- Email: [email protected]
Implementing surface plasmon resonance biosensors in drug discovery
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
Implementing surface plasmon resonance biosensors in drug discovery David G. Myszka and Rebecca L. Rich Recent improvements in instrument hardware, experimental design and data processing have made it possible to use surface plasmon resonance (SPR) biosensor technology in the discovery and development of smallmolecule drugs. The key features of SPR biosensors (i.e. real-time binding analysis and lack of labeling requirements) make this technology suitable for a wide range of applications. Current instruments have a throughput of ~100–400 assays per day, providing a complement to secondary screening. The ability to collect kinetic data on compounds binding to therapeutic targets yields new information for lead optimization. Small-molecule analysis and emerging applications in the areas of ADME (adsorption, distribution, metabolism and excretion) and proteomics have SPR biosensors poised to play a significant role in the pharmaceutical industry.
*David G. Myszka and Rebecca L. Rich Center for Biomolecular Interaction Analysis University of Utah School of Medicine 4A417 50 North Medical Drive Salt Lake City UT 84132 USA *tel: 11 801 585 5358 fax: 11 801 585 2978 e-mail: [email protected]
▼ New discoveries are often driven by the appli-
cation of new technologies.An excellent example of this is the discovery of Jupiter’s moons by Galileo’s application of the telescope. Nowhere is the application of new technologies more important than in drug discovery, especially considering the fact that the average drug takes 12–15 years to develop and costs greater than US$750 million to produce. Pharmaceutical companies are always searching for technologies that will lower development costs and decrease the lead-time to market. Surface plasmon resonance (SPR; see Glossary in Box 1) based biosensors can be implemented in many stages of the drug discovery process to increase throughput and lower costs, while generating a wealth of new information about drug candidates. SPR biosensor technology SPR biosensors measure the quantity of a complex formed between two molecules in real-time
310
without the need for fluorescent or radioisotopic labels. This lack of labeling requirement makes these instruments amenable to characterizing unmodified biopharmaceuticals, studying the interaction of drug candidates with macromolecular targets and identifying binding partners during ligand fishing experiments. There are several examples in which SPR biosensors have been used to characterize the mechanistic details of biomolecular reactions1–6; they are therefore rapidly becoming a mainstay of the modern biophysical analysis laboratory. The best indicator of the success of SPR technology is the growing number of commercially available instruments. Five companies currently manufacture a variety of optical biosensor platforms. Biacore AB (Uppsala, Sweden) released the first commercial SPR instrument in 1990 (Ref. 7). Currently, the most advanced platform, BIACORE 3000, incorporates increased sensitivity and a smaller flow cell compared with earlier models, as well as online data subtraction and microsample
Box 1. Glossary ADME Adsorption, distribution, metabolism and excretion DMSO Dimethyl sulfoxide EDTA Ethylenediaminetetraacetic acid EIA Enzyme interaction analysis ELISA Enzyme-linked immunoassay HSA Human serum albumin HEPES N-(2-hydroxyethyl)piperazine-N9-2ethanesulfonic acid HIV Human immunodeficiency virus HTK Human tyrosine kinase ka Association rate constant kd Dissociation rate constant KD Equilibrium dissociation constant RU Resonance unit SPR Surface plasmon resonance
1461-5347/00/$ – see front matter ©2000 Elsevier Science Ltd. All rights reserved. PII: S1461-5347(00)00288-1
recovery. Based on the scientific literature, BIACORE instruments are the most commonly used commercial SPR technology (approximately 90% of the 1998 and 1999 commercial SPR biosensor publications cite the use of these instruments1,6). Affinity Sensors (Franklin, MA, USA) manufactures the IAsys line of instruments, which is a cuvette-based system that uses evanescent-wave technology8. Windsor Scientific Ltd (Berks, UK) markets the IBIS system, which can be configured as a flow- or cuvette-based instrument9. Nippon Laser & Electronics Lab’s (Hokkaido, Japan) SPR-CELLIA systems10 and Texas Instruments’ (Dallas, TX, USA) integrated SPR detector Spreeta11 can be configured for a variety of applications.
(a)
The advantages of SPR biosensors Compared with many other interaction technologies, SPR biosensors exhibit several distinct advantages for characterizing molecular interactions13:
De
t
h Lig
tec
tor
θ Antibody Antigen
Response
(b) How SPR biosensors work To illustrate a typical SPR biosensor assay, the format used to characterize an antibody–antigen interaction is presented in the context of the BIACORE system. The antibody is covalently immobilized on the biosensor surface and is referred to as the ligand (Fig. 1a).The antigen (referred to as the analyte) passes over the surface through a microfluidic flow cell. SPR detectors monitor the change in the refractive index of the solvent layer near the surface induced by association and dissociation of the analyte– ligand complex formation. For a more detailed description of SPR, the reader is referred to the article by Fägerstam et al.12 An example of the real-time-derived response data, which is referred to as a sensorgram, is illustrated in Fig. 1b. During the association phase of the interaction, analyte binds to the immobilized ligand, generating an increase in response. The response intensity levels off over time as equilibrium between bound and free analyte is achieved. To monitor dissociation, the instrument automatically switches back to running buffer, making it possible to collect information about the stability of the complex. At this time, any remaining analyte–ligand complexes can usually be disrupted using mild acidic or basic washing conditions, yielding a fully functional surface that can be sampled repeatedly. The number of binding cycles on a given surface is largely dependent on the stability of the immobilized ligand. One of the most essential features of an SPR biosensor is the interface between the sensor surface and the immobilized ligand. BIACORE technology utilizes a patented carboxymethyl dextran chip, which minimizes non-specific binding of biomolecules to the surface and increases the binding capacity of the surface.The high stability of the dextran layer often allows several hundred binding cycles to be run over an immobilized ligand.
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
Equilibrium
Association Dissociation
Regeneration
Baseline
Time Pharmaceutical Science & Technology Today
Figure 1. Schematic diagram depicting a typical surface plasmon resonance (SPR) experiment. (a) The ligand (shown as an antibody) is immobilized on the biosensor chip surface. The analyte (which represents the antigen) passes through a microfluidic flow cell and SPR is used to monitor the change in refractive index caused as analyte accumulates on the biosensor surface. (b) Salient features of a typical sensorgram. Before injecting analyte, the baseline response should be stable. An increase in response during the association phase represents complex formation in real-time. Equilibrium is achieved when an equal number of analyte molecules associate with and dissociate from the surface at the same time. The surface can be washed and the decay rate of the complex obtained during the dissociation phase. Following regeneration, the binding response should return to the starting baseline position.
(1) the lack of labeling requirements allows for the analysis of almost all biomolecular systems; (2) these instruments are capable of characterizing binding reactions in real-time, thereby providing rapid, quantitative information about binding events; (3) the amount of complex at equilibrium is measured in the presence of unbound reactant, without disturbing the reaction equilibrium; (4) the stability of the immobilized ligand can be monitored by tracking the surface’s binding capacity and the baseline stability; (5) the technology can be used in qualitative formats, such as screening, and as a biophysical tool to examine interactions 311
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
(9) experiments can be performed using buffers containing organic solvents such as DMSO, which are often required when working with compounds that have low solubility in aqueous buffer systems.
(a)
i
ii
iii
iv
(b) i
Response (RU)
80
60
H3C H3C C=O
ii H3C
40 O
Progesterone
20 iii 0 0
20
40
60
80
Time (s) Pharmaceutical Science & Technology Today
Figure 2. Example of small-molecule analysis using a surface plasmon resonance biosensor. (a) Schematic of antibody–antigen assay: (i) antiprogesterone, (ii) anti-testosterone, and (iii) anti-mouse Fc, (from Pierce, Rockford, IL, USA) were immobilized on flow cells 1, 2 and 3 of a BIACORE 2000 sensor chip; flow cell 4 was left blank to serve as a reference. (b) Response data collected for progesterone (2 mM) (from Sigma Chemical, St Louis, MO, USA) injected over the antibody surfaces (running buffer contained 10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 3% DMSO, pH 7.4, at 258C and flowed at 100 ml min21). The kinetic rate and affinity constants determined from fitting the sensorgrams to a simple A1B5AB reaction model are: (i) for anti-progesterone: ka 5 2.13105 M21 s21, kd 5 3.631023 s21, KD 5 17 nM, and (ii) for anti-testosterone: ka 5 1.03105 M21 s21, kd 5 1.431021 s21, KD 5 1400 nM. Abbreviation: RU, resonance unit.
having a wide range of affinity constants (KD 5 1 mM–1 pM) and kinetic rate constants13,14 (ka 5 103–108 M21 s21, kd 5 1026 –1 s21); (6) several versions of BIACORE technology (BIACORE 3000, BIACORE 2000 and BIACORE X) can monitor simultaneously interactions occurring at multiple surfaces, providing immediate and direct comparison between different immobilized ligands and control surfaces; (7) automated instruments provide reproducible analyses while minimizing user time and effort; (8) the amounts of material required for typical analyses are low (generally requiring ~1 mg of a protein ligand to make a single surface); and 312
Target characterization The use of SPR biosensors in pharmaceutical discovery, especially the characterization of target macromolecules and biopharmaceuticals, is well documented15–20. SPR biosensors can monitor directly a molecule’s binding activity, which relates to its function. Determining kinetic and thermodynamic parameters of ligand–receptor interactions provides useful information (e.g. immobilization conditions and sample concentrations) for setting up ELISAs, immunoprecipitations and high-throughput assays. SPR biosensors also provide a rapid and reliable method to assess the quality of targets that are destined for further applications, helping guarantee that only functional targets will enter the screening phase of drug discovery. Small-molecule detection The sensitivity of SPR detectors is dependent on the refractive index of the analyte, which for proteins in general is proportional to molecular mass. In the past, this has limited standard SPR analyses to the study of analytes having molecular masses of several thousand daltons. Recent improvements in biosensor instrument hardware, experimental design21 and data processing13 have made it possible to detect routinely the binding of low-molecular-mass analytes (,500 Da). To illustrate the capabilities of detecting small molecules using SPR biosensors, we present data from experiments involving steroid hormones binding to immobilized antibodies (Fig. 2). In this assay, three monoclonal antibodies (anti-progesterone, anti-testosterone and anti-mouse Fc) were immobilized to similar levels (~15 000 resonance units, RU) on three separate flow cells within a BIACORE 2000 (Fig. 2a). The fourth flow cell remained blank as a control surface to correct for bulk refractive index changes and any non-specific binding of the analyte to the biosensor surface. Figure 2b illustrates the response data for the binding of progesterone (MWt 315 Da) to the different antibody surfaces. Several characteristics of the binding data are noteworthy: (1) The response caused by progesterone binding to the antiprogesterone antibody is readily observable (75 RU in response). This reaction reaches equilibrium rapidly and the bound progesterone dissociates slowly from the anti-progesterone surface. (2) By contrast, progesterone binds weakly to the antitestosterone surface, as indicated by a lower response at equilibrium (40 RU) and a much faster return to baseline during the dissociation phase. This is expected because the hormone recognition sites in the two antibodies are dissimilar.
PSTT Vol. 3, No. 9 September 2000
(3) Progesterone does not bind to the anti-mouse Fc antibody surface, indicating that its binding to the anti-hormone surface is specific. (4) This example illustrates how multiple surfaces can be used to compare simultaneously a compound’s selectivity for different target ligands. (5) In this case, the ligand–analyte mass ratio per binding site is ∼250, demonstrating that it is possible to observe small molecules binding to large macromolecules. (6) Estimates of binding kinetics can be obtained from fitting the biosensor data to simple interaction models (Fig. 2). In the present study, the association and dissociation rate constants of progesterone binding to the hormone antibodies differ by two- and 40-fold, respectively, resulting in an 80-fold difference in affinity. Recently published studies from other research groups further demonstrate the applicability of SPR in characterizing both high- and low-affinity small-molecule interactions22–25. For example, Kampranis et al. measured the kinetics of coumarin and cyclothialidine drugs (MWt ,700 Da) binding to wild-type and mutant DNA gyrases24. The study of these high-affinity drug–protein interactions (KD 5 1–150 nM) revealed which residues on gyrase were critical for drug binding. Analysing the opposite end of the affinity spectrum, Strandh et al. demonstrated the ability to monitor low-affinity interactions (KD ~1 mM) of saccharides (MWt ,800 Da) and antibodies using SPR25. The ability to detect weak binders is particularly useful in early screening, in which the affinity of initial leads is generally low. High-resolution screening The lack of labeling requirements and the high information content available with SPR technology makes it a particularly powerful tool in small-molecule screening. Current BIACORE instruments require sample volumes of 50–150 ml and can analyse samples in a 96-well plate format. Sample throughput is dependent on the assay, but typically ranges from 100–400 assays per day.This throughput is ideal for secondary screening applications or for characterizing combinatorial libraries. Because the assay is insensitive to non-binders, it is also possible to screen mixtures of compounds, which increases sample throughput. By providing access to binding kinetics, SPR can generate new insight into how compounds identified from primary screens bind to target molecules. Kinetics provide detailed information regarding complex formation that is not available from equilibrium-based assays. For example, two compounds with similar affinities can have dramatically different kinetic profiles (one compound might bind to a target with rapid association and dissociation rate constants whereas another
research focus
reviews
might associate and dissociate more slowly).The high-resolution kinetic information available from SPR biosensors presents the opportunity to optimize lead compounds based on binding rates, which are likely to be important aspects of a drug’s potency. An excellent example of using SPR biosensors in smallmolecule screening was reported by Markgren et al.26 This group developed a rapid and informative assay to monitor a panel of small-molecule inhibitors binding to HIV proteinase. Proteinase-binding compounds were identified and their relative binding capabilities were ranked based on their respective association and dissociation rates. SPR biosensor technology offers an additional advantage as a screening platform in that it can be used in complementary assay formats. For example, Karlsson et al. described three SPR biosensor assays that are useful in the secondary screening of drug candidates against thrombin: direct binding, surface competition and solution competition 27. The direct binding and surface competition assays yielded kinetic and affinity information, and the solution competition assay yielded complementary solution affinity data. This multifaceted approach to the analysis of thrombin inhibitors is applicable to protein and other small-molecule systems. Karlsson et al. also described a simple calibration procedure that corrects for artifacts relating to varying DMSO concentrations in the sample and running buffers, making it possible to accurately characterize weakaffinity small-molecule interactions. Assay validity When applying SPR in a routine screening laboratory, several researchers have described the advantages of the biosensor technology over enzyme immunoassays (EIAs) in the examination of biomolecular systems28,29 and the detection of small molecules30. Crooks et al. determined that the validation parameters (e.g. limits of detection and determination) for SPR biosensors were better than those obtained from EIA 30. Also, the false positive rates using the biosensor were three- to tenfold lower than from EIA. Many false positives in a multicomponent EIA occur when the compound binds to the reporter enzyme rather than to the target, which is avoided in SPR analysis.The biosensor also identified several binders that were missed by EIA.The authors further noted that information on a compound’s binding potential is available almost instantaneously using the SPR biosensor, whereas data accumulation from EIA often requires several hours. Non-specific interactions Although the previous examples highlight the potential of using SPR biosensors to characterize specific target–small-molecule interactions, the biosensor can also be useful in examining 313
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
(b) Response (kRU)
(a)
2
1
0 0
100
200
300
Time (s) Pharmaceutical Science & Technology Today
Figure 3. Non-specific binding to target proteins. (a) Schematic depiction of a compound binding non-specifically to multiple sites on an enzyme. (b) Response data from an enzyme–inhibitor interaction in which nonspecific binding was observed. The compound was injected at 10, 3.3, 1.1, 0.37, and 0.12 mM at a rate of 100 ml min21. Given the mass of the compound and target protein, the expected maximal response for a single-site interaction would be ~20 resonance units (RU), which is indicated by the dotted line. The much larger than expected binding responses and inability to saturate the surface suggest that the compound is binding non-specifically to the target protein.
non-specific interactions. For example, while the binding of compounds selected from primary enzymatic screens were being analyzed, the biosensor showed that several of the compounds bound the target non-specifically. As shown in Fig. 3,
50
Response (RU)
40 O O 30 HO 20
O
Warfarin
10 0 10 0
50
100
150
Time (s) Pharmaceutical Science & Technology Today
Figure 4. Warfarin binding to human serum albumin. Warfarin (308 Da) at concentrations of 100, 50, 25, 12.5, 6.3, 3.1, 1.1 and 0 mM were injected at a flow rate of 100 ml min21 over a surface containing 6000 resonance units (RU) of immobilized HSA in 20 mM Na2HPO4, 150 mM NaCl, 3% DMSO, pH 7.4 at 258C.
314
the target enzyme from the primary screen was immobilized on the flow cell surface and the binding of a potential inhibitor was monitored. Given the amount of enzyme bound to the surface and the enzyme:inhibitor mass ratio, the response signals generated from this interaction were far greater than what would be expected for stoichiometric enzyme–inhibitor complex formation (~20 RU; Fig. 3b). These data suggest that the compounds bound to numerous sites on the enzyme. Using isothermal titration calorimetry, it was confirmed that at these concentrations the compounds were indeed binding nonspecifically to the target at a ratio of .50:1 (D. Myszka et al., unpublished).This example illustrates an important application of the SPR biosensor as a tool to identify compounds that bind promiscuously to a target. Routine assays can be configured to screen through libraries to identify compounds with these undesirable characteristics. General pharmacological assessment In addition to the benefits of characterizing specific drug–target interactions, SPR biosensors can play an equally significant role in the general pharmacological assessment of drug candidates. Typically, a substantial amount of time and money are spent optimizing compounds for a target before examination of their physiological properties. SPR biosensor technology provides a means to acquire physiologically relevant information on a larger selection of compounds earlier in the optimization process. Routine ADME applications are being developed for SPR biosensors to determine a compound’s propensity to bind to carrier plasma proteins (e.g. serum albumin), pass through membranes and activate destructive pathways (e.g. cytochrome P450). Figure 4 demonstrates the potential to collect high-resolution binding data on small compounds interacting with human serum albumin (HSA). In this example, a series of warfarin samples were injected over an HSA surface. The binding responses are easily observable and give the expected dose dependence. By fitting the equilibrium response data to a simple binding isotherm, it was determined that HSA binds warfarin with an affinity of 17 mM (D. Myszka et al., unpublished). This affinity corresponds to a value of 97.5% bound, which agrees well with the published value (98%) measured by traditional equilibrium assays31. In a similar assay, Frostell-Karlsson et al. showed that it is possible to rank a panel of drugs as high, intermediate or low HSA binders using single drug concentrations32. In addition, the number and specificity of drugbinding sites on HSA can be determined through competition experiments with known binders. The assay described above mimics drug uptake by one component of serum, but it can be expanded to include other serum proteins, such as a1-glycoprotein32 and gamma globulins, to characterize further the
research focus
potential in vivo serum binding of small-molecule drugs. This assay will provide researchers with the ability to collect preliminary ADME data on .100 compounds per day. General applications are also being developed to assess the membrane permeability of compounds. The recent release of hydrophobic and lipophilic sensor chips (referred to as HPA and L1) for BIACORE allows for the construction of stable membrane surfaces within the flow cell. The ability to detect small molecules binding to synthetic or natural membranes captured on these surfaces has been demonstrated33–35. Danelian et al. mimicked drug adsorption to the human intestine by capturing a phospholipid preparation on an L1 chip surface and monitoring the binding response of compounds flowed over the surface35. Based on their relative responses, compounds could be divided into classes of high, medium and low membrane permeability. Applications are also being developed that use SPR biosensors to characterize the activation of metabolic pathways such as cytochrome P450. For example, Oyama et al. measured the amount of Saccharomyces cerevisiae cytochrome P450 mRNA induced by atrazine36.These researchers used immobilized DNA oligomers to measure the change in cytochrome P450 mRNA concentration through direct hybridization. They reported that this detection method proved to be far more sensitive than traditional methods, which quantitate the change in the amount of protein. The relative ease of use of SPR assays, in addition to their general applicability, makes them ideal for collecting preliminary information on the pharmacological activity of drug candidates earlier in the optimization process. Proteomics The past five years have seen the development of sophisticated methods of screening large libraries of compounds. Today’s automated technologies are capable of screening libraries containing 100 000 compounds in a week or less. Because screening is not a bottleneck in the drug discovery process anymore, attention is now focusing on identifying new targets to feed into the screening till37. Although genome database mining provides access to putative targets, it does not provide a complete picture of the target system. In the case of ligand–receptor systems for example, in silico studies cannot identify the ligands of ‘orphan’ receptors. The focus then shifts from genomics to proteomics. Because typical bioassays are not possible when the orphan’s activity is unknown, measuring the binding between the receptor and a host of plausible binding partners is often the first step in characterizing a new target system. The SPR biosensor’s ability to detect and quantify analytes from complex fluids, cell lysates and conditioned media makes this technology a logical method for orphan ligand screening38.
reviews
0.5 mm
PSTT Vol. 3, No. 9 September 2000
2.4 mm Pharmaceutical Science & Technology Today
Figure 5. Prototype microarray protein chip. The picture shows 64 spots of human serum albumin immobilized within one flow cell of a BIACORE chip, as visualized with a microscope. Protein was immobilized using standard amine-coupling chemistry applied via ink-jet technology. The visible spots represent dried protein deposits after immobilization. The spots average 0.1 mm in diameter and are spaced 0.125 mm apart (center to center); the flow cell is 2.4 mm long and 0.5 mm wide.
The popularity of implementing biosensors in the detection and characterization of binding partners is increasing. Some of the earliest SPR biosensor orphan receptor studies were carried out on kinases. For example, Lackmann et al. developed a method to identify and monitor the purity of kinase ligands39. Sakano et al. reported the use of SPR in screening cell lines to identify the natural ligand for HTK40. They immobilized HTK fusion proteins on the sensor surface and flowed concentrated conditioned media from various cell lines across the surface. Yancopoulos and co-workers used SPR in a similar screen for ligands of a tyrosine kinase orphan receptor and described in detail specific approaches for the discovery of secreted and membrane-bound ligands41. Furthermore, performing biosensor experiments in tandem with mass spectrometry provides immediate molecular weight identification of the analytes that bind to the immobilized receptor42–46. SPR-based biosensor technologies continue to advance. Using ink-jet technology combined with standard immobilization chemistries, scientists at Biacore AB have recently demonstrated the possibility of immobilizing ligands at multiple spots within a single flow cell. Figure 5 illustrates an example of a prototypical flow cell containing 64 individual spots of immobilized HSA. SPR detectors can be aligned to collect high-resolution binding information from each spot simultaneously and independently. One example of an application of these surfaces in proteomics would involve using an array of antibodies to characterize protein expression patterns within cells. Although the application of SPR arrays is very much in its infancy, this prototype experiment demonstrates how in the future it might be feasible to rapidly examine a large matrix of analyte–ligand interactions. Conclusions Over the past ten years, SPR biosensor technology has become a standard biophysical tool used to characterize macromolecular 315
reviews
research focus
targets and biopharmaceutical products. The high information content and lack of labeling requirements makes SPR technology an excellent complement to high-throughput screening methods. New biosensor applications focused on small-molecule detection are providing novel information about drug candidates during secondary screening and lead optimization. The versatility of the biosensor makes it amenable to supporting many stages of the drug discovery process, including proteomics and early ADME. As the pharmaceutical industry becomes aware of the technology’s potential, the SPR biosensor will become an indispensable tool in drug discovery.
PSTT Vol. 3, No. 9 September 2000
15
Hudson, P.J. (1999) Recombinant antibody constructs in cancer therapy.
16
Thomas, C. and Surolia, A. (1999) Kinetics of the interaction of
Curr. Opin. Immunol. 11, 548–557 endotoxin with polymyxin B and its analogs: a surface plasmon resonance analysis. FEBS Lett. 445, 420–424 17
Mangold, U. et al. (1999) Identification and characterization of potential new therapeutic targets in inflammatory and autoimmune diseases. Eur. J. Biochem. 266, 1184–1191
18
Choulier, L. et al. (1999) Kinetic analysis of the effect on Fab binding of
19
Atwell, S. et al. (1997) Structural plasticity in a remodeled
identical substitutions in a peptide and its parent. Biochemistry 38, 3530–3537 protein–protein interface. Science 278, 1125–1128
References 1
20
Myszka, D.G. (1999) Survey of the 1998 optical biosensor literature.
interleukin-2 receptor complexes created on an optical biosensor surface.
J. Mol. Recog. 12, 390–408 2
Wong, R.L. et al. (1997) Validation parameters for a novel biosensor assay
Protein Sci. 5, 2468–2478 21
which simultaneously measures serum concentrations of a humanized
3
van Regenmortel, M.H.V. (1999) Biosensors: into the 21st century.
4
Canziani, G. et al. (1999) Exploring biomolecular recognition using
295, 268–294 22
interactions. Methods 20, 319–328 23
HIV-1 protease and inhibitors using optical biosensor technology. Anal.
Malmqvist, M. (1999) BIACORE: an affinity biosensor system for
Biochem. 279, 71–78
characterization of biomolecular interactions. Biochem. Soc.Trans. 27,
24
Rich, R.L. and Myszka, D.G. Survey of the 1999 surface plasmon
8
technology. J. Mol. Recog. 11, 188–190 26
HIV-1 proteinase using optical biosensor technology. Anal. Biochem. 265,
Lowe, P.A. et al. (1998) New approaches for the analysis of molecular
340–350
recognition using the IAsys evanescent wave biosensor. J. Mol. Recog. 11,
27
between thrombin and thrombin inhibitors. Anal. Biochem. 278, 1–13 28
71, 801–805 Kondoh, M. et al. (1999) Apoptosis induction via microtubule disassembly Kukanskis, K. et al. (1999) Detection of DNA hybridization using the
123, 2755–2757 31
plasmon resonance detection applied to kinetic, binding site and concentration analysis. J. Chromatogr. 597, 397–410 13
32
Rich, R.L. and Myszka, D.G. (2000) Advances in surface plasmon
316
Myszka, D.G. et al. (1998) Extending the range of rate constants available
Ferrer, J. et al. (1998) The binding of benzopyranes to human serum albumin. A structure–affinity study. J. Protein Chem. 17, 115–119 Frostell-Karlsson, A. et al. (2000) Biosensor analysis of the interaction between immobilized human serum albumin and drug compounds for
resonance biosensor analysis. Curr. Opin. Biotechnol. 11, 54–61 14
Crooks, S.R. et al. (1998) Immunobiosensor – an alternative to enzyme immunoassay screening for residues of two sulfonamides in pigs. Analyst
Biochem. 274, 7–17 Fägerstam, L.G. et al. (1992) Biospecific interaction analysis using surface
Swanson, S.J. et al. (1999) Applications for the new electrochemiluminescent (ELC) and biosensor technologies. Dev. Biol. Stand. 97, 135–147
30
Texas Instruments Inc.TISPR-1 surface plasmon resonance biosensor. Anal. 12
Hoyer-Hansen, G. et al. (2000) Loss of ELISA specificity due to biotinylation of monoclonal antibodies. J. Immunol. Methods 235, 91–99
29
by an antitumor compound, pironetin. Biochem. J. 340, 411–416 11
Karlsson, R. et al. (2000) Biosensor analysis of drug–target interactions: direct and competitive binding assays for investigation of interactions
Wink,T. et al. (1999) Interaction between plasmid DNA and cationic polymers studied by surface plasmon resonance spectroscopy. Anal. Chem.
10
Markgren, P-O. et al. (1998) Screening of compounds interacting with
620–627
194–199 9
Strandh, M. et al. (1998) Studies of interactions with weak affinities and low-molecular-weight compounds using surface plasmon resonance
Jönsson, U. et al. (1991) Real-time biospecific interaction analysis using surface plasmon resonance and a sensor chip technology. BioTechniques 11,
Kampranis, S.C. et al. (1999) Probing the binding of coumarins and cyclothialidines to DNA gyrase. Biochemistry 38, 1967–1976
25
resonance biosensor literature. J. Mol. Recog. (in press) 7
Markgren, P-O. et al. (2000) Kinetic analysis of the interaction between
optical biosensors. Methods 19, 253–269
335–340 6
Adamczyk, M. et al. (2000) Application of surface plasmon resonance toward studies of low-molecular-weight antigen–antibody binding
Biochem. Soc.Trans. 27, 329–330
5
Morton,T.A. and Myszka, D.G. (1998) Kinetic analysis of macromolecular interactions using surface plasmon resonance biosensors. Methods Enzymol.
monoclonal antibody and detects induced antibodies. J. Immunol. Methods 209, 1–15
Myszka, D.G. et al. (1996) Kinetic analysis of ligand binding to
prediction of HSA binding levels. J. Med. Chem. 43, 1986–1992 33
Cooper, M.A. and Williams, D.H. (1999) Binding of glycopeptide
from BIACORE: interpreting mass transport-influences binding data.
antibiotics to a model of a vancomycin-resistant bacterium. Chem. Biol. 6,
Biophys. J. 75, 583–594
891–899
research focus
PSTT Vol. 3, No. 9 September 2000
34
Surdo, P.L. et al. (1999) Structural and biochemical evaluation of the
protein-tyrosine kinase HTK expressed in immature hematopoietic cells.
interaction of the phosphatidylinositol 3-kinase p85a Src homology 2 domains with phosphoinositides and inositol polyphosphates. J. Biol. Chem.
Oncogene 13, 813–822 41
22, 15678–15685 35
Danelian, E. et al. (2000) SPR biosensor studies of the direct interaction
42
femtomole to subfemtomole levels. Anal. Chem. 71, 2858–2865
Oyama, M. (2000) Highly sensitive detection of toxic chemicals by
43
measuring the quantity of induced P450 mRNAs based on SPR. Biosensors,
(BIA/MS). J. Mol. Recog. 12, 77–93
Hensley, P. and Myszka, D.G. (2000) Analytical biotechnology: sorting
44
needles and haystacks. Curr. Opin. Biotechnol. 11, 9–12 Williams, C. (2000) Biotechnology matchmaking: screening orphan
Trends Biotechnol. 18, 45–48 45
Krone, J.R. et al. (1997) BIA/MS: interfacing biomolecular interaction
46
Sönksen, C.P. et al. (1998) Combining MALDI mass spectrometry and
Lackmann, M. et al. (1996) Purification of a ligand for the EPH-like receptor HEK using a biosensor-based affinity detection approach.
40
Williams, C. and Addona,T.A. (2000) The integration of SPR biosensor with mass spectrometry: possible applications for proteomics analysis.
ligands and receptors. Curr. Opin. Biotechnol. 11, 42–46 39
Nelson, R.W. and Krone, J.R. (1999) Advances in surface plasmon resonance biomolecular interaction analysis mass spectrometry
Seattle, WA, USA
38
Nelson, R.W. et al. (1999) BIA/MS of epitope-tagged peptides directly from E. coli lysate: multiplex detection and protein identification at low-
absorbed in humans. J. Med. Chem. 43, 2083–2086
37
Davis, S. et al. (1996) Isolation of angiopoietin-1, a ligand for the TIE2 receptor, by secretion-trap expression cloning. Cell 87, 1161–1169
between 27 drugs and a liposome surface: correlation with fraction 36
reviews
analysis with mass spectrometry. Anal. Biochem. 244, 124–132
Proc. Natl.Acad. Sci. U. S.A. 93, 2523–2527
biomolecular interaction analysis using a biomolecular interaction
Sakano, S. et al. (1996) Characterization of a ligand for receptor
analysis instrument. Anal. Chem. 70, 2731–2736
In the October issue of Drug Discovery Today… Editorial – Poetry and verse: an ideal medium for scientific communication? by Raymond Rowe Update – latest news and views Endothelin-receptor antagonists: current and future perspectives Abhijit Ray, Laxminarayan G. Hegde, Anita Chugh and Jang B. Gupta Development of novel nucleoside analogues for use against drug resistant strains of HIV-1 Robert F. Rando and Nghe Nguyen-Ba Predicting human safety: screening and computational approaches Dale E. Johnson and Grushenka H.I. Wolfgang
Monitor – new bioactive molecules, combinatorial chemistry, invited profile Products
317
research focus
PSTT Vol. 3, No. 9 September 2000
Implementing surface plasmon resonance biosensors in drug discovery David G. Myszka and Rebecca L. Rich Recent improvements in instrument hardware, experimental design and data processing have made it possible to use surface plasmon resonance (SPR) biosensor technology in the discovery and development of smallmolecule drugs. The key features of SPR biosensors (i.e. real-time binding analysis and lack of labeling requirements) make this technology suitable for a wide range of applications. Current instruments have a throughput of ~100–400 assays per day, providing a complement to secondary screening. The ability to collect kinetic data on compounds binding to therapeutic targets yields new information for lead optimization. Small-molecule analysis and emerging applications in the areas of ADME (adsorption, distribution, metabolism and excretion) and proteomics have SPR biosensors poised to play a significant role in the pharmaceutical industry.
*David G. Myszka and Rebecca L. Rich Center for Biomolecular Interaction Analysis University of Utah School of Medicine 4A417 50 North Medical Drive Salt Lake City UT 84132 USA *tel: 11 801 585 5358 fax: 11 801 585 2978 e-mail: [email protected]
▼ New discoveries are often driven by the appli-
cation of new technologies.An excellent example of this is the discovery of Jupiter’s moons by Galileo’s application of the telescope. Nowhere is the application of new technologies more important than in drug discovery, especially considering the fact that the average drug takes 12–15 years to develop and costs greater than US$750 million to produce. Pharmaceutical companies are always searching for technologies that will lower development costs and decrease the lead-time to market. Surface plasmon resonance (SPR; see Glossary in Box 1) based biosensors can be implemented in many stages of the drug discovery process to increase throughput and lower costs, while generating a wealth of new information about drug candidates. SPR biosensor technology SPR biosensors measure the quantity of a complex formed between two molecules in real-time
310
without the need for fluorescent or radioisotopic labels. This lack of labeling requirement makes these instruments amenable to characterizing unmodified biopharmaceuticals, studying the interaction of drug candidates with macromolecular targets and identifying binding partners during ligand fishing experiments. There are several examples in which SPR biosensors have been used to characterize the mechanistic details of biomolecular reactions1–6; they are therefore rapidly becoming a mainstay of the modern biophysical analysis laboratory. The best indicator of the success of SPR technology is the growing number of commercially available instruments. Five companies currently manufacture a variety of optical biosensor platforms. Biacore AB (Uppsala, Sweden) released the first commercial SPR instrument in 1990 (Ref. 7). Currently, the most advanced platform, BIACORE 3000, incorporates increased sensitivity and a smaller flow cell compared with earlier models, as well as online data subtraction and microsample
Box 1. Glossary ADME Adsorption, distribution, metabolism and excretion DMSO Dimethyl sulfoxide EDTA Ethylenediaminetetraacetic acid EIA Enzyme interaction analysis ELISA Enzyme-linked immunoassay HSA Human serum albumin HEPES N-(2-hydroxyethyl)piperazine-N9-2ethanesulfonic acid HIV Human immunodeficiency virus HTK Human tyrosine kinase ka Association rate constant kd Dissociation rate constant KD Equilibrium dissociation constant RU Resonance unit SPR Surface plasmon resonance
1461-5347/00/$ – see front matter ©2000 Elsevier Science Ltd. All rights reserved. PII: S1461-5347(00)00288-1
recovery. Based on the scientific literature, BIACORE instruments are the most commonly used commercial SPR technology (approximately 90% of the 1998 and 1999 commercial SPR biosensor publications cite the use of these instruments1,6). Affinity Sensors (Franklin, MA, USA) manufactures the IAsys line of instruments, which is a cuvette-based system that uses evanescent-wave technology8. Windsor Scientific Ltd (Berks, UK) markets the IBIS system, which can be configured as a flow- or cuvette-based instrument9. Nippon Laser & Electronics Lab’s (Hokkaido, Japan) SPR-CELLIA systems10 and Texas Instruments’ (Dallas, TX, USA) integrated SPR detector Spreeta11 can be configured for a variety of applications.
(a)
The advantages of SPR biosensors Compared with many other interaction technologies, SPR biosensors exhibit several distinct advantages for characterizing molecular interactions13:
De
t
h Lig
tec
tor
θ Antibody Antigen
Response
(b) How SPR biosensors work To illustrate a typical SPR biosensor assay, the format used to characterize an antibody–antigen interaction is presented in the context of the BIACORE system. The antibody is covalently immobilized on the biosensor surface and is referred to as the ligand (Fig. 1a).The antigen (referred to as the analyte) passes over the surface through a microfluidic flow cell. SPR detectors monitor the change in the refractive index of the solvent layer near the surface induced by association and dissociation of the analyte– ligand complex formation. For a more detailed description of SPR, the reader is referred to the article by Fägerstam et al.12 An example of the real-time-derived response data, which is referred to as a sensorgram, is illustrated in Fig. 1b. During the association phase of the interaction, analyte binds to the immobilized ligand, generating an increase in response. The response intensity levels off over time as equilibrium between bound and free analyte is achieved. To monitor dissociation, the instrument automatically switches back to running buffer, making it possible to collect information about the stability of the complex. At this time, any remaining analyte–ligand complexes can usually be disrupted using mild acidic or basic washing conditions, yielding a fully functional surface that can be sampled repeatedly. The number of binding cycles on a given surface is largely dependent on the stability of the immobilized ligand. One of the most essential features of an SPR biosensor is the interface between the sensor surface and the immobilized ligand. BIACORE technology utilizes a patented carboxymethyl dextran chip, which minimizes non-specific binding of biomolecules to the surface and increases the binding capacity of the surface.The high stability of the dextran layer often allows several hundred binding cycles to be run over an immobilized ligand.
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
Equilibrium
Association Dissociation
Regeneration
Baseline
Time Pharmaceutical Science & Technology Today
Figure 1. Schematic diagram depicting a typical surface plasmon resonance (SPR) experiment. (a) The ligand (shown as an antibody) is immobilized on the biosensor chip surface. The analyte (which represents the antigen) passes through a microfluidic flow cell and SPR is used to monitor the change in refractive index caused as analyte accumulates on the biosensor surface. (b) Salient features of a typical sensorgram. Before injecting analyte, the baseline response should be stable. An increase in response during the association phase represents complex formation in real-time. Equilibrium is achieved when an equal number of analyte molecules associate with and dissociate from the surface at the same time. The surface can be washed and the decay rate of the complex obtained during the dissociation phase. Following regeneration, the binding response should return to the starting baseline position.
(1) the lack of labeling requirements allows for the analysis of almost all biomolecular systems; (2) these instruments are capable of characterizing binding reactions in real-time, thereby providing rapid, quantitative information about binding events; (3) the amount of complex at equilibrium is measured in the presence of unbound reactant, without disturbing the reaction equilibrium; (4) the stability of the immobilized ligand can be monitored by tracking the surface’s binding capacity and the baseline stability; (5) the technology can be used in qualitative formats, such as screening, and as a biophysical tool to examine interactions 311
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
(9) experiments can be performed using buffers containing organic solvents such as DMSO, which are often required when working with compounds that have low solubility in aqueous buffer systems.
(a)
i
ii
iii
iv
(b) i
Response (RU)
80
60
H3C H3C C=O
ii H3C
40 O
Progesterone
20 iii 0 0
20
40
60
80
Time (s) Pharmaceutical Science & Technology Today
Figure 2. Example of small-molecule analysis using a surface plasmon resonance biosensor. (a) Schematic of antibody–antigen assay: (i) antiprogesterone, (ii) anti-testosterone, and (iii) anti-mouse Fc, (from Pierce, Rockford, IL, USA) were immobilized on flow cells 1, 2 and 3 of a BIACORE 2000 sensor chip; flow cell 4 was left blank to serve as a reference. (b) Response data collected for progesterone (2 mM) (from Sigma Chemical, St Louis, MO, USA) injected over the antibody surfaces (running buffer contained 10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 3% DMSO, pH 7.4, at 258C and flowed at 100 ml min21). The kinetic rate and affinity constants determined from fitting the sensorgrams to a simple A1B5AB reaction model are: (i) for anti-progesterone: ka 5 2.13105 M21 s21, kd 5 3.631023 s21, KD 5 17 nM, and (ii) for anti-testosterone: ka 5 1.03105 M21 s21, kd 5 1.431021 s21, KD 5 1400 nM. Abbreviation: RU, resonance unit.
having a wide range of affinity constants (KD 5 1 mM–1 pM) and kinetic rate constants13,14 (ka 5 103–108 M21 s21, kd 5 1026 –1 s21); (6) several versions of BIACORE technology (BIACORE 3000, BIACORE 2000 and BIACORE X) can monitor simultaneously interactions occurring at multiple surfaces, providing immediate and direct comparison between different immobilized ligands and control surfaces; (7) automated instruments provide reproducible analyses while minimizing user time and effort; (8) the amounts of material required for typical analyses are low (generally requiring ~1 mg of a protein ligand to make a single surface); and 312
Target characterization The use of SPR biosensors in pharmaceutical discovery, especially the characterization of target macromolecules and biopharmaceuticals, is well documented15–20. SPR biosensors can monitor directly a molecule’s binding activity, which relates to its function. Determining kinetic and thermodynamic parameters of ligand–receptor interactions provides useful information (e.g. immobilization conditions and sample concentrations) for setting up ELISAs, immunoprecipitations and high-throughput assays. SPR biosensors also provide a rapid and reliable method to assess the quality of targets that are destined for further applications, helping guarantee that only functional targets will enter the screening phase of drug discovery. Small-molecule detection The sensitivity of SPR detectors is dependent on the refractive index of the analyte, which for proteins in general is proportional to molecular mass. In the past, this has limited standard SPR analyses to the study of analytes having molecular masses of several thousand daltons. Recent improvements in biosensor instrument hardware, experimental design21 and data processing13 have made it possible to detect routinely the binding of low-molecular-mass analytes (,500 Da). To illustrate the capabilities of detecting small molecules using SPR biosensors, we present data from experiments involving steroid hormones binding to immobilized antibodies (Fig. 2). In this assay, three monoclonal antibodies (anti-progesterone, anti-testosterone and anti-mouse Fc) were immobilized to similar levels (~15 000 resonance units, RU) on three separate flow cells within a BIACORE 2000 (Fig. 2a). The fourth flow cell remained blank as a control surface to correct for bulk refractive index changes and any non-specific binding of the analyte to the biosensor surface. Figure 2b illustrates the response data for the binding of progesterone (MWt 315 Da) to the different antibody surfaces. Several characteristics of the binding data are noteworthy: (1) The response caused by progesterone binding to the antiprogesterone antibody is readily observable (75 RU in response). This reaction reaches equilibrium rapidly and the bound progesterone dissociates slowly from the anti-progesterone surface. (2) By contrast, progesterone binds weakly to the antitestosterone surface, as indicated by a lower response at equilibrium (40 RU) and a much faster return to baseline during the dissociation phase. This is expected because the hormone recognition sites in the two antibodies are dissimilar.
PSTT Vol. 3, No. 9 September 2000
(3) Progesterone does not bind to the anti-mouse Fc antibody surface, indicating that its binding to the anti-hormone surface is specific. (4) This example illustrates how multiple surfaces can be used to compare simultaneously a compound’s selectivity for different target ligands. (5) In this case, the ligand–analyte mass ratio per binding site is ∼250, demonstrating that it is possible to observe small molecules binding to large macromolecules. (6) Estimates of binding kinetics can be obtained from fitting the biosensor data to simple interaction models (Fig. 2). In the present study, the association and dissociation rate constants of progesterone binding to the hormone antibodies differ by two- and 40-fold, respectively, resulting in an 80-fold difference in affinity. Recently published studies from other research groups further demonstrate the applicability of SPR in characterizing both high- and low-affinity small-molecule interactions22–25. For example, Kampranis et al. measured the kinetics of coumarin and cyclothialidine drugs (MWt ,700 Da) binding to wild-type and mutant DNA gyrases24. The study of these high-affinity drug–protein interactions (KD 5 1–150 nM) revealed which residues on gyrase were critical for drug binding. Analysing the opposite end of the affinity spectrum, Strandh et al. demonstrated the ability to monitor low-affinity interactions (KD ~1 mM) of saccharides (MWt ,800 Da) and antibodies using SPR25. The ability to detect weak binders is particularly useful in early screening, in which the affinity of initial leads is generally low. High-resolution screening The lack of labeling requirements and the high information content available with SPR technology makes it a particularly powerful tool in small-molecule screening. Current BIACORE instruments require sample volumes of 50–150 ml and can analyse samples in a 96-well plate format. Sample throughput is dependent on the assay, but typically ranges from 100–400 assays per day.This throughput is ideal for secondary screening applications or for characterizing combinatorial libraries. Because the assay is insensitive to non-binders, it is also possible to screen mixtures of compounds, which increases sample throughput. By providing access to binding kinetics, SPR can generate new insight into how compounds identified from primary screens bind to target molecules. Kinetics provide detailed information regarding complex formation that is not available from equilibrium-based assays. For example, two compounds with similar affinities can have dramatically different kinetic profiles (one compound might bind to a target with rapid association and dissociation rate constants whereas another
research focus
reviews
might associate and dissociate more slowly).The high-resolution kinetic information available from SPR biosensors presents the opportunity to optimize lead compounds based on binding rates, which are likely to be important aspects of a drug’s potency. An excellent example of using SPR biosensors in smallmolecule screening was reported by Markgren et al.26 This group developed a rapid and informative assay to monitor a panel of small-molecule inhibitors binding to HIV proteinase. Proteinase-binding compounds were identified and their relative binding capabilities were ranked based on their respective association and dissociation rates. SPR biosensor technology offers an additional advantage as a screening platform in that it can be used in complementary assay formats. For example, Karlsson et al. described three SPR biosensor assays that are useful in the secondary screening of drug candidates against thrombin: direct binding, surface competition and solution competition 27. The direct binding and surface competition assays yielded kinetic and affinity information, and the solution competition assay yielded complementary solution affinity data. This multifaceted approach to the analysis of thrombin inhibitors is applicable to protein and other small-molecule systems. Karlsson et al. also described a simple calibration procedure that corrects for artifacts relating to varying DMSO concentrations in the sample and running buffers, making it possible to accurately characterize weakaffinity small-molecule interactions. Assay validity When applying SPR in a routine screening laboratory, several researchers have described the advantages of the biosensor technology over enzyme immunoassays (EIAs) in the examination of biomolecular systems28,29 and the detection of small molecules30. Crooks et al. determined that the validation parameters (e.g. limits of detection and determination) for SPR biosensors were better than those obtained from EIA 30. Also, the false positive rates using the biosensor were three- to tenfold lower than from EIA. Many false positives in a multicomponent EIA occur when the compound binds to the reporter enzyme rather than to the target, which is avoided in SPR analysis.The biosensor also identified several binders that were missed by EIA.The authors further noted that information on a compound’s binding potential is available almost instantaneously using the SPR biosensor, whereas data accumulation from EIA often requires several hours. Non-specific interactions Although the previous examples highlight the potential of using SPR biosensors to characterize specific target–small-molecule interactions, the biosensor can also be useful in examining 313
reviews
research focus
PSTT Vol. 3, No. 9 September 2000
(b) Response (kRU)
(a)
2
1
0 0
100
200
300
Time (s) Pharmaceutical Science & Technology Today
Figure 3. Non-specific binding to target proteins. (a) Schematic depiction of a compound binding non-specifically to multiple sites on an enzyme. (b) Response data from an enzyme–inhibitor interaction in which nonspecific binding was observed. The compound was injected at 10, 3.3, 1.1, 0.37, and 0.12 mM at a rate of 100 ml min21. Given the mass of the compound and target protein, the expected maximal response for a single-site interaction would be ~20 resonance units (RU), which is indicated by the dotted line. The much larger than expected binding responses and inability to saturate the surface suggest that the compound is binding non-specifically to the target protein.
non-specific interactions. For example, while the binding of compounds selected from primary enzymatic screens were being analyzed, the biosensor showed that several of the compounds bound the target non-specifically. As shown in Fig. 3,
50
Response (RU)
40 O O 30 HO 20
O
Warfarin
10 0 10 0
50
100
150
Time (s) Pharmaceutical Science & Technology Today
Figure 4. Warfarin binding to human serum albumin. Warfarin (308 Da) at concentrations of 100, 50, 25, 12.5, 6.3, 3.1, 1.1 and 0 mM were injected at a flow rate of 100 ml min21 over a surface containing 6000 resonance units (RU) of immobilized HSA in 20 mM Na2HPO4, 150 mM NaCl, 3% DMSO, pH 7.4 at 258C.
314
the target enzyme from the primary screen was immobilized on the flow cell surface and the binding of a potential inhibitor was monitored. Given the amount of enzyme bound to the surface and the enzyme:inhibitor mass ratio, the response signals generated from this interaction were far greater than what would be expected for stoichiometric enzyme–inhibitor complex formation (~20 RU; Fig. 3b). These data suggest that the compounds bound to numerous sites on the enzyme. Using isothermal titration calorimetry, it was confirmed that at these concentrations the compounds were indeed binding nonspecifically to the target at a ratio of .50:1 (D. Myszka et al., unpublished).This example illustrates an important application of the SPR biosensor as a tool to identify compounds that bind promiscuously to a target. Routine assays can be configured to screen through libraries to identify compounds with these undesirable characteristics. General pharmacological assessment In addition to the benefits of characterizing specific drug–target interactions, SPR biosensors can play an equally significant role in the general pharmacological assessment of drug candidates. Typically, a substantial amount of time and money are spent optimizing compounds for a target before examination of their physiological properties. SPR biosensor technology provides a means to acquire physiologically relevant information on a larger selection of compounds earlier in the optimization process. Routine ADME applications are being developed for SPR biosensors to determine a compound’s propensity to bind to carrier plasma proteins (e.g. serum albumin), pass through membranes and activate destructive pathways (e.g. cytochrome P450). Figure 4 demonstrates the potential to collect high-resolution binding data on small compounds interacting with human serum albumin (HSA). In this example, a series of warfarin samples were injected over an HSA surface. The binding responses are easily observable and give the expected dose dependence. By fitting the equilibrium response data to a simple binding isotherm, it was determined that HSA binds warfarin with an affinity of 17 mM (D. Myszka et al., unpublished). This affinity corresponds to a value of 97.5% bound, which agrees well with the published value (98%) measured by traditional equilibrium assays31. In a similar assay, Frostell-Karlsson et al. showed that it is possible to rank a panel of drugs as high, intermediate or low HSA binders using single drug concentrations32. In addition, the number and specificity of drugbinding sites on HSA can be determined through competition experiments with known binders. The assay described above mimics drug uptake by one component of serum, but it can be expanded to include other serum proteins, such as a1-glycoprotein32 and gamma globulins, to characterize further the
research focus
potential in vivo serum binding of small-molecule drugs. This assay will provide researchers with the ability to collect preliminary ADME data on .100 compounds per day. General applications are also being developed to assess the membrane permeability of compounds. The recent release of hydrophobic and lipophilic sensor chips (referred to as HPA and L1) for BIACORE allows for the construction of stable membrane surfaces within the flow cell. The ability to detect small molecules binding to synthetic or natural membranes captured on these surfaces has been demonstrated33–35. Danelian et al. mimicked drug adsorption to the human intestine by capturing a phospholipid preparation on an L1 chip surface and monitoring the binding response of compounds flowed over the surface35. Based on their relative responses, compounds could be divided into classes of high, medium and low membrane permeability. Applications are also being developed that use SPR biosensors to characterize the activation of metabolic pathways such as cytochrome P450. For example, Oyama et al. measured the amount of Saccharomyces cerevisiae cytochrome P450 mRNA induced by atrazine36.These researchers used immobilized DNA oligomers to measure the change in cytochrome P450 mRNA concentration through direct hybridization. They reported that this detection method proved to be far more sensitive than traditional methods, which quantitate the change in the amount of protein. The relative ease of use of SPR assays, in addition to their general applicability, makes them ideal for collecting preliminary information on the pharmacological activity of drug candidates earlier in the optimization process. Proteomics The past five years have seen the development of sophisticated methods of screening large libraries of compounds. Today’s automated technologies are capable of screening libraries containing 100 000 compounds in a week or less. Because screening is not a bottleneck in the drug discovery process anymore, attention is now focusing on identifying new targets to feed into the screening till37. Although genome database mining provides access to putative targets, it does not provide a complete picture of the target system. In the case of ligand–receptor systems for example, in silico studies cannot identify the ligands of ‘orphan’ receptors. The focus then shifts from genomics to proteomics. Because typical bioassays are not possible when the orphan’s activity is unknown, measuring the binding between the receptor and a host of plausible binding partners is often the first step in characterizing a new target system. The SPR biosensor’s ability to detect and quantify analytes from complex fluids, cell lysates and conditioned media makes this technology a logical method for orphan ligand screening38.
reviews
0.5 mm
PSTT Vol. 3, No. 9 September 2000
2.4 mm Pharmaceutical Science & Technology Today
Figure 5. Prototype microarray protein chip. The picture shows 64 spots of human serum albumin immobilized within one flow cell of a BIACORE chip, as visualized with a microscope. Protein was immobilized using standard amine-coupling chemistry applied via ink-jet technology. The visible spots represent dried protein deposits after immobilization. The spots average 0.1 mm in diameter and are spaced 0.125 mm apart (center to center); the flow cell is 2.4 mm long and 0.5 mm wide.
The popularity of implementing biosensors in the detection and characterization of binding partners is increasing. Some of the earliest SPR biosensor orphan receptor studies were carried out on kinases. For example, Lackmann et al. developed a method to identify and monitor the purity of kinase ligands39. Sakano et al. reported the use of SPR in screening cell lines to identify the natural ligand for HTK40. They immobilized HTK fusion proteins on the sensor surface and flowed concentrated conditioned media from various cell lines across the surface. Yancopoulos and co-workers used SPR in a similar screen for ligands of a tyrosine kinase orphan receptor and described in detail specific approaches for the discovery of secreted and membrane-bound ligands41. Furthermore, performing biosensor experiments in tandem with mass spectrometry provides immediate molecular weight identification of the analytes that bind to the immobilized receptor42–46. SPR-based biosensor technologies continue to advance. Using ink-jet technology combined with standard immobilization chemistries, scientists at Biacore AB have recently demonstrated the possibility of immobilizing ligands at multiple spots within a single flow cell. Figure 5 illustrates an example of a prototypical flow cell containing 64 individual spots of immobilized HSA. SPR detectors can be aligned to collect high-resolution binding information from each spot simultaneously and independently. One example of an application of these surfaces in proteomics would involve using an array of antibodies to characterize protein expression patterns within cells. Although the application of SPR arrays is very much in its infancy, this prototype experiment demonstrates how in the future it might be feasible to rapidly examine a large matrix of analyte–ligand interactions. Conclusions Over the past ten years, SPR biosensor technology has become a standard biophysical tool used to characterize macromolecular 315
reviews
research focus
targets and biopharmaceutical products. The high information content and lack of labeling requirements makes SPR technology an excellent complement to high-throughput screening methods. New biosensor applications focused on small-molecule detection are providing novel information about drug candidates during secondary screening and lead optimization. The versatility of the biosensor makes it amenable to supporting many stages of the drug discovery process, including proteomics and early ADME. As the pharmaceutical industry becomes aware of the technology’s potential, the SPR biosensor will become an indispensable tool in drug discovery.
PSTT Vol. 3, No. 9 September 2000
15
Hudson, P.J. (1999) Recombinant antibody constructs in cancer therapy.
16
Thomas, C. and Surolia, A. (1999) Kinetics of the interaction of
Curr. Opin. Immunol. 11, 548–557 endotoxin with polymyxin B and its analogs: a surface plasmon resonance analysis. FEBS Lett. 445, 420–424 17
Mangold, U. et al. (1999) Identification and characterization of potential new therapeutic targets in inflammatory and autoimmune diseases. Eur. J. Biochem. 266, 1184–1191
18
Choulier, L. et al. (1999) Kinetic analysis of the effect on Fab binding of
19
Atwell, S. et al. (1997) Structural plasticity in a remodeled
identical substitutions in a peptide and its parent. Biochemistry 38, 3530–3537 protein–protein interface. Science 278, 1125–1128
References 1
20
Myszka, D.G. (1999) Survey of the 1998 optical biosensor literature.
interleukin-2 receptor complexes created on an optical biosensor surface.
J. Mol. Recog. 12, 390–408 2
Wong, R.L. et al. (1997) Validation parameters for a novel biosensor assay
Protein Sci. 5, 2468–2478 21
which simultaneously measures serum concentrations of a humanized
3
van Regenmortel, M.H.V. (1999) Biosensors: into the 21st century.
4
Canziani, G. et al. (1999) Exploring biomolecular recognition using
295, 268–294 22
interactions. Methods 20, 319–328 23
HIV-1 protease and inhibitors using optical biosensor technology. Anal.
Malmqvist, M. (1999) BIACORE: an affinity biosensor system for
Biochem. 279, 71–78
characterization of biomolecular interactions. Biochem. Soc.Trans. 27,
24
Rich, R.L. and Myszka, D.G. Survey of the 1999 surface plasmon
8
technology. J. Mol. Recog. 11, 188–190 26
HIV-1 proteinase using optical biosensor technology. Anal. Biochem. 265,
Lowe, P.A. et al. (1998) New approaches for the analysis of molecular
340–350
recognition using the IAsys evanescent wave biosensor. J. Mol. Recog. 11,
27
between thrombin and thrombin inhibitors. Anal. Biochem. 278, 1–13 28
71, 801–805 Kondoh, M. et al. (1999) Apoptosis induction via microtubule disassembly Kukanskis, K. et al. (1999) Detection of DNA hybridization using the
123, 2755–2757 31
plasmon resonance detection applied to kinetic, binding site and concentration analysis. J. Chromatogr. 597, 397–410 13
32
Rich, R.L. and Myszka, D.G. (2000) Advances in surface plasmon
316
Myszka, D.G. et al. (1998) Extending the range of rate constants available
Ferrer, J. et al. (1998) The binding of benzopyranes to human serum albumin. A structure–affinity study. J. Protein Chem. 17, 115–119 Frostell-Karlsson, A. et al. (2000) Biosensor analysis of the interaction between immobilized human serum albumin and drug compounds for
resonance biosensor analysis. Curr. Opin. Biotechnol. 11, 54–61 14
Crooks, S.R. et al. (1998) Immunobiosensor – an alternative to enzyme immunoassay screening for residues of two sulfonamides in pigs. Analyst
Biochem. 274, 7–17 Fägerstam, L.G. et al. (1992) Biospecific interaction analysis using surface
Swanson, S.J. et al. (1999) Applications for the new electrochemiluminescent (ELC) and biosensor technologies. Dev. Biol. Stand. 97, 135–147
30
Texas Instruments Inc.TISPR-1 surface plasmon resonance biosensor. Anal. 12
Hoyer-Hansen, G. et al. (2000) Loss of ELISA specificity due to biotinylation of monoclonal antibodies. J. Immunol. Methods 235, 91–99
29
by an antitumor compound, pironetin. Biochem. J. 340, 411–416 11
Karlsson, R. et al. (2000) Biosensor analysis of drug–target interactions: direct and competitive binding assays for investigation of interactions
Wink,T. et al. (1999) Interaction between plasmid DNA and cationic polymers studied by surface plasmon resonance spectroscopy. Anal. Chem.
10
Markgren, P-O. et al. (1998) Screening of compounds interacting with
620–627
194–199 9
Strandh, M. et al. (1998) Studies of interactions with weak affinities and low-molecular-weight compounds using surface plasmon resonance
Jönsson, U. et al. (1991) Real-time biospecific interaction analysis using surface plasmon resonance and a sensor chip technology. BioTechniques 11,
Kampranis, S.C. et al. (1999) Probing the binding of coumarins and cyclothialidines to DNA gyrase. Biochemistry 38, 1967–1976
25
resonance biosensor literature. J. Mol. Recog. (in press) 7
Markgren, P-O. et al. (2000) Kinetic analysis of the interaction between
optical biosensors. Methods 19, 253–269
335–340 6
Adamczyk, M. et al. (2000) Application of surface plasmon resonance toward studies of low-molecular-weight antigen–antibody binding
Biochem. Soc.Trans. 27, 329–330
5
Morton,T.A. and Myszka, D.G. (1998) Kinetic analysis of macromolecular interactions using surface plasmon resonance biosensors. Methods Enzymol.
monoclonal antibody and detects induced antibodies. J. Immunol. Methods 209, 1–15
Myszka, D.G. et al. (1996) Kinetic analysis of ligand binding to
prediction of HSA binding levels. J. Med. Chem. 43, 1986–1992 33
Cooper, M.A. and Williams, D.H. (1999) Binding of glycopeptide
from BIACORE: interpreting mass transport-influences binding data.
antibiotics to a model of a vancomycin-resistant bacterium. Chem. Biol. 6,
Biophys. J. 75, 583–594
891–899
research focus
PSTT Vol. 3, No. 9 September 2000
34
Surdo, P.L. et al. (1999) Structural and biochemical evaluation of the
protein-tyrosine kinase HTK expressed in immature hematopoietic cells.
interaction of the phosphatidylinositol 3-kinase p85a Src homology 2 domains with phosphoinositides and inositol polyphosphates. J. Biol. Chem.
Oncogene 13, 813–822 41
22, 15678–15685 35
Danelian, E. et al. (2000) SPR biosensor studies of the direct interaction
42
femtomole to subfemtomole levels. Anal. Chem. 71, 2858–2865
Oyama, M. (2000) Highly sensitive detection of toxic chemicals by
43
measuring the quantity of induced P450 mRNAs based on SPR. Biosensors,
(BIA/MS). J. Mol. Recog. 12, 77–93
Hensley, P. and Myszka, D.G. (2000) Analytical biotechnology: sorting
44
needles and haystacks. Curr. Opin. Biotechnol. 11, 9–12 Williams, C. (2000) Biotechnology matchmaking: screening orphan
Trends Biotechnol. 18, 45–48 45
Krone, J.R. et al. (1997) BIA/MS: interfacing biomolecular interaction
46
Sönksen, C.P. et al. (1998) Combining MALDI mass spectrometry and
Lackmann, M. et al. (1996) Purification of a ligand for the EPH-like receptor HEK using a biosensor-based affinity detection approach.
40
Williams, C. and Addona,T.A. (2000) The integration of SPR biosensor with mass spectrometry: possible applications for proteomics analysis.
ligands and receptors. Curr. Opin. Biotechnol. 11, 42–46 39
Nelson, R.W. and Krone, J.R. (1999) Advances in surface plasmon resonance biomolecular interaction analysis mass spectrometry
Seattle, WA, USA
38
Nelson, R.W. et al. (1999) BIA/MS of epitope-tagged peptides directly from E. coli lysate: multiplex detection and protein identification at low-
absorbed in humans. J. Med. Chem. 43, 2083–2086
37
Davis, S. et al. (1996) Isolation of angiopoietin-1, a ligand for the TIE2 receptor, by secretion-trap expression cloning. Cell 87, 1161–1169
between 27 drugs and a liposome surface: correlation with fraction 36
reviews
analysis with mass spectrometry. Anal. Biochem. 244, 124–132
Proc. Natl.Acad. Sci. U. S.A. 93, 2523–2527
biomolecular interaction analysis using a biomolecular interaction
Sakano, S. et al. (1996) Characterization of a ligand for receptor
analysis instrument. Anal. Chem. 70, 2731–2736
In the October issue of Drug Discovery Today… Editorial – Poetry and verse: an ideal medium for scientific communication? by Raymond Rowe Update – latest news and views Endothelin-receptor antagonists: current and future perspectives Abhijit Ray, Laxminarayan G. Hegde, Anita Chugh and Jang B. Gupta Development of novel nucleoside analogues for use against drug resistant strains of HIV-1 Robert F. Rando and Nghe Nguyen-Ba Predicting human safety: screening and computational approaches Dale E. Johnson and Grushenka H.I. Wolfgang
Monitor – new bioactive molecules, combinatorial chemistry, invited profile Products
317