BCL2 Inhibition Reveals a Dendritic Cell–Specific Immune Checkpoint That Controls Tumor Immunosurveillance

Oncology of the 21st century is marked by the ever-expanding number of targeted agents that act on specific oncoproteins or molecules operating downstream of such oncogene products. Such targeted agents exploit the “addiction” of cancer cells to such pathways and are generally conceived to act in a cell-autonomous fashion (1). However, the most successful innovation in clinical oncology has been the development of immune-checkpoint inhibitors (ICI) targeting CTLA-4 or the PD-1/PD-L1 interaction across multiple different cancer types (2). The unprecedented success of these treatments demonstrates the possibility to unleash the immune system for cancer treatment.

Driven by this consideration, multiple groups have developed genome-wide genetic screening methods to identify novel druggable immune checkpoints. Such screens can be designed to identify genes, the knockdown or knockout of which influences immunosurveillance. Such screens can be performed on cancer cells to identify genes that confer resistance to T cell–mediated killing (3, 4), that upregulate MHC class I molecules without upregulating PD-L1 (5), or that downregulate immunosuppressive CD47 (6). Alternatively, such screens can be performed on immune cells, in particular on T lymphocytes, to identify genes for which silencing enhances in vitro T-cell proliferation (7) and improves cancer cell killing in vitro (8), stimulates the production of specific cytokines (7, 9), augments T-cell infiltration of, and proliferation within, tumors (10), favors T-cell persistence in tumors (11), or prevents T-cell dysfunction in an immunosuppressive environment (12).

In contrast to T cells, which are relatively abundant and can be driven into proliferation, dendritic cells (DC) are relatively low-abundant and terminally differentiated, rendering genome-wide screens on primary DCs impractical. This collides with the cardinal importance that DCs play in the ignition of anticancer immune responses executed by T lymphocytes. For example, Batf3^−/− mice lacking a specific DC subset, the type 1 conventional DCs (cDC1), are unable to mount T-cell responses against viral infections (13). More­over, cancers evolving in Batf3^−/− mice do not respond to PD-1 blockade (14, 15) or other types of immunotherapy (16–18).

Here, we report the results of a genome-wide CRISPR/Cas9-based screen that involves two steps: the genetic mani­pulation of immortal (and hence infinitely expandable) DC precursors and the deimmortalization and differentiation of these cells to generate DCs that can be characterized as in vitro and in vivo. We demonstrate that BCL2, an oncoprotein that is clinically targeted for the treatment of specific hematopoietic cancers, acts as a checkpoint to restrain the function of DCs with respect to tumor immunosurveillance. Thus, BCL2 inhibition has cDC1-dependent, T lymphocyte-mediated antineoplastic effects against solid cancers that do not respond to BCL2 inhibitors in an immunodeficient context.

Recently, we developed a protocol for CRISPR/Cas9-mediated gene knockout in conditionally immortalized immature DCs, which can be limitlessly expanded before their differentiation/maturation (19). For this, we used an immature DC cell line from C57BL/6 mice, in which the SV40 large T-cell antigen (SV40LgT) is expressed under the control of a Tet-on (doxycycline-inducible) promoter and the reverse tetracycline transactivator is fused to the ligand-binding domain of a mutated glucocorticoid receptor (GR). Due to the dual blockade of retinoblastoma (RB) and tumor protein 53 (TP53) by SV40LgT, these cells are in an inducible/immortalized state[inducible immortalized DCs (iniDC)] in the presence of doxycycline and the GR agonist dexamethasone, yet are deinduced/deimmortalized by the simultaneous removal of both factors (de-iniDC; ref. 20). In contrast to iniDCs, such de-iniDCs pinocytose extracellular proteins that reach their cytosol, meaning that (like cDC1 cells; ref. 21), they become susceptible to apoptosis induction by addition of cytochrome c (Cyt c) to the culture media (Fig. 1A and B) and can present ovalbumin (OVA) protein to B3Z hybridoma cells expressing a transgenic T-cell receptor recognizing the OVA-derived peptide SIINFEKL bound to H2-K^b MHC class I molecules (Fig. 1C and D).

IniDCs that were equipped with Cas9 were expanded by culture with DEX/DOX and then infected by a lentiviral library encoding bar-coded single-guided RNAs (sgRNA) that cover the entire mouse exome. The pool of sgRNA-expressing iniDCs then were differentiated into de-iniDCs, pulsed with OVA protein, and subjected to several rounds of positive selection to purify those DCs that express high levels of the costimulatory ligand CD80 as well as SIINFEKL bound to H2-K^b. This selection was possible due to immunostaining with antibodies recognizing CD80 and SIINFEKL/H2-K^b complexes, followed by cytofluorometric sorting of the double-positive cells. The sequences encoding the sgRNAs were amplified by PCR from the nonselected cell pool and sorted cells. The abundance of sgRNAs was quantified by next-generation sequencing followed by data analysis using the MAGeCK-RRA program (22) for the identification of guidance RNAs (gRNA) associated with improved antigen presentation (Fig. 1E). A total of 715 genes (covering 3.6% of the genome-wide library) were significantly enriched in the sorted pool of cells that yielded gain-of-function phenotypes (Fig. 1F; Supplementary Table S1). Of note, a significant fraction of the genes targeted by these gRNAs were connected to 7 cellular processes that are potentially druggable for cancer treatment such as apoptosis, autophagy, cell cycle, as well as the NF-κB, tumor necrosis factor, and Toll-like receptor signaling pathways (Fig. 1G).

In the next step, we manually selected 620 gRNAs falling into these 7 categories (Supplementary Table S2). These gRNAs were individually transfected into Cas9-expressing iniDCs, followed by deinduction/deimmortalization, OVA loading of de-iniDCs, confrontation of the washed de-iniDCs with B3Z T-cell hybridomas specific for H2-K^b–bound SIINFEKL, and final measurement of IL2 production by B3Z cells (Fig. 2A). This screen yielded 34 gRNAs that significantly improved antigen presentation by de-iniDCs, among which we identified BCL2 as a prominent hit (Fig. 2B; Supplementary Table S2). We subsequently generated multiple independent de-iniDC clones lacking BCL2, all of which exhibited a gain-of-function phenotype with respect to the presentation of OVA-derived SIINFEKL to B3Z cells (Fig. 2C and D). Consistently, several KEGG pathways that cover the top hits from the genome-wide screen (Fig. 1G) matched with the 34 targets that appear in the arrayed screen (Supplementary Fig. S1A). These pathways included several apoptosis-relevant genes including those coding for BCL2 itself, the close BCL2 homolog BCLXL (Bcl2l1), BCL2 binding C-component 3 (Bbc3), three caspases (Casp2, Casp3, and Casp6) and the catalytic subunit of calpain-1 (Capn1).

Altogether, the aforementioned results identify several apoptosis-relevant genes as candidate immune checkpoints acting at the level of DCs. We decided to focus on BCL2, because it is the sole target for which an FDA/EMA-approved drug is available for the treatment of specific blood cancers (in particular, acute myeloid leukemia, chronic lymphocytic leukemia, and small lymphocytic leukemia; ref. 23).

Addition of pharmacologic inhibitors of BCL2, such as ABT737, navitoclax, and venetoclax (but not that of agents inhibiting other members of the BCL2 family such as A1331852, S63845, and WEHI539) enhanced antigen presentation by wild-type (WT) de-iniDCs (Fig. 2E) and bone marrow–derived DCs (BMDC; Fig. 2F). BCL2 inhibitors can induce both apoptosis (which depends on BAX but not ATG7) and autophagy (which depends on ATG7 but not BAX; refs. 24, 25). Accordingly, navitoclax and venetoclax induced caspase activation in WT and autophagy-deficient Atg7^−/− de-iniDCs (and less so in Bcl2^−/− and Bax^−/− de-iniDCs), as well as the autophagy-associated LC3B lipidation giving rise to electrophoretically mobile LC3B-II in WT, Bcl2^−/− and Bax^−/− (but not Atg7^−/−) de-iniDCs (Supplementary Fig. S1B and S1C). However, treatment of WT de-iniDCs with 10 μmol/L venetoclax killed only a fraction of the cells after a 24-hour treatment (∼12%). No extra cell death was induced by venetoclax in Bcl2^−/− de-iniDCs. Moreover, there was no major increase in spontaneous cell death events in Bcl2^−/− de-iniDCs compared with WT de-iniDCs (Fig. 2G).

Navitoclax and venetoclax enhanced antigen presentation by WT de-iniDCs, but not by constitutively overactive Bcl2^−/− de-iniDCs, in line with the idea that they indeed act on target. Both BCL2 inhibitors failed to enhance antigen presentation by autophagy-deficient (Atg5^−/− or Atg7^−/−) de-iniDCs, but continued to act on apoptosis-deficient (Bax^−/−) de-iniDCs, suggesting that they stimulate DC function through autophagy-related processes (but not apoptosis; Supplementary Fig. S1D and S1E), in line with prior reports that Atg5^−/− or Atg7^−/− DCs are deficient in antigen presentation (26, 27).

Bulk RNA sequencing (RNA-seq) revealed that the transcriptome of de-iniDCs globally resembles that of conventional DCs (cDC), in particular cDC1 cells (Fig. 2H). In addition, the knockout of Bcl2 caused de-iniDCs to acquire the signature of migratory cDC1s (which are CCR7⁺). Bcl2 KO and the clinically used BCL2 inhibitor venetoclax upregulated a partially overlapping set of mRNAs in de-iniDCs (Fig. 2I) that represented multiple genes involved in the immune response including the activation of cytokine responses and improved antigen processing and presentation (Fig. 2I and J). Among the 79 genes that were strongly upregulated (by ≥2 fold) in de-iniDCs by both Bcl2 knockout and venetoclax treatment, a large majority (n = 65, 82%) were connected to the type 1 interferon response (Supplementary Table S3). qRT-PCR confirmed that venetoclax treatment of WT cells as well as Bcl2 knockout induced the upregulation of mRNAs encoding type 1 interferons (Ifna1 and Ifnb1), the two subunits of the common type 1 interferon receptors (Ifnar1 and Ifnar2) and typical downstream target of type 1 interferon signaling (Cxcl10, Isg15, and Mx1). Moreover, pharmacologic or genetic inhibition of BCL2 resulted in an increase of Tmem173/Sting mRNA and its downstream effector Irf3 (Supplementary Fig. S2A). Accordingly, after venetoclax treatment or Bcl2 knockout, the abundance of cGAS protein as well as the phosphorylation of STING and IRF3 increased (Supplementary Fig. S2B and S2C). This points to the activation of the STING pathway. Indeed, the knockout of Tmem173/Sting, that of its downstream effector Irf3, as well as the knockout of Ifnar1 or Ifnar2 (which are nonredundant because they form heterodimers) abolished the capacity of venetoclax to stimulate antigen presentation by de-iniDCs (Supplementary Fig. S2D). Accordingly, the blockade of the Ifnar1 protein with a monoclonal antibody blunted the capacity of venetoclax to activate antigen presentation by de-iniDCs (Supplementary Fig. S2E). Moreover, in accord with prior records showing that BCL inhibitors can induce the release of mitochondrial DNA (mtDNA) from cells (28, 29), venetoclax treatment of WT de-iniDCs and Bcl2 KO led to an increase in double-stranded DNA in the extramitochondrial cytoplasm (Fig. S2F and S2G).

Bcl2 KO de-iniDC compared with WT controls also exhibited a higher secretion of the interleukins (IL) IL1β and IL6 (determined by ELISA), as well as higher expression of the chemokine receptors CCR2 and CXCR3, costimulatory receptors (CD80, CD83, and CD86), and activation markers (CD40, MHC class II molecules, PD-L1) than WT controls (Supplementary Fig. S3A). These findings (except the upregulation of PD-L1) could be recapitulated by adding venetoclax to WT de-iniDCs and to a lesser degree by other pharmacologic agents acting on the BCL2 family (Supplementary Fig. S3A). Of note, the neutralization of CD80 and CD86 both reduced antigen presentation of OVA to B3Z cells by navitoclax and venetoclax, suggesting that their upregulation might be functionally relevant (Supplementary Fig. S3B and S3C).

In sum, the aforementioned results support the contention that pharmacologic BCL2 inhibition can act on target to enhance the immunostimulatory potential of DCs. This effect is mediated via the activation of the type-1 interferon pathway and the upregulation of costimulatory receptors.

A prior report published in Cancer Discovery demonstrated that venetoclax increased the T-cell infiltrate of MC38 colorectal cancers subcutaneously (s.c.) implanted in C57BL/6 mice (or that of CT26 mice implanted in BALB/c mice), sensitizing such tumors to PD-1 or PD-L1 blockade (30). We investigated whether such effects would also apply to orthotopic tumors, in particular, MCA205 cutaneous fibrosarcomas and TC1 non–small cell lung cancers (NSCLC), which are not particularly susceptible to in vitro killing by BCL2 inhibitors unless very high concentrations (≥10 μmol/L) are used (Supplementary Fig. S4A and S4B). Mice bearing orthotopic MCA205 fibrosarcomas (under the skin) or TC1 NSCLC (in the lung) were treated with intraperitoneal (i.p.) injection of navitoclax or venetoclax, followed by high-dimensional immune profiling of their T cells (Supplementary Fig. S4C). PD-1 and CTLA-4 expression by CD8⁺ cytotoxic T lymphocytes (CTL) and CD4⁺Foxp3⁺ regulatory T cells (Treg) were consistently upregulated by navitoclax or venetoclax in TC1 (Supplementary Fig. S4D–S4G) or MCA205 (Supplementary Fig. S4H and S4I) bearing mice. Consistently, navitoclax and venetoclax sensitized MCA205 or TC1 tumors to PD-1 blockade, hence accentuating tumor growth reduction that can be conveniently monitored by an intrathoracic luciferase-dependent chemiluminescence signal (for TC1, Supplementary Fig. S5A–S5E) and measuring the size of s.c. tumors (for MCA205; Supplementary Fig. S5F–S5I).

Of note, navitoclax and venetoclax also upregulated the expression of the chemokine receptors CCR7 and XCR1, as well as the maturation marker MHC-II, to a variable degree on cDC1 cells (defined as CD103⁺CD11b⁻ cells within DCs, defined as the viable CD45⁺CD11c⁺F4/80⁻ MHC-II^hi population; Supplementary Fig. S6A) in the blood, tumor infiltrate, and lymph nodes from mice bearing TC1 orthotopic lung cancers (Supplementary Fig. S6B–S6D), without any signs of a reduction in absolute numbers of such DCs in those organs (Supplementary Fig. S6E–S6G). Most of these upregulations were exclusively identified on cDC1 cells but not the (CD103⁻CD11b⁺) cDC2 subpopulation. Similarly, the upregulation of CCR7, XCR1, MHC-II molecules, and CD86 on cDC1 was observed upon BCL2 inhibition with venetoclax in the MCA205 tumor model (Fig. 3A–D). Venetoclax also induced other maturation markers, including CD80 and CD83 on cDC1 (and cDC2 in some cases) cells from the blood, tumor, and both the tumor-draining and nondraining lymph nodes of mice bearing orthotopic MCA205 fibrosarcomas (Supplementary Fig. S6H and S6I). Importantly, venetoclax upregulated IL12p40 and CXCR3 (also known as CD183) specifically on cDC1 cells from tumors or tumor-draining lymph nodes; Supplementary Fig. S6J and S6K).

We validated these findings in five patients with acute myeloid leukemia (AML) that were treated with a combination of venetoclax and azacytidine. After one cycle of therapy (daily oral administration for 1 week), circulating cDC1s (defined as leukocytes lacking lymphocyte markers but expressing high levels of HLA-DR, with low levels of CD1c, CD14, CD88, and CD123, but high expression of CD141; Supplementary Fig. S7A) expressed higher levels of CCR2, CCR7, XCR1, CD5, CD86, and HLA-DQ, as determined by high-dimensional immunofluorescence cytometry (Fig. 3E). These changes were specific to cDC1 (not cDC2) cells. In PBMCs from healthy donors, the treatment with venetoclax but not azacytidine induced the activation of these markers at the surface of cDC1 (Fig. 3F–I; Supplementary Fig. S7B), suggesting that the in vivo effects on circulating cDC1s are indeed mediated by venetoclax. Of note, there was no sign of DC depletion in venetoclax-treated PBMCs (Supplementary Fig. S7C).

Altogether, these results suggest that BCL2 inhibition activates cDC1s both in tumor-bearing mice and in humans.

Because iniDCs can be limitlessly expanded, it is possible to generate large batches of de-iniDCs for their adoptive transfer into mice (19, 31). De-iniDCs labeled with the fluorescent dye PKH26 were i.v. injected into orthotopic TC1 lung cancer-bearing mice. Such cells could be detected in the lymph nodes and lung tissue up to 72 hours postinjection. Neither pretreatment of venetoclax nor Bcl2 knockout reduced the persistence of these cells in vivo (Supplementary Fig S7D–F). Both genetic and pharmacologic inhibition of BCL2 enhanced the expression of MHC-II molecules on such cells (Supplementary Fig S7G). IHC detection of de-iniDCs (which are Cas9⁺) revealed their presence in TC1 lung cancers 72 hours after injection (Supplementary Fig S7H). Importantly, Bcl2^−/− de-iniDCs were more abundant in the tumors than their WT counterparts (Supplementary Fig S7I). Cytofluorimetric analyses of the lungs and lymph nodes of such mice revealed that the injection of Bcl2^−/− de-iniDCs or venetoclax-treated WT de-iniDCs (but not that of untreated WT-de-iniDCs) increased the local presence of CD8⁺ (including granzyme B⁺) T cells (Supplementary Fig S8A and S8B). As compared with WT controls, i.v. injected Bcl2^−/− de-iniDCs were more efficient in controlling the growth of orthotopic TC1 NSCLC (Fig. 4A–D) and MCA205 fibrosarcomas (Supplementary Fig. S8C–S8F), indicating that BCL2 inhibition in DCs alone is sufficient to improve immunosurveillance. WT de-iniDCs treated with navitoclax or venetoclax in vitro also became more efficient in controlling NSCLC upon their adoptive transfer than vehicle-treated WT de-iniDCs (Fig. 4E and F).

The tumor control mediated by Bcl2^−/− de-iniDCs was further enhanced by subsequent PD-1 blockade, as shown for both TC1 (Fig. 4A–D) and MCA205 cancers (Supplementary Fig. S8C–S8F), but was lost upon depletion of T cells by means of neutralizing antibodies specific to CD4 and CD8 (Fig. 4G and H; Supplementary Fig. S8G), as well as in nude mice lacking mature T cells due to the Foxn1^nu/nu mutation (Supplementary Fig. S8H). Antibody-mediated blockade of IFNAR1 on venetoclax-treated DCs or Bcl2^−/− DCs reduced their capacity to control the growth of tumors and to extend animal survival (Supplementary Fig. S8I and S8J). Finally, the growth control of M/D-driven mammary carcinomas by radiotherapy could be enhanced by venetoclax, and this improvement was abolished upon neutralization of IFNAR1 (Supplementary Fig. S8K).

These results confirm that Bcl2^−/− DCs exhibit a gain-of-function phenotype that improves T cell–mediated cancer immunosurveillance and that depends on type 1 interferon signaling.

To prove the importance of DCs for the anticancer effects of pharmacologic BCL2 inhibitors, we used several strategies. First, we blocked the extravasation of myeloid cells (including DC) by means of a CD11b-blocking antibody (32, 33) and found that this manipulation depleted the cDC2 population in blood, tumors, and tumor-draining lymph nodes (Supplementary Fig. S9A), but did not change the number of cDC1 cells (which are CD11b⁻ as illustrated in the gating strategy in Supplementary Fig. S6; ref. 16) in the blood, but prevented cDC1 cells and in particular migratory XCR1⁺ cDC1 cells from entering the tumor bed and (partially) the tumor-draining lymph nodes (Supplementary Fig. S9B and S9C). In addition, CD11b blockade as well as T-cell depletion compromised the tumor growth–reducing effects (Fig. 5A–C) and animal survival-enhancing of navitoclax and venetoclax against TC1 lung cancers (Fig. 5D). This finding could be recapitulated for MCA205 fibrosarcomas, which failed to respond to BCL2 inhibitors upon blockade of CD11b, antibody-mediated depletion of T lymphocytes or in the context of genetically determined athymia (Supplementary Fig. S9D–S9H). Thus, not only T cells but also myeloid cells are required for pharmacologic BCL2 inhibitors to reduce tumor growth.

Next, we determined which DC subset is involved in the response to BCL2 inhibitors. Batf3^−/− mice lack cDC1 cells (13), and cancers evolving in such mice fail to respond to PD-1 blockade (14, 15) or other types of immunotherapies (16–18, 34, 35). We reconstituted lethally irradiated WT C57BL/6 mice with the bone marrow from syngeneic WT controls or Batf3^−/− mice (Fig. 6A) and confirmed that infusion of Batf3^−/− bone marrow cells caused Batf3^−/− deficiency in the spleen and bone marrow (Fig. 6B) with the consequent reduction in cDC1 cells (defined as CD11c^hiCD8α⁺MHC-II⁺; Supplementary Fig S10A) in lymphatic organs (Fig. 6C and D). Venetoclax injections controlled orthotopic TC1 lung cancers in WT bone marrow–reconstituted mice but completely failed to do so in mice that were reconstituted with bone marrow from Batf3-deficient donors (Fig. 6E–H). However, TC1 tumors established in mice with a Batf3-deficient immune system responded to venetoclax upon the adoptive transfer of WT de-iniDCs. Moreover, such tumors evolving in Batf3-deficient hosts could be partially controlled by the infusion of Bcl2^−/− de-iniDCs (Fig. 6E–H).

The aforementioned results suggest that Batf3-dependent cDC1 cells are required for the anticancer effects of BCL2 inhibitors. To confirm the involvement of cDC1 cells in the response to navitoclax and venetoclax, we resorted to another method of cDC1 depletion consisting of the repeated intravenous injection of Cyt c (21). We confirmed that i.v. injection of Cyt c (5 mg/mouse every other day) caused the selective depletion of cDC1 (including that of XCR1⁺ migratory cDC1) but no cDC2 cells in the blood, tumor, and tumor-draining lymph nodes from mice bearing MCA205 fibrosarcomas (Supplementary Fig. S9A–S9C) or TC-1 lung cancers (Supplementary Fig S10B–S10D). Importantly, Cyt c injections reduced the antitumor effects of navitoclax and venetoclax against TC1 lung cancers (Fig. 7A–D), as well as against MC205 fibrosarcomas (Supplementary Fig. S10E–S10G). Depletion of cDC1 cells also abolished the BCL2 inhibitor-induced upregulation of PD-1 and CTLA-4 on CD8⁺ CTLs and CD4⁺ T cells including Foxp3⁺ Tregs (Fig. 7E–H; Supplementary Fig. S10H–S10K). Accordingly, the synergistic interaction between venetoclax and PD-1 blockade with respect to tumor growth reduction and animal survival was lost when cDC1 were depleted by repeated Cyt c injections (Fig. 7I and J).

Altogether, these results indicate that BCL2 inhibitors require cDC1 cells to mediate their anticancer effects and cooperation with ICIs.

Most anticancer drugs have been designed to selectively kill malignant cells (efficacy) and to spare normal cells (toxicity). However, it has turned out that antineoplastics that are clinically efficient often stimulate an anticancer immune response that accounts for the long-term effects of the medication beyond therapy discontinuation (36). Prominent examples of this mode of action include anthracyclines, oxaliplatin, and taxanes that induce immunogenic cell death (ICD) of malignant cells, hence stimulating an immune response against dead-cell antigens that is mediated in the first place by DCs (that have to engulf portions of stressed and dying cancer cells) which subsequently present tumor-associated antigens to T cells recruited into the tumor microenvironment (37). However, in a certain sense, such ICD-inducing cytotoxicants act “on-target” because they primarily stress and kill malignant cells. Many other anticancer drugs appear to have additional “off-target” effects in the sense that they directly affect different immunostimulatory or immunosuppressive circuitries as this has been well documented for imatinib (which activates the DC-NK cell dialogue; ref. 37) or 5-fluorouracil that depletes myeloid-derived suppressor cells from the tumor bed (38). Our present work suggests that BCL2 inhibitors may also mediate therapy-relevant off-target effects on the immune system, in particular on DCs.

There are several lines of evidence that plead in favor of indirect, immune-dependent effects of BCL2 inhibition by the clinically approved drug venetoclax, as well as by the experimental inhibitor navitoclax. First, MCA205 and TC1 cancers largely failed to respond to treatment with navitoclax and venetoclax when different immune effectors were deficient, as shown for constitutively athymic mice, depletion of T cells in adult mice, the knockout of Batf3 (causing constitutive absence of cDC1 cells), as well as for the depletion of cDC1 cells by Cyt c injections. Second, echoing a prior report dealing with ectopic (s.c.) MC38 colorectal cancers (30), orthotopic TC1 lung cancer and cutaneous MCA205 fibrosarcomas implanted in syngeneic hosts responded to PD-1 blockade more vigorously upon pretreatment with navitoclax and venetoclax. Hence, systemic BCL2 inhibition cannot reduce the growth of solid tumors (such as MCA205, MC38, and TC1) via direct antineoplastic effects but relies on a cellular immune response.

Previous reports indicate that the BCL2 inhibitor venetoclax causes a depletion of various (B, T, and NK) circulating lymphocyte subsets after one year of treatment of patients with chronic lymphoid leukemia (39). This contrasts with another report showing that short-term treatment of healthy volunteers increases the fraction of CD4⁺ and CD8⁺ effector memory cells (TEM and TEMRA) but decreases the proportion of noneffector cells (TN and TCM), in accordance with findings obtained by in vitro treatment of circulating leukocytes (30). Moreover, venetoclax fails to impair the proliferation of activated primary human T cells in vitro (40) and rather directly activates T cells, increasing their cytotoxic potential against AML cells in vitro and in vivo through stimulatory effects on mitochondrial generation of reactive oxygen species (41). Hence, the available evidence from human studies suggests that venetoclax does not compromise immune function and rather stimulates antileukemic T-cell responses. However, no information on the prognostic or predictive value of such T cells has been reported for venetoclax-treated patients. As shown here, venetoclax activates cDC1 cells in humans both in vitro, in cultured blood leukocytes from healthy donors, and in vivo, in patients with AML, in whom circulating cDC1 cells exhibited the upregulation of several markers of DC activation.

Prior studies have demonstrated the importance of cDC1 cells for the outcome of immunotherapies and targeted therapies (42–45). In view of a possible survival deficit of BCL2-inhibited DCs (46–50), we carefully examined the possibility that BCL2 inhibition would deplete such cells. However, venetoclax treatment of tumor-bearing mice or patients with AML failed to deplete DCs, and Bcl2^−/− de-iniDCs that were transferred into NSCLC-bearing mice infiltrated tumors more efficiently than their WT counterparts. In addition, it appears that BCL2 inhibition improves the function of de-iniDCs (in vitro) and cDC1 cells (in vivo). This was coupled to multiple signs of DC activation, with an increase in the expression of chemokine receptors, cytokines, costimulatory receptors, and, perhaps most importantly, a marked stimulation of a type-1 interferon response that is likely due to the activation of the mtDNA/cGAS/STING/IRF3 pathway. Indeed, knockout or blockade of IFNAR1 was sufficient to block the enhancement of antigen presentation by BCL2 inhibition in de-iniDCs in vitro, as well as their tumor growth–controlling activity in vivo.

At the preclinical level, our study provides strong evidence that BCL2 inhibitors act on-target on DCs (and in particular the cDC1 subset) to stimulate an anticancer immune response. First, pharmacologic inhibition or genetic inhibition of BCL2 in de-iniDCs (which transcriptionally and func­tionally resemble cDC1 cells and can functionally reconstitute cDC1-deficient Batf3^−/− mice with respect to cancer immunosurveillance) enhances their antigen-presenting function, induces a marked type-1 interferon response, and upregulates DC activation markers in vitro. Second, venetoclax and navitoclax cannot enhance the function of BCL2 KO de-iniDCs any further, strongly suggesting that they act on BCL2 rather than on other proteins from the BCL2 family. Accordingly, BCLXL reportedly improves cDC1 function if overexpressed, not if inhibited, in another experimental system (51). Third, DC (and specifically cDC1) activation is also observed in mice or humans treated with BCL2 inhibitors in vivo. Fourth, adoptive transfer experiments revealed that BCL2-deficient de-iniDCs are superior to WT de-iniDCs in improving T cell–dependent immunosurveillance in mice. This superior effect was obtained upon i.v. injection of genetically engineered DCs into tumor-bearing WT mice, as well as into Batf3^−/− recipients. Fifth, when WT de-iniDCs are treated ex vivo with navitoclax or venetoclax and subsequently injected into tumor-bearing mice, they mediate enhanced cancer control. Sixth, the constitutive absence of cDC1 cells (due to the Batf3 knockout) or their acquired deficiency (due to injection of Cyt c) reduced or abolished the therapeutic effects of navitoclax or venetoclax against established solid tumors. Finally, the depletion of cDC1 cells prevented the BCL2 inhibitor-induced upregulation of CTLA-4 and PD-1 on T lymphocytes, suggesting that these T-cell effects are secondary to DC stimulation.

In sum, the available evidence suggests that (one of) the primary target(s) of BCL2 inhibitors are DCs from the cDC1 subset. BCL2 appears to act as an endogenous checkpoint of DC function, hence exemplifying a druggable target that normally restrains DC function. If BCL2 inhibitors constitute a novel class of ICIs acting on DCs, they might have broad anticancer effects (especially if combined with “classical” immunotherapies targeting PD-1 or PD-L1) that transcend their direct effects on malignant cells. Most clinical trials involving venetoclax are currently targeting hematopoietic cancers, often in combination with other drugs conceived for their cell-autonomous antineoplastic action. However, a few trials involve combinations with PD-1 or PD-L1 blockade (NCT03390296, NCT03969446, NCT04277442, and NCT05388006 listed in https://clinicaltrials.gov) for the treatment of lymphomas or leukemias. Moreover, one phase I study is evaluating the safety of venetoclax plus pembrolizumab in the treatment of PD-L1^hi NSCLC (NCT04274907). In this context, it will be important to monitor the immunostimulatory effects of BCL2 inhibitors on circulating and tumor-infiltrating cDC1 cells.

RPMI-1640 medium (cat. #61870010), DMEM (cat. #10566016), HEPES (cat. #15630056), sodium pyruvate (cat. #11360070), phosphate-buffered saline (PBS, cat. #20012027), penicillin–streptomycin (10,000 U/mL, cat. #15140122), and TrypLE Express (cat. #12604013) were purchased from Life Technologies. Fetal bovine serum (FBS, cat. #F7524), β-mercaptoethanol (cat. #M3148), dexamethasone (cat. #D0700000), doxycycline hyclate (cat. #D3000000), and PKH26 staining kit (cat. #PKH26GL-1KT) were purchased from Sigma. Recombinant murine GM-CSF (cat. #315-03) was obtained from PeproTech. Unless otherwise indicated, all plasticware was purchased from Life Sciences.

The parental iniDCs were kindly shared by Cornelia Richter and colleagues (20), iniDCs stably expressing CRISPR Cas9 (iniDC_Cas9) were established by transduction with Edit-R lentiviral CAG-Blast-Cas9 nuclease particles (cat. #VCAS10129, Horizon Discovery) followed by cloning and validation as previously published (19). All other gene-edited iniDC cell lines were generated by transfecting iniDC_Cas9 cells with specific crRNA + tracrRNA, followed by single-cell sorting and immunoblotting for knockout validation. As basic DC medium RPMI-1640 with 10% decomplemented FBS, 1 mmol/L sodium pyruvate, 10 mmol/L HEPES, and 1× penicillin/streptomycin was used. β-Mercaptoethanol (at a final concentration of 50 μmol/L) and recombinant GM-CSF (at a final concentration of 10 ng/mL) were freshly added. IniDCs and derivative cell lines are immortalized under the induction of Dex/Dox (Dex at 100 nmol/L + Dox at 2 μmol/L). Dex/Dox removal (“deinduction”) led to a halt in proliferation and differentiation into immature DCs (“de-iniDCs”) that were used for experiments. The B3Z hybridoma T cells were kindly provided by Sebastian Amigorena and maintained with DC medium supplemented with β-mercaptoethanol (50 μmol/L). MCA205 fibrosarcoma (cat. #SCC173, Sigma-Aldrich) and TC1 non–small cell lung cancer cells expressing luciferase (TC1_Luc, kindly shared by T.-C. Wu; ref. 52) were cultured with DMEM containing 10% decomplemented FBS and 1× penicillin/streptomycin. All cell lines were regularly checked for contamination with the MycoStrip Mycoplasma Detection Kit (cat. #rep-mysnc-100, InvivoGen). The cell lines were maintained in culture for a maximum of 15 passages from thawing, and fresh frozen stocks were prepared with cells in the second or third passages and confirmed Mycoplasma-free before freezing.

The Bcl2 family inhibitors including ABT-199/venetoclax (cat. #HY-15531), ABT-263/navitoclax (cat. #HY-10087), ABT-737 (cat. #HY-50907), A-1331852 (cat. #HY-19741), S63845 (cat. #HY-100741), and WEHI-539 (cat. #HY-15607A) were purchased from MedChemExpress. Lipopolysaccharides (LPS, cat. #L2654) albumin from chicken egg white (OVA, cat. #A5503), and cytochrome c from equine heart (cat. #C7752) were obtained from Sigma. Antibodies used for immunoblot and immunofluorescence such as αβ-actin (HRP, cat. #ab49900), αBCL2 (cat. #ab182858), and αLC3B (cat. #ab192890) were from Abcam; αdsDNA marker (cat. #sc-58749) were purchased from Santa Cruz Biotechnology; αATG7 (cat. #8558), αBAX (cat. #2772), and αcleaved caspase-3 (cat. #9661), αCas9 (cat. #19526), and the mouse-reactive STING pathway antibody sampler kit (cat. #16029) were from Cell Signaling Technology. In vivo neutralizing antibodies to PD-1 (cat. #BE0273), CD4 (cat. #BE0003-1), CD8 (cat. #BE0061), CD11b (cat. #BE0007), IFNAR (cat. #BE0241), and corresponding isotype controls (cat. #BE0090, BE0083) were purchased from Bio X Cell. In vitro neutralizing antibodies to mouse CD70 (cat. #104603), CD80 (cat. #1047481), CD86 (cat. #159302), IL12 (cat. #505308), and corresponding isotype controls (cat. #400502, 400940) were obtained from BioLegend. Antibodies for ELISA, including αIL1β (cat. #503502), biotin-conjugated αIL1β (cat. #515801), αIL2 (cat. #503702), biotin-conjugated αIL2 (cat. #503804), IL6 (cat. #504502), and biotin-conjugated αIL6 (cat. #504602) came from BioLegend. Mouse monoclonal antibodies used for high-dimensional flow cytometry αCCR2 BV650 (cat. #150613), αCCR7 Alexa Fluor 488 (cat. #120110), αCCR7 PE (cat. #120106), αCD3 APC (cat. #100236), αCD11c APC (cat. #117310), αCD16/32 (cat. #101302), αCD25 BV650 (cat. #100236), αCD45 Alexa Fluor 700 (cat. #103116), αCD80 PE (cat. #104708), αCD80 PercP-Cy5.5 (cat. #104722), αCD86 BV650 (cat. #105036), αCTLA-4 PE-Cy7 (cat. #106314), αF4/80_BV785 (cat. #123141), αIL12p40_APC (cat. #505206), αMHC-II BV650 (cat. #107641), αMHC-II FITC (cat. #115006), and αXCR1 BV785 (cat. #148225) were from BioLegend; αCD4 eFluor450 (cat. #48-0042-82), αCD8a PercP-Cy5.5 (cat. #45-0081-82), αCD11b eFluor450 (cat. #48-0112-82), αCD11b BUV395 (cat. #363-0112-82), αCD11c APC-eFluor780 (cat. #47-0114-82), αCD40 eFluor450 (cat. #48-0402-82), αCD45 APC-eFluor780 (cat. #103116), αCD69 E-Cy5 (cat. #15-0691-82), αCD83 PE-Cy7 (cat. #25-0839-42), αF4/80_Alexa Fluor 700 (cat. #56-4801-82), αFOXP3 FITC (cat. #11-5773-82), αPD-1 PE (cat. #12-9985-82), αPD-L1 PE (cat. #12-5982-82), αMHC-II PE (cat. #12-5321-82), and αOVA257-264 (SIINFEKL) peptide bound to H-2Kb APC (cat. #17-5743-82) came from eBioscience/Life Technologies; αCD183 BV650 (cat. #740630) and αCD103 BUV661 (cat. #750718) were purchased from BD. Flow cytometry–related monoclonal antibodies for human samples, αCD3 BV650 (cat. #300468), αCD8 APC/Fire810 (cat. #344764), αCD11b BV570 (cat. #101233), αCD14 SparkB550 (cat. #367148), αCD16 BV650 (cat. #302042), αCD19 BV650 (cat. #302238), αCD20 BV650 (cat. #302336), αCD40 PE/Cy7 (cat. #334322), αCD45 PercP (cat. #368506), αCD45RA Spark NIR685 (cat. #304 168), αCD80 PE (cat. #305208), αCD88 APC/Fire750 (cat. #344316), αCD123 PE/Daz594 (cat. #306034), αCD141 BV421 (cat. #344114), αCX3CR1 BV711 (cat. #341630), αHLA-DR PE/Fire810 (cat. #307683), and αXCR1 FITC (cat. #372612) were purchased from BioLegend; αCD1c SB436 (cat. #62-0015-42), αCD206 PP/eF710 (cat. #46-2069-41), αCD69 PE/Cy5.5 (cat. #MHCD6918), αCD274 (PD-L1) PE/Cy5 (cat. #15-5983-42), and αchickIgY-AF647 (cat. #A21449) were purchased from Life Technologies, α CD5 APC/R700 (cat. #565121), α CD11c BUV805 (cat. #742005), α CD25 BV605 (cat. #562660), α CD38 BUV615 (cat. #751138), α CD83 BV510 (cat. #563223), α CD86 BV786 (cat. #740990), α CD169 BUV661 (cat. #750363), α CD178 APC (cat. #564262), α CD192(CCR2) BUV563 (cat. #749076), α CD197(CCR7) BB755 (cat. #624391), α CD279 (PD1) BV750 (cat. #747446), and α HLA-DQ BUV395 (cat. #742614) were purchased from BD; αCD4 Cfl.YG584 (cat. #R7-20041) was purchased from CyTek; α CADM1 (cat. #CM004-3) was purchased from MBL International; α SLAN VioBlue (cat. #130-119-868) was purchased from Miltenyi Biotec. The Live/dead Yellow Fixable Dye (cat. #L34959), Hoechst 33342 (cat. #H3570), Alexa Fluor conjugated second antibody, and MitoTracker Orange CMTMRos (cat. #M7510) were from Life Technologies, and ViaDye Red Fixable Viability Dye (cat. #R7-60008) came from CyTek.

The Mouse Brie CRISPR knockout pooled library was a gift from David Root and John Doench (53). The library was obtained as lentiviral particles from Addgene (cat. #73633). For transduction, iniDC_Cas9 cells were expanded and seeded at 3 × 10⁶ cells/well in tissue culture-treated 12-well plates with complete DC medium containing β-mercaptoethanol and GM-CSF. The transduction mix was prepared by mixing 5 μg polybrene with 200 μL lentiviral particles and 800 μL DC medium for each well, corresponding to a multiplicity of infection of 0.3. The transduction mix was added dropwise into the media and was well mixed by gentle pipetting. Plates were centrifuged at 650 × g for 2 hours at room temperature, then transferred and incubated overnight at 37°C and 5% CO₂. An additional control well was prepared with the same procedure without the lentiviral particles. The next day, infected cells were detached by trypsin and transferred to T175 tissue culture flasks at a density of 2 × 10⁷ cells/flask in DC medium containing β-mercaptoethanol, GM-CSF, and Dex/Dox. The control well was transferred into a T25 flask. Forty-eight hours later, all transfected cells as well as the control cells were treated with 10 μg/mL puromycin in the same medium, which was maintained for 4 days until all cells in the control flask were killed. Supernatant in the T175 flasks containing transduced cells was replaced with fresh DC medium containing β-mercaptoethanol, GM-CSF, and puromycin, but no Dex/Dox. After 4 days of differentiation, the cells were incubated with soluble OVA (2 mg/mL in DC medium) for 18 hours and processed for immunostaining. Briefly, cells were collected by scrapping and washed once with PBS. The final cell density was adjusted to 2.5 × 10⁷ cells/mL in FACS buffer (1% BSA in PBS). An aliquot of untreated control cells was spun down, resuspended in 1 mL FBS containing 10% DMSO, and cryopreserved at −80°C. The cellular suspension was divided into aliquots of 5 × 10⁷ cells in FACS tubs, which were first incubated with 10 μg Fc blocking antibody (anti-CD16/32) for 10 minutes, then centrifuged at 400 × g for 5 minutes and resuspended in 2 mL FACS buffer containing 10 μg of anti-mouse CD80 PE and 5 μg anti-mouse SIINFEKL bound to H-2Kb APC antibodies. Cells were stained at 4°C in the dark, for 30 minutes, then centrifuged at 400 × g for 5 minutes, and finally resuspended with 2.5 mL FACS buffer for continuous cell sorting on a BD FACSAria Fusion flow cytometer (BD). Just before sorting, 1 μg/mL DAPI was added to the stained cells for viability assessment. The CD80^hiSIINFEKL^hi DAPI⁻ cells were collected, and cryopreserved at −80°C. The same procedure was repeated to get biological replicates. Genomic DNA was extracted by using the Blood and Cell Culture DNA Midi Kit (cat. #13343) from Qiagen following the manufacturer's protocol.

PCR of sgRNAs for Illumina sequencing was performed using the NEBNext Ultra II Q5 Master Mix (cat. #M0544X, New England BioLabs), with the P5&P7 primers (sequence can be found in the Addgene; cat. #73633 data sheet) synthesized by Eurofins Genomics. The PCR reaction system was optimized as 50 μL/reaction containing 1 μmol/L of P5 primer mix, 1 μmol/L of P7 primer with distinct barcode sequence, 500 ng genomic DNA, and 25 μL of the PCR master mix. PCR parameters were adapted from the NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB #E7645S/L, #E7103S/L), with an annealing temperature of 72°C and 26 denaturation–annealing–extension cycles. PCR products were purified using the QIAquick PCR and Gel Cleanup Kit (cat. #28506, Qiagen) strictly following the manufacturer's protocol. The quality control of purified PCR products was performed on an Agilent 2100 Bioanalyzer with the High Sensitivity DNA Kit (Agilent). Next-generation sequencing was performed on an Illumina NextSeq 550 System using single reads, with 20% PhiX to improve library diversity, and covered >250 reads per sgRNA. The obtained fastq files were processed by using the MAGeCK package (22, 54). The <mageck count> command was used to generate per-sgRNA read count table by matching single-end reads with sgRNA sequences provided in the brie library data sheet that can be downloaded from Addgene (cat. #73633; ref. 53). The <mageck test> subcommand was used to perform MAGeCK-RRA (22, 54) for the comparison between the sorted and unsorted condition and generated a gene_summary_txt file containing the statistical contents for the identification of hits.

All predesigned guidance RNAs (Edit-R synthetic crRNA), the nontargeting control#1 (cat. #U-007501-01-2), transactivating CRISPR RNA (tracrRNA, cat. #U-002005-5000), as well as the DharmaFECT 1 transfection reagent (cat. #T-2001-04) were purchased from Horizon Discovery. crRNA and tracrRNA were diluted as a 10 μmol/L stock solution in Tris buffer (pH 7.4). Cotransfection of the crRNA:tracrRNA duplex was performed as previously published (19, 31). IniDC_Cas9 cells were seeded in 6-well plates at 1 × 10⁶/well in 2 mL DC medium without Dex/Dox. For each transfection, 25 nmol/L crRNA and 25 nmol/L tracrRNA were mixed in 100 μl RPMI-1640 medium and incubated 5 minutes at room temperature; 10 μL of DharmaFECT 1 transfection reagent was mixed in 100 μL RPMI-1640 medium and incubated 5 minutes at room temperature; then the two solutions were mixed and incubated for another 20 minutes before being added dropwise into iniDC_Cas9 cultures. Three days after transfection, the cells were collected for FACS sorting to obtain clones carrying the KO of interest. For genetic screening purposes, the transfection reagent containing medium was replaced with fresh DC medium to let the cells recover overnight before the in vitro antigen cross-presentation assay.

BMDCs or de-iniDCs and their derivates carrying specific gene knockouts, or de-iniDCs that were transiently transfected with crRNA:tracrRNA for screening, were collected by trypsinization and diluted to 5 × 10⁵ cells/mL with DC medium containing β-mercaptoethanol and GM-CSF. 100 μL cell suspension/well was seeded in 96-well tissue culture U-bottom plates (equal to 5 × 10⁴ cells/well). The cells were treated or not with BCL2 family inhibitors as detailed in the figure legends. Then soluble OVA was added into the cell culture at a final concentration of 1 mg/mL and incubated for 4 hours at 37°C and 5% CO₂. The plates were then centrifuged at 500 × g for 5 minutes, and the supernatant was removed and replaced with 200 μL/well of RMPI 1640 medium. This step was repeated once to wash out the remaining OVA. B3Z T-cell hybridomas were collected by centrifugation of the suspension culture and diluted to 5 × 10⁵ cells/mL with DC medium. After removing the supernatant from DCs washed twice, 200 μL/well of B3Z suspension was added and coincubated with DCs for 18 hours at 37°C before collecting the supernatant by spinning the plates at 500 × g for 5 minutes and gently transferring 150 μL supernatant for the quantification of IL2 secretion by ELISA.

ELISA for IL1β, IL2, and IL6 was performed as previously published (19). In brief, the capture antibody was diluted in 1 × ELISA coating buffer (diluted with water from 5× ELISA coating buffer obtained from BioLegend) at 1/500, applied 100 μL/well in 96-well high-binding assay plates (Corning), and incubated overnight at 4°C; then the plates were washed 3 times with washing buffer (1× TBS with 0.1% Tween-20, 300 μL/well) and incubated with 150 μL/well blocking buffer (10% FBS + 1% BSA in PBS) for 1 hour at room temperature to block unspecific binding sites. Then plates were loaded with samples or serially diluted standards and incubated for 2 hours at room temperature. Afterward, the supernatant was discarded and plates were washed 4 times with washing buffer, then 100 μL/well of biotinylated detection antibody (1/500 diluted in blocking buffer) was added and incubated at room temperature for 1 hour. The supernatant was discarded, and plates were washed 4 times with washing buffer, then 100 μL/well of HRP-Avidin (cat. #405103 from BioLegend, 1/1,000 diluted in blocking buffer) was added and incubated at room temperature for 30 minutes. In the end, plates were washed 5 times and 100 μL/well of 1-Step Ultra TMB-ELISA substrate solution (cat. #34028 from Life Technologies) was added for colorization. When the top standard wells turned dark blue (generally within 10 minutes), 50 μL/well of 0.5 M H₂SO₄ was used to stop the reaction. The absorbance at 450 nm was immediately measured using a BMG FLUOstar plate reader. The exact concentration of the assayed cytokine was calculated by the standard curve and corresponding dilution factors of the sample.

Total RNA was extracted from cultured de-iniDCs (∼5 × 10⁷ cells) with the RNeasy Plus Mini Kit (Qiagen) following the manufacturer’’ instructions. RNA-seq data analysis was performed by GenoSplice Technology (www.genosplice.com). Analysis of sequencing data quality, reads repartition (e.g., for potential ribosomal contamination), inner distance size estimation, gene body coverage, and strand specificity of library were performed using FastQC v0.11.2, Picard-Tools v1.119, Samtools v1.0, and RSeQC v2.3.9. Reads were mapped using STAR v2.4.0f1 (55) on the mouse mm10 genome assembly, and read count was performed using featureCount from SubRead v1.5.0. Gene expression was estimated as described previously (56) using Human FAST DB v2018_1 annotations. Only genes expressed in at least one of the two compared conditions were analyzed further. Genes were considered as expressed if their FPKM value was greater than FPKM of 97% of the intergenic regions (background). Analysis at the gene level was performed using DESeq2 (57) using experiment ID in the DESeq2 GLM model. Genes were considered differentially expressed for fold changes ≥1.5 and P ≤ 0.05. Pathway analyses were performed using WebGestalt v0.4.4 (58) merging results from upregulated and downregulated genes only, as well as all regulated genes. Pathways and networks were considered significant with P ≤ 0.05.

We utilized publicly available raw 10× single-cell transcriptomic data from Brown and colleagues (59), downloaded from the Gene Expression Omnibus (GEO) repository (GSE137710). Using the provided metadata, we selected annotated cells and reprocessed data using Seurat's SCTransform function with default parameters. We then performed dimension reduction using RunPCA and RunUMAP with default parameters.

ssGSEA was performed using the GSVA R package with gene signatures from the single-cell RNA-seq (scRNA-seq) dataset published by Brown and colleagues that were obtained through FindAllMarkers Seurat's function. For broader myeloid cell types, the SiglecH DC cluster was considered as pDC, and cDC2 Tbet⁺, cDC2 Tbet⁻, cDC2 mixed, cDC1, CCR7hi DC clusters were gathered as cDC mega cluster. For refined DC cell types, cDC2 Tbet⁺, cDC2 Tbet⁻, and cDC2 mixed clusters were merged as cDC2.

We deconvoluted our bulk RNA-seq with the DC mouse scRNA-seq dataset using BayesPrism (60) following the standard pipeline to exclude outliers and align based on protein-coding genes. Plots were generated in R using the ggplot package.

The entire immunoblotting procedure was performed according to the standard protocol of the NuPAGE electrophoresis system (Invitrogen), and all reagents were purchased from Life Technologies if not otherwise specified. Protein extracts were obtained by lysing cells in RIPA buffer containing a protease inhibitor cocktail, 1 × SDS Loading Buffer, and 1× Sampling Reducing Buffer. Then, proteins were separated on 4%–12% NuPAGE Bis-Tris gels in the NuPAGE MES SDS Running Buffer and electro-transferred to 0.45-μm polyvinylidene fluoride (PVDF) membranes (Bio-Rad) in 1 × Tris-glycine buffer. Membranes were incubated for 1 hour in 5% BSA dissolved in TBST (Tris-buffered saline containing 0.05% Tween 20) to block unspecific binding sites, followed by incubation with the primary antibody overnight at 4°C. Membranes were then washed 5 times with TBST and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (Southern Biotechnologies), for 2 hours at room temperature. After 5 times washing with TBST, blots were subjected to chemiluminescence-based detection using the Amersham ECL Prime kit, on an ImageQuant LAS 4000 imager (GE Healthcare). Quantitation of chemiluminescence signals of bands of interest was performed with the integrated software ImageQuant TL.

WT or Bcl2^−/− de-iniDCs were seeded in poly-d-lysine-treated 96-well Assay Plate (Corning BioCoat) and let adapt overnight. Upon treatment with venetoclax for 24 hours, cells were stained with MitoTracker Orange CMTMRos dye (1/4,000 diluted in serum-free RMPI-1640 medium) for 30 minutes, and then washed once with PBS before fixing with 10% paraformaldehyde containing 5 μg/mL Hoechst 33342 for 20 minutes at room temperature. Cells were then washed twice with PBS, permeabilized with 0.1% Triton X-100/PBS for 10 minutes, and blocked with 1% BSA/PBS for 1 hour before incubated with a mouse-anti-double strain DNA antibody (dsDNA, 1/200 diluted in 1% BSA/PBS) at 4-degree overnight. Then, cells were washed twice with PBS and incubated with AlexaFluor 488 anti-mouse secondary antibody (diluted 1/250 in 1% BSA/PBS) for 45 minutes at room temperature. Cells were then washed once with PBS, and the plates were sealed with adhesive aluminum for image acquisitions using an ImageXpress Micro C automated confocal microscope (Molecular Devices) equipped with a 20 × PlanApo objective (Nikon). A built-in Custom Module Editor from MetaXpress Software was used for image analysis. Cytoplasmic and nuclear regions of interest were segmented and the mitochondrial region was defined by MitoTracker Orange staining. Fluorescence intensities, dot count, and dot area were measured within these defined regions. Cell-based measurements were transformed into well-based data with the Microsoft Excel Pivot table, which was then graphically depicted and statistically evaluated using the GraphPad Prism software. A minimum of 4 view fields/well was acquired and analyzed.

Cell death was assessed by means of the FITC-Annexin V/DAPI staining protocol. Cells were treated in 6-well plates as detailed in the figure legends, collected by trypsinization, and washed in PBS before the cell pellet was resuspended in 50 μL Annexin V binding buffer containing FITC-conjugated Annexin V (both from BioLegend). Samples were then incubated in the dark for 15 minutes before the addition of 400 μL staining buffer supplemented with 1 μg/mL DAPI. Acquisitions were performed on a BD Fortessa cytofluorometer, and data were analyzed and statistically evaluated using FlowJo.

RNA extraction from bone marrow cells and spleenocytes was performed with the GeneJET RNA Purification Kit (Life Technologies) following the manufacturers’ instructions. Reverse transcription from mRNA to cDNA was performed with the Maxima First Strand cDNA Synthesis Kit (Life Technologies), using approximately 2.5 μg total RNA as a template. RT-PCR reaction was performed on a Step­OnePlus Real-Time PCR System (Applied Biosystems) using the Power SYBR Green PCR Master Mix and corresponding settings. Gene-specific primers were designed by using the NCBI Primer-BLAST online application (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) and synthesized by Eurofins Genomics. Primer sequences are listed in Supplementary Table S4. qRT-PCR data were analyzed using the 2^ΔΔCt-method to obtain the fold change in gene expression that was normalized to expression levels of the housekeeping gene Gapdh.

All mice with the TC1 orthotopic lung cancer model and MCA205 orthotopic fibrosarcoma model were maintained at the Gustave Roussy Campus Cancer in a specific pathogen-free (SPF), environmentally controlled animal facility with 12 hours of light/dark cycles, receiving food and water ad libitum. All animal experiments were performed in compliance with the EU Directive 63/2010 and dedicated ethic protocols (Projects 2020_036 and 2021_010) that were approved by the ethical committee of the Gustave Roussy Campus Cancer, CEEA IRCIV/IGR no. 26, registered at the French Ministry of Research). Female WT C57BL/6 mice (6- to 8-week-old) and female athymic nude (nu/nu) mice were obtained from ENVIGO France. The Batf3-KO mice were maintained in the animal facility of University Hospital Erlangen. Bone marrow from WT or Batf3-KO mice was flushed from femur and tibia, dissociated into single cells, and cryopreserved at −80°C. For bone marrow transplantation, the cells were thawed in a 37°C water bath, washed with worm PBS containing 5% FBS, and resuspended in cold PBS. Five million bone marrow cells were engrafted into lethally irradiated (10 Gy) congenic recipient mice, which were maintained and monitored for two months before being used for tumor establishment. Orthotopic fibrosarcoma and NSCLC models were established as previously published (61). For the fibrosarcoma model, 5 × 10⁵ WT MCA205 cells were s.c. inoculated into the right flank of mice, which were randomly assigned into treatment groups (n = 6–8 animals per group). When tumors became palpable (surface, calculated as longest dimension × perpendicular dimension × π/4, around 20–25 mm²), mice received the treatments described below. Tumor surface was then regularly monitored, and animals bearing neoplastic lesions that exceeded 250 mm² were euthanized. For the TC1 NSCLC model, WT TC1 Luc cells (5 × 10⁵ in 100 μL PBS) were intravenously injected into mice. Tumor incidence and development were monitored by in vivo photonic imaging of tumor cell luciferase activity. When tumor incidence in the lung was detected at an exposure time of 4 minutes (6–7 days after cell injection), mice were randomized for treatment as described below. To perform bioluminescence imaging, mice were injected i.p. with 3 mg beetle luciferin potassium salt dissolved in DPBS (Promega). After 8 minutes (at peak bioluminescence signal), mice were anesthetized with vaporized isoflurane, and photons were acquired on an IVIS Lumina III imaging system (Caliper Life Sciences Inc.). In vivo imaging was conducted every 4 to 5 days with an exposure time starting with 4 minutes gradually decreasing to 1 minute when photon saturation occurred. Tumor-bearing mice showing photon saturation at 1 minute of exposure at small binning settings were euthanized.

Mice for the spontaneous breast cancer model were housed in the SPF animal facility of Weill Cornell Medical College, and all experimentation was aligned with the Guidelines for the Care and Use of Laboratory Animals guidelines and approved by the Institutional Animal Care and Use Committee (IACUC; no. 2023-0014). Six- to 9-week-old female C57BL/6J mice (Taconic Bioscience) were subcutaneously implanted with 50 mg slow-release (90 days) medroxyprogesterone acetate (MPA, M) pellets (Innovative Research of America) followed by oral gavage with 1 mg 7,12-dimethylbenz[a]anthracene (DMBA, D) in 200 μL corn oil once a week for 7 weeks after pellet implantation (62). Mice were then routinely assessed for the development of M/D-driven malignant lesions along the mammary lines, until reaching a surface area of 12–25 mm² (day 0). Mice bearing M/D-driven mammary tumors were randomly allocated to treatments included (i) vehicle control: 100 μL PEG delivered oral gavage daily on day 0 until the end of the experiment; (ii) venetoclax: 100 mg/kg delivered oral gavage in 100 μL vehicle on day 0 until the end of the experiment; (iii) focal RT: three fractions of 10 Gy each (total dose: 30 Gy, dose rate: 271 cGy/minute) delivered to the primary tumor on days 0, 1, and 2; (iv) focal RT followed by venetoclax; (v) focal RT followed by venetoclax and anti–IFNAR-1 antibody (Clone MAR1-5A3, from Bio X Cell) delivered i.p. at 20 mg/kg on day −1 and then weekly until the end of the experiment. Mice were routinely assessed for the emergence of toxicity (troublesome breathing, weight loss, anorexia and hunched posture) and tumor growth. Mice bearing M/D-driven tumors were euthanized when the tumor burden area reached 180–200 mm².

IniDC_Cas9 cells or derivates carrying gene KO were cultured in a medium without Dex/Dox to be differentiated into de-iniDCs as described before. Three days after withdrawing Dex/Dox, the lysates prepared from equal numbers of MCA205 or TC1-Luc cells (by freezing–thawing process in liquid nitrogen and a water bath, followed by sonication to ensure complete disruption of cells) were added to the cell culture and incubated for 2 hours for tumor antigen exposure. In some cases, the 3-day differentiated de-iniDC_Cas9 cells were pretreated with navitoclax or venetoclax at 5 μmol/L for 4 hours before being exposed to cancer cell lysates. Then the antigen-loaded de-iniDC_Cas9 cells were collected by trypsinization or by scrapping the cell culture layer with a plastic cell lifter. Cells were washed twice with cold PBS and passed through a 70-μm strainer to remove clogs. The single-cell suspension was diluted in cold PBS for intertumoral injection (1 × 10⁶ cells/mouse) or i.v. injection (2 × 10⁶ cells/mouse). Where indicated, the cells were incubated with an IFNAR-blocking monoclonal antibody at 10 μg per 1 × 10⁶ DCs, or with an equal quantity of isotype control antibody for 30 minutes before i.v. injection. To monitor the migration of de-iniDCs, cells were stained with the long-lasting red fluorescent dye PKH26 following the manufacturer's protocol. After injection, the lung and lung cancer-draining mediastinal lymph nodes were excised at different time points and digested to single-cell suspensions for multiplex immunofluorescence staining and flow-cytometric analysis, which allowed for subgating of DC markers and PKH26 red fluorescence. Alternatively, after i.v. injection of de-iniDCs the complete lung was excised together with the trachea, flushed with cold PBS, and then quickly infused with 4% PFA (diluted with PBS) via the trachea, which were then closed with surgical suture and immersed into a large volume of 4% PFA. Fixed lungs were embedded in paraffin and subjected to tissue sectioning and IHC.

Solvent (Sol) for chemicals is formulated as 10% Tween-80, 10% PEG400, and 4% DMSO in physiologic saline. Navitoclax and venetoclax were administrated i.p. at a dose of 50 mg/kg, following the schedule specified in the figures and corresponding legends. In case of combination with checkpoint blockade, mice received i.p. injection of either 200 μg anti–PD-1 antibody, 100 μg anti-CTLA-4 antibody, or 200 μg isotype antibody, at 8, 12, and 16 days after the first chemical treatments. For T-cell depletion, mice received i.p. injections of 100 μg anti-CD8 plus 100 μg anti-CD4 antibody, or 200 μg of isotype antibody, one day before and the same day of pharmacologic treatment or cell transfer, which were continued at a frequency of once a week for two weeks. For CD11b neutralization, mice received 100 μg anti-CD11b or equal amounts of isotype control antibody as scheduled for anti-CD4/CD8, but were treated every other day for the following 2 weeks. In some cases, mice were treated i.v. with 5 mg cytochrome C/mouse in PBS or PBS alone following the same schedule as anti-CD11b.

Orthotopic MCA205 fibrosarcoma or TC1 NSCLC cancers were established and tumor-bearing mice were treated as described above. At day 3 or day 7 after treatment, blood was collected from tumor-bearing mice via cardiac puncture (under anesthesia with vaporized isoflurane) into 2 mL centrifuge tubes containing EDTA-K. Then mice were euthanized for excising tumors and immune organs. The samples were collected in a cold RPMI-1640 medium and kept on ice until dissociation. Blood was directly subjected to erythrocyte elimination by using 1× red blood cell lysing buffer (BioLegend); spleen and lymph nodes were squeezed through 70-μm strainers (Corning) with the rubber tip of 1 mL syringe to generate single-cell solutions; tumor-bearing lungs and excised s.c. tumors were digested in an enzymic buffer containing 1 mg/mL collagenase type IV (Life Technologies) and DNase I (Sigma). The dissociated bulk cell suspension was resuspended in RPMI-1640, passed through 70-μm cell strainers, and washed twice with cold PBS. Cells from spleen and lung were further treated with 1× red blood cell lysis buffer to remove erythrocytes. Prior to surface staining of fluorescent antibodies, samples were incubated with LIVE/DEAD Yellow Fixable dye to label damaged/dead cells, and incubated with antibodies against CD16/CD32 to block Fc receptors. For multiplex staining, cells were incubated with a panel of fluorescence-conjugated antibodies for 30 minutes of surface staining in the dark. In the case of Foxp3 staining, the surface-labeled cells were permeabilized and fixed using a Foxp3/Transcription Factor Staining Buffer kit (Life Technologies), and stained with the FOXP3 FITC antibody for another 30 minutes. Otherwise, surface-labeled cells were directly fixed with 4% PFA (Sigma). After 2 times wash, the cells were kept at 4°C until flow-cytometric analysis. For the intracellular staining of IL12 p40, the dissociated cells were incubated with Brefeldin A (5 μg/mL diluted in RPMI medium) for 2 hours before surface staining. The surface-labeled cells were permeabilized and fixed using a Foxp3/Transcription Factor Staining Buffer kit, and stained with the IL12_APC antibody for another 30 minutes. Data were acquired on a BD LSRFortessa flow cytometer (BD Biosciences) and analyzed using the FlowJo software. Detailed gating strategies for the flow-cytometric analysis are provided in the corresponding Supplementary Figures.

Peripheral blood samples were collected from patients with AML, before and after treatment with venetoclax in combination with azacytidine, at the Gustave Roussy Cancer Institute. Written informed consent authorizing the blood samples to be used for research purposes was obtained from all patients, and the study was approved according to the guidelines of the protocol alpha PPP ID RCB: 2020 A03290-39 by the review board of Gustave Roussy Cancer Institute. Peripheral blood was also collected from healthy donors ages 25 to 35. Peripheral blood mononuclear cells (PBMC) were isolated by Ficoll-Paque PLUS (GE Healthcare) density gradient centrifugation following the manufacturer's protocol. PBMCs from patients with AML were cryopreserved at −80°C until staining. PBMCs from healthy donors were treated in vitro with venetoclax, azacytidine alone, or their combination (both at a concentration of 5 μmol/L) for 18 hours, and then cryopreserved as well before being stained together with the patients’ PBMCs.

For multiplex staining, frozen PBMCs were rapidly thawed in a 37°C water bath, pooled in 20 mL RPMI-1640 medium, and centrifuged at 500 × g for 5 minutes at 4°C. After discarding the supernatant, cells were resuspended in 1 mL cold PBS and transferred to 5 mL FACS tubes for additional steps. After centrifugation and removal of the supernatant, the cell pellets were resuspended with 500 μL of cold via dye red solution (1/500 diluted in PBS) and incubated at 4°C for 15 minutes. Then, 10 μL of FBS was added to each sample (equal to 2% FBS) and the cells were incubated for another 15 minutes. During this period, the antibody cocktail was prepared in two staining batches, both in the Brilliant Stain buffer (BD Biosciences) containing 2% FBS and 2 mmol/L EDTA. Batch #1 contains CCR2_BUV563, CD169_BUV661, SLAN_VioBlue, CD25_BV605, CX3CR1_BV711, PD-1_BV750, XCR1_FITC, CD206_PP/eFluor710, CCR7_BB755, CD80_PE, PD-L1_PE/Cy5, CD40_PE/Cy7, CD178_APC, CADM1, CD5_APC/R700, CD88_APC/Fire750, and CD8_APC/Fire810. Batch #2 contains AlexaFluor647 goat-anti-chicken second antibody (for CADM1), HLA-DQ_BUV395, CD38_BUV615, CD11c_BUV805, CD141_BV421, CD1c_SB436, CD83_BV510, CD3/16/19/20_BV650, CD86_BV786, CD14_SparkB550, CD45_PercP, CD4_Cfl.YG584, CD123_PE/Daz594, CD69_PE/Cy5.5, HLA-DR_PE/Fire810, and CD45RA_Spark NIR685. Following Viadye red staining, 2 mL of FACS buffer was added to each tube, and cells were spun down. The cells were then resuspended in the batch #1 antibody cocktail (100 μL/sample) and incubated at 37°C for 30 minutes. Then 2 mL FACS buffer was added and the supernatant was removed after centrifugation. Pellets were resuspended in the batch #2 antibody cocktail and incubated for 30 minutes at 4°C. After the second staining, the cells were spun down and washed twice with 2 mL FACS buffer. The stained cells were resuspended in 500 μL FACS buffer and acquired on a Cytek Aurora flow cytometer with compensations. Data analysis was performed with the FlowJo software.

Statistical significance was calculated using the GraphPad Prism software (Version 9.0.2), by means of one-way or two-way ANOVA test [with false discovery rate (FDR) or Dunnett multiple comparisons test], unpaired or paired Student t test, or Fisher exact test, as detailed in the corresponding figure legends. TumGrowth was used to analyze in vivo data (63): linear or log-transformed mixed-effects models for longitudinal comparison of tumor growth curves by type II ANOVA; cross-sectional analysis with likelihood ratio test for comparing endpoint tumor size distribution; and Cox proportional hazards regression or log-rank test for comparing survival curves. TumGrowth is freely available at Github/Kroemerlab. P values of 0.05 or less were considered to denote significance and were properly annotated in the figures.

CRISPR screen data generated are provided in Supplementary Table S1. Results from arrayed KO screens are provided in Supplementary Table S2. The bulk RNA-seq data have been deposited in the GEO database under accession number GSE218062. All other raw data generated or analyzed during this study are available upon request from the corresponding authors. No code or programs have been generated in this study.

L. Galluzzi reports grants and other support from Lytix, grants, personal fees, and other support from Promontory and Onxeo, personal fees from AstraZeneca, OmniSEQ, Longevity Labs, Inzen, Imvax, EduCom, and Boehringer Ingelheim, personal fees and other support from the Luke Heller TECPR2 Foundation, Sotio, Noxo­pharm, and other support from Ricerchiamo outside the submitted work. D. Dudziak reports grants from the German Research Foundation (DFG-TRR305, 429280966, B05, and DFG/ANR, 431402787) during the conduct of the study, as well as other support from Affimed outside the submitted work. L. Zitvogel reports personal fees and other support from EverImmune, grants from Pileje and Daiichi Sankyo, personal fees from Hookipa, and other support from Pierre Fabre during the conduct of the study, as well as personal fees from IHU Mediterranée Infections (SAB President) outside the submitted work. O. Kepp reports grants from INCA during the conduct of the study, as well as other support from Samsara Therapeutics outside the submitted work. G. Kroemer reports grants from Ligue contre le Cancer, Agence National de la Recherche, Association pour la Recherche sur le Cancer, Cancêropôle Ile-de France, Elior, the European Joint Programme on Rare Diseases (EJPRD), Gustave Roussy, the European Union, Fondation Carrefour, Institut National du Cancer, Institut Universitaire de France, The Mark Foundation, the Seerave Foundation, Investissements d'Avenir, and the European Research Council during the conduct of the study; has held research contracts with Daiichi Sankyo, Eleor, Kaleido, Lytix Pharma, PharmaMar, Osasuna Therapeutics, Samsara Therapeutics, Sanofi, Tollys, and Vascage; is on the board of directors for the Bristol Myers Squibb Foundation France; is a scientific cofounder of EverImmune, Osasuna Therapeutics, Samsara Therapeutics, and Therafast Bio; is on scientific advisory boards for Hevolution, Institut Servier, and Longevity Vision Funds; and is the inventor of patents covering therapeutic targeting of aging, cancer, cystic fibrosis, and metabolic disorders. G. Kroemer's wife, Dr. Laurence Zitvogel, has held research contracts with GSK, Incyte, Lytix, Kaleido, Innovate Pharma, Daiichi Sankyo, Pilege, Merus, Transgene, 9 m, Tusk, and Roche, was on the board of directors of Transgene, is a cofounder of EverImmune, and holds patents covering the treatment of cancer and the therapeutic manipulation of microbiota. G. Kroemer's brother, Dr. Romano Kroemer, was an employee of Sanofi and now consults for Boehringer Ingelheim. No disclosures were reported by the other authors.

L. Zhao: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. P. Liu: Conceptualization, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Mao: Data curation, investigation, visualization, methodology, writing–review and editing. S. Zhang: Data curation, formal analysis, investigation, methodology. C. Bigenwald: Resources, validation, methodology, writing–review and editing. C. Dutertre: Resources, data curation, formal analysis, methodology, writing–review and editing. C.H. Lehmann: Resources, methodology, writing–review and editing. H. Pan: Data curation, formal analysis, investigation, methodology, writing–review and editing. N. Paulhan: Data curation, methodology, writing–review and editing. L. Amon: Resources, methodology, writing–review and editing. A. Buque: Data curation, formal analysis, methodology, writing–review and editing. T. Yamazaki: Data curation, formal analysis, writing–review and editing. L. Galluzzi: Conceptualization, resources, supervision, writing–review and editing. B. Kloeckner: Data curation, formal analysis, methodology, writing–review and editing. A. Silvin: Data curation, formal analysis, methodology, writing–review and editing. Y. Pan: Data curation, methodology, writing–review and editing. H. Chen: Data curation, methodology, writing–review and editing. A. Tian: Data curation, methodology, writing–review and editing. P. Ly: Resources, methodology, writing–review and editing. D. Dudziak: Conceptualization, resources, supervision, validation, writing–review and editing. L. Zitvogel: Conceptualization, resources, supervision, funding acquisition, validation, investigation, writing–review and editing. O. Kepp: Conceptualization, resources, supervision, funding acquisition, validation, investigation, writing–original draft, writing–review and editing. G. Kroemer: Conceptualization, resources, supervision, funding acquisition, validation, investigation, writing–original draft, project administration, writing–review and editing.

G. Kroemer and L. Zitvogel are supported by the Ligue contre le Cancer (équipe labellisée); Agence National de la Recherche (ANR)—Projets Blancs; AMMICa US23/CNRS UMS3655; Association pour la Recherche sur le Cancer (ARC); Cancéropôle Ile-de-France; Fondation pour la Recherche Médicale (FRM); a donation by Elior; Equipex Onco-Pheno-Screen; European Joint Programme on Rare Diseases (EJPRD); Gustave Roussy Odyssea, the European Union Horizon 2020 Projects Oncobiome and Crimson; Fondation Carrefour; Institut National du Cancer (INCa); Institut Universitaire de France; LabEx Immuno-Oncology (ANR-18-IDEX-0001); a Cancer Research ASPIRE Award from the Mark Foundation; the RHU Immunolife; Seerave Foundation; SIRIC Stratified Oncology Cell DNA-Repair and Tumor Immune Elimination (SOCRATE); and SIRIC Cancer Research and Personalized Medicine (CARPEM). This study contributes to the IdEx Université de Paris ANR-18-IDEX-0001. O. Kepp is supported by the Ligue contre le Cancer and the DIM Elicit initative of the Île de France. This work has benefited from the facilities and expertise of the Gustave Roussy, Université Paris Saclay, UMS AMMICa, Plateforme Imagerie et Cytométrie, Villejuif, France. D. Dudziak was supported by the German Research Foundation [Deutsche Forschungsgemeinschaft (DFG); CRC1181-TPA7 261193037, DU548/5-1 420943261, TRR305 429280966, RTG2504 401821119, RTG2599 421758891], and the Bavarian State Ministry of Science and Art (Bayresq.Net-IRIS). D. Dudziak and C.H. Lehmann were funded by the Agence Nationale de la Recherche (ANR) and the DFG (DU548/6-1 431402787). C.H. Lehmann was supported by the Interdisziplinäres Zentrum für klinische Forschung (IZKF; IZKF-A87) and the German Research Foundation [Deutsche Forschungsgemeinschaft (DFG)] (RTG2504 401821119). L. Galluzzi is/has been supported (as a PI unless otherwise indicated) by two Breakthrough Level 2 grants from the US DoD BCRP (#BC180476P1; #BC210945), by a Transformative Breast Cancer Consortium Grant from the US DoD BCRP (#W81XWH2120034, PI: Formenti), by a U54 grant from NIH/NCI (#CA274291, PI: Deasy, Formenti, Weichselbaum), by the 2019 Laura Ziskin Prize in Translational Research (#ZP-6177, PI: Formenti) from the Stand Up to Cancer (SU2C), by a Mantle Cell Lymphoma Research Initiative (MCL-RI, PI: Chen-Kiang) grant from the Leukemia and Lymphoma Society (LLS), by a Rapid Response Grant from the Functional Genomics Initiative (New York, US), by startup funds from the Dept. of Radiation Oncology at Weill Cornell Medicine (New York, US), by industrial collaborations with Lytix Biopharma (Oslo, Norway), Promontory (New York, US), and Onxeo (Paris, France), as well as by donations from Promontory (New York, US), the Luke Heller TECPR2 Foundation (Boston, US), Sotio a.s. (Prague, Czech Republic), Lytix Biopharma (Oslo, Norway), Onxeo (Paris, France), Ricerchiamo (Brescia, Italy), and Noxopharm (Chatswood, Australia).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.