NOXA-dependent contextual synthetic lethality of BCL-XL inhibition and “osmotic reprogramming” in colorectal cancer

A sophisticated network of BCL-2 family proteins regulates the mitochondria-associated (intrinsic) apoptosis pathway. Antiapoptotic members such as BCL-XL or MCL-1 safeguard the outer mitochondrial membrane and prevent accidental cell death in a functionally redundant and/or compensatory manner. However, BCL-XL/MCL-1-mediated “dual apoptosis protection” also impairs response of cancer cells to chemotherapy. Here, we show that hyperosmotic stress in the tumor environment abrogates dual BCL-XL/MCL-1 protection. Hypertonicity triggers upregulation of NOXA and loss of MCL-1 and thereby enforces exclusive BCL-XL addiction. Concomitant targeting of BCL-XL is sufficient to unlock the intrinsic apoptosis pathway in colorectal cancer cells. Functionally, “osmotic reprogramming” of the tumor environment grants contextual synthetic lethality to BCL-XL inhibitors in dually BCL-XL/MCL-1-protected cells. Generation of contextual synthetic lethality through modulation of the tumor environment could perspectively boost efficacy of anticancer drugs.


Introduction
Elimination of cancer cells mainly relies on activation of the mitochondria-associated (intrinsic) apoptosis pathway. This cell death modality is tightly controlled by a complex network of BCL-2 family proteins. The effector molecules BAX and/or BAK trigger mitochondrial outer membrane permeabilization (MOMP), release of cytochrome c, caspase activation and ultimately condemn a cell to death. BAX and BAK are kept in check by BCL-2, BCL-XL and/or MCL-1 (collectively also referred to as BCL-2-like proteins) 1 . Targeting BCL-2-like proteins emerged as a therapeutic strategy and spurred development of "BH3 mimetics" 2 . This novel class of anticancer compounds mimics the function of BH3-only proteins (the natural antagonists of BCL-2-like proteins) and unlocks the intrinsic apoptosis pathway. The selective BCL-2 inhibitor ABT-199 (venetoclax) proved successful in clinical trials and is meanwhile approved for the treatment of certain leukemia 3 . However, functional redundancy and compensatory roles among the antiapoptotic BCL-2-like proteins can establish "codependencies" (e.g. BCL-XL/MCL-1) that hamper efficient cancer cell elimination when a single BCL-2-like protein is targeted. Preclinical studies demonstrated that inhibition of two or more BCL-2-like proteins was required to overcome this limitation [4][5][6][7] . Our previous work showed that cancer cells facing a hypertonic environment exhibited a lower threshold for MOMP induction [8][9][10] . We hypothesized that this could reflect hypertonicity-induced alterations in the BCL-2 family network. How hyperosmotic stress affects cancer cells has not been comprehensively investigated, but reports document enhanced cisplatin sensitivity, secretion of angiogenesis-promoting cytokines and upregulation of resistance-or metastasisassociated proteins [11][12][13][14] .
Here, we show that hypertonicity can alter codependencies of cancer cells on BCL-2-like proteins. Hyperosmotic stress shrinks dual BCL-XL/MCL-1 protection to exclusive BCL-XL addiction by upregulation of NOXA and concomitant loss of MCL-1. This renders selective BCL-XL inhibition sufficient for activation of the intrinsic apoptosis pathway. Functionally, "osmotic reprogramming" of the tumor environment grants contextual synthetic lethality to BCL-XL inhibitors in dually BCL-XL/MCL-1-protected cancer cells. Perspectively, our findings could boost the efficacy of anticancer drugs.

Hyperosmotic stress renders selective BCL-XL inhibition sufficient for cell death induction
In previous work, we showed that hyperosmotic stress facilitates MOMP induction [8][9][10] . Hyperosmotic stress (or hypertonicity) occurs when osmotically active solutes (such as NaCl) cannot passively diffuse across the plasma membrane and thus establish an osmotic pressure gradient between the extra-and intracellular space. Notably, hypertonicity and BCL-XL-selective BH3 mimetics synergized in cell death induction and abrogated the necessity of concomitant MCL-1 inhibition (Fig. 2a-c, Supplementary Table 2). In contrast, inhibition of MCL-1 was not cytotoxic (Fig. 2d). Targeting BCL-XL under hyperosmotic stress was associated with annexin-V/7-AAD positivity (Fig. 2e) and increased activity of the effector caspases 3 and 7 (Fig. 2f) while loss of BAX/BAK rescued cells from cytotoxic effects (Fig. 2g). Collectively, our data suggested that hyperosmotic stress counteracted the antiapoptotic function of MCL-1 and thereby rendered BCL-XL inhibition sufficient for apoptosis induction. Functionally, combination of BCL-XL inhibitors and hyperosmotic stress resulted in contextual synthetic lethality.
Hypertonicity-imposed contextual synthetic lethality is selective for BCL-XL-targeting drugs We assessed whether other drugs also displayed enhanced cytotoxicity under hypertonic conditions and screened a library comprising 110 well-characterized apoptosis-inducing compounds. Hypertonicity only boosted cancer cell killing of BCL-XL targeting drugs such as A1155463, A1331852, ABT-737 and navitoclax (ABT-263), but not other compounds (Supplementary Fig. 1 and Supplementary Tables 3, 4). Hyperosmotic stress apparently granted contextual synthetic lethality specifically to BCL-XL targeting drugs.

NOXA upregulation is critical for hypertonicity-enforced BCL-XL addiction
We next investigated whether the hypertonicityinduced shift from BCL-XL/MCL-1 codependency to exclusive MCL-1 addiction (Fig. 2a-c) was due to changes in the network of BCL-2 family proteins. We first focused on transcriptional changes under hyperosmotic stress. Exposing HCT116 cells to 60 mM NaCl for 1−4 h caused upregulation of BIM, NOXA and (to a lesser extent) PUMA (Fig. 3a). These early hypertonicity-induced changes did not affect interaction of BCL-XL with BAX and proapoptotic BH3-only proteins (Fig. 3b), but slightly reduced binding of BIM to MCL-1 (Fig. 3c). After 24 h, BIM and PUMA upregulation was even more prominent along with enhanced transcription of BID, MCL-1, BCL-XL, BAX and BAK (Fig. 3a). BAX is essential for MOMP in HCT116 cells and its activation is controlled by interaction with MCL-1 and BCL-XL 5,22,23 . Coimmunoprecipitation experiments, however, failed to demonstrate hypertonicity-induced changes of BCL-XL/MCL-1 interaction with BAX (Fig. 3d). We next investigated the functional relevance of hypertonicity-induced BCL-2 family protein upregulation for exclusive BCL-XL addiction. Genetic loss of the directly BAX-activating proteins BID or BIM failed to rescue NaCl/WEHI-539-treated cells 24,25 whereas loss of BAX (but not BAK) expectedly did (Fig. 3e). PUMA-deficiency somewhat attenuated NaCl/WEHI-539 cytotoxicity. Surprisingly, deficiency in the MCL-1 interacting BH3-only Protein NOXA granted almost full-blown protection to NaCl/WEHI-539 treatment ( Fig. 4a and Supplementary Fig. 2a) 26 . Notably, NOXA-deficient cells still relied on BCL-XL/MCL-1 to safeguard mitochondrial integrity as they were readily killed upon dual BCL-XL/MCL-1 inhibition (Fig. 4b). In line with the transient transcriptional upregulation of NOXA (Figs. 3a and 4f), protein levels peaked approximately 4 h after addition of NaCl and subsequently declined (Fig. 4c, e). This could reflect proteasomal degradation as NOXA is efficiently eliminated via the proteasome and proteasome inhibition consequently caused NOXA accumulation (Fig. 4d) 27 . Hyperosmotic stress did not prolong the half-life of the relatively  short-lived NOXA (Fig. 4g), again indicating that the observed transient NOXA accumulation is most probably attributable to increased transcription. Transcriptional upregulation of NOXA has previously been linked to ER stress (with subsequent activation of the transcriptional regulators ATF4 and ATF6) and/or p53 28,29 . However, exposure to 60 mM NaCl for 4 h showed no signs of ATF4 and ATF6 activation (Fig. 5a). Hypertonicity-mediated NOXA-induction also occurred in a p53-independent manner. Addition of NaCl increased p53 mRNA and protein levels (Fig. 5b, c), but p53-deficiency still allowed NOXA upregulation upon NaCl treatment (Fig. 5d). Together, our data demonstrated that transient transcriptional upregulation of NOXA by a not-yet identified Immunoprecipitates were analyzed together with the corresponding lysates by western blotting using antibodies specific for the indicated proteins. The BCL-XL inhibitor WEHI-539 and the MCL-1 inhibitor S63845 served as positive/negative controls. d HCT116 cells were challenged with NaCl (60 mM) for 24 h or left untreated. Immunoprecipitation was performed as described in (b). For (b-d), data shown are representative of at least two experiments performed. e HCT116 cells and BID-, BIM-, PUMA-, BAX-, and BAK-deficient variants thereof were challenged with the indicated concentrations of WEHI-539 for 18 h in the presence and absence of NaCl (60 mM). Data points and mean ± SEM from three independent experiments are shown. mechanism was essential for contextual synthetic lethality of hypertonicity and BCL-XL inhibition.

Accumulation of NOXA is accompanied by decline of MCL-1 levels
Thus far, we demonstrated that hypertonicity (a) facilitated MOMP induction, (b) shrank dual BCL-XL/ MCL-1 protection to exclusive BCL-XL addiction and (c) triggered upregulation of NOXA, a MCL-1 interacting BH3-only protein. We next assessed the interrelations of these observations. NOXA is capable to facilitate or induce MOMP through direct interaction with and activation of BAX or targeting MCL-1 for proteasomal degradation [30][31][32] . Coimmunoprecipitation experiments did not point to a direct NOXA/BAX interaction during hyperosmotic stress (Fig. 6a). However, hypertonicityinduced NOXA upregulation was followed by a decline in MCL-1 levels that recovered when NOXA expression at later time points returned to baseline (Fig. 4c). NOXA can interact with and target MCL-1 for proteasomal degradation [33][34][35][36] . Indeed, we observed that NOXA-deficiency significantly impaired decrease of MCL-1 levels under hyperosmotic stress (Fig. 6b). However, MCL-1 levels started to decline as early as 2 h after exposure to NaCl ( Fig. 6b and Supplementary Fig. 2b), whereas NOXA upregulation was only detectable after 4 h (Fig. 6b). Additionally, coimmunoprecipitation experiments showed reduced (rather than the expected enhanced) binding of NOXA to MCL-1 under hypertonic conditions (Fig. 3c). These observations suggested that mechanisms other than NOXA upregulation (e.g., translational repression 37 ) might account for or contribute to loss of MCL-1 during hyperosmotic stress. As hypertonicity-induced NOXA upregulation peaked approximately 4 h after addition of NaCl and subsequently declined (Fig. 4c), NOXAmediated contextual synthetic lethality of hyperosmotic stress and BCL-XL inhibitors should depend on the timing of hypertonicity-induction and BCL-XL inhibition. Indeed, NOXA-proficient cells displayed enhanced WEHI-539 cytotoxicity upon simultaneous NaCl/WEHI-539 treatment. However, preincubation with NaCl for 18 h allowed re-adjustment of NOXA levels to baseline (Fig. 4c, e) and BCL-XL inhibition was consequently not cytotoxic (Fig. 6c). NOXA-deficiency expectedly protected HCT116 cells from WEHI-539-mediated cytotoxicity in presence of NaCl. Our data thus suggested that hyperosmotic stress temporarily and inversely affected cellular levels of MCL-1 and NOXA. Functionally, this resulted in transient exclusive BCL-XL dependency.

Discussion
Our study demonstrates that hypertonicity-induced, transient upregulation of NOXA counteracts the antiapoptotic function of MCL-1 and temporarily renders dually BCL-XL/MCL-1-protected cells exclusively BCL-XLdependent. During this period, concomitant targeting of BCL-XL is sufficient to initiate MOMP (Fig. 6d). Functionally, we demonstrate contextual synthetic lethality of BCL-XL inhibitors in an exogenously modified, hypertonic tumor environment ("osmotic reprogramming") and provide insight into the underlying molecular mechanism that critically involves NOXA. This BH3-only protein determines sensitivity to BCL-XL targeting BH3 mimetics in various cancer entities 32,[38][39][40][41] . In previous work, we showed that hyperosmotic stress is capable to facilitate activation of the intrinsic apoptosis pathway [8][9][10]42 . However, in these studies we repeatedly failed to detect hypertonicity-induced disturbances in the BCL-2 protein family network. Retrospectively, this was for the simple reason that we left early, transient changes out of consideration. Instead, we looked for alterations after 18 or 24 h of NaCl treatment, when cytotoxic effects of BCL-XL inhibitor became apparent. Our reasoning that hypertonicity-induced changes in cellular levels of BCL-2 family proteins should be observable at the time of measurable cytotoxicity proved false. The current study highlights that hypertonicity-induced upregulation of NOXA and decline in MCL-1 levels are both transient events. Consequently, exclusive BCL-XL addiction and contextual synthetic lethality of BCL-XL inhibitors is restricted to a certain window (Fig. 6d). A transient decrease in cellular MCL-1 levels under hypertonic conditions was already observed earlier 37 . We also noticed that (in contrast to NOXA) hypertonicity-induced upregulation of BIMencoding mRNA (Fig. 3a) did not translate into increased protein levels (Fig. 3b,c). This could reflect translational repression and/or changes mRNA stability 37,43 . Interestingly, BCL-XL inhibitors induced apoptosis under hyperosmotic conditions in the absence of BIM (Fig. 3e). A previous study, however, convincingly demonstrated that NOXA-triggered release of BIM from MCL-1 was essential for BAX/BAK activation 44 . Our results resemble findings from another study that showed dispensability of BIM (and other BH3-only proteins) for apoptosis induction when MCL-1 and BCL-XL are inhibited 5 . Whether attenuated cytotoxicity of NaCl/WEHI-539 in PUMA-deficient cells indicates involvement of PUMA in hypertonicity-enforced BCL-XL addiction requires further studies (Fig. 3e).
Biophysical factors in the environment of tumors (such as low oxygen levels or elevated pressure) enforce adaption of cancer cells to ensure survival. The resulting cellular processes have thus far mainly been recognized as drivers of non-cell-autonomous resistance to treatment 45 . Vice versa, interfering with these environment-imposed adaptive cellular responses is often lethal for cancer cells and is known as contextual synthetic lethality 46 . Indeed, environment-based contextual synthetic lethality has been recognized as potential therapeutic target [46][47][48] . Our current study advances the concept of contextual synthetic lethality inasmuch as we exogenously modified biophysical properties in the tumor environment ("osmotic reprogramming") in a way that allowed hijacking a cancer cell's adaptive response. Therapeutic exploitation of our findings has of course limitations. For practical reasons, artificial generation of hyperosmotic stress around tumors is conceivable in solid cancers that are easily reachable, but hardly for e.g. leukemia. Notably, our data indicate that exposure of cancer cells to osmotically active solutes for just a couple of hours is sufficient to promote NOXA upregulation (Figs. 3a and 4c). Even transient hypertonicity could therefore be sufficient to render concomitant BCL-XL inhibition deleterious for cancer cells. Moreover, contextual synthetic lethality of hyperosmotic stress and selective inhibition of BCL-2-like proteins might only be achievable in cells that are (a) capable to upregulate NOXA and (b) exhibit a codependency that can be targeted by NOXA (e.g. MCL-1 or BFL-1). While we found clear evidence for hypertonicity-induced, transient transcriptional upregulation of NOXA (Fig. 3a), the underlying signaling cascades and transcription factor(s) involved remain to be determined. Our approach additionally requires codependency on BCL-2-like proteins that are druggable. There are to date no selective inhibitors for BCL-W, which can render cancer cells nonresponsive to BCL-XL and MCL-1 inhibition 4 . However, tools to individually decipher the dependencies of cancer cells on BCL-2-like proteins are available and allow (at least to some extent) to predict treatment response beforehand 2,49 . Nevertheless, our study provides evidence that reprogramming biophysical factors in the tumor environment can alter dependency of cancer cells on BCL-2-like proteins and thereby grant contextual synthetic lethality to BH3 mimetics.

Material and methods
Cell lines, antibodies and reagents

MTT-based cell viability assay
Cells were seeded in 96-well plates (2 × 10 4 cells/well) and challenged with the indicated concentrations of the indicated substances in duplicates (technical replicates). Unless indicated otherwise, cell viability was determined 18 h after stimulation using MTT staining (2 h at 37°C). Staining intensity was measured at 595 nm and the mean was calculated from the technical replicates of each experiment. The mean value for untreated controls was set to 100%. For any other condition, the MTT staining intensity is given relative to the corresponding untreated group (% of control). Data points shown are mean values (calculated from two technical replicates) of independent experiments (n ≥ 2).

Western blot analysis
Cells were harvested by centrifugation and lysed in 4× Laemmli sample buffer (8% [w/v] SDS, 0.1 M dithiothreitol, 40% [v/v] glycerol, 0.2 M Tris, pH 8.0) supplemented with phosphatase inhibitor cocktails-I and -II (Sigma). Samples were sonicated and boiled for 5 min at 96°C before proteins were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes. To block nonspecific binding sites, membranes were incubated in Tris-buffered saline containing 0.1% (v/v) Tween 20 and 5% (w/v) dry milk before primary antibodies of the specificity of interest were added. Antigen−antibody complexes were visualized using horseradish peroxidaseconjugated secondary antibodies (Dako, Hamburg, Germany) and ECL technology (Pierce, Rockford, IL, USA).

Coimmunoprecipitation
HCT116 cells (8 × 10 6 cells per condition) were challenged with the indicated substances for the indicated periods. After washing with ice-cold PBS, cell were lysed in MCBL-buffer (NP40 0.5% [v/v], 150 mM NaCl, 50 mM Tris, pH 7.4) supplemented with complete TM protease inhibitor cocktail (Roche, Mannheim, Germany) for 40 min on ice. Lysates were cleared by centrifugation (20,000 × g, 20 min, 4°C) and incubated overnight with a 1:200 dilution of BCL-XL-, MCL-1-, BAX-and NOXAspecific antibodies at 4°C. The next day, antigen-antibody complexes were precipitated using protein G agarose (Roche). After washing in lysis buffer, agarose-bound proteins were eluted by incubation at 95°C in Laemmli sample buffer (10 min) and analyzed together with the corresponding lysates by western blotting.

Flow cytometry
Cell death was assessed by annexin-V and 7aminoactinomycin D (7-AAD) staining. In brief, HCT116 cells were challenged with WEHI-539 and S63845 for 24 h or left untreated. Afterwards, cells were stained with 7-AAD and annexin-V (4°C for 15 min in the dark) and analyzed immediately using a FACSCanto flow cytometer (BD Biosciences, Heidelberg, Germany) following standard procedures 55 .