Bcl-xL inhibits IP3R-mediated Ca2+ release in living cells.
Bcl-xL has been reported to sensitize IP3Rs in living cells . Here, we evaluated the effect of Bcl-xL overexpression on IP3R function by monitoring agonist-induced Ca2+ release (Fig. 1, Fig. S1). First, we performed Ca2+ measurements in a population-based assay, using the ratiometric fluorescent Ca2+ probe Fura-2 (Fig. S1). We used trypsin, an efficient agonist of protease activated receptors 2 (PAR2) in HEK-293 cells , thereby triggering IP3 formation. We elicited IP3R-mediated Ca2+ release in Fura-2-loaded HEK-3KO cells with reconstituted rIP3R1 and we studied the impact of overexpressing Bcl-xL. For this, we transfected the cells with either P2A-mCherry or 3xFLAG-Bcl-xL-P2A-mCherry. The 3xFLAG-Bcl-xL-P2A-mCherry construct generates separate mCherry and 3xFLAG-Bcl-xL proteins due to its P2A self-cleaving sequence. In contrast to previous reports [25, 26], Bcl-xL overexpression significantly reduced the amplitude (Fig. S1a, c) and the area under the curve (AUC) of the Ca2+ signals (Fig. S1b, d) induced by both low (0.1 µM) and high (1 µM) trypsin concentrations. Moreover, similarly to our findings related to Bcl-2 and IP3R function , we noticed that the inhibitory effect of Bcl-xL overexpression was more prominent at low agonist concentrations than at high agonist concentrations.
This unexpected IP3R inhibition by Bcl-xL in population-based Ca2+ measurements prompted us to validate the effect of Bcl-xL overexpression on Ca2+ signals in single cells exposed to other agonists (Fig. 1) and to compare it with Bcl-2 overexpression, an established inhibitory modulator of IP3Rs [20, 22]. Single cell Ca2+ imaging was performed in Fura-2-loaded HEK-293 cells transfected with either P2A-mCherry, 3xFLAG-Bcl-xL-P2A-mCherry or 3xFLAG-Bcl-2-P2A-mCherry, whereby only mCherry-positive cells were analyzed. Here, we used ATP (10 µM) to trigger IP3R-mediated Ca2+ release (Fig. 1a, b, c). Similarly, we observed that Bcl-xL overexpression reduced the percentage of responding cells (Fig. 1d) and the area under the curve as representative of the extent of Ca2+ release (Fig. 1e). Interestingly, Bcl-xL appeared to dampen IP3R-mediated Ca2+ release to a similar extent as Bcl-2. We also measured the effect of Bcl-xL/Bcl-2 overexpression on Ca2+ signals elicited by another agonist, namely carbachol (Fig. 1f, g, h). We found that similarly to Bcl-2, Bcl-xL also inhibited carbachol-induced Ca2+ signals, although Bcl-xL appeared less potent than Bcl-2 (Fig. 1i, j). These results indicate that, similarly to Bcl-2, Bcl-xL inhibits IP3R-mediated Ca2+ release, irrespective of the extracellular agonist that is applied.
Full-length Bcl-xL, but not its BH4 domain, targets the LBD of IP3R1.
Next, we elucidated the interaction between Bcl-xL and IP3R. First, we compared the interaction of Bcl-xL and Bcl-2 with full-length IP3R. We overexpressed 3xFLAG-tagged Bcl-xL or Bcl-2 in HeLa cells expressing endogenous IP3Rs, immunoprecipitated Bcl-xL or Bcl-2 using anti-FLAG-coupled beads and immunoblotted for IP3Rs (Fig. 2a). This co-immunoprecipitation (coIP) analysis revealed that Bcl-xL could immunoprecipitate IP3Rs to a rather similar extend as Bcl-2, indicating that Bcl-xL and Bcl-2 display quite similar IP3R-binding properties. Bcl-2 can bind to the central, modulatory region of IP3R1, specifically to tryptic Fragment 3 [8, 12, 27]. We demonstrated that Bcl-2 binding to this site is associated with inhibition of IP3R1 activity. In addition to this, we recently discovered that Bcl-2 could also bind to the LBD of IP3R1, indicating that multiple regions are involved in IP3R1/Bcl-2 complex formation and inhibition of channel activity . We also found that Bcl-xL could target this central, modulatory region of IP3R1 though with lower efficiency than Bcl-2 . However, given the prominent inhibition of IP3Rs by Bcl-xL and the observation that this inhibition appears dependent on the agonist concentration, we asked whether Bcl-xL could also target the LBD. Thus, we used lysates from COS-7 cells that overexpressed 3xFLAG-Bcl-xL in GST-pull down experiments against purified GST-LBD and GST-Fragment 3 (representing a major part of the central, modulatory region) of IP3R1 (Fig. 2b). Our analysis revealed that similarly to Bcl-2, Bcl-xL can bind to both regions (Fig. 2c, Fig. S2a). Since GST-pulldowns are only semi-quantitative, we applied microscale thermophoresis (MST), a biophysical approach allowing to measure molecular interactions. This technique is based on detecting a change in fluorescence of a labeled target as a function of the concentration of a non-fluorescent ligand. The change in fluorescence reflects the thermophoretic movement of the fluorescent target subjected to a microscopic temperature gradient. We thus used MST to assess direct binding between purified GST-IP3R fragments and purified 6xHis-Bcl-xL to determine the binding affinity. Using MST, we demonstrated that both purified GST-LBD and GST-Fragment 3 could bind to wild-type 6xHis-Bcl-xL (Fig. 2d). The specificity of this interaction is underpinned by two negative controls, parental GST and GST-Fragment 5b that lacks the 6th TMD, previously established to be critical for Bcl-xL binding . Indeed, no binding between 6xHis-Bcl-xL and GST or GST-Fragment 5b could be detected. Furthermore, we obtained the dissociation constant for both domains with 6xHis-Bcl-xL, revealing a Kd of ~701 nM for 6xHis-Bcl-xL interaction with the GST-LBD and a Kd of ~495 nM for 6xHis-Bcl-xL interaction with GST-Fragment 3. This also indicates that wild-type 6xHis-Bcl-xL binds to both the LBD and Fragment 3.
We previously characterized the binding characteristics of the BH4 domain of Bcl-2 and Bcl-xL with Fragment 3 in detail via surface plasmon resonance (SPR) . This study showed that the BH4 domain of Bcl-2 but not the one of Bcl-xL could interact with the Fragment 3. Recently, we also identified a novel binding site for Bcl-2’s BH4 domain in the LBD, but hadn’t yet characterized its ability to interact with Bcl-xL’s BH4 domain . Thus, we examined the importance of the BH4 domain of Bcl-xL for binding to the LBD using SPR. Biotin coupled to a peptide encompassing BH4-Bcl-xL was immobilized on streptavidin chips and different concentrations of purified LBD were applied as an analyte. Background binding was determined using a peptide with a scrambled sequence and subtracted. Biotin-BH4-Bcl-2 was used as a positive control for detecting LBD binding. The association curves for 1.1 µM GST-LBD show prominent binding to BH4-Bcl-2 while its binding to BH4-Bcl-xL is much lower (Fig. 2e). Similarly to what was observed for the binding of Fragment 3 to BH4-Bcl-2 versus BH4-Bcl-xL , GST-LBD displayed a strong concentration-dependent binding to immobilized BH4-Bcl-2 , while its binding to immobilized BH4-Bcl-xL appeared much weaker (Fig. 2f).
Taken together, these data reveal that while both Bcl-2 and Bcl-xL target the same regions in IP3R, they employ different binding determinants for these interactions. In contrast to Bcl-2, which exploits its BH4 domain for binding to LBD  and Fragment 3 , Bcl-xL seems to interact with the same IP3R regions but via motifs located outside of the BH4 domain.
Residue K87 of Bcl-xL contributes to the interaction with IP3R and particularly to the binding to LBD and Fragment 3.
Our previously published results  and the data reported here (Fig. 2f) indicate that in contrast to the BH4-Bcl-2, the BH4-Bcl-xL could not be responsible for targeting LBD and Fragment 3 of IP3R1. We therefore aimed to elucidate the molecular determinants in Bcl-xL responsible for its interaction with IP3Rs. Since Bcl-xL targets the same IP3R regions as Bcl-2, we envisioned that a similar interaction surface could underlie this phenomenon. We previously showed that K17 located in the middle of the BH4-Bcl-2 was critical for binding and inhibiting IP3Rs . In the BH4-Bcl-xL, the corresponding residue is not a lysine but an aspartate, preventing its ability to bind to IP3Rs. However, in silico Bcl-2/Bcl-xL structure superposition revealed that K87, located in the BH3 domain of Bcl-xL (BH3-Bcl-xL), likely is spatially constrained in a similar manner to K17 of BH4-Bcl-2 (Fig. 3a) . Moreover, sequence analysis of Bcl-xL orthologs among main vertebrate lineages revealed that K87 is highly conserved (Fig. 3b) and thus the importance of this residue was examined further. Interestingly, this lysine is located on the opposite side of the binding pocket involved in the interaction with Bak and Bax.
First, we used confocal microscopy to assess whether altering K87 into an aspartate affected Bcl-xL’s subcellular localization. We transfected HeLa cells to express a mitochondria-or an ER- targeted RFP and compared the localization of 3xFLAG-Bcl-xL versus 3xFLAG-Bcl-xLK87D using anti-FLAG-based immunofluorescence (Fig. S3a, b, c, d). We calculated average Pearson’s coefficients above 0.75 for all conditions (Fig. S3e, f), indicating a high colocalization of Bcl-xL and Bcl-xLK87D with both the mitochondria and the ER. We also calculated average Manders’ M1 coefficient to quantify the fraction of Bcl-xL or Bcl-xLK87D overlapping with the mitochondria, being 0.8 for both Bcl-xL and Bcl-xLK87D, or the ER, being 0.6 for both Bcl-xL and Bcl-xLK87D (Fig. S3g, h). The similar coefficients calculated for the wild-type Bcl-xL and the Bcl-xLK87D indicate that the K87D mutation in Bcl-xL did not alter its subcellular localization.
Second, we tested the effect of the K87D mutation on the interaction of Bcl-xL with full-length IP3R. We therefore overexpressed 3xFLAG-tagged Bcl-xL or Bcl-xLK87D in HeLa cells expressing endogenous IP3Rs (Fig. 3c), immunoprecipitated Bcl-xL or Bcl-xLK87D using anti-FLAG-coupled beads and immunoblotted for IP3Rs. This co-immunoprecipitation analysis revealed that Bcl-xLK87D binding to the IP3R channel is severely impaired compared to wild-type Bcl-xL (Fig. 3c, d). We also immunoblotted for Bax to determine Bax binding to Bcl-xL or Bcl-xLK87D. We found that both Bcl-xL and Bcl-xLK87D could bind Bax, though Bax binding to Bcl-xLK87D appeared slightly reduced compared to its binding to Bcl-xL (Fig. 3c, e).
Third, we performed GST-pull down experiments with lysates from COS-7 cells overexpressing 3xFLAG-Bcl-xL or 3xFLAG-Bcl-xLK87D (Fig. 3f, Fig. S2b). We compared their binding to purified GST-LBD and GST-Fragment 3 of IP3R1. In comparison to wild-type Bcl-xL, the ability of Bcl-xLK87D to bind the LBD and the Fragment 3 appears significantly reduced (Fig. 3f, 3g).
Fourth, we used MST to quantitative assess the interaction of purified GST-LBD and GST-Fragment 3 with purified 6xHis-Bcl-xLK87D, similarly to the experiment performed using wild-type 6xHis-Bcl-xL (Fig. 2d). We showed that although 6xHis-Bcl-xLK87D could interact with both IP3R domains, it was with lower affinity than wild-type 6xHis-Bcl-xL (Fig. 3h). Of note, 6xHis-Bcl-xLK87D did not interact with GST (Fig. S4). Indeed, 6xHis-Bcl-xLK87D displayed higher dissociation constants than wild-type 6xHis-Bcl-xL for the interaction with the IP3R fragments. GST-LBD: Kd ~1166 nM (with 6xHis-Bcl-xLK87D) versus Kd ~701 nM (with 6xHis-Bcl-xL). GST-Fragment 3: estimated Kd ~990 nM (with 6xHis-Bcl-xLK87D) versus Kd ~495 nM (with 6xHis-Bcl-xL). Thus, the GST-pull down assays and MST analysis indicate that, compared to wild-type Bcl-xL, Bcl-xLK87D is impaired in binding to LBD and Fragment 3.
Finally, we applied an in cellulo mammalian protein-protein interaction trap (MAPPIT) assay, which is based on the functional complementation of cytokine receptor signaling . The MAPPIT data confirmed that Bcl-xL is able to interact with Fragment 3 and that the interaction was impaired by the introduction of the K87D mutation (Fig. 3i). No binding was detected with the negative control, indicating that the interaction is specific. In this assay, Bcl-xL binding to LBD could not be observed, potentially due to interference of the fusion protein to establish a functional recomplementation of the cytokine receptor.
Taken together, our data demonstrate that the K87 residue is crucial for the interaction of Bcl-xL with the IP3R, where it is involved in its binding to both the LBD and the Fragment 3.
K87 residue is critical for Bcl-xL-mediated IP3R inhibition in living cells.
Next, we examined the role of the K87 residue in Bcl-xL-mediated IP3R inhibition. We used Fura-2-loaded COS-7 (Fig. 4a-c) and HeLa (Fig. 4d-g) cells with overexpressed mCherry and either Bcl-xL or Bcl-xLK87D. We used mCherry to identify transfected cells. We first studied the effect of Bcl-xL and Bcl-xLK87D overexpression in COS cells on IP3R-mediated Ca2+ signals elicited by 500 nM ATP, a relatively high concentration provoking a Ca2+ response in about 75% of the cells. Extracellular Ca2+ was chelated with EGTA, so the reported Ca2+ signals only originate from internal stores. Under these conditions, ATP-induced Ca2+ signals appeared as a single transient (Fig. 4a). While about 75% of the cells expressing the empty vector displayed a response to ATP, only 40% of the cells expressing Bcl-xL responded (Fig. 4b). Cells expressing Bcl-xLK87D displayed similar responsiveness to ATP as empty vector-expressing cells with about 75% responding cells. Quantification of the amplitude of the ATP-induced Ca2+ transient in the responding cells yielded similar trends (Fig. 4c). Overexpression of Bcl-xL provoked a decrease in the peak [Ca2+] provoked by ATP, while overexpression of Bcl-xLK87D failed to do this.
Then, we aimed to study the effect of Bcl-xL in HeLa cells, well-known to display long-lasting Ca2+ oscillations in response to extracellular agonists [31, 32]. Here, we exposed HeLa cells to a very low [ATP] (70 nM), thereby mimicking basal, pro-survival Ca2+ oscillations and enhancing the likelihood of observing different Ca2+-signaling patterns. We could discriminate three distinct Ca2+-signaling profiles: single peak responses, long-lasting responses and baseline Ca2+ oscillations (Fig. 4d). Consistent with an inhibitory effect of Bcl-xL on IP3Rs, we found that long-lasting responses were clearly impaired upon overexpression of Bcl-xL (Fig. 4d, e). Interestingly, this effect was not observed upon overexpression of Bcl-xLK87D. Furthermore, the area under the curve (Fig. 4f) and the peak amplitude (Fig. 4g) were reduced upon overexpression of Bcl-xL, but not Bcl-xLK87D. This demonstrates that Bcl-xL’s inhibitory action on IP3Rs is critically dependent on the K87 residue.
Finally, we also determined that overexpressed Bcl-xL or its mutant did not alter the ER Ca2+ store content, by monitoring ER Ca2+ release in HeLa cells following SERCA inhibition by 1 µM thapsigargin (Fig. S5). These data are consistent with the differences observed upon ATP stimulation not being as a result of altered ER Ca2+ levels, but instead are due to the specific effect of Bcl-xL on IP3R-mediated Ca2+ release.
Purified Bcl-xL can directly suppress IP3R single-channel opening.
As all our functional data were obtained in intact cells, we also wished to provide more direct evidence for IP3R inhibition by Bcl-xL through electrophysiology. This is important because in intact cell systems Bcl-xL may have other targets besides IP3Rs that impact cytosolic Ca2+ signals. In addition, these experiments can be performed in tightly controlled conditions, including different IP3 and Bcl-xL concentrations. Therefore, we aimed to study the impact of purified Bcl-xL proteins on IP3Rs. We generated 6xHis-tagged versions of full-length Bcl-xL, Bcl-xLΔTMD and full-length Bcl-xLK87D that enabled their purification from E. coli using NiNTA columns (Fig. S6a).
Next, we tested the effect of the different recombinantly expressed and purified 6xHis-Bcl-xL variants on IP3R1 single-channel activity using the on-nucleus patch-clamp technique (Fig. 5). Channel opening in isolated nuclei obtained from DT40-3KO cells ectopically expressing IP3R1 was triggered by 1 μM of IP3 (Fig. 5a). We used purified Bcl-2ΔTMD as a benchmark (Fig. 5b), which we previously validated to inhibit IP3R1 single-channel openings . We subsequently first tested whether Bcl-xLΔTMD could inhibit the opening of IP3R1 channels induced by 1 µM IP3, but this protein failed to modulate (inhibit/sensitize) IP3R1 channels (Fig. 5c). Next, we assessed the effect of 1 µM full-length Bcl-xL (Fig. 5d), a concentration previously proposed to have a stimulatory effect on IP3R . Consistent with the data obtained in intact cells and similarly to 1 µM Bcl-2ΔTMD (Fig. 5b), application of 1 μM full-length 6xHis-Bcl-xL resulted in a significantly decreased open probability (PO) of IP3R1 channels in the presence of 1 µM IP3 (Fig. 5d). Clearly, these results contradict previous data that reported that Bcl-xL sensitizes IP3R [25, 26]. In these reports, the effect of Bcl-xL on IP3Rs was shown to exhibit a bell-shaped dependence with 1 µM of Bcl-xL optimally sensitizing IP3Rs . Hence, to ensure that we did not apply Bcl-xL at too high concentrations, we also examined the effects of 300 nM (Fig. 5e) and 100 nM (Fig. 5f) full-length Bcl-xL. These lower Bcl-xL concentrations also inhibited IP3R1 single-channel opening, though with lower potency compared to 1 µM full-length Bcl-xL. Next, we examined the effect of 1 µM full-length Bcl-xLK87D protein on IP3R1 single-channel openings activated by 1 µM IP3 (Fig. 5g). Consistent with our in vitro binding assays and our Ca2+-imaging studies in intact cells, Bcl-xLK87D failed to inhibit IP3R1 single-channel activity. Quantification of all conditions is shown in Fig. 5h. These data further demonstrate that Bcl-xL has a direct inhibitory effect on IP3R activity and that K87 is critical for this effect.
Another potential explanation could be that the conditions in which we have measured IP3R1 opening favor the detection of inhibitory effects and may miss potential sensitizing effects. We therefore measured the impact of purified Bcl-xL proteins on IP3R1 single-channel openings induced by threshold concentrations of IP3 (Fig. 5i-m). In the presence of 100 nM IP3, the PO was reduced to ~0.005 (Fig. 5i), compared to a PO of ~0.25 at 1 µM IP3 (Fig. 5a). Such conditions, which initiate threshold IP3R1 opening, should favor the detection of any potential sensitization of the channel. Nevertheless, similarly to Bcl-2ΔTMD (Fig. 5j), full-length Bcl-xL provoked a complete inhibition IP3R1 opening (Fig. 5k), while Bcl-xLK87D failed to inhibit IP3R1 opening (Fig. 5l). Quantification of all conditions is shown in Fig. 5m.
Finally, we also performed electrophysiology experiments to assess IP3R inhibition by 6xHis-Bcl-xL with high IP3 concentrations (Fig. 5n-p). Indeed, we previously demonstrated that high IP3 concentrations could compete with Bcl-2 for binding to the LBD of IP3Rs, thereby alleviating IP3R inhibition by Bcl-2 . Here, we have established that 6xHis-Bcl-xL is also able to interact with the LBD of IP3Rs, prompting us to test the effect of purified 6xHis-Bcl-xL on IP3R1 single-channel activity triggered by high concentration of IP3. With 10 µM IP3, the PO reached more than 0.65 (Fig. 5n), compared to a PO of about 0.25 at 1 µM IP3 (Fig. 5a). In those conditions, 6xHis-Bcl-xL did not alter channel activity, reflecting a loss of capacity to inhibit the channel at high [IP3] (Fig. 5o, p). These results suggest that, similarly to our observations made for Bcl-2, IP3 might compete with Bcl-xL for the LBD of IP3Rs, thereby rendering Bcl-xL less effective in inhibiting IP3Rs at high IP3 concentrations.
Given our findings were diametrically opposite to the previously reported findings, we sought to validate that the purified full-length 6xHis-Bcl-xL proteins used were properly folded and displayed bona fide anti-apoptotic functions. We first determined the CD spectrum of both Bcl-xL, which indicated that wild-type Bcl-xL, Bcl-xLK87D and Bcl-xLΔTMD had proper α-helical folding (Fig. S6b). Moreover, we performed a thermal ramping experiment. Unfolding of wild-type Bcl-xL was characterized by two apparent melting temperatures Tm1: 67 °C and Tm2: 55.47 °C, which were shifted to the left for Bcl-xLK87D (Tm1: 46.11 °C and Tm2: 53.2 °C), indicating some destabilizing effect of the mutation. These observations very much resemble the effect of K17D mutation in purified Bcl-2 . We also measured Bcl-xLΔTMD, which was characterized by one melting temperature Tm1: 76 °C, indicating that Bcl-xLΔTMD is much more stable than wild-type Bcl-xL (Fig. S6c). Next, we employed an in vitro Bax-liposome permeabilization assay, where purified Bax is incubated with liposomes encapsulating both a quencher (DPX) and a fluorophore (ANTS) (Fig. 6). Bax-pore formation can be triggered by cBid (Fig. 6a) or Bim (Fig. 6b) proteins, two “activator” BH3-only proteins. Full-length 6xHis-Bcl-xL potently inhibited cBid and Bim-triggered Bax-pore formation (IC50 of about 20 nM). Of note, 6xHis-Bcl-xLK87D too inhibited Bax-pore formation, but was less efficacious (IC50 of about 80 nM) than 6xHis-Bcl-xL (Fig. 6c, d). This might relate to the reduced Bax binding observed in the coIPs using cell lysates (Fig. 3c). Consistent with previous reports , Bcl-xLΔTMD failed to inhibit Bax-pore formation.
Overall, our electrophysiological studies provide strong evidence that recombinant Bcl-xL with validated anti-apoptotic properties directly inhibits IP3R1 single-channel opening with a critical role for K87 in Bcl-xL. Furthermore, the data suggest that Bcl-xL’s TMD is not only important for inhibiting Bax [33, 34] but also for inhibiting IP3R opening. Yet, the significance and the role of the TMD of Bcl-xL in a cellular context for IP3R modulation remains to be elucidated.
Bcl-xLK87D is impaired in protecting cells against staurosporine-induced apoptosis.
Finally, we wished to validate the importance of the IP3R/Bcl-xL interaction for the protective effects of Bcl-xL against Ca2+-dependent pro-apoptotic stimuli (Fig. 7, Fig. S7). We therefore used STS, which has been previously validated to provoke Ca2+-driven apoptosis [35-37]. Here, we assessed whether STS provoked apoptosis through IP3R-mediated Ca2+ elevations. First, we measured long-term Ca2+ dynamics in HeLa cells exposed to 0.5 µM STS for 1 hour (Fig. 7a). Using live single cell Ca2+ imaging in Fura-2-AM-loaded cells, we observed that STS triggered long-lasting Ca2+ elevations in wild-type HeLa cells. Contrary to IP3R activation with physiological agonists (Fig.1, Fig. 4), this Ca2+ release is rather slow on onset and prolonged over a long period of time. We then used a HeLa cell model in which all three IP3R isoforms have been knocked out (HeLa-3KO). In these cells, the STS-induced Ca2+-release events were virtually absent (Fig. 7a, b). Having validated that STS treatment in HeLa cells provoked long-lasting IP3R-mediated Ca2+ elevations, we determined whether IP3Rs contributed to STS-induced cell death in HeLa cells (Fig. 7c). We therefore monitored apoptotic cell death in HeLa cells exposed to 0.5 µM STS for six hours by determining the ratio of cleaved poly(ADP-ribose) polymerase (PARP) in relation to total PARP . Strikingly, in wild-type HeLa, about 90% of the total PARP was converted to the cleaved form, while only 30% of the total PARP appeared in the cleaved form in HeLa-3KO cells, indicating that IP3Rs are crucial for STS-induced cell death in HeLa cells (Fig. 7c, d). In a more general way, this is the first time that, to our knowledge, that IP3Rs were directly implicated in STS-evoked pro-apoptotic Ca2+ flux and directly linked to cell death.
Next, using live, single-cell Ca2+ imaging, we studied the impact of overexpressing Bcl-xL-P2A-mCherry and Bcl-xLK87D-P2A-mCherry on STS-induced Ca2+ elevations in Fura-2-AM-loaded wild-type HeLa cells (Fig. 7e, Fig. S7a). Ca2+ signals were measured in mCherry-positive cells. Strikingly, Bcl-xL overexpression strongly suppressed prolonged Ca2+ elevations induced by 0.5 µM STS compared to empty vector-expressing cells, while Bcl-xLK87D overexpression was much less effective (Fig. 7e, f, Fig. S7a, b). We then examined whether IP3R modulation by Bcl-xL contributed to the anti-apoptotic action of Bcl-xL (Fig. 7g). We first confirmed that the transfection of the cells with the 3xFLAG plasmids did not provoke cell death by itself (Fig. 7g, “- STS” conditions). Bcl-xL overexpression strongly suppressed PARP cleavage in wild-type HeLa cells exposed to 0.5 µM STS for 6 hours compared to empty vector-expressing cells (Fig. 7g, “+ STS” conditions). In contrast, Bcl-xLK87D overexpression was much less effective than wild-type Bcl-xL in suppressing PARP cleavage in wild-type HeLa cells (Fig. 7g, h). This suggests that Bcl-xL protects against STS through inhibition of IP3Rs, since Bcl-xLK87D is much less efficient in doing so.
We then focused on IP3R-independent cell death mechanisms. We have shown that Bcl-xLK87D binding to Bax appeared to be somewhat impaired compared to wild-type (Fig. 3c). Furthermore, the ability of Bcl-xLK87D to neutralize Bax pore formation also appeared attenuated (Fig. 6). Since STS partially acts independently of IP3Rs and since PARP cleavage also occurs in HeLa-3KO cells, though to a lesser extent (Fig. 7d), we wanted to discriminate Bcl-xL anti-apoptotic effect between IP3R inhibition versus IP3R-independent processes, such as Bax inhibition. Hence, we examined the effect of Bcl-xL and Bcl-xLK87D overexpression on STS-induced cell death in HeLa-3KO cells (Fig. 7i, j). Consistent with the ability of Bcl-xL to bind and neutralize Bax, we found that Bcl-xL could suppress STS-induced PARP cleavage in HeLa-3KO cells. Of particular interest and in contrast to the results obtained in wild-type HeLa cells, Bcl-xLK87D was equally effective as wild-type Bcl-xL in dampening STS-induced PARP cleavage in HeLa cells lacking IP3Rs. This implies that although Bax binding/inhibition is somewhat affected by the K87D mutation in Bcl-xL, there is sufficient residual Bax-binding and -inhibition capacity of Bcl-xLK87D to prevent cell death in cellulo.
Finally, we wished to assess that the K87D mutation does not affect the protection afforded by Bcl-xL towards IP3R-independent cell death mechanisms. Hence, we chose the BH3 mimetic venetoclax/ABT-199, a selective Bcl-2 inhibitor  previously established to neither interfere with the ability of Bcl-2 to inhibit IP3Rs nor to alter Ca2+ signaling [39, 40]. Venetoclax (25 µM, 24 hours) triggered ~80% PARP cleavage in HeLa cells (Fig. 7k, l). The level of PARP cleavage was similar between wild-type HeLa and HeLa-3KO, thereby validating that venetoclax indeed acted in an IP3R-independent manner. Bcl-xL and Bcl-xLK87D were equally effective in counteracting venetoclax-induced PARP cleavage by about 40-50% (Fig. 7k, l). Moreover, the anti-apoptotic effect of Bcl-xL and Bcl-xLK87D was also comparable between wild-type HeLa and HeLa-3KO. These data strongly indicate that K87D mutation impairs Bcl-xL’s protective effect against IP3R-dependent cell death but does not significantly affects its canonical anti-apoptotic function, thereby antagonizing Bax/Bak.
Bcl-xL protects breast cancer cells from IP3R-mediated cell death.
Next, by knocking down Bcl-xL in a Bcl-xL-dependent cell model, we examined whether also endogenous Bcl-xL could inhibit IP3Rs. We used a breast cancer model, the mammary gland adenocarcinoma cell line MDA-MB-231, in which Bcl-xL is important for survival  and migration . We transfected MDA-MB-231 cells with a siRNA targeting Bcl-xL, thereby lowering its protein levels by about 50% (Fig. 8a, b). Interestingly, Bcl-xL knock-down in MDA-MB-231 cells did not induce apoptosis by itself (Fig. 8a, c). This was important to exclude that any potential changes in Ca2+ signaling in cells with decreased Bcl-xL levels were a consequence of ongoing cell death rather than due to a decrease in Bcl-xL-protein levels. Using thapsigargin, we next validated in MDA-MB-231 cells that the ER Ca2+-store content is not altered following Bcl-xL depletion (Fig. 8d, e, f). Therefore, changes in agonist-induced Ca2+ signaling in Bcl-xL-depleted cells are not an indirect consequence of changes in ER Ca2+ loading. We then measured IP3R-mediated Ca2+ release elicited by ATP (0.5 µM) in individual MDA-MB-231 cells pre-treated with extracellular Ca2+ chelator EGTA, thereby ensuring that Ca2+ signals only arise from internal stores. Compared to the cells transfected with a non-target siRNA, the cells treated with a siRNA targeting Bcl-xL displayed a strikingly higher ATP-induced Ca2+ response (Fig. 8g, h). We calculated a significant increase in the number of responding cells (Fig. 8i), of the area under the curve (Fig. 8j) and of the maximal peak amplitude (Fig. 8k) in MDA-MB-231 cells in which Bcl-xL is depleted. To be certain that this effect was not due to a potential downregulation of the Bcl-xL-related Bcl-2 protein, which is a prominent inhibitor of IP3Rs, we analyzed the Bcl-2-protein levels via western blotting (Fig. S8). However, Bcl-2-protein levels were not decreased. Instead, Bcl-2-protein levels appeared increased, potentially as a compensatory mechanism that could help sustain the survival of the cells in which Bcl-xL was downregulated. Nevertheless, the overall Bcl-2-protein levels remained extremely low in these MDA-MB-231 cells, when benchmarked against the Bcl-2-protein levels present in OCI-LY-1 cells, a Bcl-2 dependent diffuse large B-cell lymphoma cell line. In any case, these data indicate that endogenous Bcl-xL suppresses IP3R activity in breast cancer cells, independently of Bcl-2 levels. To determine whether Bcl-xL could also counteract IP3R-mediated apoptotic Ca2+ release in those cells, we exposed the MDA-MB-231 cells to STS (0.5 µM) (Fig. 8l). In MDA-MB-231 cells transfected with a non-target siRNA, STS only provoked limited PARP cleavage, indicating that these cells are rather resistant to STS. However, cells treated with the siRNA against Bcl-xL displayed a prominent increase in STS-induced PARP cleavage resulting in about 50% PARP cleavage. This indicates that lowering endogenous Bcl-xL-protein levels rendered MDA-MB-231 cells very sensitive to STS-induced cell death (Fig. 8l, m). Altogether, these results reveal that endogenous Bcl-xL suppresses IP3R-mediated Ca2+ release and confers cell death protection against Ca2+-dependent cell death stimuli.