Cell‐based high‐throughput screen for small molecule inhibitors of Bax translocation

Abstract Aberrant regulation of programmed cell death (PCD) has been tied to an array of human pathologies ranging from cancers to autoimmune disorders to diverse forms of neurodegeneration. Pharmacologic modulation of PCD signalling is therefore of central interest to a number of clinical and biomedical applications. A key component of PCD signalling involves the modulation of pro‐ and anti‐apoptotic Bcl‐2 family members. Among these, Bax translocation represents a critical regulatory phase in PCD. In the present study, we have employed a high‐content high‐throughput screen to identify small molecules which inhibit the cellular process of Bax re‐distribution to the mitochondria following commitment of the cell to die. Screening of 6246 Generally Recognized As Safe compounds from four chemical libraries post‐induction of cisplatin‐mediated PCD resulted in the identification of 18 compounds which significantly reduced levels of Bax translocation. Further examination revealed protective effects via reduction of executioner caspase activity and enhanced mitochondrial function. Consistent with their effects on Bax translocation, these compounds exhibited significant rescue against in vitro and in vivo cisplatin‐induced apoptosis. Altogether, our findings identify a new set of clinically useful small molecules PCD inhibitors and highlight the role which cAMP plays in regulating Bax‐mediated PCD.

three decades have identified major pathways governing PCD initiation and execution. 3 The first of these termed the intrinsic pathway regulates cellular responses to genotoxic stress, growth factor withdrawal, calcium influx and disruption of the cytoskeletal network, while the second (extrinsic) pathway is initiated following the extracellular binding of death receptor ligands as Fas ligand, tumour necrosis factor-α (TNF-α) and TNF-related apoptosis-inducing ligand (TRAIL). 1 Although the intrinsic and extrinsic PCD pathways are regulated by distinct sets of upstream stimuli, they converge at the level of the mitochondria where interactions between pro-and anti-apoptotic Bcl-2 family members regulate mitochondrial outer membrane permeability (MOMP). 4 Once induced to their active conformation by interactions with BH3-only proteins, Bax and Bak undergo oligomerization and insertion into the mitochondrial outer membrane, promoting release of the mitochondrial second messengers such as cytochrome c, apoptosis-inducing factor (AIF), second mitochondria-derived activator of caspases (Smac)/DIABLO, and endonuclease G from the mitochondrial intermembranous space, modifies the mitochondrial matrix, and accelerates mitochondrial fission, thereby reducing levels of oxidative phosphorylation. 5 Released cytochrome c and Smac/DIABLO subsequently act to promote downstream caspase activation, while AIF and endonuclease G spawn caspase-independent modes of cell death. 3 Bax/Bak activity is opposed by the actions of anti-apoptotic Bcl-2 proteins such as Bcl-2, Bcl-xL, Bcl-w and Mcl-1 through either direct suppression of oligomerization or interaction with a unique subgroup of BH3-only proteins. [6][7][8] The critical role of Bax and Bak in PCD regulation is highlighted by studies of Bax/Bak double knockouts in which cells exhibit substantial protection from a wide array of PCD inducers. 9 Further analysis demonstrates that a majority of cell types depend principally upon Bax to execute MOMP. 10 Thus in a wide array of tissues, Bax function as the primary gatekeeper of mitochondrial MOMP, and ultimately PCD progression.
Genetic and pharmacologic manipulation of Bcl-2 protein-protein interactions in vivo has demonstrated that processes which inhibit Bax oligomerization result in a substantial enhancement of neuronal survival following acute or chronic neural injury. 11,12 Consistent with this, peptides derived from two proteins known to interact with Bax (Ku70 and Bax inhibitor-1) have been shown to exhibit some ability to suppress Bax activation and in turn apoptotic cell death. 13,14 However, the administration of such agents exhibit significant clinical challenges due to their low bioavailability and stability in vivo. The possibility of identifying small molecule Bcl-2-family modifiers is suggested from several studies aimed at enhancing apoptotic response through the inhibition of anti-apoptotic Bcl-2 family members. 15 As a result, the small molecules ABT-737 and derivative ABT-263 were identified to inhibit the activity of Bcl-2, Bcl-xL and Bcl-w interactions at nanomolar concentrations and show promise in early clinical trials. 16 Conversely, the ability to inhibit the pro-apoptotic functions of Bax is similarly highly desirable clinically in the context of neural preservation following central nervous system (CNS) damage such as stroke and spinal cord injury.
Previous studies have demonstrated the feasibility of using fully functional enhanced green fluorescent protein (EGFP)-Bax and other PCD fusion proteins as real-time detectors of apoptotic progression. 17 Given that the translocation of Bax from the cytoplasm to the mitochondria serves as an early and readily detectable indicator of functional apoptotic progression, we engineered an EGFP-Bax fusion protein to monitor the relative levels of Bax translocation in a temporal manner. Cell lines stably expressing this fusion were then examined in the context of a high-content, high-throughput chemical library screen following commitment of the cells to die in order to identify potential small molecule Bax inhibitors. From our screen of over 6000 compounds, we identified two Generally Recognized As Safe (GRAS) compounds with similar mechanism of action which promoted substantial reductions in Bax translocation following cisplatinmediated PCD stimulation. Further validation of these agents in vivo demonstrated their ability to suppress apoptotic PCD in the murine brain following cisplatin challenge. Taken together, the results demonstrate that modulation of cAMP signalling using several novel small molecule inhibitors can be used to alter levels of Bax activation and apoptotic cell death in the mammalian CNS.

| EGFP-Bax expression vector
The mouse Bax cDNA was PCR amplified from IMAGE clone 3968903 and cloned into mammalian expression vector pEGFP-C1 (Clontech Laboratories, Inc.) via BglII and EcoRI sites. All clones were sequence verified. Sigma-Aldrich Co., G8168) was utilized for selection at a concentration of 0.8 mg/mL with G418 media changed every 3 days for a period of 2 weeks prior to cloning individual sub-colonies in 24-well plates. Independent isolates were cloned and analysed for levels of EGFP-Bax expression by fluorescent microscopy. Optimal EGFP-Baxexpressing lines were expanded through three serial passages, retested and frozen for long-term cryostorage until use.

| Analysis of Bax translocation
To determine the levels of cellular Bax translocation in each cell, Cellomics ArrayScan HCS images were analysed in a blinded manner using a modified Spot Detection algorithm. Briefly, cells registering a HUI ET AL.  Figure S1). Cells undergoing PCD demonstrated redistribution of EGFP from the cell cytoplasm to punctuate localizations adjacent to the (DAPI + ) cell nucleus. Under these conditions, cells exhibit a dramatic rise in EGFP pixel intensity. Based on the fluorescence intensity in the EGFP channel, the modified Spot Detection      were recorded using a multi-well microplate reader (absorbance 570 nm, reference 690 nm; Molecular Devices, Inc., SpectraMax M2).

| In vivo cisplatin assay
Mice were injected subcutaneously on postnatal day 2 with freshly prepared cisplatin (7 mg/kg) or saline control at a volume of 10 μL/g body weight. Hit compounds or vehicle (ethanol) were then injected 1 hour following cisplatin treatment. For some animals, the caspase-3/7 inhibitor z-DEVD-fmk (Bachem Americas, Inc., 4027402) was injected 30 minutes prior to cisplatin treatment as a positive control.
For these experiments, forskolin, rolipram and DEVD were first dissolved in 100% anhydrous ethanol, then diluted in 0.9% saline to a final concentration of 4 mmol/L (25% ethanol injected at 10 µL/g body weight). Animals were killed at 24 hours following cisplatin treatment with the cerebella carefully removed and fixed in 4% paraformaldehyde with gentle agitation. Cerebella were then rinsed and prepared for paraffin wax sectioning. For each animal, a series of 7-μm parasagittal sections through the vermis were prepared 150 μm lateral to the midline at intervals of 100 μm, six sections for each series. Slides were dewaxed and prepared for TUNEL staining as described previously. 18 The numbers of TUNEL + cells were determined as a function of external germinal layer (EGL) area quantified in lobes VI and VIII of the cerebellum. All procedures were in carried out in accordance with University of Toronto Animal Protocols.

| Statistical analyses
Statistical tests including Student's t test and one-way ANOVA with Bonferroni's multiple comparison test were performed with Microsoft Excel or Prism GraphPad. Data are presented as mean ± SEM.
Results were considered statistically significant if P < 0.05.

| Analysis of EGFP-Bax expressing cells
In order to monitor Bax translocation in real time following PCD stimulation, we first constructed an expression vector producing fulllength Bax fused with EGFP at its N-terminus. Transient and stable transfection of this EGFP-Bax construct into CHO, human embryonic kidney (HEK 293T) cells, or primary murine fibroblasts confirmed correct localization and translocation of the fusion protein following PCD activation. For these studies, PCD was stimulated using staurosporine or ultraviolet C (UVC) irradiation, two well-characterized apoptotic inducers. 19 As shown in Figure 1A  Bax activity has been shown to be directly regulable by both pro-apoptotic BH3 activators and anti-apoptotic Bcl-2 family members such as Bcl-w. 7,8 We, therefore, assessed whether the translocation of this EGFP-Bax fusion protein could be functionally inhibited by expression of Bcl-w. To test this, Bcl-w was co-trans-

| Induction of Bax translocation by cisplatinmediated cell death
In order to generate a practical and reproducible PCD model amenable for high-throughput screening, we examined several chemical inducers of PCD for uniformity of response. Although analyses with staurosporine indicated that this serine/threonine kinase inhibitor induced substantial levels of Bax translocation, it was not ideal for high-throughput screening due to its notable effects in altering cell  We further tested whether Bax translocation could be enhanced by extending the analysis periods to create an even more robust assay. However, as shown in Figure 3D

| Analysis of intra-and inter-plate variability
One of the most critical features of high-throughput single-pass assay screens is the variability observed for both inter-and intraplate assay measures. Such features define both the ultimate limits of assay sensitivity and the likelihood of false-positive and false-negative readouts. As a result, estimates of intra-and inter-plate variability were determined in order to assess the relative sensitivity of the assay constructed. Replicates of stable EGFP-Bax CHO cells were plated onto six 384-well plates and subjected to PCD stimulation under the defined condition of 50 μg/mL cisplatin for 8 hours.
As indicated in Figure 4A, analysis of Bax translocation based upon

| Identified small molecule inhibitors of Bax translocation reduce executioner caspase activity and enhance cell survival
In order to assess the ability of forskolin, colforsin and rolipram to suppress the features of PCD downstream of Bax translocation, several viability assays were performed. To mirror the conditions of our high-throughput screen, cells were exposed to cisplatin (50 μg/mL) for 8 hours prior to washout and treatment of either vehicle or hit compounds (5 μmol/L) for 24 hours before the assays. As shown in Figure 6A, executioner caspase activity (cleavage of fluorogenic DEVD substrate) was significantly inhibited following treatment with forskolin, colforsin or rolipram, consistent with a role for these compounds at or upstream of Bax translocation to inhibit the progression of apoptosis. We further examined the ability of these compounds to elicit cytoprotection by measuring LDH release (plasma membrane integrity) and MTT reduction (mitochondrial activity). As shown in Figure 6B, compounds forskolin and colforsin significantly reduced the levels of cisplatin-induced LDH release, while rolipram exhibited a trend towards reductions in LDH release but did not achieve statistical significance. As shown in Figure 6C, MTT assay revealed modest enhancement of mitochondrial activity following the addition of forskolin or colforsin, but not rolipram, to cisplatin-treated cells. Previously, we have demonstrated that cerebellar granule cells display a profound dependence on caspase-3 mediated apoptosis following NMDA blockade in the early postnatal period. 18 In order to test the in vivo effectiveness of the identified compounds, we examined their potentials to protect cells of the EGL of the cerebellum from cisplatin-induced caspase-3-dependent apoptosis ( Figure 7A).
As shown in Figure 7A and B, injection of compound vehicle induced no significant increase in EGL cell death. By contrast, injection of cisplatin at 7 mg/kg induced substantial EGL cell death ( Figure 7C-D) which was substantially diminished by the application of forskolin or rolipram 1 hour following cisplatin treatment ( Figure 7E,F, respectively). Indeed, as quantified in Figure 7G, application of these agents at a concentration of 40 μmol/L at 1 hour following cisplatin treatment reduced levels of EGL cell death below that observed even fol- family proteins such as Bcl-xL and Bcl-B for use as anti-cancer therapeutics. 21,22 For these studies, we incorporated several screening features aimed at enhancing the identification of clinically useful therapeutics. First, we have constructed our primary screen around a cell-based system instead of an in vitro biochemical assay. Such systems select for compounds with appropriate membrane permeability and quickly filter out those possessing significant acute toxic effects.
The use of dynamic core markers of apoptosis such as Bax translocation also allowed direct functional testing of compounds in an unbiased manner providing a means to identify any and all agents which influence the given process regardless of mechanism rather than narrowly focusing on a particular protein-protein interaction for initial selection. We have chosen to examine the ability of small molecules to inhibit Bax translocation and enhance cell viability 8 hours subsequent to cisplatin-induced cell injury in an attempt to identify compounds which do not act on early specific initiation events in PCD per se, but rather more generalized aspects of PCD progression with the aim of targeting a more clinically realistic therapeutic window.
This approach stands in contrast to screening protocols in which functional inducer and small molecule therapeutic are tested coincident with one another, frequently identifying candidate compounds which ultimately fail in real clinical settings. 23 25,26 However, the precise molecular mechanism underlying these observations remains unclear.
Multiple pathways have been suggested, including modulation of p53 phosphorylation, altered expression of Bax and related proapoptotic proteins such as Puma, induction of BDNF expression through its cAMP responsive element binding protein and direct alteration of Bax phosphorylation (inhibits its translocation to the mitochondria). [27][28][29] In the present study, we show that the identified cAMP modulators significantly reduced the levels of downstream executioner caspase activity ( Figure 7A). As we have only monitored the activation of Bax, cAMP-independent Bax activation (eg direct interaction with p53) and Bax-independent mechanisms (ie Bak-mediated MOMP) could still have triggered cell death as illustrated by the relative minor protection against the loss of mitochondrial activity shown through MTT reduction assay ( Figure 7C).
Hence, in order to realistically rescue neurons damaged during acute neural injuries, pharmacologic modulation of multiple pathways are likely required.
In the present study, we describe the development of a robust high-throughput screening approach to identify small molecule inhibitors of Bax translocation. Such screening resulted in the F I G U R E 6 Cell death inhibition activity of compounds forskolin, colforsin and rolipram. Cells were exposed to cisplatin (50 μg/mL) for 8 h followed by washout with addition of vehicle or hit compounds (5 μmol/L) to test for efficacy in cell death inhibition. (A) Analysis of caspase-3/7 (DEVD) activity in the presence of cisplatin. All three compounds identified significantly reduced levels of cisplatin-induced executioner caspase activity. (B) Alterations of plasma membrane integrity as monitored by lactate dehydrogenase (LDH) release. Forskolin and colforsin significantly reduce the levels of LDH release compared to cisplatin controls, while rolipram exhibit a non-significant trend toward reduced LDH release. (C) Analysis of mitochondrial activity as measured by MTT reveals significant increases by forskolin and colforsin but not rolipram. (D) Assessment of cell viability by trypan blue exclusion assay at 24 h following treatment. All three compounds exhibited significant protective effects against cisplatin-mediated programmed cell death. Data are presented as normalized mean ± SEM for triplicate experiments for all treatment groups. *, **, and *** indicate statistical significance at P < 0.05, P < 0.01, and P < 0.001, respectively, by one-way ANOVA and Bonferroni's multiple comparison test between the indicated treatment groups. ns indicates not significant statistically Level of TUNEL + early postnatal cell death observed following injection of saline control. Note that in addition to TUNEL + cells (yellow open arrowheads), fluorescence is also observed in red blood cells at this wavelength (identified by their biconcave shape-white arrowheads). (B) Level of cell death observed following injections of saline control and ethanol vehicle after 1 hour. (C-D) Treatment with cisplatin resulted in a substantial rise in level of cell death within the external germinal layer (EGL) as observed by TUNEL, which was significantly reduced by DEVD pre-treatment. (E-F) Injection of forskolin at 1 hour following cisplatin treatment reduced the number of TUNEL + cells observed in the EGL as did similar treatment with rolipram. (G) Relative density of TUNEL + cells within the EGL as observed for each group quantified across four parasagittal sections with an intersection distance of 100 μm. Saline alone group showed no significant difference from saline + ethanol vehicle group. Data represent normalized means ± SEM, n = 6 animals per group. *** indicates statistical significance at P < 0.001 by one-way ANOVA and Bonferroni's multiple comparison test between the indicated treatment groups. ns indicates not significant statistically; scale bar: 100 μm identification of two small molecules regulating cAMP signalling which significantly inhibit Bax translocation at sub-micromolar concentrations. These findings demonstrate that pharmacologic manipulation of cAMP levels in vitro and in vivo alter the levels of neural apoptosis in the presence of cisplatin. However, these findings also highlight the complexities in modifying such injuries in vivo due to the regulated presence of multiple isoforms of PCD (eg necroptosis).
As such, inhibition of several distinct PCD pathways may ultimately be required to provide effective therapeutic neuroprotective responses in vivo.