Asymmetric assembly of aldose carbohydrates from formaldehyde and glycolaldehyde by tandem biocatalytic aldol reactions

Article abstract of DOI:10.1038/nchem.2321

The preparation of multifunctional chiral molecules can be greatly simplified by adopting a route via the sequential catalytic assembly of achiral building blocks. The catalytic aldol assembly of prebiotic compounds into stereodefined pentoses and hexoses is an as yet unmet challenge. Such a process would be of remarkable synthetic utility and highly significant with regard to the origin of life. Pursuing an expedient enzymatic approach, here we use engineered D-fructose-6-phosphate aldolase from Escherichia coli to prepare a series of three- to six-carbon aldoses by sequential one-pot additions of glycolaldehyde. Notably, the pertinent selection of the aldolase variant provides control of the sugar size. The stereochemical outcome of the addition was also altered to allow the synthesis of L-glucose and related derivatives. Such engineered biocatalysts may offer new routes for the straightforward synthesis of natural molecules and their analogues that circumvent the intricate enzymatic pathways forged by evolution.

Full text of DOI:10.1038/nchem.2321

ARTICLES
PUBLISHED ONLINE: 10 AUGUST 2015 | DOI: 10.1038/NCHEM.2321
Asymmetric assembly of aldose carbohydrates
from formaldehyde and glycolaldehyde by tandem
biocatalytic aldol reactions
Anna Szekrenyi1†, Xavier Garrabou1†, Teodor Parella2, Jesús Joglar1, Jordi Bujons1 and Pere Clapés1
*
The preparation of multifunctional chiral molecules can be greatly simplified by adopting a route via the sequential
catalytic assembly of achiral building blocks. The catalytic aldol assembly of prebiotic compounds into stereodefined
pentoses and hexoses is an as yet unmet challenge. Such a process would be of remarkable synthetic utility and highly
significant with regard to the origin of life. Pursuing an expedient enzymatic approach, here we use engineered -fructose-
D
6-phosphate aldolase from Escherichia coli to prepare a series of three- to six-carbon aldoses by sequential one-pot
additions of glycolaldehyde. Notably, the pertinent selection of the aldolase variant provides control of the sugar size. The
stereochemical outcome of the addition was also altered to allow the synthesis of -glucose and related derivatives. Such
L
engineered biocatalysts may offer new routes for the straightforward synthesis of natural molecules and their analogues
that circumvent the intricate enzymatic pathways forged by evolution.
arbohydrates and glycoconjugates play crucial roles in various
Suitably engineered aldolases are promising candidates in the
intercellular and intracellular functions such as cell–cell rec- search for a route towards the direct catalytic assembly of formal-
C
ognition, cell adhesion and cell signalling processes involved dehyde and glycolaldehyde. Aldolases are gaining recognition
in bacterial and viral infection, cancer metastasis and inflammatory among chemists as highly efficient asymmetric C–C bond-
reactions1–6. In addition, carbohydrates contribute to the creation of forming catalysts that act on minimally protected substrates in
structural diversity when conjugated with other biomolecules such environmentally benign aqueous media17,18. Adequately engineered
as lipids, proteins and metabolites7,8. Hence, there is growing inter- aldolases may catalyse successive C1+C2+C2 or C2+C2+C2 connec-
est in the preparation of functionally and/or stereochemically tions, thereby affording control over the size and configuration of
diverse carbohydrates and analogues that could be used as biological the sugar. In an early realization of this approach, Wong and co-
probes and in the study of their biological roles. This has led to the workers described a sequential double addition of acetaldehyde to
emergence of a vast array of assays that aim to establish their 2-chloroacetaldehyde using 2-deoxy-D-ribose 5-phosphate aldolase
therapeutic potential.
(DERA), an extremely useful and beneficial reaction in the prep-
Carbohydrate synthesis is generally approached via the modifi- aration of an atorvastatin precursor19,20. However, the installation
cation of suitable chiral starting materials. However, this requires of consecutive hydroxyl functions needs a glycolaldehyde donor,
cumbersome successive protection and deprotection steps9. which is not how aldolases function in nature. As an alternative, bio-
Alternatively, straightforward and efficient bottom-up strategies mimetic approaches based on dihydroxyacetone phosphate
for the de novo construction of polyoxygenated scaffolds may be (DHAP)-dependent aldolases have been devised to provide Cn+C3
devised through the consecutive connection of small building connections. These require the assistance of a number of enzymes
blocks, by C–C bond-forming reactions with the concomitant to generate DHAP from C3 achiral starting materials and further
installation of hydroxylated chiral positions3,10. The catalytic asym- transform the phosphorylated ketose obtained into an aldose by
metric aldol addition reaction has proved to be an extremely useful appropriate isomerases (Fig. 1b)21,22
carboligation procedure for the generation of polyoxygenated mol- An excellent candidate for our goal of synthesizing carbohydrates
.
ecules10–12. When aiming to recreate the prebiotic synthesis of from simple unprotected substrates is the D-fructose-6-phosphate
sugars, simple amino acids and inorganic catalysts were found to aldolase from Escherichia coli (FSA), a versatile class I aldolase of
assemble glycolaldehyde (1a) and formaldehyde (1b) into diverse unknown physiological function. FSA exhibits an unmatched
carbohydrates, thereby offering partial control over the sugar size, breadth of nucleophile selectivity, accepting glycolaldehyde (1a) as
configuration and enantiomeric ratio13,14. Pursuing a synthetically well as a number of hydroxylated ketones23. Wild-type FSA catalyses
useful bottom-up approach, Northrup and MacMillan synthesized stereocontrolled homo- and cross-aldol additions of glycolaldehyde,
partially protected hexoses by the sequential aldol addition of affording small sugars (for example, L-glyceraldehyde and
three synthetic glycolaldehyde equivalents, using direct organocata- D-threose) and functionally diverse analogues24. Tight control of a
lytic aldolization and a subsequent metal-catalysed directed single glycolaldehyde addition was invariably observed, hampering
Mukaiyama aldol reaction (Fig. 1a)15,16. Despite these pioneering the preparation of higher-carbon-content aldoses. The substrate
contributions, the expedient synthesis of carbohydrates from specificity of FSA has been tailored in our laboratory by a
simple unprotected substrates remains an as yet unmet challenge.
structure-guided programme of active-site engineering to expand
1Biotransformation and Bioactive Molecules Group, Instituto de Química Avanzada de Cataluña, IQAC-CSIC, Jordi Girona 18-26, 08034 Barcelona, Spain.
2Depatment Química. Servei de Ressonància Magnètica Nuclear, Universitat Autònoma de Barcelona, Bellaterra, Spain. †Present addresses: Institut für
Organische Chemie und Biochemie, Technische Universität Darmstadt, Alarich-Weiss-Strasse 4, 64287 Darmstadt, Germany (A.S.); Laboratory of Organic
Chemistry, HCI F322, ETH Hönggerberg, 8093 Zurich, Switzerland (X.G.). e-mail: pere.clapes@iqac.csic.es
*
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
1
© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2321
ARTICLES
D- or
L-proline
a
OH
*
O
b
O
*
(i) Mukaiyama reaction
(ii) Deprotection
(i) Phosphatase
(ii) Isomerase
O
*
OH
OH
DHAP
aldolase
HO
O
O
HO
OH
*
O
X
*
*
PGO PGO
PGO
+
*
*
*
*
+
HO
2-
2-
OH
OH
OPO3
OSiMe3
OH
OH
OPO3
OH
(X = OH, R)
OH
Kinase/
oxidase
Oxidase
c
HO
OH
Engineered
FSA variant(s)
OH
O
O
O
+
+
X
OH
1a
OH
1a
(X = OH, R)
Figure 1 | Bottom-up synthesis of hexoses from achiral starting materials via chemical or biocatalytic routes. a, Combination of organocatalysis and a
metal-assisted Mukaiyama aldol addition. b, Biomimetic procedure using a combination of kinase, phosphatase, oxidoreductase, aldolase and isomerase
activities. c, Novel approach based on engineered variants of D-fructose-6-phosphate aldolase from E. coli (FSA) as catalysts for the stereoselective
trimerization of glycolaldehyde.
its scope of application25,26. Mutations at positions L107 and in situ retro-aldolization of 2a. This fact, in addition to the low con-
particularly A129 were found to be crucial for its donor activity centration of acyclic 2a present in solution, made the trimerization
and selectivity, whereas the A165G mutation enhances the reactivity of 1a slower than its dimerization. In contrast, FSA variants that
of poorly reactive aldehydes25–27. These studies have provided a dramatically accelerated glycolaldehyde dimerization26, such as
toolbox of FSA variants with different synthetic capabilities as a FSAA129G, did not significantly catalyse the sugar extension to
function of the acceptor and donor substrates. In the present D-idose (Supplementary Table 4). In this manner, the pertinent
work, we capitalize on knowledge gained regarding the activity choice of FSA variants allows for the preparation of a complete,
and selectivity of FSA along the route towards the synthesis of stereochemically consistent series of three- to six-carbon aldose
aldose sugars and analogues, via stereoselective glycolaldehyde sugars, namely L-glyceraldehyde (2b), D-threose (2a), L-xylose
trimerization as well as its double addition to other acceptor (3b) and D-idose (3a), from achiral precursors (Table 1, entries 1,
aldehydes (Fig. 1c). Additional engineering efforts have been 2, 4 and 7).
devoted to expanding the substrate scope and altering the
Acetaldehyde (1d) and glycolaldehyde analogues (that is, 1e–j,
stereochemical course of the additions. In this way, the simple Table 1) were assayed as acceptors for the double addition of glycol-
carboligations that possibly enabled sugar synthesis in the prebiotic aldehyde (that is, with the C2+C2+C2 connection) to generate
world could be governed with high selectivity by a set of 6-deoxy-D-idose and 6-substituted D-idose derivatives (Table 1,
tailored enzymes.
entries 11–22). The selectivity of FSA variants towards glycolalde-
hyde as a donor ensures that the other aldehyde component will
invariably act as an acceptor substrate. Poor results were obtained
Results
A set of FSA variants with enhanced activity and selectivity towards with the available FSA variants, prompting us to further engineer
glycolaldehyde dimerization was initially assayed in the successive their acceptor binding site. The proximity of residue S166 to the
C1+C2+C2 connection of 2 equiv. of glycolaldehyde (1a) to acceptor site (Fig. 3) led us to consider the substitution of S166 by
1 equiv. of formaldehyde (1b). The formation of a double addition Gly to alleviate the steric hindrance at that position. Variants
product, L-xylose (3b), was observed with different variants (Table 1, FSAA129T/S166G and FSAA129T/A165G/S166G were found to tolerate
entries 7–10), and the most advantageous, FSAA129T/A165G, was these aldehydes satisfactorily, producing the synthesis of 6-deoxy-
selected for further studies. Under optimized experimental con- (3d), 6-O-methyl- (3e), 6-O-benzyl- (3f), 6-(benzylthio)- (3g),
ditions, FSAA129T/A165G catalysed the formation of L-xylose (3b) 6-O-phenyl (3h), 6-deoxy-6-azido (3i) and 6-deoxy-6-chloro- (3j)
with full stereoselectivity in 98% conversion (62% isolated yield) D-idose (Table 1, entries 11–20).
(Table 1, entry 7). Slow addition of 1a facilitated the sequential for-
mation of L-glyceraldehyde (2b) and L-xylose (3b) and prevented double addition of glycolaldehyde (C3+C2+C2 connection) to gen-
the dimerization of 1a (Fig. 2a,c). Next, the panel of enzymes was erate high-carbon-content aldoses. Using FSAA129T/S166G
Other simple C3 aldehydes (1c and 1k–l) were tested for the
,
assayed in the C2+C2+C2 connection of 1a. A number of FSA 6-deoxy-6-methyl-D-idose (3k) and 6-deoxy-6-(phenylmethyl)-
mutants (Table 1, entries 4–6) catalysed the stereoselective trimeri- D-idose (3l) were synthesized (Table 1, entries 21 and 22).
zation of 1a to furnish the monosaccharide D-idose (3a), achieving 3-Hydroxypropanal (1c) only produced a single glycolaldehyde
85% conversion (48% isolated yield) with FSAA129T/A165G (Table 1, addition, most probably because of the high stability of the
entry 4). Molecular models of intermediate reaction steps catalysed 4-deoxy-β-D-threo-pentopyranose hemiacetal (2c) formed after
by this double mutant show that it is well suited to bind D-threose at the first 1a addition (Table 1, entry 3).
the acceptor site and to promote its reaction with the third unit of
Minimal active-site engineering was also envisioned to gain
glycolaldehyde (Fig. 3). Thus, this is the first bottom-up, one-pot access to alternative hexose stereoisomers. Notably, inversion of
synthesis of a hexose with a single enzyme and full control of the the stereochemistry at C3 (that defined during the approach of
four stereogenic centres. In this synthesis, glycolaldehyde (1a) is the acceptor aldehyde to the FSA–donor intermediate) in the first
rapidly dimerized into D-threose (2a) (Fig. 2b,d). Then, a sub- addition of 1a would afford L-glucose, a non-caloric sweetener of
sequent second addition of 1a to the acyclic form of 2a, in equili- complex preparation by alternative procedures32. Such C3 inversion
brium with the stable hemiketal furanoid form28, produces 3a. appeared feasible, because the stereochemical approach of the alde-
The 1a needed for the second addition is obtained solely from the hyde to the aldolase–donor complex is often compromised when
2
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2321
ARTICLES
Table 1 | Single and double additions of glycolaldehyde (1a) catalysed by FSA variants.
O
O
a: R1 = CH2OH
b: R1 = H
e: R1 = CH2OCH3
f: R1 = CH2OBn
g: R1 = CH2SBn
h: R1 = CH2OPh
i: R1 = CH2N3
j: R1 = CH2Cl
: R1 = CH2CH3
HO
HO
R1
O
3
OH
OH
OH
O
O
5
4
1
2
1a
1a
c: R1 = CH2CH2OH
d: R1 = CH3
k
R1
R1
HO
l: R1 = CH2CH2Ph
OH
OH
1
2
3
O
OH
n
HO
OH
n = 1, 2a; n = 2, 2c
Entry
1
2
3
4
5
6
7
Product
FSA variant*
Conversion (%) (Reaction time, h)
Isolated yield¶ (%)
2a
2b
2c
3a
3a
3a
3b
3b
3b
3b
3d
3e
3e
3f
A129G†
98 (24)§
99 (24)§
90 (5)||
67
Wild-type†
–
#
L107Y/A129G†,‡
A129T/A165G‡
A129T/S166G
43
48
69 (24), 85 (96)||
53 (24)
#
–
#
A129T/A165G/S166G
A129T/A165G‡
A129T‡
50 (24)
–
98 (24)||
80 (24)
62
#
8
9
–
A129V‡
95 (24)
–
#
#
10
11
12
13
14
15
16
17
18
19
20
21
22
L107Y/A129G‡
A129T/A165G/S166G
A129T/A165G‡
A129T/A165G/S166G
A129T/A165G/S166G
A129T/A165G†
A129T/S166G
A129T†
90 (24)
–
70 (72)||
69 (24)
7
#
–
79 (24)||
90 (24), 94 (48)||
80 (24)
39
46
#
3f
–
3g
3g
3h
3i
3j
3k
3l
77 (24), 80 (52)||
81 (24)
39
#
–
A129T/S166G
A129T/S166G
A129T/S166G
A129T/S166G
A129T/S166G
65 (52)||
43
62
11
31
47
80 (27)||
67 (52)||
51 (48)||
58 (48)||
*Variants used that provided the highest product formation; †one single 1a addition; ‡FSA variant constructed as in ref. 26; §data from refs 24 and 26; ∥under scaled-up conditions (Supplementary Tables 4–26);
¶scaled-up reactions were performed with the respective most productive FSA variant. In entries 4 to 22 intermediate products 2 were not isolated. Purification procedures were not optimized. The isolated
carbohydrates were unequivocally characterized by NMR. Unbiased kinetic stereocontrol for si face addition of the FSA variant-bound nucleophile to the corresponding si face of the acceptor, yielding
2S,3R,4R,5R (for 3a, 3d–e, 3f, 3h–i and 3k–l), 2S,3R,4R,5S (for 3g and 3j) and 2S,3R,4S (for 3b) configured carbohydrates. No other diastereoisomers were found within the limits of NMR detection
(>96:4). Negligible NMR signals, which might be attributed to any other potential stereoisomer, could not be assigned unequivocally. To ensure that no potential stereoisomers were eliminated during
purification, all spots visible in the thin-layer chromatography analysis were separated by chromatography and carefully analysed by NMR. No other stereoisomers were detected. For additional information
concerning FSA stereochemistry see refs 24–26 and 28–30. For NMR data and structural characterization see Supplementary pages 16–91; #not isolated.
dealing with non-natural substrates26,33 and it has been successfully
Pertinent selection of FSA variants allowed the synthesis of
inverted by engineering several aldolases34. In a preliminary explora- L-glucose derivatives. Tandem addition of 1a to 1e catalysed by
tion, we observed that dimerization of 1a with FSAA165G provided FSAA129T/A165G and FSAA129T produced 6-O-methyl-L-glucose
an ∼4:1 mixture of D-threose (2a) and L-erythrose (4a), as inferred (5e), and 6-deoxy-6-chloro-L-glucose (5j) was obtained from 1a
from chromatographic and NMR analysis (Supplementary Table 28 and 1j using FSAA129T/A165G in both additions (Table 2,
and Supplementary Fig. 28b,c). FSAA165G was then subjected to Supplementary Tables 30 and 32 and Supplementary Figs 30
saturation mutagenesis at position S166, producing the variant and 32). In both cases, the first glycolaldehyde addition yielded
FSAA165G/S166P with a highly enhanced preference for L-erythrose syn/anti mixtures of products, and the second addition used the
formation (Supplementary Table 28 and Supplementary Fig. 28a–c). anti adducts selectively (that is, 4e and 4j) to produce L-glucose
Attempts to engineer the most active FSA variants26 to form derivatives as the main products.
L-erythrose (for example, library FSAL107Y/A129G/A165G/S166X) were
unsuccessful. In a second step, a set of FSA variants was assayed Discussion
for selective aldol addition of 1a to L-erythrose (4a) Life takes advantage of the modular assembly of simple building
(Supplementary Table 28). FSAA129G was the most successful blocks as an essential strategy to create molecular complexity. In
catalyst for the formation of L-glucose (5a) as the major product the case of monosaccharides, sequential addition of their most
(81% conversion, 38% isolated yield) whereas D-idose (3a) was simple precursors (1a and 1b) through enzymatic catalysis
not detected (Table 2, entry 1). The accumulation of L-glucose (‘systems biocatalysis’) may be envisioned. This was not the path fol-
(5a) was favoured by two circumstances: (1) the higher thermo- lowed by natural evolution, which instead forged highly intricate
dynamic stability of L-glucose (that is, all-trans substituted) relative pathways from central cellular metabolites. In this Article, we
to D-idose (that is, cis/trans configured) and (2) the inability of show that the direct enzymatic assembly of small aldehydes and gly-
FSAA129G to cleave L-erythrose (4a) and to use D-threose (2a) as colaldehyde into aldoses and analogues, with control of the sugar
an acceptor. Thus, the synthesis of L-glucose (5a) was accomplished size and configuration, is feasible with engineered variants of D-fruc-
in the two-step one-pot trimerization of 1a using a tandem of tose-6-phosphate aldolase (FSA). Using syn-selective FSA variants,
engineered FSA variants.
the alternation of formaldehyde or glycolaldehyde as the initial
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
3
© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2321
ARTICLES
a
1.8
1.5
1.3
1.0
0.8
0.5
0.3
0.0
b
2.0
Glycolaldehyde(1a)
D-threose(2a)
Formaldehyde (1b)
Glycolaldehyde (1a)
L-glyceraldehyde (2b)
D-threose (2a)
D-idose (3a)
1.5
1.0
0.5
L-xylose (3b)
1a added
0.0
0
6
12
18
24
0
20
40
60
80
100
Time (h)
Time (h)
c
1.8
1.5
1.3
1.0
0.8
0.5
0.3
0.0
d
1.0
0.8
0.6
0.4
0.2
0.0
0
1
2
3
4
5
6
0
1
2
3
4
5
6
Time (h)
Time (h)
Figure 2 | Time progression curves for the enzymatic synthesis of L-xylose (3b) and D-idose (3a). a–d, Time progression curves for L-xylose (3b) (a,c) and
D-idose (3a) (b,d). In a and c, the addition of 2 equiv. of 1a to 1 equiv. 1b is presented. The first addition produced 2b followed by an ensuing addition of 1a
to produce 3b. Slow addition of 1a prevented the formation of 2a. Conditions: to a solution of 1b (1 mmol) in triethanolamine buffer (10 ml, 50 mM, pH 8),
FSAA129T/A165G (20 mg, 5.2 U) was added. The reaction was started by the fed-batch addition with a syringe pump of a 400 mM 1a solution (5.0 ml,
0.42 ml h–1, 2 mmol) for 12 h. In b and d, the enzymatic trimerization of 1a is presented. Glycolaldehyde (1a) rapidly dimerizes into 2a and then a second
addition of 1a to the acyclic form of 2a produces 3a. Conditions: 1a (1.5 mmol) was dissolved in triethanolamine buffer (50 mM, pH 8, 10 ml). The reaction
was started by adding FSAA129T/A165G (30 mg, 7.8 U). Error bars are the values of the standard error of the mean of three independent experiments under
the same reaction conditions.
acceptors produces odd and even carbon products with D and L con- chemical scenario suitable for the creation and combination of
figurations, respectively. This is in contrast with the prevalence of nucleotides37. The enzymatic reactions presented in this work could
D-carbohydrates in nature, which is a consequence of the use of show that a sufficiently ‘complex polymer’ can assemble enantiopure
D-glyceraldehyde-3-phosphate as a building block. FSA is thus an carbohydrates, including thermodynamically unfavourable configur-
unprecedented biocatalyst for carbohydrate synthesis because of ations (for example, D-idose) in aqueous conditions, from formal-
the simplicity of the substrates required and the high stereoselectiv- dehyde and glycolaldehyde. It might therefore be hypothesized that
ity displayed. This compares favourably to other aldolases involved chiral polymers exhibiting defined tertiary structures could have
in sugar metabolism, which mainly act on ketoses, catalysing Cn+C3 enabled selective carbohydrate syntheses in a prebiotic system.
connections using dihydroxyacetone phosphate.
Formose sugar synthesis is commonly regarded as the source of been achieved with reported simple β-amino acid foldamers38.
carbohydrates necessary for the establishment of a prebiotic world The engineering of FSAvariants with altered substrate preferences
Remarkably, significant acceleration of the retroaldol reaction has
based on RNA molecules as self-replicating, evolvable information and stereochemical outcomes was achieved through minimal modifi-
units35. More feasible alternatives have been suggested, such as cations of the catalytic site. Indeed, this is consistent with the fact that
direct access to activated pyrimidine ribonucleosides, which would during protein evolution all the functional sequences of a protein are
bypass the separate synthesis of the free sugar and the nucleobase connected by trajectories involving changes in only one single amino-
moieties36. However, these approaches lead to a plethora of different acid residue39. Different substrates require different FSA variants,
products with detrimental effects on RNA replication. It has been which always result from a combination of a few (that is, one or
suggested that self-replicating prebiotic polymers based on chiral two) residue variations near the catalytic site. Selected mutations at
building blocks could have preceded RNA synthesis and provided a positions A165 and S166 of the active site (Fig. 3) fostered the
4
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2321
ARTICLES
c
a
b
R134
Q59
T109
Y131
T129
S166
G165
N28
K85
DTA
T27
T26
D6
Figure 3 | Molecular models of FSA-bound intermediates during formation of D-idose (3a) catalysed by FSA A129T/A165G. a–c, Active site of
FSAA129T/A165G complexed with D-threose (acyclic form, DTA) and the enamine of glycolaldehyde (a), the corresponding imine adduct (b) and the
subsequent hemiaminal intermediate (c). Mutated protein residues and non-protein moieties are shown in orange and yellow, respectively, and the rest of
the protein is shown in grey. In a, the two reactive carbon atoms are shown in a pre-reactive disposition at a distance of 3.3 Å and the D-threose carbonyl
oxygen is forming a hydrogen bond with an essential water molecule, which is also hydrogen-bonded to residues Q59, T109 and Y131. It is postulated that
this water molecule mediates proton transfer from the phenol group of residue Y131 to the acceptor aldehyde group26, once the C–C bond is formed, to
generate the corresponding iminium cation (b). Subsequent nucleophilic attack of the water on the iminium-activated carbon atom, promoted by proton
abstraction by the Y131 phenolate, would yield the hemiaminal precursor of D-idose (c). All the structures shown were modelled and energy minimized using
a previously described protocol26. The Schrödinger Suite 2014-3 was used to carry out all calculations31.
Table 2 | Synthesis of
L-glucose and C6 derivatives using one or two FSA variants.
1st step
2nd step
O
O
R1
O
3
OH
OH
R1
O
OH
OH
HO
HO
OH
O
OH
3
O
1
O
5
4
1
2
1a
1a
+
+
2
R1
R1
R1
HO
HO
FSA
variants
FSA
variants
OH
OH
OH
OH
1
2
4
5
3
a: R1 = CH2OH
e: R1 = CH2OCH3
j: R1 = CH2Cl
Entry
1
2
3
Product
FSA variant
Conversion 5/3 (%) (Reaction time, h)
Isolated yield|| of 5 (%)
5a
5e
5j
1st step: A165G/S166P*; 2nd step: A129G†
1st step: A129G/A165G†; 2nd step: A129T†
A129T/A165G†,‡
81/ND§ (240)
41 / 2 (29)
38¶
4¶
14
60 / 2 (52)
*FSA variant in this work; †FSA variant constructed as in ref. 26; ‡one-step one-pot; §not detected; ∥purification procedures were not optimized. The isolated carbohydrates were unequivocally characterized by
NMR yielding 2S,3R,4R,5S (for 5a and 5e) and 2S,3R,4R,5R (for 5j) configured carbohydrate compounds. No other diastereoisomers were observed within the limits of NMR detection (>96:4). For NMR data and
structural characterization see Supplementary pages 92–118; ¶isolated yield including both steps. Intermediate products 2 and 4 were not isolated.
double addition of 1a and provided inversion at the C3 carbon in the preparation of these intricate molecules is achieved with optimal
first 1a addition. Remarkably, a single variant can be engineered to energetic and atomic economy, avoiding both the functionalization
produce opposite selectivity at each aldol step, as in the synthesis of and activation required in alternative chemical or biocatalytic
6-chloro-L-glucose (5j). Such active site malleability demonstrates approaches. Overall, our results illustrate the utility of engineered
the great evolutionary potential for the functional diversification of enzymes in the construction of complex molecules, harnessing
FSA, which nature has not exploited, as inferred from the reduced simple substrates and biological transformations that circumvent
sequence variability observed within the FSA family23,40. Application the complexity of the natural enzymatic pathways forged
of higher-throughput screening techniques can be expected to boost by evolution.
the fast evolution of FSA and its variants towards higher activity and
Received 7 March 2015; accepted 8 July 2015;
published online 10 August 2015
stereoselectivity, and enable unprecedented non-biological reactivity.
It is also noteworthy that diverse FSA variants from the toolbox were
selective for a single glycolaldehyde addition (Supplementary Tables
14, 16, 18 and 20). This is highly important because controlling gly-
colaldehyde addition by biocatalyst selection allows the synthesis of
the mono addition adducts as final products or as intermediate
acceptors for further additions with other aldolases with alternative
donor specificity or stereochemical outcomes.
In summary, we have developed a procedure for the preparation
of a number of aldose carbohydrates and derivatives from simple
formaldehyde, glycolaldehyde and other aldehydes by biocatalytic
tandem reactions using engineered FSA variants. Remarkably, the
References
1. Weymouth-Wilson, A. C. The role of carbohydrates in biologically active natural
products. Nat. Prod. Rep. 14, 99–110 (1997).
2. Bertozzi, C. R. & Kiessling, L. L. Chemical glycobiology. Science 291,
2357–2364 (2001).
3. Stallforth, P., Lepenies, B., Adibekian, A. & Seeberger, P. H. Carbohydrates: a
frontier in medicinal chemistry. J. Med. Chem. 52, 5561–5577 (2009).
4. Wong, C.-H. Carbohydrate-based Drug Discovery Vols 1 & 2 (Wiley, 2003).
5. Campo, V. L., Aragao-Leoneti, V., Teixeira, M. B. M. & Carvalho, I.
Carbohydrates and glycoproteins: cellular recognition and drug design. New
Dev. Med. Chem. 1, 133–151 (2010).
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
5
© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CHEMISTRY DOI: 10.1038/NCHEM.2321
ARTICLES
6. Furukawa, K. et al. Fine tuning of cell signals by glycosylation. J. Biochem. 151,
573–578 (2012).
7. Galan, M. C., Benito-Alifonso, D. & Watt, G. M. Carbohydrate chemistry in drug
discovery. Org. Biomol. Chem. 9, 3598–3610 (2011).
29. Soler, A. et al. Sequential biocatalytic aldol reactions in multistep asymmetric
synthesis: pipecolic acid, piperidine and pyrrolidine (homo)iminocyclitol
derivatives from achiral building blocks. Adv. Synth. Catal. 356,
3007–3024 (2014).
8. Grunwald, P. in Carbohydrate-Modifying Biocatalysts (ed. Grunwald, P.) 29–118
(Pan Stanford, 2012).
9. Hudlicky, T., Entwistle, D. A., Pitzer, K. K. & Thorpe, A. J. Modern methods of
monosaccharide synthesis from non-carbohydrate sources. Chem. Rev. 96,
1195–1220 (1996).
30. Rale, M., Schneider, S., Sprenger, G. A., Samland, A. K. & Fessner, W.-D.
Broadening deoxysugar glycodiversity: natural and engineered transaldolases
unlock a complementary substrate space. Chem. Eur. J. 17, 2623–2632 (2011).
31. Schrödinger Suite 2014-3 (2014).
32. Guaragna, A., D’Alonzo, D., Paolella, C., Napolitano, C. & Palumbo, G. Highly
stereoselective de novo synthesis of L-hexoses. J. Org. Chem. 75,
3558–3568 (2010).
10. Mlynarski, J. & Gut, B. Organocatalytic synthesis of carbohydrates. Chem. Soc.
Rev. 41, 587–596 (2012).
11. Markert, M. & Mahrwald, R. Total syntheses of carbohydrates: organocatalyzed
aldol additions of dihydroxyacetone. Chem. Eur. J. 14, 40–48 (2008).
12. Mahrwald, R. Modern Methods in Stereoselective Aldol Reactions
(Wiley, 2013).
13. Hein, J. E. & Blackmond, D. G. On the origin of single chirality of amino acids
and sugars in biogenesis. Acc. Chem. Res. 45, 2045–2054 (2012).
14. Ruiz-Mirazo, K., Briones, C. & De La Escosura, A. Prebiotic systems chemistry:
new perspectives for the origins of life. Chem. Rev. 114, 285–366 (2014).
15. Northrup, A. B. & MacMillan, D. W. C. Two-step synthesis of carbohydrates by
selective aldol reactions. Science 305, 1752–1755 (2004).
16. Northrup, A. B., Mangion, I. K., Hettche, F. & MacMillan, D. W. C.
Enantioselective organocatalytic direct aldol reactions of α-oxyaldehydes: step
one in a two-step synthesis of carbohydrates. Angew. Chem. Int. Ed. 43,
2152–2154 (2004).
17. Clapés, P. & Garrabou, X. Current trends in asymmetric synthesis with aldolases.
Adv. Synth. Catal. 353, 2263–2283 (2011).
33. Fessner, W. -D. et al. Enzymes in organic synthesis. Part 1. Diastereoselective,
enzymatic aldol addition with L-rhamnulose- and L-fuculose-1-phosphate
aldolases from E. coli. Angew. Chem. Int. Ed. 30, 555–558 (1991).
34. Bolt, A., Berry, A. & Nelson, A. Directed evolution of aldolases for
exploitation in synthetic organic chemistry. Arch. Biochem. Biophys. 474,
318–330 (2008).
35. Orgel, L. E. Prebiotic chemistry and the origin of the RNA world. Crit. Rev.
Biochem. Mol. Biol. 39, 99–123 (2004).
36. Powner, M. W., Gerland, B. & Sutherland, J. D. Synthesis of activated pyrimidine
ribonucleotides in prebiotically plausible conditions. Nature 459,
239–242 (2009).
37. Brewer, A. & Davis, A. P. Chiral encoding may provide a simple solution to the
origin of life. Nature Chem. 6, 569–574 (2014).
38. Müller, M. M., Windsor, M. A., Pomerantz, W. C., Gellman, S. H. & Hilvert, D.
A rationally designed aldolase foldamer. Angew. Chem. Int. Ed. 48,
922–925 (2009).
18. Clapés, P. & Joglar, J. in Modern Methods in Stereoselective Aldol Reactions
(ed. Mahrwald, R.) 475–528 (Wiley, 2013).
19. Wong, C.-H. et al. Recombinant 2-deoxyribose-5-phosphate aldolase in organic
synthesis: use of sequential two-substrate and three-substrate aldol reactions.
J. Am. Chem. Soc. 117, 3333–3339 (1995).
39. Currin, A., Swainston, N., Day, P. J. & Kell, D. B. Synthetic biology for the
directed evolution of protein biocatalysts: navigating sequence space
intelligently. Chem. Soc. Rev. 44, 1172–1239 (2015).
40. Samland, A. K. & Sprenger, G. A. Transaldolase: from biochemistry to human
disease. Int. J. Biochem. Cell Biol. 41, 1482–1494 (2009).
20. Jennewein, S. et al. Directed evolution of an industrial biocatalyst: 2-deoxy-D-
ribose 5-phosphate aldolase. Biotechnol. J. 1, 537–548 (2006).
21. Durrwachter, J. R., Drueckhammer, D. G., Nozaki, K., Sweers, H. M. &
Wong, C.-H. Enzymic aldol condensation/isomerization as a route to unusual
sugar derivatives. J. Am. Chem. Soc. 108, 7812–7818 (1986).
22. Alajarin, R., Garcia-Junceda, E. & Wong, C.-H. A short enzymic synthesis of
L-glucose from dihydroxyacetone phosphate and L-glyceraldehyde. J. Org. Chem.
60, 4294–4295 (1995).
Acknowledgements
This work was supported by Spanish MINECO grants CTQ2012-31605 and CTQ2012-
32436, the Generalitat de Catalunya (2009 SGR 00281), ERA-IB MICINN, PIM2010EEI-
00607 (EIB.10.012. MicroTechEnz-EIB, www.fkit.unizg.hr/miten) and COST Action
CM1303 Systems Biocatalysis. A.S. acknowledges the CSIC for a JAE predoctoral
contract programme.
23. Samland, A. K., Rale, M., Sprenger, G. A. & Fessner, W.-D. The transaldolase
family: new synthetic opportunities from an ancient enzyme scaffold.
ChemBioChem 12, 1454–1474 (2011).
24. Garrabou, X. et al. Asymmetric self- and cross-aldol reaction of glycolaldehyde
catalyzed by D-fructose-6-phosphate aldolase. Angew. Chem. Int. Ed. 48,
5521–5525 (2009).
25. Gutierrez, M., Parella, T., Joglar, J., Bujons, J. & Clapés, P. Structure-guided
redesign of D-fructose-6-phosphate aldolase from E. coli: remarkable activity and
selectivity towards acceptor substrates by two-point mutation. Chem. Commun.
47, 5762–5764 (2011).
Author contributions
P.C. and X.G. designed the study. A.S. and X.G. performed mutagenesis, library screening,
activity measurements and synthesis of the compounds. J.B. performed the molecular
docking experiments and designed the mutations. T.P. performed and supervised the NMR
experiments and structural assignation of compounds. J.J., J.B. and P.C. supervised the
scientific work. All authors contributed to writing the paper.
26. Szekrenyi, A. et al. Engineering the donor selectivity of D-fructose-6-phosphate Additional information
aldolase for biocatalytic asymmetric cross-aldol additions of glycolaldehyde.
Chem. Eur. J. 20, 12572–12583 (2014).
27. Castillo, J. A. et al. A mutant D-fructose-6-phosphate aldolase (Ala129Ser) with
improved affinity towards dihydroxyacetone for the synthesis of
polyhydroxylated compounds. Adv. Synth. Catal. 352, 1039–1046 (2010).
28. Concia, A. L. et al. D-Fructose-6-phosphate aldolase in organic synthesis: cascade
Supplementary information and chemical compound information are available in the
online version of the paper. Reprints and permissions information is available online at
www.nature.com/reprints. Correspondence and requests for materials should be
addressed to P.C.
chemical-enzymatic preparation of sugar-related polyhydroxylated compounds. Competing financial interests
Chem. Eur. J. 15, 3808–3816 (2009).
The authors declare no competing financial interests.
6
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
© 2015 Macmillan Publishers Limited. All rights reserved