomer ratio (major component ent-14b) was opposite
to that registered for the FruAsca experiment (major compo-
nent 13b), which infers that both catalysts had preferentially
attacked the carbonyl group of identical enantiotopicity.
4
.1.2.19]) was obtained from Boehringer Mannheim. Alkaline phosphatase
from bovine intestinal mucosa ([EC 3.1.3.1], synthetic purity grade) was
purchased from Fluka. Dihydroxyacetone phosphate (DHAP) was pre-
pared from glycerol phosphate according to the published procedure.[
13]
1
888
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1888 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 6

Enzymes in Organic Synthesis
1882±1890
PCR amplification: The gene fda contained in the plasmid pBluescriptII-
fda10[ was amplified by using the standard PCR technique as described in
the literature[ but with mutating oligonucleotide primers as shown in
Scheme 1. Amplification was performed in a 100 mL reaction mixture
containing 1 ng of plasmid pBluescriptII-fda10, 300 nm each of primers,
albumin for calibration. Protein purity was analyzed by SDS-PAGE
performed according to the method of Laemmli using a 12% gel.
20]
[46]
44]
Sequence analysis: The protein sequence of the S. carnosus fructose 1,6-
bisphosphate aldolase (296 aa) and the corresponding rabbit sequence of
the muscle isozyme (364 aa), as derived from published gene sequences
200mm each of deoxynucleoside triphosphates (dNTPs), 10mm Tris-HCl
(
GenBank accession numbers X71729 and K02300, respectively), were
(
pH 8.3), 1.5mm MgCl , and 50mm KCl. The reaction mixture was heated
2
analyzed by using the program MegAlign (DNASTAR Inc.). Clustal
method with default parameters was applied for the analysis.
for 2 min at 958C, then Taq DNA polymerase (2.5 U) was added, the
solution was layered with mineral oil (100 mL) and subjected to 25 cycles of
amplification (Eppendorf Mastercycler 5330). Cycle conditions were set for
denaturation at 948C for 1 min, annealing at 508C for 2 min, and
elongation at 728C for 3 min. The mineral oil was removed and DNA
was collected by phenol/chloroform extraction and ethanol precipitation.
Analysis by 1% agarose gel electrophoresis showed the desired DNA
insert (950 bp) to be the only detectable polymer chain reaction (PCR)
product which was used for the subsequent steps without further
purification.
Analysis of hydropathy index: The hydropathy index of the fructose 1,6-
bisphosphate aldolases from S. carnosus and rabbit muscle was determined
and plotted according to the method of Kyte and Doolittle,[
implemented in the program Protean (DNASTAR Inc.).
47]
as
Evaluation of stereoselectivity: An aqueous solution (20 mL) containing
fructose 1,6-bisphosphate (25mm) and the respective aldehyde (50mm) was
adjusted to pH 6.8. After addition of triosephosphate isomerase (10 U) and
FruA (from S. carnosus or rabbit muscle; 20 U) the mixture was allowed to
stand at ambient temperature with intermittent analysis for conversion by
TLC. At about 90% conversion, enzymes were removed by short heating
to 908C and filtration through a pad of charcoal. After removal of the
volatiles by roto-evaporation at ꢂ208C under vacuum, the residue was
Construction of the expression vector pKKfda: The PCR amplified fda
insert was digested by incubation with restriction enzymes XmaI and PstI
under standard conditions. A 20 mL reaction mixture containing the DNA
2
insert (1 mg), 6mm Tris-HCl (pH 7.5), 50mm NaCl, 6mm MgCl , 1mm DTT,
2
taken up in D O for NMR analysis. Products were identified by comparison
and XmaI (1 mL, 5 U) was incubated at 378C. After 4.5 h the enzyme was
denaturated by heating at 658C for 10 min. After cooling to ambient
temperature and addition of PstI (1 mL, 10 U), the mixture was incubated at
[
25]
with reference spectra of authentic compounds, and their proportions
1
were determined by H NMR integration of key signals.
3
78C for another 4.5 h. After denaturation by heating at 658C for 10 min,
2-Hydroxypent-4-enoic acid (Æ2a): A mixture of glyoxylic acid (2.0 g,
27 mmol), powdered Sn (6.4 g, 54 mmol), and allyl bromide (4.8 mL,
54 mmol) in 50% aqueous ethanol (90 mL) was treated with ultrasound for
6 min under nitrogen (TLC analysis showed complete conversion). The
suspension was filtered and the residue washed with ethanol. The combined
filtrates were stirred over Zn dust (5 g) for 5 h, then filtered, and
concentrated. After extraction of the residue with diethyl ether (4 Â
the restricted insert was recovered by phenol/chloroform extraction and
precipitation with ethanol (2 vol) and 3m NaOAc (1/10 vol, pH 5.2). The
DNA pellets were taken up with 20 mL Tris-HCl (10mm, pH 7.8).
The vector pKK223-3 was digested with XmaI and PstI under the same
conditions. The crude digested vector was taken up in dephosphorylation
buffer (50mm Tris-HCl, pH 9.0, 0.1mm ZnCl , 1mm MgCl , 1mm
2 2
spermidine) and treated with calf intestinal alkaline phosphatase at 378C
for 1 h. The reaction was stopped by addition of 1 mL of 0.5m EDTA
solution (final concentration 10mm) and heating at 658C for 60 min.
Plasmid DNA was recovered by phenol/chloroform extraction followed by
ethanol precipitation, and the DNA pellet was taken up in 20 mL of Tris-
HCl buffer (10 mm, pH 7.8).
5
0 mL), the concentrated extract was purified by silica chromatography
using chloroform/methanol (5:1) as eluent to furnish a colorless oil (1.6 g,
1
4
1
2
1
5%); H NMR (300 MHz, CD
3
OD): d ˆ 5.83 (dddd, 1H, 4-H), 5.11 (dd,
), 5.01 (br s, 1H, OH), 4.21 (dd, 1H, 2-H),
.56-2.35 (m, 2H, 3-H); 13C NMR (100.6 MHz, CD
OD): d ˆ 175.70 (C-1),
34.44 (C-4), 118.29 (C-5), 71.52 (C-2), 39.76 (C-3). Anal: C (C, H).
c t
H, 5-H ), 5.06 (dd, 1H, 5-H
3
5
8 3
H O
Ethyl 2-hydroxypent-4-enoate (Æ2b): A mixture of ethyl glyoxylate (2.7 g,
The restricted insert (3 mL) and restricted and dephosphorylated vector
2
5
4
6.5 mmol), powdered Sn (6.3 g, 53 mmol), and allyl bromide (4.7 mL,
3 mmol) in 70% aqueous ethanol (75 mL) was treated with ultrasound for
min under nitrogen (TLC analysis showed complete conversion). The
(
1 mL) were incubated overnight at 208C with T4 DNA ligase (0.5 U) in the
appropriate buffer (66mm Tris-HCl, 5mm MgCl , 1mm dithioerythritol,
mm ATP, pH 7.5). After ligation, competent cells of E. coli JM105 strain
2
1
3
suspension was diluted with CHCl (100 mL), neutralized by careful
were transformed with the new recombinant plasmid, denominated
addition of NEt , and filtered. The organic layer was dried over MgSO ,
3
4
pKKfda, and plated onto LB agar plates supplemented with ampicillin
concentrated, and the residue was purified by distillation to yield a
�
1
(
200 mgL ). Out of six colonies randomly selected, four carried the
1
colorless liquid (2.4 g, 64%). B.p. 46 ± 508C/20 mbar; H NMR (300 MHz,
desired insert as determined by plasmid isolation and restriction analysis, as
well as by enzymatic assay for FruA activity[ after isopropyl-b-d-
thiogalactopyranoside (IPTG) induction.
CDCl
3
): d ˆ 5.82 (dddd, 1H, 4-H), 5.15 (dd, 1H, 5-H
.19 (m, 3H, 2-H, CH ), 3.7 (br s, 1H, OH), 2.57 (m, 1H, 3-H
-H ), 1.29 (t, 3H, CH
c
), 5.12 (dd, 5-H
t
), 4.31-
26]
4
3
2
a
), 2.44 (m, 1H,
13
b
3
); C NMR (100.6 MHz, CDCl
3
): d ˆ 174.52 (C-1),
Expression and purification of FruAsca: A positive clone was grown
132.68 (C-4), 118.57 (C-5), 70.12 (C-2), 61.69 (CH ), 38.72 (C-3), 14.24
2
aerobically in 200 mL of LB medium supplemented by ampicillin
(CH ). Anal C H O (C, H).
3
7
12
3
�
1
(
200 mgL ) at 378C in a shake flask at 300 rpm. When the turbidity
3
-Deoxy-d-xylo-hept-5-ulopyranosonic acid 7-phosphate (6) and 3-deoxy-
reached an OD578 of 1.0, this was used as a starter culture to inoculate a 10-
L-fermentor (B.Braun Biostat B) charged with 9.8 L of the same medium.
Cells were grown aerobically with stirring at 378C to an OD578 of about 0.7
when IPTG was added to a concentration of 0.5mm. After a further 5 ± 6 h,
cells were harvested by continuous centrifugation (Heraeus Contifuge
d-lyxo-5-heptulopyranosono-1,6-lactone 7-phosphate (7): A solution of
alkene 2a (700 mg, 6.0 mmol) in dry methanol (20 mL) was cooled to
�
788C and purged with a stream of ozone until the blue color persisted
10 min). Dimethyl sulfide (2.5 mL) was added, and the mixture was
(
allowed to warm to room temperature with stirring. When a peroxide test
proved negative (2 h), water (1 mL) was added, and the solution was
concentrated under vacuum to a residual volume of about 1 mL. Water was
added to a final concentration of 100mm. Fructose 1,6-bisphosphate (Na
salt; 110 mg, 0.3 mmol) was added to an aliquot of the aldehyde solution
2
8RS; 15000 rpm, 48C) to give a wet weight of 35 g. The pellet was
�
1
resuspended in Tris-HCl buffer (4 mLg cells) and 5000 U of benzonase
was added. Cells were disrupted by glass bead milling (IMA cell
desintegrator C), and the extracts were clarified by filtration followed by
centrifugation at 24000 g for 15 min at 48C. Total crude FruA activity,
3
(10 mL, 2.0 mmol), and the pH was adjusted to 6.8. After addition of FruA
(from S. carnosus or rabbit muscle; 10 U) and triose phosphate isomerase
(20 U), the mixture was incubated at 258C with analysis for conversion by
determined spectrophotometrically,[ amounted to 30000 U (0.86 Umg
cells; 1 U is defined to catalyze the cleavage of 1 mmol of FBP per minute at
58C). Protein purification was performed by ion exchange chromatog-
26]
� 1
2
TLC. At about 90% conversion, the solution was worked up as usual to
raphy on DEAE-Sepharose CL-6B, applying an NaCl gradient for protein
elution (100 ± 350 mm). FruA activity eluted at about 150mm, and active
fractions were pooled and concentrated by ultrafiltration (Amicon YM10
membrane) followed by ammonium sulfate (80%) precipitation (total
recovered activity 26000 U, 87%).
determine the crude product ratio by NMR analysis to be 6:7 ˆ 85:15 and
9
0:10 (FruArab and FruAsca, respectively). A similar reaction sequence was
set up but using the ester 2b instead to give a product ratio of 6:7 ˆ 60:40.
1
Compound 6: H NMR (300 MHz, D
2
O): d ˆ 4.36 (dd, 1H, 2-H), 4.28 ± 3.70
(
m, 3H, 4-, 7-H) 3.51 (d, 1H, 5-H), 2.33 (ddd, 1H, 3-Heq), 1.58 (ddd, 1H,
Protein analysis: Protein concentration was determined by the Bradford
3-Hax), J2,3ax ˆ 12.4, J2,3eq ˆ 2.0, J3ax,3eq ˆ J3ax,4 ˆ 12.0, J3eq,4 ˆ 5.0, J4,5 ˆ 9.4.
method[ with assay reagents supplied by Bio-Rad, using bovine serum
45]
Compound 7: H NMR (300 MHz, D
1
O): d ˆ 4.57 (dd, 1H, 2-H), 4.28 ± 3.70
2
Chem. Eur. J. 1999, 5, No. 6
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1889 $ 17.50+.50/0
1889

W.-D. Fessner et al.
FULL PAPER
(
3
m, 3H, 4-, 7-H) 3.58 (d, 1H, 5-H), 2.43 (ddd, 1H, 3-Heq), 1.99 (ddd, 1H,
[10] N. S. Blom, S. Tetreault, R. Coulombe, J. Sygusch, Nat. Struct. Biol.
1996, 3, 856 ± 862.
[11] S. J. Cooper, G. A. Leonard, S. M. McSweeney, A. W. Thompson, J. H.
Naismith, S. Qamar, A. Plater, A. Berry, W. N. Hunter, Structure 1996,
4, 1303 ± 1315.
-Hax); J2,3ax ˆ 10.0, J2,3eq ˆ 5.0, J3ax,3eq ˆ 13.0, J3eq,4 ˆ 1.5, J4,5 ˆ 9.4.
1
2
,3,4-Trideoxy-2,4-di-C-(hydroxymethyl)-d-talo-octos-7-ulo-1,5:7,4 -dipyr-
anose (13b) and 2,3,4-trideoxy-2,4-di-C-(hydroxymethyl)-d-gulo-octos-7-
1
[40]
ulo-1,5:7,4 -dipyranose (14b): A solution of diol 8 (400 mg, 3.1 mmol) in
MeOH (20 mL) was cooled to � 788C and purged with a stream of ozone
[
12] T. Gefflaut, C. Blonski, J. Perie, M. Willson, Prog. Biophys. Mol. Biol.
995, 63, 301 ± 340.
1
2
until a blue color persisted. Me S (1 mL) was added, and the mixture was
[13] W.-D. Fessner, G. Sinerius, Angew. Chem. 1994, 106, 217 ± 220; Angew.
Chem. Int. Ed. Engl. 1994, 33, 209 ± 212.
stirred at � 788C for 1 h. Stirring was continued at room temperature until
a peroxide test proved negative (ca. 2 h). The solvent was evaporated and
the crude aldehyde 9 was taken up in water (15 mL). To this solution was
added DHAP (4.5 mmol), and the pH was adjusted to 6.9 with 1m NaOH.
After addition of FruA from S. carnosus (200 U), the reaction was
monitored by enzymatic assay for DHAP consumption and by TLC
analysis. After complete conversion (12 h), the pH was adjusted to 8.0, and
alkaline phosphatase (100 U) was added. TLC indicated the reaction to be
complete after two days. After desalting (Dowex AG50W-X8, H ; AG1-
X8, HCO
furnish a mixture of 13b/14b in a ratio of 1:3 (200 mg, 26%). Analytical
samples were obtained by careful silica gel chromatography (CHCl
[14] R. Duncan, D. G. Drueckhammer, J. Org. Chem. 1996, 61, 438 ± 439.
[15] H.-L. Arth, W.-D. Fessner, Carbohydr. Res. 1998, 305, 313 ± 321.
[16] B. Bossow, W. Berke, C. Wandrey, BioEngineering 1992, 8, 12 ± 14.
[17] 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.
‡
[18] F. Götz, S. Fischer, K.-H. Schleifer, Eur. J. Biochem. 1980, 108, 295 ±
�
301.
3
), products were isolated by filtration through a pad of silica to
[19] H. P. Brockamp, M. R. Kula, Appl. Microbiol. Biotechnol. 1990, 34,
2
87 ± 291.
3
/
1
[20] C. Witke, F. Götz, J. Bacteriol. 1993, 175, 7495 ± 7499.
[21] H. P. Brockamp, M. R. Kula, Tetrahedron Lett. 1990, 31, 7123 ± 7126.
MeOH 5:1). Compound 13b: H NMR (500 MHz, D
2
O): d ˆ 1.19 (q, 1H,
2
J ˆ 12.2, H-3ax), 1.67 (m, 1H, H-2), 1.74 (m, 1H, H-4), 1.80 (dt, 1H, J ˆ
[22] Boehringer Mannheim, Biochemicals Catalog, 1998.
1
J
2.8, J2,3 ˆ J3,4 ˆ 4.0, H-3eq), 3.53 (t, 1H, J4,5 ˆ J5,6 ˆ 10.0, H-5), 3.69 (d, 1H,
2
[23] J. Brosius, A. Ullrich, M. A. Raker, A. Gray, T. J. Dull, R. R. Gutell,
H. F. Noller, Plasmid 1981, 6, 112 ± 118.
[24] A. Ozaki, E. J. Toone, C. H. von der Osten, A. J. Sinskey, G. M.
Whitesides, J. Am. Chem. Soc. 1990, 112, 4970 ± 4971.
[25] 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.
[26] A. U. Bergmeyer, Methods of Enzymatic Analysis, Vol. VI, 3rd ed.,
VCH, Weinheim, 1984, pp. 342 ± 350.
[27] S. J. Gamblin, G. J. Davies, J. M. Grimes, R. M. Jackson, J. A. Little-
child, H. C. Watson, J. Mol. Biol. 1991, 219, 573 ± 576.
[28] G. Hester, O. Brenner-Holzach, F. A. Rossi, M. Struck-Donatz, K. H.
Winterhalter, J. D. G. Smit, K. Piontek, FEBS Lett. 1991, 292, 237 ±
5,6 ˆ 10.0, H-6), 3.49 and 3.71 (AB, 2H
,
J ˆ 11.7, H-8), 3.57 ± 3.71 (m, 4H,
1
1
13
H-2 and H-4 ), 4.71 (d, 1H, J1,2 ˆ 8.9, H-1); C NMR (125 MHz, D
2
O): d ˆ
1
1
2
8.75 (C-3), 41.14 (C-4), 46,07 (C-2), 64.69 and 65.47 (C-2 and C-4 ), 71.98
(
C-5), 66.38 (C-8), 79.74 (C-6), 99.92 (C-1), 100.97 (C-7); MS m/z (SIMS,
‡
‡
FAB, positve-ion): 273 (5) [M‡Na] , 263 (20), 257 (49) [M‡Li] , 167 (50),
1
1
3
61 (100). Compound 14b: H NMR (500 MHz, D
2
O): d ˆ 1.65 ± 1.85 (m,
2
H, H-2,-3), 2.18 (m, 1H, H-4), 3.51 and 3.70 (AB, 2H
,
J ˆ 11.7, H-8), 3.61
2
3
1
2
and 3.72 (ABX, 2H, J ˆ 11.7, J ˆ 3.4 and 5.7, H-2 ), 4.10 (dd, 1H, J ˆ 12.0,
1
1
J
4,4'eq ˆ 2.9, H-4 eq), 3.52 (m, 1H, H-4 ax), 4.19 (dd, 1H, J5,6 ˆ 10.5, J4,5 ˆ 6.0,
13
H-5), 4.18 (d, 1H, J5,6 ˆ 10.5, H-6), 4.95 (d, 1H, J1,2 ˆ 8.9, H-1); C NMR
1
(
125 MHz, D
2
O): d ˆ 27.65 (C-3), 38.03 (C-4), 46,52 (C-2), 64.86 (C-2 ),
1
6
5.15 (C-4 ), 65.49 (C-5), 66.64 (C-8), 77.44 (C-6), 93.88 (C-1), 101.08 (C-7);
‡
‡
MS m/z (SIMS, FAB, positive-ion): 273 (54) [M‡Na] , 257 (100) [M‡Li] .
2
anose (ent-13b) and 2,3,4-trideoxy-2,4-di-C-(hydroxymethyl)-l-gulo-oc-
tos-7-ulo-1,5:7,4 -dipyranose (ent-14b): From a similar reaction of diol 8
2
42.
1
,3,4-Trideoxy-2,4-di-C-(hydroxymethyl)-l-talo-octos-7-ulo-1,5:7,4 -dipyr-
[29] D. R. Tolan, A. B. Amsden, S. D. Putney, M. S. Urdea, E. E. Penhoet,
J. Biol. Chem. 1984, 259, 1127 ± 1131.
1
[
[
[
[
[
30] D. Fischer, D. Eisenberg, Protein Sci. 1996, 5, 947 ± 955.
31] Z. Q. Zhang, D. L. Smith, Protein Sci. 1996, 5, 1282 ± 1289.
32] W. J. Lees, G. M. Whitesides, J. Org. Chem. 1993, 58, 1887 ± 1894.
33] C. J. Li, Tetrahedron 1996, 52, 5643 ± 5668.
(
430 mg, 3.40 mmol) and DHAP (3.50 mmol) but using RhuA (200 U)
instead, a mixture of ent-13b/ent-14b was obtained in the ratio of 1:3
272 mg, 32%).
(
34] N. J. Turner, G. M. Whitesides, J. Am. Chem. Soc. 1989, 111, 624 ±
6
27.
Acknowledgments
[
[
[
35] W.-D. Fessner, G. Sinerius, Bioorg. Med. Chem. 1994, 2, 639 ± 645.
36] F. Effenberger, A. Straub, Tetrahedron Lett. 1987, 28, 1641 ± 1644.
37] I. Chibata, Immobilized Enzymes. Research and Development, Wiley,
New York, 1978.
38] M. D. Trevan, Immobilized Enzymes. An Introduction and Applica-
tions in Biotechnology, Wiley, New York, 1980.
39] O. Eyrisch, W.-D. Fessner, Angew. Chem. 1995, 107, 1738 ± 1740;
Angew. Chem. Int. Ed. Engl. 1995, 34, 1639 ± 1641.
40] M. Petersen, M. T. Zannetti, W.-D. Fessner, Top. Curr. Chem. 1997,
We thank Prof. F. Götz for kindly providing us with plasmid pBluescriptII-
fda10. We are also indebted to Prof. M.-R. Kula for a reference sample of
FruAsca. This work has been supported by the Bundesminister für Bildung
und Forschung (grant 0310745), the Deutsche Forschungsgemeinschaft
grant SFB380-B25), and the Fonds der Chemischen Industrie. M.T.Z. is
grateful to the Alexander-von-Humboldt Foundation for a postdoctoral
fellowship.
[
[
[
(
1
86, 87 ± 117.
[
[
41] C. A. Grob, H. R. Pfaendler, Helv. Chim. Acta 1970, 2156 ± 2159.
42] J. Hajdu, M. Solti, P. Friedrich, Acta Biochim. Biophys. Acad. Sci.
Hung. 1975, 10, 7 ± 16.
43] J. Sambrook, E. F. Fritsch, T. Maniatis, Molecular cloning: a laboratory
manual, Cold Spring Harbor Laboratory Press, 1989.
[
[
1] B. L. Horecker, O. Tsolas, C. Y. Lai in The Enzymes, Vol. VII, 3rd ed.
Ed.: P. D. Boyer), Academic Press, New York, 1972, pp. 213 ± 258.
2] W.-D. Fessner in Microbial Reagents in Organic Synthesis (Ed.: S.
Servi), Kluwer Academic, Dordrecht, 1992, pp. 43 ± 55.
(
[
[
[
[
3] W.-D. Fessner, C. Walter, Top. Curr. Chem. 1997, 184, 97 ± 194.
4] W.-D. Fessner, Curr. Opin. Chem. Biol. 1998, 2, 85 ± 97.
5] S. Takayama, G. J. McGarvey, C.-H. Wong, Ann. Rev. Microbiol. 1997,
[
44] H. G. Gassen, G. E. Sachse, A. Schulte, PCR. Grundlagen und
Anwendungen der Polymerase-Kettenreaktion, Fischer, Stuttgart,
1
994.
51, 285 ± 310.
[
[
[
45] M. M. Bradford, Anal. Biochem. 1976, 72, 248 ± 254.
46] U. K. Laemmli, Nature 1970, 227, 680 ± 685.
47] J. Kyte, R. F. Doolittle, J. Mol. Biol. 1982, 157, 105 ± 132.
[
[
[
6] W. J. Rutter, Fed. Proc. 1964, 23, 1248 ± 1257.
7] N. Blom, J. Sygusch, Nat. Struct. Biol. 1997, 4, 36 ± 39.
8] J. Sygusch, D. Beaudry, M. Allaire, Proc. Natl. Acad. Sci. USA 1987,
84, 7846 ± 7850.
[
9] 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.
Received: June 15, 1998
Revised version: March 2, 1999 [F 1207]
1
890
ꢀ WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
0947-6539/99/0506-1890 $ 17.50+.50/0
Chem. Eur. J. 1999, 5, No. 6