Introduction

The enzyme fructosamine-3-kinase (FN3K; E.C. 2.7.1.-) has been discovered recently and is discussed in relation to repairing glucose-mediated non-enzymatic modification of proteins. The function of FN3K is seen in catalysing the ATP-dependent phosphorylation of the protein-bound fructosamine (Amadori compound) fructoselysine, which is the first stable intermediate resulting from the ‘early’ Maillard reaction between glucose and lysine, on its 3-hydroxy group to 3-phosphofructosyllysine. The phosphorylation destabilises the fructose-amine linkage leading to a spontaneous decomposition of 3-phosphofructosyllysine to the unmodified lysine residue as well as to 3-deoxyglucosulose and inorganic phosphate (Fig. 1) [1–3]. Thus, FN3K blocks further reactions of fructoselysine, which lead to the so-called advanced glycation endproducts (AGEs). The formation of fructosamines and AGEs can change the structural and functional properties of proteins adversely. This process is accelerated at higher glucose levels and has been postulated to contribute to the development of diabetic complications [4–7].

Fig. 1
figure 1

FN3K-catalysed conversion of the substrate BzGFruK to BzGpFruK followed by spontaneous decomposition of BzGpFruK to N α-hippuryllysine, 3-deoxyglucosulose (3-DG) and inorganic phosphate

Full size image

Some approaches are already known for the measurement of FN3K activity. Phosphorylated products have been detected using 31P nuclear magnetic resonance (NMR) spectroscopy [8, 9]. In order to determine the activity in erythrocytes, FN3K was partially purified from the erythrocyte lysate [10]. A second strategy is based on the use of the substrate fructose-1,4-(3-aminoethyl)-piperazine and the radiolabelled cosubstrate [γ-33P]-ATP. Quantification of the 33P-labelled phosphorylated product was performed by utilising a scintillation technique after removal of the excess of cosubstrate. Again, FN3K was partially purified from erythrocyte lysate in order to measure the activity in erythrocytes [11]. In a third approach the radiolabelled substrate 14C-deoxymorpholinofructose (14C-DMF) is used. For the scintillation analysis the phosphorylated product 14C-DMF is isolated using an anion-exchange column. Activity studies were performed using intact erythrocytes or partially purified FN3K [2, 12]. These methods are time consuming due to the use of radioactive substrates, and NMR equipment seems to be inappropriate for routine analysis. With regard to the supposed clinical relevance of FN3K, the aim of this study was to establish a convenient assay for the determination of FN3K activity in erythrocytes, which can be performed in routine laboratories. A reversed-phase high-performance liquid chromatography (RP-HPLC) method using directly diluted erythrocyte lysate without need for enzyme purification was developed, which is based on the conversion of the synthetic substrate N α-hippuryl-N ε-(1-deoxy-D-fructosyl)lysine (BzGFruK) and direct measurement of the corresponding phosphorylated metabolite.

Materials and methods

Chemicals and reagents

N α-Hippuryllysine (N α-benzoyl-glycyl-L-lysine) was obtained from Bachem (Weil, Germany). HPLC gradient grade methanol, antipain (A 6191) and leupeptin (L-2884) were from Sigma-Aldrich (Taufkirchen, Germany). Carboxypeptidase B (EC 3.4.17.2) [DFP-treated, 133 U/mg protein, 5 mg protein/mL], chlorohaemin and all other chemicals used were of analytical grade and purchased from Fluka (Taufkirchen, Germany). The water used for the preparation of buffers and solutions was obtained using a Purelab plus purification system (USFilter, Ransbach-Baumbach, Germany). A lithium–heparin-based blood collection system S-Monovette (No. 02.1065, Sarstedt) was used for the taking of blood samples and plasma separation.

Synthesis and isolation of N α-hippuryl-N ε-(1-deoxy-D-fructosyl)lysine (BzGFruK)

The substrate N α-hippuryl-N ε-(1-deoxy-D-fructosyl)lysine (BzGFruK) was prepared as described previously with N α-hippuryllysine as starting material [13]. N α-Hippuryllysine (0.62 g, 2.0 mmol) and 2.16 g (12.0 mmol) of anhydrous glucose were heated at reflux in 84 mL methanol for 4 h. The reaction mixture was evaporated to dryness at 20°C under reduced pressure and the residue was dissolved in 9.0 mL 0.2 M N-ethylmorpholine/acetic acid buffer, pH = 8.0. The pH value was adjusted to 8.0 with N-ethylmorpholine. A 27-μL aliquot of a solution of carboxypeptidase B (665 U/mL) was added, to a final activity of 2 U/mL. The solution was incubated in a screw-capped culture tube at 25°C. Hydrolysis of unreacted N α-hippuryllysine was followed by analytical HPLC and was complete after 24 h. After that, the solvent was evaporated at 40°C under reduced pressure and the solid was dissolved in 9 mL 0.01 M sodium phosphate buffer, pH = 7.0. Isolation of BzGFruK was achieved by semi-preparative HPLC with an analytical pump system from Knauer (Berlin, Germany) as described below. Semi-preparative separation was performed using a stainless steel column, 250×8 mm with guard column 30×8 mm, both filled with Knauer Eurospher 100 RP18-material of 10-μm particle size (Knauer, Berlin, Germany). Flow rate was 1.5 mL/min at 20°C, ultraviolet detection was at 280 nm. The first chromatographic stage was isocratic elution with a mixture of 0.01 M sodium phosphate buffer (pH = 7.0) and methanol 90:10 (v/v). The second stage was isocratic elution with a mixture of 0.05 M acetic acid and methanol 90:10 (v/v). BzGFruK-containing fractions were lyophilised and stored at −20°C. Purity and identity were checked using analytical RP-HPLC, 1H, 13CNMR and mass spectroscopy as well as elemental analysis. Analytical data were consistent with those reported previously [13].

Blood samples

Blood samples from a healthy, 23-year-old female volunteer were used to establish the enzyme assay. Furthermore, samples of three healthy female volunteers and four male volunteers were studied for FN3K activities. All participants were members of the Institute and consented to the studies.

Preparation of erythrocyte lysates

Human venous blood, from healthy volunteers, was taken using the heparin-based S-Monovette system (9 mL, No. 02.1065, Sarstedt). After blood collection, S-Monovette tubes were cooled immediately and centrifuged (2,000 g, 4°C, 10 min) to separate the erythrocytes. Blood plasma and the buffy coat were withdrawn using a pipette. The erythrocytes were washed three times with 4 mL 0.9% sodium chloride solution (w/v) and intermediately, centrifuged (2,000 g, 4°C, 10 min). The heparin-coated beads were removed. The erythrocyte pellet was used immediately or stored at −25°C. The lysis of erythrocytes was performed according to Delpierre et al. [1] with some modifications. Briefly, 1.0 mL erythrocyte isolate was mixed with 4.0 mL hypotonic lysis buffer containing 5 mM 4-(2-hydroxyethyl)-piperazine-1-ethanesulfonic acid (HEPES), pH = 7.5, 1 mM dithiotreitol, 1 μg/mL leupeptin, and 1 μg/mL antipain. After storage (4°C, 10 min), the erythrocyte lysate was centrifuged (2,000 g, 4°C, 10 min) and the membrane pellet was discarded. A diluted HEPES-buffered and sodium chloride containing erythrocyte lysate, referred to as HEPES-buffered erythrocyte lysate, was prepared for the measuring of the enzymatic activity of FN3K. An aliquot of the supernatant was therefore diluted 1:1 (v/v) with 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride.

Measurement of enzymatic activity

FN3K activity was quantified by following the formation of N α-hippuryl-N ε-(3-phosphofructosyl)-lysine (BzGpFruK) from the substrate N α-hippuryl-N ε-fructosyl-lysine (BzGFruK) by means of HPLC. A 500-μL aliquot of HEPES-buffered erythrocyte lysate was mixed with 25 μL 22 mM ATP solution (disodium salt, in 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride) and 25 μL 38.3 mM BzGFruK solution (in 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride) to reach final concentrations of 1 mM ATP and 1.74 mM BzGFruK. The samples were incubated for 2 h at 37°C in a water bath. The reaction was stopped by adding 450 μL of the incubated sample to the same volume of 10% (w/v) trichloracetic acid. After vigorous shaking, the samples were stored for 1 h, centrifuged (2,000 g, 20°C, 20 min), membrane-filtered (regenerated cellulose, 0.45 μm) and immediately injected into the HPLC system. The FN3K activity was expressed as mU/g haemoglobin (Hb) and one unit was defined as the amount of enzyme catalysing the formation of 1 μmol BzGpFruK/min under the conditions described above.

High-performance liquid chromatography (HPLC)

HPLC was performed with a gradient pump system from Knauer (Berlin, Germany) with online degasser, K1500 solvent organizer, K1001 pump, dynamic mixing chamber and column oven. For detection, a K2501 Knauer variable wavelength detector was used.

Separation of BzGpFruK (peak P in Fig. 2) was achieved using a stainless steel column, 250×4.6 mm, filled with Knauer Eurospher 100, RP18 material of 5-μm particle size, with integrated guard column 5×4 mm filled with the same material (Knauer, Berlin, Germany). The mobile phase consisted of 0.02 M ammonium acetate buffer, pH = 6.5 (solvent A), and methanol (solvent B). The injection volume was 50 μL, column temperature was set to 20°C and ultraviolet detection was performed at λ = 230 nm. The elution programme started isocratically with 5% B at a flow rate of 0.5 mL/min over 15 min, followed by a linear gradient from 5% B to 11% B in 23 min at a flow rate of 0.3 mL/min. An external calibration curve of BzGFruK ranging from 5  to 50 μM was used for the quantification of the formed BzGpFruK.

Fig. 2
figure 2

HPLC analysis of incubation mixtures for measurement of the enzymatic activity at t = 0 (A), after t = 2 h (B) and of a mixture with added DMF after t = 2 h (C): peak P BzGpFruK

Full size image

Desalting of the collected fractions of peak P for mass spectrometric analysis was performed with a mobile phase consisting of 0.1% (v/v) formic acid (solvent A), and 0.1% (v/v) formic acid in methanol (solvent B). A linear gradient from 10% B to 25% B in 30 min at a flow rate of 0.5 mL/min was used.

Mass spectrometry

Mass spectrometric analysis was performed with a PerSeptive Biosystems Mariner time-of-flight mass spectrometry (TOF-MS) instrument equipped with an electrospray ionisation source (ESI) in general in the positive (Applied Biosystems, Stafford, USA). Calibration of the mass scale was established using a mixture of bradykinin, angiotensin I and neurotensin.

After appropriate dilution with 1% acetic acid in 50% methanol, the sample was injected at a flow rate of 5 μL/min into the ESI source using a syringe pump for direct ESI-TOF-MS analysis.

The monoisotopic molecular masses were determined using the peak with the lowest m/z ratio (monoisotopic peak) from prominent multiple charged ions and the equation M r=z(M z−1.0078), where M r is the monoisotopic molecular mass, M z is the m/z ratio, z is the number of charges, and 1.0078 is the mass of a proton.

Studies related to the identity of BzGpFruK

The effect of the FN3K-specific inhibitor 1-deoxy-1-morpholino-fructose (DMF) was investigated in order to observe whether peak P is the expected product BzGpFruK. A 500-μL aliquot of HEPES-buffered erythrocyte lysate was mixed with 25 μL 23.0 mM ATP solution (see above), 25 μL 34.5 mM BzGFruK solution (in 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride) as well as 25 μL 460 mM DMF solution (in 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride) (DMF sample) or 25 μL of 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride (control sample). Final concentrations of 1 mM ATP, 1.5 mM BzGFruK and 20 mM DMF were reached in the DMF sample. The samples were incubated for 2 h at 37°C in a water bath and analysed using HPLC as described above. Furthermore, the control sample was analysed for mass spectrometric characterisation of peak P. Therefore, the fractions of peak P were collected manually using HPLC with the ammonium acetate buffer containing eluent system, concentrated and subsequently desalted using HPLC with the formic acid eluent system. The product was collected manually again and investigated utilising direct ESI-TOF-MS.

Studies related to the stability of BzGpFruK

In order to examine the stability at pH = 7.5, a mixture consisting of 1,500 μL HEPES-buffered erythrocyte lysate, 75 μL 22.0 mM ATP solution (see above) and 75 μL 34.5 mM BzGFruK solution (in 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride) was prepared and incubated for 4 h at 37°C in a water bath. Subsequently, 35 μmol disodium ethylendiaminetetraacetate dihydrate (Na2H2EDTA·2H2O) was added to stop the enzyme reaction (final concentration EDTA 21.2 mM). The incubation was continued at 37°C and a sample of 200 μL was taken at time t = 0, 5, 20, 24 and 28 h. The samples were immediately added to 200 μL 10% (w/v) trichloracetic acid to stabilise the BzGpFruK and then analysed by using HPLC as described above. First-order reaction kinetics for the degradation were assumed for the determination of the half-life of BzGpFruK, and the peak area A of BzGpFruK was evaluated using the ln(A t=x /A t=0) versus time plot.

In order to test the stability of BzGpFruK in the acidic solution after the protein precipitation step with trichloracetic acid, an incubation mixture for activity measurement was prepared for HPLC analysis but stored before injection in the HPLC system at 25°C for 129 h.

Determination of K m and V max

The kinetic parameters K m and V max of the FN3K-catalysed conversion of the substrate BzGFruK were graphically examined in a direct plot of initial velocity versus substrate concentration (Michaelis-Menten plot). The calculation was performed by means of non-linear regression utilising ORIGIN software.

In order to measure the initial velocity at varying substrate concentrations, mixtures consisting of 1,500 μL HEPES-buffered erythrocyte lysate, 75 μL 22.0 mM ATP solution (see above) and 75 μL BzGFruK solution (in 200 mM HEPES buffer, pH = 7.5, containing 170 mM sodium chloride) were prepared and incubated at 37°C in a water bath. The final concentrations of the substrate in the incubation mixtures were 0.097, 0.197, 387, 967 or 1,450 mM. After mixing (t = 0) and after 30, 60, 90, 120, 150 and 180 min, samples of 200 μL were taken and analysed for their BzGpFruK content by using HPLC as described above. The determination of initial velocities was performed using BzGpFruK concentration versus time plots in the linear range (steady state) from 30 to 120 min, respectively.

Measurement of haemoglobin (Hb)

The alkaline haematin D-575 method described by Zander et al. [14] was used for the determination of haemoglobin in the HEPES-buffered erythrocyte lysate. Briefly, 3.0 mL of a solution of the so-called AHD reagent (25 g/L Triton X-100 in 0.1 M sodium hydroxide solution) was mixed with 100 μL HEPES-buffered erythrocyte lysate, followed by absorbance measurement of the alkaline haematin D-575 complex at λ = 575 nm. The calibration was performed with solutions of pure chlorohaemin in AHD reagent in the concentration range from 0.017 to 0.148 mM.

Measurement of haemoglobin A1c (HbA1c)

Determination of the HbA1c content in blood samples was performed utilising the automated Bio-Rad Variant II haemoglobin A1c testing system (Bio-Rad Laboratories, Munich, Germany), which is based on HPLC using a cation-exchange column and detection at λ = 415 nm.

Results and discussion

The peptide-bound fructosamine N α-hippuryl-N ε-(1-deoxy-D-fructosyl)lysine (BzGFruK, Fig. 1) was used as substrate for the determination of fructosamin-3-kinase (FN3K) activity in erythrocytes. This compound was chosen with respect to the analysis by RP-HPLC. The structure of BzGFruK contains a benzoyl moiety, which causes the retardation on reversed-phase material and allows the UV detection at λ = 230 nm. The preparation is based on the reaction of the commercially available peptide N α-hippuryllysine with glucose in methanol. Subsequently, the resulting product BzGFruK is isolated using semi-preparative RP-HPLC [13]. The assay is based on the quantitative determination of the enzymatic conversion of the substrate to N α-hippuryl-N ε-(3-phosphofructosyl)lysine (BzGpFruK, Fig. 1). For that purpose, the erythrocytes are isolated from heparin-stabilised blood and lysed in a hypotonic medium. Subsequently, the erythrocyte lysate is diluted with HEPES-buffer pH = 7.5. The substrate is incubated in the resulting lysate in the presence of the cofactor ATP at 37°C for 2 h. In order to stop the enzyme reaction and for preparing the sample for the HPLC analysis, a protein precipitation step with trichloracetic acid is performed. Figure 2 shows the chromatograms of incubation mixtures at time zero and after an incubation time of 2 h. The during-incubation, newly formed peak P has been separated and identified as BzGpFruK. At first, the identity of peak P was investigated by analysis of an incubation mixture that contained DMF, a strong and specific inhibitor of FN3K [2]. As can be seen in Fig. 2, the peak area of peak P is significantly reduced in the chromatogram of the DMF-containing sample. For final clarification, peak P was collected manually, desalted by HPLC using a volatile eluent composition and analysed by means of ESI-TOF-MS. In the mass spectrum of peak P (Fig. 3) was dominated by a [M+H]+ ion with m/z = 550.16, which led to a monoisotopic relative molecular mass of M r=549.15. This result was consistent with the expected value for BzGpFruK (M r=549.17). Thus, the identity of the peak was regarded as sufficiently affirmed. Consequently, quantification of BzGpFruK could be used for the measurement of the FN3K activity. The determination of BzGpFruK is carried out with a calibration curve of BzGFruK, as BzGpFruK is not stable. It is supposed as convention that BzGpFruK and BzGFruK have the same extinction coefficient.

Fig. 3
figure 3

Mass spectrometric analysis of peak P (BzGpFruK) using direct ESI-TOF-MS in positive mode

Full size image

As mentioned above, BzGpFruK, like other 3-phosphofructosylamines, is not stable. The stability under the conditions during the activity assay was therefore checked. The half-life time of BzGpFruK under the conditions of the incubation (37°C, pH = 7.5) was calculated to be 11.1 h. Hence, the extent of degradation is negigibly low during the incubation time of 2 h, if the incubation time is defined in the assay conditions. Furthermore the stability was tested after the precipitation step with trichloracetic acid. No degradation was observed at 25°C during 48 h. Consequently, terminating the enzyme reaction with trichloracetic acid stabilises the analyte BzGpFruK.

In order to characterise FN3K with the substrate BzGFruK, the Michaelis-Menten kinetic parameters K m and V max are of importance. The initial reaction rates at different substrate concentrations were determined and plotted in the Michaelis-Menten plot (Fig. 4). Constant reaction rate was observed for a time period from 0.5 up to 4 h, indicating that 2 h should be sufficient for the assay. Values of K m=151 μM and V max=25 μM/h were calculated using non-linear regression. Delpierre et al. [1] obtained a K m=13.2 μM for the conversion of N ε-fructoselysine by purified FN3K from erythrocytes. Furthermore, using the substrate DMF a tenfold higher K m value for FN3K in the crude erythrocyte lysate compared to the Km value of the purified enzyme (K m=1 μM) was noticed. It was assumed that inhibiting cell components are the reason for this finding. According to that, the K m value for BzGFruK is comparable with the known value of N ε-fructoselysine. A substrate concentration of 1.74 mM was chosen, as activity measurements are run best under conditions of substrate saturation corresponding to about the tenfold K m value.

Fig. 4
figure 4

Michaelis-Menten plot for the determination of K m and V max

Full size image

The presence of sodium chloride in the incubation mixture affects the activity of the enzyme positively and therefore a HEPES-buffer containing 170 mM sodium chloride was used for activity measurement. Increasing the sodium concentration from 100 to 170 mM resulted in an increase of the reaction rate of around 10%. Further increase in sodium concentration had no effect. The observed effect is probably associated with the high chloride concentration (150 mM) found in erythrocytes [15].

In order to improve precision and comparability of the results of the activity measurements, it is important to relate the results to the content of haemoglobin. The so-called alkaline haematin D-575 method with chlorohaemin as primary standard [14] proved very useful for the determination of haemoglobin.

Based on the investigations described above the following activity assay for FN3K from erythrocytes was established. The buffered erythrocyte lysate was incubated with 1.74 mM BzGFruK in the presence of 1 mM ATP for 2 h at 37°C. As kinases use ATP as a complex with Mg2+, further studies must show whether increasing the concentration of magnesia affect the reaction rate. The formed amount of BzGpFruK is subsequently determined using RP-HPLC with UV detection. The results are related to the content of haemoglobin for standardisation and expressed as mU/g Hb, corresponding to nmol BzGpFruK/(min g Hb).

In order to check the new developed assay, FN3K activities were determined in erythrocyte lysate of three female (F1–F3) and four male (M1–M4) normoglycaemic volunteers. The participants of the study were aged 23 or 24, except one male aged 32 (M4). The measured activities varied from about 5 to 14 mU/g Hb and are shown in Fig. 5. The relative standard deviations (RSDs) of the samples ranged between 2 and 6% and corresponded with values of similar enzyme assays. As can be seen, the activities show the same order of magnitude, but differ from each other. In the lysate of the female participant F1 an activity more than twice as high as in that of the female participant F2 was detected. Significant differences exist in the male group, too: e.g. between the lysate of participant M1 and those of the other participants. Because of varying activities within the female and male group, a gender-dependency of the FN3K activity is not supposed. In summary, the participants can be divided into groups with high activity (F1, M1), middle activity (F2, M2, M3) and low activity (F3, M4).

Fig. 5
figure 5

FN3K activity in erythrocyte lysate of female (F1–F3) and male (M1–M4) subjects. Mean value±standard deviation of the procedure

Full size image

These results are in agreement with a recently published investigation. Delpierre et al. [16] found a widely interindividual variability of erythrocyte FN3K activity. The measured activities differ from each other up to a factor of four. This variability was associated with polymorphisms in the FN3K gene. In our opinion the existence of different amounts of inhibiting cell components can also be discussed. So far, the presence of such components is only supposed and their structures are therefore unknown [1]. As the participants in the present study were normoglycaemic with close HbA1C levels between 5.2 and 5.4, glucose states cannot be considered a reason for the variation of the FN3K activity. Equally, Delpierre et al. [16] could not associate the FN3K activity with HbA1c level.

The measured activities are slightly higher in this paper in comparison to the values from Delpierre et al. [16], 5–14 mU/g Hb vs. 1–4 mU/g Hb, respectively. The difference can be attributed to the different substrates, BzGFruK vs. DMF, and assay conditions, e.g. 37°C vs. 30°C, pH = 7.5 vs. pH = 7.8 as well as the diverse buffer and salt concentrations.

Interindividual variability of FN3K activity raises the question of physiological consequences linked with it. As mentioned in the introduction, the FN3K initiates a pathway leading to protein repair of fructosamine-modified proteins, hence preventing further reactions to advanced glycation endproducts. The formation of the potent glycating agent 3-deoxyglucosulose in the course of this reaction may have no adverse effect, since enzymes exist for its detoxification [17, 18]. Thus, a high activity of FN3K should be beneficial in order to avoid long-term consequences of glycation in vivo. It is assumed that differences in the activity are of particular importance in the presence of higher fructosamine levels. A low activity of FN3K could be a disadvantageous factor with regard to the development of diabetic complications. In this context, the hypothesis may be of interest whether genetically determined activity of FN3K may represent a risk factor for diabetes type II. Further investigations should deal with a putative relationship between the activity of FN3K and the level of fructosamines on one hand and individual advanced glycation endproducts in diabetes on the other. Compared to known methods described in the literature, the assay presented in this paper can be performed in all laboratories using HPLC with UV detection. Radioactive substances, radiological laboratory equipment or an NMR instrument are not required. Consequently, it should now be possible to perform FN3K analysis in a larger number of samples.

Conclusion

We have established a new and relatively simple assay for the determination of the activity of the deglycating enzyme FN3K in human erythrocyte lysate. First measurements confirmed significantly different individual activities. The good applicability of the activity assay should facilitate further studies, which examine the potentially protective role of the FN3K in the development of diabetic complications leading to new insights into this disease.