Studies of Binding by Sulfonylureas with Glyoxal- and Methylglyoxal-Modified Albumin by Immunoextraction using Affinity Microcolumns

Abstract

Diabetes is characterized by elevated levels of blood glucose, which can result in the modification of serum proteins. The modification of a protein by glucose, or glycation, can also lead to the formation of advanced glycated end-products (AGEs). One protein that can be modified through glycation and AGE formation is human serum albumin (HSA). In this study, immunoextraction based on polyclonal anti-HSA antibodies was used with high-performance affinity microcolumns to see how AGE-related modifications produced by glyoxal (Go) and methylglyoxal (MGo) affected the binding of HSA to several first- and second-generation sulfonylureas, a class of drugs used to treat type II diabetes and known to bind to HSA. With this approach, it was possible to use a single platform to examine drug interactions with several preparations of HSA. Each applied protein sample could be used over 20–50 experiments, and global affinity constants for most of the examined drugs could be obtained in less than 7.5 min. The binding constants measured for these drugs with normal HSA gave good agreement with global affinities based on the literature. Both Go- and MGo-related modifications at clinically relevant levels were found by this method to create significant changes in the binding by some sulfonylureas with HSA. The global affinities for many of the drugs increased by 1.4-fold or more; gliclazide and tolazamide had no significant change with some preparations of modified HSA, and a small-to-moderate decrease in binding strength was noted for glibenclamide and gliclazide with Go-modified HSA. This approach can be adapted for the study of other drug-protein interactions and alternative modified proteins by altering the antibodies that are employed for immunoextraction and within the affinity microcolumn.

1. Introduction

Human serum albumin (HSA) is the most abundant carrier protein in blood, with a normal concentration of 42 g/L [1]. The binding of small solutes such as fatty acids, hormones, and drugs with this carrier protein can affect the absorption, metabolism, distribution, and excretion of these agents [1–5]. HSA has a molar mass of 66.5 kDa and is composed of 585 amino acids and 17 disulfide bonds [1,2,6]. This protein also has two major binding pockets for drugs, which are often referred to as Sudlow sites I and II [1–6].

Elevated levels of glucose due to diabetes can lead to the modification of HSA through non-enzymatic glycation (i.e., leading to the production of Amadori products) or subsequent reactions that lead to advanced glycation end-products (AGEs) [3,7,8]. In the case of AGEs, active α-oxaloaldehydes such as glyoxal (Go) and methylglyoxal (MGo) can be formed that react with lysine or arginine residues on proteins, as well as the N-terminus [3,9–11]. Examples of AGEs that can occur through this process are methylglyoxal-derived hydroimidazolone isomer 1 (MG-H1) and glyoxal-derived hydroimidazolone isomer 1 (G-H1) (see Figure 1) [3,10,12]. The modification of HSA by glucose and related agents has been of recent interest because it has been demonstrated that at least some of these modifications can affect the structure of HSA and its function as a carrier agent for drugs [3].

Figure 1.

Examples of reactions involved in the formation of advanced glycation end-products (AGEs).

Sulfonylurea drugs are commonly given to patients with type II diabetes to reduce the levels of glucose in blood [7]. The general structure of a sulfonylurea and several specific compounds in this class are shown in Figure 2. The earliest drugs from this group were the first-generation sulfonylureas; examples include acetohexamide, chlorpropamide, tolazamide, and tolbutamide. These drugs were followed by the development of second-generation sulfonylureas, such as glibenclamide, gliclazide, and glipizide [13–17]. One difference in these two groups is second-generation sulfonylureas tend to have higher activities, which allows them to be given in smaller dosages than first-generation sulfonylureas and lowers the risk of side effects such as hypoglycemia [7]. All the sulfonylureas in Figure 2 are known to have significant binding to HSA, and it has been shown that these interactions occur at both Sudlow sites I and II [13–17]. One of the drugs in Figure 2 (i.e., glibenclamide) also binds to the digitoxin site of HSA [15].

Figure 2.

Structures of the sulfonylurea drugs that were examined in this study. The diagram at the top shows the basic structure for this class of drugs.

Several previous reports have used covalently immobilized HSA for binding studies with sulfonylureas and other solutes [13–17]. This study will instead use microcolumns containing polyclonal anti-HSA antibodies to non-covalently capture and retain HSA or modified forms of this protein [18,19]. The captured HSA will be used in zonal elution studies to estimate and compare the overall binding strength for these protein samples with sulfonylurea drugs, as illustrated in Figure 3. By using this immunoextraction approach, a single microcolumn may be used to study binding by several samples of HSA with a series of drugs. In addition, it should be possible to elute the captured protein and regenerate the microcolumn for use with a different preparation of HSA or a fresh portion of the same protein sample [18,19].

Figure 3.

Scheme for use of immunoextraction microcolumns and zonal elution experiments to examine the binding of drugs with HSA or modified forms of this protein. A sample containing HSA or AGE-modified HSA is first applied to the immunoextraction microcolumn. After the microcolumn has been washed to remove any non-retained agents, a small sample of the drug is injected, and its retention factor is measured. After multiple injections of the drug, the retained HSA can be eluted, and the microcolumn regenerated prior to application of a new portion or sample of HSA.

In previous studies it has been shown that the modification of HSA by glycation can affect the interactions and binding strengths of sulfonylurea drugs to this protein [13–17]. However, there is little information on how the modification of HSA with Go or MGo affects these processes. This report will seek to determine how the overall binding strength is altered in going from normal, unmodified HSA to Go- or MGo-modified HSA for a series of first- and second-generation sulfonylurea drugs by using immunoextraction based on affinity microcolumns. The binding properties of these microcolumns will be evaluated, as well as their stability and precision. These microcolumns will be used in immunoextraction to bind normal or modified HSA and to use these adsorbed proteins in zonal elution-based binding studies. The binding constants that are obtained by this approach for sulfonylurea drugs will be compared to estimates based on prior results acquired by other methods for normal, unmodified HSA. Once validated, this method will be used with the same drugs and Go- or MGo-modified HSA. The data acquired in this project should give a better insight into how AGEs may alter the binding of sulfonylurea drugs with HSA. These data should also indicate the advantages or limitations of immunoextraction microcolumns as possible tools for drug binding studies, as may be used in the future for personalized medicine in type II diabetes [14].

2. Materials & Methods

2.1. Materials

The polyclonal anti-albumin antibodies (goat, fractionated human antiserum, product A1151), Protein G-Sepharose 4B fast flow (recombinant protein expressed in E. coli), immunoglobulin G (goat, ≥ 95% pure) and HSA (essentially fatty acid free, ≥ 96%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). The racemic warfarin (≥ 98%), acetohexamide (99%), gliclazide (≥ 98%), glipizide (˃ 96%), glibenclamide (≥ 99%), and tolbutamide (99.8%) were also acquired from Sigma-Aldrich. The chlorpropamide (≥ 99%) and tolazamide (≥ 99%) were purchased from Santa Cruz Biotechnology (Dallas, TX, USA). The Nucleosil Si-300 and Si-1000 silica (7 μm, particle diameter; pore size, 300 or 1000 Å, respectively) were acquired from Macherey-Nagel (Duren, Germany). All buffers and aqueous solutions were prepared using water from a Milli-Q Advantage 10 A Water system and were filtered using 0.2 μm nylon membranes (EMD Millipore Corporation, Billerica, MA, USA).

2.2. Instrumentation

Chromatographic studies were conducted using a Jasco 2000 series HPLC system (Tokyo, Japan) and a Harvard apparatus PHD Ultra syringe pump (Hilston, MA, USA). The components of the Jasco system were as follows: a PU-2080 pump, an AS-2057 autosampler, a DG-2080 degasser, a CO-2067 column oven, and a UV-2075 Plus absorbance detector. This HPLC system also employed a Rheodyne Advantage PF six-port valve (Cotati, CA, USA), which was used to alternate flow between the Harvard syringe pump and HPLC pump. The HPLC plus syringe pump system was operated using LC-Net and ChromNav V 1.18 software from Jasco. Chromatograms were analyzed by using PeakFit 4.12 software from Jandel Scientific (San Rafael, CA, USA) and Microsoft Excel 2016 (Redmond, WA, USA).

2.2. Development of immunoextraction microcolumns

The anti-HSA immunoextraction microcolumns were prepared by using polyclonal anti-HSA antibodies that were obtained from anti-HSA goat serum. Isolation of these antibodies in the final method was performed by first dissolving 100 mg of the lyophilized anti-HSA serum in 2.0 mL of pH 7.4, 0.067 M potassium phosphate buffer. This solution was mixed by vortexing for 3 min, followed by the addition of 0.20 g Nucleosil Si-300 silica which contained ~11–12 μg of immobilized HSA. This slurry was slowly mixed for 2 h at room temperature and centrifuged at room temperature for 3 min at 7500 rpm. The supernatant was discarded, and 3 mL of pH 7.4, 0.067 M potassium phosphate buffer was added to the silica and mixed by vortexing for 1 min. This slurry was centrifuged and washed three more times in the same manner. After the last centrifugation step, 2.0 mL of pH 2.5, 0.067 M potassium phosphate buffer was added to the silica to release anti-HSA antibodies from the immobilized HSA [20]. This new mixture was vortex-mixed for 1 min and centrifuged for 3 min at 7500 rpm. The supernatant was removed and stored in a separate container. The washing and centrifugation step with the pH 2.5 buffer was repeated three times. The pH of the combined supernatant from the last set of washing steps was adjusted to pH 6.0 by adding pH 8.0, 0.067 M potassium phosphate buffer. The antibodies in this solution were used immediately for immobilization.

An alternative approach that was initially used to isolate the anti-HSA antibodies employed protein G that was immobilized to Sepharose. In this procedure, 100 mg of goat anti-HSA serum was dissolved in 2.0 mL of pH 7.4, 0.067 M phosphate buffer and mixed with 2.0 mg of a Protein G-Sepharose 4B support, followed by gentle shaking for 2 h at room temperature. The non-bound serum components were washed away from the Protein G-Sepharose 4B particles with pH 7.4, 0.067 M phosphate buffer. The captured antibodies were then removed by washing the particles with pH 2.5, 0.067 M phosphate buffer at room temperature. The pH of the extracted antibody solution was adjusted to pH 6.0, as described in the previous paragraph, for immediate use of these antibodies in immobilization.

The anti-HSA antibodies were immobilized onto Nucleosil Si-1000 through the Schiff base method, as described previously [21]. A control support, based on the same type of silica, was prepared in a similar manner but with no antibodies being added during the immobilization step. The protein content of the anti-HSA support was determined in triplicate by utilizing a micro bicinchoninic acid (BCA) protein assay kit from Thermo Fisher (Waltham, MA, USA). Goat immunoglobulin G was used as the standard, and the control support was employed as the blank.

The final anti-HSA support was packed into a 1.0 cm × 2.1 mm I.D. stainless steel microcolumn at 4000 psi (27.6 MPa) using pH 7.4, 0.067 M potassium phosphate buffer as the packing solution. The microcolumn was stored in pH 7.4, 0.067 M potassium phosphate buffer at 4°C. A control microcolumn with an identical size was prepared and stored in the same manner as the anti-HSA microcolumn. The amount of active antibodies in the anti-HSA microcolumn was measured in triplicate by frontal analysis [18] using a 5.0 μM solution of HSA in pH 7.4, 0.067 M potassium phosphate buffer that was applied at 0.10 mL/min. In this measurement, the active antibody content was determined by subtracting the difference in the midpoint of the breakthrough curves between the anti-HSA microcolumn and the control microcolumn [18].

2.3. Preparation of drug solutions

All first-generation sulfonylurea drugs (i.e., acetohexamide, chlorpropamide, tolazamide, and tolbutamide) and the second-generation sulfonylureas (i.e., gliclazide and glipizide) were dissolved at a typical concentration of 50 μM in pH 7.4, 0.067 M potassium phosphate buffer by stirring this mixture for 12 h at room temperature. Glibenclamide, due to its low solubility [22], was first dissolved in pH 11.0, 0.067 M potassium phosphate buffer by stirring for 4 h at room temperature to obtain a drug concentration of 40 μM. The pH of this solution was then slowly adjusted to pH 7.4 by adding pH 2.5, 0.067 M potassium phosphate buffer, giving a final concentration for glibenclamide of approximately 30 μM. All drug solutions were filtered prior to use in HPLC by using an Acrodisc PVDF syringe filters (0.2 μm pore size) from Waters (Milford, MA, USA). The drug solutions were stored at 4°C and used within two weeks of preparation.

2.4. Preparation of methylglyoxal- and glyoxal-modified HSA

Protein modification was performed under sterile conditions by dissolving a commercial sample of normal, unmodified and purified HSA at 42 g/L (i.e., a typical physiological concentration) [1] in pH 7.4, 0.067 M potassium phosphate buffer that contained 1 mM of sodium azide (i.e., as used to avoid bacterial growth). This HSA solution was combined with 130 nM or 210 nM of Go for the preparation of Go-HSA1 or Go-HSA2, respectively, which represented typical serum levels seen for Go in healthy adults or those with type II diabetes, respectively [12,23–26]. The same type of HSA solution was combined with 40 nM or 120 nM of MGo for the preparation of MGo-HSA1 or MGo-HSA2, with these levels of MGo corresponding to those seen in healthy adults or in pre-diabetes/type II diabetes, respectively [12,25,27]. These mixtures were incubated at 37°C over 4 weeks to mimic the exposure time of HSA to AGEs in the circulation [19,20]. The modified HSA was purified by using size exclusion columns to remove any excess or unreacted modifying agents, as described for the preparation of glycated HSA [19,20]. The modified HSA was lyophilized and stored at −80°C until further use. The level of modification for each sample was examined using fluorometric assays to determine the amount of free lysines/primary amines or free arginine residues that remained when compared to the original preparation of normal, unmodified HSA [28,29]. The extent of lysine/primary amine modification for Go-HSA1 and Go-HSA2 vs normal HSA was estimated by this approach to be 0.15 (± 0.11) and 0.55 (± 0.12) mol/mol HSA, respectively; the levels of arginine modification in the same samples were 0.20 (± 0.06) and 0.74 (± 0.05) mol/mol HSA. The lysine/primary amine modification levels for MGo-HSA1 and MGo-HSA2 were 0.43 (± 0.14) and 1.12 (± 0.15) mol/mol HSA, and their levels of arginine modification were 0.24 (± 0.10) and 0.88 (± 0.11) mol/mol HSA.

2.5. Chromatographic studies

The scheme for the immunoextraction and binding assay that was developed and used in this study is depicted in Figure 3. First, an anti-HSA immunoextraction microcolumn was equilibrated in pH 7.4, 0.067 M potassium phosphate buffer at 0.10 mL/min and 37°C for 60 min. A solution of 5.0 μM HSA or AGE-modified HSA in pH 7.4, 0.067 M potassium phosphate buffer was applied onto the immunoextraction microcolumn at 0.10 mL/min for 20 min by using a syringe pump. Once the microcolumn had been allowed to bind the applied HSA, any non-bound protein was removed by washing with pH 7.4, 0.067 M potassium phosphate buffer for 30 min at 0.10 mL/min (i.e., conditions determined by monitoring elution of the non-bound protein). A 20.0 μL sample containing 10.0 μM of the desired drug in the same pH 7.4 buffer was then injected while the absorbance of the eluent was monitored (Note: equivalent retention was seen for the injection of 2.5 μM drug samples, indicating that linear elution conditions were present) [18]. The detection wavelengths were as follows: sodium nitrate, 205 nm, tolazamide, gliclazide, and glipizide, 226 nm; tolbutamide, 228 nm; chlorpropamide, 231 nm; glibenclamide, 242 nm; acetohexamide, 252 nm; and warfarin, 308 nm. All injections were made in triplicate. After 20 injections of these drug samples had been made, the captured HSA was released and the microcolumn was regenerated by applying pH 2.5, 0.10 M potassium phosphate buffer at 0.10 mL/min for 30 min. The entire process was then repeated by applying a fresh solution of HSA or modified HSA to the immunoextraction microcolumn. Control experiments were carried out in the same manner in the absence of HSA or modified HSA in the microcolumn. Analysis of the peaks in the resulting chromatograms was conducted with Peak Fit 4.12.

3. Results and Discussion

3.1. Purification of anti-HSA antibodies for immunoaffinity column

The purification of anti-HSA antibodies from goat serum was performed by using two extraction methods based on 1) protein G that was immobilized to Sepharose or 2) HSA that was immobilized onto silica (see Section 2.2 for details). The efficiencies of these extraction methods were compared by determining the antibody/protein content of the final silica supports. The protein content when using protein G Sepharose for extraction was 69.8 (± 1.3) μg antibodies/mg silica (average, n = 3 batches), while the protein content achieved using the HSA silica for extraction was 30.8 (± 1.3) μg antibodies/mg silica (average, n = 6 batches). Each type of antibody silica was packed into a 1.0 cm × 2.1 mm I.D. microcolumn, and the amount of active anti-HSA antibodies was determined by performing frontal analysis with HSA (see Figure 4). These experiments revealed that 0.12 (± 0.01) nmol HSA could be bound by the antibody microcolumn prepared by using protein G for extraction. However, 0.69 (± 0.02) nmol HSA was initially bound by the microcolumn made with antibodies that were extracted by using the HSA silica (i.e., a value almost six-fold larger than obtained with extraction based on protein G). Although the use of protein G for extraction gave a higher total amount of immobilized antibodies, the higher activity obtained for binding to HSA when using HSA silica for extraction indicated that not all the antibodies in the original serum were specific for HSA. Based on these results, anti-HSA antibodies were purified by adsorbing them to HSA silica in the remainder of this study.

Figure 4.

Typical frontal analysis curves obtained for normal, unmodified HSA on an anti-HSA immunoextraction microcolumn (gray) and a control microcolumn (black). These results are for a 5.0 μM solution of HSA in pH 7.4, 0.067 M phosphate buffer that was applied to 1.0 cm × 2.1 mm I.D. microcolumns at 0.10 mL/min and 37°C.

Table 1 shows the results obtained when the binding capacities for Go- and MGo-modified HSA were compared to those for normal, unmodified HSA on a single anti-HSA immunoextraction microcolumn, as prepared by using HSA silica for antibody isolation. The various forms of HSA all had similar binding capacities on the microcolumn, with values in the range of 0.42 to 0.69 nmol. These results indicated that polyclonal anti-HSA antibodies could be used to capture and bind Go- and MGo-modified HSA in a manner equivalent to that for normal HSA.

Table 1.

Binding of normal HSA and AGE-modified HSA to an anti-HSA microcolumn

Type of HSA	Amount of bound HSA (nmol)a
Normal HSA
New column (1–3 protein application cycles)	0.69 (± 0.02)
Mid-term of column use (11–13 protein application cycles)	0.54 (± 0.01)
End of column use (21 protein application cycles)	0.44
Go-modified HSA
Modification at Go levels seen in controlled diabetes (Go-HSA1)	0.54 (± 0.01)
Modification at Go levels seen in advanced diabetes (Go-HSA2)	0.53 (± 0.03)
MGo-modified HSA
Modification at normal levels of MGo (MGo-HSA1)	0.42 (± 0.02)
Modification at MGo levels seen in controlled-to-advanced diabetes (MGo-HSA2)	0.49 (± 0.02)

^aThe values in parentheses represent a range of ± 1 S.D. (n = 3). The results for Go-modified HSA were obtained during the mid-term period of column use. The results for MGo-modified HSA were acquired towards the final period of column use.

There was a decrease in binding capacity of 36% for normal, unmodified HSA over 21 cycles for protein application and elution on an anti-HSA immunoextraction microcolumn, or an average decrease of 1.7% per protein application/elution cycle. This gradual change was probably caused by a loss in activity for some of the anti-HSA antibodies over multiple application and elution cycles, but may also reflect a slow loss in antibody content within the column over time. Even though this effect was relatively small, all estimates of global affinities made later in this study were corrected for this change between application cycles and/or between different types of HSA. This correction was made by directly measuring the amount of adsorbed HSA through frontal analysis during the course of each new application cycle and then using this amount in the calculation of any global affinity constants that were determined over that same cycle.

3.2. Drug-binding studies using HSA adsorbed to anti-HSA microcolumns

3.2.1. General approach

Drug-binding studies using HSA or AGE-modified HSA and sulfonylurea drugs were carried out by using zonal elution. Zonal elution is often used for the study of solute-protein interactions in more traditional modes of affinity chromatography [30,31]. This method involves the injection of a small amount of a solute onto a column while the solute’s retention time is measured [30]. Advantages of zonal elution include its use of a small amount of the injected solute, its capability of providing fast analysis times, and its ability to analyze differences in binding strength between various solutes and a ligand [30,31].

When used in later sections of this report for drug binding studies, samples of HSA or modified HSA were adsorbed by the immunoextraction microcolumns at 0.10 mL/min to maximize the capture of HSA (as determined by frontal analysis), while also minimizing the time needed for the application of a protein sample. It has been noted for this same type of microcolumn that the capture efficiency is around 82–93% for normal or glycated HSA at 0.10 mL/min [18]. Increasing the flow rate to 0.25 mL/min reduced the capture efficiency to less than 70% for normal and unmodified HSA, while using a flow rate below 0.10 mL/min gave only a slight increase in capture efficiency.

3.2.2. Stability of adsorbed HSA

The long-term stability of HSA that had been adsorbed onto an anti-HSA immunoextraction microcolumn was determined by making repeated injections of racemic warfarin. Warfarin was used because it has a well-characterized interaction at Sudlow site I of HSA, with an average association equilibrium constant of 2.4 × 10⁵ M⁻¹ for its two enantiomers at pH 7.4 and 37°C [32]. Warfarin was injected over 50 times at 0.10 mL/min onto an anti-HSA immunoextraction microcolumn containing adsorbed HSA. Some typical chromatograms that were obtained are provided in Figure 5. Changes in the retention time of warfarin of 10–11% were observed after 27 injections onto the microcolumn. Over 50 injections, the change in the retention time of warfarin was 20%. To avoid any major changes in the retention measured for other drugs, all later experiments were conducted with HSA that was applied and used over a maximum of 20 injections. Under these conditions there was a change in the retention for warfarin of only 4–5%, which was within the typical precision of the retention measurements that were made for sulfonylurea drugs on the system.

Figure 5.

Comparison of the retention for warfarin over a series of injections following a single application of normal, unmodified HSA onto an anti-HSA immunoextraction microcolumn.

3.2.3. Drug binding studies for sulfonylureas with normal, unmodified HSA

Zonal elution studies examining the overall binding between sulfonylurea drugs and normal, unmodified HSA were carried out by injecting each drug onto an anti-HSA immunoextraction microcolumn that contained adsorbed HSA. Control experiments were also conducted in which the same drugs were injected onto the immunoextraction microcolumn in the absence of any adsorbed HSA. Some typical chromatograms that were obtained are shown in Figure 6. The corresponding retention factors that were measured are provided in Table 2. Under the conditions used in this study, data for most drug injections could be obtained within 3.0–7.5 min in the presence of adsorbed HSA. The only exception was glibenclamide, which had the highest retention of the drugs and eluted within 20 min.

Figure 6.

Retention of acetohexamide on an anti-HSA immunoextraction microcolumn in the presence of (from left-to-right) no adsorbed HSA, an adsorbed sample of normal, unmodified HSA, and adsorbed samples of Go-HSA2 or MGo-HSA2. The void time of the microcolumn and system was 1.7 min, which is indicated by the dashed vertical line. In each case, the acetohexamide was injected in the presence of pH 7.4, 0.067 M phosphate buffer at 0.10 mL/min at 37°C. The chromatograms for Go-HSA2 and MGo-HSA2 have been corrected for a non-retained peak. The acetohexamide peaks have been normalized to have the same approximate area as seen for acetohexamide in the absence of adsorbed HSA.

Table 2.

Average retention factors measured for various sulfonylurea drugs on anti-HSA immunoextraction microcolumns in the presence and absence of normal HSA

Drug	Retention factor on immunoextraction microcolumna	Retention factor on HSA/immunoextaction microcolumna	Specific retention factor due to HSAb
Acetohexamide	0.23 (± 0.01)	3.10 (± 0.16)	2.87 (± 0.16)
Chlorpropamide	0.59 (± 0.04)	1.45 (± 0.08)	0.86 (± 0.09)
Glibenclamide	3.06 (± 0.05)	33.6 (± 1.5)	30.6 (± 1.5)
Gliclazide	1.19 (± 0.03)	2.36 (± 0.12)	1.17 (± 0.12)
Glipizide	0.97 (± 0.04)	4.13 (± 0.19)	3.16 (± 0.20)
Tolazamide	1.47 (± 0.03)	1.94 (± 0.10)	0.47 (± 0.10)
Tolbutamide	1.92 (± 0.15)	3.79 (± 0.24)	1.88 (± 0.28)

^aThe values in parentheses represent a range of ± 1 S.D. (n = 3).

^bThe specific retention factor due to HSA represents the difference between the retention factors that were measured for a drug in the presence of adsorbed HSA on an anti-HSA immunoextraction microcolumn and on the same microcolumn in the absence of any HSA. The values in parentheses for these results were determined by error propagation.

The specific retention factor (k) for each drug due to the adsorbed HSA was found by taking the difference between the total retention factors that were measured on the immunoextraction microcolumn in the presence and absence of HSA. Using this difference made it possible to correct for any non-specific interactions the drug may have had with the support or immobilized antibodies. This correction was minimal (i.e., 7–23% of the total binding; average, 13%) for acetohexamide, glibenclamide, and glipizide. For the other drugs that were examined, these non-specific interactions made up 41–76% (average, 56%) of the total retention; these interactions would have led to high estimates of the binding strengths for these drugs with HSA if a correction had not been made for this secondary binding. The final specific retention factors that were obtained due to HSA had relative precisions of ± 5–21% (average, ± 10%). Although this level of precision was lower than what is normally obtained when using frontal analysis or longer affinity columns for drug binding studies [33–35], it was more than sufficient for use in this present study to screen for changes that may occur in the overall binding of sulfonylurea drugs with normal HSA vs. modified HSA.

The specific retention factors were used to calculate the global affinity constant (nK_a’) for each drug with HSA by using Eq. 1 [31].

$$k=\frac{n{K}_{a^{\prime }}{m}_L}{V_m}$$ (1)

In this equation, m_L is the moles of active protein within the column, and V_m is the void volume of the column. The value of m_L was obtained by frontal analysis during the application of each fresh sample of HSA to the microcolumn, with the assumption that all the adsorbed HSA had active drug binding sites. The value of V_m was determined by injecting a non-retained marker (i.e., sodium nitrate) into the system in the presence and absence of the immunoextraction microcolumn [13–19].

Table 3 shows the values of nK_a’ that were obtained in this fashion for the sulfonylureas with normal, unmodified HSA. These values had precisions of ± 6.6–23% (average, ± 12%) and gave nK_a’ values that differed by an absolute amount of only 0.6–28% (average, 9.1%) when compared to global affinities that were calculated from binding data obtained by other methods for the same drugs with normal HSA [13–17]. In addition, the measured and reference values for nK_a’ were found to be statistically identical by a Student’s t-test at the 95% confidence level.

Table 3.

Global affinity constants (nK_a) for various sulfonylurea drugs with normal, unmodified HSA adsorbed within anti-HSA immunoextraction microcolumns^a

Drug	Experimental value for nKa (M−1)	Calculated value for nKa from literature (M−1) [Ref.]
Acetohexamide	1.73 (± 0.12) × 105	1.72 (± 0.11) × 105 [13,34]
Chlorpropamide	5.18 (± 0.68) × 104	5.90 (± 0.45) × 104 [14]
Glibenclamide	1.84 (± 0.12) × 106	1.44 (± 0.50) × 106 [15]
Gliclazide	7.08 (± 0.81) × 104	7.99 (± 1.20) × 104 [13]
Glipizide	2.41 (± 0.15) × 105	2.57 (± 0.80) × 105 [16]
Tolazamide	2.83 (± 0.64) × 104	2.80 (± 0.57) × 104 [17]
Tolbutamide	1.13 (± 0.18) × 105	1.08 (± 0.03) × 105 [13,33]

^aThe numbers in parentheses represent a range of ± 1 S.D. (n = 3). The calculated values are based on data for the specific or overall binding sites for the listed drug in the given reference. All measured and calculated values for nK_a are based on data obtained at pH 7.4 and 37°C.

Several advantages were noted for this approach versus the more common method of frontal analysis that is often used in affinity chromatography to examine the overall binding strength for drug-protein interactions [5,8]. For instance, the use of zonal elution with immunoextraction microcolumns made it possible to quickly obtain estimates of nK_a’ by using only a few injections of small drug samples. This is in contrast to the frequent need to utilize a large number of drug solutions at various concentrations in frontal analysis [14–17], which adds significantly to both the overall analysis time and amount of drug that is needed for a binding study [18,19]. The ability to periodically release the adsorbed protein and to apply a fresh portion to the immunoextraction microcolumn is another advantage of the method used in this study [18].

3.2.4. Binding studies for sulfonylureas with AGE-modified HSA

Zonal elution studies were next carried out by using adsorbed HSA that had been modified with Go or MGo. Table 1 shows the amounts of modified HSA that were adsorbed to an immunoextraction microcolumn during this process, as discussed in Section 3.1. Some typical chromatograms that were obtained with these adsorbed samples are included in Figure 6. Results were obtained with the adsorbed samples of modified HSA within 2.5–7.0 min of injection for most of the drugs examined; the exception was again glibenclamide, which had strong retention and eluted within 35 min in the presence of the MGo-modified forms of HSA.

The global affinity constants that were obtained for the sulfonylurea drugs with MGo-modified HSA are provided in Table 4. These values had precisions of ± 2.4–11% (average, ± 5.2%), which were similar to the precisions noted in Section 3.2.3 with normal, unmodified HSA. All the global affinity constants measured for MGo-HSA1 (e.g., HSA modified at levels of MGo in serum for healthy adults) were larger than the values seen for the same drugs with normal, unmodified HSA. These increases ranged from 1.35- to 3.45-fold (average, 2.20-fold) and were all significant at the 95% confidence level. The global affinities measured for MGo-HSA2 (i.e., HSA modified at MGo serum levels seen in prediabetes or type II diabetes) also increased for most of these drugs when compared to values measured with normal, unmodified HSA. For MGO-HSA2, the global affinities for drugs other than gliclazide increased by 1.64- to 2.67-fold (average, 2.18-fold) vs. normal HSA, which were significant differences at the 95% confidence level. The corresponding value for gliclazide with MGo-HSA2 gave an apparent relative increase of 1.21 (± 0.15) vs. normal HSA; this difference was not significant at the 95% confidence level but was significant at the 90% confidence level.

Table 4.

Global affinity constants for various sulfonylurea drugs with MGo-modified HSA^a

Drug	nKa for MGo-HSA1, modified at normal MGo levels (M−1)b	Relative value of nKa for MGo-HSAl vs. normal HSA	nKa for MGo-HSA2, modified at prediabetic/diabetic MGo levels (M−1)	Relative value of nKa for MGo-HSA2s vs. normal HSA
Acetohexamide	2.34 (± 0.06) × 105	1.35 (± 0.10)	4.21 (± 0.10) × 105	2.43 (± 0.18)
Chlorpropamide	7.55 (± 0.45) × 104	1.46 (± 0.20)	1.03 (± 0.05) × 105	1.99 (± 0.26)
Glibenclamide	6.19 (± 0.21) × 106	3.36 (± 0.25)	4.93 (± 0.17) × 106	2.67 (± 0.20)
Gliclazide	1.26 (± 0.08) × 105	1.79 (± 0.23)	8.55 (± 0.44) × 104	1.21 (± 0.15)c
Glipizide	6.58 (± 0.20) × 105	3.45 (± 0.28)	4.87 (± 0.15) × 105	2.55 (± 0.21)
Tolazamide	6.41 (± 0.65) × 104	2.27 (± 0.56)	4.65 (± 0.51) × 104	1.64 (± 0.41)
Tolbutamide	1.98 (± 0.11) × 105	1.75 (± 0.30)	2.05 (± 0.10) × 105	1.81 (± 0.30)

^aThese nK_a values were obtained at pH 7.4 and 37°C. The numbers in parentheses represent a range of ± 1 S.D. (n = 3).

^bThe normal MGo levels represented those seen in serum for a healthy adult [12,25,27].

^cThe two values used in this comparison were not significantly different at the 95% confidence level but were different at the 90% confidence level.

Table 5 shows the global affinity constants that were measured for the same sulfonylurea drugs with Go-modified HSA. These binding constants again had precisions of ± 4.2–26% (average, ± 10.5%). In the case of Go-HSA1 (i.e., HSA modified at levels of Go seen in serum for healthy adults), four of the sulfonylurea drugs (i.e., acetohexamide, chlorpropamide, glipizide, and tolbutamide) gave an increase in nK_a’ versus the values seen for normal, unmodified HSA; these increases spanned from 1.44- to 1.89-fold and were significant at the 95% confidence level. Glibenclamide gave a global affinity constant for Go-HSA1 that was 0.37-times the value for normal HSA, and gliclazide gave a value that was 0.73-times that of normal HSA (i.e., differences that were also significant at the 95% confidence level). Tolazamide had an apparent change in global affinity of 1.19-times the value observed for normal HSA, but this difference was not significant at either the 95% or 90% confidence levels.

Table 5.

Global affinity constants for various sulfonylurea drugs with Go-modified HSA^a

Drug	nKa for Go-HSA1, modified at normal Go levels (M−1)b	Relative value of nKa for Go-HSA1 vs. normal HSA	nKa for Go-HSA2, modified at diabetic Go levels (M−1)	Relative value of nKa for Go-HSA2 vs. normal HSA
Acetohexamide	3.21 (± 0.24) × 105	1.85 (± 0.20)	3.52 (± 0.23) × 105	2.03 (± 0.20)
Chlorpropamide	7.49 (± 0.94) × 104	1.44 (± 0.25)	7.81 (± 0.86) × 104	1.51 (± 0.25)
Glibenclamide	6.75 (± 0.54) × 105	0.37 (± 0.04)	5.57 (± 0.61) × 105	0.30 (± 0.03)
Gliclazide	5.15 (± 0.83) × 104	0.73 (± 0.14)	6.42 (± 0.48) × 104	0.91 (± 0.12)c
Glipizide	2.99 (± 0.23) × 105	1.57 (± 0.17)	9.52 (± 0.56) × 105	4.99 (± 0.48)
Tolazamide	3.38 (± 0.88) × 104	1.19 (± 0.29)c	6.58 (± 0.77) × 104	2.33 (± 0.59)
Tolbutamide	2.13 (± 0.27) × 105	1.89 (± 0.38)	7.95 (± 0.48) × 105	7.03 (± 1.17)

^aThese values were obtained at pH 7.4 and 37°C. The numbers in parentheses represent a range of ± 1 S.D. (n = 3).

^bThe normal Go levels represented those seen in serum for a healthy adult [12,23–26].

^cThe two values used in this comparison were not significantly different at either the 95% or 90% confidence levels.

Similar trends to those seen for Go-HSA1 were observed for Go-HSA2 (i.e., HSA modified at levels of Go seen in diabetic serum). Five sulfonylurea drugs (i.e., acetohexamide, chlorpropamide, glipizide, tolazamide, and tolbutamide) gave an increase in global affinity for Go-HSA2 versus normal, unmodified HSA; these increases ranged from 1.51- to 7.03-fold and were all significant at the 95% confidence level. Glibenclamide again gave a decrease in its global affinity constant, in this case corresponding to a decrease in nK_a’ to a value 0.30-times that seen for normal HSA (i.e., a difference significant at the 95% confidence level). Gliclazide gave an apparent change in the global affinity constant of 0.91-times that of normal HSA, but this difference was not significant at either the 95% or 90% confidence levels.

The results in Tables 4 and 5 were next compared to changes in binding constants that have been measured for the same sulfonylurea drugs with glycated HSA. For acetohexamide and chlorpropamide, the high affinity sites for samples of glycated HSA have been found to have overall binding constants that were up to 1.5- to 1.6-times the values for normal HSA [14,34]. The same drugs had a 1.4- to 2.4-fold increase in their global affinity for MGo-modified HSA and a 1.4- to 2.0-fold increase in affinity for Go-modified HSA. Glibenclamide has been reported to have an increase of 1.4-fold in the binding strength for its high affinity sites in a sample of glycated HSA that represented advanced diabetes [15]. The same drug in this present report gave an increase in global affinity of 2.7-fold for MGo-HSA2 and a 70% decrease for Go-HSA2.

The high affinity sites for gliclazide and glipizide have been found previously to increase in binding strength by 1.2-fold [18] or 1.2- to 2.5-fold [16], respectively, in the presence of glycated HSA. In this report, these drugs had an increase in their global affinity of up to 1.8- or 3.5-fold for the samples of MGo-modified HSA and up to a 27% decrease in global affinity (gliclazide) or up to a 5-fold increase (glipizide) when working with Go-modified HSA. Tolazamide has been found to have either a small decrease or increase in its binding strength (< 20%) at its high affinity sites at moderate or higher levels of HSA glycation [17], while increases up to 2.3-fold in global affinity were seen here for MGo- and Go-modified HSA. For tolbutamide, the binding strength for this drug at its high affinity sites has been found to increase by up to 1.4-fold for glycated HSA [33]; in this current report there was an increase in global affinity of 1.7- to 1.8-fold when using MGo-modified HSA or 1.9- to 7.0-fold when using Go-modified HSA. These comparisons indicated that AGE formation can have a large effect on the binding constants for these sulfonylurea drugs with HSA and that these effects were comparable to or larger than those seen for the modifications of HSA that occur due to early stage glycation [3,14–18,33,34].

4. Conclusion

In this study the global affinity constants for several first- and second-generation sulfonylurea drugs were measured with HSA and AGE-modified HSA by using zonal elution combined with immunoextraction and affinity microcolumns. The specific modifications that were examined were those produced on HSA by MGo and Go. The use of immunoextraction microcolumns made it possible with a single platform to examine drug interactions with several individual preparations of HSA. Only 0.42–0.69 nmol HSA were needed per application cycle, and each applied protein sample could be used over 20–50 experiments. Information on the global affinity constants for most of the injected drugs was obtained in less than 7.5 min, with even highly retained drugs giving results within 20 min. In addition, the binding constants obtained by this approach for all the drugs with normal, unmodified HSA had an average precision of ± 12% and gave good agreement with reference values estimated from the literature [13–17].

This method was used to compare binding by these drugs with normal, unmodified HSA and with HSA that had been modified with Go or MGo. It was found that both Go- and MGo-related modifications could have significant effects on the global affinity constants for these sulfonylureas with HSA. Many of these modifications resulted in stronger binding, which would lead to a corresponding decrease in the bioavailable, free fraction of the drug; however, in some cases there was weaker binding, which would produce an increase in the drug’s free fraction. It was also found that these changes in binding strength were either comparable to or larger than changes that have been noted due to early stage glycation [3,14–18,33,34]. These results should provide a better understanding of how diabetes and elevated levels of agents such as Go and MGo may alter the protein binding and effective dosage of oral antidiabetic drugs in type II diabetes. This knowledge may be used in the future to compensate for these changes in drug binding during a patient’s treatment, as part of personalized medicine [14,33–35].

Highlights.

• Affinity microcolumns were used with immunoextraction to study drug-protein binding

• Binding was measured by sulfonylureas with glyoxal/methylglyoxal-modified albumin

• This method allowed a single platform to be used with several preparations of albumin

• Global affinity constants for most of the tested drugs were obtained in less than 7.5 min

• Binding constants acquired with normal albumin had good agreement with the literature