Synthetic Lethal Interactions for Kinase Deficiencies to DNA Damage Chemotherapeutics Free

Kinases are enzymes that catalyze the addition of a phosphate from adenosine triphosphate (ATP) onto a substrate, a modification that is reversible via the enzymatic activity of phosphatases. Protein phosphorylation is one of the main post-translational modifications and can rapidly change either the activity, localization or interaction network of the target protein (1). As such, kinases regulate diverse fundamental cellular processes including cellular differentiation, cell-cycle progression, apoptosis and DNA repair, hence being implicated in several of the hallmarks of cancer. The human kinome is estimated to include some 518 kinases and, of these, 120–157 are suggested to function as drivers of cellular transformation (2). Mutations within these kinases can be either gain- or loss-of-function and can promote tumor initiation or progression, leading to a range of cancer types (3). For example, the gene PIK3CA can harbor mutations that lead to the upregulation of the AKT–mTOR pathway and promote cell growth and proliferation (2). In contrast, ATM loss-of-function mutations dysregulate the signaling of DNA damage and promote genomic instability (4, 5).

The long-standing precedence of kinases in cancer, as well as other diseases, has identified them as important drug targets. The first kinase inhibitors were discovered in the 1980s and currently, numerous are under development for different purposes. In the US alone, around 10,000 patent applications for kinase inhibitors have been filed since 2001 (2). As of 2018, 31 kinase inhibitors were approved by the FDA for cancer therapy (2). These functioned by blocking the ATP-binding domain, a region that is highly conserved, hence making these inhibitors unspecific and of low potency. Strikingly, it was not until 1998 when trastuzumab (Herceptin) became the first example of an approach to block the activity of a kinase, in the clinics. Trastuzumab is a monoclonal antibody that inhibits ERBB2 and is used for the treatment of ErbB2-positive metastatic breast cancer (6). ERBB2 is a receptor tyrosine kinase that belongs to the EGFR family. A variety of ligands bind to these receptors, activating different signaling pathways, such as those regulated by AKT and RAS. Mutations in these receptors cause over-activation of these pathways and hence, promote tumor development (6). Tyrosine kinases have also been targeted using small molecules, an example of which is imatinib (Gleevec) that was approved by the FDA in 2001. Imatinib was developed to target dysregulated tyrosine kinase activity, that results from the BCR–ABL fusion oncoprotein. However, the therapeutic benefits of such kinase inhibitors can be short-lived, due to the acquisition of resistance mechanisms. For imatinib this is mainly due to point mutations in the tyrosine kinase domain of BCR–ABL (7).

To increase the efficacy of kinase inhibitors, they can be used in combination with chemotherapeutic drugs. An important group of such chemotherapeutic agents includes those that induce DNA damage, hence functioning to stop proliferation or induce cell death (8). Since the 1960s, many chemotherapeutic agents have been developed that react chemically with DNA or block essential DNA-associated functions (9). Capecitabine (a fluoropyrimidine), as well as radiation therapy, have been used in combination with erlotinib, an EGFR small-molecule inhibitor, for the treatment of advanced pancreatic cancer (10). Moreover, the multi-kinase small-molecule inhibitor, sunitinib, has been used in combination with radiation for the treatment of a range of cancers, including renal cell carcinoma and prostate cancer (11, 12). In addition, kinase inhibitors have been shown to increase the efficacy of drugs that block the catalytic activity of DNA repair enzymes. Representative examples are the inhibitors targeting the receptor tyrosine kinase c-Met (a proto-oncogene; using foretinib and crizotinib) that synergize with inhibitors of PARP (a molecule involved in signaling DNA damage; using ABT-888 and AG014699) and hence this combination therapy may be beneficial for tumors with high c-Met expression (13).

It is noteworthy that kinases play important functions in the repair of DNA damage, or in the regulation of the cell cycle in response to DNA lesions. The DNA damage response is activated by three main kinases that are members of the PIKK family, namely ATM, ATR, and DNA-PKcs (14). Inhibitors have been developed to target these kinases and clinical trials are ongoing to test their efficacy either alone, or in combination with other treatments. Among others, the DNA-PKcs inhibitor MSC2490484A (study number: NCT02516813) as well as the ATM inhibitor AZD1390 (study number: NCT03423628) are currently being tested in combination with radiotherapy in Phase 1 trials. Similarly, the ATR inhibitors AZD6738, M6620, BAY1895344 and VX-970 are being tested either alone (NCT03718091) or in combination with a plethora of other chemotherapeutic agents, including immune checkpoint inhibition by targeting PDL-1 (study number: NCT02264678), olaparib (study number: NCT03787680), gemcitabine (NCT02595892), PARP inhibition and cisplatin (NCT02723864), as well as topotecan (NCT02487095). Other kinases have been shown to function in the regulation of the cell cycle, an essential process that regulates cellular homeostasis. Proteins like PLK1 or CHK1 and CHK2 are degraded or activated, respectively, to induce checkpoint activation (15, 16). The CHK1 inhibitor SRA737 is in Phase 1 trials either alone or in combination with gemcitabine (with or without cisplatin; study numbers: NCT02797964 and NCT02797977).

Intriguingly, the vast majority of scientific literature on kinases relates to a small number of these proteins (17), suggesting that other kinases may also function to regulate cellular pathways, including those related to DNA repair. Indeed, kinase dependencies have been studied through the use of unbiased and systematic experimental approaches and several kinases have been implicated in the repair of DNA damage (18). Such high-throughput screening approaches have included siRNAs (19–21) or small-molecule libraries (22, 23) and have unraveled new synthetic lethal interactions in cancer models, suggesting the inhibition of specific kinases, such as FGFR and CHK1, as promising therapeutic targets for osteosarcoma (19) and neuroblastoma (20), respectively. Moreover, a systematic overexpression approach for kinases, and phosphatases, identified the overexpression of tyrosine kinases to induce a MEK-dependent ERK activation that promotes drug resistance (24). In the context of DNA repair, an siRNA screen revealed that cancers with defects in the DNA repair pathways, Fanconi Anemia and homologous recombination, are hypersensitive to the inhibition of the WEE1 kinase (21). More recently, CRISPR-Cas9 (25) has emerged as a powerful approach to modify genomic loci with higher efficiency and less off-target effects and has been widely used to interrogate gene function at genome scale (26).

To investigate the roles that kinases have in response to DNA damage, we used CRISPR-Cas9 to produce an isogenic panel of cell lines deleted for all non-essential and expressed kinases (some 313 in total) in the human near-haploid cell line HAP1 (25). The cell line HAP1 is derived from the chronic myelogenous leukemia cell line KBM7 and it has been extensively used in biomedical research, especially for investigations into gene–gene and gene–drug interactions (27). Even though the HAP1 cell line carries the BCR–ABL fusion, a driver mutation of chronic myeloid leukemia, it does not depend on the translocation for growth, making it a suitable cell model for the identification of gene–drug interactions (25). This was confirmed by revealing the known hypersensitivities of 15 DNA repair defects to an array of 14 DNA damaging agents (25). The collection of kinase knock-out cell lines was designed to comprise kinases that function in different oncogenic pathways, categorized by their roles in cancer (Fig. 1A). Included in the collection is the PI3K family that has been implicated in 30%–50% of human cancers (28). Similarly, genetic alterations in other kinases such as ALK, JAK, c-KIT, FGFR1 or SRC regulate fundamental molecular mechanisms for tumor cell growth and development (29). Apart from tumor initiation, the collection of kinase-deficient cell lines was designed to also include kinases that are important for tumor survival and proliferation. This category includes members of the EGFR family, such as the ERBB family receptor tyrosine kinases, widely implicated in cancer development and progression (6, 30). Other examples include those that induce tumor cell survival, such as mTOR and the S6 and MEK families. A further category of kinases implicated in cancer includes those that are overexpressed in tumors and surrounding tissues, and hence of importance in the maintenance of tumors. These include, but are not limited to, the FGFR kinase and CK2 (2). Overall, the 313 kinases that were targeted effectively represent oncogenic kinases involved in multiple hallmarks of cancer, including rapid proliferation, growth, survival, and metastasis. Importantly, nine out of the 12 kinases that have been approved by the FDA as therapeutic targets are part of this collection (31), including BRAF, ABL1, CDK4, and CDK6. In addition, of the 23 kinases currently under evaluation as therapeutic targets in current clinical trials (31), 11 are included in this collection, such as AKT1/3, BRD4, and ERBB3.

To uncover combinatorial synthetic lethal and resistant interactions, but also to identify novel kinases in the context of the DNA-damage response and repair, these isogenic cell lines were next individually treated with 10 different DNA-damaging chemotherapeutic agents. These agents were carefully selected to span the engagement of all DNA repair pathways (Fig. 1B). The topoisomerase II inhibitors, etoposide (32) and doxorubicin (33), were chosen to induce DNA double-strand breaks, whereas DNA single-strand breaks were induced by the topoisomerase I inhibitors, topotecan (34) and camptothecin (35). Aphidicolin is a polymerase α inhibitor and it was chosen to induce DNA replication stress. Other agents that induce replication stress are hydroxyurea, that contributes to the depletion of the nucleotide pool and cytarabine, a cytidine analogue (36). Decitabine, another cytidine analogue, was chosen to mediate DNA hypomethylation (37). Carmustine was used as a DNA alkylation agent (38) and 7-hydroxystaurosporine as a CHK1/2 inhibitor, for checkpoint abrogation (39). A similar drug panel, covering all classes of chemotherapy consisting of 29 drugs, was used by Hu and colleagues (40) to map the impact on cellular survival upon depleting (using siRNA) some 625 proteins related to cancer and DNA repair.

The combination of kinase deficiency, coupled with DNA damage, led to the classification of the kinase-deficient cell lines into three different clusters of sensitivity and confirmed several known interactions of kinases to DNA damaging agents (25). The first cluster was defined by kinases that were hypersensitive to carmustine, a chemotherapeutic drug and a bifunctional alkylating agent that produces DNA monoalkylating adducts, as well as DNA intra- and interstrand cross-links. Interestingly, this cluster was significantly enriched for genes associated with increased chromatin accessibility, compared with clusters two and three. Alkylating agents, such as carmustine and temozolomide, have been reported to have a global effect on nuclear organization and chromatin structure, inducing chromatin condensation and gene silencing (41). Hence, it can be reasoned that kinase-deficient cell lines within this cluster are hypersensitive to carmustine due to alkylation-induced synthetic lethality. In support of this hypothesis, this cluster was enriched for gene ontology terms associated with the cellular response to alkylating or crosslinking agents. Such terms included the upregulation of vascular endothelial growth factor receptors (42) and induction of oxidative stress (cellular response to hydrogen peroxide and positive regulation of cytochrome c oxidase activity; ref. 43), which in turn leads to actin cytoskeleton reorganization (44). Kinases within this cluster included EPHB6, DYRK4, and MARK3. The second cluster was dominated by synthetic lethal interactions with hydroxyurea and it was enriched for terms associated with cell adhesion and apoptosis. The third cluster was enriched for kinases that displayed hypersensitivity to the DNA double-strand break inducing agents, etoposide and doxorubicin. It is worth noting that the approach taken here was limited to one cell line and future directions might include using CRISPR-Cas9 approaches targeting the human kinome across a panel of cell lines that would serve to first confirm these findings, but also to identify other potential kinases involved in the resolution of DNA damage.

Different factors can trigger cellular mechanisms of sensitivity to DNA damaging agents, such as an inability to resolve the damage, cell-cycle dysregulation, induction of apoptosis or decreased proliferation. To simultaneously assess these different possibilities in selected isogenic cell lines, a fluorescence-activated cell sorting (FACS)–based phenotypic assay was developed. This was performed using a panel of antibodies and dyes: an antibody detecting the histone variant H2AX phosphorylated at serine 139 (γH2AX) was employed as a marker of DNA damage, the DNA binding dye 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI) was used to monitor cell-cycle phases, the terminal deoxynucleotidyl transferase dUTP nick end labeling staining (also called the TUNEL assay) was used to detect DNA breaks formed during apoptosis and the 5-ethynyl-2′-deoxyuridine (a thymidine analogue) was used to measure proliferation. Twenty-five kinase-deficient cell lines from cluster one (sensitive to carmustine) were selected for further validation using this FACS-based phenotypic assay (Fig. 2A). Because carmustine can function as both a DNA alkylating and crosslinking agent, these 25 cell lines were treated with the alkylating agent temozolomide and the crosslinking agent oxaliplatin, to determine which type of lesions drive the cellular toxicity. Hierarchical clustering revealed that the hypersensitivity to carmustine was predominantly due to DNA alkylation-induced synthetic lethality and not due to the generation of DNA cross-links, since the majority of cell lines were hypersensitive to temozolomide but not to oxaliplatin.

The FACS-based approach also allowed for the identification of four kinases frequently baring loss-of-function mutations in cancer that show cellular sensitivity to temozolomide. These include EPHB6, MARK3, DYRK4, and PNCK (Fig. 2B). MARK3 was prioritized for validation, because it was the second most mutated kinase in cancer, after EPHB6 that has been previously classified as a pseudokinase. As an approach to unravel the mechanism by which MARK3-deficient cells are sensitive to temozolomide, proteomic studies were designed to identify changes in protein abundance in MARK3-deficient cells. This led to the observation that MGMT protein levels were significantly reduced in MARK3-deficient cells (Fig. 2C). The cytotoxicity of temozolomide is mediated by its addition of methyl groups at N⁷ and O⁶ sites on guanines and the O³ sites on adenines in genomic DNA. MGMT is the enzyme that reverses the methylation of the O⁶ position of guanine, hence allowing for cellular survival, therefore, cells lacking MGMT are hypersensitive to temozolomide. Temozolomide-based therapy is the standard of care for patients with glioblastoma multiforme (GBM). Resistance to this drug is mediated by high MGMT levels and several clinical studies have shown that elevated MGMT protein levels or lack of MGMT promoter methylation (and concomitant loss of promoter silencing) are associated with temozolomide resistance in some GBM tumors (45, 46). Hence, the synthetic lethal interaction between MARK3 and MGMT may hold promise for application in the clinics, as a way to revert temozolomide resistance in GBM tumors, through the development of MARK3 inhibitors. Moreover, because MARK3 itself is found to carry loss-of-function mutations in cancer, these findings suggest that such cancers would be hypersensitive to temozolomide and this gene–drug interaction might represent an unexplored avenue for their treatment.

Taken together, kinases represent an important family of enzymes, holding great potential as therapeutic targets for the treatment of cancer. Hence, investigations that systematically unravel interactions between kinases and chemotherapeutic agents are of tremendous value to the scientific community and ultimately to the clinics. Over the coming years, the outcomes of trails consisting of targeting kinases along with the administration of DNA-damaging chemotherapeutics will be known and may lead to new treatment regimes. Another exciting development is the combination of kinase inhibitors with immune checkpoint inhibition. In line with this, several clinical trials are currently investigating the combination of VEGF inhibition along with immune checkpoint inhibitors. The findings from these and related studies open the possibility for new and rational combination therapies that share a remarkable potential to unravel important clinical therapeutic benefit for patients with cancer.

No potential conflicts of interest were disclosed.

We thank Drs. Bensimon (CeMM, Austria), Nagy (CeMM, Austria), and Owusu (IRB Barcelona, Spain) as well as members of the Loizou laboratory for critically reading and commenting on this review. We also thank M. Caldera (CeMM, Austria) for curating the kinome plot. We apologize to all authors whose original research was not cited due to space limitations. J. Ferreira da Silva is funded by a DOC Fellowship (OAW25035). The Loizou laboratory is funded by two grants from the Austrian Science Fund awarded to J.I. Loizou (FWF; P29555 and P29763). CeMM is funded by the Austrian Academy of Sciences. J. Ferreira da Silva is funded by a DOC Fellowship from the Austrian Academy of Sciences (OAW25035). The Loizou laboratory is funded by two grants from the Austrian Science Fund (FWF; P29555 and P29763). CeMM is funded by the Austrian Academy of Sciences.