After informed consent, blood samples from multiple myeloma patients were collected at the University Hospital of Nantes Department of Hematology (MYRACLE study; NTC03807128) (17). For the present study, inclusion criteria were adult man/woman, diagnosis of relapsed multiple myeloma, presence of circulating plasma cells (≥5%). Blood from one patient (sPCL1) was collected at the department of Immuno-hematology of Saint-Louis Hospital (Paris). Cells from this patient were immortalized in our laboratory after being cultured in RPMI1640 with 5% FCS and 3ng/ml recombinant IL-6 and gave rise to NAN12 cell line.

Human myeloma cell lines (HMCLs, n=10) were extensively characterized as previously described (18). The XG5, XG7, MDN, NAN10 and NAN12 HMCLs were derived in our laboratory. The KMS12PE and KMM1 HMCLs were kindly provided by Dr. Otsuki (Kawasaki Medical School, Kurashiki, Japan); KARPAS-620 (K620), by Dr. Karpas (Cambridge Clinical School, Cambridge UK); ANBL6 by Dr. Jelinek (Rochester,USA) and MM1S, by Dr. S. Rosen (Northwestern University, Chicago, USA).

Venetoclax (ABT-199) and A1155463 were purchased from Selleck Chemicals GmbH, DT2216 from Clinisciences. The following antibodies were used: BCL2 (Dako, M0887), BCLXL (2764S), BAX (2772), BAK (12105), PUMA (12450) and VHL (68547) from Cell Signaling. BCLXL (sc-271121), MCL1 (sc-12756), BIK (sc-10770) from Santa Cruz. ACTIN (MAB1501) and BIM (Ab17003) from Millipore.

Mononuclear cells (MNC) were isolated from blood samples by Ficoll-Hypaque density gradient centrifugation and immediately cultured in RPMI-1640 media with 5% fetal calf serum. Plasma cells were identified using CD138 staining (anti-CD138-PE, Beckman Coulter). MNC were incubated 24h with BCL2, BCLXL BH3 mimetics or DT2216 and an untreated condition was included as a control. After CD138 staining, cell death response was measured by the loss of CD138 expression, as previously described (8). Specific cell death was expressed as the percentage relative to the untreated control condition.

Cell death in HMCL was assessed using AnnexinV-FITC staining. Fluorescence acquisition and analysis were performed using a FACsCanto (Becton Dickinson) and FlowJo software.

BH3 profiling was performed using the BIM BH3 derived peptide (0.1 μM), HRK BH3 derived peptide (10 μM) and the inert recombinant PUMA BH3-only peptide (PUMA2A) as negative control, as previously described (19). Briefly, cells were permeabilized with 0.004% Digitonin and exposed to peptides for 45 min at 27°C before fixation with 8% formaldehyde at room temperature for 15 min. After addition of neutralizing buffer (Tris 0.41 mol/L glycine pH 9.1) for 5 min, cells are stained with anti-cytochrome c–Alexa 647 (BLE612310, Ozyme) 1:40 in 0.1%Saponin/1%BSA/PBS overnight at 4°C. Loss of cytochrome c was analyzed by a flow cytometry. The quantification of cytochrome c loss induced by each peptide was analyzed by gating the cytochrome c negative population. For cytochrome c release, cells were treated with DT2216 for 15h, then permeabilized with 0.004% Digitonin and processed as in BH3 profiling.

Plasma cells were isolated from the MNC fraction using CD138 immunomagnetic beads, MS columns and MACS™ Separator (Miltenyl Biotech) following manufacture’s protocol. Briefly, cells were magnetically labeled with the CD138 immunomagnetic beads during 15 minutes at 4°C. The MS column was placed on the MACS™ Separator and pre-washed with MACS buffer. Then, cell suspension was loaded into the column and the flow-through fraction, containing the CD138 negative cells, was discarded. After 3 washes with MACS buffer, the column was removed from the separator. The retained cells were eluted as the CD138 + fraction.

Total RNA was obtained from purified CD138+ plasma cells using RNeasy mini kit (Qiagen). The mRNA expression of BCL2 family members was performed by 3’digital gene expression (DGE) RNA-sequencing protocol as previously described (20).

Cells were lysed for 40 minutes on ice in 0.5% NP40 containing buffer for immunoblotting or lysed in 1% Digitonin containing buffer for co-immunoprecipitation assays. Lysates were centrifugated at 12,000 X g for 30 min and supernatants were collected. Western blotting and immunoprecipitation reactions were performed as previously described (21). Protein expression levels were quantified using ImageLab and Image J software.

For the detection of active forms of BAX and BAK, 5x10⁵ cells were treated with DT2216 or not. After incubation, cells were fixed and permeabilized using the FOXP3 transcription factor staining buffer set (Thermo Fischer Scientific, 00-5523-00) following the manufacturer’s recommendations. Cells were incubated with the following antibodies: BAX (clone 6A7, Santa Cruz, sc-23959), BAK (clone G317-2, BD Biosciences) and a mouse Ig G1 isotypic control (Miltenyl, 130-106-545) for 30 min. After washing, cells were incubated with the Alexa fluor 647 antibodies for 30 min, washed once in PBS and resuspended in PBS-1% formaldehyde. The flow cytometry acquisition and analysis were performed on a FacsCanto (Becton Dickinson) and FlowJo software, respectively.

Correlation was assessed by the Spearman correlation method, (r) and p values are indicated.

A cohort of 13 consecutive sPCL was analyzed for the sensitivity to BH3 mimetics targeting either BCL2 (ABT-199/venetoclax) or BCLXL (A1155463) ( Table 1 ). The main molecular characteristics of sPCL were defined on the basis of DGE RNA sequencing profile and are summarized in Table 1 . sPCL 4, 6 and 8 were only sensitive to venetoclax, sPCL 1, 5 and 11 were sensitive to A1155463 and the other 7 sPCLs were resistant to both BH3 mimetics ( Table 1 ). Analysis of cell death in sPCL samples upon BH3 mimetics treatment is illustrated for sPCL1, sensitive to A1155463 and for the non-sensitive sPCL3 sample ( Supplementary Figure 1 ). Additionally, sensitivity to A1155463 was paired with BH3 profiling using BIM-BH3 derived peptide, to evaluate the global priming and HRK-BH3 derived peptide, specific for BCLXL, in four sPCL samples; sPCL1 and sPCL5 were sensitive to A1155463 and sPCL6 and sPCL8 sensitive to venetoclax. This analysis showed 91 to 99% cytochrome c release after exposure to BIM-derived peptide in the four analyzed samples, reflecting their high capacity to undergo mitochondrial apoptosis, according to their sensitivity to the respective BH3 mimetic. However, only cells from sPCL1 and sPCL5 permeabilized their mitochondria upon the HRK-BH3 peptide (71 and 76% cytochrome c release, respectively), confirming the BCLXL primed profile of these samples ( Figure 1A ). In contrast, cells from sPCL6 and sPCL8 released 12 and 30% of cytochrome c upon exposure to HRK-BH3 peptide, in agreement with their resistance to A1155463 ( Figure 1A ).

We further analyzed the mRNA BCL2 family expression according to BH3 mimetics response. We found that BCL2 mRNA levels did not correlated with venetoclax sensitivity in contrast to BCL2/BCL2L1 (p=0.0437), BCL2/BCL2L1+MCL1 (p=0.0142) and BCL2/MCL1 (p=0.0036) mRNA ratios, the latter ratio correlated the best with venetoclax response, ( Figure 1B ). Of note, the sensitivity to A1155463 significantly correlated with BCL2L1 (coding for BCLXL) mRNA levels (p=0.006), though the ratio of BCL2L1/MCL1 (p=0.0018) mRNA appeared the most potent marker associated with A1155463 sensitivity ( Figure 1C ). Nevertheless, the MCL1 mRNA levels itself did not correlate with the sensitivity of any of both BH3 mimetics tested. ( Supplementary Figure 2 ).

Since the sPCLs sensitive to A1155463 were venetoclax resistant, targeting BCLXL by the specific BCLXL Protac degrader DT2216 could be of interest for this particular sPCL subgroup. To study the specificity and the ability of DT2216 to degrade BCLXL, we took advantage of myeloma cell lines (HMCLs) and in particular of NAN12 cell line, which has been immortalized in presence of IL-6 from the sPCL1. Because DT2216 relies on Von Hippel-Lindau (VHL) E3 ligase to achieve BCLXL protein degradation, we first analyzed VHL expression in 10 HMCLs and CD138+ cells from 3 sPCL (1, 6 and 8). Due to alternative codon initiation, VHL protein is expressed as two isoforms of 18 and 24 kDa (23), which were detected in all HMCL tested (n=10), as well as in sPCL cells ( Figure 2A ). However, the levels of the VHL E3 ligase did not correlate with DT2216 sensitivity (r=0.1593, p=0.604) ( Figure 2A , Supplementary Table 1 ), suggesting that cell death induced by DT2216 was not influenced by the level of expression of VHL protein. We next analyzed the degradation of BCLXL in 4 different cell lines treated with various doses of DT2216. After 48hrs of treatment, 20 to 50% of BCLXL was already degraded at 25nM DT2216, reaching a major degradation of at least 80% of BCLXL observed at 150nM in all HMCLs tested ( Figure 2B ). Finally, we confirmed that DT2216 selectively degraded BCLXL, even after 72 hours at the dose of 150nM, without modifying the levels of MCL1 or BCL2 ( Figure 2C ).

We further compared the sensitivity to DT2216 and A1155463 in HMCLs (n=10) and primary cells from sPCLs (1, 6, 8 and 13). We found a strong and significant correlation between the sensitivity to these two compounds (r=0.8699, p=0.0001, Spearman test). Three cell lines (ANBL6, MM1S and NAN12) and 1 sPCL (sPCL1) were highly sensitive to DT2216 with LD₅₀ values ranging from 15 to 100 nM ( Figure 3A , Supplementary Table 1 ). Because 150nM of DT2216 almost completely degraded BCLXL (at least 80%) in all myeloma cells studied, only cells that were significantly killed at this dose or less were considered as BCLXL dependent. Moreover, DT2216 sensitive cells were resistant to venetoclax ( Supplementary Table 1 ) and therefore dependent on BCLXL. One cell line (KARPAS 620) had a particular sensitivity profile characterized by an intermediate sensitivity to DT2216 (LD₅₀ = 231 nM) and a high sensitivity to venetoclax (LD_50 = 5 nM). This result indicates a co-dependence on BCLXL as well as BCL2, which may rely on its high BCL2 protein level ( Figure 3B ).

We also analyzed the expression of the main BCL2 anti-apoptotic proteins in relation to DT2216 response. DT2216 sensitive myeloma cells expressed the highest levels of BCLXL and moderate levels of either BCL2, MCL1 or both. In opposition, resistant cells were characterized by a lower BCLXL and higher BCL2 and/or MCL1 protein expression ( Figure 3B ). We further demonstrated that DT2216 sensitivity strongly correlated with BCLXL protein expression (r= -0.7496, p=0.0067) as well as the ratio of BCLXL/MCL1+BCL2 protein (r=-0.7483, p=0.007) ( Figure 3C ). Although to a lesser extent, the protein ratio of BCLXL/MCL1 also correlated with induction of cell death by DT2216 (r=-0.6224, p=0.0347), while no correlation was found with the ratio of BCLXL/BCL2. At the same time, neither BCL2 nor MCL1 showed any correlation with DT2216 cell death induction ( Figure 3C ). To go deeper into the mechanism of cell death induction by DT2216, we investigated its impact on the mitochondrial apoptotic pathway. Indeed, we demonstrated cytochrome c release ( Figure 3D ), caspase-9 and consequently caspase-3 activation under DT2216 treatment in sensitive NAN12 cells. In contrast, no caspase activation was observed in the resistant XG7 cell line ( Figure 3E ). Of note, using antibodies against the conformational active forms of BAK and BAX, we demonstrated that upon DT2216 treatment both effectors were readily activated from 24hrs (27% BAK and 29% BAX) in NAN12 cells. Activation that clearly progressed at 36 hrs, since around 50% of cells displayed both activated forms. As expected, no activation of BAK or BAX was observed in the resistant XG7 cells ( Figure 3F ).

Finally, we investigated the nature of BCLXL complexes in the sensitive NAN12 cell line by co-immunoprecipitation assays. Our results revealed that BCLXL was complexed with both BH3-only (BIK, BIM) and effector proteins (BAX, BAK) ( Figure 3G ). Quantification of pro-apoptotic proteins before and after BCLXL immunoprecipitation indicated that 86% of BIK, 77% of BIM and around 50% of BAX and BAK effectors expressed in NAN12 were sequestered by BCLXL. Thus, these findings supported the activation of BAX and BAK effectors in DT2216 cell death induction, according to the unified model of the BCL2 members interaction (1).