Studies and analysis of drug-target interactions by affinity chromatography and related techniques: A review

Abstract

The characterization of drug-target interactions is a key component of drug discovery, testing, and development. Affinity chromatography is one approach that can be used for this type of analysis. For instance, this may be done by using an immobilized target as a stationary phase and a drug as the applied solute. This review will discuss the various ways in which affinity chromatographic methods have been used to examine drug-target interactions, with an emphasis on high-performance methods. The general principles of this approach and factors to consider in its use for drug-target interaction analysis will first be examined. Methods based on zonal elution or frontal analysis for binding and competition studies will then be discussed. Various techniques for kinetic studies will next be considered, along with approaches that employ secondary binding agents and hybrid techniques. In each case, the general principles and theory of an approach will be given along with examples of its use in drug-target interaction studies. Advantages or limitations of each approach will be provided as well. This information should make it possible in the future to extend these techniques to other drug-target systems of interest in biomedical research and drug testing or development.

Graphical abstract

Highlights

• •Affinity chromatography can be used in many ways to study drug-target interactions.
• •Zonal elution and frontal analysis methods are often used for this purpose.
• •Various affinity chromatographic methods for kinetic studies are also available.
• •Secondary capture agents or hybrid affinity methods can also be used.
• •The applications, advantages and limitations for each of these methods are discussed.

1. Introduction

Exploring drug-target interactions is vital for modern drug discovery and therapeutics [[1], [2], [3], [4], [5], [6], [7], [8], [9]]. These interactions involve the binding of drugs to targets such as proteins, enzymes, receptors, and nucleic acids [2,4,8,9]. Characterizing these interactions is crucial for determining the molecular mechanisms that underly drug action and safety; this is also vital for drug discovery and development [1]. A few examples include the analysis of how kinase inhibitors can target certain signalling pathways to impede tumor progression during cancer treatment or of how drugs that target neurotransmitter receptors or ion channels in neurological disorders can help restore normal brain function [[3], [4], [5]]. It is also essential to have information on drug-target interactions to identify off-target effects that can lead to adverse reactions and toxicity [6]. Tools for interaction studies can further be used to identify new drug targets and potential therapies, as is needed in conditions such as Alzheimer's disease, diabetes, and autoimmune disorders [[7], [8], [9]]. In addition, the analysis of drug-target interactions may allow for the customization of medications and therapeutic strategies for personalized medicine [8].

Various methods have been employed in the past to study drug-target interactions. These approaches have involved both biophysical measurements and computational strategies [1,2,[9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21]]. Some experimental techniques for this work have included spectroscopic methods based on absorbance, fluorescence or fluorescence polarization, surface plasmon resonance (SPR), circular dichroism, and nuclear magnetic resonance (NMR) [1,2,11,20,21]. Other methods that have been employed are equilibrium dialysis, ultrafiltration, calorimetry, stopped flow analysis, capillary electrophoresis, mass spectrometry (MS), liquid chromatography (LC), and LC-MS [1,2,12,15,[17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28]]. Proteomics-based techniques such as thermal proteome profiling and chemoproteomics have also been developed for studying drug-target interactions within complex biological systems [[13], [14], [15]]. Computational methods that have been used for this work are molecular docking and molecular dynamics simulations, such as utilized to examine interactions between potential drug candidates and biological targets [10,16]. However, each of these methods has its own limitations, creating an ongoing need for fast, precise, and robust methods that can work with small amounts of materials and examine the binding of drugs with potential targets [1,2,9,10,[16], [17], [18], [19], [20]].

Affinity chromatography (AC), especially in the form of high-performance affinity chromatography (HPAC), is one alternative technique that can be utilized to study drug-target interactions [1,[17], [18], [19],22]. AC and HPAC make use of a stationary phase that consists of a biologically-related agent (e.g., a drug or its target) that is immobilized or placed within a chromatographic support [17,19] (see Fig. 1). In traditional AC, this support is a low or medium performance material such as a carbohydrate or polymeric resin, while in HPAC a material like HPLC-grade silica, a monolith or high-performance polymer is employed [19]. Advantages of using AC and HPAC for studies of biochemical interactions include their specificity, ability to study binding or kinetics under nearly physiological conditions, and their compatibility with many analysis and detection formats (e.g., absorbance, fluorescence, MS), usually in a label-free manner [1,17,18,22]. These methods are especially useful for studying weak or transient interactions, which can be challenging to examine by other techniques [17,18]. In addition, it is possible with HPAC to often reuse the same immobilized agent for many experiments, which provides good reproducibility, reduces the amount of agent needed per study, and minimizes the effects of batch-to-batch variations in this agent [17,18]. The speed, precision, and ease of automation of HPAC also make this method useful for rapid and high-throughput measurements of drug–target interactions [[17], [18], [19]].

Fig. 1.

General model for using affinity chromatography (AC) or high-performance affinity chromatography (HPAC) in drug-target interaction studies, as illustrated using a column with an immobilized form of the target and a soluble drug that is injected or applied to this column. The terms used in this figure are the same as used and defined in Eqs. (1), (3): D: drug; T: target; K_a: association equilibrium constant; k_a: second-order association rate constant; k_d: first-order dissociation rate constant.

This review will examine ways in which AC has been used to study drug-target interactions, with an emphasis on HPAC methods. The principles of this approach and factors to consider in its use for interaction analysis will first be discussed. The various formats that may be used in drug-target interaction studies will then be presented, such as methods based on zonal elution or frontal analysis, schemes for kinetic analysis, and hybrid techniques. The general theory of each approach will be provided along with examples of its applications in binding and interaction analysis. The advantages or potential limitations of each method will be given as well, thus providing information that can be used for extension of these tools in the future to other drug-target systems of interest in pharmaceutical science or biomedical research.

2. General considerations for drug-target interaction studies in AC

The following general model is often used in AC and HPAC to describe the reversible interaction between a drug (D) and its target (T), in which one of these components is typically present as the stationary phase in an affinity column [22,23] (see Fig. 1).

$$D+T\begin{array}{c}{k}_{\mathrm{a}}\\ \rightleftharpoons \\ k_{\mathrm{d}}\end{array}DT$$ (1)

where

$$K_{\mathrm{a}}=\frac{\left[DT\right]}{\left[D\right]\left[T\right]}\text{or}{K}_{\mathrm{d}}=\frac{\left[T\right]\left[D\right]}{\left[DT\right]}$$ (2)

$$K_{\mathrm{a}}=\frac{k_{\mathrm{a}}}{k_{\mathrm{d}}}=\frac{1}{K_{\mathrm{d}}}$$ (3)

In this simple 1:1 model, D can bind reversibly with T to form the complex DT. The terms K_a and K_d represent the association equilibrium constant and dissociation equilibrium constant for this interaction, with these terms being inversely related to each other. The rate of binding between D and T is described by the second-order association rate constant k_a, and the rate of dissociation for DT is described by the first-order dissociation rate constant k_d. More complex models, such as involving n interaction sites for D with T, can be developed and written in a similar manner [23].

The proper use of AC and HPAC to study drug-target interactions requires the careful consideration of several factors [17,18,22]. The first of these is the technique that will be used for the binding or rate studies, as will be addressed in the following sections [1,22,23]. Other general factors that need to be considered are the immobilization method, application and elution conditions, and type of detection that will be used, as well as the amounts of drug, target, and sample that are available for these experiments [22]. The selection of these factors, and particularly the immobilization method, will also depend on whether the binding studies in AC and HPAC are desired to take place with both the target and drug in solution (i.e., a homogeneous method) or with one of these agents immobilized in the column (i.e., a heterogeneous method) [18,22,23].

Immobilization in AC and HPAC is the procedure of attaching a binding agent, or affinity ligand, to a support for use as a stationary phase [24]. If either the drug or target is to be placed on a support for use in a heterogenous method, it is important to select and use coupling conditions that avoid undesirable effects like improper orientation, steric hindrance, or multisite attachment of the immobilized agent [18,24]. Many schemes are available for covalent immobilization through residues like amine, sulfhydryl, carboxyl or carbohydrate groups on a given agent; the choice of such a method will be determined by the number and location of such groups on the target or drug that is to be immobilized and the effect of their modification on the interaction that is to be examined [24]. It is also possible to immobilize a binding agent to a support through physical adsorption (e.g., through electrostatic interactions, hydrophobic forces, or hydrogen bonding) or biospecific adsorption to a secondary immobilized agent (e.g., the binding of a biotin-labeled drug or target to a support containing immobilized streptavidin) [19,24]. In some cases it may be possible to physically trap a soluble form of the binding agent in a column, such as might be done through the use of encapsulation within sol-gels or entrapment within silica supports that have been treated with large capping agents to keep the binding agent in the support's pores [18,19,24].

The support to be used in the AC or HPAC method for drug-target interaction studies is another key factor to consider [19,25]. Traditional AC has typically used large-diameter supports based on carbohydrates (e.g., agarose or cellulose) or hydrophilic polymers (e.g., polymethacrylate or polyacrylamide) [19,25]. These materials can possess large pores, good chemical and pH stability, and low non-specific binding for biological agents, while also providing groups that allow for easy immobilization of various targets [24,25]. However, these materials tend to have limited mechanical stability at higher pressures and slow mass transfer properties, making them more suitable for preparative applications or sample pretreatment than higher-speed analytical applications [17,19,25]. In HPAC, the supports that are used are instead more rigid and efficient materials such as small diameter porous silica, glass beads or polymeric particles, as well as inorganic or organic monoliths [17,19,25,26].

The application and elution conditions that are used in AC and HPAC for interaction studies are other items that are important to consider [27]. The application buffer will usually consist of a mobile phase that has the solvent composition, ionic strength, polarity, and pH that is desired for the drug-target system during their analysis [17,27]. If the interaction being studied is weak-to-moderate in strength (i.e., K_a of 10⁶ M⁻¹ or less), it may be possible to use the application buffer alone for both sample application and elution of the retained drug or target (i.e., isocratic elution) [17,19,27]. If stronger binding is present, a competing agent may be needed to elute the retained component through competition and mass action, giving a technique known as biospecific elution [26,27]. For systems with quite strong binding, a separate elution buffer may be needed with a different pH or composition from the application buffer to lower the binding strength and promote release of the retained substance, a method known as non-specific elution [26,27]. After either biospecific elution or non-specific elution, the application buffer would then be run again through the system to regenerate it prior to the next application of a drug or target sample [18,26,27].

The amount of drug and target that is available for study will be an additional item to consider during the selection of AC or HPAC techniques for examining a drug-target interaction [22]. For instance, techniques like zonal elution and ultrafast affinity extraction, as discussed later in this review, require only a small amount of drug or solute for a binding study, while the method of frontal analysis requires a much larger amount to allow saturation of the target and measurement of a binding isotherm [17,18,22,23]. In addition, some of these methods may require the amount of applied sample to be much smaller than the amount of an immobilized target, as is needed to establish linear elution conditions, while other techniques can be used under non-linear conditions and with an amount of sample that approaches or exceeds that of the binding agent [22,23].

3. Zonal elution methods for drug-target interaction studies

3.1. General principles of zonal elution in AC

Zonal elution is probably the most common technique employed in AC and HPAC for studying the binding of solutes like drugs with proteins and other targets [18,19,22,28]. In this method, a small quantity of the drug or solute is injected onto an affinity column that contains an immobilized form of the target [22]. This is usually done in presence of a mobile phase with an appropriate pH and composition to mimic the desired binding conditions for the drug-target interaction [22,28]. A fixed concentration of an additive or competing agent may also be present in the mobile phase. The retention time (t_R) of the drug or solute is then determined on both the affinity column with the immobilized target and on a control column with no binding agent present. The retention factor (k) is then found on each column by combining this retention data with the measured void time of the column (t_M) by using the relationship k = (t_R - t_M)/t_M (Note: for small columns, it is important to use a value of t_M that is only for the column and that has been corrected for the void time due to other system components) [22,28,29]. The specific retention factor (k′) for the drug/solute due to binding with the immobilized target can then be found by taking the difference in the retention factors that have been determined on the affinity column and control column [28,29].

3.2. Zonal elution in AC under linear elution conditions for studies of drug-target interactions

In zonal elution, the retention observed for a solute injected onto an affinity column will be directly related to the interaction of the solute with the immobilized binding agent. If the amount of injected solute is small compared to the amount of the immobilized agent (i.e., linear elution conditions are present) and these interactions are relatively fast on the timescale of the experiment, the specific retention factor k′ for a solute or drug that has one binding region on an immobilized target can be linked to the association equilibrium constant (K_a) for their interaction, as shown in Eq. (4) [22,23].

$$k^{\prime }=K_{\mathrm{a}}\frac{m_T}{V_{\mathrm{M}}}$$ (4)

In this equation, the effective concentration of the immobilized target in the column is represented by the term (m_T/V_M), where m_T is the moles of active binding sites for the injected solute on the target, and V_M is the void volume of the column. An equivalent relationship can be written for a solute that has n independent binding sites on an immobilized target [22,23].

$$k^{\prime }=\frac{\left(n_1{K}_{\mathrm{a}1}+n_2{K}_{\mathrm{a}2}+\dots +n_{\mathrm{n}}{K}_{\text{an}}\right)m_T}{V_M}=\frac{\left(n{K}_a^{´}\right)m_T}{V_M}$$ (5)

In this more general expression, the association equilibrium constants for the drug/solute at binding sites 1 through n are given by the terms K_a1 through K_an, with n₁ through n_n being the relative amounts (in mol/mol of target or binding agent) for each type of site. The summation of the terms $n_1{K}_{\mathrm{a}1}$ through $n_{\mathrm{n}}{K}_{\text{an}}$ in eq. (5) can also be represented by nK´_a, or the global affinity constant, which indicates the overall binding strength of the drug or solute with the target due to the combined effect of all the available binding sites [18,22,23].

Eqs. (4), (5) show that the specific retention measured by AC or HPAC and zonal elution for a drug with an immobilized target will be determined by the binding constants for this interaction, the total amount of binding sites that are available for the drug on the target, and the number and types of these sites in terms of their relative amounts and affinities for the drug [22,23]. These features have made retention measurements and k’ values that are obtained by zonal elution useful in providing a convenient and precise means of evaluating and comparing the overall binding of drugs with different targets or of many drugs with the same target [[30], [31], [32], [33], [34], [35], [36], [37], [38]]. This is especially true if the immobilized target is known to be present in a highly active form (e.g., as obtained through use of optimized covalent immobilization, biospecific adsorption, or entrapment) (see example in Fig. 2) [[29], [30], [31], [32]].

Fig. 2.

Use of zonal elution and high-performance affinity chromatography (HPAC) to screen drug-target binding, as illustrated for injections obtained at 0.50 mL/min on a 2.0 cm × 2.1 mm i.d. microcolumn containing the entrapped protein DJ-1. The results are shown for (A) sodium nitrate (i.e., a non-retained solute) and isatin (a known binder for DJ-1) or (B) a predicted non-binding ligand and a predicted binding ligand for DJ-1, as identified through in silico screening. Reproduced from Ref. [31] with permission.

Recent examples of zonal elution being used in this manner with HPAC include studies of binding by antidiabetics, antihypertensives, antibiotics, antipsychotics, peptides and hormones with normal or modified forms of the transport proteins human serum albumin (HSA) and α₁-acid glycoprotein (AGP) [[28], [29], [30],37,38]. This approach has also been recently used to examine binding by the protein DJ-1 against potential drug candidates and binding partners for this therapeutic target, as had previously been selected through in silico screening (Fig. 2) [31]. Zonal elution has further been used to study interactions by antibiotics and other drugs with humic acid [32,33], to examine binding by various drugs with the β₂-adrenoreceptor, histamine H1 receptor, and 5-hydroxytryptamine 1A receptor [[34], [35], [36]]. In addition, expanded and modified equations for zonal elution have been recently developed and used to examine the simultaneous binding of two drugs to the same target or of the same drug to multiple targets [[39], [40], [41]]. These latter techniques have been employed to examine the binding of bosentan plus ambrisentan or macitentan with endothelin A receptor and the interactions of valsartan, candesartan, and telmisartan with angiotensin II type 1 and type 2 receptors [40,41].

Another way zonal elution can be utilized in AC and HPAC is to estimate the relative extent of solute binding to an immobilized agent [22,23]. This makes use of the fact that, at true equilibrium, the retention factor for an injected solute can be written as shown in Eq. (6).

$$k=b/f=b/\left(1–b\right)$$ (6)

In Eq. (6), b is the bound fraction of the solute with the immobilized binding agent, and f is the free fraction of the solute present in the corresponding region of the mobile phase. Because the sum of b and f must be equal to 1, the ratio b/f can also be written as b/(1 - b) [23]. It is assumed in this expression that linear elution conditions are present, that a true equilibrium is present at the center of the solute's peak, and that the immobilized binding agent is highly active [22]. Use of Eq. (6) has been applied to estimate the bound and free fractions of various drugs in the presence of HSA [42,43].

3.3. Competition studies using zonal elution in AC for drug-target interaction studies

A common way in which zonal elution is used in AC and HPAC to study drug-target interactions is through competition studies [22,23,28]. In this type of experiment, a probe solute is injected into the affinity column in the presence of a mobile phase that contains a known concentration of a drug or possible competing agent [22,28]. The change in the retention of the injected solute is then monitored while varying the concentration of drug and competitive agent. The change in retention is then used to determine whether the solute and competing agent bind to the same sites on the immobilized target, or if they have allosteric effects on each other, and to measure the binding constants or coupling constants that are present for these interactions [22,28,44,45].

The retention data that are acquired through a competitive binding zonal elution experiment can be fit to various equations to determine the types of interactions that are present between the solute and mobile phase additive, as well as the physical constants that describe these interactions [22,28,44]. For instance, if the injected probe solute (P) and drug (D) used as a mobile phase additive compete directly at a single type of site on an immobilized target (T), the change in retention that is seen for P can be described by Eq. (7), in which the reciprocal of the retention factor for P is plotted versus the concentration of drug and competing agent, [D], giving a linear relationship [22,28].

$$\frac{1}{k_P}=\frac{K_D\left[D\right]}{K_P{C}_T}+\frac{1}{K_P{C}_T}$$ (7)

In this equation, K_P and K_D are the association equilibrium constants of the target for the injected probe and drug or competing agent, respectively; k_P is the retention factor for the probe; and C_T is the effective concentration of active target in the affinity column. This equation assumes P has no other binding sites on the target, and that a correction has been made for non-specific binding of P to the column or system. If competition at a single site on the target is present between P and D, the ratio of the slope to the intercept that is obtained from Eq. (7) can be used to give the association equilibrium constant for D at this site on the target [22,28]. Recent applications of this method include its use to characterize the interactions of antidiabetic drugs and cathinones with HSA or modified forms of this protein [[46], [47], [48], [49]]. The same technique has been utilized to screen arginase inhibitors and to study binding by a bioactive compound from the tree Tetradium ruticarpum to the α1A adrenoreceptor [50,51].

Plots made according to Eq. (7) or related expressions can also be used to examine more complex interactions between a drug and probe compound for a common target [22,28]. If there is no competition between the probe and mobile phase additive, a plot made according to this equation will give a response with a slope of zero. If multisite binding or a negative allosteric interaction is present between D and P, this plot will have a positive slope and non-linear response; a negative slope will be produced if a positive allosteric interaction is present [22]. In these situations, alternative models and equations can be fit to the zonal elution data to obtain more information on the equilibrium constants and coupling constants that are present for these interactions [22]. For instance, both Eq. (7) and related expressions for systems with multiple binding sites and/or allosteric effects have been used to develop affinity maps to describe the complex interactions of some antidiabetic drugs with HSA or glycated/modified HSA (see Fig. 3) [46,47].

Fig. 3.

Characterization of complex drug-target interactions by using high-performance affinity chromatography (HPAC), zonal elution and competition studies carried out under linear elution conditions. This example shows an affinity map that was generated through zonal elution to show the interactions of two antidiabetics (pioglitazone and rosiglitazone) at several possible binding sites on normal human serum albumin (HSA) or glycated HSA. The term K_a is the association equilibrium constant for binding by the given drug at the specified region of HSA. Reproduced from Ref. [46] with permission.

3.4. Non-linear zonal elution methods in AC for drug-target interaction studies

Although most zonal elution experiments are done under linear elution conditions, it is possible to use this method under non-linear conditions to examine solute-target binding [[52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66]]. In this approach, the amount of injected solute is no longer much less than that of the immobilized target. This means the retention time and retention factor will now vary as the amount of solute is altered [[52], [53], [54], [55], [56]]. One advantage of using non-linear conditions is it is now easier to detect the analytes. A disadvantage is that the description of solute retention under these conditions is much more complex than when the solute amount is small compared to the target. For example, when using non-linear zonal elution, a non-Gaussian peak is produced with a position and shape that is related to the amount of injected solute versus the amount of active target in the column [[57], [58], [59], [60], [61], [62], [63], [64], [65], [66]].

One way drug-target binding has been examined with zonal elution and under non-linear equations is by fitting the peak profiles for injected drugs or solutes to Eq. (8) [57,58,[62], [63], [64]].

$$y=\frac{a_0}{a_3}\left[1-e^{\left(\frac{{-a}_3}{a_2}\right)}\right]\left[\frac{\left(\sqrt{\frac{a_1}{x}}\right)I_1\left(\frac{2\sqrt{a_{1}x}}{a_2}\right)e^{-x-a_1/a_2}}{1-T\left(\frac{a_1}{a_2},\frac{x}{a_2}\right)\left[1-e^{-a_3/a_2}\right]}\right]$$ (8)

In this equation, y is the response at a given time, and x is the reduced retention time. The term I₁ is a modified Bessel function, and T is the switching function. The parameters a₀ through a₃ are used to fit Eq. (8) to the experimental peak. These best-fit parameters can, in turn, be utilized to obtain the equilibrium and rate constants for the interaction between the injected solute and the immobilized binding agent. The association equilibrium constant for this interaction can be determined by using the relationship K_a = a₃/C₀, where C₀ is related to the analyte's concentration, the sample volume, and the column's dead volume. The apparent dissociation rate constant (k_d,app) can be found from k_d,app = 1/a₂ t_M, where t_M is the column void time [[62], [63], [64]].

Non-linear zonal elution has been utilized with Eq. (8) in both binding and kinetic studies of drug-target interactions [[58], [59], [60], [61], [62], [63], [64], [65], [66]]. This method has been used to determine the rate constants for various drugs with receptors such as the endothelin A receptor, muscarinic-3 acetylcholine receptor, and cysteinyl-leukotriene receptor [56,60,61]. Eq. (8) has also been employed to investigate the binding behavior and to generate quantitative structure-activity relationships for the interactions of nicotinic acetylcholine receptor with phencyclidine, 18-methoxycoronaridine, and bupropion plus verapamil [[62], [63], [64]]. Non-linear zonal elution and peak fitting have further been utilized to examine the binding for such agents as bambuterol, clorprenaline, ephedrine hydrochloride, methoxyphenamine, morin, mulberroside, propranolol, salbutamol, and tulobuterol with the β₂-adrenoceptor [53,65,66].

A second technique based on non-linear zonal elution that has been used for drug-target binding studies is the injection-amount dependent method [67]. In this method, various known moles of the drug or solute per injection (n_b) are placed into an affinity column containing the target. The retention factor (k) of the drug/solute is then measured and used along with the value of n_b to make a plot of (k n_b)/(1 + k) versus (k V_M), where V_M is the void volume. The expected result is a linear relationship in which the association equilibrium constant for the drug-target interaction and moles of binding sites on the target can be obtained from the slope and intercept [67]. This technique has been used to screen compounds that may bind to calcium sensing receptor and to characterize binding by rosmarinic acid with cysteinyl leukotriene receptor type 1 [54,55].

4. Frontal analysis methods for drug-target interaction studies

Frontal analysis is a useful tool in AC and HPAC for obtaining information on both how strongly a solute may interact with a stationary phase and the binding capacity of the stationary phase for the solute [22,36,49,51,[68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87]]. In this method, a known concentration of a drug or solute is continuously introduced to a column that contains an immobilized target as the stationary phase. As the solute passes through the column, it binds to sites on the target. The amount of solute that elutes from the column will increase as more sites on the target become occupied, resulting in the formation of a breakthrough curve for the solute [22,68]. Examples of such curves are given in Fig. 4A [79]. If relatively fast association and dissociation between target and solute are present on the time scale of this process, the position of the breakthrough curve will be related to the solute concentration, the amount of available binding sites, and the association equilibrium constant(s) for the solute-target interaction [22,68,72].

Fig. 4.

Characterization of a drug-target interaction by frontal analysis and high-performance affinity chromatography (HPAC), as demonstrated through the binding of R-propranolol to an affinity column containing immobilized very low density lipoprotein (VLDL). (A) The breakthrough curves that were obtained for various concentrations of R-propranolol that were applied at 0.50 mL/min, pH 7.4, and 37 °C onto a 100 mm × 2.1 mm inner diameter column containing VLDL. (B) The fit of the resulting data to a selected binding model for this interaction; the inset gives the corresponding residual plot for this fit. Reproduced from Ref. [79] with permission.

Once frontal analysis data have been acquired at various solute concentrations, these can be compared to various binding isotherms to provide information on the equilibrium constants and number plus amount of binding sites that are involved in these interactions, as shown in Fig. 4B [22,68,79]. Examples of these equations for a system with 1:1 binding between an applied drug and immobilized target are shown in Eqs. (9), (10) [22,68,[75], [76], [77]].

$$m_{Tapp=}\frac{m_T{K}_{\mathrm{a}}\left[D\right]}{\left(1+K_{\mathrm{a}}\left[D\right]\right)}$$ (9)

$$\frac{1}{m_{Tapp}}=\frac{1}{m_T{K}_{\mathrm{a}}\left[D\right]}+\frac{1}{m_T}$$ (10)

In these expressions, m_Tapp is the apparent moles of the applied drug or solute that are required to reach the breakthrough time at drug/solute concentration [D], and m_T is the true total moles of binding sites that are present on the target for the drug. In the case of Eq. (9), a non-linear fit is made to a plot of m_Tapp versus [D] to obtain the values of m_T and the association equilibrium constant for this interaction, K_a. For the double-reciprocal transform of Eq. (9) that is shown in Eq. (10), a plot of 1/m_Tapp versus 1/[D] would instead be made and used to provide the values of m_T and K_a from the slope and intercept [22,76]. Similar expressions to Eq. (9) can be derived and used for systems that have more complex binding, such as multisite interactions [22,76,77,79].

There are many examples of frontal analysis being employed for the study of drug-target interactions. For instance, this method has been used to characterize the binding of buspirone, serotonin, and hypidone with 5-hydroxytryptamine 1A receptor [36]. This approach has also been employed to characterize binding to low versus high affinity sites for various antidiabetics to HSA or modified forms of this protein [49]. Other applications have included the screening of potential inhibitors for arginase and characterization of binding by drugs to lipoproteins [51,78,79].

Another way frontal analysis can be used is to examine the competition of solutes for the same target [68,73,74]. This experiment is performed by adding both a competitive binding agent and the solute or drug of interest to the mobile phase that is passed through the column. The change in retention time or volume of the drug/solute's breakthrough curve is then measured as a function of competing agent concentration. If there is direct competition between a drug/solute and competition agent, this will result in a decrease in the breakthrough time for the drug/solute as the competing agent's concentration is increased. This change can be described by the following equation [73,74].

$$V_{\mathrm{R}}-V_{\min }=\frac{P}{\left(K_{\mathrm{d},IT}+\left[I\right]\right)}$$ (11)

In Eq. (11), K_d,IT is the dissociation equilibrium constant between the competing agent (I) and immobilized target, V_R is the breakthrough volume of the solute, and V_min is the breakthrough volume of the solute when its binding to the target is fully suppressed. The term P is related to the number of active binding sites and the ratio of K_d,IT versus K_d,DL, the dissociation equilibrium constant for the drug/solute with the target. This technique is often used by combining frontal analysis with mass spectrometry, as shown in Fig. 5 [74] and discussed later in this review [70,73]. Examples of applications have included use of this equation and method to study binding of dual agonists to peroxisome proliferator-activated receptors and characterization of the interactions by quercetin and related compounds to histone deacetylase SIRT6 [73,74].

Fig. 5.

Breakthrough curves obtained in competition and displacement studies by frontal analysis, combined with mass spectrometry, for a given ligand (R-3) that was applied at a fixed concentration of 100 nM and at 2.5 μL/min to a 40 cm × 100 μm inner diameter capillary containing an immobilized peroxisome proliferator-activated receptor and in the presence of (a) no competing or displacing agent and (b–d) increasing amounts of the competing/displacing agent (compound 2 in the given reference). Adapted from Ref. [74] with permission.

Stepwise frontal analysis is a recent variation of frontal analysis in which a sequence of drug/solute solutions ranging from low to high concentrations are applied to an affinity column containing an immobilized target [39,[82], [83], [84], [85], [86], [87]]. Unlike traditional frontal analysis, in this method the column is not rinsed with mobile phase containing no solute between each new solute application. The elimination of this washing step saves both cost and time for the analysis [82,83]. The chromatograms that are obtained in this process have a ladder-shaped pattern [82]. The dissociation equilibrium constant for an applied drug/solute with an immobilized target can be found from this data by using an expression like the one given in Eq. (12) for a system with a 1:1 interaction [39,84].

$$\frac{1}{\left[DT\right]}=\frac{K_{\mathrm{d}}}{\left[T\right]}\frac{1}{\left[D\right]}+\frac{1}{\left[T\right]}$$ (12)

In this relationship, [DT] is the concentration of drug-target complex obtained at a given applied concentration of the drug [D]. The term [T] is the initial concentration of the target, and $K_{\mathrm{d}}$ is the dissociation equilibrium constant for D with T. According to Eq. (12), a system with 1:1 binding should give a linear fit to a plot of $\frac{1}{\left[DT\right]}$ versus $\frac{1}{\left[D\right]}$ and allow $K_{\mathrm{d}}$ to be obtained through combined use of the slope and intercept [39,84].

Stepwise frontal analysis has been used in AC and HPAC to examine several types of drug-target interactions. For instance, this method has been used to study the binding of warfarin or digitoxin with HSA and of verapamil or tamsulosin with AGP [39,84]. This method has further been used to examine the interactions of drugs such as oxymetazoline, silodosin, tamsulosin, and the bioactive compound Schizandrin A with the cell membrane receptor alpha 1A adrenergic receptor; the binding of atenolol, esmolol and metoprolol with β1-adenoreceptor; and the interactions of clorprenaline, methoxyphenamine, salbutamol, and terbutaline with β2-adenoreceptor [83,[85], [86], [87]].

5. Kinetic methods for studying drug-target interactions

Kinetic studies can provide crucial insights into the rates and mechanisms of drug-target interactions for use in drug development, understanding the pharmacokinetics and pharmacodynamics of a therapeutic candidate, and in examining disease mechanisms [1,22,88]. Various approaches are available by which AC and HPAC can be used to obtain information on the rates of drug-target binding and dissociation. These methods include approaches based on peak fitting, band-broadening measurements or peak profiling, split-peak measurements, and peak decay analysis [1,22,88].

5.1. Peak fitting methods

Peak fitting under non-linear conditions and during zonal elution is one way information can be obtained on the kinetics of drug-target binding by AC and HPAC [57,58,[62], [63], [64],[88], [89], [90], [91]]. As shown earlier in eq. (8), this process can provide both the association equilibrium constant for drug-target binding and the apparent dissociation rate constant for the drug-target complex [58,[62], [63], [64]]. This technique has been used to examine the binding and dissociation for various drugs with β₂-adrenoceptor and cysteinyl-leukotriene, endothelin A, muscarinic-3 acetylcholine, and nicotinic acetylcholine receptors [53,56,[60], [61], [62], [63], [64], [65], [66]].

Peak fitting can also be used with frontal analysis to examine the interaction kinetics for a drug or solute with its target [52,[89], [90], [91], [92]]. As an example, the association rate constant (k_a) for such an interaction can be obtained from the apparent association rate constant (k_a,app) that is measured during frontal analysis, as demonstrated by Eq. (13) [89].

$$\frac{1}{k_{\mathrm{a},\text{app}}}=\frac{q_x{V}_{\mathrm{M}}}{F{n}_{mt}}+\frac{1}{k_{\mathrm{a}}}$$ (13)

In this equation, n_mt is the global mass transfer coefficient, which is dependent on the support particle size and dimensions of the column; F is the flow rate; V_M is the column void volume; and q_x is the loading capacity of the column per unit volume of mobile phase. The association rate constant for this system can be obtained from the intercept for a plot of 1/k_a,app versus q_x [89]. This technique has been used to estimate the association rate constants for anti-HSA antibodies with HSA [89,90].

Another way peak fitting can be employed for kinetic studies with frontal analysis is by using eq. (14) to estimate the dissociation rate constant (k_d) for a drug-target complex [52].

$$k_{\mathrm{d}}=\frac{2\left(V_D-V_D^{\ast }\right)}{d\sigma D^2/dF}$$ (14)

The terms shown in this equation include V_D, which is the breakthrough volume for the retained drug; $V_D^{\ast }$, the breakthrough volume for a non-retained solute; F, the flow rate; and σD², which is the variance of the breakthrough curve for the drug. A slope (dσD²/dF) is obtained when a plot of σD² versus F is made, which can then be used with other terms in eq. (14) to calculate the value of k_d [52,92]. The dissociation rate constant for the sugar p-nitrophenyl-a-d-mannopyranoside with the lectin concanavalin A has been estimated by this method [92].

5.2. Band-broadening measurements

Band-broadening measurements are a second route for conducting kinetic studies of drug-target interactions by AC and HPAC [52,88,91,[93], [94], [95], [96]]. This method is typically performed under linear elution conditions and by obtaining estimates of the various contributions to the overall plate height and extent of band-broadening for a drug injected onto an affinity column [52,93]. This involves a consideration of the total measured plate height versus the individual plate height terms due to eddy diffusion plus mobile phase mass transfer, stagnant mobile phase mass transfer, stationary phase mass transfer, longitudinal diffusion, and extra-column band-broadening [52,[93], [94], [95], [96]]. The goal is to determine the value of the plate height due to stationary phase mass transfer (H_s), which can then be used to obtain the dissociation rate constant (k_d) between the drug and immobilized target through use of Eq. (15) [52,93,96].

$${\mathrm{H}}_{\mathrm{s}}=\frac{2\mathrm{u}k}{k_{\mathrm{d}}{\left(1+k\right)}^2}$$ (15)

In this equation, u is the linear velocity of the mobile phase and k is the retention factor of the drug. One way the value of the dissociation rate constant can be obtained is by preparing a plot of H_s versus (u k)/(1 + k)², which should result in a linear relationship with a slope equal to 2/k_d [94,95].

The plate height method has been used to characterize the interaction kinetics for the separate chiral forms of D/L-tryptophan and R/S-warfarin with HSA [94,95]. This approach has further been used to study the binding of sugars with concanavalin A [93]. This method has also been employed to see how changes in conditions such as temperature, pH, solvent polarity, and ionic strength influence the binding of D/L-tryptophan with HSA and the effect of temperature on the dissociation rate constants for R/S-warfarin with this protein [94,96].

Peak profiling is another band-broadening method that has been utilized for studying dissociation kinetics of drug-target systems [65,[97], [98], [99], [100], [101], [102], [103], [104]]. This approach involves measuring the retention and overall band-broadening of an injected drug or solute on both an affinity column that contains the target and on a control column with no target present. These measurements are made at a known set of flow rates and under linear elution conditions. The results are then used with expressions like those in eqs. (16), (17) to estimate the apparent dissociation rate constant (k_d,app) for a drug and target, as applied in this case to a system with a single type of interaction site [[97], [98], [99]].

$$k_{\mathrm{d},\text{app}}=\frac{2t_M^2\left(t_R-t_M\right)}{t_M^2{\mathrm{\sigma }}_R^2-{\mathrm{t}}_R^2{\mathrm{\sigma }}_M^2}$$ (16)

$${\mathrm{H}}_R-{\mathrm{H}}_M={\mathrm{H}}_{\mathrm{s}}=\frac{2\mathrm{u}k}{k_{\mathrm{d},\text{app}}{\left(1+k\right)}^2}$$ (17)

In Eq. (16), ${\mathrm{\sigma }}_R^2$ and ${\mathrm{\sigma }}_M^2$ are the variances measured for the peak widths of the drug in the presence of the column with the target and on a control column, or for a non-retained solute on the affinity column. The terms $t_M$ and $t_R$ represent the void time and drug's retention time on the affinity column [99,101]. The value of k_d,app is often determined with Eq. (16) by using data from individual flow rates. In Eq. (17), data from multiple flow rates are instead used by measuring both the retention factor (k) for the drug and the total plate height on the affinity column (${\mathrm{H}}_R$) versus the control column (${\mathrm{H}}_M$) over a set of linear velocities (u). A plot is then made of (${\mathrm{H}}_R$ - ${\mathrm{H}}_M$) versus (u k)/(1 + k)², which should give a straight line with a slope of 2/k_d,app [99,100]. Expressions related to Eq. (17) have been derived for more complex systems, such as those involving multiple types of interactions of the drug with the affinity column [97,98,[100], [101], [102]].

Many applications have been reported in recent years for peak profiling in HPAC [52,65,97,98,[100], [101], [102], [103], [104]]. This method has been used to examine the following types of interactions: carbamazepine, imipramine, and l-tryptophan with HSA; ephedrine hydrochloride, isoprenaline hydrochloride, methoxyphenamine, salbutamol, and terbutaline with β₂-adrenoceptor; and chlorpromazine, disopyramide, imipramine, lidocaine, propranolol, and verapamil with AGP (see Fig. 6) [52,65,97,100,101,103]. This approach has further been used to simultaneously measure the dissociation rate constants for the chiral forms of phenytoin metabolites with HSA [102]. Peak profiling has been used as well to determine the dissociation constants for several drugs with β-cyclodextrin [88,104]. In general, the use of peak profiling or band-broadening measurements for kinetic studies tends to be best-suited for studying interactions with relatively fast dissociation rates and weak-to-moderate binding affinities [88,97,98,[100], [101], [102]]. In addition, both methods ideally require linear elution conditions and need a correction for extra-column band broadening to ensure accurate and reliable kinetic measurements [52,88].

Fig. 6.

Use of peak profiling and high-performance affinity chromatography (HPAC) to examine the interaction kinetics for several drugs injected over various flow rates at pH 7.4 and 37 °C onto a 1.0 cm × 2.1 mm inner diameter column containing α₁-acid glycoprotein (AGP), with the resulting data being plotted according to Eq. (17). H_s: plate height due to stationary phase mass transfer; u: linear velocity of the mobile phase; and k: retention factor of the drug. Adapted from Ref. [103] with permission.

5.3. Split-peak method

The split-peak method is a method that can be utilized in AC and HPAC to examine the association rates of drug-target binding [22,93,95,[105], [106], [107], [108], [109], [110]]. This method is based on the “split-peak effect”, in which a small fraction of an applied drug or solute can elute from an affinity column without binding to the target, even when the target is present in excess of the solute [95]. The extent of this effect can be enhanced by either increasing the application flow rate or reducing the analyte's residence time within the column. When this effect occurs under linear elution conditions, the association rate constant ($k_{\mathrm{a}}$) of an injected drug with an immobilized target can be acquired by using the following expression [95].

$$-\frac{1}{\ln \left(f\right)}=F\left(\frac{1}{k_{\mathrm{m}}{V}_{\mathrm{e}}}+\frac{1}{k_{\mathrm{a}}{\mathrm{m}}_T}\right)$$ (18)

In Eq. (18), F is the flow rate, $V_{\mathrm{e}}$ is the excluded or interparticle volume of the column, and $f$ is the free fraction of drug or solute that elutes from the column without being retained. The terms $k_{\mathrm{m}}$ and $m_T$ (also sometimes given as $m_L)$ are the forward mass transfer rate constant and the moles of active binding sites on the target within the column. According to Eq. (18), a plot of $-1/\ln \left(f\right)$ versus F under linear elution conditions should give a straight line with a slope that is related to a stagnant mobile phase mass transfer term, $1/\left(k_{\mathrm{m}}{V}_{\mathrm{e}}\right)$, and a term that describes the rate of analyte adsorption to the stationary phase, $1/\left(k_{\mathrm{a}}{m}_T\right)$. However, if solute adsorption to the target is the rate-limiting step in retention, this slope reduces to the latter term and can be used with a known value of $m_T$ to obtain $k_{\mathrm{a}}$ for binding of the drug and target [52,91,95]. Related equations that describe this effect under non-linear conditions and adsorption-limited kinetics are also available [105,107,108].

The split-peak method was initially employed to study the association rates for binding of rabbit immunoglobulin G (IgG) to immobilized protein A [95]. The same method was later used to examine the association kinetics of IgG-class antibodies with protein G, or combinations of protein A and G (see Fig. 7) [105,106]. Split-peak studies conducted under non-linear elution conditions have been used to estimate $k_{\mathrm{a}}$ values in such systems as the binding of HSA with anti-HSA antibodies, 2,4-dichlorophenoxyacetic acid (2,4-D) to anti-2,4-D antibodies, and l-thyroxine with anti-l-thyroxine aptamers [52,[107], [108], [109], [110]]. Most of these applications have involved interactions with moderate-to-high affinities and slow dissociation rates [52,91]. A key advantage of the split-peak method is that it only requires peak area measurements, making it easier to perform and more accurate than other methods for kinetic studies that instead rely on peak variances or fitting peak profiles [52,91,95]. However, the effectiveness of this approach does depend on the selection or optimization of various experimental conditions (e.g., flow rate, column size, and the amount of target) to create an observable split peak effect and measurable free solute fraction [52,91].

Fig. 7.

Analysis of association rates of immunoglobulin G (IgG)-class antibodies with protein A and protein G by the split-peak method. (A) The non-retained peaks and split-peak effect for rabbit IgG that was injected at flow rates of up to 2.0 mL/min on 6.35 mm × 2.1 mm inner diameter columns containing protein G or a control support. (B) The free fraction (f) that was measured for rabbit IgG as a function of flow rate and analyzed according to Eq. (18) for 6.35 mm × 2.1 mm inner diameter columns containing (top-to-bottom) protein G, a mixture of protein G and protein A, or protein A alone. Adapted from Ref. [105] with permission.

5.4. Peak decay method

The peak decay method is an approach in AC and HPAC for determining the dissociation rate constant ($k_{\mathrm{d}}$) for a drug-target interaction [22,69,88,93,[111], [112], [113], [114]]. One way this method can be carried out is by first applying the drug or solute to a column containing an immobilized target. A mobile phase with a high concentration of a competing agent is then introduced to bind the target and prevent any dissociated drug from rebinding to the column [111]. Alternatively, a relatively large amount of drug or solute can be applied and used with a flow rate and column size that prevent the solute from rebinding once it has dissociated from the target [112,113]. As the eluent is monitored, a decay curve is produced during release of drug from the column. When this curve is converted into a logarithmic form, it approaches the following response that can be used to obtain $k_{\mathrm{d}}$ [22,93].

$$\ln \left(\frac{d{m}_{Ee}}{dt}\right)=\ln \left(m_{Eo}{k}_{\mathrm{d}}\right)-k_{\mathrm{d}}t$$ (19)

The term $m_{Ee}$ in Eq. (19) represents the moles of drug/solute that elute from the column at time t, and $m_{Eo}$ is the initial moles of drug/solute that were retained by the immobilized target. According to Eq. (19), a plot of $\ln \left(d{m}_{Ee}/dt\right)$ versus time $t$ should yield a linear relationship, where the slope can be used to obtain the dissociation rate constant for the drug and target [52,91,114].

The peak decay method has been used to investigate dissociation kinetics of the sugar 4-methylumbelliferyl α-d-mannopyranoside from concanavalin A in the presence of 4-methylumbelliferyl α-d-galactopyranoside as a competing agent [111]. This method has also been used without a competing agent to determine dissociation rate constants for drugs such as acetohexamide, amitriptyline, diazepam, imipramine, lidocaine, nortriptyline, quinidine, tolbutamide, verapamil, and warfarin from HSA or AGP (see Fig. 8) [112,113]. This technique has further been employed to estimate $k_{\mathrm{d}}$ values for l-thyroxine from anti-thyroxine antibodies or aptamers, IgG-class antibodies from protein G, and 2,4-D from anti-2,4-D antibodies during the selection of elution conditions for immunoaffinity and bioaffinity chromatography [106,109,110]. As can be seen from these examples, the peak decay method has been used with systems that may have either weak-to-moderate or strong binding [106,[109], [110], [111], [112], [113]]. This method does require conditions in which solute dissociation from an immobilized target is the rate-limiting step during elution. To ensure these conditions are met, factors such as the flow rate, column size, support material, and amount of injected solute or competing agent concentration can be adjusted and optimized [52,88,91,111].

Fig. 8.

Use of the peak decay method to examine drug-target dissociation kinetics. The results in this example were obtained at pH 7.4 and 37 °C for racemic warfarin that was applied to and eluted at 4 mL/min from 1 mm × 4.6 mm inner diameter silica monolith columns that contained human serum albumin (HSA) or a control support. (A) The original chromatograms. (B) The same data after being converted into a logarithmic form for analysis according to Eq. (19). Adapted from Ref. [113] with permission.

6. AC methods for drug-target interaction studies using a secondary capture agent

Each of the methods in the following sections used a target that was immobilized in some manner within an affinity column. However, there are other forms of AC and HPAC in which a secondary capture agent is present in the column and both the drug and target are in solution. Two approaches that fall into this category are normal-role affinity chromatography and ultrafast affinity extraction [18,22].

6.1. Normal-role affinity chromatography

Normal-role affinity chromatography is a type of zonal elution in which a drug or target is used as either a mobile phase additive or the injected compound, while a secondary capture agent for one of these components is present in the affinity column [18,93,115,116]. This is typically done by using the drug (D) as the injected component, the target (T) as a mobile phase additive, and a secondary capture agent for the drug as the immobilized ligand (L) (see Fig. 9) [18]. If D can bind to both T and L and has a much lower concentration than either of these other components, the retention factor (k) for D in the absence of any non-specific binding is described by Eq. (20) [116].

$$\frac{1}{k}=\frac{1}{K_{\mathrm{a},DL}\left[L\right]}+\frac{K_{\mathrm{a},DT}\left[T\right]}{K_{\mathrm{a},DL}\left[L\right]}$$ (20)

Fig. 9.

General scheme for drug-target interaction studies based on normal-role affinity chromatography. D: drug; L: immobilized secondary capture agent and ligand; T: target. Adapted from Ref. [18] with permission.

The term K_a,DL in this equation is the association equilibrium constant between D and L, while K_a,DT is the corresponding association equilibrium constant between D and T. If D has 1:1 binding with both T and L, a plot of 1/k versus [T] that is made according to eq. (20) should produce a straight line with an intercept that gives K_a,DL and a slope/intercept ratio that provides K_a,DT [116]. Related equations can be obtained for more complex systems, such as those with multiple binding sites [115,116].

Normal-role elution was one of the earliest forms of AC that was used for drug-target studies [115]. This technique was originally used to examine the binding by targets such as enzymes and proteins with immobilized analogs of their inhibitors, substrates, or binding partners. A few examples included studies of the binding of ribonuclease and Staphylococcus nucleases to nucleotides, as well as studies of antibody-antigen interactions [115]. In work based on HPAC, this method has been utilized to examine the binding of β-cyclodextrin, as a mobile phase additive, with several drugs and on a column with immobilized HSA as a secondary capture agent for the drugs [116]. In each of these examples, normal-role elution has been used with a soluble target and immobilized ligand that compete or bind to the same drug. It has also been necessary for these interactions to be relatively fast on the time scale of the chromatographic analysis and that linear elution conditions be present for the injected solute [115,116]. In addition, this method has always required some means of monitoring the injected drug in the presence of the target in the mobile phase [116]. An advantage noted for this method is that it is relatively fast and can allow a single column to be used for examining binding by drugs with multiple types of targets [115,116].

6.2. Ultrafast affinity extraction

Ultrafast affinity extraction, or UAE, is another method that can use a secondary capture agent to study a drug-target interaction [[117], [118], [119], [120], [121], [122], [123], [124], [125], [126], [127]]. This method can also be employed to measure the free fractions of drugs and other solutes in standard solutions and biological or clinical samples [[117], [118], [119], [120]]. In this technique, injections of a drug or solute and mixtures of this drug/solute plus their target are made onto a small HPAC column that contains a secondary capture agent for the drug/solute, as shown in Fig. 10A [18,91,119]. This is typically done using a flow rate and column size that produce a sample residence time in the column on the millisecond-to-second time scale. These conditions are used to minimize and control dissociation of the drug from its complex with the target as their mixture passes through the column [[117], [118], [119], [120]].

Fig. 10.

Analysis of drug-target interactions by ultrafast affinity extraction. (A) General scheme for drug-target interaction studies based on ultrafast affinity extraction. (B) Use of ultrafast affinity extraction to examine the binding and rates of drug-target interactions. The results in (B) are for injections of warfarin plus soluble human serum albumin (HSA) at pH 7.4, 37 °C and various flow rates onto a 1.0 cm × 2.1 mm inner diameter affinity column that also contains HSA as an immobilized capture agent. The inset in (B) shows a plot of the same data that was obtained when using to Eq. (21). D, drug; T, target; t, column residence time; F_t, apparent free fraction for drug at residence time t. Adapted from Refs. [18,124] with permission.

Ultrafast affinity extraction can be used for determining both the dissociation rate constant and binding affinity for a drug-target interaction that is taking place in a solution-phase sample [118,119,125]. Eq. (21) is one expression that can be used for this type of analysis [118].

$$\ln \frac{1}{\left(1-F_t\right)}=k_{\mathrm{d}}t-\ln \left(1-F_0\right)$$ (21)

In this equation, $F_0$ is original fraction of free solute in the sample at equilibrium, and $F_t$ is the apparent free solute fraction that is measured at a given flow rate and sample residence time in the column, t. The term $k_{\mathrm{d}}$ is the dissociation rate constant of the drug or solute from the target in the injected sample. According to eq. (21), a plot of $\ln \left[1/\left(1-F_t\right)\right]$ versus $t$ for a system with pseudo-first order dissociation should result in a linear relationship where the slope provides $k_{\mathrm{d}}$ and $F_0$ can be estimated from the intercept [118]. The value of $F_0$ can, in turn, be used with the known total concentration of drug and target in the sample to estimate the association equilibrium constant or global affinity constant for the interaction of these components in solution [118,125].

Ultrafast affinity extraction has been used with HPAC in several studies to characterize the interactions between drugs and hormones with solute proteins such as HSA, AGP, sex hormone binding globulin, and equine serum albumin, as illustrated in Fig. 10B [103,[117], [118], [119],122,[124], [125], [126]]. These systems have had affinities ranging from 10³ to 10⁹ M⁻¹ and dissociation rate constants spanning from 10⁻² to 10¹ s⁻¹ [88,103,[122], [123], [124]]. The advantages of ultrafast affinity extraction noted in these applications have included its short analysis times, low sample volume requirements, and ability to conduct a label-free analysis of drug-target interactions in solution [88,91]. A limitation has been the need to carefully select and adjust factors such as the flow rate and column size to provide conditions that are suitable for the capture and measurement of free solute fractions by this approach [91,125].

7. Hybrid techniques using AC for drug-target interaction studies

Hybrid or hyphenated methods combine two or more distinct analytical techniques to produce a new approach capable of providing chemical information that could not be obtained by either of the original methods alone. For instance, AC has frequently been coupled with MS for the analysis and characterization of drug-target interactions (see prior example in Fig. 5) [74]. The combination of frontal analysis-based AC with MS for drug-target studies is often referred to as frontal affinity chromatography-mass spectrometry (FAC-MS) [[128], [129], [130], [131], [132]]. A major advantage of this combination is MS allows the identification of compounds solely based on their mass-to-charge ratios and can be used as a label-free detection system in applications such as the screening of drug leads for a given target [128,129]. However, it is necessary to also use a mobile phase for AC that is compatible with MS for such work [129]. In FAC-MS, a target such as a protein is immobilized onto a suitable support and packed into a column [133]. A mixture of drug candidates that may bind the target is then continuously introduced to the column. As a drug candidate binds, the target will approach saturation of its binding sites and the excess drug candidate will elute from the column, giving a breakthrough curve. Non-retained or weakly retained compounds will elute first from the column, and stronger binding compounds will appear last [134]. This enables FAC-MS to determine the relative binding strength of each compound in the mixture [133].

Several applications have been reported for FAC-MS in drug-target binding studies. This method has been used with binding agents and targets that have including enzymes, antibodies, growth receptor mimics, and lectins [128,[133], [134], [135], [136], [137], [138], [139], [140], [141], [142], [143], [144], [145], [146], [147], [148]]. For instance, this method has been used to distinguish between competitive ligands for the ATP and substrate sites of protein kinase C [131]. FAC-MS has also been adopted as an effective tool for differentiating the binding of stereoisomers, as used to examine the interactions of sialyl-lacto-N-tetrose b and sialyl-lacto-N-tetrose c with Polyporus squamosus lectin [137]. This method has further been utilized to study catalytic activity of the enzyme N-acetylglucosaminyltransferase V and the binding of inhibitors to dihydrofolate reductase [136,138].

Some applications of FAC-MS have been reported to be capable of screening up to 10,000 compounds per day per instrument, making it attractive as a high-throughput tool for drug discovery [138]. However, this technique is not applicable for drugs with slow association and dissociation rates. In addition, the measured affinities may deviate from their values in solution if the target does not fully retain its native structure after immobilization, a situation which can be avoided by careful optimization of immobilization or by using an entrapment-based approach [136,[139], [140], [141], [142]]. Although FAC-MS can allow the estimation of binding affinities for multiple compounds in a single experiment, these values may be underestimated if several ligands in the mixture compete for the same binding site. Even in this case, the relative order of binding strengths that is obtained for compounds in a mixture is usually correct, as can be used to select the most promising drug candidates for further development [138].

8. Advantages and limitations of AC methods for drug-target interaction studies

The previous sections in this review examined the principles and applications for various techniques that can be used in AC to study drug-target interactions. An overview of these methods is provided in Table 1 [1,18,19,22,23,28,36,49,51,52,[68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87], [88],93,[115], [116], [117], [118], [119], [120], [121], [122], [123], [124], [125], [126], [127]]. This table also provides a summary of the relative advantages and limitations for each of these methods, as will be compared and discussed further in this section.

Table 1.

Summary of AC methods for the analysis of drug-target interactions.^a

Method & Principle	Advantages	Limitations
Zonal elution	•Common technique to examine reversible drug-target interactions. •Can easily determine the overall extent of binding and affinity of a solute with its target, as well as variations in this binding with temperature, mobile phase content, and type of drug or target's structure. •Can be used to determine the location of a drug's binding regions on a target and site-specific affinities for the drug. •Requires less time and drug than frontal analysis. •Can be used to provide fast screening of drug-target and drug-drug interactions.	•Requires drug and target to have relatively fast interactions on the time scale of the chromatographic process. •Requires AC column with active, immobilized target and a control column. •Typically requires linear elution conditions and small amounts of injected solute, although methods based on non-linear elution are now available.
Retention is measured to provide information on an interaction between a small plug of a drug or probe solute that is injected onto an affinity column with an immobilized form of the target; an additive or competing agent may also be present in the mobile phase. [18,19,22,23,28]
Frontal analysis	•Complementary approach to zonal elution by providing information on overall drug-target binding. •Can determine equilibrium constants, number of binding regions, and effects of temperature, mobile phase content, type of drug and target's structure. •Has been used in hybrid methods with mass spectrometry to study drug-solute competition for a target and to screen drug candidates.	•Requires drug and target to have relatively fast interactions on the time scale of the chromatographic process for measuring binding constants. •Requires AC column with active, immobilized target and a control column. •Typically uses larger amount of drug or solute and requires longer analysis times than zonal elution, although newer methods and column formats have reduced these requirements.
Drug is continuously introduced to a column containing an immobilized target; the breakthrough curve of the drug is analyzed to obtain information on the drug-target interaction. [22,36,49,51,[68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87]]
Kinetic methods	•Many approaches available for measuring binding or dissociation rates for drug-target interactions. •Some approaches can be used with the same data as obtained by zonal elution or frontal analysis. •Overall set of techniques can examine a wide range of rate constants, including values not easily accessible by other kinetic methods.	•Requires AC column with active, immobilized target and a control column. •Some methods require special conditions, such as for column size and flow rate.
A column with an immobilized target is used through approaches such as peak fitting, band-broadening studies, split-peak or peak decay analysis to study the interaction rates of the target with an applied drug. [1,22,52,88,89]
Secondary capture methods	•Target remains in solution and does not need to be immobilized. •Methods can be used for both binding and kinetic studies. •Needs only small amounts of drug and target. •Suitable for high-throughput applications (UAE).	•Requires AC column with secondary capture agent for drug, but which will not bind the target, and a control column. •Must be able to detect the drug in the presence of the soluble target (normal-mode AC). •Requires special column size and flow rate conditions (UAE).
Drug and target interactions are examined in solution by using either a secondary immobilized agent that competes for drug binding (normal-role AC) [18,93,115,116] or that can retain the unbound drug in drug/target samples (UAE). [[117], [118], [119], [120], [121], [122], [123], [124], [125], [126], [127]]

^aAC: affinity chromatography; UAE: ultrafast affinity extraction.

The first method listed in Table 1 is zonal elution, which was described earlier as a common technique to examine reversible drug-target binding by using small injections of a drug or probe solute onto an AC column containing an immobilized form of the target [18,19,22,28]. This method is a useful and convenient tool for providing detailed information on how a drug binds to a target. This information can include the overall extent of binding, the affinity of the drug with the target at selected binding regions, and the locations of these regions. Data can further be obtained on how the binding changes with variations in temperature, mobile phase content, the type of drug or the target's structure [22,28]. Another advantage of this method is it requires less time and drug than frontal analysis, another common AC method in Table 1 for evaluating drug-target binding [22,28]. It is also possible when using zonal elution as a tool with affinity microcolumns to provide relatively fast screening of drug-target and drug-drug interactions (i.e., on the minute time scale) [22,28]. This approach does require that the drug and target have relatively fast interactions on the time scale of the chromatographic process, thus allowing a local equilibrium to be present at the center of the injected drug or solute's peak [22,23,28]. Another potential limitation is this method requires an AC column that contains an immobilized and active target, as well as a corresponding control column in which no target is present (i.e., to correct for any binding by drug to the support) [22,28]. In the past, most zonal elution methods have used linear elution conditions and small amounts of injected solute to study drug-target binding [22,28]. However, several approaches based on non-linear elution and that can employ larger amounts of drug or solute are now available as well for this work [[52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66]].

The second group of AC methods discussed in this review were those based on frontal analysis [22,36,49,51,[68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84], [85], [86], [87]]. As described in Table 1, in this technique a drug is continuously introduced to a column that contains an immobilized target, with the drug's breakthrough curve then being utilized to provide information on the drug-target interaction [22]. As was true for zonal elution, a key advantage of frontal analysis is its ability to provide detailed information on the interactions taking place between a drug and the target of interest. In this regard, frontal analysis is complementary to zonal elution in that it provides information on the overall binding of a drug with a target [22,68,72]. Like zonal elution, frontal analysis can be used to determine equilibrium constants for these interactions and the effects of variations in temperature, mobile phase content, the type of drug or the target's structure [22,68]. However, frontal analysis can further provide information on the number and general types of binding regions that may be present for the drug with a target [68,72]. Another advantage of frontal analysis is it has frequently been used in hybrid methods with mass spectrometry (i.e., FAC-MS) to study drug-solute competition or to screen drug candidates for a target [[128], [129], [130], [131], [132]]. Many of the limitations for frontal analysis in AC are the same as for zonal elution. For example, frontal analysis requires that fast interactions be present between the drug and target if this method is to be used for measuring binding constants [22]. Also, frontal analysis requires an AC column that contains an active form of the target and a control column to correct for non-specific binding to the support [22,68]. One further limitation for frontal analysis is that it typically uses a much larger amount of drug or solute and requires longer analysis times than zonal elution; however, newer approaches based on stepwise frontal analysis or affinity microcolumns have helped reduce both these requirements [39,49,70,[82], [83], [84], [85], [86], [87],98].

The third set of AC methods in Table 1 are those that can be used in kinetic studies of a drug-target interaction. As discussed earlier, there are many approaches that are available for such work, including peak fitting, band-broadening studies, and split-peak or peak decay analysis [1,22,52,88,89]. Each of these techniques requires an AC column that contains the immobilized target, as well as a corresponding control column [22]. Some of these methods, such as peak fitting or band-broadening measurements, can often be used with the same data that are obtained through zonal elution or frontal analysis studies of drug-target binding [52,88]. Other techniques (i.e., the split-peak method or peak decay analysis) require more specialized conditions (e.g., small columns and/or fast flow rates) [22,52,88]. This combined set of tools can be employed to examine a wide range of rate constants. This range includes some values (e.g., rate constants for weak-to-moderate binding systems with relatively fast dissociation) that are difficult to obtain by other methods for kinetic studies [88].

The fourth set of AC methods in Table 1 are those in which the column contains a secondary capture agent for examining a solution-phase drug-target interaction. One method in this category is normal-role AC, in which the secondary immobilized agent is allowed to compete with a soluble form of the target for binding to a drug [18,93,115,116]. A second approach in this group is ultrafast affinity extraction, or UAE; in this method, a column containing the secondary binding agent is used to capture and retain the unbound form of a drug in samples that may contain both this drug and the target [[117], [118], [119], [120], [121], [122], [123], [124], [125], [126], [127]]. There are several important advantages to this set of techniques. For instance, in both normal-role AC and ultrafast affinity extraction the target remains in solution and is not immobilized [18]. Also, each of these methods can be used for either binding or kinetic studies and requires only small amounts of the drug and target [18,115]. In addition, ultrafast affinity extraction can provide results within seconds or minutes, making it suitable for high-throughput applications [91,119]. A limitation of these methods is they do need an AC column with a secondary capture agent that can bind to the desired drug but not the target [18]. Normal-role AC has the additional limitation that it must be possible to monitor the drug in the presence of the target in solution [18,115,116]. For ultrafast affinity extraction, an additional requirement is the need for special column size and flow rate conditions to obtain effective capture of the drug's non-bound fraction in applied samples [18,91,119].

9. Conclusions

This review presented an overview of how AC and HPAC can be used to examine the binding strength and kinetics of drug-target interactions. The general principles, applications, and advantages or limitations of each method were discussed (see summary in Table 1). Methods that could be used for binding constant measurements or in competition studies included several formats based on zonal elution or frontal analysis. Kinetic studies could be conducted by techniques such as peak fitting, band-broadening measurements, and the split-peak method or peak decay methods. Approaches in AC and HPAC that use secondary capture agents for interaction studies, such as normal-role elution or ultrafast affinity extraction, were considered as well. The combined use of AC or HPAC with MS in hybrid techniques was also discussed. The information and examples of applications provided in this review should allow for the future extension of these methods to other drug-target systems. This, in turn, should allow for even greater use of AC, HPAC, and related methods in studying interactions of interest in drug discovery, drug development, and biomedical research.

CRediT authorship contribution statement

David S. Hage: Supervision, Formal analysis, Writing – original draft, Funding acquisition, Writing – review & editing, Project administration, Conceptualization. Sadia Sharmeen: Investigation, Writing – original draft, Writing – review & editing. B.K. Sajeeb: Investigation, Writing – original draft. Harshana Olupathage: Writing – original draft, Investigation. Md Masudur Rahman: Writing – original draft, Investigation. Isaac Kyei: Writing – original draft, Investigation. Samiul Alim: Writing – original draft, Investigation. Nigar Sultana Pinky: Investigation, Writing – original draft.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.