Expression of anti-apoptotic Bcl-2 family members in GSC-ECLs
To evaluate the endogenous expression levels of anti-apoptotic factors, we performed RT-qPCR and Western blot analysis of Bcl-2, Bcl-w, Bcl-xL, and Mcl-1 in GSC-ECLs (named G01, G02, G03, G07, G08, and G09) and in untransformed neural progenitors (NP). We found that Bcl-2 expression is more abundant in NPs, while Bcl-w showed higher expression levels in GSC-ECLs. Although the abundance of Bcl-xL and Mcl-1 is similar between NP and GSC-ECLs, the G03 cell line exhibited higher expression levels of Bcl-xL, and G01 cell line showed increased Mcl-1 expression at both mRNA and protein levels (Fig. 1a,b and Supplementary Fig. S1). These differences in the expression levels of anti-apoptotic Bcl-2 family members between GSC-ECLs further support the well-known inter-tumor heterogeneity described in GBM. Finally, we also examined the transcriptional profile of these anti-apoptotic factors in each GSC-ECL. As shown in Fig. 1c, despite the variability in GBM, we found that Bcl-xL and Mcl-1 represent the most abundantly expressed anti-apoptotic factors in all tested GSC-ECLs while Bcl-w, followed by Bcl-2, exhibit the lowest expression levels.
ABT-263 and WEHI-539 enhance GSC-ECL sensitivity to chemotherapeutic agents in a cell line-specific manner
As one of our main objectives was to evaluate the effectiveness of BH3-mimetics in sensitizing GSC-ECLs to chemotherapeutic agents, we defined concentrations of TMZ, CCNU and VCR at which they exert their primary effect but fail to induce excessive cell death. We established that the experimental conditions to achieve these requirements were 250 mM TMZ, 20 mM CCNU and 0.5 mM VCR for 48 hours. At these conditions, we observed only a slight increase in cell death. Even so, TMZ and CCNU induced H2AX phosphorylation (gH2AX), indicative of DNA damage. VCR, on the other hand, led to the appearance of abnormally shaped cells, suggestive of the depolymerization of the microtubular network. Additionally, these chemotherapeutic drugs also caused the expected cell cycle arrest (Supplementary Fig. S2). Therefore, despite the lack of promoting extensive cell death in these conditions, these drugs were effective in exerting their mechanism of action.
As a first approach, we selected two BH3 mimetics, ABT-263 and WEHI-539, to target either a subset (Bcl-2, Bcl-W, and Bcl-xL) or individual (Bcl-xL) pro-survival proteins and evaluated their effect on GSC-ESCL cell viability. In response to single treatments with these BH3 mimetics, G01, G03, G08, and G09 cell lines underwent a significant increase in cell death (Fig. 2a, black bars). To determine if these BH3 mimetics sensitize GSC-ECLs to chemotherapeutic agents, we also combined these BH3-mimetics with TMZ, CCNU, or VCR. In all cases, except for G07 cells, we found that the presence of ABT-263 or WEHI-539 significantly potentiated the extent of cell death compared to that triggered by each chemotherapeutic alone, although each cell line showed a differential sensitivity. In this sense, G03 cell line showed a considerably higher susceptibility to these BH3-mimetics than the other cell lines. On the contrary, G07 cell line showed resistance to these BH3-mimetics (Fig. 2a). Of note, except in the G01 cell line, ABT-263 treatment had a more pronounced effect on cell death than WEHI-539 in all tested conditions. However, this difference was strikingly higher in the G03 cell line (Fig. 2a).
Finally, to evaluate if the observed cell death is due to the activation of apoptosis, we assessed the activity of caspase-3 in two GSC-ECLs. As expected, we observed an increase in caspase-3 activity in treated cells, which could account for the observed changes in cell death (Fig. 2b).
Bcl-xL is a major regulator of GSC-ECL viability
To dissect the contributions of the anti-apoptotic factors targeted by ABT-263 (Bcl-2, Bcl-w, and Bcl-xL) on GSC-ECL viability, we conducted siRNA-mediated knockdowns for Bcl-2, Bcl-w, or Bcl-xL in GSC-ECLs (Supplementary Fig. S3). Only the silencing of Bcl-xL increased cell death (Fig. 3a and Supplementary Fig. S4). This result is consistent with the fact that Bcl-xL is one of the most abundant anti-apoptotic Bcl-2 family members and that its inhibition by WEHI-539 suffices to increase the sensitivity of the tested GSC-ECLs to antineoplastic drugs (Fig. 2a).
As previously shown, ABT-263 frequently led to higher cell death than WEHI-539, particularly in the G03 cell line (Fig. 2a). Thus, we hypothesized Bcl-2 and/or Bcl-w support cell viability when Bcl-xL is inhibited. To address this issue, we transfected cells with Bcl-2 and/or Bcl-w specific siRNAs in the presence of WEHI-539. Although we observed a trend towards increased cell death when these factors (principally Bcl-w) were silenced in the presence of WEHI-539 in all cell lines, the effect was much stronger in the G03 cells (Fig. 3b and Supplementary Fig. S5). Therefore, the higher efficacy of ABT-263 compared to WEHI-539 in G03 (Fig. 2a) would indicate a dependency of this cell line on Bcl-w and, less so, on Bcl-2, which may serve as a backup mechanism to survive when Bcl-xL activity is compromised. In accordance, in the G01 cell line, where ABT-263 and WEHI-539 caused a similar effect, we observed a low contribution of Bcl-2 and/or Bcl-w to cell survival (Supplementary Fig. S5).
Noxa expression correlates with GSC-ECL sensitivity to ABT-263 and WEHI-539
As previously described, each GSC-ECL displays a differential sensitivity to ABT-263 and WEHI-539 (Fig. 2a). To estimate the degree of BH3-mimetic sensitivity in each cell line, we calculated the difference (Δ) in the percentage of cell death by subtracting the cell death observed in untreated cells from the one observed in BH3-mimetic-treated cells (e.g. ABT-263-induced cell death – basal cell death). In both cases, G01 and G03 were the most sensitive cell lines to BH3-mimetic treatments (Fig. 4a).
Even though we have found that Bcl-xL inhibition has an important impact on cell death induction, we did not find any clear correlation between its abundance and GSC-ECL sensitivity to ABT-263 or WEHI-539. Thus, considering that BH3-mimetics emulate BH3-only proteins, we analyzed possible relationships between the expression of these factors and the sensitivity of GSC-ECLs to ABT-263 or WEHI-539. Notably, among the BH3-only factors (Supplementary Fig. S6) we found Noxa expression levels (Fig. 4b,c) significantly correlate with the degree of sensitivity to BH3-mimetics (Fig. 4d), being G01 and G03 the cell lines that exhibit the highest expression levels of this BH3-only (Fig. 4b,c).
Impairment of Mcl-1 activity strongly potentiates WEHI-539-induced cell death in GSC-ECLs
At this point, we wondered whether the well-known Noxa binding partner, Mcl-1, was also responsible for the differential sensitivity of GSC-ECLs to ABT-263 and WEHI-539. To address this issue, we silenced this anti-apoptotic factor using a siRNA-specific sequence (Supplementary Fig. S3). In accordance with the high expression levels of Mcl-1, we found that this anti-apoptotic factor also contributes to GSC-ECL survival (Fig. 5). Even though Mcl-1 silencing alone slightly lowered the cell viability only in two GSC-ECLs, the combination with WEHI-539 was synergistic, inducing a notable increase of cell death in all tested cell lines (Fig. 5). To further evaluate the involvement of Mcl-1 in regulating GSC-ECL viability, we exposed cells to the Mcl-1 inhibitor S63845 in the presence or absence of WEHI-539, combined or not with chemotherapeutic agents (Fig. 6). S63845 treatment as a single agent causes a slight increment in cell death. However, as occurred with Mcl-1 silencing, the combination of S63845 with WEHI-539 provokes a marked increase in cell death. Therefore, the co-inhibition of the most abundantly expressed anti-apoptotic factors, Mcl-1 and Bcl-xL (Fig. 1c), causes substantial cell death in GSC-ECLs.
Finally, to confirm that Noxa interacts with Mcl-1, we performed a co-immunoprecipitation assay. As expected, Noxa binds Mcl-1 in GSC-ECLs (Supplementary Fig. S7).
Noxa expression determines sensitivity to ABT-263 and WEHI-539
To confirm that Noxa contributes to the induction of cell death triggered by these BH3-mimetics (either in the presence or absence of chemotherapeutic agents), we conducted gene silencing experiments using a Noxa specific-siRNA (Supplementary Fig. S3). For this purpose, we selected four cell lines that exhibited different basal expression levels of Noxa (G01 and G03 high; G02 and G09 low).
In the G03 cell line, down-regulation of Noxa significantly rescued cells from cell death induced by either ABT-263 or WEHI-539 (Fig. 7a). However, when cells were treated with chemotherapeutic agents in the absence of BH3 mimetics, silencing of Noxa did not affect cell death (Supplementary Fig. S8). Strikingly, in the G01 cell line, which also expresses relatively high basal levels of Noxa, its silencing promotes a much lesser decrease in cell death than in the G03 cell line (Fig. 7a). This could be explained by the fact that the G01 cell line expresses higher levels of Mcl-1 than the G03 cell line (Fig. 1). On the contrary, G02 and G09 cell lines showed no differences in cell death after silencing Noxa. (Fig. 7). This result is in agreement with the fact that these cell lines express barely detectable levels of Noxa (Fig. 4b,c). As a whole, these findings demonstrate that Noxa expression levels not only correlate with sensitivity to Bcl-xL inhibition (Fig. 4d), but also contribute to the induction of cell death promoted by these BH3 mimetics.
To investigate whether BH3-mimetic treatments affect Noxa expression in the Noxa-dependent GSC-ECLs (G01 and G03), we measured mRNA and protein levels in untreated and treated cells. In the G03 cell line, ABT-263, but not WEHI-539, significantly induces Noxa expression (Fig. 7b,c). Notably, this induction coincides with the increase of cell death mediated by ABT-263 in this cell line (Fig. 2). Contrarily, in the G01 cell line, neither WEHI-539 nor ABT-263 induces Noxa expression (Fig. 7b,c).
Finally, as previously shown, the participation of Bcl-2 and Bcl-w in the G03 cell line is more relevant than in other cell lines (Fig. 3b and Supplementary Fig. S5), explaining the higher sensitivity of this cell line to ABT-263 than to WEHI-539 (Fig. 2a). Considering that Bcl-xL and Mcl-1 are the most relevant anti-apoptotic Bcl-2 family members in GSC-ECLs and that high levels of Noxa can inhibit Mcl-1, we hypothesized that in WEHI-539-treated G03 cells, Bcl-2 and Bcl-w become relevant due to the simultaneous inhibition of both Bcl-xL (due to WEHI-539) and Mcl-1 (due to the high levels of endogenous Noxa). To test this hypothesis, we evaluated the effect of silencing Bcl-2 and Bcl-w in WEHI-539-treated cells (as in Fig. 3b), but this time also silencing Noxa, therefore keeping Mcl-1 uninhibited. As shown in Fig. 7d, in WEHI-539-treated G03 cells, in a context of uninhibited Mcl-1, Bcl-2 and Bcl-w are no longer as relevant. As expected, in the G09 cell line, where the levels of Noxa are considerably lower, its silencing did not affect the contribution of Bcl-2 and Bcl-w to cell survival (Fig. 7d).
The sensitizer role of Noxa depends on Mcl-1 activity in GSC-ECLs
Having determined that Noxa influences GSC-ECL sensitivity to Bcl-xL inhibition, we explored the underlying mechanism of this response. To evaluate whether Noxa acts as an activator or a sensitizer BH-3 only, we down-regulated its expression in a context in which Mcl-1 activity was inhibited. If Noxa behaves as an activator, the abrogation of Mcl-1 would increase cell death, partly by facilitating Noxa-mediated Bax/Bak oligomerization. Hence, silencing Noxa in this context would decrease cell death. On the contrary, if Noxa functions as a sensitizer, the lack of Mcl-1 activity would render Noxa irrelevant, as the only function of Noxa would be the neutralization of Mcl-1. As a consequence, Noxa silencing would not affect cell death. To discern between these two alternatives, we pharmacologically inhibited Mcl-1 (S63845) and Bcl-xL (WEHI-539) and evaluated the effect of silencing Noxa. Importantly, S63845 binds to Mcl-1 in the same binding groove that Noxa does30, promoting its displacement. As shown in Fig. 8a, the down-regulation of Noxa in the presence of S63845 did not decrease cell death. Therefore, these results indicate that Noxa acts as a sensitizer through its role as an Mcl-1 inhibitor in GSC-ECLs.
Having observed that the complete inhibition of Mcl-1 by S63845 renders Noxa irrelevant, we wondered if the partial decrease in the expression levels of Mcl-1 could also affect the relevance of Noxa. To find the answer, instead of inhibiting Mcl-1 with S63845, we silenced Mcl-1 and repeated the same experiments. Surprisingly, in this context, Noxa down-regulation caused a significant decrease in cell death in all tested cell lines (Fig. 8b), even in G02 and G09, in which Noxa silencing did not affect cell survival when Mcl-1 expression was unmodified (Fig. 7a). Importantly, in all tested cell lines, Noxa silencing led to a more pronounced decrease in cell death than that observed in cells with unmodified Mcl-1 expression (Fig. 8b and Supplementary Table S1). Therefore, Noxa becomes more relevant in the context of a diminished amount of Mcl-1. Of note, in all the studied cell lines, the basal expression levels of Mcl-1 transcripts far exceed those of Noxa (Supplementary Fig. S9).Thus, by lowering Mcl-1 expression, Noxa would now be able to bind and inactivate a higher proportion of this anti-apoptotic factor. Therefore, the relevance of Noxa as a sensitizer depends not only on its expression levels but also on those of Mcl-1, since when the expression of Mcl-1 decreases, the influence of Noxa on cell death increases.