Applied Microbiology and Biotechnology

, 71:253

Microbial aldolases as C–C bonding enzymes—unknown treasures and new developments


  • Anne K. Samland
    • Institut für MikrobiologieUniversität Stuttgart
    • Institut für MikrobiologieUniversität Stuttgart

DOI: 10.1007/s00253-006-0422-6

Cite this article as:
Samland, A.K. & Sprenger, G.A. Appl Microbiol Biotechnol (2006) 71: 253. doi:10.1007/s00253-006-0422-6


Aldolases are a specific group of lyases that catalyze the reversible stereoselective addition of a donor compound (nucleophile) onto an acceptor compound (electrophile). Whereas most aldolases are specific for their donor compound in the aldolization reaction, they often tolerate a wide range of aldehydes as acceptor compounds. C–C bonding by aldolases creates stereocenters in the resulting aldol products. This makes aldolases interesting tools for asymmetric syntheses of rare sugars or sugar-derived compounds as iminocyclitols, statins, epothilones, and sialic acids. Besides the well-known fructose 1,6-bisphosphate aldolase, other aldolases of microbial origin have attracted the interest of synthetic bio-organic chemists in recent years. These are either other dihydroxyacetone phosphate aldolases or aldolases depending on pyruvate/phosphoenolpyruvate, glycine, or acetaldehyde as donor substrate. Recently, an aldolase that accepts dihydroxyacetone or hydroxyacetone as a donor was described. A further enlargement of the arsenal of available chemoenzymatic tools can be achieved through screening for novel aldolase activities and directed evolution of existing aldolases to alter their substrate- or stereospecifities. We give an update of work on aldolases, with an emphasis on microbial aldolases.


The making and breaking of carbon–carbon (C–C) bonds is at the heart of synthetic organic chemistry, with the aim to synthesize larger and more complex compounds out of small and simple starting materials. C–C-bonding—especially in the synthesis of carbohydrates and sugar-derived compounds—using the arsenal of organic chemistry is often hampered by the necessity of using tedious and time-consuming methods of iterative protection and deprotection steps, taking care of the various hydroxyl groups which are present in carbohydrates. Moreover, stereospecificities of organic chemical syntheses of carbohydrates are often less than satisfactory.

In living cells, C–C splitting and bonding steps are essential in both catabolism and anabolism of a wide variety of carbohydrates and of some keto acids. Thus, nature provides various enzymes which form or cleave C–C bonds. These enzymes have attracted the interest of synthetic chemists in recent years, and an increasing number of reports can be seen which especially deal with the synthesis of complex molecules like (rare) sugars and sugar-derived compounds.

Enzymes provide several advantages compared to organic chemistry: (1) They can be highly specific for their substrates (chemospecific). No side products are formed. (2) They are stereospecific. (3) They are regiospecific. No protection chemistry is needed. (4) Reactions can be performed under mild conditions (ambient temperature and pressure, no extreme pH, no toxic solvents, and no heavy metal catalysts).

For asymmetric C–C bonding, mainly aldolases, transferases like transaldolases and transketolases, thiamine diphosphate-dependent enzymes (for acyloin condensations), and oxynitrilases are used. Of these, aldolases are the most important group of asymmetric C–C-bonding enzymes, which are currently used in bio-organic chemistry. Aldolases are a specific group of lyases that catalyze the reversible stereo-selective addition of a donor compound (nucleophile) onto an acceptor compound (electrophile). While the donor compound for aldolases is mostly invariable (see Fig. 1 for details), the acceptor(s) can vary in chain length from C1 to C6, as long as they fulfill the minimum requirement of e.g., an (α-hydroxy)-aldehyde (Rose and O’Connell 1969; Machajewski and Wong 2000). The utilization of aldolases for catalytic asymmetric aldol reactions in chemoenzymatic syntheses has been reviewed in-depth over the years, mostly from the viewpoint of bio-organic chemists (for reviews and further references, see Horecker et al. 1972; Wong and Whitesides 1983; Bednarski et al. 1989; Toone et al. 1990; Toone and Whitesides 1991; Straub et al. 1990; Wong 1995; Fessner and Walter 1997; Takayama et al. 1997; Fessner 1998; Machajewski and Wong 2000; Fessner and Helaine 2001; Silvestri et al. 2003; Breuer and Hauer 2003; Sukumaran and Hanefeld 2005). The present contribution therefore intends to review more recent advances in the use of microbial aldolases, with an emphasis on the literature from the year 2000 onwards.
Fig. 1

Overview of the main groups of aldolases classified by their specific donor substrate. FBP represents fructose 1,6-bisphosphate; TBP represents tagatose 1,6-bisphosphate; FucA represents fuculose 1-phosphate aldolase; RhaA represents rhamnulose 1-phosphate aldolase; FSA represents fructose 6-phosphate aldolase; KDPG represents 2-keto-3-deoxy-6-phosphogluconate; KDO8P represents 3-deoxy-d-manno-octulosonate 8-phosphate; DAHP represents 3-deoxy-d-arabino-heptulosonate 7-phosphate; NeuAc represents neuraminic acid; DERA represents 2-deoxy-d-ribose 5-phosphate aldolase; SHMT represents serine hydroxymethyl transferase. Organic residue (R), H or phosphate group (R1), phosphate group (P), newly formed stereocenter(s) by C–C bond formation (asterisks)

Different classes of aldolases

Up to now, more than 30 aldolases are known (Machajewski and Wong 2000), and as far as we know, most—if not all—organisms contain aldolases (EC 4.1.2.x). They are usually categorized based on their requirement for nucleophiles (donor compounds in aldolization reactions; see Fig. 1). Dihydroxyacetone phosphate (DHAP), glycine, acetaldehyde, pyruvate, and phosphoenolpyruvate (PEP) can serve as donors. So-called class I aldolases (Rutter 1964) need no cofactor, as their reaction mechanism of donor activation involves the formation of a Schiff base at a conserved lysine residue in the active site of the enzyme (Horecker et al. 1972; Gefflaut et al. 1995; Machajewski and Wong 2000). Class II aldolases are dependent on a metal ion (mainly Zn2+, in some cases also Fe2+ or Co2+), which acts as a Lewis acid and activates the donor substrate. Therefore, class II aldolases can be inhibited by EDTA. They are often more stable than class I aldolases, which is important for synthetic purposes. Class II aldolases occur only in prokaryotes or lower eukaryotes (Gefflaut et al. 1995).

Until recently, class I aldolases were thought to occur mainly in higher eukaryotes (Machajewski and Wong 2000). However, more recent findings of a variety of class I aldolases in prokaryotes have challenged this opinion (references in Schürmann and Sprenger 2001; Siebers et al. 2001; Thomson et al. 1998). Interestingly, class I and II fructose 1,6-bisphosphate (FBP) aldolases can occur in the same organism, as shown for Escherichia coli (Thomson et al. 1998). This means that the same reaction product can be formed by apparently unrelated enzymes that show hardly any sequence identity to each other. Thus, the two classes of aldolases can be regarded as examples for convergent enzyme evolution (Rutter 1964; Marsh and Lebherz 1992).

Although aldolases are highly specific for their donor substrate (see Fig. 1), the variability in the acceptor substrate makes it possible to use them for the synthesis of more than one aldol product. An additional advantage is their stereospecificity, which results in the formation of one stereoisomer only and is difficult to control in organic chemical syntheses. In most cases, the stereospecificity is determined by the enzyme, and therefore the configuration of the product is highly predictable.

DHAP-dependent aldolases

DHAP-dependent aldolases constitute the most important group with respect to their use in biocatalytic applications. The most prominent example of a DHAP-dependent aldolase is the “textbook” enzyme, FBP aldolase or “aldolase” (or FruA; EC In glycolysis, it catalyzes the cleavage of FBP into two three-carbon fragments, DHAP and glyceraldehyde 3-phosphate (GAP). But “aldolase” works in both glycolysis and gluconeogenesis of most organisms. If provided with its donor compound, DHAP, aldolase catalyzes the formation of FBP, with GAP as the accepting aldehyde (Meyerhof et al. 1936). As the equilibrium of this reaction lies strongly on the side of aldol formation, the enzyme can be utilized in vitro for synthetic purposes.

FBP aldolase is highly specific for DHAP, but accepts a large variety of acceptor aldehydes (Rose and O’Connell 1969; Bednarski et al. 1989), as do other DHAP aldolases. In general, unhindered aliphatic aldehydes and α-heteroatom-substituted aldehydes, including different aldoses (C3–C5), are suitable substrates. Phosphorylated substrates are usually preferred over nonphosphorylated ones (Rose and O’Connell 1969; Bednarski et al. 1989; Machajewski and Wong 2000). The aldol reaction thus leads to an elongation by three carbon atoms and the formation of the respective 1-phosphate of α-ketosugars.

DHAP-dependent aldolases create two new stereogenic centers with (usually) excellent enantio- and diastereo-selectivities (Schoevaart et al. 2000a). While four stereoconfigurations at the C-atoms 3 and 4 are principally possible, FBP aldolase is strictly specific for the 3S,4R stereoconfiguration. The other three possible stereoisomers at C-atoms 3 and 4 can be obtained by other DHAP aldolases (see below, Fig. 2).
Fig. 2

Complementary stereochemistry of DHAP-dependent aldolases. Organic residue (R), phosphate group (P)

Traditionally, rabbit muscle aldolase (RAMA) has been most widely investigated for structure–function relationships and is in use for chemoenzymatic syntheses, as the enzyme is commercially available (Bednarski et al. 1989; Toone and Whitesides 1991; Gefflaut et al. 1995; Machajewski and Wong 2000). RAMA is a class I aldolase and forms a tetrameric structure. RAMA forms sugars and sugar analog compounds with the same strict 3S,4R stereoconfiguration as it does with its natural substrates. In vitro, RAMA can utilize various (α-hydroxy)-aldehydes, including sterically nonhindered aliphatic aldehydes, aldehydes with substitutions at the α-heteroatom (Bednarski et al. 1989), and monosaccharides and their derivatives (Machajewski and Wong 2000). Aromatic aldehydes, sterically hindered aldehydes, and α,β-unsaturated aldehydes are usually not substrates (Bednarski et al. 1989).

Due to the fact that RAMA and various FBP aldolases from microbial origins catalyze (rather) stereo- and regio-specific aldolization reactions with a given specific donor compound (e.g., DHAP), but allowing a rather promiscuous range of aldehyde acceptor substrates, they have already been used as biocatalysts for a series of preparative organic syntheses in vitro (see below).

Microbial DHAP aldolases

The class II FBP aldolase from E. coli is commercially available (Fessner 1998); this aldolase appears to have a similar acceptor substrate range as RAMA, but is operationally more stable. The E. coli enzyme forms a dimer, which is typical for class II aldolases. Additionally, class I microbial DHAP aldolases have been described in recent years. The FBP aldolase from Staphylococcus carnosus is a class I aldolase but forms a stable monomer; it has been described as a useful biocatalyst (Zannetti et al.1999; Dinkelbach et al. 2001). Recently, archaeal class I FBP aldolases were described (Lorentzen et al. 2003; Siebers et al. 2001) which show similarity to the class I FBP aldolase from E. coli (Thomson et al. 1998).

From catabolic pathways for the rare sugars l-fucose, l-rhamnose, or d-tagatose/galactitol/N-acetylgalactosamine (tagatose 1,6-bisphosphate pathway), aldolases are known which also utilize DHAP as a donor (Fig. 1) but result in the other three possible stereoconfigurations at carbon atoms C3 and C4 of the respective aldol products (Fig. 2). This means that l-fuculose 1-phosphate aldolase (EC forms the 3R,4R (Ozaki et al. 1990; Fessner et al. 1991) and l-rhamnulose 1-phosphate aldolase (EC the 3R,4S stereoconfiguration. Finally, tagatose 1,6-bisphosphate aldolase (EC builds preferentially 3S,4S bonds. Although this enzyme tends to form mixtures of diastereomers (references in Machajewski and Wong 2000), which is a certain drawback for its use in chemoenzymatic syntheses.

However, in theory, as these four aldolase enzymes are available in recombinant form, all four stereoconfigurations can be obtained in syntheses. Similar to the FBP aldolase, these other three DHAP-dependent aldolases accept a wide range of aldehyde acceptor compounds (Fessner et al. 1991; Garcia-Junceda et al. 1995; Fessner 1998; Schoevaart et al. 2000a; Fessner and Helaine 2001).

Another DHAP aldolase may be involved in the catabolism of myo-inositol in Bacillus subtilis and Klebsiella strains. The enzyme IolJ presumably uses DHAP and malonic semialdehyde as acceptor (Yoshida et al. 2004). However, the use of this interesting enzyme in bio-organic syntheses has not been evaluated so far, and its substrate spectrum is unknown.

Synthetic uses for DHAP aldolases

For synthetic organic chemists, the natural aldol reaction of DHAP aldolases is only of limited interest. Using simple chemically synthesized aldehydes (formaldehyde, glycolaldehyde, or d-glyceraldehyde), short-chain sugar phosphates can be obtained (e.g., erythrulose 1-phosphate, xylulose 1-phosphate, fructose 1-phosphate; Bednarski et al. 1989; Gefflaut et al. 1995). If labeled acceptors are used, e.g., deuterated, 13C-, or 14C-labeled, the respective labeled sugar phosphates can be formed, which can be used for biosynthetic labeling or transport experiments. The 1-phosphate group can be removed by the use of alkaline or acidic phosphatase (Bednarski et al. 1989; Espelt et al. 2003; Guanti et al. 2000) to obtain the unphosphorylated product, which is the desired product in most applications.

As long as the acceptor compound is not too sterically hindered, a variety of heteroatoms can be introduced, e.g., amino, nitro, azido, phosphono, sulfo, chloro, fluoro, deoxy, or other groups. This allows the chemoenzymatic synthesis of the respective sugars or sugar analogs carrying heteroatoms (amino, nitro, azido, phosphono, sulfo etc.). Subsequent chemical steps can deliver iminocyclitols (azasugars), or educts for further chemical steps (Bednarski et al. 1989; Azéma et al. 2000; Guanti et al. 2000; Zhu and Li 2000; Guanti et al. 2001; Mitchell et al. 2001; Schuster et al. 2001).

The resulting products can then be of use both in basic science and for applied purposes. For example, substrate analogs containing a phosphonate group can help to elucidate structure–function relationships in protein crystal structures. Various C6-labeled dansyl fructose analogs were obtained by aldolases; these analogs can be used in inhibition studies of the glucose transport of Trypanosoma brucei, which causes sleeping disease (Azéma et al. 2000). Other applications are the inhibition of sugar-converting enzymes as glycosidases and glycosyltransferases. Iminocyclitols (azasugars) are a quite important group of compounds available with DHAP-dependent aldolases, as iminocyclitols are potential inhibitors of glycosidases. Thus, they find applications as new antibiotics, antimetastatic, antihyperglycemic, or immunostimulating agents. The synthesis of iminocyclitols from DHAP and N-benzyloxycarbonyl-aminoaldehydes using different DHAP-dependent aldolases in emulsion systems has been described recently (Fig. 3a; Espelt et al. 2003, 2005). Others (Mitchell et al. 2001) report on the synthesis of iminocyclitols, which contain a phosphonate group; this phosphonate might mimic the pyrophosphate of substrates of glycosyltransferases and glycosidases. Along the same line, the synthesis of phosphonylated dihydroxypyrrolidines was reported (Fig. 3a; Schuster et al. 2001) as potential inhibitors for fucosyltransferase. Romero and Wong (2000) describe the synthesis of tetrahydroxylated pyrrolizidine alkaloids by a stereoselective aldol reaction and a subsequent bis-reductive amination.
Fig. 3

Applications of aldolases in chemoenzymatic synthesis. The examples shown are: a DHAP-dependent aldolases fuculose 1-phosphate aldolase, RAMA, and rhamnulose 1-phosphate aldolase in the synthesis of iminocyclitols (Schuster et al. 2001; Espelt et al. 2003, 2005) and complex ring structures like the microbial elicitors (−)-syringolides (Chenevert and Dasser 2000), and pancratistatin (Phung et al. 2003). b An important application of DERA is the sequential aldol reaction to statins like epothilone A and atorvastatin (DeSantis et al. 2003). c NeuAc aldolase. Organic residue (R), phosphate group (P), newly formed stereocenter(s) by C–C bond formation (asterisks)

The binding of Sialyl Lewis X to selectins is important for cell–cell recognition in eukaryotes. l-Fucose analogs can mimic these interactions and can be used as potential inhibitors for selectins. Fessner et al. (2000) have synthesized such analogs. Also mannosylphosphates can mimic for Sialyl Lewis X (Lin et al. 1999). Guanti et al. (2000, 2001) describe the synthesis of ω-phosphonic deoxysugars, which are valuable chiral building blocks for the synthesis of various phosphorus-containing analogs of biologically active substances. Also, branched-chain ketoses can be synthesized starting from achiral substances (David 1999). Zhu and Li (2000) produced perfluoroalkylated sugars. Due to their amphiphilic nature, these compounds could be used as surfactants and emulsifiers for biomedical applications. Even more complex substances, like the microbial elicitors (−)-syringolides (Chenevert and Dasser 2000) and pancratistatin (Phung et al. 2003), can be obtained with the help of aldolases (Fig. 3a). Crestia et al. (2001, 2004) have synthesized 3-deoxy 2-ulosonic acids, like sialic acid, 3-deoxy-d-manno-octulosonate (KDO), 3-deoxy-d-arabino-heptulosonate, and derivatives thereof. These can also be produced by pyruvate- or PEP-dependent aldolases (see below). But in contrast to those, RAMA and transketolase form the C5–C6 bond. This allows for a greater variability at C4.

Limitations and drawbacks of DHAP-dependent aldolases

As shown above, DHAP-dependent aldolases are now widely used, at least in academic laboratories. However, their application still suffers from some serious limitations and drawbacks. As they are specific for DHAP as the donor substrate [dihydroxyacetone (DHA) is a very poor donor; Gefflaut et al. 1995], the aldol reaction leads always to a phosphorylated product, which then requires a dephosphorylation step if an unphosphorylated product is desired. Moreover, due to the similar physicochemical properties of DHAP and sugar phosphates, purification of the product by chromatographic steps is hampered. In addition, the product is always a ketose, while some of the most desired products would be the corresponding aldehyde. This problem at least can be solved by the subsequent use of isomerases (Durrwachter et al. 1986).

The main problem, however, is DHAP itself, as it is expensive, instable, and difficult to synthesize. While DHAP is a regular metabolite of glycolysis, its shear price as commercial compound is prohibitive for applications beyond academic interest. Knowing this, advances to provide DHAP by chemical synthesis or by multienzyme approaches have recently been made (for references, see Fessner and Walter 1992; Machajewski and Wong 2000; Schoevaart et al. 2000b; Franke et al. 2003). Schoevaart et al. (1999) describe the synthesis of 5-deoxy-5-ethyl-d-xylulose where DHAP is synthesized in situ by a phytase and glycerol 3-phosphate oxidase. The reaction is controlled by pH. By this means, the phytase can be used again to dephosphorylate the product at the end. Sanchez-Moreno et al. (2004) describe a more straightforward way, where DHA is phosphorylated by a kinase and ATP is recycled using an acetate kinase.

Another possibility is to circumvent the use of DHAP by taking transaldolase (TAL), which is a transferase from the pentose phosphate pathway. It catalyzes the transfer of an activated dihydroxyacetone moiety (C3) from a donor ketose, fructose 6-phosphate in the natural reaction, to an acceptor aldehyde, erythrose 4-phosphate (Fig. 1). Sedoheptulose 7-phosphate and fructose also serve as donors of a dihydroxyacetone moiety, which can be transferred to essentially the same acceptors as with RAMA and yielding the 3S,4R aldol products (Horecker et al. 1972; Sprenger et al.1995; Schörken 1997). In recent years, transaldolases from different microbial origins as E. coli (Sprenger et al. 1995), B. subtilis, Thermotoga maritima (Schürmann and Sprenger 2001), Methanocaldococcus jannaschii (Soderberg and Alver 2004), and others have been characterized. All contain a conserved lysine residue in the active site and belong to the class I aldolases. TalB of E. coli forms a dimer, whereas the B. subtilis enzyme forms a decamer (Schürmann and Sprenger 2001). The B. subtilis enzyme is more stable at higher temperatures than TalB of E. coli.

However, there are several disadvantages. Presently, no transaldolases that can be used to provide the other three possible stereoconfigurations are known. Moreover, the transaldolase reaction starts with two educts (donor and acceptor) and yields two products that are in equilibrium. Products and educts have similar physicochemical properties, and hence, are difficult to separate from each other. Finally, dihydroxyacetone as a donor for transaldolases is used only at a marginal activity (Horecker et al. 1972; Samland and Sprenger, unpublished observations).

Fructose 6-phosphate aldolase uses dihydroxyacetone as donor

Is there no aldolase around which can directly use dihydroxyacetone at decent rates for aldolization? Recently, two fructose 6-phosphate aldolase isoenzymes (“TALC,” FSA) were reported from the E. coli genome (Schürmann and Sprenger 2001). Initially annotated as transaldolase isoenzymes, these enzymes were found to actually split fructose-6-P into glyceraldehyde-3-P and DHA. However, the reverse reaction is performed at even higher rates. With DHA as the donor, the aldolization product yields the 3S,4R configuration. The physiological role of both isoenzymes remains unclear. But, in vitro FSA was successfully used as a tool to provide rare sugars from the condensation of DHA with various aldehydes (Schürmann et al. 2002). As dihydroxyacetone can be utilized as a donor, nonphosphorylated products can be achieved. Interestingly, DHAP is not a donor. Moreover, also hydroxyacetone (acetol) can act as a donor. This allows the synthesis of the novel class of 1-deoxy-sugars as aldolization products (Schürmann et al. 2002).

FSA is a thermostable decameric enzyme and shows structural and sequential similarity to the transaldolases of B. subtilis and T. maritima (Schürmann and Sprenger 2001; Thorell et al. 2002).

Microbial PEP- and Pyruvate-dependent aldolases

Aldolases, which use pyruvate or PEP (Fig. 1), are almost exclusively of microbial origin. Pyruvate-dependent aldolases mostly have catabolic functions in vivo, while PEP-dependent enzymes are involved in the biosynthesis of α-keto acids. Both catalyze an elongation by three carbon atoms. PEP- and pyruvate-dependent aldolases form or split 3-deoxy-2-keto acids and have recently become of increasing importance, as sialic acid derivatives are used for cancer therapy and as anti-infectives. Both PEP- and pyruvate-dependent aldolases are not strictly acceptor-specific. Thus, a wide range of deoxy-sugars or sugar acids becomes synthetically available (Allen et al. 1992; Shelton et al. 1996; Henderson et al. 1998; Sundaram and Woodard 2000).

Aldolases which use pyruvate as donor substrate are N-acetylneuraminic acid (NeuAc) aldolase, 2-keto-3-deoxygluconate (KDG) aldolase, 2-keto-3-deoxy-6-phosphogluconate 6-phosphate (KDPG) aldolase, and various others which are specific for analogs of KDPG like 2-keto-3-deoxy-6-phosphogalactonate (Allen et al. 1992). The most interest is paid to NeuAc aldolase, which catalyzes the synthesis of sialic acid from pyruvate and N-acetylmannosamine (ManNAc) (Fig. 3c). As the equilibrium lies on both sides to the same extent, an excess of pyruvate has to be used to push the reaction in favor of the condensation reaction. The excess pyruvate can cause difficulties in the purification of the product. C5 and C6 aldehydes are good acceptor substrates (Wong et al. 1995). Shorter aldehydes do not work as well. The stereochemistry depends on the substrate used and hence is not as predictable as with DHAP-dependent aldolases (Allen et al. 1992).

Different approaches have been described to synthesize the expensive educt ManNAc from N-acetyl-d-glucosamine. Blayer et al. (1999) made use of the epimerization in alkali pH, whereas others (Kragl et al. 1991) applied an epimerase for this step. Also, a renin binding protein was used (Lee et al. 2004). A one-pot synthesis of (3-13C)-labeled NeuAc analogs was reported by Miyazaki et al. (2000). Here, ManNAc was gained by the retro reaction of NeuAc aldolase. The equilibrium was shifted towards the cleavage reaction as pyruvate was consumed by lactate dehydrogenase and a cofactor regenerating system. These reactions were stopped prior to the addition of (3-13C)-labeled pyruvate.

To optimize NeuAc aldolases for C4 aldehydes as substrates and the synthesis of potent influenza inhibitors like zanamivir, a directed evolution approach has been recently carried out to design novel NeuAc aldolases (Woodhall et al. 2005). The new enzymes can be used to synthesize ternary amides. Wada et al. (2003) changed the stereospecificity of NeuAc aldolase from d- to l-KDO by error-prone PCR (Fig. 4).
Fig. 4

Stereospecificity of aldolases altered by directed evolution. The groups of Berry and Wong demonstrated the usefulness of directed evolution for altering the substrate- and stereospecificity of aldolases. This is shown here for tagatose 1,6-bisphosphate aldolase (upper panel; Williams et al. 2003) and NeuAc aldolase (lower panel; Wada et al. 2003)

KDG aldolase from Sulfolobus sulfataricus uses pyruvate and glyceraldehyde as substrates. It can also take other nonphosphorylated substrates (Buchanan et al. 1999). On the other hand, it shows a lack of specificity and synthesizes KDG and KDGal (Lamble et al. 2003). Stereoselectivity could be introduced, however, by using acetonide-substituted aldehyde substrates (Lamble et al. 2005b). By directed evolution, KDPG aldolase was optimized to accept nonphosphorylated substrates and change its selectivity from d- to l-glyceraldehyde (Fong et al. 2000).

PEP-dependent aldolases transfer the enolpyruvyl moiety to acceptors like arabinose 5-phosphate (3-deoxy-d-manno-octulosonate-8-phosphate synthase; KDO8P synthase) or erythrose 4-phosphate [3-deoxy-d-arabino-heptulosonate-7-phosphate synthase; (DAHP synthase)], the initial step of the general aromatic amino acid pathway in eubacteria (shikimic acid pathway). Substrate ambiguities are known for DAHP synthases from E. coli (Sundaram and Woodard 2000) and Pyrococcus furiosus (Schofield et al. 2005). DAHP synthase can be utilized for the synthesis of different monosaccharides (Sundaram and Woodard 2000). Recently, by directed evolution of KDPGal aldolases from enteric bacteria, a mutant pyruvate-dependent aldolase was selected, which, in vivo, took over the role of a DAHP synthase in an E. coli DAHP-deficient mutant (Ran et al. 2004).

2-Deoxy-d-ribose 5-phosphate aldolase, an acetaldehyde-dependent aldolase

In the group of acetaldehyde-dependent aldolases, only one enzyme is known so far, namely, 2-deoxy-d-ribose 5-phosphate aldolase (DERA) (Fig. 1), found in the DNA salvage pathway of many microorganisms. DERA catalyzes the synthesis of d-2-deoxyribose 5-phosphate from acetaldehyde and GAP (Figs. 1 and 3b) and is a class I aldolase (Machajewski and Wong 2000). The donor substrate specificity of DERA is not as strict as with the other aldolases mentioned above. Besides acetaldehyde, propanal, acetone, and fluoroacetone are accepted; therefore, a chain elongation by two or three carbon atoms is possible (Barbas et al. 1990).

Another difference compared to the other aldolases described above is the fact that both substrates and the product are aldehydes. Due to this circumstance, DERA can perform sequential aldol reactions (Fig. 3b), first reported by Gijsen and Wong (1994). Starting from three achiral C2 aldehydes, 2, 4, 6-trideoxyhexoses can be formed. The reaction is driven by the formation of a stable hemiacetal of the condensation product, which can be oxidized to the corresponding lactone. The resulting 1,3-polyols are very useful synthons and are employed, for example, in the syntheses of the anticancer drugs epothilone A and C, iminocyclitols, or statins (Machajewski and Wong 2000). Statins are inhibitors of the 3-hydroxy-3-methyglutaryl-coenzyme A reductase, which catalyzes the committed step in cholesterol synthesis, and hence, they are used as cholesterol-lowering agents. Acetaldehyde derivatives like glycolaldehyde, where the diproduct can form a stable hemiacetal, are poor substrates for the sequential aldol reaction. The sequential aldol reaction can also be combined with other aldolases like RAMA or NeuAc aldolases. A combination of DERA and RAMA allows for the synthesis of 5-deoxyketoses (Gijsen and Wong 1995). Liu and Wong (2002) investigated the range of different acceptor aldehydes that can be used for the synthesis of pyranoses in a sequential aldol reaction. These pyranoses were selectively alkylated in the synthesis of the fragments of epothilone A and C (Fig. 3b; Koeller and Wong 2001; Liu and Wong 2002).

Using a rational mutagenesis approach (DeSantis et al. 2003), the DERA mutant S238D was selected, which accepts nonphosphorylated substrates and 3-azidopropinaldehyde as a new substrate. This mutant was used for the synthesis of atorvastatin (Fig. 3b; Koeller and Wong 2001; DeSantis et al. 2003; Liu et al. 2004). Recently, Greenberg and coworkers discovered a new DERA in an environmental genomic library screen. By upscaling of the process, they achieved a product at a 100-g scale and with high enantiomeric excess (>99.9%), which should allow a more cost-effective process for statin synthesis (Greenberg et al. 2004).

Glycine-dependent aldolases

The fourth group are glycine-dependent aldolases, which form β-hydroxy-α-amino carbonic acids. Glycine-dependent aldolases work with pyridoxal phosphate (PLP) as cofactor (Fig. 1). The two known members of this group are serine hydroxymethyltransferase and threonine aldolase. Various d- and l-threonine aldolases have been analyzed for their substrate and stereospecificity and possible applications (Kimura et al. 1997; Liu et al. 2000a; Paiardini et al. 2003). β-Hydroxy-α-amino carbonic acids can be used for the synthesis of building blocks of antibiotics (vancomycin), immunosuppressants (cyclosporin), and drugs for Parkinson’s disease therapy (Kimura et al. 1997; Liu et al. 2000b). Recently, a PLP-dependent threonine transaldolase was described from Streptomyces cattleya, which is involved in the biosynthesis of the rare amino acid 4-fluorothreonine (Murphy et al. 2001).

Conclusions and perspectives

While being potentially important tools for asymmetric aldol reactions, aldolases have still not reached the mainstream chemical laboratories. Apart from more general reasons, as the insufficient supply, the price of enzymes, and their weakness in nonaqueous reaction conditions, aldolases can be regarded as not sufficient for many chemical reactions, as they require expensive donor compounds such as DHAP or PEP.

The high stereospecificity may even be regarded as a disadvantage, as it prevents the formation of the opposite stereoconfigurations, if these are of interest. Synthetic chemists often wish to have all possible stereoconfigurations of one compound as pure products. How then can the other stereoconfigurations be achieved? This can be done by the directed evolution of enzymes (Fig. 4), as nicely shown by the groups of Toone for pyruvate aldolases (Wymer et al. 2001; Griffiths et al. 2004), Fersht for NeuAc aldolase (Joerger et al. 2003), Wong for KDPG aldolases and NeuAc aldolase (Fong et al. 2000; DeSantis et al. 2003; Wada et al. 2003; Franke et al. 2004; Hsu et al. 2005), and Berry for NeuAc aldolase and DHAP aldolases (Williams et al. 2003; Hao and Berry 2004; Woodhall et al. 2005).

How can these newly acquired stereoconfigurations be made visible? Here, the method of using mock substrates like umbelliferyl-, coumaryl- or nitrophenyl-aldols may be of future worth, as they allow us to work in high-throughput systems with spectrophotometric or fluorometric multiple-well-reader detectors (Gonzalez-Garcia et al. 2003).

Which other aldolases are around? Of course, many aldolases from microbial sources are still to be discovered. Screening for novel aldolase activities has already been successful in the past when microbial sources were used (Machajewski and Wong 2000). This may reflect the high metabolic diversity in microorganisms. It may be conceived that further aldolases are to be detected from the wealth of degradation pathways that are present in bacteria. For example, recently, a phenylserine aldolase (glycine-dependent) was found in a Pseudomonas putida strain (Misono et al. 2005). This aldolase could be used to form l-phenylserine from benzaldehyde and glycine, albeit as a mixture of the threo- and erythro-forms.

Where to look for novel aldolases? Certainly, we can look for these aldolases in “weird” sugar metabolic pathways, as in the archaeal versions (Sulfolobus solfataricus, Thermoproteus tenax) of the Entner–Doudoroff pathways, which involves phosphorylated, unphosphorylated, or a mixture of phosphorylated and unphosphorylated steps (Buchanan et al. 1999; Lamble et al. 2003, 2005a). Next, of course, these aldolases can be searched for in degradation pathways of xenobiotics such as tetralin (a class II aldolase ThnF from Sphingopyxis macrogoltabida; Fig. 5; Hernaez et al. 2002), toluene (P. putida, Cho et al. 2000), or carbazole (pyruvate aldolase CarE from Pseudomonas resinovorans, Inoue et al. 2004), which, as an ultimate step, often show the splitting of an organic acid into pyruvate and an aldehyde compound similar to the ones outlined in Fig. 5. In the degradation pathway of different aromatic compounds, hydratases and aldolases are involved. These aldolases accept different, rather bulky aldehydes, but the products are not always chiral. No further investigations for their use in biocatalysis have been made (Fig. 5; Hernaez et al. 2002; Keck et al. 2002; Liu et al. 2002, 2003; Horinouchi et al. 2003; Muraki et al. 2003; Inoue et al. 2004). Several aldolases of those pathways, however, are known or presumed to be bifunctional enzymes (Hara et al. 2003). The nonaldolase function can lead to the removal of the aldehyde moiety, which would counteract aldol product formation. Examples are a bifunctional transaldolase/phosphoglucoseisomerase from Gluconobacter oxydans (Sugiyama et al. 2003). In another case, a trans-o-hydroxybenzylidenepyruvate hydratase/aldolase from a naphthalene-degrading organism was successfully used as a biocatalyst for aldol condensation of pyruvate with various benzaldehydes (Eaton 2000).
Fig. 5

Aldolases involved in the degradation of cyclic and aromatic compounds. The degradation pathways of tetralin (Hernaez et al. 2002) and aniline (Liu et al. 2002) are presented as examples. The last enzymatic step is catalyzed by pyruvate-dependent aldolases

The mining of genome data may further contribute to the arsenal of aldolases, as, due to recent advances in genetic engineering, more and more microbial sources of aldolases became available and the cloning and overexpression of these enzymes has become standard. So, microbial aldolases as unknown treasures from microbial metabolic pathways may still show their future merit in other fields, such as in bio-organic chemistry or in the pharmaceutical industry.

Copyright information

© Springer-Verlag 2006