Chapter 1

Introduction to Surface Plasmon Resonance

Richard B. M. Schasfoorta
a Medical Cell BioPhysics Group (MCBP), MIRA Institute, University of Twente, The Netherlands. E-mail: r.b.m.schasfoort@utwente.nl


This tutorial chapter is written for students, laboratory technicians, and scientists who are new to the field of applying surface plasmon resonance (SPR) technology to biomolecular interaction analysis. It describes the basis of how to apply SPR technology for measuring biomolecular interactions label free and in real time. Not only are the general aspects of constructing an SPR assay outlined but also the general kinetic and affinity aspects of biomolecular interactions are explained briefly.


1.1 Introduction to Surface Plasmon Resonance

During the years following the introduction of the first commercial surface plasmon resonance (SPR) instrument (Biacore) in 1990, the number of publications that include data collected from commercial biosensors increased to more than 20 000 papers by 2016 (PubMed data), as shown in Figure 1.1.

Fig. 1.1 SPR papers in PubMed: total >20 000. 1990–1995: 217 articles; 1995–2000: 1190 articles; 2000–2005: 3493 articles; 2005–2010: 6958 articles; 2010–2015: 8617 articles.

Not only the number of publications but also advances in the technology led to an improvement in detection sensitivity by roughly 100-fold up to 10−8 RIU (refractive index unit) or 0.01 RU (resonance or response unit). The range of affinity and kinetic data that can be determined has been extended at least 200-fold as a consequence of the increased sensitivity and due to improvements in data analysis. The number of independent channels or spots grew from four channels in 1990 (Biacore) to at least 192 flow-controlled spots in the new IBIS MX96 imaging instrument and more than 10 000 drop-spotted ligands in SPR imaging instruments from various manufacturers (e.g. Plexera). The carboxymethylated dextran surface introduced in 1990,1 still the first choice for many applications, has been complemented with a range of other surfaces (see Chapter 6). Systems for dedicated applications have been introduced by various manufacturers as complements to all-purpose research instrumentation,2 and the impact of SPR biosensors on biomolecular interaction studies is growing continuously. With improved experimental design and advanced data analysis methods, high-quality data for the determination of kinetic parameters of biomolecular interaction phenomena can be obtained. These data promise additional insights not only into the affinity of biomolecular pairs but also into the mechanisms of molecular binding events, which will be important for function–regulatory protein interaction studies in order to unravel the exciting processes in living species.

1.2 What is a Biosensor?

The term biosensor was introduced around 1975, relating to the exploitation of transducer principles for the direct detection of biomolecules at surfaces. Currently the most prominent example of a biosensor is the glucose sensor, reporting glucose concentration as an electronic signal, e.g. based on a selective enzymatic process. According to the current definition, in biosensors the recognition element (ligand) of the sensor or the analyte should originate from a biological source.Biosensors are analytical devices comprising a biological element (tissue, microorganism, organelle, cell receptor, enzyme, antibody) and a physicochemical transducer. Specific interaction between the target analyte and the biological material produces a physicochemical change detected by the transducer. The transducer then yields an analog electronic signal proportional to the amount (concentration) of a specific analyte or group of analytes.Anthony P. F. Turner (Editor, Biosensors and Bioelectronics)

Application of SPR-based sensors to biomolecular interaction monitoring was first demonstrated in 1983 by Liedberg et al.3 A historical overview of the use of the phenomenon for biosensor applications is given in Chapter 2. To understand the excitation of surface plasmons, let us start with a simple experiment.

1.2.1 A Simple Experiment

Consider the experimental set-up depicted in Figure 1.2. When polarized light is shone through a prism on a sensor chip with a thin metal film on top, the light will be reflected by the metal film acting as a mirror. On changing the angle of incidence and monitoring the intensity of the reflected light, one observes that the intensity of the reflected light passes through a minimum (Figure 1.2, line A). At this angle of incidence, the light will excite surface plasmons, inducing surface plasmon resonance, causing a dip in the intensity of the reflected light. Photons of p-polarized light can interact with the free electrons of the metal layer; inducing a wave-like oscillation of the free electrons, thereby reducing the reflected light intensity.

Fig. 1.2 Schematic experimental set-up of surface plasmon resonance excitation. A sensor chip with a gold coating is placed on a hemisphere (or prism). Polarized light shines from the light source (star) on the sensor chip. Reflected light intensity is measured in the detector (disk). At a certain angle of incidence (φ), excitation of surface plasmons occurs, resulting in a dip in the intensity of the reflected light (A). A change of refractive index at the surface of the gold film, will cause an angle shift from A to B.

The angle at which the maximum loss of the reflected light intensity occurs is called resonance angle or SPR-dip. The SPR-dip angle is dependent on the optical characteristics of the system, e.g. on the refractive indices of the media on both sides of the metal, usually gold, and are explained in detail in Chapter 2. Whereas the refractive index at the prism side does not change, the refractive index in the immediate vicinity of the metal surface will change when accumulated mass (e.g. proteins) adsorb on the thin gold layer. Hence the SPR conditions are changing and the real-time shift of the SPR angle is suited to provide information on the kinetics of, e.g., protein adsorption on the surface.

1.2.2 From Dip to Real-time Measurement

Surface plasmon resonance is an excellent method for monitoring changes in the refractive index in the near vicinity of the metal surface. When the refractive index changes, the angle at which the intensity minimum is observed will shift as indicated in Figure 1.2, where line A depicts the original plot of reflected light intensity versus incident angle, and B is the plot after the change in refractive index. SPR is not only suited to measure the difference between these two states, it can also monitor the change in time, if one follows in time the shift of the resonance angle at which the dip is observed. Figure 1.3 depicts the shift of the dip in time, a so-called sensorgram. If this change is due to a biomolecular interaction, the kinetics of the interaction can be studied in real time.

Fig. 1.3 A sensorgram: the angle at which the dip is observed versus time. First, no change occurs at the sensor and a baseline is measured with the dip at SPR angle (A). After injection of the sample (arrow), biomolecules will adsorb on the surface, resulting in a change in refractive index and a shift of the SPR angle to position B (association). The adsorption–desorption process can be followed in real time and the amount of adsorbed species can be determined. Dissociation of the analyte bound to the ligand occurs when the sample containing the analyte is exchanged again with the system buffer.

SPR sensors investigate only in a very limited vicinity in a fixed volume at the metal surface. The penetration depth of the electromagnetic field (so-called evanescent field) at which a signal is observed typically does not exceed a few hundred nanometers, decaying exponentially with the distance from the metal layer at the sensor surface. The penetration depth of the evanescent field is a function of the wavelength of the incident light, as explained in Chapter 2.

SPR sensors lack intrinsic selectivity: all refractive index changes in the evanescent field will result in a change of the signal. These changes can be due to a refractive index difference of the medium, e.g. a change in the buffer composition or concentration, or temperature effects; non-specific and specific adsorption of material on the sensor surface can also cause refractive index changes. The amount of adsorbed species can be determined after injection of the original baseline buffer as shown in Figure 1.3. To allow selective detection at an SPR sensor, its surface needs to be modified with ligands suited for selective capturing of the target compounds (the analyte) but which are not prone to adsorb any other components present in the sample or buffer media.

1.3 How to Construct an SPR Assay

Now we have a basic understanding of the SPR signal and how to measure it in time. We know that the sensor surface needs to be modified to permit selective capturing and thus selective measurement of a target compound or analyte. In the following, we will learn more about an SPR measurement. First, the steps of an SPR assay will be discussed from immobilization through analysis to regeneration in a measurement cycle. Next, we become acquainted with standard 1 : 1 interaction models, including kinetics, followed by examples of assay formats. Finally, a short outlook is provided on the basics of the instrumentation.

1.3.1 The Steps of an Assay

In the simplest case of an SPR measurement, a target component in solution or analyte is captured by the capturing element or so-called ligand. The ligand is permanently or temporarily immobilized on the sensor surface (Figure 1.4) prior to the measurement of the analyte interaction. Various sensor surfaces with immobilized ligands are commercially available, and many more can be custom made, as explained in Chapter 6 by Erk Gedig.

Fig. 1.4 Schematic representation of direct detection: the analyte is captured by the ligands (Y) immobilized on the sensor surface. Accumulation of the analyte results in a refractive index change in the evanescent field detected by SPR. Here the ligand is immobilized in a hydrogel.

In the simplest case, the event of capturing the analyte by the ligand gives rise to a measurable signal; this is called direct label free detection. Figure 1.5 shows the sensor signal step-by-step in the measurement cycle with direct detection.

Fig. 1.5 Sensorgram showing the phases of an analysis cycle. (1) Buffer is in contact with the sensor (baseline). (2) Continuous injection of sample solution (association phase). A refractive index (RI) bulk shift occurs when the system buffer RI is different from the analyte buffer RI. (3) Injection of buffer (dissociation phase); ΔR indicates the measured response due to the bound target compound. (4) Removal of bound species from the surface during injection of regeneration solution (regeneration step) followed by a new analysis cycle.

Each measurement starts with conditioning of the sensor surface with a suitable buffer solution (1). It is vital to have a stable baseline before the capturing event starts. Then common mode effects such as temperature or hydrogel swelling no longer fluctuate. At this point, the sensor surface contains the active ligands, ready to capture the target analytes. On injecting the solution containing the analyte (2), these molecules are captured on the surface. Other components of the sample might also adhere to the sensor surface; without the selection of a suitable ligand, this adherence will be non-specific. In this step, the adsorption kinetics of the analyte molecule can be determined in a real-time measurement. Next, buffer is injected onto the sensor and the non-specifically bound components are flushed off during the so-called dissociation phase (3). As indicated in Figure 1.5, the accumulated mass can be obtained from the SPR response (ΔR). The double-headed arrow represents the specific response (ΔR). Also in this step, dissociation of the analyte starts, enabling the kinetics of the dissociation process to be studied. Finally, a regeneration solution is injected, which breaks the specific binding between analyte and ligand (4). If properly anchored to the sensor surface, the ligands remain on the sensor, whereas the target analytes are quantitatively removed. In order to perform many tests with the same sensor surface, it is vital to use a regeneration solution that leaves the activity of the ligands intact, as the analysis cycle is required to take place repeatedly hundreds, sometimes even thousands, of times. Again, buffer is injected to condition the surface for the next analysis cycle. If the regeneration is incomplete, remaining accumulated mass causes the baseline level to be increased. A typical example of a repeated injection of the same analyte (8×) over eight channels/spots is shown in Figure 1.6. The instrument is generating raw data, which should be zeroed and referenced to obtain a “clean” sensorgram of the biomolecular interaction only [to “scrub” the data that can be performed in Scrubber 2.0 software (see Chapter 9)].

Fig. 1.6 Processing (scrubbing) raw data of a real experiment of eight repeated analyte injections exposed to eight channels/spots. (a) Non-zeroed, non-referenced overlay plot; (b) zeroed, non-referenced overlay plot; (c) zeroed, referenced overlay plot; (d) zeroed, referenced serial plot. Observe that the initial slopes of RU shift for all spots are the same. (mass transport-controlled regime, see Section 1.3.1). Only the Rmax value per spot differs (see Section 1.3.3). Data generated in the IBIS MX96.

A typical referenced and zeroed sensorgram is shown in Figure 1.7 with the phases of an analysis cycle of Figure 1.5.

Fig. 1.7 For accurate measurement of the kinetic parameters, it is preferred that the analyte is injected at increasing concentrations (a). The serial injections can be aligned to present the data in an overlay plot (b). Software (e.g. Scrubber 2) calculates the kinetic parameters from the overlay plot using global fitting (all curves in the overlay plot are fitted simultaneously). Note that the initial slopes are different (see Sections 1.3.1 and 1.3.3). Data generated in the IBIS MX96.

Referencing means that at least two channels/spots are measured, one with ligand and the other without ligand as a reference channel. The referenced (= subtracted) signal is shown in Figure 1.7 and quality features are given in Figure 1.8. 1. Baseline phase: Initially, baseline buffer is in contact with the sensor surface to establish the baseline. For sensor calibration purposes, the injection of a calibration liquid (e.g. a tuned glycerol percentage spiked in baseline buffer) can be incorporated in this phase (not shown) in order to compensate for the RI bulk shift of the analyte buffer. 2. Association phase: Sample containing the target compound is injected; the capturing elements on the sensor surface bind the target compound, resulting in complex formation. 3. Dissociation phase: Upon injection of baseline=system buffer, target compounds (and also non-specifically bound molecules) dissociate from the surface. 4. Regeneration phase: The regeneration solution (e.g. low-pH buffer) is injected to remove the remaining bound target compounds (not shown in Figure 1.8).

Fig. 1.8 Example of a referenced and zeroed overlay plot of various analyte concentrations (same as Figure 4.22, courtesy of Arnoud Marquart).

After this phase, the cycle is completed and a new experiment can start by establishing the baseline again. If remaining accumulated mass is present, the baseline level will increase. Because of the different refractive indices of regeneration liquids and the difficulty of referencing caused by swelling or shrinking of the protein-loaded hydrogel (often by the pH shift–salt step), it is not necessary to show the real-time data of the regeneration process. It is sufficient to measure the baseline shift in the system buffer after the regeneration phase. Regeneration buffer scouting protocols as explained in Section 1.5.2 can be applied to find the optimal regeneration conditions.

Often SPR measurements are carried out to determine the kinetics of a binding process. For realistic results, it is vital to avoid immobilization changing the ligand in such a way that would influence its strength or affinity toward the target (analyte) compound. In addition, kinetic experiments can provide information on the equilibrium dissociation constants and the rate constants (on- and off-rate) and on the thermodynamics, e.g. on the binding energy of processes. Examples of kinetic binding curves are provided in Chapter 4 by Arnoud Marquart and the effects of distributed ligand immobilization and kinetic theory can be found in Chapter 5 by Huaying Zhao and Peter Schuck.

1.3.2 Concentration Determination

Apart from kinetic and thermodynamic studies, SPR measurements can also be used for the determination of the concentration of an analyte in a sample (quantitative analysis). In this case, first different concentrations of the analyte are applied in separate analysis cycles. The sensorgrams measured at different concentrations give an overlay plot similar to that depicted in Figure 1.9, with the plateaus of the association step increasing at increasing analyte concentration.4 However, at a certain high analyte concentration the SPR response saturates.

Fig. 1.9 Typical overlay plot of sensorgrams from serial diluted analyte concentrations. Just after injection at t0, a sample specific binding of the analyte occurs and mass transport to the surface is rate limiting and linearly dependent on the concentration. From the slopes for a positive control (dR/dt), the concentration of an unknown sample can be determined. During the association phase, the number of unbound ligand molecules decreases and dissociation takes place, causing bending of the binding curve. The off-rate constant or dissociation constant (kd) can be determined just after injecting system buffer=baseline buffer at t1.

A calibration curve can be constructed by simply plotting the response (ΔR) after a certain time interval (t1) versus the concentration, but it depends on the affinity of the interaction. For example, if the concentration of the analyte in the sample is very high, the undiluted sample will yield results on the upper plateau range of the calibration curve. However, diluted solutions might yield points along the lower, concentration-dependent sections of the calibration curve and the concentration of the target compound can be determined.

A calibration-free concentration analysis (CFCA) can be carried out if the binding of the molecules is partly mass transport limited. This is explained briefly in Section 1.4.2 and in Chapter 7 by Robert Karlsson et al. (Section 7.3.3 and Figure 7.12).

As mentioned above, SPR sensing means detection of refractive index changes at the sensor surface, which in practice translates to the amount of mass deposited at the sensor surface. Direct detection is only possible if the capturing event of the analyte brings about measurable refractive index changes. This is easier to achieve if the molecular weight (MW) of the analyte is large (i.e. around 1000 Da or higher). However, for small molecules to produce a measurable refractive index change, large numbers would be required, making the analysis intrinsically less sensitive. If the analyte is a small molecule (MW<1000 Da), the response will be close to the system noise of the instrument (see Chapter 7). Only if high ligand densities are applied in combination with a highly sensitive instrument will the low-MW compound shift the refractive index sufficiently for SPR detection.

Detection of small molecules can also be carried out using a different strategy. Most often, small molecules are detected in a competition or inhibition assay format. In all immune assay formats, not only is the lower detectable concentration limited, but also the physical number of immobilized elements on the sensor surface, which provide a maximum limiting value. Discussions of the different assay formats can be found in Section 1.5 and detection levels in Chapter 7.

1.3.3 Determination of Kinetic Parameters

The most prominent benefit of direct detection using SPR biosensor technology is the determination of the kinetics of (bio)molecular interactions. Reaction rates (ka and kd) and equilibrium constants (KD) of interactions can be determined, e.g. the interaction A+B→AB, where A is the analyte and B is the ligand immobilized on the sensor surface, can be followed in real time with SPR technology (note that in other parts of this book, e.g. Chapter 4, the ligand B is denoted L).

Table 1.1 contains the most relevant kinetic parameters, the association constant (ka) and dissociation constant (kd) for the simplest case A+B→AB. The association constant represents the reaction rate of the complex (AB) formation, giving the number of complexes formed per unit time at unit concentrations of A and B. As soon as the complex AB is formed on the sensor surface, its dissociation can commence. The dissociation rate constant (kd) expresses the number of AB complexes dissociating per unit time. Note that the unit dimensions for the association and dissociation rates are different and can vary with the stoichiometry of the complex. The typical range of the association and dissociation constants shows large variations and is dependent on, e.g., temperature.

Table 1.1 Definitions of the most relevant kinetic parameters: the association and dissociation constants.
Association rate constant, ka Dissociation rate constant, kd
Definition A+B→AB AB→A+B
Description Reaction rate of AB formation: number of AB complexes formed per unit time at unit concentrations of A and B Dissociation rate of AB: number of AB complexes dissociating per unit time
Units L mol−1 s−1 s−1
Typical range 103–107 10−1–5×10−6

When association of A and B starts, no product AB is yet present at the sensing surface. At this point, the rate of the association reaction is highest and that of the dissociation reaction is lowest. As the process progresses, more and more of the AB complex is produced enhancing the rate of dissociation. Equilibrium is reached when the rates of the association and dissociation reactions are equal; definitions are given in Table 1.2. As can be seen, the equilibrium association and dissociation constants, which represent the affinity of an interaction, have a reciprocal relationship with each other. Note that the dissociation equilibrium constant has a capital letter in the subscript (KD). The effects of parameters such as temperature are described in Chapter 2.

Table 1.2 Definitions of the equilibrium association and dissociation constants.
Equilibrium association constant, KA Equilibrium dissociation constant, KDa
Definition [AB]/[A][B]=ka/kd [A][B]/[AB]=kd/ka
Description Affinity to association: high KA; high affinity to associate Stability of AB: high KD; low stability of AB
Units L mol−1 mol L−1 (M)
Typical range 103–1012 10−3–10−12
a KD has the units of concentration in moles per liter (M) and is the preferred term to express affinity. Note the capital D in the subscript.

The rate constants (Table 1.1) and equilibrium constants (Table 1.2) of (bio)molecular interactions provide information on the strength of association and the tendency for dissociation. Various aspects of the kinetics, models, and calculation of affinity constants are described in Chapters 4 and 5.

1.3.4 Basics of Instrumentation

To study biomolecular interactions using SPR does not require a detailed understanding of the optical phenomena. It is sufficient to know that SPR-based instruments use an optical method to measure the refractive index near a sensor surface (within ∼300 nm from the surface). SPR instruments comprise three essential units integrated in one system: optical unit, liquid handling unit, and the sensor surface. The features of the sensor chip have a vital influence on the quality of the interaction measurement. The sensor chip forms a physical barrier between the optical unit (dry section) and the flow cell (wet section).

SPR instrumentation can be configured in various ways to measure the shift of the SPR-dip. In general, three different optical systems (see Chapter 3) are used to excite surface plasmons: systems with prisms, gratings, and optical waveguides. Most widespread are instruments with a prism coupler, also called instruments in the “Kretschmann configuration”.5 In this configuration, which is shown in Figure 1.2, a prism couples p-polarized light into the sensor coated with a thin metal film. The light is reflected onto a detector that measures its intensity, using a photodiode or a camera. All configurations share the same intrinsic phenomenon: the direct, label-free, and real-time measurement of refractive index changes at the sensor surface. SPR sensors offer the capability of measuring low levels of chemical and biological compounds near the sensor surface. Sensing of a biomolecular binding event occurs when biomolecules accumulate at the sensor surface and change the refractive index by exchanging the background electrolyte or water molecules. Protein molecules have a higher refractive index than water molecules (Δn≈10−1). The sensitivity of most SPR instruments is in the range of Δn≈10−5–10−8 or 10–0.01 pg mm−2 of proteinous material at the sensor surface. Often in real-time biosensing, absolute values are not a prerequisite – only the change is monitored as a result of biospecific interactions at the sensor surface. Detailed descriptions of a selection of commercial instruments are given in Chapter 3.

The selectively accumulated mass on the sensor chip surface (which is generally expressed in pg mm−2) correlates linearly with the change in the refractive index near the sensor surface measured by the SPR instrument.6 A rule of thumb is that for an instrument that uses light of wavelength of 670 nm, 1 ng mm−2protein accumulation gives a signal of about 1000 RU SPR angle shift. In Biacore instruments, ∼8.2 resonance units (RU) correspond to 1 millidegree (m°) SPR angle shift. One resonance unit (RU) corresponds to exactly 10−6 RIU by definition. When light of another wavelength is used, then the angle shift should be calibrated. A preferred calibration of instruments is that the output signal of an SPR instrument corresponds ultimately to RU shifts (and not to wavelength or percent reflectivity shifts) calibrated with known RIU glycerol–water mixtures for better comparing ligand densities and kinetics of biomolecular interactions between different instruments.One resonance unit (RU) corresponds to exactly 10−6 RIU

1.4 Kinetics of Biomolecular Interactions

Evanescent wave biosensors offer an easy way to measure the kinetics of the reversible binding of a biomolecule from solution to a ligand (typically another biomolecule) immobilized on the sensor surface. Although theoretical aspects are treated in depth in Chapters 4 and 5, a brief analysis of kinetics is described here. The conventional treatment starts with a simple 1 : 1 interaction model,7 equivalent to the Langmuir model, which is the simplest physically plausible isotherm based on three assumptions: The binding is specific to ligands attached to the surface. All sites are equivalent and the ligands are uniformly divided. The ability of an analyte to bind to its ligand is independent of the degree of occupation of neighboring sites. If these conditions are met, the dynamic equilibrium is given by image file: BK9781782627302-00001-t1.tif (1.1) assuming that A is the molecule binding from solution (analyte) and B is the species immobilized on the sensor surface (ligand) (or L=ligand). The forward and reverse reaction rates are described by the adsorption (association) rate constant (ka) and the desorption (dissociation) rate constant (kd), respectively. The association process results in the formation of the complex [AB] on the sensor surface and is described by image file: BK9781782627302-00001-t2.tif (1.2) while the dissociation rate of the complex [AB] is given by image file: BK9781782627302-00001-t3.tif (1.3) Once a dynamic equilibrium is established, the rates of the two processes are equal, i.e. ka[A][B]=kd[AB] (1.4) Hence the equilibrium constants can be expressed by the rate constants according to image file: BK9781782627302-00001-t4.tif (1.5) and image file: BK9781782627302-00001-t5.tif (1.6) where KA and KD are the affinity constant and the equilibrium dissociation constant, respectively. This formalism is mathematically identical with that for the treatment of the interaction in a homogeneous phase. However, at the solid/liquid interface the transport (diffusion and convection) of A from the bulk solution to the interface must be taken into account (see Chapter 5). Because KD has the dimensions of concentration (moles per liter) and relates to the amount that has been bound with respect to the free analyte concentration, it is recommended to apply KD and not KA.

1.4.1 Mass Transport-controlled Kinetics

In most SPR systems, the following reaction takes place: image file: BK9781782627302-00001-t6.tif (1.7) This reaction describes two events. First, analyte is transported out of the bulk solution to the surface of the sensor. Second, the actual analyte–ligand interaction takes place (see Figure 1.10). Both events have their own rate constants km and ka/kd. When the diffusion of the analyte from the bulk to the surface is slower than the binding rate of the analyte to the ligand, a shortage of analyte at the surface occurs. In this situation, ka (and kd) is limited by the mass transport. The apparent ka is slower than the real ka. For kinetic measurements, mass transport limitation (MTL) is obviously unwanted and should be minimized as much as possible. However, for concentration measurements, MTL can be exploited since under full or partial MTL the binding rate is proportional to the analyte concentration as can be observed in Figure 1.9. It is this property of MTL that allows concentration measurements without the need for a calibration curve or calibration standard [see calibration-free concentration analysis (CFCA) as explained briefly in the next section and in Section 7.3.3 and Figure 7.12].

Fig. 1.10 Parameters that determine mass transport limitation (MTL).

Therefore, in other words, if the surface concentration of B (ligand) is very high, and the mass transport rate km is small compared with the association rate constant ka, i.e. kmka[B], in the extreme boundary condition the interaction is controlled by the mass-transport rate only. Then the complex formation rate is dependent solely on the bulk concentration of analyte A, and the binding signal increases linearly with time: image file: BK9781782627302-00001-t7.tif (1.8) This can be used for the determination of the concentration of analyte A, since the slope of the initial stage of the binding curve is proportional to the analyte concentration. However, the rate of binding should be totally limited by mass transport, the degree of which can be determined with the calibration-free concentration analysis (CFCA) method. Theoretically, the linear range of the dose–response curve has no limitation at the lower concentration side. If the reaction rate is fully mass transport limited, the sensor surface acts like an infinite sink and [Asurface]=0. In this case, km for all practical situations can be described by8 image file: BK9781782627302-00001-t8.tif (1.9) where D is the diffusion coefficient, h and b are the height and the width of the flow cell, respectively, v is the volumetric flow rate, and x is the distance from the flow cell entrance.

1.4.2 Calibration-Free Concentration Analysis (CFCA)

As indicated in the previous section, there is a linear relationship of the slope with respect to the concentration only when the interaction process is 100% mass transport limited. However, a 100% mass transport-limited interaction often cannot be reached and the binding is both mass transport limited and rate (kinetically) limited (note in Figure 1.6c how early the slope of lower density spots deviates from that of higher density spots for the same analyte concentration). By applying two different flow rates, the degree of mass transport limitation can be experimentally determined. This CFCA method was recently introduced by Karlsson and Roos9 (see also Section 7.3.3 and Figure 7.12).

For CFCA measurements, the initial binding rate of an analyte is measured at two different flow rates, i.e. under two different conditions of partial MTL. The two conditions are chosen such that they maximize the difference in degree of MTL while maintaining data quality. In order for CFCA to function correctly, it is preferred that a large amount of ligand is immobilized. Also, the ligand must be stable and a proper regeneration buffer must be used, leaving the ligand intact. From the two initial binding rates obtained, the degree of MTL can be determined by calculating the QCratio, as explained by Pol et al.10 For a certain analyte (e.g. an IgG molecule) the molecular weight is known and a flow cell characteristic parameter can be calculated. An unknown functional concentration of analyte, e.g. in supernatants, can then be calculated from image file: BK9781782627302-00001-t9.tif (1.10) where FCMW is a flow cell constant.

Analyte may be present in complex matrices such as culture supernatants or mixtures, where the exact functional concentration determination is difficult. One may establish the activity of the analyte solution by measuring the concentration by means of A280 (absorbance at 280 nm) and comparing this with the CFCA value. In this way, a “percent activity” value can be obtained for the analyte solution (100×[A]CFCA/[A]A280).

In SPR imaging instruments for multiplex analysis of biomolecular interactions (e.g. the IBIS MX96), only one ligand spot is necessary to obtain the analyte concentration, which may then be used for the entire array, e.g. for early-stage screening experiments such as affinity ranking of monoclonal antibody panels.

To summarize, CFCA allows the functional concentration of an analyte to be determined by SPR without the need for a calibration curve. In practice, this leads to a flow cell constant, FCMW, which can be used to calculate an unknown analyte concentration with the same molecular weight by measuring the initial slope of the association phase under two different mass transport-limited conditions set by the flow rate (see also Section 7.3.3 and Figure 7.12).

1.4.3 Interaction-controlled Kinetics

If the mass transport rate is much larger than the association rate constant, or if the surface concentration of the immobilized species is low, i.e. kmkon [B], then [Asurface]=[Abulk] and the binding rate can be expressed as image file: BK9781782627302-00001-t10.tif (1.11) The surface concentration of the free binding site, [B], is the difference between the concentration of the complex at saturation, [ABmax], and the current complex concentration, [AB]: [B]=[ABmax]−[AB] (1.12)

Combining eqn (1.11) and (1.12), and considering that the response R scales linearly with the complex concentration [AB], one obtains image file: BK9781782627302-00001-t11.tif (1.13) where c0 is the concentration of the analyte ([A]bulk) and Rmax is the saturation signal where all functional ligand molecules bound analyte molecules. Theoretically, this means that the Rmax signal can be reached only at unlimited high analyte concentrations where the degree of dissociation can be fully neglected. Be aware that [AB]max or [B]max corresponds to the maximal loading of the ligands with analyte, while Rmax corresponds to the maximal theoretical SPR signal when analyte molecules occupy all functional ligand molecules. In Figure 1.11, a serial (a) and an overlay plot (b) of binding curves from two spots with a high and a low Rmax value are exposed to the same analyte series of injections (IBIS MX96 data).

Fig. 1.11 (a) Serial plot (referenced and zeroed) of eight analyte injections in two channels (spots). The Rmax values or analyte capacity (functional ligand density) of the two spots are different. (b) Overlay plot. Data generated with the IBIS MX96.

Hence the Rmax value represents the functional ligand density on the surface. The solution of eqn (1.12) yields image file: BK9781782627302-00001-t12.tif (1.14) This is shown schematically in Figure 1.12 in the association phase.

Fig. 1.12 Equilibrium analysis based on a 1 : 1 Langmuir model. Only when the sensorgram shows equilibrium [%Rmaxversus time (s)] can equilibrium analysis be applied. Note that if the concentration has the value of KD the Y-value (%Rmax) is at 50% at equilibrium.

For the dissociation phase, c0=0, hence image file: BK9781782627302-00001-t13.tif (1.15) and the solution becomes [cf.Figure 1.12 (dissociation phase)] R=R0 ekdt (1.16)

Eqn (1.15) and (1.16) can be used to obtain ka and kd from a single set of association/dissociation experiments with various analyte injections using global non-linear curve fitting.

1.4.4 Equilibrium Analysis

Once a dynamic equilibrium is reached, the net effect of the association and dissociation process is zero, i.e. image file: BK9781782627302-00001-t14.tif (1.17) where Req is the equilibrium response at a given analyte concentration c0. Therefore, the equilibrium signal reflects the affinity constant KA and dissociation constant KD of the interaction couple. This can be converted to a format that resembles the 1 : 1 Langmuir isotherm: image file: BK9781782627302-00001-t15.tif (1.18) image file: BK9781782627302-00001-t16.tif (1.19)

The isotherm is an S-shaped curve if using a logarithmic axis for the concentration, as shown in Figure 1.13. Upon application of an analyte concentration of c0=KD, Req is half of the saturation response Rmax. This is an important observation and the reason why the affinity equilibrium dissociation constant KD should be used and not KA. Thus the KD value expresses the concentration at equilibrium where 50% of the ligands are occupied by an analyte molecule (=½Rmax).

Fig. 1.13 The % Req/Rmax ratio as a function of the concentration of the analyte [A] can be plotted e.g. on a linear scale (a) or semilogarithmic scale (b). Note that the equilibrium dissociation constant depends only on the Req/Rmax ratio and not on the actual value of Rmax (in RU).

Although this pseudo-first-order kinetic model has been used successfully in qualitative studies (such as in the demonstration of interactions between biomolecules), the determination of the kinetic rate constants of binding is often complicated by the fact that most binding curves deviate from the single exponential time course expected for a simple pseudo-first-order reaction. Apart from the experimental causes (e.g. sample depletion, noise, drift, impurity), major concerns regarding the deviation are focused on mass transport/rebinding effects, multivalent interactions/avidity effects, heterogeneity in the immobilized ligands/matrix effects, and complex binding mechanisms. It is recommended that with complex kinetics deviating from 1 : 1 binding the experimental conditions should be changed in such a way that it will approach and will be closer to the simple 1 : 1 binding model. On improving the experimental design (e.g. by using high flow rates and low surface capacities) and applying advanced analysis algorithms {e.g. global analysis11 and distribution analysis of the constants (Chapter 5), fitting association and dissociation phase data for a series of concentrations simultaneously}, the contribution of most of these effects can be minimized or even corrected.12

1.5 Buffer Solutions for Measuring the Analysis Cycle

1.5.1 Baseline or System Buffer

The running, background, baseline, or system buffer should create optimal physiological conditions for the binding of the analyte to the ligand. For biomolecular interactions, the baseline or system buffer is usually a physiological buffer with sufficient salt and (near) neutral pH. Phosphate-buffered saline (PBS) is a standard system buffer. Alternatively, 10 mM HEPES, pH 7.4–0.15 M NaCl (HBS) is often used with small amounts of ions such as potassium added. The addition of a surfactant, e.g. Tween-20 (0.03–0.075%) to the buffer is advantageous to minimize non-specific binding. The surfactant not only enhances the ratio between specific and non-specific binding but also helps to prevent adsorption of air bubbles on the surface. However, surfactants should not be added if hydrophobic surfaces are used for non-covalent attachment of membrane-bound components. Sometimes 3 mM EDTA is added to trap remaining bivalent cations (e.g. Mg2+ or Ca2+), which may interfere with the carboxylate groups in the hydrogel layer. Blocking components may help to reduce non-specific binding, e.g. 3% bovine serum albumin (BSA) or human serum albumin (HSA) if patient serum is used. In order to optimize the system buffer for dedicated biomolecular interaction experiments, the ABA inject protocol (as explained in Section 7.3.6 and Figure 7.13) can be used to find the optimal system buffer. When advanced microfluidic valves are involved in the instrument, the buffer solutions should be prepared with Milli-Q quality water and should preferably be filtered through a 0.22 μm filter and degassed before use. Tuning the refractive index of the background buffer to mimic the sample and using the baseline buffer for dilution of the sample is a way to minimize the bulk refractive index effect. Not measuring a bulk refractive index step is better than correcting the bulk RI effect by advanced referencing.

1.5.2 Regeneration Buffer

For repeated use of the same sensor chip, the surface should be regenerated by removal of analyte and any other non-covalently bound material. However, the ligand should be kept intact and should not be inactivated or denatured during the regeneration phase. Commonly used solutions for regeneration include low-pH buffers, e.g. 10 mM glycine·HCl, pH 2.5, or 100 mM phosphoric acid, pH 3, or a high-salt solution, e.g. 3 M MgCl2. Optimal sensor regeneration includes a pH shock, and regeneration is preferably performed as two subsequent steps of, for instance, 30 s each, rather than one step of 60 s. Often in the transition the regeneration process takes place. If a negatively charged gel-type sensor is used, not only will the specific bond between ligand and analyte break during low-pH regeneration, but also the hydrogel could collapse, with the analyte being more or less squeezed out of the hydrogel. If the ligand or analyte cannot withstand low pH, sometimes a highly alkaline pH is used, e.g. 10 mM NaOH (pH >11). Alternative regeneration solutions have a high salt concentration, and the salts used include chaotropic agents that are chemicals, such as urea and guanidine hydrochloride, that disrupt hydrogen bonding in aqueous solutions. It should be noted that concentrated solutions of these agents may denature proteins, because they also interfere with hydrophobic interactions, and the functional ligand density as determined with the Rmax value will decrease accordingly. Geuijen et al. described the testing of various regeneration buffers.13

Furthermore, from the sensorgram without measuring the regeneration phase, the effectiveness of the regeneration procedure can be determined by tracking the baseline levels in a concatenated series of injections. Incomplete regeneration will result in residual analyte and/or non-specifically bound components on the sensor surface, which will increase the baseline, whereas too harsh sensor regeneration will result in a decreased binding capacity of the sensor in subsequent analysis cycles. In multiplex array-based instruments where a single regeneration buffer is injected, a regeneration buffer scouting protocol should be applied to group the ligands for the optimal regeneration buffer (not too mild or too harsh but just right). A general protocol for antibody–analyte interactions is to start with MgCl2 (3 M) followed by Gly·HCl (100 mM), pH 3.0, phosphoric acid (100 mM), pH 3.0, Gly·HCl, pH 2.5, and finally Gly·HCl, pH 2.0. Multiple antibodies should be grouped for the optimal regeneration conditions and ranking the affinity and epitope binning experiments should be performed using these conditions (see Chapter 8).

1.6 SPR-based Immunoassays

In general, an immunoassay is a laboratory technique based on the binding between an antigen and its homologous antibody in order to identify and quantify a specific antigen or antibody in a sample. In classical immunoassays, the determination of the concentration of an analyte relies on signals generated by various labels (fluorescent dyes, enzymes, or radioisotopes) attached to antigens or antibodies.14 Labeling may disrupt the binding sites involved in the interaction. Moreover, labeling induces heterogeneity of the biomolecular interaction because in most cases labeling of a specific molecule (e.g. an antibody) is not homogeneously distributed. In addition, the label itself might interact with the capturing ligands, leading to false positives. A way to circumvent some of these problems is to use a labeled secondary binding molecule, but this extra step requires an additional, high-affinity binding compound and it will also increase the analysis time required. An SPR-based biosensor measures protein–protein binding directly as a shift in surface-bound masses and therefore it can be applied as an immunoassay.

Figure 1.14 shows different assay formats that are appropriate for SPR, including direct, competitive, inhibition, and sandwich assays.

Fig. 1.14 Immunoassay formats commonly used in SPR measurements. (a) Direct assay. The ligand (antibody) is immobilized on the sensor surface and interaction with the analyte (here antigen) yields a detectable refractive index shift. (b) Competition assay for measuring small molecules where direct capture of the antigen yields an insufficient refractive index shift, whereas the conjugated antigen is large enough for a measurable refractive index shift. (c) Inhibition assay where the analyte is the same molecule as the immobilized ligand. Antibody is added to the sample in small excess. The analyte forms conjugates with the antibody, inhibiting the binding to the sensor surface. (d) Sandwich assay with secondary antibody.

1.6.1 Direct Assay

The direct assay is described by A+B→AB. In this simplest type of assay, antibodies directed against the antigen (= target compound) are immobilized on the sensor surface (= ligand). Sample solution containing the antigen (analyte) is then incubated with the sensitized sensor surface. The signal increase resulting from antigen binding correlates with the amount of antigen in the sample. Direct assays can also be designed with the antigen coupled to the surface (= ligand) and detection of the binding of the specific antibody (= analyte).

1.6.2 Competition Assay

This type of assay is typically applied for the detection of low molecular weight antigens that do not generate a sufficient signal, while accumulating at the sensor surface. In this assay format, specific antibodies are immobilized on the sensor surface and sample solution that contains the antigen is mixed with a conjugated antigen. Because of its high molecular weight, the conjugate enhances the SPR angle shift. The antigen–conjugated antigen mixture is incubated with the sensor surface. The difference in signal between a reference sample containing only conjugated antigen and the sample solution indicates the amount of antigen in the sample. In this assay, high antigen concentrations in the sample will result in low signals (less conjugated antigen can be bound). Kinetically, this type of assay shows deviations because the diffusion constants of the analyte and the conjugated analyte will differ significantly. Only equilibrium analysis can be applied.

Competition assays are often used for the detection of toxic compounds. The maximum signal is generated when no free (toxic) analyte is present. When the signal is too low, possibly the analyte is present in the sample or the ligands have been denatured or poisoned by the sample and are no longer active while the analyte is not present. Both outcomes are harmful and will require action. If the competition assay shows a response of the unconjugated antigen as in the direct assay, then detailed calibration procedures have to be performed. Preferably in SPR-based competition immunoassays, the conjugated antigen is a high refractive index label (e.g. a latex bead or gold nanoparticle) loaded with the antigen. Equilibrium analysis is the preferred approach.

1.6.3 Inhibition Assay

In this assay format, the target antigen is immobilized on the sensor surface. Sample solution containing the antigen is mixed with specific antibodies in excess and incubated with the sensor surface. Antibodies will bind both to antigen in solution and to antigen that is bound to the sensor surface. The difference in signal between a blank sample that does not contain the antigen and the sample solution indicates the amount of antigen in the sample. In this assay, high antigen concentrations in the sample result in low signals (fewer antibodies remain to bind to the surface). Because antibodies have high molecular weight, their binding is directly detected. Both competition and inhibition assays need various injections of ratio dilutions of the antigen with fixed concentrations of the competitor/inhibitor.

1.6.4 Sandwich Assay with Secondary Antibody and Signal Enhancers

In sandwich assays, antibody molecules against the analyte are immobilized on the sensor surface, capturing the analyte molecules when sample solution is incubated with the sensor surface. In the next injection, a secondary antibody binds specifically with either the antigen or the antigen-bound antibody. The antigen is captured by a sandwich of two antibodies. Only very high affinity capture antibodies should be used, in order to avoid a mixture of affinities of each component in the sandwich assay. Several steps can be built in, which complicates the analysis, however, but it may increase the sensitivity and specificity dramatically.

Often, a highly specific goat/rabbit/sheep anti-mouse immunoglobulin G (IgG) is immobilized as the first capturing agent, which traps a monoclonal (mouse) antibody for the antigen. Streptavidin–biotin linkers are often used because of the high affinity constant, which has limited interference with the rate and equilibrium constants of the analyte–ligand pair. The increase in signal (as a result of antigen binding) in a limited time window is proportional to the amount of antigen in the sample. Washing the surface with buffer is followed by the injection of a secondary antibody. The high molecular weight of the secondary antibody is usually sufficient for monitoring the binding process. The calculated Rmax value of this injection represents the functional antigen density as bound to the sensor surface in the previous step. For further signal enhancement, antibody conjugates with an enzyme generating insoluble precipitates or colloidal gold as refractive index label can be used.15 An advanced application is described in Section 12.6.2 regarding a signal enhancement cascade for boosting the dynamic range.

1.7 How to Read This Book

Although most of the chapters can be read as stand-alone literature on different aspects of SPR technology, this handbook aims to provide the reader with total coverage of the basics of biomolecular interaction sensing and its applications and most relevant developments at the time of reviewing. The next chapter provides the history and a description of the physics of surface plasmons and SPR in its original form.

The description of SPR instrumentation and a survey of currently available commercial products from various companies follow in Chapter 3. An introduction on how to obtain kinetic information from SPR measurements and get a feeling for the curves based on the “SPR pages” on the Internet can be found in Chapter 4 by Arnoud Marquart. In Chapter 5, distribution analysis is explained by Huaying Zhao and Peter Schuck. Chapter 6, by Erk Gedig, brings the reader closer to the surface architecture and chemical design strategies for SPR assay protocols.

Specific application areas are highlighted in subsequent chapters. The latest advanced information on Biacore instruments for fragment and low molecular weight compound analysis is discussed in Chapter 7 by Robert Karlsson et al. Chapter 8, by Koen Wagner, describes the full characterization of antibodies regarding ranking the affinity, epitope binning, and mapping to the single amino acid level. In Chapter 9, by Noah Ditto and Josh Eckman, modern software tools for processing of raw data and the analysis of biomolecular interactions for affinity determination and epitope binning are treated. Because of the important role of biolayer interferometry (BLI) in label-free interaction analysis, Chapter 10, by David Apiyo, deals the various aspect of BLI technology. Chapter 11, by Sylvie Ricard-Blum et al., is an application chapter about protein interaction networks and is a tutorial for good SPR analysis practice. The final Chapter 12 summarizes SPR technology in general and gives an outlook on future trends including SPR cytometry, which concerns new cell-based applications.

1.8 Questions

How can you reduce mass transport-limited interactions? Eqn (1.19) expresses the Req value for various concentrations of the analyte. Calculate the values of Req in %Rmax for c0=1/10KD, KD and 10KD. Why is it so important for equilibrium analysis that the response of a signal spot shows exactly the same sensitivity to a common refractive index shift (e.g. by glycerol) as a reference spot? How do you check that these responses are similar for various glycerol concentrations? The Rmax value represents the functional ligand density. Explain this and how you can determine the Rmax value. Explain why the concentration of an analyte in a direct assay should be determined in the MTL regime and in a sandwich assay by determining Rmax of the sandwicher.

References

  1. B. Johnsson , S. Löfås and G. Lindquist , Anal. Biochem., 1991, 198 , 168 —277 Search PubMed .
  2. J. M. McDonnell Curr. Opin. Chem. Biol., 2001, 5 , 572 —577 CrossRef CAS PubMed .
  3. B. Liedberg , C. Nylander and I. Lundström , Sens. Actuators, 1983, 4 , 299 —304 CrossRef CAS .
  4. R. Karlsson and A. Fält , J. Immunol. Methods, 1997, 200 , 121 —133 CrossRef CAS PubMed .
  5. E. Kretschmann Z. Phys., 1971, 241 , 313 —324 CrossRef CAS .
  6. R. Karlsson and A. Fält , J. Immunol. Methods, 1997, 200 , 121 —133 CrossRef CAS PubMed .
  7. P. W. Atkins Physical Chemistry , 1998, Search PubMed .
  8. P. Schuck Annu. Rev. Biophys. Biomol. Struct., 1997, 26 , 541 —566 CrossRef CAS PubMed .
  9. R.Karlsson and H.Roos, US Patent Application Publication US20140141529, May 22nd, 2014.
  10. E. Pol , H. Roos , F. Markey , F. Elwinger , A. Shaw and R. Karlsson , Anal. Biochem., 2016, 510 , 88 —97 CrossRef CAS PubMed .
  11. T. A. Morton , D. G. Myszka and I. M. Chaiken , Anal. Biochem., 1995, 227 , 176 —185 CrossRef CAS PubMed .
  12. D. G. Myszka , T. A. Morton , M. L. Doyle and I. M. Chaiken , Biophys. Chem., 1997, 6 , 127 —137 CrossRef .
  13. K. P. M. Geuijen et al. Anal. Biochem., 2014, 454 , 38 —40 CrossRef PubMed .
  14. The Immunoassay Book , D. WildStockton Press, New York, 1994, Search PubMed .
  15. G. A. J. Besselink et al. Anal. Biochem., 2004, 333 , 165 —173 CrossRef PubMed .

Footnotes

  1. This is a theoretical assumption. In Chapter 5, consequences of the ligand heterogeneity are explained.
  2. Some software programs sometimes apply [B]max and others apply Rmax. The latter is the recommended nomenclature to express the functional ligand density based on the analyte response for the specific analyte–ligand interaction.

© The Royal Society of Chemistry 2017 (2017)