Directed evolution of an aldolase for the parallel synthesis of sialic acid mimetics
Thomas Woodhall,
†,‡
Gavin Williams,
†,§
Alan Berry
†,§
and Adam Nelson*
,†,‡
The exquisite selectivity and efficiency of enzymes has served as an inspiration to chemists for many years.[1] These remarkable properties enable enzymes to guide the assembly of complex products from mixtures of reactants present in low concentration (nM-

M).[2] Indeed, high levels of substrate specificity, and stereo- and chemoselectivity are hallmarks of enzymatic catalysis.
The most useful catalysts to the synthetic chemist, however, are those which have broad application. Indeed, Sharpless has commented on the synthetic virtues of the asymmetric dihydroxylation in terms of its remarkable scope: “It [OsO
4
] reacts only with olefins and it reacts with all olefins (slight poetic licence here)”.[3] The range of substrates accepted by many enzymes is rather more restricted than this, limiting synthetic utility. The power of directed evolution has been brought to bear on this problem[4] and has, for example, been used to create an amine oxidase with broad substrate specificity and high enantioselectivity[5] and an aldolase which modifies the stereochemical course of C-C bond formation.[6,7]
We set the challenging objective of broadening the substrate specificity of a carbon
 carbon bond-forming enzyme, namely sialic acid aldolase ( N -acetylneuraminic acid aldolase), sufficiently for application in the parallel synthesis of sialic acid mimetics. Sialic acid aldolase catalyses the reversible aldol condensation between pyruvate and Nacetylmannosamine, 3 , to give sialic acid, 4
(Scheme 1). Although a number of hexoses and pentoses and their analogues are substrates for this enzyme, condensations involving shorter aldehydes are less promising:
L
- and
D
-erythrose and threose react at between 0.3% and 5% of the rate of N -acetyl mannosamine, and two- and threecarbon aldehydes are not substrates.[8] The substituted dihydropyran 2 is an influenza A sialidase inhibitor[9] whose activity was optimised from the first potent inhibitor of influenza sialidases,
Zanamivir, 1 .[10] We decided, therefore, to attempt to use directed evolution to generate an
1

aldolase with sufficiently broad substrate specificity to convert four-carbon aldehydes of general structure 5 into the corresponding sialic acid mimetics 6 (Scheme 1).
HO
HO
H
O CO
2
H
HO
AcHN
HN NH
2
1
NH
OH OH O
HO
OH NHAc
+
3 (open form)
HO
2
C
O pyruvate
O
O CO
2
H
N
AcHN
2
HN
NH
NH
2
O OH
HO
HO
OH
3 , ManNAc
NHAc sialic acid aldolase
HO directed evolution
HO
H
O
OH
CO
2
H
HO
AcHN
OH
4 sialic acid
R
1
R
2
N
OH O
+
HO
2
C mutant enzyme R
1
O
O
OH
CO
2
H
O X
O
N
R
2
X
5 pyruvate
6
OH
Scheme 1. Directed evolution of an aldolase for application in the parallel synthesis of sialic acid mimetics
We used a semi-rational approach to evolve a synthetically useful enzyme. Analysis of the X-ray crystal structure[11] of the Haemophilus influenzae sialic acid aldolase in complex with the inhibitor 4-oxo sialic acid revealed three residues which contacted the C
7
C
9
side chain: the residues in the corresponding E. coli protein , Asp191, Glu192 and Ser208, were selected for saturation mutagenesis (see Panel A, Figure 1). Since the aim of the investigation was to evolve an aldolase with broad specificity, we designed a screening substrate, 12 , with relatively large R
1
and R
2
(= n Pr); it was hypothesised that an active site which was able to accommodate this rather bulky substrate would also be able to accept a wide range of smaller substrates as well. Furthermore, the tolerance of a wide range of C
2
-substituted aldehydes by the wild-type enzyme[8] was expected to be preserved since this substituent (-NHAc in N -acetyl mannosamine) is solvent exposed.[11] A schematic diagram of the designed screening substrate in complex with the mutant enzyme in shown in Panel B, Figure 1.
2

A
Glu192
N
H
O
2
C
Asp191
O
H
O
O
O
H
O
H
N
OH O N
Lys165
CO
2
H
H
O NHAc
HO
Ser208
B
N
O
OH
OH
OH N
Lys165
CO
Figure 1. Schematic diagrams of the active sites of sialic acid aldolases.
Panel A: Residues which have been shown by X-ray crystallography to interact with
C
7
C
9
of 4-oxo-sialic acid.
Panel B: Schematic diagram of the active site of an evolved enzyme in which a hydrophobic pocket has been sculpted to accommodate the dipropylaminocarbonyl group of the screening substrate.
2
H
The screening substrate 12 was prepared using the synthetic sequence outlined in Scheme 2. The lactone 7 , readily available by oxidative cleavage of isoascorbic acid,[12] was opened with dipropylamine to give the

-hydroxyamide 8 . Swern oxidation of 8 , and indium-mediated coupling with ethyl

-bromomethyl acylate 13 ,[13] gave the

-hydroxyamides 9 ( anti : syn 77:33), an outcome which is consistent with Felkin

Anh-controlled[14] attack on the intermediate aldehyde. Acetonide deprotection and ester hydrolysis gave the screening substrate 12 .
O
O
H
O O
H a
(65%)
N
O
O O
OH b, c N
O
O O
OH
CO
2
Et
7 8
N
O
O O
OH
CO
2
Et
anti-9 , 45% d (81%) anti : syn 77:23
syn-9 , 13% n Pr
2
N
O
O
OH
CO
2
NH
4 f, g
(97%)
HO
12
OH
O n Pr
2
N
HO
O
OH
CO
2
Et e
(51%)
N
O
O O
OH O
CO
2
Et
Br
CO
2
Et
OH
11 10 13
Scheme 2. Synthesis of the screening substrate 12 . Reagents and conditions: a) n Pr
2
NH, MeOH; b) DMSO, (COCl)
2
, Et
3
N, CH
2
Cl
2
; c) In, 13 , 1:1 THF

H
2
O; d) O
3
, MeOH,

78 °C then Me
2
S; e) 1:1 TFA

H
2
O; f) Ba(OH)
2
, EtOH

H
2
O; g) (NH
4
)
2
SO
4
, H
2
O. DMSO = dimethylsulfoxide, THF = tetrahyrofuran, TFA = trifluoroacetic acid.
The library of enzymes produced by saturation mutagenesis of residues 191, 192 and 208 were screened for useful synthetic activity on the premise that mutant enzymes which were able to cleave the screening substrate 12 would also, of course, be able to catalyse the forward reaction as well.[6]
Libraries of proteins were screened in 96-well plates by analysing thermally treated, crude cell
3

lysates in which His-tagged mutant proteins had been overexpressed to ca.
40% of total protein content. A coupled enzyme assay was used in which the cleavage of the screening substrate to generate pyruvate was detected spectroscopically at 340 nm by lactate dehydrogenase-catalysed reduction of pyruvate with concomitant oxidation of NADH. In this way, we identified a mutant enzyme, containing the Glu192Asn mutation (E192N), for which substrate specificity
[( k cat
/K m( 12 )
)/( k cat
/K m(sialic acid)
)] had been switched 630-fold. This mutant had k cat
/K m for the screening substrate, 12 , 50-fold higher than the wild-type enzyme; indeed, the mutant catalysed the cleavage of the screening substrate 12 six times more effectively than the wild type enzyme catalysed the cleavage of sialic acid!
The synthetic utility of the evolved E192N enzyme was investigated using the crude aldehydes
18 generated by ozonolysis of the corresponding

,

-unsaturated amides 15 , prepared in parallel from the known[15]

,

-unsaturated acid 14 (see Scheme 3 and Table). With the tertiary amides
15a-g (entries 1-7; Table), the enzymatic reactions reached completion well within 3 days, and, after purification by ion exchange chromatography, the sialic acid mimetics 16 / 17a-g were obtained in
37-66% yield over the two steps from the corresponding

,

-unsaturated amides 15 . In each case, the products 16 / 17 were obtained as ca. 80:20 mixtures of epimers ( 16 : 17) which is consistent with thermodynamic stereochemical control.[16]
O
HO
O O
14 a, b
R
1
R
2
N
O
OH
OH
15 a : R
1
=R
2
=Et b : R
1
=Me; R
2
= n Pr c : R
1
=R
2
= n Pr e c, d d : R
1
: R
=R
2
1
,R f : R
1
,R
2
2
= n
O
R
1
R
2
N
HO
Bu
= -(CH
= -(CH
2
2
)
)
4
5
-
-
O
OH
CO
2
H
R
1
16
OH g : -(CH
2
)
2
O(CH
2
)
2
h : R
1
= t Bu; R
2
=H i : R
1
= n Pent; R
2
=H
R
2
N
O
O
OH
CO
2
H
HO
17
OH j : R
1
= c Hex; R
2
=H k : R
1
=Ph; R
2
=H l : R
1
=( R )-CHMeBn m : R
1
=( S )-CHMeBn
Scheme 3. Parallel synthesis of the sialic acid mimetics 16 / 17 (see Table 2). Reagents and conditions: a) EDC, HOBt,
R
1
R
2
NH, CH
2
Cl
2
; b) 1:1 TFA

H
2
O; c) O
3
, MeOH then Me
2
S; d) 2 x 10
-2
mol% E192N, pyruvate, pH 7.4 buffer. EDC =
1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride, HOBt = 1-hydroxybenzotriazole hydrate, TFA = trifluoroacetic acid.
R
1
R
2
N
O
OH
OH
O
18
4

Table : Synthetic utility of the mutant sialic acid aldolase Glu192Asn entry Alkene R 1 R 2 Yield a Products Time
15 15 [%] [day]
1
2
3
4
5
6
7
8
9
10
11
12
13
15a
15b
15c
15d
15e
15f
15g
15h
15i
15j n
Et
Pr n Pr n Bu n
Et
Me n Pr
Bu
-(CH
2
)
4
-
-(CH
2
)
5
-
-(CH
2
)
2
O(CH
2
)
2
t Bu H n Pent c Hex
H
H
15k Ph H
15l ( R )-CHMeBn H
15m
14 ent-15c
( S )-CHMeBn n Pr n
H
Pr
61
73 e
78
88
66
59
68
79
81
90
67
74
68
78 g
16/17a
16/17b
16/17c
16/17d
16/17e
16/17f
16/17g
16/17h
16/17i
16/17j
16/17k
16/17l
16/17m
ent-16c/
3 82:18
3 82:18
3
3
3
3
3
14
14
14
14
14
14
7
16:17 d
82:18
82:18
82:18
79:21
82:18
45:55
70:30
60:40
60:40
60:40
60:40
64:36
Yield
16/17 c,d
37
42
42
66
48
55
47
[%]
35 [55]
[13] f
[29] f
[19] f
[30] f
[35] f
32 h
ent-17c a Yield of purifed product over two steps from the acid 14 .
b Determined by integration of the 500 MHz 1 H
NMR spectrum; c Yield of purified product over two steps from the corresponding alkene 15 ; [Yields in parentheses were determined by integration of the 500 MHz 1 H NMR spectrum of the crude reaction mixture]; d Products were obtained as mixtures of the two pyranose and the two furanose forms: for 16 , ca.
75:8:8:7 and for 17 , ca.
13:7:44:36; e 53:47 mixture of rotomers; f The ratio of the two pyranose and two furanose forms was not determined. g The enantiomeric acid ent-14 was used. h The enantiomeric starting material ent 15c (R 1 =R 2 = n Pr) was used.
The enzymatic reactions of the secondary amide substrates 15h-m were less efficient (entries 8-
13, Table). In each case, the required products were observed after 14 days by analysis of crude reaction mixtures by mass spectrometry and 500 MHz
1
H NMR spectroscopy. The yields of the products 16 / 17h-m , determined by 500 MHz 1 H NMR spectroscopy, were, however, low (19-55%).
After purification, the tertbutyl amide 16h / 17h was obtained in 35% yield over two steps from 15h as a 45:55 mixture of epimers.
Remarkably, the enantiomeric aldehyde ent-15c , prepared analogously from D -lyxose,[17] was also a substrate for the evolved aldolase (entry 14, Table). After 7 days, the enzymatic reaction was worked up and a 32% yield of the enantiomeric mimetic ent-16c / ent-17c was obtained. The catalysis of the aldol reaction was not as efficient in this case, however, since thermodynamic equilibration between the epimeric products was not complete (compare the 82:18 ratio of epimers, entry 3, with the 64:36 ratio of epimers, entry 14).
5

In summary, we have re-engineered an aldolase which was exploited in the parallel synthesis of sialic acid mimetics. The mutant enzyme was most efficient at catalysing the synthesis of the tertiary amides 16a-g / 17a-g , presumably, a consequence of the screening assay used: it is well known that “You get what you screen for”[18] and our assay would have selected for enzymes able to accept a range of sterically varied tertiary amides. The novel enzyme could be used to catalyse the interconversion of dynamic combinatorial libraries,[19] for example in the discovery of functional sialic acid mimetics such as paramyxovirus sialidase inhibitors.[20]
†
Astbury Centre for Structural Molecular Biology, University of Leeds, Leeds, UK, LS2 9JT
‡
School of Chemistry, University of Leeds, Leeds, UK, LS2 9JT
§
School of Biochemistry and Microbiology, University of Leeds, Leeds, UK, LS2 9JT
*Address for correspondence: School of Chemistry, University of Leeds, Leeds, UK, LS2 9JT
[1] For example, see: W. J. Albery, J. R. Knowles, Angew. Chem., Int. Ed. Engl. 1977 , 16 , 285-
293.
[2] See: X. Li, D. R. Liu, Angew. Chem., Int. Ed. 2004 , 43 , 4848-4870.
[3] For a transcript of his Nobel lecture, see: K. B. Sharpless, Angew. Chem., Int. Ed. 2002 , 41 ,
2024-2032.
[4] (a) M. Alexeeva, R. Carr, N. J. Turner, Org. Biomol. Chem. 2003 , 1 , 4133-4137; (b) K. A.
Powell, S. W. Ramer, S. B. del Cardayre, W. P. C. Stemmer, M. B. Tobin, P. F. Longchamp,
G. W. Huisman, Angew. Chem., Int. Ed. 2001 , 40 , 3948-3959; (c) M. T. Reetz, K.-E. Jaeger,
Topics Curr. Chem.
1999 , 200 , 31-57; (d) K.-E. Jaeger, T. Eggert, Curr. Opin. Biotech. 2004 ,
15 , 305-313.
[5] R. Carr, M. Alexeeva, A. Enright, T. S. C. Eve, M. J. Dawson N. J. Turner, Angew. Chem.,
Int. Ed. 2003 , 42 , 4807-4810.
6

[6] G. J. Williams, S. Domann., A. Nelson, A. Berry, Proc. Natl. Acad. Sci. USA 2003 , 100 , 3143-
3148.
[7] See also: M. Wada, C.-C. Hsu, D. Franke, M. Mitchell, A. Heine, I. Wilson, C.-H. Wong,
Bioorg. Med. Chem. 2003 , 11 , 2091-2098.
[8] C.-H. Wong and G. M. Whitesides, Enzymes in synthetic organic chemistry , 1994 , Permagon,
Oxford.
[9] (a) P. W. Smith, S. L. Sollis, P. D. Howes, P. C. Cherry, I. D. Starkey, K. N. Cobley, H.
Weston, J. Scicinski, A. Merritt, A. Whittington, P. Wyatt, N. Taylor, D. Green, R. Bethell, S.
Madar, R. J. Fenton, P. J. Morley, T. Pateman, A. Beresford, J. Med. Chem. 1998 , 41 , 787-
797; (b) N. R. Taylor, A. Cleasby, O. Singh, T. Skarzynski, A. J. Wonacott, P. W. Smith, S. L.
Sollis, P. D. Howes, P. C. Cherry, R. Bethell, P. Colman, J. Varghese, J. Med. Chem. 1998 ,
41 , 798-807.
[10] M. von Itzstein, W.-Y. Wu, G. B. Kok, M. S. Pegg, J. C. Dyason, B. Jin, T. van Phan, M. L.
Smythe, H. F. White, S. W. Oliver, P. M. Colman, J. N. Varghese, D. M. Ryan, J. M. Woods,
R. C. Bethell, V. J. Hotham, J. M. Cameron, C. R. Penn, Nature 1993 , 363 , 418-423.
[11] J. A. R. G. Barbosa, B. J. Smith, R. DeGori, H. C. Ooi, S. M. Macuccio, E. M. Campi, W. R.
Jackson, R. Brossmer, M. Sommer, M. C. Lawrence, J. Mol. Biol. 2000 , 303 , 405-421.
[12] N. Cohen, B. L. Banner, A. J. Laurenzano, L. Carozza, Org. Synth. 1985, 63 , 127-135.
[13] (a) T. H. Chan, C. J. Li, J. Chem. Soc., Chem. Comm.
1992, 747-748; (b) M. D. Chappell, R.
L. Halcomb, Org. Lett.
2000 , 2 , 2003-2005; (c) C.-J. Li, T.-H. Chan, Tetrahedron , 1999 , 55 ,
11149-11176 .
[14] (a) M. Chérest, H. Felkin, N. Prudent, Tetrahedron Lett.
1968 , 2199-2204; (b) N. T. Anh,
Top. Curr. Chem. 1980 , 88 , 145-162.
[15] V. Jaeger, B. Haefele, Synthesis 1987 , 801-806.
[16] C. H. Lin, T. Sugai, R. L. Halcomb, Y. Ichikawa, C. H. Wong, J. Am. Chem. Soc 1992 , 114 ,
10138-10145.
7

[17] B. V. Lao, S. Lahiri, J. Carbohydr. Chem. 1996 , 15 , 975-984.
[18] F. H. Arnold, Acc. Chem. Res. 1998 , 31 , 125-131.
[19] (a) R. J. Lins, S. L. Flitsch, N. J. Turner, E. Irving, S. A. Brown, Angew. Chem., Int. Ed. Engl.
2002 , 41 , 3405-3407; (b) R. J. Lins, S. L. Flitsch, N. J. Turner, E. Irving, S. A. Brown,
Tetrahedron 2004 , 60 , 771-780.
[20] For application of dihydropyrancarboxamides similar to 2 in the inhibition of paramyxovirus sialidase, see: P. Chand, Y. S. Babu, S. R. Rowland, T.-H. Lin, PCT Int. Appl. 2002 , WO
2002076971.
Keywords
Biotransformations, Directed evolution, Aldol reaction, Enzymes
Table of contents
Directed evolution can be used to broaden the substrate specificity of an aldolase sufficiently for use in parallel syntheses. Here, the substrate specificity of sialic acid aldolase was switched 630fold, and the mutant enzyme was applied in the synthesis of fourteen sialic acid mimetics.
R
1
R
2
N
O
OH
OH
1. O
3 then Me
2
S
2. mutant aldolase
R
1
R
2
N
O
O
OH
CO
HO
OH
14 examples
2
H
8