- Email: [email protected]
Using surface plasmon resonance to directly identify molecules in a tripeptide library that bind tightly to a vancomycin chip
ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 336 (2005) 172–177 www.elsevier.com/locate/yabio
Using surface plasmon resonance to directly identify molecules in a tripeptide library that bind tightly to a vancomycin chip Ming-Chung Tseng, Yen-Ho Chu* Department of Chemistry and Biochemistry, National Chung Cheng University, Chia-Yi, Taiwan 621, Taiwan, ROC Received 9 June 2004 Available online 19 November 2004
Abstract This paper describes a procedure, based on direct binding, for identifying tight-binding ligands for a receptor immobilized on a sensor chip from an array of equimolar tripeptides using surface plasmon resonance. Vancomycin and a library of 96 tripeptides, with molecular weight ranging from 316 to 560 Da, were used as a model system to illustrate the procedure. A consensus structure of the strongest interacting peptides consisted of D -Ala at the C terminus and aromatic amino acid in the penultimate position. Ligands having this structure bound more tightly to vancomycin than the known D -Ala-D -Ala peptide. The throughput of our continuous assay is 96 compounds in 3.3 h, and the sample consumption is less than 2 lg per peptide and 1 ng for vancomycin. This procedure should be applicable to peptide libraries of greater complexity than that used here and to mixtures of small organic compounds. Ó 2004 Elsevier Inc. All rights reserved. Keywords: Surface plasmon resonance; Library screening; Vancomycin; Biomolecular recognition
We are evaluating surface plasmon resonance (SPR)1 as an assay technique for use in identifying ligands and drug leads [1–3]. Among the strategies now used for discovering lead compounds are those based on screening collections of natural products (such as fermentation broths and plant extracts) or libraries of synthetic compounds [4–7]. These methods all require efficient assays [8–14]. This paper outlines an SPR-based assay for simultaneously screening all compounds in a library for their ability to bind to an immobilized receptor. Our assay is directly applicable to libraries in which constituents are available separately and in which the receptor is in limiting supply. In this application, our procedure is useful in shortening the total time required for the analysis of *
Corresponding author. Fax: +886 5 2721965, +886 5 2721040. E-mail addresses: [email protected], [email protected] (Y.-H. Chu). 1 Abbreviations used: SPR, surface plasmon resonance; DMSO, dimethyl sulfoxide; RU, resonance units. 0003-2697/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.10.035
binding. In addition, the assay is, in principle, also suitable for guiding a search through a mixture in which isolation of a component of unknown structure is required. The chip-based SPR technology has proven useful for both qualitative and quantitative biomolecular interaction analyses in real time without the need for fluorescent tags or radioactive labels [15–17]. This technology typically involves, first, the immobilization of a receptor to, for example, the carboxymethyl dextran surface of a sensor chip and, seconds, the search for its corresponding binding ligands by monitoring changes in SPR that reflect changes in the refractive index of the solution close to the surface of a sensor chip, which are caused by ligand binding to the immobilized receptor. This refractive index is directly related to the mass concentration in the surface layer and increases when ligand binds to the immobilized receptor. In this work, our library screening experiments were performed under continuous and controlled flow conditions using a microfluidic system in BIAcore3000 for sample delivery to the sensor surface.
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
Materials and methods
173
Screening of a library of 96 equimolar tripeptides using SPR
Materials Vancomycin, from Streptomyces orientalis, was a generous gift from Eli Lilly and Company (Indianapolis, IN, USA) through Dr. Richard C. Thompson. Synthesis of 96 tripeptides Peptides used for this study were prepared in parallel fashion using standard Fmoc solid-phase peptide chemistry. The integrity of individual peptide as the lyophilized powder was confirmed by reversed-phase HPLC. These compounds were diluted from DMSO stock solutions to their final concentration (100 lM) in running buffer. Due to its high refractive index, the effect of DMSO in samples and running buffer were experimentally calibrated for surface plasmon resonance measurements [18]. Instrumentation and buffers BIAcore3000 from BIAcore AB (Uppsala, Sweden) was used in this work. These SPR experiments were performed at 25 °C. The HBS-EP buffer (10 mM Hepes, pH 7.4; 150 mM NaCl; 3 mM EDTA; 0.005% (v/v) P20 surfactant) was first degassed, filtered, and then used as the running buffer for library screening. The Kd measurements of identified tripeptidyl ligands with immobilized vancomycin were carried out also in HBS-EP buffer. Immobilization of vancomycin The immobilization of vancomycin onto the sensor chip was carried out by following the recommended procedure from the manufacturer [19]. Sensor chip CM5 was used for analysis in this work. The CM5 chip consists of a carboxymethyl-modified dextran polymer linked to a gold-covered glass support. Vancomycin at 1 mM in 10 mM sodium acetate, pH 6.0, was immobilized to the sensor chip, using the chemistry of amine coupling. HBS-EP running buffer was used during immobilization. The surface was first activated by injecting a solution containing 0.2 M N-ethyl-N-dimethylaminopropyl carbodiimide and 50 mM N-hydroxysuccimide (NHS) for 7 min. Vancomycin was then injected for 7 min and the surface was blocked by injecting 1 M ethanolamine at pH 8.5 for 7 min. Finally, 50 mM NaOH was injected in three 30-s pulses to wash off noncovalently bound vancomycin and to stabilize the baseline. Final immobilization levels were typically between 600 and 700 RU. The immobilization procedure was followed by several washes with buffer to equilibrate the vancomycin surface in the running buffer. The unmodified dextran in channel one was used as a reference surface.
Library compounds were placed in a 96-microwell titer plate. Using the autosampler in BIAcore3000, each compound (100 lM) in the tripeptide library was programmed and injected sequentially (5 lL) over the reference and vancomycin flow cells at a flow rate of 15 lL/min. Each sample cycle consisted of a 20-s injection of compound. Data obtained in the reference flow cell were automatically subtracted from data obtained in the vancomycin flow cell. For peptides identified as ligands for vancomycin in this study, high-resolution mass spectrometric measurements [FAB-HRMS (M + H)] were carried out to confirm the identity of each peptide [high-resolution FAB mass spectra were collected on a JMS-700 double focusing mass spectrometer (JEOL, Tokyo, Japan) with a resolution of 8000 (5% valley definition). For FAB mass spectra, the source accelerating voltage was operated at 10 kV with Xe gun, using 3-nitrobenzyl alcohol as the matrix]: calcd. for Ac-KDFDA 407.229, found 407.229; calcd. for Ac-KDYDA 423.224, found 423.224; calcd. for Ac-KDWDA 446.240, found 446.241; calcd. for AcKDVDA 359.229, found 359.230; calcd. for Ac-KDKDA 388.256, found 338.257; calcd. for Ac-KDHDA 397.220, found 397.219; calcd. for Ac-KDQDA 388.220, found 388.220; calcd. for Ac-KDMDA 391.202, found 391.201; calcd. for Ac-KDADA 331.198, found 331.197; calcd. for Ac-KDSDA 347.193, found 347.193; calcd. for AcKDEDA 389.204, found 389.205; calcd. for Ac-KDTDA 361.209, found 361.208; calcd. for Ac-KDNDA 374.204, found 374.204; calcd. for Ac-KDLDA 373.245, found 373.244; calcd. for Ac-KDYDS 439.219, found 439.219; calcd. for Ac-KDFDS 423.224, found 423.224. For dissociation constant measurements of identified soluble peptides with immobilized vancomycin, triplicate binding experiments were conducted.
Results and discussion We illustrate the methodology using vancomycin and a small library of peptides. Vancomycin is an antibiotic that inhibits the cell wall biosynthesis of gram-positive bacteria by specifically binding to cell wall peptidoglycans that terminate in D -Ala-D -Ala [20]. The interaction of vancomycin and D -Ala-D -Ala-containing peptides has been extensively studied in solution, and the quantitative measurements of vancomycin interaction with individual immobilized peptides by SPR have recently been reported [21]. Vancomycin was chosen as a model receptor for the reason of its ease of chemical manipulation and because there is a substantial literature concerning its binding pocket and the peptidyl ligands that bind tightly to it [20]. Moreover, it is a fast-on
174
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
and off binding interaction and belongs to the micromolar binding system. Here, we screened a group of 96 unlabeled peptides of Ac-KDX1DX2 form (X1 = A, D, E, F, H, K, L, M, N, P, Q, S, T, V, W, Y; X2 = A, F, G, S, W, Y) for their affinity to vancomycin. The structure of Ac-KDX1DX2 tripeptides was used in this study because it resembled that of the C terminus of the gram-positive bacterial cell wall precursor, ADEKDADA, and it could be readily detected by SPR upon binding. As compared with the C-terminal DX2 in Ac-KDX1DX2, 16 D -amino acids were selected for use in the DX1 position with the hope that the diversity at this position may lead to a greater chance to identify the strongest binding ligands (our previous results in peptide library screening [22–24] indicated that vancomycin appeared to accommodate more amino acids in the DX1 position than in the DX2 position). Among 6 chosen D -amino acids in C-terminal DX2, A, G, and S were included because they were part of the binding elements previously discovered from screening of various peptide libraries [22–24] and F, W, and Y are aromatic amino acids. The screening procedure described in this work was performed with tripeptides arrayed in a 96-well plate and vancomycin immobilized on a CM-5 sensor chip, under conditions in which individual peptidyl ligand was programmed to directly interact with the receptor (Fig. 1). The binding site on vancomycin remained fully accessible after immobilization of vancomycin onto the sensor surface via its only primary amine group. When the ligand was injected over the sensor chip, any change in mass concentration on the surface resulting from an interaction was detected as a positive response expressed
in resonance units. The automated sample injections and continuous display of RU as a function of time gave a complete screening sensorgram (Fig. 1). Because it is a fast-on and -off binding system, all compounds tested bound reversibly to immobilized vancomycin. Although weaker signals were expected from the binding of the soluble small ligands onto the immobilized large receptor, the economy with which SPR uses materials and the minute quantity of the receptor needed permit our method to be practical. In our case, a series of 96 continuous runs required only 3 mL of buffer (for a flow rate of 15 lL/min) and an average of 1.5 lg for each tripeptide (with an average molecular weight of 429), and the amount of receptor needed is less than 1 ng for vancomycin for the entire series. Fig. 1 shows a SPR sensorgram resulting from direct binding of the library of 96 Ac-KDX1DX2 tripeptides to vancomycin immobilized on a chip; the insets are detailed sensorgrams for direct interactions of Ac-KDX1G and Ac-KDX1DS (16 each) sublibraries with the receptor. The results in Fig. 1 clearly demonstrate that vancomycin binds most tightly with peptides consisting of D -Ala at the C terminus. This finding is consistent with that of previous studies [22–24]. Among 16 Ac-KDX1DA studied, tripeptides having aromatic amino acids (D Phe, D -Tyr, and D -Trp) at the penultimate position gave the highest binding RU responses and only 2 peptides (DX1 = D -Asp and D -Pro) were not ligands for vancomycin. We also observed that, though relatively weak in RU response, several tripeptides in Ac-KDX1DS and Ac-KDX1G sublibraries bound specifically with vancomycin (the insets in Fig. 1). Among them, Ac-KDFDS and Ac-KDYDS tripeptides produced most significant
Fig. 1. Simultaneous screening of 96 equimolar tripeptides, Ac-KDX1DX2 (X1 = A, D, E, F, H, K, L, M, N, P, Q, S, T, V, W, Y; X2 = A, F, G, S, W, Y), to a vancomycin chip using surface plasmon resonance. The insets detail the binding interaction of Ac-KDX1DG and Ac-KDX1DS tripeptides to vancomycin. The throughput of this library screening is 96 compounds in 3.3 h.
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
175
RU responses upon binding. All these results clearly illustrate that SPR is a useful biospecific screening technique and well suited to direct and continuous binding interaction between immobilized vancomycin and soluble peptidyl ligands of low molecular weights. Our results presented here are in agreement with previous reports that the SPR can offer simple and rapid (i.e., yes or no) answers about the binding of analytes, as potential ligands, in a library to an immobilized receptor [1–3,25]. In this study, an equimolar ligand concentration of 100 lM was chosen for use in the experiments for the goal of screening library compounds over the entire affinity range; that is, at this concentration the responses from compounds with binding affinities of lM or lower could be readily detected. In addition, because many compounds in the library may be of limited solubility and we wanted to minimize possible nonspecific and secondary (if any) binding interactions, compound concentrations of higher than 100 lM were not used. To confirm the screening results, we quantitatively measured the binding of the identified tripeptides, at concentrations ranging from 0 to 100 lM, to the immobilized vancomycin under the equilibrium conditions (Figs. 2B and 3B). Scatchard analysis of the amount Fig. 3. Binding of Ac-KDYDA tripeptide to vancomycin immobilized on a sensorchip. (A) Scatchard plot of the binding data. The value for the equilibrium dissociation constant of the binding system is 12 lM. RU is the SPR response at equilibrium on the surface. (B) Sensorgrams of Ac-KDYDA binding to immobilized vancomycin. The concentration of Ac-KDYDA in the buffer is indicated by [Ac-KDYDA].
Fig. 2. Binding of Ac-KDFDA tripeptide to vancomycin immobilized on a sensorchip. (A) Scatchard plot of the binding data. The measured value for the equilibrium dissociation constant of the binding system is 6.6 lM. RU is the SPR response at equilibrium on the surface. (B) Sensorgrams of Ac-KDFDA binding to immobilized vancomycin. The concentration of Ac-KDFDA in the buffer is indicated by [Ac-KDFDA].
of peptide bound at the surface as a function of the concentration of peptide in the buffer gave the value for the equilibrium dissociation constant (Figs. 2A and 3A). Representative examples of the two most tight-binding peptides, Ac-KDFDA and Ac-KDYDA, binding with immobilized vancomycin are shown in Figs. 2 and 3, respectively. The values of the dissociation constants are ranked and given in Table 1. It is noted that, in Scatchard plots, values of the equilibrium dissociation constants (Kd) obtained from the slopes (K d 1 ) and from the intercepts of both the abscissa and the ordinate (RU1 max and RUmax K d , respectively) are similar, if not completely identical. For example, the Scatchard analysis of Ac-KDFDA binding with vancommycin gave the Kd value of 7.4 ± 1.1 (from slope) or 7.6 ± 0.9 (from intercepts). The second tight-binding ligand Ac-KDYDA, upon association with vancomycin, produced the Kd value of 12 ± 2 (from slope) or 11.6 ± 2.3 (from intercepts). Results in Table 1 clearly show that peptides having the structure of the Ac-KDX1DA form bind most tightly with immobilized vancomycin (entries 1–14). Most significantly, among 96 tripeptides studied, peptides consisting of D -Ala at its C terminus and
176
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
Table 1 Dissociation constants (Kd) for ligands identified and selected from a library of 96 tripeptides Entry
Peptide
MW (Da)
Kd (lM)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Ac-KDFDA Ac-KDYDA Ac-KDWDA Ac-KDVDA Ac-KDKDA Ac-KDHDA Ac-KDQDA Ac-KDMDA Ac-KDADA Ac-KDSDA Ac-KDEDA Ac-KDTDA Ac-KDNDA Ac-KDLDA Ac-KDYDS Ac-KDFDS Ac-KDDDA Ac-KDPDA
406 422 445 358 387 396 387 390 330 346 388 360 373 372 438 422 374 356
7.4 ± 1.1 12 ± 2 24 ± 1 28 ± 1 44 ± 1 49 ± 2 55 ± 2 65 ± 3 65 ± 3 80 ± 3 86 ± 1 87 ± 4 92 ± 5 95 ± 2 129 ± 4 126 ± 1 3300 ± 500 2600 ± 500
D -aromatic
residue in the penultimate position are the best ligands in binding with vancomycin. As previously described, Ac-KDDDA and Ac-KDPDA peptides were hardly recognized by vancomycin (entries 17 and 18).
Conclusion The SPR-based biosensor screening methodology described here provides an efficient approach with which to search for tight-binding ligands in libraries of small organic molecules. The main advantages of this technology are that binding is monitored directly and continuously without the use of labels, sample consumption is low, and the screening analysis is rapid and automated. The results obtained with this procedure are reproducible and the binding affinities obtained in this work are correlated with established methodologies and literature results. Currently, the most serious experimental uncertainty in considering it for examination of a receptor appears to be the molecular weight of drug-like compounds studied, although the binding of ligands of much smaller molecular weights in principle can be detected using the competitive binding method. Our screening methodology does not require a biological activity: it is a pure binding assay. This method is practical, it provides necessary binding information, and complicated analysis is avoided. The throughput of our continuous assay presented is 96 compounds in 3.3 h.
Acknowledgments We gratefully acknowledge support of this work by the Ministry of Education (Taiwan, ROC) through an
instrumentation fund (91CE093) and in part by the Taichung Science-Based Industrial Park (Taiwan, ROC) (NSC92-2218-E-194-021) and the National Science Council (Taiwan, ROC) (NSC93-2113-M-194019, NSC92-2751-B-001-014, and NSC92-2218-E-194015). We especially thank Ms. Vickie Ma of Amersham Biosciences (Taiwan, ROC) for many insightful suggestions and valuable discussions. M.C.T. is a recipient of the NSC Predoctoral Fellowship. References [1] R.L. Rich, Y.S.N. Day, T.A. Morton, D.G. Myszka, Highresolution and high-throughput protocols for measuring drug/ human serum albumin interactions using BIAcore, Anal. Biochem. 296 (2001) 197–207. [2] R. Karlsson, M. Kullman-Magnusson, M.D. Hamalainen, A. Remaeus, K. Andersson, P. Borg, E. Gyzander, J. Deinum, Biosensor analysis of drug–target interactions: direct and competitive binding assays for investigation of interactions between thrombin and thrombin inhibitors, Anal. Biochem. 278 (2000) 1– 13. [3] C.P. Woodbury Jr., D.L. Venton, Methods of screening combinatorial libraries using immobilized or restrained receptors, J. Chromatogr. B 725 (1999) 113–137. [4] M. Patek, P. Safar, M. Smrcina, E. Wegrzyniak, K. Bjergarde, A. Weichsel, P. Strop, 2D and 3D spatially addressed arrays for high-throughput automated synthesis of combinatorial libraries, J. Comb. Chem. 6 (2004) 43–49. [5] D.A. Horton, G.T. Bourne, M.L. Smythe, The combinatorial synthesis of bicyclic privileged structures or privileged substructures, Chem. Rev. 103 (2003) 893–930. [6] B. Yan, A.W. Czarnik, Optimization of Solid-Phase Combinatorial Synthesis, Marcel Dekker, New York, 2002. [7] H. Fenniri, Combinatorial Chemistry: A Practical Approach, Oxford University Press, New York, 2000. [8] T. Cserhati, E. Forgacs, Z. Deyl, I. Miksik, A. Echardt, Binding of low molecular mass compounds to proteins studied by liquid chromatographic techniques, Biomed. Chromatogr. 17 (2003) 353–360. [9] P. Englebienne, A.V. Hoonacker, M. Verhas, High-throughput screening using the surface plasmon resonance effect of colloidal gold nanoparticles, Analyst 126 (2001) 1645–1651. [10] Y.-H. Chu, X. Zang, J. Tu, Affinity capillary electrophoresis: from binding measurement to combinatorial library screening, J. Chin. Chem. Soc. 45 (1998) 713–720. [11] D. Hage, S. Tweed, Recent advances in chromatographic and electrophoretic methods for the study of drug–protein interactions, J. Chromatogr. 699 (1997) 499–525. [12] P. Wu, M. Brasseur, U. Schindler, A high-throughput STAT binding assay using fluorescence polarization, Anal. Biochem. 249 (1997) 29–36. [13] S.B. Shuker, P.J. Hajduk, R.P. Meadows, S.W. Fesik, Discovering high-affinity ligands for proteins: SAR by NMR, Science 274 (1996) 1531–1534. [14] N.D. Cook, Scintillation proximity assay: a versatile highthroughput screening technology, Drug Discov. Today 1 (1996) 287–289. [15] R.L. Rich, D.G. Myszka, A survey of the year 2002 commercial optical biosensor literature, J. Mol. Recognit. 16 (2003) 351–382. [16] R.L. Rich, D.G. Myszka, Survey of the year 2001 commercial optical biosensor literature, J. Mol. Recognit. 15 (2002) 352–376. [17] M.A. Cooper, Optical biosensors in drug discovery, Nat. Rev. Drug Discov. 1 (2002) 515–528.
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177 [18] A. Frostell-Karlsson, A. Remaeus, H. Roos, K. Andersson, P. Borg, M. Hamalainen, R. Karlsson, Biosensor analysis of the interaction between immobilized human serum albumin and drug compounds for prediction of human serum albumin binding levels, J. Med. Chem. 43 (2000) 1986–1992. [19] B. Johnsson, S. Lofas, G. Lindquist, Immobilization of proteins to carboxy methyl dextran-modified gold surfaces for biospecific analysis in surface plasmon resonance sensors, Anal. Biochem. 198 (1991) 268–277. [20] D.H. Williams, B. Bardsley, The vancomycin group of antibiotics and the fight against resistant bacteria, Angew. Chem. Int. Ed. 38 (1999) 1172–1193. [21] J. Rao, L. Yan, B. Xu, G.M. Whitesides, Using surface plasmon resonance to study the binding of vancomycin and its dimer to self-assembled monolayers presenting D -Ala-D -Ala, J. Am. Chem. Soc. 121 (1999) 2629–2630.
177
[22] Y.-H. Chu, D.P. Kirby, B.L. Karger, Free solution identification of candidate peptides from combinatorial libraries by affinity capillary electrophoresis/mass spectrometry, J. Am. Chem. Soc. 117 (1995) 5419–5420. [23] Y.-H. Chu, Y.M. Dunayevskiy, D.P. Kirby, P. Vouros, B.L. Karger, Affinity capillary electrophoresis-mass spectrometry for screening combinatorial libraries, J. Am. Chem. Soc. 118 (1996) 7827–7835. [24] Y.M. Dunayevskiy, Y.V. Lyubarskaya, Y.-H. Chu, P. Vouros, B.L. Karger, Simultaneous measurement of nineteen binding constants of peptides to vancomycin using affinity capillary electrophoresismass spectrometry, J. Med. Chem. 41 (1998) 1201–1204. [25] P. Gomes, E. Giralt, D. Andreu, Surface plasmon resonance screening of synthetic peptides mimicking the immunodominant region of C-S8c1 foot-and-mouth disease virus, Vaccine 18 (2000) 362–370.
Using surface plasmon resonance to directly identify molecules in a tripeptide library that bind tightly to a vancomycin chip Ming-Chung Tseng, Yen-Ho Chu* Department of Chemistry and Biochemistry, National Chung Cheng University, Chia-Yi, Taiwan 621, Taiwan, ROC Received 9 June 2004 Available online 19 November 2004
Abstract This paper describes a procedure, based on direct binding, for identifying tight-binding ligands for a receptor immobilized on a sensor chip from an array of equimolar tripeptides using surface plasmon resonance. Vancomycin and a library of 96 tripeptides, with molecular weight ranging from 316 to 560 Da, were used as a model system to illustrate the procedure. A consensus structure of the strongest interacting peptides consisted of D -Ala at the C terminus and aromatic amino acid in the penultimate position. Ligands having this structure bound more tightly to vancomycin than the known D -Ala-D -Ala peptide. The throughput of our continuous assay is 96 compounds in 3.3 h, and the sample consumption is less than 2 lg per peptide and 1 ng for vancomycin. This procedure should be applicable to peptide libraries of greater complexity than that used here and to mixtures of small organic compounds. Ó 2004 Elsevier Inc. All rights reserved. Keywords: Surface plasmon resonance; Library screening; Vancomycin; Biomolecular recognition
We are evaluating surface plasmon resonance (SPR)1 as an assay technique for use in identifying ligands and drug leads [1–3]. Among the strategies now used for discovering lead compounds are those based on screening collections of natural products (such as fermentation broths and plant extracts) or libraries of synthetic compounds [4–7]. These methods all require efficient assays [8–14]. This paper outlines an SPR-based assay for simultaneously screening all compounds in a library for their ability to bind to an immobilized receptor. Our assay is directly applicable to libraries in which constituents are available separately and in which the receptor is in limiting supply. In this application, our procedure is useful in shortening the total time required for the analysis of *
Corresponding author. Fax: +886 5 2721965, +886 5 2721040. E-mail addresses: [email protected], [email protected] (Y.-H. Chu). 1 Abbreviations used: SPR, surface plasmon resonance; DMSO, dimethyl sulfoxide; RU, resonance units. 0003-2697/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.10.035
binding. In addition, the assay is, in principle, also suitable for guiding a search through a mixture in which isolation of a component of unknown structure is required. The chip-based SPR technology has proven useful for both qualitative and quantitative biomolecular interaction analyses in real time without the need for fluorescent tags or radioactive labels [15–17]. This technology typically involves, first, the immobilization of a receptor to, for example, the carboxymethyl dextran surface of a sensor chip and, seconds, the search for its corresponding binding ligands by monitoring changes in SPR that reflect changes in the refractive index of the solution close to the surface of a sensor chip, which are caused by ligand binding to the immobilized receptor. This refractive index is directly related to the mass concentration in the surface layer and increases when ligand binds to the immobilized receptor. In this work, our library screening experiments were performed under continuous and controlled flow conditions using a microfluidic system in BIAcore3000 for sample delivery to the sensor surface.
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
Materials and methods
173
Screening of a library of 96 equimolar tripeptides using SPR
Materials Vancomycin, from Streptomyces orientalis, was a generous gift from Eli Lilly and Company (Indianapolis, IN, USA) through Dr. Richard C. Thompson. Synthesis of 96 tripeptides Peptides used for this study were prepared in parallel fashion using standard Fmoc solid-phase peptide chemistry. The integrity of individual peptide as the lyophilized powder was confirmed by reversed-phase HPLC. These compounds were diluted from DMSO stock solutions to their final concentration (100 lM) in running buffer. Due to its high refractive index, the effect of DMSO in samples and running buffer were experimentally calibrated for surface plasmon resonance measurements [18]. Instrumentation and buffers BIAcore3000 from BIAcore AB (Uppsala, Sweden) was used in this work. These SPR experiments were performed at 25 °C. The HBS-EP buffer (10 mM Hepes, pH 7.4; 150 mM NaCl; 3 mM EDTA; 0.005% (v/v) P20 surfactant) was first degassed, filtered, and then used as the running buffer for library screening. The Kd measurements of identified tripeptidyl ligands with immobilized vancomycin were carried out also in HBS-EP buffer. Immobilization of vancomycin The immobilization of vancomycin onto the sensor chip was carried out by following the recommended procedure from the manufacturer [19]. Sensor chip CM5 was used for analysis in this work. The CM5 chip consists of a carboxymethyl-modified dextran polymer linked to a gold-covered glass support. Vancomycin at 1 mM in 10 mM sodium acetate, pH 6.0, was immobilized to the sensor chip, using the chemistry of amine coupling. HBS-EP running buffer was used during immobilization. The surface was first activated by injecting a solution containing 0.2 M N-ethyl-N-dimethylaminopropyl carbodiimide and 50 mM N-hydroxysuccimide (NHS) for 7 min. Vancomycin was then injected for 7 min and the surface was blocked by injecting 1 M ethanolamine at pH 8.5 for 7 min. Finally, 50 mM NaOH was injected in three 30-s pulses to wash off noncovalently bound vancomycin and to stabilize the baseline. Final immobilization levels were typically between 600 and 700 RU. The immobilization procedure was followed by several washes with buffer to equilibrate the vancomycin surface in the running buffer. The unmodified dextran in channel one was used as a reference surface.
Library compounds were placed in a 96-microwell titer plate. Using the autosampler in BIAcore3000, each compound (100 lM) in the tripeptide library was programmed and injected sequentially (5 lL) over the reference and vancomycin flow cells at a flow rate of 15 lL/min. Each sample cycle consisted of a 20-s injection of compound. Data obtained in the reference flow cell were automatically subtracted from data obtained in the vancomycin flow cell. For peptides identified as ligands for vancomycin in this study, high-resolution mass spectrometric measurements [FAB-HRMS (M + H)] were carried out to confirm the identity of each peptide [high-resolution FAB mass spectra were collected on a JMS-700 double focusing mass spectrometer (JEOL, Tokyo, Japan) with a resolution of 8000 (5% valley definition). For FAB mass spectra, the source accelerating voltage was operated at 10 kV with Xe gun, using 3-nitrobenzyl alcohol as the matrix]: calcd. for Ac-KDFDA 407.229, found 407.229; calcd. for Ac-KDYDA 423.224, found 423.224; calcd. for Ac-KDWDA 446.240, found 446.241; calcd. for AcKDVDA 359.229, found 359.230; calcd. for Ac-KDKDA 388.256, found 338.257; calcd. for Ac-KDHDA 397.220, found 397.219; calcd. for Ac-KDQDA 388.220, found 388.220; calcd. for Ac-KDMDA 391.202, found 391.201; calcd. for Ac-KDADA 331.198, found 331.197; calcd. for Ac-KDSDA 347.193, found 347.193; calcd. for AcKDEDA 389.204, found 389.205; calcd. for Ac-KDTDA 361.209, found 361.208; calcd. for Ac-KDNDA 374.204, found 374.204; calcd. for Ac-KDLDA 373.245, found 373.244; calcd. for Ac-KDYDS 439.219, found 439.219; calcd. for Ac-KDFDS 423.224, found 423.224. For dissociation constant measurements of identified soluble peptides with immobilized vancomycin, triplicate binding experiments were conducted.
Results and discussion We illustrate the methodology using vancomycin and a small library of peptides. Vancomycin is an antibiotic that inhibits the cell wall biosynthesis of gram-positive bacteria by specifically binding to cell wall peptidoglycans that terminate in D -Ala-D -Ala [20]. The interaction of vancomycin and D -Ala-D -Ala-containing peptides has been extensively studied in solution, and the quantitative measurements of vancomycin interaction with individual immobilized peptides by SPR have recently been reported [21]. Vancomycin was chosen as a model receptor for the reason of its ease of chemical manipulation and because there is a substantial literature concerning its binding pocket and the peptidyl ligands that bind tightly to it [20]. Moreover, it is a fast-on
174
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
and off binding interaction and belongs to the micromolar binding system. Here, we screened a group of 96 unlabeled peptides of Ac-KDX1DX2 form (X1 = A, D, E, F, H, K, L, M, N, P, Q, S, T, V, W, Y; X2 = A, F, G, S, W, Y) for their affinity to vancomycin. The structure of Ac-KDX1DX2 tripeptides was used in this study because it resembled that of the C terminus of the gram-positive bacterial cell wall precursor, ADEKDADA, and it could be readily detected by SPR upon binding. As compared with the C-terminal DX2 in Ac-KDX1DX2, 16 D -amino acids were selected for use in the DX1 position with the hope that the diversity at this position may lead to a greater chance to identify the strongest binding ligands (our previous results in peptide library screening [22–24] indicated that vancomycin appeared to accommodate more amino acids in the DX1 position than in the DX2 position). Among 6 chosen D -amino acids in C-terminal DX2, A, G, and S were included because they were part of the binding elements previously discovered from screening of various peptide libraries [22–24] and F, W, and Y are aromatic amino acids. The screening procedure described in this work was performed with tripeptides arrayed in a 96-well plate and vancomycin immobilized on a CM-5 sensor chip, under conditions in which individual peptidyl ligand was programmed to directly interact with the receptor (Fig. 1). The binding site on vancomycin remained fully accessible after immobilization of vancomycin onto the sensor surface via its only primary amine group. When the ligand was injected over the sensor chip, any change in mass concentration on the surface resulting from an interaction was detected as a positive response expressed
in resonance units. The automated sample injections and continuous display of RU as a function of time gave a complete screening sensorgram (Fig. 1). Because it is a fast-on and -off binding system, all compounds tested bound reversibly to immobilized vancomycin. Although weaker signals were expected from the binding of the soluble small ligands onto the immobilized large receptor, the economy with which SPR uses materials and the minute quantity of the receptor needed permit our method to be practical. In our case, a series of 96 continuous runs required only 3 mL of buffer (for a flow rate of 15 lL/min) and an average of 1.5 lg for each tripeptide (with an average molecular weight of 429), and the amount of receptor needed is less than 1 ng for vancomycin for the entire series. Fig. 1 shows a SPR sensorgram resulting from direct binding of the library of 96 Ac-KDX1DX2 tripeptides to vancomycin immobilized on a chip; the insets are detailed sensorgrams for direct interactions of Ac-KDX1G and Ac-KDX1DS (16 each) sublibraries with the receptor. The results in Fig. 1 clearly demonstrate that vancomycin binds most tightly with peptides consisting of D -Ala at the C terminus. This finding is consistent with that of previous studies [22–24]. Among 16 Ac-KDX1DA studied, tripeptides having aromatic amino acids (D Phe, D -Tyr, and D -Trp) at the penultimate position gave the highest binding RU responses and only 2 peptides (DX1 = D -Asp and D -Pro) were not ligands for vancomycin. We also observed that, though relatively weak in RU response, several tripeptides in Ac-KDX1DS and Ac-KDX1G sublibraries bound specifically with vancomycin (the insets in Fig. 1). Among them, Ac-KDFDS and Ac-KDYDS tripeptides produced most significant
Fig. 1. Simultaneous screening of 96 equimolar tripeptides, Ac-KDX1DX2 (X1 = A, D, E, F, H, K, L, M, N, P, Q, S, T, V, W, Y; X2 = A, F, G, S, W, Y), to a vancomycin chip using surface plasmon resonance. The insets detail the binding interaction of Ac-KDX1DG and Ac-KDX1DS tripeptides to vancomycin. The throughput of this library screening is 96 compounds in 3.3 h.
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
175
RU responses upon binding. All these results clearly illustrate that SPR is a useful biospecific screening technique and well suited to direct and continuous binding interaction between immobilized vancomycin and soluble peptidyl ligands of low molecular weights. Our results presented here are in agreement with previous reports that the SPR can offer simple and rapid (i.e., yes or no) answers about the binding of analytes, as potential ligands, in a library to an immobilized receptor [1–3,25]. In this study, an equimolar ligand concentration of 100 lM was chosen for use in the experiments for the goal of screening library compounds over the entire affinity range; that is, at this concentration the responses from compounds with binding affinities of lM or lower could be readily detected. In addition, because many compounds in the library may be of limited solubility and we wanted to minimize possible nonspecific and secondary (if any) binding interactions, compound concentrations of higher than 100 lM were not used. To confirm the screening results, we quantitatively measured the binding of the identified tripeptides, at concentrations ranging from 0 to 100 lM, to the immobilized vancomycin under the equilibrium conditions (Figs. 2B and 3B). Scatchard analysis of the amount Fig. 3. Binding of Ac-KDYDA tripeptide to vancomycin immobilized on a sensorchip. (A) Scatchard plot of the binding data. The value for the equilibrium dissociation constant of the binding system is 12 lM. RU is the SPR response at equilibrium on the surface. (B) Sensorgrams of Ac-KDYDA binding to immobilized vancomycin. The concentration of Ac-KDYDA in the buffer is indicated by [Ac-KDYDA].
Fig. 2. Binding of Ac-KDFDA tripeptide to vancomycin immobilized on a sensorchip. (A) Scatchard plot of the binding data. The measured value for the equilibrium dissociation constant of the binding system is 6.6 lM. RU is the SPR response at equilibrium on the surface. (B) Sensorgrams of Ac-KDFDA binding to immobilized vancomycin. The concentration of Ac-KDFDA in the buffer is indicated by [Ac-KDFDA].
of peptide bound at the surface as a function of the concentration of peptide in the buffer gave the value for the equilibrium dissociation constant (Figs. 2A and 3A). Representative examples of the two most tight-binding peptides, Ac-KDFDA and Ac-KDYDA, binding with immobilized vancomycin are shown in Figs. 2 and 3, respectively. The values of the dissociation constants are ranked and given in Table 1. It is noted that, in Scatchard plots, values of the equilibrium dissociation constants (Kd) obtained from the slopes (K d 1 ) and from the intercepts of both the abscissa and the ordinate (RU1 max and RUmax K d , respectively) are similar, if not completely identical. For example, the Scatchard analysis of Ac-KDFDA binding with vancommycin gave the Kd value of 7.4 ± 1.1 (from slope) or 7.6 ± 0.9 (from intercepts). The second tight-binding ligand Ac-KDYDA, upon association with vancomycin, produced the Kd value of 12 ± 2 (from slope) or 11.6 ± 2.3 (from intercepts). Results in Table 1 clearly show that peptides having the structure of the Ac-KDX1DA form bind most tightly with immobilized vancomycin (entries 1–14). Most significantly, among 96 tripeptides studied, peptides consisting of D -Ala at its C terminus and
176
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177
Table 1 Dissociation constants (Kd) for ligands identified and selected from a library of 96 tripeptides Entry
Peptide
MW (Da)
Kd (lM)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Ac-KDFDA Ac-KDYDA Ac-KDWDA Ac-KDVDA Ac-KDKDA Ac-KDHDA Ac-KDQDA Ac-KDMDA Ac-KDADA Ac-KDSDA Ac-KDEDA Ac-KDTDA Ac-KDNDA Ac-KDLDA Ac-KDYDS Ac-KDFDS Ac-KDDDA Ac-KDPDA
406 422 445 358 387 396 387 390 330 346 388 360 373 372 438 422 374 356
7.4 ± 1.1 12 ± 2 24 ± 1 28 ± 1 44 ± 1 49 ± 2 55 ± 2 65 ± 3 65 ± 3 80 ± 3 86 ± 1 87 ± 4 92 ± 5 95 ± 2 129 ± 4 126 ± 1 3300 ± 500 2600 ± 500
D -aromatic
residue in the penultimate position are the best ligands in binding with vancomycin. As previously described, Ac-KDDDA and Ac-KDPDA peptides were hardly recognized by vancomycin (entries 17 and 18).
Conclusion The SPR-based biosensor screening methodology described here provides an efficient approach with which to search for tight-binding ligands in libraries of small organic molecules. The main advantages of this technology are that binding is monitored directly and continuously without the use of labels, sample consumption is low, and the screening analysis is rapid and automated. The results obtained with this procedure are reproducible and the binding affinities obtained in this work are correlated with established methodologies and literature results. Currently, the most serious experimental uncertainty in considering it for examination of a receptor appears to be the molecular weight of drug-like compounds studied, although the binding of ligands of much smaller molecular weights in principle can be detected using the competitive binding method. Our screening methodology does not require a biological activity: it is a pure binding assay. This method is practical, it provides necessary binding information, and complicated analysis is avoided. The throughput of our continuous assay presented is 96 compounds in 3.3 h.
Acknowledgments We gratefully acknowledge support of this work by the Ministry of Education (Taiwan, ROC) through an
instrumentation fund (91CE093) and in part by the Taichung Science-Based Industrial Park (Taiwan, ROC) (NSC92-2218-E-194-021) and the National Science Council (Taiwan, ROC) (NSC93-2113-M-194019, NSC92-2751-B-001-014, and NSC92-2218-E-194015). We especially thank Ms. Vickie Ma of Amersham Biosciences (Taiwan, ROC) for many insightful suggestions and valuable discussions. M.C.T. is a recipient of the NSC Predoctoral Fellowship. References [1] R.L. Rich, Y.S.N. Day, T.A. Morton, D.G. Myszka, Highresolution and high-throughput protocols for measuring drug/ human serum albumin interactions using BIAcore, Anal. Biochem. 296 (2001) 197–207. [2] R. Karlsson, M. Kullman-Magnusson, M.D. Hamalainen, A. Remaeus, K. Andersson, P. Borg, E. Gyzander, J. Deinum, Biosensor analysis of drug–target interactions: direct and competitive binding assays for investigation of interactions between thrombin and thrombin inhibitors, Anal. Biochem. 278 (2000) 1– 13. [3] C.P. Woodbury Jr., D.L. Venton, Methods of screening combinatorial libraries using immobilized or restrained receptors, J. Chromatogr. B 725 (1999) 113–137. [4] M. Patek, P. Safar, M. Smrcina, E. Wegrzyniak, K. Bjergarde, A. Weichsel, P. Strop, 2D and 3D spatially addressed arrays for high-throughput automated synthesis of combinatorial libraries, J. Comb. Chem. 6 (2004) 43–49. [5] D.A. Horton, G.T. Bourne, M.L. Smythe, The combinatorial synthesis of bicyclic privileged structures or privileged substructures, Chem. Rev. 103 (2003) 893–930. [6] B. Yan, A.W. Czarnik, Optimization of Solid-Phase Combinatorial Synthesis, Marcel Dekker, New York, 2002. [7] H. Fenniri, Combinatorial Chemistry: A Practical Approach, Oxford University Press, New York, 2000. [8] T. Cserhati, E. Forgacs, Z. Deyl, I. Miksik, A. Echardt, Binding of low molecular mass compounds to proteins studied by liquid chromatographic techniques, Biomed. Chromatogr. 17 (2003) 353–360. [9] P. Englebienne, A.V. Hoonacker, M. Verhas, High-throughput screening using the surface plasmon resonance effect of colloidal gold nanoparticles, Analyst 126 (2001) 1645–1651. [10] Y.-H. Chu, X. Zang, J. Tu, Affinity capillary electrophoresis: from binding measurement to combinatorial library screening, J. Chin. Chem. Soc. 45 (1998) 713–720. [11] D. Hage, S. Tweed, Recent advances in chromatographic and electrophoretic methods for the study of drug–protein interactions, J. Chromatogr. 699 (1997) 499–525. [12] P. Wu, M. Brasseur, U. Schindler, A high-throughput STAT binding assay using fluorescence polarization, Anal. Biochem. 249 (1997) 29–36. [13] S.B. Shuker, P.J. Hajduk, R.P. Meadows, S.W. Fesik, Discovering high-affinity ligands for proteins: SAR by NMR, Science 274 (1996) 1531–1534. [14] N.D. Cook, Scintillation proximity assay: a versatile highthroughput screening technology, Drug Discov. Today 1 (1996) 287–289. [15] R.L. Rich, D.G. Myszka, A survey of the year 2002 commercial optical biosensor literature, J. Mol. Recognit. 16 (2003) 351–382. [16] R.L. Rich, D.G. Myszka, Survey of the year 2001 commercial optical biosensor literature, J. Mol. Recognit. 15 (2002) 352–376. [17] M.A. Cooper, Optical biosensors in drug discovery, Nat. Rev. Drug Discov. 1 (2002) 515–528.
Using SPR to identify molecules bound to vancomycin chip / M.-C. Tseng, Y.-H. Chu / Anal. Biochem. 336 (2005) 172–177 [18] A. Frostell-Karlsson, A. Remaeus, H. Roos, K. Andersson, P. Borg, M. Hamalainen, R. Karlsson, Biosensor analysis of the interaction between immobilized human serum albumin and drug compounds for prediction of human serum albumin binding levels, J. Med. Chem. 43 (2000) 1986–1992. [19] B. Johnsson, S. Lofas, G. Lindquist, Immobilization of proteins to carboxy methyl dextran-modified gold surfaces for biospecific analysis in surface plasmon resonance sensors, Anal. Biochem. 198 (1991) 268–277. [20] D.H. Williams, B. Bardsley, The vancomycin group of antibiotics and the fight against resistant bacteria, Angew. Chem. Int. Ed. 38 (1999) 1172–1193. [21] J. Rao, L. Yan, B. Xu, G.M. Whitesides, Using surface plasmon resonance to study the binding of vancomycin and its dimer to self-assembled monolayers presenting D -Ala-D -Ala, J. Am. Chem. Soc. 121 (1999) 2629–2630.
177
[22] Y.-H. Chu, D.P. Kirby, B.L. Karger, Free solution identification of candidate peptides from combinatorial libraries by affinity capillary electrophoresis/mass spectrometry, J. Am. Chem. Soc. 117 (1995) 5419–5420. [23] Y.-H. Chu, Y.M. Dunayevskiy, D.P. Kirby, P. Vouros, B.L. Karger, Affinity capillary electrophoresis-mass spectrometry for screening combinatorial libraries, J. Am. Chem. Soc. 118 (1996) 7827–7835. [24] Y.M. Dunayevskiy, Y.V. Lyubarskaya, Y.-H. Chu, P. Vouros, B.L. Karger, Simultaneous measurement of nineteen binding constants of peptides to vancomycin using affinity capillary electrophoresismass spectrometry, J. Med. Chem. 41 (1998) 1201–1204. [25] P. Gomes, E. Giralt, D. Andreu, Surface plasmon resonance screening of synthetic peptides mimicking the immunodominant region of C-S8c1 foot-and-mouth disease virus, Vaccine 18 (2000) 362–370.