Threonine Aldolases

CHAPTER THREE

Threonine Aldolases Sarah E. Franz, Jon D. Stewart1 Department of Chemistry, University of Florida, Gainesville, Florida, USA 1 Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Threonine Aldolases Utilized for Chemical Synthesis 3. Summary of Reactants and Products for Threonine Aldolases 3.1 Glycine/alkyl aldehydes 3.2 Glycine/aryl aldehydes 3.3 Other amino acid donors (D-Ala, D-Ser, and D-Cys) 4. Structural Studies of Threonine Aldolases 4.1 T. maritima L-threonine aldolase 4.2 E. coli L-threonine aldolase 4.3 Other threonine aldolase structures 5. Protein Engineering Studies of Threonine Aldolases 5.1 Improving catalytic activity 5.2 Improving thermostability 5.3 Improving stereoselectivity 5.4 Introducing and optimizing threonine aldolase activity into a novel scaffold 6. Conclusions and Future Outlook References

58 60 61 61 84 85 85 91 91 92 93 94 94 95 97 99 99

Abstract Threonine aldolases catalyze the pyridoxal phosphate-dependent condensation between small amino acids (principally glycine) and aldehydes such as acetaldehyde. Carbon–carbon bond formation involves forming two adjacent chiral centers. As a rule, threonine aldolases are very stereoselective for a-carbon configuration but show modest selectivity at the b-carbon. On the other hand, these enzymes accept a wide variety of synthetically useful acceptor aldehydes, making them important additions to the synthetic toolkit. This review briefly summarizes the reaction mechanism and then lists all published synthetic reactions by threonine aldolases as of early 2014. The current state of the art in crystallographic and protein engineering studies of these enzymes is also presented.

Advances in Applied Microbiology, Volume 88 ISSN 0065-2164 http://dx.doi.org/10.1016/B978-0-12-800260-5.00003-6

#

2014 Elsevier Inc. All rights reserved.

57

58

Sarah E. Franz and Jon D. Stewart

1. INTRODUCTION Aldol condensations are one of the most common ways that nature accomplishes carbon–carbon bond formation and/or cleavage. The reaction is widely applicable since many common metabolites—especially carbohydrates—contain aldehyde or ketone moieties and this allows them to function as either enol(ate) nucleophiles or electrophiles. Because the same molecule can serve as both donor and acceptor, the scope of accessible products is very large. The aldol addition product is usually favored, although the exact equilibrium position is dictated by the relative thermodynamic stabilities of the reactants and products as well as their concentrations. A variety of aldolases have evolved to facilitate these conversions in both primary and secondary metabolic pathways. Native aldolases generally tolerate little or no variation in the enol(ate) partner and this is often used to classify these enzymes. By contrast, aldolases usually accept a wide variety of aldehyde electrophiles, which forms the foundation of their synthetic versatility. Interest in using aldolases for nonnative substrates grew rapidly in the 1980s after a seminal publication by Wong and Whitesides (1983). Carbohydrates and their derivatives were logical targets for the first-generation syntheses involving aldolases because the reactions closely mimicked their normal metabolic roles and substrate acceptance was simplified. In this regard, dihydroxyacetone phosphate (DHAP)-dependent aldolases found particular utility and many ingenious applications were developed using a diverse range of aldehyde acceptors. The subsequent identification of four stereocomplementary DHAP aldolases that provided each of the four possible diastereomeric aldol addition products allowed this technology to mature into a well-accepted synthetic methodology (for recent reviews, see Brovetto, Gamenara, Mendez, & Seoane, 2011; Clapes, Fessner, Sprenger, & Samland, 2010). The major drawback of DHAP-dependent aldolases is their nearcomplete specificity for DHAP. This narrows the scope of accessible products to those containing this substructure (or those derivable by subsequent transformations of the DHAP moiety) and motivated a search for aldolases that accept other enol(ate) donors. These efforts yielded pyruvate-dependent aldolases and 2-keto-3-deoxygluconate aldolase, N-acetylneuraminic acid (NeuAc) lyase, and 2-keto-3-deoxy-6-phosphogluconate aldolase, all of which have been applied to organic synthesis (examples are summarized in Brovetto et al., 2011; Clapes et al., 2010). All of the aforementioned aldolases follow chemical mechanisms that involve either a metal ion-stabilized enol(ate) or a synthetically equivalent

59

Threonine Aldolases

enamine intermediate (utilizing an active-site Lys side chain that initially forms a Schiff’s base with the donor carbonyl). Pyridoxal phosphate (PLP)dependent aldolases follow a fundamentally different pathway (Scheme 3.1). These enzymes first establish a Schiff’s base between the substrate’s amino group and PLP (referred to as an external aldimine). The cationic pyridinium ring facilitates deprotonation on the a-carbon to the amine by an enzyme general base, yielding a highly resonance-stabilized anion. This nucleophile adds to the aldehyde acceptor, thereby forming the CdC bond and yielding a Schiff’s base complex between the aldol product and the PLP cofactor. The catalytic cycle is completed by an analogous Schiff’s base exchange (transaldimination) that transfers the cofactor from the product back to the active-site Lys side chain (referred to as an internal aldimine). To date, all known PLP-dependent aldolases utilize a-amino acids as their native substrates (typically glycine, serine, or threonine). In particular,

Lys H HO R1

Aldol product

H3N

NH3 R2 R1 H

O

R2 CO2

CH3

H

N H

Amino acid donor

R1

OPO3= H3N

CO2

PLP / Lys conjugate

B

B

NH3

H

CO2

R1

CO2 N

O CH3

OH

N

H H

N

H

O

OPO3=

OPO3= N H

CH3

PLP / product conjugate

N H PLP / amino donor conjugate O R2

O H B NH3 R2 H CO2 R1

O H B

H R2 CO2 R1

NH3

H

N

O CH3

H

H =

OPO3 N H

PLP-stabilized anion form 2

Scheme 3.1 Threonine aldolase mechanism.

N

O CH3

H OPO3=

N H

PLP-stabilized anion form 1

H Aldehyde acceptor

60

Sarah E. Franz and Jon D. Stewart

threonine aldolases have emerged as useful enzymes for organic synthesis since the aldol reaction creates two new, adjacent stereocenters (Eq. 3.1). These enzymes have been divided into four classes based on their stereochemical preferences: high-specificity L- and D-threonine aldolases and low-specificity L- and D-threonine aldolases. The high-specificity enzymes can be further subdivided into the threonine and allo-threonine subtypes. It should be noted that even those enzymes designated as “low-specificity” are in fact highly selective for a particular a-carbon configuration; the “low-specificity” term arises because they yield a mixed population of b-carbon configurations. R1 H3 N

Threonine aldolase

O CO2

R1 = H, CH3, CH2OH, CH2SH

+ R 2

H

R2

* OH

R1 H3 N

*

CO2

(1)

ð3:1Þ

The possibility of forming only one enantiomer (out of the four potential products) starting from simple, inexpensive building blocks has motivated most of the efforts in this research area. This review briefly summarizes the range of substrates and products that have been employed with threonine aldolases and then describes our structural knowledge of these enzymes and the efforts to use this information to increase their substrate range and stereoselectivities. For previous reviews in this area, see Liu, Dairi, et al. (2000), Liu, Odani, et al. (2000), and Du¨ckers, Baer, Simon, Gr€ oger, and Hummel (2010).

2. THREONINE ALDOLASES UTILIZED FOR CHEMICAL SYNTHESIS A handful of threonine aldolases have dominated the published synthetic applications, particularly the L-threonine aldolases from Aeromonas jandaei (Fesko, Uhl, Steinreiber, Gruber, & Griengl, 2010; Liu, Dairi, Kataoka, Shimizu, & Yamada, 1997), Candida humicola (Vassilev, Uchiyama, Kajimoto, & Wong, 1995), Pseudomonas putida (Steinreiber, Fesko, Mayer, et al., 2007; Steinreiber, Fesko, Reisinger, et al., 2007), Streptomyces coelicolor (Gwon & Baik, 2010), and Escherichia coli (Gutierrez et al., 2008; Gwon et al., 2012; Kimura, Vassilev, Shen, & Wong, 1997; Sagui, Conti, Roda, Contestabile, & Riva, 2008) along with D-threonine aldolases produced by Alcaligenes xylosoxidans (Liu, Dairi, et al., 2000; Liu, Odani, et al., 2000; Steinreiber, Fesko, Mayer, et al., 2007; Steinreiber, Fesko, Reisinger, et al., 2007), Pseudomonas sp. (Fesko et al., 2010), and Xanthomonas oryzae (Kimura et al., 1997). In addition to these bona fide

Threonine Aldolases

61

threonine aldolases, the Hilvert group has developed a mutant alanine racemase from Geobacillus stearothermophilus that endows the variant with D-threonine aldolase activity (Fesko, Giger, & Hilvert, 2008; Toscano, Mu¨ller, & Hilvert, 2007). This alanine racemase is evolutionarily related to D-threonine aldolases, and the mutation removed one of the two acid– base groups required for Ala epimerization. All of these workhorse enzymes have been cloned and overexpressed in E. coli at high levels, which simplifies their use in chemical synthesis. The Griengl group recently created and surveyed a larger collection of these enzymes in hopes of uncovering examples with higher diastereoselectivities (Fesko, Reisinger, et al., 2008). Whether the “Cb diastereoselectivity problem” can best be overcome by testing additional wild-type isolates or by applying protein engineering technologies to existing threonine aldolases awaits experimental testing.

3. SUMMARY OF REACTANTS AND PRODUCTS FOR THREONINE ALDOLASES We have compiled a complete list of synthetic applications of threonine aldolases that were published by early 2014 (Tables 3.1–3.4). The examples have been grouped first by the amino acid donor nucleophile (glycine, alanine, serine, or cysteine; Tables 3.1–3.4, respectively). Within each table, examples are ordered by increasing size and structural complexity of the aldehyde acceptor (alkyl aldehydes followed by aryl aldehydes). In most cases, aldol products were not isolated from the reaction mixtures and only fractional conversions based on chromatographic analysis (typically HPLC) are available. Stereochemical purities were almost universally assessed by chromatographic separations. Finally, it should be noted that large excesses of the amino acid were typically employed to drive reactions toward the desired aldol product. For these reasons, the fractional conversion achieved in a given example should only be taken as a rough guide with regard to estimating the synthetic feasibility of a preparative-scale reaction.

3.1. Glycine/alkyl aldehydes Simple alkyl aldehydes can be converted to L-anti-products with high diastereoselectivities by the E. coli L-threonine aldolase (Table 3.1, entries 4, 11, and 16) (Kimura et al., 1997). The active site of this enzyme can accommodate relatively large n-alkyl aldehydes, although the antidiastereoselectivity and fractional conversion decline as aldehyde size increases (Table 3.1, entries 24 and 30). To date, a D-threonine aldolase with comparable levels of diastereoselectivity has not been identified,

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor Aldehyde acceptor

Product

O CH3

OH H

CO2

CH3

Enzyme

Yield or conversion % de

Entry References

C. humicola L-TA

30–40%

N/A

1

Vassilev et al. (1995)

P. putida L-TA

n.d.

4% (anti)

2

Steinreiber, Fesko, Mayer, et al. (2007)

E. coli L-TA

40% (24 h) 82% (anti)

3

Kimura et al. (1997)

E. coli L-TA

35% (3 h)

99% (anti)

4

Kimura et al. (1997)

A. xylosoxidans D-TA

n.d.

2% (anti)

5

Steinreiber, Fesko, Mayer, et al. (2007)

X. oryzae D-TA

60% (3 h)

6% (syn)

6

Kimura et al. (1997)

X. oryzae D-TA

50% (24 h) 6% (syn)

7

Kimura et al. (1997)

C. humicola L-TA

30%

n.d.

8

Vassilev et al. (1995)

P. putida L-TA

71%

28% (syn)

9

Steinreiber, Fesko, Mayer, et al. (2007)

E. coli L-TA

18% (24 h) 97% (anti)

10

Kimura et al. (1997)

E. coli L-TA

15% (3 h)

11

Kimura et al. (1997)

NH3

OH CH3

CO2 NH3

O CH3

OH H

CH3

CO2 NH3

99% (anti)

OH CH3

NH3 O CH3

OH H

H3C

NH3

OH

H3C

CH3

NH3

CH3 OH H

9% (syn)

12

Steinreiber, Fesko, Mayer, et al. (2007)

X. oryzae D-TA

37%

33% (syn)

13

Kimura et al. (1997)

P. putida L-TA

55%

42% (syn)

14

Steinreiber, Fesko, Mayer, et al. (2007)

E. coli L-TA

10% (3 h)

76% (anti)

15

Kimura et al. (1997)

E. coli L-TA

1% (24 h)

99% (anti)

16

Kimura et al. (1997)

Y265A Ala racemase

0%

n.d.

17

Fesko, Giger, and Hilvert (2008)

X. oryzae D-TA

49%

86% (syn)

18

Kimura et al. (1997)

A. xylosoxidans D-TA

24%

>95% 19 (syn)

Steinreiber, Fesko, Mayer, et al. (2007)

P. putida L-TA

94%

10% (syn)

20

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

65%

54% (syn)

21

Steinreiber, Fesko, Mayer, et al. (2007)

CO2

CH3

CH3 O

52%

CO2

CH3

CH3

A. xylosoxidans D-TA CO2

CO2

CH3

NH3 CH3 OH CH3

CO2 NH3

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de O CH3 4

OH H

P. putida L-TA

92%

31% (syn)

22

Steinreiber, Fesko, Mayer, et al. (2007)

E. coli L-TA

16% (24 h) 26% (anti)

23

Kimura et al. (1997)

E. coli L-TA

7% (3 h)

84% (anti)

24

Kimura et al. (1997)

X. oryzae D-TA

31% (3 h)

22% (syn)

25

Kimura et al. (1997)

X. oryzae D-TA

23% (24 h) 28% (syn)

26

Kimura et al. (1997)

A. xylosoxidans D-TA

42% (30% DMSO)

68% (syn)

27

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

33% (no cosolvent)

73% (syn)

28

Steinreiber, Fesko, Mayer, et al. (2007)

P. putida L-TA

25%

9% (syn)

29

Steinreiber, Fesko, Mayer, et al. (2007)

E. coli L-TA

2%

44% (anti)

30

Kimura et al. (1997)

X. oryzae D-TA

3%

38% (syn)

31

Kimura et al. (1997)

A. xylosoxidans D-TA

12%

55% (syn)

32

Steinreiber, Fesko, Mayer, et al. (2007)

CO2

CH3

Entry References

4

NH3

OH CO2

CH3 4

NH3

OH

O CH3 6

H

CO2

CH3 6

NH3 OH CO2

CH3 6

NH3

OH

O CH3 8

H

P. putida L-TA

29%

23% (anti)

33

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

<1%

n.d.

34

Steinreiber, Fesko, Mayer, et al. (2007)

P. putida L-TA

33%

15% (anti)

35

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

<1%

n.d.

36

Steinreiber, Fesko, Mayer, et al. (2007)

P. putida L-TA

11%

15% (anti)

37

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

<1%

n.d.

38

Steinreiber, Fesko, Mayer, et al. (2007)

CO2

CH3 8

NH3

OH CO2

CH3 8

NH3

O CH3 9

OH H

CO2

CH3 9

NH3 OH CO2

CH3 9

NH3

O CH3 10

OH H

CO2

CH3 10

NH3 OH CO2

CH3 10

NH3 Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de O CH3 4

9

OH H

CH3

Entry References

C. humicola L-TA

30%

n.d.

39

Vassilev et al. (1995)

P. putida L-TA

50%

93% (syn)

40

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

30%

97% (syn)

41

Steinreiber, Fesko, Mayer, et al. (2007)

C. humicola L-TA

<5%

n.d.

42

Vassilev et al. (1995)

P. putida L-TA

65%

40% (syn)

43

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

26%

73% (syn)

44

Steinreiber, Fesko, Mayer, et al. (2007)

CO2 4

9

NH3 OH

O F

H

F

CO2 NH3 OH

F

CO2 NH3 OH

O Cl

H

Cl

CO2 NH3

OH Cl

CO2 NH3

O Br

OH Br

H

P. putida L-TA

20%

73% (syn)

45

Steinreiber, Fesko, Mayer, et al. (2007)

A. xylosoxidans D-TA

6%

82% (syn)

46

Steinreiber, Fesko, Mayer, et al. (2007)

C. humicola L-TA

>75%

n.d.

47

Vassilev et al. (1995)

E. coli L-TA

13%

0%

48

Sagui et al. (2008)

E. coli L-TA

30%

20% (syn)

49

Gutierrez et al. (2008)

C. humicola L-TA

>75%

84% (anti)

50

Vassilev et al. (1995)

E. coli L-TA

36%

88% (anti)

51

Kimura et al. (1997)

X. oryzae D-TA

80%

40% (syn)

52

Kimura et al. (1997)

CO2 NH3

OH Br

CO2 NH3

O N3

OH CO2

N3

H

NH3

OH

O O

H

O

CO2 NH3

OH O

CO2 NH3

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de

C. humicola L-TA

45–75%

84% (anti)

53

Vassilev et al. (1995)

C. humicola L-TA

45%

86% (anti)

54

Vassilev et al. (1995)

E. coli L-TA

18%

40% (syn)

55

Gutierrez et al. (2008)

C. humicola L-TA

30%

n.d.

56

Vassilev et al. (1995)

E. coli L-TA

29%

10% (syn)

57

Gutierrez et al. (2008)

CH3 NH3

E. coli L-TA

54%

64% (syn)

58

Gutierrez et al. (2008)

OH

E. coli L-TA

40%

68% (syn)

59

Gutierrez et al. (2008)

O O

O

OH O

H

3

Entry References

O

CO2

3

NH3 O Pth

Cbz

O

N H

H N

OH H

Pth

O H

Cbz

N H

H N

O

CO2 NH3

OH CO2 NH3

Pth

H N

O H

Pth

H N

OH CO2 NH3

Cbz

H N

O H

Cbz

H N

CH3

Cbz

H N

O H CH3

Cbz

H N

OH CO2

CO2 CH3 NH3

O

OH CO2

H

C. humicola L-TA

10–30%

n.d.

60

Vassilev et al. (1995)

E. coli L-TA

10% (30% DMSO)

44% (anti)

61

Kimura et al. (1997)

E. coli L-TA

5% (no cosolvent)

66% (anti)

62

Kimura et al. (1997)

X. oryzae D-TA

16%

74% (syn)

63

Kimura et al. (1997)

C. humicola L-TA

53%

6% (anti)

64

Kimura et al. (1997)

E. coli L-TA

10%

88% (anti)

65

Kimura et al. (1997)

X. oryzae D-TA

45% (3 h)

29% (syn)

66

Kimura et al. (1997)

X. oryzae D-TA

35% (25 min)

64% (syn)

67

Kimura et al. (1997)

E. coli L-TA

11%

0%

68

Gutierrez et al. (2008)

NH3

OH CO2 NH3

O

OH H

O

CO2

O

NH3

OH CO2

O

NH3

O Cbz

N H

OH

H

Cbz

N H

CO2 NH3 Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de O

OH H

S

Entry References

C. humicola L-TA

>75%

n.d.

69

Vassilev et al. (1995)

C. humicola L-TA

10%

n.d.

70

Vassilev et al. (1995)

C. humicola L-TA

10–30%

n.d.

71

Vassilev et al. (1995)

C. humicola L-TA

10%

n.d.

72

Vassilev et al. (1995)

E. coli L-TA

67%

0%

73

Sagui et al. (2008)

E. coli L-TA

34%

n.d.

74

Sagui et al. (2008)

CO2

S

NH3 CH3 O

CH3 OH H

S

CO2

S

NH3 CH3 CH3O S

CH3 CH3 OH H

CO2

S

NH3 EtO

O

EtO

O H

S

O

OH CO2

S

NH3 O O2C

OH H

O2C

CO2 NH3

O O2C

OH H

O2C

CO2 NH3

OH

O H O CH3

O

O

CH3

CH3

O

OH O CH3

O

CH3

H

O CH3

O

H

O CH3

O

O H

E. coli L-TA

70%

0%

77

Kimura et al. (1997)

X. oryzae D-TA

84%

76% (syn)

78

Kimura et al. (1997)

C. humicola L-TA

30%

n.d.

79

Miura and Kajimoto (2001)

NH3

O CH3

Ph O

Kimura et al. (1997)

CO2

O

Ph

76

CH3

OH

CH3

40% (anti)

NH3

O

O

CH3

73%

CO2

O

O

X. oryzae D-TA

NH3

OH

CH3

Kimura et al. (1997)

CH3

O O

75

CO2

O CH3

92% (anti)

CH3

H CH3

35%

NH3

O O

E. coli L-TA CO2

O

OH CO2 NH3

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de OH

O H

T. maritima L-allo-TA

25% (5 min)

20% (syn)

80

Fesko, Reisinger, et al. (2008)

P. putida L-TA

80% (30 min)

21% (syn)

81

Steinreiber, Fesko, Reisinger, et al. (2007)

P. aeruginosa L-TA

80% (30 min)

21% (syn)

82

Fesko, Reisinger, et al. (2008)

A. jandaei L-allo-TA

30% (5 min)

27% (syn)

83

Fesko, Reisinger, et al. (2008)

P. putida L-TA

40% (1 min)

>30% 84 (syn)

Fesko, Reisinger, et al. (2008)

P. aeruginosa L-TA

10% (1 min)

>30% 85 (syn)

Fesko, Reisinger, et al. (2008)

E. coli L-TA

9% (24 h)

60% (syn)

86

Kimura et al. (1997)

E. coli L-TA

3% (3 h)

71% (syn)

87

Kimura et al. (1997)

A. jandaei L-allo-TA

<20% (<1 min)

Anti

88

Fesko, Reisinger, et al. (2008)

S. cerevisiae L-low-TA

60% (5 h)

22% (anti)

89

Fesko, Reisinger, et al. (2008)

CO2 NH3

Entry References

S. cerevisiae L-low-TA

4% (1 min) 40% (anti)

90

Fesko, Reisinger, et al. (2008)

C. humicola L-TA

45%

40% (anti)

91

Vassilev et al. (1995)

B. bronchiseptica

10% (5 min)

70% (anti)

92

Fesko, Reisinger, et al. (2008)

S. pomperoyl D-low-TA 80% (5 days)

21% (syn)

93

Fesko, Reisinger, et al. (2008)

X. oryzae D-TA

10% (24 h) 73% (syn)

94

Kimura et al. (1997)

X. oryzae D-TA

11% (3 h)

74% (syn)

95

Kimura et al. (1997)

Y265A Ala racemase

17% (24 h) 76% (syn)

96

Fesko, Giger, and Hilvert (2008)

Y265A Ala racemase

10% (3 h)

97% (syn)

97

Fesko, Giger, and Hilvert (2008)

A. xylosoxidans D-TA

79%

98% (syn)

98

Steinreiber, Fesko, Reisinger, et al. (2007)

L-low-TA

OH CO2 NH3

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de F

F

O

OH

P. putida L-TA

68%

35% (syn)

99

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

68%

95% (syn)

100

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

90%

52% (syn)

101

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

27%

67% (syn)

102

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

79%

34% (syn)

103

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

6%

35% (syn)

104

Steinreiber, Fesko, Reisinger, et al. (2007)

CO2

H

Entry References

NH3 F

OH CO2 NH3

Cl

Cl

O

OH CO2

H

NH3

Cl

OH CO2 NH3

Br

Br

O

OH CO2

H

NH3

Br

OH CO2 NH3

NH2 OH

NH2 O

Y265A Ala racemase

<1%

n.d.

105

Fesko, Giger, and Hilvert (2008)

P. putida L-TA

99%

32% (syn)

106

Steinreiber, Fesko, Reisinger, et al. (2007)

E. coli L-TA

93% (24 h) 42% (anti)

107

Kimura et al. (1997)

E. coli L-TA

46% (3 h)

68% (anti)

108

Kimura et al. (1997)

X. oryzae D-TA

89% (3 h)

44% (syn)

109

Kimura et al. (1997)

A. xylosoxidans D-TA

18%

65% (syn)

110

Steinreiber, Fesko, Reisinger, et al. (2007)

Y265A Ala racemase

1%

>97% 111 (syn)

Fesko, Giger, and Hilvert (2008)

P. putida L-TA

64%

27% (syn)

112

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

54%

81% (syn)

113

Steinreiber, Fesko, Reisinger, et al. (2007)

CO2

H

NH3

NO2 O

NO2 OH CO2

H

NH3

NO2 OH CO2 NH3

OH

O

CO2

H

NH3

F

F OH CO2 NH3 F

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de OH

O

P. putida L-TA

69%

30% (syn)

114

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

60%

85% (syn)

115

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

63%

55% (syn)

116

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

43%

71% (syn)

117

Steinreiber, Fesko, Reisinger, et al. (2007)

CO2

H

Entry References

NH3 Cl

Cl

OH CO2 NH3 Cl O

OH CO2

H

NH3 Br

Br OH CO2 NH3 Br

O

OH H

NH3 OH

E. coli L-TA

43%

46% (syn)

118

Kimura et al. (1997)

P. putida L-TA

56%

51% (syn)

119

Steinreiber, Fesko, Reisinger, et al. (2007)

X. oryzae D-TA

53% (24 h) 46% (syn)

120

Kimura et al. (1997)

X. oryzae D-TA

54% (3 h)

48% (syn)

121

Kimura et al. (1997)

Y265A Ala racemase

3%

70% (syn)

122

Fesko, Giger, and Hilvert (2008)

A. xylosoxidans D-TA

76%

86% (syn)

123

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

74%

21% (syn)

124

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

90%

80% (syn)

125

Steinreiber, Fesko, Reisinger, et al. (2007)

Y265A Ala racemase

55% (24 h) 85% (syn)

126

Fesko, Giger, and Hilvert (2008)

Y265A Ala racemase

20% (3 h)

127

Fesko, Giger, and Hilvert (2008)

CO2

OH OH CO2 NH3 OH

OH

O

CO2

H

NH3

NO2

NO2 OH CO2 NH3 NO2

93% (syn)

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de O

OH

CH3

E. coli L-TA

17%

20% (anti)

128

Kimura et al. (1997)

X. oryzae D-TA

25%

14% (anti)

129

Kimura et al. (1997)

P. putida L-TA

51%

29% (syn)

130

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

42%

91% (syn)

131

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

57%

17% (syn)

132

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

26%

86% (syn)

133

Steinreiber, Fesko, Reisinger, et al. (2007)

CO2

H

NH3

CH3

OH CO2 NH3

CH3 OH

O

CO2

H F

F

NH3

OH CO2 F O

NH3

OH CO2

H Cl

Cl

NH3 OH CO2

Cl

Entry References

NH3

OH

O

P. putida L-TA

47%

14% (syn)

134

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

12%

74% (syn)

135

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

11%

36% (syn)

136

Steinreiber, Fesko, Reisinger, et al. (2007)

C. humicola L-TA

30%

40% (anti)

137

Vassilev et al. (1995)

A. xylosoxidans D-TA

15%

70% (syn)

138

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

79%

24% (syn)

139

Steinreiber, Fesko, Reisinger, et al. (2007)

E. coli L-TA

53% (24 h) 6% (anti)

140

Kimura et al. (1997)

E. coli L-TA

35% (3 h)

141

Kimura et al. (1997)

CO2

H Br

Br

NH3 OH CO2

Br

NH3

OH

O

CO2

H HO

HO

NH3 OH CO2

HO

O

OH CO2

H O2N

NH3

O2N

NH3

28% (anti)

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de OH

X. oryzae D-TA

88% (24 h) 10% (syn)

142

Kimura et al. (1997)

X. oryzae D-TA

72% (3 h)

16% (syn)

143

Kimura et al. (1997)

Y265A Ala racemase

36% (24 h) 40% (syn)

144

Fesko, Giger, and Hilvert (2008)

A. xylosoxidans D-TA

31%

75% (syn)

145

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

68%

53% (syn)

146

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

63%

99% (syn)

147

Steinreiber, Fesko, Reisinger, et al. (2007)

CO2 NH3

O2N

O

OH CO2

H O CH3

O S O

CH3

NH3

S O OH

CO2 O S

CH3

NH3 O

Entry References

O

OH

H O S H2N O

O H2N

S O

P. putida L-TA

92%

24% (syn)

148

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

53%

>90% 149 (syn)

Steinreiber, Fesko, Reisinger, et al. (2007)

P. putida L-TA

<1%

n.d.

150

Steinreiber, Fesko, Reisinger, et al. (2007)

S. coelicolor L-TA

n.d.

14% (syn)

151

Gwon and Baik (2010)

V81I/R241C/Y306C n.d. S. coelicolor L-TA

21% (syn)

152

Gwon and Baik (2010)

R241C/A287V S. coelicolor L-TA

n.d.

21% (syn)

153

Gwon and Baik (2010)

Y306C S. coelicolor

n.d.

26% (syn)

154

Gwon and Baik (2010)

n.d.

28% (syn)

155

Gwon and Baik (2010)

CO2 NH3

OH CO2 O S H2N

HO

O

OH

O HO

NH3

H

HO HO

CO2 NH3

L-TA

Y36C/Y306C/ R316C S. coelicolor L-TA

Continued

Table 3.1 Synthetic applications of threonine aldolases using glycine as the donor—cont'd Yield or Aldehyde acceptor Product Enzyme conversion % de

OH

O O

O

O

O

O CH3O CH3O

CH3O

Gwon and Baik (2010)

Y39C/T306C/A48T S. coelicolor L-TA

43% n.d. (fourfold of (syn) wt)

157

Gwon and Baik (2010)

E. coli L-TA

71%

60% (syn)

158

Gwon et al. (2012)

A. xylosoxidans D-TA

<1%

n.d.

159

Steinreiber, Fesko, Reisinger, et al. (2007)

Y265A Ala racemase

5%

70% (syn)

160

Fesko, Giger, and Hilvert (2008)

P. putida L-TA

15%

16% (syn)

161

Steinreiber, Fesko, Reisinger, et al. (2007)

A. xylosoxidans D-TA

16%

46% (syn)

162

Steinreiber, Fesko, Reisinger, et al. (2007)

CO2 NH3 OH

CH3O

156

NH3

CH3O CH3O

38% (syn)

CO2

OH H

n.d.

NH3

OH H

Y39C/Y306C S. coelicolor L-TA

CO2

HO HO

Entry References

CO2 NH3

OH

O H

NH3

F

F

E. coli L-TA

21% (24 h) 40% (anti)

163

Kimura et al. (1997)

E. coli L-TA

19% (3 h)

50% (anti)

164

Kimura et al. (1997)

X. oryzae D-TA

49% (24 h) 30% (syn)

165

Kimura et al. (1997)

X. oryzae D-TA

42% (3 h)

52% (syn)

166

Kimura et al. (1997)

E. coli L-TA

40%

32% (syn)

167

Kimura et al. (1997)

X. oryzae D-TA

60%

22% (syn)

168

Kimura et al. (1997)

CO2

NO2

NO2

OH CO2 NH3

F NO2

OH

O N

H

CO2

N

NH

NH NH3

OH N

CO2 NH NH3

Alkyl aldehydes are shown first, in the approximate order of increasing size and structural complexity. Aryl aldehydes follow, also in the order of increasing size and structural complexity. When a given reaction has been carried out by more than one threonine aldolase, entries are arranged in the order of increasing anti-selectivity, followed by increasing syn-selectivity for both L- and D-selective aldolases. n.d., not determined or not reported. de, diastereomeric excess = % major diastereomer % minor diastereomer.

84

Sarah E. Franz and Jon D. Stewart

although the A. xylosoxidans enzyme can show good syn-selectivity in favorable cases (e.g., Table 3.1, entry 19). Threonine aldolases generally tolerate relatively bulky and highly substituted aldehyde acceptors. a-Halo-, a-alkoxy-, and a-amino moieties are acceptable, even when the latter are derivatized by large protecting groups such as benzyl and Cbz (Table 3.1, entries 40–59). The main drawback is that diastereomeric mixtures are usually obtained and the preferences for Cb-stereochemistry are relatively modest. Aldehyde acceptors with b-substituents—even relatively large ones—are also tolerated by threonine aldolases (Table 3.1, entries 60–78). The general conclusion is that nearly all alkyl aldehydes can serve as acceptors for glycine; however, it is likely that a mixture of diastereomers will be obtained at the b-carbon. Preexisting chiral centers in the aldehyde have modest impacts on diastereoselectivity (see Table 3.1, entries 57–59 and 75–79).

3.2. Glycine/aryl aldehydes Because it is a critical component of some semisynthetic b-lactam antibiotics, phenylserine has been a key target for threonine aldolases (Table 3.1, entries 80–98). A number of L-threonine aldolases yield the desired target from benzaldehyde and glycine, although the diastereoselectivity is incomplete. By contrast, A. xylosoxidans D-threonine aldolase produces the D-syn-product with both high conversion and excellent diastereoselectivity (Table 3.1, entry 98). This is one of the more successful uses of threonine aldolases in preparative synthesis. Interestingly, the engineered Ala racemase also gives very good D-syn-diastereoselectivity in this reaction, although this is tempered by poor fractional conversion (Table 3.1, entries 96 and 97). A variety of monosubstituted benzaldehydes have been tested as glycine acceptors by a variety of threonine aldolases (Table 3.1, entries 99–149). In many cases, the A. xylosoxidans D-threonine aldolase affords good diastereoselectivity as does the Ala racemase point mutant. The P. putida L-threonine aldolase also accepts a wide variety of monosubstituted benzaldehydes; however, the diastereoselectivities are generally poor to moderate. The Yamada group published an early study focused on 4-methylthiophenylserine involving the same substrate pair as entries 146 and 147, but employing a low-specificity D-threonine aldolase from Arthrobacter sp. DK-38 (Liu et al., 1999). Unfortunately, the diastereomeric purity of the product was not reported.

Threonine Aldolases

85

Next to phenylserine itself, 3,4-dihydroxyphenylserine has been the most popular application of threonine aldolases in preparative synthesis since this compound has been used to treat Parkinson’s disease (Table 3.1, entries 151–158). Earlier synthetic routes involved difficult separations of diastereomeric mixtures; the possibility that an enzyme could directly yield only the desired material was a major impetus in exploring threonine aldolases as an alternative. Interestingly, some further derivatization of the hydroxyl groups of 3,4-dihydroxybenzaldehyde can be tolerated, for example, 3,4methylenedioxy or 3,4-dimethoxy (Table 3.1, entries 160–162), but monomethoxy analogs were not accepted by the enzyme (Steinreiber, Fesko, Mayer, et al., 2007; Steinreiber, Fesko, Reisinger, et al., 2007).

3.3. Other amino acid donors (D-Ala, D-Ser, and D-Cys) After screening a variety of native threonine aldolases, the Griengl group identified two enzymes that accepted more complex amino acids in addition to glycine (Fesko et al., 2010). Despite forming only L-threonine and analogs when glycine was the donor, the A. jandaei enzyme could also accept D-Ala as a substrate. More surprisingly, L-Ala was not accepted. A panel of representative aldehydes was tested as partners for D-Ala, and the results paralleled those observed for Gly with the same aldehydes (Table 3.2). Despite the relatively poor diastereoselectivities (caused by mixtures at the Cb chiral center), these results are significant since a quaternary, nonracemizable center is created at the a-carbon with very high enantioselectivities. This remains a difficult challenge in organic synthesis. In addition to the A. jandaei enzyme, a D-threonine aldolase from Pseudomonas sp. also utilized D-Ala and a variety of aldehyde acceptors, albeit with modest diastereoselectivities. The same pair of aldolases described in the preceding text also accepted D-Ser and D-Cys as nucleophiles in aldol reactions (Tables 3.3 and 3.4, respectively) (Fesko et al., 2010). While neither fractional conversion nor diastereoselectivities were high, the ability to exercise high control over a quaternary a-carbon center is an equally impressive achievement.

4. STRUCTURAL STUDIES OF THREONINE ALDOLASES In addition to the extensive catalytic characterization studies carried out with threonine aldolases, our understanding of their structures has also increased in recent years. Several representative crystal structures are known, although some key enzymes used widely for synthesis remain unsolved. Interestingly, L- and D-threonine aldolases are structurally and evolutionarily

Table 3.2 Synthetic applications of threonine aldolases using alanine as the donor Aldehyde acceptor Product Enzyme Yield or conversion % de O CH3

A. jandaei L-TA

OH

Fesko et al. (2010)

A. jandaei L-TA

26% (anti) 3

Fesko et al. (2010)

Pseudomonas sp. D-TA 32%

66% (syn) 4

Fesko et al. (2010)

A. jandaei L-TA

8% (anti)

5

Fesko et al. (2010)

Pseudomonas sp. D-TA 84%

33% (syn) 6

Fesko et al. (2010)

A. jandaei L-TA

6% (anti)

Fesko et al. (2010)

6%

CO2

CH3

OH CO2

CH3 H3C H3N

O 4

42% (syn) 2

H3C CH3 NH3

CH3

CH3

Pseudomonas sp. D-TA 54% CH3

OH

H

Fesko et al. (2010)

CO2

CH3 H3N

O

46% (anti) 1

NH3

OH

CH3

20%

CO2

CH3 CH3

H

Entry References

CH3

OH H

58%

CO2

CH3 4

CH3 NH3 OH CH3

CO2 4

H3N O

CH3

OH H

CO2 CH3 NH3

35%

7

OH

Pseudomonas sp. D-TA 11%

65% (syn) 8

Fesko et al. (2010)

A. jandaei L-TA

35% (syn) 9

Fesko et al. (2010)

Pseudomonas sp. D-TA 21%

95% (syn) 10

Fesko et al. (2010)

A. jandaei L-TA

60%

7% (anti)

Fesko et al. (2010)

Y265A Ala racemase

12% (24)

65% (syn) 12

Fesko, Giger, and Hilvert (2008)

Pseudomonas sp. D-TA 36%

76% (syn) 13

Fesko et al. (2010)

Y265A Ala racemase

80% (syn) 14

Fesko, Giger, and Hilvert (2008)

CO2 H3N Cl

O

Cl

CH3

OH

24%

CO2

H

CH3 NH2 Cl

OH CO2 H3N

O

CH3

OH

11

CO2

H

CH3 NH3 NO2

NO2 OH CO2 H3N NO2

CH3

5% (3 h)

Alkyl aldehydes are shown first, in the approximate order of increasing size and structural complexity. Aryl aldehydes follow, also in the order of increasing size and structural complexity. When a given reaction has been carried out by more than one threonine aldolase, entries are arranged in the order of increasing anti-selectivity, followed by increasing syn-selectivity for both L- and D-selective aldolases. n.d., not determined or not reported. de, diastereomeric excess = % major diastereomer % minor diastereomer.

Table 3.3 Synthetic applications of threonine aldolases using serine as the donor Aldehyde acceptor Product Enzyme Yield or conversion O CH3

OH

Entry

References

A. jandaei L-TA

6%

65% (anti)

1

Fesko et al. (2010)

Pseudomonas sp. D-TA

23%

11% (anti)

2

Fesko et al. (2010)

A. jandaei L-TA

30%

45% (anti)

3

Fesko et al. (2010)

Pseudomonas sp. D-TA

43%

24% (anti)

4

Fesko et al. (2010)

A. jandaei L-TA

10%

40% (anti)

5

Fesko et al. (2010)

Pseudomonas sp. D-TA

<1%

n.d.

6

Fesko et al. (2010)

CO2

CH3

H

% de

NH3 OH

OH CO2 OH

CH3 H3N O CH3 4

OH H

CO2

CH3 4

NH3 OH OH CO2

CH3 4

H3N

OH

OH

O

CO2

H

NH2 OH

OH CO2 H3N

OH

O

OH

A. jandaei L-TA

15%

65% (anti)

7

Fesko et al. (2010)

Pseudomonas sp. D-TA

<5%

23% (anti)

8

Fesko et al. (2010)

CO2

H

NH3 OH NO2

NO2

OH CO2 H3N

OH

NO2 Alkyl aldehydes are shown first, in the approximate order of increasing size and structural complexity. Aryl aldehydes follow, also in the order of increasing size and structural complexity. When a given reaction has been carried out by more than one threonine aldolase, entries are arranged in the order of increasing anti-selectivity, followed by increasing syn-selectivity for both L- and D-selective aldolases. n.d., not determined or not reported. de, diastereomeric excess = % major diastereomer % minor diastereomer.

Table 3.4 Synthetic applications of threonine aldolases using cysteine as the donor Aldehyde acceptor Product Enzyme Yield or conversion O

OH

% de

Entry

References

A. jandaei L-TA

27%

18% (anti)

1

Fesko et al. (2010)

Pseudomonas sp. D-TA

33%

20% (anti)

2

Fesko et al. (2010)

A. jandaei L-TA

30%

12%(anti)

3

Fesko et al. (2010)

Pseudomonas sp. D-TA

39%

6% (anti)

4

Fesko et al. (2010)

CO2

H

NH2 SH OH CO2 H3N

SH

OH

O

CO2

H

NH3 SH

NO2

NO2

OH CO2 H3N

SH

NO2 Alkyl aldehydes are shown first, in the approximate order of increasing size and structural complexity. Aryl aldehydes follow, also in the order of increasing size and structural complexity. When a given reaction has been carried out by more than one threonine aldolase, entries are arranged in the order of increasing anti-selectivity, followed by increasing syn-selectivity for both L- and D-selective aldolases. n.d., not determined or not reported. de, diastereomeric excess = % major diastereomer % minor diastereomer.

Threonine Aldolases

91

distinct, the former belonging to the aspartate aminotransferase family and the latter to the alanine racemase family. Currently, much more is known about the structures of L-threonine aldolases than their D counterparts.

4.1. T. maritima L-threonine aldolase The first crystal structure of a threonine aldolase was reported by Kielkopf and Burley in 2001 and 2002 (PDB codes 1JG8 and 1M6S) (Kielkopf & Burley, 2002). This is a low-specificity enzyme produced by a thermophilic organism. Its amino acid sequence is similar to those of several L-threonine aldolases that have been applied to chemical synthesis (including those from A. jandaei, E. coli, P. aeruginosa, and S. cerevisiae). In addition to the native enzyme, cocrystallized forms with either Gly (PDB code 1LW5) or L-allo-Thr (PDB code 1LW4) were also solved. External aldimine formation with either amino acid did not change the overall protein structure apart from a small rotation of the PLP ring. Of the four active sites in the homotetrameric enzyme, two were fully occupied with substrate/product, a third was partially occupied and the fourth showed only the resting form (the internal aldimine, in which the PLP cofactor forms a Schiff’s base with a Lys side chain). Whether the diversity of active-site occupancies reflects kinetic cooperativity or was an artifact of the crystallization process remains unknown. The b-hydroxyl group of bound L-allo-Thr interacted with both an active-site His side chain (residue 83) and an ordered water molecule that was in turn hydrogenbonded to the phosphate of PLP. This suggests that the side chain of His 83 might act as a general acid–base group in the catalytic mechanism.

4.2. E. coli L-threonine aldolase Very recently, Safo, Contestabile, and coworkers solved several X-ray crystal structures of the low-specificity L-threonine aldolase from E. coli (di Salvo et al., 2014). The overall structure of the unliganded enzyme (PDB code 4LNJ) was similar to that of the T. maritima L-threonine aldolase described in the preceding text (Kielkopf & Burley, 2002). One key difference is that the active sites of the homotetrameric E. coli enzyme are composed of residues that converge from three subunits.1 In common with other enzymes in this family, each monomer consists of two domains with the PLP cofactor located at their interface and bound via a Schiff’s base with Lys 197. Cocrystallizing E. coli L-threonine aldolase with L-Ser yielded a mixture of glycine bound to PLP by a Schiff’s base in the active site along with a 1

Active sites in most homologues are formed at the dimer interface between two subunits.

92

Sarah E. Franz and Jon D. Stewart

fraction in which the cofactor was covalently bound to Lys 197 (PDB code 4LNM) (di Salvo et al., 2014). External aldimine formation did not change the overall protein structure, although the PLP ring rotated slightly within the active site, a phenomenon observed previously in other PLP-dependent enzymes. When E. coli L-threonine aldolase was cocrystallized with L-Thr (PDB code 4LNL), one active site contained the glycine/PLP covalent complex and the other contained a mixed population of this species along with a mixture of PLP covalently bound to L-Thr and L-allo-Thr. Importantly, the side-chain hydroxyl groups of both amino acids were located in the same position; their respective methyl groups occupied different locations. This has important ramifications for the catalytic mechanism since the aldol/retro-aldol reaction involves direct acid–base interactions with this functional group. It also suggests ways to modify the active site to enhance b-carbon stereoselectivity. The b-hydroxyl group makes hydrogen bonds with both His 83 and His 126, suggesting that these might play a role as general acid/base catalysts. Surprisingly, the mutation of either His 83 or His 126 alone was tolerated by the enzyme, although some variants at these positions were unstable and precipitated after extended storage. Moreover, replacing His 126 with Asn or Phe actually increased kcat values by up to threefold, although compensating changes in KM blunted the impact somewhat. These results argue against a direct acid–base role by either His side chain, which was further supported by the retention of some catalytic activity in a double His 83/His 126 mutant. At this time, the identity of the group responsible for b-hydroxyl protonation/deprotonation remains a mystery. Two replacements were also made for Phe 87, which had been proposed to be a key determinant of substrate specificity. The properties of both the Ala and Asp mutants were not in accord with predictions, leaving this as another open issue. The authors also pointed out that the active site is much larger than needed to accommodate the relatively small substrate acetaldehyde. Furthermore, L-allo-Thr is not a recognized metabolite in E. coli. It is therefore possible that the true physiological role of this enzyme—despite its being named “L-Thr aldolase”—may actually involve different (and possibly multiple) substrates in the native host rather than Thr and/or allo-Thr.

4.3. Other threonine aldolase structures In addition to the more complete studies described in the preceding text, several other threonine aldolase crystal structures have been solved as part

Threonine Aldolases

93

of structural genomics projects with little or no additional data available. One example from Leishmania major was reported in 2011 (PDB code 1SVV). A second example was described in 2010, a threonine aldolase from Listeria monocytogenes EGD-E (PDB code 3PJ0). No published synthetic applications of these enzymes have appeared in the literature as of early 2014. The final structural entry, which lacks a formal literature citation, is the phenylserine aldolase from P. putida (PDB code 1V72).

5. PROTEIN ENGINEERING STUDIES OF THREONINE ALDOLASES Recent years have witnessed significant improvements in protein engineering methodologies, high-throughput screening, and selections along with structure determination using X-ray crystallography. These developments have made it possible for even smaller academic laboratories to undertake protein mutagenesis projects aimed at improving the performance of biocatalysts in synthetic applications. Surprisingly, only a relatively few examples where these techniques have been applied to threonine aldolases have been published as of early 2014, and these are summarized in the succeeding text. The selection and/or screening methodology dictates the number of variants that can be screened, and this is nearly always the limiting factor in protein engineering studies. In cases where L-threonine is the desired product, enabling growth of a threonine auxotroph in minimal medium can be used to select the desired variants. Even when substrates beyond L-threonine itself are targeted, this can be a useful “prescreen” since it can easily be applied to libraries containing up to 1010 variants.2 The danger is that the best mutant for a novel conversion may have lost the ability to accept glycine and/or acetaldehyde and would therefore be missed in such a native activity “prescreen.” It is also more challenging to devise growth-based highthroughput assays that directly interrogate stereoselectivity (both enantioselectivity and diastereoselectivity) without resorting to GC or HPLC analyses. In such cases, library sizes are practically limited to <1000 members unless pooling/deconvolution strategies are employed (Bougioukou, Kille, Taglieber, & Reetz, 2009).

2

In these cases, the library size is limited primarily by transformation efficiency.

94

Sarah E. Franz and Jon D. Stewart

5.1. Improving catalytic activity Lee and coworkers devised a growth-based selection for threonine aldolases with greater catalytic efficiencies based on the observation that aldehydes such as acetaldehydes depress the growth rate of E. coli (Lee, Kang, & Lee, 2003). In the synthetic direction, threonine aldolases consume aldehydes; by depleting the local medium of the toxic substrate, transformed E. coli cells grow at correspondingly faster rates (positive selection). In principle, this is a generally applicable strategy that should allow the most active threonine aldolase variants to predominate. The authors also considered a negative selection based on an analogous strategy (using the acetaldehyde from threonine degradation to inhibit cell growth). Unfortunately, this proved impossible to implement in practice since the levels of acetaldehyde never rose to toxic levels. The positive selection strategy was applied to P. aeruginosa L-threonine aldolase with the goal of increasing its catalytic efficiency (Lee et al., 2003). Error-prone PCR yielded a library of ca. 20,000 colonies that were grown in the presence of 20 mM acetaldehyde under conditions where the cloned threonine aldolase variants were overexpressed. The initial selection provided 10 hits, and plasmid DNA was isolated from each. After retransformation, 5 of the 10 plasmids retained the ability to confer highlevel acetaldehyde resistance and these were further characterized. The best variant exhibited a twofold improvement in catalytic activity as compared to the wild type. Despite the relatively modest impact on catalytic activity, the selection method devised in this study may be useful in other protein engineering studies. In cases where expanded substrate range is desired, supplementing the growth medium with both acetaldehyde and the amino donor might allow one to identify the desired variants.3 This selection scheme should also be applicable to other aldehyde acceptors, although the precise concentrations needed for cell toxicity will need to be established empirically for each substrate.

5.2. Improving thermostability As described previously, the stereoselective synthesis of 3,4dihydroxyphenylserine has been an important synthetic application for 3

In practice, it may also be necessary to reduce glycine levels to favor reaction with the amino donor of interest. This can be accomplished by employing a stringent glycine auxotroph and supplementing the growth medium with limiting concentrations of glycine. Hilvert and coworkers have developed one example of a glycine auxotrophic host strain that might be useful in this strategy (Giger, Toscano, Bouzon, Marlie`re, & Hilvert, 2012).

Threonine Aldolases

95

threonine aldolases. Baik and coworkers cloned and expressed an L-threonine aldolase from S. coelicolor A3(2) as the basis for their strategy (Baik, Yoshioka, Yukawa, & Harayama, 2007). While the wild-type enzyme provided an acceptable reaction rate and stereoselectivity, its longevity under process conditions was too low for practical use. Error-prone PCR was therefore used to introduce random changes throughout the entire length of the protein. Approximately, 15,000 clones were individually screened for the ability to degrade L-threonine after a 65  C heat treatment step (using a colorimetric assay for the acetaldehyde by-product). This search yielded eight variants that appeared to be more thermostable than the wildtype enzyme; four were chosen for additional studies. All four had only single amino acid changes and all four changes were unique. The best mutant (His 177 Tyr) retained 86% of the original activity after 20 min at 60  C; under similar conditions, the wild type retained only 11%. Importantly, greater thermostability was not achieved at the expense of catalytic activity, and the His 177 Tyr variant had essentially the same steady-state kinetic values as the parent protein. In whole-cell format, the improved variant performed at the same level for 20 successive batch reactions and provided a final product concentration of 4 g/l. Because the targeted level of enzyme improvement was reached after one generation of mutagenesis and selection, the beneficial mutations were not examined combinatorially nor were additional random mutations added to the best first-generation variants. In the absence of data, one must speculate as to whether the observed thermostabilities are the best that can be reached. The major drawback to the final process is that the molar yield of the final product was only 0.7%, based on the aldehyde added (glycine was present in vast excess).

5.3. Improving stereoselectivity Enzymes are by nature homochiral catalysts, and the ability to direct reactions into single product enantiomers is one of their most important attributes. As noted previously, threonine aldolases create two adjacent stereocenters, and as a general rule, these enzymes are highly selective at the a-carbon. By contrast, they often have relatively lower diastereoselectivity, which is manifest by a mixture of configurations at the b-carbon. Reversibility also plays a role in governing stereoselectivity. For any chiral center, a racemic mixture is always the thermodynamic minimum. When more than one stereocenter is present (as is the case for

96

Sarah E. Franz and Jon D. Stewart

threonine aldolases), a diastereomeric mixture of enantiomers nearly always occurs at equilibrium. While the precise composition depends on the actual product structure, it is rare that a single diastereomer predominates at equilibrium for most synthetically interesting targets. For these reasons, it is almost always essential that preparative reactions be carried out under kinetically controlled conditions and the reverse (retro-aldol) reaction should be avoided as much as possible. The main drawback is that the yield of the desired product is almost always low when reactions are limited to far-fromequilibrium conditions. Several years ago, the Griengl group carefully analyzed the properties of representative threonine aldolases (Fesko, Reisinger, et al., 2008). This study included members of all four available classes: high-specificity L-threonine, L-allo-threonine, and D-threonine types and a low-specificity L-threonine type. The formation of phenylserine from glycine and benzaldehyde was chosen as the model system. As expected, all four threonine aldolase types yielded the same thermodynamic mixture of products (60:40 syn:anti) after extended reaction times. Of the four classes, only the high-specificity D-threonine aldolase provided high diastereoselectivity under kinetic conditions in the early phase of the reaction; the rest gave mixtures of products from the start. Interestingly, the high-specificity D-threonine aldolase reached the equilibrium product mixture only after 5 days (compared to 5 h for the other three aldolase types). This observation is critical since it demonstrates conclusively that high diastereoselectivity combined with good product yield is possible using threonine aldolases. Put another way, Griengl’s study showed conclusively that rapid b-carbon epimerization is not an intrinsic flaw of threonine aldolases. This implies that the problem can be removed from other threonine aldolases using the appropriate protein engineering. Griengl and coworkers subsequently carried out an NMR study to understand why some threonine aldolases catalyzed rapid product epimerization at the b-carbon while others did not (Fesko, Reisinger, et al., 2008). 13C-Labeled syn-product was mixed in a 60:40 ratio with unlabeled anti-product in the presence of glycine, benzaldehyde, and enzyme that matched their equilibrium values. Based on the known chemical mechanism for threonine aldolases, the relevant species can be deduced (Scheme 3.2). Both b-carbon epimerization and the back-reaction to free glycine proceed via the cofactor-stabilized anion that results from retro-aldol cleavage of the syn-product external aldimine. The NMR study yielded the relative rates of 13C label transfer from the syn-product to the anti-product and to free glycine, which can also be described as the partition ratio for the

97

Threonine Aldolases

External aldimine formation / breakdown OH

HO CO2

**

NH3

H

H O

HO CO2

**

NH3

C–C bond cleavage / formation

O P N H

CH3

OH

CO2

**

N

H

CH3

H

H

CO2

**

H O CH3

CO2

Internal aldimine formation / breakdown

O

O H

**

H O

Aldehyde binding / dissociation

O P N H

CO2

**

H

N

H O CH3

N

O P

H3N

CO2

* *

N H

N

O P

Partition ratio

N H

Scheme 3.2 Threonine aldolase kinetic pathway.

cofactor-stabilized anion. In comparing data from the four threonine aldolases, three yielded partition ratios ranging from 0.5 to 20 (epimerization/ back-reaction). As expected, these values correlated with the initial diastereoselectivities of the reactions. Because it was not possible to measure microscopic rate constants under the experimental conditions, the relative net rate constants could not be further decomposed into the individual contributions from the multiple steps that occur in each branch. It might be possible to measure the relative contributions of proton transfer steps and aldehyde binding/release by incorporating additional isotopic labels. This would provide extremely valuable guidance for future protein engineering studies by focusing the improvements on the most relevant step(s) in the reaction pathway.

5.4. Introducing and optimizing threonine aldolase activity into a novel scaffold In 2003, the Hilvert group reported that a single amino acid substitution was sufficient to convert a PLP-dependent alanine racemase from G. stearothermophilus into a threonine aldolase (Seebeck & Hilvert, 2003). The specific mutation (Tyr 265 to Ala) was designed to allow a histidine side chain (His 83) to act as an acid–base group for oxyanion protonation/ deprotonation in the aldol reaction. In addition, the smaller Ala side chain created additional active-site volume to allow larger substrates to bind. While the catalytic activity was relatively modest when assessed against the standard glycine/benzaldehyde benchmark reaction, the mutant was more than 5 orders of magnitude more efficient than the starting racemase. The mutant enzyme was also highly selective for the D-configuration at the a-carbon. The preference, if any for b-carbon stereochemistry, was not reported.

98

Sarah E. Franz and Jon D. Stewart

In a follow-up study, the ability of the Tyr 265 Ala mutant to accept a-methyl substrates was assessed (Seebeck, Guainazzi, Amoreira, Baldridge, & Hilvert, 2006). This is an important application since aldol condensations that yield quaternary centers remain particularly challenging. Steady-state kinetic constants were measured for three substrates (Scheme 3.3). Interestingly, the steady-state kinetic values for a-hydrogen and a-methyl (2R,3S)-diastereomers were very similar, implying that steric bulk at the a-carbon was well tolerated. This may be a consequence of the evolutionary heritage of the enzyme, which originally bound Ala. The D-anti analog had a ca. 10-fold lower kcat value, but KM was also decreased by a similar extent so that the kcat/KM ratio was nearly the same. Given the large number of microscopic rate constants in the catalytic cycle of threonine aldolases, it is very difficult to determine which individual step(s) was impacted and a deeper understanding will require future pre-steady-state kinetic investigations. To support future protein engineering studies, the Hilvert group recently developed a new growth-based selection system for threonine aldolases (Giger, Toscano, Bouzon, Marlie`re, & Hilvert, 2012). Rather than target threonine, their strategy uses an engineered E. coli strain with four simultaneously inactivated genes essential for glycine biosynthesis. This “clean” glycine auxotroph can only grow in minimal medium when supplemented with the amino acid or when retro-aldol activity by a cloned threonine aldolase yields glycine. The advantage of this selection is that the substrate of interest can be directly interrogated. This includes selection for stereoselectivity if a diastereomerically pure threonine analog added to the growth medium. The systems’ only limitation is that glycine must be the amino acid partner in the aldol/retro-aldol reaction. The utility of this screen was demonstrated by creating a library of simultaneous random amino acid replacements at four positions in a previously uncharacterized L-threonine aldolase from Caulobacter crescentus CB15. The data revealed that only one of the four amino acids (His 91) was absolutely essential for catalytic activity. H

HO H

HO

H

H

OH

CO2

CO2

CO2

NH3

CH3 NH3

CH3 NH3

(2R,3S) D-syn

(2R,3S) D-syn

(2R,3R) D-anti

kcat

5.7/min

2.8/min

0.16/min

KM

8.5 mM

2.8 mM

0.14 mM

Scheme 3.3 Aldol products from a mutant Ala racemase.

Threonine Aldolases

99

6. CONCLUSIONS AND FUTURE OUTLOOK Threonine aldolases have been clearly established as useful catalysts for asymmetric organic synthesis. Their ability to control the stereochemistry at the a-carbon is excellent and they accept a diverse array of acceptor aldehydes. On the other hand, these enzymes have several drawbacks that must be overcome before they can be employed routinely. The lack of stereochemical control at the b-carbon is a significant problem that detracts from synthetic utility. As Griengl’s work has shown, it is possible to decouple Cb epimerization from retro-aldol cleavage. Enhancing these properties is an obvious target for protein engineering efforts. The establishment of several selection/screening methodologies for threonine aldolases should simplify these studies. The other major drawback of threonine aldolases is that high substrate concentrations are usually needed to drive the conversion to products (and avoid equilibrating conditions that erode diastereomeric purities). While glycine is inexpensive, many aldehydes of synthetic interest do not share this trait and this practically limits the range of usable substrates. One possibility is to employ coupled enzyme systems that further convert the aldol product in an effectively irreversible reaction, for example, by lipase-mediated acylation or redox conversions. Similar strategies have proven quite useful in transaminations and may provide inspiration to this field as well (for a recent review, see Tufvesson et al., 2011 and references therein).

REFERENCES Baik, S.-H., Yoshioka, H., Yukawa, H., & Harayama, S. (2007). Synthesis of L-threo-3,4dihydroxyphenylserine (L-threo-DOPS) with thermostabilized low-specific L-threonine aldolase from Streptomyces coelicolor A3(2). Journal of Microbiology and Biotechnology, 17, 721–727. Bougioukou, D. J., Kille, S., Taglieber, A., & Reetz, M. T. (2009). Directed evolution of an enantioselective enoate-reductase: Testing the utility of iterative saturation mutagenesis. Advanced Synthesis and Catalysis, 351, 3287–3305. Brovetto, M., Gamenara, D., Mendez, P. S., & Seoane, G. A. (2011). C-C bond forming lyases in organic synthesis. Chemical Reviews, 111, 4346–4403. Clapes, P., Fessner, W. D., Sprenger, G. A., & Samland, A. K. (2010). Recent progress in stereoselective synthesis with aldolases. Current Opinion in Chemical Biology, 14, 154–167. di Salvo, M., Remesh, S. G., Vivoli, M., Ghatge, M. S., Paiardini, A., D’Aguanno, S., et al. (2014). On the catalytic mechanism and stereospecificity of Escherichia coli L-threonine aldolase. FEBS Journal, 281, 129–145. Du¨ckers, N., Baer, K., Simon, S., Gr€ oger, H., & Hummel, W. (2010). Threonine aldolases— Screening, properties and applications in the synthesis of non-proteinogenic b-hydroxya-amino acids. Applied Microbiology and Biotechnology, 88, 409–424.

100

Sarah E. Franz and Jon D. Stewart

Fesko, K., Giger, L., & Hilvert, D. (2008). Synthesis of b-hydroxy-a-amino acids with a reengineered alanine racemase. Bioorganic and Medicinal Chemistry Letters, 18, 5987–5990. Fesko, K., Reisinger, C., Steinreiber, J., Weber, H., Schu¨rmann, M., & Griengl, H. (2008). Four types of threonine aldolases: Similarities and differences in kinetics/thermodynamics. Journal of Molecular Catalysis B: Enzymatic, 52, 19–26. Fesko, K., Uhl, M., Steinreiber, J., Gruber, K., & Griengl, H. (2010). Biocatalytic access to a, a-dialkyl-a-amino acids by a mechanism-based approach. Angewandte Chemie, International Edition, 49, 121–124. Giger, L., Toscano, M. D., Bouzon, M., Marlie`re, P., & Hilvert, D. (2012). A novel genetic selection system for PLP-dependent threonine aldolases. Tetrahedron, 68, 7549–7557. Gutierrez, M. L., Garrabou, X., Agosta, E., Servi, S., Parella, T., Joglar, J., et al. (2008). Serine hydroxymethyl transferase from Streptococcus thermophilus and L-threonine aldolase from Escherichia coli as stereocomplementary biocatalysts for the synthesis of b-hydroxy-a, o-diamino acid derivatives. Chemistry—A European Journal, 14, 4647–4656. Gwon, H.-J., & Baik, S.-H. (2010). Diastereoselective synthesis of L-threo-3,4dihydroxyphenylserine by low-specific L-threonine aldolase mutants. Biotechnology Letters, 32, 143–149. Gwon, H.-J., Yoshioka, H., Song, N.-E., Kim, J.-H., Song, Y.-R., Jeong, D.-Y., et al. (2012). Optimal production of L-threo-2,3-dihydroxyphenylserine (L-threo-DOPS) on a large scale by diastereoselectivity-enhanced variant of L-threonine aldolase expressed in Escherichia coli. Preparative Biochemistry and Biotechnology, 42, 143–154. Kielkopf, C. L., & Burley, S. K. (2002). X-ray structures of threonine aldolase complexes: Structural basis of substrate recognition. Biochemistry, 41, 11711–11720. Kimura, T., Vassilev, V. P., Shen, G.-J., & Wong, C.-H. (1997). Enzymatic synthesis of b-hydroxy-a-amino acids based on recombinant D- and L-threonine aldolases. Journal of the American Chemical Society, 119, 11734–11742. Lee, S.-J., Kang, H.-Y., & Lee, Y. (2003). High-throughput screening methods for selecting L-threonine aldolases with improved activity. Journal of Molecular Catalysis B: Enzymatic, 26, 265–272. Liu, J.-Q., Dairi, T., Itoh, N., Kataoka, M., Shimizu, S., & Yamada, H. (2000). Diversity of microbial threonine aldolases and their application. Journal of Molecular Catalysis B: Enzymatic, 10, 107–115. Liu, J. Q., Dairi, T., Kataoka, M., Shimizu, S., & Yamada, H. (1997). L-allo-threonine aldolase from Aeromonas jandaei DK-39: Gene cloning, nucleotide sequencing, and identification of the pyridoxal 50 -phosphate binding lysine residue by site-directed mutagenesis. Journal of Bacteriology, 179, 3555–3560. Liu, J. Q., Odani, M., Dairi, T., Itoh, N., Shimizu, S., & Yamada, H. (1999). A new route to L-threo-3-[4-methylthiophenylserine], a key intermediate for the synthesis of antibiotics: Recombinant low-specificity D-threonine aldolase-catalyzed stereospecific resolution. Applied Microbiology and Biotechnology, 51, 586–591. Liu, J. Q., Odani, M., Yasuoka, T., Dairi, T., Itoh, N., Kataoka, M., et al. (2000). Gene cloning and overproduction of low-specificity D-threonine aldolase from Alcaligenes xylosoxidans and its application for production of a key intermediate for Parkinsonism drug. Applied Microbiology and Biotechnology, 54, 44–51. Miura, T., & Kajimoto, T. (2001). Application of L-threonine aldolase-catalyzed reaction to the preparation of protected 3R,5R-dihydroxy-L-homoproline as a mimetic of idulonic acid. Chirality, 13, 577–580. Sagui, F., Conti, P., Roda, G., Contestabile, R., & Riva, S. (2008). Enzymatic synthesis of o-carboxy-b-hydroxy-a-amino acids. Tetrahedron, 64, 5079–5084. Seebeck, F. P., Guainazzi, A., Amoreira, C., Baldridge, K. K., & Hilvert, D. (2006). Stereoselectivity and expanded substrate scope of an engineered PLP-dependent aldolase. Angewandte Chemie, International Edition, 45, 6824–6826.

Threonine Aldolases

101

Seebeck, F. P., & Hilvert, D. (2003). Conversion of a PLP-dependent racemase into an aldolase by a single active site mutation. Journal of the American Chemical Society, 125, 10158–10159. Steinreiber, J., Fesko, K., Mayer, C., Reisinger, C., Schu¨rmann, M., & Griengl, H. (2007). Synthesis of g-halogenated and long-chain b-hydroxy-a-amino acids and 2-amino-1,3diols using threonine aldolases. Tetrahedron, 63, 8088–8093. Steinreiber, J., Fesko, K., Reisinger, C., Schu¨rmann, M., van Assema, F., Wolberg, M., et al. (2007). Threonine aldolases—An emerging tool for organic synthesis. Tetrahedron, 63, 918–926. Toscano, M. D., Mu¨ller, M. M., & Hilvert, D. (2007). Enhancing activity and controlling stereoselectivity in a designed PLP-dependent aldolase. Angewandte Chemie, International Edition, 46, 4468–4470. Tufvesson, P., Lima-Ramos, J., Jensen, J. S., Al-Haque, N., Neto, W., & Woodley, J. M. (2011). Process considerations for the asymmetric synthesis of chiral amines using transaminases. Biotechnology and Bioengineering, 108, 1479–1493. Vassilev, V. P., Uchiyama, T., Kajimoto, T., & Wong, C.-H. (1995). L-threonine aldolase in organic synthesis: Preparation of novel b-hydroxy-a-amino acids. Tetrahedron Letters, 36, 4081–4084. Wong, C. H., & Whitesides, G. M. (1983). Synthesis of sugars by aldolase-catalyzed condensation reactions. Journal of Organic Chemistry, 48, 3199–3205.