Rapid and High-Sensitivity Cell-Based Assays of Protein−Protein

Feb 24, 2010 - The new fragments from click beetle luciferase were applied for GPCR−β-arrestin interaction assays (Figure 2A). In fact, GPCR is a f...
0 downloads 0 Views 4MB Size
Download PDF
Anal. Chem. 2010, 82, 2552–2560

Rapid and High-Sensitivity Cell-Based Assays of Protein-Protein Interactions Using Split Click Beetle Luciferase Complementation: An Approach to the Study of G-Protein-Coupled Receptors Naomi Misawa,†,‡ A. K. M. Kafi,† Mitsuru Hattori,† Kenji Miura,‡ Kenji Masuda,§ and Takeaki Ozawa*,†,‡,| Department of Chemistry, School of Science, The University of Tokyo, 7-3-1 Bunkyo-ku, Hongo, Tokyo 113-0033, Japan, ProbeX Inc., 4-1-4 Bunkyo-ku, Hongo, Tokyo 113-0033, Japan, PRESTO, Japan Science and Technology Agency, 5-3 Chiyoda-ku Yonbanchou, Tokyo 102-8666, Japan, and Tsuruga Institute of Biotechnology, Toyobo Co., Ltd., 10-24, Toyo-cho, Tsuruga, Fukui, 914-0047, Japan To identify biologically relevant compounds in basic biology and drug discovery processes, rapid quantitative methods for elucidating protein-protein interactions have become necessary. We describe a novel optical technique for monitoring protein-protein interactions in living cells based on complementation of split luciferase fragments from click beetle (Brazilian Pyrearinus termitilluminans). A new pair of amino-terminal and carboxy-terminal fragments of the luciferase was identified using semirational library screening, demonstrating achieved markedly higher sensitivity and signal-to-background ratio. The identified fragments were applied to the study of five G-protein coupled receptors (GPCR) that interact with β-arrestin on the plasma membrane. By generating cell lines stably expressing the GPCRs and β-arrestin connected with the luciferase fragments, we demonstrated rapid and sensitive screening of potential chemicals that act on GPCRs using a 96-well microtiter plate format. The screening time was reduced to 5-10 min after ligand stimulation. The maximum response became more than 15-fold higher than the background signal. This luciferase complementation method also enabled accurate spatial and temporal analyses of interactions in single living cells using bioluminescence microscopy. These GPCR assays will facilitate developments of high-throughput screening systems in a multiwell plate format. Furthermore, using specific proteins of interest, the novel fragments of luciferase will provide different assay methods for the study of many intracellular signals in living cells and animals. Formation of protein-protein complexes is a central mechanism for generation of biological regulatory specificity. Compared with a single protein acting independently, complexes comprising different proteins can accomplish much more numerous biological * To whom correspondence should be addressed. E-mail: ozawa@ chem.s.u-tokyo.ac.jp. Tel.: 81-3-5841-4351. Fax: 81-3-5802-2989. † The University of Tokyo. ‡ ProbeX Corporation. § Tsuruga Institute of Biotechnology. | Japan Science and Technology Agency.

2552

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

functions. Many genetic methods have been developed to investigate the complicated network of protein-protein interactions and to screen chemicals that regulate the interactions specifically. In yeast or mammalian two-hybrid based assays,1,2 two proteins of interest induce a transcriptional activation, which can then be detected with a specific reporter protein. Because two-hybrid systems are in vivo assays, they offer advantages over in vitro biochemical methods. Indeed, some protein-protein interactions require specific post-translational modifications of the proteins and/or specific cofactors in the cellular microenvironment. However, the two-hybrid system works only in the nucleus that is close to the reporter gene. For that reason, techniques for identifying and studying protein-protein interactions in the cytosol have been developed to overcome this disadvantage. Mammalian protein-protein interaction trap (MAPPIT) is one applied twohybrid technique3,4 that enables detection of protein-protein interactions in the cytosol of living mammalian cells. Proteins of interest are linked to signaling deficient cytokine receptor chimeras. Interaction of proteins and ligand stimulation restores intracellular signaling, which ultimately engenders the transcription of a reporter under control of a specific promoter. The MAPPIT has been used for identifying novel proteins from cDNA library and for making high-throughput interactome analysis in model organisms.5 However, MAPPIT has difficulty, in principle, analyzing membrane proteins’ interactions or interactions in a specific organelle. (1) Fields, S.; Song, O. Nature 1989, 340, 245–246. (2) Fearon, E. R.; Finkel, T.; Gillison, M. L.; Kennedy, S. P.; Casella, J. F.; Tomaselli, G. F.; Morrow, J. S.; Van Dang, C. Proc. Natl. Acad. Sci. U. S. A. 1992, 89, 7958–7962. (3) Eyckerman, S.; Verhee, A.; der Heyden, J. V.; Lemmens, I.; Ostade, X. V.; Vandekerckhove, J.; Tavernier, J. Nat. Cell Biol. 2001, 3, 1114–1119. (4) Eyckerman, S.; Lemmens, I.; Catteeuw, D.; Verhee, A.; Vandekerckhove, J.; Lievens, S.; Tavernier, J. Nat. Methods 2005, 2, 427–433. (5) Simonis, N.; Rual, J. F.; Carvunis, A. R.; Tasan, M.; Lemmens, I.; HirozaneKishikawa, T.; Hao, T.; Sahalie, J. M.; Venkatesan, K.; Gebreab, F.; Cevik, S.; Klitgord, N.; Fan, C.; Braun, P.; Li, N.; Ayivi-Guedehoussou, N.; Dann, E.; Bertin, N.; Szeto, D.; Dricot, A.; Yildirim, M. A.; Lin, C.; de Smet, A. S.; Kao, H. L.; Simon, C.; Smolyar, A.; Ahn, J. S.; Tewari, M.; Boxem, M.; Milstein, S.; Yu, H.; Dreze, M.; Vandenhaute, J.; Gunsalus, K. C.; Cusick, M. E.; Hill, D. E.; Tavernier, J.; Roth, F. P.; Vidal, M. Nat. Methods 2009, 6, 47–54. 10.1021/ac100104q  2010 American Chemical Society Published on Web 02/24/2010

A more general approach for detecting protein-protein interactions includes methods based on protein-fragment complementation because this system shares several characteristics that set it apart from the methods described above for the study of protein-protein interactions in cells. Protein-fragment complementation methods are based on the fusion of complementary fragments of a reporter protein to two proteins of interest. Fragments of the reporter protein are brought within proximity; they spontaneously refold to generate a detectable signal if the two test proteins interact. Some noticeable complementation techniques and their related methods are split ubiquitin,6-8 dihydrofolate reductase,9,10 β-galactosidase,11,12 β-lactamase,13 green fluorescent protein and its variants,14,15 firefly luciferase,16-18 renilla luciferase,19,20 and click beetle luciferase.21 These methods present a strong advantage over the two-hybrid techniques in that the protein-protein interactions can be detected at the right place where the interactions occur in living cells and are therefore widely used now for basic studies in biology and drug discovery.22 Of the protein-fragment complementation methods, the use of split luciferases is particularly advantageous because of their applicability in imaging protein-protein interactions in intact cells and by direct application to small living animals because of tissue attenuation of emitted photons.23,24 Moreover, because luciferases emit light from a chemical reaction, they require no high-energy excitation source and do not suffer from background noise and tissue damage in signal measurements.25 We first demonstrated the basic concept using split firefly luciferase fragments for detection of protein-protein interactions; since then, others have successively demonstrated imaging of a particular protein-protein interaction in (6) Dunnwald, M.; Varshavsky, A.; Johnsson, N. Mol. Biol. Cell 1999, 10, 329– 344. (7) Johnsson, N.; Varshavsky, A. Proc. Natl. Acad. Sci. U. S. A. 1994, 91, 10340–10344. (8) Stagljar, I.; Korostensky, C.; Johnsson, N.; te Heesen, S. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 5187–5192. (9) Pelletier, J. N.; Arndt, K. M.; Pluckthun, A.; Michnick, S. W. Nat. Biotechnol. 1999, 17, 683–690. (10) Pelletier, J. N.; Campbell-Valois, F. X.; Michnick, S. W. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 12141–12146. (11) Rossi, F.; Charlton, C. A.; Blau, H. M. Proc. Natl. Acad. Sci. U. S. A. 1997, 94, 8405–8410. (12) Blakely, B. T.; Rossi, F. M.; Tillotson, B.; Palmer, M.; Estelles, A.; Blau, H. M. Nat. Biotechnol. 2000, 18, 218–222. (13) Wehrman, T.; Kleaveland, B.; Her, J. H.; Balint, R. F.; Blau, H. M. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 3469–3474. (14) Hu, C. D.; Kerppola, T. K. Nat. Biotechnol. 2003, 21, 539–545. (15) Ozawa, T.; Nogami, S.; Sato, M.; Ohya, Y.; Umezawa, Y. Anal. Chem. 2000, 72, 5151–5157. (16) Ozawa, T.; Kaihara, A.; Sato, M.; Tachihara, K.; Umezawa, Y. Anal. Chem. 2001, 73, 2516–2521. (17) Paulmurugan, R.; Umezawa, Y.; Gambhir, S. S. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 15608–15613. (18) Luker, K. E.; Smith, M. C.; Luker, G. D.; Gammon, S. T.; Piwnica-Worms, H.; Piwnica-Worms, D. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 12288– 12293. (19) Kaihara, A.; Kawai, Y.; Sato, M.; Ozawa, T.; Umezawa, Y. Anal. Chem. 2003, 75, 4176–4181. (20) Paulmurugan, R.; Gambhir, S. S. Anal. Chem. 2003, 75, 1584–1589. (21) Hida, N.; Awais, M.; Takeuchi, M.; Ueno, N.; Tashiro, M.; Takagi, C.; Singh, T.; Hayashi, M.; Ohmiya, Y.; Ozawa, T. PLoS One 2009, 4, e5868. (22) Michnick, S. W.; Ear, P. H.; Manderson, E. N.; Remy, I.; Stefan, E. Nat. Rev. Drug Discov. 2007, 6, 569–582. (23) Massoud, T. F.; Paulmurugan, R.; De, A.; Ray, P.; Gambhir, S. S. Curr. Opin. Biotechnol. 2007, 18, 31–37. (24) Massoud, T. F.; Gambhir, S. S. Genes Dev. 2003, 17, 545–580. (25) Contag, C. H.; Bachmann, M. H. Annu. Rev. Biomed. Eng. 2002, 4, 235– 260.

living mice.17,18 Subsequently, several pairs of luciferase fragments were identified such as split renilla luciferases19,20 and split Gaussia luciferase,26 which are suitable for studying protein-protein interactions in living cells and animals. Recently, we developed novel luciferase fragments from click beetles in green (Brazilian Pyrearinus termitilluminans; Emerald Luc; ELuc) and in red (Caribbean Pyrophorus plagiophthalamus; CBR) to visualize dynamic protein-protein interactions in living Xenopus embryos.21 The use of click beetle luciferase has the strong advantage of brightness; the photon count of click beetle luciferase is estimated as being over 10-fold higher than those of firefly luciferases. In addition, the spectrum of firefly luciferase is known to change in a pH-dependent manner, whereas click beetle luciferases have a property of pH independence of the spectra.27 Although the current methods present great utility, click beetle luciferase fragments have certain limitations in terms of sensitivity and the signal-to-background ratio. The split fragments used in our previous study failed to produce sufficient levels of signal-tobackground ratio.21 In fact, a small molecule of rapamycin, which induces binding of FK506-binding protein (FKBP) to FKBPbinding domain (FRB), led to restoration of luciferase activity. However, the signal changes between the presence and absence of rapamycin were only 20-fold because of higher background bioluminescence of spontaneous complementation between luciferase fragments. Further improvements in the sensitivity and signal-to-background ratio were necessary to facilitate basic studies of protein-protein interactions and to develop a high-throughput screening system for chemical libraries based on protein-protein interactions. In this work, we developed novel luciferase fragments from click beetle luciferase using semirational library screening with FKBP and FRB proteins. The best pair of luciferase fragments showed brighter bioluminescence and a higher signal-to-background ratio over the previous one. We applied the newly fragmented luciferase to the study of interactions of five G-proteincoupled receptors (GPCRs) with β-arrestin on the plasma membrane because GPCRs are active in most organs and present widely various opportunities as therapeutic targets in areas including diabetes, cancer, obesity, inflammation, and pain.28,29 Generating cell lines harboring GPCRs and β-arrestin connected with the luciferase fragments, we demonstrated that the luciferase fragments are useful for screening potential chemical candidates that act on GPCRs and for monitoring the ligand-induced β-arrestin-GPCR interactions in living mammalian cells. MATERIALS AND METHODS Materials. DNA polymerase, restriction, modification enzymes, and the cDNA library of human brain were obtained from Takara Bio Inc. (Japan). The cDNA of endothelin receptor type B (EDNRB) was obtained from GeneCopoeia Inc. (Rockville, MD). The cDNA of click beetle luciferase (ELuc), cDNA libraries of human placenta, and Emerald Luciferase assay reagent were (26) Remy, I.; Michnick, S. W. Nat. Methods 2006, 3, 977–979. (27) Viviani, V. R.; Arnoldi, F. G.; Neto, A. J.; Oehlmeyer, T. L.; Bechara, E. J.; Ohmiya, Y. Photochem. Photobiol. Sci. 2008, 7, 159–169. (28) Alkhalfioui, F.; Magnin, T.; Wagner, R. Curr. Opin. Pharmacol. 2009, 9, 629–635. (29) Conn, P. J.; Christopoulos, A.; Lindsley, C. W. Nat. Rev. Drug Discov. 2009, 8, 41–54.

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2553

obtained from Toyobo Co. Ltd. (Japan). Rapamycin, endothelin 1 (ET-1), and BIM23052 were purchased from Wako Pure Chemical Industries Ltd. (Japan). From Peptide Institute Inc. (Japan), (Pyr1)apelin-13 was obtained. We used with somatostatin from Calbiochem-Novabiochem AG (UK), gastrin-1 from GenScript USA Inc., and BIM23056 from Sigma-Aldrich Japan K.K. We used DRAQ5 dye obtained from Biostatus Ltd. (Leicestershire, UK) and mouse anti-Myc-Tag polyclonal antibody from Cell Signaling Co. Ltd. (Japan). Mouse anti-V5 antibody and donkey antimouse IgG antibody tagged with AlexaFluoro 488, Hoechst 33342, and expression vectors of pcDNA3.1/myc-His (B) and pcDNA4/V5His (B) were purchased from Invitrogen Corp. (Carlsbad, CA). Construction of Mammalian Expression Vectors. The cDNAs used for the plasmid construction were generated using the standard polymerase chain reaction (PCR) with gene specific primer and Pyrobest DNA polymerase. The cDNA of Apelin receptor (AGTRL1) was amplified from a human placenta cDNA library. The cDNAs of β-arrestin (type 2), β2-adrenergic receptor (ADRB2), cholecystokinin B receptor (CCKBR), and somatostatin type 2 receptor (SSTR2) were amplified from the human brain cDNA library. The PCR fragments were subcloned into restricted enzyme sites of mammalian expression vectors, pcDNA3.1/mycHis (B) or pcDNA4/V5-His (B) (Supporting Information Figure S1). All PCR fragments were sequenced using a genetic analyzer (ABI310; Applied Biosystems). For transfection of the cells, the expression vectors were purified using a DNA purification system (Wizard Plus Minipreps; Promega Corp.). Cell Culture, Transfection, and Generation of Stable Expression Cell Lines. The human embryonic kidney (HEK293) cells were cultured in Dulbecco’s modified eagle’s medium (DMEM, high glucose) supplemented with 10% fetal bovine serum (FBS), 100 unit/mL penicillin, and 100 µg/mL streptomycin at 37 °C in an incubator with 5% CO2. Transfection of plasmids into the cells was performed with a reagent (TransIT Transfection; TransIT-LT1; Mirus Co., TX). For the establishment of the stable expression cell line, the transfected cells were screened with 0.8 mg/mL Geneticin (G418) or both 0.8 mg/mL Geneticin (G418) and 0.04 mg/mL Zeocin in the culture medium. Bioluminescence Assays of FKBP-FRB and GPCR-βArrestin Interactions. In the transient transfection assays, HEK293 cells transfected with FKBP fused to N-terminal ELuc and FRB fused to C-terminal ELuc were incubated in 96-well microtiter plates and 1.0 µM rapamycin was added to the culture medium 1 day before bioluminescence assays. The luciferase activity was measured using a 96-well microplate reader (TriStar LB941; Berthold Technologies GmbH and Co. KG, Germany) with 100 µL/well of the Emerald Luciferase assay reagent as a substrate. The time for measuring each luciferase activity was set at 2 s/well. For the HEK293 cells that stably expressed GPCR and β-arrestin fused to split ELuc fragments, the cells were cultured with a phenol red free DMEM supplemented with 1% FBS on 96well microtiter plates and incubated at 37 °C in 5% CO2 overnight. The cells were stimulated with different concentrations of ligands incubated for a given time at 37 °C in 5% CO2. The ELuc activities were measured using 100 µL/well of the Emerald Luciferase assay reagent. All measurements were performed five or six times with different wells of culture plates, 2554

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

after which the values were averaged and presented along with standard deviations. Immunostaining. Stable expression cell lines were cultured on a cover glass coated with poly L-lysine at 37 °C in 5% CO2 for 24 h. The cells were stained with Hoechst 33342 (1.0 µg/mL) for 30 min and washed with phosphate buffer saline (PBS). Cells were fixed with 3.7% formaldehyde in the culture dish at 37 °C for 15 min and washed with PBS. The cells were permeabilized with 0.2% TritonX-100 in PBS for 5 min, followed by washing three times with PBS. The cells were blocked by 0.2% fish skin gelatin (FSG) in PBS for 1 h at room temperature. For the staining of GPCRs, the buffer was exchanged to PBS (0.2% FSG) containing 1/500 dilution of mouse Anti-V5 antibody and incubated for 1 h at room temperature, shaking gently. After washing the cells with PBS, the cells were filled with PBS (0.2% FSG) containing 1/2000 dilution of a donkey antimouse IgG labeled with AlexaFluoro 488 and incubated for 1 h at room temperature. For β-arrestin staining, 1/500 dilution of mouse Myc antibody and AlexaFluoro 488 donkey antimouse IgG were used, respectively, as first and second antibodies. Cells were washed with PBS (0.2% FSG) and fixed on a cover glass with Mowiol. The cells labeled with fluorescent molecules were observed under a confocal fluorescence microscope (FV-1000; Olympus Corp.). Live Cell Imaging. The stable cell lines originated from HEK293 cells were plated on 35 mm culture dishes and incubated in DMEM with 1% FBS for 20-24 h. Before taking bioluminescence images, the nuclei of the cells were stained with DRAQ5 in DMEM with 1% FBS for 15 min. Subsequently, 10 mM D-luciferin was added to the culture dish and incubated at 37 °C for 30 min. Bioluminescence and fluorescence images of live cells were taken with an upright fluorescence and bioluminescence microscope (BX61; Olympus Corp.), which was used with a 20× dipping objective (0.40 NA) and a 60 W metal-halide light source for DRAQ5 and a filter set: a band-pass filter (560 ± 20 nm) for excitation and a long-pass filter (>610 nm) for emission. For bioluminescence imaging, we used only an emission band-pass filter (536 ± 10 nm). Emitted light from the sample was passed through a lens attachment (U-TV0.25XC; Olympus Corp.) in front of a camera. Digital images were acquired using a cooled (set at -80 °C) electron multiplying charge-coupled device (EM-CCD) camera (ImagEM; Hamamatsu Photonics K.K.). The filters and camera control were adjusted automatically using software (Meta Morph; Universal Imaging Corp., Downingtown, PA). Stray light was cut off by turning off the electric system and covering it tightly with foil. Bioluminescence images were acquired every 5 min, in which interval bright field and fluorescence images were taken, respectively, for 2 min and 300 ms. Obtained images were analyzed using imaging software (Meta Morph; Universal Imaging Corp.). RESULTS AND DISCUSSION Small Library Screening for Identifying a Novel Pair of Luciferase Fragments. In our previous study, we chose a pair of fragments of luciferase from click beetle (ELuc) which has the capability to complement upon protein-protein interactions in living cells.21 Because the 3D structure of ELuc has not been clarified to date, the pair of fragments was found based on a comparison between the amino acids sequences of firefly luciferase and those of ELuc. Our group and others have well examined the

Figure 1. Identification of a pair of luciferase fragments from rational library screening. (A) Schematic structures of the constructs: ELucN, N-terminal fragment of luciferase; ELucC, C-terminal fragment of luciferase. (B) Analysis of rapamycin-induced bioluminescence changes. Different combinations of the luciferase fragments were coexpressed in HEK293 cells, and the bioluminescence in the presence of rapamycin was evaluated. The asterisk denotes the maximum change in the bioluminescence obtained with a pair of N-terminal (1-415) and C-terminal (394-542) fragments of ELuc.

dissection sites of split firefly luciferase fragments16,18,21 the best complement pair was from 1 to 416 amino acids as the N-terminal fragment and from 398 to 550 as the C-terminal one. According to this information, the split ELuc fragments of 1-413 amino acids (N-terminal fragment) and 394-542 amino acids (C-terminal fragment) were designed and used for protein-protein interaction analysis. Although absolute photon counts of the complement ELuc fragments were higher than the firefly luciferase fragments when using a pair of interacting proteins, FKBP and FRB, the exposure time of the EM-CCD camera required more than 1 min to obtain an image of real-time protein-protein interactions. In addition, the sensitivity did not meet the criteria for evaluating protein-protein interactions on a 96-well microtiter plate format. To improve the sensitivity, we then created small libraries of N-terminal and C-terminal fragments of ELuc, both near the dissection sites determined in the previous study. The N-terminal fragment of 1-406 amino acids were extended one by one in their amino acids and 12 cDNAs of N-terminal fragments (the end from 406 to 417) were prepared, each of which was connected with the cDNA of FKBP (Figure 1A). Similarly, the cDNAs of 25 C-terminal fragments from 389-542 to 413-542 amino acids were also generated and connected with the cDNA of FRB. All

combinations of the cDNA constructs were coexpressed in HEK293 cells. Then their luminescence intensities were examined in the presence and absence of 1.0 µM rapamycin using a microplate reader (Figure 1B). We used the higher concentration of rapamycin than the normal (∼10 nM) because the properties of rapamycin such as hydrophobicity and chemical instability might lower the effective concentration in the cytosol. The results showed that different photon counts originated from complement luciferase activities of N-terminal and C-terminal fragments of ELuc. The maximum changes in bioluminescence were obtained from the cells that had been cotransfected with a pair of the N-terminal (1-415) and C-terminal (394-542) fragments. Recovery of the bioluminescence from the fragments was efficient, with a 70-90-fold greater signal after exposure to rapamycin. The absolute photon counts of the bioluminescence signal were 3-fold higher than the previous pair of the luciferase fragments. The background level of biluminescence was higher than that of untransfected cells. The background bioluminescence of the ELuc fragments may be originated from overlapping the fragments of amino acids, 394-415. Such an overlapping effect on the background bioluminescence has been previously reported using firefly luciferase.18 Results also showed that single amino-acid differences Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2555

of the dissection sites strongly affect the reconstitution efficiency of the luciferase fragments. For example, single amino-acid differences such as the C-terminal fragments of amino acids, 393-542 or 395-542, complement the N-terminal fragment of 1-415 amino acids, causing an insufficient recovery of bioluminescence compared to the best pair of fragments. Some combinations of ELuc fragments revealed no complementation, even in the presence of rapamycin. The combination of N-terminal ELuc (1-415) and C-terminal ELuc (394-542), with the highest signals, was selected for additional development of GPCR-β-arrestin interaction assay methods. Generation of Stable Cell Lines for GPCR-β-Arrestin Interaction Assays. The new fragments from click beetle luciferase were applied for GPCR-β-arrestin interaction assays (Figure 2A). In fact, GPCR is a family of seven-transmembrane receptors, which sense extracellular molecules and which induce intracellular signaling transductions. The β-arrestin is an adapter protein that binds to desensitize GPCRs. The interaction of β-arrestin and GPCRs thereby regulates a second wave of signaling and is an important event for evaluating the activity of GPCRs and drug screening. Of many kinds of GPCRs, a somatostatin receptor (SSTR2) was selected for trial experiments. It is wellknown that N-terminal amino acids of SSTR2 is necessary for targeting to the endoplasmic reticulum. It is also evident that C-terminal tail of β-arrestin is important for the interaction with GPCRs. Because of these facts, ligation sites of SSTR2 and β-arrestin were limited; the C-terminal end of SSTR2 was connected with the C-terminal fragments of ELuc, whereas the N-terminal end of β-arrestin was connected with the N-terminal fragment of ELuc. The fusion proteins were coexpressed in the HEK293 cells, which were cultured on 96-well microtiter plates. The bioluminescence changes upon injection of somatostatin were examined using a microplate reader. Total photon counts measured in 2 s were found to be 8.0-fold stronger than the background noise level in the presence of 1.0 × 10-6 M somatostatin (Figure 2B). No significant signal from the unstimulated cells was observed, indicating that ligand-induced activation of SSTR2 and its successive binding to β-arrestin caused complementation of the luciferase fragments without steric hindrance. Optimization of linkers used for the fusion proteins is crucial for enhancing the folding and flexibility of luciferase fragments. We examined different lengths of flexible linkers composed of (Gly-Gly-Gly-Gly-Asn)n repeat (Supporting Information Figure S2). The highest bioluminescence signal was obtained when we used a pair of linkers: 21 amino acids for the N-terminal luciferase and 7 amino acids for the C-terminal one. Hereafter, the linker lengths were used for the construction of luciferase-fused proteins. Next, we generated HEK293 cells, which stably express β-arrestin connected with N-terminal ELuc fragment in the cytosol. The cDNA was transfected in HEK293 cells and the cells constitutively expressing the fusion protein were obtained after 20 days of selection in a G418-containing growth medium. The expression was examined using transient transfection of SSTR2 fused to the C-terminal ELuc fragment. We examined bioluminescence for more than 100 clones and obtained a specific clone, of which bioluminescence increased significantly upon injection 2556

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

Figure 2. Assays based on somatostatin-induced GPCR-β-arrestin interaction. (A) Schematic diagram of the bioluminescent probe used to monitor the GPCR-β-arrestin interaction. The N-terminal and C-terminal fragments of ELuc are fused, respectively, to β-arrestin and GPCR. Binding of a ligand to GPCR triggers phosphorylation of GPCR, thereby inducing its interaction with β-arrestin. This interaction brings the N-terminal ELuc into proximity with the C-terminal ELuc, and bioluminescence activity is recovered. (B) Bioluminescence upon injection of somatostatin using HEK293 cells transiently coexpressed with ELucN-β-arrestin and SSTR2-ELucC. The transfected cells were cultured on 96-well microtiter plates and stimulated with 1 × 10-6 M somatostatin for 12 min. Photon counts were evaluated for 2 s/well using a microplate reader (n ) 6). (C) Bioluminescence upon injection of 1 × 10-6 M somatostatin using HEK293-ARRB2 cells transiently expressed with SSTR2-ELucC. Photon counts were evaluated for 2 s/well using a microplate reader (n ) 6). (D) Localization of β-arrestin in HEK293-ARRB2 cells.

of somatostatin (Figure 2C). The intracellular localization of β-arrestin was examined using immunostaining. The results showed that β-arrestin localized in the cytosol without a nucleus in the absence of somatostatin (Figure 2D), which was consistent with the localization analysis described in a previous report.30 From these results, we concluded that the HEK293 cells stably expressing the fusion protein of β-arrestin and N-terminal fragment of ELuc, named HEK293-ARRB2, are useful for constructing a higher-sensitivity SSTR2 assay system. (30) Oakley, R. H.; Laporte, S. A.; Holt, J. A.; Caron, M. G.; Barak, L. S. J. Biol. Chem. 2000, 275, 17201–17210.

Figure 3. Characterization of cell lines stably expressing ELucN-β-arrestin and SSTR2-ELucC. (A) Dose-response curves for somatostatin and its analogues based on the SSTR2-β-arrestin interactions. The cells were cultured on 96-well microtiter plates and stimulated for 12 min. The mean luminescence intensities were determined at each ligand concentration (n ) 6). (B) Time course analysis of SSTR2-β-arrestin interactions. The stably expressed cells in each well were treated with 1.0 × 10-6 M somatostatin; at the indicated time points, luciferase activities were measured in the subsets of cells to determine the time course of the interaction (n ) 6). (C) Immunocytochemical images of HEK293-ARRB2-SSTR2 cells. The cells were incubated for 12 min in the absence or presence of 1.0 × 10-7 M somatostatin. The SSTR2 and β-arrestin were recognized, respectively, by anti-V5 and anti-myc antibodies. The nuclei were stained with Hoechst 33342; superimposed images are shown. Bar: 20 µm.

Generation and Characterization of a HEK293-ARRB2SSTR2 Cell Line for GPCR Assays. Using the HEK293-ARRB2 cell line, we further constructed a cell line that expressed SSTR2 fused to the C-terminal fragment of ELuc. After transfection of the cDNA, we picked up more than 50 subcloned cells and obtained a cell line that stably expressed the SSTR2 connected with C-terminal ELuc fragment on the plasma membrane (named HEK293-ARRB2-SSTR2). To examine characteristics of the newly generated HEK293ARRB2-SSTR2 cell line, we first studied the quantitative detection of the somatostatin-induced SSTR2-β-arrestin interaction in the cells. The cells were harvested on 96-well microtiter plates and were stimulated with each concentration of somatostatin for 12 min. The bioluminescence was measured for 2 s/well; their raw data are shown in Figure 3A. The photon counts from bioluminescence increased concomitantly with increasing somatostatin concentrations from 3.0 × 10-9 to 3.0 × 10-7 M. No change in the photon counts was observed at concentrations lower than 1.0 × 10-9 M. The maximum response in the presence of somatostatin was 30-fold stronger than that in the absence of somatostatin. No remarkable bioluminescence was found in the presence of analogues, BIM23052 or BIM23056 (Figure 3A), demonstrating that this assay format is useful for selective and quantitative analyses of the extent of ligand activities based on SSTR2-β-arrestin interactions. The response time after stimulation of somatostatin is important for high-throughput screening. To examine the time-dependent changes in photon counts, we investigated recovery of the bioluminescence at different time points after stimulation of

somatostatin. Upon stimulation of 1.0 × 10-6 M somatostatin, the bioluminescence increased rapidly and reached a plateau within 15 min (Figure 3B). Thereafter, the bioluminescence caused a gradual decrease over time. The decrease in the bioluminescence might have originated from a decrease in the number of SSTR2 on the plasma membrane and/or decomposition of SSTR2 and β-arrestin inside the cells. Reportedly, after stimulation with a specific ligand, GPCR internalizes on the plasma membrane and partly moves into a proteasome pathway.30 To understand the dynamic movement of SSTR2 and β-arrestin, we examined the localization of the proteins in the presence and absence of somatostatin using the immunostaining technique. In somatostatinunstimulated cells, SSTR2 was clearly distributed on the plasma membrane, whereas β-arrestin was distributed in the cytosol (Figure 3C). In the presence of somatostatin, a portion of SSTR2 was found to be internalized in the cytosol and to have formed a dot-like structure. Similarly, β-arrestin connected with the ELuc fragment formed several aggregates in the cytosol, indicating that the complex of SSTR2-β-arrestin caused internalization into the cytosol within a few minutes. This internalization might be attributed to the gradual decrease of photon counts after stimulation of somatostatin. The exact mechanism for this remains to be clarified. Time-Lapse Imaging of SSTR2-β-arrestin Interactions in Single Living Cells. Next, we examined the capability of the ELuc fragments for imaging SSTR2-β-arrestin interactions in real time. Prior to taking bioluminescent images, HEK293-ARRB2-SSTR2 cells were cultured in a 35 mm dish for 24 h. To identify individual cells, the nuclei were stained with a nuclear staining dye, DRAQ5. Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2557

Figure 4. Real time bioluminescence images of SSTR2-β-arrestin interactions. Stable cell lines expressing ELucN-β-arrestin and SSTR2-ELucC were cultured on a dish and stained with a nucleus staining fluorescent dye, DRAQ5, for identifying individual cells. The cells were treated with 1.0 × 10-7 M somatostatin; the injection time was defined at 0 min. Bioluminescence images were acquired every 5 min, in which interval bright field and fluorescence images were taken using a EM-CCD camera. The obtained bioluminescence (green) and fluorescence (red) images were superimposed on the bright field images. Bar: 50 µm.

The fluorescence images were obtained immediately after acquisition of bioluminescence images, and individual images were overlaid with each bright field image. Upon stimulation of the cells with 1.0 × 10-7 M somatostatin, a strong and rapid increase in bioluminescence was detected only in the area of the plasma membrane of the cells (Figure 4 and Supporting Information Movie S1). Successively, the localization of the bioluminescence showed dot-like structures that were almost identical to results obtained using the immunostaining study (Figure 3C). During the observation time, no bioluminescence was detected in the nuclei of the cells. On the basis of these results, we concluded that the bioluminescence intensity of the novel ELuc fragments is sufficiently strong to be detected at the single-cell level using an optimized fluorescence and bioluminescence microscope equipped with an EM-CCD camera. Application to Different GPCRs. To show the general usefulness of the HEK293-ARRB2 stable cells, we developed several cell lines that stably expressed GPCRs of adrenergic β2 receptor (ADRB2), apelin receptor (AGTRL1), endothelin receptor type B (EDNRB), and cholecystokinin B receptor (CCKBR), each of which was connected with the C-terminal fragment of ELuc. For their characterization, each receptor was stained with AlexaFluoro 488; then, their localization was examined under a fluorescence microscope (Figure 5A). All the receptors localized on the plasma membrane, although some portion of the receptor was in the endoplasmic reticulum. Next, we examined responses of bioluminescence against each specific ligand, isoproterenol for ADRB2, (pyr1)-apelin-13 for AGTRL1, gastrin-1 for CCKBR, and endothelin-1 for EDNRB. The study of ligand-dependent bioluminescence revealed that the bioluminescence increased concomitantly with the increased concentration of the ligand; the maximum response was obtained as 15-40-fold higher than the background bioluminescence (Figure 5B). It is particularly interesting that time-dependent analysis of photon counts upon the ligand stimulation showed different response of the bioluminescence (Figure 5C). For EDNRB, the bioluminescence upon endothelin-1 stimulation increased faster than the bioluminescence for the SSTR2. The maximum response reached at 5-10 min after stimulation of the ligands. Decreases in the bioluminescence after the maximum response were also rapid, and photon counts became less than half of the maximum response at 60 min. In contrast, responses of the bioluminescence for AGTRL1 and CCKBR were much slower than the response for EDNRB. Maximum responses were 2558

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

obtained at 20 min after stimulation of the corresponding ligands; thereafter, the bioluminescence decreased gradually. Although the maximal response was reached at 20 min, large changes in bioluminescence were obtained within 10 min. Furthermore, we examined bioluminescence changes at a single cell level (Supporting Information Movies S2-S4). Of the four cell lines examined, cell lines, which express ADRB2, AGTRL1, and CCKBR, showed remarkable changes in bioluminescence under the microscope. Before adding the ligands, spontaneous bioluminescence from self-complementation of ELuc fragments was not detected. Upon stimulation, stable cell lines showed large increases in bioluminescence. Their localization of the bioluminescence was observed only in the cytosol. In addition, all the cell lines showed remarkable morphological changes in the bright field images after appearance of bioluminescence. The morphological changes may be caused by dynamic movement of the cellular cytoskeleton which was induced by the GPCR ligands or undesirable buffer conditions. These results demonstrate that the HEK293-ARRB2 cells are generally used for many kinds of GPCRs and that the stable cell lines thus generated have unique properties of rapid response of bioluminescence and a high signal-tobackground ratio. The rapid detection of GPCR-β-arrestin interactions in this study, particularly with low background signals, is a strong advantage for extending this method into a high-throughput assay format. For rapid GPCR assays, many methods have been developed to date: one existing method is fluorescent proteinfragment complementation to assess GPCR-β-arrestin association.31 This approach measured the redistribution of the GPCR-βarrestin complex in cytosol using either a plate reader or confocal microscopy. The GPCR assay allowed assessment of interactions occurring at a cellular or subcellular level. Although such advantages exist, at least 30-60 min for chromophore formation is necessary, which hampers rapid screening of GPCR assays. A related technique is β-galactosidase fragment complementation for the study of GPCR-β-arrestin interactions.32 This assay format supported a homogeneous assay conducted in a microtiter plate; GPCRs activation generated a luminescence signal. Because the luminescence is originated from two-step enzymatic reactions of (31) Yu, H.; West, M.; Keon, B. H.; Bilter, G. K.; Owens, S.; Lamerdin, J.; Westwick, J. K. Assay Drug Dev. Technol. 2003, 1, 811–822. (32) von Degenfeld, G.; Wehrman, T. S.; Hammer, M. M.; Blau, H. M. Faseb J. 2007, 21, 3819–3826.

Figure 5. Characterization of cell lines stably expressing GPCRs and β-arrestin. (A) Localization analysis of β2-adrenergic receptor (ADRB2), apelin receptor (AGTRL1), cholecystokinin B receptor (CCKBR), and endothelin receptor type B (EDNRB) by the immunostaining. The GPCRs were recognized by anti-V5 antibody and visualized using AlexaFluoro 488 (green). The nuclei were stained with Hoechst 33342 (blue). Superimposed images are shown. Bar: 20 µm. (B) Dose-response curves for each ligand. The cells were cultured on 96-well microtiter plates and stimulated for 10 min. The mean luminescence intensities were determined at each ligand concentration (n ) 6). (C) Time course analysis of GPCR-β-arrestin interactions. The stably expressed cells in each well were treated with 1.0 × 10-7 M isoproterenol for ADRB2, 1.0 × 10-6 M (pyr1)-apelin-13 for AGTRL1, 1.0 × 10-8 M gastrin-1 for CCKBR, and 1.0 × 10-7 M endothelin-1 for EDNRB. Then, at the indicated time points, luciferase activities were measured in the subsets of cells to determine the time course of the interaction (n ) 6).

β-galactosidase followed by luciferase, more than 30 min were required to obtain 20-fold induction of the luminescence. A more widely used technology is bioluminescence resonance energy transfer (BRET) assay.33 The assay time to obtain a meaningful bioluminescence is, in principle, identical to the present method. However, the use of BRET as a means to characterize GPCR oligomerization and β-arrestin interactions is controversial. Because of spectral overlaps between bioluminescence and fluorescence, rigorous data analyses are needed to distinguish nonspecific interaction from true oligomeric protein interactions. Luker et al. have recently studied CXCR4-β-arrestin interaction by using firefly luiferase complementation system.34 Bioluminescence was quantified 5-10 min after addition of ligands. The time required for the assay on a microplate format is almost comparable to the present assay system. However, the increase in bioluminescence was only 7-fold upon ligand stimulation. In comparison to such existing methods, the present method presents strong advantages in terms of the detection periodsless than 10 minsa high signal(33) Molinari, P.; Casella, I.; Costa, T. Biochem. J. 2008, 409, 251–261. (34) Luker, K. E.; Gupta, M.; Luker, G. D. Anal. Chem. 2008, 80, 5565–5573.

to-background ratio of the bioluminescence, and high sensitivity of absolute photon counts. Moreover, it is a first demonstration that the GPCR-β-arrestin interactions can be visualized directly in real time at a single cell level. This spatiotemporal imaging of the GPCR-β-arrestin complex is also a strong advantage over fluorescent protein-fragment complementation, in which no temporal information is provided. CONCLUSIONS We identified a new pair of split luciferase fragments from click beetles using a semirational combinatorial screening. We demonstrated that novel luciferase fragments provided a significant improvement of the signal-to-background ratio and absolute photon counts. We applied the luciferase fragments to the development of GPCR assays in cultured mammalian cells. Generating the cell lines that stably expressed the luciferase fragment-fused GPCRs and β-arrestin, we demonstrated the applicability for time-course and quantitative assessments of the effects of ligands on the receptors. These stable cell lines are immediately useful for assay and screening of chemicals without Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

2559

the need for cumbersome handling of gene transfer or for control of protein expression. Bioluminescence imaging in live cells further demonstrated the utility of the new luciferase fragments for spatiotemporal studies of the interactions on the cell membrane. Unique properties of novel luciferase fragments, such as faster response than the previous method, are applicable to the different assay formats to screen various chemical compounds for drug discovery in accordance with the intended use. Consequently, the luciferase fragments might greatly accelerate to develop cell-based assays of high-throughput screening of chemicals that act on a specific protein-protein interaction in living cells.

Corporation (JST), New Energy Industrial Technology Development Organization (NEDO), and in part by the Global COE Program and grants (S0801035), MEXT, Japan. T.O. was supported in part by the Kato Memorial Foundation.

ACKNOWLEDGMENT This work was supported by the Japan Society for the Promotion of Science (JSPS), Japan Science and Technology

Received for review January 14, 2010. Accepted February 11, 2010.

2560

Analytical Chemistry, Vol. 82, No. 6, March 15, 2010

SUPPORTING INFORMATION AVAILABLE Figure S1 and S2 and Movies S1-S4. This material is available free of charge via the Internet at http://pubs.acs.org.

AC100104Q