Creation of haemoglobin A1c direct oxidase from fructosyl peptide oxidase by combined structure-based site specific mutagenesis and random mutagenesis

The currently available haemoglobin A1c (HbA1c) enzymatic assay consists of two specific steps: proteolysis of HbA1c and oxidation of the liberated fructosyl peptide by fructosyl peptide oxidase (FPOX). To develop a more convenient and high throughput assay, we devised novel protease-free assay system employing modified FPOX with HbA1c oxidation activity, namely HbA1c direct oxidase (HbA1cOX). AnFPOX-15, a modified FPOX from Aspergillus nidulans, was selected for conversion to HbA1cOX. As deduced from the crystal structure of AnFPOX-15, R61 was expected to obstruct the entrance of bulky substrates. An R61G mutant was thus constructed to open the gate at the active site. The prepared mutant exhibited significant reactivity for fructosyl hexapeptide (F-6P, N-terminal amino acids of HbA1c), and its crystal structure revealed a wider gate observed for AnFPOX-15. To improve the reactivity for F-6P, several mutagenesis approaches were performed. The ultimately generated AnFPOX-47 exhibited the highest F-6P reactivity and possessed HbA1c oxidation activity. HbA1c levels in blood samples as measured using the direct assay system using AnFPOX-47 were highly correlated with the levels measured using the conventional HPLC method. In this study, FPOX was successfully converted to HbA1cOX, which could represent a novel in vitro diagnostic modality for diabetes mellitus.


Obtaining the template enzyme for modification
In the course of screening our fugal library, fructosyl peptide oxidase (AnFPOX-1) of Aspergillus nidulans KH125 was selected because of its highest reactivity ratio for fructosyl valine (F-V) over -fructosyl lysine-F-K) (F-V/-F-K signal ratio: 28.4). The amino acid sequence of wild -type AnFPOX-1 shared high identity with those of known group I FPOX species including Eupenicillium terrenum FPOX (89%) 1 and Phaeosphaeria nodorum FPOX (83%) 2 .
The recombinant protein of AnFPOX-1 was successfully prepared using the Escherichia coli DH5 strain harbouring the expression plasmid pTrc-AnFPOX-1. AnFPOX-1 exhibited highly specific oxidation activity for F-V (0.557 U/mg) and weak reactivity for -F-K (0.012 U/mg), supporting the reported group I FPOX nature of AnFPOX-1 3 . Conversely, AnFPOX-1 displayed no reactivity for fructosyl valyl histidine (F-VH), a target fructosyl substance in the current HbA1c enzymatic assay.

Improving enzymatic characteristics via random mutagenesis
The enzymatic specification of AnFPOX-1 was improved using oxidation activity for F-VH and thermal stability as indices for modification via random mutagenesis. AnFPOX-2, generated by introducing an S71Y mutation in AnFPOX-1, exhibited significant F-VH oxidation activity for the first time. After subsequent rounds of random mutagenesis using the pTrc-AnFPOX-2 plasmid as a template, AnFPOX-15 was generated by introducing an additional ten mutations. The F-VH oxidation activity of AnFPOX-15 was improved by 7.5-fold compared with that of AnFPOX-2 (Table S1). No mutants generated in these trials exhibited reactivity for fructosyl peptides longer than F-VH and, let alone for whole HbA1c molecule.
Concurrently, the thermal stability of the generated mutants was evaluated by measuring the residual activity after heat treatment at 55°C for 15 min. The thermal stability of AnFPOX-15 was improved by 17.4-fold compared with that of AnFPOX-1 (Table S1).

Activity-based screening using a fungal library
Fructosyl substrates (F-V, F-VH and -F-K) were synthesised and purchased from Peptide Institute, Inc. (Osaka, Japan) Our fungal library (122 strains) was subjected to enzyme screening by evaluating the reactivity ratio for fructosyl substrates (F-V/-F-K). Each strain was inoculated in YPD liquid medium and cultured with shaking at 30°C and 200 rpm for 2 days. Grown mycelia were collected via filtration and homogenised after freezing using liquid nitrogen. By adding 10 mM potassium phosphate buffer (KPB, pH 7.0) containing protease inhibitor cocktail, internal proteins were extracted as the samples for reactivity evaluation.
For the evaluation of reactivity, a colorimetric method using oxidase and peroxidase (POD) was employed. A 190 L mixture containing of 10 mM KPB (pH7.0), 3.5 U/L POD, 0.5 mM 4-aminoantipyrine, 0.5 mM N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3methylaniline and 20 L of 1 mM substrate (F-V or -F-K, final concentration of 0.09 mM) was mixed in a multi-well plate. Then, 10 L of the enzyme samples were added to initiate the reaction. The absorbance change at 550 and 690 nm (reference) during incubation for 10 min at 30°C was measured using an Infinite F200 (Tecan, Männedorf, Switzerland). Signal intensities detected using F-V and -F-K were compared, and the strain that exhibited a high F-V/-F-K ratio was selected as a candidate.

Preparing recombinant enzyme
Aspergillus nidulans (KY125) was cultured on potato dextrose agar medium, and the genomic DNA of the cultured mycelia was prepared using NucleoSpin Plant II (TaKaRa Bio, Japan). The hypothetical FPOX sequence of A. nidulans genomic DNA was identified via a Blast search against a DNA database (http://www.aspgd.org/) using the cDNA sequence of fungal FPOX (BAD00186.1| fructosyl peptide oxidase [Coniochaeta sp. NISL 9330]) as a query. Referring to this sequence, the primer pair of AnFPOX-F (5′-AATGCCATGGCGCCCCGAGCCAACACCAAAATC-3′, underline denotes an Nco I site and bold indicates an artificial base replacement to introduce an Nco I site) and AnFPOX-R primer (5′-AATGGGATCCCTACATCTTTGCCTCATTCCTCCAC-3′, underline denotes a Bam HI site) was designed, reflecting the 5′ or 3′ sequence of the hypothetical FPOX sequence. Using these primers, the genomic DNA sequence of the putative FPOX was amplified using KOD-plus DNA polymerase (Toyobo, Japan) following the manufacturer's protocol. The PCR fragment was purified and digested by Nco I and Bam HI endonucleases to introduce the digested product into the multi-cloning site of the pTrc99A vector (Amersham Pharmacia). After sequence analysis, the determined sequence and that in the database were compared to identify intron sequences.
The cDNA sequence of FPOX was constructed via iterative overlap extension PCR procedures 4 using the primer pairs shown in Table S2. The cDNA fragment was introduced into the pTrc99A vector to construct the expression plasmid, pTrc-FPOX-1. E.
coli DH5competent cells were transformed by pTrc-FPOX-1 to develop a recombinant FPOX expression system. Cultured cells were collected via centrifugation, suspended in 10 mM KPB (pH 7.0) and disrupted by sonication. The homogenate was centrifuged and filtrated using a 0.22 m membrane to prepare a cell-free extract. The obtained cell-free extract was subjected to ammonium sulphate precipitation at 60% saturation to afford a pellet, and the pellet was The purity of the sample was confirmed via SDS-PAGE. Protein contents were determined by measuring the absorbance at 280 nm using a U-3000 spectrophotometer and assuming that E280 = 1.48 (AnFPOX-1) corresponded to 1 mg/mL 5 . The contents of other mutants were also determined in the same manner using each E280 values.

Random mutagenesis via error-prone PCR
Random mutagenesis of AnFPOX-1 was conducted using a Gene Morph II Random Mutagenesis Kit (Stratagene, San Diego, CA, USA) following the manufacturer's protocol using the primers AnFPOX-F and AnFPOX-R. In total, 300 ng of the template plasmid in 50 L of the PCR mix achieved the desired error rate (1 base error/kb). Amplified PCR fragments were purified and introduced into a multi-cloning site of the expression vector. E. coli DH5was transformed by the expression vector and cultured in LB agar plate medium (50 mg/L Amp.). Grown colonies were used for evaluation as a random mutagenesis library. Approximately 5000 clones were used to evaluate the enzymatic characteristics following the aforementioned method.

Activity assay of generated mutant enzymes
The oxidation activity for F-VH (F-VH oxidation activity) was evaluated using the same method employed for screening the fungal library. Oxidation activity was calculated via the generation of hydrogen peroxide using the calibration curve prepared from measurements using samples with known concentrations of hydrogen peroxide. One unit of activity was defined as the quantity of enzyme necessary to catalyse the formation of 1 mol of hydrogen peroxide per minute. Specific activity (U/mg) was defined as the activity catalysed by 1 mg of protein.

Evaluation of the thermal stability of mutant enzymes
The cell-free extract of FPOX-expressing E. coli or purified enzymes was subjected to heat treatment using the Gene Amp PCR System 9700 (Applied Biosystems, Foster City, CA, USA) for 10 min at 50 °C (for crude extract on screening) or 15 min at 55 °C (for purified enzymes on evaluation of enzymatic specification). Residual activity was calculated from the signal intensity after heat treatment comparing with that for nontreated samples using F-V as the substrate.

Table S1 Summary of fructosyl valyl histidine (F-VH) reactivity and thermal stability of the AnFPOX mutants
*Bold indicates mutations newly introduced in the mutant. **Specific activity (U/mg) is presented as fold changes relative to that of AnFPOX-2, which was set as 1.0.

Mutant
Mutation (number of mutation)*    Black arrows indicate the primer pairs used for the amplification of target fragment.

Figure S3. Thermal stability of AnFPOX mutants
Thermal stability of generated mutants were evaluated by measuring the residual reactivity after heat treatment at 55 o C for 15 min using F-V as a substrate. Data are mean ± SD (n =3).   The active site of AnFPOX-21/FSA structure was shown as cartoon model. A355 residue was shown as a stick in red. FSA and FAD are shown in cyan and magenta respectively in stick models. The figure was generated by PyMOL programme.