, 68:179

Chromatographic Framework to Determine the Memantine Binding Mechanism on Human Serum Albumin Surface


  • Firas Ibrahim
    • Equipe des Sciences Séparatives et Biopharmaceutiques (2SB/EA-3924), Laboratoire de Chimie Analytique, Faculté de Médecine PharmacieUniversité de Franche-Comté
    • Equipe des Sciences Séparatives et Biopharmaceutiques (2SB/EA-3924), Laboratoire de Chimie Analytique, Faculté de Médecine PharmacieUniversité de Franche-Comté
  • Claire André
    • Equipe des Sciences Séparatives et Biopharmaceutiques (2SB/EA-3924), Laboratoire de Chimie Analytique, Faculté de Médecine PharmacieUniversité de Franche-Comté

DOI: 10.1365/s10337-008-0675-6

Cite this article as:
Ibrahim, F., Guillaume, Y. & André, C. Chroma (2008) 68: 179. doi:10.1365/s10337-008-0675-6


In this work, the interaction of memantine with human serum albumin (HSA) immobilized on porous silica particles was studied using a biochromatographic approach. The determination of the enthalpy change at different pH values suggested that the protonated group in the memantine–HSA complex exhibits a heat protonation with a magnitude around 65 kJ mol−1. This value agrees with the protonation of a guanidinium group, and confirmed that an arginine group may become protonated in the memantine–HSA complex formation. The thermodynamic data showed that memantine–HSA binding, for low temperature (<293 K), is dominated by a positive entropy change. This result suggests that dehydration at the binding interface and charge–charge interactions contribute to the memantine–HSA complex formation. Above 293 K, the thermodynamic data ΔH and ΔS became negative due to van der Waals interactions and hydrogen bonding which are engaged at the complex interface. The temperature dependence of the free energy of binding is weak because of the enthalpy–entropy compensation caused by a large heat capacity change, ΔCp = − 3.79 kJ mol−1 K−1 at pH = 7. These results were used to determine the potential binding site of this drug on HSA.


Column liquid chromatographyHuman serum albuminMemantine

1 Introduction

N-Methyl-d-aspartate (NMDA) receptor antagonists have therapeutic potential in several central nervous system disorders, including neuroprotective treatment in chronic neurodegenerative diseases, and symptomatic treatment in other neurologic diseases [1].

Memantine, an NMDA antagonist, has been recently approved for the treatment of moderate to advanced Alzheimer’s disease (AD) [2]. Memantine is a low-affinity voltage-dependent uncompetitive antagonist for glutamatergic NMDA receptors [3, 4]. By binding to the NMDA receptor with a higher affinity than Mg2+ ions, memantine is able to inhibit the prolonged influx of Ca2+ ions which form the basis of neuronal excitotoxicity [5]. The low affinity and rapid off-rate kinetics of memantine at the level of the NMDA receptor-channel, however, preserves the physiological function of the receptor as it can still be activated by the relatively high concentrations of glutamate released following depolarization of the presynaptic neuron [6]. Age-related changes in physiology and organ function alter drug pharmacokinetics. In addition, older persons take more medications in treating multiple disorders, increasing the risk of drug–drug and drug–disease interactions [7]. Thus, the expanded pharmacokinetics studies, e.g. binding on plasmatic proteins, are important for drugs which are taken by aging patients as the drugs of Alzheimer’s disease. Memantine binds on plasmatic proteins to about 45%, it crosses the blood-brain barrier but its cerebrospinal fluid level is 20% to 50% lower than the serum level due to albumin binding in serum [810]. HSA is the most abundant protein in blood and can reversibly bind a large number of pharmacological substances. Few specific binding sites are present on HSA [11, 12]. The most important sites are benzodiazepine and warfarin binding sites. He et al. [12] have determined the three dimensional structure of HSA and have shown that these two binding sites are located in hydrophobic cavities in subdomains IIA and IIIA. Site I is formed as a pocket in subdomain IIA and involves the lone tryptophan of the protein (Trp214). The inside wall of the pocket is formed by hydrophobic side chains, whereas the entrance to the pocket is surrounded by positively charged residues. Site II corresponds to the pocket of subdomain IIIA, which has almost the same size as Site I, the interior of cavity is constituted of hydrophobic amino-acid residues and the cavity exterior presents two important amino-acids residues (Arg410 and Tyr411) [13, 14]. Two common methods that have traditionally been used in evaluating the binding of drugs to albumin include equilibrium dialysis and ultrafiltration [1517]. These two methods suffer of several disadvantages, as the long periods of time which are required to establish an equilibrium during the dialysis process [15, 16]. Furthermore, it is necessary to correct for the alterations in free and bound analyte concentrations that occur during the dialysis procedure [15]. Ultrafiltration requires less time to perform, but like dialysis it requires the use of a labeled drug and/or an additional analysis step for the actual measurement of the final free drug concentration. In addition, the effects of analyte adsorption to the ultrafiltration membrane must be considered [15, 16]. Other problems include difficulties with temperature changes during the separation and problems when working with highly bound drugs [15]. Because of these limitations, there has been continuing research to find better, faster and more convenient approaches for the analysis of drug–protein binding. One such approach involves the use of affinity chromatography (AC) [18]. AC is a liquid chromatography (LC)-based method in which the stationary phase consists of an immobilized biologically-related ligand. In the case of solute–albumin studies, this ligand consists of serum albumin which has been adsorbed or covalently linked to a support like silica. The association constants of many ligands have been determined by zonal elution [19] or frontal analysis [20]. The thermodynamic process involved in the binding have also been studied [2123]. One advantage of utilizing AC for solute–protein studies is the ability of this method to reuse the same ligand preparation for multiple experiments (small amount of protein is needed for a large number of studies), this helps to give good precision by minimizing run-to-run variations. Other advantages include the ease with which AC methods can be automated and the relatively short periods of time that are required in AC for most solute binding studies. The fact that the immobilized protein is continuously washed with an applied solvent is yet another advantage of AC [24, 25]. HSA was the model ligand used in a great number of studies. The main advantage of using HSA is the data available for its interaction with a wide range of organic and inorganic compounds [18].

In this study, AC was used to determine and quantify the forces driving the association between memantine and HSA by studying the energetic changes of this association as both a function of temperature and pH. Moreover, the number of protons linked to this memantine binding reaction of HSA was calculated. These results were used for estimating the binding site of memantine on HSA.

2 Experimental

2.1 Reagents and Operating Conditions

Memantine (Fig. 1) and diazepam were purchased from Sigma (Paris, France), water was obtained from an Elgastat option water purification (Odil Talant, France) fitted with a reverse osmosis cartridge. Sodium dihydrogenophosphate and di-natriumhydrogenophosphate were obtained from Prolabo and Merck (Paris, France) respectively. The mobile phase consisted of 0.1 M sodium phosphate buffer adjusted at different pH varying between 5.0 and 7.0 (5.0, 5.5, 6.0, 6.5, and 7.0). Experiments were carried out over the temperature range 278–308 K (278, 283, 293, 298, 303 and 308 K), and the mobile phase flow-rate was 0.3 mL min−1. The memantine (20 μL) was injected three times at each temperature and pH. Once the measurements were completed at the maximum temperature, the column was immediately cooled to ambient conditions to minimize the possibility of any unfolding of the immobilized HSA.
Fig. 1

Memantine structure

2.2 Apparatus

The LC system consisted of a Shimadzu LC- 10ATvp pump (Champs sur Marne- France), a Rheodyne 7125 injection valve (Cotati, California, USA) fitted with a 20 μL sample loop, and a Shimadzu UV–Visible detector. A ChromTech HSA column (Interchim, Montluçon, France) (150 mm × 4 mm I.D., 5 μm particle size) was used where HSA was covalently bound onto spherical 5 μm silica particles. The temperature was controlled with an Interchim oven TM701 (Monluçon, France).

3 Result and Discussion

3.1 Bulk Solvent pH Effects

Valuable information about the processes driving the memantine–HSA association mechanism can be further gained by examining the effect of pH on memantine retention. The memantine retention on the albumin stationary phase can be evaluated using the retention factor k:
$$ k = \left( {t - t_{\rm o} } \right)/t_{\rm o} $$
where (t) is the retention time of memantine and (to) is the column void time. The void time was determined using the mobile phase peak. As well, the memantine retention factor can be related to the association constant K between HSA and memantine as follows:
$$ k = \Phi K $$
where Φ is equal to the ratio of the active binding site number in the column over the void volume of the chromatographic column. When the pH of the bulk solvent changed, a full description is essential, which explicitly maintains conservation of mass of each species taking into account binding of H+ to human serum albumin (HSA), memantine (M) and the complex HSA-M:
$$ {\text{HSA}}\left( {H^ + } \right)A + M\left( {H^ + } \right)B + n_{H + } H^ + \leftrightarrow {\text{HSA}} - M\left( {H^ + } \right)C $$
where nH+ = C – (B) is the number of protons linked to this memantine binding reaction of albumin. The association constant of this equilibrium was given by:
$$ K = \left[ {{\text{HSA}} - M} \right]/\left[ {{\text{HSA}}} \right]\left[ M \right]\left[ {H^ + } \right]^{nH + } $$
Equation 4 can be written as:
$$ K = K_0 /\left[ {H^ + } \right]^{nH + } $$
where K0 is the K value for [H+]=1 M. Taking the logarithm of Eq. 5 gives:
$$ \log K = \log K_0 - n_{H + } \log \left[ {H^ + } \right] $$
As, −log[H+] = pH, Eq. 6 can be rewritten as:
$$ \log K = \log K_0 + n_{H + } { }{\rm pH} $$
Derivation of Eq. 7 gives:
$$ \partial \log K/\partial {\rm pH} = n_{H + } $$
Combining Eqs. 2 and 8 the following is obtained:
$$ \partial \log k'/\partial {\rm pH} = n_{H + } $$
The logarithm of the retention factor k was plotted against pH, when the bulk solvent pH increased from 5.0 to 7.0, for a wide variation range of temperature (278–308 K) (Fig. 2). These plots were linear for all temperatures with correlation coefficients r higher than 0.96, and showed that the binding affinity increased linearly with pH. This increase of the binding affinity came from two aspects of effects, one from the albumin and another from the drug. Although the influence of the buffer pH on the secondary structure of albumin is small, the rigidity of the albumin molecule will be somewhat affect, and the changes of charge on the entrance of the binding pocket would influence on the access of the drug to the binding site in some extent [26, 27]. On the other hand the ionization state of the drug would be different with the variation of the bulk solvent pH, and thus, affected the binding affinity of the drug. From Eq. 9, the slope of the curve log k versus pH gives the number of protons (nH+) at the drug–HSA interface implied in the binding process. For example, at 298 K, the value of nH+ was 0.25. The positive values of nH+ obtained at all the pH values reflected the uptake of protons when drug bound to HSA, and mean that one or more groups of ligand or/and albumin should increase its pKa as a consequence of binding and uptake of protons [28]. Several residues of HSA are likely candidates for proton acceptance, as the residue histidine, arginine or tyrosine, as well as the amine group of memantine. Many studies have demonstrated that pH-induced alterations in the binding sites of protein molecule play an important role in the changes of ligand binding to protein [2932]. Xie et al. [31] have shown that the binding ability of morin to HSA decreased with the increase of buffer pH which showed that the level of protonation played an important role during the drug–protein binding process. Also, Bagnost et al. [32] demonstrated that the affinity of nor-NOHA to the arginase enzyme was high and increased slightly with the pH and this binding mechanism was accompanied by a protonation of the histidine residue in the binding site of the enzyme. Thus, there are several reasons that can explain these changes of pKa values. For instance, a variation in the micropolarity of the environment surrounding the side chains of certain active side residues as a result of drug binding is a possibility. Alternatively, a protonated form could be stabilized by forming hydrogen bond with a neighbouring group. Furthermore, the magnitude of the corresponding heat protonation ∆HH+ in the memantine–HSA complex was determined using the following relation [3234]:
Fig. 2

log k versus pH for memantine at = 298 K

$$ \Delta H_{H + } = - 2.3RT^2 \left( {\partial n_{H + } /\partial T} \right) $$
The plot nH+ versus temperature showed a constant values for < 298, and increased linearly for ≥ 298 with a correlation coefficient r higher than 0.96 (Fig. 3). Using the slope of the second part of the plot, the corresponding magnitude of the heat protonation of the protonated group in the memantine–HSA complex was determined around 65 kJ mol−1 (Eq. 10). This value agrees with the heat protonation of a guanidinium group [35, 36], and confirmed that an arginine group may become protonated in the complex. This result was agreed with other previous studies which demonstrated that the binding of ligand to protein might be accompanied with a protonation in one or more residues in the binding site [31, 32, 37].
Fig. 3

Temperature dependence of the linked protons (per mol albumin), nH+

3.2 Thermodynamic Analysis

The temperature dependence of the memantine retention factor is given by the well known thermodynamic relation [38, 39]:
$$ \partial \ln k'/\partial T = \Delta H/RT^2 $$
where ∆H is the binding enthalpy and R is the gas constant. The analysis of the thermodynamics was carried out by measuring the memantine retention factor in the temperature range (293–308 K) with various pH (5 ≤ pH ≤ 7) of the bulk solvent. The van’t Hoff plots for the memantine exhibit a significant non-linear behaviour as shown in (Fig. 4). Similar non-linear van’t Hoff plots were obtained for other solute–albumin binding processes in previous studies [40, 41]. These non-linear plots indicated changes in the solute retention mechanism with temperature. From these plots and using Eq. (11) the ∆H values were determined (Table 1). ∆H depends linearly on the temperature in the range (278–308 K) with correlation coefficients r higher than 0.99 (Fig. 5). At low temperatures (<293 K), the binding enthalpy contributes non-favourably to the free energy of binding. At about 293 K, the enthalpy change of association was nil and above this value became negative indicating that the complex formation is enthalpically governed. This means that van der Waals interactions and hydrogen bonding (both characterized by negative enthalpy changes at these temperatures) are engaged at the complex interface confirming strong albumin–ligand hydrogen bond [4244]. In the temperature range 278 to 293 K, as temperature increases, the binding enthalpy becomes less endothermic (more favourable). As can be seen in Fig. 5 the enthalpy change decreases quickly with temperature due to a large negative heat capacity changes. At pH = 7 the ∆Cp value was −3.79 kJ mol−1.K−1. this value was determined from the slope of linear temperature dependence of enthalpy changes (Fig. 5). As well, the entropy change ∆S was determined from the ∆H obtained and using the value of ∆G calculated from the relation:
Fig. 4

Temperature dependence of the ln k values of memantine at pH = 6.5
Fig. 5

Temperature dependence of ΔH for the binding of memantine on HSA at various pH values of bulk solvent

Table 1

Thermodynamic parameters ∆H (kJ mol−1) and ∆S (J mol−1 K−1) for the memantine binding to HSA at pH = 7 and for six temperatures

T (K)

H kJ mol−1

S J mol−1 K−1



















$$ \Delta G = - { }RT\left( {\ln k' - \ln \Phi } \right) $$
The ∆S value was then calculated using the following equation:
$$ \Delta S = - { }\Delta G\left( T \right)/T + \Delta H\left( T \right)/T $$
TH and TS (reference temperatures at which ∆H and ∆S are nil) for the binding of memantine to HSA were around 293 K at all pH values of bulk solvent. About this temperature the entropy change was ≈ nil. Below this temperature value, the binding process is therefore accompanied by a positive entropy change (Table 1), which also strongly depends on temperature, while ∆G changes slightly with temperature (Fig. 6) because of the enthalpy–entropy compensation [32]. This behaviour has been found in many ligand–protein interactions [4548]. At low temperature, below 293 K, the positive enthalpy change and positive entropy change of binding upon complex formation can be justified by charge–charge interactions and hydrophobic forces [49, 50]. This is in agreement with the findings reported by several authors who demonstrated that hydrophobic and electrostatic interactions play an important role in the binding of drugs with an acidic or basic character to HSA [42, 51]. Memantine is both highly basic (pKa = 10.42) and lipophilic (log P = 3.28) [52] suggesting that ionic interactions thanks to its basic primary amine group, and hydrophobic interactions due to its hydrophobic rings will take place when memantine is included in the HSA binding cavity. In the association of a protein to a ligand, several contacts between non-polar groups of memantine and HSA are engaged. Thus, substantial fraction of polar and non-polar surface is buried in the complex formation which is thus accompanied by negative heat-capacity changes of the system. Many studies [53, 54] have suggested that ∆Cp may be described as a phenomenon in hydration terms, pointing out that changes in vibrational modes apparently contribute slightly to ∆Cp. Similarly, Connely and coworkers have shown that the heat capacity of ligand binding can be approximated by contributions arising from dehydration of solvents exposed groups [5557]. The interaction between apolar groups of memantine and HSA requires the dehydration of both protein and the drug and there is an entropic gain from the transfer of interfacial water into the bulk solvent. Assuming that ∆Cp value is due principally to the hydrophobic effect [58] and that the decrease in heat capacity per mol of water lost is, on average, 24 J mol−1.K−1 [59], one can calculate that, at pH = 7, about 157 water molecules are released. As well, the enthalpy and heat capacity values provide an estimation of solvent accessibility changes during the binding. Murphy and Freire have suggested the following equations for ∆Cp and ∆H60 (enthalpy change at 60 NC) [53].
$$ \Delta C_p = 1.88\Delta ASA_{ap} - 1.09\Delta ASA_p $$
$$ \Delta H_{60} = -35.3\Delta ASA_{ap} + 131\Delta ASA_p $$
Fig. 6

Temperature dependence of the thermodynamic parameters for the binding of memantine-albumin at pH = 7

where ∆Cp, ∆H60 and ∆ASA are in J.K−1 mol−1, J mol−1 and Å2 units, respectively [5360]. ∆ASAap and ∆ASAp represent the changes in non-polar and polar areas exposed to solvent (accessible surface area) that take place upon albumin-memantine binding. The temperature of 60 °C in the expression is the mean value of the denaturation temperature of the model proteins used in the analysis. At pH = 7 using ∆H60 = −147.62 kJ mol−1, assuming a ∆Cp = −3.79 kJ mol−1K−1, the changes in accessible surface areas are ∆ASAap = − 3,168 Å2 and ∆ASAP = − 1981 Å2. Therefore, the results of Murphy’s approach indicated that the surface area buried on complex formation comprises 62% nonpolar surface and 38% polar surface. The amount of non-polar surface involved appeared too large to be accounted for in “rigid body” association. That could justify the accessible surface area value calculated. At low temperature below ≈293 K, ∆H and ∆S values remained positive since the contributions of the desorption of the solvent molecules overweight that of the memantine adsorption on the albumin surface. At 293 K ∆≈ 0, it appears so that there should be some source of negative entropy compensating the positive entropy of dehydration. The overall entropy change (∆S) at 293 K can be split up in the following way [32, 54]:
$$ \Delta S = \Delta S_{\rm hydr} + \Delta S_{\rm trans} + \Delta S_{\rm specific} $$
where ∆Shydr is the contribution by the hydrophobic effect. ∆Strans accounts for the reduction in the overall rotational and translational degrees of freedom, as well as the immobilization of amino acid side chain at the complex interface. ∆Sspecific describes system-specific contributions such as reduction of main chain mobility and entropic contributions from polar interactions. ∆Shydr can be estimated from
$$ \Delta S_{\rm hydr} = 1.35\Delta C_p .\ln \left( {T/386} \right) $$
where ∆Cp (in J mol−1 K−1) is the measured heat capacity change, T the absolute temperature and 386 the reference temperature at which the entropy of transfer of non-polar liquids to water vanishes [54]. For memantine-albumin complex at 293 K and pH = 7 we obtained ∆Shydr = + 1.41 kJ mol−1 K−1. From S≈ 0 kJ mol−1.K−1 at 293 K, we calculatead that ∆Strans +∆Sspecific = − 1.41 kJ mol−1 K−1. For a great number of bimolecular association reactions, ∆Strans has been thought to contribute −0.21 kJ mol−1 K−1 of rotational and translational entropy [54]. Hence, the remaining entropic loss of −1.20 kJ mol−1 K−1 must be contributed by the loss in the conformational restrictions of memantine and HSA. This unfavourable conformational change entropy could proceed from fixation of side chains at the interface and structural changes in the interacting molecules upon complex. Our results showed that adaptive conformational transitions are associated with the memantine–HSA complex formation where both components are able to adjust their recognition surfaces in order to maximize complementarities through tightly packed contacts involving Coulomb interactions and hydrogen bonding [60, 61]. Following previous studies from Guillaume’s group concerning pH and temperature effects on dansyl amino acids–HSA binding mechanism [40, 44, 62], our present results can be explained by suggesting that memantine binds to albumin Site II, where these hydrophobic groups occupy the nonpolar interior of the cavity and its primary amine group interacts with charged (Arg410) and polar (Tyr411) groups on the cavity rim forming electrostatic and hydrogen-bond. In order to confirm that memantine mainly bound on HSA Site II, a zonal elution approach [25, 63] was carried out. For this, the binding of a well known ligand on HSA Site II, i.e., diazepam, was examined by injecting small amounts of diazepam into the HSA column while a known concentration of memantine was applied to the column in the mobile phase. By examining how the mobile phase concentration of the memantine additive affects the retention of the diazepam injected solute, information will be gained on the type of competition (i.e., allosteric interaction or direct competition) which is occurring between the two solute molecules. Eleven memantine concentration values were included in the concentration range 0–10 μM. The retention factor of diazepam declined with increasing the concentration of the competitive agent, i.e., memantine, as follows [25, 6365]:
$$ 1/k_{\rm dia} = K_M \left[ M \right]/\left( {K_{\rm dia} \left[ {S_{\rm tot} } \right]} \right) + 1/\left( {K_{\rm dia} \left[ {S_{\rm tot} } \right]} \right) $$
where kdia, Kdia and KM are respectively the diazepam retention factor and the equilibrium affinity constants with HSA for diazepam and memantine, [Stot] is the effective concentration of common binding sites, and [M] is the memantine concentration in the mobile phase. (Fig. 7) presents the plot 1/kdia versus the memantine concentrations [M] at 293 K (Eq. 18). As can be seen in Fig. 7, a decrease in retention of diazepam was observed as increasing amounts of memantine were added to the mobile phase. It appears that the plot 1/kdia versus the memantine concentration is linear (r² > 0.998). This plot gave a good agreement between the experimental intercept (i.e. kdia when no competing agent (memantine) was present) and the intercepts which were determined by linear regression of the entire data set (kexperimental = 11.42 ≈ kpredicted ≈ 11.51). The agreement between these values is significant since it indicates the displacement of memantine by diazepam was though a direct, rather than an allosteric mechanism of competition and confirmed that memantine interacts reversibly with a single type of equivalent binding site, i.e., Site II [25, 64].
Fig. 7

Plot of 1/kdia vs [M] (μM) in the mobile phase (0.1 M sodium phosphate buffer, pH = 7.0)

4 Conclusion

In this paper the memantine binding mechanism to human serum albumin was analyzed. This binding was accompanied with a proton uptake which can be attributed to an increase in the pKa of one or more groups of the memantine and/or HSA in the complex at the entire range of pH studied. The protonation heat showed that an arginine group of albumin binding site may become protonated in the complex. The binding was temperature dependent. In a low temperature domain (<293 K) it was entropically driven, indicating a contribution from hydrophobic effect due to the release of water molecules when memantine and HSA associated. Above 293 K, the thermodynamic data ∆H and ∆S became negative due to van der Waals interactions and hydrogen bonding which are engaged at the complex interface. By the use of known correlations between the heat capacity change and the burial of non-polar surface area, the surface area that is burried in the memantine–HSA complex was estimated. These results associated with the use of a zonal elution approach demonstrated that memantine seemed to be good candidate as ligand for the HSA Site II (indole-benzodiazepine site).

Copyright information

© Vieweg+Teubner | GWV Fachverlage GmbH 2008