Modular, Antibody-free Time-Resolved LRET Kinase Assay Enabled by Quantum Dots and Tb3+-sensitizing Peptides

Fluorescent drug screening assays are essential for tyrosine kinase inhibitor discovery. Here we demonstrate a flexible, antibody-free TR-LRET kinase assay strategy that is enabled by the combination of streptavidin-coated quantum dot (QD) acceptors and biotinylated, Tb3+ sensitizing peptide donors. By exploiting the spectral features of Tb3+ and QD, and the high binding affinity of the streptavidin-biotin interaction, we achieved multiplexed detection of kinase activity in a modular fashion without requiring additional covalent labeling of each peptide substrate. This strategy is compatible with high-throughput screening, and should be adaptable to the rapidly changing workflows and targets involved in kinase inhibitor discovery.

Scientific RepoRts | 6:28971 | DOI: 10.1038/srep28971 We previously reported phosphorylation-sensitive lanthanide binding peptides as specialized substrates for tyrosine kinases 20 . As also reported by others 21,22 , these substrates chelate lanthanide ions directly upon phosphorylation, eliminating the need for chemical labeling with a separate lanthanide chelate [22][23][24] , resulting in higher lanthanide luminescence intensity and longer luminescence lifetime 20,21,23 . The in silico workflows developed in our lab have ensured the optimal kinase specificity as well as lanthanide binding affinity simultaneously for biosensors that are newly designed 25 or engineered from existing substrates 26 , providing the foundation of multiplexed kinase assay. We have explored the design and application of such sequences 25 for novel time-resolved luminescence kinase assays in TRL and TR-LRET forms 20,27 for a variety of kinases involved in cancer signaling, including a dual-plexed approach using small molecule fluorophores to differentiate between substrates 27 . While our previous approach is functional and high-throughput compatible, its modularity was not optimal-requiring covalent fluorophore labeling and purification of each individual peptide substrate. Here we report a more flexible strategy for a multiplexed, antibody-free kinase assay using TR-LRET between quantum dot (QD) fluorophores and phosphorylation-dependent lanthanide-sensitizing peptide biosensors. Because of the broad and continuous absorption spectra of QDs, which are highest in the UV to short wavelength visible range regardless of emission color (Fig. 1a), the luminescence emission from Tb 3+ is efficiently exploited and provides more flexible LRET pair options (Fig. 1a) than conventional organic fluorophores.

Results and Discussion
As depicted in Fig. 1b, QD-peptide conjugates could be readily prepared via the interaction between tetravalent streptavidin and biotin. Streptavidin-coated 28 QD605 or QD655 were incubated with SAStide (GGDEEDYEEPDEPGGK b GG, a Syk biosensor 25 ) or SFAStide-A (GGEEDEDIYEELDEPGGK b GG, a Src family kinase biosensor 25 ) at room temperature in HEPES buffer (pH = 7.5) for 1 hr. To confirm the formation of the conjugates, agarose gel electrophoresis 14 was performed (Fig. 1c,d). QD-biosensor conjugates exhibited increased electrophoretic mobility relative to unlabeled QDs. QDs were saturated (exhibiting no further electrophoretic mobility changes) when biosensor/QD ratio reached about 200:1. Conjugate formation was resistant to high concentrations of urea (Fig. 1e), whether the urea addition was prior to or after the incubation process. The QD-biosensor conjugates were also subject to ligand exchange tests, in which the pre-made conjugates showed no signal loss caused by ligand exchange with another labeled QD or excess amount of free biosensors in buffer during a 4 hr incubation (Figures S12 and S13), indicating their high stability (deriving from the high affinity of the streptavidin-biotin interaction).
The peptide biosensors were designed to chelate free Tb 3+ cation when phosphorylated by tyrosine kinases as mentioned above and described in our previous publications 20, 25,26 . Once the conjugates were prepared, luminescence emission was generated upon UV excitation, from both the QD (short-lived fluorescence) and the peptide-chelated Tb 3+ (long-lived luminescence) (Fig. 2). Indeed, steady state emission of the conjugates was dominated by inherent QD fluorescence regardless of peptide phosphorylation state as expected (Fig. 3a), given the much higher extinction coefficients of QDs 28 (4.4 × 10 6 cm −1 M −1 for QD605 at 350 nm and even higher for QD655) than that of Tb 3+ chelates 29 (~1.2 × 10 4 cm −1 M −1 at 327 nm). However, since QD fluorescence lifetimes are typically less than a hundred nanoseconds, QD inherent emission was greatly reduced by applying a sufficient time gate between excitation and detection (50 μ s or longer), while the luminescence emission from Tb 3+ remained strong. Since Tb 3+ bound to phosphorylated biosensors had higher luminescence intensity and longer lifetime 20 than the unphosphorylated conjugate, phosphorylation of QD-biosensor conjugates resulted in stronger LRET emission from the QDs (Figs 3a, S10 and S12). Background QD emission in unphosphorylated conjugate samples was low, and was a combined result of residual QD steady-state fluorescence 30 , free Tb 3+ binding by buffer components, and unphosphorylated biosensor, which had far lower emission than for the phosphorylated biosensor (as previously described 20,25,26 and observed here-arising from lower affinity for Tb 3+ cation and quenching of Tb 3+ luminescence by the water molecules that occupy the leftover ligand sites in the unphosphorylated chelate) (Figures S11 and S12). Such background emission could be alleviated by optimizing the reaction buffer components/concentrations and instrumental parameters for sufficient delay times. At  (a) Emission spectra comparing steady-state and time-resolved emission of the QD-biosensor conjugates used in this study. Steady state emission from QD605-pSAStide (10 nM) and QD655-pSFAStide-A (10 nM) conjugate are dominated by inherent QD emission. However, time-resolved LRET emission from both conjugates showed a significant difference between emission spectra of the phospho-and unphospho-conjugates. (b) In order to mimic the experimental conditions of kinase assay (as discussed in Figure S11), a calibration curve for QD605-SAStide was established in kinase reaction buffer. (c) In order to mimic the experimental condition of multiplexed kinase assay (as discussed in Figure S11), a calibration curve for QD655-SFAStide-A was established in kinase reaction buffer. (d) Emission spectra of samples from b) showed that TR-LRET emission increased as the percent of pSAStide (proportional to total biosensor) increased. (e) Emission spectra of samples from (c) showed that TR-LRET emission increased as the percent of pSFAStide-A (proportional to total biosensor) increased. For (b,c) the QD-biosensor conjugates were diluted to 20 nM with 100 μ M Tb 3+ in 10 mM HEPES buffer (pH = 7.5) and other kinase reaction reagents as described in Supporting Information. Time-resolved emission spectra were recorded using built-in monochromator on a Synergy 4 microplate reader with 250 μ s delay and 1000 μ s integration. Excitation wavelength: 266 nm. Calibration curves were established using luminescence reading from a 605/10 and a 655/10 filter accordingly with a 265/10 excitation filter. 10 nM concentration in the reaction buffer (to mimic multiplexed kinase reaction conditions as discussed in Figure S11), the signal/background for LRET emission from phosphorylated vs. unphosphorylated conjugates was 3:1 when delay time was 250 μ s, and could be even higher with longer delay time or higher concentration ( Figure S15). Time-resolved emission of both QD605-SAStide and QD655-SFAStide-A conjugates (prepared at biosensor: QD ratio of 100:1 to ensure saturation of the QD surface) showed linear increase with increasing proportion of phosphorylated conjugates, demonstrating that kinase assays using QD-biosensor conjugates can be appropriately quantified according to particular assay conditions and components (Fig. 3b,e). The proportion of phosphorylated conjugate was quantified by either the area under curve (AUC) of QD emission peak spectra or by using readings from designated emission filters, and calibration curves were constructed for quantitative detection of the amount of phosphorylation (Fig. 3b,c). Even for proportions of phosphorylated product of as low as 10% (considered the appropriate regime for kinetics experiments), the Z′ factors calculated for both conjugate (Tables S1-S3) were > 0.5, above the threshold that is generally considered the compatibility requirement for high-throughput screening.
An advantage of the QD-biosensor conjugates shown here is that the TR-LRET kinase assay could be set up with flexible protocols in which the conjugates were prepared either prior to or after the kinase reaction. In vitro Src and Syk activity assays were performed according to the procedure described in Supporting Information. Briefly, 20 nM of substrate, as either pre-made QD-biosensor conjugates or free biosensor peptide, was incubated with the reaction buffer for 5 min, and the reaction was started by addition of kinase(s). At selected time points, aliquots of the reaction mixture were taken out and quenched in Tb 3+ -containing detection buffer to a final volume of 100 μ L. Time-resolved, quantitative readout was done by either monochromator-based spectral scan or filter-based detection using a Biotek Synergy4 plate reader. In the single conjugate kinase assay, TR-LRET analysis of each conjugate displayed a corresponding increase in QD emission signal as a result of biosensor phosphorylation (Fig. 4a,c,e,g). Kinase activity was readily reported by both the conjugate formed prior to the reaction (Fig. 4e,g) and after the reaction (when the quenched aliquot was incubated with QD to capture biotinylated peptide) (Fig. 4a,c), showing that a QD conjugate strategy could be easily adapted to rapidly changing assay needs without prior chemical preparations such as covalent fluorophore labeling.
For multiplexed detection, QD605-SAStide and QD655-SFAStide-A conjugates were both prepared prior to the assay and then combined in reaction buffer before kinases were added (Fig. 4b,d,f,h). TR-LRET spectra for the conjugate mixture were collected by the same method above. To minimize the cross interference caused by the overlap of QD605 background emission with QD655 absorption, delay time for QD605-SAStide detection was increased to 400 μ s. In the presence of either Src or Syk, both conjugates showed increasing LRET emission corresponding to the specific respective kinase (Fig. 4b,d,f,h) used over the course of the reactions. Selective phosphorylation of QD605-SAStide by Syk (Fig. 4b,d) and selective phosphorylation of QD655-SFAStide-A by Src (Fig. 4f,h) were demonstrated in the mixture of both conjugates, and the result was in agreement with our previously published work on these two substrates 20,25 . TR-LRET emission can be quantified in percent of phosphorylation for either single conjugate assay (Fig. 4c,g) or multiplexed assay (Fig. 4d,h) using corresponding calibration curves (Figs 3b,c and S16). These results demonstrated the feasibility of using QD-biosensor conjugates in multiplexed TR-LRET kinase assays with a robust readout and flexible procedures.
In conclusion, this study has implemented QDs as the LRET acceptor in a time-resolved LRET kinase assay strategy that uses phosphorylation-sensitive Tb 3+ -chelating synthetic peptides, demonstrated with tyrosine kinases Src and Syk. Previous studies have suggested that QDs are not ideal LRET acceptors in steady state detection 31 , indicating the importance of using time-resolved detection for the assays discussed in this study. The major advantage of this strategy is that mainstream TR-LRET assays used in kinase inhibitor screening practices today are highly dependent on chemically-labeled, customized antibodies 1,5 , which are high-cost and require significant, and unique, investment to implement per target. By using validated lanthanide binding peptide biosensors for the kinases of interest 25 , which are inexpensive to produce and do not require any labeling other than standard incorporation of biotinyl-Lysine during their synthesis, the strategy reported here enables flexible combinations of QD/biosensor LRET pairs for assaying a broad, expandable list of kinases 25 in a cost efficient manner. Additionally, this approach is more modular than our previously reported dual-plexed LRET assay 27 , since the choice of LRET donor and acceptor can be decided immediately before the assay without any special chemical labeling planned ahead. The simple nature of this antibody-free assay also renders it more adaptable to the rapidly changing needs of new assay targets, and thus it could be a valuable tool for drug discovery efforts.

Methods and Materials
Peptide synthesis. Peptides were prepared to > 90% purity using standard Fmoc solid phase peptide synthesis procedure as described in the Supporting Information. Briefly, peptides were synthesized using a Symphony X peptide synthesizer (Protein Technologies, USA) on Rink-amide-MBHA resin (Protein Technologies, USA). The synthesis was performed using Fmoc (9-fluorenylmethoxy-carbonyl)-protected amino acids (Protein Technologies, USA) with coupling reagent 2-(6-chloro-1H-benzotria-zole-1-yl)-1,1,3,3-tetramethylaminiumhexafluorophosphate (HCTU, Protein Technologies, USA) in the presence of N-methylmorpholine (NMM, Protein Technologies, USA) in dimethylformamide (DMF, Iris Biotech Gmbh, Germany). A 20% piperidine solution in DMF was used for Fmoc deprotection. Peptide cleavage and side-chain deprotection were performed simultaneously by using 10   Calbiochem, USA) buffer (pH = 7.5). The final solutions had 10 nM QDs, various amount of peptides, and 2.4 M urea when indicated. After incubation for 1 hr at room temperature, glycerol (100%, Macron Fine Chemicals, USA) was added to each sample with a final concentration of 5% (v/v). The QD-biosensor conjugates were then loaded to a 10 cm long 1% (w/v) agarose (Invitrogen, USA) gels (4 μ L sample per well) in 1 × TAE buffer (Thermo Fisher Scientific, USA), and run for 60 min at 100 V using BioRad PowerPac Basic power supply. Gels were then imaged using a Gel Logic 112 imaging system (Carestream, USA).
Fluorescence/luminescence measurements. Fluorescence/luminescence emission spectra were measured on a Synergy4 plate reader (Biotek, USA) at room temperature in 384-well black plates (Fluortrac 200, Greiner bio-one, Germany). The emission spectra were collected between 450 and 650 nm with 2 nm increments using the built-in monochromator. Filter-based detection using 550/10, 605/10, 655/10 emission filters (Omega Optical, USA) was also applied when necessary. The excitation wavelength was set to 266 nm by built-in monochromator, or a 265/10 excitation filter (Omega Optical, USA) with 250 μ s delay time unless otherwise indicated. Details are described in the Supporting Information.
Kinase assay. QDs were incubated with the biosensors in HEPES buffer for 1 hr to form the conjugates.
Alternatively, QDs could also be added to the quenched samples after the kinase assay. Final kinase reaction buffer (e.g. 15 nM Src, 5 μ M peptides, 100 μ M adenosine triphosphate (ATP), 10 mM MgCl 2 , 0.2 μ g/μ L BSA, 25 mM HEPES, pH 7.5) were incubated at 37 °C, and the reaction was initiated by the addition of kinase. At selected time points, 40 μ L aliquots were taken and quenched in detection buffer (40 μ L 6 M urea, 10 μ L 1 mM TbCl 3 and 10 μ L 1 M NaCl). Time-resolved luminescence emission spectra were collected as described in Supporting Information.