Method and apparatus for analysis of protein-protein interaction

Abstract

Disclosed are a method for analyzing an activation state of a signaling pathway in a cell or tissue through protein-protein interaction analysis, a method for selecting a tailored personal therapeutic agent and/or monitoring efficacy of a therapeutic agent using the analysis method, and a device for use therein.

Claims

1. A method for measuring activation of a signaling pathway in a cell or a tissue, comprising steps (1), (2), (3), (4), and (5), or steps (1), (2), (3), (4-1), (4-2), and (5): (1) preparing a substrate having a first protein immobilized thereto by adding a test sample containing the first protein to the substrate; (2) adding and reacting the prepared first protein-immobilized substrate with a maker-conjugated second protein; (3) measuring a signal from the reactant obtained in step (2); (4) measuring a signal value per unit amount of the first protein in the test sample added in step (1) based on the signal measured in step (3); (4-1) obtaining a signal value per unit amount of the test sample added in step (1) based on the signal measured in step (3); (4-2) obtaining a signal value per unit amount of the first protein contained in the test sample based on the signal value per unit amount of the test sample as measured in step (4-1); (5) comparing a result obtained in step (4) or (4-2) with that obtained in a reference sample, wherein the test sample is the cell, the tissue, lysate, homogenate, or extract of the cell or the tissue, or a body fluid comprising the cell or the tissue, which are all isolated from a mammalian subject, the reference sample comprises a normal cell, a cell having a known activation level of a signaling pathway in which the first protein is involved, or a cell isolated from a subject having a known activation level of a signaling pathway in which the first protein is involved, and the first protein is involved in the signaling pathway, and the second protein interacts with the first protein, wherein the signal value per unit amount of the first protein of step (4) or (4-2) is a quantitative value of a signal or signal intensity obtained by dividing the signal value measured in step (3) or (4-1) by weight or concentration of the first protein in the test sample added in step (1).

2. The method of claim 1, wherein the marker is at least one selected from the group consisting of a small-molecule compound, a protein, a peptide, and a nucleic acid, all of which generate a signal that can be measured through detection of an enzymatic reaction, fluorescence, luminescence, or radiation.

3. The method of claim 1, wherein the marker in the step (2) is at least one selected from the group consisting of a small-molecule compound, a protein, a peptide, and a nucleic acid, all of which generate fluorescence; and the step (3) of measuring a signal is carried out using a total internal fluorescence microscope, a fluorescence camera, or a combination thereof.

4. The method of claim 3, wherein the fluorescence camera is set to have an exposure time of about 0.001 sec to about 1 sec per frame.

5. The method of claim 1, wherein the first protein and the second protein are each independently at least one selected from proteins involved in the signaling pathway in the cell or tissue, and the second protein is a protein that is located downstream of the first protein in the signaling pathway.

6. The method of claim 5, wherein the first protein is a cell membrane protein.

7. The method of claim 6, wherein the first protein is at least one selected from the group consisting of receptor tyrosine kinases, toll-like receptors, G-protein-coupled receptors (GPCR), transferrin receptors, low-density lipoprotein (LDL) receptors, ROS1; BCR-Abl1 fusion proteins; non-receptor kinases; GTPases; hormone receptors; anti-apoptotic proteins; and immune checkpoint proteins.

8. The method of claim 1, wherein: the first protein and the second protein are each independently of two or more kinds of proteins involved in the signaling pathway in the cell or tissue, the second protein being located downstream of the first protein in the signaling pathway, and the value, obtained in step (4) or (4-2), per unit amount of the first protein contained in the test sample is a sum of the values respectively obtained for each of the two or more kinds of the first protein and each of the two or more kinds of the second protein.

9. A method for screening a first protein as a target of a therapy suitable for application to a subject, comprising steps (1), (2), (3), (4), and (5), or steps (1), (2), (3), (4-1), (4-2) and (5):(1) preparing a substrate having a first protein immobilized thereto by adding a test sample containing the first protein to the substrate; (2) adding and reacting the prepared first protein-immobilized substrate with a maker-conjugated second protein; (3) measuring a signal from the reactant obtained in step (2); (4) measuring a signal value per unit amount of the protein in the test sample added in step (1) based on the signal measured in step (3), or (4-1) obtaining a signal value per unit amount of the test sample added in step (1) based on the signal measured in step (3); and (4-2) obtaining a signal value per unit amount of the first protein contained in the test sample based on the signal value per unit amount of the test sample as measured in step (4-1), wherein the steps (1), (2), (3), and (4), or steps (1), (2), (3), (4-1), and (4-2) are carried out for each of two or more different first proteins; and (5) comparing the results obtained in steps (4) or (4-2) for the two or more different first proteins, wherein the test sample is a cell, a tissue, a cell or tissue lysate, homogenate, or extract, or a body fluid, which are all isolated from the subject, and the first protein is involved in the signaling pathway, and the second protein interacts with the first protein, wherein the signal value per unit amount of the first protein of step (4) or (4-2) is a quantitative value of a signal or signal intensity obtained by dividing the signal value measured in step (3) or (4-1) by weight or concentration of the first protein in the test sample added in step (1).

Description

DESCRIPTION OF DRAWINGS

(1) FIG. 1 is a schematic diagram of a method for measuring a single molecular protein interaction.

(2) FIG. 2 is a graph showing the result of confirming a target protein (first protein) immobilized on the substrate.

(3) FIG. 3 is a fluorescence image showing protein interaction after injection of a fluorescence-labeled interacting protein (a second protein).

(4) FIG. 4 is a graph showing quantified results of the number of PPI complexes observed in FIG. 3.

(5) FIG. 5 is a graph showing changes in the number of PPI complexes according to the amount of injected cell lysate.

(6) FIG. 6 is a schematic diagram showing the process of quantifying the first protein through a single-molecule sandwich ELISA.

(7) FIG. 7 is a graph showing specificities measured by single-molecule sandwich ELISA.

(8) FIG. 8 is a graph showing changes in the number of PPI complexes according to the types of cells (red {circle around (1)} vs light blue {circle around (3)}) and states (red {circle around (1)} vs black {circle around (2)}).

(9) FIG. 9 is a graph showing changes in the number of PPI complexes for various target RTKs (first proteins) according to cell conditions.

(10) FIG. 10 is a graph showing changes of PPI complex according to the EGFR mutation state and the ratio of EGFR activated per each cell based on the changes.

(11) FIG. 11 shows the results of performing the same method for EGFR in FIG. 10 on HER2 and HER3.

(12) FIG. 12 is a heatmap showing the result of measuring the interactions between EGFR, MET, HER2, HER3 (the first protein) and the downstream signal transduction protein (the second protein) in each cell line.

(13) FIG. 13 is a graph quantitatively showing the results shown in FIG. 12 (left and middle), and a graph showing the reactivity results of AZD9291, which is a EGFR targeted anticancer drug (right).

(14) FIG. 14 is a graph showing the correlation between the reactivity (left, y axis) and the activation score (left side, x axis) of the EGFR targeted anticancer drug (AZD9291), and the diversity (right) of the targeted anticancer response according to the genotype.

(15) FIG. 15 is a heatmap showing the intensities of HER2 and HER3 signals in breast cancer cell lines.

(16) FIG. 16 is a graph showing expression level of HER2 (upper) and HER3 (middle), which are conventional biomarkers for predicting the drug response of breast cancer cell line to trastuzumab, and inhibition level of cell growth (bottom) by trastuzumab.

(17) FIG. 17 is a graph showing the correlation between the PPI score measured using the HER2 or HER3 signal and the drug response to trastuzumab (log GI.sub.50).

(18) FIG. 18 is a heatmap showing the PPI complex signal results of EGFR, MET, HER2, and HER3 with three downstream signal transduction proteins measured in the PDTX mouse model.

(19) FIG. 19 is a graph showing EGFR expression level (upper) and activation score (bottom) calculated using the EGFR expression level in the PDTX mouse model.

(20) FIG. 20 is a graph showing the results of measuring changes in tumor size by administering gefitinib to a PDTX mouse model.

(21) FIG. 21 is a graph showing the correlation between tumor growth inhibition by gefitinib and EGFR activation score in the PDTX mouse model.

(22) FIG. 22 is a graph showing EGFR PPI complex counts measured in tissues before and after being treated with gefitinib in a PDTX mouse model, respectively.

(23) FIGS. 23a to 23i show drug responses to EGFR targeted inhibitor in a lung cancer PDTX model, wherein

(24) 23a is a schematic diagram illustrating a process for preparing a PDTX model,

(25) 23b is a graph showing tumor volume changes in PDTX model when treated with vehicle or indicated EGFR-specific inhibitor, wherein the tumor volume changes are measured in lung adenocarcinoma PDTXs (PDTX-A1 to A3) treated with osimertinib (5 mg per kg of weight daily) and lung squamous cell carcinoma (SQCC) PDTXs (PDTX-S1˜S5) treated with gefitinib (50 mg per kg of weight daily) (population of each PDTX test group is 3 or more),

(26) 23c is a graph showing PPI complex counts (the number of PPI complexes) for the downstream signal proteins of the indicated receptor tyrosine kinases (RTK; EGFR, HER2, HER3 and MET),

(27) 23d is a graph showing the EGFR expression levels in 8 PDTX (A1 to A3 and S1 to S5) individuals, which are normalized to EGFR expression level in A549 cells (control group),

(28) 23e and 23f are graphs with a tumor growth inhibition ratio (%) on the y axis and values obtained by dividing EGFR PPI sum of PDTX models (e) and SQCC PDTX models (f) by the EGFR level on the x axis,

(29) 23g is a graph showing changes in PPI complex counts (the number of PPI complexes between EGFR and the second protein indicated on x axis) when treated with gefitinib every day for 15 days,

(30) 23h is a graph showing the degree of tumor growth in PDTX-S1 (n=2) with co-treatment of gefitinib and BKM120 for 15 days, compared to single treatment, and

(31) 23i is a graph in which x axis shows values obtained by dividing the PPI sum by the EGFR level in all 8 PDTX (A1 to A3 and S1 to S5) individuals, and y axis shows the tumor growth inhibition ration (Error bars: s.d.).

(32) FIGS. 24a to 24d show examples of application of single-molecule co-IP and single-molecule immunolabeling to human tumor samples, wherein

(33) 24a shows human tumor tissues obtained by tumor resection surgery of two tumor patients (P1 and P2),

(34) 24b is a graph showing expression levels and PTM level (immunolabelling level) of 10 proteins measured by single-molecule immunolabeling, and PPI level of 10 protein-protein pairs obtained by performing single-molecule co-IP to each of the 10 samples using a high-efficiency single molecule imaging system, wherein PC9 cells (for EGFR), HCC827 cells (for MET), and SKBR3 cells (for HER2 and HER3) are respectively used as positive controls,

(35) 24c is a graph showing PPI complex counts for the indicated RTK in P1 and P2, and

(36) 24d is a graph showing changes in PPI complex counts of PLCgammaSH2 and Grb2 when PGFN1 treatment is performed after pulling-down of EGFR on surface (Error bars: s.d.).

(37) FIGS. 25a to 25h show the characteristics of PDTX-models (n=3),

(38) 25a to 25c are graphs showing MET levels (a), HER2 levels (b), and HER3 levels, compared to the levels of MET, HER2, and HER3 in HCC827 cells (for MET) and SKBR3 cells (for HER2 and HER3) (Error bars: s.d.; n=5), indicating that none of the RTKs is overexpressed,

(39) 25d is an image showing immunohistochemical staining (IHC) results of EGFR measured representatively in 5 SQCC PDTX models, wherein the expression of EGFR was determined by calculating EGFR H-score by a magnification rule,

(40) 25e is a scatter diagram showing the correlation between EGFR level measured by single-molecule immunolabeling and EGFR H-score, indicating that the IHC H-score shows complete linear correlation with total EGFR expression level measured by single-molecule immunolabeling,

(41) 25f and 25g are scatter diagrams showing the correlation between tumor growth inhibition and EGFR level (g) and PPI sum (h) in SQCC PDTX models,

(42) 25h shows Immunoblot analysis results of PDTX-S2 treated with vehicle or gefitinib, wherein after treatment with gefitinib for 15 days, the phosphorylation (pEGFR) of the 1068.sup.th residue, tyrosine, of EGFR (pEGFR) is completely removed (disappeared) and phosphorylation of AKT and S6K (pAkt and pS6K) is also inhibited by gefitinib, indicating that the tumor growth inhibitory effect in PDTX-S2 model is obtained by inhibiting the EGFR/AKT/mTOR/S6K signaling pathway by gefitinib treatment.

(43) FIGS. 26a and 26b show the effects of gefitinib treatment on PDTX-S1 and PDTX-S2 models, wherein

(44) 26a is a graph showing changes in EGFR level when treated with gefitinib for 15 days (Error bars: s.d.; n=5),

(45) 26b is a graph showing the inhibition degree of EGFR PPI by gefitinib treatment, wherein EGFR PPI complex count in A549 cell is used as a negative control (Error bars: s.d.; n=5).

(46) FIG. 27 is a perspective view schematically illustrating a multi-well according to an embodiment.

(47) FIG. 28 is a diagram showing the multi-well turned upside down.

(48) FIGS. 29 and 30 show processes of manufacturing the multi-well.

(49) FIG. 31 is a graph showing GFP counts in the multi-wells manufactured by an example and a comparative example, respectively (Y-axis: GFP counts).

(50) FIGS. 32 and 33 are graphs showing GFP counts in the multi-wells in multi-well A of an embodiment (FIG. 32) and multi-well B of a comparative example (FIG. 33), in which no antibody is immobilized.

(51) FIG. 34 is a graph showing GFP counts in multi-well A of an embodiment depending on immobilization of an antibody on the well.

(52) FIG. 35 is a graph showing GFP counts in multi-well A of an embodiment according to amount of cell sample.

(53) FIGS. 36 and 37 are graphs showing GFP counts in multi-well A of an embodiment depending on immobilization of an antibody on the well.

(54) FIG. 38 illustrates numbered multi-well A of an embodiment.

MODE FOR INVENTION

(55) A better understanding of the present invention may be obtained through the following examples which are set forth to illustrate, but are not to be construed as limiting the present invention.

Example 1: Preparation of First Protein (EGFR, MET, HER2, and HER3)

(56) EGFR, MET, HER2, and HER3 were selected as first proteins. From lysates resulting from the lysis of cell lines (e.g., cancer cell lines) containing the proteins, the first proteins were obtained. This process will be explained in detail below:

(57) 1.1. Cell Lysate Preparation

(58) 1.1.1. Cell Line Preparation

(59) A cell line was seeded in an amount of 2×10.sup.6 cells and cultured in a medium (RPMI1640, high glucose (Thermo 11965-092)). When reaching 90% confluency or higher in a 100-pi culture dish, the cells were collected and aliquoted into two 1.5 ml tubes. After centrifugation (5 min×15,000 g), the culture medium was discarded and the cell pellets were frozen at −80° C. for storage.

(60) Cell lines prepared are listed in Table 1, below:

(61) TABLE-US-00001 TABLE 1 Cell Line Source/Accession number Lung Cancer PC9 CVCL_B260 Cell Line HCC4006 ATCC, CRL-2871 HCC827 ATCC, CRL-2868 H1650 ATCC, CRL-5883 HCC4006-ER CVCL_S746 HCC827-GR5 CVCL_V622 H1666 ATCC, CRL-5885 H2291 ATCC, CRL-5939 A549 ATCC, CCL-185 H358 ATCC, CRL-5807 YU-105 Yonsei University, derived from patient HCC827-GR13 Yonsei University PC9-GR CVCL_S706 YU-101 Yonsei University, derived from patient H1975 ATCC, CRL-5908 Breast Cancer SKBR3 ATCC, HTB-30 Cell Line BT474 ATCC, HTB-20 HCC1419 ATCC, CRL-2326 HCC2218 ATCC, CRL-2343 MDA-MB-453 ATCC, HTB-131 HCC1954 ATCC, CCRL-2338 SKBR3-HR Seoul National University Hospital SKBR3-LR Seoul National University Hospital MCF7 ATCC, HTB-22 T47D ATCC, HTB-133 MDA-MB-231 ATCC, HTB-26

(62) 1.1.2. Cell Lysate Preparation

(63) A cell lysis buffer was prepared to have the composition of 50 mM Tris-HCl (pH 7.4), 1% (v/v) Triton X-100, 150 mM NaCl, 1 mM EDTA, 10% (v/v) glycerol, protease inhibitor cocktail (Sigma, P8340) 100×, and tyrosine phosphatase inhibitor cocktail (Sigma, P5726) 100×.

(64) The cell aggregates of the cell line sample prepared in Example 1.1.1 were broken up by pipetting. The prepared cell lysis buffer was added in an amount of 200 μl per tube to the pipetted cell line sample. The sample was then stored for 30 min in a cold block (0-4° C.) on ice for a reaction during which the cells were physically mixed by pipetting at regular intervals of 10 min to incite the surfactants to perform cell lysis.

(65) After 30 min of the cell lysis reaction, centrifugation was conducted (10 min, 15,000 g, 4° C.). Then, the pellet was discarded and the supernatant was filtered through a membrane having pores with a size of 0.2 μm. The filtrate was transferred into a new tube and stored until a subsequent experiment.

(66) The cell lysate was found to have a total protein concentration of about 5-10 mg/ml as measured by a total protein quantitation method (Bradford, BCA, DC protein assay, etc.).

(67) 1.2. Tissue Lysate Preparation

(68) 1.2.1. Construction of Patient-Derived Tumor Xenograft Model

(69) A lung squamous cell carcinoma (SQCC) patient-derived tumor xenograft was granted from a Yonsei University research team. Patient-derived tumor xenografts (PDTXs) were constructed. In brief, female severe combined immunodeficient mice (NOG) and nude mice (nu/nu mice; Orient Bio), both 6 to 8 weeks old, were prepared. All animal experiments were conducted according to the guidelines set forth by the Institutional Animal Care and Use Committee (IACUC). A clinical tumor sample derived from a patient was cut into a fragment of 3 mm or less in size and subcutaneously implanted to the flank of each of the prepared NOG mice. Tumor growth rates in the subcutaneous tissue were obtained by measuring tumor sizes twice a week with calipers. When grown to have a diameter of about 1.5 cm, the tumor tissue was excised and sectioned into small fragments (hexahedra with each side about 5 mm long). The sectioned tissue was reimplanted into different mice to sequentially acquire individuals subsequently developing tumors. The mice that retained the patient-derived tumor were designated F0 and the mice subsequently developing tumors derived from F0 were designated FI, F2, F3, F4, and the like, sequentially. A vehicle (PBS) or gefitinib was administered to mice having the 3.sup.rd-generation subsequent tumor (F3) before use in experiments. Intraperitoneal injection of gefitinib or a vehicle at a dose of 50 mg/kg into the prepared PDTXs was performed once a day. Fifteen days after gefitinib injection, tumor tissues were collected from the PDTXs and monitored for PPI and expression level change.

(70) 1.2.2. Tissue Lysate Preparation

(71) The tumor tissue obtained in Example 1.2.1. was prepared in an amount of about 20 mm.sup.3. A greater volume may be acceptable.

(72) The lysis buffer prepared in Example 1.1.2 was added in an amount of about 300 μl per 20 mm.sup.3 of the prepared tumor tissue and subjected to a reaction for one hour in a 4° C. refrigerator while rotating. In this regard, the tissue was cut as finely as possible with operating scissors to realize a large surface area per volume, thereby maximizing the efficiency of the chemical reaction with the surfactant in the lysis buffer.

(73) After one hour of the reaction described above, centrifugation was conducted (10 min, 15,000 g, 4° C.). Thereafter, the precipitate (pellet) was discarded and the supernatant was filtered through a membrane having pores with a size of 0.2 μm. The filtrate was transferred into a new tube and stored until a subsequent experiment.

(74) The tissue lysate was found to have a total protein concentration of about 5-10 mg/ml as measured by a total protein quantitation method (Bradford, BCA, DC protein assay, etc.).

Example 2: Preparation of Second Protein

(75) In this Example, illustration is made of a process of preparing a second protein that is in the form of a fluorescent-protein-attached protein downstream of the first protein prepared in Example 1.

(76) The second proteins illustrated in this Example are summarized in Table 2, below.

(77) TABLE-US-00002 TABLE 2 Expres- sion vector Protein Accession number Marker (Source) PLC- A nucleic acid molecule coding eGFP pEGFP-N1 gamma- for a sequence of amino acids or SH2 545-765 on PLCg (NM_013187.1) pEGFP-C1 or a sequence of amino acids vector 540-765 on NP_002651.2 was (Clontech) cloned. Grb2 A nucleic acid molecule coding for a full a. a. sequence of Grb2 (NP_002077.1) or an a. a. sequence of SH2 (57-155), SH3-SH2 (1-154), or SH2-SH3 (57-217) was cloned P85-alpha A nucleic acid sequence coding for a full a. a. sequence of human p85a (NP_852664.1) or an a. a. sequence of N-SH2 (333-428), C-SH2 (624-718), or tandem SH2 (333-718) or for a full a. a. sequence of mouse p85a (P26450) or an a. a. sequence of N-SH2 (333- 428), C-SH2 (624-718), or tandem SH2 (333-718) was cloned

(78) For preparing the second proteins, HEK293 cell line (ATCC) and HeLa cell line (ATCC), both of which express low levels of the first proteins prepared in Example 1, were obtained.

(79) An expression vector for each of the second proteins listed in Table 2 was introduced into the prepared HEK293 cells or HeLa cells, which were then cultured to express the second proteins. After cultivation for 24 hours, the cells were harvested and aliquoted in appropriate amounts for storage at −80° C.

(80) To the cells, a lysis buffer (50 mM Tris-HCl (pH 7.4), 1% Triton X-100, 150 mM NaCl, 1 mM EDTA, 10% glycerol, protease inhibitor cocktail (Sigma, P8340) 100×, tyrosine phosphatase inhibitor cocktail (Sigma, P5726) 100×) was added. To begin with, the lysis buffer was added in an amount of 60 μl to 5×10.sup.5 cells because a high concentration of the surfactant (Triton X-100) might interfere with interaction between proteins.

(81) Cell aggregates in the reaction mixture thus obtained were broken up by pipetting, and the detached cells were then stored for 30 min in a cold block (0-4° C.) on ice for a reaction during which the cells were physically mixed by pipetting at regular intervals of 10 min to incite the surfactant to do cell lysis.

(82) After 30 min of the cell lysis reaction, centrifugation was conducted (10 min, 15,000 g, 4° C.). Then, the precipitate (pellet) was discarded and the supernatant was transferred into a new tube and stored until a subsequent experiment.

(83) To 140 μl of PBS, 60 μl of the obtained aqueous solution was added. As a result, a solution of the second protein containing 0.3% Triton X-100 could be obtained. Concentrations of the fluorescent protein attached to the second protein were measured using a fluorimeter. As a result, three kinds of the second proteins were found to range in concentration from 400 to 1000 nM.

Example 3: Preparation of Substrate

(84) A coverslip was immersed in a 1 M KOH solution and then washed in a sonicator (20-30 min). Thereafter, the coverslip was washed well with deionized water and then with a piranha solution (sulfuric acid: hydrogen peroxide=2:1-3:1 (v/v)). The washed coverslip was coated sequentially with aminopropyl silane and PEG.

(85) After reaction for two hours, the coverslip was washed with deionized water and stored at 20° C. in such a way that the PEG-coated surface was not in contact with any matter until use.

(86) Simultaneously, a channel-type quartz substrate or a well-type acryl substrate was prepared. For the quartz substrate, washing and PEG coating processes were carried out with reference to the aforementioned procedure for the coverslip.

(87) After being constructed, the acryl substrate was immersed in deionized water and washed by sonication. The washed acryl substrate was immersed in a 5% BSA solution, reacted for two hours to prevent nonspecific protein adsorption, and then stored at −20° C. until use.

(88) In the following test, the prepared coverslip and acryl substrate were used. The coverslip and the acryl substrate were thawed and assembled before testing. Alternatively, after PEG coating, the coverslip and the acrylate substrate were assembled and stored in an assembled form at −20° C. until use. Immediately before use, the assembly was thawed.

Example 4: Imaging of Protein-Protein Interaction Between 1.SUP.st .and 2.SUP.nd .Protein

(89) To the prepared substrate, the avidin-lineage protein Neutravidin (Thermo, A2666) was fed at a concentration of 0.1 mg/ml. After 5 min of reaction at room temperature, the substrate was washed twice with 30 μl of PBS buffer.

(90) An antibody against the first protein to be targeted was added to the prepared substrate. In this context, the antibody was in a biotinylated form. The concentration of the antibody was suitably controlled according to antigen-antibody affinity (dissociation constant, KD). In this Experimental Example, the antibody was used at a concentration of 2 ug/ml, with a reaction time of 5 min given thereto. In the case where an antibody was not conjugated with biotin, a secondary antibody might be used to attach the anti-first-protein antibody to the substrate.

(91) The anti-first-protein antibodies used are summarized in Table 3, below.

(92) TABLE-US-00003 TABLE 3 First protein Antibody Source EGER anti-EGFR antibody MS-378-B0, Thermo MET anti-MET antibody Ab89297, abcam HER2 anti-HER2 antibody OP39, Calbiochem HER3 anti-HER3 antibody 66201, R&D systems

(93) The antibody-treated substrate was washed twice with 30 μl of PBS buffer. To the prepared substrate was added the cell lysate solution or tissue lysate solution containing the first protein prepared in Example 1. The reaction time was set to 15 min because an antigen-antibody reaction might increase for 15 min, but might decrease in efficiency over time after 15 min.

(94) After the reaction, the substrate was washed with a PBS buffer containing 0.05% (v/v) Tween 20. The 0.05% Tween 20 helps prevent hydrophobic regions of membrane proteins from collapsing as well as reducing nonspecific binding.

(95) Subsequently, the second protein lysate solution obtained in Example 2 was added to the substrate. The concentration of the second protein in the first protein lysate solution was set to 1-50 nM (about 30 nM) on the basis of the fluorescent protein. A second protein concentration of 100 nM or higher increases background noise in a florescence microscope, acting as a hindrance to the measurement of accurate fluorescent signals.

(96) The substrate was fixed on and imaged by a fluorescence microscope to obtain data for the first protein/second protein, respectively.

Example 5: Protein Complex (PPI Complex) Analysis

(97) A PPI complex was analyzed on the basis of the toolkit provided by the MATLAB program (MathWorks).

(98) The fluorescent images obtained in Example 4 was stored in a 16-bit unsigned integer format. Fluorescent signals came from eGFP (enhanced green fluorescent protein). For use in observing the signals, a laser was given a wavelength of 488 nm. The laser power was adjusted to 2 mW to maintain the luminescence of eGFP for about 11 sec. Of the entire frames (30 frames), early and middle frames were discarded, and an average of the images of three frames (22.sup.nd-25.sup.th frames) was taken to generate one image. This process was repeatedly conducted at various positions within a well to acquire a total of five images with which the following procedures were then performed. The first about 20 frames were discarded in order to select a section where unnecessary signals (autofluorescence) generated from the null surface of the substrate disappears, and eGFP signals are maintained. Such a selected section may vary depending on imaging condition/established equipment state. In this Example, an EMCCD (Electron-multiplying charge-coupled device; Andor iXon Ultra 897 EX2 (DU-897U-CS0-EXF)) camera was used to obtain fluorescence images with an exposure time of 0.1 per frame and an EMCCD gain value of 40.

(99) In order to remove noises, the following procedure was conducted for each frame:

(100) (a) A start is made from the upper left. One frame was composed of 512×512 pixels. A median value was obtained from 11×11 pixels in the region of a distance of 11 pixels in the right direction from the reference pixel by a distance of 11 pixels in the left direction from the reference pixel. This median value was subtracted from the value of the reference pixel [(Intensity_pixel)−(medianIntensity_11×11neighborhood)]. This procedure was conducted for every 512×512-pixel region. Through median filtering, pepper & salt noise was removed.

(101) (b) The processed image was made smooth by Gaussian smoothing at sigma=0.7, size=5×5.

(102) (c) A threshold was set. Thereby, a pixel value less than a threshold is increased to the threshold value (using an algorithm of searching for local maxima in MATLAB toolkit). Through this process, the local maxima that were not created by fluorescent signals can be removed from the image. A threshold value of 70 was used in the imaging condition of this Example.

(103) Signals from a fluorescent protein are detected as a localized point spread function (PSF). A number of PSF (physical value) is a PPI complex score between the first protein and the second protein to be measured (biological value). The PSF value was converted into a PPI complex score, which is a biological value, via the following procedures:

(104) (a) Positions of local maxima were obtained (for example, i.sup.th row, j.sup.th column pixel). As described above, single-molecule fluorescence signals are localized and formed at a particular position (about 5×5 pixel size, 1 pixel=0.167 micrometers, under the current observation equipment). Hence, the discovery of local maxima allows the selection of individual PSF. This can be obtained using a toolkit provided by MATLAB.

(105) (b) A process of determining whether or not the local maxima obtained in (a) were actually generated from PSF was performed. To begin with, a minimum intensity value of the local maxima was defined. Analysis was conducted only in the case where the local maxima obtained above were greater than the minimum intensity value. The minimum intensity value used in this Example was 75, and may vary depending on laser power/exposure time/established equipment state. Information of 5×5 pixels based on the finally obtained local maximum coordinates was retrieved and the centroid of intensity was determined in the 5×5 pixels. Here, if the obtained centroid of intensity deviated by 0.5 pixels or greater from the existing local maximum coordinate (if 2D symmetry for the PSF pattern disappeared), the fluorescent signal was determined to be abnormal and excluded from analysis.

(106) (c) Only the PSF that passed all test conditions were selected for determination of the coordinates and the total number. Herein, the total number of PSF obtained are the number of the PPI complex.

(107) Every file photographed under the same condition was subjected to the above procedure to obtain numbers of PSF which were then aggregated and used to calculate a mean and standard deviation. This value fmally represents the number of PPI complex in a particular condition (expressed as “Number of single PPI complexes” in the drawing).

Example 6: Determination of PPI Strength, PPI Score, and Activation Score

(108) In the graph where concentrations of the cell lysate (see Example 1.1.2) are set forth on the X axis and PPI complex values measured at concentrations of the cell lysate (see Example 5) are set forth on the Y axis, the slope, that is, [(PPI complex)/(number of PPI complex per unit concentration of cell lysate (1 μg/ml))] was defined as PPI strength (or PPI slope) between the first and the second protein in the cell. All PPI strengths obtained for each cell line were summed. The sum of PPI strengths was defined as a PPI score for the cell line. The sum of PPI strengths (or a PPI score) accounts for a total of PPI of the first protein and the second protein tested at a unit concentration of cell lysate of each cell line.

(109) The sum of PPI strengths can be calculated according to the following equation:

(110) $\mathrm{Sum}\mathrm{of}PP{I}_{\mathrm{cancer}\mathrm{cell}}^{1\mathrm{st}\mathrm{protein}}=\underset{k=2\mathrm{nd}\mathrm{protein}}{.Math.}{\left(PPI\mathrm{strength}\right)}_k^{1\mathrm{st}\mathrm{protein}}$

(111) (1.sup.st protein: RTK (EGFR, MET, HER2, or HER3 for lung cancer; and HER2 and HER3 for breast cancer);

(112) (2.sup.nd protein: downstream protein (PLC-gamma-SH2, Grb2, p85-alpha)).

(113) In order to express relative PPI scores among cell lines, the PPI score of a particular cell line (hereinafter referred to as “reference cell line”; PC9 cells among lung cancer cell lines and SKBR3 cells among breast cancer cell lines were used in this Example) was set to 1, on the basis of which PPI scores obtained in other cells (cells other than the reference cell, hereinafter referred to as “test cell”) were normalized. In this regard, the value obtained for each cell line was defined as a normalized PPI score for the cell line.

(114) In addition, a total amount of the first protein in each cell lysate (for example, RTK (EGFR, MET, HER2, or HER3) for lung cancer; and HER2 and HER3 for breast cancer) was measured. As the total amount of first protein, a value obtained by dividing a quantitative result from Sandwich ELISA, quantitative western blot, etc. using each antibody (see Table 3) by the weight of the cell lysate (weight of total protein in the cell lysate) was determined.

(115) A value obtained by dividing the PPI score or normalized PPI score by the total amount of the first protein was defined as an activation score. Here, in order to express relative activation scores among cell lines, the activation score of a reference cell line (PC9 cells for lung cancer cell lines and SKBR3 cells for breast cancer cell lines) was set to 1 on the basis of which activation scores obtained in test cell lines were normalized. In this regard, the value obtained for each cell line was defined as a normalized PPI activation for the cell line.

(116) In the case where a PPI score is obtained in a PDTX (patient-derived tumor xenograft) mouse model, a negative background measured may be subtracted from the value obtained in the above-described manner so as to reduce background noise. For the negative background, a normal tissue from the same patient or a cancer cell lysate with a normal EGFR gene may be used. For example, in A549 cells with a normal EGFR gene, interaction between EGFR and each downstream protein is measured to obtain a PPI score. This is set forth as a negative background. A final PPI score is calculated by subtracting the negative background from the PPI score obtained in each PDTX mouse model.

Example 7: Heatmap Construction

(117) In addition to the digitalization of data in Example 5, a heatmap was constructed to give supplemental data for analytical decision. The heatmap is an option for representing data, but is not intended to limit the analysis of data.

(118) Of X and Y axes, one is given to second proteins (downstream proteins) while the other is for kinds of cell lines to create a 3×16 lattice structure (number of second proteins (a total of 3: see Table 2 (p85-alpha, Grb2, and PLC-gamma-SH2))×number of cell lines (a total of 15 (lung cancer cell lines) or a total of 11 (breast cancer cell lines): see Table 1)). This lattice structure was created for each first protein (four lattice structures (EGFR, MET, HER2, and HER3; lung cancer), or two lattice structure (HER2 and HER3; breast cancer) (for HER3, only p85-alpha was used as a second protein). Then, color and brightness were determined for each lattice according to the PPI strength obtained between the first protein and the second protein in corresponding cells to construct a heatmap (for example, a deep black color, a moderately bright red color, and a bright green color may be given in increasing order of PPI strength, but the color and brightness may not be standardized, but arbitrarily determined via tests by a researcher). Regardless of which cell is set forth as a reference, the relative difference between cell lines is not changed.

Example 8: Drug Responsiveness and Correlation Between PPI Score and Activation Score

(119) With reference to the drawing, the test results in Examples 1 to 7 are explained as follows:

(120) FIG. 1 is a schematic view of a method for measuring single-molecule protein interaction. The left panel schematically illustrates that neutravidin, an anti-RTK antibody, and a cell or tissue lysate to be analyzed are added in that order to a polyethylene glycol-coated substrate and then washed, through which an RTK protein target is fixed to the substrate, as shown in the middle panel. In the right panel, there is a representation that a fluorescence-labeled interacting protein is added, followed by observing and quantitating a fluorescent signal to measure the level of single-molecule protein interaction.

(121) FIG. 2 gives graphs showing identification results of target proteins (first proteins) immobilized on substrates. Assays were conducted with reference to the methods described in Examples 4 and 5. EGF treatment was performed in an amount of 100 ng/μl for 3 min. The left graph of FIG. 2 shows a result after EGFR in H1666 cells was immobilized on a substrate via an antibody (MA5-13266, ThermoFisher) against an extracellular domain of EGFR and the immobilization was identified with an antibody (#4267, Cell signaling technology) against a cytoplasmic domain of EFGR while the middle graph of FIG. 2 gives a result after an antibody against HER2 was added to identify whether or not EGFR formed a dimer with HER2. In the right graph of FIG. 2, there is a result after an antibody against Shc1 was added to identify whether EGFR formed a dimer with Shc1.

(122) According to the addition of an antibody to a substrate, the target RTK protein (first protein) is (expressed as +) or is not (expressed as −) immobilized on the substrate. Through the results, suitable antibodies can be selected to attach various target proteins to a substrate. Further, it was identified that not only single target proteins, but also protein conjugates existing in the body, like EGFR-HER2 or EGFR-Shc1, can be immobilized on a substrate.

(123) FIG. 3 gives images showing protein interaction after a fluorescence-labeled interacting protein (second protein) was added to a first-protein-immobilized substrate. The PPI complex observed as described in Example 5 is represented as a localized point spread function (PSF) and was selected using a computer algorithm. A fluorescent signal was observed only after a downstream protein was added. Green circles represent observed PPI complexes.

(124) FIG. 4 is a graph in which the amounts of PPI complexes observed in FIG. 3 were quantified. When a downstream protein (second protein) is added (PLCgammaSH2, Grb2, and p85-alpha on x axis), a high level of PPI complex was selectively observed. In contrast, a low level of signals was observed in the absence of an antibody against the target RTK protein (EGFR) (black bar) or a downstream protein (buffer on X axis). The signal observed in both cases can be understood to be background noise.

(125) FIG. 5 shows graphs in which the number of PPI complexes increases with increasing amount of cell lysate. It is observed that the amount of PPI complexes (Y axis) linearly increases as the amount of the cell lysate including a target RTK protein (first protein: EGFR) increases (X axis). This analysis allows quantitative comparison of PPI complexes between samples in a given amount of a particular cell lysate.

(126) FIG. 6 is a schematic view illustrating a process of quantitating a first protein by single-molecule sandwich ELISA. The process of attaching a target RTK protein (first protein) to a substrate is the same as in FIG. 1. Instead of a fluorescence-labeled downstream protein (second protein), a second antibody recognizing the target RTK protein may be added to quantitate the target RTK protein. Here, the RTK protein should have the second antibody (detection antibody) and should have respective different antibody-recognizing sites (epitopes) for the first antibody (pull-down antibody) used to immobilize the RTK protein to the substrate and for the second antibody. From a fluorescence-labeled antibody recognizing the second antibody, the amount of the RTK protein immobilized on the surface can be measured using a single-molecule technique (see Example 5).

(127) FIG. 7 is a graph showing a specificity result obtained through single-molecule sandwich ELISA. It was observed that the absence of even one of the components shown in the schematic view of FIG. 6 resulted in poor single-molecule sandwich ELISA signal results.

(128) FIG. 8 gives graphs showing changes in the number of PPI complexes according to kinds (red {circle around (1)} vs. sky blue {circle around (3)}) and states (red {circle around (1)} vs. black {circle around (2)}) of cell lines. It was observed that the activation of a target RTK protein (first protein; EGFR) existing in cells by a corresponding ligand (EGF+) resulted in a higher level of PPI complexes at the same amount than otherwise. In addition, the presence of an activity mutation in the target RTK (PC9, sky blue) was observed to increase the number of PPI complexes with the target RTK.

(129) FIG. 9 gives graphs showing changes in the number of PPI complexes per unit concentration of a sample (PPI slope) according to various target RTKs (first proteins). A change in the number of PPI complexes between ligand stimulation (gray) and non-stimulation (black) of the target proteins (first proteins) EGFR, MET, HER2, and HER3 was quantitatively measured using the single-molecule co-IP technique described in Example 6. Based on the result, the activity of target RTK can be measured through PPI complex quantitation.

(130) FIG. 10 gives graphs showing changes in the number of PPI complexes according to EGFR mutation states and the ratios of activated EGFR per cell, calculated on the basis of the changes. The upper graph shows interaction results between EGFR and downstream proteins according to individual cell lines as analyzed by the PPI complex measurement method and the lower graph shows quantitated levels of activated EGFR per cell (absolute occupancy (%)) that are obtained by measuring an expression level of EGFR per cell using single-molecule sandwich ELISA (see FIG. 6) and dividing the expression level by the interaction result.

(131) FIG. 11 is a graph showing absolute occupancy (%) obtained by applying the same method as for EGFR in FIG. 10 to HER2 and HER3. Very poor activity was detected for HER2 whereas HER3 shows very high activity.

(132) FIG. 12 is a heatmap showing interaction (signal strength) between EGFR, MET, HER2, and HER3 (first proteins) and proteins downstream thereof (second proteins) for lung cancer cell lines (Example 7). Color indicators for signal strength are given below the heatmaps.

(133) FIG. 13 is a graph showing a sum of quantified values of respective signal strengths between EGFR (first protein) and three different second proteins out of the results of FIG. 12 (left and middle) and responsiveness of individual cell lines to the EGFR-targeted anticancer agent AZD9291 (osimertinib) (IC50; a concentration at which cell viability decreases by 50%, compared to no treatment).

(134) Bar colors account for the groups divided according to EGFR gene mutations at the right side. The activation scores were found to have more significant correlation with drug responsiveness (IC50) than the PPI scores (the higher the activation score, the lower the IC50 value (higher drug responsiveness)).

(135) FIG. 14 is a graph showing correlation between responsiveness (Y axis) and activation scores of an EGFR-targeted anticancer agent (AZD9291) (left) and a variety of responsiveness of the targeted anticancer agent according to gene types (right). In this graph, it is observed that the activation scores exhibit high correlation (r=0.85) with the responsiveness of AZD9291 and although the genes are the same type, different responsiveness may be obtained by conventional EGFR genetic assays.

(136) FIG. 15 is a heatmap showing signal strength (interaction) between HER2 and HER3 (first proteins) and downstream proteins (second proteins) in breast cancer cell lines (Example 7).

(137) FIG. 16 is a graph showing expression levels of HER2 (upper) and HER3 (middle), which are the biomarkers conventionally used to predict the responsiveness of the anticancer agent trastuzumab in breast cancer cell lines and degrees of the trastuzumab-induced cell growth inhibition (lower).

(138) FIG. 17 gives graphs showing correlations between PPI scores measured using HER2 or HER3 signals and trastuzumab responsiveness (log GI50). PPI scores (r=0.91) are found to have higher correlation with trastuzumab responsiveness, compared to expression levels of the conventional biomarkers HER2 (r=0.54) and pHER2 (r=0.44) (lower graphs).

(139) FIG. 18 is a heatmap showing PPI complex signals between EGFR, MET, HER2, and HER3 and three different downstream proteins thereof, as measured in tissue lysates from PDTX mouse models (Example 1.2) (n=5; expressed as PDTX-1, -2, -3, -4, and -5).

(140) FIG. 19 gives graphs showing expression levels of EGFR in tissue lysates obtained from PDTX mouse models (Example 1.2) (upper) and activation scores calculated using the expression levels of EGFR (results of FIG. 18) (lower).

(141) FIG. 20 gives graphs showing changes of tumor size in gefitinib (50 mg/kg)-injected PDTX mouse models (Example 1.2.1) in comparison with results in vehicle (PBS)-administered groups. For PDTX-2, although the expression level of EGFR was low, a high activation score was detected (see FIG. 19) and an excellent anti-tumor effect was obtained (see FIG. 20). These results indicate that the activation scores (that is, activated EGFR ratios) are in closer correlation with drug responsiveness than EGFR expression levels.

(142) FIG. 21 is a graph showing correlation between degrees of gefitinib-induced tumor growth inhibition and EGFR activation scores in PDTX mouse models (Example 1.2.1). As described above, it was observed that there was significant correlation (r=0.96) between degrees of gefitinib-induced tumor growth inhibition and EGFR activation scores.

(143) FIG. 22 is a graph showing numbers of EGFR PPI complexes per unit concentration of each of tissue lysate samples obtained from PDTX mouse models (Example 1.2.1) before and after gefitinib (50 mg/kg) injection. Significantly reduced numbers of EGFR PPI complexes were counted in the tissues after gefitinib injection, indicating that gefitinib induces EGFR signaling suppression.

Example 9

(144) 9.1. Preparation of Antibody and Reagent

(145) In order to pull down respective proteins, the following antibodies were employed: anti-EGFR antibody (MS-378-B0 ThermoFisher), anti-MET antibody (ab89297 Abcam), anti-HER2 antibody (BMS120BT ThermoFisher), anti-HER3 antibody (BAM348 R&D systems) mCherry (ab34771 Abcam), and anti-KRas antibody (sc-521 Santa Cruz).

(146) As respective detection antibodies for corresponding proteins and PTMs (post-translational modifications), the following antibodies were employed: anti-EGFR antibody (4267 Cell signaling), anti-EGFR (pTyr 1068) antibody (ab32430 Abcam), anti-EGFR (pTyr 1086) antibody (ab32086 Abcam), anti-EGFR (pTyr 1173) antibody (4407 Cell signaling), anti-MET antibody (8494 Cell signaling), anti-HER2 antibody (MA5-15050 ThermoFisher), anti-HER2 (pTyr1221/1222) antibody (2243 Cell signaling), anti-HER3 antibody (ab32121 Abcam), anti-HER3 (pTyr 1289) antibody (Cell signaling technology, cat. No. 4791), anti-Grb2 antibody (ab32037 Abcam), anti-Shc1 antibody (ab33770 Abcam), anti-Shc1 (pTyr 239/240) antibody (ab109455 Abcam), anti-HSP90 antibody (PA3-013 ThermoFisher), anti-MIG6 antibody (11630-1-AP Proteintech), anti-GAPDH antibody (3906 Cell signaling), and anti-c-Cbl antibody (2179 Cell signaling).

(147) Biotinylated anti-mouse immunoglobulin G (IgG) (405303 BioLegend) and Cy3-conjugated anti-rabbit IgG (111-165-046 Jackson ImmunoResearch) antibodies were used as secondary antibodies.

(148) Western blotting was conducted using the following antibodies: anti-EGFR (pTyr 1068) antibody (2234 Cell signaling), anti-EGFR antibody (2232 Cell signaling), anti-Erk (pThr202/Tyr204) antibody (9106 Cell signaling), anti-Erk antibody (4696 Cell signaling), anti-Akt antibody (4060 Cell signaling), anti-Akt antibody (4691 Cell signaling), anti-S6K (pSer235/236) antibody (4858 Cell signaling), anti-S6K antibody (2217 Cell signaling), and anti-actin antibody (ab8227 Abcam).

(149) EGFR was stimulated (3 min) using 100 ng/ml EGF (PHG0311L Life technologies).

(150) Gefitinib (S1025 Selleckchem), osimertinib (S7297 Selleckchem), BKM120 (S2247 Selleckchem), dabrafenib (S2807 Selleckchem), and trastuzumab (A1046 BioVision) were used to measure PPI changes in lung adenocarcinoma cells and HER2-/HER3-PPI in breast cancer cells, cell viability based on MTT assay, and tumor growth in PDTX models.

(151) 9.2. Cell Culture

(152) All cell lines were cultured in an RPMI1640 medium (22400-105 Life technologies) supplemented with 10% (w/v) fetal bovine serum (26140-079 Life technologies), 10 μg/ml gentamicin (15710-063 Life technologies), 100 units/ml penicillin, and 100 μg/ml streptomycin (15140-122 Life technologies). PC9-GR (gefitinib-resistant cell line Accession No. CVCL_S706), HCC827-GR5 (gefitinib-resistant cell line Accession No. CVCL_V622), and HCC4006-ER (erlotinib-resistant cell line Accession No. CVCL_S746) cell lines were cultured in the presence of 100 nM gefitinib or erlotinib. All cell lines were cultured at 37° C. in a 5% CO.sub.2 atmosphere in a humidified incubator. The cultured cells were washed with cold phosphate buffered saline (PBS). Cells were rapidly collected using 1 ml of cold PBS and a scraper (90020 SPL Life Science). A cell suspension obtained from one petri dish (diameter 100 mm) was divided into 3-4 aliquots. These aliquots were centrifuged at 4° C. and 3,000×g for 5 min. The supernatant was discarded and the pellet was stored at −80° C. until use.

(153) 9.3. Construction of eGFP-Labeled Prey Protein and Transfections

(154) Rat PLCγ.sub.SH2 cDNA including a tandem SH2 domain (amino acids 542 to 765 of NM_013187.1) was isolated directly from a Rat cDNA library using BglII and EcoRI. cDNAs of Grb2 (human Grb2; Addgene 46442), p85α (mouse p85α Addgene 1399), Shc1 (human Shc1, Addgene 73255), Eat2 (human Eat2, Addgene 46423), APCS (human APCS, Addgene 46477), Nck1 (human Nck1, Addgene 45903), and SOS1 (human SOS1, Addgene 32920) were excised using restriction enzymes corresponding to restriction sites in their respective plasmids. eGFP-tagged CARM1 (human CARM1) and EGFR genes were provided by Seoul National University (Korea) and KAIST (Korea), respectively. All the cDNAs were cloned into pEGFP-C1 (Clontech Laboratories) to construct corresponding eGFP-labeled prey proteins. W36K, R86M, and W193K point mutations were introduced into a Grb2 gene to afford respective Grb2 mutants N*-, SH2*-, and C*-construct. An EGFR mutant was constructed by deleting E746-A750 from an EGFR gene or substituting a lysine at position 858 with arginine on an EGFR gene.

(155) The plasmids obtained above were introduced into HEK293 cells by electroporation using Neon transfection system (MPK5000 Life technologies) according to the manufacturer's instruction. For this, 30 μg of a plasmid DNA was mixed with 100 μl of a HEK293 cell suspension containing about 2×10.sup.6 cells. Two 950V electric pulses were applied to the HEK293 cells (with a duration of 35 ms for each pulse). Twenty four hours after transfection, transfected cells were harvested and stored at −80° C.

(156) 9.4. Lung Cancer Patient-Derived Tumor Xenograft Model

(157) All animal studies were conducted according to the guidelines set forth by the Institutional Animal Care and Use Committee (IACUC). Female severe combined immunodeficient mice (NOG) and nude mice (nu/nu mice; Orient Bio), both 6 to 8 weeks old, were prepared. A clinical tumor sample (obtained from a lung adenocarcinoma patient or a lung squamous cell carcinoma (SQCC) patient) was cut into a fragment of about 3 mm in size and subcutaneously implanted to the flank of each of the NOG mice. One to four months after the implantation, a tumor was observed in the implanted region. Tumor growth rates in the subcutaneous tissue were obtained by measuring tumor sizes twice a week with calipers. When grown to have a diameter of about 1.5 cm, the tumor tissue was excised and sectioned into small fragments (each 5 mm.sup.3 in volume). The sectioned tissue was reimplanted into different mouse groups to acquire subsequent tumors. The mice that retained the patient-derived tumor were designated F0 and subsequent generations having subsequent tumors derived from F0 were designated FI, F2, F3, and the like, sequentially (see FIG. 23A). The 3.sup.rd generation mice (F3) were used in treatment with a vehicle (PBS), osimertinib, or gefitinib.

(158) Of the patient-derived tumor xenografts (PDTXs) thus obtained, the mice (F3; n=3) engrafted with the tumor derived from a lung adenocarcinoma patient were designated PDTX-A1, PDTX-A2, and PDTX-A3 and the mice (F3; n=5) engrafted with the tumor derived from a lung SQCC patient were designated with PDTX-S1, PDTX-52, PDTX-S3, PDTX-S4, and PDTX-S5, respectively.

(159) Osimertinib and gefitinib or a vehicle were intraperitoneally injected once a day at respective doses of 5 and 50 mg/kg into the patient-derived tumor xenografts (PDTXs). Fifteen days after drug injection, tumor tissues were excised from the PDTXs and monitored for PPI and expression level change.

(160) 9.5. Single-Molecule Co-IP and Immunolabeling Imaging

(161) For a detailed protocol of single-molecule co-IP and immunolabeling imaging, reference was made to “Lee, H. W. et al. Real-time single-molecule coimmunoprecipitation of weak protein-protein interactions. Nat. Protoc. 8, 2045-2060, (2013)”. NeutrAvidin (10 μl of 0.1 mg/ml; A2666 Life technologies) was put in individual reaction chambers. After 10 min of incubation, uncoupled NeutrAvidin was removed. A miniaturized imaging chamber was immersed in a PBS-filled reservoir and completely washed by shaking 100 times in lateral directions. After complete removal of PBS, biotinylated pull-down antibodies were incubated for 10 min on the NeutrAvidin-coated surface to form a layer. For a MET antibody, a biotinylated secondary antibody (alpha-mouse IgG) was used to recognize the primary antibody. After the chamber was washed with PBS, a cancer cell or tumor tissue extract was applied to the antibody-coated surface. After 15 min, uncoupled extracts were removed and the chamber was immersed in a PBS-filled reservoir supplemented with 0.05% (v/v) Tween 20.

(162) For single-molecule co-IP imaging, a transformed HEK293 cell extract was diluted with a 30 nM eGFP-tagged probe protein and then loaded to an imaging chamber. The chamber was positioned on a TIRF microscope and eGFP florescence was recorded on EMCCD (20 frames; 100-ms exposure).

(163) In the single-molecule immunolabeling imaging, a dye-labeled detection antibody was used instead of the eGFP-labeled probe protein for five frames. To avoid overlap between the detection antibody and the pull-down antibody, selection was made of a detection antibody that has an epitope at a tyrosine residue on a cytoplasmic kinase region or tail. The detection antibody was labeled directly with Alexa488 (MET antibody) or indirectly with a Cy3-labeled secondary antibody (EGFR, HER2, HER3, and pTyr antibody). After recording fluorescence of 5 or 20 frames (0.1 sec exposure) in a TIFF stack, fluorescent spots were counted to measure numbers of single-molecule PPI complexes or immunolabeled proteins. A mean value and standard deviation of single-molecule counts was obtained from 10 different positions within the same reaction chamber.

(164) 9.6. PPI Complex and Immunolabeled Protein Counting

(165) TIFF files obtained by fluorescence imaging were analyzed using a custom GUI (written in MATLAB (MATLAB 2016a, MathWorks)). From three frames (17-19 for eGFP9 and 3-5 for Cy3 and Alexa488), local maxima having an intensity representative of single PPI complexes or immunolabeled proteins were identified. For background correction, an image obtained by spatial median-filtering (11×11 pixel) was subtracted from an original image according to frames. The images thus obtained were averaged and subjected to thresholding before use in detecting local maxima (with custom MATLAB GUI).

Example 10: Prediction of Responsiveness of PDTX to EGFR-Targeted Inhibitor

(166) 10.1. Prediction of Responsiveness of Lung Adenocarcinoma Xenograft PDTX to Osimertinib

(167) It was ascertained that PPI metrics of the HER-family receptors are tightly correlated with the drug responsiveness of cancers and examination was made to determine whether or not single-molecule immunolabeling or co-IP analysis is applicable to the screening of certain cancer that has responsiveness to HER-family receptor-targeted therapy (that is, on which HER-family receptor-targeted therapy shows an anticancer effect). To this end, lung adenocarcinoma patient-derived tumor xenograft mice (PDTXs: PDTX-A1-PDTX-A3 of FIG. 23) were created.

(168) These lung adenocarcinoma PDTXs were observed to have activation mutation in the EGFR gene (exon 19 or L858R mutation).

(169) After 30 days of treatment of the lung adenocarcinoma PDTXs (PDTX-A1-PDTX-A3; 3 or more mice each, the following results are represented by average values) with osimertinib (5 mg per 1 kg of weight daily), tumor sizes were measured and compared with a control (vehicle administered). The results are depicted in the left panels of FIG. 23b. As is understood from data of the left panels of FIG. 23b, PDTX-A1-PDTX-A3 showed a significant reduction of tumor size by treatment with osimertinib (A1>A2>A3).

(170) In addition, PPI complexes between each of EGFR, HER2, HER3, and MET receptors and each of the downstream proteins PLCgammaSH2, Grb2, and p85-alpha in each PDTX (PDTX-A1-PDTX-A3) were counted, and the results are depicted in the left panels of FIG. 23c (expressed as PPI count in FIG. 23c). As shown in FIG. 23c, PPI complex counts between EGFR and three different downstream proteins were in the order of A1>A2>A3, coinciding with the behavior of tumor size reduction by treatment with the EGFR inhibitor, osimertinib. This result indicates that PPI complex counts between a target protein and a downstream protein in lung adenocarcinoma PDTX models exhibit significant correlation with an anticancer effect of a therapeutic agent targeting the target protein.

(171) In addition, expression levels of EGFR and other receptors (MET, HER2, and HER3) in 8 PDTX (A1-A3 and S1-S5) individuals were measured, and the expression level of each of the receptors was normalized to that of a control (EGFR: A549, MET: HCC827-GR5, and HER2 and HER3: SKBR3). The results are depicted in FIGS. 23d and 25a-c.

(172) As shown in FIG. 23c, the PDTX models (A1-A3) did not exhibit significant PPI complex counts for MET, HER2, or HER3 receptors, but exhibited somewhat significant PPI complex counts for EGFR. This result implies that PDTXs exhibit oncogene addiction to EGFR signaling at protein and PPI levels.

(173) Next, as proven by normalized PPI counts (numbers of PPI complexes per unit concentration of first protein; corresponding to activation scores) in lung adenocarcinoma cell lines, examination was made to determine whether the responsiveness of PDTX to osimertinib therapy can be predicted. In each of the PDTX models (A1-A3) treated for 30 days with osimertinib (5 mg per kg of weight daily), tumor growth inhibition rates (tumor growth inhibition (%)=[(ΔV.sub.vehicle−ΔV.sub.gefitinib/|ΔV.sub.vehicle]×100; ΔV.sub.vehicle: difference in tumor volume between post- and pre-treatment with a vehicle; ΔV.sub.gefitinib: difference in tumor volume between post- and pre-treatment with gefitinib) on the Y-axis were plotted versus PPI sum/EGFR levels (PPI sum: PPI score, PPI sum/EGFR level: Activation score) on the X-axis, and the results are depicted in FIG. 23e. As can be seen in FIG. 23e, normalized PPI counts (PPI sum/EGFR level; activation score) were observed to have high correlation with the tumor growth inhibition of osimertinib (r=1).

(174) 10.2. Prediction of Responsiveness to Osimertinib in Lung Squamous Cell Carcinoma (SQCC) Xenograft PDTX

(175) As much as 29% of lung adenocarcinoma cases are related to the sensitizing mutation of EGFR, whereas only 0.5% of lung SQCC cases have a sensitizing mutation. Accordingly, there are currently no suitable biomarkers for EGFR-targeted therapy for lung SQCC.

(176) In this Example, five lung SQCC PDTXs (PDTX-S1-PDTX-S5) were created and subjected to single-molecule immunolabeling and co-IP profiling (Example 9.5). All the five PDTXs (PDTX-S1-PDTX-S5) were found to exhibit minimal levels of MET, HER2, and HER3 receptor proteins and PPI complexes related thereto (the right panels of FIG. 23c and FIGS. 25a-c).

(177) Meanwhile, total EGFR counts (EGFR level) and EGFR PPI complex counts were detected at significant levels (FIGS. 23c and 23d, and FIGS. 25d and 25e).

(178) These results are deemed to result from the fact that lung SQCC often depends on EGFR with respect to proliferation signaling.

(179) 10.3. Prediction of Responsiveness to Gefitinib in Lung Squamous Cell Carcinoma (SQCC) in PDTX

(180) After 15 days of the treatment of five PDTXs (PDTX-S1-PDTX-S5) with gefitinib, tumor growth inhibition rates were measured and are depicted in the right panels of FIG. 23b. Of PDTX-S1-PDTX-S5, S1 and S2 were identified to show significant tumor inhibition effects. In the various PDTX individuals tested, the PPI complex count normalized to an EGFK level (activation score) was identified again to have very high correlation (Spearman correlation of 0.9) with tumor growth inhibition (FIG. 23f and FIG. 25f-h).

(181) The data obtained above suggest that the normalized PPI count is tightly correlated with the responsiveness of non-small cell lung cancer to an EGFR-targeted inhibitor. To understand that a normalized PPI count other than a sum of PPI counts has high correlation with responsiveness to an anti-tumor drug, two lung SQCC PDTXs (PDTX-S1 and PDTX-S2) which had shown characteristic signaling phenotypes and gefitinib responsiveness were particularly observed.

(182) PPI complex counts were measured in PDTX-S1 and PDTX-S2 before gefitinib treatment and 15 days after gefitinib treatment, and are depicted in FIG. 23g and FIG. 26a-b.

(183) As is understood from the data, PDTX-S1 retained a detectable level of EGFR PPI complex counts particularly with the regulatory p85a subunit of PI3K. In contrast, PDTX-S2 showed EGFK PPI complex counts that were reduced or could not be discriminated, compared with a negative control using A549 cells (FIGS. 23g and 26). Accordingly, the gefitinib dose (50 mg/kg) used in the experiments completely regulated a hyperactive but smaller pool of EGFRs in PDTX-S2 to induce the shrinkage of corresponding cancers. In contrast, the same gefitinib dose could not regulate EGFK activity, but allowed significant EGFK overexpression in PDTX-S1. The results account for the reason why PDTX-S2 with a normalized PPI count exhibits a more outstanding responsiveness to gefitinib than PDTX-S1 with a high EGFK level and total PPI count.

(184) Based on the result that PDTX-S1 shows increased EGFR-p85α interaction, PDTX-S1 was treated with the PI3K inhibitor BKM120 (50 mg/kg) (FIG. 23h).

(185) As is understood from the data, BKM120 exhibited a more potent tumor growth inhibition effect at the same dose (50 mg/kg) than gefitinib. A combination therapy of gefitinib and BKM120 (gefitinib (50 mpk (mg per 1 kg weight)/BKM120 (50 mpk)) elicited tumor shrinkage. This result suggests that the single-molecule co-IP profiling, which employs different downstream proteins to examine PPI, can find useful applications in selecting target signaling pathways and designing a combination therapy strategy of two or more drugs.

(186) 10.4. Prediction of Responsiveness in Non-Small Cell Cancer Xenograft PDTX

(187) Normalized PPI complex counts obtained from two kinds of non-small cell lung cancer xenograft models PDTX-A1-A3 and PDTX-S1-S5 were pulled down and compared in single plots (FIG. 23i).

(188) Despite a difference in genetic alteration and cancer subtype, the data points exhibit a uniform pattern, making a fair coincidence with the tumor growth inhibition with a Spearman correlation of 0.95.

(189) The data suggests that the normalized PPI complex count can be a gauge of EGFR signal strength and thus can act as an efficacy prediction marker for EGFR-targeted therapies.

Example 11. Single-Molecule Immunolabeling and Co-IP Profiling of Human Patient Sample)

(190) Using a micro-chamber and a high-throughput single-molecule imaging system proposed in the present invention, tumor tissues of human patients were characterized (FIG. 24).

(191) A typical cryogenic lysis protocol developed for PDTX specimens was applied to two lung adenocarcinoma patient tissues surgically excised from lung adenocarcinoma patients (Yonsei University Severance Hospital) (P1 and P2 in FIG. 24a).

(192) Briefly, the prepared patient tissues were homogenized (about 0.6 cm) and immersed in liquid nitrogen. After further homogenization, PBS was added for complete dissolution. Centrifugation afforded a pellet which was then incubated at 4° C. in PBS while continuously mixing. After centrifugation, the supernatant was taken.

(193) For each of the lysates obtained respectively from the tissues with a size of 15 mm.sup.3 (P1) and 18 mm.sup.3 (P2) in the above-mentioned manner, 10 different PPI levels (PPI complex values in P1 and P2 relative to the PPI complex of 1.0 in a positive control) and 10 different protein and PTM (post-translational modifications) levels (expression) were measured (see Immunolabeling of Example 9.5) (FIG. 24b). As positive controls, use was made of PC9 cells for EGFR, HCC827 cells for MET, and SKBR3 cells for HER2 and HER3.

(194) Although both the tumor tissues (P1 and P2) had an exon19 mutation (exon19 deletion mutation) in the EGFR gene, only sample P1 showed significant EGFR PPI complex counts (FIGS. 24b and c and Table 4).

(195) TABLE-US-00004 TABLE 4 Biopsy Deposits EGFR type date genotype Treatment Response Patient Surgical 2013 Δexon19 Gefitinib PR: 2014 Nov. P1 resection Dec. 27 25~2016 Mar. 28 PD: 2016 Jul. 19 Patient 2014 Erlotinib SD: 2014 Apr. P2 Feb. 11 16~2015 May 15 PD: 2015 May 21 (PR = Partial response, PD = Progressive disease)

(196) With respect to other receptors, namely HER receptors and MET receptors, significant PPI counts were observed in neither of the two samples (FIGS. 24b and c).

(197) Patient P1 maintained a partial response (PR: according to response evaluation criteria in solid tumors (RECIST)) for about 1.5 years before progressive disease (PD) designation, whereas Patient P2 maintained stable disease (SD) for one year before PD designation.

(198) Lastly, PDTX-A1, which showed the highest activity of an EGFR signal, and the mutant EGFR (exon 19 deletion) derived from the human patient sample P1 were subjected to in-vitro dephosphorylation by treatment with PIPN1 (FIG. 24d).

(199) After dephosphorylation, the binding of eGFP-labeled PLCgamma.sub.SH2 and Grb2 to the mutant EGFR complex was measured and the measurements are depicted in FIG. 24d. As shown in FIG. 24d, the dephosphorylation (+PTPN1) almost completely stopped the binding of PLCgamma.sub.SH2, which reliess fully on the pTyr-SH2 domain interaction, whereas Grb2 binding counts were maintained even after 50% or higher and 80% or higher dephosphorylation. The data suggests that the pTyr-independent signaling mechanism of mutant EGFR works in practice in the PDTX models and surgical tumors.

Example 12: Test for Effect of Adhesive-Application Modality

(200) A multi-well A in which an adhesive E (UV epoxy) was applied to the outer brim of a contact surface between a receiving part 153 and a substrate 300 (test group) and a multi-well B in which an adhesive (UV epoxy) is applied to the entire contact surface between a receiving part 153 and a substrate 300 (comparative group) were each fabricated and the wells were numbered as shown in FIG. 38:

(201) All the wells were surface treated with biotin, with no antibodies immobilized thereto.

(202) A green fluorescent protein (GFP, Clontech)-tagged Grb2 protein (NP_002077.1) was added in an amount of 30 nM to each well, incubated at room temperature (23-27° C.) for 5-10 min, and washed with a washing solution containing PBS (with 0.05% (v/v) Tween 20). GFP counts left in the wells were measured by imaging to quantitate the Grb2 protein. In this regard, a washing solution containing a low concentration (e.g., about 0.1% (v/v) or less) of a non-ionic surfactant (e.g., Tween20, Triton x-100, etc.) is preferred (hereinafter, the same is applicable to protein quantitation). GFP counts were measured by counting fluorescent spots on fluorescence images obtained using an EMCCD (Electron-multiplying charge-coupled device; Andor iXon Ultra 897 EX2 (DU-897U-CS0-EXF)) camera with an exposure time of 0.1 sec per frame and an EMCCD gain value of 40.

(203) Measurements of the GFP counts are depicted in FIG. 32 (multi-well A) and FIG. 33 (multi-well B). In FIGS. 32 and 33, each # number means the position of a well given the corresponding number.

(204) For multi-well A, as shown in FIG. 32, GFP counts that remained in the wells after washing appeared at low levels similarly over the entire well. In contrast, GFP counts that remained in the wells of the multi-well B after washing were measured to be about 15- to 20-fold greater in some wells (#5, #6, and #7) than the other wells.

(205) Because antibodies were not immobilized to the wells, the GFP that remained in the wells after washing were deemed to be attributed to non-specific binding. Therefore, the results of FIGS. 32 and 33 show that the multi-well A in which an adhesive is applied to the outer brim of the receiving part allows the non-specific binding of proteins at a far lower level than the multi-well B in which an adhesive is applied to the entire contact surface between the receiving part and the substrate. A high non-specific binding level of proteins in the multi-well B results from the fact that a part of the adhesive (epoxy) applied to the entire contact surface between the receiving part and the substrate infiltrates into the through-hole so that the adhesiveness of the epoxy exerts an unintended influence to the protein-protein reaction to induce non-specific binding. On the other hand, the adhesive that is applied to the outer brim of the receiving part in the multi-well A cannot infiltrate into the through-hole or only a very small amount of the adhesive infiltrates into the through-hole. Thus, there are no or very slight effects on the protein-protein interaction, thereby resulting in a very low non-specific binding level of proteins.

(206) In order to further identify the induction of specific protein reactions in the multi-well A, a GFP antibody was not immobilized in three wells #1, #5, and #9, but in the remaining nine wells #2, #3, #4, #6, #7, #8, #10, #11, and #12 the above experiment was conducted again to measure GFP counts.

(207) Measurements of GFP counts obtained above are depicted in FIG. 34. As shown in FIG. 34, GFP counts in the nine wells #2, #3, #4, #6, #7, #8, #10, #11, and #12 where a GFP antibody was not immobilized were remarkably higher than those in the three wells #1, #5, and #9 where a GFP antibody was not immobilized. The result identifies again that specific protein interactions are induced in the multi-well A where the adhesive is applied to the outer brim of the receiving part.

Example 13: Protein-Protein Interaction Measurement

(208) A multi-well A was prepared in the same manner as in Example 12 and an anti-EGFR antibody (MS-378-B0, Thermo) was immobilized to the surface of the multi-well A. To each antibody-immobilized well was added a cell sample that was obtained by lysing a cell line having a high expression level of EGFR (H1666; ATCC, #CRL-5885) in a cell lysis buffer (50 mM Tris-HCl (pH 7.4), 1% (v/v) Triton X-100, 150 mM NaCl, 1 mM EDTA, 10% (v/v) glycerol, protease inhibitor cocktail (Sigma, P8340) 100×, tyrosine phosphatase inhibitor cocktail (Sigma, P5726) 100×), followed by incubation at room temperature (23-27° C.) for 15 mM and then by washing with a washing solution containing PBS (with 0.05% (v/v) Tween 20) to capture EGFR on the well surface. GFP-tagged p85-a protein (NP_852664.1) was added to each well having EGFR captured on the surface thereof before imaging.

(209) GFP counts left in the wells were measured with reference to the method of Example 12 to examine protein-protein interaction between EGFR and p85-a.

(210) GFP counts (mean values of five measurements) are plotted against amounts of the cell sample in FIG. 35. GFP counts (mean values of five measurements) and standard deviations are given in Table 5:

(211) TABLE-US-00005 TABLE 5 Cell sample amount (mg/ml) GFP count (Mean) Standard deviation 0 87.5 16.90858 0.03 604.8 27.67129 0.05 974.8 34.54273 0.08 1348 68.6343 0.1087.5 1651.8 61.36937

(212) As shown in FIG. 35 and Table 5, when the multi-well A was used, protein-protein interactions were explicitly measured to have a linear proportional correlation with cell sample concentrations, and the measurements obtained through repeated experiments were observed to have relatively low levels of standard derivations.

(213) Two multi-wells A were prepared in the same manner as in Example 12. Of the two, one was not coated with a GFP antibody (blank) whereas the other was coated (GRB2-GFP). To each well, 100 pM GFP-GRb2 was added and then signals were measured in the same manner as in Example 1.

(214) The results obtained above are depicted in FIGS. 36 and 37. In the multi-well A, as is understood from the data of FIGS. 36 and 37, almost no signals were detected when the well were not coated an antibody for capturing a target protein (GFP) whereas signals were very well detected with low standard deviations (GRB2-GFP) when the antibody was applied. These results indicate that the use of the multi-well A is almost not accompanied by a false positive result.