Moderate to high throughput in vitro binding kinetics for drug discovery
This review provides a concise summary for state of the art, moderate to high throughput in vitro technologies being employed to studydrug-target binding kinetics. These technologies cover a wide kinetic timescale spanning up to nine orders of magnitude from milliseconds to days. Automated stopped flow measurestransient and (pre)steady state kinetics from milliseconds to seconds. For seconds to hours timescale kinetics we discuss surface plasmon resonance-based biosensor,global progress curve analysis for high throughput kinetic profiling of enzyme inhibitors and activators, and filtration plate-based radioligand or fluorescentbinding assays for receptor binding kinetics. Jump dilution after pre-incubation is the preferred method for very slow kinetics lasting for days. The basic principles,best practices and simulated data for these technologies are described. Finally, the application of a universal label-free technology, liquid chromatography coupledtandem mass spectrometry (LC/MS/MS), is briefly reviewed. Select literature references are highlighted for in-depth understanding. A new reality is dawning whereinbinding kinetics is an integral and routine part of mechanism of action elucidation and translational, quantitative pharmacology for drug discovery.
Binding Kinetics, Bk, Stopped Flow, Surface Plasma Resonance, SPR, Global Progress Curve Analysis, GPCA, Radioligand Binding, Jump Dilution, Review
Binding kinetics (BK) is a universal molecular phenomenon for all biochemical processes. Thus, all metabolic pathways are mediated by enzymatic turnovers that occur with the initial substrate binding to and final product being released from the enzyme. Similarly, all regulatory mechanisms are accomplished by the binding of a regulatory molecule to its effector molecule and consequent conformational changes. Finally, all signal transduction pathways are a cascade of obligatory binding events beginning with the initial interaction of the signaling molecule with its receptor. Proteins in particular perform a host of biological functions, serving for example as enzymes, receptors, transporters, regulatory and signal transduction components, defense mechanisms, and structural, contractile and motor elements. Protein function is in turn controlled in response to various stimuli by the binding of their natural ligands or substrates, as well as by allosteric ligands that modulate their activity. Such ligands can be small molecules, peptides, proteins or nucleic acids. Importantly, the actions of these native ligands can be phenocopied and improved upon by synthetic molecules (pharmacological agents) that also act as orthosteric and allosteric inhibitors and activators of their targets. Defining the parameters that comprise protein/ligand interactions is central to our understanding of both modern biochemistry and the drug discovery process.
All drugs manifest their efficacy and toxicity by binding to “appropriate” and “inappropriate” targets with various levels of occupancy and duration. The manner in which binding kinetics can fine tune pharmacodynamics has recently been recognized (1-5), most notably using the concept of drug-target residence time (6-13). Binding kinetics encompasses not only molecular association and dissociation; it also broadly includes reversible isomerization via conformational dynamics and irreversible inactivation via catalysis, aggregation or precipitation. Binding kinetics, together with enzyme turnover kinetics, provides a quantitative description and mechanism of action (MOA) understanding for all pathways in any biological system. Multiparametric models are known to exhibit both “sloppiness” and “stiffness”, i.e.,insensitivity and sensitivity, respectively, of model response to parameter changes (14-18). Critical nodes of interactions exist for the complex web of an interactome (19-22) and complementary experimental design can help to determine most or all parameters reliably (14, 18). Identifying such critical interaction nodes and quantitatively measuring the associated kinetics by in vitro and in vivo methods to predict model responses constitutes an essential task for systems biology and systems pharmacology.
A number of in vitro technologies are available for measuring kinetics on the various timescales encountered in biomedical research and drug discovery (Table 1). For ultrafast, nanosecond timescale kinetics, time-resolved fluorescence is used to monitor protein folding and conformational dynamics as well as for ultrafast enzyme reactions. This ultrafast technology is out of the scope of the present review and may be elaborated in a future update of this review. For fast, millisecond to second timescale kinetics, stopped flow and quenched flow are used to monitor transient and (pre)steady state binding and catalytic events. We will focus on the automated stopped flow method. For ordinary second to hour timescale kinetics, biosensors of various detection schemes such as surface plasmon resonance (SPR) (23, 24), SPR imaging (25-28), biolayer interferometry (29), back scattering interferometry (30-32), resonant waveguide grating and dual polarization interferometry (33), plasmonic-based electrochemical impedance imaging (34), DNA-based switchSENSE (35-38), and second harmonic generation imaging (39) are used to monitor biomolecular interactions and, for some platforms, also conformational dynamics. Biosensor methods complement traditional radioligand binding assays performed in the scintillation proximity or microtiter plate filtration assay formats. High throughput kinetic profiling of enzyme inhibitors/activators or receptor antagonists/agonists by global progress curve analysis has been developed (40, 41). The kinetic size-exclusion chromatography method is able to monitor kinetics on the timescale of seconds to minutes (42, 43). For slow timescale kinetics that takes hours to days, preincubation followed by jump dilution is used to monitor the very slow dissociation of a binary complex (44). Finally, liquid chromatography coupled tandem mass spectrometry (LC/MS/MS) can be used in conjunction with many of the above techniques.
|Kinetic timescale||Prevailing method||Key advantages||Key requirement|
|Nanoseconds||Time-resolved fluorescence||Dynamics/binding/function based ultrafast kinetics||Fluorescence reflecting the rapid kinetics|
|Milliseconds to seconds||Stopped or quenched flow||Binding/function based fast kinetics, low to moderate throughput||Specialized instrument|
|Seconds to hours||Surface plasmon resonance or other biosensors||Binding based moderately fast to slow kinetics, moderate throughput||Specialized instrument|
|Global progress curve analysis||Binding/function based moderately fast to slow kinetics, high throughput||Continuous detection preferred, though not obligatory|
|Radioligand or fluorescent ligand binding assay||Binding based moderately fast to slow kinetics, high throughput||Radiolabeling or fluorescent labeling|
|Hours to days||Jump dilution||Binding/function based very slow kinetics, moderate to high throughput||Long stability of complex|
In this review, we outline the basic operating principles and key advantages and requirements of six prevailing technologies commonly used in our laboratories for moderate to high throughput in vitro kinetics studies in drug discovery. We also share the best practices for each technology. We illustrate the timescale and observation of each technology with simulated data. Readers interested in other in vitro and in vivo technology platforms (such as calorimetry-based kinetics (45-47), single molecule kinetics (48-53) or positron emission tomography based kinetics (54-60)) should consult the references cited herein and elsewhere.
3. In vitro technologies for binding kinetics
3.1. Automated stopped flow method
Stopped flow instrumentation for fast kinetics was developed more than 75 years ago (61, 62) and available for general use for more than 50 years (63). Until recently, it was not easily adaptable to high throughput data generation and analysis. The main difficulties preventing adaptation to higher throughput screening included a tedious sample handling requirement for accurate use of the instrument, relatively large sample consumption, and difficulty of rapid data analysis. Recent applications of robotic automation for liquid handling coupled to automatic computational analysis with instruments such as the Auto SF-120 from KinTek Corp. are significantly enhancing the throughput of stopped flow data acquisition and analysis, providing the impetus for us to discuss this technique in the present review.
The design of stopped flow instruments has been described extensively elsewhere (61, 63-65). Briefly, a stopped flow apparatus consists of two or more loading syringes containing separate reactants and a stop syringe that rapidly stops the flow of the reactants in a detection chamber. Generally, the monitoring of millisecond (ms) timescale reactions requires rapid and sensitive detection normally achieved through optical methods (66), e.g.absorbance, fluorescence, static light scattering, or circular dichroism. Alternate detection methods such as NMR (67, 68), EPR (69), and small angle X-ray scattering (70) also have been used, but these are generally reserved for analysis of slower reactions. Most commercially available optical detection-based instrumentation has a dead time in the 0.5 – 2 ms range allowing for measurement of reactions with millisecond half-lives (See Figure 1). While the absolute limits of detection are predicated on the intensity of the optical signal change, under some circumstances association rate constants, kon, on the order of 1 × 108 M-1s-1, can be reliably determined (64, 65).
The best practices for stopped flow assay design are listed in Table 2. For single turnover or transient enzyme kinetics, a sub-stoichiometric amount of substrate, compared to enzyme concentration, should be used. For (pre)steady state enzyme kinetics and simple binding kinetics, up to saturating level of substrate or ligand is used. For the simplest cases, the kinetics of the binding interaction between two reactants is measured directly by the difference in optical signature between the free and bound forms of the reactants (Figure 2A). In a 1-step binding interaction where one reactant is in large excess, this interaction follows a pseudo-first order exponential change. Kinetic constants can be characterized monitoring the change in kobs (the apparent pseudo-first order rate constant for reaching steady state or equilibrium) that occurs as the concentration of one of the reactants is varied. Plotting the kobs versus reactant concentration will yield a straight line with the slope indicating the kon and the Y-intercept representing the koff. A hyperbolic line would indicate a 2-step kinetic mechanism such as association before or after an isomerization step resulting in higher or lower asymptote, respectively (12). From this method of analysis, the value of kon can be well determined while the value of koff may be associated with large error since association and dissociation often occur within significantly different timescales. Accurate analysis of koff is limited by the reactant concentrations required to enable detection, while still ensuring pseudo-first order kinetics of association. In practice this limits direct determination of koff, and by extension Kd, to binding interactions with rapid dissociation (koff > 0.01 s-1) (64).
|Design of mixing||Pseudo-first order for (pre) steady state kinetics|
|Stoichiometry of binding||Sub-stoichiometric ratio for transient, single turnover kinetics; excess for dose-dependent binding or turnover kinetics|
|Concentration||Adequate for signal/noise ratio(S/N > 10)|
|Timescale of detection||Exponential, capturing >4 halftimes|
|Mode of detection||Fluorescence, fluorescence polarization, absorbance, light scattering|
|Temperature control||Thermostatic for reaction chamber and reagents|
|Data analysis||Concentration-dependent kobs or amplitude; concentration-independent dissociation, isomerization or inactivation|
As noted above (Table 1 and Figure 1), stopped flow methodology provides a solution based method for monitoring reactions occurring at millisecond to second scales. Also notable, for interactions that are associated with a change in optical signal of sufficient magnitude for detection with this instrumentation, direct measurement of association of reactants can be made. This enables reactions in biological systems to be studied in complexities ranging from whole organisms (71) to purified proteins. Despite the broad range of systems that can be monitored by stopped flow there are certain limits to the technique that are not shared by the remaining methods discussed below. Due to the reliance on optical detection, interactions displaying nonspecific optical signals can interfere with detection. Also for most commercially available equipment, volumes approaching 100 µL are necessary for each reaction measured and in many cases, high concentrations of reactants (~µM) are required for a sufficient optical signature to be monitored.
While rapid kinetics can be monitored continuously in real time by stopped flow method, quenched flow method can also be employed to prepare effectively quenched fast reaction aliquots which are subsequently monitored discontinuously by a variety of offline detection schemes, most notably the universal, label-free mass spectrometry (see Section 3.6).
3.2. Surface plasmon resonance based biosensors
There are various platforms of biosensors as mentioned earlier. This review focuses on the most widely used surface plasmon resonance (SPR) based biosensors, such as BIAcore series of GE Healthcare and, more recently, Pioneer series of SensiQ Technologies employing Taylor dispersion to create a continuous concentration gradient spanning three orders of magnitude. The basic principle of SPR has been reviewed elsewhere (23, 24). Briefly, under total internal reflection conditions from a glass prism coated with gold film, the photons in incident light at a certain resonance angle will have the same momentum as the surface plasmons (i.e., free conduction electrons on the gold surface), resulting in resonant energy transfer from photons to plasmons and a concomitant loss of reflection intensity at this resonance angle. This surface plasmon resonance angle is sensitive to the reflective index changes within ~100 nm of the gold surface, such as on the adjacent carboxymethyl dextran layer where biomolecular interaction occurs. The SPR angular shifts as a function of the refractive index changes due to binding and mass increases form the basis of SPR detection. The detection area is usually a small fraction of a nanoliter on the biosensor chip. The detection is continuous, at a frequency of up to 40 Hertz for one interaction spot. This detection speed along with the dispersion at solvent change interface sets the upper or fast limit of kinetics to be < 1 s-1. The lower or slow limit of kinetics depends on the baseline drift and is usually around 10-4 s-1. The above two limits set the timescale of SPR-based binding kinetics on the orders of seconds to a few hours. The instrument setup for injection volume and flow rate usually allows association time lasting several minutes, which enables kon to be measured reliably between ~103 and 107 (up to 108 for proteins) M-1 s-1. Such kon values, when coupled with the dynamic range of koff (10-4 to 10-1 s-1), allow SPR-measurable affinity constants in the range of ~ 1pM to ~0.1 mM. This range may be extended to < 1pM with properly designed competition binding SPR.
The best practices for SPR-based binding kinetics are summarized in Table 3. More instructions and applications can be found in a book on SPR (72). Despite the diverse offering of biosensor chips, we recommend the nitrilotriacetic acid (NTA) derivatized chips and Streptavidin or NeutrAvidin (deglycosylated avidin) immobilized directly to CM chips by amine coupling or captured indirectly on CAP chips via DNA hybridization. This is because of the widespread use of generally non-perturbing histidine (e.g.,decahistidine (73)) and biotin tags on recombinant proteins (74, 75), as well as the stability of the capture. The NTA surface can also be easily regenerated by 200 mM imidazole (which removes the His-tagged protein) or 350 mM EDTA (which removes both nickel and the His-tagged protein, and therefore nickel addition must precede protein capturing for next run). To prevent leaching, nickel NTA-captured His-tagged proteins may be further covalently, but sparingly, tethered to the matrix using very brief (seconds) and dilute amine coupling reagents. Covalent immobilization may inadvertently compromise the binding competence of the protein and should be used with caution. Streptavidin-or NeutrAvidin-coated chips may also be used to firmly capture biotinylated (preferably site-specific, mono-biotinylated) biomolecules, which gives no leaching due to the extremely high affinity binding and ultra-slow dissociation. Streptavidin indirectly captured via DNA hybridization provides not only stable immobilization of biotinylated proteins, but easy regeneration (such as with BIAcore biotin CAPture kit). Amine coupling under acidic conditions (pH 4-5.5), while universally applicable to untagged or non-biotinylated proteins, suffers from the potential of damaging some or all protein binding to some or all compounds. If amine coupling must be used, it is recommended to use a large molar excess of binding site-protective compound with Kd in sub-μM range and devoid of a reactive primary or secondary amine group under acidic condition that does not irreversibly denature the protein (as easily assessed by a thermal shift assay (76) in relevant buffers).
|Choice of biosensor chips||Nickel NTA or streptavidin/NeutrAvidin chip for capturing high-affinity and non-covalent His10-tagged or mono-biotinylated proteins, respectively|
|Level of immobilization||As low as detection limit permits and as high as invivo target density dictates|
|Concentrations of analytes||Adequate to see apparent curvature in association for at least two sensorgrams|
|Multi-cycle or single cycle||Optimally designed single cycle is preferred, especially if regeneration is difficult|
|Dissociation length||Adequate dissociation (at least four halftimes unless very slow)|
|Regeneration||Use only when needed or shown to be non-harmful. Removal of His-tagged protein from NTA chip or biotinylated protein from CAP chip before re-capturing for subsequent cycle|
|Data analysis||Simple 1:1 binding model with mass transport limit, unless there is compelling reasons to use more complex models (such as heterogeneous analyte model for binding by racemic or otherwise competitive analytes)|
|Parameter quality control||Overall good fitting within limits, Chi-square<1%Rmax, U(uniqueness) value <15, kt (flow-rate corrected mass transport rate constant) >kon*Rmax|
The level of immobilization should be as low as possible to allow final binding signals < 20 RU for small molecules and < 100 RU for large proteins binding kinetics. Preferably, biosensor chips with low density or thin matrix are used to minimize mass transport limitation and rebinding as well as surface heterogeneity (77). An exception to this rule is encountered when a high density of proteins on the biosensor surface is required to mimic the in vivo protein density at target site. Ideally, the immobilized protein should be homogenous, not prone to aggregation, equally binding-competent before and after immobilization and sufficiently stable to endure extended runs. The concentrations of analytes to be injected should be judiciously chosen before the run (definitely before an optimized re-run). This can be difficult without some prior knowledge of the binding kinetics. Future generations of SPR instrument should employ real time data analysis and automated decision making options in the experimental design to optimize the very first kinetic run and to minimize subsequent re-runs. For steady state affinity determination of fast-on and fast-off kinetics, analyte concentrations should bracket the steady state affinity constant. But for slow binding kinetics, one may have to use significantly more concentrated analytes to induce observable curvature during the association phase. Even with extended association at low flow rate (which may exacerbate mass transport limitation), steady state may not be readily achieved for slow binding analytes. As a rule of thumb, at least two well-curved sensorgrams in both association and dissociation phases are needed to permit reliable measurement of kinetic constants (78). The dissociation phase should be long enough to cover at least four halftimes, unless the dissociation is very slow and near the detection lower limit. In our experience, it is best to run ample buffer blanks throughout the runs, including in the beginning of the run to stabilize the baseline.
Biosensor surface regeneration is not needed for fast dissociating analytes. If needed with slow dissociation or irreversible binding, regeneration conditions for the NTA and CAP chips are the most straightforward, as mentioned earlier. Inclusion of a known allosteric compound that associates and dissociates rapidly (within seconds) and substantially enhances the dissociation of a slow binding analyte is useful for regeneration. Unless regeneration is demonstrated to have negligible effect on the immobilized protein and binding kinetics, repetitive regeneration should be minimized. One way of accomplishing this is by single-cycle kinetics, i.e.multiple, stacked up injections of increasing concentrations of analyte are followed by a single adequately long dissociation. Care must be taken with single-cycle multiple injections since each successive injection will have progressively less free protein to bind (thus the apparent Rmax is dwindling). It is prudent to choose the concentrations so that at least two well-curved association phases are observed with good magnitude of signal increase (compared to noise).
The data analysis of SPR sensorgrams are performed by global curve fitting. The veracity of the kinetic constants depends on the quality of the raw data. Nonspecific binding to biosensor matrix should be insignificant to minimize distortion of kinetic data after reference sensorgram subtraction. Attention must be paid to the buffer blank runs’ offset drift and chronologically close blanks should be used for referencing. Not all data contribute equally to the analysis. Spikes around injection junctions are best removed. Long stretches of data (such as a long drifting baseline after a rapid dissociation) which contribute little to, or worse, distort, the overall fitting should be cropped out. Well-curved sensorgrams whose curvatures are properly emphasized in the fitting permit reliable analysis. For very slow dissociating compounds, only association rate may be determined. For fast-on and fast-off compounds that give straight up and down changes in sensorgrams, only the steady state affinity constant may be determined. For extremely fast-on compounds or with high density protein immobilization, mass transport limitation may hinder the accurate determination of association and dissociation rate constants. For kinetics exhibiting heterogeneous, two or more sites binding (especially with amine coupling method of immobilization), the reliability of determination will decrease with more floating parameters, even if the proper heterogeneous ligand model is used for curve fitting. Antibody bivalence will cause avidity due to rebinding to the same antigenic site. Antigen multivalence will cause “sticky mouse trap” phenomenon due to repeated rebinding to adjacent sites by the same injected, “bipedal walking” antibody. Protein binding kinetics with racemic or otherwise competitive analytes require curve fitting using the heterogeneous analyte model.
The final fit lines should adequately describe the overall majority behavior of binding kinetics. Preferably, the residuals should be randomly distributed with chi-square no more than ~1% of the Rmax. The kinetic constants should be non-correlated, yielding a uniqueness (U) value less than 15. Verify that the flow rate-corrected mass transport rate constant (kt) is more, ideally far more, than the product of kon and Rmax such that mass transport limitation is considered insignificant. Shown in Figure 2B is simulated biosensor data for binding kinetics with fast (blue curve) and slow (red curve) on-rate and off-rate near the detection boundary by SPR.
3.3. Global progress curve analysis
There are two types of kinetics assays, exclusively binding-based (such as SPR and radioligand binding) or jointly binding and function-based. Global progress curve analysis (GPCA), while equally applicable to exclusively binding-based, time-and dose-dependent assay (see Section 3.4 and also (40)), here refers specifically to a jointly binding and function-based assay that monitors the time-and dose-dependent enzyme inhibition or activation in the presence of enzyme substrate(s) and a test compound. Any binding event that is observably slow compared to the enzyme turnover kinetics will exhibit time-and dose-dependence that reflects the slow binding kinetics. That is the foundation for GPCA. A book chapter has been devoted to the details of experimental design and data analysis of this method (41). We summarize a few key points for best practices in Table 4.
|Continuous detection||Design a chromogenic, fluorogenic or mass detectable product (for signal increase) or substrate (for signal decrease)|
|Enzyme stability||Stable during assay timeframe, best >>2 hours|
|Concentrations of enzyme||As low as required by adequate signal and sustained steady state over the assay time course|
|Concentrations of substrate||Near Km, or well above Km (if competitive) or below Km (if uncompetitive) to decrease apparent affinity of compound|
|Concentrations of compound||Adequate to generate >2 good curves, bracketing Ki and Ki* values|
|Time of detection||Long enough to observe some good curvature and yet no-inhibitor control remains linear|
A continuous detection of unquenched reactions is preferred over discontinuous detection of effectively quenched reaction aliquots, but sufficiently time-resolved data from discontinuous detection will permit reliable analysis. A continuous detection is best achieved with a signal increase assay monitoring chromogenic or fluorogenic product formation, but a signal decrease assay monitoring the disappearance of substrate is acceptable. Liquid chromatography coupled mass spectrometry (LCMS) can also be used for label-free detection of product and/or substrate. Another prerequisite for GPCA is the extended stability of enzyme over the time course of the monitored turnover reactions. The concentration of the enzyme should be low enough to allow for both reliable determination of high potency compounds under non-tight binding condition and sustained steady state during the assay time frame, but not too low to avoid enzyme sticking to assay wells or substantial dissociation of oligomeric enzyme. The concentration of substrate used is preferably near Km value, especially when prior knowledge about the mode of compound binding (orthosteric or allosteric) is not available. Sometimes, higher concentration of substrate may be needed to sustain steady state condition or to weaken the competitive compound’s apparent potency to avoid tight binding condition and allow reliable potency determination. Other times, lower substrate concentration, at the risk of shorter lived steady state turnover kinetics, may be used to weaken the uncompetitive compound’s apparent potency.
Compound concentrations should bracket Ki and, for 2-step binders, also Ki* to allow reliable determination of binding affinity. For slow binding compounds exhibiting koff < 10-3 s-1, compound concentrations will have to be substantially higher than Ki and Ki* to observe within a convenient time window the exponential curvature of time-dependent progress curves from initial to final steady state. As a result, the optimal concentrations of compound required for reliable determination of kinetic constants may not overlap with those required for reliable determination of binding affinity. To allow simultaneous determinations of both affinity and kinetics by GPCA, the compound concentrations may have to vary over a wide range. In addition, the time course must be adequately long to allow observable curvature from initial to final steady state for at least two compound doses. It is important to use an optimal combination of enzyme and substrate concentrations such that a sustained steady state in the absence of compound will be maintained.
The data analysis by global curve fitting of both time-and dose-dependent enzyme turnover data yield both affinity and koff under optimal conditions for 1-step binders (e.g. Figure 2C). The kon value can then be calculated. When koff of a 1-step binder is too small (< 10-5 s-1), only the association rate constant kon (= koff/Ki for inhibitors or koff/Kd for activators) may be determined, unless with longer monitoring. When koff of a 1-step binder is too large (> 10-2 s-1), only Ki or Kd may be determined, unless with much faster monitoring (such as stopped flow). For 2-step binders, when the slow backward isomerization is too slow (k6 < 10-5 s-1), only Ki or Kd for the first and fast step may be determined, unless with longer monitoring. When k6 is too large (> 10-2 s-1), only Ki* or Kd* for the overall affinity may be determined, unless with much faster monitoring. We routinely use commercial software platforms such as GraphPad Prism and KinTek Explorer (79) for global curve fitting analysis.
3.4. Radioligand binding assay
Radioligand binding remains a preferred method for investigating the interaction of ligands with their target protein (for reviews see (80-82)). It is an extremely versatile technique that does not require costly instrumentation and can be performed as an automated assay with moderate or high throughput. The main advantage of this technique is that it can be applied to the study of not only soluble and purified proteins but also to membrane preparations, extracts or whole cells. While this section discusses radioligand-based binding assays, the general principle universally applies to all ligand binding assays with or without radioactive isotope or fluorescent label. For example, a kinetic probe competition assay (kPCA) based on time-resolved fluorescence resonance energy transfer (TR-FRET) detection has been developed for high throughput binding kinetics (40).
The tracer used in radioligand binding studies should allow accurate detection of low levels of bound ligand (Table 5). Small molecule ligands are most commonly labeled with 3H and peptide ligands with 125I, although other radioisotopes (14C, 33P and 35S) are also employed. Tritiated ligands generally behave identically to their unlabeled counterparts, have long radiochemical half-lives (12.5 years) and specific activities in the range of 20 to 80 Ci/mmol (depending on the number of tritium atoms incorporated in the molecule). These properties allow sufficient signal to study binding interactions having nanomolar affinities. Iodinated ligands, on the other hand, have short radiochemical half-lives (60 days) and specific activities of around 2000 Ci/mmol. Caution needs to be taken regarding whether the incorporation of 125I results in altered binding properties vs. non-iodinated analogs, unless the radioligand is used primarily in competition assay to determine binding kinetics for unlabeled compounds (see later). Due to its high specific activity, low receptor densities and ligand binding interactions having picomolar affinities can be experimentally detected with 125I, 33P and 35S labeling (83, 84).
|Radiolabel||3H, 14C, 33P, 35S, 125I, high specific activity, sufficient half life|
|Radioligand concentration||Adequate protein binding and signal|
|Radioligand binding affinity||Near protein concentration|
|Nonspecific binding ligand||Cold ligand chemically different from radioligand to minimize isotope dilution effect on the specific binding site|
|Protein concentration||As low as assay detection limit permits to avoid radioligand depletion (“tightbinding” condition)|
|Sampling frequency||Adequate to cover >90% curvature (or4 halftimes)|
|Dissociation after wash out or jump dilution||Add cold ligand to minimize rebinding|
Radioligand binding studies require parallel incubations of radioligand with receptor alone (for total binding) and in the presence of sufficient unlabeled ligand (for nonspecific binding) to occupy all the specific (but not nonspecific) binding sites of available receptors. If possible, it is best practice to use a displacing agent that is structurally different from the radioligand (thus competing for only the specific binding sites) and not the cold version of the radioligand that not only eliminates specific binding but probably reduces nonspecific binding due to isotopic dilution. The non-specific binding accounts for binding of the ligand with a much lower affinity to other components of the receptor-containing preparation and/or binding to materials used in the experiment. Thus, specific binding, or binding to the receptor, is defined as the binding to the target that it is displaceable by a large excess (usually 100 times its Kd) of cold ligand.
Another important point to consider in setting up a radioligand binding assay is the choice of technique (vacuum filtration, dialysis, centrifugation, gel filtration or precipitation) for adequate separation of bound and free ligand. Vacuum filtration is the most widely used with membrane preparations and whole cells (85). The bound ligand is retained on glass-fiber filters and the free ligand passes through. Following initial filtration, the filters are usually rinsed with cold washing buffer to reduce non-specific binding to the filter and to improve the signal to noise ratio. This method is very reproducible and quick. However, the removal of free ligand necessarily promotes dissociation of the ligand from the receptor, thus, this technique is not suitable for ligands with fast off-rates. Gel filtration method is one of the best choices for soluble and purified receptors and exhibits excellent recovery of protein-ligand complexes and also less opportunity for dissociation (86). In addition to this, there are homogeneous radioligand binding methods that do not involve washing steps or removal of free radioligand. One of the most often used is the Scintillation Proximity Assay (SPA). It is a mix and read bead-based assay, can be adapted to soluble targets as well as membrane-bound receptors and has a high throughput. Moreover, association and dissociation experiments can also be performed, and it is an excellent technology to determine fast on and off rates by cycle counting that may be difficult to obtain by other methods (87, 88).
In kinetic experiments, the binding of one or more concentrations of radioligand is monitored at different time points and analyzed to estimate association (kon) and dissociation (koff) rate constants. The simplest scenario is the determination of kon and koff of the radiolabeled compound (89, 90). The standard analyses of ligand binding experiments assume that the concentration of free radioligand is constant during the experiment, which means that most of the ligand remains free and only a small fraction (less than 10%) is bound even at equilibrium (pseudo-first order conditions). The first best practice is to use a low concentration of receptor (<< Kd), as low as permitted but still giving reliable detection, to avoid “tight binding” condition under which significant radioligand depletion would cause non-pseudo first order kinetics. Then one should design a concentration-and time-dependent matrix by using increasing concentrations of the radioligand (thus, achieving varying levels of receptor occupancy) and monitoring each binding reaction over time until >90% equilibrium is achieved. If this approach is adhered to, a traditional re-plot of the observed apparent rate constants (kobs) toward achieving equilibrium (obtained by simple mono-exponential curve fitting to each kinetic curve, see Figure 2D, green curve) as a function of total (non-depleted) ligand concentrations should yield a straight line with a slope equal to kon and a Y-intercept equal to koff. More simply and robustly, a global curve fitting approach (similar in spirit to GPCA method in section 3.3) is highly recommended for simultaneous analysis of all concentration-and time-dependent curves to derive kon and koff directly. It is also possible to determine kon in an association experiment when only a single concentration of radioligand is used but it requires an independent determination of the maximum number of binding sites, Bmax, present in the experiment. In any case, if the pseudo-first order conditions are incorrectly assumed, the value of kon will be underestimated. The direct determination of the dissociation constant, koff, is less complicated. Once ligand and receptor binding have achieved equilibrium, the dissociation of the complex can be initiated by jump dilution into a solution containing a large excess of cold competing ligand (see section 3.5 for more details). The jump dilution and the excess competing ligand should be able to minimize rebinding of the dissociated labeled ligand. A failure to completely dissociate the ligand-receptor complex within a finite period of observation indicates the possibility of slow reversibility (slower than the observation time window), irreversibility, or allosteric binding. There should be good agreement of the Kd calculated in kinetic experiments with the Kd obtained in saturation equilibrium experiments if there is only one binding site, the binding is fully reversible and there are no tight binding or gross experimental errors. To alleviate tight binding for super high affinity compounds, one may consider a “double competition” setup where both a hot radioligand and a cold, structurally identical or different, competitor ligand of known affinity are used at proper concentrations to weaken the apparent binding affinities of compounds to be near or preferably well above the protein concentration while maintaining good radioligand binding signals.
Kinetic rate constants of unlabeled ligands can also be determined by radioligand binding competition, but they require prior determination of the association and dissociation rate constants for preferably an orthosterically competitive radioligand. A mathematical model was described (91) to determine the kinetics of binding of competitive ligands based on the experimental data of the binding kinetics of a dose-fixed radioligand in the presence of simultaneously added, dose-varied unlabeled ligand. In these studies, the presence of the unlabeled compound will alter the time course of the binding of the radioligand. If the competitor dissociates faster than the radioligand, the association of the radioligand will slowly and monotonically approach its equilibrium in time (see Figure 2D, red curve). If the competitor dissociates slower than the radioligand, the association curve will be biphasic and more radioligand will be bound at some intermediate time points than at equilibrium (see Figure 2D, blue curve).The two-step competition binding approach, or delayed association method, is another indirect method to estimate off-rates of unlabeled competitors (92). In this case, the receptor is preincubated with the competitive unlabeled ligand, the samples are washed off of free ligand, and subsequently the time course of the association of the radioligand is monitored. Radioligand binding can also be coupled to downstream functional readout to measure binding kinetics and drug-target residence time (93).
Competition binding kinetics may also be used to discriminate between allosteric and orthosteric modulators. In general, it is not always possible to determine if an inhibitory effect on radioligand association kinetics is allosteric or orthosteric except when an alteration of the dissociation rate of the radioligand (without the complication of rebinding) would be indicative of allosteric modulation (94). Similarly, a characteristic change of association rate constants in the presence of increasing levels of orthosteric or allosteric modulators is helpful to discern the mode of binding, since an orthosteric modulator will perpetually decrease the apparent kon value while an allosteric modulator will either not change or asymptotically decrease or increase the apparent kon values. As a result of the above changes to apparent kon and koff values, the apparent binding affinity changes of radioligand due to the presence of another allosteric or orthosteric ligand will also exhibit distinct patterns. Thus, an orthosteric competitor will perpetually weaken the radioligand affinity, while an allosteric modulator will either not change (by noncompetitor), or asymptotically increase (by uncompetitor) or asymptotically decrease (by allosteric competitor) the radioligand affinity. A monovalent ligand that orthosterically but partially competes with one of the two sites of a heterobivalent ligand may be mistaken as an allosteric modulator (95).
3.5. Jump dilution
To determine koff of a very slow dissociating compound from its target, jump dilution followed by often discontinuous monitoring over the course of hours to days is the preferred method of choice. Jump dilution relies on the physical chemical principle that pre-established equilibrium, upon massive dilution (usually > 100-fold), will re-equilibrate to a new state. Thus, when a preformed compound/target complex is diluted massively into a solution to minimize re-binding between them, the compound will exhibit a first order dissociation kinetics (Figure 2E). There are several issues to pay attention to for jump dilution (Table 6).
|Preincubation stoichiometry, concentrations, time||Nearly stoichiometric, up to~mM level, >30-min|
|Fold of jump dilution||>100-fold|
|Post-dilution condition||Apparent Kd or Ki >>(compound) ≈(enzyme) > enzyme’s low limit of detection|
|Minimizing rebinding||Use substrate near Km or well above Km (if competitive) or below Km (if uncompetitive); useexcess cold ligand|
|Monitoring time||Adequate to cover >90% dissociation (or four halftimes)|
|Monitoring method||Radioactivity, enzyme activity, MW of free ligand|
|Stability||Protein-ligand complex must be sufficiently stable over the course of monitoring|
The preformed complex between compound and its target protein should preferably be initiated with stoichiometric levels that reflect the true binding stoichiometry (1:1, 2:1, 1:2, etc). Preincubation using lopsided, non-stoichiometric levels of enzyme and tight binding inhibitor may suffer from diminished window of detection upon jump dilution (44). The concentrations during preincubation should be high enough (up to μM) to enable rapid equilibrium of binding between the compound and its target. The preincubation time should be adequately long to assure equilibrium binding. As a general rule of thumb for most compounds, preincubation at > 1 μM for > 30-min should be used. For very slow binding compounds, higher concentrations and longer time may be needed to achieve equilibrium binding. For 2-step binding inhibitors whose initial rapid binding is followed by a slow noncovalent isomerization or covalent modification, pre-incubation time is best varied for 30-min and many hours (up to overnight) to ascertain that reversible equilibrium or irreversible inactivation has been achieved.
The fold of jump dilution should be large enough to minimize compound rebinding, in conjunction with additional measures. It is best to dilute so massively that the final residual compound concentration is well below the apparent Kd or Ki and the final protein target concentration is still above low limit of detection (if enzyme turnover is used to monitor inhibitor/enzyme dissociation). For high affinity compounds (Kd or Ki < 1 nM), it may not be practical to dilute the compound concentration below its affinity constant and may require additional measures to decrease the apparent affinity and minimize rebinding. For enzymatic monitoring following jump dilution, high concentrations (>> Km) of competitive substrate or low concentrations (<< Km) of uncompetitive substrate can be used to decrease the apparent compound binding affinity so that the final residual compound is well below the apparent binding affinity. For non-enzymatic monitoring following jump dilution, it is best to use excess cold ligand (for radioligand monitoring) or orthosterically competitive ligand (for non-radioactive monitoring such as mass spectrometry). The monitoring time should preferably cover at least four halftimes unless the dissociation is extremely slow and the protein target stability is not sufficient to allow longer monitoring. For very slowly dissociating enzyme inhibitors, it is important to include a control reaction spiked with the same residual final compound concentration as achieved after jump dilution. This control reaction provides an accurate measure of the final steady state velocity after full enzyme recovery from inhibition post jump dilution so that the exponential fitting will have the proper constraint to the final state.
For protein target that associates to form dimer, trimer or higher order oligomer, jump dilution may be complicated if there is rapid, significant dissociation of the protein target oligomer. The final protein concentration needed to maintain the oligomeric state may create a very tight binding condition for a high affinity ligand such that jump dilution experiment becomes impractical. In this case, alternative methods of binding kinetics will have to be considered.
Traditionally, quantitation of bound and/or free ligand in binding studies was performed by radiolabeling the ligand of interest to a known specific activity with one of several biomolecule-compatible radioisotopes including 3H, 14C, 35S, 33P or 125I. Radiolabeled ligands, in combination with modern radiochemical detection methods, allow for ligand detection in the pM to fM range (for details, see Section 3.4). While extremely valuable, radiochemical binding assays are limited by the need to successfully prepare a suitable radioligand for each compound of interest, the necessity of working with hazardous radioactive material and the expense associated with the resultant waste. The magnitude of these issues increases with the number of ligands of interest. Fluorescently labeled ligands have been used as an alternative, but the dye substituent is often bulky and hydrophobic, which can alter ligand binding properties (see for example (96)), unless such labeled ligands are used primarily in competition binding assays to examine the binding of unlabeled compounds.
A method in which bound and/or free ligand could be quantitated with comparable sensitivity in a label-free manner would be clearly advantageous. Analyte quantitation using liquid chromatography coupled tandem mass spectrometry (LC/MS/MS) on a triple quadrupole mass spectrometer (QqQ-MS) solves this problem. This section reviews this universal, label-free detection method and its applications to binding kinetics in drug discovery. In retrospect, this is a clear extension of the widespread use of LC/MS/MS to study the ADME properties of drug candidates (see for example (97)). LC/MS/MS offers unparalleled specificity, sensitivity and speed for analyte determination in complex mixtures. The key to this method is the combination of a high resolution liquid chromatography system (generally operated using reverse phase conditions) coupled in line to a QqQ-MS. The system, when used in selected (or multiple) reaction monitoring mode, allows the specific ligand of interest (and an internal standard) to be identified and quantitated with extremely high specificity and sensitivity using a unique combination of three parameters: (1) LC retention time; (2) Parent ion m/z peak (selected in quadrupole 1) and (3) A major fragment ion m/z (selected in quadrupole 3). What follows is a brief review of publications describing the application of LC/MS/MS to evaluate the binding of specific ligands to their receptor targets. A discussion of high throughput screening applications of MS to identify and characterize (unknown) ligands obtained by affinity selection methods is beyond the scope of this discussion (98), as is the characterization of non-covalent receptor/ligand complexes by electrospray ionization MS (99) and the determination of in vivo receptor occupancy by LC/MS/MS (100).
Application of LC/MS/MS analysis to ligand binding studies was first reported in 2003 (101). In this report, LC/MS/MS was employed to develop a competition binding assay for membrane bound dopamine D1 receptor utilizing the SCH 23390 ligand at ~ Kd levels (0.5 nM) in combination with LC/MS/MS compatible buffers. The increase in free SCH 23390 as a consequence of increasing competitor concentration was monitored by LC/MS/MS following removal of the receptor/ligand complex by ultracentrifugation; LC/MS/MS and radiochemical competition assay derived Ki values were in excellent agreement. This early protocol was improved in subsequent papers from the Wanner laboratory. One study used the dopamine D2 receptor and spiperone as the probe ligand and employed solid phase extraction to remove non-volatile binding buffer components and reduce matrix effects; haloperidol was incorporated as an internal standard to compensate for process variability and non-quantitative analyte recovery (102). Asubsequent study, using the membrane bound GABA transporter mGAT1 and its high affinity ligand NO 711, described the LC/MS/MS-based binding protocol used most widely today, which incorporates the following modifications: (1) Per-deuterated NO 711 was used as a heavy atom labeled internal standard; (2) 96-well glass fiber filter plates in combination with a vacuum manifold were used to isolate transporter bound from free ligand; (3) Wash steps were employed to eliminate all free ligand and interfering buffer components from the protein/ligand complex entrapped on the filter; and (4) Bound ligand was liberated from its receptor by drying the plate, followed by protein denaturation and ligand elution with an organic solvent (typically methanol) at concentrations consistent with direct injection onto the LC/MS/MS (103). In this seminal publication, the authors first demonstrated the application of MS binding technology to measure kon and koff values, Kd and Bmax, and also developed and utilized a competition MS binding assay to define ligand structure activity relationships (SAR). NO711 binding to mGAT1 under more physiological conditions was also described, demonstrating the applicability of the improved LC/MS/MS binding protocol to lower affinity (Kd = 150 nM) ligands (104). The Wanner laboratory also focused on increasing the throughput of MS based binding experiments by utilizing MALDI/MS/MS (105) and by employing shorter LC columns/run times during LC/MS/MS analysis (106). Additional studies exploited increased incubation volumes to lower target concentration while maintaining sufficient bound ligand for LC/MS/MS detection; this maneuver can be used to measure high affinity interactions while avoiding the “tight binding” conditions (107). This paper is also notable in that it describes the use of a single ligand to probe three distinct monoamine transporters. An alternative approach to multiplexed assays employing LC/MS/MS analysis (several ligands/receptors in a single incubation) has also been described (108).
The application of LC/MS/MS methods to study ligand binding to soluble proteins is less well described. At least three publications describe the development and utilization of competition binding assays in which unbound ligand was quantitated following removal of the enzyme/ligand complexes by ultrafiltration (109-111). In our laboratory, we generally employ an alternative method in which the protein/ligand complex is isolated from free ligand using 96-well plates containing size exclusion chromatography media. The protein/ligand complex in the eluate is then dissociated by addition of organic solvent and analyzed by LC/MS/MS. We have used this approach to measure kon and koff values, Kd and Bmax, and also for competition MS binding assay to define SAR for soluble receptor orthosteric and allosteric ligands. An alternative is to precipitate the protein/ligand complex (using PEG or ammonium sulfate) and isolate/wash the complex using 96-well glass fiber filter plates, similar to the popular method used for membrane-associated proteins (103).
Though highlighted here mostly for its application to equilibrium binding studies, LC/MS/MS has broad utility as a detection method for experiments designed to explore binding kinetics. Table 7 summarizes some best practices that should be incorporated into any experimental design employing this technique. First and foremost, the inclusion of an appropriate internal standard as early as possible in the sample preparation process is indispensable as a means to correct for sample to sample variation; it also serves as a useful tool during troubleshooting exercises. Where available, stable isotope labeled (2H, 13C or 15N) internal standards are preferred, but in their absence a suitable chemical analogue of the ligand being studied is often sufficient. If samples are to be prepared and stored prior to analysis or if the analytical runtime is long, it is wise to investigate sample stability under autosampler conditions. This is easily done by preparing a small number of replicate samples at relevant concentrations along with a suitable standard curve; these samples are analyzed at time zero and then maintained in the autosampler and re-injected along with a fresh standard curve at a subsequent time point to bracket the expected timeframe for storage or analysis. Acceptable agreement in the measured concentrations will establish post-processing stability. Samples of high complexity such as crude cell lysates and membrane preparations washed with detergents will contain matrix elements that can build up over time and contribute to deteriorating performance of the mass spectrometer. Simple strategies to mitigate the impact of this issue include the aforementioned internal standard, additional sample clean up prior to LC/MS/MS analysis (via solid phase or liquid-liquid extraction to isolate the analyte from undesired matrix elements) and the use of flow-switching fluidics to direct only the portion of the chromatogram where the analyte is known to elute to the mass spectrometer and to divert the remainder of the LC cycle to waste. Finally, it is inevitable that the need will arise to push the limit of detection as low as possible. In these cases, rigorous tuning of conditions affecting analyte ionization and fragmentation by an experienced mass spectrometrist will be helpful in achieving the optimal signal to noise ratio. This, coupled with advancements in mass spectrometers and the application of micro-and nanoflow LC, has made sub-nanomolar limits of detection increasingly common.
|Quantitation||Isotope-dilution using stable-labeled internal standard (where available); external standard curve prepared in sample matrix|
|Sample processing/instrumental variation||Include an appropriate stable-labeled or chemical analog internal standard|
|Post-processing sample stability||Maintain samples in a temperature-controlled autosampler; evaluate and document stability over expected analysis time|
|Ruggedness||Consider clean-up strategies for complex sample matrices (liquid-liquid or solid-phase extraction); make use of integrated instrument fluidics to divert LC flow to waste during void volume elution and column cleaning/regeneration portions of gradient|
|Limit of detection||Optimize ionization mode (negative vs positive ESI; APCI and APPI); low-flow enhances electrospray ionization|
|ESI: Electrospray ionization, APCI: Atmospheric pressure chemical ionization, APPI: Atmospheric pressure photoionization|
In summary, the application of LC/MS/MS methodology provides an analytical method that enables the analysis of both equilibrium and kinetic binding parameters of virtually any ligand in a label-free manner to both membrane bound and soluble proteins with comparable sensitivity and specificity, but arguably increased experimental flexibility, versus radioligand or kPCA based binding approaches.
4. Future directions
It has taken the scientific community quite some time to grapple with the necessity of understanding binding kinetics. Recognition of the importance of drug-target residence time, a celebrated component of binding kinetics, is now entering its second decade. With the enabling technologies for measuring in vitro binding kinetics covering timeframes over nine orders of magnitude as reviewed here, coupled with additional in vitro and in vivo kinetic profiling reported in literature, we anticipate a more intentional and proper use of these technologies by scientists in advancing their quantitative and translational pharmacology studies. One fundamental gap to achieving this goal is the lack of awareness regarding the pervasive role binding kinetics plays from everyday activity and binding assays to ultimate drug efficacy and safety. It has been demonstrated that residence time within the same series of chemical or biological molecules can be tunable, resulting in differentiated impact on efficacy and safety (112, 113). We encourage pharmacologists and the modeling and simulation community to actively incorporate binding kinetics into mechanistic and translational PKPD models (114, 115) and dynamically interact with experimentalists to fine tune their models. Modeling can also reciprocally benefit experimentalists by providing early identification of the critical binding network nodes and relevant kinetic parameters that will be essential to predict drug efficacy and safety. We anticipate that drug discovery will be increasingly grounded in quantitative systems biology and pharmacology and, as a result, yield a more bountiful medicine chest that efficaciously and safely tailors to the various medical needs of humanity.
We gratefully acknowledge the managerial support from Drs. Maria Webb, George Addona and John Hunter.
Abbreviations: Binding kinetics (BK), association rate constant (kon), dissociation rate constant (koff), mechanism of action (MOA), drug-target residence time, pharmacokinetics (PK), pharmacodynamics (PD), stopped flow, surface plasmon resonance (SPR), global progress curve analysis (GPCA), radioligand binding, Scintillation Proximity Assay (SPA), time-resolved fluorescence resonance energy transfer (TR-FRET) based kinetic probe competition assay (kPCA), jump dilution, liquid chromatography coupled tandem mass spectrometry (LC/MS/MS), drug discovery.
21. Randhawa, V., P. Sharma, S. Bhushan & G. Bagler: Identification of key nodes of type 2 diabetes mellitus protein interactome and study of their interactions with phloridzin. Omics, 17, 302-17 (2013)
22. Wong, C. F. & S. Bairy: Drug design for protein kinases and phosphatases: flexible-receptor docking, binding affinity and specificity, and drug-binding kinetics. Curr Pharm Des, 19, 4739-54 (2013)
28. Yanase, Y., T. Hiragun, T. Yanase, T. Kawaguchi, K. Ishii & M. Hide: Application of SPR imaging sensor for detection of individual living cell reactions and clinical diagnosis of type I allergy. Allergol Int, 62, 163-9 (2013)
29. Concepcion, J., K. Witte, C. Wartchow, S. Choo, D. Yao, H. Persson, J. Wei, P. Li, B. Heidecker, W. Ma, R. Varma, L. S. Zhao, D. Perillat, G. Carricato, M. Recknor, K. Du, H. Ho, T. Ellis, J. Gamez, M. Howes, J. Phi-Wilson, S. Lockard, R. Zuk & H. Tan: Label-free detection of biomolecular interactions using BioLayer interferometry for kinetic characterization. Comb Chem High Throughput Screen, 12, 791-800 (2009)
31. Haddad, G. L., S. C. Young, N. D. Heindel, D. J. Bornhop & R. A. Flowers, 2nd: Back-scattering interferometry: an ultrasensitive method for the unperturbed detection of acetylcholinesterase-inhibitor interactions. Angew Chem Int Ed Engl, 51, 11126-30 (2012)
32. Jepsen, S. T., T. M. Jorgensen, W. Zong, T. Trydal, S. R. Kristensen & H. S. Sorensen: Evaluation of back scatter interferometry, a method for detecting protein binding in solution. Analyst, 140, 895-901 (2015)
33. Daghestani, H. N. & B. W. Day: Theory and applications of surface plasmon resonance, resonant mirror, resonant waveguide grating, and dual polarization interferometry biosensors. Sensors (Basel), 10, 9630-46 (2010)
34. Liang, W., S. Wang, F. Festa, P. Wiktor, W. Wang, M. Magee, J. LaBaer & N. Tao: Measurement of small molecule binding kinetics on a protein microarray by plasmonic-based electrochemical impedance imaging. Anal Chem, 86, 9860-5 (2014)
35. Knezevic, J., A. Langer, P. A. Hampel, W. Kaiser, R. Strasser & U. Rant: Quantitation of affinity, avidity, and binding kinetics of protein analytes with a dynamically switchable biosurface. J Am Chem Soc, 134, 15225-8 (2012)
36. Langer, A., P. A. Hampel, W. Kaiser, J. Knezevic, T. Welte, V. Villa, M. Maruyama, M. Svejda, S. Jahner, F. Fischer, R. Strasser & U. Rant: Protein analysis by time-resolved measurements with an electro-switchable DNA chip. Nat Commun, 4, 2099 (2013)
37. Langer, A., W. Kaiser, M. Svejda, P. Schwertler & U. Rant: Molecular dynamics of DNA-protein conjugates on electrified surfaces: solutions to the drift-diffusion equation. J Phys Chem B, 118, 597-607 (2014)
38. Langer, A., M. Schraml, R. Strasser, H. Daub, T. Myers, D. Heindl & U. Rant: Polymerase/DNA interactions and enzymatic activity: multi-parameter analysis with electro-switchable biosurfaces. Sci Rep, 5, 12066 (2015)
42. Bao, J., S. M. Krylova, L. T. Cherney, J. C. Le Blanc, P. Pribil, P. E. Johnson, D. J. Wilson & S. N. Krylov: Pre-equilibration kinetic size-exclusion chromatography with mass spectrometry detection (peKSEC-MS) for label-free solution-based kinetic analysis of protein-small molecule interactions. Analyst, 140, 990-4 (2015)
43. Bao, J., S. M. Krylova, L. T. Cherney, J. C. LeBlanc, P. Pribil, P. E. Johnson, D. J. Wilson & S. N. Krylov: Kinetic size-exclusion chromatography with mass spectrometry detection: an approach for solution-based label-free kinetic analysis of protein-small molecule interactions. Anal Chem, 86, 10016-20 (2014)
44. Copeland, R. A., A. Basavapathruni, M. Moyer & M. P. Scott: Impact of enzyme concentration and residence time on apparent activity recovery in jump dilution analysis. Anal Biochem, 416, 206-10 (2011)
48. Burzomato, V., M. Beato, P. J. Groot-Kormelink, D. Colquhoun & L. G. Sivilotti: Single-channel behavior of heteromeric alpha1beta glycine receptors: an attempt to detect a conformational change before the channel opens. J Neurosci, 24, 10924-40 (2004)
51. Qian, H. & L. M. Bishop: The chemical master equation approach to nonequilibrium steady-state of open biochemical systems: linear single-molecule enzyme kinetics and nonlinear biochemical reaction networks. Int J Mol Sci, 11, 3472-500 (2010)
58. Shidahara, M., Y. Ikoma, J. Kershaw, Y. Kimura, M. Naganawa & H. Watabe: PET kinetic analysis: wavelet denoising of dynamic PET data with application to parametric imaging. Ann Nucl Med, 21, 379-86 (2007)
65. Martin, S. R. & M. J. Schilstra: Rapid Mixing Kinetic Techniques. In: Protein-Ligand Interactions: Methods and Applications. 2nd ed. Eds: M. A. Williams&T. Daviter. Humana Press, London, UK (2013)
66. Wang, R.-Y.: Rapid Scan, Stopped-Flow Kinetics. In: Applications of Physical Methods to Inorganic and Bioorganic Chemistry. Eds: R. A. Scott&C. M. Lukehart. John Wiley & Sons Ltd, Hoboken, NJ (2007)
68. Hoeltzli, S. D. & C. Frieden: Refolding of (6-19F)tryptophan-labeled Escherichia coli dihydrofolate reductase in the presence of ligand: a stopped-flow NMR spectroscopy study. Biochemistry, 37, 387-98 (1998)
69. Sienkiewicz, A., A. M. da Costa Ferreira, B. Danner & C. P. Scholes: Dielectric resonator-based flow and stopped-flow EPR with rapid field scanning: Amethodology for increasing kinetic information. J Magn Reson, 136, 137-42 (1999)
71. Hubert, J. F., L. Duchesne, C. Delamarche, A. Vaysse, H. Gueune & C. Raguenes-Nicol: Pore selectivity analysis of an aquaglyceroporin by stopped-flow spectrophotometry on bacterial cell suspensions. Biol Cell, 97, 675-86 (2005)
77. Schuck, P. & H. Zhao: The role of mass transport limitation and surface heterogeneity in the biophysical characterization of macromolecular binding processes by SPR biosensing. Methods Mol Biol, 627, 15-54 (2010)
83. Garcia-Calvo, M., H. G. Knaus, O. B. McManus, K. M. Giangiacomo, G. J. Kaczorowski & M. L. Garcia: Purification and reconstitution of the high-conductance, calcium-activated potassium channel from tracheal smooth muscle. J Biol Chem, 269, 676-82 (1994)
84. Garcia-Calvo, M., J. Lisnock, H. G. Bull, B. E. Hawes, D. A. Burnett, M. P. Braun, J. H. Crona, H. R. Davis, Jr., D. C. Dean, P. A. Detmers, M. P. Graziano, M. Hughes, D. E. Macintyre, A. Ogawa, A. O’Neill K, S. P. Iyer, D. E. Shevell, M. M. Smith, Y. S. Tang, A. M. Makarewicz, F. Ujjainwalla, S. W. Altmann, K. T. Chapman & N. A. Thornberry: The target of ezetimibe is Niemann-Pick C1-Like 1 (NPC1L1). Proc Natl Acad Sci U S A, 102, 8132-7 (2005)
85. Shrikhande, A., C. Courtney, D. Smith, M. Melch, M. McConkey, J. Bergeron & S. K. Wong: Fully automated radioligand binding filtration assay for membrane-bound receptors. Biotechniques, 33, 932-7 (2002)
87. Berry, J., M. Price-Jones & B. Killian: Use of scintillation proximity assay to measure radioligand binding to immobilized receptors without separation of bound from free ligand. Methods Mol Biol, 897, 79-94 (2012)
90. Sykes, D. A., M. R. Dowling & S. J. Charlton: Measuring receptor target coverage: a radioligand competition binding protocol for assessing the association and dissociation rates of unlabeled compounds. Curr Protoc Pharmacol, Chapter9, Unit914 (2010)
92. Packeu, A., M. Wennerberg, A. Balendran & G. Vauquelin: Estimation of the dissociation rate of unlabelled ligand-receptor complexes by a ‘two-step’ competition binding approach. Br J Pharmacol, 161, 1311-28 (2010)
93. Vauquelin, G.: Determination of drug–receptor residence times by radioligand binding and functional assays: experimental strategies and physiological relevance. Med. Chem. Commun., 3, 645-651 (2012)
94. Vauquelin, G. & I. Van Liefde: Radioligand dissociation measurements: potential interference of rebinding and allosteric mechanisms and physiological relevance of the biological model systems. Expert Opin Drug Discov, 7, 583-95 (2012)
95. Vauquelin, G., D. Hall & S. J. Charlton: ‘Partial’ competition of heterobivalent ligand binding may be mistaken for allosteric interactions: a comparison of different target interaction models. Br J Pharmacol, 172, 2300-15 (2015)
97. Zhang, J., W. Z. Shou, M. Vath, K. Kieltyka, J. Maloney, L. Elvebak, J. Stewart, J. Herbst & H. N. Weller: An integrated bioanalytical platform for supporting high-throughput serum protein binding screening. Rapid Commun Mass Spectrom, 24, 3593-601 (2010)
99. Liu, Y. H., L. Ramanathan, B. Malcolm, G. Njoroge, T. Y. Chan & B. N. Pramanik: Screening and rank ordering of reversible mechanism-based inhibitors of hepatitis C virus NS3 protease using electrospray ionization mass spectrometry. J Mass Spectrom, 46, 764-71 (2011)
100. Need, A. B., J. H. McKinzie, C. H. Mitch, M. A. Statnick & L. A. Phebus: In vivo rat brain opioid receptor binding of LY255582 assessed with a novel method using LC/MS/MS and the administration of three tracers simultaneously. Life Sci, 81, 1389-96 (2007)
103. Zepperitz, C., G. Hofner & K. T. Wanner: MS-binding assays: kinetic, saturation, and competitive experiments based on quantitation of bound marker as exemplified by the GABA transporter mGAT1. ChemMedChem, 1, 208-17 (2006)
107. Grimm, S. H., G. Hofner & K. T. Wanner: MS Binding Assays for the Three Monoamine Transporters Using the Triple Reuptake Inhibitor (1R, 3S)-Indatraline as Native Marker. Chem Med Chem, 10, 1027-39 (2015)
108. de Jong, L. A., C. M. Jeronimus-Stratingh & T. I. Cremers: Development of a multiplex non-radioactive receptor assay: the benzodiazepine receptor, the serotonin transporter and the beta-adrenergic receptor. Rapid Commun Mass Spectrom, 21, 567-72 (2007)
109. Li, C. M., Y. Lu, S. Ahn, R. Narayanan, D. D. Miller & J. T. Dalton: Competitive mass spectrometry binding assay for characterization of three binding sites of tubulin. J Mass Spectrom, 45, 1160-6 (2010)
110. Mulabagal, V. & A. I. Calderon: Development of an ultrafiltration-liquid chromatography/mass spectrometry (UF-LC/MS) based ligand-binding assay and an LC/MS based functional assay for Mycobacterium tuberculosis shikimate kinase. Anal Chem, 82, 3616-21 (2010)
111. Mulabagal, V. & A. I. Calderon: Development of binding assays to screen ligands for Plasmodium falciparum thioredoxin and glutathione reductases by ultrafiltration and liquid chromatography/mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci, 878, 987-93 (2010)
112. Bradshaw, J. M., J. M. McFarland, V. O. Paavilainen, A. Bisconte, D. Tam, V. T. Phan, S. Romanov, D. Finkle, J. Shu, V. Patel, T. Ton, X. Li, D. G. Loughhead, P. A. Nunn, D. E. Karr, M. E. Gerritsen, J. O. Funk, T. D. Owens, E. Verner, K. A. Brameld, R. J. Hill, D. M. Goldstein & J. Taunton: Prolonged and tunable residence time using reversible covalent kinase inhibitors. Nat Chem Biol, 11, 525-31 (2015)
113. Louvel, J., D. Guo, M. Agliardi, T. A. Mocking, R. Kars, T. P. Pham, L. Xia, H. de Vries, J. Brussee, L. H. Heitman & A. P. Ijzerman: Agonists for the adenosine A1 receptor with tunable residence time. ACase for nonribose 4-amino-6-aryl-5-cyano-2-thiopyrimidines. J Med Chem, 57, 3213-22 (2014)
114. Ramsey, S. J., N. J. Attkins, R. Fish & P. H. van der Graaf: Quantitative pharmacological analysis of antagonist binding kinetics at CRF1 receptors in vitro and in vivo. Br J Pharmacol, 164, 992-1007 (2011)
115. Walkup, G. K., Z. You, P. L. Ross, E. K. Allen, F. Daryaee, M. R. Hale, J. O’Donnell, D. E. Ehmann, V. J. Schuck, E. T. Buurman, A. L. Choy, L. Hajec, K. Murphy-Benenato, V. Marone, S. A. Patey, L. A. Grosser, M. Johnstone, S. G. Walker, P. J. Tonge & S. L. Fisher: Translating slow-binding inhibition kinetics into cellular and in vivo effects. Nat Chem Biol, 11, 416-23 (2015)