A Flow Injection Kinase Assay System Based on Time-Resolved

Junko Hirata, Camiel F. de Jong, Maarten M. van Dongen, Joost Buijs, Freek Ariese, Hubertus Irth, and Cees Gooijer*. Laser Center, Department of Analy...
2 downloads 0 Views 125KB Size
Download PDF
Anal. Chem. 2004, 76, 4292-4298

A Flow Injection Kinase Assay System Based on Time-Resolved Fluorescence Resonance Energy-Transfer Detection in the Millisecond Range Junko Hirata, Camiel F. de Jong, Maarten M. van Dongen, Joost Buijs, Freek Ariese, Hubertus Irth, and Cees Gooijer*

Laser Center, Department of Analytical Chemistry and Applied Spectroscopy, Vrije Universiteit Amsterdam, de Boelelaan 1083, 1081 HV Amsterdam, The Netherlands

A flow injection analysis (FIA) system for biochemical assays using time-resolved fluorescence resonance energy transfer (TR-FRET) in the millisecond time scale was developed. As a model system, we studied a kinase assay, measuring the phosphorylation of poly(GT)-biotin (substrate) by a receptor tyrosine kinase (epidermal growth factor receptor). A streptavidin labeled with XL665 (SA-XL665)sthe acceptorswas coupled to the biotin moiety, and an antiphosphotyrosine antibody labeled with europium cryptate (Ab-EuK)sthe donorswas coupled to the phosphorylated tyrosine group(s). Long-lived FRET can only occur if the substrate is successfully phosphorylated. For the time-resolved detection of such long-lived luminescence phenomena in a flow system, the repetition rate of the excitation source plays a crucial role. Good results were obtained for a small-sized commercially available quadrupled Nd:YAG laser emitting at 266 nm with a repetition rate of 7.8 kHz and a pulse width of 0.3 ns. The long-lived emissions of the donor at 625 nm and that of the acceptor at 665 nm were monitored simultaneously with two photomultipliers, using a delay time of 50 µs and a gate time of 75 µs to exclude background fluorescence interferences. In the FIA experiments, the Ab-EuK concentration was 6 nM and the substrate concentration and SA-XL665 concentrations were 7 nM. By monitoring the intensity changes at 625 and 665 nm, the inhibition of tyrosine kinase by tyrphostin AG1478 was studied and an IC50 value of 5.1 ( 0.4 nM obtained. The screening of synthetic and natural chemical sources is the starting point in drug discovery, the main goal being the identification of promising pharmacologically active compounds. High-throughput screening (HTS) technologies have been developed and implemented that are able to test tens of thousands of compounds or more per day for their activity in various assay types, ranging from receptor binding and enzyme inhibition to whole-cell assays. A large variety of assay formats has been described, with fluorescence detection in various modes currently * Corresponding author. Tel.: E-mail: [email protected].

+31-20-4447540. Fax:

+31-20-4447543.

4292 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

being the most important detection technique.1-4 Furthermore, there is a strong tendency toward homogeneous assays, i.e., assays where no separation between free and bound reporter molecules is required. Despite the inherent selectivity of fluorescence detection, fluorescence-based screening assays are hampered by background interferences particularly when complex samples, e.g., natural extracts or biological matrixes, have to be analyzed. Background interferences often lead to a decreased assay sensitivity and the assignment of false positives, thereby compromising the efficiency of HTS assays. To reduce background interferences, various fluorescence detection modes have been developed that are more selective than direct fluorescence detection. Examples are fluorescence polarization,5-7 fluorescence correlation spectroscopy,8-10 time-resolved fluorescence including long-lived lanthanide luminescence,11-13 and fluorescence resonance energy transfer (FRET).14-16 The latter detection mode has gained in importance (1) Jager, S.; Garbow, N.; Kirsch, A.; Preckel, H.; Gandenberger, F. U.; Herrenknecht, K.; Rudiger, M.; Hutchinson, J. P.; Bingham, R. P.; Ramon, F.; Bardera, A.; Martin, J. J. Biomol. Screening 2003, 8, 648-659. (2) Kulmala, S.; Suomi, J. Anal. Chim. Acta 2003, 500, 21-69. (3) Biran, I.; Walt, D. R. Anal. Chem. 2002, 74, 3046-3054. (4) Evans, C. A.; Miller, S. J. Curr. Opin. Chem. Biol. 2002, 6, 333-338. (5) Parker, G. J.; Law, T. L.; Lenoch, F. J.; Bolger, R. E. J. Biomol. Screening 2000, 5, 77-88 (6) Fowler, A.; Swift, D.; Longman, E.; Acornley, A.; Hemsley, P.; Murray, D.; Unitt, J.; Dale, I.; Sullivan, E.; Coldwell, M. Anal. Biochem. 2002, 308, 223231. (7) Harris, A.; Cox, S.; Burns, D.; Norey, C. J. Bimol. Screening 2003, 8, 410420. (8) Zemanova´, L.; Schenk, A.; Valler, M. J.; Nienhaus, G. U.; Heilker, R. Drug Dicovery Today 2003, 8, 1085-1093. (9) Moore, K. J.; Turconi, S.; Ashman, S.; Ruediger, M.; Haupts, U.; Emerick, V.; Pope, A. J. J. Biomol. Screening 1999, 4, 335-353. (10) Koltermann, A.; Kettling, U.; Bieschke, J.; Winkler, T.; Eigen, M. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 1421-1426. (11) Kiviniemi, M.; Nurmi, J.; Turpeinen, H.; Lo¨vgren, T.; Ilonen, J. Clin. Chem. 2003, 36, 633-640. (12) Karvinen, J.; Laitala, V.; Ma¨kinen, M. L.; Mulari, O.; Tamminen, J.; Hermonen, J.; Hurskainen, P.; Hemmila¨, I. Anal. Chem. 2004, 76, 14291436. (13) Ankelo, M.; Westerlund-Karlsson, A.; Ilonen, J.; Knip, M.; Savola, K.; Kankaanpa¨a¨n, P.; Merio ¨, L.; Siitari, H.; Hinkkanen, A. Clin. Chem. 2003, 49, 908-915. (14) Wahler, D.; Reymond, J. L. Curr. Opin. Biotechnol. 2001, 12, 535-544. (15) Philipps, B.; Hennecke, J.; Glockshulber, R. J. Mol. Biol. 2003, 327, 239249. 10.1021/ac049465o CCC: $27.50

© 2004 American Chemical Society Published on Web 06/26/2004

in recent years. FRET assays involve a donor-acceptor pair of fluorescent labels that are in most approaches attached to interacting species; currently FRET is mainly used for the screening of protein-protein interactions but also finds applications in the screening of enzymes and receptors. Most FRET measurements are based on fluorescence donors that emit light on the 10-ns scale, i.e., the same time scale typically associated with sample components causing background fluorescence. Recently, the Mathis group reported a novel way of performing FRET measurements by utilizing a donor that provides long-lived luminescence, lasting as long as ∼1 ms, without the need to deoxygenate the sample solution.17 The acceptor is a strongly fluorescent compoundsas is usual in FRETswith an absorption spectrum that should properly overlap with the donor emission spectrum. Interestingly, however, in a good approximation the acceptor luminescence observed after FRET has the same long lifetime as the donor luminescence (as in the well-known phenomenon of “delayed fluorescence”). This technology has been developed for immunoassays, enzyme assays, binding assays, protein-protein interactions, and cell surface receptor studies. To enable time-resolved FRET, special europium cryptate compounds are available that can be used for labeling of, for example, proteins and active ligands. Europium cryptate has a long-lived luminescence, and the result is a slow transfer (microto millisecond scale) of excitation energy to the acceptor. This allows measuring the FRET emission of the acceptor after a relatively long delay, without any interference from the short-lived medium background or from fluorescence of the acceptor. Direct excitation of the acceptor leading to short-lived acceptor fluorescence does not result in any signal under time-resolved detection conditions. The time-resolved FRET (TR-FRET) methods described so far have been developed for micro titer plate (MTP)-based assay techniques. While MTP-based assays represent the majority of HTS formats currently used in drug discovery, the static format of these assays essentially prevents a link to other analysis systems such as LC-MS, required to determine the chemical structure of active compounds in complex mixtures such as natural extracts. In recent years, we have described several methodologies where HTS assays are carried out in a continuous-flow format in order to measure simultaneously the chemical and biochemical characteristics of bioactive compounds.18-20 In contrast to MTP assays where fluorescence plate readers are required for assay readout, flow-based assays are based on flow-through detectors such as standard HPLC fluorescence detectors. In the present paper, we describe the development of a continuous-flow TR-FRET assay system. Next to the fundamental aspects of continuous-flow TR-FRET, we describe the development of homogeneous flow injection analysis assay using a tyrosine kinase, epidermal growth factor receptor (EGFR), as a model. EGFR is a membrane protein and plays a crucial role in signal tranduction pathways.21 Since EGFR is involved with oncogenesis (16) Stauffer, S. R.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125, 6977-6985. (17) Bazin, H.; Trinquet, E.; Mathis, G. Rev. Mol. Biotechnol. 2002, 82, 233250. (18) Derks, R. J. E.; Hogenboom, A. C.; van der Zwan, G.; Irth, H. Anal. Chem. 2003, 75, 3376-3384. (19) Hogenboom, A. C.; de Boer, A. R.; Derks, R. J. E.; Irth, H. Anal. Chem. 2001, 73, 3816-3823. (20) Hirata, J.; Ariese, F.; Gooijer, C.; Irth, H. Anal. Chim. Acta 2003, 478, 1-10.

and tumor cell proliferation, it has recently become an interesting target for drug discovery.22-24 Inhibition effects on the kinase assay reaction will also be studied. Emphasis is on the development of a detection system based on pulsed laser excitation and compatible with the features of TR-FRET and with the requirements of a flow dynamic system. EXPERIMENTAL SECTION Chemicals. Human epidermal growth factor (EGF), human EGFR, 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES), potassium fluoride, adenosine 5′-triphosphate (ATP), manganese chloride, magnesium chloride, polyoxyethylenesorbitan monolaurate (Tween 20), and tyrphostin AG1478 were purchased from Sigma-Aldrich (Zwijndrecht, The Netherlands). Ethylenediaminetetraacetic acid (EDTA) was purchased from Janssen Chimica (Geel, Belgium). Dimethyl sulfoxide (DMSO), spectrophotometric grade 99%, was obtained from Acros Organics (Geel, Belgium). Biotin-conjugated poly(Glu, Tyr) substrate (poly(GT)-biotin), streptavidin labeled with XL665 (SA-XL665), and antiphosphotyrosine antibody labeled with europium cryptate (Ab-EuK) were kindly donated by CIS Bio International (Bagnols-sur-Ce`ze, France). All aqueous solutions were prepared with water purified with a Milli-Q system from Millipore (Bedford, MA). Preparation of Samples. The EGF stock solution was stored at -20 °C and the EGFR stock solution at -80 °C. The ATP mix solution, containing the final concentration of 100 µM ATP, 30 mM Mg2+, and 6 mM Mn2+, was freshly made prior to use. The stock solutions of poly(GT)-biotin and Ab-EuK were both aliquoted and stored at -20 °C. The SA-XL665 stock solution was also aliquoted and stored at -80 °C. All solutions were thawed prior to use. For a kinase assay, 50 mM HEPES buffer (pH 7.4) was used. As a TR-FRET detection buffer, 50 mM HEPES (pH 7.0) including 100 mM potassium fluoride and 0.1% (v/v) Tween 20 were used. Tyrphostin AG1478 (inhibitor) was dissolved in DMSO and stored at -80 °C. Prior to performing the inhibitor assays, the stock solution was thawed and diluted with DMSO. Kinase Assay. The principle of the kinase assay is illustrated in Figure 1(a). EGFR was used for phosphorylation of poly(GT)biotin (substrate).25 At first, EGFR was activated by EGF; 5 µL of the 10 µM EGF solution was added to 5 µL of the EGFR solution (0.05 units/µL) and the resultant mixture incubated in a thermostated water bath at 30 °C for 10 min. Then 5 µL of the poly(GT)biotin solution (the final concentration was 7 nM, unless specified otherwise) and 15 µL of the ATP mix solution were added and incubated at 30 °C for 10 min unless specified otherwise. Then, 5 µL of the EDTA solution (final concentration 70 mM) was added to the solution in order to stop the kinase assay reaction. This procedure led to a partial phosphorylation of the tyrosine residues; ∼1.3% of the tyrosine residues was phosphorylated. Binding Reaction. Following the kinase assay, the binding reaction was performed (Figure 1b). Then, 17.5 µL of an Ab-EuK (21) Hackel, P. O.; Zwick, E.; Prenzel, N.; Ullrich, A. Curr. Opin. Cell Biol. 1999, 11, 184-189. (22) Levitzki, A.; Gazit, A. Science 1995, 267, 1782-1788. (23) Fabbro, D.; Ruetz, S.; Buchdunger, E.; Cowan-Jacob, S. W.; Fendrich, G.; Liebetanz, J.; Mestan, J.; O’Reilly, T.; Traxler, P.; Chaudhuri, B.; Fretz, H.; Zimmermann, J.; Meyer, T.; Caravatti, G.; Furet, P.; Manley, P. W. Pharmacol. Ther. 2002, 93, 79-98. (24) Levitzki, A. Eur. J. Cancer 2002, 38, S11-S18. (25) Ellis, A. G.; Nice, E. C.; Weinstock, J.; Levitzki, A.; Burgess, A. W.; Webster, L. K. J. Chromatogr., B 2001, 754, 193-199.

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4293

Figure 2. Schematic to illustrate the FIA system. The pump is connected to a superloop (10 mL) to deliver a carrier buffer at a flow rate of 150 µL/min. Samples are injected and excited in a 4-µL flow cell by a 266-nm Nd:YAG laser. The emissions at 625 and 665 nm are recorded by two identical photomultipliers and a dual-channel photon counter. To achieve optimum signal-to-background several filters are inserted; F1, 335-nm cutoff filter; F2, 580-nm cutoff filter; F3, 625-nm interference filter; F4, 665-nm interference filter.

Figure 1. Schematic to illustrate the kinase assay principle and the TR-FRET detection method.

solution (final concentration 6 nM, unless specified otherwise) and 17.5 µL of a SA-XL665 solution (final concentration 7 nM, unless specified otherwise) were prepared in the TR-FRET detection buffer, separately. The Ab-EuK solution and the SAXL665 solution were added to the kinase assay solution (total volume of 70 µL). After the solution was gently vortexed, it was incubated for 10 min at 30 °C. Inhibitor Assay. At first, EGF, EGFR, and the inhibitor tyrphostin AG1478 in 8% DMSO (v/v, final concentration of the total assay solution) or 8% DMSO only (blank solvent control experiment) were incubated for 10 min. After the incubation, the procedure was the same as described in the sections on kinase assay and binding reaction. Measurement of Excitation and Emission Spectra. An AbEuK solution and a SA-XL665 solution were prepared in the TRFRET detection buffer; the final concentrations of Ab-EuK and SA-XL665 were 12.5 and 100 nM, respectively. Both the kinase assay and the binding reaction were performed using the same protocols as used for the flow injection analysis (FIA) experiments. The excitation and emission spectra of the samples were recorded at specific time points in a thin quartz cuvette (dimension 1 mm × 10 mm) on a Perkin-Elmer luminescence spectrometer LS-50B (Perkin-Elmer, Beaconsfield, U.K.). To record the emission spectra, the sample was excited at 310 nm, corresponding with the excitation maximum, using a pulsed Xe lamp. The delay time was set at 100 µs and the gate time at 2 ms. The data acquisition was performed by the software package FL WinLab version 3.00 running under Windows 2000. Flow Injection Analysis. The FIA setup is schematically shown in Figure 2. A syringe pump PHD 2000 (Harvard Apparatus 4294 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

Inc., Holliston, MA) was first connected to a 10-mL Superloop (Pharmacia Biotech, Uppsala, Sweden). The Superloop was used for delivery of the buffer solution in order to avoid unwanted interferences to the system from pump materials. The flow rate was set to 150 µL/min. A Gilson (Villiers-le-Bel, France) autoinjector 234 equipped with a Rheodyne (Cotati, CA) six-port injection valve (injection volume 4 µL) was fitted for the injection of the samples (dissolved in the same buffer system). The light from a quadrupled Nd:YAG NanoUV laser (Uniphase, Meylan, France), emitting at 266 nm with an average output power of 5.4 mW, a pulse width of 0.34 ns, and a repetition rate as high as 7.8 kHz was focused into the 4-µL flow cell by a quartz lens with a focal distance of 3.5 cm and a cylindrical quartz lens with a focal distance of 2 cm. The rectangular spot matched the dimension (size 1 × 4 mm) of the flow channel as well as possible; the flow cell was positioned perpendicularly to the laser beam. The flow cell was masked with black rubber in order to avoid detection of emission from the cell walls. Two red-sensitive 9558 QA photomultiplier tubes (EMI, Middlesex, U.K.) used for detection were connected to a 456H power supply operated at 1100 V. One of the photomultipliers was provided with a 625 ( 8 nm interference filter to collect the emission of Ab-EuK (reference signal). The other photomultiplier with a 665 ( 5 nm interference filter collected the SA-XL665 emission (TR-FRET signal). Both photomultipliers were equipped with a 7-cm-diameter quartz lens with a focal distance of 9 cm; 335- and 580-nm cutoff filters were used to block scattered excitation light and ambient light. Photon counter thresholds were set at 10 mV. The data acquisition was performed by software (homemade) running under Windows 2000. RESULTS AND DISCUSSION TR-FRET Excitation and Emission Spectra. Before constructing the continuous-flow system, batch experiments were carried out to study the TR-FRET system and the accompanying spectral changes. The excitation and emission spectra of the donor Ab-EuK, recorded in a cuvette (excitation by means of a pulsed Xe lamp at 310 nm, using a delay time of 100 µs and a gate time of 2 ms), are depicted in Figure 3. From the excitation spectrum

Figure 3. (A) Excitation spectrum of Ab-EuK (emission wavelength, 620 nm), (B) long-lived fluorescence emission spectrum of Ab-EuK, (C) Long-lived fluorescence emission spectrum of SA-XL665, and (D) long-lived fluorescence emission spectrum of the kinase assay solution. All spectra were recorded in the buffer ([HEPES] ) 50 mM, [KF] ) 100 mM, pH 7.0; delay time, 100 µs; gate time, 2 ms). The excitation wavelength for the emission spectra was 310 nm.

in Figure 3A, it is obvious that the donor can be excited reasonably efficiently at 266 nm, although it is not the optimum excitation wavelength. In Figure 3B, the typical europium cryptate (concentration 12.5 nM) emission spectrum is seen (emission maximums: 583, 604, 620, and 700 nm). A fluorescence spectrum of 100 nM SA-XL665 in the TR-FRET buffer was also obtained using the same delay and gate time as in Figure 3B. After a 100-µs delay, there is no measurable emission left from the 100 nM SA-XL665 or from the buffer solution (Figure 3C), whereas the normal fluorescence spectrum of unbound SA-XL665 (when measured without delay) shows a maximum at 665 nm (not shown). The strength of TR-FRET as a minimal-background detection technique will be obvious from these spectra: despite the high fluorescence yield of SA-XL665 and even at a high concentration of 100 nM, under time-resolved conditions no signal is observed at 665 nm. After performing the standard kinase assay (reaction time 10 min), the solution was incubated with Ab-EuK (donor) and SAXL665 (acceptor) for 10 min (after 10 min of binding reaction time, the signal reached a plateau; data not shown). As illustrated in Figure 1, the streptavidin moiety of the acceptor molecule is bound to the biotin moiety of the substrate, while the antiphosphotyrosine part of the donor is connected to the phosphorylated tyrosine residues. In Figure 3D, the europium emission has decreased compared to Figure 3B and a new band at 665 nm shows up, i.e., at the maximum emission wavelength of SA-XL665. Without phosphorylation this new band did not appear. This delayed emission of SA-XL665 is observed only if there is substantial energy transfer from the Ab-EuK (donor) to the SA-XL665 (acceptor). For that process to be efficient at nanomolar concentrations, donor and acceptor need to be bound to the same substrate. Apparently, the flexibility of the peptide chain is sufficient for the donor and acceptor to come within the Fo¨rster distance within the lifetime of the excited state of the donor. Monitoring the long-lived emission at 665 nm thus makes it possible to measure the phosphorylation of the substrate without separating the products in the solution.

Laser-Excited TR-FRET Detection in a Flow System. It should be realized that the nitrogen laser-based detection system used in the micro titer plate TR-FRET bioassay cannot be simply adopted for a flow dynamic system as in FIA or eventually HPLC. In flow systems the flow cell dimensions should be small enough to exclude band broadening: in our setup, we used a cylindrically shaped flow cell with a volume of 4 µL (internal diameter 1 mm, length 4 mm). Even at a flow rate as low as 150 µL/min as utilized below, a plug of 4 µL would reside only during ∼1.6 s in the flow cell. It will be obvious that for such systems a pulsed laser like a nitrogen laser with a low repetition rate of typically 10 Hz is not appropriate: only 16 shots per sample can be measured. Furthermore, low repetition rates and accompanying higher peak powers tend to cause unwanted side effects such as photodegradation or possible cell damage. Also, the extinction coefficient of the AbEuK donor is rather low at 337 nm. That is the reason why for this work a recently commercialized quadrupled Nd:YAG laser such as the nano UV laser was tested: because of its repetition rate of 7.8 kHz, ∼1.2 × 104 shots/sample can be recorded, i.e., an improvement by a factor of almost 3 orders of magnitude. For our present purpose, too high repetition rates should be avoided as well: the time span between two pulses should be of the same order of magnitude as the TR-FRET lifetimes. For the nano UV laser, the pulse-to-pulse time span of 128 µs matches well with the luminescence lifetime of ∼500 µs. In practice, both photon counters were set at a delay time of 50 µs and a gate time of 75 µs. The utilization of a delay time of 50 µs would seem exaggerated, in view of the fact that the laser pulses have a width of only 0.3 ns and fluorescence background (direct emission from SAXL665, proteins, impurities) will last no longer than 100 ns. We found, however, that the strong fluorescence signal, mainly coming from directly excited, unbound SA-XL665, caused a perturbation of the dual-channel photoncounter thatsdepending on its intensityslasted up to 50 µs, which explains the use of the above delay. This also explains why the SA-XL665 concentration Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4295

was kept at the low-nanomolar level, see below. The average output of the quadrupled Nd:YAG laser, though being only 5.4 mW (at 266 nm), caused serious problems (e.g., bubble formation due to overheating, Ab-EuK degradation) if directly focused in the flow cell at a spot size of ∼0.2 × 0.2 mm2. This can readily be understood in view of the high spectral irradiances provided by such a pulsed laser system. Overheating and photodegradation could be avoided by defocusing the laser and creating a rectangular spot of about 1 mm wide and 4 mm high, guaranteeing optimum irradiation of the flow cell and resulting in 100-fold reduction of irradiance. Under these conditions, saturation does not occur. It should be noted that excitation at 266 nm does not pose a real problem, although the molar absorptivity at this wavelength is somewhat lower than at the excitation maximum of the Eu-cryptate complex at 310 nm (see Figure 3A). Of course, the use of wavelengths longer than 266 nm would be more favorable since that could reduce the impact of interferences; fortunately, under time-resolved measurement conditions, interferences were only of minor importance. Certain additives (e.g., proteins, buffers) that absorb at 266 nm could cause inner filtering and should not be added at high concentration. For that reason, Tween 20 was selected as surfactant/blocking agent instead of Triton X-100 or bovine serum albumin (BSA). Optimization of Flow Injection Analysis. In terms of costs per assay, the amounts of donor and acceptor should be kept as low as possible. On the other hand, the donor and acceptor concentrations should not be too low compared with that of the substrate, to avoid the situation that donor and acceptor would end up bound to different substrate molecules and not give a TRFRET signal. In addition, if the concentration of the acceptor is too high, its fluorescence will be too strong so that it could contribute to the background noise at 665 nm. At first, the influence of the Ab-EuK concentration was investigated using a substrate concentration of 7 nM and an acceptor concentration of 1 nM. As expected, the signals at 625 nm (reference signal) and at 665 nm (FRET signal) increased linearly upon enhancing the Ab-EuK concentration from 0 to 6 nM (not shown). Since there is on average less than one phosphorylated tyrosine residue per substrate, an excess of donor would remain unbound and would not lead to stronger TR-FRET signals. For a sufficiently high signal at 665 nm and still a relatively good cost performance, a 6 nM concentration of the Ab-EuK was selected. Next, the influence of the concentration of SA-XL665 was optimized. Streptavidin has four binding sites to biotin (see Figure 1b); therefore, four different SA-biotin complexes exist (SA/ biotin: 1:1, 1:2, 1:3, 1:4) in a solution. Using a 6-7 nM concentration of the SA-XL665 solution (the substrate concentration is 7 nM), the complexation between SA and biotin is 99% completed26 and there is no need to add a larger excess of SA-XL665. In Figure 4, at a 2 nM SA-XL665 concentration, the FRET signal has already reached a plateau value, in agreement with the multiple binding sites at SA. It was decided to use a 7 nM concentration of SAXL665 for further experiments. Figure 5 illustrates the standard operation of the flow injection analysis; each injection is repeated three times. Injections I refer to the complete kinase assay; injections A-H are control experiments in which one or more ingredients were left out. Two traces (26) Green, N. M. Methods Enzymol. 1990, 184, 51-67.

4296 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

Figure 4. Influence of the concentration of SA-XL665 on the peak area at 665 nm; [poly(GT)-biotin] ) 7 nM; [Ab-EuK] ) 6 nM. The injection volume was 4 µL. All experiments were carried out three times, applying triplicate injections for each experiment. Error bars show the standard deviation.

are shown: at 625 nm the emission of Ab-EuK is recorded (reference signal) and, simultaneously, at 665 nm the delayed emission from SA-XL665 (TR-FRET signal). In the ideal situation, the TR-FRET signal at 665 nm should be completely zero if not all the essential constituents of the kinase assay are present, whereas in the presence of all reagents, it should be strong. In our setup, we are not far away from this objective: significant signals are only obtained for the injections denoted as (I), while the weak responses discernible for the injections F-H must be due to diffusion-controlled FRET between unbound donor and acceptor in solution. These findings confirm that FRET is only efficient if both the donor and the acceptor are bound to the same phosphorylated substrate. In solution at these low concentration levels, the average distance between unbound Ab-EuK and SAXL665 is too large for FRET to occur efficiently and the longlived 665-nm signal is very low, but not completely absent, as can be understood as follows. In the case of conventional FRET occurring at a nanosecond time scale, energy transfer between unbound donor and acceptor can only occur at high concentrations and can be fully neglected at nanomolar levels because the average distances are of the order of several hundred nanometers, and diffusion distances are limited to a few nanometers. In our case, however, the much longer lifetime of the donor means that via diffusion donor and acceptor can come within the Fo¨rster distance of 5-10 nm, even when the diffusion coefficients are relatively low due to the large macromolecules attached. This remaining background is of a fundamental nature, not distinguishable from the signal by means of time discrimination, and will determine the final detection limit of the method. Interestingly, the 625-nm trace only shows a response if AbEuK is a constituent of the injected solution. Although there is an obvious variation in peak heights, careful analysis reveals that the variation in peak areas is relatively small, less than 8%. Apparently, the long-lived excitation and emission characteristics are rather independent from influences of other chemicals. The 625-nm signals obtained in series I are ∼30% lower in intensity, which can be readily explained since under these conditions energy is transferred to the acceptor.

Figure 5. Flow injection analysis profile of 625-nm emission (top) and 665-nm emission (bottom), showing three repetitive injections. (A) Blank solution (50 mM HEPES, 100 mM KF), (B) 6 nM Ab-EuK, (C) 7 nM SA-XL665, (D) 6 nM Ab-EuK and 7 nM poly(GT)-biotin, (E) 7 nM SA-XL665 and 7 nM poly(GT)-biotin, (F) 7 nM SA-XL665, 6 nM Ab-EuK and 7 nM poly(GT)-biotin without additives, (G) kinase assay without EGFR/EGF (all other concentrations are same as in (F)), (H) 6 nM Ab-EuK and 7 nM XL665, and (I) kinase assay (the concentrations of the chemicals are the same as in (F)).

Figure 6. Effect of inhibition. Flow injection profile for (A) kinase assay solution without inhibitor and (B) kinase assay solution with inhibitor (100 nM tyrphostin AG1478). Both solutions contain 6 nM Ab-EuK, 7 nM SA-XL665, 7 nM poly(GT)-biotin, and 8% DMSO.

Of course the signals at 665 and 625 nm are interrelated. The signal decrease is governed by ΦT, the efficiency of energy transfer, since for the intensity at 625 nm the following equation applies:27

I(625) ∝ I0DcDLΦ°D(1 - ΦT)

(1)

wherein I0 is the intensity of the excitation light, D is the molar extinction coefficient of Ab-EuK at 266 nm, cD is the Ab-EuK concentration, L is the optical path length, and Φ°D is the luminescence efficiency of the donor in the absence of acceptor. The signal intensity at 665 nm, under the experimental conditions at hand obeys the following equation:27

I(665) ∝ I0DcDLΦTΦ°A

(2)

wherein the symbols have the same meaning as in eq 1, while Φ°A is the acceptor fluorescence quantum yield. Comparison of (27) Valeur, B. Molecular fluorescence; principles and applications; Wiley-VCH: New York, 2002.

Figure 7. Effect of varying the concentration of tyrphostin AG1478 inhibitor on the EGFR-induced kinase phosphorylation reaction. Conditions are given in the Experimental Section. The data given are means of three experiments, each injected in triplicate; error bars show the standard deviation.

eqs 1 and 2 reveals that the decrease at 625 nm is not necessarily identical to the increase at 665 nm, since Φ°A * Φ°D and furthermore the detection efficiencies are not identical. Inhibition Experiments. Even more interesting is Figure 6, showing the effect of an inhibitor, i.e., tyrphostin AG1478, present at a concentration of 100 nM. Since this inhibitor of the phosphorylation reaction with EGFR can hardly be solved in water, 8% DMSO (v/v) was added to the kinase assay. Figure 6 shows the results after incubation for 10 min in the absence (solvent control) and in the presence of inhibitor. Obviously, the inhibitor causes a significant signal reduction at 665 nm, the remaining signals being close to the noise level. Simultaneously, the signals in the 625-nm trace become roughly 30% more intense, in line with an almost complete absence of energy transfer from the unbound donor. Tyrphostin AG1478 is a strong and specific inhibitor for EGFR.28 To obtain the IC50, various concentrations of tyrphostin (28) Osherov, N.; Levitzki, A. Eur. J. Biochem. 1994, 225, 1047-1053.

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4297

AG1478 were tested (Figure 7). Peak areas of the signals at 665 nm were plotted as a function of the concentration of tyrphostin AG1478. All data points are based on three independent experiments (and for each experiment three injections), and error bars show the standard deviation. The IC50 value was obtained as (5.1 ( 0.4) × 10-9 M, which is in line with the literature value of 3 nM.28 CONCLUSIONS By using a simple time-resolved detection system based on a robust, small-sized laser emitting at 266 nm at an appropriate repetition rate of 7.8 kHz, the TR-FRET technique can be readily adapted to flow dynamic systems. The setup was sufficiently sensitive for reagent concentrations at the low-nanomolar level. The system could be used to measure the inhibition of the phosphorylation reaction by tyrphostin AG1478 and determine the

4298

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

IC50. The method outlined here shows all the advantages inherent to TR-FRET as known from bioassay micro titer plate results; timeresolved detection allows for the efficient rejection of fluorescence background from interfering components in the sample and from the reagents. Current work is aimed at further reduction of the background and coupling the system to liquid chromatography. ACKNOWLEDGMENT The authors thank the Dutch Foundation for the Advancement of Science (NWO-CW) for financial support (Grant 99032) and Dr. G. Mathis of CIS Bio International (France) for providing us with the TR-FRET reagents. Received for review April 7, 2004. Accepted May 18, 2004. AC049465O