Immune escape modulating compounds in cancer treatment

Abstract

The invention relates to the use of AT7867, MC1568, ACY-1215, AZD-4547, NPK76- II-72-1, tivozanib, SAR-245409, HG-9-91-01, BIX-01294, UNC-669, WZ-7043, dacomitinib, CC-401, UNC-0638, epigallocatechin gallate (-), SB-239063, GDC-0879, KIN001-260, quizartinib, CGK-733, UNC-1215, XMD11-85H, garcinol, PF-573228, veliparib, VE-821, NVP-TAE684, GSK-J2, masitinib, KU-55933, KIN001-127, KIN001- 270, vandetanib, CG-930, GNF-2, decitabine, SGI-1776, amuvatinib, BS-181, CTB, MGCD-265, olaparib, ZM-447439, AGK-2, bosutinib, JNJ-38877605, RG-108, SYK- inhibitor, AZ-20, and OSI-930 drugs, which aim to suppress immune system escape mechanisms in cancer treatment and reduce the proliferation and survival abilities of cancer cells by interacting with proteins involved in immune escape processes and / or neighbouring proteins of these proteins, in T-cell-based immunotherapies.

Description

[0001] DESCRIPTION

[0002] IMMUNE ESCAPE MODULATING COMPOUNDS IN CANCER TREATMENT

[0003] Technical Field of the Invention

[0004] The invention relates to AT7867, MC1568, ACY-1215, AZD-4547, NPK76-II-72-1 , tivozanib, SAR-245409, HG-9-91 -01 , BIX-01294, UNC-669, WZ-7043, dacomitinib, CC-401 , UNC-0638, epigallocatechin gallate (-), SB-239063, GDC-0879, KIN001 -260, quizartinib, CGK-733, UNC-1215, XMD1 1 -85H, garcinol, PF-573228, veliparib, VE- 821 , NVP-TAE684, GSK-J2, masitinib, KU-55933, KIN001 -127, KIN001 -270, vandetanib, CG-930, GNF-2, decitabine, SGI-1776, amuvatinib, BS-181 , CTB, MGCD- 265, olaparib, ZM-447439, AGK-2, bosutinib, JNJ-38877605, RG-108, SYK-inhibitor, AZ-20, OSI-930 drugs that will be used to target neighbouring proteins of proteins that play a role in the immune escape system in treating cancer and / or increasing the effectiveness of treatments by suppressing the escape of tumour cells from the immune system.

[0005] State of the Art

[0006] Cancer continues to be the most important problem of our age in terms of science, economy and humanity. Although significant momentum has been gained in the fight against cancer, the expected leap has not yet occurred. Although the basic mechanisms of cancer are very simple and understandable, the rapid adaptation of cancer to treatments constitutes the real bottleneck and treatments need to change rapidly, similar to tumours. Therefore, the immune system's response to changes in environmental factors in particular requires an evolutionary process. Within this framework, drugs that will increase the sensitivity of the immune system in the fight against cancer have become widespread in recent years. However, it has been observed that cancer reappears after immunotherapy and more advanced studies are needed in this area that can explain and prevent cancer's escape from the immune system.

[0007] Non-small cell lung cancer (NSCLC) is one of the leading causes of cancer-related deaths in the world. Lung adenocarcinoma, the most common type of NSCLC, is characterised by intense lymphocytic infiltration. This suggests that the immune system plays an active role in the formation and development of adenocarcinomas

[0001] . In particular, new information on immune escape and cancer immune responses has paved the way for new treatment strategies for cancer immunotherapy. New immunomodulatory agents, including anti-CTLA4 and anti-PD1 / PDL1 monoclonal antibodies, have been investigated as monotherapy in metastatic cancer [2], However, one of the first clinical applications, anti-PD1 activity, showed low response rates ranging from 18% to 45% [3]. Early attempts to functionally characterise these interactions through survival analyses resulted in inconsistent and highly heterogeneous response rates. In general, dense lymphocytic infiltration has been associated with long-term prognosis, whereas data at the cellular level suggest that CD4 and CD8 cytotoxic T lymphocytes are both protective and unrelated to prognosis [4], This situation highlights the need for a more comprehensive understanding of tumour-immune interactions specific to lung adenocarcinoma.

[0008] The long-term results of clinical and biological studies in finding drugs and drug targets have necessitated new research approaches. In this context, systematic analyses encompassing multiple independent lung adenocarcinoma datasets may allow a better understanding of the role of each cell in lung adenocarcinoma and NSCLC. A recent study has significantly associated ZNF71 expression with T cell activities for the diagnosis and prognosis of NSCLC using RNA sequencing and proteomic profiles of patient tumours [5]. From the multi-omics network structure created for the ZNF71 gene, MEK1 / 2, VEGFR and IGF-1 R inhibitors PD-198306, U-0126, ZM-306416, PQ- 401 drugs have been proposed to treat NSCLC. Inhibition of MEK protein is applied in many tumours, in a recent study, the activities of 1 1 ,808 molecules against MEK protein were evaluated by a computational screening for NSCLC treatment [6]. Two compounds, DB012661 and DB07642, showed better performance in all screening analyses; free energy calculations indicated that the compounds are important therapeutic candidates to overcome drug resistance against both MEK and PIM1 proteins.

[0009] Recent studies have better characterised the role of immune infiltration in lung adenocarcinoma and systematically characterised tissue-specific tumour-immune interactions. However, there are no studies yet that suggest drugs targeting tumour- immune interactions using network-based drug repositioning / repurposing approaches and free binding energy calculations. These methods provide the opportunity to determine the specific locations of drugs that can be repositioned in cancer treatment in the immune escape system network. In this way, it is possible to provide more effective use of drugs in the state of the art by targeting the pathways of proteins that play a role in immune escape.

[0010] For example, no association was found between drugs known in the art such as NPK76-II-72-1 , SAR-245409, UNC-669, WZ-7043, dacomitinib, SB-239063, GDC- 0879, CGK-733, UNC-1215, XMD11 -85H, GSK-J2, masitinib, KU-55933, KIN001 -270, vandetanib, CG-930, GNF-2, SGI-1776, amuvatinib, BS-181 , CTB, bosutinib, JNJ- 38877605, RG-108, AZ-20 and OSI-930 and any protein involved in the immune escape system. This situation indicates that the potential use of many drugs known in the state of the art in cancer treatment is limited and shows that the effects of drugs on the immune system should be better defined for an effective treatment. It is difficult to provide effective use in cancer treatment without evaluating the potential locations and effects of these drugs on immune escape pathways.

[0011] On the other hand, although it has been suggested that AT7867, MC1568, tivozanib, HG-9-91 -01 , BIX-01294, garcinol, VE-821 , KIN001 -127, decitabine, ZM-447439 and AGK-2 drugs known in the state of the art may be associated with the immune escape system, the mechanisms by which these drugs are associated with immune escape have not been clearly demonstrated. This deficiency provides insufficient evidence for the use of these drugs as immunotherapy for cancer in the state of the art and reveals the need to examine the effects of the drugs on the immune system in more detail. A better understanding of the specific effects of these drugs on the immune escape pathway is critical for identifying potential secondary uses of these drugs and developing more effective treatment strategies.

[0012] In the current state of the art, drugs that interact with a type of immune escape structure associated with lung cancer include compounds such as CC-401 , UNC-0638, epigallocatechin gallate (-), KIN001 -260, quizartinib, PF-573228, veliparib, NVP- TAE684, ACY-1215, AZD-4547, MGCD-265, olaparib, and SYK-inhibitor. Although there are some studies indicating that these drugs may directly affect immune escape processes in the context of lung cancer, the mechanisms through which this effect occurs have not been demonstrated in detail within the scope of the current state of the art. In particular, the evidence on the effect of lung cancer on immune escape mechanisms is insufficient, and more advanced analyses are needed to understand the specific effects of these drugs on the immune system.

[0013] The patent numbered 2022 / 014783 in the current art is related to a computational machine learning model that predicts protein function with a small amount of data, but it has not been applied to predict proteins that may play a role in immune escape for lung adenocarcinoma and may be drug targets. In addition, the model in the document in question does not include elements such as network-based drug repositioning, molecular dynamics simulations, and verification of drug efficacy in immune co-culture cell models. Therefore, the patent document numbered 2022 / 014783 cannot be used alone to determine proteins that may play a role in immune escape for lung adenocarcinoma and may be drug targets.

[0014] Although certain effects of these drugs have been suggested in the suppression of immune escape structures in the context of cancer and specifically lung cancer, sufficient technical evaluations of these effects have not been made in terms of improving the recognition of cancer cells by the immune system or preventing immune escape. Network and pathway-based analyses regarding the specific targeting of pathways that cause immune escape, especially in lung cancer, remain lacking. Although some preliminary data on how these drugs modulate immune responses are presented in the relevant literature, comprehensive studies examining the interactions of these drugs with the immune system in cancer treatment in more detail are needed. Due to these deficiencies, there is a need for effective drugs identified through network and pathway-based analyses that can develop more effective treatment strategies by targeting immune escape mechanisms and aim to increase the sensitivity of the immune system to cancer cells.

[0015] Brief Description and Aims of the Invention

[0016] The invention includes drugs that can be used to suppress the escape of tumour cells from the immune system in cancer types such as lung adenocarcinoma. Among the drugs evaluated within the scope of the invention are compounds such as AT7867, MC1 568, ACY-1215, AZD-4547, NPK76-II-72-1 , tivozanib, SAR-245409, HG-9-91 -01 , BIX-01294, UNC-669, WZ-7043, dacomitinib, CC-401 , UNC-0638, epigallocatechin gallate (-), SB-239063, GDC-0879, KIN001 -260, quizartinib, CGK-733, UNC-1215, XMD1 1 -85H, garcinol, PF-573228, veliparib, VE-821 , NVP-TAE684, GSK-J2, masitinib, KU-55933, KIN001 -127, KIN001 -270, vandetanib, CG-930, GNF-2, decitabine, SGI-1776, amuvatinib, BS-181 , CTB, MGCD-265, olaparib, ZM-447439, AGK-2, bosutinib, JNJ-38877605, RG-108, SYK-inhibitor, AZ-20, and OSI-930. These drugs have been determined to target immune escape mechanisms in order to develop treatment strategies to increase the effectiveness of the immune system.

[0017] The main aim of the invention is to develop new drugs that can be used in T-cell-based immune therapy. For this purpose, drugs that are repositioned through the effect of immune escape proteins on signalling pathways and biological interaction networks have been determined to be used in cancer treatment. These drugs have been tested on target proteins with bioinformatics and molecular modelling analyses and detailed in terms of their interactions with proteins with three-dimensional modelling and molecular docking analyses. Molecular dynamics simulations have been used to determine the most suitable protein-drug complexes and verify their effectiveness. The effectiveness of drugs has been tested in lung adenocarcinoma-specific immune coculture cell models and their effects on cancer cells have been verified. With these methods, the specific locations of drugs evaluated within the scope of the state of the art in cancer treatment in the immune escape system network have been determined and they have been repositioned to target the pathways of proteins that play a role in immune escape. In this way, drugs that are effective through immune escape protein neighbours in T-cell based immune therapy in cancer treatment have been identified.

[0018] Description of Drawings

[0019] Figure 1 : The network of immune escape proteins formed (*: other interconnecting proteins used in the interaction of immune escape proteins).

[0020] Figure 2: Target distribution of drugs.

[0021] Figure 3: Ligand enrichment performance of LeDock on CSF1 R, ROC Curve.

[0022] Figure 4: Distribution of docking scores for 50 drugs, listed from left to right, top to bottom, according to their anonymous codes. Figure 5: Molecular dynamics simulation analysis (a) Displacement of protein backbone atoms over 20 ns, (b) Radius of gyration measurements to assess the compactness of the complex structure, (c) Displacement of ligand heavy atoms.

[0023] Figure 6: Intraprotein dielectric constant optimisations of binding free energy predictions based on MM-PBSA calculations.

[0024] Figure 7: Analysis steps applied in selecting candidate drugs for in vitro experiments.

[0025] Figure 8: Effects of cytotoxic doses of Drug 29, Drug 19 and Drug 24 on lung cancer and T-cells on cell proliferation and spheroid volume (A: Effect of Drug 29 on A549 Cell, B: Effect of Drug 19 on A549 Cell, C: Effect of Drug 24 on A549 Cell, D: Cell proliferation of cytotoxic doses of Drug 29 on T-Cells, E: Cell proliferation of cytotoxic doses of Drug 19 on T-Cells, D: Cell proliferation of cytotoxic doses of Drug 24 on T- Cells, G: Effect of Drug 29 on spheroid volume, H: Effect of Drug 19 on spheroid volume, I: Effect of Drug 24 on spheroid volume).

[0026] Figure 9: Images of spheroid samples pretreated with Drug 29, Drug 24, Drug 19 drugs 24 hours before in the co-culture model at the 6th and 72nd hours, respectively

[0027] Figure 10: Time-dependent effects of drug doses on spheroid volumes for (a) Drug 29, (c) Drug 19, (c) Drug 24 and (d) 72nd hour spheroid sizes for all three drug candidates.

[0028] Detailed Description of the Invention

[0029] The invention relates to drugs determined to be used in immune therapy in cancer treatment as a result of determining and repositioning drugs targeting neighbouring proteins of immune escape structures to increase the effectiveness of the immune system in cancer types such as lung adenocarcinoma. In this context, the selection of specific drugs for cancer types that exhibit immune system escape characteristics such as lung adenocarcinoma was made with bioinformatic analyses and network-based drug repositioning approaches. In this process, neighbours of proteins involved in immune escape mechanisms were detailed and drugs that will strengthen the immune system response were determined. The invention includes immune escape neighbours that have not been determined in the state of the art and drugs that will be used in cancer treatment by exhibiting binding properties to these neighbours, especially in order to increase the effectiveness of T-cell-based immune therapies.

[0030] Drugs determined to be used in T-cell-based immune therapy in cancer treatment are given in Table 1 .

[0031] Table 1. The drugs determined to be used in T-cell based immune therapy in the treatment of cancer, which are the subject of the invention

[0032] In one embodiment of the invention, the drugs given in Table 1 are used in T-cell-based immune therapy in the treatment of non-small cell lung cancer.

[0033] The invention also includes drugs that will suppress the immune escape system by binding to the proteins given in Table 2 for use in T-cell-based immune therapy in cancer treatment. In one embodiment of the invention, these drugs are used in T-cellbased immune therapy in the treatment of non-small cell lung cancer.

[0034] The drugs determined to be used in T-cell-based immune therapy in the treatment of cancer subject to the invention are given in Table 2 together with their target protein pairs. In one embodiment of the invention, these drugs are used to suppress the immune escape system in T-cell-based immune therapy in the treatment of non-small cell lung cancer by binding to the proteins they target. Table 2. The subject of the invention is drug-target protein pairs for drugs determined to be used in T-cell based immune therapy in cancer treatment.

[0035] The repositioning of the drugs in question was achieved as a result of three-stage analysis processes. The Drug Repositioning (DR) algorithm developed in the first stage calculates the similarity between the biological network structure formed by proteins that cause immune escape and the modules in the functional interaction network that a drug affects in the cell; as a result, it determines drugs that can suppress immune escape proteins. The second stage of the invention determines the physical protein targets of the identified candidate drugs, obtains their 3D structures, more accurate detection of targets with molecular docking and determines the most consistent protein-drug complexes by performing molecular dynamics simulations of protein-drug complexes. Within the scope of the third stage, the validation of the treatment of lung adenocarcinoma with immune therapy of candidate drugs was performed with the created 2D / 3D immune co-culture cell models. These analysis processes are explained in detail below. The DAL algorithm developed in the first stage of the invention calculates the similarity between the biological network structure formed by proteins that cause immune escape and the modules in the functional interaction network that a drug affects in the cell.

[0036] The DR algorithm provides new drug treatment recommendations by applying both biological interactome and genomic transcriptome analyses. The algorithm will receive the network topology (IEP module) of signalling pathways known to be associated with the immune escape protein (IEP) cluster as input. In order for the DR method to work, the network topology of biological pathways and drug data must be processed appropriately, these applied processes are explained in detail below.

[0037] Functional Connectivity Network (FCN) is a weighted protein-protein interaction (PPI) network used to represent the biological function similarity of human proteins. The FCN containing data on the interaction of the proteins to be used was obtained from the literature [7], The FCN used consists of a total of 20,790 proteins (nodes) and 21 ,952,150 interactions (connections). The weight value of each connection expresses the biological function similarity between two proteins, and these values are between 0 and 1. These interactions between proteins with very low functional similarity, i.e., those with a connection weight of 0 - 0.1 between them, were removed from the FCN. After this process, a total of 15,002 proteins and 334,225 connections remained in the FCN. This network, which will be called FCN in the rest of the study, is the network structure that has undergone this filtering process.

[0038] As input to the DR method, the set of immune escape proteins (IEP) related to lung cancer should be represented in an interaction network. The IEP set given as input will be converted into a weighted graph structure (IEP module). Thus, the functional interactions between the genes in the IEP cluster can be expressed more effectively with the gene information represented in the FCN. For this purpose, mapping was performed using only the genes that are directly neighbouring in the FCN. This network structure was briefly named as “Immune Escape Proteins Network” (IEP).

[0039] When a total of 68 proteins identified as causing immune escape in lung cancer were mapped on the FCN according to their directly interacting neighbours, 48 proteins were connected to each other as neighbours in the network. Three of the remaining 20 proteins are not included in FCN. The other 17 proteins were analysed for their ability to connect to the existing 48-protein network with the help of 1 intermediate protein. As a result of this analysis, the intermediate proteins with the highest biological function similarities were selected among the intermediate proteins found, and 15 of the 17 proteins were connected to the 48-protein network with the help of 1 intermediate protein. The remaining 2 proteins were not connected to this network. In this case, as seen in Figure 1 , a total of 78 proteins and an IEP consisting of 205 interactions were obtained. The expression values obtained from the RNA-sequencing data of these proteins were obtained as z-scores by performing differential gene analysis on the lung adenocarcinoma (LUAD) dataset from The Cancer Genome Atlas (TCGA) project.

[0040] Transcriptome profiles (differentially affected mRNA gene transcripts) of drug (compound) candidates were obtained from the database of the LINCS L1000 project [8]. The gene lists and the regulation amounts of genes included in these profiles are matched as the drug-affected proteins module on FCN. Level-5 transcriptome profiles of each compound were used for the lung cancer cell line (A549) from the L1000 project. The mRNA expression values in the raw data are expressed as z-scores. Drugs with 24-hour application experiments were selected in the A549 cell line. There are generally 6 different doses (0.04, 0.12, 0.37, 1.1 1 , 3.33 and 10.0 micromolar) for each drug. A total of 1604 drug-different dose combination samples were formed for the A549 cell line.

[0041] After detecting significant genes (differentially changed gene lists) by applying z-score filtering in each drug file, lists of drugs and drug-affected genes (IEP) were created. To perform this operation, 3 different filtering values are considered. The IEP list for each drug was created according to three different z-score filtering: Genes with absolute values greater than 1 , absolute values greater than 1 .5 and absolute values greater than 2 were selected. The number of genes included in the IEP lists formed according to these threshold values for all drugs was analysed and the distributions were examined accordingly. As a result, it was decided to use the IEP lists formed by genes with z-scores greater than |1 |. Because when other threshold values were used, the number of genes remaining in the IEP was quite low.

[0042] After filtering each drug according to its z-score, each IEP formed was mapped individually on the FCN. Since the aim was to create an independent interaction network specific to a drug, the functional interactions between genes affected by the treatment of this drug could be represented more effectively after this mapping process. For this purpose, mapping was applied according to the neighbours that directly interacted on the FCN. As a result of these analyses, 1604 drug-specific interaction networks (DEP) were obtained.

[0043] The first method used to calculate the network (module) similarity between the immune escape protein network (IEP) and the drug-specific interaction network (DEP) is a similarity score calculation based on the “ModuleSim” method in the literature [9], This method was developed in the original study to calculate the gene-based similarity between two different disease modules. In this invention, it was updated to calculate a similarity score between the drug-specific DEP and the immune escape protein network (IEP) and reformulated in a way that the z-score values of the genes can be integrated into it.

[0044] The first DR algorithm calculates the result as a “Module Similarity Score”. The original “ModuleSim” method did not take into account the values of the nodes (node scores) while considering the edge weights that hold the biological similarity of the network structure. The new module developed in the invention has been updated to take into account the fold-change values that express the high regulation (up-regulation) or low regulation (down-regulation) information of the relevant protein (node) when calculating the similarity score. Fold change scores are expressed with the z-scores of the proteins that show statistically significant changes.

[0045] Module Similarity Score calculation basically consists of four steps. In the first step of the calculation, the fold change values of the protein are integrated (Equation 1 ). Equation (1 )

[0046] The FCi and FC2 values in Equation 1 represent the fold change scores of the relevant genes in the IEP or DEP modules. The sp(gi,g2) value here calculates a similarity score between two genes using the shortest network distance between the genes. A and b values are fixed values given by the user, A and b values are fixed as 1 and 0.5 in the invention, respectively. When examining the similarity of common genes between IEP and DEP, FC1 and FC2 values are different and the absolute value of the sum of both values is taken. The absolute value taken is given as an exponent to the value of 0.5, so the sim(gi,g2) value of the proteins that will suppress each other will be high. The “0.5 IFC1 + FC21” section in the sim(gi, g2) equation has been added within the scope of the invention.

[0047] Let the network information (DEP or DEP) in the invention be represented in the G module. The relationship of genes with a disease is calculated using Equation 2. (Equation 2)

[0048] In Equation 2, the disease association score of a g gene is calculated by taking the average of the similarity scores between the genes in the G module and the gv gene.

[0049] Let Gi={gn, gi2, ... , gim}, be two different biological network modules consisting of m genes and G2={g2i, g22, ... , g2n} be two different biological network modules consisting of n genes. The relationship between these two modules is calculated by Equation 3.

[0050] In Equation 3, a similarity score based on the shortest path of the two modules is calculated by considering all possible gene pairs within the two modules. In order to estimate the similarities of the modules more consistently, the relationship score between Gi and G2 is normalised by dividing it by the average relationship score between them (Equation 4).

[0051] Modulesimilarity (6^2)= (Equation 4) The “Module Similarity” score calculated in Equation 4 indicates the similarity between modules G1 and G2. A high module similarity score indicates that there is a close biological similarity (connection) between G1 and G2. In all equations above, the IEP module is represented by the G1 cluster and the DEP module is represented by the G2 cluster, and the similarity score between the two given modules is calculated. In order to ensure that the contribution of the Module Similarity score (MSS) obtained for each drug to the results is equal, these calculated scores are standardised between 0-1 using min-max normalisation. The formula in Equation 5 is used when calculating min-max normalisation. Drugs with a normalisation value greater than 0.5 are considered candidate drugs. The reason for choosing the threshold value as 0.5 is to obtain 50% of the data distributed in the range of 0-1 .

[0052] I^gp > ModuleSimilarityScoreoftheDrug-min(ModuleSimilarityScore) (Equation normmax(ModuleSimilarityScore)-min(ModuleSimilarityScore)

[0053] 5)

[0054] The first DR algorithm calculates the MSS-norm score for one IEP given as input, separately for all different DEPs of all drugs in the dataset. The highest potential drug repositioning candidates were suggested among drugs with an MSS-norm score of 0.5 and greater.

[0055] Within the scope of the network centrality-based algorithm, three different network centrality metrics were applied to calculate the overlap score based on topological similarity between the Immune Escape Proteins Network (IEP) and the drug-affected proteins network (DEP): Adamic-Adar Coefficient, Preferential Attachment and Random Walk. Previously, these metrics did not use edge weights while working on the network. These metrics have been used in the literature for the analysis of social networks for many years, but they were used for the first time for drug repositioning purposes within the scope of this invention by adding some adaptations. The metrics are uniquely adapted to use the weights of interactions within the network and to process differentially varying gene information as z-scores.

[0056] Adamic-Adar Coefficient: Adamic-Adar Coefficient [9] is calculated based on the degrees of common neighbours of two genes.

[0057] Sxy= S ze(xnY) iwy (Equation 6)

[0058] In Equation 6, where x and y are different genes, Sxy represents the similarity between genes x and y. z is the set of common genes that have neighbourhoods with both x and y. kz is the total number of neighbours of gene z in the network. Since the similarity between only two given genes is obtained with Equation 6, when Sxy is calculated for all genes in the network, a similarity matrix is formed. In order to create a similarity score for only one gene from this matrix, the method of adding all values was used. As can be seen in Equation 7, the similarity score (Adamic-Adar coefficient) of a gene ‘x’ is expressed by the sum of its similarities with all other genes ‘y’.

[0059] Sx= SyeN Sxy(Equation 7)

[0060] In Equation 7, Sx is the similarity score of gene x. ‘N’ includes all genes in the network for which calculations are made. Sxy represents the similarity between genes x and y, and ‘y’ represents other genes in the network.

[0061] After obtaining the Adamic-Adar coefficient (V Sx) on a gene basis, a new formulation was created as follows in order to include gene expression information in this calculation:

[0062] If x e (IEP A DEP) then

[0063] If x e (DEP) then

[0064] If x e (IEP) then

[0065] In these formulas, z(drug,x) represents the z-score (RNA-seq measurement) of gene x after drug application, and z(disease,x) represents the z-score obtained for the disease information of gene x. The “z(drug,x)+ z(disease,x)” part gives the absolute value of the sum of these values. The main aim of this formulation is that if the z-score obtained for a gene for drug and disease is of opposite sign and the sum is close to 0, this value increases the Adamic-Adar coefficient (Sx) obtained from the metric, otherwise it decreases. Thus, if the common genes in the immune escape (disease) network and the drug network have values close to damping each other in terms of RNA-sequencing measurements, this situation will cause the relevant gene to score higher in the ranking, and in other cases, its score will decrease. If the relevant gene is present in only one of the two networks, there will be no damping of this gene, thus the value of this gene will decrease and it will move down in the ranking. It is known in the literature that the expression level of the target immune escape protein in the disease and drug network should be in the opposite direction (high vs. low), this biochemical observation has been added to all network metrics in the most appropriate way to mathematically support the higher score of drugs that inhibit the target protein.

[0066] The preferential binding metric is basically calculated as in Equation 8.

[0067] SXy=kx*ky (Equation 8)

[0068] In Equation 8, when x and y are different genes, Sxy represents the similarity between genes x and y. kxand kyrepresent the number of neighbours (degree) of genes x and y, respectively. A special transformation has been applied for this criterion. Instead of the number of neighbours of each gene, it has been transformed into the sum of the weights of the interactions with the neighbours it is connected to.

[0069] Sxy=Wx*Wy(Equation 9)

[0070] In Equation 9, Wx and Wyare equal to the sum of the edge weights of the neighbours of the x and y genes in the network, respectively. The z-score adaptation to this formula was made only on the z-scores obtained for the network used. Because, for example, when DEP is used, both the values of the x and y genes will be calculated on the same network, namely DEP, in this case, the z-scores of the x and y genes were obtained from the z-scores obtained for the drug, and this formula was updated.

[0071] If Part(x)={ Z(x) is negative, then

[0072] If Part(x)={ Z(x) is positive, then H4*Z(x)}

[0073] If Part(y)={ Z(y) is negative, then

[0074] If Part(y)={ Z(y) is positive, then V / y*Z(y)}

[0075] Sxy=Part(x)*Part(y) (Equation 10)

[0076] In these formulas, each gene is first processed with the sum of the edge weights ( 1 / 14, 1 / 1 ) coming from the network according to its own z-score (Z(x), Z(y)), and then their multiplication as in the classical preferential binding formula gives the result of “Sxy”. This addition applied is an original contribution to the invention, it has not been done before in the literature.

[0077] After this stage, the method of adding all the values was used to create a similarity score for a gene. It was determined that it would not be correct to rank both drug genes and disease genes from largest to smallest for the calculation of the area under the curve (AUC). Because when the z-score, which is the value showing the change caused by drug treatment of a gene, is positive and goes up in the drug gene ranking, if the z-score of the same gene observed due to disease is negative, it can go down in the disease gene ranking. In this case, the fact that both rankings are opposite to each other makes more sense biologically. For this reason, in the combined AUC calculations, after sorting the AUC. DEP and AUC.IEP values for the drug network (DEP) and disease gene network (IEP) in decreasing and increasing values, respectively, their products were calculated as Combined-AUC.

[0078] The PageRank algorithm, which is a random walk adaptation, was used for the Random Walk (PageRank) metric. Random walk works for one node (gene) at a time, but PageRank works for the entire network at once. For this purpose, the "Page_rank" function in the "igraph" package was used, and performance was tested with different 'damping' parameters. As a result of these tests, the 'damping' parameter was fixed as '0.75'. After calculating the PageRank score (PR(x)) for each gene using this function, the same adaptation used in the Adamic-Adar coefficient was applied, and gene expression information was included in this calculation as follows.

[0079] If x e (IEP A IEPA) then

[0080] If x G (IEPA) then

[0081] If x e (IKPA) then

[0082] Similarity scores were calculated for all proteins in DEP with two different network centrality metrics, then proteins were ranked from largest to smallest according to their similarity scores. Two different lists ranked in this way were created for each drug, and the results were verified individually for each drug. Proteins that are members of IEP and listed above I below a threshold value (the highest score is p proteins: top-p) were checked. Proteins that are both ranked higher than top-p and are members of IEP will be recorded as true positive (TP) proteins. True negatives (TN) are ranked lower than top-p and do not include IEP proteins. False positive (FP) proteins are listed higher than top-p and do not include IEP members. False negatives (FN) are listed lower than top-p and include proteins in IEP. For each drug, TP, TN, FP, and FN values are determined from a list ranked by a threshold (top-p) value that completes a single confusion matrix. When True Positive Rate and False Positive Rate are calculated from a confusion matrix, a point on the ROC curve is represented. 100 different threshold values (1 %, 2%, 99%, 100% of the protein count for each drug) were applied to obtain 100 separate measurements on the ROC curve. Then, the area under the ROC curve for each drug was calculated as the AUC score value. When DEP is used in network metrics and IEP is used as the golden-reference to create the confusion matrix, the calculated performance value is defined as AUC. DEP. When the same idea is applied in reverse, when network metrics are run on IEP, DEP is used as the golden- reference to calculate the AUC. DEP score. These two calculated AUC values were multiplied and integrated to calculate a combined AUC score as in Equation 11 .

[0083] Combined. AUC= AUC. DEP *AUC.IEP (Equation 1 1 )

[0084] A separate Combined-AUC score is calculated for each drug and for each network metric, drugs are ranked from the highest to the lowest according to this Combined- AUC score. Drugs with the highest score for each network metric (Combined-AUC > 0.85) are presented as new recommendations of the drug repositioning algorithm.

[0085] Drugs whose scores were calculated with two different drug repositioning methods were selected as candidates according to different criteria. The results of the module similarity algorithm were filtered by normalising them between 0-1 , drugs with MSS- Norm value greater than 0.5 were selected first. Drugs were ranked according to Combined-AUC score in the results of three metrics based on network centrality (Adam-Adar Coefficient, Preferential Binding), all drugs with this score greater than 0.85 were selected. During the detection of drugs commonly recommended by the two algorithms, all drugs with MSS-Norm value greater than 0.4 were checked. All candidate drugs selected according to these criteria are listed in Table 3. Among the metrics based on network centrality, the most candidates were suggested by Adam- Adar Coefficient. 6 drugs were suggested jointly by the two drug repositioning algorithms.

[0086] Table 3. Common and different drugs detected by the DR algorithm, three metrics based on network centrality (Adam-Adar Coefficient, Preferential Attachment, Random Walk) and the Module similarity algorithm (MSS-Norm)

[0087] The final list of 50 drugs was obtained and target protein studies of these drugs were conducted. The 50 drugs were sequentially scanned in 4 different databases, Drugbank, Cancerrx, Binding DB and Pubchem, and the codes of the detected targets are listed in Table 2. Targets could not be found in the relevant databases for 10 of the 50 drugs. For the remaining 40 drugs, an average of 7.8 protein targets were detected, with at least one target. The distribution of target numbers for drugs is given in Figure 2. The detected targets for drugs are listed in Table 2.

[0088] The codes of all PDB entries for each Uniprot entry of target proteins were given in Table 10. A PDB structure for each target protein listed in Table 10 was selected for molecular docking analysis. During this selection, criteria such as the quality of the structure (resolution, R-factor) were taken into account, and in addition, structures that covered the full-length protein sequence to the greatest extent were prioritized. Finally, all cross-PDB references for each entry were examined and, if available, complex structures (structures that interact with at least one compound) were prioritized. Prioritizing complex PDBs is generally due to the low global docking (blinding) performance of molecular docking algorithms

[0010] . The PDB codes of the target protein structures selected for docking are listed in the “PDB” column given in Table 10. In addition, whether these targets are complexes formed with at least one ligand is given in the column named in Table 10. Structures with the value “L” in this column are complex structures. The binding region to be used during molecular docking for these structures was written to the configuration files created for docking separately for each target. Selected PDBs (rows with empty column in Table 10) that were not detected as complexes were not included in the docking analysis. Apart from this, all selected structures were cleaned of their heterogeneous atoms (water, ligand, cofactor, etc.) before docking predictions. In addition, symmetric units of protein structures were deleted and docking preparations were done using the asymmetric units. Drug molecules were also obtained from Pubchem in sdf format, converted to 3-dimensional structures with openbabel software and saved in mol2 format after being minimized. Molecular docking studies were performed against 124 targets for a library of 50 drugs.

[0089] Data sets were created and tested for the validation of the molecular docking algorithm planned to be used during the obtaining of drug-target complex 3-dimensional structures. The performance of the algorithm to be used with the selected drug list and targets against immune escape targets was analysed in a ligand set. It was planned to evaluate the prediction performance of the molecular docking software to be used against target proteins or their analogues. Among well-known and highly cited docking programs such as Autodock4, Autodock Vina and DOCK6, LeDock, which has been shown to be successful in ligand enrichment tasks

[0011] . was tested. It has been shown that LeDock is quite successful in enriching active ligands from a bait ligand set, which is one of the most important criteria in the performance evaluation of docking programs, for different proteins

[0011] . For example, it has been observed that LeDock enriched active ligands of a protein called RPA, which plays an important role in DNA damage response pathways, with an enrichment factor of 20. In line with these findings, in a recent study comparing the performance of different docking programs, LeDock successfully predicted 81 % of the protein-ligand pairs

[0012] ,

[0090] Table 4. Ligand enrichment libraries

[0091] One of the important criteria frequently used to evaluate molecular docking algorithms is the ligand enrichment method

[0013] . In this evaluation, the extent to which the algorithms highlight active ligands compared to physically similar decoy ligands is analysed. In this context, decoy ligand libraries consisting of active and conjugated decoy ligands to be used are often obtained from “Directory of Useful Decoys” (DUD- E) and screened against targets. A target protein that has previously been associated with immune escape is selected for the protein used in this step. This protein is macrophage colony stimulating factor receptor (CSF1 R), which has been associated with immune escape, and CSF1 R inhibitors have shown potential in supressing immune escape in clinical trials

[0014] , Table 4 shows the number of active and decoy ligands in the ligand library belonging to this target. Accordingly, approximately 13 thousand ligands were downloaded from DUD-E in SMILES format with a dilution ratio of 50:1 and converted to 3D format using OpenBabel software. In this step, the algorithm named LeDock was first scanned considering the examples listed above. As a result of this enrichment analysis, LeDock performance was evaluated with the area under the ROC curve analysis and the study was selected for use (Figure 3).

[0092] The library of 50 drugs was cross-docked to 124 targets using LeDock. Overall, a total of 6200 docking runs were performed, each run producing a single pose was produced for each pair. The results of the docking analysis was listed in Table 7. When the distribution of docking scores was examined according to drugs, it was observed that some drugs particularly produced low docking scores against the 124 targets than others (Figure 4). The obtained protein-drug complexes were first ranked according to their LeDock scores and a total of 50 protein-drug complexes with high potential for interaction with target proteins were advanced to dynamic analyses.

[0093] When the drugs and targets in the top 50 were determined, it was observed that 16 out of 50 different drugs and 26 out of 124 different PDBs entered the top 50 (Table 5). 32% of the 50 drugs were represented by at least one complex. As a result of their screening against 124 different targets, it was also noted that two of these drugs (drugs 3 and 29) were represented among the top 50 complexes with a total of 5 different complexes. These results were evaluated as an output for the detection of possible protein targets of the 50 drugs identified within the scope of Stage 1 and listed in Table 1 , which have the potential to play a role in immune escape.

[0094] Table 5. PDB and drug codes representing the top 50 complexes with the highest docking score.

[0095] Table 2 lists the targets of 40 out of 50 drugs identified by two network-based drug repositioning models. These models suggest that the 50 drugs in Table 1 could be effective against target that are associated with immune escape and / or linked to immune escape pathways. Hence, 50 drugs in Table 1 are considered as potential therapeutic agents for lung adenocarcinoma and potentially other cancers by modulating immune escape. Their potential interference with the immune escape pathways through their listed targets in Table 1 suggest their potentail in treating other human diseases where immune escape pathways are implicated. This discovery highlights that the targets of 40 identified drugs (Table 2) also hold promise as pharmacological targets for cancer treatment through immune escape modulation. Beyond these specific drugs, targeting these proteins with other therapeutic approaches may also offer anticancer benefits. The complex structures of the 50 drug-target pairs with the highest raw and normalised scores were selected to be used in the next stage, molecular dynamics simulations (Table 7). In the list in Table 7, the complexes formed by the drugs identified in the previous stage with their targets were written as “c” (control). From this perspective, the first drug that was previously identified as a target and gave the highest docking score was identified in the 53rdplace. This shows that there are candidate drug-target complexes in the higher ranks in the docking list compared to the control complexes and indicates the potential of the selected complexes in terms of interaction in this respect. In addition, when the docking scores calculated as “score_norm” and normalised according to the number of heavy atoms for each drug are examined, it is observed that high scores do not only come from large ligands. The coordinates of the complex structures for the 50 drug-target pairs with the highest raw and normalised scores were obtained from the dock extension files obtained with LeDock and included in the dynamic analyses.

[0096] The 50 drug-target complex structures selected according to the LeDock docking score were considered to be used in molecular dynamics simulations. Dynamic analyses were performed with the NAMD molecular dynamics algorithm. All systems were simulated using the nPT thermodynamic assembly for 20 ns, at a constant temperature of 310 Kelvin and a constant pressure of 1 atm, and with periodic boundary conditions using the CHARMM36 force field (Huang and MacKerell, 2013) and the NAMD molecular dynamics algorithm (Phillips et al., 2005). To generate the topologies of drugs, one of the general force fields, cGenFF, was used

[0015]

[0016] . For electrostatic energy calculation, the “Particle Mesh Ewald” (PME) summation method

[0017] was used with the TIP3P model for water molecules

[0018] . The system coordinates were recorded every 2 ps. All simulation trajectories and structures were analysed with Visual Molecular Dynamics (VMD) and / or Chimera programs

[0019]

[0020] . Each simulation trajectory was analysed in terms of root mean square deviations (RMSDs), fluctuations and radius of gyratione. 37 of the selected complexes were successfully analysed with molecular dynamics simulations and 34 with binding free energy calculations. All systems were energy minimized in 1 ,000 steps in the first stage. All dynamic analyses were performed at TUBITAK ULAKBIM, High Performance and Grid Computing Center (TRUBA resources). The stability of the complexes was assessed according to tRMSD of the backbone atoms (C, N, Ca, 0) for the protein and all atoms except hydrogen for the ligands (Figure 5a) obtained during the simulation period and is shown in Figure 5a. The number of systems that were simulated for 20 ns and whose trajectories were obtained without any problems is 37. First of all, all of these complexes produced lower docking scores than the control (Table 7). In addition, the majority of the complexes showed stable complex dynamics after MD simulations.

[0097] The displacement of the protein and ligand chain atoms during the simulation period (RMSD- Figure 5a, 5c) and the compactness of the complex (RGYR- Figure 5b) were analysed. Accordingly, all complexes showed stable dynamics except for the 3 complexes given in Figure 5. Given in Figure 5; When the dynamic trajectories for the 3 complexes consisting of the PDB-drug pairs 3swr / KIN001 -270, 6qcn / KIN001 -270 and 7b7r / SAR-245409 were examined, it was seen that the docking poses were not preserved and thus, they were excluded from the binding free energy predictions.

[0098] The binding free energy of the complexes obtained was estimated based on the Molecular Mechanics / Poisson-Boltzmann Surface Area (MM-PBSA)

[0021] method. For this purpose, 10-to-40 different structures were sampled at every 250-to-1000 ps of the last 10 ns when the molecular dynamics simulations were stable, and these different complex conformations were used for binding free energy calculations. For this calculation, the parmed, chamber and antemm-pbsa programs in AMBER tools were used. The binding free energy calculation was performed with MM-PBSA.py

[0022] , In this analysis, optimisations were performed for both the number of conformations analysed and the dielectric constant for the protein. The binding free energy (AGbind) was calculated using the individual energy components given in the equation;

[0099] AGbind = EvdW + Eelec + Gpolar + Gnonpolar (Equation 12)

[0100] Here, the change in free energy is calculated by calculating the electrostatic (Eeiec) and van der Waals (Evdw) and polar (Gpoiar) and non-polar (Gnonpolar) contributions separately. The parameters of the MM-PBSA model are obtained from the most widely used article of this tool

[0021] , The obtained energy values are listed and the drug and target pairs with the highest binding free energy are selected. In order to estimate the relative binding free energies of drug-target proteins with molecular dynamics simulations using the Molecular Mechanics / Poisson-Boltzmann Surface Area (MM / PBSA) method, the relative binding free energies calculated with the MM-PBSA method were ranked and the drugs were determined to be tested in the experimental steps according to the relative binding free energy values after the docking scores.

[0101] Accordingly, in the last 10 ns of the MD simulations, which were determined to be stable, a total of 10- to 40 structures (every 250 and 1000 ps) were sampled. During the MM / PBSA calculation, the model parameters were taken as the same as those used in the literature

[0022] , Obviously, the dielectric constant (E) for the protein was studied separately as 1 , 2 or 4 for the protein interior, and for the water solvent it was taken as 80 throughout all calculations, and the temperature used for the PB calculation was determined as 310. The MM-PBSA. py tool in AmberTools is used for these calculations

[0023] .

[0102] Table 6. MM-PBSA optimisation results

[0103] The results obtained as a result of the optimisation of the dielectric constant are given in Figure 6 and listed in Table 6. When the constant is set to 1 , the binding scores were as positive. However, changing the constant to 2 or 4 has produced negative binding free energies to the expected negative figures. At this point, the optimisation was made based on the values of the target-drug complex at the top shown in Figure 6. The target 4drh and the drug 7 pair was taken as the control complex. It is expected to have negative binding free energy estimates through MM-PBSA calculations for this complex to indicate binding. At this point, the intraprotein solvent where strong binding occurs was taken as 4 and the binding free energy scores of the remaining target-drug complexes are listed. Apart from this, the optimisations include the number of complex conformations included in the calculations. 10-to-40 different conformations taken at intervals of 250 to 1000 ps from the last 10 ns were analyzed. The calculation shown in Figure 6 was shown for 10 different conformations. As a result of the optimisations, it was noted that the complex conformation numbers did not have a significant effect on the binding free energies. This is an expected result considering the stability of the protein chain at the end of the simulations. The number of conformations to be used in the calculations was determined as 20, trading-off between the calculation time and number of confomations.

[0104] According to the binding free energy predictions, most of the drugs showed stronger binding than the control (Table 6 and Figure 6). When the binding affinity score calculated in the same evaluation of the studied control complex was compared, it was seen that the first 27 of the 33 complexes showed stronger binding than the control. As a result, the drug and target pairs listed in Table 7 were chosen as candidates.

[0105] Table 7. Scores obtained from molecular docking and binding free energy calculations of target and drug complexes

[0106] The drug targets in the list were evaluated in terms of whether they were immune- escape proteins previously known in the literature and, if not, in terms of their neighbour numbers in immune-escape networks. In this way, both binding free energy estimates and neighbour statuses with proteins in immune-escape networks were taken into account for the selection of drugs to be tested in experimental analyses. Considering the target-drug list highlighted in Table 8, it was observed that none of the targets have been associated with immune-escape pathways yet. Therefore, these targets were searched within the immune-escape pathways from the network-based analyses. The targets that were identified as the first neightbout of any immune escape proteins in network was listed in Table 9. Accordingly, 18 targets do not have any close itneractions with the immune escape protein, i.e. were identified as the first neighbours of any immune escape protein However, the primary neighbourhoods of the remaining 15 targets, ranging from 1 to 9, were determined with proteins known to play a role in immune-escape. Target-drug pairs with a functional role in immune-escape and strong binding potential were studied for use in the next analysis. Table 8. Immune-escape neighbourhood assessments of target-drugs obtained in

[0107] Phase 2

[0108] For the list of 50 drugs given in Table 1 , all analyses were performed within the scope of Stage 2, and the protein targets with the highest binding free energy and the relevant candidate drugs are listed in Table 7. The workflow for selecting drugs for the next stage experiments is summarised in Figure 7. None of the protein targets in the “Target protein PDB” column in Table 9 are directly known immune escape proteins. The protein that physically binds to a drug does not have to be the protein whose activity is directly targeted to be changed in the cell, the proteins that are the physical partners of the drug may also have tasks such as transporting it to certain regions in the cell. Therefore, the expression of the proteins that are physical targets and the proteins that they interact with secondarily in metabolic or signalling pathways in the cell (downstream pathway partners) mostly changes after drug treatment [24,25]. Based on this observation, the immune escape proteins that the physical target proteins directly interact with in the FCN (the basic interaction network used in the Drug Repositioning method in Stage 1 ) were investigated.

[0109] In the selection of candidate drugs, first of all, the drug target proteins in Table 8 were identified in the FCN, and if the drug target was not included in the FCN, this drugtarget pair was removed from the list (Figure 7). In the second process, immune escape proteins that directly interact with the physical protein target of the drug were identified, and these neighbouring proteins were listed in the “Immune escape protein neighbours” column in Table 9. At the end of the second process, 8 drugs and 12 unique protein targets that could bind to them remained in the candidate list. In the third process, a literature review was conducted on the candidate drugs in Table 9, and drugs that had not been used before for lung cancer immune therapy were selected to suggest unique treatment candidates. In the final process, the candidates in Table 9 were ranked according to their binding free energy and total target protein count. The candidate drugs to be used in in vitro experiments within the scope of Stage 3 were selected from those that have not been used before in lung cancer immunotherapy, have been associated with the immune response via neighbouring proteins, have high target protein binding free energy and have a high target number. As a result, 8 different drug candidates were identified as input for Stage 3. Three of these 8 candidate drugs (Drug 19, Drug 24 and Drug 29) were obtained and in vitro experiments were performed.

[0110] Table 9. Interaction information of potential protein targets of candidate drugs with immune escape proteins in the functional interaction network (FCN).

[0111] For in vitro testing of selected candidates, firstly Human lung carcinoma A549 cell line was obtained from German Collection of Microorganisms and Cell Cultures (DSMZ). This cell line was maintained in DMEM / F-12 complete culture medium supplemented with 10% heat-inactivated foetal bovine serum (FBS), 1 % penicillin-streptomycin (p / s) at 37°C and 5% CO2 in a humidified incubator.

[0112] For the isolation of peripheral blood mononuclear cells, Peripheral blood mononuclear cells (PBMC) were isolated from 7 ml of heparinised peripheral whole blood taken from healthy volunteers by Ficoll Paque density gradient separation method. 7 ml of peripheral blood was gently transferred to form phase on an equal volume of Lymphocyte separation medium (Cat no. J0100-840, Cegrogen, Germany). Following centrifugation at 1200 g for 20 minutes, mononuclear cells that appeared as clouds in the interphase were collected. The collected cells were washed twice in RPMI 1640 supplemented with 10% FBS, 1 % P / S and 1% L-glutamine. The cells were counted by trypan blue method and resuspended in EasySep TM Buffer (Cat no. 20144, STEMCELLTM TECHNOLOGIES, Canada) at 1 x 10 6 cells per ml.

[0113] T cells were isolated from samples enriched with peripheral blood mononuclear cells by negative selection method according to the EasySep TM Human T Cell Isolation Kit (Cat no. 17951 , STEMCELL TECHNOLOGIES, Canada) protocol. 50 pl isolation cocktail was seeded into cells suspended in 1 ml EasySep buffer and incubated for 5 minutes to complete the antibody-antigen relationship. Then, Rapidsphere TM was added to the solution and the polystyrene tube was transferred to the EasySep Magnet (Cat no. 18000, STEMCELL TECHNOLOGIES, Canada) and incubated for another 5 minutes. With a single continuous move, the solution in the polystyrene tube was transferred to the 15 ml tube and T cells were isolated. For the activation of T cells, 1 ml of ImmunoCultTM-XF T Cell Expansion Medium (cat no. 10981 , STEMCELL TECHNOLOGIES, Canada) supplemented with 100 lU / ml Human Recombinant IL-2 (cat no. 92590, CHEMICON, USA) selected by negative selection were resuspended. To activate T cells, 25 pl / ml of ImmunoCultTM Human CD3 / CD28 T cell activator cocktail (cat no. 10971 , STEMCELL TECHNOLOGIES, Canada) was added to the T cell suspension and incubated at 37°C, 5% CO2 for 3 days. T cells were counted every 3 days by trypan blue method and IL-2 in the growth medium was renewed every 3 days according to the manufacturer's protocol. T cell activation was assessed by cell morphology during trypan blue and membrane receptor phenotype by flow cytometry on the third day after activation.

[0114] For the verification of T cell activation, activation of cytokine-induced T cells was assessed by cell size and membrane receptor profile. 1 x 106 stimulated cells were resuspended in 1 ml PBS to assess cell profile and activation. Cells were labelled with anti-CD25, anti-HLA-DR, anti-CD45, anti-CD16+ 56 and 7-AAD fluorescently labelled antibodies, all supplied by Beckman Coulter. Following 20 minutes of incubation at room temperature, measurements were taken with 20000 events per sample on the Navios EX Flow Cytometer and analysis was performed with Kaluza Analysis 2.1 software.

[0115] For the formation of NSCLC spheroids, A549 cells were seeded into the wells of a 48 well-plate at a ratio of 1 :5 in matrigel (Corning, America) cell suspension at a rate of 10,000 cells per droplet and turned upside down. After overnight incubation, the plate was orientated and completed media was added on top.

[0116] In order to obtain the growth curves of the spheroids and to determine the platelet stages, they were imaged every 2 days with bright field and Z-stack functions for 20 days following the formation of the spheroids. Before switching to the co-culture model, the medium was changed every 2 days and incubated for 7 days in order for the spheroids to complete their natural growth process (to reach the platelet stage).

[0117] The resazurin-based PrestoBlue viability reagent was then used to evaluate the effects of Drug 29, Drug 19 and Drug 24 on the viability of lung cancer A549 and T cells. A549 and T cells were seeded in 96 well-plates at 1x104 cells per well in 100 pl and incubated for 24 hours. A549 and T cells were then treated with different concentrations of Drug 29, Drug 19 and Drug 24. At the end of the experiment, 10 pl of PrestoBlue solution was added to each well for 3 hours in the dark at 37°C. Absorbance was measured at 570 nm and 600 nm reference wavelengths on the Varioskan Lux Microplate Reader (Thermo Fisher Scientific, VLBL00D2, USA). All analyses were performed in triplicate and all experiments were repeated at least three times.

[0118] One day before co-culture application (8th day of incubation), A549 cells were pretreated with 16, 128 and 1024 nM doses of candidate drugs Drug 29, Drug 19 and PF-58 for 24 hours. Following pretreatment, spheroids were imaged with bright field and Z-stack functions of ZEISS LSM800 confocal microscope on the 9th day. Spheroid groups were created using CD3 / CD28 and IL-2 activated T cells in a 1 :20 ratio to create co-culture conditions. Co-culture was continued to be photographed at 6, 12, 24, 48, 72 and 96 hours. After imaging, the width and length of the spheroids were measured in pm units with ZEISS blue software and their volumes were calculated with the formula 4 / 3 x fl x ((h x I) 1 / 2 ) 3. The volumes of the spheroids before co-culture were accepted as the zeroth moment and the change in their volumes depending on time was evaluated by calculating the fold changes within themselves. In this context, the group in which Anti-PD1 was applied to T cells 2 hours before co-culture was defined as the positive control, only activated T cells were defined as the negative control and the group without co-culture was defined as the experimental control group to show that the spheroids did not shrink in their natural processes.

[0119] The changes between drug candidates, drug doses and control groups were investigated with Kruskal-Wallis test and pairwise comparisons were performed with Mann-Whitney U test. The time-dependent change of cell death or a spheroid volume was investigated with Wilcoxon test. The difference between the groups calculated with a p-value of 0.05 was considered statistically significant.

[0120] As a result of in-silico analyses, targets and drug candidates associated with immune regulators or primary neighbours were determined among the agents. From this candidate list, the three drugs with the highest interaction amount according to the number of targets and molecular docking score: Drug 29, Drug 19 and Drug 24 were selected for in vitro validation of the findings.

[0121] In the first stage, the effects of possible cytotoxic doses of Drug 29, Drug 24 and Drug 19 candidate drugs on lung cancer or T cells on cell proliferation and spheroid volume were investigated. Drug 19 does not show a significant difference on T cell proliferation. Drug 29 (16-1024 nM) and Drug 24 (64-1024 nM) drugs were observed to cause a decrease in T cell viability, but the maximum decrease remained at the 10% limit (Figure 8d, 8e, 8f). Similarly, it was shown that the drug candidates did not cause cell death exceeding 10% in A549 cell viability (Figure 8a, 8b, 8c).

[0122] In NSCLC spheroids, it was confirmed that the drugs caused a decrease in spheroid growth rate compared to the control group (cytostatic effect) but did not show a cytotoxic effect that caused a decrease in spheroid volume compared to the baseline (Figure 8g, 8h, 8i). In this context, the effects of Drug 29, Drug 19 and Drug 24, determined with bioinformatic strategies, on the immune response were investigated in the dose range (16-1024 nM) that affected T cell and A549 cell death by less than 10% and did not cause a decrease in A549 spheroid volume, independently of the immune regulatory system.

[0123] T cells began to surround the spheroids by infiltrating the matrix at the 6thhour of coculture. The first significant change in spheroid volume was measured at the 24thhour. However, the change can be observed most clearly at the 72ndand 96th hours in Figure 9 and Figure 10. When A549 spheroids were examined independently of the natural cytotoxic process of T cells (experimental control / cold tumour), their volumes expanded by an average of 15.46% at the 72ndhour. In our group, where only the natural cytotoxic effects of T cells were observed without any treatment, the spheroid volume decreased by 19.76% to 80.24% (P=0.0313). In line with our study, it was aimed to use immunomodulatory agents to increase T cell cytotoxicity in this hot tumour model. In this context, in-silico analysis results showed that the ratio of the average spheroid volumes to the initial volume of the drugs Drug 29, Drug 24, and Drug 19, which were associated with the immune response and had the highest target number and interaction probability according to the docking score, was measured as 57.47%, 63.36%, and 62.22% at 128 nM doses at the 72ndhour of treatment, respectively. In this context, it was determined that all three drugs significantly increased T cell cytotoxicity compared to the untreated group (P=0.0011 ; P=0.0130; P=0.0238, respectively). Anti-PD1 agent used in the clinical treatment of lung cancer was used as a positive control to compare the effectiveness of these treatment strategies. In Anti-PD1 treatment, the spheroid volume decreased to 59.36% at 72 hours, and when the effects of Drug 29, Drug 24, and Drug 19 on spheroid volume were compared, it was shown that there was no significant difference between these treatment approaches and Anti-PD1 treatment (P=0.058; P=0.050; P=0.080, respectively).

[0124] Sphericity in spheroid morphology and cloudy appearance due to cell death in the peripheral spheroid structure are important parameters indicating structural deterioration. In NSCLC spheroids that were not co-cultured, the borders can be distinguished as a clear line and cloudy appearance due to cell debris is not observed at the end of 72 hours. In the untreated hot tumour group, an average of 19.75% reduction in spheroid volume due to T cell cytotoxicity was observed at 72 hours, but the spheroid borders were preserved. In our positive control, cemiplumab (Anti-PD1 ), group, the spheroid volume decreased by an average of 40.64% at the end of the 72ndhour and a cloudy image was formed, making it difficult to evaluate the spheroid sphericity. Peripheral borders became unclear in spheroids treated with our candidate drugs Drug 29, Drug 24 and Drug 19 at a dose of 128 nM.

[0125] In this context, when we compared the spheroid volumes in our three-dimensional cancer T cell co-culture model over time, it was confirmed in our in vitro model that the 128 and 1024 nM doses of Drug 29, Drug 24 and Drug 19 increased T cell cytotoxicity and reduced the spheroid volume compared to the negative control group (untreated co-culture group) during the 72-hour treatment period. It was confirmed that the 128 and 1024 nM doses of all three drugs reduced the spheroid volume more than the negative control and cold tumour (non-co-culture) models.

[0126] Previously unknown target proteins were identified for the drugs revealed by network and structure-based repositioning. These drug and protein pairs are given in Table 2. The list of all PDB entries listed in the PDB database for the Uniprot entry of the target proteins is given in Table 10. The target proteins listed in Table 10 are protein targets that indirectly regulate mechanisms that play a role in regulating cancer's escape from the immune system. Drug molecules that target these proteins are indirect inhibitor candidate drugs for checkpoints in the immune system. These drugs and target proteins may be new drug candidate molecules and targets to be used in cancer immunotherapy, and in particular, target proteins have been shown to indirectly regulate previously unknown immune escape pathways. Targeting specific proteins that are not found in the immune escape pathway but are identified with known immune escape proteins by this description with any pharmacological strategy(ies) can be used for cancer immunotherapy.

[0127] Table 10. Target Proteins and Cross-Reference PDB codes

[0128]

[0129]

[0130] REFERENCES

[0131] [1] Zdanov S., Mandapathil M., Abu Eid R., et aL, "Mutant KRAS Conversion of Conventional T Cells into Regulatory T Cells," Cancer Immunol Res., vol. 4, no. 4, pp. 354-365, 2016. doi :10.1 158 / 2326-6066.CIR-15-0241 .

[0132] [2] Rolfo C., Sortino G., Smits E., et aL, "Immunotherapy: is a minor god yet in the pantheon of treatments for lung cancer?," Expert Rev Anticancer Ther., vol. 14, no. 10, pp. 1 173-1 187, 2014. doi:10.1586 / 14737140.2014.952287.

[0133] [3] Rizvi NA., Mazieres J., Planchard D., et al., "Activity and safety of nivolumab, an anti-PD-1 immune checkpoint inhibitor, for patients with advanced, refractory squamous non-small-cell lung cancer (CheckMate 063): a phase 2, single-arm trial," Lancet Oncol. , vol. 16, no. 3, pp. 257-265, 2015. doi :10.1016 / S1470- 2045(15)70054-9.

[0134] [4] Al-Shibli KL, Donnem T., Al-Saad S., Persson M., Bremnes RM., Busund LT., "Prognostic effect of epithelial and stromal lymphocyte infiltration in non-small cell lung cancer," Clin Cancer Res., vol. 14, no. 16, pp. 5220-5227, 2008. doi : 10.1 158 / 1078-0432.CCR-08-0133.

[0135] [5] Ye Q., Hickey J., Summers K., et aL, "Multi-Omics Immune Interaction Networks in Lung Cancer tumourigenesis, Proliferation, and Survival," Int J Mol ScL, vol. 23, no. 23, p. 14978, 2022. doi:10.3390 / ijms232314978.

[0136] [6] Thirunavukkarasu MK., Suriya LL, Rungrotmongkol T., Karuppasamy R., "In Silico Screening of Available Drugs Targeting Non-Small Cell Lung Cancer Targets: A Drug Repurposing Approach," Pharmaceutics, vol. 14, no. 1 , p. 59, 2022. doi : 10.3390 / pharmaceutics14010059.

[0137] [7] Linghu B., Snitkin ES., Hu Z., Xia Y., Delisi C., "Genome-wide prioritization of disease genes and identification of disease-disease associations from an integrated human functional linkage network," Genome Biol., vol. 10, no. 9, p. R91 , 2009. doi:10.1 186 / gb-2009-10-9-r91 .

[0138] [8] Subramanian A., Narayan R., Corsello S., et aL, "A Next Generation Connectivity Map: L1000 Platform and the First 1 ,000,000 Profiles," Cell, vol. 171 , no. 6, pp. 1437-1452. e17, 2017.

[0139] [9] Ni P., Wang J., Zhong P., Li Y., Wu FX., Pan Y., "Constructing Disease Similarity Networks Based on Disease Module Theory," IEEE / ACM Transactions on Computational Biology and Bioinformatics, vol. 17, no. 3, pp. 906-915, May-Jun 2020.

[0140]

[0010] Cinaroglu S. S., Timucin E., "Comparative Assessment of Seven Docking Programs on a Nonredundant Metalloprotein Subset of the PDBbind Refined," J Chem Inf Model, vol. 59, no. 9, pp. 3846-3859, 2019.

[0141]

[0011] Cross JB., Thompson DC., Rai BK., et al., "Comparison of several molecular docking programs: pose prediction and virtual screening accuracy," J Chem Inf Model, vol. 49, no. 6, pp. 1455-1474, 2009.

[0142]

[0012] Wang Z., Sun H., Yao X., et al., "Comprehensive evaluation of ten docking programs on a diverse set of protein-ligand complexes: the prediction accuracy of sampling power and scoring power," Phys Chem Chem Phys., vol. 18, no. 18, pp. 12964-12975, 2016.

[0143]

[0013] Mysinger MM., Carchia M., Irwin JJ., Shoichet BK., "Directory of useful decoys, enhanced (DUD-E): better ligands and decoys for better benchmarking," J Med Chem., vol. 55, no. 14, pp. 6582-6594, 2012.

[0144]

[0014] Hu-Lieskovan S., Patnaik A., Eisenberg P., et al., "Phase 1 / 2a study of double immune suppression blockade by combining a CSF1 R inhibitor (pexidartinib I PLX3397) with an anti PD-1 antibody (pembrolizumab) to treat advanced melanoma and other solid tumours," Annals of Oncology, vol. 26, p. viii5, 2015.

[0145]

[0015] Vanommeslaeghe K., MacKerell AD., Jr., "Automation of the CHARMM General Force Field (CGenFF) I: bond perception and atom typing," J Chem Inf Model, vol. 52, no. 12, pp. 3144-3154, 2012.

[0146]

[0016] Vanommeslaeghe K., Raman EP., MacKerell AD., Jr., "Automation of the CHARMM General Force Field (CGenFF) II: assignment of bonded parameters and partial atomic charges," J Chem Inf Model, vol. 52, no. 12, pp. 3155-3168, 2012.

[0147]

[0017] Harvey MJ., De Fabritiis G., "An Implementation of the Smooth Particle Mesh Ewald Method on GPU Hardware," J Chem Theory Comput, vol. 5, no. 9, pp. 2371-2377, 2009.

[0148]

[0018] Price DJ., Brooks CL., 3rd, "A modified TIP3P water potential for simulation with Ewald summation," J Chem Phys., vol. 121 , no. 20, pp. 10096- 10103, 2004.

[0019] Humphrey W., Dalke A., Schulten K., "VMD: visual molecular dynamics," J Mol Graph., vol. 14, no. 1 , pp. 33-38, 27-38, 1996.

[0149]

[0020] Pettersen EF., Goddard TD., Huang CO., et al., "UCSF Chimera — a visualization system for exploratory research and analysis," J Comput Chem., vol. 25, no. 13, pp. 1605-1612, 2004.

[0150]

[0021] Genheden S., Ryde U., "The MM / PBSA and MM / GBSA methods to estimate ligand-binding affinities," Expert Opin Drug Discov., vol. 10, no. 5, pp. 449-461 , 2015.

[0151]

[0022] Kumari R., Kumar R., Open Source Drug Discovery, C., Lynn A., "g_mmpbsa--a GROMACS tool for high-throughput MM-PBSA calculations," J Chem Inf Model, vol. 54, no. 7, pp. 1951 -1962, 2014.

[0152]

[0023] Miller BR., 3rd, McGee TD., Jr., Swails JM., et al., "MMPBSA.py: An Efficient Program for End-State Free Energy Calculations," J Chem Theory Comput., vol. 8, no. 9, pp. 3314-3321 , 2012.

[0153]

[0024] Iskar M., Zeller G., Blattmann P., et al., "Characterisation of drug-induced transcriptional modules: towards drug repositioning and functional understanding," Mol. Syst. Biol., vol. 9, p. 662, 2013.

[0154]

[0025] Isik Z., Baldow C., Cannistraci C., et al., "Drug target prioritization by perturbed gene expression and network information," Sci Rep, vol. 5, p. 17417, 2015.

Claims

CLAIMS1. A drug for use in T-cell-based immune therapy in cancer treatment, wherein the drug is selected from the group comprising:• 4-(4-chlorophenyl)-4-[4-(1 H-pyrazol-4-yl)phenyl]piperidine• (E)-3-[4-[(E)-3-(3-fluorophenyl)-3-oxoprop-1 -enyl]- 1 -methylpyrrol-2-yl]- N-hydroxyprop-2-enamide• N-[7-(hydroxyamino)-7-oxoheptyl]-2-(N-phenylalanilino)pyrimidine-5- carboxamide• N-[5-[2-(3,5-dimethoxyphenyl)ethyl]-1 H-pyrazol-3-yl]-4-[(3R,5S)-3,5- dimethylpiperazin-1 -yl]benzamide• 4-[(9-cyclopentyl-5,8-dimethyl-6-oxo-7,8-dihydropyrimido[4,5- b][1 ,4]diazepin-2-yl)amino]-3-methoxy-N-(1 -methylpiperidin-4- yl)benzamide• 1 -[2-chloro-4-[(6,7-dimethoxyquinolin-4-yl)oxy]phenyl]-3-(5-methyl-1 ,2- oxazol-3-yl)urea• 2-amino-8-ethyl-4-methyl-6-(1 H-pyrazol-5-yl)pyrido[2,3-d]pyrimidin-7- one• 1 -(2,4-dimethoxyphenyl)-3-(2,6-dimethylphenyl)-1 -[6-[4-(4- methylpiperazin-1 -yl)anilino]pyrimidin-4-yl]urea• N-(1 -benzylpiperidin-4-yl)-6,7-dimethoxy-2-(4-methyl-1 ,4-diazepan-1 - yl)quinazolin-4-amine• (5-Bromopyridin-3-yl)(4-(pyrrolidin-1 -yl)pi peridi n- 1 -yl)methanone• 3-(2,6-dichloro-3,5-dimethoxyphenyl)-7-[4-(diethylamino)butylamino]-1 - methyl-4H-pyrimido[4,5-d]pyrimidin-2-one• (E)-N-[4-(3-chloro-4-fluoroanilino)-7-methoxyquinazolin-6-yl]-4- pi peridin-1 -ylbut-2-enamide• 3-[3-(2-piperidi n- 1 -yloxethoxy)phenyl]-5-(1 H-1 ,2,4-triazol-5-yl)- 1 H- indazole• 2-cyclohexyl-6-methoxy-N-(1 -propan-2-ylpiperidin-4-yl)-7-(3-pyrrolidin- 1 -ylpropoxy)quinazolin-4-amine• [(2R,3R)-5,7-dihydroxy-2-(3,4,5-trihydroxyphenyl)-3,4-dihydro-2H- chromen-3-yl] 3,4,5-trihydroxybenzoate• 4-[4-(4-fluorophenyl)-5-(2-methoxypyrimidin-4-yl)imidazol-1 - yl]cyclohexan-1 -ol• 2-[4-[(1 E)-1 -(hydroxyimino)-2,3-dihydroinden-5-yl]-3-(pyridin-4- yl)pyrazol-1 -yl]ethanol• 2-amino-6-[2-(cyclopropylmethoxy)-6-hydroxyphenyl]-4-piperidin-4- ylpyridin-3-carbonitrile• 1 - (5-tert-butyl- 1 ,2-oxazol-3-yl)-3-[4-[6-(2-morpholine-4- ylethoxy)imidazo[2,1 -b][1 ,3]benzothiazol-2-yl]phenyl]urea• 2,2-diphenyl-N-[2,2,2-trichloro-1 -[(4-fluoro-3- nitrophenyl)carbamatothiolamino]ethyl]acetamide• [3-an i li no-4- (4-pyrrolidi n- 1 -ylpiperidin- 1 -ylcarbonyl)phenyl]-(4-pyrrolidin- 1 -ylpi peridi n- 1 -yl)methanone• 4-[(9-cyclopentyl-6-oxo-5-propan-2-yl-7,8-dihydropyrimido[4,5- b][1 ,4]diazepin-2-yl)amino]-3-methoxy-N-(1 -methylpiperidin-4- yl)benzamide• (1 S,3Z,5R,7R)-3-[(3,4-dihydroxyphenyl)-hydroxymethylidene]-6,6- dimethyl-5,7-bis(3-methylbut-2-enyl)-1 -[(2S)-5-methyl-2-prop-1 -ene-2- ylhex-4-enyl]bicyclo[3.3.1 ]nonane-2, 4, 9-trione• 6-[[4-[(3-methylsulfonylphenyl)methylamino]-5- (trifluoromethyl)pyrimidin-2-yl]amino]-3,4-dihydro-1 H-quinolin-2-one• 2-[(2R)-2-methylpyrrolidin-2-yl]-1 H-benzimidazole-4-carboxamide• 3-amino-6-(4-methylsulfonylphenyl)-N-phenylpyrazine-2-carboxamide• 5-chloro-2-N-[2-methoxy-4-[4-(4-methylpiperazin-1 -yl)pi peridin-1 - yl]phenyl]-4-N-(2-propan-2-ylsulfonylphenyl)pyrimidine-2,4-diamine• 3-[[2-pyridin-3-yl-6-(1 ,2,4,5-tetrahydro-3-benzazepin-3-yl)pyrimidin-4- yl]amino]propanoic acid• 4-[(4-methylpiperazin-1 -yl)methyl]-N-[4-methyl-3-[(4-pyridin-3-yl-1 ,3- thiazol-2-yl)amino]phenyl]benzamide• 2-morpholin-4-yl-6-thianthren-1 -ylpyran-4-one• N-[5-[5-(4-acetylpiperazine-1 -carbonyl)-4-methoxy-2- methylphenyl]sulfanyl-1 ,3-thiazol-2-yl]-4-[(3-methylbutan-2- ylamino)methyl]benzamide• N-[5-[[6-[3-(1 ,3-dioxoisoindol-2-yl)phenyl]pyrimidin-4-yl]amino]-2- methylphenyl]methanesulfonamide• N-(4-bromo-2-fluorophenyl)-6-methoxy-7-[(1 -methylpiperidin-4- yl)methoxy]quinazolin-4-amine• 4-[(9-[(3R)-oxolan-3-yl]-8-(2,4,6-trifluoroanilino)purin-2- yl]amino]cyclohexan-1 -ol• 3-[6-[4-(trifluoromethoxy)anilino]pyrimidin-4-yl]benzamide• 4-amino-1 -[(2R,4S,5R)-4-hydroxy-5-(hydroxymethyl)oxolan-2-yl]-1 ,3,5- triazin-2-one• N-[(1 -methyl pi peridi n -4-yl) methyl] -3-[3-(trifluoromethoxy)phenyl]imidazo[1 ,2-b]pyridazin-6-amine• N-(1 ,3-benzodioxol-5-ylmethyl)-4-([1 ]benzofuro[3,2-d]pyrimidin-4- yl)piperazine-1 -carbotioamide• 5-N-(6-aminohexyl)-7-N-benzyl-3-propan-2-ylpyrazolo[1 ,5-a]pyrimidine- 5,7-diamine• N-(4-chloro-3-(trifluoromethyl)phenyl)-2-ethoxybenzamide• N-[[3-fluoro-4-[2-(1 -methylimidazol-4-yl)thieno[3,2-b]pyridin-7- yl]oxyphenyl]carbamatothioyl-2-phenylacetamide• 4-[[3-[4-(cyclopropanecarbonyl)piperazine-1 -carbonyl]-4- fluorophenyl]methyl]-2H-phthalazin-1 -one• N-[4-[[6-methoxy-7-(3-morpholin-4-ylpropoxy)quinazolin-4- yl]amino]phenyl]benzamide• (E)-2-cyano-3-[5-(2,5-dichlorophenyl)furan-2-yl]-N-quinolin-5-ylprop-2- enamide• 4-[(2,4-dichloro-5-methoxyanilino)-6-methoxy-7-[3-(4-methylpiperazin- 1 -yl)propoxy]quinoline-3-carbonitrile• 6-[dif luoro-[6-( 1 -methylpyrazol-4-yl)-[1 ,2,4]triazolo[4,3-b]pyridazin-3- yl]methyl]quinoline• (2S)-2-(1 ,3-dioxoisoindol-2-yl)-3-(1 H-indol-3-yl)propanoic acid• 2-[[7-(3,4-dimethoxyphenyl)imidazo[1 ,2-c]pyrimidin-5-yl]amino]pyridine- 3-carboxamide• (3R)-4-[2-(1 H-i ndol-4-yl)-6-( 1 -methylsulfonylcyclopropyl)pyrimidin-4-yl]- 3-methylmorpholine• 3-(quinolin-4-ylmethylamino)-N-[4-(trifluoromethoxy)phenyl]thiophene- 2-carboxamide.

2. A drug according to claim 1 for use in T-cell based immune therapy for the treatment of non-small cell lung cancer.

3. A drug according to Claims 1 or 2, wherein the drug is any one of• 4-(4-chlorophenyl)-4-[4-(1 H-pyrazol-4-yl)phenyl]piperidine, targeting the protein identified by UNIPROT code P17612 or P61925 or P31751 or P49841 , or• (E)-3-[4-[(E)-3-(3-fluorophenyl)-3-oxoprop-1 -enyl]- 1 -methylpyrrol-2-yl]- N-hydroxyprop-2-enamide, targeting the protein identified by UNIPROT code Q94F81 or Q9ZTP8 or Q13547 or P56524, or• 1 -[2-chloro-4-(6,7-dimethoxyquinolin-4-yl)oxyphenyl]-3-(5-methyl-1 ,2- oxazol-3-yl)urea, targeting the protein identified by UNIPROT code P08684 , or P02768 , or P17948 , or P35968 , or P35916 , or P10721 , or P09619 , or P08183 , or Q9UNQ0 , or P36888 , or P16234 , or Q13882 , or Q02763 , or P1 1362 , or P08581 , or• 2-amino-8-ethyl-4-methyl-6-(1 H-pyrazol-5-yl)pyrido[2,3-d]pyrimidin-7- one, targeting the protein identified by UNIPROT code P48736 or P42336 or 000329 or P42338 or P78527 or P42345 or Q8N122 or Q9BVC4 or P01584 or ACA62796 or P09992 or P13699, or• 1 -(2,4-dimethoxyphenyl)-3-(2,6-dimethylphenyl)-1 -[6-[4-(4- methylpiperazin-1 -yl)anilino]pyrimidin-4-yl]urea, targeting the protein identified by UNIPROT code P08684 or P01584 or P57059 or Q9Y2K2 or Q9H0K1 , or• N-(1 -benzylpiperidin-4-yl)-6,7-dimethoxy-2-(4-methyl-1 ,4-diazepan-1 - yl)quinazolin-4-amine, targeting protein identified by UNIPROT code Q9H9B1 , or Q96KQ7 , or AAI07878 , or P42345 , or P84022 , or ACA62796 , or CAG38738 , or BAA01847 , or P39748 , or P01574 , or AAH15051 , or P42858 , or P01579 , or Q9Z148 , or P09992 , or P13699, or• (5-bromopyridin-3-yl)(4-(pyrrolidin-1 -yl)pi peridi n-1 -yl)methanone, targeting the protein identified by UNIPROT code Q9Y468 or Q96JM7, or• 3-(2,6-dichloro-3,5-dimethoxyphenyl)-7-[4-(diethylamino)butylamino]-1 - methyl-4H-pyrimido[4,5-d]pyrimidin-2-one, targeting the protein identified by UNIPROT code P13699 or ACA62796, or• (E)-N-[4-(3-chloro-4-fluoroanilino)-7-methoxyquinazolin-6-yl]-4- piperidin-1 -ylbut-2-enamide, targeting the protein identified by UNIPROT code P10635 , or P02768 , or P08684 , or P1 1712 , or P22309 , or P08183 , or Q9UNQ0 , or 015245 , or P00533, or• 2-Cyclohexyl-6-methoxy-N-(1 -propan-2-ylpiperidin-4-yl)-7-(3-pyrrolidin- 1 -ylpropoxy)quinazolin-4-amine, targeting the protein identified by UNIPROT code Q96KQ7 or Q9H9B1 or P26358 or Q9Y657 or ACA62796 or P09992, or• [(2R,3R)-5,7-dihydroxy-2-(3,4,5-trihydroxyphenyl)-3,4-dihydro-2H- chromen-3-yl] 3,4,5-trihydroxybenzoate, targeting the protein identified by UNIPROT code P35869 or P26358 or Q86XF0, or• 4-[4-(4-fluorophenyl)-5-(2-methoxypyrimidin-4-yl)imidazol-1 - yl]cyclohexan-1 -ol, targeting the protein identified by UNIPROT code P45984 or P06239 or Q16539 or Q15759 or 015264 or P53778 or P00533 or P17252 or ACA62796 or P01579, or• 2-[4-[(1 E)-1 -(hydroxyimino)-2,3-dihydroinden-5-yl]-3-pyridin-4- ylpyrazol-1 -yl]ethanol, targeting the protein identified by UNIPROT code P15056 , or P04049 , or P27361 , or P28482 , or P49674 , or Q56UN5 , or Q8N4C8 , or Q9BVS4 , or 094804 , or Q9H2G2 , or P48730 , or 000238 , or 095819 , or P21860 , or Q9UKE5 , or P36896, or• 2-amino-6-[2-(cyclopropylmethoxy)-6-hydroxyphenyl]-4-piperidin-4- ylpyridine-3-carbonitrile, targeting the protein identified by UNIPROT code 014920 or 0151 1 1 or ANW61984 or P01574 or P01584 or ACA62796 or P09992 or P13699, or• 1 - (5-tert-butyl- 1 ,2-oxazol-3-yl)-3-[4-[6-(2-morpholin-4-yl- ethoxy)imidazo[2,1 -b][1 ,3]benzothiazol-2-yl]phenyl]urea, targeting the protein identified by UNIPROT code P36888 or P35869 or P26358 or Q86XF0, or• 2,2-diphenyl-N-[2,2,2-trichloro-1 -[(4-fluoro-3- nitrophenyl)carbamatothiolamino]ethyl]acetamide, targeting the proteinidentified by UNIPROT code Q13315 or Q13535 or P51450 or P04637 or P01579, or• [3-an i li no-4- (4-pyrrolidi n- 1 -ylpiperidin- 1 -carbonyl)phenyl]-(4-pyrrolidin- 1 -ylpiperidin-1 -yl)methanone, targeting the protein identified by UNIPROT code Q96JM7 or Q9Y468 or P1 1229 or ACA62796 or Q05BQ5 or Q12888, or• 4-[(9-cyclopentyl-6-oxo-5-propan-2-yl-7,8-dihydropyrimido[4,5- b][1 ,4]diazepin-2-yl)amino]-3-methoxy-N-(1 -methylpiperidin-4- yl)benzamide, targeting the protein identified by UNIPROT code ACA62796 or AAI07878 or AAH65243 or Q6IN02 or P09992 or P13699, or• 6-[[4-[(3-methylsulfonylphenyl)methylamino]-5- (trifluoromethyl)pyrimidin-2-yl]amino]-3,4-dihydro-1 H-quinolin-2-one, targeting the protein identified by UNIPROT code Q05397 , or 035346 , or P01584 , or P50613 , or P06493 , or P14635 , or P07947 , or P08684 , or P01574 , or ACA62796 , or Q66H76 , or P13699 , or Q9BV86, or• 2-[(2R)-2-methylpyrrolidin-2-yl]-1 H-benzimidazole-4-carboxamide, targeting the protein identified by UNIPROT code P09874 or Q9UGN5, or• 3-amino-6-(4-methylsulfonylphenyl)-N-phenylpyrazine-2-carboxamide, targeting the protein identified by UNIPROT code Q13535, or• 5-chloro-2-N-[2-methoxy-4-[4-(4-methylpiperazin-1 -yl)pi peridin-1 - yl]phenyl]-4-N-(2-propan-2-ylsulfonylphenyl)pyrimidine-2,4-diamine, targeting the protein identified by UNIPROT code P07332 or Q9UM73 or 015075, or• 2-morpholin-4-yl-6-thiantren-1 -ylpyran-4-one, targeting the protein identified by UNIPROT code Q13315 or P42338 or P78527 or P42336 or P42345, or• N-[5-[5-(4-acetylpiperazine-1 -carbonyl)-4-methoxy-2- methylphenyl]sulfanyl-1 ,3-thiazol-2-yl]-4-[(3-methylbutan-2- ylamino)methyl]benzamide, targeting the protein identified by UNIPROT code P08684 or P01584 or ACA62796 or ANW61984 or P13699, or• N-[5-[[6-[3-(1 ,3-dioxoisoindol-2-yl)phenyl]pyrimidin-4-yl]amino]-2- methylphenyl]methanesulfonamide, targeting the protein identified by UNIPROT code P50750, or• N-(4-bromo-2-fluorophenyl)-6-methoxy-7-[(1 -methylpiperidin-4- yl)methoxy]quinazolin-4-amine, targeting protein identified by UNIPROT code P15692 , or P00533 , or P33527 , or Q9UNQ0 , or P07949 , or Q01740 , or P31513 , or Q13882 , or Q02763 , or P02763 , or P02768 , or P08684 , or 015244 , or P08183, or• 4-[[9-[(3R)-oxolan-3-yl]-8-(2,4,6-trifluoroanilino)purin-2- yl]amino]cyclohexan-1 -ol, targeting the protein identified by UNIPROT code P53779 or P45984 or P45983 or ACA62796 or P09992 or P13699, or• 3-[6-[4-(trifluoromethoxy)anilino]pyrimidin-4-yl]benzamide targeting the protein identified by UNIPROT code P00519 or P1 1274 or 1 IEP_A or AAI07878 or P01584 or P08684 or ANW61984 or AAI14949 or P43116 or AAH00750 or BAC78637 or BAC78638 or P00520 or P04156, or• 4-amino-1 -[(2R,4S,5R)-4-hydroxy-5-(hydroxymethyl)oxolan-2-yl]-1 ,3,5- triazin-2-one, targeting the protein identified by UNIPROT code P26358 , or P27707 , or P30085 , or P15531 , or P22392 , or P32320 , or Q9Y6K1 , or Q9UBC3 , or 000337, or• N-(1 ,3-benzodioxol-5-ylmethyl)-4-([1 ]benzofuro[3,2-d]pyrimidin-4- yl)piperazine-1 -carbothioamide, targeting the protein identified by UNIPROT code P10721 or P08581 or P07949 or P16234 or P36888 or Q06609, or• 5-N-(6-aminohexyl)-7-N-benzyl-3-propan-2-ylpyrazolo[1 ,5-a]pyrimidine- 5,7-diamine, targeting the protein identified by UNIPROT code P50613 or P51946 or P51948 or P24941 or 060563 or P50750 or P24864 or Q00535 or Q15078 or ACA62796 or P49336 or P06493 or P14635 or P01574 or P1 1802 or P24385 or Q00534 or P09992 or P13699, or• N-[[3-fluoro-4-[2-(1 -methylimidazol-4-yl)thieno[3,2-b]pyridin-7- yl]oxyphenyl]carbamothioyl]-2-phenylacetamide, targeting the protein identified by UNIPROT code P35968 or P08581 or Q99835 or P06239 or 014965 or P36888 or P00519 or P08631 or P42684 or P12931 orQ08345 or 095819 or P1 1274 or P07948 or Q13882 or P42685 or P06241 or 000444 or Q16832 or Q13470 or P01584 or Q04912 or P43405 or P09619 or ACA62796, or• 4-[[3-[4-(cyclopropanecarbonyl)piperazine-1 -carbonyl]-4- fluorophenyl]methyl]-2H-phthalazin-1 -one, targeting the protein identified by UNIPROT code P09874 or P08684 or P20813 or Q9UGN5 or Q9Y6F1 or P20815 or P08183 or Q9UNQ0, or• N-[4-[[6-methoxy-7-(3-morpholin-4-ylpropoxy)quinazolin-4- yl]amino]phenyl]benzamide, targeting the protein identified by UNIPROT code Q6DE08 or Q96GD4 or 014965 or P00533 or ACA62796 or AAI07878 or XP_001349989 or ABH03417 or P01579, or• (E)-2-cyano-3-[5-(2,5-dichlorophenyl)furan-2-yl]-N-quinolin-5-ylprop-2- enamide, targeting the protein identified by UNIPROT code Q8IXJ6 or P49798 or P01574 or AAH18745 or AAH15051 or Q96EB6 or Q9BTU6 or Q9BV86 or Q9NXA8 or Q9NTG7, or• 4-(2,4-dichloro-5-methoxyanilino)-6-methoxy-7-[3-(4-methylpiperazin-1 - yl)propoxy]quinoline-3-carbonitrile, targeting the protein identified by UNIPROT code P1 1274 , or P00519 , or P07948 , or P08631 , or P12931, or P08183 , or P08684 , or P24941 , or Q02750 , or P36507 , or Q9Y2U5 , or Q13555 , or P10632, or• (2S)-2-(1 ,3-dioxoisoindol-2-yl)-3-(1 H-indol-3-yl)propanoic acid, targeting the protein identified by UNIPROT code P48775, or• 2-[[7-(3,4-dimethoxyphenyl)imidazo[1 ,2-c]pyrimidin-5-yl]amino]pyridine- 3-carboxamide, targeting the protein identified by UNIPROT code P43405 or ACA62796 or P01584 or P01574 or P08684 or 0151 11 or Q12851 or P13699, or• (3R)-4-[2-(1 H-i ndol-4-yl)-6-( 1 -methylsulfonylcyclopropyl)pyrimidin-4-yl]- 3-methylmorpholine, targeting the protein identified by UNIPROT code Q13535 or P42345 or P08684 or P01584 or P42336 or P01579, or• 3-(quinolin-4-ylmethylamino)-N-[4-(trifluoromethoxy)phenyl]thiophene- 2-carboxamide, targeting protein identified by UNIPROT code P10721 or P17948.

Images