An SPR-Based Screening Method for Agonist Selectivity for Insulin

ACS eBooks; C&EN Global Enterprise ..... Magnetic “Fishing” Assay To Screen Small-Molecule Mixtures for Modulators of Protein−Protein Interactio...
0 downloads 0 Views 81KB Size
Download PDF
Anal. Chem. 2000, 72, 6-11

An SPR-Based Screening Method for Agonist Selectivity for Insulin Signaling Pathways Based on the Binding of Phosphotyrosine to Its Specific Binding Protein Taishi Yoshida,† Moritoshi Sato,†,‡ Takeaki Ozawa,†,‡ and Yoshio Umezawa*,†,‡

Department of Chemistry, School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan, and Japan Science and Technology Corporation, Tokyo, Japan

A new screening method was developed that evaluates physiologically relevant chemical selectivity of agonists for insulin-signaling pathways. Phosphorylation (pY939) by an insulin-activated insulin receptor of a target peptide (Y939) derived from an insulin receptor substrate-1 (IRS1) and its subsequent binding to another downstream target, the SH2 domain of PI-3 kinase (SH2N), were detected by surface plasmon resonance (SPR) spectrometry. This method is based on competitive binding of SH2N to pY939 either in a solution or on the gold surface of the SPR sensor chip. With increasing the concentration of pY939 in solution by the insulin-induced kinase reaction of insulin receptor, SH2N bound to pY939 in solution increases and the one on the sensor chip decreases, thereby causing a decrease in the SPR signal. The amount of thus-detected complex pY939-SH2N was found to depend on added insulin concentrations, confirming that the method utilized part of the sequential transduction mechanism of the insulin-signaling pathways. The kinase activity of insulin receptor-agonist complexes increased in the order of IGF-II < IGF-I < insulin, and neither vanadium ions nor thiazolidine-type medicines for NIDDM, troglitazone and pioglitazone, directly acted on both the kinase reaction of insulin receptor or the binding of pY939 to SH2N. The present approach will thus become a general method for screening agonists for one specific pathway in tyrosine phosphorylation of IRS-1 in insulin signaling, which is regulated by specific protein-protein interaction between a phosphorylated tyrosine in IRS-1 and its corresponding SH2 domain-containing protein such as PI-3 kinase, Grb2-Sos, or SHP2. Screening of agonists concerning their physiologically relevant selectivity is very important for biological studies as well as for pharmaceutical needs. The conventional binding assays can neither give sufficient information on the agonist activity nor discriminate between agonists and antagonists. Recently, alterna* To whom correspondence should be addressed: (phone) +81-3-58414351; (fax) +81-3-5841-8349; (e-mail) [email protected]. † The University of Tokyo. ‡ Japan Science and Technology Corporation.

6 Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

tive methods have been emerging for evaluating chemical selectivity of agonists on the basis of physiologically relevant signaltransduction mechanisms by receptor proteins for generating analytical signals.1-9 Typical examples of such new types of biosensing are shown in ion-channels and transporter proteins embedded in lipid-bilayer membranes.4-7 These sensors utilize the corresponding transmembrane signals, such as ion-channel currents or active membrane transport. The tyrosine kinase signaling system, which exists widely and plays important roles in the cell, was also used for evaluating physiologically relevant chemical selectivity of insulin agonists.2,10 When insulin binds to its receptor, the insulin-receptor complex undergoes phosphorylation of tyrosine residues in endogenous substrate proteins such as insulin receptor substrate-1 (IRS-1).11,12 Upon tyrosine phosphorylation of IRS-1, phosphotyrosines serve as a beacon that attracts src homology-2 (SH2) domain-containing proteins such as phosphatidyl inositol 3-kinase (PI-3 kinase),13 Ras guanine-nucleotide-releasing complex (Grb2Sos),14 or src homology-containing phosphatase-2 (SHP2).15 These (1) Radecka, H.; Nakanishi, J.; Hirano, A.; Sugawara, M.; Umezawa, Y. J. Pharm. Biomed. Anal. 1999, 19, 205-216. (2) Ozawa, T.; Sato, M.; Sugawara, M.; Umezawa, Y. Anal. Chem. 1998, 70, 2345-2352. (3) Ozawa, T.; Kakuta, M.; Sugawara, M.; Umezawa, Y.; Ikura, M. Anal. Chem. 1997, 69, 3081-3085. (4) Sugawara, M.; Hirano, A.; Reha´k, M.; Nakanishi, J.; Kawai, K.; Sato, H.; Umezawa, Y. Biosens. Bioelectron. 1997, 12, 425-439. (5) Sugao, N.; Sugawara, M.; Minami, H.; Uto, M.; Umezawa, Y. Anal. Chem. 1993, 65, 363-369. (6) Ottenbacher, D.; Ja¨hnig, F.; Go ¨pel, W. Sens. Actuators, B 1993, 13, 173175. (7) Minami, H.; Sugawara, M.; Odashima, K.; Umezawa, Y.; Uto, M.; Michaelis, E. K.; Kuwana, T. Anal. Chem. 1991, 63, 2787-2795. (8) Kiefer, H.; Klee, B.; John, E.; Stierhof, Y.-D.; Ja¨hnig, F. Biosens. Bioelectron. 1991, 6, 233-237. (9) Gotoh, M.; Tamiya, E.; Momoi, M.; Kagawa, Y.; Karube, I. Anal. Lett. 1987, 20, 857-870. (10) Okada, Y.; Yokono, K.; Katsuta, A.; Yoshida, M.; Morita, S.; Irino, H.; Goto, T.; Baba, S.; Roth, R. A.; Shii, K. Anal. Biochem. 1998, 257, 134-138. (11) White, M. F.; Kahn, C. R. J. Biol. Chem. 1994, 269, 1-4. (12) Keller, S. R.; Lienhard, G. E. Trends Cell Biol. 1994, 4, 115-119. (13) Backer, J. M.; Myers, M. G.; Shoelson, S. E.; Chin, D. J.; Sun, X. J.; Miralpeix, M.; Hu, P.; Margolis, B.; Skolnik, E. Y.; Schlessinger, J.; White, M. F. EMBO J. 1992, 11, 3469-3479. (14) Sun, X. J.; Crimmins, D. L.; Myers, M. G.; Miralpeix, M.; White, M. Mol. Cell. Biol. 1993, 12, 7418-7428. (15) Kuhne´, M. R.; Pawson, T.; Lienhard, G. E.; Geng, G. S. J. Biol. Chem. 1993, 268, 11479-11481. 10.1021/ac990795w CCC: $19.00

© 2000 American Chemical Society Published on Web 11/20/1999

Figure 1. Principle for the present physiologically relevant insulin assay system. 11-Mercaptoundecanoic acid (MUA) is self-assembled on the gold surface attached to the glass support, and pY939 is immobilized on the MUA monolayer. (a) At low concentration of insulin, SH2N binds to pY939 on the gold surface, which causes a large change in SPR signals. (b) At a high concentration of insulin, the concentration of pY939 peptide in a sample solution increases by the insulin-induced kinase reaction of insulin receptor. Consequently, SH2N tends to bind more to the solution of pY939 competitively than to the immobilized pY939 on the gold surface, and therefore, small change in SPR signal results.

SH2 domain-containing proteins act as switching units by specifically recognizing the phosphotyrosine residues. Each SH2 domain-containing protein recognizes not only phosphotyrosine but also specific amino acid residues around the phosphotyrosine: YMXM for PI-3 kinase,16 YXNX for Grb2-Sos,17 or YIDL for SHP2.15 This specificity of binding of those SH2 domaincontaining proteins to phosphotyrosines and surrounding amino acid residues contributes to the sorting out of the signal for one particular function among multifunctional signals for tyrosine phosphorylation of IRS-1 in insulin signaling.18 In our previous paper, the amount of phosphotyrosine in IRS-1 generated by the kinase reaction of insulin receptor was measured with a monoclonal anti-phosphotyrosine antibody for evaluating chemical selectivity of agonists.2 If the specificity of the binding of signalregulating phosphotyrosine to its corresponding SH2 domaincontaining protein is added one step downstream from phosphorylation of tyrosine residues of IRS-1, the sensing method thus designed will be capable of analyzing agonists with selectivity for one particular insulin-signaling pathway among multifunctional signals. This is, in fact, the purpose of this study. In this paper, a new screening method is described for evaluating agonist selectivity for insulin-signaling pathways governing activation of PI-3 kinase, which is based on the insulinsignaling process, i.e., the binding of insulin to its receptor, phosphorylation of IRS-1, and its subsequent binding of the SH2 domain of PI-3 kinase. The principle is schematically shown in Figure 1. Insulin receptor serves as a primary receptor for insulin and Y939 as a target peptide. The Y939 peptide, a synthetic 12amino-acid residue, consists of a tyrosine-phosphorylation site of (16) Rodorf-Nikolic, T.; Van Horn, D. J.; Chen, D.; White, M. F.; Backer, J. M. J. Biol. Chem. 1995, 270, 3662-3666. (17) Myers, M. G.; Wang, L. M.; Sun, X. J.; Zhang, Y.; Yenush, Y.; Schlessinger, Y.; Pierce, J. H.; White, M. F. Mol. Cell Biol. 1994, 14, 3577-3587. (18) Pawson, T.; Scott, J. D. Science (Washington, D.C.) 1997, 278, 2075-2080.

IRS-1 and a binding domain of IRS-1 to PI-3 kinase.19,20 SH2N is the N-terminal SH2 domain of PI-3 kinase and binds to phosphorylated Y939 (pY939).21 The binding of SH2N to this pY939 is monitored by surface plasmon resonance (SPR) spectrometry. When a kinase reaction of insulin receptor is induced by an added agonist and part of Y939 in solution is thereby phosphorylated, SH2N in the solution competitively binds either to pY939 on the sensor chip or to the one in solution. Upon increasing the added agonist concentration, the pY939 generated by the kinase reaction of insulin receptor increases in the solution. Consequently, the binding of SH2N to solution pY939 will be increased and that immobilized on the sensor chip decreased. This process is detected by SPR. The amount of SH2N-pY939 complex generated by the kinase reaction of insulin receptor thus is reflected as changes in the observed SPR signals and is a selective and sensitive measure for agonists regulating the activity of PI-3 kinase in insulin signaling. EXPERIMENTAL SECTION Materials. Human insulin was purchased from Peptide Institute, Inc. (Osaka, Japan). Recombinant human insulin-like growth factor-I (IGF-I) was purchased from Funakoshi Co. (Tokyo, Japan). Recombinant human insulin-like growth factor-II (IGF-II) was kindly provided as a gift from Wakunaga Co. (Hiroshima, Japan). Chemically synthesized HPLC purified Y939 peptide, which consists of the amino acid sequence of SEEYMNMDLGPC (expressed by one-letter abbreviations) and tyrosine-phosphorylated Y939 (pY939) were purchased from Sawady Technology Co. (Tokyo, Japan). These Y939 and pY939 peptides differ from their native amino acid sequences in insulin receptor substrate-1 (IRS1) by an added cysteine residue, which is needed for biotinylation of the synthetic peptides at the C-terminal end of both peptides. N-hydroxysuccinimide (NHS) was obtained from Wako Pure Chemical Industries (Osaka, Japan). (()-5-[4-(6-Hydroxy-2,5,7,8tetramethylchroman-2-ylmethoxy) benzyl]-2,4-thiazolidinedione (CS-045, troglitazone) was kindly provided as a gift from Sankyo Co. (Tokyo, Japan). (()-5-[p-[2-(5-Ethyl-2-pyridyl)ethoxy] benzyl]2,4-thiazolidinedione hydrochloride (AD-4833, pioglitazone) was kindly donated by Takeda Chemical Industries (Osaka, Japan). 2-[4-(2-Hydroxy-ethyl)-1-piperazinyl] ethanesulfonic acid (HEPES) and N-ethyl-N′-[(dimethylamino)propyl]carbodiimide (EDC) were obtained from Dojindo Laboratories (Kumamoto, Japan). 11Mercaptoundecanoic acid (MUA) and biotinylated pY939 were synthesized as previously described.2,22 The SH2-domain-transformed Escherichia coli (DH5R) was kindly provided as a gift from Drs. Masato Kasuga and Wataru Ogawa, Kobe University (Kobe, Japan). Chinese hamster ovary cells overexpressing human insulin receptor (CHO-HIR cells) was kindly gifted from Drs. Takashi Kadowaki and Kohjiro Ueki, the University of Tokyo (Tokyo, Japan). Other salts and solvents used were all of the highest purity available. All aqueous solutions were prepared with Milli-Q grade (>18.2 MΩ resistance) water obtained with a Milli-Q Plus System (Millipore Corp., Bedford, MA). (19) Piccone, E.; Case, R. D.; Domchek, S. M.; Hu, P.; Chaudhuri, M.; Backer, J. M.; Schlessinger, J.; Shoelson, S. E. Biochemistry 1993, 32, 3197-3202. (20) Nishiyama, M.; Wands, J. R. Biochem. Biophys. Res. Commun. 1992, 183, 280-285. (21) Yonezawa, K.; Kasuga, M. J. Biol. Chem. 1992, 267, 25958-25966. (22) Bain, C. D.; Troughton, E. B.; Tao, Y. T.; Evall, J.; Whitesides, G. M.; Nuzzo, R. G. J. Am. Chem. Soc. 1989, 111, 321-325.

Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

7

SPR Measurements. All surface plasmon resonance (SPR) measurements were performed on a SPR sensor system of DKK Co. (Tokyo, Japan). The sensor system used SPR to probe refractive-index changes in a flow cell due to the binding of molecules to an immobilized ligand. The change in refractive index detected by the SPR sensor system was expressed as an arbitrary unit called the resonance unit (RU).23 In the present study, the flow rate of all solutions through the flow cell adjacent to the SPR sensor chip was always kept at 50 µL/min at 25.0 °C. Extraction of SH2 Domain Protein and Human Insulin Receptor. The N-terminal SH2N domain of the p85 subunit of PI-3 kinase (SH2N) was obtained as previously described.21,24 SH2N was expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli (DH5R) and purified by affinity chromatography on a glutathione agarose. The purity of the SH2N was evaluated by Coomassie stained SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and a 36 KDa single band was obtained. The extracted SH2N was stored at -80 °C until use. Insulin receptor was extracted from Chinese hamster ovary cells overexpressing human insulin receptor (CHO-HIR cells) as previously described.24,25 The insulin receptor extracted from CHOHIR cells was purified by affinity chromatography on a WGAsepharose. The eluent containing insulin receptor was concentrated by pressured dialysis using a Diaflo ultrafiltration membrane PM-30 (Amicon Inc., Beverly, MA). The concentrated insulin receptor suspension thus prepared was divided into each 200-µL portion and stored at -80 °C until use. Preparation of Sample Solutions. Each sample solution was prepared in a microtube. Forty five microliters of a solution consisting of 0.05% Triton X-100, 150 mM NaCl, 100 µM ATP, 5 mM MnCl2, 10 µg of insulin receptor (0.16 pmol), 10 mM HEPES/ NaOH (pH 7.4) and each concentration of agonist was incubated in the microtube for 1 h at 4 °C. For phosphorylation of Y939 peptide by the insulin receptor, a 30-µL portion of 10 µM Y939 solubilized in HEPES buffer (0.05% Triton X-100, 150 mM NaCl, and 10 mM HEPES/NaOH, pH 7.4) was added into the microtube, and the mixture was shaken on a mixer (Tomy Seiko Co., Tokyo, Japan) for 2 h at 37 °C. To terminate the kinase reaction by masking Mn2+ with EDTA, a HEPES buffer solution containing EDTA was added to achieve a final concentration of 7.5 mM.24 Upon addition of SH2N to this solution, it was equilibrated at 4 °C for 1 h before being used for SPR measurements. Immobilization of pY939 on a Gold Surface. The surface of gold substrates for sensor chips was modified with a selfassembled monolayer of mercaptoundecanoic acid (MUA), by immersing the sensor chip into 1 mM MUA in EtOH for more than 10 h at 4 °C. After setting the MUA-modified sensor chip onto a glass prism of the SPR sensor system, the modified sensor surface was equilibrated with a conditioning solution containing 0.05% Triton X-100, 150 mM NaCl, and 10 mM HEPES/NaOH (pH 7.4). The flow rate of solutions adjacent to the sensor chip was kept at 50 µL/min. The carboxyl group of MUA on the gold surface was activated for amidation by 350 µL of an NHS/EDC mixture (0.05 M NHS, 0.2 M EDC in Milli-Q water) for 7 min. (23) Granzow, R.; Reed, R. Biotechnology 1992, 10, 390-393. (24) Sato, M.; Ozawa, T.; Yoshida, T.; Umezawa, Y. Anal. Chem. 1999, 71, 39483954. (25) Gazit, A.; Yaish, P.; Gilon, C.; Levitzki, A. J. Med. Chem. 1989, 32, 23442352.

8

Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

Avidin was coupled with MUA via free amino groups of avidin by injecting 350 µL of 125 µg/mL avidin solution in the 5 mM maleate buffer (pH 6.0) over the sensor chip for 7 min.26 After immobilizing avidin, 350 µL of 200 µM ethanolamine solution was injected over the sensor chip for 7 min to mask free carboxyl groups of MUA. To immobilize pY939 on the gold surface via an avidin-biotin complex, 0.15 nM biotinylated pY939 (biotin-pY939) dissolved in 7 mM phosphate buffer (pH 7.4) was injected over the sensor chip for 5 min.2 The pY939-modified gold surface thus made was washed with a 1 mM HCl solution containing 500 mM NaCl for 3 min, following which the surface was equilibrated with the conditioning solution. Calibration of SPR Signals for the Binding of pY939 to SH2N. The observed SPR signal, RU, for the binding of SH2N to biotinylated pY939 immobilized on the sensor chip was expressed as ∆RU, which was obtained as the difference in the observed SPR signals between sample solutions with and without SH2N. When a pY939-immobilized sensor chip was repeatedly used for measuring SPR signals, ∆RU changed even for the same sample solutions because of unexpected alteration in experimental conditions such as nonselective adsorption of proteins onto the sensor chip or slight change of temperature in the flow cell. To correct these possible drifts in ∆RU values, two different standard solutions were occasionally measured once in every 10 times as follows: The two standard solutions both containing 100 nM SH2N but one without and one with 1 µM pY939 were used as the standard for correction, and the ∆RU value measured for each sample solution was normalized against the difference of ∆RU values for these two standard solutions, which was termed as Rp:

Rp(%) )

S - S1 × 100 S0 - S1

(1)

where S is ∆RU for respective sample solutions, S0 is the one for the standard solution without pY939, and S1 is the one with 1.0 µM of pY939. All measurements for sample solutions were made in triplicate or more, and the results were expressed as means ( SD. P values were determined by an unpaired Student’s t-test to compare mean values; P < 0.10 was considered to be statistically significant. RESULTS AND DISCUSSION (a) SPR Response for Insulin. Upon binding of insulin to its receptor, Y939 is phosphorylated, and this phosphorylated Y939 (pY939) binds to SH2N. The binding event of SH2N to pY939 for insulin signaling was monitored by the competitive binding reaction of SH2N to either biotinylated pY939 on the SPR sensor chip or to pY939 generated by the kinase reaction of insulin receptor in solution. A typical time profile of the SPR signal is shown in Figure 2. After equilibrating the sensor chip ((1) in Figure 2) where the observed SPR signal (RU) was defined as 0 RU, a sample solution containing 100 nM SH2N and pY939 generated by the insulin-induced kinase reaction of insulin receptor was injected over the sensor chip for 5 min. Upon introducing 1.0 × 10-12 M of insulin, the SPR signal initially (26) Johnson, B.; Lo¨fås, S.; Lindquist, G. Anal. Biochem. 1991, 198, 268-277.

Figure 2. Typical time profiles of the absolute angle of SPR signals. The surface-immobilized pY939 was equilibrated with a running buffer containing 0.05% Triton X-100, 0.5% BSA, 150 mM NaCl, 7.5 mM EDTA, HEPES/NaOH (pH 7.4) ((1) and (4)). Sample solutions containing 100 nM SH2N and pY939 formed by the insulin-dependent kinase reaction of 10 mg of insulin receptor were passed through the pY939-immobilized gold surface (2). Concentrations of insulin in the sample solution were 1.0 × 10-12 M (plain line) and 9.8 × 10-8 M (dashed line). After the end of each injection, the pY939immobilized sensor chip was regenerated with 0.5% SDS solution (3).

decreased for 20 s ((2)-(a) in Figure 2) and subsequently increased ((2)-(b)). This initial decrease in the SPR signal may originate from the bulk refractive-index changes between the running buffer and the sample solutions and the following increase in SPR signals may be due to the binding of SH2N to biotinpY939 immobilized on the sensor surface. When the concentration of insulin was 9.8 × 10-8 M, the increase in the SPR signals was lower (dashed line in Figure 2), indicating that SH2N in the sample solution competitively bound more to solution pY939 than to the immobilized pY939 on the surface. The former pY939 was generated by an insulin-dependent kinase reaction of insulin receptor. To dissociate SH2N from biotin-pY939 on the sensor chip, 0.5% SDS solution was injected for 1 min ((3) in Figure 2) and the sensor surface was equilibrated with the running buffer until the SPR signal returned to the 0 RU((4) in Figure 2). It took a relatively long time in terms of regaining the original SPR base signal level to regenerate the surface. This may be due to the high concentration of SDS that was used for desorption of SH2N from pY939, which required more time to totally replace with the running buffer. For the sample solution containing 1.0 × 10-12 M insulin, the continuous rise in the SPR signals reached a plateau in 10 min (data not shown). Upon increasing the insulin concentrations, a longer time was required for this process. The extent of this time interval seemed inconvenient for repetitive measurements of many analytes. Analyte-dependent SPR signals were therefore sampled at a fixed time of 5 min after starting the injection of sample solutions, and the SPR signals thus obtained were empirically calibrated by comparing them with the ones obtained for known concentrations of insulin under the identical experimental conditions. Figure 3 shows the dependence of the SPR signals on the concentration of insulin. The Rp values on the y-axis show normalized SPR signals calculated according to eq 1 (see Experimental Section). Upon increasing the concentration of

Figure 3. Effect of insulin concentration on Rp values. The concentrations of free insulin in sample solutions were 1.0 × 10-12 M, 1.0 × 10-11 M, 1.6 × 10-10 M, 8.0 × 10-9 M, and 9.8 × 10-8 M, respectively.

insulin, the Rp value decreased. This is due to the fact that pY939 generated by the kinase reaction of insulin receptor increased upon increasing the concentration of insulin and that SH2N bound to biotin-pY939 decreased by the competitive reaction. It is thus confirmed that the present sensing system can detect the binding of pY939 to SH2N through insulin signaling by the competitive reaction as illustrated in Figure 1. The relationship between the analyte concentration and Rp values was not sigmoidal. It has been shown that typical kinase assays of insulin receptor with [γ-32P]ATP and synthetic substrates such as random copolymers of poly(Glu:Tyr)(4:1) are insulindependent sigmoidal 32P incorporation into the substrates. This discrepancy may be due to the difference in the detection methods for the phosphorylated substrates: In the present method, the detection of the phosphorylated substrates was based on the competitive reaction of SH2N between biotin-pY939 on the chip and pY939 in sample solutions. Some complicated kinetics in the competitive reaction on the sensor surface might have caused this nonsigmoidal relationship between the insulin concentration and the SPR signal. The exact reason for this, however, remains to be worked out. (b) SPR Response for Insulin-like Growth Factors. IGF-I and IGF-II are peptide hormones having structures similar to insulin and bind to insulin receptor like insulin. They thereby induce a postreceptor insulin signaling such as phosphorylation of insulin receptor itself and of its endogenous substrates, as well. To evaluate agonist selectivity of these insulin-like growth factors on the insulin-signaling pathway, the amounts of pY939 generated by the IGF-I- and IGF-II-induced kinase reaction of insulin receptor were measured as a function of their concentrations. The results were shown in Figure 4. Upon increasing the concentration of IGF-I from 8.5 × 10-10 to 9.8 × 10-8 M, the Rp values decreased. At a concentration lower than 8.5 × 10-10 M, no change in the Rp values was observed. In the case of IGF-II, the Rp values decreased when its concentration went from 3.8 × 10-10 to 9.8 × 10-7 M. These decreases in the Rp values in both cases are due to the fact that the amount of pY939 generated by the kinase reaction of insulin receptor increased upon increasing the concentration of these agonists. Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

9

Table 1. Effect of Agonist on Rp Values

Figure 4. Effect of agonist concentration on Rp values. The concentrations of free IGF-I in the sample solutions were 8.4 × 10-12 M, 8.4 × 10-11 M, 8.5 × 10-10 M, 9.1 × 10-9 M, 9.8 × 10-8 M and 9.8 × 10-7 M, respectively. The concentrations of free IGF-II in the sample solutions were 3.0 × 10-12 M, 3.1 × 10-11 M, 3.8 × 10-10 M, 1.0 × 10-9 M, 8.1 × 10-9 M, 9.8 × 10-8 M, and 9.8 × 10-7 M, respectively.

Binding affinities of IGF-I, IGF-II, and insulin toward insulin receptor have been evaluated;27,28 the dissociation constants (Kd) for IGF-I, IGF-II, and insulin are 11, 0.9, and 0.23 nM, respectively. At the concentrations of their dissociation constants, one-half of total insulin receptors are occupied by their respective agonists. In the present study, differences in the kinase activity of agonistbound insulin receptor were evaluated at each agonist concentration of Kd values by comparing the obtained Rp values for each agonist. The resulting Rp values for IGF-I, IGF-II, and insulin were 55.8 ( 5.7, 71.0 ( 4.7, and 44.7 ( 9.8, respectively. The Rp values at agonist concentration which give their Kd values increased in the order of IGF-II < IGF-I < insulin, indicating that kinase activities of insulin receptor induced by these agonists increased in this order. Considering that binding affinity of these agonists toward insulin receptor increases in the order of IGF-I < IGF-II < insulin, the order of IGF-I and IGF-II was found to be reversed. This suggests that the present approach is based on the very magnitude of the amount of phosphorylated Y939 induced by each agonist, which reflects not only the binding affinity for the agonists but also the ability of signal transduction, i.e., phosphorylation of Y939 and its subsequent binding to SH2N. (c) SPR Response for Anti-diabetic Medicines such as Vanadium Ions and Thiazolidines. An insulin-like effect of vanadate and vanadyl ions has been reported to increase the amount of endogenous phosphotyrosine content and stimulate glucose uptake in various kinds of cells.29-31 The possibility of using these ions for antidiabetic medicines has been examined;32 exact mechanisms are, however, yet to be known for why these (27) Schumacher, R.; Soos, M. A.; Schlessinger, J.; Brandenburg, D.; Siddle, K.; Ullrich, A. J. Biol. Chem. 1993, 268, 1087-1094. (28) Sakano, K.; Enjoh, T.; Numata, F.; Fujiwara, H.; Marumoto, Y.; Higashihashi, N.; Sato, Y.; Perdue, J. F.; Fujita-Yamaguchi, Y. J. Biol. Chem. 1991, 266, 20626-20635. (29) Miralpeix, M.; Gil, J.; Rosa, J. L.; Carreras, J.; Bartrons, R. Life Sci. 1989, 44, 1491-1497. (30) Dubyak, G. R.; Kleinzeller, A. J. Biol. Chem. 1980, 255, 5306-5312. (31) Schechter, Y.; Karlish, S. J. D. Nature (London) 1980, 255, 5306-5312. (32) Poucheret, P.; Verma, S.; Grynpas, M. D.; McNeill, J. H. Mol. Cell. Biol. 1998, 188, 73-80.

10 Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

agonist

Rp value (%)

no agonist insulin 1 µM troglitazone 10 µM pioglitazone 10 µM VO2+ 100 µM H2VO4- 100 µM

85.1 ( 6.3 9.3 ( 6.2 74.8 ( 6.3 84.0 ( 7.1 83.0 ( 8.1 82.4 ( 5.4

ions induce the above insulin-mimetic effect. The effect of vanadate and vanadyl was therefore examined with the present SPR method on the phosphorylation of Y939 peptide and the binding of pY939 to SH2N. The results are shown in Table 1. Significant change in the Rp values was not observed with or without vanadyl or vanadate, indicating that both of these ions neither effected the tyrosine kinase activity of insulin receptor nor the following binding event of pY939 to SH2N. Vanadate and vanadyl are known to be protein phosphotyrosine phosphatase (PTPase), alkaline phosphatase, Ca2+/Mg2+ ATPase and Na+/K+ ATPase inhibitors.33-37 The exact mechanism of how the vanadium ions inhibit those phosphohydrolyzing enzymes remains to be worked out. It is, however, widely believed that these vanadium ions bind to the phosphotyrosine binding site of the enzymes because of their structural similarities with phosphate.38 Though SH2N used in this study as well as PTPase recognizes phosphotyrosine, the effect of vanadate on the binding of phosphotyrosine to SH2 domains has not been reported. The present study confirmed that vanadate and vanadyl take roles neither of agonists of phosphotyrosine nor of antagonists. Thiazolidine type medicines such as troglitazone and pioglitazone are known as oral hypoglycemic agents for noninsulindependent diabetes mellitus (NIDDM).39,40 The effects of these medicines on the tyrosine-kinase signaling of insulin receptor were evaluated by the present method. The obtained results are shown in Table 1. A significant difference in the Rp values was not observed for sample solutions with or without 10 µM troglitazone or pioglitazone, showing that these medicines affect neither the kinase reaction of insulin receptor nor the binding of pY939 to SH2N. Studies with rat fat cells and rat fibroblasts transfected with human insulin receptor have suggested that hyperglycemia reduces the insulin receptor kinase activity, and both troglitazone and pioglitazone in the presence of insulin are capable of increasing this reduced tyrosine kinase activity in the cell.41,42 It was shown by our previous study that these thiazolidine-type (33) Elberg, G.; Li, J.; Shechter, Y. J. Biol. Chem. 1994, 269, 9521-9527. (34) Swarup, G.; Cohen, S.; Garbers, D. L. Biochem. Biophys. Res. Commun. 1982, 107, 1104-1109. (35) Ramasarma, T.; Crane, F. L. Curr. Top. Cell. Regul. 1981, 20, 247-301. (36) Macara, I. G. Trends Biochem. Sci. 1980, 5, 92-94. (37) Cantley, L. C.; Resh, M. D.; Guidotti, G. Nature (London) 1978, 272, 552554. (38) Fantus, I. G.; Tsiani, E. Mol. Cell. Biochem. 1998, 182, 109-119. (39) Sohda, T.; Momose, Y.; Meguro, K.; Kawamatsu, Y.; Sugiyama, Y.; Ikeda, H. Arzneim.-Forsch. 1990, 40, 37-42. (40) Yoshioka, T.; Fujita, T.; Kanai, T.; Aizawa, Y.; Kurumada, T.; Hasegawa, K. J. Med. Chem. 1989, 32, 421-428. (41) Kellerer, M.; Kroder, G.; Tippmer, S.; Berti, L.; Kiehn, R.; Mosthaf, L.; Ha¨ring, H. U. Diabetes 1994, 43, 447-453. (42) Mu ¨ llaer, H. K.; Kellerer, M.; Ermel, B.; Mu ¨ hlho ¨fer, A.; Obermaier-Kusser, B.; Vogt, B.; Haring, H. U. Diabetes 1991, 40, 1440-1447.

medicines do not directly enhance the kinase activities of the insulin receptor;2 the result obtained in this study is consistent with the previous work and suggests, moreover, that neither troglitazone nor pioglitazone affected the binding of SH2N to pY939; instead, they may, rather, regulate insulin receptorindependent signaling pathways in a manner to that of DNA transcriptional regulators including peroxisome proliferatoractivated receptor γ (PPAR γ).43 (d) Significance of the Present Method. Noninsulin-dependent diabetes mellitus (NIDDM) involves progressive development of insulin resistance and a defect in insulin secretion, which leads to overt hyperglycemia. Several studies have shown modest decreases in the number of insulin receptors among tissues or cells, decreases in insulin-stimulated receptor tyrosine kinase activity, and defects in receptor-mediated IRS phosphorylation or PI 3-kinase activation.44,45 Thus, a subset of NIDDM patients have clear defects in insulin signaling that might, in theory, be overcome by treatment aimed at augmenting the receptor function. The discovery of orally active small molecules that mimic insulin’s effects could eventually lead to alternative therapies for this disorder. Recently, a new molecule, L-783,281, which activates tyrosine kinase of insulin receptor, was reported:46 This molecule was selected from among fifty thousand other candidate compounds by a cell-based screening assay system; incubation of cultured cells that overexpress the insulin receptor, immunopurification of insulin receptor, and assays for measuring tyrosine kinase activity toward the exogenous substrate of poly(Glu:Tyr)(4:1) with [γ-32P]ATP. In contrast to this conventional cell-based radioisotope approach, new screening methods for insulin-like substances have been developed that are based on insulin-signaling pathways: Okada et al. proposed a method to determine the ability of the insulin-like substances to stimulate the autophosphorylation of insulin receptor, which was measured by a two-site immunofluorometric assay using monoclonal anti-insulin receptor antibodies and europium-labeled anti-phosphotyrosine antibodies.10 Ozawa et al. detected a step one step downstream of the autophos(43) Komers, R.; Vrana, A. Physiol. Res. (Prague) 1998, 47, 215-225. (44) Goodyear, L. J.; Giorgino, F.; Sherman, L. A.; Carey, J.; Smith, R. J.; Dohm, G. L. J. Clin. Invest. 1995, 95, 2195-2204. (45) Caro, J. F.; Sinha, M. K.; Raju, S. M.; Ittoop, O.; Pories, W. J.; Flickinger, E. G.; Meelheim, D.; Dohm, G. L. J. Clin. Invest. 1987, 79, 1330-1337. (46) Zhang, B.; Salituro, G.; Szalkowski, D.; Li, Z.; Zhang, Y.; Royo, Y.; Vilella, D.; Diez, M. T.; Palaez, F.; Ruby, C.; Kendall, R. L.; Mao, X.; Griffin, P.; Calaycay, J.; Zierath, J. R.; Heck, J. V.; Smith, R. G.; Moller, D. E. Science (Washington, D.C.) 1999, 284, 974-977.

phorylation of insulin receptor, i.e., phosphorylation of its target peptide derived from IRS-1, which was measured with peroxidaselabeled anti-phosphotyrosine antibody.2 In both cases, nonradiolabeled immunoassay techniques were used for the detection of phosphorylated insulin receptor or its substrates after addition of insulin-like agonists to the receptor. In the present system, phosphorylation of Y939 peptide and its subsequent binding to the SH2 domain in PI-3 kinase are detected, the signal of which covers insulin signaling pathways even further downstream of those described above. This method neither needs to isolate phosphorylated Y939 as the analyte nor any amplification steps needed for the immunoassay techniques. The method could, therefore, replace the above-mentioned timeconsuming procedures for the cell-based screening or immunoassay technique.2,10,46 CONCLUSION A new screening method was developed that can evaluate the physiologically relevant agonist selectivity for insulin-signaling pathways governing activation of PI-3 kinase. This method is based on competitive binding of SH2N to pY939 either in a solution generated by insulin receptor or on the gold surface, which was monitored by the SPR technique. The amount of SH2N binding to pY939 on the surface depended on the concentration of insulin, confirming that the sequential transduction mechanism is based on the insulin signaling pathway. The present assay method also revealed that the observed order of selectivity of IGF-I and IGFII for the kinase activity of insulin receptor was changed compared with their binding affinities for insulin receptor. The method presented in this paper will be applicable for rapid screening of insulin-mimetic substances that activate particular insulin-signaling pathways and thus become very important not only for fundamental biological studies but also for pharmaceutical needs. ACKNOWLEDGMENT We thank D. Kuboshima for experimental help at an early stage of this work. This work has been supported by CREST (Core Research for Evolutional Science and Technology) of JST (Japan Science and Technology) and by Grants from the Ministry of Education, Science and Culture, Japan to Y.U. A research fellowship from the Japan Society for the Promotion of Science (JSPS) to M.S. is gratefully acknowledged. T.O. wishes to thank the Sumitomo Foundation for financial support. Received for review July 20, 1999. Accepted October 20, 1999. AC990795W

Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

11