Advancing the Development of Glycated Protein Biosensing Technology

Abstract

Research advances in biochemical molecules have led to the development of convenient and reproducible biosensing molecules for glycated proteins, such as those based on the enzymes fructosyl amino acid oxidase (FAOX) or fructosyl peptide oxidase (FPOX). Recently, more attractive biosensing molecules with potential applications in next-generation biosensing of glycated proteins have been aggressively reported. We review 2 such molecules, fructosamine 6-kinase (FN6K) and fructosyl amino acid-binding protein, as well as their recent applications in the development of glycated protein biosensing systems. Research on FN6K and fructosyl amino acid-binding protein has been opening up new possibilities for the development of highly sensitive and proteolytic-digestion-free biosensing systems for glycated proteins.

Glycated proteins in blood, such as hemoglobin A1c (HbA1c) and glycated albumin, are formed by a nonenzymatic reaction (called glycation or Maillard reaction) between glucose and proteins, proceeding through a Schiff base intermediate to produce a relatively stable product. HbA1c results from glycation of the hemoglobin β subunit N-terminal valine residue, whereas glycated albumin results from glycation of the amino group on the side chain of internal lysine residues. HbA1c and glycated albumin are considered major diagnostic indicators of diabetes and are routinely measured for assessing treatment effectiveness. Due to the erythrocyte’s long lifetime and the slow continuous and essentially irreversible characteristics of the glycation process,¹ the relative amount of HbA1c reflects the average blood glucose concentration of the preceding 1-2 months.^2,3 Glycated albumin levels, which are not affected by the long lifetime of erythrocytes, reflect the average blood glucose concentration of the preceding 2-3 weeks.³ While HbA1c has been the gold standard for assessing long-term control of glycemic levels in diabetes patients, glycated albumin is often the preferred glycemic indicator for diabetes patients undergoing hemodialysis⁴ and patients with hematologic disorders.⁵

Glycated protein concentrations have been measured in clinical laboratories by a number of different systems, such as high-performance liquid chromatography (HPLC), affinity chromatography, and immunoassay. As an attractive alternative to these conventional measurement systems, enzyme assay systems based on the enzyme fructosyl amino acid oxidase (FAOX) or fructosyl peptide oxidase (FPOX) have recently been commercially available.^6-11 FAOXs catalyze the oxidative degradation of fructosyl valine and/or ϵ-fructosyl lysine (ϵ-FK), which are the degradation products of HbA1c and glycated albumin, respectively (Figure 1A). FPOXs show relatively high activity toward fructosyl valyl histidine, which corresponds to the glycosylated N-terminal dipeptide derived from HbA1c. FAOX/FPOX-based measurement systems are expected to become a major component of glycated protein sensing since they are rapid, reproducible, and suitable for automated analyzers. Among the various biological molecules related to monitoring glycated proteins, only oxidases have been utilized in actual clinical practice. Our group has been engaged in the development of a variety of molecules and principles for glycated protein biosensing.^12-18 Especially, recent engineering studies succeeded in converting FAOX and FPOX into fructosyl amino acid dehydrogenase and fructosyl peptide dehydrogenase, respectively, by introducing mutations.^19,20 However, the FAOXs/FPOXs and the corresponding engineered dehydrogenases show relatively high K_m values (10^-3 to 10^-2 M) and cannot react with intact glycated proteins. There therefore remains a need for a more sensitive and simple analytical method for measuring glycated proteins, ideally without proteolytic pretreatment processes, which are unavoidable in currently available enzymatic methods.

Figure 1.

The chemical reactions catalyzed by deglycation enzymes. (A) Oxidation catalyzed by fructosyl amino acid oxidase/fructosyl peptide oxidase. (B) Phosphorylation catalyzed by fructosamine 3-kinase and spontaneous decomposition. (C) Phosphorylation and subsequent deglycation catalyzed by fructosamine 6-kinase and deglycase, respectively.

Kinases and binding proteins that recognize glycated amino acids and glycated proteins have also recently been reported.^21-26 These proteins have been reported to have low K_m values for their substrates and the ability to recognize intact glycated proteins, giving them great potential as novel biosensing molecules. Engineering of these proteins is expected to expand the development of more sensitive and simple biosensing systems for glycated proteins as alternatives to the FAOX/FPOX-based systems. Here, we review these next-generation biosensing molecules, fructosamine kinase and fructosyl amino acid-binding protein.

Fructosamine Kinase

Property of Fructosamine 3-Kinase and Fructosamine 6-Kinase

Glycation is one of the major nonenzymatic mechanisms occurring in various natural environments, including in living organisms, which have therefore acquired various deglycation systems. One such deglycation system, fructosamine kinase, can be divided into 2 types based on the catalytic mechanism: fructosamine 3-kinase (FN3K) and fructosamine 6-kinase (FN6K).

FN3Ks, which are found in mammals, cleave fructosamines in a 2-phased process²¹ that starts with the phosphorylation at the C3 of the fructosyl moiety, generating the rather unstable fructosamine 3-phosphate. This subsequently undergoes an autocatalytic degradation to 3-deoxyglucosone, inorganic phosphate, and the amino compound that originally reacted with glucose (Figure 1B).^27-29 Some FN3K homologues were also discovered in several organisms. Instead of reacting with fructosamines, these FN3K-related proteins (FN3K-RP) were found to react with C3 epimers, such as psicosamines, ribulosamines, and erythrulosamines.^30,31 FN3Ks/FN3K-RPs may be attractive sensing molecules since they have the ability to react with intact glycated proteins, however, the development of analytical methods employing FN3K/FN3K-RP have not yet been reported.

FN6Ks have been identified in Escherichia coli, Bacillus subtilis, and Arthrobacter aurescens.^22-24 In these bacteria, FN6K catalyzes the production of fructosamine 6-phosphate by phosphorylating fructosamine at the C6 of the deoxyfructose moiety. Following phosphorylation, deglycase catalyzes the degradation of fructosamine 6-phosphate to free amino acid and glucose 6-phosphate (Figure 1C). The genes encoding FN6K (frlD) and the deglycase (frlB) are located on the same operon. Extensive studies in E. coli and B. subtilis have led to an understanding of the physiological role of FN6K and deglycase as key enzymes in the catabolic pathway of naturally occurring fructosamines, which is supported by reports that growth on Amadori products as sole carbon and nitrogen source is possible for E. coli²² and B. subtilis.³² Based on conserved primary structural features, FN6Ks belong to the PfkB ribokinase family, which comprises enzymes that phosphorylate primary alcohols.^32-35

FN6Ks from E. coli, B. subtilis, and A. aurescens do not react with any sugars or free amino acids, and are thus specific for fructosamines (Table 1). This substrate specificity provides the possibility of FN6Ks utilization as novel molecular recognition elements for sensing glycated proteins. FN6Ks show lower K_m values (10^-5 to 10^-4 M) than FAOXs/FPOXs (generally 10^-3 to 10^-2 M), while FN6Ks and FAOXs/FPOXs have comparable V_max values. An additional attractive property of FN6Ks is that soluble protein can be obtained in high yields by recombinant expression in E. coli; 110 to 270 mg of purified FN6K has been obtained from 1 liter of culture.^22,36

Table 1.

Enzymatic Properties of Reported Fructosamine 6-Kinases.

Substrate	E. coli22,23		B. subtilis23		A. aurescens24
	Km (mM)	Vmax (U/mg)	Km (mM)	Vmax (U/mg)	Km (mM)	Vmax (U/mg)
ϵ-FK	1.8 ×10−2	30	14	7.0	6.0 ×10−1	1.8
Fructosyl valine	>20	4.0 ×10−2 a	1.0 ×10−1	3.0	>20	4.2 ×10−2 a
Fructose	>50	10 ×10−2 b	na	na	>100	20 ×10−2 c
Glucose	nd	nd	na	na	nd	nd
Lysine	na	na	na	na	nd	nd

na, data are not available; nd, not detected. Activity with 20 mM^a, 50 mM^b, or 100 mM^c substrate.

The substrate specificity of the B. subtilis FN6K differs from those of the E. coli and A. aurescens FN6Ks, although they share 50% similarity in amino acid sequences.²⁴ The FN6K from B. subtilis shows broader substrate specificity, catalyzing the phosphorylation of both ϵ-FK and α-glycated amino acids, such as fructosyl valine. In contrast, the FN6Ks from E. coli and A. aurescens preferentially catalyze the phosphorylation of ϵ-FK (Table 1).^23,24 Furthermore, the FN6K from E. coli was reported to recognize intact glycated albumin as a substrate and to catalyze the production of protein-fructosamine 6-phosphate.³⁷ Considering the high amino acid sequence similarities between the 3 FN6Ks, there is a possibility that the A. aurescens FN6K recognizes intact glycated albumin and the B. subtilis FN6K recognizes both intact glycated albumin and HbA1c. Further investigation on the FN6Ks’ substrate specificities against glycated proteins is expected.

Application and Engineering of Fructosamine 6-Kinase

Because intact glycated proteins cannot be substrates for FAOXs/FPOXs, a proteolytic pretreatment is always included in FAOX/FPOX-based measurement systems. Because FN6Ks can react with glycated proteins, as demonstrated for the E. coli FN6K,³⁷ they can be utilized as novel recognition elements for glycated protein measurement systems without the necessity of incorporating proteolytic pretreatment.

The enzymatic activity of FN6Ks can be easily measured by colorimetric assay by coupling with pyruvate kinase, pyruvate oxidase, and peroxidase reactions.^24,36 The measurement is based on the monitoring of the ADP generated from ATP by the FN6K-catalyzed phosphorylation of fructosamine. The generated ADP is subsequently subjected to sequential enzymatic reactions with pyruvate kinase, pyruvate oxidase, and peroxidase. The quantity of H₂O₂ produced by pyruvate oxidase and the quinoneimine dye produced by peroxidase correlates with fructosamine concentration (Figure 2), and as a result, the increase of absorbance was observed with increasing substrate concentration. In the assay using ϵ-FK (hydrolytic product of glycated albumin) as substrates, positive correlation between the absorbance and the ϵ-FK concentration was observed in range from several µM to several hundred µM (Figure 3).³⁶ This suggested that the measurement of glycated albumin pretreated with protease will show similar results and be able to cover the clinical range of glycated albumin by diluting the samples in reaction buffer, since the estimated molar concentration ranges of glycated albumin in human blood is approximately 30-770 µM. In addition, using this measurement system, we have succeeded in the application of recombinant E. coli FN6K for the detection of glycated albumin without the use of protease digestion (data submitted for publication). The colorimetric change based on quinoneimine dye production increased with increasing glycated albumin concentration. This will be the first report demonstrating that intact glycated proteins can be measured without proteolytic digestion.

Figure 2.

Principle of colorimetric assay for glycated protein employing fructosamine 6-kinase. The measurement of ADP generated by the phosphorylation of fructosamine by fructosamine 6-kinase with ATP. ADP is subjected to sequential enzymatic reactions with pyruvate kinase, pyruvate oxidase, and peroxidase in which quinoneimine dye produced by peroxidase correlates with fructosamine concentration.

Figure 3.

Calibration curve for ϵ-fructosyl lysine by colorimetric assay (n = 2, error bar: SD).³⁶ Measurement of absorbance at 546 nm was carried out with fructosamine 6-kinase (FN6K) from E. coli and various concentrations of ϵ-fructosyl lysine (ϵ-FK), prepared by incubating glucose and lysine, in the reaction mixture (1 mM ATP, 1 mM MgCl₂, 500 U/ml pyruvate kinase, 100 U/ml pyruvate oxidase, 1.5 mM N,N-bis(4-sulfobutyl)-3-methylaniline, 2 U/ml peroxidase, 1.5 mM 4-aminoantipyrine, 25 mM HEPES buffer [pH 7.1]) at 25°C.³⁶ The x-axis represents final concentrations of ϵ-FK in the reaction mixture.

Three-dimensional crystal structure of FN6K is not yet available. Elucidation of the 3D structure of the FN6K substrate-binding site is expected as an important milestone for engineering FN6Ks. Our group modified the substrate specificity of the E. coli FN6K after creating a 3D structural model by homology modeling³⁶ using as a template the crystal structure of the Thermus thermophilus 2-keto-3-deoxygluconate kinase (PDB ID: 1V1A),³⁸ which shares high amino acid sequence similarity. The 2-keto-3-deoxygluconate kinase is a sugar kinase of the PfkB ribokinase family, of which FN6K is also a member. Based on the 3D structural model, a site-directed mutagenesis investigation of the putative active-site region led to the identification of several amino acid residues that likely play important roles in the enzyme reaction. To further investigate the residues impacting substrate specificity, a primary structure alignment of FN6Ks from E. coli, B. subtilis, and A. aurescens, as well as a putative FN6K from A. tumefaciens was performed. The residue corresponding to Met220 of the E. coli FN6K was identified as a residue that plays an important role in substrate recognition. Substitution of Met220 to Leu, the residue at the corresponding position in the B. subtilis FN6K, which shows a different substrate specificity, resulted in 7-fold increased activity for fructosyl valine and 6-fold decreased activity for ϵ-FK, thus increasing the specificity for fructosyl valine by 40-fold.³⁶ The research has provided valuable information on the molecular recognition mechanism behind the ability of FN6Ks to distinguish between ϵ-FK and fructosyl valine. This is expected to greatly contribute to future engineering of FN6K specificity, thus enabling the development of biosensing systems specific for glycated albumin or HbA1c without proteolytic digestion.

Fructosyl Amino Acid-Binding Protein

Property of Fructosyl Amino Acid-Binding Protein

Another molecular recognition element that has potential to become a major component of next-generation biosensing systems for glycated proteins is fructosyl amino acid-binding protein. Fructosyl amino acid-binding proteins are bacterial substrate-binding proteins (SBPs), which are monomeric proteins consisting of 2 domains linked by a hinge region.³⁹ Ligand binding to bacterial SBPs leads to a large conformational change that can be easily monitored by changes in fluorescence from either autofluorescence or an appropriately positioned fluorescent label.^40,41 In the research on the development of glucose sensor, sensing technologies using SBPs such as a galactose/glucose-binding protein and concanavalin A are well established and have been evaluated for continuous glucose monitoring in preclinical studies and human clinical studies.^42-44 However, measurement technology for glycated proteins employing SBPs has only recently been reported.

The presence of binding protein for fructosyl amino acid was first reported in 1994, which was isolated from the water soluble fraction of Pseudomonas sp. extract.⁴⁵ This protein, called Amadori product-binding protein (ABP), was reported to have a binding constant of 1.49 µM toward glycated ϵ-aminocaproate and no enzymatic activity. ABP was therefore considered to be a novel transport protein or permease. The ABP was suggested to preferentially bind to ϵ-fructosyl amino acids rather than α-fructosyl amino acids, however, the binding to glycated albumin was not detected.⁴⁵ In addition, the amino acid sequence of ABP has not been determined. The application of ABP as a sensing molecule for glycated protein is therefore considered limited.

Additional fructosyl amino acid-binding proteins have been recently identified in Agrobacterium tumefaciens and Arthrobacter sp FV1-1.^25,26 The fructosyl amino acid-binding protein from Arthrobacter sp FV1-1 shares 64% amino acid sequence similarity with the fructosyl amino acid-binding protein from A. tumefaciens. The A. tumefaciens fructosyl amino acid-binding protein serves as a key component of the ABC transporter in the santhopine (fructosyl glutamine) catabolism system and is encoded by a gene (socA) located on the santhopine catabolism operon.²⁵ Some fructosyl amino acid-binding protein homologues have been identified in other Arthrobacter sp in homology analysis,²⁶ suggesting that fructosyl amino acid-binding proteins may be broadly distributed among bacteria, both Gram-negative and Gram-positive.

Tryptophan-related autofluorescence can be used to monitor ligand binding for fructosyl amino acid-binding proteins. The fructosyl amino acid-binding protein from A. tumefaciens binds preferentially to α-fructosyl amino acids, such as fructosyl glutamine and fructosyl valine, resulting in calculated K_d values of 0.2 µM and 0.6 µM, respectively (Table 2).²⁵ The fructosyl amino acid-binding protein from Arthrobacter sp, which shares high similarity of amino acid sequence with the fructosyl amino acid-binding protein from A. tumefaciens, has a similar ligand specificity (Table 2).²⁶ These fructosyl amino acid-binding proteins do not bind to ϵ-FK, glucose, fructose, glutamine, and valine. Moreover, the fructosyl amino acid-binding proteins show extremely high affinity for ligands, with K_d values of 10^-7 M, compared to the FAOXs/FPOXs, which generally have K_m values of 10^-3 to 10^-2 M. Furthermore, our evaluation of intact HbA1c binding ability using surface plasmon resonance (SPR) measurement with an apo-HbA1c-immobilized sensor chip resulted in increased response with increasing fructosyl amino acid-binding protein concentration (data not shown), suggesting that fructosyl amino acid-binding protein has the ability to bind intact HbA1c. The high ligand specificity and affinity for α-fructosyl amino acids and the ability to bind intact HbA1c provides the possibility of fructosyl amino acid-binding protein utilization as novel molecular recognition elements for HbA1c sensing.

Table 2.

Ligand Specificity of Reported Fructosyl Amino Acid-Binding Proteins.

Ligand	Kd (µM)
	A. tumefaciens25,26	A. aurescens26
Fructosyl valine	0.6	0.6
Fructosyl glutamine	0.2	0.3
ϵ-FK	nd	nd
Fructose	nd	nd
Glucose	nd	nd
Valine	nd	nd

nd, not detected.

Application and Engineering of Fructosyl Amino Acid-Binding Protein

The advantage of using fructosyl amino acid-binding proteins to measure glycated proteins is that the fructosyl amino acid-binding proteins show extremely high affinity for fructosyl valine compared to FAOXs/FPOXs. Fructosyl amino acid-binding proteins have the potential of being applied in highly sensitive fluorescent sensing systems employing fluorescent probes (Figure 4). A novel fluorescent sensing system to measure fructosyl valine was developed employing the A. tumefaciens fructosyl amino acid-binding protein, which was engineered by site-direct mutagenesis and modified with a fluorescent probe.⁴⁶ An Ile166Cys mutant of fructosyl amino acid-binding protein was created to insert a candidate site for labeling with a thiol-reactive fluorophore, since Ile116 was expected to be located close to the active site without contributing to substrate binding, based on the fructosyl amino acid-binding protein 3D structural model constructed using as a template the crystal structure of the E. coli glutamine-binding protein (PDB ID: 1WDN).⁴⁷ The fructosyl amino acid-binding protein Ile166Cys mutant, which was modified with the environmentally sensitive fluorophore acrylodan, showed a stable fluorescence signal. The increase in fluorescence intensity was observed with increasing fructosyl valine concentration between 17 nM and 2.5 µM, whereas no change in fluorescence intensity was detected with the addition of up to 3.7 M ϵ-FK.⁴⁶ The sensitivity to fructosyl valine is 8-fold greater than when using autofluorescence with the unconjugated protein, making it over 100-fold more sensitive than FAOX/FPOX-based detection systems. Using this measurement system, we have succeeded in the detection of HbA1c in hemoglobin sample pretreated by protease digestion (Figure 5).

Figure 4.

Diagrammatic illustration of fluorescence sensing by fructosyl amino acid-binding protein. With the open conformation of fructosyl amino acid-binding protein, the fluorophore is in a polar environment and fluorescence is suppressed. In the fructosyl valine-bound closed conformation, fluorophore environment in the binding site changes to a hydrophobic one, thereby increasing fluorescence intensity.

Figure 5.

Determination of HbA1c from digested hemoglobin samples based on acrylodan-conjugated fructosyl amino acid-binding protein (n ≥ 2, error bar: SD). HbA0 and HbA1c, purified from human blood, were separately hydrolysed with proteases (thermolysin and carboxypeptidase Y) and solutions at various percentages of HbA1c were prepared with mixing 2 hydrolyzed samples. The measurements were carried out with 0.1 mg/ml total hemoglobin, 1 µM acrylodan-conjugated fructosyl amino acid-binding protein in 10 mM MOPS buffer (pH7.0) at 25°C. Categories of “prediabetes” and “diabetes” are based on the American Diabetes Association’s Standards of Care.⁴⁹

Following this success, the measurement of fructosyl valyl histidine, corresponding to the N-terminal fructosyl peptide derived from HbA1c, was also investigated with the acrylodan-conjugated Ile166Cys mutant of fructosyl amino acid-binding protein. The increase in fluorescence intensity was observed with increasing fructosyl valyl histidine concentration, while no increase in the fluorescence intensity was observed with the addition of valyl histidine (Figure 6). Fructosyl amino acid-binding protein was thus demonstrated to enable fluorescent measurement of fructosyl valyl histidine in the range of several µM to several hundred µM, with a K_d value of 30 µM. Since the estimated molar concentration ranges of HbA1c in human blood is 30-310 µM, it is expected that the measurement of HbA1c will show similar results although the dynamic range may be different from Figure 6 due to difference of substrate molecular mass. Further detailed investigations on fructosyl amino acid-binding protein’s HbA1c binding ability based on the fluorescent detection system are expected to accelerate the development of biosensing systems specific for HbA1c without proteolytic digestion.

Figure 6.

Correlation between fluorescence intensity and the concentration of fructosyl valyl histidine (circles) and valyl histidine (squares) (n ≥ 2, error bar: SD). The measurements were carried out with 0.6 µM acrylodan-conjugated fructosyl amino acid-binding protein and various concentrations of valyl histidine or fructosyl valyl histidine, prepared by incubating glucose and valyl histidine, in 100 mM MOPS buffer (pH 7.0) at 25°C . The x-axis represents final concentrations of valyl histidine or fructosyl valyl histidine in the assay buffer.

Conclusion and Prospects

We reviewed the recent progress in the development of biosensing molecules for glycated proteins, focusing on the novel molecular recognition elements fructosamine kinases and fructosyl amino acid-binding proteins, which have abilities that will help us overcome the hurdle we are currently facing with existing enzymatic measurement systems. Compared to the currently employed enzymes FAOX and FPOX, FN6K and fructosyl amino acid-binding protein show extremely high affinity for glycated amino acids and the ability to react with or bind to intact glycated proteins. Their unique properties will enable the development of new sensing systems for detecting glycated proteins with high sensitivity and making proteolytic digestion unnecessary.

The availability of FN6K and fructosyl amino acid-binding protein remains limited, although FN6K homologues have been reported in Bacillus amyloliquef-aciens, Bacillus megaterium, Agrobacterium vitis, and Agrobacterium tumefaciens,⁴⁸ and fructosyl amino acid-binding protein homologues have been reported in Arthrobacter aurescens, Arthrobacter globiformis, and Arthrobacter phenanthrenivorans.²⁶ Further investigations on these putative FN6Ks and fructosyl amino acid-binding proteins and comparison of their properties will improve our understanding of FN6K/fructosyl amino acid-binding protein substrate recognition mechanism for fructosyl valine, ϵ-FK and intact glycated proteins.

The additional advantage of sensing systems based on enzymes like FN6Ks is that they have the potential of being applied in amperometric sensing systems employing technology that has already been well established in glucose sensors based on glucose dehydrogenase. On the other hand, the additional advantage of sensing systems based on SBPs like fructosyl amino acid-binding protein is that they have the potential of being applied in fluorescent sensing systems employing the technology that has already been well established in glucose sensors based on galactose/glucose-binding protein and concanavalin A. This can lead to the development of highly sensitive and convenient biosensing systems for point-of-care treatment, self-monitoring, implantable continuous-monitoring, or in vivo nano-sensing applications for glycated proteins.

FN6Ks and fructosyl amino acid-binding proteins are attractive second-generation sensing elements that have a great potential in the development of highly sensitive and proteolytic-digestion-free biosensors for glycated proteins. The further biomolecular engineering of FN6Ks and fructosyl amino acid-binding proteins and the combination with existing sensing technologies will accelerate major advances in glycated protein biosensing research.