Kinetic analysis of a high-affinity antibody/antigen interaction performed by planar waveguide fluorescence immunosensor

Abstract

Methods based on optical biosensors for the investigation of biomolecular interactions between high-affinity antibodies and antigens has advanced over the last years. In this study, experimental and data analysis protocols were developed to determine the rate and equilibrium constants by using multiplexed planar waveguide fluorescence immunosensor. Four different antibodies were used as models in the test system. Antigen–BSAs were immobilized on the sensor surface, and the binding of specific antibodies labeled with Cy5.5 under certain conditions was measured. Sets of binding curves obtained with different antibody concentrations were analyzed by using the numerical integration of differential rate equations and global fitting applying by the one to one reaction model. As a result, the kinetic rate constants (k_a and k_d) and affinity (K_D) can be determined for each antibody interaction under identical conditions. An indirect competitive immunoassay simulation model is also presented in the paper and shows the relationship between detection response, initial concentration of antibody and affinity constant. By analyzing the results and fitted in the indirect competitive immunoassay model, the optimized concentration of antibody and working ranges for detection can be estimated. Thus, the application of waveguide fluorescence immunosensors for protein interaction analysis is a promising and high throughput tool for obtaining data on the binding behavior between antibodies and antigens, and support the optimization in immunoassay.

---

1. Introduction

Protein interactions, such as monoclonal antibodies, are important in medical diagnostics, biochemical research, and food/environment monitoring.^1,2 To qualify and quantify antigen/antibody interactions in research and routine applications, optical biosensors were developed and used to overcome the drawbacks caused by calorimetry, electrophoresis, and analytical ultracentrifugation, which provide only low-resolution data and low throughput. Interaction analysis by optical biosensor technology has been commercially available since 1990.³ To date, the most commonly applied techniques are surface plasmon resonance (SPR)^4,5 and bio-layer interferometry (BLI).^6–9 SPR technology can be used to measure complex formation without labeling the reactants in real time and provide detailed information about the reaction kinetics.¹⁰ Compared with microfluidic SPR, which commonly delivers samples to a stationary sensor chip, BLI technology is established in an open shaking microwell plate format without microfluidics. However, SPR and BLI both are expensive and complex for analysis to maintain high sensitivity. Recently, many new methods are developed, based on carbon nanotube field-effect transistor,¹¹ TIRF-protein binding microarray,¹² that are rapid, sensitive in the analysis of low molecules interactions.

In this work we wanted to develop a feasible method to study antibody and antigen interactions on the basis of planar waveguide fluorescence biosensor.^13,14 The objective of achieving spatially-resolved excitation and fluorescence collection from fluorescently-labeled antibodies that are locally bound at a planar interface can be met by the evanescent field excitation of fluorophores. The excitation light was guided by total internal reflections within the transducer structure. This process results in an evanescent wave, which allows the excitation fluorophore bound to the transducer surface. The total internal reflection fluorescence (TIRF) principle allows the selective detection of surface-bound fluorophores, and the on-line monitoring of binding was superior to that of the direct illumination of the active area of transducers.^15,16 A planar transducer, which is manufactured as an integrated part of fluidics systems, was also preferred to a fiber-based system.¹⁷

Given our general interest in the development of planar waveguide fluorescence-biosensor applications, we design a method that uses antigens to capture Cy5.5 antibodies on the sensor surface. Furthermore, antigen–bovine serum albumin (BSA) surfaces can be recycled by a simple regeneration step. Therefore, the use of fully automated biosensors can enable the screening of single or multiple samples on the same sensor surface.

2. Materials and methods

2.1 Reagents and materials

Bovine serum albumin (BSA), N-(4-maleimidobutyryloxy) succinimide (GMBS), 3-mercaptopropyl-trimethoxysilane (MTS), and 1-ethyl-3-(3-dimethylamino-propyl). All other reagents, unless specified, were supplied by Beijing Chemical Agents. Monoclonal anti-melamine antibody, monoclonal anti-aflatoxin M1 antibody, monoclonal anti-bisphenol A antibody, monoclonal anti-sulfadimidine antibody, hapten conjugate BSA–melamine, BSA–aflatoxin M1, BSA–bisphenol A, and BSA–sulfadimidine were purchased from Shijiazhuang Solarpex Biotechnology Co., Ltd. The antibodies were also labeled Cy5.5, as previously described.

2.2 Instrumentation

The schematic of a fluorescence immunosensor is shown in Fig. 1. This instrument mainly comprises three parts: a planar waveguide biosensing platform, a flow injection system, and a control and data processing system. In the platform, the K9 glass biochip is made of 2.0 mm-thick surface-polished sheet glass (60 mm × 15 mm) with a beveled face was used for biosensing analysis. Light from a laser diode (Huanyuan-Star Laser Ltd., Beijing) with 635.0 nm wavelength and 10 mW power was coupled into the beveled edge of the biochip and propagated alongside the sensitive area of the biochip by the TIRF. In each TIRF point, the evanescent wave on-field generated on the surface of the biochip interacts with the surface-bound fluorescence-labeled biomoleculars, thus leading to the excitation of fluorophores, which were collected by high-numerical-aperture (NA = 0.46) plastic fibers located opposite the sensor positions. The collected fluorescence was then filtered by a high pass filter (HF01-700, CDHC-Optics, China) to reject lost and scattered laser lights. The fluorescence was then detected by photodiodes through lock-in detection. All reagents were delivered by a flow delivery system operated with a peristaltic pump. The control of fluid delivery system and data acquisition and processing were automatically performed by a computer. To ensure the activity and stability of bioreagents, the antibody storage and pre-reaction loop were maintained in two individual thermostats, where the temperatures were respectively adjusted to 4 and 37 °C.

Fig. 1 Schematic set-up of PWFI: (A) sensor chip, optics system and embedded computer (B) flow delivery system.

2.3 Sensor chip preparation

The sensor chips were cleaned in a freshly prepared piranha solution (concentrated H₂SO₄: 30% H₂O₂, 2 : 1, v/v) for approximately 60 min, rinsed with Milli-Q water, and dried by nitrogen. After drying under a nitrogen flow, 30 mL toluene with 2% MTS were applied to the surface and reacted for up to 2 h at room temperature. The silanized surface was successively rinsed with toluene dried in a stream of nitrogen to ensure the uniformity of the self-assembled monolayer. The thiol group of the silane was activated by immersing the chip in 2 mmol L⁻¹ GMBS (dissolved in ethanol) for 1 h, the thiol group of the silane was activated with a heterobifunctional cross linker, followed by ethanol and water washing and then drying in a nitrogen stream. This method provided a non-fouling background. Thereafter, antigen–BSA conjugate (0.5 mg mL⁻¹ in PBS) was dropped on the specific binding sites, placed in Petri dishes, and incubated overnight at a temperature of 4 °C. The epsilon amino groups on proteins were successfully bound to the succinimide group on the GMBS. Finally, the other unreacted active groups were blocked by 1 h incubation in 2 mg mL⁻¹ BSA.

2.4 Covalent labeling of antibody with Cy5.5-NHS ester

According to the manufacturer's recommendations, first, we should dilute a small sample of the antibody solution with 0.1 M NaHCO₃ so that the absorbance at 280 nm can be measured. Calculate the total amount of antibody required for labelling. Then, prepare a 10 mg mL⁻¹ solution of Cy5.5-NHS ester in dimethyl sulfoxide (DMSO). Calculate the volume needed to give the desired ratio of Cy5.5-NHS ester to antibody (e.g. 20 : 1), and add this gradually to the antibody solution while stirring. Stir the solution for a further 45 minutes at room temperature in the dark. To separate the free dye, dialyse the antibody against 1 L of 0.01 mol L⁻¹ PBS/0.01% sodium azide for 4 hours at room temperature, repeating this step as before using an overnight incubation at +4 °C. Next, filter the labelled antibody solution through a 0.22 μm syringe filter. Finally, dilute an aliquot of labelled antibody solution with 0.0 mol L⁻¹ PBS/0.01% sodium azide for the dual absorbance measurements at 280 nm (for protein) and at 675 nm (for Cy5.5; the molar extinction coefficient is 250 000 M⁻¹ cm⁻¹ at this wavelength). Correct the calculation for the absorbance of Cy5.5 at 280 nm; this is approximately 18% of the absorbance at 675 nm.

2.5 Biosensor experiments

A freshly prepared sensor chip that was micro-arrayed with melamine, aflatoxin M1, bisphenol, and sulfadimidine (as described in the section “Sensor chip preparation”) was placed in the flow cell. All biosensor experiments were performed in 10 mM PBS buffer containing 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM NaH₂PO₄, and pH 7.4 at 25 °C. To generate the kinetic binding data, 200 μL antibody solutions that contain melamine antibody-Cy5.5, aflatoxin M1 antibady-Cy5.5, bisphenol A antibody-Cy5.5, and sulfadimidine antibody-Cy5.5 were prepared at different concentrations. The mixture was then delivered into the sample cell for different times, which allowed the antibodies to bind onto the surface of the antigen immobilized biochip in the sample cell. To reduce the effect of free antibody in solutions and its non-specific adsorption on the detection result, the fluorescence signal was detected after the mixture was washed with 0.01 mol L⁻¹ PBS solutions. After each detection cycle, the sensor surface was regenerated by 3 min incubation in 0.5% SDS buffer (pH = 1.9) to break the antibody–antigen association. The sensor surface was then washed with 0.01 mol L⁻¹ PBS buffer. All measurements were performed in triplicate.

2.6 Kinetic data analysis

A bimolecular interaction between an antigen and antibody occurs provided that a single binding site is present on the antibody.^18,19 This binding is expressed in eqn (1):

Simple bimolecular:

where AgAb* is the antibody–antigen complex, Ab* is the labeled free antibody, Ag is the antigen, k_a is the association rate constant (liters per mole per time), and k_d is the dissociation rate constant (time⁻¹). The rate of change in the concentration of labeled bound antibody is described by the following:where [Ag] is the concentration of the antigen (M), [Ab*] is the concentration of the dissociated labeled antibody (M), is the rate of change of molar concentration of AgAb* (moles per liter per time).

For the biosensor, the change in instrument response R is proportional to the amount of bound ligate (eqn (3)).

The integration of eqn (3):

[Ab^*₀] is the initial concentration of antibody-Cy5.5, R_t is the response at time t, and R_max is the maximal response.

The modeled data for each binding mechanism were generated by the numerical integration of their differential rate equations by using Origin Lab 8.5.

The individual signal responses at each concentration and the maximum responses (R_max) were calculated as an average of three independent measurements.

The association rate constant (k_a) is defined as the rate of complex formation per second in a 1 mole solution of two reaction partners. The dissociation rate constant (k_d) indicates the stability of this complex.

2.7 An indirect competitive immunoassay model

There were two steps during an indirect competitive immunoassay method. First, the monoclonal antibodies solution should mix with the targets (antigens) for a period of time, so called pre-incubation. Second, the mixture was delivered into the sensors for several minutes which allowed antibodies with free binding sites left in the pre-incubation process binding with antigen immobilized.

The Section 2.5 has showed that the affinity constant K_D is calculated by the ratio of k_d/k_a

When equilibrium,

[Ab₀] = [Ab] + [AgAb], (6)

[Ag₀] = [Ag] + [AgAb], (7)

[Ab₀] is the initial concentration of antibody, [Ag₀] the initial concentration of antigen (targets).

Bring the eqn (6) and (7) into the eqn (5), get the equation as followings:

[AgAb]² − ([Ag₀] + [Ab₀] + K_D)[AgAb] + [Ag₀][Ab₀] = 0 (8)

Calculate the eqn (8), thus get results:

Bring the eqn (9) into the eqn (6),

Since the antibody can bind with two antigens, the eqn (10) need corrections. Based on the standard binomial distribution, the concentration of antibody binding with two antigens ([Ag₂Ab]) is according to [AgAb] and [Ab₀].

[Ag₂Ab] = [AgAb]²/(2[Ab₀]) (11)

Only the antibody has active sites can bind with the antigens immobilized, the concentration of antibody with active sites ([Ab_bind]) can be figured out in the followings:

[Ab_bind] = [Ab₀]/2 − [Ag₂Ab] (12)

Finally bring the eqn (9) and (11) into the eqn (12) and get the results.

3 Results and discussions

3.1 Planar waveguide fluorescence immunosensor-antibody/antigen assay

To establish this technology for antibody/antigen interaction analysis, sensor performance was adjusted. The antigen–BSA coated sensor chip were incubated with various antibody-Cy5.5 at different concentrations to measure the corresponding association and dissociation profiles of melamine, aflatoxin M1, bisphenol A, and sulfadimidine antibodies diluted in PBS. In the preliminary experiments, different concentrations were tested for each antibody formulation. The specific response at a low concentration (1 nM) was 2209 mV for aflatoxin M1 antibody-Cy5.5, when the time is 600 s. At low concentrations, no valid response of aflatoxin M1 antibody-Cy5.5 was wholly detected. A concentration of aflatoxin M1 antibody-Cy5.5 above 160 nM resulted in a signal plateau and, no further increase of the response signal can be measured. To perform comparable experiments, a concentration range between 1 and 160 nM was defined. Similar to the other three antibodies, a concentration range between 0.1 and 10 nM was defined for both bisphenol A antibody-Cy5.5 and sulfadimidine antibody-Cy5.5 analysis, respectively. A concentration range of 0.4 nM to 25 nM was defined for melamine antibody-Cy5.5. The individual association curves are shown in Fig. 2. Each selected antibody suspension interacted in a concentration- and lipid-dependent manner. The calculated R_max values for all antibodies at each concentration were globally fit and are shown in Fig. 3 according to the logistic model.

Fig. 2 Association curves of aflatoxin M1, melamine, bisphenol A, sulfadimidine in different concentrations. Curves represent the mean values of triplicate measurements of each concentration. (A) Sensor chip immobilized with antigens–BSA (concentration of 0.05 mg mL⁻¹), (B) sensor chip immobilized with antigens–BSA (concentration of 0.5 mg mL⁻¹).

Fig. 3 Simulated R_max values and equilibration curves for triplicate determinations of aflatoxin M1, melamine, bisphenol A, sulfadimidine in different concentrations. (A) Sensor chip immobilized with antigens–BSA (concentration of 0.05 mg mL⁻¹), (B) sensor chip immobilized with antigens–BSA (concentration of 0.5 mg mL⁻¹).

3.2 Kinetic analysis

To calculate the kinetic parameters of the Abs/Ags interaction, points that were outliers were removed from the data set. The same interaction was studied over the low and high density surfaces (concentrations of antigen–BSA conjugates for immobilization were 0.05 mg mL⁻¹, 0.5 mg mL⁻¹). Therefore, the binding events occurring at the surfaces can be characterized using a single k_a and k_d value.^20,21 Each surface was assigned a different R_max value because different amounts of antigen–BSA were immobilized over each surface. Although the concentration series were performed over the same antigen–BSA surfaces, they are assigned distinct R_max values to account for slight variations in surface activity.

To determine the affinity of each interaction, the antibody-Cy5.5 binding data obtained should be normalized by dividing each response curve by the corresponding antigen capture level (R_max). This approach ensures the global fit of the antibody-Cy5.5 binding kinetics of antigens from variable surface densities. The data shown in Fig. 2 and 4 show the antibody-Cy5.5 responses before and after normalization, respectively.

Fig. 4 Binding data was divided by the capture level of the corresponding antigen to generate the normalized profiles. Aflatoxin M1 (A), melamine (B), bisphenol A (C), sulfadimidine (D).

The normalized antibody-Cy5.5 binding data were fit globally to a simple interaction model.²² Each antibody/antigen interaction was defined in Origin Lab 8.5 as a different species. Accordingly, the kinetics of antigen binding to each antibody were mathematically described in the model by a set of local kinetic parameters and a global maximum binding capacity or R_max. In Table 1, the calculated rate and affinity constants of the different antibodies are summarized. Sulfadimidine antibody/sulfadimidin had the highest affinity constant k_a of 5.19 (10⁶ M⁻¹ s⁻¹), whereas aflatoxin M1 antibody/aflatoxin M1 had the smallest k_a of 0.37 (10⁶ M⁻¹ s⁻¹).

Table 1 Kinetic rate constants and affinities were determined for the antibody-Cy5.5/antigen interactions (n = 6)

Analyst	ka (10^6 M^−1 s^−1)	kd (10^−3 s^−1)	KD (nM)
Aflatoxin M1	0.37(±0.01)	0.30(±0.05)	0.81(±0.13)
Melamine	1.09(±0.18)	3.60(±0.19)	3.40(±0.68)
Bisphenol A	3.36(±0.55)	4.74(±0.98)	1.10(±0.14)
Sulfadimidine	5.19(±0.76)	5.25(±0.79)	1.01(±0.09)

---

3.3 Reproducibility

The major concerns in evaluating an immunosensor are stability and reusability, which would decrease both time and cost per assay and avoid calibration on each measurement. Because analytes are too small to be directly immobilized onto the chip, BSA-analytes conjugates were used as recognition elements that covalently attached to surface with a heterobifunctional reagent, which could prevent the compromise of immobilized molecules' binding properties. Flowing a 0.5% SDS (pH = 1.9) solutions for three minutes efficiently regenerated all the surfaces without damaging the bio-functionality of the immobilized bioreagents. The reproducibility of the test performance was evaluated with melamine antibody/melamine (3.2 nM, n = 20) to explore the reliability of the planar waveguide fluorescence based assay. The corresponding R_{t = 600} during the association phase was 7363 mV with a relative standard deviation (R.S.D.) of 2.79% (shown in Fig. 5). The R.S.D. was defined as the standard deviation divided by the mean and multiplied by 100%. The resulting R_t=600 values and the corresponding standard deviation for melamine, aflatoxin M1, bisphenol A, and sulfadimidine antibodies at all concentrations showed excellent reproducibility (Fig. 5). These results demonstrated that the established assay is reusable for more than 20 times. Through a large number of immunoassays, no significant change in performance (less than 5% decrease).

Fig. 5 Bars and error bars represent the mean R values and standard deviation for determinations of aflatoxin M1, melamine, bisphenol A, sulfadimidine at 5 nM, 3.2 nM, 1 nM, 1 nM (n = 20). Inset shows representative response for each determinations after regeneration.

3.4 Specificity among the four antigens and antibodies

The specificity among the four Ags and Abs was determined and then evaluated using the cross-reactivity rate (CR, %). The Ag–Ab CR was calculated based on the corresponding changes in fluorescence intensity (FI) using the multiplexed planar waveguide fluorescence immunosensor. With the exception of the corresponding Ag–Ab at 100%, other CRs were calculated using the rate of fluorescence intensities (FI) between the two hybridization effects. For instance, the CR value between BPA-BSA and AFM1-Ab was counted using eqn (14).

All the CRs are shown in Fig. 6. Based on the results, the CRs of four Ags (AFM1, BPA, Mel, SM2) were lower than 8%. Therefore, their affinities were comparatively high.

Fig. 6 Cross-reactivities among the four complete antigens and antibodies.

3.5 Concentration of antibody in immunoassay

Nowadays, particular attention has been paid to develop immunoassays or immunosensors such as surface plasmon resonance, quartz crystal microbalance, optical detection methods including fluorescence and chemiluminescence, and electrochemical method, in order to achieve sensitive and specific determination of biomolecules and proteins, applying the high affinity antigen/antibody interaction. In most immunoassays experiments, the concentration of antibody added initially is usually an important parameter, which would definitely affect the limit and range of detection. The Section 2.5 has showed the relationship between the responses, initial concentration of antibody, affinity constants. The affinity constant K_D is calculated by the ratio of k_d/k_a. When the R_t = 1/2R_max, , thus [Ab^*₀] = K_D.

Use the affinity constants (K_D) of the different antibodies (summarized in Table 1) and fit in the indirect competitive immunoassay model (eqn (13)), when [Ab^*₀] = K_D. Analyzing the simulation results (see in Fig. 7), the optimized concentration of antibody-Cy5.5 added initially could be K_D, which also all meet the MRLs set by FAO or EPA. Through the data in real detection should make some discount, the simulated results still can be used in the optimization of concentration.

Fig. 7 Simulated standard calibration curves for aflatoxin M1, melamine, bisphenol A, sulfadimidine when [Ab₀] = K_D.

4. Conclusion

Antibody/antigen interaction measurements by using planar waveguide fluorescence immunosensor are appropriate for the analysis of antibody/antigen interactions in a simple and reproducible manner. The simulation model for an indirect competitive immunoassay is also presented in the paper and shows the relationship between detection response, initial concentration of antibody and affinity constant. Thus, the application of planar waveguide fluorescence immunosensor for antibody/antigen interaction analysis is a promising and high throughput tool for obtaining data which could support the optimization of antibody concentration in most immunoassays.

Acknowledgements

This research is supported by the Major Scientific Equipment Development Project of China (2012YQ030111). Thanks to the special fund of State Key Joint Laboratory of Environment Simulation and Pollution Control (12L01ESPC).

References

1. C. Milstein, BioEssays, 1999, 21, 966–973.
2. C. Souriau and P. J. Hudson, Expert Opin. Biol. Ther., 2003, 3, 305–318.
3. L. G. Fägerstam, A. Frostell, R. Karlsson, M. Kullman, A. Larsson, M. Malmquist and H. Butt, J. Mol. Recognit., 1990, 3, 208–214.
4. D. G. Myszka, M. D. Jonsen and B. J. Graves, Anal. Biochem., 1998, 265, 326–330.
5. R. Karlsson and A. Failt, J. Immunol. Methods, 1997, 200, 121–133.
6. R. Barbour and M. P. Bova, Bioanalysis, 2012, 4, 619–622.
7. C. M. Maragos, Mycol. Res., 2011, 27, 157–165.
8. Y. Abdiche, D. Malashock, A. Pinkerton and J. Pons, Anal. Biochem., 2008, 377, 209–217.
9. J. Wallner, G. Lhota, D. Jeschek, A. Mader and K. Vorauer-Uhl, J. Pharm. Biomed. Anal., 2013, 72, 150–154.
10. M. A. Takacs, S. J. Jacobs, R. M. Bordens and S. J. Swanson, J. Interferon Cytokine Res., 1999, 19, 781–789.
11. S. Sorgenfrei, C.-Y. Chiu, R. L. Gonzalez Jr, Y.-J. Yu, P. Kim, C. Nuckolls and K. L. Shepard, Nat. Nanotechnol., 2011, 6, 126–132.
12. A. J. Bonham, T. Neumann, M. Tirrell and N. O. Reich, Nucleic Acids Res., 2009, 37, e94.
13. S. Y. Rabbany, R. T. Piervincenzi, A. W. Kusterbeck, R. Bredehorst and F. S. Ligler, Anal. Lett., 1998, 31, 1663–1675.
14. P. A. Lowe, T. J. Clark, R. J. Davies, P. R. Edwards, T. Kinning and D. Yeung, J. Mol. Recognit., 1998, 11, 194–199.
15. A. Brecht, A. Klotz, C. Barzen, G. Gauglitz, R. D. Harris, G. R. Quigley, J. S. Wilkinson, P. Sztajnbok, R. Abuknesha, J. Gascón, A. Oubiña and D. Barceló, Anal. Chim. Acta, 1998, 362, 69–79.
16. J. Tschmelak, G. Proll, J. Riedt, J. Kaiser, P. Kraemmer, L. Bárzaga, J. S. Wilkinson, P. Hua, J. P. Hole, R. Nudd, M. Jackson, R. Abuknesha, D. Barceló, S. Rodriguez-Mozaz, M. J. L. d. Alda, F. Sacher, J. Stien, J. Slobodník, P. Oswald, H. Kozmenko, E. Korenková, L. Tóthová, Z. Krascsenits and G. Gauglitz, Biosens. Bioelectron., 2005, 20, 1499–1508.
17. H. L. Guo, X. H Zhou, Y. Zhang, B. D. Song, L. H. Liu, J. X. Zhang and H. C. Shi, Sens. Actuators, B, 2014, 19, 114–119.
18. D. G. Myszka, X. Y. He, M. Dembo, T. A. Morton and B. Goldstein, Biophys. J., 1998, 75, 583–594.
19. T. Bravman, V. Bronner, K. Lavie, A. Notcovich, G. A. Papalia and D. G. Myszka, Anal. Biochem., 2006, 358, 281–288.
20. M. J. Cannon, G. A. Papalia, I. Navratilova, R. J. Fisher, L. R. Roberts, K. M. Worthy, A. G. Stephen, G. R. Marchesini, E. J. Collins, D. Casper, H. Qiu, D. Satpaev, S. F. Liparoto, D. A. Rice, I. I. Gorshkova, R. J. Darling, D. B. Bennett, M. Sekar, E. Hommema, A. M. Liang, E. S. Day, J. Inman, S. M. Karlicek, S. J. Ullrich, D. Hodges, T. Chu, E. Sullivan, J. Simpson, A. RaWque, B. Luginbühl, S. N. Westin, M. Bynum, P. Cachia, Y. J. Li, D. Kao, A. Neurauter, M. Wong, M. Swanson and D. G. Myszka, Anal. Biochem., 2004, 330, 98–113.
21. T. A. Morton and D. G. Myszka, Methods Enzymol., 1998, 295, 268–294.
22. D. G. Myszka and T. A. Morton, Trends Biochem. Sci., 1993, 23, 149–150.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra01073c

---