Research ArticleCancer

Activation of concurrent apoptosis and necroptosis by SMAC mimetics for the treatment of refractory and relapsed ALL

See allHide authors and affiliations

Science Translational Medicine  18 May 2016:
Vol. 8, Issue 339, pp. 339ra70
DOI: 10.1126/scitranslmed.aad2986

Giving leukemia a SMAC

Second mitochondria-derived activator of caspases, or SMAC, is a protein involved in apoptosis, a mechanism of cell death that is commonly targeted by cancer therapies. SMAC mimetics are drugs designed to mimic the action of SMAC. Now, a pair of related articles provides insights into the effects of SMAC mimetics in leukemia. For acute lymphocytic leukemia, McComb et al. show that a SMAC mimetic called birinapant works best when it can activate two different types of cell death: apoptosis and necroptosis. For acute myelocytic leukemia, Brumatti et al. show that birinapant is particularly effective when combined with a caspase inhibitor, which shuts off the apoptotic pathway and promotes cell death by necroptosis. These findings should be helpful for identifying patients most likely to benefit from treatment with SMAC mimetics and selecting effective treatment combinations for these patients.


More precise treatment strategies are urgently needed to decrease toxicity and improve outcomes for treatment-refractory leukemia. We used ex vivo drug response profiling of high-risk, relapsed, or refractory acute lymphoblastic leukemia (ALL) cases and identified a subset with exquisite sensitivity to small-molecule mimetics of the second mitochondria-derived activator of caspases (SMAC) protein. Potent ex vivo activity of the SMAC mimetic (SM) birinapant correlated with marked in vivo antileukemic effects, as indicated by delayed engraftment, decreased leukemia burden, and prolonged survival of xenografted mice. Antileukemic activity was dependent on simultaneous execution of apoptosis and necroptosis, as demonstrated by functional genomic dissection with a multicolored lentiCRISPR approach to simultaneously disrupt multiple genes in patient-derived ALL. SM specifically targeted receptor-interacting protein kinase 1 (RIP1)–dependent death, and CRISPR-mediated disruption of RIP1 completely blocked SM-induced death yet had no impact on the response to standard antileukemic agents. Thus, SM compounds such as birinapant circumvent escape from apoptosis in leukemia by activating a potent dual RIP1-dependent apoptotic and necroptotic cell death, which is not exploited by current therapy. Ex vivo drug activity profiling could provide important functional diagnostic information to identify patients who may benefit from targeted treatment with birinapant in early clinical trials.


Despite aggressive application of targeted chemotherapy, acute lymphoblastic leukemia (ALL) patients with persistent minimal residual disease (MRD), drug refractory disease, and/or relapsed disease still have a poor prognosis (1, 2). With increasing knowledge of the genomic landscape of ALL, disease-specific oncogenic lesions amenable to personalized treatment approaches are increasingly identifiable (3). Nonetheless, the efficacy of such targeted agents can be undermined by general deregulation of cell death pathways. Failure to undergo chemotherapy-induced apoptosis constitutes a key mechanism for drug resistance and clonal escape (4). Thus, alternative strategies to reactivate cell death pathways are being actively pursued, but so far, none of these have reached wide clinical application (5, 6).

The cellular inhibitor of apoptosis proteins (IAPs) (cIAP1, cIAP2, and XIAP) are often overexpressed in cancer and contribute to drug resistance (7, 8). IAPs inhibit programmed cell death through a number of mechanisms, including direct inhibition of caspases (9) and ubiquitination of receptor-interacting protein kinase 1 (RIP1) (10). Whereas RIP1 can act as a potent activator of multiple forms of programmed cell death (11), its ubiquitination by cIAPs mediates proinflammatory signaling through nuclear factor κB (NF-κB) (10, 12), and RIP1 can flip roles to become an inhibitor of apoptotic and necroptotic cell death (1315). Thus, RIP1 acts as a key convergence point between prodeath, prosurvival, and proinflammatory signals, as detailed by a number of recent reviews (1618). To short-circuit the ability of cIAPs to rewire prodeath into prosurvival signals, small-molecule peptidomimetics of the second mitochondria-derived activator of caspases (SMAC) have been developed. These SMAC mimetics (SM) promote cell death through inhibition of IAPs (19, 20). SM compounds have been demonstrated to induce degradation and direct inhibition of IAPs and drive cell death as single agents in a number of cancer cell lines (21, 22). The potential for application of promising SM compounds in leukemia has yet to be fully examined.

Here, we show that a subset of ALL samples exhibits high sensitivity to SM-induced cell death in vitro and in vivo, with median inhibitory concentration (IC50) values below 100 nM. We use clustered regularly interspaced short palindromic repeats (CRISPR)–based genetic disruptions and pharmacologic interference to show that SM induces simultaneous activation of RIP1-mediated apoptosis and necroptosis in a manner not similarly exploited by current chemotherapeutic drugs. Our data provide strong evidence for high antileukemic activity of SM in relapsed ALL and suggest functional drug response profiling as a tool to identify those patients who are most likely to benefit from SM treatment. Such ex vivo drug profiling could be applied to improve personalized treatment decisions for individual patients as we recently showed for the BCL-2 inhibitor venetoclax in transcription factor 3–hepatic leukemia factor (TCF3-HLF)–positive ALL (23).


Drug profiling identifies ALL with high sensitivity to SM

Given our previous identification of RIP1-dependent necroptosis as a mechanism to overcome drug resistance in ALL (24), we were interested in investigating additional drugs that could activate RIP1-dependent cell death. On the basis of promising SM activity in a large-scale ALL drug screening platform previously described (23, 25), we performed focused testing of SM response on a set of 51 patient-derived B cell precursor ALL xenografts enriched for samples from relapsed and drug-resistant disease (table S1, A and B). ALL xenografts retain important genotypic, clonotypic, and phenotypic characteristics of primary patient samples and thus constitute a logical model to investigate refractory disease (26, 27). Birinapant is a highly active SM compound that contains two cIAP binding moieties and drives rapid degradation of cIAPs (28). Leukemic response to birinapant varied greatly, with IC50 values in the low nanomolar range (<250 nM) in 17 of 51 cases (Fig. 1A). No toxicity was observed toward nontumorigenic mesenchymal stromal support cells used in this coculture assay, suggesting the existence of a therapeutic window (fig. S1, A and B). Samples from high-risk or relapsed patients (labeled as very high risk, high risk, or relapse) were among highly birinapant-sensitive cases. Similar response profiles were found for the monovalent SM compound, LCL161 (6), although with reduced potency (fig. S1C). In addition, cases with moderate sensitivity to SM were also detected in T cell ALL (fig. S1D).

Fig. 1. Drug profiling identifies a cohort of ALL highly sensitive to SM.

(A) ALL xenografts derived from standard risk (SR), high risk (HR), very high risk (VHR), and relapse (R) patients were treated with varying doses of birinapant for 48 hours. IC50 values were calculated on the basis of viability staining and automated microscopy. (B to D) NSG mice were injected intravenously with 106 SM-sensitive leukemic cells and treated starting on day 2 after injection with birinapant (30 mg/kg) daily for two 2-week blocks separated by a 2-week recovery. Graphs show the proportion of human CD45+ (hCD45+) cells within the nucleated blood cell fraction of vehicle-treated (red lines) and birinapant-treated (blue lines) animals as measured by flow cytometry. (E to G) Leukemic mice were treated as above with a single 2-week block of birinapant treatment starting at engraftment of >30% hCD45+ cells as determined by flow cytometry (treated group in green). (H to J) Survival plots for vehicle controls or mice treated before or after engraftment for three different SM-sensitive patient-derived xenografts. n ≥ 3 mice per group. (K to P) Experiments were performed similarly to the ones above but using patient samples with low in vitro SM sensitivity. n ≥ 2 mice per group.

We next evaluated the antileukemic activity of SM in vivo using xenografted mice. Birinapant treatment initiated shortly after ALL transplantation delayed disease progression in three cases identified as highly SM-sensitive (Fig. 1, B to D). Complete responses, defined by the absence of leukemic cells within the blood detectable by flow cytometry, could also be achieved in these three cases when birinapant treatment was initiated after more than 30% leukemic engraftment (Fig. 1, E to G), which corresponds to extensive extramedullary and complete bone marrow involvement. Treatment rapidly decreased the leukemic load in peripheral blood (Fig. 1, E to G) and delayed disease progression to an experimental endpoint of >50% (Fig. 1, H to J) or >30% and >20% (Fig. 1, O and P), of human leukemia cells in the nucleated peripheral blood fraction. Endpoints were defined according to sample-specific engraftment kinetics. The in vivo efficacy of birinapant in delaying progression was less pronounced in two samples with lower SM sensitivity in vitro (Fig. 1, K to P). Notably, we did not detect in vivo activity of LCL161 in two highly sensitive patient samples, perhaps because of the lower potency of LCL161 (fig. S1, E to J). Thus, our in vitro drug profiling identified a subset of primary ALL samples with high ex vivo SM sensitivity (IC50 < 250 nM for 17 of 51 primary ALL samples tested), which predicted antileukemic efficacy in vivo in the xenograft model.

SM treatment induces varying activation of both apoptosis and necroptosis

To analyze the mechanism by which SM compounds induce cell death, we examined SM dose-response curves of SM-sensitive patient samples in the presence of pharmacological inhibitors of apoptosis (pan-caspase inhibitor zVAD) and/or necroptosis (RIP1 kinase inhibitor necrostatin-1). Initial experiments in two highly birinapant-sensitive leukemic samples showed that rescue from SM-induced cell death varied greatly between patient samples. Specifically, R-03 cells required simultaneous application of zVAD and necrostatin-1 for rescue from birinapant (Fig. 2A), whereas VHR-10 cells showed no significant rescue with the addition of both death inhibitors (Fig. 2B). To further investigate the heterogeneity of birinapant response, we screened a larger cohort of patient samples using these death inhibitors. The contribution of apoptosis and/or necroptosis varied greatly among patient samples (Fig. 2, A to J), revealing the following death phenotypes: necroptotic samples were rescued by necrostatin-1 alone (Fig. 2, A, C, and D), mixed death phenotype samples were rescued only by a combination of zVAD and necrostatin-1 (Fig. 2E), and some samples were not rescued by any combination of inhibitors (Fig. 2, B and F). None of the samples tested showed a purely apoptotic phenotype, and inhibition of RIP1 kinase with necrostatin-1 alone actually decreased cell viability in a small number of cases, which was not subsequently rescued by zVAD (Fig. 2F). Similar cell death patterns were detected for birinapant and LCL161 (Fig. 2, G to J). Together, these data indicate that SM treatment induces variable activation of apoptotic and necroptotic cell death in SM-sensitive patient samples.

Fig. 2. SM induce both apoptosis and necroptosis.

(A and B) Patient-derived ALL cells in coculture with mesenchymal stem cells were treated with varying doses of birinapant alone (black) or in combination with 25 μM zVAD (blue), 25 μM necrostatin-1 (Nec) (red), or both zVAD and necrostatin (green). Curves were calculated from the mean ± SEM of three experiments and show the relative number of live cells compared to dimethyl sulfoxide controls. (C to F) Experiments were performed as above with a larger number of patient samples. Curves show the number of live cells at the indicated concentrations in a single experiment performed in duplicate for each patient sample. (G) The viability of patient samples at the birinapant IC50 in the presence or absence of death inhibitors is shown. Patients are grouped on the basis of the pattern of rescue achieved with the death inhibitors. (H) Pie chart shows the proportion of patient cells grouped as necroptotic (red), mixed cell death (green), no rescue (gray), or necrostatin-1–sensitive (orange). (I to J) Patient samples were screened as above, using LCL161 instead of birinapant (n = 1 experiment in duplicate × 15 patient samples).

Differential expression of TNFα, RIP1, RIP3, or IAPs does not predict SM response

SM treatment–induced expression of tumor necrosis factor–α (TNFα) is an important driver of SM-induced cell death (29). To investigate the role of TNFα expression in the response to SM, we performed a quantitative polymerase chain reaction (qPCR) analysis of TNFα mRNA with and without birinapant treatment on 16 patient samples. There was no significant difference in the expression of TNFα in samples with high or low sensitivity to birinapant, and an increase in TNFα expression upon birinapant treatment for 6 hours could only be detected in 2 of 16 cases tested (median increase of 1.156- and 0.9475-fold, respectively; Fig. 3A). We also did not observe significant phosphorylation of p65, a marker for canonical NF-κB activation by TNFα, after SM treatment in SM-sensitive cells (Fig. 3B). Blocking TNFα signaling with a specific blocking antibody only partially decreased the response to SM in sensitive cells (Fig. 3C). In contrast, TNFα blockade was highly effective in a control experiment using a patient sample that only became SM-responsive when exogenous TNFα ligand was added (fig. S2A). Overall, these data indicate that ALL responsiveness to SM is not determined by differential TNFα expression and seems only partly dependent on ligand signaling.

Fig. 3. Differential expression of TNFα, RIP1, apoptotic, or necroptotic mediators cannot explain ALL response to SM.

(A) SM-sensitive xenograft cells were placed in monoculture with or without birinapant treatment and examined after 6 hours, after which cells were lysed and examined for TNFα expression by reverse trancription PCR. ns, not significant. (B) Xenograft cells were examined by Western blotting for p65 phosphorylation at the indicated time points of birinapant incubation as shown. (C) SM-sensitive xenograft cells were cocultured with mesenchymal stem cells and treated with varying doses of birinapant in the presence (red lines) or absence (black lines) of TNFα blocking antibody (1 μg/ml). After 48 hours, cells were stained for viability and examined by automated microscopy. Graphs show the mean ± SEM from three experiments performed in duplicate. (D) A panel of 12 patient-derived xenografts with varying SM sensitivity was examined by Western blotting to assess the expression of various apoptotic or necroptotic mediators.

We next tested whether varying expression of key players of the apoptotic and necroptotic response might explain the differential response to SM in patient samples. By Western blotting, we observed no consistent correlation of cIAP1, cIAP2, XIAP, CASP8, FADD, RIP3, MLKL, or RIP1 protein expression with sensitivity to SM (Fig. 3D). Furthermore, SM treatment induced efficient degradation of cIAP1 in both sensitive and nonsensitive samples (fig. S2B). These data indicate that neither differential expression of mediators of the apoptotic and/or necroptotic death programs nor SM-induced degradation of cIAP1 correlate with the response to SM in primary ALL.

RIP1 drives concurrent SM-induced apoptosis and necroptosis

In leukemic cell lines, only moderate response to single agent birinapant was detected in two of five cell lines tested (Fig. 4A). Consistent with previous observations in other cell lines (21, 30), addition of TNFα greatly sensitized most ALL cell lines to SM (Fig. 4B). In contrast to patient-derived cells, apoptotic responses predominated in cell lines, with only Jurkat cells recapitulating a mixed death phenotype (Fig. 4C). This mixed cell death phenotype does not appear to be driven by underlying clonal differences within the population because a single-cell clone line of Jurkat cells responded with similarly mixed cell death (fig. S3A). Thus, rather than consisting of different populations that respond by either necroptosis or apoptosis, we observe Jurkat cells to activate apoptosis and necroptosis simultaneously (fig S3, B and C).

Fig. 4. RIP1 drives SM-induced apoptosis and necroptosis.

(A) Leukemic cell lines were treated with varying doses of birinapant (Bir) as shown, for 48 hours. Viability was assessed using CCK-8 colorimetric cell viability assay. (B) Leukemic cells were similarly treated with varying doses of birinapant in the presence of human TNFα (10 ng/ml). (C) Leukemic cell lines were treated with birinapant (500 nM) and TNFα (10 ng/ml) in the presence or absence of zVAD (blue), necrostatin-1 (red), or both (green). Viability was assessed as above and normalized to untreated control. (D) Four different leukemic cell lines were transduced with lentiCRISPR-EGFP targeting RIP1. Cells were subjected to two rounds of sorting to generate cell lines with ≥90% disrupted RIP1 expression as assessed by Western blotting. WT, wild type. (E and F) RIP1-WT and RIP1-CRISPR cells from partially SM-sensitive cell lines were treated with varying doses of SM as shown, for 48 hours. Cell viability was assessed using CCK-8 colorimetric viability assay. (G to J) RIP1-WT (dashed red line) and RIP1-CRISPR (solid black line) cell lines were treated as above with varying doses of birinapant in the presence of TNFα (50 ng/ml). (K) Jurkat cells were transduced with lentiCRISPR carrying varying fluorescent tags and single-guide RNA (sgRNA) sequences targeting CASP8, FADD, RIP3, and/or MLKL. Various single- and double-knockout lines were purified through at least two rounds of sorting to generate cell lines with disrupted expression as shown in the Western blot. (L to O) Apoptotic CRISPR (L), necroptotic CRISPR (M), or apoptotic/necroptotic CRISPR cells (N and O) were treated with varying doses of birinapant in the presence of TNFα (50 ng/ml). Viability was assessed at 48 hours after treatment using CCK-8. (P) Various CRISPR cell lines were treated with birinapant and TNFα alone (black) or in combination with zVAD (blue), necrostatin-1 (red), or both (green) as shown. All graphs show the mean ± SEM of at least three experiments performed in duplicate.

To explore the genetic pathways involved in the response to SM, we have adapted a lentiCRISPR gene disruption approach (31, 32). Using fluorescence-activated cell sorting (FACS) of cells transduced with fluorescently labeled lentiCRISPR vectors, we were able to disrupt single or multiple genes simultaneously without single-cell cloning steps (fig. S3D). To assess the role of RIP1 in the response to SM, we first targeted RIP1 for genetic disruption in four leukemic cell lines (Fig. 4D). Loss of RIP1 resulted in a potent rescue from SM-induced cell death in both partially SM-sensitive (Fig. 4, E and F) and SM/TNFα-sensitive cell lines independently of the mode of cell death (Fig. 4, G to J). Consistent with previous reports (33), loss of RIP1 did not inhibit moderate cell killing by TNFα alone (Fig. 4, H and J). These results show that SM-induced death is strictly dependent on RIP1 function.

To dissect the mediators of apoptosis and necroptosis downstream of RIP1, we disrupted apoptotic mediators (FADD or CASP8), necroptotic mediators (RIP3 or MLKL), or a combination thereof in Jurkat cells. Sorting of multicolor lentiCRISPR cell subsets yielded single- and double-knockout cell lines with no need for single-cell cloning (Fig. 4K). Targeted disruption of apoptotic (Fig. 4L) or necroptotic mediators (Fig. 4M) resulted in little or no rescue of cell death. Only combined disruption of both apoptosis and necroptosis blocked the response to SM treatment (Fig. 4, N and O). Combined pharmacological inhibition and genetic disruption also confirmed that cells fully switch to the alternative pathway when a single pathway is inhibited (Fig. 4P and fig. S3E). In the leukemic cell lines observed above to be apoptotic (658w, Nalm6, and CEM-C7), SM-induced death was rescued completely by the disruption of a single apoptosis gene (fig. S3F). Western blotting shows that these apoptotic leukemia cell lines lack RIP3 expression, whereas the amounts of RIP1, CASP8, FADD, and MLKL were similar compared to the mixed phenotype Jurkat cells, explaining the lack of necroptosis observed within these cells (fig. S3G). Thus, functional dissection using lentiCRISPR reveals genetic requirements for distinct death phenotypes in ALL and supports the notion that SM can drive activation of RIP1-dependent apoptotic and necroptotic cell death.

To validate RIP1 as the mediator of SM-induced cell death in vivo, we next transduced primary SM-sensitive ALL samples with lentiCRISPR targeting RIP1 (LC-EGFP-RIP1). After engraftment, we detected a consistent subpopulation of LC-EGFP-RIP1+ cells not present in untransduced controls (Fig. 5A). Western blotting shows that enhanced green fluorescent protein (EGFP)–positive cells have a clear decrease of RIP1 expression relative to EGFP and untransduced control cells after sorting (Fig. 5B, left). The proportion of EGFP+/RIP1-disrupted cells remained stable during engraftment, indicating that, similar to leukemic cell lines, RIP1 loss does not confer a growth disadvantage for these leukemia subtypes (Fig. 5, A to C, and fig. S4A). This is in contrast to previous observations in normal bone marrow, where RIP1 deficiency can cause bone marrow failure (34).

Fig. 5. RIP1 is required for SM-induced apoptosis and necroptosis in leukemia xenografts in vivo.

(A) Highly SM-sensitive patient-derived xenograft cells (106) (R-03) were transduced with lentiCRISPR-EGFP targeting RIP1 for 2 days and then injected into NSG mice. At weekly intervals after transplantation, the proportions of hCD45+ and GFP+ cells were assessed by flow cytometry. Treatment with birinapant (30 mg/kg) was initiated at day 14 and continued until the end of the experiment. (B) Cells from mice sacrificed before treatment (left) and after treatment (right) with birinapant were sorted by EGFP expression and examined for RIP1 expression by Western blotting. Control cells (Con) were untransduced xenograft cells derived from the same patient sample. (C) The total engraftment of hCD45+/hCD19+ leukemia (left axis, solid gray line) and the proportion of EGFP+ among gated hCD45+/hCD19+ cells (right axis, dashed green line) over the course of the experiment is shown; n = 2 mice. (D to F) Targeted RIP1 disruption was performed in additional patient-derived xenograft cells (VHR-10) as above. Mice were examined for engraftment and EGFP expression in hCD45+ cells and RIP1 expression as described above; n = 2 mice. (G to K) Additional patient samples were treated as above and examined for outgrowth of EGFP+ cells; n = 1 mouse per patient. (L and M) Highly SM-sensitive primary cells (R-03) were transduced with lentiCRISPRs targeted against CASP8, FADD, RIP3, and/or MLKL for 2 days and then injected into NSG mice. Treatment with birinapant was initiated at day 46 and continued until the end of the experiment. Mouse peripheral blood was examined for engraftment and fluorescent markers at varying time points. Graphs show pregated populations to exclude triple- and quadruple-positive cells. (N to P) Fold change (N) in proportion of exclusive single- and double-positive populations from initial detection over the course of the experiment (graph shows the mean result from two mice of each type). Populations of WT, double-knockout, and quadruple-knockout cells from above were expanded through xenograft and examined by Western blotting (O) and in vitro response to birinapant (P) (graph shows the mean from three experiments performed in duplicate).

As hypothesized, birinapant treatment resulted in a selective expansion of the EGFP+/RIP1-deficient hCD19+ leukemic cells (Fig. 5, A to C). Similar expansion of EGFP+/RIP1-deficient cells after SM treatment was observed in seven SM-sensitive patient samples (Fig. 5, A to K). After birinapant treatment, both EGFP+ and EGFP subpopulations exhibited loss of RIP1 (Fig. 5, B and E), likely because we were unable to differentiate cells expressing low amounts of EGFP from truly EGFP cells. Consistent with this, we were able to detect integrated RIP1-specific sgRNA in sorted EGFP cells after birinapant treatment (fig. S4B). We also confirmed that there was no loss of RIP1 after treatment of untransduced patient cells (fig. S4C). Overall, these results show that RIP1 expression is not required for leukemic propagation and is essential for SM-induced death in vivo.

We next recapitulated our double-knockout experiments to demonstrate parallel induction of apoptosis and necroptosis in vivo. Colored lentiCRISPR constructs targeting key mediators of both pathways (LC-EGFP-RIP3, LC-mCherry-MLKL, LC-BFP-FADD, and LC-RFP657-CASP8) were transduced into an SM-sensitive ALL xenograft (R-03). At 39 days after injection of transduced cells, we detected subpopulations of all expected cell types, with the curious exception of LC-RFP657-CASP8 (Fig. 5, L and M, and fig. S4, D and E). Double-positive cells bearing lentiCRISPRs targeting apoptotic and necroptotic genes (RIP3/FADD or MLKL/FADD) were initially extremely rare but expanded after SM treatment (Fig. 5, L to N). In contrast, cells with single or double knockout for two necroptosis genes (LC-EGFP-RIP3+ and LC-mCherry-MLKL+) showed marginal or no outgrowth after treatment (Fig. 5, L to N). Control cells with nontargeted lentiCRISPRs also exhibited no selection after treatment (fig. S4F). To confirm genetic disruption in double- and quadruple-knockout cells, we isolated cells from the spleen, sorted and expanded them by xenografting, and found that the cells showed the expected loss of protein expression and the presence of sgRNA sequences (Fig. 5O and fig. S4G). In addition, xenografts with targeted disruption of MLKL/FADD, RIP3/FADD, or MLKL/RIP3/FADD/CASP8 showed significantly (P < 0.0001) decreased responses to birinapant in vitro, whereas sorted wild-type cells showed a response similar to untransduced control cells (Fig. 5P). These results provide a proof of concept that multicolored lentiCRISPR gene targeting can be used for focused in vivo dissection of signaling pathways directly in primary patient cells and offer an alternative to established functional genomic strategies (35).

RIP1 cell death pathways are not exploited by standard antileukemic therapies

Having generated RIP1-deficient patient-derived ALL samples (fig. S5A), we next compared the in vitro drug response in wild-type and RIP1-deficient cells derived from two steroid-resistant high-risk samples and one steroid-sensitive standard risk sample. We observed a more than 50-fold increase in the IC50 of RIP1-deficient cells for birinapant and LCL161 (Fig. 6A). In contrast to reports implicating RIP1-dependent signaling in the response to DNA-damaging agents such as vincristine, doxorubicin, or etoposide (36), loss of RIP1 in primary ALL samples did not affect response to these drugs (Fig. 6B). In a panel of 14 antileukemic drugs, the response to DNA stress agents (etoposide, doxorubicin, vincristine, and topotecan), frontline antileukemic drugs (dexamethasone, cytarabine, 6-thioguanine, 6-mercaptopurine, or bortezomib), or preclinical compounds (ABT-199, obatoclax, or JQ-1) was unaffected by deletion of RIP1 (Fig. 6, B to D, and fig. S5, B and C). We also observed mostly unchanged sensitivity to a wider panel of chemotherapeutic drugs in the 658w-LC-EGFP-RIP1 leukemic cell line that we generated as above (fig. S5D). RIP1-deficient 658w cells showed a somewhat attenuated response to some drugs, particularly to obatoclax, but this was not seen in RIP1-deficient patient-derived leukemia cells (Fig. 6D). Overall, these data indicate that RIP1-dependent death is not exploited by standard chemotherapeutics, identifying a distinct vulnerability targeted by SM in refractory and relapsed ALL.

Fig. 6. RIP1 cell death pathways are not exploited by standard antileukemic therapies.

(A to C) WT and RIP1-CRISPR cells generated as described in Fig. 5 from primary samples R-03, VHR-10, and SR-03 were cocultured with mesenchymal stromal cells and challenged with increasing doses of various chemotherapeutic drugs for 48 hours. Cell viability was assessed by CyQUANT staining to mark live cells and was examined by automated microscopy. Survival curves from patient samples for three WT (broken black lines) and three corresponding RIP3-deficient cells (green lines) show the mean viability ± SEM (performed in duplicates). (D) The IC50 for the response of WT (open symbols) and RIP1-CRISPR (filled green symbols) cells from three primary samples (VHR-10, R-03, and SR-03) to various drugs is shown. Experiments show the mean from three independent experiments performed in duplicate. Inset shows the mean fold change in IC50 for RIP1-deficient cells ± SEM. (E) Model showing the central role of RIP1 in mediating the induction of apoptosis and necroptosis in response to SM.


There is an urgent need for the development of new treatment approaches for patients with relapsed ALL who do not respond to current therapy (37). To date, only a few second-line drugs are available for salvage treatment of chemoresistant disease. Here, we have shown that birinapant, an SM compound currently in phase 2 clinical trials for application in various solid and hematological malignancies (20), induces potent leukemic cell death through parallel activation of RIP1-dependent apoptosis and necroptosis (Fig. 6E) and has strong antileukemic efficacy in vivo in a subset of primary ALL, including cases with relapsed and refractory disease.

To dissect the signaling pathway underlying the antileukemic potency of birinapant, we have adapted a lentiCRISPR gene disruption approach previously used for large-scale screening of knockout cell lines and tumor modeling in mice (31, 32). We provide proof of concept that multicolored lentiCRISPRs can be directly applied for genetic dissection in vitro and in vivo in primary human cancer. Given recent improvements in humanized mouse models for studying human disease (38), the development of multicolored CRISPR-Cas9 lentiviral constructs provides an alternative to previously established in vivo functional genomic strategies (35, 39, 40), with specific focus on dissecting the complex interaction of several genes that contribute to a specific phenotype. Our demonstration that simultaneous disruption of two distinct cell death pathways downstream of RIP1 is necessary to block SM response underlines the power of this approach for disentangling genetic redundancy.

The ability of SM-treated ALL cells to switch between apoptotic and necroptotic programmed cell death is reminiscent of parallel pathways of death previously observed in T cell expansion and contraction (41), suggesting that engaging both death pathways could represent a distinct vulnerability in lymphoid malignancies. Such dual death activation could circumvent clonal escape, which is a driver of drug resistance (8, 42) from either death mechanism. We consistently detected a contribution of the necroptotic pathway to SM-induced cell death in primary patient material, with no sample undergoing a purely apoptotic cell death, supporting the notion that the necroptotic pathway may constitute a particular Achilles’ heel for targeted therapy in refractory disease (24). Upstream of this dual death activation, our data pinpoint RIP1 as the critical target of SM-induced cell death. RIP1-deficient ALL displayed no defect in the response to several frontline chemotherapeutic agents, providing a strong argument that RIP1 deficiency is unlikely to be selected for in resistant cases that have been subjected to multiple rounds of chemotherapy. Consistent with this, we observed no consistent variation in the expression of RIP1 within patient-derived samples. Our finding that RIP1-deficient cells show normal sensitivity to DNA-damaging agents (doxorubicin and etoposide) is in stark contrast to a number of reports detailing activation of the cytosolic RIP1-dependent ripoptosome complex in response to DNA damage in other cancer cell line models (36, 43, 44). The reason for this discrepancy is not readily clear but could perhaps be ascribed to cell type differences because previous studies used cell lines derived from solid tumor models. Strategies to combine SM with RIP1-activating immunostimulatory agents, such as interferon (45, 46) or pathogen-associated molecular patterns (47), may be highly effective, as has recently been demonstrated using oncolytic viruses or immunostimulatory agents in breast cancer xenografts (48).

Immunodeficient mice used for xenografts of primary human ALL lack a functional immune system. In this setting, the influence of functional immune cells cannot be studied. At present, there are no syngeneic mouse models available to model drug-resistant or relapsed ALL. As for many other preclinical studies of this kind, the potential of new agents can only be ultimately assessed in appropriately planned clinical trials. The importance of the IAPs as chemotherapeutic targets has long been recognized (49), but potential toxicity and low single-agent efficacy of monovalent SM compounds have so far hindered their clinical application (19, 50).

Our data imply that strong antileukemic activity of SM will depend on adequate selection of SM-sensitive ALL patients. Despite the clear implication of RIP1 and its downstream mediators in the leukemic response to SM, expression of these proteins is not predictive of SM sensitivity. Although extended genetic and functional screening may yet reveal predictive markers of SM response, our results thus far highlight the need for alternative means to identify sensitive patients. Ex vivo drug response profiling could provide an important functional layer of information for clinical application of SM treatment. Similar approaches are increasingly being incorporated into diagnostic workflows to identify candidate drugs for the treatment of relevant patient groups (23, 51, 52).

Together, our results provide a rationale for further development of birinapant and other SM compounds for experimental therapy in high-risk and relapsed ALL. The fact that these agents trigger a potent dual activation of both apoptosis and necroptosis, which is not exploited by contemporary chemotherapy, provides a strong argument to evaluate SM compounds for personalized treatment of refractory and relapsed ALL patients.


Study design

The goal of this study was to assess the antileukemic potential of SM in drug-resistant and relapsed ALL ex vivo and in vivo on the basis of the initial finding of specific antileukemic activity of SM against primary ALL samples. Furthermore, we aimed to delineate the mode of action on a molecular level and to identify the key players that control and mediate SM-induced cell death. To achieve this, we used in vivo antileukemic activity assays in xenograft models and developed a CRISPR-based genome editing methodology that allowed simultaneous disruption of several candidate genes in primary human tumor samples. Sample sizes were chosen on the basis of previous experience that included statistical evaluation.

Human samples

Xenografts were obtained from primary human ALL samples recovered from cryopreserved bone marrow aspirates of patients enrolled in the ALL-BFM 2000, ALL-BFM 2009, and ALL-REZ-BFM 2002 studies. Informed consent was given in accordance with the Declaration of Helsinki, and approval was granted by the Ethics Commission of the Kanton Zürich (approval no. 2014-0383). Samples were classified as standard risk, medium risk, high risk, very high risk, morphological nonresponders, or relapse samples according to the clinical criteria used in the ALL-BFM 2000 study (24, 26).


Multicolored lentiCRISPR plasmids were derived from the lentiCRISPR v1 plasmid (Addgene catalog #49535) (31). Puromycin resistance cassette was removed using Nhe I and Mlu I restriction enzymes, and plasmids were cloned with fluorescent EGFP, TagBFP (Addgene catalog #44247), mCherry, or RFP657 (Addgene catalog #31959) sequences by standard restriction cloning technique. Cloning of sgRNA into lentiCRISPR plasmids was performed with a single-tube restriction and ligation protocol (see the Supplementary Materials and Methods).

A number of sgRNA sequences were screened for gene disruption activity in cell lines by Western blotting, and the most effective sequences were chosen for further experiments. The specific sgRNAs used are listed in table S2.

Drugs and chemicals

Recombinant human TNFα was purchased from Gibco/Life Technologies (catalog #PHC3011); TNFα neutralizing antibody was purchased from Cell Signaling (catalog #7321); birinapant (TL32711) was purchased from Selleckchem (catalog #S7015) for in vitro studies and from ChemieTek (catalog #CT-Biri) for in vivo studies; Z-VAD-FMK was purchased from ApexBio (catalog #A1902); necrostatin-1 was purchased from BioVision (catalog #2263-1); and LCL161 was purchased from Novartis. Vincristine (catalog #S1241), mitoxantrone (catalog #S2485), idarubicin (catalog #S1228), cytarabine (catalog #S1648), topotecan (catalog #S1231), methotrexate (catalog #S1210), doxorubicin (catalog #S1208), clofarabine (catalog #S1218), obatoclax (catalog #S1057), dactolisib (catalog #S1009), prednisolone (catalog #S1737), dexamethasone (catalog #S1322), cyclophosphamide (catalog #S1217), and bortezomib (catalog #S1013) were purchased from Selleckchem. 6-Thioguanine (catalog #A4882), 6-mercaptopurine (catalog #852678), and asparaginase (catalog #A3809) were purchased from Sigma. DAPT was purchased from Axon Medchem (catalog #Axon1484).

For live-cell microscopy, Jurkat cells were treated as indicated and stained directly with propidium iodide and CellEvent Caspase-3/7 substrate (catalog #C10423, Life Technologies) for 30 min. Live-cell microscopy was then performed using Zeiss Axio Observer fluorescence microscope under cell culture conditions (37°C and 0.5% CO2) to track changes in cell viability over time.

Biochemical assays

Complete methods for in vitro viability measurement, Western blotting, and FACS staining are provided in the Supplementary Materials and Methods.

In vivo experiments

Patient-derived ALL cells (106) were transplanted intravenously into immunodeficient NSG mice. For in vivo experiments, birinapant was dissolved in 12% Captisol (Ligand Pharmaceuticals) with 0.1% tris (pH 6.8). Birinapant (30 mg/kg) was given daily by intraperitoneal injection during the indicated periods. For LCL161 studies, LCL161 was dissolved in 100 mM sodium acetate buffer adjusted to pH 4.3 to 4.6 with HCl. The leukemic engraftment was measured weekly by FACS quantification of human leukemic cells in peripheral blood after red blood cell lysis and staining with hCD19-PE-Cy7 (catalog #302208, BioLegend), hCD45–Alexa Fluor 647 (catalog #304018, BioLegend), and mCD45–eFluor 450 (catalog #48-0451-82, eBioscience). Age- and sex-matched animals were assigned randomly to the treatment or vehicle [water with 12% Captisol and 0.1% tris (pH 6.8) for birinapant studies and 100 mM sodium acetate buffer for LCL161 studies] groups. Results were derived from direct analysis of leukemic engraftment in blood, and thus, no blinding was necessary. Animal sample sizes were chosen to minimize the number of animals used because interanimal variation in leukemic engraftment was generally low. No animals were excluded from analysis. In vivo experiments were approved by the veterinary office of the Canton of Zürich.

For lentiCRISPR gene disruption in patient-derived xenografts, cells were placed in monoculture in RPMI + 10% fetal bovine serum and incubated at 37°C for 24 hours. Cells were then exposed to high-titer lentiCRISPR vector (multiplicity of infection ≥ 1) in the presence of polybrene (10 ng/ml). After 24 to 48 hours, cells were washed three times with phosphate-buffered saline and injected directly into NSG mice. Mice were then treated with birinapant as described in the text, and peripheral blood was examined at varying time points for expression of hCD19 and appropriate fluorescent markers.


For in vivo birinapant treatment experiments, the significance of divergence of survival curves was determined by Kaplan-Meier survival analysis and Mantel-Cox test. For in vitro analyses, we used Student’s t test to detect significant divergence between test conditions. Statistical analysis was performed using GraphPad Prism 5 software package.


Materials and Methods

Fig. S1. ALL patient samples are sensitive to LCL161 in vitro but not in vivo.

Fig. S2. TNF blocking effectively rescues TNF-dependent sensitization to birinapant.

Fig. S3. Birinapant induces dual activation of apoptosis and necroptosis in ALL cell lines.

Fig. S4. In vivo birinapant treatment selects CRISPR-driven RIP1-deficient cells.

Fig. S5. RIP1 is not required for response to a wide range of chemotherapeutic agents.

Table S1. Patient characteristics.

Table S2. sgRNA-targeting sequences used for gene disruptions.


Acknowledgments: We thank T. Radimerski of Novartis for providing LCL161 and W.W.-L. Wong for critical comments on the manuscript. Funding: This work was supported by the Stiftung Kinderkrebsforschung Schweiz, the MAM Funds of the Children’s Research Centre of the University Children’s Hospital Zürich, the Empiris Foundation, the clinical research focus program Human Hemato-Lymphatic Diseases of the University of Zürich, the Swiss Cancer League (KFS 3609-02-2015), the Novartis Foundation for Biomedical Research, the Swiss National Science Foundation (SNF 310030–133108), the Canadian Institutes for Health Research, the Forschungskredit of the University of Zürich (FK-14-016), and the Fondation Panacée. Author contributions: S.M., J.A.-G., B.M., and L.H. performed experiments and analyzed data; G.C., M. Stanulla, M. Schrappe, A.v.S., and C.E. provided samples and clinical data; S.M., J.A.-G., J.-P.B., and B.C.B. wrote the manuscript; J.-P.B. and B.C.B. conceived and supervised the study. Competing interests: J.-P.B. served on a pediatric advisory board for Roche. All other authors declare that they have no competing interests.

Stay Connected to Science Translational Medicine

Navigate This Article