Written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki. Peripheral blood mononuclear cells (PBMCs) were isolated from peripheral blood of CLL patients (table 1) or buffy coats of (age-matched) HDs (table 2) from Sanquin Blood Supply (Amsterdam, The Netherlands) using Ficoll-Plaque (VWR). All samples were cryopreserved in liquid nitrogen and CLL samples used had a purity of CD5⁺CD19⁺ of at least 85%. CLL patients were selected on their diagnosis (CD5⁺CD19⁺CD23⁺ lymphocytosis), treatment history or chemoresistance.

The CD3xCD19 DART (JNJ-64052781, also known as MGD011) molecule was constructed as described.24 The CD3xCD19 DART is composed of anti-CD3 sequence XR32 and anti-CD19 sequence BU12. The control DART molecule, CD3xFITC, in which the variable domain sequences of the antifluorescein mAb 4-4-20 replaces the CD19 arm, was engineered in a similar manner. DART molecules were expressed transiently in CHO-S cells and purified to homogeneity by using protein A.

CLL cells, 3T3 and 3T40 (expressing CD40L) fibroblasts and Ramos cells were cultured in Iscove’s Modified Dulbecco’s Medium (Thermo Fisher Scientific). JeKo-1 and Daudi cells were cultured in RPMI 1640 medium (Thermo Fisher Scientific). Medium was supplemented with 10% fetal calf serum and 1% penicillin/streptomycin.

PBMCs were washed with PBA (phosphate buffered saline, 0.5% BSA and 0.02% sodium azide) and stained using fluorescently labeled antibodies for 20 min on ice. The following antibodies were used: CD3 Alexa Fluor (AF)700 (eBioscience), CD3 allophycocyanin (APC) (eBioscience), CD3 V500 (BD Biosciences), CD4 Phycoerythrin (PE)-Cy7 (BD Biosciences), CD4 AF594 (Biolegend), CD4 Brilliant Violet (BV) 605 (BD Biosciences), CD5 PE (eBioscience), CD8 PerCPCy5.5 (BD Biosciences), CD8 V450 (BD Biosciences), CD8 PE-Cy7 (eBioscience), CD19 APC (BD Biosciences), CD20 FITC (BD Biosciences), CD25 APC (BD Biosciences), CD25 BV786 (BD Biosciences), CD71 FITC (eBioscience), CD107a PE-Cy7 (BD Biosciences), interferon γ (IFNγ) Brilliant Ultraviolet (BUV) 395 (BD Biosciences), tumor necrosis factor-α (TNFα) AF700 (BD Biosciences), Granzyme B AF647 (BD Biosciences), Perforin AF488 (Sony) and cleaved poly ADP-ribose polymerase (PARP) PE (BD Biosciences). To exclude dead cells, Fixable Viability Dye eFluor 780 or Fixable Viability Dye eFluor 506 (Thermo Fisher Scientific) was used according to manufacturer’s instructions. For staining of intracellular cytokines or cleaved PARP, cells were fixed and permeabilized using the Fixation/Permeabilization Solution Kit (BD Biosciences). After antibody staining, samples were washed using PBA and acquired on BD FACS Canto or LSR Fortessa flow cytometer and analyzed with FlowJo v10.

3T3 control or 3T40 fibroblasts were irradiated and plated as described.26 After 2 hours of adherence, CLL cells were added to the 3T3 or 3T40 fibroblasts. After 24–48 hours, CLL cells were removed from the fibroblasts and used as target cells in a cytotoxicity assay (see below). Effective CD40L stimulation of the CLL cells was confirmed by testing resistance of the 3T3 and 3T40 stimulated CLL cells to 10 nM ABT-199 for 24 hours.

Cell-lines or primary CLL samples were labeled with cell trace violet (CTV) (Thermo Fisher Scientific) according to manufacturer’s instructions and cocultured with HD PBMCs or CLL derived (autologous) T cells in different effector to target (E:T) ratios. Where indicated, T cells were pretreated for 2 hours with 1 µM concanamycin A (CMA) or dimethyl sulfoxide (DMSO) as a control. The treated T cells were thoroughly washed before using in the coculture. 10 or 20 µM of the pan-caspase inhibitor QVD-OPh (APExBIO) was added to the cocultures where indicated. Viability of the target cells was assessed using TO-PRO-3 (Invitrogen) and MitoTracker Orange (Invitrogen) using Flow Cytometry. Specific lysis of target cells was calculated as (% target cell death in treated sample - % cell death target cells in medium control)/(100 – % cell death target cells in medium control) * 100%. Samples were excluded when cell death in medium controls exceeded 50%.

PBMCs from HD or CLL patients were labeled with CTV. PBMCs were incubated in the presence of 100 ng/mL CD3xCD19 DART or stimulated with CD3 (clone 1XE) and CD28 (clone 15E8) antibodies. After 2 days activation was measured and after 4 days proliferation and activation were measured by flow cytometry as described above.

PBMCs from HD or CLL patients were incubated in the presence of 100 ng/mL CD3xCD19 DART or stimulated with CD3 (clone 1XE) and CD28 (clone 15E8) antibodies for 2 days. Brefeldin A (10 ug/mL, Invitrogen), GolgiStop (BD Biosciences) and anti-CD107a PE-Cy7 were added 4–6 hours before assessment of degranulation and cytokine production by flow cytometry as described above.

Bcl-2-associated X (BAX) and Bcl-2 homologous antagonist/killer (BAK) knockout (KO) JeKo-1 cells were generated using CRSIPR/Cas9 technology. Details regarding single guide RNAs (sgRNAs) targeting BAK and BAX have been published before.27 Each sgRNA was cloned in the lentiCRISPRv2 puro plasmid, a gift from Brett Stringer (Addgene #98290). Lentivirus was produced and used to infect the cell-lines. After selection with puromycin (Sigma) and single cell cloning, knockouts were confirmed by SDS-PAGE with BAK and BAX antibodies (Cell Signaling).

To test for normal distribution a D’Agostino-Pearson test or a Shapiro-Wilk test if n=4, was used. P values were calculated by using two-sided paired or unpaired t tests, Wilcoxon test, Mann-Whitney U test, one-way analysis of variance (followed by Tukey’s post hoc test) or Friedman test (followed by Dunn’s post hoc test). Statistical analysis was performed using Graphpad PRISM V.8.0.2 with significance set at P<0.05.

CD3xCD19 DART (JNJ-64052781 or MGD011) is an Fc bearing bispecific molecule with an approximate molecular weight of 109 kDa and is composed of an anti CD3 (clone XR32) and anti CD19 domain (BU12).24 Previously it has been shown that this DART is specific for CD19, since CD19^- cell lines are not affected.25 To assess whether this DART molecule treatment leads to robust T cell activation, age-matched HD PBMCs were cultured in presence of 100 ng/mL CD3xCD19 DART, and activation was compared with direct TCR stimulation with anti-CD3/CD28 antibodies. Robust T cell proliferation was observed when treating HD PBMCs with the CD3xCD19 DART for 4 days, equal to stimulation with anti-CD3/CD28 antibodies for CD4⁺ T cells (figure 1A). CD8⁺ T cells showed increased proliferation after treatment with CD3xCD19 DART (figure 1A). Furthermore, exposure to CD3xCD19 DART led to T cell activation as indicated by upregulation of activation markers CD25 and CD71, the degranulation marker CD107a, production of IFN-γ, TNF-α, perforin, and granzyme B in both CD4⁺ and CD8⁺ T cells (figure 1B–G and online supplementary file S1A). Enhanced expression of CD25 and the effector molecules was similar in the CD3xCD19 DART conditions compared with anti-CD3/CD28 stimulation (figure 1B–G). As a proof-of-principle that CD3xCD19 DART induces T cell-mediated killing of CD19⁺ targets, we cocultured HD T cells with the CD19⁺ mantle cell lymphoma cell-line JeKo-1, in the presence of increasing concentrations of the functional CD3xCD19 DART or a control DART molecule (CD3xFITC). JeKo-1 was chosen because of its coexpression of CD5 and CD19 and loss of TP53, also a feature of high-risk CLL.28 Exposure of JeKo-1 cells to HD T cells in a 1:1 effector to target ratio in the presence of CD3xCD19 DART caused lysis of target cells already after 24 hours, which was further augmented when increasing CD3xCD19 DART concentrations, prolonging exposure up to 4 days (figure 1H), or increasing E:T ratios (figure 1I). Exposure to the control DART molecule did not result in lysis of the CD19⁺ targets (figure 1H–I). CD3xCD19 DART mediated lysis of CD19-positive cell-lines was also observed using the Burkitt lymphoma cell-line Daudi, and was confirmed using T cells derived from different donors (online supplementary figure S1B,C).

To determine whether exposure to CD3xCD19 DART also leads to lysis of primary CLL cells, HD T cells were incubated with increasing concentrations of the CD3xCD19 DART or control DART molecule in the presence of primary CLL cells for 24 hours or 4 days. Killing of the primary CLL cells could already be observed after 24 hours in a 1:1 E:T ratio, and was dependent on the concentration of CD3xCD19 DART, the E:T ratio, and duration of exposure (4 days) (figure 2A–B and online supplementary S1D–G). Since the prognosis of CLL patients is dependent on the immunoglobulin heavy chain variable region (IGHV) mutational status, presence of TP53 aberrations and/or whether patients are relapsed or refractory after chemotherapy, we tested whether CD3xCD19 DART could induce lysis in unmutated IGHV, mutated IGHV CLL, cells derived from chemorefractory patients and TP53 mutated and/or 17 p deleted patients. Lysis on CD3xCD19 treatment was observed in all subgroups of patients (figure 2C and online supplementary figure S1H). Although, low numbers, first evidence is available that patients who become refractory on new agents such as ibrutinib are still sensitive to CD3xCD19 mediated lysis (online supplementary figure S1I). Taken together, using HD T cells, primary CLL cells can be efficiently lysed by using CD3xCD19 DART irrespective of IGHV mutation status, TP53 aberrations or sensitivity to chemotherapy.

T cells from CLL patients are known to be dysfunctional with regard to activation and proliferation.14 15 18 To test whether CD3xCD19 DART exposure could overcome this dysfunctional phenotype, CLL PBMCs were cultured in the presence of 100 ng/mL CD3xCD19 DART or anti-CD3/CD28 antibodies. As we and others have previously reported,14 18 TCR stimulation of CLL-derived T cells results in impaired proliferation and activation (as measured by upregulation of CD25 and CD71) compared with age-matched HD T cells (figure 3A,B, online supplementary figure S1J). Remarkably, addition of the CD3xCD19 DART induced robust proliferation in both CD4⁺ and CD8⁺ T cells derived from CLL patients (figure 3A), similar to HD T cells. In line with this, exposure to CD3xCD19 DART led to activation of CD4⁺ and CD8⁺ CLL derived T cells as indicated by upregulation of activation markers CD25 and CD71, the degranulation marker CD107a, production of IFN-γ, TNF-α, perforin and granzyme B (figure 3B–G and online supplementary S1J). Whereas impaired T cell activation was measured on TCR stimulation with anti-CD3/CD28 antibodies compared with healthy controls, CD3xCD19 DART treatment led to equal or even improved activation of CLL derived T cells (figure 3B, online supplementary figure S1J). Also, production of IFN-γ, TNF-α and perforin on CD3xCD19 DART treatment in HD and CLL-derived T cells was equal (figure 3D–G), yet degranulation was significantly increased in both CD4⁺ and CD8⁺ T cells in CD3xCD19 treated CLL derived T cells compared with HD (figure 3C). Interestingly, whereas both CLL and HD CD8⁺ T cells produce granzyme B on TCR or CD3xCD19 DART, CLL derived CD4⁺ T cells lack behind compared with HD derived CD4⁺ T cells (figure 3G). Nevertheless, granzyme B expression increased significantly in CLL derived CD4⁺ T cells on CD3xCD19 DART treatment compared to anti-CD3/CD28 stimulation (figure 3G). Taken together, while TCR stimulation with anti-CD3/CD28 antibodies results in poor T cell activation, proliferation and degranulation in T cells of CLL patients, treatment with CD3xCD19 DART induces fully functional T cell responses in CLL samples.

Since it was determined that CD3xCD19 DART exposure leads to enhanced T cell activation, proliferation and production of effector molecules compared with TCR stimulation, we assessed whether T cells of CLL patients are also able to lyse CLL cells. Isolated CD4⁺ or CD8⁺ T cells from CLL patients were cocultured separately or mixed (CD4:CD8 ratio of 1:1) with JeKo-1, CLL cells from an allogeneic donor, or autologous CLL cells in the presence of 100 ng/mL CD3xCD19 DART. CLL derived T cells lysed JeKo-1 cells very efficiently (figure 4A). CD4⁺ T cells alone were equally capable of lysing JeKo-1 or allogeneic CLL cells as CD8⁺ T cells, and CD4⁺ and CD8⁺ together. Although slightly lower compared with lysis of JeKo-1 cells, similar results were found for lysis of allogeneic CLL cells (figure 4A). In the autologous setting, the combination of CD4⁺ and CD8⁺ T cells showed more efficient lysis of autologous CLL cells than either subset alone (figure 4A). The more efficient killing of Jeko-1 compared with primary CLL cells might at least in part be explained by higher expression of CD19 on JeKo-1 compared with primary CLL cells (figure 4B). Coculture of CLL T cells (CD4 and CD8 cells combined, CD4:CD8 ratio as present in donor unknown), with primary CLL and increased dosages of CD3xCD19 DART could not improve killing of primary CLL cells, as a plateau in lysis of CLL cells was already reached at 10 ng/mL CD3xCD19 DART (figure 4C). In conclusion, CD3xCD19 DART treatment leads to potent lysis of JeKo-1 cells and (autologous) primary CLL cells mediated by patient-derived T cells.

The selective Bcl-2 inhibitor venetoclax results in deep responses in blood and bone marrow but seems less effective in eradicating lymph node (LN) residing CLL cells. This might be due to local upregulation of anti-apoptotic molecules such as Bcl-XL and MCL-1.29 We, therefore, assessed effectiveness of CD3xCD19 DART against CLL cells that have acquired an anti-apoptotic profile, after in vitro stimulation by CD40L, which induces upregulation of Bcl-XL and MCL-1, and leads to venetoclax resistance (online supplementary figure S2A).29 On exposure to the CD3xCD19 DART, CD40L-stimulated CLL cells were still effectively lysed by HD T cells (figure 5A). Importantly, CLL cells of a venetoclax-refractory patient could still be lysed on CD3xCD19 DART exposure (online supplementary figure S1I). Furthermore, Bcl-2 overexpressing Ramos cells, which are resistant to apoptotic stimuli such as CD95 or B cell receptor triggering,30 31 were sensitive to CD3xCD19 DART treatment (figure 5B). These results indicate that a block in intrinsic apoptosis does not prevent CD19⁺ cells from being targeted by CD3xCD19 DART.

CMA inhibited CD3xCD19 DART mediated cell death of CLL cells (figure 5C), implying that killing depends on the perforin and granzyme pathway.32 Reported mechanisms of granzyme-mediated lysis of target cells are not conclusive. Initiation of the intrinsic apoptosis pathway through cleavage of Bid, but also direct activation of caspase 3 have been described.33–35 To test dependency of CD3xCD19 DART-mediated cell death on the intrinsic apoptosis pathway, BAX/BAK KO JeKo-1 cell-lines were generated (online supplementary figure S2B). Intrinsic apoptosis blockade was confirmed by resistance to venetoclax treatment after 24 hours (figure 5D). Yet, HD T cells caused similar lysis of the BAX/BAK KO and wild type cell-lines (figure 5E), implying CD3xCD19 mediated cell death does not depend on the intrinsic apoptosis pathway. We next tested involvement of caspases by using the pan-caspase inhibitor QVD-OPh. QVD-OPh rescued venetoclax-induced but not CD3xCD19 DART mediated cell death (figure 5F,G). Similar results were obtained using CD40L stimulated CLL or JeKo-1 as target cells (online supplementary figure S2C–E). These data imply that the mechanism of DART molecule mediated T cell-induced cell death is, at least in part, non-apoptotic. This was further confirmed by the induction of PARP cleavage in JeKo-1 and CLL cells after CD3xCD19 DART exposure, which could be fully blocked by addition of QVD-OPh although the cells still died (figure 5H,I). Taken together, CD3xCD19 DART mediated lysis is dependent on granule exocytosis and can potently induce lysis in venetoclax resistant cells, which seems to be independent of BAX/BAK and caspase activity, and therefore seems non-apoptotic.