Apoptotic priming in senescence predicts specific senolysis by quantitative analysis of mitochondrial dependencies

Abstract

Cellular senescence contributes to a variety of pathologies associated with aging and is implicated as a cellular state in which cancer cells can survive treatment. Reported senolytic drug treatments act through varying molecular mechanisms, but heterogeneous efficacy across the diverse contexts of cellular senescence indicates a need for predictive biomarkers of senolytic activity. Using multi-parametric analyses of commonly reported molecular features of the senescent phenotype, we assayed a variety of models, including malignant and nonmalignant cells, using several triggers of senescence induction and found little univariate predictive power of these traditional senescence markers to identify senolytic drug sensitivity. We sought to identify novel drug targets in senescent cells that were insensitive to frequently implemented senolytic therapies, such as Navitoclax (ABT-263), using quantitative mass spectrometry to measure changes in the senescent proteome, compared to cells which acquire an acute sensitivity to ABT-263 with senescence induction. Inhibition of the antioxidant GPX4 or the Bcl-2 family member MCL-1 using small molecule compounds in combination with ABT-263 significantly increased the induction of apoptosis in some, but not all, previously insensitive senescent cells. We then asked if we could use BH3 profiling to measure differences in mitochondrial apoptotic priming in these models of cellular senescence and predict sensitivity to the senolytics ABT-263 or the combination of dasatinib and quercetin (D + Q). We found, despite being significantly less primed for apoptosis overall, the dependence of senescent mitochondria on BCL-XL was significantly correlated to senescent cell killing by both ABT-263 and D + Q, despite no significant changes in the gene or protein expression of BCL-XL. However, our data caution against broad classification of drugs as globally senolytic and instead provide impetus for context-specific senolytic targets and propose BH3 profiling as an effective predictive biomarker.