A novel strategy for enzymatic synthesis of 4-hydroxyisoleucine: identification of an enzyme possessing HMKP (4-hydroxy-3-methyl-2-keto-pentanoate) aldolase activity


  • Editor: Sergio Casella

Correspondence: Sergey V. Smirnov, Ajinomoto-Genetika Research Institute, 1st Dorozhny pr. 1, Moscow, 113545, Russia. Tel.: +7 495 315 37 66; fax: +7 495 315 00 01; e-mail: servasmi@rol.ru


A two-step enzymatic synthesis process of 4-hydroxyisoleucine is suggested. In the first step, the aldol condensation of acetaldehyde and α-ketobutyrate catalyzed by specific aldolase results in the formation of 4-hydroxy-3-methyl-2-keto-pentanoate (HMKP). In the second step, amination of HMKP by the branched-chain amino acid aminotransferase leads to synthesis of 4-hydroxyisoleucine. An enzyme possessing HMKP aldolase activity (asHPAL) was purified 2500-fold from a crude extract of Arthrobacter simplex strain AKU 626. Sequencing of the asHPAL structural gene showed that the purified enzyme belongs to the HpcH/HpaI aldolase family. The 4-hydroxyisoleucine was synthesized in vitro from acetaldehyde, α-ketobutyrate and l-glutamate using a coupled aldolase/branched-chain amino acid aminotransferase bienzymatic reaction.


4-Hydroxyisoleucine (4HIL) is a natural nonproteinogenic amino acid possessing insulinotropic biological activity (Sauvaire et al., 1998; Broca et al., 1999, 2004). It was extracted from fenugreek seeds (Trigonella foenum-graecum) (Fowden et al., 1973) and its absolute stereo configuration was determined as (2S, 3R, 4S) (Alcock et al., 1989). 4HIL increases glucose-induced release of insulin. In contrast to several types of pharmacological drugs that have been used for the treatment of type II diabetes (e.g. sulfonylureas), the insulin response mediated by 4HIL is strictly dependent on the glucose concentration. This unique property of 4HIL allows us to avoid undesirable side-effects such as hypoglycemia in the therapy of type II diabetes (Jackson & Bessler, 1981; Jennings et al., 1989). Thus, 4HIL seems a promising dietary supplement in the treatment and prevention of this chronic disease. Because of the currently high incidence of type II diabetes and the unfavorable prognosis of its spread in the future, the development of an industrial process of 4HIL synthesis is highly timely.

Several methods of chemo enzymatic synthesis of 4HIL have been reported. A 39% overall yield of 4HIL has been achieved in an eight-step synthesis process described by Wang et al. (2002). The key step of this synthesis is the biotransformation of ethyl-2-methylacetoacetate to ethyl (2S, 3S)-2-methyl-3-hydroxybutanoate with Geotrichum candidum. A six-step chemo enzymatic synthesis process of 4HIL with total control of stereochemistry has been proposed (Rolland-Fulcrand et al., 2004). The last step of this process is the enzymatic resolution of an N-phenylacetyl lactone derivative using immobilized penicillin acylase G.

Here we have suggested the novel two-step (full enzymatic) synthesis of 4HIL from acetaldehyde, α-ketobutyrate and l-glutamate by using activities of two enzymes: 4-hydroxy-3-methyl-2-keto-pentanoate aldolase (HPAL) and branched-chain amino acid aminotransferase (BCAT) (Figs 1 and 3a). We report on the purification and identification of the enzyme possessing HPAL activity.

Figure 1.

 The ability to utilize 4HIL as a sole carbon source may be used as a ‘selection marker’ for HPAL-producing bacteria. The putative 4HIL breakdown pathway is shown. Coupling activities of HPAL (hypothetical HMKP aldolase) and BCAT (in the presence of α-ketoglutarate) results in conversion of 4HIL into a mixture of l-glutamate, acetaldehyde and α-ketobutyrate (HPAL/BCAT bienzymatic reaction). Acetaldehyde and α-ketobutyrate can be oxidized by activities of coenzyme A-(CoA) acetylating acetaldehyde dehydrogenase (ADH, EC and pyruvate dehydrogenase complex (PDH, EC, respectively. The resulting propionyl-CoA can be converted into pyruvate and succinate by activities of enzymes involved in the 2-methylcitric acid cycle (Horswill & Escalante-Semerena, 1999). As a result, all carbon atoms of the 4HIL molecule are included in intermediates of the tricarboxylic acid cycle (TCA). Thus, a microorganism that is able to utilize 4HIL as a sole carbon source is a potential producer of HPAL.

Figure 3.

 Synthesis of 4HIL in the course of the HPAL/BCAT bienzymatic reaction. (a) Chemical equilibria established in the course of the HPAL/BCAT bienzymatic reaction are shown. All ΔG′0 values are calculated as described in Mushkambarov (1996). HPLC analysis of the standard solution containing 4HIL and α-aminobutyrate (b), control reaction mixture (without HPAL) (c) and bienzymatic reaction mixture (d); (e–g) LC/MS analysis (MS detection mode, compounds with Mr=147 were only detected) of the same samples. Peak abbreviations: 1, l-glutamate; 2–6, quinolinilurea; 3–6, aminoquionoline; 4, α- aminobutyrate; 5, (2S, 3R, 4R)-4HIL; 6, (2S, 3R, 4S)-4HIL; 7, (2R, 3S, 4R)-4HIL; 8, unidentified compound.

Materials and methods


Most chemicals were purchased from Sigma-Aldrich. The (2S, 3R, 4S)-4HIL was extracted from fenugreek seeds as described in Fowden et al. (1973) and used as the standard for HPLC and LC/MS analyses. Chromatographic resins and HPLC columns were obtained from Amersham Pharmacia Biotech.

Bacteria and culture conditions

Arthrobacter simplex strain AKU 626 was obtained from the AKU (Faculty of Agriculture, Kyoto University) culture collection of Kyoto University. Novosphingobium aromaticivorans strain B-9294 (ATCC 12444) was obtained from the ATCC Bacteriology Collection. Strains MG1655[Ptac-rbs-YfaU] and MG1655[Ptac-rbs-YhaF] were obtained from the AGRI (Ajinomoto-Genetika Research Institute) cultural collection.

Frozen stock of Ar. simplex AKU 626 cells was inoculated in 3 mL of LB broth and cultured aerobically with shaking at 34°C for 12 h. Then, 0.1 mL of this culture was inoculated in 5 liters of LB broth [25 × (200 mL in a 1-liter flask)] and grown at 34°C with agitation for 24 h. Cells were harvested by centrifugation (16 000 g, 20 min) at 4°C and stored at −70°C until required. Obtained biomass was used for purification of an enzyme possessing 4-hydroxy-3-methyl-2-keto-pentanoate (HMKP)-aldolase activity (asHPAL).

Selection of bacteria utilizing 4HIL as sole carbon source

Commercially available dry fenugreek seeds were moisturized with sterile distilled water (1 g seeds per 1 mL H2O) and incubated with occasionally stirring at 30°C for about 30 min. Then, an aliquot of the seed washout (100 μL) was transferred to an agar plate of no-carbon M9 medium supplemented with 4HIL (5 mM) as a carbon source and incubated at 30°C for 72 h. Two well-growing morphologically identical bacterial colonies were detected on the selective plate. Both isolated microorganisms were identified as Arthrobacter globiformis based on phylogenetic analysis of 16S rRNA gene sequences (Liu et al., 1997).

The same procedure was carried out to isolate bacteria utilizing 4HIL from several soil samples. As a result, three environmental microbes belonging to the genera Paracoccus, Achromobacter and Pseudomonas were selected.

Cloning, expression and purification of BCAT from Bacillus subtilis

The ywaA gene encoding BCAT (Berger et al., 2003) was amplified by PCR using chromosomal DNA of B. subtilis as a template and oligonucleotides P1 (5′-CTGACCATGGGCAGCAGCCATCATCATCATCATCACAGCAGCGGCACTAAACAAACAATTCGCGTTGAATTG-3′) and P2 (5′- CTGAGCGGCCGCTTACTTGCTTTCAGTCAGCGCTG-3′) as primers. The resulting PCR fragment was digested with NcoI and ligated with plasmid pET-15(b+), which was previously digested with the same restriction enzyme. An aliquot of the reaction mixture obtained was used as a template for PCR amplification with oligonucleotides P3 (5′-TAATACGACTCACTATAGGG-3′) and P2 as primers. The resulting PCR fragment containing the ywaA gene under the control of the T7 promoter was cleaved by XbaI and NotI and ligated into the pET-22(b+) vector. The resulting plasmid pET-HT-IlvE-BSU was introduced in strain BL21(DE3) (Novagen, Madison, WI). The his6-tag-YwaA protein was purified by routine affinity chromatography using 1 mL HisTrap® column.

Analytical methods

HPLC analysis

The amino acids were determined using the Accq-Tag method (Waters). Synthesis of amino acid derivatives and their separation were performed according to manufacturer's recommendations. Separation was carried out at 20°C using an Agilent 1100 series chromatograph equipped with LUNA C18 (2) column (150 × 2 mm, 3 μm; Phenomenex).

LC/MS analysis

Synthesis of amino acid derivatives was performed in the same manner as for HPLC analysis. Samples were separated using an Alliance 2695 high-pressure chromatograph (Waters) equipped with LUNA C18 (2) column (150 × 2 mm, 3 μm; Phenomenex) at 26°C. The mobile phase contained 10 mM CH3COONH4 at pH 6.8. The flow rate was 0.3 mL min−1. One microliter of the sample was injected into the chromatograph. Detection was performed by positive mode electrospray ionization (ESI) in a Quattro-Micro tandem quadrupole mass spectrometer (Waters-Micromass). The source conditions were: capillary voltage, 2.0 kV; cone voltage, 20 V; source temperature, 119°C; dissolution temperature, 450°C; cone gas flow rate, 60 L h−1; dissolution gas flow rate, 659 L h−1. All chromatograms were recorded in SIR data type.

Enzyme assay

The HPAL/BCAT bienzymatic reaction was carried out as follows. Each reaction mixture (50 μL) contained 100 mM l-glutamate (pH 8 adjusted by NaOH), 100 mM α-ketobutyrate, 100 mM acetaldehyde, 2 mM MgCl2, 35 μg purified his6-tag-YwaA and sample of the aldolase preparation. Under the reaction conditions used, the rate of HMKP amination was assumed to be proportionate to the rate of its synthesis. Thus, the specific activity was indirectly estimated by the rate of 4HIL formation. Reaction mixtures were incubated at 37°C for various times (5–30 min). For each reaction, the final concentration of 4HIL was determined by HPLC analysis. Based on these data the rate of 4HIL production was estimated. One unit of HPAL activity corresponds to the formation of 1 nmol 4HIL per minute.

Purification of asHPAL

All chromatographic procedures were carried out using an ÄKTAbasic100 system (Amersham Pharmacia Biotech). The purification protocol included the following steps.

Step 1. Frozen cells were thawed and resuspended in 30 mL buffer A [50 mM KH2PO4 (pH 7.4) adjusted by KOH] supplemented with 1mM phenylmethylsulfonyl fluoride. Cells were disrupted by 3–5 passages through a French pressure cell (maximum pressure 2500 Psi) followed by centrifugation (14 000 g, 4°C, 20 min) to remove cell debris. The protein preparation obtained was passed through a Sephadex G-15 column (1.6 × 25 cm) equilibrated with buffer A.

Step 2. Fifty milliliters of protein preparation obtained from Step 1 were applied to a DEAE-Sepharose column (1.6 × 25 cm) equilibrated with buffer A. Elution was carried out at a flow rate of 2.5 mL min−1 by a linear NaCl gradient (0–0.5 M) in buffer A (10 column volumes). Each 10-mL fraction was collected. Active fractions eluted within the 0.35–0.37 M NaCl concentration interval were pooled and desalted as described in ‘Step 1’.

Step 3. The protein preparation obtained from Step 2 was applied to a 1.6-mL Sourse15Q column equilibrated with buffer A. The elution was carried out at a flow rate 1 mL min−1 by a linear NaCl gradient (0–0.5 M) in buffer A (40 column volumes). Each 2-mL fraction was collected. Active fractions eluted within the 0.31–0.35 M NaCl concentration interval was pooled.

Step 4. The protein concentration in the preparation obtained from Step 3 was adjusted to 0.8 mg mL−1 by dilution in buffer A, and then ammonium sulfate was added up to the final concentration of 1.5 M. The protein preparation obtained was applied to a 1-mL Resource PHE column equilibrated with buffer A containing 1.5 M ammonium sulfate. Elution was carried out at a flow rate of 1 mL min−1 by a linear gradient from 1.5 to 0 M of ammonium sulfate in buffer A (30 column volumes). Each 1-mL fraction was collected. Active fractions were eluted within the 0.19–0.1 M ammonium sulfate concentration interval and pooled.

Step 5. One milliliter of protein preparation obtained from Step 4 was applied to a Superdex 200 HR 10/30A column equilibrated with buffer A containing 100 mM NaCl. Isocratic elution was performed at a flow rate of 0.5 mL min−1. Each 1-mL fraction was collected. Active fractions were pooled.

Western blotting and amino acid sequencing

The purified sample was electrophoresed in a 12% sodium dodecyl sulfate (SDS) polyacrylamide gel and electroblotted on to a Sequi-Blot PVDF membrane (Bio-Rad) using a Trans-Blot SD cell (Bio-Rad). The N-terminal amino acid sequence of the immobilized asHPAL was determined by means of automated Edman degradation with a model 491cLC Protein Sequencer (Applied Biosystems).

PCR-based amplification and nucleotide sequencing of the asHPAL structural gene

A 0.9-kb DNA fragment of chromosome of Ar. simplex AKU 626 was amplified using oligonucleotides DP [5′-CGGCCTCCTGTTTAGCTCCCGATGCCITT(T/C)CCIGTIGA(A/G)(C/T)TICCIGA(T/C)AA(T/C)TT(T/C)-3′] and FP [5′-CGGCCTCCTGTTTAGCTCCCGATG-3′] as primers.

The amplification procedure was carried out as follows.

Step 1. Reaction mixture I (40 μL) contained 0.1 μg of the purified chromosomal DNA of Ar. simplex AKU 626 and 20 pmol DP primer. The PCR I protocol used was 50 cycles of 95°C for 10 s, 53°C for 20 s and 72°C for 40 s. An amplified 0.9-kb DNA fragment was purified using 1% agarose gel.

Step 2. Reaction mixture II contained 0.01 μg of DNA fragment amplified at Step 1 and 10 pmol FP primer. The PCR II protocol consisted of 25 cycles of 95°C for 10 s, 65°C for 20 s and 72°C for 40 s. The amplified 0.9-kb DNA fragment was recovered from 1% agarose gel and digested with NcoI. Because of the resulting DNA fragment was flanked by two identical sequences in opposite orientation (see Fig. 2), we could not directly use the FP primer for sequencing of the whole amplified fragment. Therefore, it was digested by NcoI and then each of the 0.6- and 0.3-kb fragments obtained was sequenced using primer FP. Based on the determined sequences of the internal fragments, primers IP1 [5′-ACTTGGGTGCGCAAAACCTCATGG-3′) and IP2 (5′-CCATGAGGTTTTGCGCACCCAAGT-3′) were designed and applied to the complete sequencing of the 0.9-kb fragment.

Figure 2.

 Amplification of the asHPAL structural gene. The 3′-flanking nucleotide sequence (R) of the degenerate DP primer encodes the determined N-terminal amino acid sequence of asHPAL, at positions from 1 (Pro) to 10 (Phe). The 5′-flanking nucleotide sequence (L) coincides with sequence of the FP primer and includes an ATG start codon. Inosine was used at the third (wobble) position of each codon where any of the four bases might be required. (inosine can form base pairs with cytidine, thymidine and adenosine).

Cloning and expression of the BphF aldolase

A 0.8-kb DNA fragment of the pNL1 plasmid from N. aromaticivorans B-9294 was amplified using PCR with oligonucleotides SVS93 [5′-GGTCACAAGCTTGTAATATAGGAGGGCGACAATGCAAACACCCG-3′) and SVS94 (5′-GGTCACGAATTCTTAGACTCCGCCCTGCGCGGCGATGC-3′) as primers. A piece of cell culture from the agar plate was used as a source of template DNA for the PCR procedure. An amplified DNA fragment containing the bphF gene was digested with HindIII and EcoRI and ligated with pUC19 plasmid previously digested with the same restriction enzymes. The resulting plasmid pUC19-bphF was introduced into strain MG1655. Specific HPAL activity measured in crude extract of recombinant strain MG1655 [pUC19-bphF] was about 10 U mg−1.

Purification of the YhaF, YfaU and BphF aldolases

YhaF, YfaU and BphF aldolases were purified from crude extracts of recombinant strains MG1655[Ptac-rbs-YhaF], MG1655[Ptac-rbs-YfaU] and MG1655 [pUC19-bphF], respectively, by using a uniform purification scheme. This included two anion exchange chromatographies and precipitation by ammonium sulfate (data not shown). Specific HPAL activities of the aldolase preparations obtained were 1500 U mg−1 (YhaF), 1000 U mg−1 (YfaU) and 800 U mg−1 (BphF).

Nucleotide sequence accession number

The determined nucleotide sequence was registered in the GenBank under accession no. EF117323.

Results and discussion

Screening of the microorganisms possessing HPAL activity

The reversible HPAL/BCAT bienzymatic reaction may be considered as a part of the putative 4HIL breakdown pathway, which results in formation of acetyl-coenzyme A and propionyl-coenzyme A (Fig. 1). Because the expression of HPAL is a necessary criterion in realizing the proposed 4HIL degradation route, the ability to utilize 4HIL as a sole carbon source (4HIL+ phenotype) may be used as a ‘selection marker’ for the HPAL-producing bacterium. Because the 4HIL+ phenotype may be the result of metabolic adaptation of some microorganisms to its specific environment (e.g. excess 4HIL), the fenugreek seeds have been chosen as a ‘source’ of the desired microbes. We expected that the abundance of 4HIL may serve as a ‘nutrient attractor’ for the bacteria, which are able to ‘colonize’ the seed surface and produce enzymes involved in 4HIL catabolism. As a result, an environmental bacterium utilizing 4HIL (as well as α-ketobutyrate and acetaldehyde) as a sole carbon source and producing HPAL was isolated and identified as Ar. globiformis. However, specific HPAL activity detected in crude cell lysates of this strain was rather low (about 0.06 U mg−1), and this appreciably complicated the HPAL purification procedure. Nevertheless, the result obtained suggested that microbes closely related to Ar. globiformis can express homologous enzymes possessing HPAL activity. Searches of the most effective HPAL-producing strains were carried out among bacteria related to the genus Arthrobacter. For this purpose, we tested HPAL activity in crude extracts of 10 species from the AKU and AGRI collections. As a result, Ar. simplex AKU strain 626 was selected as a high HPAL producer (specific HPAL activity assayed in crude cell lysates of this strain was 0.6 U mg−1). In addition, we tried to isolate environmental microbes possessing the 4HIL+ phenotype from several soil samples. Three HPAL-producing bacteria belonging to the genera Paracoccus, Achromobacter and Pseudomonas were selected (specific HPAL activities assayed in crude cell lysates of these strains were estimated as 1, 0.8 and 0.4 U mg−1, respectively). In order to make the final choice of the HPAL-producing strain, the stereochemistry of the HPAL/BCAT bienzymatic reaction (see above) was investigated. Partially purified aldolases from four selected strains were used in the bienzymatic reaction. We found that coupling of activities of BCAT and aldolases from Paracoccus, Achromobacter and Pseudomonas strains led to about equimolar synthesis of the (2S, 3R, 4S) and (2S, 3R, 4R)-4HIL isomers, while using aldolase from Ar. simplex AKU 626 resulted in preferable formation (about threefold) of the (2S, 3R, 4S)-4HIL isomer. The result suggested that the enzyme from strain AKU 626 exhibits stereoselectivity during aldol condensation of acetaldehyde and α-ketobutyrate. Thus, our next aim was to purify and identify HPAL from AKU 626.

Purification and preliminary characterization of asHPAL

The enzyme possessing HPAL activity was 2400-fold purified from crude cell extract of A. simplex AKU 626 strain as shown in Table 1. The determined N-terminal amino acid sequence of the purified protein was: PFPVELPDNFAKRVTDSDSAQVGLFI. The native protein showed a molecular mass of 180 kDa by gel filtration and of 27 kDa by SDS-polyacrylamide gel electrophoresis (PAGE) analysis, indicating that asHPAL is a hexameric protein. Divalent metal ions such as Zn2+, Mg2+ and Mn2+ increased activity, whereas EDTA completely inhibited the enzyme (data not shown). Thus, it was concluded that asHPAL belongs to the type II (metal-dependent) class of aldolases.

Table 1.   Summary of purification of asHPAL
StepTotal protein (mg)Total activity (U)Specific activity (U mg−1)Purification (fold)Yield (%)
1. Desalted crude cell lysate15129030.61100
2. DEAE-Sepharose14401304845
3. SOURCE 15Q440210016745
4. RESOURCE PHE0.11421419236516
5. Superdex 200 HR 10/30A0.0460149024837

Identification of the asHPAL

To amplify the asHPAL structural gene, ‘inoculating’ PCR 1 was carried out using only the special degenerate primer (DP) (Fig. 2). Its primary structure consisted of two parts: a 3′-flanking sequence encoded the determined N-terminal amino acid sequence at positions 1–10, and a 5′-flanking sequence coincided with the structure of the flanking primer (FP) that had been synthesized separately. To reduce the degeneracy of the DP primer, the ‘neutral’ inosine has been used at the ‘wobble’ position of each coding triplet where any of the four bases might be required (Takahashi et al., 1985). The PCR 1 protocol has been designed to provide: (1) a specific annealing of the DP primer on the 5′-end of the asHPAL gene and (2) a nonspecific (fortuitous) annealing of the same primer on the unique site localized about 100 bp downstream of the asHPAL structural gene. As a result, a 0.9-kb DNA fragment containing the asHPAL structural gene was obtained. Sequencing of the amplified DNA fragment revealed a 780-bp (including TAG stop codon) ORF (asHPAL) encoding a protein of 259 amino acids with a calculated molecular weight of 27 228 Da. A search for the nearest homologues of asHPAL revealed that the purified enzyme belongs to the HpaI/HpcH-aldolase family, which includes three functional subgroups: 2,4-dihydroxyhept-2-ene-1,7-dioic acid aldolase (HHDE aldolase), 2-keto-3-deoxyglucarate aldolase (KDG aldolase) and 4-hydroxy-2-oxovalerate aldolase (HKP aldolase). We assumed that HPAL activity is an attribute of any enzyme that belongs to the HpcH/HpaI aldolase family. This hypothesis was confirmed by testing HPAL activity of YhaF (KDG aldolase) and YfaU (putative HHDE aldolase) enzymes from Escherichia coli (Izard & Blackwell, 2000; Wright et al., 2002) and BphF (putative HKP-aldolase) proteins from N. aromaticivorans (Romine et al., 1999). In each case coupling activities of purified aldolase and BCAT led to synthesis of 4HIL from acetaldehyde, α-ketobutyrate and l-glutamate (Table 2).

Table 2.   Relative production of the 4HIL stereoisomers in HPAL/BCAT bienzymatic reaction dependent on the aldolase used and reaction pH
AldolasepH*4HIL stereo isomer production (mM)
[2S, 3R, 4S][2S, 3R, 4R]ρ
  • *

    Reaction mixture (50 μL) contained: 200 mM l-glutamate, 2 mM MgCl2 (2 mM ZnCl2 in reaction with asHPAL and his6-tag-asHPAL), 100 mM α-ketobutyrate, 300 mM acetaldehyde, 2 mM MgCl2 (pH 6, 7, 9 adjusted with NaOH), his6-tag-YwaA (1.5 mg mL−1) and aldolase preparation (1 mg mL−1). Reaction mixtures were incubated at 37°C for 1 h (until apparent chemical equilibrium was established).

  • The data shown are mean values of triplicate measurements.

  • The relative production of 4HIL isomers was calculated as [2S, 3R, 4S]/[2S, 3R, 4R].


Preliminary characterization of the HPAL/BCAT bienzymatic reaction

Based on the numerous experimental data concerning BCAT substrate selectivity (Mamer & Reimer, 1992; Hall et al., 1993) we considered that the S-(l)-configuration of the 4HIL chiral C2 carbon atom is only under strict control during the enzymatic amination of HMKP. Thus, if an enzymatic aldol condensation of acetaldehyde and α-ketobutyrate is not a stereo-specific process, the synthesis of a mixture of four 4HIL diastereoisomers (2S, 3S/R, 4S/R) is expected. LC/MS analysis of the bienzymatic reaction mixture revealed only two compounds with a molecular weight of 147 Da, whose synthesis was strictly dependent on the HPAL activity (Fig. 3). The HPLC analysis method allowed us to separate 4HIL diastereoisomers (data not shown). Peak 6 had the same retention time as those observed for the 4HIL standard solution and was associated with 4HIL isomer (2S, 3R, 4S). Peak 5 was identified as the (2S, 3R, 4R)-4HIL isomer. The stereo configuration at the C3 chiral carbon atom thus is fixed during HMKP enzymatic synthesis.

We used three homologs of asHPAL (each presents one of three functional subgroups of the HpaI/HpcH-aldolase family) to model the HPAL/BCAT bienzymatic reaction. The ratio of equilibrium concentrations of two 4HIL isomers (ρ parameter) has been used as a stereoselectivity characteristic of the 4HIL bienzymatic synthesis (Table 2). The HKP aldolase (BphF)/BCAT reaction led to virtually nonselective synthesis of two 4HIL isomers irrespective of the pH used. KDG aldolase (YhaF)/BCAT and HHDE aldolase (YfaU)/BCAT reactions resulted in an almost twofold preferable formation of the (2S, 3R, 4S)-4HIL isomer at pH 9. In contrast, the asHPAL/BCAT reaction led to a sixfold preferable synthesis of the (2S, 3R, 4S)-4HIL isomer at pH 6 and only twofold at pH 9. The results suggested that BphF, YhaF and YfaU aldolases catalyze nonselective synthesis of (3R, 4S/R)-HMKP isomers, whereas acetaldehyde/α-ketobutyrate aldol condensation catalyzed by asHPAL leads to preferable formation of the (3R, 4S)-HMKP isomer. These data allow us to consider asHPAL as a novel type of enzyme relating to the HpaI/HpcH-aldolase family. Because only the (2S, 3R, 4S)-4HIL isomer exhibited insulinotropic biological activity, further searches for novel aldolases with highest possible stereospecificity will be helpful in construction of an optimal 4HIL synthesis process.

The HPAL/BCAT bienzymatic reaction comprises four chemical equilibria and formally is equivalent to direct condensation of acetaldehyde and α-aminobutyrate (Fig 3a). Because of the absolute ΔG0′ value of this reaction is quite low (about 5 kJ mol−1), the final yield of 4HIL in the bienzymatic reaction was not high. (For example, the theoretically predicted equilibrium concentration of 4HIL is 42 mM under the reaction conditions described in Table 2). This is the main bottleneck in the bienzymatic synthesis of 4HIL. However, this problem can be solved in the context of the actual biotechnological process (e.g. via separation of the reaction products and recirculation of the unreacted substrates).