Reprogramming RIPK3-induced cell death in malignant B cells promotes immune-mediated tumor control

Abstract

Cancer therapies that induce immunogenic cell death such as necroptosis can arm the immune system to clear residual tumor cells. Receptor-interacting serine/threonine protein kinase 3 (RIPK3) signaling can trigger necroptosis, orchestrating immune responses to solid tumors. Whether RIPK3 activity can be exploited for controlling B cell malignancies remains unclear. Here, we establish a strategy to manipulate RIPK3 activity in malignant B cells and promote immune-mediated tumor control. By controlling and visualizing RIPK3 signaling using intravital imaging, we established that RIPK3 activation promoted apoptosis and rapid macrophage engulfment but failed to induce necroptosis due to limited MLKL expression in B-lineage cells. RIPK3-induced cell death could be diverted toward necroptosis with type I IFN and caspase inhibition. Exploiting these findings, we showed that a combination therapy activating RIPK3 with SMAC mimetics while suppressing caspase activity and providing type I IFN resulted in immune-mediated control of B cell tumors. Hence, reprogramming RIPK3 activity represents an attractive therapeutic opportunity to target B cell malignancies.

A cell death reprogramming therapy that promotes robust necroptosis leads to immune-mediated clearance of malignant B cells.

INTRODUCTION

Cancer therapies aim at maximizing the eradication of tumor cells by either direct or indirect mechanisms. Chemotherapeutic regimens or targeted therapies directly induce tumor apoptosis but are often challenged by resistance mechanisms. Tumor development is intrinsically associated with alterations in cell death pathways, with escape from apoptosis being considered a hallmark of cancer (1). Additional resistance mechanisms can accumulate because of the selective pressure imposed by successive treatment regimens (2). Reprogramming cell death pathways has therefore emerged as an interesting approach to bypass these primary or acquired resistance mechanisms. For example, strategies that trigger necroptosis have been proposed to overcome resistance to apoptosis observed in some patients treated for acute lymphoblastic leukemia (3). Another key issue is the interplay between cell death and the immune response. In this respect, immunogenic forms of cell death may endow the immune system with the ability to clear residual tumor cells (4–6).

Among the various cell death players, receptor-interacting serine/threonine protein kinase 3 (RIPK3) represents an attractive target for cell death pathway reprogramming. Although RIPK3 is well known for its canonical function as necroptosis inducer (7), it was originally identified as an apoptosis inducer (8, 9). RIPK3 phosphorylates mixed lineage kinase domain-like (MLKL) protein, inducing its oligomerization and translocation to cellular membranes where it promotes pore formation and membrane disruption (10, 11). RIPK3 can also induce caspase-8 activation and apoptosis (12). RIPK3 can trigger necroptosis or apoptosis depending on its kinase activity (13, 14), local target availability (15), and/or the presence of posttranslational modifications (16).

RIPK3 has been shown to favor necroptotic cell death in solid tumors, leading to efficient antitumor immune responses (17, 18). Necroptosis is considered a highly immunogenic cell death due to the release of intracellular mediators after membrane disruption and the active production of inflammatory cytokines within necroptotic corpses (4, 19). How RIPK3 activation dictates cell fate decision between necroptosis and apoptosis in hematological malignancies remains largely unknown.

Here, we used an activatable system to show that specific RIPK3 activation in B cell tumors induces apoptosis in vivo. We find that activating RIPK3 while simultaneously inhibiting caspases can moderately favor necroptosis and show that this phenomenon is considerably amplified by type I interferon (IFN), which increases the expression of MLKL. Exploiting these findings, we show that combining small-molecule mimetics of the second mitochondria-derived activator of caspase (SMAC) protein (20), with the pan-caspase inhibitor emricasan, as well as IFN-β, maximizes RIPK3-induced necroptosis and antitumor immunity providing the most efficient strategy to control tumor B cell burden in vivo. Thus, reprogramming of RIPK3 activity in B cell malignancies represents an attractive therapeutic approach to effectively control B cell tumors.

RESULTS

RIPK3 activation induces apoptosis in B cell tumors

RIPK3 is a major hub in programmed cell death that has been widely studied in solid tumors. To dissect how RIPK3 signaling affects malignant B cells, we relied on an activatable RIPK3-2xFKBP chimeric protein (actRIPK3), upon aggregation with the B/B dimerizer reagent (fig. S1A) (17, 21). RIPK3 oligomerization has been shown to force RIPK3 activation and induce cell death (21). This system offers the possibility to selectively activate RIPK3 signaling with high temporal control. We introduced actRIPK3 in two distinct B cell tumor models, an Abl-driven acute B cell leukemia cell line (pro-B cells) and in a Myc-driven B cell lymphoma (Eμ-Myc) (fig. S1, B and C). Activation of RIPK3 by B/B dimerizer–mediated oligomerization in vitro triggered rapid cell death in both B cell tumor models, with most cells dying within 6 hours (fig. S1D).

To characterize the cell death pathway(s) associated with RIPK3 activation, cells were examined for activated caspase-3 expression in the presence of the fixable Live Dead viability dye to assess membrane permeabilization. As shown in fig. S1E, RIPK3 activation, upon addition of B/B dimerizer, resulted in robust caspase-3 activity, a hallmark of apoptosis, in both pro-B and Eμ-Myc tumors. Most tumor cells also incorporated the vital dye, which is expected at late stages of apoptotic cell death (fig. S1, E and F). We also observed rapid membrane blebbing, a morphological feature of cells undergoing apoptosis and, at later stages, membrane permeabilization as detected by 4′,6-diamidino-2-phenylindole (DAPI) incorporation (fig. S1G). Overall, our results suggest that selective RIPK3 activation in B cell malignancies is characterized by rapid apoptosis with little evidence for nonapoptotic cell death. Of note, RIPK3 activation in the NIH-3T3 fibroblast cell line transduced with activatable RIPK3 construct induced necroptosis characterized by membrane permeabilization in the absence of caspase-3 activity and induction of phospho-MLKL (fig. S2). Our results with B cell tumors also contrast with previous observations made on solid tumors where necroptosis was the dominant cell death pathway triggered by forced RIPK3 activation (18).

It remained essential to determine the fate of B cell tumors undergoing RIPK3 activation in vivo. We tested whether selective RIPK3 activation could be achieved in tumor-bearing mice treated with a single dose of B/B dimerizer (Fig. 1A). We observed a substantial decrease in tumor burden as early as 3 hours postinjection and a complete elimination of tumor cells 3 days later (Fig. 1B). Thus, B/B dimerizer–mediated oligomerization could be performed efficiently in vivo, and RIPK3 activation induced rapid and efficient tumor cell death. To track tumor cell fate in vivo upon RIPK3 activation, we transduced actRIPK3-bearing B cell tumors with a fluorescence resonance energy transfer (FRET)–based reporter for caspase-3/7 activity (Fig. 1C) (22). Mice harboring these pro-B cell tumors were subjected to intravital two-photon imaging of the bone marrow (BM). We observed a burst of apoptotic events between 30 and 60 min after B/B dimerizer injection with the vast majority of tumor cells becoming positive for the caspase reporter (FRET^neg) (Fig. 1, D and E, and movie S1). Similar results were observed using mice bearing Eμ-Myc B cell tumors (Fig. 1, D and E). To assess the fate of B cells undergoing apoptosis at the tumor site, we visualized tumor-associated macrophages during RIPK3 activation in B cell tumors. As shown in Fig. 1 (F to H), fig. S3, and movie S2, B cell tumors first activated caspase-3/7 (FRET^neg) and were then rapidly engulfed by tumor-associated macrophages. Macrophage-mediated capture of apoptotic tumors was both very efficient (Fig. 1F and movie S2) and rapid, typically occurring in less than 15 min postcaspase activation (Fig. 1H and fig. S3). In sum, our data demonstrate that selective activation of RIPK3 in two models of B cell malignancy leads to rapid cell apoptosis and subsequent macrophage-mediated scavenging.

Fig. 1. RIPK3 signaling promotes apoptosis in B cell malignancies in vivo.

(A and B) Mice were injected with actRIPK3-pro-B cells and treated with B/B dim or control vehicle. (A) Experimental scheme for actRIPK3 induction in vivo. (B) Number of tumor cells in the blood and BM at 3 hours or 3 days posttreatment. Representative of two independent experiments with three mice per group in each experiment. Each dot represents one mouse. (C) Experimental setup of BM intravital imaging in mice bearing tumors expressing FRET-based caspase-3/7 reporter (DEVD). (D) Time-lapse images showing apoptotic events in tumor-bearing mice (DEVD–actRIPK3–pro-B or DEVD–actRIPK3–Eμ-Myc) receiving B/B dim. Live tumor cells (magenta) and apoptotic tumors (blue). Scale bars, 20 μm. Representative of 16 movies (obtained in four independent mice) treated with B/B (16 movies) and 8 movies (in two independent mice) treated with the control vehicle. (E) Kinetics of appearance of FRET^neg tumors for individual movies and corresponding means. (F) Intravital images from the BM of mice bearing DEVD–actRIPK3–pro-B cell tumors showing phagocytosis of tumor cells by F4/80⁺ macrophages (green) upon apoptosis induction. Macrophages are visualized by intravenous (iv) injection of anti-F4/80 antibody. Scale bars, 20 μm. Examples of cell (white arrowheads) undergoing apoptosis and subsequent phagocytosis are shown. Scale bars, 10 μm. Representative of 16 movies (obtained in four independent mice). (G) Kinetics of FRET loss and phagocytosis in 50 representative tumor cells following B/B dim administration. Horizontal lines represent individual cells. (H) Timing of FRET loss relative to phagocytosis events. Cumulative events in 4-min time intervals. Data in [(G) and (H)] were compiled from 16 movies acquired from four independent mice and representing n = 248 events. Data are expressed as mean ± SEM. Unpaired Student’s t test (B) or two-way analysis of variance (ANOVA) (E) was used for statistical analysis (E). ns, not significant; **P < 0.01 and ***P < 0.001. FACS, fluorescence-activated cell sorting.

Redirecting the fate of RIPK3 activation from apoptosis to necroptosis in vivo

Because necroptosis is typically associated with stronger antitumor immune responses as compared to apoptosis, we sought to promote necroptotic cell death in B cell tumors. RIPK3-mediated apoptosis has been shown to require caspase-8 activation (13, 15). Accordingly, RIPK3 oligomerization in the presence of the pan-caspase inhibitor zVAD-fmk substantially reduced cell death induction in vitro (Fig. 2A). A small but sizeable fraction of tumor cells (15 to 20%) underwent cell death despite efficient caspase inhibition, indicating that redirection to nonapoptotic cell death was possible but relatively inefficient (Fig. 2, A to C). This nonapoptotic cell death corresponded morphologically to necroptosis with cell enlargement, intact nucleus, absence of blebbing, and uptake of the vital dye (Fig. 2D). Western blot analyses confirmed that hallmarks of apoptosis (caspase-3/8 cleavage) were dominant upon B/B dimerizer treatment alone, while features of necroptosis (up-regulation of phospho-MLKL in the absence of caspase cleavage) were detected upon B/B dimerizer plus zVAD-fmk (Fig. 2E). We observed a similar shift toward low levels of necroptosis by generating Casp8^−/− B cell tumors and treating the cells with the B/B dimerizer (Fig. 2F and fig. S4A).

Fig. 2. RIPK3 activation in the presence of caspase inhibition triggers moderate levels of necroptosis in B cell tumors.

(A) Cell death induction in actRIPK3-pro-B cells pretreated for 30 min with the pan-caspase inhibitor zVAD-fmk followed by B/B dim stimulation. (B and C) actRIPK3-pro-B cells were cultured with B/B dim in the presence or absence of zVAD-fmk. (B) Representative plots showing the double staining act-Casp3/LD in actRIPK3-pro-B cells treated in vitro with the indicated conditions. Numbers represent the percentage of gated cells. LD, fixable Live Dead viability dye. (C) Quantification of cells with act-Casp3 and/or LD staining in the indicated conditions. Results are representative of five independent experiments. Data are expressed as mean ± SEM. (D) Time-lapse images showing morphological changes in actRIPK3-pro-B cells stimulated with B/B dim in the presence or absence of zVAD-fmk. Scale bars, 10 μm. Data are representative of three independent experiments. A representative cell death in the absence or presence of zVAD-fmk is shown with white arrowheads. (E) Expression of cell death–related proteins in actRIPK3-pro-B cells upon treatment of with B/B dim alone or in combination with zVAD-fmk. The molecular weight for each protein is indicated by arrows. The lower band corresponds to phospho-MLKL (pMLKL). (F) actRIPK3-pro-B cells with genetic ablation of caspase-8 by CRISPR-Cas9 were cultured with B/B dim for 3 hours. Quantification of cells with act-Casp3 and/or LD staining. Data are representative of three independent experiments. Data are expressed as mean ± SEM. One-way ANOVA was used for statistical analysis (A). ***P < 0.001.

We next evaluated the ability to monitor necroptosis in vivo and performed intravital imaging of the BM in mice with established actRIPK3-Casp8^−/− pro-B cell tumors. Tumors also expressed the CFP^(DEVD)YFP reporter that can reveal both apoptotic events (due to the cleavage of the DEVD peptide) and cell digestion within phagocytes [due to the loss of yellow fluorescent protein (YFP) fluorescence at low pH] as we have detailed previously (Fig. 3A) (23). Contrasting with what we observed with caspase-8–proficient tumors for which RIPK3 activation led to apoptosis (loss of FRET) followed by macrophage engulfment (Fig. 1, F to H, and movie S2), RIPK3 activation in Casp8^−/− tumors led to rapid tumor engulfment without detectable apoptosis measured by FRET loss (Fig. 3, B to D, and movie S3). However, change in reporter activity was seen after phagocytosis most likely reflecting loss of YFP fluorescence in the acidic phagosomal environment as reported previously (Fig. 3, B to D, and movie S3) (23). The distinct sequences of events observed in caspase-8–proficient and caspase-8–deficient tumors upon RIPK3 activation (Figs. 1, F to H, and 3, B to D) are consistent with the induction of apoptotic and necroptotic cell death in vivo, respectively. Both types of cell death triggered clearance by tumor-associated macrophages within 10 to 20 min.

Fig. 3. Reprogramming RIPK3-induced cell death in vivo.

BM intravital two-photon imaging was performed in mice bearing DEVD-actRIPK3-Casp8^−/− pro-B cell tumors before and immediately after B/B dim administration. (A) Experimental setting and predicted fate for DEVD reporter fluorescence in cells undergoing necroptosis and subsequently phagocytosed. Necroptotic cells are not expected to lose FRET before phagocytosis but are expected to lose FRET in the acidic environment of phagocytic lysosomes due to a drop of YFP [but not cyan fluorescent protein (CFP)] fluorescence at low pH. Ab, antibody. (B) Time-lapse images showing the in vivo fate of DEVD-actRIPK3-Casp8^−/− pro-B cells upon B/B dim administration. Macrophages (shown in green) were labeled by injection of a fluorescent anti-F4/80 antibody. Note that upon B/B dim administration, tumor cells undergo phagocytosis in the absence of caspase-3/7 activity and subsequently lose FRET after phagocytosis. Examples of individual cells (white arrowheads) undergoing phagocytosis are shown. Scale bars, 10 μm. Representative of n > 15 movies obtained from three independent mice. (C) Kinetics of FRET loss and phagocytosis in tumor cells lacking caspase-8. Each horizontal line represents a single cell. Fifty representative cells are shown for Casp8^−/− tumors. (D) Quantification of the timing of FRET loss relative to the phagocytosis events for tumor lacking caspase-8. Results are compiled from 10 movies. (E and F) Mice were injected with pro-B cell tumors expressing or lacking caspase-8 and treated with a single dose of B/B dim or control vehicle 5 days later. (E) Experimental scheme. (F) Number of tumor cells in the blood of B/B dim–treated mice versus control at 24 hours posttreatment. Representative of two independent experiments with n = 6 mice per group in each experiment. Each dot represents one mouse. Unpaired Student’s t test was used for statistical analysis (F). ***P < 0.001.

We next compared the extent of RIPK3-mediated induction of apoptosis or necroptosis in vivo by injecting the B/B dimerizer in mice bearing caspase-8–proficient or caspase-8–deficient actRIPK3–pro B cell tumors, respectively. While RIPK3-mediated induction of apoptosis was highly efficient (>99% of circulating tumor elimination), RIPK3-mediated triggering of necroptosis only led to twofold reduction in tumor load (Fig. 3, E and F). In sum, our results provided direct in vivo evidence that blockade of caspase activity can redirect RIPK3 activation toward necroptosis but only in a fraction of malignant B cells.

Low MLKL expression characterizes nontransformed and malignant B-lineage cells

To clarify the low level of necroptosis observed in B cell tumors, we examined the level of key molecules associated with cell death pathways (Fig. 4A). When comparing to mouse embryonic fibroblasts (MEFs) and bone marrow–derived macrophages (BMDMs), pro-B cell tumors expressed very low levels of MLKL and the highest levels of caspase-8 (Fig. 4B). We next assessed whether poor MLKL expression was a shared feature of malignant B cells across species. First, we analyzed MLKL expression on human B cell tumors using bulk RNA sequencing (RNA-seq) data from the Cancer Cell line Encyclopedia (24). Most (16 of 19) analyzed B cell tumors demonstrated very low expression of MLKL as compared to the THP-1 monocytic cell line known to express MLKL robustly (Fig. 4C) (25). In addition, analyzing single-cell RNA-seq data from patients with acute lymphoblastic leukemia (ALL) and diffuse large B cell lymphoma (DLBCL) (26, 27) confirmed the low levels of MLKL expression in patient malignant B cells (fig. S5, A to D). Low MLKL expression was also a feature of nontransformed B-lineage cells as observed in bulk RNA-seq of murine immature and mature B cells from the Immgen database (Fig. 4D) (28). Low expression of MLKL on mature (splenic) B cells was confirmed at the protein level (Fig. 4E). Last, analyzing single-cell RNA-seq data from human BM mononuclear cells (10X Genomics), we again observed low MLKL expression in developing B cells but not in common lymphoid progenitors (CLPs) (Fig. 4, F and G). To test whether the lack of MLKL protein expression in B cells was associated with low chromatin accessibility, we analyzed public ATACseq dataset (29). Low MLKL expression was associated with reduced chromatin accessibility in a region that overlap with the transcription start site (TSS) and exon 1 of MLKL in immature and mature murine B cells (Fig. 4H). Conversely, CLPs and monocyte/macrophages that expressed higher levels of MLKL protein showed evidence of high chromatin accessibility in these regions. Together, these data suggested that low expression of MLKL in B-lineage cells observed in mice and humans may limit the ability of malignant B cells to engage RIPK3-mediated necroptosis.

Fig. 4. Low expression of MLKL is a hallmark of B-lineage cells.

(A) Expression of Casp8, FADD, RIPK1, RIPK3, and MLKL in whole-cell extracts from pro-B cell tumors by Western blot. MEFs and BMDMs were used as a positive control. Data are representative of three independent experiments. (B) Protein quantification from (A) expressed as a fold change over MEF expression. Results are compiled from three independent experiments. (C) MLKL expression from bulk RNA-seq (Cancer Cell Line Encyclopedia database) performed on 19 human B cell lines originating from ALL, Burkitt lymphoma, and DLBCL. THP-1 cells were used as positive control. (D) MLKL expression from bulk RNA-seq (ImmGen database) in different stages of murine B cell development. Peritoneal macrophages and endothelial cells were used as positive controls. GC, germinal center. (E) MLKL protein expression assessed on murine splenic B cells, pro-B tumors, and MEFs. (F) t-SNE representation of single-cell RNA-seq (10X Genomics public datasets) from frozen human BM mononuclear cells. B cell development stages are highlighted. (G) Violin plots of MLKL expression in various stages of human B cell development from (F). Myeloid cells were used as positive control. (H) Chromatin accessibility of the MLKL locus in mouse primary cells was assessed by ATACseq. The gray area highlights changes in ATACseq peaks within the TSS and exon 1. Wilcoxon rank sum test was used for statistical analysis (G). ***P < 0.001.

Manipulating RIPK3-induced cell death in B tumor cells with type I IFN

We hypothesized that increasing MLKL expression may favor necroptosis in B cell tumors. IFNs have been shown to up-regulate MLKL transcripts and protein expression in various cell lines (30–33). We also observed increased transcription of MLKL in response to IFNs with B lymphoma cells (fig. S6). We noted a substantial increase in MLKL protein expression in pro-B cell tumors upon treatment with either IFN-α, IFN-β, and IFN-γ with the highest effect being seen with IFN-β (Fig. 5A). We therefore triggered RIPK3 activation in B tumor cells pretreated with IFN-β in the presence or absence of caspase inhibition. Tumor cell death triggered by RIPK3 activation was accelerated by IFN-β pretreatment (Fig. 5B). In the presence of caspase inhibition, the addition of IFN-β strongly promoted cell death with 80 to 90% of tumor cells dying upon RIPK3 activation (Fig. 5B). We observed that apoptosis was still the dominant pathway triggered by RIPK3 in the presence of IFN-β. However, robust necroptosis was achieved by RIPK3 activation combined with IFN-β and caspase inhibition (Fig. 5, C and D). The generation of MLKL-deficient B cell tumors confirmed that the cell death pathway promoted by the combined action of RIPK3, IFN-β, and caspase blockade was completely MLKL dependent (Fig. 5, E and F, and fig. S4B). Enhanced necroptosis in the presence of IFN-β and caspase inhibition was further supported by imaging (Fig. 5G) and by inhibition of RIPK3 kinase activity (fig. S7).

Fig. 5. Manipulating RIPK3-induced cell death by type I IFNs in vitro.

(A) Expression by Western blot of Casp8, RIPK1, RIPK3, and MLKL in whole-cell extracts from pro-B cell tumors pretreated with type I or type II IFNs. Data are representative of two independent experiments. MEF cells were used as positive control. (B) Kinetics of cell death in actRIPK3-pro-B cells treated or not with IFN-β for 24 hours and cultured later with B/B dim in the presence or absence of zVAD-fmk. (C) Representative plots showing the staining act-Casp3/LD in actRIPK3-pro-B cells stimulated for 3 hours with B/B dim with or without zVAD-fmk and IFN-β pretreatment. Numbers indicate the percentage of gated cells. (D) Frequency of cells expressing act-Casp3 and/or incorporating LD from (C). (E and F) actRIPK3-pro-B cells expressing or lacking MLKL were cultured with B/B dim with or without zVAD-fmk and IFN-β pretreatment. Cell death was assessed by flow cytometry. (E) Representative plots and (F) quantification. Results in [(C) to (F)] are representative of three independent experiments. Data are expressed as mean ± SEM. (G) Time-lapse images showing morphological changes in actRIPK3-pro-B cells pretreated with IFN-β and exposed to B/B dim with or without zVAD-fmk. Scale bars, 10 μm. Data are representative of three independent experiments. (H) Scheme summarizing the type and magnitude of cell death induced by RIPK3 signaling in pro-B tumors in the presence or absence of caspase inhibition and IFN-β.

As summarized in Fig. 5H, apoptosis appeared to be the default pathway induced by RIPK3 in these tumors, but redirection of cell death toward the necroptotic pathway can be achieved in vitro by inhibiting caspases and maximized in the presence of type I IFN. The robust induction of necroptosis was associated with higher chemokine release compared to apoptosis (fig. S8) and with increased macrophage activation in coculture (fig. S9, A to C). Furthermore, when caspase-8–sufficient or caspase-8–deficient actRIPK3–pro-B tumors were established in vivo and treated with the B/B dimerizer, we again observed higher levels of macrophage activation in conditions favoring necroptosis (fig. S9, D and E). Together, these results highlight the preferential contribution of necroptosis to promote immune activation.

Exploiting RIPK3-associated cell death pathways to control B cell tumors

Given the known immunogenicity of necroptotic cell death, our results suggested a possible benefit of combining RIPK3 activation with type I IFN and caspase inhibition to achieve optimal antitumor immune activity against malignant B cells. We therefore sought to translate these results and establish a combined therapy to optimally target B cell tumors. We used SMAC mimetics (SM) that promote cell death by inducing degradation and direct inhibition of cellular inhibitor of apoptosis proteins (IAPs) (20). We found that in vitro treatment with the bivalent SM BV6 induced moderate levels of cell death (Fig. 6A) in pro-B cells that was dependent on RIPK3 (Fig. 6A and fig. S4C). We therefore combined SM as an activator of RIPK3 with IFN-β and emricasan (a clinical grade pan-caspase inhibitor). Combining SM with emricasan or type I IFN increased the magnitude of tumor cell death, and associating the three drugs achieved maximal effects (Fig. 6B). Again, the effects mediated by the combined treatment were largely dependent on RIPK3 (Fig. 6B). As shown in Fig. 6C, the association of SM, IFN-β, and emricasan was the most efficient strategy to trigger robust necroptosis. We extended these observations to human cells by showing that the addition of IFN-β on human tumor B cell lines also boosted necroptosis induced by SM and caspase inhibition (fig. S10).

Fig. 6. A combined therapy to optimize RIPK3-induced cell death in B cell tumors in vivo.

(A to C) Pro-B cells expressing or lacking endogenous RIPK3 pretreated (or not) with IFN-β were cultured in the presence or absence of SM BV6 and pan-caspase inhibitor emricasan (Emri). (A and B) Frequency of cell death without (A) or with (B) IFN-β pretreatment. (C) Frequency of cells expressing act-Casp3 and/or incorporating LD dye. (D and E) WT mice bearing pro-B tumors were treated with three doses of IFN-β and three doses of SM birinapant (Biri) with or without Emri. (D) Experimental design. (E) Frequency of tumor cells among CD45⁺ cells in the blood, lymph nodes (LNs), and BM 14 days after tumor inoculation. Each dot represents one mouse. Data are representative of three independent experiments with n = 5 mice per group in each experiment. ip, intraperitoneal. (F) Survival of WT or Rag2^−/− B6 mice harboring pro-B cell tumors and treated with a combination of birinapant, emricasan, and IFN-β. n = 12 to 14 mice per group compiled from two independent experiments. (G and H) Induction of tumor-specific CD8⁺ T cell responses upon cell death reprograming therapy. (H) H-2K^b-OVA tetramer staining was performed 14 day post–pro-B–OVA cell injection in mice treated with the indicated drug combination (n = 4 mice per group). Representative of two independent experiments. Data are expressed as mean ± SEM. One-way ANOVA was used for statistical analysis [(A), (B), (E), and (H)]. **P < 0.01, and ***P < 0.001.

To translate our findings in vivo, mice bearing pro-B cell tumors were treated with the bivalent SM birinapant alone or in combination with IFN-β and/or emricasan (Fig. 6D). Treatment with SM alone had no detectable impact on tumor load (Fig. 6E). Tumor control was somewhat improved by combining SM with either IFN-β or emricasan, but only the tritherapy substantially reduced tumor burden in different anatomical sites, including blood, lymph nodes, and BM (Fig. 6E). As shown in fig. S11, the tritherapy appeared to spare nonmalignant immune cells, suggesting that malignant B cells are specifically sensitive to the cell death reprogramming tritherapy.

To assess the involvement of the immune system in the tumor control elicited by the tritherapy, we compared the survival of treated tumor-bearing wild-type (WT) or Rag2^−/− hosts (Fig. 6F). We found that the treatment cured two-thirds of the WT mice but none of the Rag2^−/− mice, revealing a major role for the adaptive immune system in the therapeutic efficacy. The tritherapy induced an accumulation of activated T cells prone to produced IFN-γ and tumor necrosis factor–α (TNF-α) (fig. S12, A to D). Moreover, mice cured by the tritherapy were protected against a secondary challenge with the same tumor cell line, indicative of long-term immune memory (fig. S12E).

To further analyze T cell responses induced by the treatment, we used pro-B cell tumors expressing the ovalbumin (OVA) model antigen and analyzed antigen-specific CD8⁺ T cells using major histocompatibility complex peptide tetramers. As shown in Fig. 6 (G and H), the highest levels of tumor-specific T cells were achieved by the tritherapy as compared to any other drug combinations tested. Thus, the cell death reprogramming strategy not only induced robust tumor cell death but also elicited a potent immune response, essential for tumor clearance. In sum, our results highlight that RIPK3 is a central node for B cell tumor death and establish a therapeutic strategy relying on the combined action of RIPK3 activation, type I IFN, and caspase inhibition to optimize tumor cell death and antitumor immune responses.

DISCUSSION

Here, we explored how RIPK3 signaling in B cell tumors could be manipulated to promote potent antitumor immunity. Contrasting with what has been reported with other cell types (17, 18, 21), forced RIPK3 activation triggered B tumor cell death almost exclusively via apoptosis. Reprograming RIPK3 activity toward necroptosis required caspase inhibition but remained relatively inefficient due to the poor expression of MLKL in B cells. The addition of type I IFN substantially improved MLKL expression and necroptosis induction. On the basis of these findings, we identify an optimal combination therapy aimed at activating RIPK3, blocking caspase and providing type I IFN that optimizes antitumor immune responses and tumor regression in vivo.

Dissecting the role of specific cell death pathways is challenging but can be facilitated by activatable molecule such as FKBP fusion proteins. Activatable forms of RIPK3, caspase-8, or Fas-associated death domain protein (FADD) have been generated to study the consequences of immunization with cells induced to die in vitro (6, 17). We show here that it is possible to extend these approaches in vivo and effectively induce RIPK3 activation in established tumors, with precise temporal control. The use of intravital microscopy offered the possibility to track the fate of tumor cells undergoing RIPK3 activity at the tumor site in real time. These experiments formally established that RIPK3 activation induces robust apoptosis in B cell tumors followed by rapid macrophage engulfment. In settings where caspase-8 is absent, tumor cells died by necroptosis, and cells were also rapidly engulfed by tumor-associated macrophages. This is consistent with the fact that necroptotic cells expose phosphatidylserine after MLKL membrane translocation (34). It has been shown that necroptotic cells continue to produce inflammatory cytokines several hours after membrane permeabilization, a feature that contributes to their inflammatory potential (19). Our in vivo imaging revealed that cells undergoing necroptosis may not persist as isolated necroptotic corpses for very long, especially in macrophage-rich environments (such as tumors). Therefore, inflammation associated with cell death may depend not only on the type of cell death but also on the speed at which cellular corpses are cleared by macrophages.

Our results suggest that strategies aimed at promoting necroptosis in solid tumors (5, 18) need to be adjusted for B cell malignancies in which apoptosis is the prevalent pathway activated by RIPK3. Low expression of MLKL in B-lineage cells combined with robust expression of inhibitors of necroptosis (i.e., caspase-8) likely restrain necroptotic cell death in B cell malignancies and could explain the synergistic effects mediated by caspase inhibition in the presence of type I IFN. Low MLKL expression in B cells could be reversed by IFNs, hypothetically through modification of chromatin accessibility as observed at other interferon stimulated gene (ISG) loci (35) and increased transcription (fig. S6) (30).

Treatment of B cell lymphoma and multiple myeloma with type I IFNs has provided therapeutic benefit in preclinical models and in patients. A response rate of 15% has been reported in multiple myeloma patients treated with type I IFNs as a monotherapy, yet the underlying mechanisms remained controversial (36). Our results may provide a mechanistic basis for the beneficial effects provided by type I IFNs since type I IFNs increased the magnitude of cell death by necroptosis and the capacity of the host to mount an efficient immune response. It is possible that the tritherapy acts on immune cells in addition to malignant cells. We did not observe notable immune cell death during the tritherapy, possibly due to the relative selectivity of SM toward malignant over healthy cells as malignant cells tend to express high levels of IAPs and/or display dysregulated cell death pathway and increased dependence on prosurvival signals. Nevertheless, we cannot exclude that other effects are induced on host cells. In particular, the presence of IFNs may both promote tumor cell death and induce myeloid cell activation with potential additive or synergistic effects.

Our results demonstrate that the cytotoxic activity of SM in B cell tumors is dependent on RIPK3, possibly because the targeting of IAP proteins can stabilize RIPK1 and promote RIPK1/RIPK3 interactions and RIPK3 activation (10). The development of SM represents an attractive approach to engaging tumor cell death pathways, but their use as monotherapy has yielded limited results, highlighting the need for combined treatments (37). Previous in vitro studies have reported the benefit of combining SM with type II IFN (38) or TNF-α (3) to promote necroptosis or with IFN-α to enhance apoptosis in renal cell carcinoma (39). In models of acute myeloid leukemia (AML), SM synergized with caspase inhibition (40). In light of these important studies on SM, the present work provides several key findings. We found that SM and caspase inhibition are not sufficient to treat B cell malignancies in contrast to what was shown with AML. This is largely due to an intrinsic resistance to necroptosis in B cells uncovered here and that could be overcome by the addition of type I IFN. Notably, previous studies on SM have not explored the role of the immune system, largely focusing on in vitro experiments or immunodeficient mouse models.

In the context of B cell tumors, both type I IFN and caspase inhibition were critical to uncover the therapeutic benefit of SM. Such a cell death reprogramming therapy targeting RIPK3 may be particularly suited for patients primarily refractory to or relapsing after standard chemotherapy regimen as these treatments are mostly RIPK1 independent (3).

Guided by the findings made with the activatable RIPK3, we established that a combined therapy including type I IFNs, the SM birinapant and the caspase inhibitor emricasan (all of which have been individually subject to clinical evaluation) promote effective control of B cell tumors in vivo encouraging further clinical investigations. This therapy elicited a strong T cell responses and promoted clearance of B cell tumors in most treated mice. The fact that this therapy was not effective in Rag2^−/− mice highlights the importance of the immune response to eliminate tumor cells that were not directly killed by the drug combination. The induction of effective antitumor immunity by tumor cell undergoing necroptosis may originate from the release of danger molecules during membrane permeabilization and/or on the RIPK3-dependent production of inflammatory cytokine/chemokine. Future studies may help clarify the role of macrophage and dendritic cell subsets in inducing antitumor immunity during cell death reprogramming therapy. As illustrated here, strategies that optimize the type and magnitude of cancer cell death while boosting the associated immune response represent an attractive avenue to treat B cell malignancies.

MATERIALS AND METHODS

Mice

Male CD45.2 C57BL/6J (B6) mice were purchased from Envigo. Male Rag2^−/− B6 mice were bred in our animal facility under specific pathogen–free conditions. All mice used were 8 to 12 weeks old. All animal studies were approved by the Institut Pasteur Safety Committee in accordance with French and European guidelines (CETEA 2017-0038).

Cells and constructs

An immortalized pro-B cell line was established by transducing BM cells from WT CD45.1 B6 mice with a retrovirus encoding viral Abelson kinase (41). A lymphoma B cell line was isolated from male Eμ-myc mice that spontaneously develop Burkitt-like lymphomas (42). These cell lines were retrovirally transduced with a prodeath construct consisting in an activatable version of a full-length murine RIPK3 chimeric protein based on a previously described model (21) (actRIPK3-pro-B and actRIPK3-Eμ-Myc cells). Briefly, murine RIPK3 sequence was cloned upstream of two copies of FKBP carrying the F36V mutation. This chimeric protein was cloned into a pMSCVpuro retroviral vector containing T2A ribosome–skipping sequences upstream of mCherry fluorescent protein. Transduced cells were selected for 1 week in puromycin (2 μg/ml) and then sorted on the basis of mCherry expression. Pro-B and Eμ-Myc cell lines expressing activatable RIPK3 construct were also retrovirally transduced to express a FRET-based reporter for caspase-3/7 activity [CFP^(DEVD)YFP probe], named DEVD (22). Pro-B cells were also retrovirally transduced to express chicken OVA and sorted to generate a pro-B–OVA cell line. Cells were cultured in RPMI 1640 medium–GlutaMAX (Gibco) supplemented with 10% heat-inactivated fetal bovine serum, penicillin (50 U ml⁻¹), streptomycin (50 μg ml⁻¹), 1 mM sodium pyruvate, 10 mM Hepes, and 50 μM 2-mercaptoethanol (Gibco) and maintained at 37°C and 5% CO₂. The NIH-3T3 fibroblast cell line, expressing activatable RIPK3 construct, has been described previously (17). NIH-3T3 and MEFs were cultured in Dulbecco’s modified Eagle’s medium (Gibco) supplemented as described for RPMI. Human B cell lines: SUDHL4 (DLBCL), Ramos, and Raji (Burkitt lymphoma), purchased at American Type Culture Collection, were cultured in supplemented RPMI as described above. All cell lines were routinely tested and determined to be free of mycoplasma (Venor-Gem Advance Mycoplasma Detection kit, Minerva Biolabs).

CRISPR-Cas9 gene editing

The pro-B cell line lacking endogenous RIPK3, caspase-8, or MLKL was generated by genome editing following direct delivery of CRISPR-Cas9 system as a ribonucleoprotein (RNP) complex (Integrated DNA Technologies). The sequences of crRNA used to target RIPK3, caspase-8, or MLKL were the following: GCGGAGGGTTCAAGCTGTGT, GTGGGATGTAGTCCAAGCAC, and GCACACGGTTTCCTAGACGC, respectively. Briefly, RIPK3 or caspase-8 or MLKL-specific RNP complexes were generated by annealing an equimolar concentration of each specific oligo crRNA with the common tracrRNA using a slow slope reaching 23°C followed by incubation at room temperature during 10 min with S.p. HiFi Cas9 Nuclease V3. One million pro-B cells were resuspended in SF nucleofection solution (Lonza) with the corresponding RNP complex, transferred to nucleofection cuvette strips, and electroporated using the DN-100 program (4D-Nucleofector X Unit, Lonza). Transfected cells were further cultured for 48 to 72 hours in complete RPMI 1640 medium at 32°C in 5% CO₂ to force genome editing (43) before resuspension in supplemented fresh medium. Effective RIPK3, caspase-8, and MLKL deletion was confirmed by Western blotting.

Cell death stimulation in vitro

ActRIPK3–pro-B and actRIPK3–Eμ-Myc cells, stably transduced with an activatable RIPK3-based prodeath construct, were incubated in complete RPMI 1640 medium containing different concentrations of the chemical reagent B/B dimerizer (Takara) from 5 to 500 nM during 1, 3, or 6 hours. Dimerization of the FKBP fusion proteins in the presence of B/B dimerizer forces RIPK3 oligomerization, activation, and cell death induction. RIPK3-induced cell death was also examined in cells pretreated for 30 min with the pan-caspase inhibitor zVAD-fmk (50 μM, R&D Systems) and later cultured in the presence of the B/B dimerizer (500 nM). In some conditions, cells were pretreated for 24 hours with recombinant mouse IFN-β (2.5 ng/ml; BioLegend) before being cultured with the B/B dimerizer in the presence or absence of zVAD-fmk and RIPK3 kinase inhibitor GSK′843 (5 μM; MedChemExpress). Pro-B cells expressing or lacking endogenous RIPK3 (RIPK3^−/− pro-B) were treated with BV6 (5 μM; HY-16701, CliniSciences) to favor apoptosis induction or BV6 (5 μM) plus emricasan (5 μM; HY-10396, CliniSciences) for necroptosis induction during 16 hours. To examine the effect of IFN-β in the magnitude of the induced cell deaths, pro-B or RIPK3^−/− pro-B cells were cultured with IFN-β (2.5 ng/ml) for 24 hours before the addition of BV6 or BV6 plus emricasan. SUDHL4, Ramos, and Raji cells were cultured with recombinant human IFN-β (1 ng/ml; Peprotech) for 24 hours before adding BV6 (5 μM) plus emricasan (5 μM) during 16 hours.

Cell death stimulation in vivo

Male B6 mice were injected intravenously with 3 × 10⁶ of DEVD–actRIPK3–pro-B or DEVD–actRIPK3–Eμ-Myc cells. Tumors were first established in the BM and later disseminated in the blood, spleen, and peripheral lymph nodes. Five days after tumor injection, when circulating tumor cells were detected, tumor-bearing mice received a single dose of B/B dimerizer intravenously following the manufacturer’s recommendations (Takara). The stock solution of B/B dimerizer was prepared at 62.5 mg/ml in ethanol. Mice received the B/B dimerizer (10 mg/kg) prepared in an injection solution consisting of 4% of ethanol, 10% polyethylene glycol, molecular weight 400, and 1.7% Tween in water. All injections were administered to mice within 30 min of dilution into the injection solution. Control vehicle mice received the same solution without the B/B dimerizer. Tumor burden analyses were performed at 3, 24 hours, and at day 3 after treatment in the blood and BM. When indicated, male B6 mice were injected intravenously with 3 × 10⁶ of DEVD–actRIPK3–pro-B lacking caspase-8. The number of tumor cells was measured in 60 μl of blood, and the absolute number was calculated using counting beads assuming a total blood volume of 2 ml. Tibia BM and spleen tumor count was measured in the whole organs, and absolute count was calculated using counting beads. Mice were examined and killed in case of prostration, weakness, tousled hair, or a weight loss of >10%. To assess the antileukemic effect of SMAC mimetic (SM), WT or Rag2^−/− B6 mice bearing pro-B tumors were treated with birinapant (10 mg/kg; CliniSciences) in association with emricasan (5 mg/kg; CliniSciences) and 2 μg of IFN-β, all already used as a single agent in clinical trials. Mice received three doses of IFN-β or phosphate-buffered saline intravenously every 2 days starting at day 2 after tumor injection. Three doses of birinapant with or without emricasan were administered intraperitoneally every 2 days starting at day 3 after tumor inoculation. Tumor burden analyses were performed at day 14 after tumor injection, and survival was monitored for up to 2 months. The induction of tumor-specific T cell responses was measured by K^b-OVA tetramer staining at day 14 after inoculation of pro-B–OVA cells.

Multiplex protein assay for measurement of chemokine released by dying cells

DEVD–actRIPK3–Eμ-Myc cells were pretreated for 24 hours with recombinant mouse IFN-β (2.5 ng/ml). Cells were later plated and cultured with the B/B dimerizer (500 nM) in the presence or absence of zVAD-fmk (50 μM). Culture supernatants were collected at 3, 6, and 9 hours after adding the B/B dimerizer. Supernatants were processed with the Cytokine Mouse Magnetic 21-Plex ProcartPlex Panel (Invitrogen) following the manufacturer’s protocol. Data analysis was conducted in a Bio-Plex 200 instrument with Bio-Plex Manager software (Bio-Rad). Data values were scaled to 0 to 1 range using the following formula: (value − V_minimum)/(V_maximum − V_minimum).

BMDM cocultures with dying tumor B cells

For phagocytosis assay, BMDMs were generated by macrophage colony-stimulating factor differentiation during 7 to 10 days. A total of 5 × 10⁵ BMDMs were seeded in a 24-well plate. Then, IFN-β pretreated DEVD–actRIPK3–pro-B cells expressing or lacking caspase-8 were added in each well (1:1 ratio). Tumor B cells were induced to die by adding the B/B dimerizer (500 nM). The percentage of phagocytosis was measured as the frequency of mCherry⁺ cells in the F4/80⁺ gate.

Cell preparation and flow cytometry

BM single-cell suspensions were obtained from tibias, spleen, and blood of tumor-bearing mice and controls. For all staining protocols, cell suspensions were incubated with FcR-blocking reagent before cell surface staining (anti-CD16/32 monoclonal antibody, clone 93, BioLegend). Stainings were performed with the following antibodies: CD19–Alexa Fluor 647 (clone 1D3, BioLegend), CD45.1-BV421 (clone A20, BioLegend), CD45.2–fluorescein isothiocyanate (clone 104, BioLegend), TCRβ-PC7 (clone H57-597, BioLegend), Ly6C-BV785 (clone HK1.4, BioLegend), Ly6G-BUV395 (clone 1A8, BD Biosciences), Ly6G-BV650 (clone 1A8, BD Biosciences), CD8-BV785 (clone 53-6.7, BioLegend), CD44-BV421 (clone IM7, BioLegend), CD80–phycoerythrin (PE; clone 16-10A1, BioLegend), CD86-BUV661 (clone GL1, BD Biosciences), I-A/I-E–BUV496 (clone 2G9, BD Biosciences), F4/80-PE/cyanine7 (clone BM8, BioLegend), CD90.2-BV650 (clone 30-H12, BD Biosciences), NK1.1-BV650 (clone PK136, BD Biosciences), CD11b-BUV395 (clone M1/70, BD Biosciences), and CD64-APC (clone X54-5/7.1, BioLegend). For identification of membrane permeabilized dead cells, Live Dead fixable viability dye eFluor780 (eBioscience) was used. For tetramer staining, cell suspensions were incubated for 1 hour at 4°C with APC-H2-K^b-OVA_257–264 and PE-H2-K^b-OVA_257–264 tetramer (provided by National Institutes of Health Tetramer Core Facility). For intracellular staining, viable, dying, or dead tumor cells stained with the Live Dead dye were fixed and permeabilized using the Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer’s recommendations and stained with Alexa Fluor647–anti–cleaved caspase-3 monoclonal antibody (clone C92-605, BD Biosciences). For intracellular cytokine staining, cells were restimulated ex vivo with phorbol 12-myristate 13-acetate (20 ng/ml; Sigma-Aldrich) and ionomycin (0.5 μg/ml; Sigma-Aldrich) for 4 hours in the presence of brefeldin A (1 μg/ml; BD Biosciences) and stained with IFN-γ–BUV737 (clone XMG1.2, BD Biosciences) and TNF-α–Alexa Fluor 488 (clone MP6-XT22). Analyses were performed with an LSR Fortessa II (BD Biosciences) or a CytoFLEX LX (Beckman Coulter) cytometer and analyzed using FlowJo software version 10.8.1 (BD Biosciences).

Immunoblot analysis

Cells were lysed in radioimmunoprecipitation assay extraction buffer (Thermo Fisher Scientific) supplemented with a protease inhibitor cocktail (Thermo Fisher Scientific) at a concentration of 1 × 10⁶ cells in 30 μl of buffer for 20 min on ice. Lysates were obtained by centrifugation at maximum velocity for 15 min at 4°C. The pellet was discarded, and the supernatant was kept for further analysis. Protein concentration was measured using a BCA protein assay (Thermo Fisher Scientific). Equivalent micrograms of proteins were resuspended in 4× Laemmli sample buffer (Bio-Rad) and separated using 4 to 15% Mini-Protean TGX Stain-Free gels (Bio-Rad), under reducing conditions. Proteins were then transferred on polyvinylidene difluoride membranes (Bio-Rad) using a semidry system (Transblot Turbo Transfer System). After blocking with tris-buffered saline 1% Western-blotting blocking reagent (Merck), membranes were incubated overnight at 4°C with primary antibodies. After washing, membranes were incubated with a horseradish peroxidase (HRP)–conjugated appropriate secondary antibody. Detection of bound antibodies was revealed using the Clarity Western ECL substrate (Bio-Rad), and images were acquired with ImageQuant LAS 4000 Mini (GE Healthcare). Quantifications were done using Image Lab software (Bio-Rad). Primary antibodies for Western blot were anti-mouse: RIPK3 (polyclonal, Novus Biologicals), RIPK1 (clone 38/RIP, BD Biosciences), caspase-8 (clone 4927, Cell Signaling), cleaved caspase-8 (9429, Cell Signaling), caspase-3 (9662, Cell Signaling), FADD (clone 10L6M0, Thermo Fisher Scientific), MLKL (clone 3H1, Merck), phospho-MLKL (D6E3G, Cell Signaling), and β-actin (clone 13E5, Cell Signaling). Secondary antibodies including HRP-conjugated goat anti-rabbit immunoglobulin G (IgG) (H + L), HRP-conjugated goat anti-mouse IgG (H + L), and HRP-conjugated goat anti-rat IgG (H + L) were all from the Jackson Laboratory.

Bulk RNA-seq analysis of available mouse and human data

MLKL expression data from sorted primary murine cells were obtained from the Immunological Genome Project (ImmGen, https://immgen.org) in the murine RNA-seq section (http://rstats.immgen.org/Skyline/skyline.html). MLKL expression data from human cell lines were obtained from Cancer Cell Line Encyclopedia (https://sites.broadinstitute.org/ccle). The expression of MLKL is depicted as normalized counts.

Bulk RNA-seq data from IFN-γ–stimulated Eμ-Myc cells were obtained from GSE140191 (44). The data were processed and analyzed using the standard DESeq2 pipeline (45). Differential expression analysis comparing IFN-γ–stimulated and untreated Eμ-Myc cells was performed using DESeq2. Genes with a log₂ fold change of >1 and an adjusted P value of <0.05 were considered significantly differentially expressed.

Single-cell RNA-seq of human BM mononuclear cells

Raw counts from two adult frozen BM mononuclear cell single-cell RNA-seq datasets were downloaded from the following 10x genomics website: https://10xgenomics.com/resources/datasets/frozen-bmm-cs-healthy-control-1-1-standard-1-1-0 and https://10xgenomics.com/resources/datasets/frozen-bmm-cs-healthy-control-2-1-standard-1-1-0. Raw counts from seven B cell acute lymphoblastic leukemia (B-ALL) patients were downloaded from Gene Expression Omnibus (GEO) (GSE134759) (26). Raw counts from four patients with DLBCL were downloaded from GEO (GSE182436) (27).

The count matrix from each dataset was fed into Seurat (46) and analyzed independently with the standard pipeline. “NormalizeData” function with default parameters was applied to normalize the expression level of genes in each single cell. The “ScaleData” function was used to scale and center gene expression matrices after regressing out heterogeneity associated with mitochondrial contamination. The two datasets were merged using canonical correlation analysis. To perform clustering, the dimensionality of the data was determined by calculating relevant principal components using the ElbowPlot function. The t-SNE representation was used to visualize the single-cell transcriptional profile in two-dimensional (2D) space. Annotation of the clusters was performed using marker genes. The expression of MLKL, in given subsets, is depicted using Violin Plots showing ALRA-imputed expression values.

ATACseq analysis of mouse primary cells

ATACseq analysis of MLKL locus was performed on public datasets (29). Bigwig files were downloaded from GEO (GSE 100738), and ATACseq profiles were visualized using Integrative Genomics Viewer (47).

In vitro confocal imaging

In vitro live imaging was performed using an inverted microscope (Nikon.Ti2E) equipped with a Yokagawa CSU-W1 spinning disk and 40×/1.15 water objective. Imaging was performed in the presence of DAPI at 37°C and 5% CO_2.

Intravital two-photon imaging

Mice bearing tumors were subjected to BM two-photon intravital imaging before and after injection with the B/B dimerizer, as previously described (48). Briefly, mice were anesthetized with a mixture of xylazine (Rompun; 10 mg/kg) and ketamine (Imalgene; 100 mg/kg). The scalp was removed, the skin was incised to expose the bone, and a coverslip was fixed above the frontoparietal suture. During imaging, mice were supplied with oxygen, and temperature was maintained at 37°C using a heated pad. Two-photon imaging was performed using an upright microscope (FVMPE-RS, Olympus), a 25×/1.05 numerical aperture water dipping objective equipped with an objective heater and using FV31S-SW software (Olympus). An Insight DeepSee dual laser (Spectra-Physics) tuned at 820 nm was used for excitation. Macrophages were visualized in vivo by intravenous injection of an Alexa Fluor 594–conjugated anti-F4/80 monoclonal antibody (BioLegend). The following filters were used: cyan fluorescent protein (CFP; 483/32), YFP (542/27), and Alexa Fluor 594 (624/40). Time-lapse sequences were created by scanning a 30- to 50-μm-thick tissue volume using 5-μm Z-steps and 60-s intervals. Videos were processed and analyzed using Fiji software (ImageJ 2.3.0). Figures and videos based on two-photon microscopy are 2D maximum intensity projections of 3D datasets.

Statistical analyses

For each experiment, the number of independent experiments, replicates, and statistical tests used is indicated in the figure legends. All statistical tests were performed using Prism v.9.2.0 (GraphPad). Data are expressed as mean ± SEM. Unpaired Student’s t test, one-way analysis of variance (ANOVA), and Wilcoxon rank sum test were used as indicated in individual figure legends. Significance was defined by a P < 0.05. P values were reported as stars: *P < 0.05, **P < 0.01, and ***P < 0.005; ns is nonsignificant.

Supplementary Materials

The PDF file includes:

Figs. S1 to S12

Legends for movies S1 to S3

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S3