A Mammalian Genetic System To Screen for Small Molecules Capable

Apr 13, 2004 - fused with the inhibitory KRAB domain as a suppressor. The binding of the suppressor to the tet operator entirely blocked expression of...
1 downloads 0 Views 247KB Size
Anal. Chem. 2004, 76, 2922-2927

A Mammalian Genetic System To Screen for Small Molecules Capable of Disrupting Protein-Protein Interactions Hui-Fen Zhao,† Taira Kiyota,† Shafinaz Chowdhury,† Enrico Purisima,† Denis Banville,† Yasuo Konishi,† and Shi-Hsiang Shen*,†,‡

Health Sector, Biotechnology Research Institute, National Research Council of Canada, 6100 Royalmount Avenue, Montreal, Quebec, Canada H4P 2R2, and Department of Medicine, McGill University, Montreal, Quebec, Canada, H3G 1A4

A mammalian two-hybrid system was developed for highthroughput screening of compounds that disrupt specific protein-protein interactions. The existing mammalian systems are unsatisfactory for drug screening due to nonregulated expression of interacting proteins. To construct a tightly regulated system, the tetracycline repressor was fused with the inhibitory KRAB domain as a suppressor. The binding of the suppressor to the tet operator entirely blocked expression of two interacting proteins. When both the inducer doxycycline and drugs were added to the culture, the reporter gene was either activated by interaction of the paired proteins with ineffective drugs or remained silent due to disruption of the protein interactions by the effective drugs. We demonstrate that interactions of the type I receptor for TGFβ with FKBP12 and the epidermal growth factor receptor (EGFR) with p85 are effectively disrupted by FK506 and EGFR kinase inhibitor AG1478, respectively. The power of this system for drug screening was further demonstrated by rapid identification of inhibitors from a druglike library for the receptor kinases. Protein-protein interactions are involved in most biological processes and cellular functions. Elucidation of protein interactions in the cell constitutes an important approach to functional genomics. The most powerful technology for the detection of protein-protein interaction is the yeast two-hybrid system that was first described by Fields and Song.1 Protein-protein interactions can be inhibited by trans-acting molecules such as dissociating peptides or small molecules that may serve as therapeutic agents. At present, the selection of dissociation molecules for particular protein-protein interactions is executed using yeast reverse two-hybrid systems2-4 or a bacterial genetic system.5,6 Although the yeast and bacterial systems are useful for some genetic selections, they are not ideal for the * Corresponding author: (e-mail) [email protected]; (fax) 514-496-6319. † National Research Council of Canada. ‡ McGill University. (1) Fields, S.; Song, O. Nature 1989, 340, 245-246. (2) Leanna, C. A.; Hannink, M. Nucleic Acids Res. 1996, 24, 3341-3347. (3) Vidal, M.; Brachmann, R. K.; Fattaey, A.; Harlow, E.; Boeke, J. D. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 10315-10320. (4) Shih, H. M.; Goldman, P. S.; DeMaggio, A. J.; Hollenberg, S. M.; Goodman, R. H.; Hoekstra, M. F. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13896-13901. (5) Zhang, Z.; Zhu, W.; Kodadek, T. Nat. Biotechnol. 2000, 18, 71-74. (6) Park, S. H.; Raines, R. T. Nat. Biotechnol. 2000, 18, 847-851.

2922 Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

screening of therapeutic small molecules as drug permeability in yeast or in bacteria is very different from that of mammalian cells. In addition, various posttranslational modifications required for particular protein interactions are limited in both yeast and bacteria. In fact, the potential of drug candidates needs to be primarily evaluated in mammalian cells. Presently there are only a few reports of mammalian two-hybrid systems for protein interactions.7-9 However, these existing mammalian systems are not suitable for high-throughput reverse two-hybrid screens of small molecules due to the fact that the two interacting proteins are constitutively expressed in the reported mammalian systems. Such unregulated constitutive expression of the two interacting proteins in the existing systems can immediately activate the reporter gene, and the inhibitory effect of a drug thus cannot be evaluated by the repressive extent of the reporter gene as it has already been fully expressed before addition of drugs in the systems. To perform high-throughput screening (HTS) of potential therapeutic molecules that disrupt protein-protein interactions in mammalian cells, it is essential that small molecules be introduced into the system before the synthesis of the two interacting proteins is induced in cells for interaction. Theoretically, the constructs for expression of the two interacting proteins may be transiently transfected with drugs together into the cells in each well for screening. However, performing such transient transfection of the two constructs into individual wells is unpractical for high-throughput drug screens. Hence, we have developed a mammalian genetic system that is particularly useful for HTS of small molecules capable of preventing protein-protein interaction. The two interacting proteins are fused either with a DNA binding domain (BD) or with a transcriptional activation domain (AD). The expression of the genes encoding these two interacting proteins is tightly regulated before the addition of drugs. Three reporters, namely, the green fluorescent protein (GFP), β-lactamase (β-Lac), and luciferase (Luc), were employed to complement each other for detection of different potent drugs in disruption of the interacting proteins. EXPERIMENTAL PROCEDURES Construction of Expression Vectors. The reporter plasmid of pG5CMV-GFP-IRES-β-Lac/Luc (Figure 1D) was constructed (7) Luo, Y.; Huang, B.; Leandro, S.; Pyan, D. U.S. Patent 6, 114, 111, 2000. (8) Shioda, T.; Andriole, S.; Yahata, T.; Isselbacher, K. J. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 5220-5224. (9) Eyckerman, S.; Verhee, A.; Van der Heyden, J.; Lemmens, I.; Van Ostade, X.; Vandekerckhove, J.; Tavernier, J. Nat. Cell Biol. 2001, 3, 114-120. 10.1021/ac035396m CCC: $27.50 Published 2004 Am. Chem. Soc. Published on Web 04/13/2004

Figure 1. Schematic representation of the mammalian reverse twohybrid system for selection of small molecules capable of disrupting a protein-protein interaction. The plasmids employed in this study are schematically presented. X and Y, represent specific interacting proteins; TATA, the mini CMV promoter sequence; IRES, internal ribosome entry site; and Gal4 BE, the Gal4 binding elements.

using pG5CAT (Clontech) as backbone. The synthetic IRES sequence,10 the human cytomegalovirus (CMV) minipromoter, and the reporter genes of GFP and β-lactamase or luciferase were inserted into pG5CAT by PCR-based cloning. For construction of pCMV-TetR-KRAB (Figure 1A), the TetR sequence from pUHD15.111 was ligated with the synthetic nuclear localization sequence (NLS) of seven amino acids12 in frame and inserted into pCDNA3.1 (Invitrogen). Then the KRAB domain of human Kox1 gene13 was joined with TetR-NLS in frame in the plasmid. Finally, the whole expression cassette of pCMV-TetRKRAB sequence with the selective marker was isolated and inserted into pG5CMV-GFP-IRES-β-Lac/Luc to generate pG5CMVGFP-IRES-β-Lac/Luc-TetR-KRAB for stable transfection into the cells. For construction of pCMV-tetOP-AD‚X-IRES-BD‚Y (Figure 1B), the CMV core promoter sequence (pCDNA3.1, nucleotides 230-780) was inserted into pUHD 10.311 before the 7X Tet operator sequence. The interacting protein sequence “X” (RI/p85) and “Y” (FKBP/EGFR) were joined with Vp16 transcriptional AD (Clontech) and Gal4 DNA BD (Clontech) in frame, respectively. These interacting protein sequences, together with the IRES sequence, were cloned into the modified pUHD 10.3 by a PCRbased method. Finally, the hygromycin gene was also cloned into the above plasmid for stable transfection selection. Cell Culture and Stable Cell Lines. To establish a stable cell harboring both TetR-KRAB (Figure 1A) and the reporter genes (Figure 1D) from pG5CMV-GFP-IRES-β-Lac/Luc-TetRKRAB (not shown in Figure 1), the human embryonic kidney (10) Chappell, S. A.; Edelman, G. M.; Mauro, V. P. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 1536-1541. (11) Gossen, M.; Bujard, H. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 5547-5551. (12) Kalderon, D.; Roberts, B. L.; Richardson, W. D.; Smith, A. E. Cell 1984, 39, 499-509. (13) Deuschle, U.; Meyer, W. K.; Thiesen, H. J. Mol. Cell Biol. 1995, 15, 19071914.

epithelial cell line 293A cells (ATCC) were plated in a dish (60 mm) a day prior transfection. The first stable transfection was carried out at 80% cell confluence with 5 µg of the linearized plasmid (pG5CMV-GFP-IRES-β-Lac/Luc-TetR-KRAB) mixed with 20 µL of Superfect (Qiagen). At 48 h posttransfection, cells were trypsinized and diluted 1:20 in the culture medium containing 800 µg/mL G418 for selection. Positive clones (20 out of 51) were first identified by their ability to activate the reporter gene in the presence of Dox following transient cotransfection with pCMVtetOP-AD-X-IRES-BD-Y for protein interaction. Two clones, which were highly induced by doxycycline (Dox), were further stably transfected with plasmid pCMV-tet OP-AD-p85-IRES-BD-EGFR and selected with both G418 (400 µg/mL) and hygromycin (150µg/ mL). Among 50 stable clones examined, 22 clones were found to be Dox-inducible with varied luciferase activities. Clone D-3, which displayed the highest induction of luciferase activity, was chosen for HTS. Among others, this stable cell line has been stably responsive to Dox for many generations. For transient expression of interacting proteins, several million cells stably transfected with pG5CMV-GFP-IRES-β-Lac/Luc-TetRKRAB were transiently tansfected with pCMV-tet OP-AD-X-IRESBD-Y overnight and then divided into 96-/384-well plates containing both Dox and drugs. After an additional 24-36-h incubation, the expression level of the reporters was measured. Compound Screening. Cells stably transfected with constructs for expression of the interacting proteins and the reporter (GFP) were cultured in 100/50 µL DMEM medium. Druglike compounds dissolved in DMSO were dispensed into the cells in 96-/384-well plates at a final concentration of 10 µM (0.5% DMSO) and induced with Dox (1 µg/mL) for 36 h. Plates were read. To eliminate false-negative results caused by compound autofluorescence, compounds in plates were also read without induction with Dox as a mock screen. After comparing the results in the mock screen, compounds reporting fluorescence values significantly below the mean generated from all compounds were selected for retesting in the CMV-promoter directed GFP expression system for measuring compound toxicity to cells. Any potential inhibitory compounds evaluated from the CMV-GFP system were secondly screened with cells stably harboring the Luc reporter using the same assay conditions. Detection of GFP Reporter Expression in Cells. To directly assay the expression of GFP, cells were cultured in 96-/384-well plates (low-fluorescence assay plates) with 100/50 µL of DMEM medium. Dox was added at a final concentration of 1 µg/mL. Following 36-h induction with Dox, the medium in each well was removed robotically. The intensity of GFP was directly measured at 485-nm excitation and 525-nm emission with a fluorescence multiwell plate reader (CytoFluor II, PerSeptive Biosystems, or Wallac Victor V Stacker, Perkin-Elmer). Luciferase Activity Assay. To assay Luc activity, cells in 96-/ 384-well plates were grown in 100/50 µL of DMEM medium and induced with Dox at a final concentration of 1 µg/ml for 24 h for protein expression. After induction, media in the plates were poured out and cells were lysed in 20 µL of lysis buffer (Promega) for 20 min. Luc activity was determined using the Dual-Luciferase Reporter Assay System (Promega) according to manufacturer’s instructions. Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

2923

Immunoblotting. A431 cells were plated 1 day prior to experiments. After serum starvation for 12 h, the cells were treated for 2 h with the indicated concentrations of BR20T2-B1, followed by the stimulation of EGF (100 ng/mL) for 5 min. The treated cells were lysed with the protein sample buffer (10% glycerol, 0.2 M Tris, pH 6.8, 5% β-mercaptoethanol, 3% SDS) and boiled for 5 min. The cell lysates were subjected to 10-12% SDS-PAGE and transferred to nitrocellulose membrane. One membrane was immunoreacted with anti-Akt-phosphoserine-S473 antibody and then reprobed with anti-AKT antibody after stripping the membrane. Another membrane was blotted with anti-phosphotyrosine antibodies and reprobed with anti-EGFR antibody after stripping the blot. Growth in Soft Agar. Twelve-well tissue culture plates were covered with 1 mL of 0.5% agarose growth media (DMEM plus 10% FBS) in the absence or presence of different concentrations of kinase inhibitors. Cells (2 × 104) were plated in duplicate in a 1-mL suspension of 0.35% low-melting agarose media as the solid underlay. After 2 h at 4 °C, DMEM media containing 10% FBS were added to each well and refreshed twice weekly. For kinase inhibitor treatment of cells, the same concentrations of the drugs were added to the solid underlay, overlay agarose media, and the liquid top layer of each individual well. Cells were incubated at 37 °C and photographed after 10 days. RESULTS AND DISCUSSION Principle. To develop a mammalian genetic system for drug screening, we used the tetracycline repressor (TetR) to inhibit expression of interacting proteins when TetR binds to the tet operator (OP) located in the promoter. However, unlike in Escherichia coli, we found that in mammalian systems TetR itself was not able to completely block the transcription of the gene in the system. To enhance the inhibitory capability, we fused the TetR with the KRAB domain of human Kox1 (Figure 1A), which can silence promoter activity.13 When the TetR-KRAB bound to the OP in the absence of tetracycline, it indeed entirely suppressed expression of the interacting proteins X and Y, which are respectively fused with BD and AD and linked by an internal ribosome entry site (IRES)10 (Figure 1B). Dox, a derivative of tetracycline, abolished the dimerized binding of the TetR-KRAB to the OP and induced the expression of interacting proteins (Figure 1C) to activate the reporter genes (Figure 1D). When small molecules and Dox are added together, any potential drug can be selected as the disruption of the interaction between proteins X and Y keeps the reporter gene silenced (Figure 1E). Both the reporter gene and the TetR-KRAB suppressor construct were integrated into chromosomes of the cells. The construct coding for the two interacting proteins is preferentially stably transfected with TetRKRAB and the reporter plasmids together in cells. However, since the TetR-KRAB tightly regulates expression of the interacting proteins, the construct for the interacting proteins can be transiently transfected into a batch of several million cells first, and the transfected cells then can be divided into multiple-well plates containing both drugs and inducer Dox for performing HTS. Disruption of TGFR (RI)-FKBP12 Interaction by FK506 in the Mammalian Reverse Two-Hybrid System. One of the well-characterized protein-protein interactions identified by the yeast two-hybrid screen is the interaction between the cytoplasmic domain (RI) of the transforming growth factor-β (TGFβ) type I 2924 Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

Figure 2. Disruption of specific protein-protein interactions in 293A cells by the corresponding drugs. (A) Cells stably expressing the suppressor TetR-KRAB, the GFP reporter gene, AD-RI-BD-FKBP12 were treated with 1 µg/mL Dox, 100 nM FK506 and 2.5 µM AG1478 for 48 h, respectively, and expression of the GFP reporter gene was evaluated following the treatment. (B) Cells stably expressing TetRKRAB, the GFP reporter gene, AD-p85-BD-EGFR were treated as in (A) for 48 h, and expression of the GFP reporter gene was evaluated in response to the drug treatments. The data in both (A) and (B) represent the average and standard deviation from three independent experiments.

receptor and the immunophilin FKBP12.14 FKBP12 is a target of the macrolide FK506 that competes with the TGFβ receptor type I for binding to FKBP12 and hence interferes with the interaction between the receptor RI and FKBP12.14 We tested our mammalian system with this protein-protein interaction pair to evaluate whether the effect of FK506 on disrupting the interaction can be detected. In this regard, the cytoplasmic domain (RI) of the receptor type I was fused with the AD of VP16, and FKBP12 was fused with the BD of Gal4. A stable cell line harboring the constructs for the TetR-KRAB and the GFP reporter gene, and the construct of pCMV-tetOP-AD-RI-IRES-BD-FKBP12, was selected by both G418 and hygromycin. The intensity of GFP expression was measured following 36-h treatment of the cells with Dox and FK506. Figure 2A shows that in the absence of Dox expression of GFP in control (DMSO) cells was virtually undetectable. Addition of Dox induced expression of the interacting proteins and dramatically activated expression of the reporter gene. However, when the drug FK506 was added together with Dox, expression of GFP was almost completely abolished. The kinase inhibitor AG1478,15 an unrelated compound, exerted no significant effect on the Dox-induced GFP expression. Similar (14) Wang, T.; Donahoe, P. K.; Zervos, A. S. Science 1994, 265, 674-676. (15) Levitzki, A.; Gazit, A. Science 1995, 267, 1782-1788.

Figure 3. Disruption of protein-protein interactions in 293A cells visualized by the expression of GFP. Cells stably transfected with vector TetR-KRAB and the GFP reporter gene were further transiently transfected with constructs either for expression of FKBP12 and the receptor RI (a-d) or for expression of the EGFR and p85 (e-h). After 12 h following the transfection, the inducer Dox and the drugs were added to the culture. Photomicrographs of the cells showing expression of GFP were taken 36 h after the treatment of the cells with the drugs. The concentration of AG1478 used was 2.5 µM and that of FK506 was 100 nM.

results were obtained when pCMV-tetOP-AD-RI-IRES-BD-FKBP12 was transiently expressed in a stable cell line harboring the construct of the suppressor TetR-KRAB and the GFP reporter gene (Figure 3a-d). These data demonstrate that, in our system, FK506 effectively disrupted the interaction of FKBP12 with the receptor RI and that the disruption of the interacting proteins by FK506 was readily detected by the GFP reporter in live cells. The stable cell line expressing this pair of interacting proteins could be used to screen for additional candidate immunosuppressant drugs. Disruption of EGFR and p85 Interaction by the Kinase Inhibitor AG1478. The system was further examined with another pair of interacting proteins, the epidermal growth factor receptor (EGFR) and the p85 subunit of phosphatidylinositol 3-kinase. It is known that the cytoplasmic domain of the EGFR family can be autophosphorylated on tyrosine residues when overexpressed in the nucleus.16 p85 is one of the signaling molecules that bind to the tyrosine-phosphorylated residues on EGFR.17 AG1478 is a potent inhibitor for the EGFR kinase.15 We reasoned that the inhibition of the EGFR kinase activity with AG1478 abolishes tyrosine autophosphorylation of the receptor, resulting in elimination of the interaction of p85 with the EGFR. We therefore examined whether the inhibitory activity of AG1478 could be detected by this mammalian system. As before, a stable cell line harboring the construct of the TetR-KRAB and the GFP reporter gene, and plasmid pCMV-tetOP-AD-p85-IRES-BD-EGFR, was selected with the appropriate antibiotics. The intensity of GFP expression was monitored following 48-h treatment of the cells with Dox and the drugs. Figure 2B shows that expression of GFP (16) Xie, Y.; Hung, M. C. Biochem. Biophys. Res. Commun. 1994, 203, 15891598. (17) Hu, P.; Margolis, B.; Skolnik, E. Y.; Lammers, R.; Ullrich, A.; Schlessinger, J. Mol. Cell Biol. 1992, 12, 981-990.

was insignificant in control (DMSO) cells. Dox induced a highlevel expression of GFP that was not affected by FK506. However, addition of AG1478 to the culture almost completely antagonized the Dox-induced activation of the GFP reporter gene, indicating that the inhibitor abolished the interaction between the EGFR and p85. We also performed a transient transfection of pCMV-tetOPAD-p85-IRES-BD-EGFR into a stable cell line only harboring the construct of TetR-KRAB and the GFP reporter gene and obtained similar results (Figure 3e-h). These results suggest that this mammalian system can be used not only to screen for molecules directly interfering with specific protein interactions but also to indirectly screen for enzyme inhibitors as long as the enzymatic activities affect the interaction of the enzyme with its partner proteins. Thus, this mammalian system provides a convenient and powerful tool for cell-based screening of candidate inhibitors for a variety of enzymes, such as proteases, phosphatases, and kinases, particularly disease-related kinases such as EGFR, PDGFR, Her2, ZAP70, Src-like kinases, and others. Identification of BR20T2-B1 as a Molecule Interrupting the EGFR-p85 Interaction. To demonstrate the power and usefulness of this mammalian system for HTS, a library consisting of ∼20 000 drug-like/drug-related compounds was first screened for potential molecules capable of disrupting the EGFR-p85 interaction in 293A cells stably transfected with the constructs for expressions of the interacting proteins and the GFP reporter (data not shown). The false-negative disruptors generated by compound autofluorescence were corrected by comparing the compounds in a mock screen, and false-positive disruptors caused by compound toxicity to cells were eliminated by comparing quenching compounds in a CMV promoter-directed GFP expression screen. Compounds that exhibited signals significantly lower than the mean were secondly screened in cells stably transfected with the Luc reporter (data not shown). Among others, BR20T2-B1 was confirmed to dissociate the interaction in the Luc-reporter screen. Figure 4A shows that BR20T2-B1 disrupted the EGFR-p85 interaction with an IC50 value of ∼5 µM. BR20T2-B1 did not exert a significant effect on the interaction of PDGFR with p85 even at a concentration of 10 µM (data not shown). The same concentration of BR20T2-B1 also did not inhibit the growth of 293 cells in the experimental conditions (data not shown). Interestingly, AG1478, although as a potent inhibitor for the EGFR kinase possessing an IC50 of as low as 3 nM in vitro assay,15 exerted the disruption of the EGFR-p85 interaction with an IC50 value as high as 0.5 µM in this mammalian system (Figure 4B). Such a high IC50 value for AG1478 in vivo may be caused by its permeability or its stability in the cells. There are two possible mechanisms by which BR20T2-B1 disrupts the EGFR-p85 interaction. One mechanism is that BR20T2B1, as with AG1478, could act as an inhibitor of the EGFR kinase and block tyrosine phosphorylation of the receptor, thus abolishing its interaction with p85. The other mechanism may involve its direct interference with the interaction between the two proteins. To examine whether BR20T2-B1 can act as an inhibitor of the EGFR kinase, A431 cells, which overexpress the EGFR, were grown in the presence of BR20T2-B1 at various concentrations and stimulated with EGF for activation of the receptor. Figure 4C shows that BR20T2-B1 displayed an inhibitory effect on the autophosphorylation of both the EGFR and Akt-S473 in a doseAnalytical Chemistry, Vol. 76, No. 10, May 15, 2004

2925

Figure 4. Effects of BR20T2-B1 on the EGFR-mediated signaling pathway. (A) and (B) BR20T2-B1 and AG1478 disrupt the EGFR-p85 interaction. Cells stably expressing the suppressor TetR-KRAB, the Luc reporter gene, and AD-p85-BD-EGFR were treated with 1 µg/mL Dox, and BR20T2-B1/AG1478 at the indicated concentrations for 24 h, respectively. Expression of the Luc reporter gene was evaluated in response to the drug treatments. The data represent the average and standard deviation from three independent experiments. The inset in (A) is the structure of BR20T2-B1. (C) Inhibition of the EGFR-activated autophosphorylation by BR20T2-B1. A431 cells were grown on six-well plates for 24 h. After serum starvation for 12 h, the cells were incubated with BR20T2-B1 for 2 h at the indicated concentrations. The cells then were stimulated with 100 ng/mL EGF for 5 min. The cells were lysed and subjected to Western blotting with anti-phosphotyrosine antibodies or with anti-Akt-phosphoserine-S473 antibody. After stripping, the blots were reprobed with anti-EGFR and anti-AKT antibodies, respectively. (D) Model of the binding mode of BR20T2-B1 in the ATP binding cavity of EGFR. The coordinates of EGFR were taken from the crystal structure of the tyrosine kinase domain of the receptor.18 Schematic ribbon diagram of the overall fold of EGFR around the ATP binding cavity is shown. Helices are presented in yellow, β strands in orange, loops in cyan, and the inhibitor BR20T2-B1 in capped stick model. Hydrogen bonds between BR20T2-B1 and the backbone of the Met 767 in capped stick are shown with a purple solid line.

dependent manner with IC50 of approximately 1.0 and 0.5 µM, respectively. The in vitro kinase assay using recombinant EGFR confirmed its inhibitory activity with an IC50 value of ∼0.5 µM for the kinase (data not shown). These results suggest that BR20T2-B1 acts as an inhibitor of the EGFR kinase to block tyrosine phosphorylation of the kinase, thus disrupting the interaction between EGFR and p85, rather than interfering directly with the proteinprotein interaction between EGFR and p85. Based on the structure of the tyrosine kinase domain of EGFR,18 BR20T2-B1 was modeled into the ATP binding cavity of 2926 Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

EGFR. Figure 4D shows that BR20T2-B1 can occupy the ATP binding cavity of the tyrosine kinase as expected on the basis of the assumed competitive behavior of the inhibitor with ATP. The function of BR20T2-B1 in the EGFR-activated signaling pathway was further explored by examining its inhibitory effect on the colony formation of A431 cells in soft agar. A431 cells expressing EGFR were treated with various concentrations of BR20T2-B1 and AG1478. As shown in Figure 5, both BR20T2-B1 (18) Stamost, J.; Sliwkowski, M. X.; Eigenbrot, C. J. Biol. Chem. 2002, 277, 46265-46272.

In conclusion, we have demonstrated the efficacy of this mammalian genetic system for dissociation of two well-characterized interaction pairs with the defined compounds either interfering directly with the interaction (R1/FKBP12) or acting as a kinase inhibitor and indirectly blocking the interaction (EGFR/p85). The power of the system was further demonstrated by rapid identification of BR20T2-B1 as a dissociative molecule for the EGFR-p85 interaction. The system is currently utilized in HTS and several potent inhibitors for the receptor tyrosine kinases have been identified through screening of combinatorial libraries.

Figure 5. Inhibition of A431 colony formation in soft agar by BR20T2-B1 and AG1478. Twenty thousand cells per well were seeded on a six-well plate in soft agar. BR20T2-B1 and EGFR inhibitor AG1478 at the indicated concentrations were applied in the growth medium from the top. After 10 days, the colonies were photographed (magnification 100×).

and AG1478 inhibited the colony formation of A431 cells in soft agar in a dose-dependent manner.

ACKNOWLEDGMENT We thank Robin Green and Mark Slater for comments on the manuscript, Dr. Zhenbao Yu for valuable discussion and advice, Patrice Bouchard for HTS, and Denis L’Abbe and Normand Jolicoeur for technical assistance. This work was supported in part by The Natural Science and Engineering Research Council of Canada Grant 0GP0183691. Received for review November 25, 2003. Accepted March 4, 2004. AC035396M

Analytical Chemistry, Vol. 76, No. 10, May 15, 2004

2927