A method to identify physical, chemical and biological factors that trigger Bcl-xL-mediated apoptosis

We demonstrated recently that different viruses and transfected viral RNA or plasmid DNA killed human non-malignant cells sensitized with Bcl-xL-specific inhibitor A-1155463. Here, we show that DNA-damaging agent 4-nitroquinoline-1-oxide (4NQO) killed human non-malignant as well as malignant cells and the small roundworm C. elegans when combined with A-1155463, but not with Bcl-2or Mcl-1-specific agents. The synergistic effect of 4NQO-A-1155463 combination was p53 dependent and was associated with the release of Bad and Bax from Bcl-xL, indicating that Bcl-xL linked DNA damage response pathways, p53 signalling and apoptosis. Other anticancer drugs (i.e. amsacrine, SN38, cisplatin, mitoxantrone, dactinomycin. dinaciclib, UCN-01, bortezomib, and S63845), as well as birth-control drug 17α-ethynylestradiol, immunosuppressant cyclosporin, antiviral agent brincidofovir, DNA binding probes MB2Py(Ac), DB2Py(4) and DBPy(5) and UV radiation also killed A-1155463-sensitized non-malignant cells. Thus, we established a method to identify physical, chemical and biological factors, which trigger Bcl-xL-mediated apoptosis. The method could be used in the development of novel anticancer therapies based on systemic Bcl-xL-specific inhibitor and local radiation, oncolytic virus infection or chemotherapy.


INTRODUCTION
Apoptosis is a tightly regulated process that kills cells with severely damaged or invasive DNA, RNA, and proteins (Elmore, 2007. When apoptosis is inhibited, cells that should be eliminated may persist and become malignant (Wong, 2011). B-cell lymphoma 2 (Bcl-2) family of proteins are key players in apoptosis (Kale, Osterlund et al., 2018). For example, Bcl-xL, Bcl-2 and Mcl-1 are anti-apoptotic, whereas Bax, Bak and Bad are proapoptotic members of the family (Denisova, Kakkola et al., 2012, Fernandez, Sebti et al., 2018, Kuivanen, Bespalov et al., 2017. Interaction between pro-and anti-apoptotic proteins determine the fate of a cell. In particular, alteration of the interactions could lead to release of Bax and Bak which could form a pore in the mitochondrial outer membrane that allowed cytochrome C to escape into the cytoplasm and activate the caspase cascade (Shamas-Din, Kale et al., 2013).
Several chemical inhibitors were developed to bind anti-apoptotic of Bcl-2 proteins and induce cancer cells death (Adams, Clark-Garvey et al., 2018). These small molecules belong to several structurally distinct classes (Fig. S1a). First class of Bcl2i includes ABT-737 and its derivatives  and ; second class includes WEHI-539 and its derivatives, A-1331852 and A-1155463. There are other classes of Bcl-2 inhibitors containing structurally similar molecules, such as S63845 and S64315, S55746 and A1210477.
The inhibitors have different affinities to Bcl-2 protein family members (Fig. S1b). For example, ABT-199 has a high affinity to Bcl-2; WEHI-539, A-1331852 and A-1155463 are specific for Bcl-xL; S63845, S64315, S55746 and A1210477 have high affinity to Mcl-1; whereas ABT-263 binds Bcl-2, Bcl-xL and Bcl-w with similar affinity. Importantly, S63845 and several other Bcl-2 inhibitors are currently in clinical trials against blood cancers and solid tumours (Casara, Davidson et al., 2018, King, Peterson et al., 2017, Korycka-Wolowiec, Wolowiec et al., 2019, Timucin, Basaga et al., 2019) (NCT02920697). These drugs provide opportunities for treatment of hematologic and other types of cancers, but also create new challenges associated with emerging drug resistance of cancer cells and toxicity for non-cancer cells (e.g. thrombocytopaenia).
To obtain additive or synergistic effects, enhance efficacy of treatment and combat genetically heterogeneous cancers, Bcl-2 inhibitors were combined with other anticancer drugs ( Fig. S1c; bcl2icombi.info) (Chen, Jin et al., 2011, Haikala, Anttila et al., 2019, Jeong, Oh et al., 2019, Shen, Li et al., 2018. The drug combinations were also used to lower the dose of Bcl-2 inhibitors to overcome resistance and toxicity issues for non-malignant cells (Adams et al., 2018). Dozens of the drug combinations have been reported to be active in vitro (cell culture, patient-derived cells or organoids) and in vivo (patient-derived xenograft mouse models). In addition, 109 combinations (excluding combinations with biological agents) were in clinical trials. However, only ABT-199 in combination with azacytidine, decitabine or cytarabine was approved for the treatment of acute myeloid leukaemia (AML).
Clinical trials with 31 combinations have been terminated, withdrawn or suspended due to adverse effects or other issues. These include trials with ABT-263 plus bendamustine and rituximab in patients with relapsed diffuse large B-cell lymphoma, ABT-263 plus abiraterone acetate with or without hydroxychloroquine in patients with progressive metastatic castrate refractory prostate cancer, as well as ifosfamide,carboplatin,etoposide), or plus rifampin, or plus bortezomib and dexamethasone, ixazomib and carfilzomib (NCT02471391, NCT01423539, NCT03064867, NCT01969682, NCT02755597, NCT03314181, NCT03701321). The identification of the adverse effects prior to clinical trials is challenging but could save lives for many cancer patients.
We noticed that many Bcl-2 inhibitors have been combined with compounds, which damaged cellular DNA, RNA or proteins by targeting DNA replication, RNA transcription and decay, as well as protein signalling and degradation pathways. Here, we shed new light on the mechanisms of actions of such combinations and establish a method for identification of chemical, physical and biological factors which trigger Bcl-xL-mediated apoptosis.
First, we showed that combination of DNA-damaging agent 4-nitroquinoline-1-oxide (4NQO) together with Bcl-xL-specific inhibitor A-1155463, but not drugs alone, were toxic for human malignant and non-malignant cells, as well as for C. elegans. The synergistic effect of the drug combination was dependent on p53 expression and associated with the release of Bad and Bax from Bcl-xL. Second, we demonstrated that several anticancer drugs (i.e. amsacrine, SN38, cisplatin, mitoxantrone, dactinomycin. dinaciclib, UCN-01, bortezamib, and S63845), as well as birth-control drug 17αethynylestradiol, immunosuppressant cyclosporin, antiviral agent brincidofovir, DNA binding probes MB2Py(Ac), DB2Py(4) and DBPy(5), and UV radiation triggered Bcl-xl-mediated apoptosis in nonmalignant cells. Finally, we discussed the application of the method for the development of novel options for treatment of cancer and other diseases.

Toxicity of A-1155463-4NQO combination in C. elegans
We tested DNA-damaging agent 4NQO (Walker & Sridhar, 1975) plus Bcl-xL-specific inhibitor A-1155463 in C. elegans. The worms treated with A-1155463-4NQO combination died faster than those treated with either A-1155463 or 4NQO (Fig. 1a, b). Moreover, the worms treated with the drug combination exhibited defects in reproduction and development (Fig. 1c). The Bliss synergy scores for adult development, L4 development and egg hatching were 19, 21 and 24 respectively. Treatment with A-1155463 or 4NQO alone did not affect these stages. Thus, combination of 4NQO with A-1155463 had a severe impact on the C. elegans lifespan, reproductive system and development.  After 24 h cell viability was measured using the CTG assay. Synergy scores were quantified based on the ZIP model.

Bax from Bcl-xL in RPE cells
Immunoblot analysis of whole-cell extracts showed that expression of p53, a key regulator of DNAdamage response and Bcl-xL-dependent apoptosis, was induced after 2 h of treatment with 4NQO, but not with A-1155463 in RPE cells (Fig. 3a). Confocal microscopy confirmed this observation. Of note, p53 accumulated in the nucleus of 4NQO-treated cells (Fig. 3b).

Immuno-precipitation of Bcl-xL-interacting partners showed that A-1155463 displaced Bad and
Bax from Bcl-xL in RPE cells after 2 h of treatment (Fig. 3a). Confocal microscopy revealed that Bad dislocated from mitochondria, whereas Bax accumulated in the nucleus of A-1155463-treated cells (Fig. 3c,d). were obtained 2 h after treatment. Proteins were immuno-precipitated by anti-Bcl-xL antibody. P53, Bcl-xL, Bad, Bax, and GAPDH were analysed using Western blotting in WCE and immunoprecipitates.
(b-d) RPE cells were treated as for (a). Two hours after treatment cells were fixed, and p53, Bcl-xL, Bad, Bax and Tom40 (mitochondria) were stained with corresponding antibodies. Nuclei were stained with DAPI. Cells were imaged using a confocal microscopy. Representative images (n=8) were selected. Scale bar, 20 μm.
The 4NQO-A-1155463 combination induced expression of p53. Moreover, the combination displaced Bad and Bax from Bcl-xL retaining both proteins in the cytoplasm of RPE cells (Fig. 3). Both events could be critical for initiation of apoptosis (Kakkola, Denisova et al., 2013, Lai, Chi et al., 2007. Of note, treatment of RPE cells for 2 h with either 4NQO, or A-1155463 or in combination did not substantially affect general transcription and translation, as well as expression of several apoptotic proteins and phosphorylation of protein kinases (Fig. S2).

p53 is dispensable for A-1155463-4NQO synergy
To confirm the role of p53 in A-1155463-4NQO synergy, we used malignant HCT116 TP53 -/cells, which lacked p53 expression (Fig. 4a). A-1155463-4NQO combination had substantially lower effect on viability, death and early apoptosis of TP53 -/cells ( Fig. 4b-d). By contrast, A-1155463-4NQO combination killed malignant HCT116 TP53 +/+ cells, at the same concentration as RPE cells. These results indicated that p53 was essential for A-1155463-4NQO synergy. 0.1% DMSO. 2 h after treatment p53 and GAPDH were analysed using western blotting of whole-cell extracts. (b) Cells were treated as for (a). Cell viability was measured by the CTG assay. Mean ± SD, n = 3. (c) Cells were treated as for (a). Cell death was measured by the CTxG assay. Mean ± SD, n = 3. (d) Cells were treated as for (a). Apoptosis was measured by the Annexin V assay. Mean ± SD, n = 3.

RNA-and protein-damaging agents for RPE cells
We next tested combinations of A-1155463, A-133852, ABT-199, and S63845 with 39 DNA, RNA and protein damaging agents in RPE cells. The experiment revealed that A-133852 and A-1155463 in combination with these agents were highly synergistic, indicating that Bcl-xL was essential for induction of apoptosis in non-malignant cells under chemical insults (Fig. 5a, b). Importantly, hit 8 compounds induced expression of p53 after 2 h of treatment (Fig. 5c). Interestingly, combinations of S63845 with A-1155463 or A-133852, but not with ABT-199, were highly synergistic and toxic for nonmalignant cells indicating that Mcl-1 or another potential cellular target could be damaged by S63845 (Fig. 5d). cell viability was measured using the CTG assay. The compounds with synergy >7.5 were selected and plotted against their targets. ZIP method was used to calculate synergy scores for all panels.
After 24 h cell viability was measured by the CTG assay (n=3).
Given that DB2Py(4) and DB2Py(5) differed from DBA(n) molecules by the pyrrole fragment and the N-methylpiperazine end group (instead of dimethylaminopropylamide; Fig. S3), these differences appeared to be essential for initiation of apoptosis in A-1155463-sensitized RPE cells. Moreover, dimerization of B2Py(n) was not essential for Bcl-xL-mediated apoptosis, since both MB2Py(Ac) and DB2Py(n) had similar synergy scores. Most probably, MB2Py(Ac) bound minor groove of cellular DNA and altered pol II transcription, which led to accumulation of aberrant RNA transcripts, that triggered apoptosis in A-1155463-sensitized RPE cells. Thus, the MB2Py(Ac), DB2Py(4) and DB2Py (5) possessed DNA-binding activity that could interfere with general transcription.
We further exploited Bcl-xL-mediated apoptosis for identification of chemical agents that damaged cellular DNA, RNA or proteins. We tested 48 drugs commonly dispensed in Norway and 50 safe-inman broad-spectrum antiviral agents. We found that 17α-ethynylestradiol, bortezomib, cyclosporin and brincidofovir primed A-1155463-sensitized RPE cells for apoptosis (Bliss synergy score >6).
These results indicated that 17α-ethynylestradiol, cyclosporin and brincidofovir could damage cellular DNA, RNA or proteins and subsequently could induce apoptosis (Fig. 6b,c). Moreover, we confirmed that bortezomib impaired protein degradation, which could be associated with accumulation of damaged proteins, which triggered Bcl-xL-mediated apoptosis (Kim, Song et al., 2014).

Exploiting p53-dependent Bcl-xL-mediated apoptosis for identification of physical factors that damage cellular DNA, RNA or proteins
We next exposed A-1155463-sensitized and non-sensitized RPE cells to UVB and UVC radiation, which similarly to the 4NQO treatment, induced lesions in DNA (Walker & Sridhar, 1975). After 24 h the cell viability was measured using the CTG assay. We observed that already 8 sec of exposure UV killed A-1155463-sensitized but not non-sensitized RPE cells (Bliss synergy scores: 31.2 and 23.8, respectively; Fig. 7). This indicated that UVB and UVC triggered Bcl-xL-mediated apoptosis. Figure 7. UV radiation kills A-1155463-sensitized, but not non-sensitized RPE cells. RPE cells were exposed to UVB (a) and UVC (b) radiation for the indicated times and covered with medium containing 1 μM A-1155463 or 0.1% DMSO. After 24 h, viability of cells was measured using the CTG assay. Mean ± SD, n = 3.
In addition, we demonstrated that DNA-damaging agent 4NQO killed human non-malignant and malignant cells, as well as small roundworm C. elegans, which were sensitized with A-1155463. The cell death was dependent on the concentration of both agents. The synergistic effect of 4NQO-A-1155463 combination was p53 dependent and was associated with the release of Bad and Bax from Bcl-xL. Our results suggest that intracellular receptors sensed damaged DNA, transmitted this information via p53 to anti-apoptotic Bcl-xL, which released pro-apoptotic Bax and Bad to trigger cell death. Thus, Bcl-xL protein could serve as 'safety fuse', which mediated apoptosis, when concentration of damaged cellular factors reached critical levels (Fig. 8).
We showed recently that non-malignant cells infected with viruses or transfected with viral RNA or plasmid DNA were also sensitive to Bcl-xL-specific inhibitor A-1155463 (Bulanova, Ianevski et al., 2017, Kakkola et al., 2013. This indicate that invasive RNA or DNA could also trigger Bcl-xL-mediated apoptosis (Fig. 8).  (Table S2) (Li, Wang et al., 2019, Wu, Schiff et al., 2014. Moreover, ABT-737 sensitized chronic lymphocytic leukaemia cells to reovirus and vesicular stomatitis oncolysis (Samuel, Beljanski et al., 2013). Further in vivo studies are needed to exploit locally-induced Bcl-xL mediated apoptosis for cancer treatment.
In conclusion, we developed a method for identification of physical, chemical and biological factors that induced Bcl-xL-mediated apoptosis. This method could provide novel opportunities for discovery of therapeutic and adverse effects of anticancer medications. It also could be used to identify conditions, which would prevent other non-communicable and communicable diseases as well as aging, given that cells with high concentration of damaged DNA, RNA or proteins or with invasive DNA and RNA molecules would be eliminated from an organism receiving a Bcl-xL-specific inhibitor.

Bcl2iCombi database
We reviewed developmental status of Bcl-2 inhibitors and their combinations with other anticancer therapeutics. We summarized the information in freely accessible database (bcl2icombi.info). The drug annotations were obtained from PubChem, DrugBank, DrugCentral, PubMed and clinicaltrials.gov databases (Table S1) (Kim, Chen et al., 2019, Ursu, Holmes et al., 2019, Wishart, Feunang et al., 2018. The database summarizes activities and developmental stages of the drug combinations. The database allows interactive exploration of drug-drug and drug-target interactions. A feedback form is available on the website. The database will be updated upon request or as soon as a new drug combination is reported.  (7) were synthesized as described previously (Koval et al., 2018).
Monomeric MB2Py(Ac) and dimeric DB2Py(4) and DB2Py(5) benzimidazole-pyrroles were also described previously (Ivanov et al., 2013). A library of 527 approved and emerging investigational oncology drugs were from the collection of the Institute of Molecular Medicine Finland, FIMM (www.fimm.fi/en/services/technology-centre/htb/equipment-and-libraries/chemical-libraries). A library of 48 drugs commonly dispensed in Norway was assembled based on Norwegian Prescription Database (www.norpd.no). Table S1 lists these compounds, their suppliers and catalogue numbers. A library of 50 safe-in-man broad-spectrum antivirals was published previously (Ianevski, Zusinaite et al., 2018). To obtain 10 mM stock solutions, compounds were dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich, Steinheim, Germany) or milli-Q water. The solutions were stored at −80 °C until use.
For lifespan experiments, gravid adult worms were placed on NGM plates containing either A-1155463 (10 µM), 4NQO (10 µM), combination A-1155463 (10 µM) + 4NQO (10 µM) or vehicle control and seeded with OP50 to lay eggs. Progeny were grown at 20°C through the L4 larval stage and then transferred to fresh plates in groups of 30-35 worms per plate for a total of 100 individuals per experimental condition. Animals were transferred to fresh plates every 2-4 days thereafter and examined every other day for touch-provoked movement and pharyngeal pumping, until death.
Worms that died owing to internally hatched eggs, an extruded gonad or desiccation due to crawling on the edge of the plates were censored and incorporated as such into the data set. Survival curves were created using the product-limit method of Kaplan and Meier. The log-rank (Mantel-Cox) test was used for statistical analysis.
A series of toxicity experiments, including fecundity, egg hatching, larval development, were conducted using N2 Bristol isolate C. elegans, which were cultivated as previously described (Brenner, 1974) and maintained at 20°C. Briefly, for the toxicity assay animals were initially synchronised by bleaching gravid adults (adult day 1-4) to extract the eggs. Eggs were placed on nematode growth medium (NGM) plates seeded with Escherichia coli (OP50). L4 larvae were subsequently transferred onto fresh OP50-seeded NGM plates and allowed to grow to adulthood. Ten adult day 1 worms (n = 30-50/experimental condition) were transferred onto assay NGM plates with OP50 containing either A-1155463 (1, 10, 100 µM), 4NQO (1, 10, 100 µM), combination A-1155463 (10 µM) plus 4NQO (2, 10, 20 µM) or vehicle control. Animals laid eggs for three hours. Subsequently, adults were removed from the plate and the frequency of eggs laid was quantified as a measure of reproduction and egglaying capacity of worms. The following day the frequency of unhatched eggs and L1 larvae were counted in order to evaluate the efficiency of egg hatching. Subsequently, 36 h later, development to L4 larvae was assessed, as a measure of larval growth. Finally, growth of L4 larvae to adulthood was quantified after 16 h.
The toxicity assay was conducted at 20°C on 10 ml NGM plates seeded with 100 µl OP50 from an overnight culture. Drug compounds and vehicle solvents were dissolved in a total volume of 200 µl, sufficient to cover the entire surface of the plate, and were dried at room temperature for 1-2 h prior to the transfer of worms. Each chemical concentration was tested three-five times. Statistical analysis was conducted using One-Way ANOVA followed by Tukey's multiple comparison test.

Real-time impedance assay
RPE cells were grown at 37°C in 16-well E-Plates to 90% confluency. The plates were supplied with golden electrodes at the bottom of the wells and a weak electrical current was constantly applied to the cell medium. The changes in the cell adherence indexes (CI) were monitored by the xCELLigence real-time drug cytotoxicity system (ACEA Biosciences, San Diego, USA) as described previously (Solly, Wang et al., 2004). When cells reached 90% confluency, 1 µM 4NQO, 1 µM A-1155463 or their combination were added. Control cells were treated with 0.01% of DMSO. CI were normalized and monitored for another 24 hours.

Cell viability assay
Approximately, 4x10 4 RPE or HCT116 cells per well of 96-well plate were treated with A-1155463, 4NQO or both compounds. Control cells were treated with 0.01% of DMSO. After 24 or 48 h, respectively, the viability of cells was measured using the Cell Titer Glo assay (CTG, Promega).
Luminescence was measured using PerkinElmer Victor X3 or Synergy Mx plate readers.

Cell toxicity assay
Approximately, 4x10 4 HCT116 cells per well of 96-well plate were treated with 1 µM 4NQO, 1 µM A-1155463 or their combination. Control cells were treated with 0.01% of DMSO. After 24 h, the death of cells was detected using the Cell Toxicity Assay (CTxG, Promega). Fluorescence was measured using PerkinElmer Victor X3 Reader.
Activation of apoptosis was assessed using RealTime-Glo Annexin V Apoptosis and Necrosis Assay (Promega). Luminescence was measured using PerkinElmer Victor X3 Reader.

Apoptosis and phospho-kinase arrays
Approximately, 1x10 6 RPE cells per well of 6-well plate were treated with 1 µM 4NQO, 1 µM A-1155463 or their combination. Control cells were treated with 0.01% of DMSO. After 2 h, relative levels of apoptosis-related proteins and protein kinase phosphorylation were determined using proteome profiler human apoptosis and human phospho-kinase array kits, respectively, as described in the manuals (R&D Systems, Minneapolis, MN, USA). Membranes were scanned using Odyssey Li-Cor system.

Metabolic labelling of cellular RNA and proteins
Approximately, 1x10 5 RPE cells per well of 12-well plate were treated with 1 µM 4NQO, 1 µM A-1155463 or their combination dissolved in 500 l cell growth medium. Control cells were treated with 0.01% of DMSO. The medium was supplemented with 3 l of [alpha-P32]UTP (9,25 MBq, 250 Ci in 25 l). Cells were incubated for 2 h at 37 °C. Cells were washed twice with PBS. Total RNA was isolated using RNeasy Plus extraction kit (Qiagen, Hilden, Germany). RNA was separated on 1% agarose gel. Gel was dried. Total RNA was detected using Ethidium Bromide. 32 P-labeled RNA was monitored using radioautography and visualised using Typhoon 9400 scanner (GE Healthcare).
In a parallel experiment, the compounds or DMSO were added to 500 μl cysteineand methioninefree DMEM medium (Sigma-Aldrich, Germany) containing 10% BSA and 3 µl [ 35 S] EasyTag Express protein labelling mix (7 mCi, 259 MBq, 1175 Ci/mmol in 632 ml; Perkin Elmer, Espoo, Finland). After 2 h of incubation at 37 °C cells were washed twice with PBS, lysed in 2 × SDS-loading buffer and sonicated. Lysates were loaded and proteins were separated on a 10% SDS-polyacrylamide gel. 35 Slabelled proteins were monitored using radioautography and visualised using a Typhoon 9400 scanner (GE Healthcare).

Synergy calculations
RPE cells were treated with different concentrations of a Bcl-2 inhibitor and another drug. After 24 h cell viability was measured using CTG assay. To test whether the drug combinations act synergistically, the observed responses were compared with expected combination responses. The expected responses were calculated based on Bliss reference model using SynergyFinder webapplication (Ianevski, He et al., 2017). For in vitro combinatorial experiments where the whole doseresponse matrix was measured, a normalized Bliss reference model, i.e. Zero Interaction Potency (ZIP) model was utilized.
Immuno-precipitation and immuno-blotting RPE cells remained untreated or were treated with 1 μM A-1155463, 1 μM 4NQO or their combination for 2 h. Cells were lysed in buffer containing 20 mM Tris-HCl, 0.5% NP-40, 150 mM NaCl, 1.5 mM MgCl2, 10 mM KCl, 10% Glycerol, 0.5 mM EDTA, pH 7.9 and protease inhibitor cocktail (Sigma). Bcl-xL-associated factors were immuno-precipitated using rabbit anti-Bcl-xL antibody (Cell Signaling Technology) immobilized on the magnetic Protein G Dynabeads (Thermo Fisher Scientific). Normal rabbit IgG (Santa Cruz Biotechnology, sc-2025) was used to control immunoprecipitation with an equal volume of cell lysate. The immobilized protein complexes together with the appropriate wholecell extract inputs were subjected for SDS-PAGE followed by immunoblotting.