CN116590375A - Method for screening drugs for tumor diseases, program product, high-content imaging analysis system, method for analyzing phase separation - Google Patents

Abstract

The application relates to a method for screening medicines for tumor diseases, a program product, a high content imaging analysis system and a method for analyzing phase separation conditions. The method includes seeding cells comprising a target fusion protein associated with a tumor condition of interest, comprising PS and DBD and comprising a marker molecule, into an aperture plate comprising an array of apertures, forming an aggregate that is recognizable by imaging. Partitioning a plurality of small molecule compounds into individual wells; with a microscopic imaging system, each well is continuously photographed in several different fields of view for a predetermined period of time to generate time-series images of the several fields of view. The attenuation of the amount of intracellular aggregates in each well was analyzed accordingly. Screening out the pores of which the attenuation condition of the amount of the intracellular aggregates meets the predetermined condition, and using the correspondingly distributed small molecule compounds as candidate drugs for the target tumor disorder. Therefore, the dynamic action of a large number of different small molecules relative to the aggregate can be continuously and comprehensively monitored in a high flux, and candidate medicines can be efficiently and accurately screened.

Description

Method for screening drugs for tumor diseases, program products, high-content imaging analysis system, method for analyzing phase separation conditions

Technical Field

The present application relates to the field of drug screening, and more specifically, to a method for screening drugs for tumor diseases based on cell microscopic imaging, a program product, a high-content imaging analysis system, and a method for analyzing phase separation conditions.

Background Art

Chromosomal rearrangement is a mutational event caused by DNA breaking and rejoining at the wrong location, which sometimes produces fusion proteins. Many fusion proteins, such as BCR-ABL found in chronic myeloid leukemia, can lead to tumorigenesis. In addition, some fusion proteins associated with carcinogenesis are composed of a DNA binding domain (DBD) from one protein and a phase separation prone domain (PS) from another protein (which can be simply referred to as PS-DBD fusion proteins). The fusion between the phase separation prone structure and the DNA binding domain may lead to a "super transcription factor" that recruits components of the transcription factory through phase separation that is not regulated by upstream signals.

In previous studies, the phase separation ability of PS-DBD fusion proteins has occasionally been mentioned, such as the EWS-FLI1 fusion protein in Ewing sarcoma and the NUP98-HOXA9 fusion protein in acute myeloid leukemia (AML). The FET protein family includes three members, FUS, EWS, and TAF15. They all contain a conserved N-terminal IDR that can drive liquid-liquid phase separation (LLPS). The IDR of FET proteins is fused to the DBD of the ETS (E26 transformation sequence) family transcription factors to form FET-ETS fusion proteins. Such fusion proteins can form biomolecular aggregates to recruit RNA polymerase II, resulting in abnormal gene expression and increased carcinogenicity.

Previous studies have speculated that some small molecules may increase the fluidity of biomolecular aggregates to alleviate diseases (Patel, A. et al. A liquid-to-solid phase transition of the ALS protein FUS accelerated by disease 715 mutation. Cell 162, 1066-1077 (2015)). However, it is currently impossible to accurately monitor the effects of small molecules on aggregates, and there is a lack of methods for efficiently screening candidate drugs for target tumor diseases from a large number of small molecules. This application is proposed to solve the above problems in the prior art.

Summary of the invention

What is needed is a method for screening drugs for tumor diseases, a program product, a high-content imaging analysis system, and a method for analyzing phase separation conditions, which can make aggregates formed by fusion proteins available for imaging identification, and conduct continuous and comprehensive high-throughput monitoring of the dynamic effects of a large number of different small molecules on the aggregates, so as to efficiently and accurately screen candidate drugs for target tumor diseases.

According to the first scheme of the present application, a method for screening drugs for tumor conditions is provided. The method comprises the following steps. Cells containing a target fusion protein are inoculated into a well plate comprising a well array, wherein the cells contain a nucleic acid molecule encoding a target fusion protein, the target fusion protein is associated with a target tumor condition, contains a phase separation tendency domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to form aggregates that can be imaged and identified. A variety of small molecule compounds are distributed to each well. Using a microscopic imaging system, each well is continuously photographed in a predetermined time period with several different fields of view to generate time series images of several fields of view of each well. The processor analyzes the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well. The processor screens out the wells whose attenuation of the amount of intracellular aggregates meets the predetermined conditions, and uses the small molecule compounds distributed to the screened wells as candidate drugs for the target tumor condition.

According to the second scheme of the present application, a computer program product is provided, which includes computer executable instructions that can be stored and/or downloaded to a non-transitory storage medium. When the computer executable instructions are executed by a processor, a method for screening drugs for tumor conditions is performed, comprising the following steps. Time series images of several fields of view of each well of a well plate including a well array are obtained. The time series images are generated by continuously photographing each well in a predetermined time period with several different fields of view after performing the following steps: cells containing a target fusion protein are inoculated into a well plate including a well array, wherein the cells contain a nucleic acid molecule encoding a target fusion protein, the target fusion protein is associated with a target tumor condition, contains a phase separation tendency domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to form aggregates that can be imaged and identified; a plurality of small molecule compounds are distributed to each well. Then, based on the time series images of at least one field of view of each well, the amount of intracellular aggregates in each well is analyzed. Attenuation conditions. Wells whose attenuation conditions of the amount of intracellular aggregates meet predetermined conditions are screened, and small molecule compounds assigned to the screened wells are determined as candidate drugs for the target tumor disorder.

According to the third scheme of the present application, a high-content imaging analysis system is provided, comprising an automatic high-speed microscopic imaging component and at least one processor. The automatic high-speed microscopic imaging component is configured to continuously shoot each well of a well plate including a well array with several different fields of view within a predetermined time period after the following steps are performed, and generate time series images of several fields of view of each well: inoculating cells containing a target fusion protein into a well plate including a well array, wherein the cells contain a nucleic acid molecule encoding a target fusion protein, the target fusion protein is associated with a target tumor condition, contains a phase separation tendency domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to form aggregates that can be imaged and identified; a plurality of small molecule compounds are distributed into each well. The at least one processor is configured to execute a method for screening drugs for tumor conditions according to various embodiments of the present application.

According to the fourth scheme of the present application, a method for analyzing the phase separation state of a target fusion protein comprising a phase separation tendency domain (PS) and a DNA binding domain (DBD) is provided, comprising the following steps. An image of a cell comprising the target fusion protein is collected, wherein the cell comprises a nucleic acid molecule encoding the target fusion protein, and the target fusion protein further comprises a marker molecule so as to form aggregates that can be imaged and identified. The amount of intracellular aggregates in the image is identified, and the phase separation state of the PS-DBD fusion protein in vivo and/or in vitro is determined based on the amount of aggregates identified, and the larger the amount of aggregates, the more severe the phase separation state.

The present application comprehensively investigated oncogenic fusion events and found more than 1,500 potential PS-DBD fusion proteins that can be used as targets for diagnosing diseases associated with abnormal gene expression (e.g., tumor conditions), and specifically identified hundreds of new cancer targets. In addition, the present application also established a method for screening drugs for tumor conditions based on computer vision of a microscopic imaging system (also referred to herein as DropScan), a program product, a high-content imaging analysis system, and a method for analyzing phase separation conditions. It enables the aggregates formed by the fusion protein to be imaged and identified, and a large number of different small molecules are continuously and comprehensively monitored for dynamic attenuation regulation of the aggregates, so as to efficiently and accurately screen candidate drugs for the target tumor condition, and the candidate drugs are used as lead candidates for clinical treatment drugs for tumor conditions. The present application successfully used DropScan to identify some small molecules that can partially or completely dissolve aggregates formed by FET-ETS cancer fusion proteins. In particular, LY2835219 reduces the number of aggregates by upregulating lysosomal activity, and then rescues the related abnormal gene expression.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings, which are not necessarily drawn to scale, the same reference numerals may describe similar components in different views. The same reference numerals with letter suffixes or different letter suffixes may represent different instances of similar components. The accompanying drawings generally illustrate various embodiments by way of example and not limitation, and together with the description and claims, serve to illustrate the disclosed embodiments. Such embodiments are illustrative and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.

FIG. 1 shows a flow chart of a method for screening drugs for tumor diseases according to a first embodiment of the present application.

FIG. 2 shows a schematic diagram of a process for obtaining more than 1,500 potential PS-DBD fusion proteins from multiple databases according to the second embodiment of the present application.

FIG3 is a schematic diagram showing the interconnectedness of PS-DBD fusions according to the third embodiment of the present application, wherein a fusion network of a central protein with more than 10 fusion partners is shown.

FIG4( a ) shows a schematic diagram of inoculating cells containing the target fusion protein, U2OS/mCherry-FUS-ERGmut cells, into a well plate for continuous imaging in a method for screening drugs for tumor diseases according to the fourth embodiment of the present application.

FIG4( b ) shows a schematic diagram of performing segmentation of U2OS/mCherry-FUS-ERGmut cells and aggregates on each image in the method for screening drugs for tumor diseases according to the fifth embodiment of the present application.

FIG4( c ) shows a graphical representation of an exponential model fitted based on the distribution curve of the fraction of aggregate-rich cells in each well within the predetermined time period in the method for screening drugs for tumor diseases according to the sixth embodiment of the present application.

FIG4( d ) shows a scatter plot of drug effects calculated for 1777 small molecules and R-squared as a fitting effect parameter in the method for screening drugs for tumor diseases according to the seventh embodiment of the present application.

FIG. 4( e ) shows a ranking diagram of the drug effects of 144 screened positive small molecules in the method for screening drugs for tumor diseases according to the seventh embodiment of the present application.

FIG. 5( a) to FIG. 5( m) are schematic diagrams showing the FUS-ERG fusion protein driving phase separation in vivo and in vitro according to an embodiment of the present application, wherein:

FIG5( a ) shows a schematic diagram (top) and disorder prediction (bottom) of FUS, ERG, and FUS-ERG fusion protein, with the rectangle representing the DBD in ERG;

FIG5( b ) shows a live cell image of Hela cells heterologously expressing mCherry-FUS-ERG, DNA was stained with Hoechst33342 (blue), scale bar, 10 μm;

FIG5( c ) shows a schematic diagram of the OptoIDR method, wherein the OptoIDR element is composed of the target IDR, mCherry, and Cry2olig (which is a mutant form of the Arabidopsis CRY2 PHR domain);

FIG5( d ) shows representative images of U2OS living cells expressing mCherry-Cry2olig (top) and FUS IDR-mCherry-Cry2olig (bottom) fusion proteins, with cells stimulated with 488 nm laser;

Figure 5(e) shows quantification of droplet formation kinetics after stimulation as in Figure 5(d), with n ≥ 30 cells per construct;

FIG5( f ) shows confocal fluorescence images of EGFP-FUS-ERG recombinant protein at specified concentrations, scale bar, 10 μm;

Fig. 5(g) shows the fusion of aggregates formed by EGFP-FUS-ERG (40 μM), scale bar, 2 μm;

Figure 5 (h) shows a representative fluorescence image of FRAP of aggregates formed by EGFP-FUS-ERG (40 μM), scale bar, 2 μm;

FIG5( i ) shows the quantitative analysis of FRAP data of aggregates formed by EGFP-FUS-ERG (40 μM) (n=9, mean±standard deviation);

Fig. 5(j) shows the imaging results of 40 μM EGFP-FUS-ERG mixed with 50 nM 306-bp dsDNA, 306-bp dsDNA containing 25×GGAA and labeled with Cy5, scale bar, 10 μm;

Figure 5(k) shows a representative fluorescence image of FRAP of a polymer formed by EGFP-FUS-ERG (40 μM) and 306-bp dsDNA (50 nM, 25×GGAA, labeled with Cy5), scale bar, 2 μm;

FIG5( l ) shows the quantitative analysis of the data of the condensate formed by EGFP-FUS-ERG (40 μM) and 306-bp dsDNA (50 nM, 25×GGAA, labeled with Cy5) (n=6, mean±SD);

Figure 5(m) shows live cell images of Hela cells heterologously expressing the DNA binding domain mutant form of mCherry-FUS-ERG (Figure 5(b)), mCherry-FUS-ERGmut. DNA was stained with Hoechst 33342 (blue), scale bar, 10 μm.

FIG. 6( a) to FIG. 6( c) show the analysis diagrams of in vitro purified DNA and FUS-ERG according to the eighth embodiment of the present application, wherein:

FIG6( a) shows SDS-PAGE analysis of MBP-EGFP-FUS-ERG purified from Escherichia coli according to the eighth embodiment of the present application;

FIG6( b ) shows a schematic diagram of agarose electrophoresis analysis of purified 306-bp 25×GGAA dsDNA (labeled with cy5) according to the eighth embodiment of the present application.

FIG6( c ) shows the sequence of 306-bp dsDNA according to the eighth embodiment of the present application, in which 25×GGAA bases are highlighted.

FIG. 7( a) to FIG. 7( y ) show schematic diagrams of BRD9 IDR/COL17A1 IDR/MLLT1IDR/EGFP-SFPQ IDR/EGFP-MED15 IDR fusion protein driving phase separation in vivo and in vitro according to the ninth embodiment of the present application, wherein:

Figure 7 (a) shows representative images of U2OS cells expressing the indicated BRD9 IDR-mCherry-Cry2olig fusion proteins, and the cells were stimulated with a 488 nm laser to activate OptoIDR, scale bar, 20 μm;

Figure 7(b) shows quantification of droplet formation kinetics after stimulation as in Figure 7(a), with n ≥ 30 cells per construct;

FIG7( c ) shows confocal fluorescence images of Alexa 488-labeled BRD9 IDR protein at the indicated concentrations, scale bar, 10 μm;

FIG7( d ) shows FRAP of aggregates formed by Alexa 488-labeled BRD9 IDR protein at the following concentrations (BRD9 IDR (150 μM), scale bar, 2 μm;

FIG7( e ) shows the quantitative analysis of the FRAP data from FIG7( d ) (n=6-12, mean±standard deviation);

Figure 7(f) shows representative images of U2OS cells expressing the indicated COL17A1 IDR-mCherry-Cry2olig fusion proteins, and the cells were stimulated with 488 nm laser to activate OptoIDR, scale bar, 20 μm;

Figure 7(g) shows quantification of droplet formation kinetics after stimulation as in Figure 7(f), with n ≥ 30 cells per construct;

FIG7( h ) shows confocal fluorescence images of Alexa 488-labeled COL17A1 IDR protein at the indicated concentrations, scale bar, 10 μm;

FIG. 7( i ) shows FRAP showing aggregates formed by Alexa 488-labeled COL17A1 IDR protein at the following concentrations (COL17A1 IDR (100 μM), scale bar, 2 μm;

FIG7( j ) shows the quantitative analysis of the FRAP data from FIG7( i ) (n=6-12, mean±standard deviation);

Figure 7(k) shows representative images of U2OS cells expressing the indicated MLLT1 IDR-mCherry-Cry2olig fusion proteins, and the cells were stimulated with 488 nm laser to activate OptoIDR, scale bar, 20 μm;

Figure 7(l) shows quantification of droplet formation kinetics after stimulation as shown in Figure 7(k), with n≥30 cells per construct;

FIG7( m ) shows confocal fluorescence images of Alexa 488-labeled MLLT1 IDR protein at the indicated concentrations, scale bar, 10 μm;

FIG7( n ) shows FRAP of aggregates formed by Alexa 488-labeled MLLT1 IDR protein at the following concentrations (MLLT1 IDR (25 μM), scale bar, 2 μm;

FIG7( o ) shows the quantitative analysis of the FRAP data from FIG7( n ) (n=6-12, mean±standard deviation);

Fig. 7(p) shows representative images of U2OS cells expressing the indicated EGFP-SFPQ IDR-mCherry-Cry2olig fusion proteins, cells were stimulated with 488 nm laser to activate OptoIDR, scale bar, 20 μm;

Figure 7(q) shows quantification of droplet formation kinetics after stimulation as in Figure 7(p), n ≥ 30 cells per construct;

FIG7 (r) shows confocal fluorescence images of Alexa 488-labeled EGFP-SFPQ IDR protein at the indicated concentrations, scale bar, 10 μm;

Figure 7(s) shows FRAP of aggregates formed by Alexa 488-labeled EGFP-SFPQ IDR protein at the following concentrations (FPQ IDR (10 μM), scale bar, 2 μm;

FIG7( t ) shows the quantitative analysis of the FRAP data from FIG7( s ) (n=6-12, mean±standard deviation);

Fig. 7(u) shows representative images of U2OS cells expressing the indicated EGFP-MED15 IDR-mCherry-Cry2olig fusion proteins, and the cells were stimulated with 488 nm laser to activate OptoIDR, scale bar, 20 μm;

Figure 7(v) shows quantification of droplet formation kinetics after stimulation as shown in Figure 7(u), with n≥30 cells per construct;

FIG7 ( w ) shows confocal fluorescence images of Alexa 488-labeled EGFP-MED15 IDR protein at the indicated concentrations, scale bar, 10 μm;

FIG7( x ) shows FRAP of aggregates formed by Alexa 488-labeled EGFP-MED15 IDR protein at the following concentrations (MED15 IDR (10 μM), scale bar, 2 μm;

FIG. 7( y ) shows the quantitative analysis of the FRAP data from FIG. 7( x ) (n=6-12, mean±s.d.).

FIG. 8( a)-FIG. 8( b) show a validation diagram of a selected PS-DBD fusion protein according to the ninth embodiment of the present application, wherein:

FIG8( a) shows a schematic diagram of SDS-PAGE analysis of a designated PS-DBD fusion protein purified from Escherichia coli according to the ninth embodiment of the present application;

FIG8( b ) shows a schematic diagram of the PS, DBD and PS-DBD fusion protein and disorder prediction according to the ninth embodiment of the present application (on the left side of each dotted box), and the rectangle represents DBD.

Figure 9(a) shows a GSEA analysis diagram of the DBD target of PS-DBD fusion tumor samples in the TCGA database according to the tenth embodiment of the present application. For each sample with PS-DBD fusion, samples of the same tumor type without such fusion are selected as the background set.

Figure 9(b) shows a GSEA analysis diagram of the FLI1 target of the EWS-FLI1 fusion cell lines in the CCLE database according to the eleventh embodiment of the present application, in which a bone cell line without EWS-FLI1 fusion is used as the background, and six cell lines with EWS-FLI1 fusion are shown to have significantly increased downstream transcription.

FIG. 10( a) to FIG. 10( c) are schematic diagrams showing phase separation analysis of a specified IDR according to the twelfth embodiment of the present application, wherein:

FIG. 10( a ) shows a representative image of U2OS cells expressing a specified IDR-mCherry-Cry2olig fusion protein according to the twelfth embodiment of the present application, wherein the cells are stimulated with a 488 nm laser to activate OptoIDR, and the scale bar is 20 μm;

FIG10( b ) shows a representative image of U2OS cells expressing a specified IDR-mCherry-Cry2olig fusion protein without 488 nm laser stimulation according to the twelfth embodiment of the present application, with a scale bar of 20 μm;

FIG10( c ) shows a pie chart showing a summary of OptoIDR results according to the twelfth embodiment of the present application.

Figures 11(a)-11(j) show the structural formulas, images before treatment, and images 6 hours after treatment of 10 representative candidate drugs for tumor diseases screened out in the method for screening drugs for tumor diseases according to the thirteenth embodiment of the present application, and the scale bar is 50μm.

12 shows images of wells in which the cell survival conditions after being dispensed with ON123300 are inferior to the predetermined survival conditions in the method for screening drugs for tumor disorders according to the fourteenth embodiment of the present application.

FIG. 13( a) to FIG. 13( h) show schematic diagrams of LY2835219 dissolving aggregates by activating lysosomes according to the fifteenth embodiment of the present application, wherein:

FIG. 13( a ) is a comparative diagram showing a cell image not treated with LY2835219 and a cell image treated with LY2835219 for 6 hours according to the fifteenth embodiment of the present application;

FIG13( b) is a schematic diagram showing the dynamic quantitative effect of LY2835219 on droplets of U2OS/mCherry-FUS-ERGmut cells of a specified concentration according to the fifteenth embodiment of the present application, specifically, showing the decreasing curve of the cell fraction rich in droplets at different doses over time after being distributed to LY2835219;

FIG13( c ) shows a Lysotracker green (lysosomal green fluorescent probe) staining image of U2OS/mCherry-FUS-ERGmut cells after being treated with 7.5 μM LY2835219 for 4 hours according to the fifteenth embodiment of the present application, wherein the DMSO treatment group is used as a negative control, and the scale bar is 50 μm;

FIG13( d ) shows a quantitative schematic diagram of the relative intensity of Lysotracker green (lysosomal green fluorescent probe) in U2OS/mCherry-FUS-ERGmut cells treated with 7.5 μM LY2835219 for 0, 3 and 6 hours by flow cytometry according to the fifteenth embodiment of the present application, ***P<0.001, for two-sample t-test, the error bars represent standard deviations;

FIG13( e ) shows a live cell image of U2OS/mCherry-FUS-ERGmut cells treated with the lysosomal inhibitor Baf-A1 (200 nM) and/or LY2835219 (7.5 μM) according to the fifteenth embodiment of the present application, wherein the scale bar is 50 μm;

FIG. 13( f ) shows live cell images of U2OS/mCherry-FUS-ERGmut cells transfected with lysosomal reporter plasmid YFP-LAMP1 and then treated with DMSO or 7.5 μM LY2835219 for a specified time according to the fifteenth embodiment of the present application, wherein the scale bar is 20 μm;

Figure 13(g) shows live cell images of U2OS/mCherry-EWS-FLI1, U2OS/mCherry-EWS-FLI1mut, U2OS/mCherry-NUP98-HOXD13mut and U2OS/mCherry-NUP98-HOXA9mut treated with 10 μM LY2835219 for the indicated time according to the fifteenth example of the present application, the lower and right figures show the quantitative effect of 10 μM LY2835219 on each cell line, and the scale bar is 50 μm;

13( h ) shows live cell images of U2OS cells transfected with GFP-FUS IDR, GFP-NUP98 IDR or EGFP-TDP43 respectively before and after treatment with 7.5 μM LY2835219 for 6 hours according to the fifteenth embodiment of the present application, and the scale bar is 20 μm.

FIG. 14( a) to FIG. 14( c) show schematic diagrams of LY2835219 reducing the amount of aggregates and activating lysosomes according to the sixteenth embodiment of the present application, wherein:

FIG. 14( a ) shows a live cell image of U2OS/mCherry-FUS-ERGmut cells treated with LY2835219 at a specified concentration according to the sixteenth example of the present application, with a scale bar of 50 μm;

FIG. 14( b ) shows a Lysotracker green staining image of U2OS cells after being treated with 7.5 μM LY2835219 for 4 hours according to the sixteenth embodiment of the present application, with a scale bar of 50 μm;

14( c ) shows a diagram of Western immunoblot analysis of U2OS/mCherry-FUS-ERGmut treated with 7.5 μM LY2835219 for the indicated time according to the sixteenth example of the present application, wherein the fusion protein was detected with an anti-mCherry antibody and an anti-GAPDH antibody was used as an internal control.

FIG. 15( a) to FIG. 15( f) show schematic diagrams of LY2835219 reducing the number of droplets and rescuing abnormal gene expression according to the seventeenth embodiment of the present application, wherein:

FIG15( a ) shows the results of a dual luciferase reporter gene assay in cells treated with 7.5 μM LY2835219 for a specified time according to the seventeenth embodiment of the present application, HEK293T cells were transfected with a dual luciferase reporter plasmid and an EWS-FLI1 fusion protein expression vector, wherein EV represents an empty vector without an introduced fusion gene, ***P<0.001, for a two-sample t-test, the error bars represent standard deviations;

FIG15( b ) shows a schematic diagram of identifying EWS-FLI1 targets in the disclosed perturbation data according to the seventeenth embodiment of the present application, wherein the expression level of the EWS-FLI1 target in the EWS-FLI1 overexpressing cells is higher than that in the FLI1 overexpressing cells and the control cells, which is highlighted in bold, and NC represents a control without overexpression;

FIG. 15( c ) shows a volcano plot of RNA-seq data of differentially expressed genes in U2OS/mCherry-EWS-FLI1 cells before/after treatment with 7.5 μM LY2835219 for 3 hours and 6 hours according to the seventeenth embodiment of the present application;

FIG15( d ) shows the relative expression level of the EWS-FLI1 target in U2OS/mCherry-EWS-FLI1 cells before/after treatment with 7.5 μM LY2835219 according to the seventeenth embodiment of the present application, wherein the transcriptome of wild-type U2OS cells without EWS-FLI1 fusion was used as a control sample to calculate the fold change of expression;

Figure 15(e) shows the RT-qPCR results of the EWS-FLI gene expression level in U2OS/mCherry-EWS-FLI cells treated with 7.5 μM LY2835219 for a specified time according to the seventeenth embodiment of the present application, for each target gene, the mRNA level was normalized with GAPDH, wherein NS indicates no significance, *P<0.05, **P<0.01, ***P<0.001, two-sample t-test, and the error bars represent standard deviations;

FIG. 15( f ) is a schematic diagram showing the principle of how the PS-DBD fusion protein according to the seventeenth embodiment of the present application drives abnormal phase separation in relevant cancers.

FIG. 16( a)-FIG. 16( b) show schematic diagrams of LY2835219 partially rescuing abnormal gene expression in A-673 cells according to the eighteenth embodiment of the present application, wherein:

FIG. 16( a ) shows a volcano plot of RNA-seq data of differentially expressed genes in A-673 cells before/after treatment with 7.5 μM LY2835219 for 3 hours and 6 hours according to the eighteenth embodiment of the present application;

FIG16( b) shows the relative expression level of the EWS-FLI1 target in A-673 cells before/after treatment with 7.5 μM LY2835219 according to the eighteenth embodiment of the present application, wherein the transcriptome from the CCLE bone cell line without EWS-FLI1 fusion is used as a control sample to calculate the fold change of expression;

FIG. 17 shows a schematic diagram of the structure of the backbone plasmid pCMV-mCherry-Age I-FUS-ERG-Age I used in the examples of the present application.

FIG. 18 shows a schematic diagram of the structure of the backbone plasmid pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig used in the examples of the present application.

FIG. 19 shows a schematic diagram of the structure of the backbone plasmid pETDuet used in the examples of the present application.

Figure 20 shows a schematic diagram of the structure of the backbone plasmid pRSFDual used in the examples of the present application.

DETAILED DESCRIPTION

In order to enable persons with ordinary knowledge in the art to understand the features and effects of the present invention, the following is a general description and definition of the terms and expressions mentioned in the specification and the scope of the patent application. Unless otherwise specified, all technical and scientific words used in the text have the common meanings understood by those skilled in the art for this application. In the event of a conflict, the definition in this specification shall prevail.

Method for screening drugs for tumor diseases

Fig. 1 shows a flow chart of a method for screening drugs for tumor diseases according to the first embodiment of the present application, wherein the method shown is also referred to as DropScan herein. As shown in Fig. 1 , the method comprises the following steps.

In step 101, cells containing a target fusion protein are seeded into a well plate comprising a well array, wherein the cells contain a nucleic acid molecule encoding a target fusion protein, the target fusion protein is associated with a target tumor disorder, contains a phase separation prone domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to be able to form aggregates that can be identified by imaging.

As used herein, the term "target fusion protein" refers to a protein translated from a fusion gene produced by two proteins with specific domains in a gene fusion event in a tumor, wherein the fusion gene is mainly composed of a fusion of a protein with a phase separation tendency domain (PS) and a protein with a DNA binding domain (DBD). For a fusion protein obtained by fusion and translation of two genes, the portion of the protein translated from the gene at the 5' end of the fusion gene is called the head, and the portion translated from the gene at the 3' end of the fusion gene is called the tail. If the head of the fusion protein carries a phase separation tendency domain (PS) and the tail carries a DNA binding domain (DBD), or conversely, the tail carries a phase separation tendency domain (PS) and the head carries a DNA binding domain (DBD), it is called a target fusion protein. For example, the SFPQ-TFE3 fusion protein in renal cancer is produced by the translation of the fusion gene of the SFPQ protein with a phase separation tendency domain and the TFE3 protein with a DNA binding domain; another example is the EWS-FLI1 fusion protein in Ewing sarcoma, which is produced by the translation of the fusion gene of the EWSR1 protein with a phase separation tendency domain and the FLI1 protein with a DNA binding domain.

As used herein, the term "phase separation prone domain (PS)" refers to at least one of an intrinsic disordered region (IDR), a low complexity region (LCR), and a prion-like domain (PLD), wherein the intrinsic disordered region can be detected by the ESpritz tool, the low complexity region can be detected by the SEG tool, and the prion-like domain can be detected by the PLAAC tool. Taking the human protein EWS as an example (Uniprot ID Q01844), its 1st amino acid to the 56th amino acid is an intrinsic unneeded region, its 182nd amino acid to the 266th amino acid is a low complexity region, and its 1st amino acid to the 306th amino acid is a prion-like domain.

As used herein, the term "DNA binding domain (DBD)" refers to a domain in a protein that can bind directly or indirectly to DNA, such as the bHLH domain in the transcription factor AHR and the bZIP domain in the transcription factor ATF1.

For example, the structure of the fusion protein can be found in Appendix 1.

Acquisition of PS-DBD fusion protein

In some embodiments, as shown in Figure 2, more than 1500 potential PS-DBD fusion proteins are obtained from multiple databases. These PS-DBD fusion proteins can be used as target fusion proteins for use in methods intended to screen drugs for target tumor disorders, based on their association with the occurrence of target tumor disorders (e.g., concomitant occurrence, significant changes associated with the latter occurrence, corresponding significant changes dependent on changes in the latter status, etc.), or causing the occurrence of target tumor disorders (having a causal relationship with the target tumor disorder).

Phase separation-associated regions include intrinsically disordered regions (IDRs), low complexity regions (LCRs), and prion-like domains (PLDs). IDRs are flexible segments of proteins that lack unique, native structural patterns. LCRs are segments of proteins that contain a composition that is biased toward a few types of amino acids. PLDs are characterized by regions that contain a strong bias toward a few specific types of amino acids, such as asparagine, glutamine, tyrosine, and glycine residues. All three sequences have been found to be enriched in phase-separating proteins.

In order to determine the phase separation part of the fusion protein, as shown in Figure 2, the IDR structure, LCR structure and PLD structure were searched in the sequences of the fusion proteins (gene pairs of fusion genes) in the FusionGDB database, the ChimerKB database and the Mitelman database. In order to detect IDR, the Espritz-Disprot tool was downloaded from the website http://old.protein.bio.unipd.it/espritz/, and a 5% false discovery rate (FDR) was used as the dividing line. In order to identify LCR, the SEG tool was downloaded from http://manpages.ubuntu.com/manpages/bionic/man1/ncbi-seg.1.html, and the default parameters of the ncbi-seg software were used as the threshold. For PLD, the PLAAC software was downloaded from http://plaac.wi.mit.edu/, and the fragment containing the PLD core region predicted by PLAAC was extracted. For those fusion proteins with breakpoint information in FusionGDB, the protein was considered to contain a phase separation-related region when more than 50% of the sequence in the head or tail was predicted to be IDR, or more than 40% of the sequence was predicted to be LCR, or included the PLD core region. For those fusions from ChimerKB and Mitelman without detailed breakpoint positions, the full-length protein was used as the head or tail part of the fusion protein when evaluating its phase separation-related region.

For the DNA binding part, the DNA binding domain and its corresponding PFAM ID were obtained from the Nucleic Acid-Protein Interaction Database (NPIDB). The HMM profiles of the corresponding entries were downloaded from the PFAM database, and the hmmsearch module of the HMMER software was used to search the above HMM profiles in the fusion protein sequences obtained from the FusionGDB, ChimerKB and Mitelman databases. Fragments with E values <10^-5 detected by hmmsearch were considered as candidate DBD fragments. Similarly, for those fusions from FusionGDB, the head or tail sequences were used to detect the DBD fragments. For those fusions from ChimerKB and Mitelman, when detecting the DBD fragments, the full-length protein was used as the head or tail part of the fusion protein.

Finally, the PS-related region and DBD fragment determined in the above steps are respectively in the head sequence and tail sequence of the fusion protein, or conversely, in the tail sequence and head sequence of the fusion protein, and the fusion protein is defined as a PS-DBD fusion protein (hereinafter sometimes also referred to as a PS-DBD fusion).

Specifically, 110 potential PS-DBD fusion proteins were obtained from FusionGDB, and 1424 potential PS-DBD fusion proteins were obtained from ChimerKB and Mitelman databases. PS-DBD fusion proteins are divided into three categories, IDR-DBD, LCR-DBD and PLD-DBD. Since the IDR, LCR and PLD regions are not mutually exclusive, these three types of fusions show substantial overlap. For example, 47 fusions belong to all three of the above categories. Using the method shown in Figure 2, a total of 1472 unique PS-DBD fusion proteins were obtained, all of which can be used as target fusion proteins to screen drugs for target tumor diseases.

FIG3 shows a schematic diagram of the interconnectedness of PS-DBD fusions according to an embodiment of the present application, wherein a fusion network of a central protein with more than 10 partners is shown. Each edge between two nodes represents a fusion event between a protein with phase separation potential and a protein with DNA binding ability. A subnetwork of a central protein is associated with a related cancer type. For example, the N-terminus of NUP98 is fused to 27 different proteins with DBD, many of which are associated with AML. Interestingly, many of the highlighted central hubs are proteins with phase separation potential, such as NUP98, EWS, and FUS. In addition, although not as concentrated as PS proteins, some DBD proteins are also often fused to multiple PS proteins. For example, MiT transcription factor family protein TFE3 can be fused to SFPQ, MED15, or PRCC, and all of these fusions are associated with renal cell carcinoma.

Abnormal phase separation of some fusion proteins can promote tumorigenesis by enhancing transcriptional activation, also known as pathogenic fusion proteins. For example, the IDR in NUP98-HOXA9 drives the formation of phase-separated aggregates, which is essential for the development of abnormal chromatin loops and cancer. Another similar example is EWS-FLI1, in which the PLD in EWS drives the generation of phase-separated aggregates, which is essential for recruiting the BAF complex and activating abnormal transcription. It can be expected that many of the above 1472 candidate PS-DBD fusion proteins promote tumorigenesis through abnormal phase separation, and thus can serve as target fusion proteins for drug screening methods for the tumors they promote.

In some embodiments, the labeling molecule includes various fluorescent labels, such as but not limited to mCherry fluorescent labels, etc., so that the formed aggregates can be imaged and identified. In some embodiments, cells containing the target fusion protein are, for example, U2OS cells, Hela cells, etc.

Regarding step 101, for example, as shown in FIG. 4( a), U2OS cells stably expressing mCherry-FUS-ERGmut can be inoculated into a well plate of a 384-well array, and the cells contain a nucleic acid molecule encoding FUS-ERG as a target fusion protein. For another example, other cells (e.g., selected from U2OS cells and 293T cells) other than U2OS cells stably expressing mCherry-FUS-ERG or mCherry-EWS-FLI1mut (rather than mCherry-FUS-ERGmut) can also be inoculated into a well plate of a 384-well array. The FUS-ERG fusion protein is a homologous fusion protein of EWS-FLI1 (see FIG. 5( a)), which is found in acute myeloid leukemia (AML) and Ewing's sarcoma and is associated with poor clinical outcomes.

The amino acid sequence of mCherry-FUS-ERGmut is as follows (SEQ ID NO: 1):

MKVSKGEEDNMAIIKEFMRFKVHMEGSVNGHEFEIEGEGEGRPYEGTQTAKLKVTKGGPLPFAWDILSPQFMYGSKAYVKHPADIPDYLKLSFPEGFKWERVMNFEDGGVVTVTQDSSLQDGEFIYKVKLRGTNFPSDGPVMQKKTMGWEASSERMYPEDGALKGEIKQRLKLKDGGHYDAEVKTTYKAKKPVQLPGAYNVNIKLDIT SHNEDYTIVEQYERAEGRHTGGMDELYKTMKGGSGGSGGSGGSLEMASNDYTQQATQSYGAYPTQPGQGYSQQSSQPYGQQSYSGYSQSTDTSGYGQSSYSSYGQSQNTGYGTQSTPQGYGSTGGYGSSQSSQSSYGQ QSSYPGYGQQPAPSSTSGSYGSSSQSSSYGQPQSGSYSQQPSYGGQQQSYGQQQSYNPPQGYGQQNQYNSSSGGGGGGGGGGNYGQDQSSMSSGGGSGGGYGNQDQSGGGGSGGYGQQDRGGRGRGGSGGGGGGGGGGYNRSSGGYEPRGRGGGRGGRGGMGGSDRGGFNKFGGSGQIQLWQFLLELLSDSS NSSCITWEGTNGEFKMTDPDEVARRWGERKSKPNMNYDKLSDALDYYYDKNIMTKVHGKRYAYKFDFHGIAQALQPHPPESSLYKYPSDLPYMGSYHAHPQKMNFVAPHPPALPVTSSSFFAAPNPYWNSPTGGIYPNTRLPTSHMPSHLGTYY*

The amino acid sequence of mCherry-EWS-FLI1mut is as follows (SEQ ID NO: 2):

MKVSKGEEDNMAIIKEFMRFKVHMEGSVNGHEFEIEGEGEGRPYEGTQTAKLKVTKGGPLPFAWDILSPQFMYGSKAYVKHPADIPDYLKLSFPEGFKWERVMNFEDGGVVTVTQDSSLQDGEFIYKVKLRGTNFPSDGPVMQKKTMGWEASSERMYPEDGALKGEIKQRLKLKDGGHYDAEVKTTYK AKKPVQLPGAYNVNIKLDITSHNEDYTIVEQYERAEGRHTGGMDELYKTMKGGSGGSGGSGGSLEMASTDYSTYSQAAAQQGYSAYTAQPTQGYAQTTQAYSAYTAQPTQGYAQTTQAYGQQSYGTYGQPTDVSYTQAQTTATYGQTAYATSYGQPPTGYTTPTAPQAYSQPVQGYGTGAYDTTTATVTTTQASYAAQSAYGTQP AYPAYGQQPAATAPTRPQDGNKPTETSQPQSSTGGYNQPSLGYGQSNYSYPQVPGSYPMQPVTAPPSYPPTSYSSTQPTSYDQSSYSQQNTYGQPSSYGQQSSYGQQSSYGQQPPTSYPPQTGSYSQAPSQYSQQSSSYGQQNPSYDSVRRGAWGNNMNSGLNKSPPLGGAQTISKNTEQRPQPDPY QILGPTSSRLANPGSGQIQLWQFLLELLSDSANASCITWEGTNGEFKMTDPDEVARRWGERKSKPNMNYDKLSLALLYYYDKNIMTKVHGKRYAYKFDFHGIAQALQPHPTESSMYKYPSDISYMPSYHAHQQKVNFVPPHPSSMPVTSSSFFGAASQYWTSPTGGIYPNPNVPRHPNTHVPSHLGSYY*

Mechanisms of association between PS-DBD fusion proteins and tumor pathology: phase separation, aggregate formation, and abnormal genes Express

Taking the FUS-ERG fusion protein as an example, which includes FUS as PS and ERG as DBD, it can drive the occurrence of phase separation. Specifically, the FUS part undergoes phase separation, while the ERG part contains a DBD that recognizes the GGAA microsatellite sequence.

By expressing mCherry-FUS-ERG (stable or transient) in cells (such as Hela cells, 293T cells or U2OS cells), it was found to co-localize with chromatin (Figure 5(b)) and show a small amount of aggregates in the nucleus. Specifically, the FUS-ERG fusion protein undergoes phase separation both in vivo and in vitro to form aggregates, which are bound to chromatin and show a recognizable morphology that deviates from the spherical shape, similar to the wetting behavior of a liquid.

In some embodiments, the phase separation of FUS-ERG fusion proteins was evaluated and verified using the OptoIDR system. The OptoIDR system consists of an IDR of interest, an mCherry fluorescent marker, and a mutant of an aggregation-enhanced form of the photolyase domain of the Arabidopsis Cry2 protein (Cry2olig). Under blue light stimulation, Cry2olig forms oligomers, which also causes IDR oligomerization and increases the valence of IDRs. This may drive phase separation at concentrations below the LLPS critical concentration, as shown in Figure 5(c). As shown in Figures 5(d) and 5(e), blue light stimulation causes obvious phase separation in cells expressing OptoFUS, indicating that FUS IDR does have the potential to undergo phase separation.

Next, FUS-ERG tagged with recombinant enhanced green fluorescent protein (EGFP) was purified (Figure 6(a)). Under physiological buffer conditions without crowding agents, EGFP-FUS-ERG phase separated at submicromolar concentrations (Figure 5(f)). Aggregate fusion experiments and fluorescence recovery after photobleaching (FRAP) experiments showed that the aggregates had liquid properties (see Figures 5(g), 5(h), and 5(i)). By incubating a DNA fragment containing 25 copies of the GGAA microsatellite sequence with EGFP-FUS-ERG (Figures 6(b) and 6(c)), it was found that the DNA substrate was recruited into the aggregates, and these aggregates became irregular (Figure 5(j)). FRAP analysis showed that these aggregates still retained a certain degree of fluidity (Figures 5(k) and 5(l)). These aggregates can be identified from images, for example, they can appear as gel-like, flowable irregular liquid forms, etc.

Among the above 1472 candidate PS-DBD fusion proteins, in addition to the FUS-ERG fusion protein, many other fusion proteins also drive, occur or undergo abnormal phase separation in vivo and in vitro.

For example, BRD9 is a bromodomain-containing protein that plays an important role in chromatin remodeling and transcriptional regulation. BRD9 is a fusion partner of TERT (telomerase reverse transcriptase) in hepatocellular carcinoma. Based on the OptoIDR system, BRD9 IDR promoted the occurrence of phase separation (see Figure 7(a) and Figure 7(b)). In addition, the purified recombinant Alexa 488-labeled BRD9IDR protein (Figure 8(a)) formed spherical droplets with fluidity (see Figure 7(c), Figure 7(d) and Figure 7(e)). Therefore, BRD9IDR does promote phase separation in vivo and in vitro. Previous studies have shown that in liposarcoma expressing another TERT fusion protein (BRD9-TERT), the expression level of TERT mRNA is about 100 times higher than that in liposarcoma without this fusion protein. Activation of TERT is necessary for the unlimited proliferation of tumor cells. That is, the phase separation activity of BRD9 promotes TERT activation through fusion.

COL17A1 (collagen type XVII α1 chain) is a collagen transmembrane protein and a key component of the epidermal anchoring complex. COL17A1 is fused to the CST complex subunit SNT1 in gastric adenocarcinoma. The CST complex binds to single-stranded DNA with high affinity in a sequence-independent manner. COL17A1 IDR also promotes phase separation in the OptoIDR system. In addition, the purified recombinant COL17A1 IDR protein (see Figure 8(a)) forms aggregates with gel properties. Therefore, COL17A1 can promote phase separation both in vivo and in vitro. The phase separation potential of STN1-COL17A1 may promote tumorigenesis such as gastric adenocarcinoma.

MLLT1 (also known as ENL) is a histone acetylation reader that is particularly important for cell proliferation and oncogenic transcriptional regulation in AML. MLLT1 is often found fused to MLL (also known as KMT2A) in leukemia. MLLT1 IDR promotes phase separation in the OptoIDR system. In addition, purified MLLT1 IDR protein (see Figure 8(a)) formed liquid aggregates. That is, MLLT1 can drive phase separation both in vivo and in vitro, suggesting that abnormal phase separation of MLL-MLLT1 fusions may play a role in leukemogenesis such as AML.

SFPQ is a pre-mRNA splicing factor. Chromosomal aberrations involving the fusion of SFPQ with TFE3 may be the cause of papillary renal cell carcinoma. SFPQ IDR promotes phase separation in the OptoIDR system. Purified EGFP-SFPQ IDR protein (see Figure 8(a)) forms aggregates and exhibits liquid-like properties (not shown).

MED15 is another fusion partner of TFE3 (see Figure 8(a)), and this fusion event is associated with the development of renal cell carcinoma. Purified EGFP-MED15 IDR protein can form aggregates with liquid properties in vivo and in vitro (not shown).

As exemplified above, a large number of PS-DBD fusion proteins undergo phase separation in cells, and the phase separation and the aggregates formed as a result thereof are associated with the occurrence of certain tumor disorders, and various PS-DBD fusion proteins can be used as target fusion proteins in drug screening methods for these associated tumor disorders (i.e., target tumor disorders). In some embodiments, the target tumor disorders include but are not limited to malignant tumors.

Aggregates identified from cell images are caused by phase separation, so that the phase separation status of the PS-DBD fusion protein (as a target fusion protein) and/or the change in the status of the target tumor disease of the cell can be determined by observing the change in the amount of aggregates in the cell. For example, the greater the amount of aggregates in the cell, the more severe the phase separation status of the PS-DBD fusion protein in vivo and/or in vitro. Abnormal phase separation is the operating mechanism of cancer-related PS-DBD fusion. The present application proposes a new approach to combat related cancers by changing the phase separation behavior through small molecules, and accordingly, PS-DBD fusions may represent new cancer drug targets. For another example, a reduction in the amount of aggregates in the cell means an alleviation of the status of the target tumor disease of the cell, that is, a good therapeutic effect of the target tumor disease.

The phase separation experienced by the target fusion protein drives aberrant gene expression, and a reduction in the amount of the aggregates is associated with a reversal of aberrant gene expression.

For example, EWS-FLI1 and NUP98-HOXA9 can alter downstream transcription and drive tumorigenesis. To determine whether downstream transcriptional changes caused by PS-DBD fusions are common, transcriptome data from patients with PS-DBD fusions in The Cancer Genome Atlas (TCGA) were analyzed. Data from 116 patients with PS-DBD fusions were extracted from the TCGA library. The direct targets of PS-DBD fusions are not yet clear, so target genes of DBD were used as a surrogate. Target genes of DBD proteins were collected from enrichrlibraries to construct a target gene list for 462 DBD proteins. Combining TCGA transcriptome data and target gene lists, downstream transcription levels of 21 PS-DBD fusions in 27 patients were obtained. The GSEA algorithm was used to evaluate the downstream expression levels of PS-DBD fusions, and samples from the same cancer type were used as background controls. Among the 27 tumor samples, 13 fusion events were associated with significant changes in downstream transcription, of which 11 were upregulated and 2 were downregulated (Figure 9(a)). This result suggests that PS-DBD fusion protein can indeed alter downstream transcription in tumor samples.

Next, it was assessed whether the PS-DBD fusion protein could alter downstream expression in established cancer cell lines. Data for 10 EWS-FLI1 fusion cell lines were obtained by analyzing cell line transcriptome data in the CCLE database 56. In all 10 cell lines, the expression levels of FLI1 target genes were significantly upregulated (Figure 9(b)). In addition, dual luciferase analysis was used to detect the relative luciferase activity in cell lines with FET-ETS fusions. The promoter region of the luciferase reporter gene carries 25 copies of GGAA, which is the target sequence of ERG and FLI1. Both FUS-ERG and EWS-FLI1 enhanced the relative activity of luciferase.

As above, by analyzing patient data containing PS-DBD fusion information and downstream expression in cancer cell lines, as well as using luciferase reporter analysis, it was verified that PS-DBD fusion proteins can indeed alter the expression of target genes, thereby driving abnormal gene expression.

By understanding the association mechanism between PS-DBD fusion protein and target tumor disease, including phase separation and its driving of abnormal gene expression, and the positive correlation between the amount of aggregate recognition and the phase separation status, different PS-DBD fusion proteins can be selected for various target tumor diseases, and then small molecule compounds that significantly attenuate the amount of aggregates, that is, small molecule compounds that significantly reverse the phase separation status, can be selected as candidate drugs for target tumor diseases. Please note that this application proposes an efficient and intuitive phenotypic screening method for candidate drugs for target tumor diseases. Regardless of the mechanism and method by which the small molecule compound dissolves and attenuates the aggregates, regardless of whether it regulates the aggregates by direct physical interaction or indirect action, as long as the amount of aggregates is reduced, it is considered to have therapeutic efficacy for the target tumor disease.

In some embodiments, when the target tumor disorder is a tumor associated with FET fusion, a blood disease associated with NUP98 or MLL fusion, a TFE3 fusion-associated renal cell carcinoma, and a HMGA2 fusion-associated lipoma, a member derived from the FET-ETS fusion protein family is selected as the target fusion protein.

Specifically, the correspondence between the target fusion protein and the target malignant tumor includes at least one of the following. If the target fusion protein contains NUP98-HOXA9, the corresponding target malignant tumor is acute myeloid leukemia. If the target fusion protein contains SFPQ-TFE3, MED15-TFE3 and/or PRCC-TFE3, the corresponding target malignant tumor is renal cell carcinoma. If the target fusion protein contains BRD9-IDR or BRD9-TERT, the corresponding target malignant tumor is hepatocellular carcinoma. If the target fusion protein contains TRIO-TERT, the corresponding target malignant tumor is liposarcoma. If the target fusion protein contains STN1-COL17A1, the corresponding target malignant tumor is gastric adenocarcinoma. If the target fusion protein contains MLL-MLLT1, the corresponding target malignant tumor is leukemia. If the target fusion protein contains SFPQ-TFE3, the corresponding target malignant tumor is papillary renal cell carcinoma.

Introducing mutations in the DBD of the target fusion protein

In some embodiments, the DBD of the target fusion protein contains mutations, which can make the imaging recognition of the formed aggregates more significant than the imaging recognition of the aggregates formed by the DBD not containing mutations. The PS-DBD fusion protein was transiently expressed in Hela cells, and a DBD mutant version (PS-DBDmut) of each PS-DBD was also generated. The live cell image of the ectopically expressed Hela cell is shown in Figure 8 (b), wherein the DNA is stained with Hoechst 33342 (blue), the scale bar is 10 μm, and the upper live cell image in each dotted box shows the image of the Hela cell expressing mCherry-PS-DBD, while the lower live cell image in the corresponding dotted box shows the image of the Hela cell expressing mCherry-PS-DBDmut (a mutant version with a mutation introduced into the DNA binding domain). It can be seen that in most of the dotted boxes, the aggregates formed by the latter are nearly spherical (or spherical) droplets (see Figure 5 (m)), or converge into larger clumps, so that the imaging recognition is better than the former. The aggregates formed by the former are bound and attached to chromatin (or chromosomes). The introduction of mutations into the DNA binding domain releases the binding, thereby releasing the aggregates and converging into droplets.

For TFE3 fusion proteins, SFPQ/MED15-TFE3 and its DBD mutant form produced aggregates with different morphologies in cells, and the latter morphology was easier to identify (see Figure 8(b)). These data suggest that the PS-TFE3 fusion protein found in renal cell carcinoma may have a phase separation-dependent mechanism in the carcinogenic process. The EWS-FLI1 fusion protein formed some aggregates similar to FUS-ERG, while EWS-FLI formed many droplets in cells (see Figure 8(b)). The other two EWS fusion proteins, EWS-ERG and EWS-ATF1, have both shown obvious phase separation droplets in the absence of DBD mutations (Figure 8(b)). The NUP98 fusion protein forms aggregates in cells (Figure 8(b)). In other words, for some PS-DBD fusion proteins, as long as the morphological recognition of the aggregates is high enough, mutations in the DBD may not be introduced, and mutations can be selectively introduced according to needs.

It was noted that some PS-DBD fusion proteins, such as PRCC-TFE3, did not undergo phase separation ( FIG. 8( b ) ).

Using OptoIDR analysis on more PS protein candidates (Figure 10(a) and Figure 10(b)), it was found that OptoHES4 formed aggregates under blue light stimulation. However, there are still some OptoIDR-modified proteins that do not respond to blue light and do not show obvious aggregates. It is speculated that these proteins have relatively weak phase separation ability or no phase separation behavior, and the stimulation intensity is not enough to activate it. For all OptoIDR results, it was found that 20% of the proteins showed aggregates before stimulation, 40% of the OptoIDR-modified proteins phase separated and showed aggregates under blue light stimulation, and 40% of the OptoIDR-modified proteins did not respond to blue light stimulation (see Figure 10(c)). Overall, these results indicate that more than half of the PS-DBD fusion proteins found in cancer patients tend to phase separate, and abnormal phase separation may amplify or inhibit downstream signaling pathways, thereby affecting cell physiology and being associated with the occurrence of cancer, and can be used as target fusion proteins in step 101.

In step 102, a plurality of small molecule compounds are distributed to each well. As shown in FIG4(a), more than 1700 small molecule compounds used as anticancer drugs from the APExBIO Anti-Cancer Compound Library Plus can be distributed in parallel to five well plates, that is, 384*5 wells. That is, by using five well plates and executing the process shown in FIG1, high-throughput screening of more than 1700 drugs can be performed in parallel for the target tumor disease. The small molecule compounds for distribution to each well for subsequent screening can be randomly selected from the small molecule compound library, for example, can be selected from a drug small molecule library for treating tumor diseases or malignant tumor diseases.

In some embodiments, for the target tumor condition, drugs for treating other conditions other than the tumor condition or malignant tumor are used as the small molecule compounds to be distributed into each well for screening. The other conditions are associated with the target fusion protein, or are associated with the abnormal phase separation of the target fusion protein. Other conditions, such as ALS, frontotemporal dementia (FTD), Alzheimer's disease (AD), Parkinson's disease (PD), Huntington's disease (HD), spinocerebellar ataxia (SCA), and myotonic dystrophy type 1 (DM1) and type 2 (DM2), are all known to be associated with pathological deposits, and candidate drugs for the target tumor condition can also be screened from the small molecules in the drug small molecule library of these conditions via the method shown in Figure 1. Benefiting from the high-throughput processing efficiency of the present application, this screening can be a random screening. The clinical trial of a drug is a long process. By screening candidate drugs for the target tumor disease from the drug library of other diseases that have been approved or even used for a long time, once successfully screened, the potential risk of the selected drug to the human body can be significantly reduced (because these drugs may have been used for a long time, and the side effects have been fully verified), and the manufacturing cost is significantly reduced. In addition, with the aggravation of the aging of the population, many other diseases may occur with the target tumor disease, which is conducive to doctors to make more accurate combination drug regimens. If these other diseases are also associated with the same or homologous target fusion protein corresponding to the target tumor disease, or are associated with abnormal phase separation of the target fusion protein, the probability of successful screening is high, and key screening can also be performed. In other words, if other diseases correspond to the same target point as the target tumor disease - the same or homologous target fusion protein, the former drug can also be efficiently reused in the target tumor disease through screening and verification.

In some embodiments, for the target malignant tumor condition, drugs for treating other malignant tumor conditions other than the target malignant tumor condition are used as the small molecule compounds to be distributed to each well for screening. For example, the target malignant tumor includes sarcoma, and the other malignant tumor conditions include any one of breast cancer, leukemia, kidney cancer, liver cancer, gastric cancer and thyroid cancer. Benefiting from the high-throughput processing efficiency of the present application, this screening can be a random screening, similar to screening from a drug library of other diseases, so that the potential risk of the selected drug to the human body can be significantly reduced, the manufacturing cost of the drug can be significantly reduced, and it is beneficial for doctors to make more accurate combined drug regimens. For example, if other malignant tumor conditions are associated with the target fusion protein, or are associated with the abnormal phase separation of the target fusion protein, there is a higher probability that candidate drugs for the target malignant tumor condition can be screened out from its drug library, and focused screening can also be performed. That is to say, if other malignant tumor conditions correspond to the same target point-the same or homologous target fusion protein as the target malignant tumor condition, the former drug can also be efficiently reused in the target malignant tumor condition through screening verification.

In step 103, a microscopic imaging system is used to continuously photograph each well with several different fields of view within a predetermined time period to generate time series images of several fields of view of each well. The microscopic imaging system can be, for example, an automatic high-speed microscopic imaging component equipped with a high-content imaging analysis system, but is not limited thereto. A microscopic imaging system that can continuously photograph and image the cells in each well can be used as the microscopic imaging system. In some embodiments, the microscopic imaging system can use a microscope as an imaging component (as in a high-content imaging analysis system), but is not limited thereto, and imaging components of other imaging technologies such as microfluidic chips can also be used.

As shown in FIG4(a), a high-content imaging analysis system can be used to image multiple fields of view of each well after small molecule compounds are dispensed into each well, and each field of view can be continuously imaged in a long-term imaging mode for 6 hours.

The high-content imaging analysis system generally includes an automatic high-speed microscopic imaging component and at least one processor, and the at least one processor can be configured to execute the method for screening drugs for tumor diseases according to the various embodiments of the present application. High-content image analysis refers to the simultaneous detection of the effects of different conditions on cell morphology, growth, differentiation, migration, apoptosis, metabolic pathways and signal transduction under submicroscopic morphology while maintaining the structural and functional integrity of living cells, obtaining a large amount of relevant information from a single experiment to determine its biological activity and potential toxicity. Its detection range includes but is not limited to: cell counting, protein expression, cell apoptosis, protein translocation, cell viability, cell migration, receptor endocytosis, cytotoxicity, cell cycle and signal transduction. The at least one processor can be further configured to perform fully automatic image analysis and data management and other processing.

In this way, during the entire imaging process, time series images are generated for each well where various small molecule compounds act, so that the kinetic changes of aggregate behavior in each well can be monitored. Based on the analysis of the kinetic changes of aggregate behavior in each well, the influence of individual differences in aggregate behavior between wells at each sampling time point is eliminated, so that candidate drugs can be screened more efficiently and accurately.

In step 104, the processor analyzes the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well. In step 105, the processor screens out the wells whose attenuation of the amount of intracellular aggregates meets the predetermined conditions, and uses the small molecule compounds allocated to the screened wells as candidate drugs for the target tumor disease. According to the phase separation-aggregate association mechanism described above, the attenuation effect on the amount of aggregates means the reversal of the phase separation condition, and also means the therapeutic effect on the target tumor disease. If the attenuation of the amount of aggregates meets the predetermined conditions, it is considered to have therapeutic efficacy for the target tumor disease, and then the corresponding small molecule compound can be selected as its candidate drug. In some embodiments, the attenuation of the amount of aggregates may include the attenuation ratio of the amount of the current aggregates in the same well relative to the amount of aggregates at the reference time, the attenuation ratio of the current aggregate-rich cell fraction in the same well relative to the aggregate-rich cell fraction at the reference time, and the like.

Through the screening method shown in FIG1 , the aggregates formed by the fusion protein can be imaged and identified, and the dynamic attenuation regulation of a large number of different small molecules relative to the aggregates can be continuously and comprehensively monitored at high throughput. It is worth noting that the screening method reveals the dynamics of aggregate changes by analyzing the time series of images, effectively and directly quantifies the effects of small molecules on aggregates such as droplet morphology, and can be used for large-scale, highly parallel compound screening. This analysis method based on time series image-aggregate dynamic changes eliminates the influence of the large difference in the proportion of droplet-rich cells at different time points in different wells (see FIG11 (a)-FIG11 (j)), so that even if the difference is large, candidate drugs can still be efficiently and accurately screened for the target tumor disease. The candidate drug can be used as a lead candidate for clinical treatment drugs for tumor diseases.

In some embodiments, continuously photographing each well with several different fields of view within a predetermined time period specifically includes: dividing each well into several independent fields of view; for each well, sequentially photographing images of each field of view to obtain time series images of each field of view within a predetermined time period. For a long photographing time of, for example, 6 hours, during which living cells will move, if a single field of view is photographed, the cell distribution in the photographed image is prone to randomness (for example, the number of cells in the field of view changes dramatically, the number of cells is scarce, etc.), thereby causing deviations from the true dynamic changes of the behavior of the aggregates. Continuously photographing images with multiple fields of view, and performing a joint comparative analysis of the attenuation conditions of the intracellular aggregates therein, such as superposition or averaging, can eliminate the influence of distribution randomness, so that the analysis of the true dynamic changes of the behavior of the aggregates in each well is more accurate and more robust.

In some embodiments, analyzing the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well specifically includes performing the following steps for each well.

For the images of each sampling time point of the time series images of each field of view, the segmentation of cells and aggregates is performed, and the amount of aggregates of each cell is determined. Specifically, the amount of aggregates of each cell can be obtained by associating the segmented aggregates with the segmented cells. For example, the segmentation of cells and aggregates can adopt various image segmentation algorithms, including but not limited to various segmentation learning networks (such as convolutional neural networks), edge-based segmentation algorithms, threshold-based segmentation algorithms, etc.

For the images at each time point of each visual field, the proportion of cells whose aggregate amount exceeds the enriched aggregate threshold is determined, and the average is calculated based on the visual field as the aggregate-rich cell fraction of the well at each sampling time point, thereby obtaining a distribution curve of the aggregate-rich cell fraction of the well within the predetermined time period. For each visual field, the proportion of cells whose aggregate amount exceeds the enriched aggregate threshold is determined and then averaged based on the visual field to characterize the attenuation of the amount of intracellular aggregates. Compared with directly using the attenuation of the amount of aggregates, or first calculating the average amount of aggregates based on the visual field and then calculating the aggregate-rich cell fraction, the influence of the imbalance of cell numbers in different visual fields can be eliminated, so as to prevent the aggregate state of the visual field with a large number of cells from dominating and covering up the aggregate state of other visual fields, and at the same time, the influence of the distribution contingency caused by cell movement in different visual fields can be eliminated, so that the characterization of the attenuation of the amount of intracellular aggregates is more comprehensive, thoughtful and accurate.

Based on the distribution curve of the aggregate-rich cell fraction of the well within the predetermined time period, an exponential model is fitted. Based on the fitted exponential model, the fitting effect parameter and the attenuation ratio of the aggregate-rich cell fraction are calculated. The applicant has found that the fitted exponential model can achieve optimized regression and eliminate local disturbances relative to the aggregate-rich cell fraction at each time point on the distribution curve. Not only is the attenuation ratio of the aggregate-rich cell fraction calculated based on this more accurate and more robust, but the reliability of the attenuation ratio calculation of the distribution curve can also be tested by the fitting effect parameter. Among them, the predetermined condition includes that the attenuation ratio of the aggregate-rich cell fraction reaches the first predetermined condition and the fitting effect parameter is better than the second threshold value, that is, the calculation result of the attenuation ratio of the aggregate-rich cell fraction is sufficiently reliable and reaches the first predetermined condition, and it is considered that the dissolution attenuation effect of the small molecule compound distributed to the screened well on the amount of aggregates (also means the reversal effect on the phase separation state) reaches a therapeutic effect, so that it can avoid guiding the screening of candidate drugs for target tumor diseases based on unreliable attenuation ratio calculation results.

In some embodiments, if the influence of distribution contingency caused by cell movement in the field of view is small, similar processing can also be performed based on a single field of view. Specifically, for the images of the time series images of each field of view at each sampling time point, the segmentation of cells and aggregates is performed, and the amount of aggregates of each cell is determined. For the images of each sampling time point of a single field of view, the proportion of cells whose aggregate amount exceeds the enriched aggregate threshold is determined as the aggregate-rich cell fraction of the hole at each sampling time point, thereby obtaining the distribution curve of the aggregate-rich cell fraction of the hole in the predetermined time period. Based on the distribution curve of the aggregate-rich cell fraction of the hole in the predetermined time period, an exponential model is fitted. Based on the fitted exponential model, the fitting effect parameter and the attenuation ratio of the aggregate-rich cell fraction are calculated. Wherein, the predetermined condition includes that the attenuation ratio of the aggregate-rich cell fraction reaches the first predetermined condition and the fitting effect parameter is better than the second threshold.

In some embodiments, the following method can be used to perform step-by-step targeted segmentation of cells and aggregates on each image. The first algorithm can be used to segment cells. Identify the point structure in the image, and perform enhancement with the representative size of the aggregate as the target size; then use the second algorithm to segment the aggregates whose intensity value is higher than the representative intensity value of the background by a third threshold. A 6-hour shooting of a well plate can generate up to 10G or even 50G of image data. By first using the first algorithm to segment the cells, a cell mask (binary matrix) can be formed, which significantly improves the subsequent image processing speed and reduces the computational load. The point structure corresponding to the aggregate is bright (and the distribution may be uneven or dispersed due to noise), and the image background is also bright. The contrast between the two is not high enough. By performing enhancement with the representative size of the aggregate (e.g., 5-10 pixels) as the target size, the aggregate points with insufficient contrast can be efficiently identified by simple threshold division with the representative intensity value of the background.

For example, the first algorithm automatically determines the intensity threshold for segmenting the cell foreground and the cell background and performs cell segmentation accordingly. Compared with the use of learning networks or edge-based recognition, the segmentation based on the intensity threshold of the foreground and the background has a smaller computational load and a faster computational speed, and the segmentation effect of the cells in the field of view image of the hole is also good. The second algorithm automatically determines the representative intensity value of the robustness of the aggregate background, and the third threshold is 3-5 times the standard deviation of the representative intensity value of the background. In this way, after the cells are segmented, the enhanced aggregates can be efficiently and comprehensively identified using such a third threshold setting.

Referring to FIG. 4( b ), for each image captured by the high-content microscope, the image can be normalized, for example, the intensity value of each pixel is divided by 0.05 to generate a new image. For each image, the Otsu algorithm can be first used to segment cells with a target diameter of 70 to 250 pixels. The Otsu algorithm can quickly and automatically determine the threshold for segmenting the foreground and background of cells, and is particularly suitable for the segmentation of dozens to hundreds of cells in a single image, which happens to be the number of cells usually in a single field of view of each well.

As shown in Figure 4(b), the dot structure in the image is enhanced, and the target diameter is set to 5 pixels, that is, the droplet is enhanced. For example, a white tophat filter is used to enhance the image, and its enhancement radius is set to be close to the representative radius of the droplet.

Then, referring to FIG. 4( b ), a robust background algorithm can be used to detect droplets with a diameter of 5 to 10 pixels, wherein structures with an intensity higher than the representative intensity value of the background, such as 4 standard deviations, are identified and segmented as droplets. The droplets present in a single image are usually in the order of tens of thousands, and the brightness of different droplets is not as uniform as that of cells. The robust background algorithm is used to determine whether the structure is a droplet by considering the relative comparison relationship between the structure and the surrounding brightness. Specifically, the robust background algorithm can first remove the brightest and darkest pixel intensities in the image to eliminate outliers and true signals so that all remaining pixels represent only the background intensity level. Then, the mean and standard deviation of the background intensity are calculated. Then, the threshold is calculated by adding the mean to n*standard deviation, n=3, 4 or 5, and those above the threshold are considered to be foreground droplets. The robust background algorithm can be executed in the entire image or in any local area of the image as needed.

Next, the identified droplets are associated with the cells identified in the previous step to obtain the number of droplets per cell.

For each well plate, the number of droplets per cell at the initial time point can be obtained. All cells at the initial time point are sorted by their droplet number, and the number of droplets at the top 15% is used as the threshold for defining droplet-enriched cells. Then, for all wells in the plate, the proportion of cells with droplet numbers exceeding the droplet enrichment threshold is monitored over time. The time series of the proportion of droplet-enriched cells after drug treatment can be well fitted by the following exponential model, the curve of which is shown in Figure 4(c).

The exponential model is defined by the following formula (1), which describes the fraction of droplet-enriched cells (also the fraction of aggregate-enriched cells in each well) as a function of time after drug treatment:

Where _c0 and _c1 are constants that need to be regressed in the exponential model, t is the action time of the small molecule compound, and Y(t) is the fraction of aggregate-rich cells in each well, which can be the fraction of aggregate-rich cells based on a single field of view or the fraction of aggregate-rich cells averaged over multiple fields of view.

In some embodiments, the fitting effect parameter is R square, and the second threshold is 0.2-1.0. Specifically, see the scatter plot of R square (example of fitting effect parameter) and log ₂ drug effect (example of derivative parameter of attenuation ratio of aggregate-rich cell fraction) of more than 1777 small molecule compounds shown in Figure 4 (d). Specifically, the first predetermined condition is the first predetermined number of attenuation ratios reaching 0.2 and arranged in order of attenuation ratio, and the attenuation ratio is defined as the ratio of the attenuated aggregate-rich cell fraction calculated using the fitted exponential model to the initial aggregate-rich cell fraction. The small molecule compound assigned to the well that meets the first predetermined condition and the fitting effect parameter is better than the second threshold can be used as a candidate drug for the target tumor condition.

For example, based on the scatter plot in Figure 4(d), 144 small molecule compounds corresponding to the regions with R-square greater than 0.2 (good fitting effect) and drug effect less than 0.5 (the attenuation of the droplet-rich cell fraction under the action of the drug exceeds 50%) can be extracted as candidate compounds for further study.

After manually checking the cell status and the number of droplets after treatment, the top 10 drugs ranked in order of attenuation ratio among the 144 candidate drugs can be selected for retesting (see FIG. 4( e)). In some embodiments, the method further includes: manually or automatically identifying the survival status of cells in each well, determining the wells whose survival status is inferior to the predetermined survival condition; and excluding the small molecule compound assigned to the well from the candidate drugs for the target tumor disease.

For example, automatically identifying the survival status of cells in each well specifically includes: using an image processing algorithm to determine at least one of the cell morphology and the rate of reduction of the number of cells; when the deviation of the identified cell morphology from the spherical shape is higher than a deviation threshold and/or the rate of reduction of the number of cells is higher than a reduction threshold, identifying the cell as having a survival status worse than a predetermined survival condition. In this way, toxic small molecule compounds can be excluded through manual or automatic screening.

Of the top 10 small molecule compounds shown in Figure 4(e), eight compounds showed a significant reduction in the number of aggregates (see Figures 11(a)-11(j)). Among them, LY2835219 (also known as abetamoxiclib mesylate) is included, which is a selective cyclin-dependent kinase (CDK) inhibitor that targets the cyclin D1/CDK4 and cyclin D1/CDK6 cell cycle pathways. In addition, LY2835219, a potent pan-PIM kinase inhibitor CX-6258, and two epidermal growth factor receptor inhibitors EKB-569 and AZD-9291 showed effective aggregate dissolution (see Figures 11(a), 11(i), and 11(b)). In addition, some hit compounds, such as UNC 0631, may reduce the number of droplets by triggering cell death (see Figure 11(h)). In summary, using DropScan, we were able to identify a panel of aggregation-modulating compounds with varying effectiveness at a screening throughput of nearly 2,000 small molecules at a time.

The target tumor disorder can be various tumor disorders associated with FET-ETS fusion proteins, including at least one of Ewing's sarcoma (associated with EWS-FLI1 fusion protein), mesothelioma (associated with FUS-ATF1 fusion protein), malignant melanoma (associated with EWS-ATF1 fusion protein), desmoplastic small round cell tumor (associated with EWS-ERG fusion protein) and leukemia (associated with FUS-ERG fusion protein). Accordingly, in step 101, various FET-ETS fusion proteins can also be used as target fusion proteins. Accordingly, the candidate drugs for the target tumor disorder obtained by the screening method include at least one of the following:

LY2835219 (Abetamoxiclib mesylate), the structural formula of which is

CHIR-124, whose structural formula is

EMD-1214063, whose structural formula is

CX-6258, whose structural formula is

Pelitinib (EKB-569), whose structural formula is

AZD-9291, whose structural formula is

GDC-0941, whose structural formula is

SB743921, whose structural formula is

These are also the eight small molecule compounds shown in Figures 11(a) to 11(j) that significantly reduce the number of aggregates. As mentioned above, Figures 11(a) to 11(j) are candidate drugs obtained by screening using FET-ETS fusion proteins (for example, using U2OS/mCherry-FUS-ERGmut cells).

It can be concluded that LY2835219 ranks first among the hit compounds (see Figure 4(e)). LY2835219 effectively dissolves aggregates in reporter cell lines expressing Ewing sarcoma fusions and partially rescues the abnormal expression of target genes. It can be used as a candidate drug for FET fusion-related Ewing sarcoma.

Therapeutic mechanism of LY2835219

Considering that LY2835219 is a CDK4/6 inhibitor, other CDK4/6 inhibitors were searched in the small molecule library and more than four inhibitors were found. Among them, purvalanol B, palbociclib isesionate and LEE011 failed to dissolve aggregates, while ON123300 caused massive cell death (see Figure 12).

These data indicate that LY2835219 does not rely on CDK4//6 inhibitor activity to induce aggregate dissolution. Next, LY2835219 was tested in a concentration range of 1 to 10 μM. The amount of aggregates decreased in a dose-dependent manner over time (see Figure 13 (b) and Figure 14 (a)-Figure 14 (c)).

It can be noted that cells treated with LY2835219 contain many vacuolar structures (see Figure 13(a)). A previous study reported that lysosomes swell to form vacuolar structures in cells treated with LY2835219. To confirm this, the reporter cell line was stained green using the lysosome tracking method before and after LY2835219 treatment. Consistent with the previous study, vacuolar structures co-localized with lysosomes (see Figure 13(c)), and the intensity of lysosomes was significantly enhanced (see Figure 13(c) and Figure 13(d)). These results are consistent with the concept that LY2835219 treatment leads to lysosomal acidification.

In addition, U2OS cells also showed a significant increase in lysosomal intensity in response to LY2835219 (Figure 14(b)). Therefore, a specific inhibitor of lysosomes, bafilomycin A1 (Baf-A1), was co-added with LY2835219 to assess whether Baf-A1 reduces the activity of LY28305219. High-content imaging of cells was performed during treatment with LY2835219 or Baf-A1 alone or both, and it was noted that no vacuoles were formed in cells cultured with Baf-A1 and LY2875219 (see Figure 13(e)). In addition, Baf-A1 significantly reduced the dissolution rate of LY2835219 (see Figure 13(e)).

Further, the lysosomal reporter gene YFP-LAMP1 (lysosomal associated membrane protein 1) was introduced into the fusion cell line and imaged for a long time under small molecule treatment. In this experiment, it was found that before small molecule treatment, the aggregates of mCherry-FUS-ERGmut were highly colocalized with lysosomes. LY2835219 treatment increased this colocalization, dissolved the aggregates, and significantly enlarged the lysosomes (see Figure 13 (f)). In summary, these data indicate that LY2835219 may partially dissolve the fusion protein aggregates in the cell by activating lysosomal acidification.

Next, we tested the effect of LY2835219 on aggregates generated by a number of phase separation prone proteins: EWS-FLI1, EWS-FLI1mut, NUP98-HOXA9mut, NUP98 HOXD13mut, the N terminus of FUS (aa 1-212), the N terminus of NUP98 (aa1-515), and full-length TDP43. LY2835219 treatment resulted in the dissolution of all four PS-DBD fusion aggregates (Figure 13(g)), but not the three non-fusion proteins (Figure 13(h)). Interestingly, LY2835219 showed specificity for PS-DBD fusion proteins in dissolving aggregates.

In addition to being able to effectively dissolve FET-ETS droplets, the inventors found that LY2835219-mediated droplet dissolution can reverse the aberrant expression of the fusion target gene.

The EWS-FLI1 fusion significantly enhanced the relative luciferase activity, and in the luciferase assay, EWS-FLI1 was more effective than FUS-ERG. Therefore, for further functional testing, luciferase activity was monitored using EWS-FLI1 and LY2835219 treatment. It was found that the relative luciferase activity decreased over time and decreased by 50% after 12 hours of treatment (Figure 15(a)).

RNA sequencing was performed on U2OS/mCherry-ESW-FLI1 cells before and after treatment with LY2835219 to verify whether LY2835219 treatment would reduce the expression of EWS-FLI1 target genes. Putative target genes of EWS-FLI1 were extracted from a previously published transcriptome dataset. In short, 356 genes whose expression levels in EWS-FLI overexpressing cells were significantly higher than those in FLI1 overexpressing cells and control cells were considered to be EWS-FLI1 target genes (Figure 15(b)). By comparing the expression levels of 356 EWS-FLI1 target genes before and after LY2835219 treatment, it was found that 26 target genes were significantly downregulated after 3 hours of treatment, and only 2 target genes were upregulated (Figure 15(c)). Further, the relative expression levels of cells expressing EWS-FLI1 and wild-type cells before and after LY2835219 treatment were calculated. Consistent with the above findings, the relative expression level of the EWS-FLI1 target gene was significantly reduced after LY2835219 treatment (see FIG. 15( d )). In addition, the inventors repeated the above experiment on the Ewing sarcoma cell line A-673 and reached a consistent conclusion.

Furthermore, real-time fluorescence quantitative polymerase chain reaction (RT-qPCR) was performed to verify the downregulation pattern of EWS-FLI1 target genes after LY2835219 treatment. After LY2835219 treatment, the expression levels of most tested genes decreased, and this process was time-dependent (see Figure 15(e)). Overall, the luciferase, RT-qPCR and RNA sequencing data jointly confirmed that LY2835219 can indeed rescue the abnormal gene expression induced by fusion oncoproteins and is a candidate drug for the treatment of these FET fusion-related Ewing sarcomas or other cancers.

In many cancer cells, the genome and transcriptome encode fusion proteins with PS and DBD. These fusion proteins can phase separate on chromatin and act as new transcription factors to alter the expression of downstream target genes. Abnormal transcriptomes are often associated with cancer. Phase separation regulating drugs can interfere with aggregates and reverse abnormal transcription patterns. These drugs are expected to be used as therapeutic drugs for fusion-related cancers. The mechanism of action can be seen in Figure 15(f): Under normal circumstances, genes are expressed normally; when chromosome rearrangements occur to produce in-frame PS-BDB fusions, phase separation of the target DNA site leads to abnormal gene expression, which is associated with cancer, and small molecules that dissolve phase-separated aggregates can rescue abnormal expression.

Computer program product

According to an embodiment of the present application, a computer program product is provided, which includes computer executable instructions that can be stored and/or downloaded to a non-transitory storage medium.

When the computer executable instruction is executed by the processor, the method for screening the medicine of tumor disease according to each embodiment of the present application is executed. The processor can be a processing device including one or more general processing devices (such as a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), etc.). More specifically, the processor can be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor running other instruction sets, or a processor running a combination of instruction sets. The processor can also be one or more special processing devices, such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a system on chip (SoC), etc. As will be understood by those skilled in the art, in some embodiments, the processor can be a special processor, rather than a general processor.

The following is a brief description of an example of the method. It should be noted that various examples of the method for screening drugs for tumor diseases according to various embodiments of the present application can be combined here and will not be repeated here.

For example, a method for screening drugs for tumor conditions executed by a processor may include the following steps.

Acquire time series images of several fields of view of each well of a well plate including a well array, wherein the time series images are generated by continuously photographing each well with several different fields of view within a predetermined time period after executing the following steps (these steps may not be executed by a processor, but represent the operating state of the well when the image is acquired): inoculate cells containing a target fusion protein into a well plate including a well array, wherein the cells contain a nucleic acid molecule encoding the target fusion protein, the target fusion protein is associated with a target tumor disorder, contains a phase separation prone domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to be able to form aggregates that can be identified by imaging; distribute a plurality of small molecule compounds into each well.

Based on the time series images of at least one field of view of each well, the amount decay status of the intracellular aggregates in each well is analyzed.

Wells whose attenuation conditions of the amount of intracellular aggregates meet predetermined conditions are screened, and small molecule compounds assigned to the screened wells are determined as candidate drugs for the target tumor disorder.

In some embodiments, analyzing the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well specifically includes: for each well: for the images of each sampling time point of the time series images of each field of view, performing cell and aggregate segmentation, and determining the amount of aggregates of each cell; for the images of each time point of each field of view, determining the proportion of cells whose aggregate amount exceeds the enriched aggregate threshold, and averaging based on the field of view as the aggregate-rich cell fraction of the well at each sampling time point, thereby obtaining a distribution curve of the aggregate-rich cell fraction of the well within the predetermined time period; fitting an exponential model based on the distribution curve of the aggregate-rich cell fraction of the well within the predetermined time period; and calculating the fitting effect parameter and the attenuation ratio of the aggregate-rich cell fraction based on the fitted exponential model, wherein the predetermined condition includes that the attenuation ratio of the aggregate-rich cell fraction reaches a first predetermined condition and the fitting effect parameter is better than a second threshold.

In some embodiments, the fitting effect parameter is R squared, and the second threshold is 0.2-1.0.

In some embodiments, the first predetermined condition is the first predetermined number of cells whose attenuation ratio reaches 0.2 and are arranged in order of attenuation ratio, and the attenuation ratio is defined as the ratio of the aggregate-rich cell fraction after attenuation calculated using a fitted exponential model to the initial aggregate-rich cell fraction.

In some embodiments, the method for screening drugs for tumor conditions further comprises: automatically identifying the survival status of cells in each well, determining wells whose survival status is worse than a predetermined survival condition; and excluding the small molecule compound assigned to the well from candidate drugs for the target tumor condition.

In some embodiments, automatically identifying the survival status of cells in each well specifically includes: using an image processing algorithm to determine at least one of the cell morphology and the rate of reduction in cell number; when the deviation of the identified cell morphology from a spherical shape is higher than a deviation threshold and/or the rate of reduction in cell number is higher than a reduction rate threshold, identifying the cell as having a survival status worse than a predetermined survival condition.

In some embodiments, the exponential model is defined by the following formula (1):

where _c0 and _c1 are constants to be regressed in the exponential model, t is the exposure time of the small molecule compound, and Y(t) is the fraction of aggregate-rich cells in each well.

In some embodiments, performing segmentation of cells and aggregates on each image specifically includes: using a first algorithm to segment cells; identifying point structures in the image and performing enhancement with the representative size of the aggregates as the target size; using a second algorithm to segment aggregates whose intensity values are higher than the representative intensity value of the background by a third threshold.

In some embodiments, the first algorithm automatically determines an intensity threshold for segmenting cell foreground and cell background and performs cell segmentation accordingly, the second algorithm automatically determines a representative intensity value for the robustness of the aggregate background, and the third threshold is 3-5 times the standard deviation of the representative intensity value of the background.

In some embodiments, the first algorithm comprises an Otsu algorithm and the second algorithm comprises a robust background algorithm.

High-content imaging analysis system

According to an embodiment of the present application, a high-content imaging analysis system is provided, comprising: an automatic high-speed microscopic imaging component, which is configured to continuously photograph each well of a well plate including a well array with several different fields of view within a predetermined time period to generate time series images of several fields of view of each well after the following steps are performed: inoculating cells containing a target fusion protein into a well plate including a well array, wherein the cells contain a nucleic acid molecule encoding the target fusion protein, the target fusion protein is associated with a target tumor condition, contains a phase separation prone domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to be able to form aggregates that can be identified by imaging; dispensing a plurality of small molecule compounds into each well; and at least one processor, which is configured to execute a method for screening drugs for tumor conditions according to various embodiments of the present application.

Various examples of methods for screening drugs for tumor diseases according to various embodiments of the present application can be combined here and will not be described in detail here.

Method for analyzing the phase separation of target fusion protein

According to an embodiment of the present application, a method for analyzing the phase separation status of a target fusion protein is provided. The analysis method may include the following steps.

An image of a cell containing the target fusion protein is acquired, wherein the cell contains a nucleic acid molecule encoding the target fusion protein, and the target fusion protein further contains a labeling molecule so as to form aggregates that can be identified by imaging.

The amount of intracellular aggregates in the image is identified, and the phase separation status of the PS-DBD fusion protein in vivo and/or in vitro is determined based on the amount of the identified aggregates. The larger the amount of aggregates, the more serious the phase separation status.

In some embodiments, the aggregates are in the form of droplets or gels.

In some embodiments, the DBD of the target fusion protein comprises mutations such that the imaging recognition of the aggregates formed is more significant than the imaging recognition of aggregates formed when the DBD does not comprise the mutations.

The phase separation status of the target fusion protein and its association mechanism with aggregates have been described in detail in the above section "The association mechanism between PS-DBD fusion protein and tumor disease status: phase separation, aggregate formation and abnormal gene expression " and will not be repeated here.

In some embodiments, the mutation is a point mutation, and is introduced into a site having DNA binding activity in the DBD of the target fusion protein to be analyzed. The introduction of the mutation into the DBD has been described in detail in the above section " Introducing mutations into the DBD of the target fusion protein ", which will not be repeated here.

In some embodiments, the site with DNA binding activity in the DBD of the target fusion protein to be analyzed is determined based on the information on DNA binding activity in the Swiss-Prot database. Specifically, if the Swiss-Prot database reports that a mutation at a site in the region where the domain of the target protein is located affects its ability to bind to DNA, or a mutation at a site in the same domain of other proteins affects its ability to bind to DNA, and the site can be sequence aligned to the target protein, such a site is considered to be a site with DNA binding activity in the DBD.

Example

Hereinafter, the present application will be described in detail by way of examples. However, the examples provided herein are only for illustrative purposes and are not intended to limit the present application.

Unless otherwise specified, the experimental methods used in the following examples are all conventional methods.

Unless otherwise specified, the materials and reagents used in the following examples can be obtained from commercial sources.

The scientific instruments involved in the embodiments are as follows:

Pipette (EPPENDORF);

DNA electrophoresis device (Six One Biological Company);

DNA gel electrophoresis imaging system (Liuyi Biotechnology Co., Ltd.);

High temperature and high pressure sterilizer (ZEALWAY);

Electronic analytical balance (Shanghai Youke Company);

UV spectrophotometer (Beijing Junyi Dongfang Company);

Nanodrop (ThermoFisher Scientific);

PCR nucleic acid amplifier (BIO-RAD);

Real-time fluorescence quantitative PCR instrument (BIO-RAD)

Electric constant temperature water bath (Beijing Yongguangming Company);

Metal bath (Dalong Xingchuang Company);

Clean bench (Shanghai Zhicheng Company);

Tabletop refrigerated centrifuge (BECKMAN);

Low temperature refrigerated centrifuge (EPPENDORF);

Ultra-low temperature freezer (ThermoFisher Scientific);

protein electrophoresis apparatus (BIO-RAD);

Protein semi-dry transfer instrument (BIO-RAD);

Biological safety cabinet (ThermoFisher Scientific);

Flow cytometer (BD Biosciences);

Cell culture incubator (ThermoFisher Scientific);

Bacterial incubator (Zhichu Instrument Co., Ltd.);

Ultrasonic cell disruptor (ASONE);

High pressure cell disruptor (ATS);

Protein purification system AKTA (GE Healthcare);

Nickel ion affinity chromatography column Ni-NTA resin (Genscript);

HiTrap Q HP (GE Healthcare);

Superdex 200 Increase 10/300 chromatography molecular sieve column SD200 (GE Healthcare);

Fluorescence inverted microscope (NIKON);

High-resolution confocal microscope (NIKON A1R HD25);

Super-resolution confocal microscope A1/SIM/STORM (NIKON);

Echo550 nanoliter acoustic pipetting system (Labcyte);

Laser confocal high-content imaging microscope Opera Phenix (PerkinElmer);

Multifunctional microplate reader EnVision (PerkinElmer).

The reagents involved in this embodiment are as follows:

FastPfu polymerase (TransGen Biotech), dNTPs solution (TransGen Biotech), restriction endonucleases Age I-HF, Spe I-HF, Nco I-HF, Xho I-HF (NEB), HiFi DNA Assembly Master Mix (NEB), agarose (Biowest), DNA molecular weight Trans DNA Marker (TransGen Biotech), DNA electrophoresis TAE buffer (Prile), DNA electrophoresis loading buffer (TransGen Biotech), plasmid extraction kit (CWBIO), DNA agarose gel recovery kit (CWBIO), large-scale DNA product purification kit (TIANGENBIOTECH), 4S Green nucleic acid dye (Sangon Biotech).

RNA extraction reagent Trizol (ThermoFisher Scientific), reverse transcription reagent One-StepgDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech), qPCR reagent Transgen Green qPCR SuperMix (TransGen Biotech).

NaCl (Beijing Chemical Industry), MgCl ₂ (Sinopharm Group), KCl (Sinopharm Group), sodium phosphate (Sinopharm Group), hydrochloric acid (Sinopharm Group), Na ₂ HPO ₄ (Sinopharm Group), NaH ₂ PO ₄ (Sinopharm Group), calcium phosphate (Sigma), Tris (Sigma), glycine (Sigma), glycerol (Sigma), imidazole (Sigma), HEPES (Sigma), ethanol (Sigma), IPTG (Inalco), EDTA (Sigma), ammonium acetate (Sigma), sodium acetate (Amresco), glacial acetic acid (Beijing Chemical Plant), isopropanol (Beijing Chemical Plant), dithiothreitol DTT (Inaclo), maltose (Amresco), guanidine hydrochloride (MPBiomedicals), 3-morpholinepropanesulfonic acid MOPS (Solarbio), sodium dodecyl sulfate SDS (Amresco), Ni-NTA Resin (ThermoFisher Scientific), polyacrylamide precast gel (Genscript), three-color prestained protein molecular weight standard Marker (YEASEN), Coomassie brilliant blue staining solution (Genscript), Coomassie brilliant blue destaining solution (Genscript), skim milk powder (ThermoFisher Scientific), Tween-20 (Amresco), Nonidet P40 (Amresco), β-mercaptoethanol (Amresco), PMSF protease inhibitor (Sigma), protease inhibitor cocktail (Beyotime), phosphate buffered saline (PBS) (HyClone), Coomassie brilliant blue R250 (MACKLIN).

Example 1: Establishment of stable cell line structure and related experimental procedures

Cell culture: The cells HEK293T, Hela, U2OS and A-673 involved in this article all used high-glucose DMEM medium (HyClone, SH30243.01), added with 10% fetal bovine serum FBS (Gibco, 10099-141), 1% penicillin-streptomycin double antibody (Gibco, 15140122), and cultured in a constant temperature and moisturizing incubator at 37°C and 5% CO _2. When performing cell screening and culturing stable cell lines, the culture medium is additionally added with puromycin or corresponding antibiotics at a final concentration of 1 μg/ml. When the cells grow to a density of 85%, trypsin Trypsin is used for digestion and passage. All operations on cells expressing OptoIDR structures were performed under red light to avoid ambient light exposure.

The original backbone plasmid involved in the plasmid construction of this example is as follows

(1)pCMV-mCherry-Age I-FUS-ERG-Age I

The structure of pCMV-mCherry-Age I-FUS-ERG-Age I is shown in Figure 17. As shown in Figure 17, the backbone plasmid pCMV-mCherry-Age I-FUS-ERG-Age I is a lentivirus expression vector, prokaryotic cells are ampicillin resistant, and eukaryotic cells are puromycin resistant. The target gene can be inserted into the vector linearized by endonuclease Age I, and the mCherry fluorescent label carried on the vector can trace the target protein. The plasmid can be used to package lentivirus, construct a stable cell line, and can also be transiently transfected to detect the expression status of the target protein.

The full-length sequence of pCMV-mCherry-Age I-FUS-ERG-Age I is shown in SEQ ID NO: 3.

(2)pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig

The structure of pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig is shown in Figure 18. As shown in Figure 18, the backbone plasmid pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig is modified from Addgene plasmid #10122 (Shin et al., 2017) and is a eukaryotic cell expression vector. The target gene can be inserted into the vector linearized by endonuclease Spe I, and the mCherry-Cry2olig element carried on the vector can be used in the OptoIDR experiment. Through blue light stimulation, the state of the fusion protein is observed to evaluate whether the target protein has phase separation potential.

The full-length sequence of pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig is shown in SEQ ID NO:4.

(3) pETDuet vector

The structure of pETDuet vector is shown in Figure 19. As shown in Figure 19, the backbone plasmid pETDuet (His tag-MBP-TEV) is an Escherichia coli prokaryotic expression plasmid, ampicillin resistance, purchased from Novagen. The target gene can be inserted into the vector linearized with restriction endonucleases Nco I and Xho I, and connected after the recombinant tobacco etch virus protease (Tobacco Etch Virus, TEV) sequence. The plasmid can also be further transformed by inserting enhanced green fluorescent protein (Enhanced Green Fluorescent Protein, EGFP) after the TEV protease site, and the target gene is then connected after EGFP. The plasmid can obtain a fusion protein after induction expression by isopropyl-β-D-thiogalactoside (IPTG) in Escherichia coli for in vitro phase separation experiments.

(4) pRSFDual vector

The structure of the pRSFDual vector is shown in Figure 20. As shown in Figure 20, the backbone plasmid pRSFDual (His tag-EGFP) is modified from the original vector pRSFDual, and EGFP is connected after the histidine tag His tag. The plasmid is an E. coli prokaryotic expression vector, kanamycin resistant, purchased from Novagen. The target gene can be inserted into the recombinant vector linearized with the restriction endonuclease XhoI, and the target protein fused with His tag-EGFP can be expressed in E. coli after IPTG induction for in vitro phase separation research.

The construction process of the lentiviral transfer vector encoding the fusion protein is as follows:

The original lentiviral vectors pCMV-mCherry-Age I-FUS-ERG-Age I and pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig involved in this study were from the laboratory of Professor Li Pilong of Huazhong University. The genes involved and their mutations such as EWS-FLI1/EWS-FLI1mut or EWS IDR were synthesized by Genewise (which has been acquired by Ascent). The construction process first linearizes the pCMV-mCherry-Age I-FUS-ERG-Age I vector with restriction endonuclease Age I-HF (NEB, R3552S), and linearizes the pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig vector with endonuclease Spe I-HF (NEB, R3133S), and purifies the linearized vector after gel electrophoresis; then, PCR is used to obtain the target gene fragment such as EWS-FLI1/EWS-FLI1mut or EWS IDR, and the target fragment is purified after gel electrophoresis; then, the recombinant reagent is used according to the product instructions. HiFi DNA Assembly Master Mix (NEB, E2621L) was used to recombinate the linearized vector and the target fragment. Fusion genes such as EWS-FLI1/EWS-FLI1mut were recombined with the linearized vector pCMV-mCherry-AgeI-FUS-ERG-Age I. Fusion genes such as EWS IDR were recombined with the linearized vector pHR-Spe I-FUS IDR-Spe I-mCherry-Cry2olig. Finally, the recombinant products were transformed into competent Stbl3 (Kangti Life Science, KTSM110L), and after recovery culture, they were evenly spread on plates with corresponding resistance. After overnight culture at 37°C, single clones were picked for sequencing and identification.

The specific process of cell transfection is as follows:

According to the experimental requirements, cells are inoculated into culture plates or culture dishes with different pore sizes and materials, and transfection experiments are performed when the cell density reaches 50-80%. The specific experimental method is as follows (taking Lipo8000 transfection reagent as an example):

1) Before transfection, cells are seeded in a culture dish (a 6-well plate is used as an example below) at an appropriate density according to cell type, size, and growth rate, and transfection experiments are performed when the cell density reaches about 75%;

2) Before transfection, replace the culture medium with 2 ml of pre-warmed fresh culture medium;

3) Take a clean and sterile centrifuge tube, add 125 μL Opti-MEM (Gibco, 31985062) medium, add 2.5 μg endotoxin-free plasmid DNA, and mix them evenly with a pipette; then add 4 μL Lipo8000 transfection reagent and mix them evenly with a pipette;

4) (Optional) Leave at room temperature for 10-20 minutes;

5) Add the transfection reagent-DNA mixture evenly into the culture dish and mix gently;

6) Continue culturing for 24-48 hours, and then detect the transfection effect by appropriate methods, such as Western blot, flow cytometry FACS, fluorescence detection, etc., or add appropriate antibiotics to screen stable transfected cell lines.

The stable cell lines constructed in this example are all stable cell lines infected with lentivirus, and the plasmids used are two backbone plasmids pMD2.G (Addgene, #12259) and psPAX2 (Addgene, #12260).

The method for constructing a stable cell line is as follows (taking the operation of transfection reagent PEI in a 6 cm culture dish as an example):

1) Inoculate HEK293T cells into a 6 cm culture dish and start transfection when the cell density reaches about 50%;

2) Before transfection, replace the culture medium with 6 ml of pre-warmed culture medium without antibiotics;

3) Prepare the transfection mixture: add 4 μg of lentiviral plasmid containing the target gene, 2.4 μg psPAX2 and 1.6 μg pMD2.G to 800 μL Opti-MEM, mix well, add 18 μL transfection reagent PEI, mix gently, and incubate at room temperature for 20 minutes;

4) Add the transfection mixture evenly to the culture dish, mix gently and place in a cell culture incubator for culture;

5) 8-12 hours after transfection, replace the culture medium with fresh medium and add sodium pyruvate and 1 mM non-essential amino acids at a final concentration of 1 mM;

6) Every 12-18 hours thereafter, collect the virus liquid from the culture medium supernatant, filter it with a 0.45 μm syringe filter, and add it to the cells to be infected (e.g., U2OS) (density of about 50%). The HEK293T cells that packaged the virus continue to be cultured with fresh culture medium supplemented with sodium pyruvate and non-essential amino acids; the virus supernatant can be collected continuously for 3-5 times;

7) After collecting the viral supernatant and infecting the cells (e.g., U2OS) multiple times, replace with fresh culture medium and continue culturing for 24 hours;

8) Based on the corresponding antibiotic labeling or fluorescent labeling, flow cytometry or continuous antibiotic screening can be used to enrich stable cell lines.

The U2OS/mCherry-FUS-ERGmut stable cell line used in this example is a monoclonal cell line. After obtaining a stable cell line pool, 192 monoclonal cells were flow-sorted and the monoclonal cell line was selected based on the imaging effect after expanded culture. The other cell lines involved are all non-monoclonal stable cell lines. The specific construction process is to replace the target plasmid in the third step of this process, and the rest of the process is consistent.

The sequence information of all inserted genes involved in this embodiment is provided in Table 1 below.

Table 1

The plasmid information related to prokaryotic protein expression purification and mammalian cell transfection involved in this embodiment is shown in Table 2 below.

Table 2 Plasmid information related to this example

The live cell staining involved in this embodiment mainly includes the staining of cell nuclei with Hoechst and the staining of lysosomes with LysotrackerGreen.

When staining the nuclei of living cells, Hoechst 33342 nucleic acid dye (Invitrogen, H3570) can be used for staining. Hoechst can penetrate the cell membrane and emit blue fluorescence after binding to dsDNA. The specific experimental method is as follows:

1) Inoculate the cells into a suitable culture dish for culture;

2) Dilute Hoechst 33342 staining solution 1:2,000 with PBS to prepare staining working solution;

3) Aspirate the culture medium and add the staining working solution;

4) Incubate in dark for 5-10 minutes;

5) Aspirate the staining solution and gently rinse the cells three times with PBS;

6) Add culture medium, observe cell staining under a microscope, and keep images;

When staining lysosomes in living cells, Lysotracker Green (Invitrogen, L7526) can be used for staining. Lysotracker Green is a green dye that can penetrate the cell membrane and can stain the acidic compartments in living cells. The maximum excitation/emission wavelengths are approximately 504/511 nm, respectively. The specific experimental method is as follows:

1) Take a small amount of Lysotracker Green and add it to the cell culture medium at a ratio of 1:13,333-1:20,000 for use. The final concentration is 50-75nM and pre-incubate at 37°C.

2) Remove the cell culture medium, add the pre-incubated Lysotracker Green staining solution, and incubate with the cells at 37°C for 30 minutes;

3) Remove the Lysotracker Green staining solution and rinse the cells three times with PBS;

4) Add fresh culture medium, perform fluorescence observation, and keep the image.

The reagents used in the dual fluorescein reporter experiment involved in this example are Dual- Reporter Assay System 10-Pack (Promega, E6921). The specific experimental steps are as follows:

1) Take one well of a 24-well plate as an example, transfect HEK293T cells, the transfection reagent is Lipo8000, 25μL Opti-MEM + 1.2μL Lipo8000 + 606ng plasmid per well, the plasmid components are as follows:

300 μg mCherry-EWS-FLI1

300 μg 25x GGAA-promoter firefly luciferase (obtained from Qi Zhi's laboratory, School of Life Sciences, Peking University)

6μg Renilla luciferase---internal reference (obtained from Qi Zhi's laboratory, School of Life Sciences, Peking University)

The control group was transfected with a plasmid without fusion protein;

2) 24 hours after transfection, carefully rinse the transfected cells with PBS;

3) Prepare 1x Passive lysis buffer (PLB) and dilute 5x PLB stock solution with sterile water; store the 5x PLB stock solution in aliquots at -80°C;

4) Add 100 μL 1x PLB to the washed transfected cells and fully lyse for about 15 minutes. Collect the lysate and centrifuge to remove the supernatant for later use.

5) Preparation of Luciferase Assay Reagent II (LAR II): Add Luciferase Assay Buffer II to the freeze-dried Luciferase Assay Substrate, dissolve thoroughly, and store in aliquots at -80°C.

6) Aliquot 50x Stop& Substrate (SGS), stored in -80℃ freezer;

7) Stop& Buffer (SGB), stored in -80℃ refrigerator;

8) Add one volume of SGS to 50 volumes of SGB to prepare Stop & Reagent(SGR);

9) Add 100 μL of substrate LAR II to the dedicated detection 96 plate, and then add 20 μL of cell lysis buffer and mix well;

10) Place the test plate in a multifunctional microplate reader for testing. The detailed procedure is to perform two 10-second readings after a 2-second delay. This reading is the Firefly luciferase reading;

11) Take out the test plate, add 100 μL of substrate SGR and mix thoroughly, then put it into the multifunctional microplate reader for detection again. This reading is the Renilla luciferase reading.

Each sample was replicated three times and each experiment was repeated three times.

When calculating, the reading of Firefly luciferase is corrected by using the reading of Renilla luciferase, and then compared with the control group without transfection of fusion protein to calculate the ratio. This ratio represents the transcription enhancement ability of the fusion protein as a transcription factor on the downstream target gene.

The protein expression and purification process involved in this embodiment is as follows:

(1) Prokaryotic expression of proteins

1) The vector containing the target gene is transformed into BL21 competent cells and cultured overnight after plating;

2) Pick a single clone on the LB plate, add it to 5 ml of LB liquid medium containing the corresponding antibiotic, and culture it in a shaker at 37°C and 220 rpm for 4 hours;

3) Transfer 5 ml of the cultured bacterial solution to 200 ml of fresh LB liquid medium containing antibiotics and continue culturing at 37°C and 200 rpm for 3 hours;

4) Transfer 200 ml of the cultured bacterial liquid to 800 ml of fresh LB liquid medium containing antibiotics and continue culturing at 37°C and 200 rpm;

5) When the bacteria enter the logarithmic growth phase (OD600 = 0.8), add the inducer IPTG to a final concentration of 0.5 mM

6) Incubate overnight at 16°C and 200 rpm for 16 hours.

(2) Collecting bacteria and lysing

1) Collect the bacteria by centrifugation at 4,392 g for 30 minutes at 4°C in a high-speed centrifuge;

2) Remove the supernatant and add 30 ml of cell lysis buffer per liter of bacterial liquid to resuspend;

3) Ultrasonic disruption of the bacteria in an ice bath for 30 minutes, under the following conditions: ultrasonic power 50 W, ultrasonic frequency 20 kHz, cycle of 2 seconds for ultrasonic disruption and 4 seconds for cooling.

(3) Protein Ni-NTA initial purification

1) Centrifuge at 47,850 g for 30 minutes at 4°C to collect the bacterial supernatant;

2) Equilibrate Ni-NTA resin with 30 ml cell lysate;

3) Add bacterial lysate supernatant to the equilibrated Ni-NTA resin, mix gently and incubate at 4°C for 30 minutes to allow the protein to fully bind to Ni;

4) Allow the lysate to pass through the Ni-NTA resin by gravity at 4°C and collect the flow-through;

5) Wash the Ni-NTA resin with 30 ml of cell lysis buffer at 4°C to elute impurities;

6) Wash the Ni-NTA resin with 30 ml of nickel ion column affinity chromatography wash buffer at 4°C to elute impurities and non-specifically bound proteins;

7) Elute the Ni-NTA resin with 30 ml of nickel ion column affinity chromatography elution buffer at 4°C;

8) Collect the eluate and use SDS-PAGE to detect protein expression.

(4) Secondary protein purification

1) Dilute the eluate with 120 ml of protein diluent;

2) Purify the protein using a HiTrap Q HP anion exchange column, using a gradient elution of protein diluent (from 100% to 0%) and a high salt solution (from 0% to 100%);

3) Collect the eluate and identify the target protein using SDS-PAGE;

4) Ultrafiltration and concentration of high purity (>95%) target protein eluate;

5) using SD200 and gel filtration buffer to further purify the protein;

6) Collect the eluate and identify the target protein using SDS-PAGE;

7) Ultrafiltration concentrates the high-purity (>95%) target protein and measures the protein concentration;

8) The concentrated protein was quickly frozen with liquid nitrogen and stored at -80°C for future use.

The protein fluorescent labeling involved in this embodiment is as follows:

Alexa Fluor ^TM 488 NHS Ester is a bright green fluorescent dye that is water-soluble and pH-stable. It can be excited at 488nm to produce stable fluorescence signals. Alexa Fluor ^TM 488 NHS Ester can form stable amide bonds with amino groups on proteins, mainly primary amines.

The small molecule dye is dissolved in DMSO at a concentration of 1 mg/ml. When used, the small molecule dye and the target protein are mixed at a molar ratio of 1:1 and incubated at room temperature for 1 hour. Centrifuge the MicroSpin G-50 separation column at 5,000 rpm for 10 minutes, discard the pre-loaded equilibrium solution in the column, add the labeled protein to the separation column, and centrifuge at 5,000 rpm for 10 minutes. The labeled protein solution is in a centrifuge tube.

The DNA fluorescent labeling process involved in this embodiment is as follows:

Cy5 is a far-red dye with an excitation wavelength of 633nm or 647nm. The significant advantage of this dye is its low autofluorescence in biological samples in this spectral region.

In this paper, when synthesizing primers, the 5' end of the forward primer is labeled with Cy5, followed by normal PCR amplification, purification of the DNA product to obtain the labeled DNA fragment, and dilution to the required concentration for use.

The 306 bp DNA sequence (SEQ ID NO:47) incubated with the FUS-ERG fusion protein in vitro is as follows: CTCGACTAGGTTTTCCTTATGCTGAGAATTCCAGGTCCTGGAAGAAGAAAAAGAGAAAGAAAGAGAGAGAAGGAGTGAGAGGGAGGGAGGGAGGGAGGGAGGGA GGAAGGAAGGAAGGAAGGAAGGAAGGAAAGGAAGGAAGGAAGGAAGG AAGGAAGGAAAGGAAGGAAGGAAGGA AGGAAGGAAGGAAGGAAGGAAGGAA AAGAAACAGCAAAAAAAGAAAGAGGGAGGATGGGAGGGAGGGAAAAAGTAAAAATGATTCTGTATCAGCTGGTATATACCAACAACTAGC, the underlined part is the 25x GGAA sequence.

The in vitro protein phase separation experimental process involved in this embodiment is as follows:

Protein in vitro phase separation experiments were all performed in 384-well low-adsorption multiwell plates (Cellvis, P384-1.5H-N). For proteins with MBP tags, 1 μL of 1 mg/ml TEV protease was directly added to 10 μL of the protein solution to be tested and incubated at room temperature for half an hour.

For the experiment to detect the interaction between FUS-ERG and 25x GGAA double-stranded DNA, 50 nM of DNA fragments were added to the FUS-ERG phase-separated droplets, and observation began after incubation for 5 minutes.

The imaging of protein in vitro phase separation experiments was performed using a Nikon A1R HD25 high-resolution confocal microscope.

The fluorescence photobleaching recovery experiment process involved in this embodiment is as follows:

Fluorescence photobleaching recovery experiment FRAP, using Nikon A1R HD25 high-resolution confocal microscope, bleaching experiment with the same excitation light as the luminescent group carried by the target protein or DNA, that is, 488nm was used to bleach the protein with EGFP label, and 647nm was used to bleach the Cy5 labeled DNA. Each group of experiments was repeated at least three times, with a laser intensity of 50% and a bleaching time of 1 second. Images were collected at different intervals according to the recovery speed until the fluorescence intensity curve no longer changed. The fluorescence intensity change of the phase change droplet in the unbleached area was used to correct the bleaching effect that occurred naturally during the imaging process, and the fluorescence intensity change of the background part without phase change droplet was used to subtract the background fluorescence intensity. Data processing was performed using NIS-Elements AR Analysis, ImageJ and GraphPadPrism 8.0.

The cell fluorescence imaging process involved in this embodiment is as follows:

Live cell imaging can be performed 24 hours after cell transfection. The imaging culture dish used is a 35mm glass-bottomed quadrant (In Vitro Scientific, D35C4-20-1-N). Use a NIKON A1R HD25 high-resolution confocal microscope or a NIKON super-resolution confocal microscope A1/SIM/STORM, and image the live cell working platform at 37°C with constant temperature and humidity and 5% CO _2. Image processing uses NIS-Elements AR Analysis.

The OptoIDR experiment was optimized from the literature (Shin et al., 2017). This experiment uses the NIKON super-resolution confocal microscope A1/SIM/STORM. For the IDR protein to be detected, the image before processing is first collected at 561nm, and then stimulated once with a 488nm laser point scan with a fluorescence intensity of 0.5-3% according to the phase separation ability of the detected protein, and then an image is collected every 3 seconds at 561nm for about 3 minutes. This experiment needs to be kept away from light throughout the process, and each group of proteins is repeated at least twice.

During quantitative processing, the Identify Primary Objects module of CellProfiler software was used to identify phase-separated droplets with a diameter of 5-20 pixels in the image. For the identified droplet area, this paper used CellProfiler to measure the fluorescence intensity of this area and the total fluorescence intensity of the entire image, and then used the ratio of these two intensities as a measure of protein phase separation in the image, thereby measuring the phase separation ability of the protein in a time series.

The DropScan experimental process involved in this embodiment is as follows:

This application builds a DropScan experimental method to discover small molecule compounds that affect protein phase separation.

The specific operations are as follows:

The constructed monoclonal cell line U2OS/mCherry-FUS-ERGmut was inoculated into the CellCarrier-384Ultra microplate (PerkinElmer) at a density of 7,000-8,000 per well, and then cultured overnight. The wells without cells were supplemented with an equal volume of sterile water or PBS to slow down the evaporation of the edge wells, which may cause inaccurate volume and affect the experimental results. The next day, the prepared small molecule compounds were added to the cells at a final concentration of 10μM using the Echo550 nanoliter acoustic pipetting system (Labcyte), and then high-content imaging was performed immediately using a laser confocal high-content imaging microscope. Two fields of view were collected for each well under a 40x water microscope, and photos were taken one by one. A new cycle was started immediately after one round, and the photos were taken for about 6 hours. For each field of view, it was a long-term photo for up to 6 hours.

Small molecule information

The small molecule library used for intracellular screening in this example is APExBIO Anti-Cancer-Compound-Library Plus, which is ordered and managed by the Pharmaceutical Experiment Platform of Tsinghua University. The concentration of the small molecule in the stock solution is 10 millimolar, and the solvent is dimethyl sulfoxide (DMSO). Unless otherwise specified, the final concentration of the small molecule in the experiment is 10 micromolar.

In the subsequent small molecule verification experiments, ten small molecules, LY2835219, CX-6258, EMD-1214063, Pelitinib (EKB-569), GDC-0941, AZD-9291, UNC 0631, CHIR-124, SB743921 and Ursolicacid, were all taken from the APExBIO Anti-Cancer-Compound-Library Plus molecular library of Tsinghua University's School of Pharmacy.

In the experiment proving that the function of LY2835219 is independent of the activity of CDK4/6 inhibitors, the small molecule compounds involved were all from the small molecule library APExBIO Anti-Cancer-Compound-Library Plus, and the cell images treated with small molecules during screening were directly used to display the results.

When investigating the potential mechanism of LY2835219 function, the lysosomal inhibitor Baf-A1 was used at a final concentration of 200 nanomolar.

The protein immunoblotting experiment (Western blot) process involved in this embodiment is as follows:

The total protein of cells was extracted using the Minute ^TM Total Protein Extraction Kit for Animal Cultured Cells/Tissues (Invent). The total protein can be directly detected by downstream Western blot.

The specific steps of Western blot are as follows:

1) Electrophoresis: Use pre-made gradient gel. After protein loading, run at 110V for about 1.5 hours to perform immunoblotting detection according to the size of the detected protein.

2) Transfer: This experiment is semi-dry transfer. Both the nitrocellulose membrane (NC membrane) and the transfer filter paper need to be fully soaked in freshly prepared pre-cooled transfer solution. Then the transfer sandwich from bottom to top is the filter paper, NC membrane, protein gel and filter paper. Air should be removed as much as possible to avoid affecting the transfer effect. The transfer conditions are constant current 250mA, 65 minutes;

3) Blocking: After the transfer, cut the membrane into a suitable size, put it into PBST solution containing 5% skimmed milk powder, and block it at room temperature on a shaker for about 1.5 hours;

4) Primary antibody: After blocking, rinse the NC membrane three times with PBST solution on a shaker for 5 minutes each time, then place the NC membrane in the primary antibody solution and incubate overnight at 4°C;

5) Secondary antibody: After the primary antibody is completed, the NC membrane is rinsed three times with PBST solution on a shaker for 5 minutes each time, and then the NC membrane is placed in the secondary antibody solution and incubated on a shaker at room temperature for 1 hour;

6) Color development: After the secondary antibody is completed, rinse the NC membrane with PBST solution on a shaker three times, 5 minutes each time, place the membrane in an imager, evenly add the color development solution, and then start imaging.

The RT-qPCR experimental process involved in this embodiment is as follows:

In this embodiment, RNA extraction was performed using the Trizol method, and the specific steps were as follows:

1) Rinse the cells with pre-cooled PBS free of DNA/RNase, then add 300 μL Trizol directly to the culture dish to lyse the cells and enrich RNA, and pipette the cell Trizol lysate into a centrifuge tube free of DNA/RNase;

2) Add 2 μL of linear acrylamide to the centrifuge tube to a final concentration of 10-20 μg/ml and mix thoroughly;

3) Incubate at room temperature for 5 minutes;

4) Add 60 μL of chloroform, mix thoroughly for 15 seconds, and incubate at room temperature for 5 minutes;

5) Centrifuge at 14,000 rpm for 15 min at 4°C;

6) Carefully transfer the supernatant (about 160-200 μL, about 60% of the volume of Trizol) to a DNase/RNase-free centrifuge tube;

7) Add 1 volume of pre-cooled isopropanol, mix thoroughly and place the centrifuge tube at -20°C for at least 2 hours, preferably overnight;

8) Centrifuge at 14,000 rpm for 30 min at 4°C and carefully remove the supernatant;

9) Rinse the precipitate with 300 μL 75% ethanol and incubate at room temperature for 5-10 minutes to remove any residual guanidine salt;

10) Centrifuge at 14,000 rpm for 10 min at 4°C;

11) Carefully remove the supernatant. You can use a pipette to absorb the remaining ethanol and air-dry at room temperature for 5-10 minutes.

12) Add 50 μL of commercial RNase-free water, leave at room temperature for 5 minutes, and pipette repeatedly to fully dissolve;

13) Measure RNA concentration using Nanodrop and store at -80°C;

The extracted RNA was then reverse transcribed using the full-length gold One-Step gDNA Removal and cDNA Synthesis SuperMix (AT311-02), first follow the table below.

Then continue to operate according to the following conditions, as shown in the following table:

The cDNA obtained by reverse transcription was diluted to 80 μL, and the reagent used for qPCR was Transgen PerfectStart ^™ Green qPCR SuperMix (AQ601-01). The 10 μL system is shown in the following table:

Each sample was tested in three technical replicates and two experimental replicates, and the samples were placed in CFX Real-time PCR system (BIO-RAD) for detection. The qPCR primers involved in this example are shown in the following table:

result:

FUS-ERG fusion protein drives phase separation in vivo and in vitro

The phase separation behavior of EWS-FLI1 has been extensively studied. The inventors first turned their attention to the homologous fusion protein of EWS-FLI, FUS-ERG (Figure 5(a)), which is found in AML and Ewing's sarcoma and is associated with poor clinical prognosis. FUS-ERG has been reported to enhance target gene expression. The inventors wanted to know whether FUS-ERG, like EWS-FLl1, could also drive the occurrence of phase separation.

It is well known that in FUS-ERG, part of FUS undergoes phase separation, while part of ERG contains a DBD that recognizes the GGAA microsatellite sequence. The inventors transiently expressed mCherry-FUS-ERG in Hela cells. It was found to colocalize with chromatin (Figure 5(b)) and show a small amount of aggregates in the nucleus. This observation suggests that FUS-ERG either has a very small phase separation potential or that FUS-ERG aggregates wet the chromatin, thereby losing their spherical morphology. Next, the present application attempts to distinguish between these two possibilities. To this end, the inventors used the OptoIDR system established by Brangwynne and colleagues (Shin, Y. et al. Spatiotemporal control of intracellular phase transitions using light-activated optodroplets. Cell 168, 159-171 (2017)) to evaluate the phase separation potential. In brief, the OptoIDR system consists of an IDR of interest, an mCherry fluorescent marker, and an aggregation-enhanced form of the photolyase domain of the Arabidopsis Cry2 protein (Cry2olig). Upon stimulation with blue light, Cry2olig forms oligomers, which also leads to IDR oligomerization and increases the valence of IDRs. This may drive phase separation at concentrations below the critical concentration for LLPS (Figure 5(c)). Consistent with previous work (Wei, M.T. et al. Nucleated transcriptional condensates amplify gene expression. Nat. Cell Biol. 22, 1187-1196 (2020); Shin, Y. et al. Spatiotemporal control of intracellular phase transitions using light-activated optodroplets. Cell 168, 159-171 (2017)), blue light stimulation caused obvious phase separation in cells expressing OptoFUS (Figure 5(d), Figure 5(e)), indicating that FUS IDR does have the potential to undergo phase separation.

Next, the inventors purified FUS-ERG labeled with recombinant enhanced green fluorescent protein (EGFP) (Figure 6 (c)). Under physiological buffer conditions without crowding agents, EGFP-FUS-ERG phase separated at submicromolar concentrations (Figure 5 (f)). Aggregate fusion experiments and fluorescence recovery after photobleaching (FRAP) experiments showed that the aggregates had liquid properties (Figure 5 (g), Figure 5 (h), Figure 5 (i)). The present application incubated a DNA fragment containing 25 copies of the GGAA microsatellite sequence with EGFP-FUS-ERG (Figure 6 (b), Figure 6 (c)) and found that the DNA substrate was recruited into the aggregates, which became irregular (5 (j)). FRAP analysis showed that these aggregates still retained a certain fluidity (5 (k), 5 (l)).

Taking into account the above observations, the inventors wanted to know whether FUS-ERG really undergoes phase separation in cells, and the resulting aggregates are confined to chromatin and deformed. Next, the inventors introduced point mutations into ERG to disrupt DNA binding. As expected, the mutant fusion protein (FUS-ERGmut) formed many spherical droplets in cells (5 (m)). In summary, these results indicate that FUS-ERG undergoes phase separation both in vivo and in vitro. Due to chromatin binding, FUS-ERG droplets in vivo deviate from a spherical morphology, similar to the wetting behavior of liquids (Morin, J.A.et al.Surface condensation of a pioneer transcription factor on DNA.BioRxiv, 311712 (2020)).

Pervasive anomalous phase separation in PS-DBD fusion proteins

The data of this application show more than 1000 potential PS-DBD fusion proteins (Figure 2). In order to study whether the abnormal phase separation caused by these fusions is more common than currently understood, the inventors randomly selected more fusion proteins for further study.

BRD9 is a bromodomain-containing protein that plays an important role in chromatin remodeling and transcriptional regulation. BRD9 is a fusion partner of TERT (transcription of telomerase reverse transcriptase) in hepatocellular carcinoma. Based on the OptoIDR system, BRD9IDR promoted the occurrence of phase separation (Figure 7(a), Figure 7(b)). In addition, the purified recombinant Alexa 488-labeled BRD9IDR protein (Figure 8(a)) formed spherical droplets with fluidity (Figure 7(c), Figure 7(d), Figure 7(e)). Therefore, BRD9 IDR does promote phase separation in vivo and in vitro.

COL17A1 (collagen type XVII α1 chain) is a collagen transmembrane protein and a key component of the epidermal anchoring complex. COL17A1 is fused to the CST complex subunit SNT1 in gastric adenocarcinoma. The CST complex binds to single-stranded DNA with high affinity in a sequence-independent manner. COL17A1 IDR also promotes phase separation in the OptoIDR system (Figure 7(f), Figure 7(g)). In addition, the purified recombinant COL17A1 IDR protein (Figure 8(a)) forms aggregates with gel properties (Figure 7(f), Figure 7(i), Figure 7(j)). Therefore, COL17A1 can promote phase separation both in vivo and in vitro. The phase separation potential of STN1-COL17A1 may contribute to its tumorigenesis.

MLLT1 (also known as ENL) is a histone acetylation reader that is specifically required for cell proliferation and oncogenic transcriptional regulation in AML. MLLT1 is frequently found fused to MLL (also known as KMT2A) in leukemia. MLLT1 IDR promotes phase separation in the OptoIDR system. In addition, purified MLLT1 IDR protein (Figure 8(a)) formed liquid aggregates (Figure 7(m), Figure 7(o)).

SFPQ is a pre-mRNA splicing factor. Chromosomal aberrations involving the fusion of SFPQ and TFE3 may be the cause of papillary renal cell carcinoma. SFPQ IDR promotes phase separation in the OptoIDR system (Figure 7(p), Figure 7(q)). Purified EGFP-SFPQ IDR protein (Figure 8(a)) forms aggregates and exhibits liquid-like properties (Figure 7(r), Figure 7(s), Figure 7(t)). MED15 is another fusion partner of TFE3 in renal cell carcinoma. The data of this application show that MED15 (Figure 8(a)) can form aggregates with liquid properties in vivo and in vitro.

In addition, the present application transiently expressed these fusion proteins in Hela cells (Figure 8(b)). Since DNA binding may distort the spherical morphology of phase-separated aggregates, the present application also generated a DBD mutant version (PSDBDmut) of each PS-DBD. The data of the present application show that these fusion proteins can form aggregates in cells. For TFE3 fusion proteins, SFPQ/MED15-TFE3 and its DBD mutant forms produce aggregates in cells (Figure 8(b)). These data suggest that the PS-TFE3 fusion protein found in renal cell carcinoma may have a phase separation-dependent mechanism in carcinogenesis. The EWS-FLI1 fusion protein formed some aggregates similar to FUS-ERG, while EWS-FLI formed many droplets in cells (Figure 8(b)). The other two EWS fusion proteins, EWS-ERG and EWS-ATF1, both showed obvious phase-separated droplets, and the DBD had no mutations (Figure 8(b)). The NUP98 fusion protein formed aggregates in cells (Figure 8(b)).

The inventors noticed that some PS-DBD fusion proteins, such as PRCC-TFE3, did not undergo phase separation (Figure 8(b)). The inventors then used OptoIDR analysis to evaluate more PS protein candidates (Figure 10(a), Figure 10(b)). The inventors found that OptoHES4 formed aggregates under blue light stimulation. But there are still some OptoIDR-modified proteins that do not respond to blue light and do not show obvious aggregates. The inventors speculate that this is because these proteins have relatively weak phase separation ability or no phase separation behavior, and the stimulation intensity is not enough to activate it. For all OptoIDR results, the inventors found that 20% of the proteins showed aggregates before stimulation, 40% of the OptoIDR-modified proteins phase separated under blue light stimulation, and 40% of the OptoIDR-modified proteins did not respond to blue light stimulation (Figure 10(c)). Overall, these results show that more than half of the PS-DBD proteins found in cancer patients tend to phase separate. Abnormal phase separation may amplify or inhibit downstream signaling pathways, thereby affecting cell physiology.

Example 2: Downstream expression analysis of TCGA data

To verify that PS-DBD fusion promotes transcription of downstream target genes, the inventors visited TCGA patients whose fusion genes were included in the FusionGDB database. Transcriptome data, gene expression TPM matrix and corresponding clinical information of all TCGA tumor samples were obtained from UCSC Xena (Goldman, M.J. et al. Visualizing and interpreting cancer genomics data via the xena platform. Nat. Biotechnol. 38, 675-678 (2020)). Then, the inventors obtained target gene information of 462 DBD proteins from the enrichr library, which came from three data sources: (1) ChIP seq data from ENCODE and ChEA databases; (2) transcriptome data of DBD proteins that experienced perturbations (overexpression or knockout) from the GEO database; and (3) PWM binding matrices of DBDs from the TRANSFAC and JASPAR databases.

Of these 462 DBD proteins, 55 were associated with PS-DBD fusion genes, and 27 TCGA samples contained these fusion genes. For each sample containing a PS-DBD fusion, the inventors used it as a foreground set and selected samples of the same tumor type without fusion as a background set. For each gene, the present application calculated the Z score to normalize its expression level using the following formula:

Where _Xj represents the expression level of gene j in samples containing PS-DBD fusion gene. μ( _Yj ) represents the mean expression level of gene j in background samples, and σ( _Yj ) represents the standard deviation. The GSEA algorithm was then used to evaluate the downstream expression level of PS-DBD fusion gene based on the Z score. The target of DBD protein was used as the gene set.

result:

PS-DBD fusions alter downstream transcription

Several PS-DBD fusions, such as EWS-FLI1 and NUP98-HOXA9, are known to alter downstream transcription and drive tumorigenesis (Chong, S. et al. Imaging dynamic and selective low-complexity domain interactions that control gene transcription. Science 361, eaar2555 (2018); Zuo, L. Y. et al. Loci-specific phase separation of FET fusion oncoproteins promotes gene transcription. Nat. Commun. 12, 1491 (2021); Wei, M. T. et al. Nucleated transcriptional condensates amplify gene expression. Nat. Cell Biol. 22, 1187-1196 (2020); Terlecki-Zaniewicz, S. et al. Biomolecular condensation of nup98 fusion proteins drives leukemogenic gene expression. Nat. Struct. Mol. Biol. 28, 190-201 (2021)). In order to determine whether the downstream transcriptional changes caused by PS-DBD fusion are common, this application analyzed the transcriptome data of patients with PS-DBD fusions in the Cancer Genome Atlas (TCGA). Data of 116 patients with PS-DBD fusions were extracted from the TCGA library. The direct target of PS-DBD fusions is not yet clear. Therefore, the target genes of DBD proteins are used as an alternative. The target genes of DBD proteins are collected from enrichr libraries (Xie, Z. et al. Gene set knowledge discovery with enrichr. Curr. Protoc. 1, e90 (2021)).

Example 3: Compound Screening

For in vitro screening, the Anti-Cancer-Compound-Library Plus library Plus (from the Pharmaceutical Technology Center of Tsinghua University) was used. The final concentration analyzed was 10 μM. For subsequent retest analysis, 10 compounds were extracted from the library (LY2835219, CX-6258, EMD-1214063, Pelitinib (EKB-569), GDC-0941, AZD-9291, UNC 0631, CHIR-124, SB743921 and Ursolic acid). LY2835219 was purchased back from APExBIO as a 20 mM stock in DMSO. The other compounds were prepared as 10 mM stocks in DMSO. The final concentration of all compounds analyzed was 10 μM.

High-content imaging. Compounds were screened for their effects on U2OS/mCherry-FUS-ERGmut cells in PerkinElmer CellCarrier-384 plates using the DropScan assay. 7,000-8,000 cells were seeded per well and incubated overnight. Compounds were then added to the cells using a Labcyte ECHO 550Liquid Handle at a final concentration of 10 μM. Images were captured immediately after the addition of the small molecules. Two fields of view per well were captured using the Opera Phenix High-Content Screening System equipped with a 40× NA 1.1 water immersion objective. The imaging process was repeated at least 30 times or the analysis lasted at least 6 hours.

For Lysotracker green staining images, U2OS and U2OS/mCherry-FUS-ERGmut cells were treated with DMSO or compounds for approximately 4 hours and then stained with Lysotracer green (Invitrogen, Cat. L7526). Multiple fields of view per well were captured using the Opera Phenix High Content Screening System equipped with a 60× water immersion objective.

Image quantification and analysis. For each image taken by high-content microscopy, the application divided its intensity by 0.05 to generate a new image. For each image, the Otsu algorithm in CellProfiler was used to segment cells with a target diameter size of 70 to 250 pixels. To describe the characteristics of droplets, the application enhanced the point structures in the image using the Enhance Features module in CellProfiler, with the target diameter set to 5 pixels. Then, a robust background algorithm was used to detect droplets with a diameter of 5 to 10 pixels, with a signal 4 standard deviations above the background. The identified droplets were associated with the cells identified in the previous step using the Relate Objects module in CellProfiler. Based on the above results, the number of droplets per cell was obtained.

Then, for each plate, the number of droplets per cell at the initial time point is available. All cells at the initial time point are sorted by their droplet number, and the droplet number at the top 15% position is used as the threshold to define droplet-enriched cells. Then, for all wells in the plate, the proportion of cells with droplet numbers exceeding the droplet enrichment threshold is monitored over time. Based on these data, a time series of the proportion of droplet-enriched cells after drug treatment is derived. Then, this application proposes an exponential model to describe the fraction of droplet-enriched cells as a function of time after drug treatment:

Where _c0 and _c1 are constants that need to be regressed in the exponential model, t is the action time of the small molecule compound, and Y(t) is the fraction of aggregate-rich cells in each well, which can be the fraction of aggregate-rich cells based on a single field of view or the fraction of aggregate-rich cells averaged over multiple fields of view.

Finally, two measures are calculated from the fitted model. The first is the drug effect, defined as Y(t_max)/(0). Here Y(0) is calculated from the fitted model at t=0, and Y(t_max) is calculated from the fitted model at t=t_max. The second is R-squared, which measures the goodness of fit of the exponential model above.

result:

Identification of aggregation modulator compounds by computational vision-assisted high-content screening

As the PS-DBD fusion protein undergoes phase separation and changes target gene expression, the inventors attempted to identify small molecules that regulate phase separation to reverse abnormal gene expression. Reported small molecule screening protocols only compare cell images treated with different molecules at any end time point, including established positive/negative controls. When using these protocols, kinetic information about aggregate changes is lost. In addition, the heterogeneity between holes can lead to inevitable uncertainty and/or noise. In addition, the related staining and washing steps are cumbersome to perform and often lead to large systematic errors. Here, the present application has designed a more effective strategy, called DropScan, for screening aggregates that regulate small molecules.

In DropScan, the inventors used high-content imaging screening in long-term imaging mode (Figures 4(a) to 4(c)). Specifically, U2OS cells stably expressing mCherry-FUS-ERGmut (U2OS/mCherry-FUS-ERGmut) were inoculated into 384-well plates. After small molecule compounds were distributed into each well, multiple fields of each well were imaged, and the same fields were repeatedly imaged within 6 hours (Figures 4(a) to 4(c), model and imaging). In this way, changes in the behavior of intracellular aggregates in the same set of fields can be monitored throughout the imaging process. In this study, a library of 1,777 anti-cancer drugs (APExBIO Anti-Cancer Compound Library Plus) was screened in five 384-well plates, and time series images were generated for each drug.

For each image, cells and droplets were segmented using Otsu and robust background algorithms, and the number of droplets per cell was obtained (Figure 4(a) to Figure 4(c), image processing and feature extraction). For each drug, the number of droplets per cell was monitored over time, and the proportion of cells with droplet numbers exceeding the droplet-enriched threshold was calculated accordingly. An exponential model was then fitted to describe the fraction of cells enriched in droplets as a function of drug treatment time (Figure 4(a) to Figure 4(c), quantification and hit identification).

To evaluate the effect of drugs on droplet dissolution, the present application calculated two measurements, drug effect and R-squared, according to the above exponential model (see the Materials and Methods section for details), and displayed a scatter plot of drug effect and R-squared for 1777 compounds (Figure 4(d)). Among them is LY2835219 (also known as abetamoxiclib mesylate), which is a selective cyclin-dependent kinase (CDK) inhibitor that targets the cyclin D1/CDK4 and cyclin D1/CDK6 cell cycle pathways. In addition, LY2835219, a potent pan-PIM kinase inhibitor CX-6258, and two epidermal growth factor receptor inhibitors EKB-569 and AZD-9291, showed effective aggregate dissolution (Figures 11(a) to 11(j)). In addition, some hit compounds, such as UNC 0631, may reduce the number of droplets by triggering cell death (Figures 11(a) to 11(j)). In summary, the present application successfully used DropScan to identify a group of aggregation modulating compounds with different effectiveness.

Example 4: Verification of LY2835219

Luciferase assay. The firefly luciferase reporter gene contains 25 copies of the GGAA motif in a minimal promoter (kindly provided by Qi Zhi Laboratory, Peking University). 293T cells were co-transfected with FUS-ERG or EWS-FLI1, 25×GGAA firefly luciferase plasmid, and Renilla luciferase plasmid (Promega, Cat. E6921). After ~24 hours of transfection, a dual luciferase assay including firefly luciferase and Renilla luciferase was performed according to the manufacturer's instructions for the luciferase assay system (Promega, Cat. E1960). For each experiment, the fluorescence intensity of firefly luciferase was normalized by the Renilla luciferase intensity and compared with the normalized intensity of the empty vector. The relative activity of firefly luciferase reflects the transactivation ability of the tested TF at the GGAA microsatellite-driven promoter. Each experimental condition was repeated three times.

Western blotting. The antibodies used in this study were Anti-mCherry (CST, Cat.43590), Anti-GAPDH (Easybio, Cat.BE0023), F(ab')2-Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary antibody, HRP (Invitrogen, Cat.A24537), and Goat anti-Mouse IgG (H+L) Secondary Antibody, HRP (Invitrogen, Cat.62-6520). The established U2OS/mCherry-FUS-ERGmut cells were treated with 7.5 μM LY2835219 for 0 h, 2 h, 4 h, and 6 h, respectively. Proteins were purified from cell samples using the Minute ^TM Animal Culture Cell/Tissue Total Protein Extraction Kit. Proteins were separated by SDS-PAGE and transferred to PVDF membranes. The membranes were incubated with primary antibodies overnight, followed by incubation with corresponding HRP-labeled IgGs. Finally, the results were visualized using enhanced chemiluminescence reagents.

RT-qPCR. Established U2OS/mCherry-EWS-FLI1 cells were treated with 7.5 μM LY2835219 for 0, 3, and 6 hours. Total RNA was purified from cells using Trizol (Invitrogen, Cat. A33252) and quantified using Nanodrop. One-step gDNA removal and cDNA synthesis SuperMix (TransGen, Cat. A T311-02) was used to reverse transcribe 1 μg of total RNA into cDNA. 1 μl of 1:4 cDNA dilution was used for PCR on the BIO-RAD CFX real-time PCR system. Quantitative PCR was performed using Green qPCR SuperMix (TransGen, Cat. AQ601-01). Three independent replicates were performed for each gene and two independent replicates were performed for each treatment. Data were normalized using GAPDH and the normalized fold change for each target gene was calculated.

RNA-seq. Whole-genome gene expression analysis was performed using RNA sequencing (RNA-seq). Established U2OS cells and A-673 cells were treated with 7.5 μM LY2835219 for 0 h, 3 h, and 6 h, respectively. Untreated U2OS cells and treated cell samples were collected in Trizol and delivered to Annoroad GeneTechnology Company (Beijing, China). Total mRNA was enriched with oligo(dT) beads, followed by fragmentation and reverse transcription with random primers. cDNA was purified, and its 5' and 3' ends were repaired and ligated with adapters. The ligated cDNA was amplified by PCR and subjected to 150-nt double-end sequencing using the Illumina Novaseq 6000 S4 system (Annoroad).

The valid data files were quantified using the unmatched quantification software Kallisto, and the pre-built index constructed from the Ensembl reference transcriptome was downloaded from the Kallisto transcriptome index website. The TPM values of different transcripts were calculated using default parameters, and the TPM value of a gene was calculated as the sum of the TPM values of all transcripts of the gene.

In order to verify whether LY2835219 inhibits the expression of EWS-FLI1 target genes, the present application calculated the fold change of EWS-FLI target gene expression levels in U2OS cell lines treated with 7.5 μM LY2835219 for 0 hours, 3 hours and 6 hours compared with wild-type U2OS cells. For A-673 cells, the target gene expression levels of 10 bone cell lines were extracted from the CCLE database as controls. Then, the fold change of EWS-FLI1 target gene expression levels in A-673 cell lines treated with 7.5 μM LY2835219 for 0 hours, 3 hours and 6 hours compared with the control expression level was calculated.

In the experiment to exclude that the function of LY2835219 is independent of the activity of CDK4/6 inhibitors, the small molecule compounds involved were all from the small molecule library APExBIO Anti-Cancer-Compound-Library Plus, and the cell images treated with small molecules during screening were directly used to display the results.

The reporter cell line was stained green by lysosome tracking method before and after LY2835219 treatment. The specific process is shown in the specific operation process of live cell staining (lysosome staining by Lysotracker Green) in Example 1.

When investigating the potential mechanism of LY2835219 function, the lysosomal inhibitor Baf-A1 was used at a final concentration of 200 nanomolar.

result:

LY2835219 dissolves aggregates by activating lysosomes

LY2835219 ranks first among the hit compounds (Figure 4 (e)). This application chooses to study the molecular mechanism of its effect on aggregate dissolution. Considering that LY2835219 is a CDK4/6 inhibitor, this application searched for other CDK4/6 inhibitors in the library and found more than four inhibitors. Among them, purvalanol B, palbociclibisesionate and LEE011 failed to dissolve aggregates, while ON123300 caused a large number of cell deaths (Figure 12). These data indicate that LY2835219 induces aggregate dissolution through a CDK4//6 independent mechanism. Next, LY2835219 was tested in a concentration range of 1 to 10 μM. The amount of aggregates decreased in a dose-dependent manner over time (Figure 13 (b) and Figure 14 (a)).

The inventors noted that cells treated with LY2835219 contained many vacuolar structures (Figure 11 (a) and Figure 12). A previous study reported that in cells treated with LY2835219, lysosomes expanded to form vacuoles (Hino, H. et al. Abemaciclib induces atypical cell death in cancer cells characterized by formation of cytoplasmic vacuoles derived from lysosomes. Cancer Sci. 111, 2132-2145 (2020)). To confirm this, the present application used lysosome tracking to stain the reporter cell line green before and after LY2835219 treatment. Consistent with previous studies, vacuolar structures co-localized with lysosomes (Figure 13 (c)), and the intensity of lysosomes was significantly enhanced (Figure 13 (c), Figure 13 (d)). These results are consistent with the concept that LY2835219 treatment leads to lysosomal acidification (Hino, H. et al. Abemaciclib induces atypical cell death in cancer cells characterized by formation of cytoplasmic vacuoles derived from lysosomes. Cancer Sci. 111, 2132-2145 (2020)).

In addition, the response of U2OS cells to LY2835219 also showed a significant increase in lysosomal intensity (Figure 14 (b)). Therefore, the specific inhibitor of lysosomes, bafilomycin A1 (Baf-A1), was added together with LY2835219 to evaluate whether Baf-A1 reduces the activity of LY28305219. The present application performed high-content imaging of cells during treatment with LY2835219 or Baf-A1 alone or both, and noted that no vacuoles were formed in cells cultured with Baf-A1 and LY2875219 (Figure 13 (e)). In addition, Baf-A1 significantly reduced the dissolution rate of LY2835219 (Figure 13 (e)).

The present application further introduces the lysosomal reporter gene YFP-LAMP1 (lysosomal associated membrane protein 1) into the fusion cell line and performs long-term imaging under small molecule treatment. In this experiment, the inventors found that before small molecule treatment, mCherry-FUS-ERGmut aggregates were highly colocalized with lysosomes. LY2835219 increased this colocalization, dissolved aggregates, and significantly enlarged lysosomes (Figure 13 (f)). In summary, these data indicate that LY2835219 may partially dissolve fusion protein aggregates in cells by activating lysosomal acidification.

Next, the specificity of LY2835219 in dissolving aggregates was evaluated. The present application tested the effect of LY2835219 on aggregates produced by many phase separation prone proteins: EWS-FLI1, EWS-FLI1mut, NUP98-HOXA9mut, NUP98HOXD13mut, the N terminus of FUS (aa 1-212), the N terminus of NUP98 (aa 1-515), and full-length TDP43. LY2835219 treatment resulted in the dissolution of all four PS-DBD fusion aggregates (Figure 13 (g)), but the three non-fusion proteins (Figure 13 (h)) were not dissolved.

Reversing abnormal phase separation rescues aberrant gene expression

The above results show that LY2835219 can effectively dissolve FET-ETS droplets. The present application further explores whether LY2835219-mediated droplet dissolution can reverse the abnormal expression of the fusion target gene. The present application found that the relative luciferase activity decreased over time and decreased by 50% after 12 hours of treatment (Figure 15 (a)).

In order to understand whether LY2835219 treatment will reduce the expression of EWSFLI1 target genes, the present application performed RNA sequencing on U2OS/mCherry-ESW-FLI1 cells before and after treatment with LY2835219. The putative target genes of EWS-FLI1 were extracted from a previously published transcriptome dataset (Boulay, G. et al. Cancer-specific retargeting of baf complexes by a prion-like domain. Cell 171, 163-178 (2017)). In short, 356 genes whose expression levels in EWS-FLI overexpressing cells were significantly higher than those in FLI1 overexpressing cells and control cells were considered as EWS-FLI1 target genes (Figure 15 (b)). By comparing the expression levels of 356 EWS-FLI1 target genes before and after LY2835219 treatment, the present invention found that 26 target genes were significantly downregulated after 3 hours of treatment, and only 2 target genes were upregulated (Figure 15 (c)). The present application further calculated the relative expression levels of EWS-FLI1 expressing cells and wild-type cells before and after LY2835219 treatment. Consistent with the above findings, the relative expression levels of EWS-FLI1 target genes were significantly reduced after LY2835219 treatment (Figure 15(d)).

The present application also performed real-time fluorescence quantitative polymerase chain reaction (RT-qPCR) to verify the downregulation pattern of EWS-FLI1 target genes after LY2835219 treatment. After LY2835219 treatment, the expression levels of most tested genes decreased, and this process was time-dependent (Figure 15(e)). Overall, the luciferase, RT qPCR and RNA sequencing data showed that LY2835219 can rescue abnormal gene expression induced by fusion oncoproteins. The results of the present application suggest that LY2835219 may be a potential drug for the treatment of these FET fusion-related Ewing sarcomas or other cancers. This is consistent with previous studies, LY2835219 showed multifaceted anti-tumor effects in preclinical Ewing sarcoma models (Dowless, M. et al. Abemaciclib is active in preclinical models of ewing sarcoma via multipronged regulation of cell cycle, DNA methylation, and interferon pathway signaling. Clin. Cancer Res. 24, 6028-6039 (2018)). The data of this application reveal the possible molecular mechanism of the beneficial effects of LY2835219 on Ewing sarcoma.

In many cancer cells, the genome and transcriptome encode fusion proteins with phase separation-prone domains and DBDs. These fusion proteins can phase separate on chromatin and act as novel transcription factors, altering the expression of downstream target genes. Abnormal transcriptomes are often associated with cancer. Phase separation-regulating drugs can interfere with aggregates and reverse abnormal transcription patterns. These drugs are expected to be therapeutic drugs for fusion protein-related cancers (Figure 15(f)).

Schedule 1

KMT2A-MLLT1 in Appendix 1 is the MLL-MLLT1 mentioned above; EWSR1 in Appendix 1 is the EWS mentioned above.

Claims (47)

1. A method for screening drugs for tumor diseases, comprising: Inoculating cells containing a target fusion protein into a well plate comprising a well array, wherein the cells contain a nucleic acid molecule encoding a target fusion protein, the target fusion protein being associated with a target tumor disorder, containing a phase separation prone domain (PS) and a DNA binding domain (DBD) and further containing a marker molecule so as to be able to form aggregates that can be identified by imaging; Dispense a variety of small molecule compounds into individual wells; Using a microscopic imaging system, continuously photographing each well in a predetermined period of time with a plurality of different fields of view, so as to generate time series images of the plurality of fields of view of each well; The processor analyzes the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well; The processor selects wells whose attenuation conditions of the amount of intracellular aggregates meet predetermined conditions, and uses the small molecule compounds allocated to the selected wells as candidate drugs for the target tumor disease. 2. The method according to claim 1, characterized in that the target fusion protein is associated with the occurrence of the target tumor condition, or causes the occurrence of the target tumor condition. 3. The method according to claim 1, characterized in that the target fusion protein undergoes phase separation in the cell, and the phase separation and the aggregates formed thereby are associated with the occurrence of the target tumor condition. The method of claim 1 , wherein the target tumor condition comprises a malignant tumor. 5. The method of claim 4, wherein the phase separation experienced by the target fusion protein drives abnormal gene expression, and a reduction in the amount of the aggregates is associated with a reversal of the abnormal gene expression. 6. The method according to claim 1 is characterized in that the DNA binding domain of the target fusion protein contains a mutation, so that the imaging recognition of the formed aggregates is more significant than the imaging recognition of the aggregates formed when the DNA binding domain does not contain the mutation. 7. The method according to claim 1, characterized in that the aggregates are in the form of droplets or gel. 8. The method according to claim 1, characterized in that continuously photographing each hole with a plurality of different fields of view within a predetermined time period specifically comprises: Divide each well into several independent fields of view; For each well, images of each field of view are taken sequentially to obtain a time series image of each field of view within a predetermined time period. 9. The method according to claim 8, characterized in that the step of analyzing the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well specifically comprises, for each well: For each image of each sampling time point of the time series image of each field of view, performing segmentation of cells and aggregates, and determining the amount of aggregates of each cell; For the images at each time point in each field of view, the proportion of cells whose aggregate amount exceeds the aggregate enrichment threshold is determined, and the fraction of aggregate-rich cells in the well at each sampling time point is averaged based on the field of view, thereby obtaining a distribution curve of the fraction of aggregate-rich cells in the well within the predetermined time period; fitting an exponential model based on the distribution curve of the fraction of aggregate-rich cells for the well over the predetermined time period; Based on the fitted exponential model, the fitted effect parameter and the decay ratio of the aggregate-rich cell fraction were calculated. The predetermined conditions include that the decay ratio of the aggregate-rich cell fraction reaches a first predetermined condition and the fitting effect parameter is better than a second threshold. 10. The method according to claim 8, characterized in that the step of analyzing the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well specifically comprises, for each well: For the images of the time series images of each field of view at each sampling time point, performing segmentation of cells and aggregates, and determining the amount of aggregates of each cell; For the images at each sampling time point of a single field of view, determining the proportion of cells whose aggregate amount exceeds the aggregate enrichment threshold, as the aggregate-rich cell fraction of the well at each sampling time point, thereby obtaining a distribution curve of the aggregate-rich cell fraction of the well within the predetermined time period; fitting an exponential model based on the distribution curve of the fraction of aggregate-rich cells for the well over the predetermined time period; Based on the fitted exponential model, the fitted effect parameter and the decay ratio of the aggregate-rich cell fraction were calculated. The predetermined conditions include that the decay ratio of the aggregate-rich cell fraction reaches a first predetermined condition and the fitting effect parameter is better than a second threshold. 11. The method according to claim 9 or 10, characterized in that the fitting effect parameter is R square. 12. The method according to claim 9 or 10, characterized in that the second threshold is 0.2-1.0. 13. The method according to claim 9 or 10 is characterized in that the first predetermined condition is the first predetermined number of cells whose attenuation ratio reaches 0.2 and are arranged in order of attenuation ratio, and the attenuation ratio is defined as the ratio of the aggregate-rich cell fraction after attenuation calculated using a fitted exponential model to the initial aggregate-rich cell fraction. 14. The method according to any one of claims 1-13 is characterized in that it also includes: manually or automatically identifying the survival status of the cells in each well, determining the wells whose survival status is worse than the predetermined survival condition; and excluding the small molecule compound allocated to the well from the candidate drugs for the target tumor disease. 15. The method according to claim 14, wherein automatically identifying the survival status of cells in each well specifically comprises: Determine at least one of the morphology of the cells and the rate of reduction of the number of cells using an image processing algorithm; When the deviation of the morphology of the identified cells from a spherical shape is higher than a deviation threshold and/or the reduction rate of the number of cells is higher than a reduction rate threshold, the cells are identified as having a survival condition worse than a predetermined survival condition. 16. The method according to claim 9 or 10, characterized in that the exponential model is defined by the following formula (1): where _c0 and _c1 are constants to be regressed in the exponential model, t is the exposure time of the small molecule compound, and Y(t) is the fraction of aggregate-rich cells in each well. 17. The method according to claim 9 or 10, characterized in that performing segmentation of cells and aggregates on each image specifically comprises: using a first algorithm to segment cells; Identify point structures in the image and perform enhancement using the representative size of the aggregate as the target size; Using the second algorithm, aggregates whose intensity values are higher than the representative intensity value of the background by a third threshold are segmented. 18. The method according to claim 17 is characterized in that the first algorithm automatically determines the intensity threshold for segmenting cell foreground and cell background and performs cell segmentation based on it, the second algorithm automatically determines the representative intensity value of the robustness of the polymer background, and the third threshold is 3-5 times the standard deviation of the representative intensity value of the background. 19. The method according to claim 17, characterized in that the first algorithm comprises an Otsu algorithm, and the second algorithm comprises a robust background algorithm. 20. The method according to claim 17 is characterized in that determining the amount of aggregates of each cell specifically comprises: associating the segmented aggregates with the segmented cells to obtain the amount of aggregates of each cell. 21. The method according to claim 9 or 10, characterized in that the photographing and the analysis are performed via a high-content cell imaging analysis system. 22. The method according to any one of claims 1-13 is characterized in that it also includes: when the target tumor condition is a tumor associated with FET fusion, a blood disease associated with NUP98 or MLL fusion, a renal cell carcinoma associated with TFE3 fusion, and a lipoma associated with HMGA2 fusion, selecting a member derived from the FET-ETS fusion protein family as the target fusion protein. 23. The method of claim 22, wherein, in the case where the target tumor disorder is a tumor disorder associated with FET-ETS fusion, including at least one of Ewing's sarcoma (EWS-FLI1), mesothelioma (FUS-ATF1), malignant melanoma (EWS-ATF1), desmoplastic small round cell tumor (EWS-ERG) and leukemia (FUS-ERG), the candidate drug for the target tumor disorder comprises at least one of the following: LY2835219 (Abetamoxiclib mesylate), the structural formula of which is CHIR-124, whose structural formula is EMD-1214063, whose structural formula is CX-6258, whose structural formula is Pelitinib (EKB-569), whose structural formula is AZD-9291, whose structural formula is GDC-0941, whose structural formula is SB743921, whose structural formula is 24. The method according to claim 23, characterized in that the target fusion protein comprises mCherry-FUS-ERGmut or mCherry-EWS-FLI1mut, and the cells are selected from U2OS cells and 293T cells, The amino acid sequence of mCherry-FUS-ERGmut is shown in SEQ ID NO: 1. The amino acid sequence of mCherry-EWS-FLI1mut is shown in SEQ ID NO:2. 25. The method of claim 23, wherein LY2835219 is used as a drug candidate for FET fusion-associated Ewing sarcoma. 26. The method according to claim 4, characterized in that the correspondence between the target fusion protein and the target malignant tumor includes at least one of the following: The target fusion protein comprises NUP98-HOXA9, and the target malignant tumor is acute myeloid leukemia; The target fusion protein comprises SFPQ-TFE3, MED15-TFE3 and/or PRCC-TFE3, and the target malignant tumor is renal cell carcinoma; The target fusion protein comprises BRD9-IDR or BRD9-TERT, and the target malignant tumor is hepatocellular carcinoma; The target fusion protein comprises TRIO-TERT, and the target malignant tumor is liposarcoma; The target fusion protein comprises STN1-COL17A1, and the target malignant tumor is gastric adenocarcinoma; The target fusion protein comprises MLL-MLLT1, and the target malignant tumor is leukemia; The target fusion protein comprises SFPQ-TFE3, and the target malignant tumor is papillary renal cell carcinoma. 27. The method according to claim 1, further comprising: for a target tumor condition, using a drug for treating a condition other than the tumor condition or malignant tumor as the small molecule compound to be distributed into each well for screening. 28. The method according to claim 1, further comprising: for a target malignant tumor condition, using a drug for treating other malignant tumor conditions other than the target malignant tumor condition as the small molecule compound to be dispensed into each well for screening. 29. The method of claim 28, wherein the target malignant tumor comprises a sarcoma, and the other malignant tumor conditions comprise any one of breast cancer, leukemia, kidney cancer, liver cancer, gastric cancer, and thyroid cancer. 30. The method of claim 27, wherein the other disease is associated with the target fusion protein, or is associated with abnormal phase separation of the target fusion protein. 31. The method of claim 28, wherein the other malignant tumor condition is associated with the target fusion protein, or is associated with abnormal phase separation of the target fusion protein. 32. A computer program product comprising computer executable instructions that can be stored and/or downloaded to a non-transitory storage medium, wherein when the computer executable instructions are executed by a processor, a method for screening a drug for a tumor condition is performed, comprising: Acquiring time series images of several fields of view of each well of a well plate including a well array, the time series images being generated by continuously photographing each well in several different fields of view within a predetermined time period after performing the following steps: inoculating cells containing a target fusion protein into the well plate including a well array, wherein the cells contain a nucleic acid molecule encoding the target fusion protein, the target fusion protein is associated with a target tumor disorder, contains a phase separation prone domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to be able to form aggregates that can be identified by imaging; dispensing a plurality of small molecule compounds into each well; Based on the time series images of at least one field of view of each well, analyzing the attenuation of the amount of intracellular aggregates in each well; Wells whose attenuation conditions of the amount of intracellular aggregates meet predetermined conditions are screened, and small molecule compounds assigned to the screened wells are determined as candidate drugs for the target tumor disorder. 33. The computer program product according to claim 32, wherein the step of analyzing the attenuation of the amount of intracellular aggregates in each well based on the time series images of at least one field of view of each well specifically comprises, for each well: For each image of each sampling time point of the time series image of each field of view, performing segmentation of cells and aggregates, and determining the amount of aggregates of each cell; For the images at each time point in each field of view, the proportion of cells whose aggregate amount exceeds the aggregate enrichment threshold is determined, and the fraction of aggregate-rich cells in the well at each sampling time point is averaged based on the field of view, thereby obtaining a distribution curve of the fraction of aggregate-rich cells in the well within the predetermined time period; fitting an exponential model based on the distribution curve of the fraction of aggregate-rich cells for the well over the predetermined time period; Based on the fitted exponential model, the fitted effect parameter and the decay ratio of the aggregate-rich cell fraction were calculated. The predetermined conditions include that the decay ratio of the aggregate-rich cell fraction reaches a first predetermined condition and the fitting effect parameter is better than a second threshold. 34. The computer program product of claim 33, wherein the fitting effect parameter is R squared, and the second threshold is 0.2-1.0. 35. The computer program product according to claim 33 is characterized in that the first predetermined condition is the first predetermined number of cells whose attenuation ratio reaches 0.2 and are arranged in order of attenuation ratio, and the attenuation ratio is defined as the ratio of the aggregate-rich cell fraction after attenuation calculated using a fitted exponential model to the initial aggregate-rich cell fraction. 36. The computer program product according to any one of claims 32-35, characterized in that the method for screening drugs for tumor diseases also includes: automatically identifying the survival status of cells in each well, determining the wells whose survival status is worse than the predetermined survival condition; and excluding the small molecule compound assigned to the well from the candidate drugs for the target tumor disease. 37. The computer program product according to claim 36, wherein automatically identifying the viability of cells in each well specifically comprises: Determine at least one of the morphology of the cells and the rate of reduction of the number of cells using an image processing algorithm; When the deviation of the morphology of the identified cells from a spherical shape is higher than a deviation threshold and/or the reduction rate of the number of cells is higher than a reduction rate threshold, the cells are identified as having a survival condition worse than a predetermined survival condition. 38. The computer program product of claim 33, wherein the exponential model is defined by the following formula (1): where _c0 and _c1 are constants to be regressed in the exponential model, t is the exposure time of the small molecule compound, and Y(t) is the fraction of aggregate-rich cells in each well. 39. The computer program product of claim 33, wherein performing segmentation of cells and aggregates on each image comprises: using a first algorithm to segment cells; Identify point structures in the image and perform enhancement using the representative size of the aggregate as the target size; Using the second algorithm, aggregates whose intensity values are higher than the representative intensity value of the background by a third threshold are segmented. 40. The computer program product according to claim 39 is characterized in that the first algorithm automatically determines the intensity threshold for segmenting cell foreground and cell background and performs cell segmentation based on it, the second algorithm automatically determines the representative intensity value of the robustness of the polymer background, and the third threshold is 3-5 times the standard deviation of the representative intensity value of the background. 41. The computer program product of claim 39, wherein the first algorithm comprises an Otsu algorithm and the second algorithm comprises a robust background algorithm. 42. A high-content imaging analysis system, comprising: An automatic high-speed microscopic imaging component is configured to continuously photograph each well of a well plate including a well array with several different fields of view within a predetermined time period to generate time series images of several fields of view of each well after the following steps are performed: inoculating cells containing a target fusion protein into a well plate including a well array, wherein the cells contain a nucleic acid molecule encoding the target fusion protein, the target fusion protein is associated with a target tumor disorder, contains a phase separation prone domain (PS) and a DNA binding domain (DBD) and further contains a marker molecule so as to be able to form aggregates that can be identified by imaging; dispensing a plurality of small molecule compounds into each well; and At least one processor configured to execute the method for screening drugs for tumor conditions according to any one of claims 32-41. 43. A method for analyzing the phase separation state of a target fusion protein comprising a phase separation prone domain (PS) and a DNA binding domain (DBD), comprising: Acquiring an image of a cell comprising the target fusion protein, wherein the cell comprises a nucleic acid molecule encoding the target fusion protein, and the target fusion protein further comprises a labeling molecule so as to form aggregates that can be identified by imaging; and The amount of intracellular aggregates in the image is identified, and the phase separation status of the PS-DBD fusion protein in vivo and/or in vitro is determined based on the amount of the identified aggregates. The larger the amount of aggregates, the more serious the phase separation status. 44. The method according to claim 43, characterized in that the aggregates are in the form of droplets or gel. 45. The method according to claim 43 is characterized in that the DNA binding domain of the target fusion protein contains a mutation, so that the imaging recognition of the formed aggregates is more significant than the imaging recognition of the aggregates formed when the DNA binding domain does not contain the mutation. 46. The method according to claim 45, characterized in that the mutation is a point mutation, and is introduced into a site having DNA binding activity in the DBD of the target fusion protein to be analyzed. 47. The method according to claim 46 is characterized in that the site with DNA binding activity in the DBD of the target fusion protein to be analyzed is determined based on the information on DNA binding activity in the Swiss-Prot database.